Assessing the Economic Consequences of Two Cannabis Policy Options

Home , Cannabis in Australia, Cannabis in France, Cannabis in New Zealand

Assessing the economic consequences of two cannabis policy options

Marian Shanahan, BA (Hons), MA (Econ)

A thesis submitted in accordance with the requirements for admission to the degree of Doctor of Philosophy

Drug Policy Modelling Program National Drug and Alcohol Research Centre School of Public Health and Community Medicine Faculty of Medicine, University of New South Wales Sydney, Australia

March 2011

DECLARATION OF ORIGINALITY

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………......

Date ……………………………………………......

iii

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in the whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral thesis only).

I have either used no substantial portions of copyright material in my thesis or I have obtained permission to sue copyright material: where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.’

Signed ......

Date ......

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.

Signed ......

Date ......

Abstract

The research in this dissertation addresses the question of costs and benefits of two policy options for cannabis in the context of New South Wales, Australia. The first policy is the current one, where cannabis is illegal, although a cannabis cautioning program is available for the use or possession of a small amount of cannabis. In the second policy option, cannabis is a legal but a highly regulated good. As no legalised– regulated policy currently exists, as part of this research the policy framework was developed with the objective of minimising the harms associated with the use of cannabis and with the policy itself.

The societal value and preferences for the policies were evaluated in two ways. First, through the use of a traditional cost benefit analysis (CBA) and secondly, with a discrete choice experiment (DCE). The CBA estimated important costs and benefits in monetary terms for each policy with the results presented as a net social benefit. The DCE assessed the preferences for different policies for cannabis among a representative sample of the population. Potential trade-offs between key outcomes including the prevalence of consumption, expenditures by the criminal justice system and health care systems related to cannabis were explored.

The results from the CBA indicate there is no clear difference in the net social benefit between the two policy options, although there is a potential revenue gain for government in the legalised–regulated option. The results from the DCE reveal that compared to the current policy there is a moderate preference for legalisation of cannabis among a community sample and strong preference against the complete criminalisation of cannabis. The results also demonstrate the trade-offs between different harms and benefits as well as the interactions between personal characteristics and the policy preferences.

The findings from this analysis of the costs and benefits of two cannabis policy options will start to redress some of the evidence gaps that arise when making public policy in this area.

ACKNOWLEDGMENTS

I would like to acknowledge all of those who came before me in the cannabis policy debate. I have learned much from the many passionate and articulate arguments made by both sides. I also would like to thank those who completed the many surveys on which this work was built.

This work would not have been possible without the funding from an ARC Discovery Grant and the Colonial Foundation Trust, and I am grateful for their financial support.

Throughout this endeavour many people have supported me, both at work and personally. To my friends and colleagues in DPMP and at NDARC who have listened, and provided much advice along the way, I thank you. I would especially like to acknowledge Annie Bleeker who provided a considered and enthusiastic introduction into the world of cannabis both in Australia and in the Netherlands. To the other community of people in my life, I could not have done it without you – you listened, debated, provided encouragement and stepped up when needed, often without being asked. I definitely could not have completed this work without all your support.

Thanks to my supervisors Alison Ritter, Glenn Salkeld and Karen Gerard for your insights, support, encouragement and guidance, but in particular I would like to thank Alison for making this thesis possible and for providing an enthusiastic and supportive environment in which to work.

I would like to dedicate this thesis to my father. He did not get to see the end product but he was always interested in the ‘project’ and he widely shared his pride.

Last but not least, I want to acknowledge my husband Paul’s love and support. I could not have done it without you. I love you.

Contents

Declaration of originality...... iii Copyright Statement ...... iv Abstract ...... v Acknowledgement ...... vi Contents ...... vii List of tables ...... xii List of figures ...... xv List of abbreviations ...... xvi

Chapter 1: Introduction ...... 1

Chapter 2: Policy alternatives ...... 9 2.1 International Conventions ...... 9 2.2 Possible cannabis policy options ...... 11 2.3 An overview of existing cannabis laws ...... 19 2.4 Selecting policies for comparison ...... 25 2.5 Summary ...... 27

Chapter 3: Economic theory and its application in this thesis ...... 28 3.1 Introduction ...... 28 3.2 Existing literature on economic evaluations of cannabis policies ...... 29 3.3 Various methods for evaluating the societal impact of policy changes ...... 32 3.4 Cost benefit analysis: an overview ...... 35 3.4.1 .. Decision rules of a CBA ...... 37 3.4.2 .. Methods of valuing benefits ...... 38 3.4.3 .. Steps in a CBA ...... 42 3.5 Summary ...... 50

vii

Chapter 4: Cannabis: rates and quantity consumed ...... 51 4.1 Introduction ...... 51 4.2 What is cannabis?...... 52 4.3 Current cannabis consumption ...... 55 4.3.1 .. Measuring cannabis consumption ...... 58 4.3.2 .. Estimating the number of users and quantity consumed ...... 60 4.4 Cannabis consumption under legalisation...... 64 4.4.1 .. Estimation of consumption under legalisation ...... 69 4.5 Impact of price changes and using price elasticity to estimate demand ...... 71 4.5.1 .. Price elasticity literature ...... 72 4.6 Estimating the retail value of cannabis ...... 74 4.7 Discussion ...... 75

Chapter 5: Cannabis and its impact on the CJS ...... 78 5.1 Introduction ...... 78 5.1.1 .. Pathways ...... 78 5.2 Documenting existing laws and regulations pertaining to cannabis in NSW ...... 79 5.3 The criminal justice system (CJS) as it pertains to cannabis ...... 81 5.4 Measuring and valuing the resources used in the CJS ...... 81 5.4.1 .. Literature review: Issues when estimating CJS costs ...... 83 5.4.2 .. Top-down approaches...... 84 5.4.3 .. Micro costing approaches ...... 87 5.5 NSW Police Force ...... 88 5.5.1 .. Police data ...... 89 5.5.2 .. Methods of police proceeding ...... 89 5.5.3 .. Cannabis offence data ...... 94 5.5.4 .. Survey of police ...... 95 5.6 Magistrates early referral into treatment (MERIT) ...... 101 5.7 New South Wales courts ...... 102 5.7.1 .. Court data ...... 103 5.7.2 .. Outcomes of court appearance ...... 106 5.7.3 .. Cost of court–related activity for cannabis ...... 106 5.8 Other costs: prosecution and Legal Aid ...... 108 5.9 Corrective Services ...... 108 5.9.1 .. Penalties ...... 108

viii

5.9.2 .. Costs related to penalties...... 109 5.10 Total cost to the CJS for enforcing cannabis laws ...... 111 5.11 Costs to the individual ...... 112 5.12 Discussion ...... 113

Chapter 6: The public health perspective: a legalised–regulated framework..... 116 6.1 Dealing with regulation ...... 117 6.2 The regulatory framework ...... 120 6.2.1 .. The purchaser/consumer ...... 120 6.2.2 .. The grower ...... 123 6.2.3 .. Distributor / retailer ...... 125 6.2.4 .. Government(s) ...... 129 6.3 Costs of production, distribution and sale of cannabis ...... 131 6.3.1 .. Growing cannabis under an illegal system ...... 131 6.3.2 .. Estimating the resource implications of the various regulations ...... 134 6.3.3 .. Individuals ...... 135 6.4 Grower...... 137 6.4.2 .. Estimating revenues ...... 145 6.5 The distributor and retailer ...... 147 6.5.1 .. Operational costs and complying with regulations ...... 147 6.6 Cost to government of enforcing these regulations...... 151 6.7 Cost to government of a regulatory agency ...... 153 6.8 Summary ...... 155 6.9 Discussion ...... 156

Chapter 7: Contingent valuation: quantifying stigma from a criminal offence.. 160 7.1 Introduction ...... 160 7.2 Background ...... 161 7.3 Methods ...... 163 7.3.1 .. Survey ...... 164 7.3.2 .. Analyses ...... 168 7.4 Results ...... 169 7.4.1 .. Demographics, characteristics and attitudes ...... 170 7.4.2 .. Societal WTP ...... 173 7.4.3 .. Regression analyses ...... 177 ix

7.5 Discussion ...... 180

Chapter 8: Other benefits, harms and costs related to cannabis policies ...... 184 8.1 Health effects and resource implications ...... 184 8.1.1 .. Overview ...... 184 8.1.2 .. Cannabis use disorder ...... 186 8.1.3 .. Treatment for dependence ...... 187 8.1.4 .. Mental health ...... 190 8.1.5 .. Low birth weight newborns ...... 194 8.1.6 .. Accidents ...... 195 8.1.7 .. Respiratory cancers and other lung disease ...... 198 8.2 Cognitive impairment and impact of cannabis use on educational attainment and earnings of youth ...... 200 8.3 Utility gained from cannabis use ...... 204 8.4 Productivity ...... 207 8.5 Discussion ...... 208

Chapter 9: Results of the CBA...... 210 9.1 Introduction ...... 210 9.2 Methods for dealing with uncertainty ...... 211 9.3 Results of the CBA ...... 212 9.4 Sensitivity analyses ...... 216 9.5 Discussion ...... 220 9.5.1 .. Caveats and limitations ...... 224 9.5.2 .. Conclusion ...... 227

Chapter 10: A discrete choice experiment: policy options for cannabis ...... 228 10.1 Introduction ...... 228 10.2 Theory ...... 231 10.3 Methods ...... 232 10.3.1 Model specification ...... 232 10.3.2 Survey ...... 233 10.3.3 Analyses ...... 244 10.4 Results ...... 248

10.4.1 Multinomial logit (MNL) results ...... 250 10.4.2 Mixed logit results (ML) ...... 250 10.4.3 Utility scores ...... 257 10.5 Discussion ...... 259 10.6 Conclusion ...... 262

Chapter 11: Discussion ...... 264

References ...... 272

Appendix ...... 303

Appendix - Chapter 2 ...... 304 An overview of the existing legal status of cannabis in various countries

Appendix - Chapter 4 ...... 314 Estimating current rates of use males Estimating current rates of use females Estimating new consumption under legalisation males Estimating new consumption under legalisation females

Appendix– Chapter 5 ...... 319 Cannabis – crime pathways Police Survey Description of penalties received for cannabis offences

Appendix – Chapter 7 ...... 332 Contingent valuation survey

Appendix– Chapter 8 ...... 335 Cannabis use disorder annd dependence data

Appendix– Chapter 10...... 337 DCE design from NGENE DCE survey Mixed logit code Final model results (all interactions)

Tables

Table 1: Chapters and data components and types of analyses ...... 5 Table 2: Summary of the policy paradigms ...... 12 Table 3: Summary of potential responses against those found in possession of small amount of cannabis ...... 21 Table 4: Relationship between consequences and policy options ...... 46 Table 5: Costs and (dis)benefits of cannabis policies ...... 49 Table 6: Frequency of cannabis use in NSW by age (males and females combined)* 61 Table 7: Summary of the amount of cannabis consumed per occasion of use ...... 62 Table 8: Total annual consumption of cannabis in NSW: Summary of results ...... 63 Table 9: Reasons for never/no longer using cannabis...... 67 Table 10: Per cent who stated they would start or increase cannabis use by age and current use category if cannabis was legal ...... 68 Table 11: Prevalence of use in NSW by age group with and without legalisation ...... 70 Table 12: Estimating the total consumption of cannabis consumed under legalisation framework ...... 70 Table 13: Street value of cannabis ...... 74 Table 14: Cannabis offences ...... 86 Table 15: Implications of different assumptions when allocating police expenditure . 87 Table 16: Definitions of methods by which police may proceed against offender...... 90 Table 17: Frequency of cannabis offences by age groups ...... 94 Table 18: Methods of proceeding against for cannabis offences ...... 95 Table 19: Comparing across Local Area Command: time in minutes for completion of various activities per officer ...... 97 Table 20: Police survey: Total time per encounter for each type of police encounter . 98 Table 21: Total FTEs and costs by age category, offence type and method of proceeding ...... 99 Table 22: Other police costs...... 101 Table 23: Characteristics of cannabis offences by court type (2006) ...... 104 Table 24: Frequency and distribution of cannabis offence by age category (principal and non-principal) ...... 105 Table 25: Characteristics of offenders with and without previous non-cannabis offences within past ten years ...... 106 Table 26: Outcome of court appearance by type of court (all cannabis offences) ...... 106 Table 27: Estimates of total court costs by offence type ...... 107 Table 28: Estimates of DPP and Legal Aid costs ...... 108 xii

Table 29: Penalties for cannabis offence by type of court ...... 109 Table 30: Costs of penalties ...... 110 Table 31: Total cost of penalties for cannabis offences ...... 111 Table 32: Total government expenditures on CJS to enforce cannabis laws ...... 111 Table 33: Sensitivity analyses: varying key inputs ...... 112 Table 34: Total fines for cannabis offence by type of cannabis offence and court type113 Table 35: Personal costs ...... 113 Table 36: One estimate of costs and income for a single growing operation ...... 133 Table 37: Costs to the individual as a result of the regulation ...... 136 Table 38: Estimating electricity costs for indoor growing ...... 138 Table 39: Summary of cost of growing in greenhouse ...... 139 Table 40: Additional annual costs per grower as result of regulations ...... 141 Table 41: Number of greenhouses required ...... 142 Table 42: Cost of producing dried cannabis per greenhouse ...... 144 Table 43: Potential payments to growers by price and annual production of dried cannabis ...... 145 Table 44: Unit cost for estimating costs of retail outlets ...... 149 Table 45: Cost estimates for retail outlets ...... 150 Table 46: Costs to government of enforcing regulations ...... 151 Table 47: Estimate of annual costs for regulatory agency ...... 154 Table 48: Consumer information campaigns ...... 154 Table 49: Summary of costs of complying with and enforcing regulations (range) ... 155 Table 50: Potential revenue and expenses ...... 155 Table 51: Characteristics and demographics (n=875) ...... 171 Table 52: Educational attainment...... 172 Table 53: Mean WTP for payment card, final payment and 1% trim ...... 174 Table 54: WTP to avoid a criminal record by scenario (trimmed data) ...... 175 Table 55: Wilcoxon Rank Sum (WTP) – 1% Trim...... 175 Table 56: Total societal willingness to pay options ...... 176 Table 57: Differences between those who changed their answers and those who did not ...... 177 Table 58: Regression analysis using data from the final revised stated WTP ...... 178 Table 59: Comparing payment card, open-ended and trimmed data ...... 179 Table 60: Projected increase in cannabis use disorder ...... 187 Table 61: Costs of treatment for cannabis use disorders in NSW 2007 ...... 189

xiii

Table 62: Additional episodes and costs of treatment: two estimates for CUD ...... 190 Table 63: Number of new heavy and light cannabis users ...... 192 Table 64: Number of additional persons with schizophrenia and psychosis ...... 193 Table 65: Schizophrenia and psychosis: additional health care costs ...... 194 Table 66: Additional low birth weight newborns and related health care costs ...... 195 Table 67: Number of persons in accidents attributable to cannabis use, NSW, 2007 196 Table 68: Additional health care costs related to traffic accidents ...... 197 Table 69: Estimates of losses related to fatal accidents...... 197 Table 70: Estimates of new tobacco smokers under legalisation ...... 199 Table 71: Potential costs related to additional tobacco smokers ...... 200 Table 72: Estimates of the value of a loss of educational attainment* ...... 203 Table 73: Estimating the value of pleasurable consumption ...... 207 Table 74: Summary of total costs, total benefits and net revenues for both models .. 213 Table 75: Summary of costs and benefits* ...... 215 Table 76: Sensitivity analysis of varying the costs ...... 219 Table 77: Attributes and levels ...... 235 Table 78: Policy attribute: example of effects coding ...... 246 Table 79: Basic descriptive information ...... 249 Table 80: Did people only choose or never choose status quo? ...... 250 Table 81: Mean for attributes by alternative ...... 250 Table 82: Main results...... 253 Table 83: Significant interactions and random parameters ...... 256 Table 84: Relative utilities scores –constructing scenarios for sensitivity analyses ... 258

xiv

Figures

Figure 1: The spectrum of drug control regimes (MacCoun and Reuter, 2001a) ...... 14 Figure 2: Spectrum of Drug Control Approaches ...... 18 Figure 3: Rates of recent cannabis use in Australia in those aged 14+ ...... 56 Figure 4: Percentage of NSW students who reported using cannabis at least once in past twelve months ...... 59 Figure 5: Estimated number who consume cannabis by age category and frequency of use (Methods 1 and 2) ...... 71 Figure 6: The substance-defined crime pathway for NSW (a simplified flow chart) .. 82 Figure 7: Example of an illegal growing operation ...... 131 Figure 8: Survey respondents ...... 170 Figure 9: Comparisons of age ...... 172 Figure 10: Distribution of weekly income ...... 173 Figure 11: Distribution of revised responses to WTP question ...... 174 Figure 12: Australian trends in treatment by principal drug of concern ...... 188 Figure 13: Demand curve for cannabis ...... 206 Figure 14: NSB main estimates from Monte Carlo simulation...... 214 Figure 15: Examining impact on the NSB of various assumptions regarding (dis)benefits (millions $) ...... 217 Figure 16: Percentage change in government expenditures when key resources are varied ...... 218 Figure 17: The impact of the additional (dis)benefits of tobacco smoking on the legalised–regulated option ...... 219 Figure 18: Example of one profile ...... 239 Figure 19: Rates of cannabis use among the survey respondents and the population 248 Figure 20: Distribution of coefficients in the sample for the criminal status option .. 254

Abbreviations

ADIS Alcohol and Drug Information Service AIHW Australian Institute of Health and Welfare AUD Australian dollar BCR Benefit Cost Ratio BOCSAR Bureau of Crime Statistics and Research CAN No bail court attendance required CBA Cost benefit analysis CBD Cannabidiol CBN Cannabinol CEA Cost effectiveness analysis CEP cannabis eradication program CJS Criminal justice system CI Confidence intervals COPS Computerised Operational Policing System CV Contingent valuation CUD Cannabis use disorder DCE Discrete choice experiment DRG Diagnostic related group DUCO Drug Use Careers of Offenders DUMA Drug Use Monitoring in Australia EDRS Ecstasy and Related Drug Reporting System reports EMCDDA European Monitoring Centre for Drugs and Drug Addiction FTE Full time equivalent Gm Gram HREC Human Research Ethics Committee IDRS Illicit Drug Reporting System IID independent and identically distributed Kg Kilogram LAC Local Area Command MERIT Magistrates Early Referral into Treatment ML Mixed logit

xvi

MNL Multinomial logit NDSHS National Drug Strategy Household Survey NGO Non Government Organisation NMDS National Minimum Data Set NNP Number needed to prevent NSB net social benefit NSW New South Wales PE Price elasticities QALY Quality Adjusted Life Year TDPF Transform Drug Policy Foundation RTA Roads and Transport Authority SQ Status Quo THC Tetrahydrocannabinol VOSL Value of a statistical life VOSLY Value of a statistical life year WA Western Australia WTP Willingness to pay

xvii

Chapter 1

The Journey is the Reward (Chinese proverb)

Chapter 1 : Introduction

Cannabis use is widespread in the Australian community, with one in three (33.5%) Australians aged 14 and older having consumed it in their lifetime, and almost one in ten having done so in the last year (referred to as ‘recent’ use) (Australian Institute of Health and Welfare, 2008a). Among recent users, 14.9% reported using daily and 19.8% used at least weekly (Australian Institute of Health and Welfare, 2008a). This consumption of cannabis occurs while the possession and/or use of cannabis remains illegal, although various Australian jurisdictions have laws providing for some police discretion for handling individuals found in possession of a small amount of cannabis.

Changes to the legislative status of cannabis are frequently proposed with some advocating for liberalisation of the laws, including legalisation, while others advocate for total prohibition. Still others recognise the limitations of existing data and argue additional research is required as a prerequisite to the rational consideration of any change in public policies for managing cannabis (Strang et al., 2000; MacCoun and Reuter, 2001b; Hall and Pacula, 2003; Hall and Lynsky, 2009). An ongoing challenge in determining the ‘best’ public policy for cannabis is the conflict between the rights of the individual and the public good. For some there is no conflict, for example John Stuart Mill’s principle is that the only time the state should intervene against a person’s will is to prevent harm to others (Leitzel, 2008). Even with exceptions to protect children, Mill’s interpretation of this harm principle would arguably not have allowed the prohibition of drugs (Leitzel, 2008). On the other hand, the collectivism of public policy measures often override individual autonomy even though there may be little benefit to that individual but great benefit to the population (Baum, 2008). Some argue that invoking the precautionary principle as a method of protecting society in the face of the lack of scientific knowledge about the impact of the drug laws actually does more harm than good (Nutt, 2009). Still others argue the laws are paternalistic and an infringement of individual rights (van Ree, 1999).

1 Chapter 1

The debate on how prohibitive drugs policy should be, in particular cannabis policy, has previously been centred on arguments associated with liberty and harm. It is often emotive and not fully informed by evidence (Nadelmann, 1989). Advocates of legalisation have argued that maintaining the criminal status of cannabis encourages criminal activity, necessitates contact with illicit drug sellers, leads to individuals acquiring criminal records/penalties for use of small amounts of cannabis, results in taxation losses, and increases the costs of enforcement (Friedman, 1972; Wodak et al., 2002; Englesman, 2003). Advocates of total prohibition have countered by arguing that prohibition leads to higher prices, lower consumption of cannabis, better health status, and improved productivity (Tonry and Wilson, 1990; Wilson, 1990; Drummond, 2002). Another perspective gaining prominence in the literature has been a focus on public health measures with the goal of minimising the social costs of illicit drugs (Nadelmann, 1992; Haden, 2004; Haden, 2008; Room et al., 2008; van den Brink, 2008; Macleod and Hickman, 2010). Such an approach may result in a different legal framework for each illicit drug; it may legalise and heavily regulate a drug; or limit who has access, and under what conditions they have access to the drug while also providing prevention and education programs on safe methods of use. Importantly it might be a middle ground between the advocates for the free market and those who support criminalisation (Haden, 2004).

Although the process of policy development is complex, with scientific evidence being one of a number of influences (Lin and Gibson, 2003), the formation of cannabis policies have previously been made largely with only partial or no economic evidence (Wodak et al., 2002; Kisely, 2005). This is despite the fact that economic analyses can provide very useful information to the decision-making process. Little is known about the economic implications of policy interventions pertaining to cannabis consumption (MacCoun and Reuter, 2001b; Miron, 2005; Pollack and Reuter, 2007) and hence there has been limited information to inform cannabis policy.

The economic theory of consumer behaviour assumes that individuals are rational decision-makers who seek to maximise their preferences. In a market economy, the consumer’s sovereignty is often considered paramount, that is, the consumer is considered free to choose. However, often the state intervenes when there are perceived externalities related to this behaviour. These externalities may be related to harms that occur to those who do not participate (in this case those who do not use cannabis) or 2 Chapter 1

when participants do not recognise the self-harm (denial, lack of information). With cannabis (and other illicit drugs) the consumer’s ability to chose has been removed (or at least moved to criminal status) by the state, as has any assurance over quality of the product which accompanies legally sanctioned goods. On the other hand, alcohol and tobacco despite both being responsible for significant health and social harms (Collins and Lapsley, 2008) are legal and the question is not whether the drugs should be legal or not but rather how to develop more effective interventions to decrease harms while permitting adults to consume if they wish.

Previous research in this area has not attempted to examine fully the societal costs and benefits of various cannabis policies but rather focused on one or two impacts of the policy change such as on the criminal justice system (Brooks et al.; Baker and Goh, 2004; Bates, 2004; Miron, 2005; Kilmer et al., 2010); the impact on potential taxes (Miron, 2005; Kilmer et al., 2010 ), or the potential impact on use (Donnelly et al., 1999; Cameron and Williams, 2001; Williams, 2004; Williams and Mahmoudi, 2004; Miron, 2005; Williams and Skeels, 2006a; Kilmer et al., 2010). Others have focused on treatment outcomes (French et al., 2002a), or the impact of cannabis on driving (Fergusson and Horwood, 2001a; Laumon et al., 2005; Mann et al., 2007; Mann et al., 2008) but none appear to have combined the wider costs and benefits of cannabis policies, nor has any research included the individuals’ costs and benefits.

Through the estimation of the current societal costs related to cannabis and the estimation of relative costs and benefits of two separate cannabis policy options, this research will address this gap. An illicit drug market will never deliver a socially optimal solution with respect to drug policy as the market is unconcerned by the distribution of resources and effects (equity) and failures in the market (including externalities). CBA is one approach to assessing public policy decisions in the absence of a market solution. This work will be undertaken using a conventional cost benefit analysis (CBA). In doing this, the evidence-base for cannabis-related policy decision- making in Australia will be expanded and several contributions are made to the literature.

The result of a traditionally estimated CBA is the net social benefit, a number (or a range of numbers), which if greater than zero, suggests that the new policy relative to the status quo is a more efficient use of societal resources. The validity of this

3 Chapter 1

assumption is contingent on all essential costs and benefits being included in the model. Another way of assessing societal preferences is to quantify stated choices for various policies using a discrete choice experiment. Such an experiment was conducted and the results are compared to those of the traditional CBA.

A multidisciplinary approach is required to address the complexity of this topic. While the foundations are in health economics, the work here crosses into economics of the firm, the application of economics to the criminal justice system, criminology, epidemiology and the quantification of drug use, and the study of regulations.

The remainder of this introductory chapter serves as a guide to the thesis. Each chapter is listed in Table 1 with a brief description of its purpose, the type of data used in each chapter and the method of analyses. Most, but not all chapters, start with a literature review pertaining to the topic of that chapter. Chapters 2 and 3 construct the rationale for selecting the policies for comparison, and the use of CBA as method of analysis.

Chapters 4 through 9 generate, and finally aggregate the data for the two alternative policies in the CBA model. In any study of policy a comprehensive approach to detecting the policy effects is required. In the case of cannabis policy, there are numerous policy impacts. These include changes in whether a criminal record results and changes in use of cannabis, for example. In the case of the latter, changes in use of cannabis will need to be explored in terms of consequences of use. Some impacts are direct and some indirect. An example of an indirect effect is the number of dependent cannabis users. The rate of dependence among users is assumed constant with the policy change, but the numbers who consume cannabis are shown to increase with the policy change thus the number of dependent users increase, whereas the relationship between criminal record and policy are directly related.

The objective of Chapter 2 is to provide an overview of existing cannabis policies internationally in the context of International Treaties and Conventions for cannabis. The review of existing policies draws heavily on some of the many recent reviews describing cannabis laws and how they are implemented for many western democratic countries and a few Asian and African countries (European Monitoring Centre for Drugs and Drug Addiction, 2007; Room et al., 2008). From this review there were three fundamental messages: i) in many countries the actual enforcement of laws differs

4 Chapter 1

from what is on the legal statutes; ii) as a result of cultural, societal, and legal system differences the usefulness of any micro level comparisons across different countries with different systems would be limited; and iii) there is no country in the world that has fully legalised cannabis to provide a case study.

Table 1: Chapters and data components and types of analyses Chapter Purpose of chapter Main data Type of data/ analysis 1. Introduction • Briefly introduce topic, • Not applicable overview of chapters 2. Cannabis policy • Review cannabis policy • Not applicable literature; select policies for comparison 3. Economic • Review economic • Not applicable theory and CBA theory; describe methods of CBA; review literature on economic evaluations of cannabis policy 4. Cannabis • Generate frequency and • 2007 NDSHS* • Analysis of unit consumption quantity of current and • Published data on record data estimates projected cannabis use cannabis per joint • Descriptive and prices statistical analysis

5. Criminal justice • Estimate the resource • NSW police data • Primary data system implications of • Police activity collection enforcing current survey • Secondary data cannabis laws, • NSW court unit • Descriptive including: record data statistical o Police resources for analysis enforcing cannabis offences o Quantification of Court and Corrective Services activities attributable to cannabis offences 6. Legalised– • Develop and cost a • Unit costs • Published and regulated legalised and regulatory • Estimates of grey literature framework framework for cannabis production costs • Secondary data for NSW • Licensing and analysis enforcement costs 7. Willingness to • Quantify societal • Internet survey • Primary data pay to avoid willingness to pay to data collection stigma avoid stigma from a • Non-parametric criminal record analysis • Regression analysis (OLS/WLS) 8. Other benefits, • Estimate the value of the • Consumption data • Secondary data harms and costs additional consumption from Chapter 4 analysis of cannabis under • Costs of treatment legalisation • Rates of psychosis • Quantify health care • Educational

5 Chapter 1

expenditures related to attainment cannabis dependence, and other health care consequences • Estimate the impact of the increased use by adolescents on educational impact 9. Results of the • Summation of results • Chapters 4 to 8 • Descriptive CBA Calculate net social analysis benefit (NSB) & benefit • Monte Carlo cost ratio (BCR) simulation 10. Discrete choice • Develop and conduct a • Internet survey • Primary data experiment survey to quantify data collection societal preferences for • DCE survey cannabis policies • Descriptive • Regression analysis: MNL and ML 11. Discussion • Discusses the overall • Not applicable findings and contrasts results from Chapters 9 and 10 *NDSHS National Drug Strategy Household Survey

After reviewing the literature, two distinct cannabis policy options (current NSW policy and a legalised cannabis policy) were chosen for the cost benefit analysis (CBA). The next step in Chapter 2 was to review other proposed legislative frameworks to assist with the development of the criteria for the legalised policy option. Here again there were important lessons—there is no single agreed upon model for legalising cannabis and none of the proposed models had moved forward to actually consider, if cannabis were to be legalised, what would be involved in developing and enforcing such regulations. As a result, it is apparent that while the literature had much to offer, in order to be included in a CBA, a detailed regulated–legalised framework for cannabis needed to be developed (see Chapter 6) within this thesis.

Chapter 3 provides a review of the existing literature on the economic impact of cannabis policies and methods of evaluating complex policies. An argument is made for using CBA as the evaluation method and a description of the economic theory underpinning it is provided. The chapter concludes with an outline of the CBA study to be conducted in this thesis including documenting the types of costs and benefits to be included.

The purpose of Chapter 4 is to provide an estimate of the current consumption (prevalence and quantity) of cannabis in NSW and projections of future use under the

6 Chapter 1

legalisation. Key findings from this chapter are used throughout the thesis to populate the CBA model including when establishing the size of the market and determining regulatory costs in Chapter 6. Estimating the quantity of cannabis consumed, in the absence of sales data, requires an estimation of the prevalence of use, frequency of use, and quantity consumed on each occasion of use. This was done using a variety of data sources. The primary data source was the Australian National Drug Strategy Household Survey data (2007) from which the number of users, frequency of recent use, their stated behaviour change should cannabis be legalised, and the number of joints consumed on a given use day were obtained. The rates of use of cannabis were also used to project changes in educational attainment and subsequent value of lost wages among those who initiate cannabis use as adolescents, and to estimate the increased personal benefit from the use of cannabis following the change of legal status.

Chapter 5 investigates the economic costs of enforcing current cannabis policies in NSW. This chapter is in two parts; a review of the existing evidence and a costing study. The review critiques studies which have attempted to quantify the criminal justice costs pertaining to enforcing cannabis laws. The costing study estimates the costs of those offences which are classified as ‘substance defined’ in Pernanen’s (2002) pathway. The costing study, using micro costing methods represents a first attempt to quantify policing costs related to cannabis, and in addition, uses one year of court data to assess the costs of court and any subsequent sanctions.

Chapter 6 describes the characteristics for one potential formulation of a legalised– regulated market for cannabis. The underlying framework adopted here for the legalised–regulated model attempts to minimise personal and social harms related to cannabis. As such, this framework takes a public health approach, and does not treat cannabis as if it were a harmless substance but rather as a psychoactive drug with the potential for causing dependence in some users, impacting decision-making and ability to drive and operate machinery while under the influence. Various regulations for the consumer, grower and retailer are presented in Chapter 6, as are the costs to each of the stakeholders, and to government for enforcing these regulations.

Chapter 7 explores one of the personal benefits of moving from a criminalised to legalised policy. A contingent valuation study was conducted to value stigma from a criminal record. This study quantifies the societal value of stigma related to a criminal

7 Chapter 1

record for the possession or use of cannabis. This empirical chapter is a rare example of the results of a contingent valuation study being incorporated into a CBA (Borghi, 2007).

In Chapter 8 other costs and benefits of the cannabis policies are considered. This chapter draws heavily on the literature and work done elsewhere but is essential to complete the compilation of the costs and benefits. It includes estimates of health care costs under the current legislative regime, and then projects an increase in demand for treatment for dependence as well as other health consequences as a result of cannabis use. These health consequences include schizophrenia and psychosis, low birth weight newborns whose mothers were using cannabis, and the consequences of traffic accidents. Benefits include the impact on educational attainment and change in productivity among young cannabis users. The value of the additional cannabis consumed was included as a proxy for additional benefit gained.

Chapter 9 presents the total costs and benefits and the net social benefit for the two policy alternatives. Monte Carlo simulations are conducted to provide potential ranges for the results. The results of the sensitivity analysis are presented and the limitations and the interpretation of the findings of the CBA are discussed.

Chapter 10 sets out to explore an alternative way of valuing policy through quantifying societal preferences for different types of cannabis policies using a discrete choice experiment (DCE). The DCE explores the feasibility of using this approach to elicit preferences for different policies for cannabis while considering the various trade-offs between health harms, criminal justice expenditures, rates of cannabis use and the location of purchase. Four policy frameworks were included (criminalisation, cannabis cautioning, a civil penalty for cannabis use, and the legalisation and regulation of cannabis). This chapter is the first use of DCE in the drugs policy field.

The final chapter contrasts the differences in findings between the traditional CBA and the DCE and discusses the implications of the results.

8 Chapter 2

Chapter 2 : Policy alternatives

Prior to the discussion of economic theory, its relevance to this thesis and the description of the CBA model in Chapter 3, this chapter sets out the context of the two policies selected for comparison. The two policies are: 1) cannabis is legalised but regulated, and 2) cannabis is illegal but there are mediating diversion programs such as cannabis cautioning for possession and use offences. This chapter provides an introduction to International Drug Conventions within which cannabis policies exist, and reviews various policy options including notional options suggested in the literature and those implemented in various international jurisdictions. The chapter concludes with the motivation for the two options compared in this thesis.

Under prohibition, all activities related to the use of cannabis are illegal, and all persons engaging in the activity may be prosecuted under criminal law, and receive criminal penalties and a recorded criminal conviction if found guilty (McDonald et al., 1994). The rationale often provided for this approach is that tougher police responses should decrease the availability of cannabis and lead to increased prices.

Those arguing for less punitive policies make the point that cannabis offences cause crowding of the criminal justice system as evidenced by the data showing that over 70% of drug-related crimes in Australia are for cannabis (Australian Crime Commission, 2007). Additionally, increased enforcement changes how drugs are sold for instance, by forcing supply activities indoors where it is difficult and expensive to enforce or by improving some neighbourhoods but shifting problems elsewhere. Others make the point that prohibition increases violence associated with drug markets whereas disputes in a legal market are usually settled in a court (Nadelmann, 1992; Health Officer Council of British Columbia, 2005). And finally, it is contended that prices do not appear to have increased (Clements, 2004; Health Officer Council of British Columbia, 2005; Rolles et al., 2006) and cannabis remains easy or relatively easy to obtain (Australian Institute of Health and Welfare, 2008a).

2.1 International Conventions

There are a number of international treaties on illicit drugs, under which cannabis (marijuana, hashish and cannabis oil) falls. Australia is a signatory to these treaties

9 Chapter 2

(McLaren and Mattick, 2006). The first international control occurred when cannabis was integrated into the ‘International Opium Convention’ in 1925 after which cannabis use, possession, production, distribution and sale were subject to criminal controls for all signatory countries (Room et al., 2008). Then, under the 1961 United Nations Single Convention on Narcotic Drugs, cannabis was elevated to the status of a substance whose properties might give rise to dependence and presents a serious risk of abuse (Ballotta et al., 2008). Being a signatory to this convention allowed for the adoption by countries of any special measures of control deemed necessary, including prohibition of use (Ballotta et al., 2008) conditional on constitutional limitations (Room et al., 2008). As the original wording was seen as giving some countries too much leeway, the United Nations Convention Against Illicit Traffic of 1988 stipulated that signatories establish a criminal offence for possession of drugs, including cannabis, for the purposes of trafficking, and for the possession for personal consumption (European Monitoring Centre for Drugs and Drug Addiction, 2007).

The point is made by Ballotta and colleagues (2008) that while the conventions allow for flexibility of response, which may be conditional on existing legal systems and constitutional limitations, jurisdictions are meant to respond in good faith in the stringent control of cannabis (and other psychoactive drugs). They suggest that there is “..no doubt about the severity requested towards cannabis and it is evident that signatory countries cannot allow non-medical use of cannabis, such as in a hypothetical legalisation regime, without renouncing the UN Conventions. They must set measures to discourage, prevent or — if considered necessary — prohibit and punish personal use of cannabis”. (p 104)

In 2009 the International Narcotics Control Board indicated that in its opinion:

“[T]he 1988 Convention requires that illicit possession of controlled substances must be prohibited, but it does not require criminal prosecution for small quantities. At times, drug possession can serve as a pretext to detain an otherwise dangerous or suspect individual, but otherwise, the law must allow for non-custodial alternatives when a police officer stumbles upon small amounts of drugs. It is important that the incident be documented and the opportunity availed to direct the user to treatment if required, but it is 10 Chapter 2

rarely beneficial to expend limited prison space on such offenders.” (INCB, 2009)

In summary, under these conventions, cannabis is subject to criminal controls for the cultivation, distribution, sale, and potentially for possession and use. However, there are ongoing debates as to what is actually meant by these conventions as it pertains to possession and use, and currently countries choose, under their own constitutional requirements, to interpret the international treaties differently.

Before proceeding further it is helpful to be clear on some terminology. Commonly used words to describe approaches to cannabis include ‘criminal’, ‘de facto’, ‘depenalisation’, ‘decriminalisation’, and ‘legalisation’. In this thesis, the term ‘legalisation’ is used to describe the situation where the use, supply and cultivation of cannabis would be completely legal for adults. The terms ‘decriminalisation’ and ‘depenalisation’, which are often used interchangeably, are not used herein to describe specific policies, but rather descriptions of the actual policies are used (i.e. civil penalties, cautioning, not arresting users, referring to treatment). The term ‘de facto’ is often used to describe the status of cannabis policy in the Netherlands and will be described in more detail later in the chapter. ‘Criminal’ is used to refer to those situations where there are no mediating policies for cannabis—that is when all types of cannabis offences are subject to a criminal charge.

2.2 Possible cannabis policy options

There have been many reviews and task forces exploring policy options for cannabis, each has provided a number of potential policy regimes for cannabis and other illicit drugs. This section provides an overview of some of these notional policy options (McDonald et al., 1994; MacCoun and Reuter, 2001a; Haden, 2002; Englesman, 2003; Haden, 2004; Health Officer Council of British Columbia, 2005; Rolles et al., 2006; Rolles, 2009; Law Commission, 2010). The policy options are presented in a framework (Table 2) with the various options listed from high to low on the level intensity of restriction with an attempt to line up similar legislative options; but this does not suggest that they are necessarily equivalent.

Table 2: Summary of the policy paradigms Australian National Task MacCoun (MacCoun et King County (King (Nadelmann, 1989) Transform Drug BC Public Health Force on Cannabis al., 1996; MacCoun and County Bar Association, Policy Foundation / Hayden (Haden, (McDonald et al., 1994) Reuter, 2001a) 2005) (Rolles et al., 2006; 2002; Health Rolles, 2009) Officer Council of British Columbia, 2005) Total prohibition Pure prohibition Prohibition Prohibition Criminalisation Prohibition with civil penalties Prohibitory prescription Prohibition with Depenalisation adaptations such as drug courts, de-policing, diversion, cannabis expiation notices Maintenance Safe administering and Medical De facto prescription of hard drugs prescription decriminalisation Partial prohibition Regulatory prescription Over-the-counter Decriminalisation pharmacy sales Positive licence Licensed premises Prescription Negative licence Legalising cannabis by Right of access Coffee shops Legalisation with licensing and regulating, model (Dutch style) product taxing, forbidding restrictions advertising Regulation Adult market Expand legalisation to all Legalisation with drugs product restrictions and market restrictions on consumer Free availability Free market Supermarket model Unlicensed sales Free market

12 Chapter 2

There are considerable similarities across the various regimes, with most ranging from prohibition through to legalisation with some variation in the intermediate options provided. While there are other frameworks in the literature, these examples provide an overview of the most commonly suggested options.

In 1992, the Australian National Task Force on Cannabis was established by the National Drug Strategy Committee and it subsequently presented five possible legislative options for cannabis: total prohibition, prohibition with civil penalties for minor offences; partial prohibition; regulation; and free availability (McDonald et al., 1994). Each is described briefly:

a. Total prohibition—Under this policy cannabis use, possession, cultivation, importation, sale and distribution would be a criminal offence with large scale growing, importation and trafficking incurring larger penalties.

b. Prohibition with civil penalties—Under this option a civil fine would apply to possession and cultivation of small amounts of cannabis for personal use while criminal sanctions would still apply to larger quantities and trafficking. Such a policy recognises that criminal prosecution encroaches into individual rights and freedoms while allowing for a policy which, although recognising the harms of the cannabis and discouraging its use, avoids some of the disadvantages of total prohibition.

c. Partial prohibition—This policy would attempt to control production and distribution whist avoiding the costs of criminalising the use of the drug. There would be no penalties for personal use but sales or cultivation would be prohibited. There may also be other restrictions such as age restrictions, limits on the potency of cannabis, restrictions on use in public, and on the amount which could be cultivated.

d. Regulation—A regulated market would involve a regulatory structure for some or all of the following: cultivation, production, and distribution. The report observed there is no existing structure for regulating cannabis but there is much to learn from the regulatory framework for tobacco, alcohol, firearms and explosives.

13 Chapter 2

e. Free availability—Under this option there would be no regulation or controls on the growing, selling or using of cannabis. It is of note that some discussions of free availability equate this to the purchase of coffee – but even coffee has some regulations, quality control and so on.

MacCoun and Reuter (2001) take a broader approach in their generation of a framework designed to encompass all illicit drugs. As a result some regimes may not be relevant for cannabis but for completeness all are presented here. The eight regimes cover the strictly legal options but also include potential medical options such as prohibitory prescription or regulatory prescription (see Figure 1) with some overlap across the options.

Figure 1: The spectrum of drug control regimes (MacCoun and Reuter, 2001a) Regime Model

Pure prohibition—no use allowed legally

Prohibitory prescription—only available for narrow therapeutic purpose

Maintenance—prescribed for relief of addiction; strictly administered Prohibitory

Regulatory prescription—self-administered under prescription

Positive licence—available for any adult with a licence

Prescription Negative licence— available for any adult for whom the privilege has not been revoked

Regulatory Regulated adult market— available to any adult e.g. alcohol

Free Market—available to any individual ,e.g. caffeine

Under the prohibitory regime, the policies move from the total prohibitory (where no use is legally allowed) through to a model where all non-prescribed use is prohibited. A prohibitory prescription model might be implemented for medical use of cannabis, where cannabis may be prescribed, subject to Therapeutic Goods Administration approval, for terminally or seriously ill persons for the anti-spasmodic, anti-emetic or analgesic effects but all other use would be prohibited. In the alternative prescription regimes such as the maintenance model, regulatory prescription would likely not be applicable for cannabis as there is currently no maintenance treatment available.

14 Chapter 2

Two licensing models are presented in the regulatory regime — the positive licence model (originally suggested by Kleiman, 1992) requires that an individual demonstrates knowledge of safe use in order to receive a licence to use, possess, supply or grow cannabis. A negative licence model recognises the right to consume, but that right can be removed. The regulated adult market controls to various degrees the dimensions of production, supply and use.

King County Bar Association in Seattle, Washington undertook a three-year project to examine the history of drugs and drug laws in the USA, international lessons, the international legal framework, and alternative legal models (King County Bar Association, 2005). The options presented in their report incorporate several policies already outlined but in addition include options already being employed such as drug courts, and police officers disregarding possession for personal use. The legal options in their report are:

a. Total Prohibition with:

o Drug courts o De-policing—police disregard possession of small amounts of drugs o Diversion o Cannabis expiation notices. b. Safe administration and prescription of ‘hard’ drugs.

c. Legalising cannabis but licensing and regulating commerce, taxing sales of cannabis but forbidding advertising.

d. Expand (c) to all drugs with state authorities determining the laws for production, distribution, and sales (King County Bar Association, 2005).

The Transform Drug Policy Foundation (TDPF) report begins with the premise that prohibition is a policy that is unworkable and will ultimately be replaced by a different paradigm with government regulating and controlling markets (Rolles et al., 2006; Rolles, 2009). The TDPF review was prepared to support this premise and provides arguments for those who are supporting their advocacy role. The TDPF currently recommends that psychoactive drugs be subject to tight legal and social regulations in recognition of the potential harms of psychoactive drugs, with some activities remaining prohibited and provide the following policy options, many of which are similar to those

15 Chapter 2

listed above. Different options would be used for different drugs depending upon their potential harms.

a. Medical prescription (possibly involving supervised use) for the most risky drugs. b. Over-the-counter pharmacy sales from qualified pharmacists. c. Licensed sales (as with off-licences for tobacconists) with various available tiers of licensing conditions applied as appropriate. d. Licensed premises (i.e. pubs or Dutch-style coffee shops). e. Unlicensed sales (similar to coffee) for low risk drugs (Rolles et al., 2006; Rolles, 2009).

Nadelmann (1992) approached the issue of legal status of drugs with the question “how do we best maximise the benefits of the free market while minimising the risks and how do we best retain the advantages of drug prohibitions while minimising the direct and indirect costs?” He also makes the point that legalisation and prohibition are not simply polar extremes, for example, there are restrictions on the sale of alcohol and the purchase and use of tobacco is becoming increasingly restricted. Rather the concepts of “able to purchase the good over the counter” versus “requiring permission from a doctor or government body to obtain the substance” and “relying on public health policies and education” versus using “criminal sanctions to control the misuse of drugs” are employed to distinguish between legalisation and prohibition.

Potential policies are described as on a spectrum with strictly prohibitionist and highly punitive policies on one end and an unregulated market on the other end with a wide range of regulated structures in the middle. Nadelmann (1992) offers three policies between which there will be other variants:

a. Prohibitionist mode— as described above.

b. Right of access model—such a model recognises the consumers’ right to purchase the drug while restricting the ease of availability through mechanisms such as price or the necessity to obtain prescriptions. The ‘mail order’ model is one method suggested. It restricts access while allowing the right of purchase. In this model, adults are allowed to possess small amounts of any drug for personal consumption but are also entitled to obtain it from a reliable, legal,

16 Chapter 2

regulated source which is responsible and liable for the quality of the product (Nadelmann, 1989). It is argued that such a model retains the essence of the legalisation regime, diminishing the impact of the black market and the harms from the unregulated production of illicit drugs as well as eliminating the infringements on individual freedoms (Nadelmann, 1989). Such a model permits the introduction of public health measures such as those which have been introduced for alcohol and tobacco.

c. The supermarket model—such a model would see drugs widely available and being purchased at prices that reflects the retailers’ costs, a reasonable profit margin and sales taxes.

A group of Canadian public health physicians expanded upon the public health approach articulated by Nadelmann. Their argument is presented using a graphical approach, first portraying psychoactive substance use on a spectrum (not shown) from beneficial use through problematic use to chronic dependence and then secondly, by illustrating the direction of harms (Health Officer Council of British Columbia, 2005).

The point is made that as the intensity of regulatory restrictions increase along the solid black line (Figure 2) from legal to total prohibition, the negative health effects decrease (at a population level) but as criminal punishments increase the social drug-related harms increase.

The authors suggested that the existing situation in British Columbia for legal drugs such as alcohol and tobacco is ‘current a’ and illicit drugs is ‘current b’. These authors propose that drug control policies should aim to reduce harmful use, to minimise negative health effects to the individual and to limit secondary drug related harms to society. They proposed that shifting regulations for licit substances to the right ‘proposed (a)’ in Figure 2 and regulations for illicit drugs to the left ‘proposed (b)’ would result in more control on licit substances in terms of where, when and by whom they are used, thereby decreasing the population level harms.

17 Chapter 2

Figure 2: Spectrum of Drug Control Approaches

Current: Tobacco & alcohol: Current: Illicit drugs - organised crime, for-profit supply; private industry, marketing to unregulated, no age limits. Proposed - becomes increase demand, age restrictions, government restricted supply; regulated for quality, purity location restrictions. Proposed: potency, age and other restrictions, access to health for increased regulations harm reduction and treatment

Current (a) Current (b) Proposed (a) Proposed (b)

Legal Prohibition

Decreasing negative health effects Increasing negative social drug- at the population level related harms

Intensity of regulatory restriction Intensity of criminal punishments increases increases

Source: (Health Officer Council of British Columbia, 2005), Adapted from Figures 2 &3)

The move to the left for currently illicit drugs, would see them becoming legal for possession and use but available only from a tightly restricted government monopoly supplier with promotion or advertising not permitted, age restrictions on use, as well as regulations for purity, quality, and potency. In addition, a full range of public health strategies including assessing harms, disease surveillance, monitoring and treatment of harms, policy development, additional resource allocation, building of public information, and regulation of service and products would need to occur (Health Officer Council of British Columbia, 2005).

In summary, this review provided descriptions of various policy options and some insights as to where each are situated on the level of restriction. While individually the policy options provided limited details on what a given policy should entail, taken together they provide a range of ideas to consider when weighing the preferred characteristics of a cannabis policy. For example, Nadelmann (1992) and Haden (2004, 2006) provided concrete ideas for policy which incorporate public health objectives; the TDPF suggestions clearly articulate the potential for a highly regulated market; and

18 Chapter 2

MacCoun and Reuter (2001) and Nadelmann (1992) pose the idea of a licence for users. These and other details of the various policy options will be incorporated in Chapter 6.

2.3 An overview of existing cannabis laws

Examining cannabis policies across a number of jurisdictions reveal a number of significant points. What becomes immediately apparent is that there is no ‘common’ policy across countries and in some instances even within a country. For example, in Australia, United States and Switzerland policies vary by state or canton. Secondly, cultural, societal, and legal system differences across countries make any inter-country comparisons of cannabis policies and their potential outcomes difficult. In addition, in many countries the actual enforcement of laws differs from what is on the statutes and sorting this out is difficult if not impossible.

Some jurisdictions, in the belief that more harm may occur as a result of using law enforcement (particularly imprisonment) to manage drug offences, have begun to introduce harm-reduction strategies alongside a prohibitionist stance. Some of these strategies include: diversion, cautioning, treatment, drug courts and civil penalties (King County Bar Association, 2005; Room et al., 2008). The types of responses available to the courts, prosecutors or police for possession of a small amount of cannabis are listed in Table 3 for a number of countries in Europe, the Americas, and Oceania.

Despite its classification as a narcotic drug by the United Nations, how cannabis is controlled or managed varies considerably across countries and remains highly controversial (European Monitoring Centre for Drugs and Drug Addiction, 2007) with several examples of inconsistencies between the laws and how the laws are interpreted or implemented (European Monitoring Centre for Drugs and Drug Addiction, 2007). The most obvious example is the Netherlands, where the ‘main’ drug law is the Opium Act (European Monitoring Centre for Drugs and Drug Addiction, 2007) which is strictly prohibitionist in nature. The sale and supply of listed drugs (including cannabis) are illegal with possession, manufacture and cultivation included under this law (Kerley, 2004). Possession is subject to imprisonment for a maximum of six months. However, arresting and criminalising users possessing small quantities for personal use of any drug is not regarded as a priority for law enforcement. This is specified mainly by guidelines issued by the Office of the Public Prosecutor (European Monitoring Centre

19 Chapter 2

for Drugs and Drug Addiction, 2007). Under these guidelines the behaviours of use and possession of five or less grams of cannabis are not criminalised in the Netherlands although drug use is not permitted in schools or on public transportation. ‘Coffee shops’ have emerged as the ‘official/unofficial’ sales channel for cannabis, albeit under strict conditions (European Monitoring Centre for Drugs and Drug Addiction, 2007). This is commonly referred to as de facto legalisation. However, supply and cultivation of cannabis remain illegal and in fact are resolutely prosecuted with a maximum penalty of four years imprisonment and/or a €67 000 fine for importing or exporting any quantity of cannabis (European Monitoring Centre for Drugs and Drug Addiction, 2007). This leaves the Netherlands in the position where it is legal to buy cannabis but not legal to grow or supply it. Each cannabis shop is limited to 500 grams in the shop at a given time (European Monitoring Centre for Drugs and Drug Addiction, 2007). Sale or possession of more than 30 grams of cannabis may result in a penalty of up to two years for manufacture, including cultivation, transportation, sale, possession and storage (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Differences between the intent of law and actual outcomes are also evident in Austria where the possession of cannabis is illegal, and anyone who possesses, produces, or supplies drugs to somebody else can be sentenced to up to six months in prison or be required to pay a fine. Despite this it appears that prosecution will not occur if the amount possessed is small, not for trafficking purposes, and does not involve a minor (Room et al., 2008). In France, cannabis possession is a criminal offence, with no legal distinction between different narcotic substances although judicial authorities have the right to decide how to deal with offences, leading again to differences between the law and its implementation. Drug users in France might be arrested and charged or simply held in the police station up to 48 hours with most released after a few hours (European Monitoring Centre for Drugs and Drug Addiction, 2007). Differences between the laws on the books and implementation also appear to exist in Belgium where the laws were changed in 2001, to introduce some consistency in the consequences for the possession of cannabis for personal use. While cannabis remains illegal, the civil fines are meant to be €75–125 for first time possession. However it seems that often a warning may be given if the offender attends treatment (European Monitoring Centre for Drugs and Drug Addiction, 2007; Room et al., 2008).

20 Chapter 2

Table 3: Summary of potential responses against those found in possession of small amount of cannabis Prison Treatment Fine (Civil) No prosecution Europe Austria ✔ ✔ ✔ Belgium ✔ ✔ (Civil) Britain ✔ ✔ ✔Sanctions/ cautioning Czech ✔Sanctions Croatia ✔ ✔ Denmark ✔ ✔ Finland ✔ ✔ ✔ France ✔ ✔ ✔ ✔ Held in police station for up to 48 hours / warnings Germany ✔ ✔ Prosecutor decides if to prosecute Greece ✔ ✔often nothing ✔ as no treatment places available Italy ✔ Warnings/ Prohibit leaving residence/ lose drivers licence Netherlands ✔ (On books/ ✔ (On books/ ✔ not used) not used) Norway ✔ ✔ ✔ (Civil) Portugal ✔ ✔ (Civil) ✔ Commission for the Dissuasion of Drug Addiction (CDA), community service Spain ✔ ✔ (Civil) ✔ Russia ✔ (Civil) ✔ Sweden ✔ Switzerland ✔ ✔ ✔ In some cantons Americas Argentina ✔ Brazil ✔ ✔ ✔Community service / suspension of drivers licence Canada ✔ ✔ ✔ ✔Community service / suspension of drivers licence if plead guilty Columbia ✔ Mexico ✔ United States ✔ ✔ ✔ Medical cannabis permitted in some US states India ✔ ✔ Lassi (a local drink) permitted Indonesia ✔ ✔ Pakistan ✔ ✔ Oceania Australia ✔ ✔ ✔ (Civil) ✔Warnings/cautions New Zealand ✔ ✔ ✔Warnings /diversion Sources: (Hall and Babor, 2000; Kisely, 2005; European Monitoring Centre for Drugs and Drug Addiction, 2007; Hughes and Stevens, 2007; Room et al., 2008; Pease-Watkin, 2010)

21 Chapter 2

Examining Table 3 carefully, the range of options available both across countries but also within countries becomes apparent (additional information on individual country’s policies can be found in Appendix-Chapter 2). Even within the European Union, where some standardisation might be expected, none is apparent. The options range from civil fines with no criminal record, to fines with a criminal record, treatment, incarceration, or no prosecution. The same can be said for United States where the overarching federal law on the legal status of cannabis in the US is total prohibition (Hall and Pacula, 2003) but every state has its own drug legislation from depenalisation of the possession of cannabis for personal use, treating it as misdemeanour or regarding it as civil offence (Room et al., 2008). Still other states allow those convicted of a non- violent drug offence to participate and complete drug treatment programs which if not completed may result in incarceration (Room et al., 2008).

Other countries also have a range of options available as a response to cannabis offences. Austria for example, has scope for the prosecution with fine or incarceration or no prosecution at all (European Monitoring Centre for Drugs and Drug Addiction, 2007). In Finland, the use of cannabis is a criminal offence and while the maximum penalty for an ordinary drug offence is two years imprisonment, possession for own use is punishable by a fine or maximum of six months imprisonment. Most often a fine is issued, but prosecution and punishment may be waived if the offence is considered insignificant or if the suspect has sought treatment (European Monitoring Centre for Drugs and Drug Addiction, 2007). Portugal also has a focus on treatment with its recently introduced Commission for the Dissuasion of Drug Addiction but may also issue civil fines (Hughes and Stevens, 2007).

In Canada, the Public Prosecution Service has a key role as it is responsible for prosecution (Kisely, 2005). Since 1996, conditional sentencing for some drug offences has occurred, where if the accused pleads guilty s/he may receive treatment or community service (Room et al., 2008). Although cannabis possession offences have increased, the rates of prosecution by the Public Prosecution Service have declined to as low as 35% of offences in some areas of the country (Kisely, 2005).

Russia and Sweden provide an illustration of how different motivations and processes can appear to result in apparently similar responses. Russia, in a reaction to a large cannabis offence burden on the courts, changed the laws in 2004 so that the possession

22 Chapter 2

of no more than 10 times the amount of a ‘single dose’ is considered an administrative infraction rather than a criminal offence. Punishment is a fine of no more than 40,000 rubles (~$1,380 AUD) or community service (Schreck, 2004). In Sweden the objective was different; this country has strongly pursued the goal of a drug free society through the use of increased penalties for drug offences, police powers to urine-test suspected drug users and the maintenance of a large number of treatment places to which suspected drug users may be coerced to attend (Hall and Babor, 2000). Despite this ‘zero tolerance’ approach, the possession of cannabis for personal use is treated as a minor offence punishable by a fine with the amount contingent on income (European Monitoring Centre for Drugs and Drug Addiction, 2007). It appears that Russia’s desire to decrease its court burden and Sweden’s zero tolerance approach both result in fines.

The previous example of Canada highlights another issue faced when attempting to understand the relationship between the law and the actual outcome. Although cannabis possession is treated as a criminal offence in Canada, prosecutors have discretion as to the ultimate response but information on what leads to their decisions is unknown (Kisely, 2005). It may be the circumstances of the arrest, the individual offender, the availability or not of treatment or community services, or the ideology of the individual prosecutor that results in variation in the rates of prosecution.

Recent changes to laws

There has been recent liberalisation of drug laws in some South and Central American countries which may be a portent of future changes elsewhere. To date there has not been any evaluation of the outcomes from these changes. Following a challenge to the Argentinean drug laws by a 19 year old who was convicted of possession of two grams of cannabis, and sentenced to 45 days in prison, the Argentinean Supreme Court ruled it unconstitutional to prosecute offences involving the use of cannabis in private. This ruling was ratified by the government (Carlos Hildago, 2009). As elsewhere, cannabis is illegal in Columbia but the Columbian Supreme Court has ruled that personal possession small quantities (< 20gm) of any psychoactive drug is legal (Room et al., 2008) but this was subsequently limited to only in the home. In December 2009, the Columbian Congress tried to prohibit the possession and carrying of all drugs, including cannabis, but as the political situation is currently in flux the necessary changes to the Columbian Constitution have not occurred (Pease-Watkin, 2010). In Mexico, the

23 Chapter 2

Senate has approved a bill decriminalising possession of small amounts of narcotics for personal use (Angel Gutierrez, 2009). This encompasses what police were already doing, that is ignoring possession of small amounts of drugs as their resources are overextended, in an attempt to deal with the ongoing drug wars.

California may be the closest to the legalisation of cannabis. The use of cannabis for medicinal purposes is permitted in California (and other US states) with a letter from a willing doctor. Under these conditions almost anyone can obtain cannabis for medicinal purposes. Some see this as quasi back door legalisation that has resulted in the decline in the price of cannabis (Kilmer et al., 2010). Californians recently (November, 2010) defeated Proposition 19, which if passed would have permitted adults 21 and over to possess up to an ounce of cannabis, to consume it in a non-public place as long as no children were present, and to grow it in small private plots. Proposition 19 which was defeated (with 53.5% voting against it) would have authorised local governments to regulate sales and cultivation of cannabis (Drug Policy Alliance, 2010).

As elsewhere, cannabis use, possession, supply or cultivation is illegal in Australia although currently four jurisdictions (Western Australia, South Australia, Australian Capital Territory and the Northern Territory) have civil fine programs operating for the possession of a small amount of cannabis (McLaren and Mattick, 2006). The fines, the amount of cannabis permitted under the scheme, and the amount of discretion under which police operate vary between the various jurisdictions (McLaren and Mattick, 2006; Room et al., 2008). In other Australian jurisdictions (New South Wales, Queensland, Victoria and Tasmania) there are cannabis cautioning schemes. In addition there are a number of diversion programs. The existing legislation and implementation are further discussed in the chapter on the criminal justice system (Chapter 5).

From this brief overview it is apparent that although many countries have implemented programs that comply with the international treaties there is considerable variation in the intent and the implementation of the laws. Many countries provide for the possibility of incarceration but most have either formal or informal alternatives such as treatment, payment of a fine, or community service with only a few—indeed just a small fraction in western countries—actually imprisoned for only cannabis possession for personal use (Caulkins and Kleiman, 2007). The response by many countries has been a pragmatic one (Fazey, 2003), maintaining cannabis as illegal but either not

24 Chapter 2

implementing the law (Netherlands) or by giving discretion to the prosecution (Canada, Germany) and /or to the police (Australia, France, Britain, and New Zealand).

The Netherlands is often held up as an example of how cannabis should be managed in the Australian system but Kerley (2004) in his comparison of criminal law and drug law enforcement between Australia and the Netherlands makes some salient points in concluding that legally the Dutch approach is not appropriate for Australia. Kerley (2004) argues that analysing laws in the absence of context results in little understanding of how an issue is managed by the criminal justice system. In fact, at a superficial level the prosecution powers which exist in Australia are also present in the Netherlands but it is at a deeper level where the differences in the regulations and discretion which exist in the Netherlands are unique to the legal culture in the Netherlands. Further the current management of cannabis “reflects a unique legal culture in the Netherlands that evolved in accordance with particular social and political developments in twentieth century Dutch society”(Kerley, 2004. p36).

Accepting that the Dutch approach is not appropriate for Australia, based not only on cultural differences, but also on the legal inconsistency that it is legal to buy but not to supply cannabis one needs to turn elsewhere for an alternative policy.

2.4 Selecting policies for comparison

As mentioned earlier, two policies were selected for comparison in the CBA. The first policy selected was that of the current (as of 2007) New South Wales (NSW) cannabis laws. With the across country and between country variation in laws, enforcement and outcomes there is not a common or usual means by which cannabis offences are managed although there does appear to be some common objectives, that of limiting harms to the users and decreasing the demand on the courts (Room et al., 2008). Such are the objectives of the current cannabis policy in NSW (Baker and Goh, 2004). It is in keeping with the international treaties, where all cannabis activities are illegal but it does not require criminal prosecution for small quantities, and allows for non-custodial alternatives when police happen upon small amounts of drugs (INCB, 2009). The NSW program was designed to divert first-time and minor drug offenders from the criminal justice system and was based on the principles of harm reduction (Baker and Goh, 2004). While cannabis is illegal in NSW, police have the discretion to caution adult

25 Chapter 2

offenders for a first or second use or possession offence plus there are informal warnings and cautions available for juveniles (Baker and Goh, 2004). In addition, the choice of current NSW policy (as of 2007) as one option for the CBA addresses Kerley’s (2004) point pertaining to analysis occurring out of context. Access to NSW Police, court staff, and treatment providers enabled interaction during the research for this thesis thus permitting the analyses to occur in an environment of insight as to how issues are actually managed by the criminal justice system.

The choice of a legalisation option arises from the initial motivation for this thesis, the publication of the report The Budgetary Implications of Marijuana Prohibition (Miron, 2005) where the case for legalisation of cannabis was made based on potential savings to the criminal justice system and projections of taxation revenue for government. This report and its recommendations were supported by 500 American economists in an open letter to the President of the United States and other legislators (Marijuana Policy Project, 2006). While specific limitations of this report are found in more detail in Chapters 3 and 5, what is relevant here is that, in my view, there are major limitations to the assumptions, the data and the methods in Miron’s report. Limitations include how criminal justice costs were estimated; the absence of a comparator; the apparent assumption that there were no costs related to the implementation of legalised cannabis; plus the failure to address potential health and education harms related to increased cannabis use. Additionally, no attempt was made to ascertain the costs to, the benefits for, or the preferences of individuals. While Miron (2005) did not intend for his work to be a summary of all the costs and benefits, the implicit inference is that only taxation income and criminal justice savings are ultimately what is important when making cannabis policy decisions. In this thesis a much broader perspective is taken to explore other government sectors as well as the benefits and harms to individuals.

A key point made in the Miron (2006) report was that cannabis is not legal in any country and this was easily confirmed (MacCoun and Reuter, 2001a; Room et al., 2008; Kilmer et al., 2010), although there was recently an attempt to legalise cannabis in California (Kilmer et al., 2010). This lack of an existing policy from which to start the analysis means not only that decisions must be made on the form of the legalised policy (does it look more like coffee, alcohol, or prescription drugs?) but also requires several assumptions on outcomes of the policy.

26 Chapter 2

The characteristics of a legalised but regulated option for cannabis for inclusion in the CBA was constructed based on the public health approach articulated by Nadelmann (1992) and further expanded by others (McDonald et al., 1994; Haden, 2002; Haden, 2004; Rolles et al., 2006; Haden, 2008; Rolles, 2009). The policy includes positive licensing as outlined by MacCoun and Reuter (2001), a modern version of Nadelmann’s mail-order model using the internet, disallowing promotion and advertising (Haden, 2004; King County Bar Association, 2005), monopoly provision (Haden, 2004), potency and quality monitoring (Haden, 2004; Rolles, 2009 ), age restrictions and restrictions on location of consumption (Babor et al., 2003). The legalised option will be further discussed and the resources implications estimated in Chapter 6. This hypothetical legalised–regulated policy is in direct conflict with the international treaties discussed above, a matter which is noteworthy but not further considered herein.

2.5 Summary

In summary, this chapter has briefly outlined some of the myriad of policy options for cannabis. The choice of a legalised option was determined by the original research question however it became clear upon reviewing the literature that it would be necessary to define and clarify the components of such a policy as no concrete example existed. The comparator, the existing NSW cannabis policy is but one example of a policy which complies with the international treaties in maintaining the illegality of cannabis but with associated attempts to limit harms to the users and decreasing the demand on the courts through the introduction of a cannabis cautioning program.

Many have made the point that additional research is required as a prerequisite to the rational consideration of public policies (Strang et al., 2000; MacCoun and Reuter, 2001b; Hall and Pacula, 2003; Hall and Lynsky, 2009). Undertaking a CBA of two cannabis policies, one of which is a modelled legalised option, while contentious, will start to address some of the methodological challenges regarding what should be considered when making public policy in this area. Some of these challenges include which costs, which benefits, how they are best quantified and whose preferences should be included. The next chapter begins to address some of these questions.

27 Chapter 3

Chapter 3 : Economic theory and its application in this thesis

3.1 Introduction

In Chapter 2 the various policies for cannabis were reviewed and two legislative policy alternatives were selected for the CBA. In this chapter, the existing literature on economic evaluations of cannabis policies are reviewed as are the various methods of evaluating complex policies and the rationale provided as to why a cost benefit analysis (CBA) is an appropriate method to compare policies. The economic theory behind CBA is discussed including various approaches to quantifying benefits. Finally, the chapter concludes with an outline of the CBA to be conducted in this thesis, including documenting the types of costs and benefits to be included. The model is static with the assumption of equilibrium having been reached with a policy change. Data was analysed for a one year period, and all costs are in 2007 Australian dollars. The various components for populating the model are described in Chapters 4 to 8 with the final steps and the results of the model in Chapter 9.

At the time of the conception of this thesis, and in the years since, an ongoing debate has continued with respect to the potential harms of cannabis, particularly schizophrenia (Macleod et al., 2004; Moore et al., 2007; Degenhardt et al., 2008; Macleod and Hickman, 2010) and an increased recognition of the need for treatment for cannabis users (Hart, 2005). Alongside this there are some who continue to raise concerns around the use of the criminal justice system to manage cannabis; the failure to recognise drug use as a human right (van Ree, 1999); the uncertainty around quality and potency of cannabis (Poulsen and Sutherland, 2000); and consumers having contact with illicit drug dealers. This leads to arguments for liberalisation or even legalisation of cannabis (Wodak et al., 2002; Haden, 2008; van den Brink, 2008; Rolles, 2009). At the same time, recognition of the harms related to the risky consumption of tobacco and alcohol, in particular binge drinking, have been prominent in the public discourse (Chikritzhs et al., 2009; Laslett et al., 2010b) with additional restrictions imposed on the locations of consumption, the locations of purchase, the increased enforcement of regulations by police, and increases in taxation of both alcohol and tobacco (Freeman et al., 2008; NSW Health, 2008; Jones et al., 2009; Communities Office of Liquor, 2010). All of these circumstances, alongside the lack of existing research make it an opportune

28 Chapter 3

time to explore the costs and benefits of cannabis policies using both economic and public health methods.

3.2 Existing literature on economic evaluations of cannabis policies

Changes to the legislative status of cannabis are frequently proposed but there is remarkably little known about the economic implications of these proposed interventions (MacCoun and Reuter, 2001b; Miron, 2005; Pollack and Reuter, 2007). Most discussions on policy changes are often made with only partial or no economic evidence (Wodak et al., 2002; Kisely, 2005).

A recommendation made by a number of eminent cannabis researchers in a recent review of cannabis policy was that future policy should be grounded in science and, in part, be based on the total economic impact of the drug and the policy (Room et al., 2008). An assessment of the economic impact was not part of that review. This is plainly evident in the report as while there are several pages devoted to describing the policies in various countries, only four economic analyses were identified and they examined only the government costs of civil penalty schemes. None of these are economic analyses with true comparisons nor do they fully explore the costs and benefits of the policies.

A systematic and wide-ranging literature search of peer-reviewed publication databases (Econlit, PubHealth, Medline, CINAHL, Criminal Justice Abstracts) using search terms ‘cannabis policy’ or ‘marijuana policy’ and ‘cost benefit analysis’, or ‘economic evaluation’ resulted in no published peer-reviewed articles with a quantification of the costs and benefits of moving from one legal option to another for cannabis (or for any currently illicit drug). There are economic evaluations of treatment for cannabis but none which evaluated changes in legislative policies. There are however, two sorts of publication that do move this debate forward; those which describe the need for a CBA of drug policy, and provide insights into the methodological challenges (MacCoun et al., 1996; van Dijk, 1998; Hall and Babor, 2000; MacCoun and Reuter, 2001a; Hall and Pacula, 2003; Hall and Lynsky, 2009; Rolles, 2009) and those found in the grey literature (Brooks et al., 1999; May et al., 2002; Bates, 2004; Miron, 2005; May et al., 2007; Gieringer, 2009; Kilmer et al., 2010) which have estimated some of the various costs to government, with or without estimating changes in the use of cannabis. A focus

29 Chapter 3

of much of the grey literature is on potential cost savings to the criminal justice system (Brooks et al., 1999; May et al., 2002; Bates, 2004; Miron, 2005; May et al., 2007). Several use crude top-down estimates of police and court costs (Bates, 2004; Miron, 2005) and potential lost transfers from the federal government to the state (Bates, 2004) while others estimate police costs using micro costing (Brooks et al., 1999; May et al., 2002; Baker and Goh, 2004; May et al., 2007) but do not attempt to measure other costs or benefits.

An article by van Dijk (1998) promises much in its title “The narrow margins of the Dutch drug policy: a cost-benefit analysis” but while it discusses the importance and methods of a CBA it does not fulfil the promise of the title. The costs to the Dutch criminal justice system are estimated in dollar terms but the remainder of the potential consequences are scored with either a plus or a minus sign indicating for each consequence the expected movement from the current policy (de facto legal) to either re-criminalisation or legalisation. The pluses and minuses are summed, and the author concludes that there is evidence that the status quo is preferred. There are a number of limitations – firstly, the assumption is made that there would be no increase in the use of cannabis, thus no increased demand for health care related to cannabis. The other serious issue with this paper is that the authors, without providing any evidence to convince the reader of the validity, or a theoretical construct, give each of their categories an equal weighting when summing the plusses and minuses.

Bates (2004) examined the economic consequences of eliminating prohibition (legalisation) of cannabis in Alaska. The focus of the study was on the potential savings to the criminal justice system but also included estimates of lost economic output, impact on family and social service budgets and whether there would be a loss in federal grants and other funding (Bates, 2004). On describing the methods and rationale for a CBA, Bates states “In order for prohibition to pass the first elementary test of cost/benefit analysis, it must be shown that prohibition significantly reduces consumption of the substance. It is not merely enough that consumption decreases, but that the decrease is significant enough to warrant the expenditure” (Bates, 2004) p18. The author then goes on to make the point that as there has been little or no change in cannabis use when other jurisdictions moved from criminalisation to decriminalisation (not what was being considered by Bates) that prohibition would not decrease use and concluded “Prohibition does not achieve its principal objective in any meaningful way”. 30 Chapter 3

There are two problems here—firstly, decriminalisation (when cannabis possession/use is not permitted but there are no criminal proceedings), is not the same as legalisation and one would not expect the same effects. Secondly, the test for a CBA is not about whether use is reduced but rather whether the net costs are less than the overall benefits.

A small body of work has explored the cost savings to the criminal justice system of moving from a more restrictive to a less restrictive policy (Brooks et al., 1999; May et al., 2002; Baker and Goh, 2004; Crime Research Centre, 2007; May et al., 2007). Each study was limited in its scope and did not go beyond examining cost savings to the criminal justice system.

Miron (2005, 2010) focuses on the impact on the criminal justice system and the potential taxation revenues should cannabis be legalised in the United States thus ignoring the potential health care costs related to dependence, and other health-related consequences of the use and misuse of cannabis. Also not included are the potential negative impacts on educational attainment from heavy use of cannabis as an adolescent (Miron, 2005). Missing from both Bates’ (2004) and Miron’s (2005, 2010) studies are the costs of establishing, and enforcing the regulations of a legalised industry (i.e. controlling sales to minors, setting and maintaining standards for purity, potency, advertising standards).

More recently, Kilmer and colleagues, in response to California’s Proposition 19 attempted to estimate the impact of legalisation on consumption and public budgets (Kilmer et al., 2010). This work, while focusing on California, adds to the literature by improving on the methods used by both Miron and Bates in the estimation of the costs of enforcing cannabis laws in California. They also provide estimates of the production costs of cannabis in both legal and semi-legal contexts, and estimate the impact on use of a substantive price decrease. Kilmer et al. (2010) concluded that under any of the proposed models in California, the pre-tax price of cannabis will decline substantially, consumption will increase, but taxation revenues are difficult to predict as there is considerable uncertainty around demand and taxation avoidance.

In conclusion, although some work has been done evaluating the costs and benefits in some sectors, there has been no full economic evaluation of cannabis policy.

31 Chapter 3

3.3 Various methods for evaluating the societal impact of policy changes

This section briefly discusses some alternative methods of evaluating policy changes and lays out the rationale for the choice of a cost benefit analysis in this thesis. When either the intervention or the setting in which the intervention is occurring is complex, the evaluation is also necessarily complicated. There are a number of forms of evaluation methods to consider when undertaking an evaluation of a policy which involves criminal justice, health, education, government regulation, personal preferences and possibly productivity. As the objective is to assess the relative societal costs and benefits of two policies for cannabis the method must permit combining the benefits across multiple sectors in a meaningful way.

Adler (2006) categorises methods of analysis into four types: 1) non-welfare focused; 2) welfare narrowly focused; 3) welfare widely focused; and 4) hybrid. The evaluation methods in the first category, the non-welfare-focused evaluations do not meet the objective of assessing the societal costs and benefits thus are not suitable. For example, one form of non-welfare analysis is a financial evaluation which is a cash flow analysis of a project or a program which does not use shadow prices or include externalities, and may not quantify outcomes (Commonwealth of Australia, 2006). Another method, feasibility analysis demonstrates whether a project could be undertaken given available resources, and may assess the return on a capital investment but again does not value the wider social benefits.

Adler’s second category of methods, are classified as narrowly focused welfare methods, where narrow appears to reflect the choice in outcome measure. Most of the methods in this category focus on a small set of measures of wellbeing and ignore all other outcomes. For example, the outcome of one study may be life–years saved, the outcome of another may be a gain in quality of life and the third, educational attainment; all are relevant when considering societal welfare but individually (narrow focus) they do not achieve the objective of a societal approach for a policy that extends across health, criminal justice, and education. Individually, the narrow-focused research may return results which are at odds with overall welfare. While the outcomes from individual studies may be a component of what is considered when examining social welfare, they may not share a common denominator which means they cannot be summed.

32 Chapter 3

The third of Adler’s categories are widely focused welfare approaches. CBA is included under the widely focused assessment of welfare as is equilibrium analysis. Equilibrium analysis is used for addressing macro-economic issues such as design of tax system (Adler and Posner, 2006) but not particularly useful for the question being addressed here. Conversely, CBA is poorly suited to constitutional issues and not appropriate for macro-economic issues but is well suited to evaluating social policy (Elvik, 2001). Intuitive balancing, as presented by Adler and Posner (2004) is a method of assessing the value of a program by assessing the relative trade-offs between a set of pre-established criteria. ‘Balancing with explicit trade-off rates’ is a variant of intuitive balancing but has explicitly specified quantitative trade-off rates between criteria. While these may be valid methods of assessing benefits of a program the trade-off rates between various criteria are established outside the evaluation (Adler and Posner, 2006). Should such rates exist for cannabis policy they would lack transparency. The use of such rates, if they existed, might also reflect what the evaluator, or decision-makers thought ought to occur rather than the views of those directly affected by the policy.

Within the field of health economics a different debate occurs over methods. Here the debate is often between ‘welfarist’ and ‘extra-welfarist’. And this reflects the choice between Adler’s hybrid methods (cost effectiveness and cost utility analysis) and a CBA. Those who are welfarist believe that “the ‘goodness’ of any situation … be judged solely on the basis of the utility levels attained by individuals in that situation” (Hurley, 1998) p377). Those who are extra-welfarist reject the exclusive focus on individual utility and broaden the evaluative space to include other measures such as health gain, patient satisfaction or caregiver burden (Brouwer et al., 2008). Further, extra-welfarists “do not define the output of healthcare in terms of preferences for health vis-a-vis other goods, but according to its contribution to health itself, i.e. they wish to maximize health as against overall welfare” (Gyrd-Hansen, 2005) p425). Extra- welfarists also argue there is room for equity adjustments in estimating potential benefits whereas welfarists judge benefits solely on the utility levels gained by individuals (Hurley, 1998).

The choice between welfarist and extra-welfarist has implications for the method of economic evaluation. Cost benefit analyses, with a monetary valuation of all individual benefits, has its foundation in welfare theory. Cost effectiveness analysis (CEA) with a QALY outcome is perceived as extra-welfarist. The point has been made that at least 33 Chapter 3

theoretically a CBA with QALYs can be undertaken using a welfarist approach (Brazier et al., 2007). However, there are a number of methodological challenges which have yet to be resolved including how to manage interactions between different health states with the current imposition of additive separability between health states being unrealistic. In addition, off the shelf quality of life instruments are not available and likely not possible (Brazier et al., 2007) given the welfarist view that it is the individual’s own utility which is of interest. There are also ongoing statistical validity issues around the estimation of QALY weights, which are resolved when data are aggregated, but still exist at the individual level (Brazier et al., 2007). Similarly it is argued that CEA benefits can be valued monetarily but again there are a number of concerns such as the cost per QALY in the CEA does not capture non-health benefits, and although non-health costs can be included, non-health benefits are often ignored (Brazier et al., 2007).

At a practical level, cost effectiveness analyses which measure only a single natural outcome such as life-years saved, cases detected or prevented, are not suitable when there are multiple important and possibly conflicting benefits and harms. This was demonstrated in a study of different drug treatments where multiple outcomes were measured (Sindelar et al., 2004). Comparisons of these outcomes for standard and enhanced treatment found that the magnitude of the change in outcomes over the course of treatment varied by type of outcome thus leading to very different incremental cost effectiveness ratios (Sindelar et al., 2004). Cost-utility analyses while quantifying both changes in lives saved and changes in quality of life include only health–related productivity changes or other non-health outcomes. A broader set of measures are necessary when assessing changes in the legal status of cannabis. Until methods of constructing a composite outcome measure are developed, CBA is the most appropriate tool when there are multiple important outcomes (Cartwright, 2000; French et al., 2002b; Sindelar et al., 2004).

To summarise, CBA was selected as the method for the evaluation of two policies as there are multiple potential outcomes (use, cannabis use disorder, psychosis, mortality and morbidity from motor vehicle accidents, low birth weight newborns, poor educational attainment, stigma) across multiple sectors (i.e. health, criminal justice, education) and through the monetary valuation of outcomes CBA has the potential to encompass this. Methods such as ‘intuitive balancing’ which may also encompass all 34 Chapter 3

potential outcomes rely on the existence of trade-off rates between outcomes that do not yet exist, and even if they did they may not reflect individual preferences but rather reflect the preferences of policy makers (Brouwer et al., 2008). Others make the point that CBA is the only normative framework which explicitly assesses the costs and benefits of social policies (Vining and Weimer, 2010).

3.4 Cost benefit analysis: an overview

The objective of a CBA is to value all the costs and benefits of a policy or program in monetary terms with the broad purpose of assisting with social decision-making (Boardman et al., 2001) by providing evidence on the efficient allocation of resources in areas where, for one reason or another, private markets cannot or do not achieve this outcome (Commonwealth of Australia, 2006). CBA can be particularly useful when there are multiple and conflicting outcomes of a policy (Elvik, 2001; Vining and Weimer, 2010). Both of these criteria apply to cannabis policies. Here the choice between policies does not sit neatly in the private market framework, and there are multiple and conflicting outcomes (i.e. stigma, enjoyment, health, education) between the two policies.

When market failures exist the price may not reflect the opportunity cost of the good; or in terms of policy when decision-makers fail to consider all costs and benefits of a policy they may not be making the best decision on the efficient allocation of resources (Vining and Weimer, 2010). CBA can be used to tackle issues of allocative efficiency which is the overall efficient allocation of resources.

CBA has its foundations in welfare economics, the branch of economics that addresses normative questions (what should be) as compared to positive economics (making predictions without value judgements). Normative, in the context of CBA, assumes that: i) social welfare is made up from the welfare (or utilities) of each individual within society and ii) individuals are the best judges of their own welfare (consumer sovereignty) (Drummond et al., 2005). The central objective of welfare economics is to provide a coherent ethical framework for making meaningful statements about whether some states are socially preferable to others (Boadway and Bruce, 1984). Importantly “...the welfare economist wishes to determine the desirability of a particular policy – not

35 Chapter 3

in terms of his or her own values, but in terms of some explicitly stated ethical criteria” (Boadway and Bruce, 1984).

The central concept that lies behind CBA is Pareto optimality which is defined as ‘no person should be worse off under an alternative program compared to prior to its introduction’. Any measure that allows an individual to judge themselves better or worse off can be subject to this rule, thus at a practical level, if a single person is dissatisfied with an alternative scenario then it is not an unambiguous improvement in social welfare—it is not Pareto optimal (Tsuchiya and Williams, 2001). The consequence of this is that it would be virtually impossible to introduce a program or policy which was deemed beneficial to society. In response to this issue, in the late 1930s, recognising that the relative amount of benefits enjoyed by the winners and losers could vary, economists were disturbed by the view that “each individual had an equal capacity for enjoyment and that the gains and losses among different individuals could be directly compared” (Mishan and Quah, 2007).

Under this view, the movement to the alternative program, say legalisation of cannabis, would not be considered a Pareto optimal solution if, for example, one program resulted in a large number of individuals each gaining considerable benefits (the freedom to use cannabis legally) while losses ensued to a small number of individuals (additional adolescents not achieving their maximum level of education because they were enticed to use cannabis at a young age). Additionally, it was argued that interpersonal comparisons of utility were unscientific, and while this latter point was acknowledged, it was argued by Kaldor that the interpersonal comparisons could be made irrelevant. What was really important was whether a policy led to an increase in aggregate real income (benefits) then “the economists’ case for the policy is quite unaffected by the comparability of individual satisfaction since in all cases it is possible to make somebody better off or at any rate to make some people better off without making anybody worse off” (Zerbe Jr et al., 2006). Whether this compensation took place was not seen as relevant; what was important was whether there could hypothetically be compensation from the winners to the losers. The test is whether the potential winners could offer to the potential losers sufficient income so the losers would be as well off as they were at baseline (Tsuchiya and Williams, 2001). This criterion, referred to as the Kaldor-Hicks criterion does not mean the redistribution actually takes place and is often referred to as the potential compensation test (PCT). If the PCT is passed then the 36 Chapter 3

project is determined to be allocatively efficient. The Kaldor-Hicks criterion separates efficiency and equity leaving the latter to the politicians.

In applying CBA to this thesis, given social welfare is comprised of utilities of each individual member of society, coercive policies or restrictions on the individual’s choice to consume cannabis implies some loss of individual welfare compared to an alternative allowing more freedom of choice (Godfrey, 2006). From a welfarist perspective, the status quo (the currently illegal with cautioning) would only be preferred to legalisation with regulation if the external costs associated with cannabis consumption are lower and more than compensate for the loss of welfare from the restriction of individual choice.

3.4.1 Decision rules of a CBA

The primary goal of a CBA is to identify those projects or policies with the highest net social benefit (NSB) (Pearce et al., 2006). The NSB is the sum of all benefits minus sum of all costs and is mathematically presented as follows.

Given i = 1….I possible investments, over t time periods

b − ct t )()( n i i ∑ NSB = t=1 t−1 i (1 + r)

Where: bi(t) = benefits in money terms derived in year t

ci(t) costs in money terms in year t 1/(1+r) = discount factor at annual interest rate r n= lifetime of project

Once the NSB is calculated, all those policy options where the NSB is greater than zero are ranked according to their NSB from lowest to highest (Drummond et al., 2005; Commonwealth of Australia, 2006), with the preferred option being that with the highest NSB. In this study, as only current year costs and benefits are included the equation reduces to:

NSB= � ∑� �� 37 Chapter 3

The use of only a single year’s costs and benefits is a limitation of this research and the implications of this are discussed in Chapter 9.

Another rule frequently used to determine the preferred option is the Benefit-Cost Ratio (BCR) (Commonwealth of Australia, 2006):

BCR= Present value of benefits divided by Present value of costs.

When BCR is used, the BCR must be greater than one for a project to be accepted. If the question at hand is whether or not to either accept/reject a given project then BCR may be sufficient. However, if there is a choice between mutually exclusive projects, the BCR and NSB may rank projects differently.

Although the BCR rule is easy to interpret and may be appealing, it has major limitations (Ableson, 2000; Commonwealth of Australia, 2006) as it is very sensitive to how costs are defined. For example, in the BCR whether the (dis)benefits are added to the costs or included in the benefits stream can result in very different recommendations (Ableson, 2000; Drummond et al., 2005; Commonwealth of Australia, 2006). In this project which has mutually exclusive policies, and several (dis)benefits, the NSB will be the main decision rule, although the BCR will be reported1. By using the NSB rule any arbitrary decisions as to whether a certain item is a change in costs or a (dis)benefit are avoided. The opportunity costs of resources are included on the cost side of the equation and on the benefit side are those items which are utility-bearing change (+ or -).

The results of a CBA are often not the sole source of information a decision-maker may need to consider but it does provide quantitative comparisons of alternatives, together with supporting information for benefits or inputs that could not otherwise be quantified (Commonwealth of Australia, 2006).

3.4.2 Methods of valuing benefits

As indicated above, one of the key challenges in any CBA is valuing all benefits in monetary terms. The valuation of resources, while time consuming, is methodologically comprehensive using opportunity cost methods. The resources are first identified then quantified and finally a unit price applied (Drummond et al., 2005). The use of

1 The terms ‘benefit’ and ‘(dis)benefit” will be used in this thesis to reflect positive and negative consequences of moving from one policy to another. 38 Chapter 3

opportunity costs in a CBA to estimate the resource implications of policies provides decision-makers with the assurance that the resource implications of non-marketable goods are, where feasible, included (Drummond et al., 2005).

There are two main approaches to valuing benefits—the human capital approach (HCA) and willingness to pay (WTP). This thesis uses multiple methods of valuing benefits but most fall within the willingness-to- pay approach. Both revealed preferences and stated preference WTP methods are employed to value the benefits within the context of the two policy options.

Briefly, the HCA can be used either as the sole measure of the value of all outcomes or alternatively, as a means of measuring only the value of productivity changes. In the latter, a monetary value is assigned to a productivity loss related to mortality or morbidity (Drummond et al., 2005). HCA as a sole measure of value is often criticised for its absence of any link to welfare theory; its inequitable treatment of the elderly and those who are currently not employed (both young and old); and its inherent assumption that the objective of health care (or any program) is to maximise national productivity (Johannesson and Jonsson, 1991) while mostly ignoring the value of leisure time or unpaid work.

Revealed preferences are based on observations of how consumers respond to changes in price and income (Commonwealth of Australia, 2006). In this research, there are two factors to consider when shifting from criminalised cannabis to a legalised cannabis policy. The real price of cannabis arguably decreases as the risk of arrest decreases (Williams, 2004), even though money price is being held constant in this current research at the median NSW street price (Black et al., 2007; Phillips and Burns, 2008). Secondly, the demand curve shifts out in response to the demand shock of legalising the system. The increase in quantity consumed as a response to these two effects is estimated in Chapter 4. Other methods, not used herein, of obtaining revealed preferences include: i) Hedonic prices, where the value of the good is the sum of the value of its attributes such as in housing prices which are used to reflect the benefits of neighbourhood amenities; ii) Travel cost analysis; and iii) defensive household behaviour such as the purchasing of insurance (Drummond et al., 2007).

39 Chapter 3

Stated preference techniques provide an alternative to revealed preferences and there are two common forms of eliciting stated preferences, both of which are used in this research. They are contingent valuation (Chapter 7) and discrete choice experiments (Chapter 10). A description of each method is found below, with an additional literature review included in the relevant chapters.

Contingent valuation method (CV) is a survey technique that can be used to value the benefits of a policy or good where a market price is not available (Drummond et al., 2005). CV is a choice-based approach where respondents are asked to think about the actual market and to reveal the maximum amount they would be willing to pay for a good/policy or alternatively, the minimum they would require to forgo the gain (Drummond et al., 2005; Amaya-Amaya et al., 2008). The theoretical underpinnings of CV lie in utility maximisation with an income constraint. Contingent valuation, as a tool to measure benefits, is based on the premise that the maximum amount of money an individual is willing to pay for a good or service is a measure of the utility or satisfaction that good provides them (McIntosh, 2010). CV studies have been used broadly, for example in health (Pellegrini and Jeanrenaud; Zarkin et al., 2000; Olsen and Smith, 2001; Schwarzinger et al., 2009; Weimer et al., 2009), environment (Pearce et al., 2006), and transport economics (Hensher, 1993; Elvik, 2001; Hensher et al., 2009), and in the assessment of crime prevention programs (Cohen et al., 2004; Cohen, 2007).

Despite CV studies being conducted widely (Olsen and Smith, 2001) there remains unresolved several methodological challenges (Smith, 2006; Frew, 2010a; McIntosh, 2010). Challenges to conducting a CV include ensuring that the framing of the question and scenario are comprehensible and relevant to the respondents; presentation of the question in such a way as it does not bias the responses; and selection of the method of electing the monetary value (Smith, 2006; Frew, 2010a; McIntosh, 2010). Failure to address these issues may lead to invalid results but the use of focus groups, and pre- and pilot testing can help in addressing many issues.

Critics of CV methods often denounce the results of CV studies. Reasons for this are often related to the hypothetical nature of the question and lack of reproducibility. Often the respondents are being asked to value a new good or service or one which they typically are not required to pay for (Olsen and Smith, 2001; Frew, 2010b). Thus respondents may overstate their ability to pay (Johannesson et al., 1996) in an attempt to

40 Chapter 3

please, or because they recognise it is a hypothetical question (McIntosh, 2010). Alternatively they fail to recognise the budgetary impact such a decision may have. A related criticism of CV is the relationship between willingness and ability to pay. Those who can, often choose to pay more which may subsequently bias against low income groups or programs targeted at these groups. However, if societal valuations are of interest the use of a representative sample will go some way to address this latter issue (Johannesson et al., 1996).

It has been suggested that the continued use of CV studies in health economics is due to its strong theoretical foundations in welfare theory and the valuation of benefits in monetary units (Olsen and Smith, 2001). In this work, which goes beyond health outcomes, WTP was selected as one tool to value benefits, both because of its theoretical foundations and its usefulness in valuing non-health benefits (stigma). In this research individuals were asked outright what they would be willing to pay to avoid the stigma from a criminal record from possession of cannabis. Additional discussion as to how the various methodological challenges of CV were dealt with is found in Chapter 7. The results from Chapter 7 are incorporated into the final summation of the costs and benefits in Chapter 9.

The second stated choice method is a discrete choice experiment (DCE). DCE is a method of eliciting preferences which is based in random utility theory. When conducting a DCE individuals are provided with a series of choices. Each choice selected implies that it provides a higher level of utility than the alternative (Gerard, et al. 2003). DCEs, through the generation of the choice data, can create a hypothetical market with which to explore areas of interest such as preferences for a good, service or policy (Lancsar and Louviere, 2008). In the DCE, preferences are elicited by providing a series of profiles (choices) to respondents. Each profile is comprised of a set of attributes describing the good of interest with a level for each attribute, where the levels vary across the profiles. Each choice made implies that it provides a higher level utility than the alternative. Previously, utility scores from DCE have been accepted as cardinal measures, however recent work demonstrates that these scores are ordinal, and that while any monotonic transformation of the utility function results in the same ordering, the relative difference between the scores are not known (De Bekker-Grob et al., 2010 ). The inclusion of a price proxy allows movement from an ordinal measure to a cardinal measure (De Bekker-Grob et al., 2010 ). Additionally, Best Worst Scaling can be used 41 Chapter 3

if a price proxy is not available to address problems that result when different dimensions have different underlying scales (Flynn et al., 2007; De Bekker-Grob et al., 2010 ). In this thesis, where the purpose is to compare the preferences for different attributes of policy options the utility scores are treated as ordinal.

The multiple choices made by each respondent in a DCE mean that each response cannot be considered independent and this must be accounted for in the analysis. But the multiple responses in a DCE also allow for exploration of preference heterogeneity. As preferences for drugs policy have been shown to vary within the community (Matthew-Simmons et al., 2008) the DCE analysis uses mixed logit methods to explore preference heterogeneity across individuals (Hensher and Green, 2003; Hall et al., 2006; Hole, 2008).

Chapter 10 uses the DCE approach to elicit preferences for different policies for cannabis and the trade-offs between health harms, criminal justice expenditures, rates of use of cannabis and the location of purchase. The theory, methods and results of the DCE which explored preferences for conducted are further described in Chapter 10.

3.4.3 Steps in a CBA There are a number of basic steps undertaken in any economic evaluation (Ableson, 2000; Boardman et al., 2001; Commonwealth of Australia, 2006; Zerbe Jr and Bellas, 2006). They include: a. Clarifying the question of interest and the context b. Clarifying the policy options including the base line comparator c. Clarifying the scope and objectives of the study including identifying the constraints (boundary) of the CBA d. Setting out the assumptions e. Identifying possible welfare impacts (costs and benefits) f. Quantifying and valuing the costs and benefits g. Discounting future values to obtain present values h. Testing for uncertainty and risk i. Calculating the net present value and interpreting the findings.

42 Chapter 3

In the following sections of this chapter, steps ‘a’ through ‘e’ will be discussed. The actual quantifying and valuing of costs and benefits occurs in Chapters 4 through 8 with Chapter 9 containing the results.

3.4.3.1 Clarifying the question, the context, the policy, the objectives and identifying the constraints (boundary) of the CBA

Although the list of steps above is presented as a set of separate steps, some have to be considered concurrently. For example, articulating the question for a CBA necessitates understanding the policy context, issues raised by the various policies as well as understanding the constraints and the limitations of the method.

The question, the policy, and the objectives have been addressed in Chapter 2 but it is worth restating them. The objective of the CBA is to assess the costs and benefits of two separate policies for cannabis from a societal perspective to assess which has the highest net social benefit (NSB). The broad perspective taken includes the cannabis user, the grower, retailer, government and the wider society. The policy options are the status quo in NSW where cannabis is illegal (with a cannabis cautioning program for possession/use of a small amount of cannabis) and a legalised but regulated alternative.

The policy context can be considered in two ways, one is the geographical region (or population) and the other is the social and political milieu. The geographical area, the state of NSW, was selected for a number of reasons as discussed previously: it is one jurisdiction where cannabis is currently illegal, but there are mediating programs that limit the negative impact of criminal sanctions; there was access to data to use for the model; and last but not least there was access to a wide range of individuals who provided information and insights as to how the current policies are implemented and enforced.

Other exclusions are further outlined at the conclusion of this chapter but there are two key constraints that are central to understanding the boundaries of this CBA. The first is that the potential national and international political ramifications of adopting a legalised framework under the current 1961 United Nations Single Convention on Narcotic Drugs are in no way quantified in this work. Evidence from the Netherlands suggests that the pressure from neighbouring countries, the United States and the United Nations Office on Drug Control would be substantial should a move to legalisation

43 Chapter 3

occur (van Dijk, 1998). These costs while recognised as being significant are not included in this study, and neither is the issue of potential cross border smuggling.

The second important boundary is that this thesis does not assess is the impact on the consumption of alcohol or other illicit drugs should cannabis become legalised. The impact may be significant, depending on whether alcohol is a complement or a substitute for cannabis, and the extent to which any increase in cannabis use impacts on individual budget constraints. While both boundary constraints are interesting and important it was beyond the scope of this thesis to address them in detail.

3.4.3.2 Identifying the costs and benefits

The notion of opportunity cost underpins cost benefit analysis. The analyst undertaking a cost benefit analysis asks ‘what is the value of the facility in its best alternative use?’ If the answer to the question is zero, the costs should be considered ‘sunk’, and may be disregarded, irrespective of the financial cost of the facility in times past. Text books and monographs make the point that all possible benefits should be included in a CBA (Boadway and Bruce, 1984; Ableson, 2000; Boardman et al., 2001; Elvik, 2001; Commonwealth of Australia, 2006; Zerbe Jr and Bellas, 2006), but in reality actual (dis)benefits included in a CBA are often less comprehensive than the cataloguing of the costs. In the health economics literature, Drummond et al. (2005) suggest there are three types of outcomes to be valued: intangible benefits which are the value of improved health per se to the individual consumer of a program; future health care costs avoided; and increased productive output due to improved health status. Likewise, the Australian Handbook on Cost Benefit Analysis (2006) suggests there are three main types of benefits. Two are similar to Drummond et al. (2005), improvement in health and productivity benefits, but the third category is referred to as amenity and includes improvements in non-market goods such as recreational experiences or the quality of life. This somewhat broader view reflects the wider target audience (i.e. transport, environmental economics) of this handbook (Commonwealth of Australia, 2006). Reporting on a review of the use of CBA in criminal justice literature, Roman (2004) suggests many of the studies in this area of research focus only on the impact on the criminal justice and the drug treatment sectors with some studies including the reduction in crime enforcement costs, changes in public revenue receipts and the private benefits of reduced criminal victimisation (Roman, 2004).

44 Chapter 3

While each of these authors provide useful inputs for deciding what should be included in a CBA of cannabis policies, all reflect the context of the writer. For example, Drummond’s list provides the scope to include wider benefits but primarily reflects health outcomes; the Australian handbook on CBA has less on health and more on amenities and productivity, while the paper from the criminal justice system reflects what is important to that sector. None of these lists appear sufficient to reflect the wide- ranging costs and benefits which need to be considered in a full assessment of cannabis policy.

Others writing the drug policy field provide a more comprehensive list of potential costs and benefits (MacCoun and Reuter, 2001b; MacCoun and Reuter, 2001a; Godfrey, 2006). In Drug War Heresies, (Table 6.1) MacCoun and Reuter (2001) offer a taxonomy of drug-related harms according to category (health, social and economic functioning and criminal justice) and who bears the risk (user, dealer, intimates, employers, neighbourhood and society). The purpose of this table was to encompass the harms related to a wide array of illicit drugs and as such includes many consequences, such as HIV and other injecting-related illnesses (MacCoun and Reuter, 2001a) that are not germane to cannabis. Nor does this table provide any indication of any potential benefits.

Godfrey (2006) categorises potential costs and consequences as follows: Costs a. direct intervention costs b. costs to other agencies c. costs to the individual and families. Consequences d. those benefits to the individual drug user and /or family from reduced drug problems e. resource savings—health care, criminal justice expenditure, reduced resources from social care from reduced drug- related problems f. other value created—increased productivity, reduced accidents related to third parties, value from reduction in crime, decrease in spread of infectious diseases, potential impacts on future drug use and harms g. adverse consequences or spill over effects of policies directed at one problem that might increase others (Godfrey, 2006). 45 Chapter 3

There are both direct (D) and indirect (I) effects of the policies. While both direct and indirect effects of the policies are important, when evaluating drug policy, disentangling the (dis)benefits from the policy change and the (dis)benefits from the use of the drug itself is essential within the welfare model (Godfrey, 2006). For example, changes in benefits that may occur from a policy change, such as shifting from cannabis being illegal to legal, might include decreasing the strain on the criminal justice system, decreasing the negative perceptions of law enforcement, and lessening the harms to individuals of having a criminal record. On the other hand, if the policy change results in increased drug use leading to increased dependence and increased treatment demand, this is a consequence of the drug effect. Both the direct impact of policy and the indirect effects, such as through the drug itself, are important and need to be included but clear attribution of the pathway is necessary for the model. This is illustrated in Table 4. The existing knowledge on the relationship between the use of cannabis (a direct effect) and the indirect effects (i.e. psychosis, low birth weight newborns, and motor vehicle accidents) are incorporated into this work.

Table 4: Relationship between consequences and policy options Consequence (direct/ Pathway Policy: NSW Policy: indirect effect of the status quo legalised– policy) regulated Health   Dependence (I)   Use of Psychosis (I)   cannabis LBW(I)   (D) Motor vehicle accidents (I)  

Educational attainment (I)   Enjoyment (I)   Fines (D) Policy   Criminal record (D) Policy   Stigma (D) Policy   Licence –Users and grower Policy N/A  (D)

Potential costs and benefits for this work were identified from a wide range of sources, including: drug policy (Nadelmann, 1989; McDonald et al., 1994; MacCoun et al., 1996; MacCoun and Reuter, 2001b; MacCoun and Reuter, 2001a; Haden, 2002; Wodak et al., 2002; Haden, 2004; Shanahan et al., 2004; Sindelar et al., 2004; Rolles, 2009), health consequences from cannabis use (Hall and Babor, 2000; Hall and Swift, 2000; Teesson et al., 2002; Carr et al., 2003a, ; Andrews and Tolkien II Team, 2006; Burns et al., 2006; Access Economics., 2008; Calabria et al., 2008; Hickman et al., 2010; Ngui

46 Chapter 3

and Shanahan, 2010); stigma (Lott, 1992; Waldfogel, 1994; Rasmusen, 1996; Lenton et al., 1999 ; Lenton and Heale, 2000; Pager, 2003; Funk, 2004; Ahern et al., 2007; McKeganey, 2010); traffic accidents (Laumon et al., 2005; Biecheler et al., 2008; Mann et al., 2008; Crancer and Crancer, 2010); education attainment (Pacula et al., 2003b; Room et al., 2008; van Ours and Willams, 2009; Horwood et al., 2010; McCaffrey et al., 2010); and consumption (Wilkins et al., 2002; Pudney et al., 2006 ; Legleye et al., 2008; Kilmer and Liccardo Pacula, 2009).

The categorising of costs and benefits in this thesis follows that of Godfrey with some modifications to fit the context of the study. Each of the potential costs and benefits were classified into one of the categories above, and are listed in Table 5. Those items marked with an asterisk are not quantified in this thesis, and for the remainder the chapter number where they are found is included. Those costs and benefits not included and the rationale for their exclusion is discussed briefly below. The five categories of costs and consequences are as follows:

1. Direct intervention costs for both policies—for the current cannabis laws in NSW this includes policing costs, costs of court time, prosecutors, Legal Aid, corrective services, diversion programs, and juvenile justice. For the legalised and regulatory option this includes the costs to the grower, the distributor/retailer, the individual and to the regulatory agency charged with enforcing the regulations. Activities will include: policing, standards, licensing, price setting, eradicating the existing black market, and drug driving programs.

2. Costs to other agencies and personal costs—costs to the health care system, which are an indirect effect of the policies, that occurs as a result of cannabis use are included here. Such costs include treatment for dependence and other health consequences (psychosis, low birth weight babies). Also included in this category are prevention programs such as mass media communications and web- based programs.

Also included are the costs of defence attorneys, fines, and parents’ lost productivity while attending court for juvenile cautioning or juvenile justice conferencing.

47 Chapter 3

3. (Dis)benefits to the individual or family from a change in policy—included in this category is the quantification of the impact of stigma due to a criminal record, personal wellbeing from use.

4. Other value (productivity, accidents to third parties)—this category captures injuries to others from accidents as a result of cannabis use; the impact on educational attainment and subsequent earnings of young persons who start using cannabis at a time when the brain is still developing.

5. Adverse events / spill over—these are the consequences which occur in one area while trying to make changes in another. While three separate implications (social norms, international and political costs, and impact on use of other substances have been identified) none have been included in this study.

Table 5: Costs and (dis)benefits of cannabis policies Direct intervention costs (both Costs to other agencies (Dis)benefits to the Other value Adverse events /spill over policies) (resource savings/costs for individual or family from (productivity, agencies/ government; change in policy accidents to third individuals, families, firms parties) etc.) Criminal justice system [Ch 5, 6 &8] Prevention programs [Ch 8] • Impact on number of • Accidents /injuries to Impact on social norms [Ch 10] • Police- investigation, arrests, admin, o Mass media persons with criminal third parties as a • Change in youth norms towards court, follow-up o Web record & potential stigma consequence of cannabis ** • Courts o School based from criminal record [Ch increased cannabis • Impact on educational attainment • Juvenile justice • Health care costs [Ch 8] 7] use [Ch 7,8] and subsequent earnings [Ch8] • MERIT/diversion programs Cannabis treatment [Ch 8], • Impact on crowding • Prosecution/ Legal Aid • Other health consequences in CJS [Ch 5] • Corrective services Grower [Ch 6] Personal [Ch 5, 7,8] • Value of the enjoyment • Strained international • Growers permit • Health care costs from cannabis use [Ch 8] relationships** • Time to negotiate sales contract • Fines • Political consequences** • Cost of complying with NSW • Legal defence costs workplace laws and agricultural • Parents’ lost productivity regulations (attending court with children) • Testing for potency Distributor /retailer [Ch 6] • Increased rate of uptake of • Attitudinal changes Impact on use of other substances [Ch 8, • Infrastructure costs treatment as a result of not cannabis use 9] • Staffing/ training being illegal (separate becoming more • Tobacco ** • Website sales from increased use) [Ch 8] acceptable among • Alcohol consumption** • Transportation youth - use increases • Other drugs** [Ch 4] Consumer • Search time / risk ** • Impact on family • Licence/course [Ch 6] function as a result of increased cannabis use ** Enforce regulations [Ch 6] • Information – Change in • Impact on workplace • Police[Ch 6] information about product; productivity** • Regulatory body labels indicating potency, (licensing/standards etc.) [Ch 6] potential harms and free • Price setting [Ch 4,6 ] from contaminants** • Contract negotiation [Ch 6] • Black market [Ch 6] • Drug driving testing programs [Ch 8]

(** not quantified in this study) 49 Chapter 3

The costs and benefits included in this CBA are discussed further in the various chapters as identified in Table 5 and will not be discussed further here. Those identified but not included (i.e. marked with an asterisk in Table 5) in the CBA are: • Value of information • Family function as a result of increased cannabis use • Search time/risk • Workplace productivity • Strained international relationships • Political consequences • Other substance use

o tobacco o alcohol consumption o gateway effect The primary reason for the exclusion of these categories listed above is the lack of available data. Estimation of each of these would require additional sub-studies which are beyond this thesis. The potential impact of their exclusion will be discussed in the results of the CBA, Chapter 9.

3.5 Summary

This chapter has provided the justification for the use of conventional CBA to compare a wide array of costs and benefits for two policies to manage cannabis. Further it has been argued that a CBA can be used to examine the issue of allocative efficiency through the compensation test (PCT). The review of the literature demonstrates there have been no CBA of policy options that have explored the wider implications of costs and benefits with most focusing only on cost savings to police. This study while including costs to police and courts includes costs to the health care system, the cost to consumers, growers and to government of both complying with and enforcing the regulatory system.

The following chapters set out the quantification of the cost and benefit components of the CBA models. The next chapter estimates the current use of cannabis and provides projections for increased use under a legalised policy option.

50 Chapter 4

Chapter 4 : Cannabis: rates and quantity consumed

4.1 Introduction

The previous chapters have provided an introduction to cannabis policies and the economic framework for this thesis. This chapter is the first of five chapters which lay out the data required for the CBA model. The purpose of Chapter 4 is to provide estimates of current consumption and projections of future consumption of cannabis in NSW. These estimates, in the absence of sales data, require many assumptions and different sources of data. The results of this chapter will be a key input for subsequent chapters, for example, in Chapter 8 when estimating increased demand for treatment as a consequence of increased use, and in Chapter 6 when estimating the number of hectares of cannabis needed to be cultivated to supply the market.

As will be illustrated, others have estimated the consumption and value of cannabis in their respective countries, and globally, but this chapter adds to the literature in two ways. The first is to provide estimates of current cannabis consumption for NSW. The second and more important addition is to provide a range of estimates of consumption under a legalised framework. Evidence to date suggests that moving from a system where cannabis use is completely criminalised to one where cannabis is illegal but possession/use of a small amount of cannabis does not routinely result in a criminal offence, does not result in a significant increase in cannabis consumption (Donnelly et al., 1999; MacCoun and Reuter, 2001a; Williams, 2004; Hughes and Stevens, 2007; Room et al., 2008; Kilmer and Liccardo Pacula, 2009; Kilmer et al., 2010). But we do not yet know the impact of legalising cannabis as no jurisdiction has yet legalised all aspects (use, supply, cultivate) of cannabis. If cannabis were to be legalised, the question remains as to whether there would be a reassessment of the private costs and benefits from consumption which then leads to a significant change in consumption. Key unknowns include:

• Consumer response to changes in private non-monetary costs (risk, search time, societal attitudes) • Consumer response to changes in monetary price of cannabis • Supplier responses to the changes in risk, prices, and societal attitudes.

51 Chapter 4

Before considering these issues it is necessary to derive estimates not only of prevalence of use but also frequency and quantity of cannabis consumed. The base estimates of prevalence, frequency and quantity consumed per occasion of consumption derived here utilise representative Australian data. International data are employed in the sensitivity analysis. Although as Kilmer and Pacula (2009) point out in their assessment of a global market for cannabis (p 12) “the need to draw on Australian data to predict market estimates for the UK and part of Canada demonstrate the dearth of country specific information…” suggesting that the quality of international data is limited and, given different patterns of drug consumption may not always be relevant.

The outline of this chapter is as follows: first, to inform the remainder of the thesis a description of cannabis, a discussion of issues of potency and other important key characteristics are provided. This is followed by the results of a literature review on methods of estimating cannabis consumption. This review provides the information subsequently used to estimate the current cannabis consumption in NSW. The next step addresses the issue of the potential impact of legalising cannabis on prevalence and consumption. A retail value is estimated holding the price constant at $20 per gram. The chapter concludes with a discussion of the implications and limitations of these results.

4.2 What is cannabis?

Cannabis is a psychoactive drug from the plant belonging to the family Cannabaceae, the genus Cannabis and the species Cannabis sativa and its variants (Clarke and Watson, 2002; Hillig, 2005). The plant occurs in both male and female forms, with only the female plant having the cannabinoids which are responsible for the psychoactive effect. There are around 40 different cannabinoids most of which are not psychoactive. The cannabinoids which are the most prominent are:

• Tetrahydrocannabinol (Δ9 THC) is the main ingredient responsible for the psychoactive effects (Ashton, 2001). It is found in the resin that covers the flowering tops and upper leaves (Hall and Pacula, 2003). • Δ8 THC is found in low concentrations (Ashton, 2001) with ongoing debate as to whether it is less potent than THC (Ashton, 2001) or more potent (Cervantes, 2002).

52 Chapter 4

• Cannabidiol (CBD) is found in most varieties of cannabis but the concentration can vary from a minimal amount through to 95%, and is often responsible for the sedative effect (Cervantes, 2002). • Cannabinol (CBN) is produced as the THC breaks down, and cannabis with high levels of CBN can result in grogginess or feeling sluggish (Cervantes, 2002). • Tetrahydrocannabivarin (THCV) is associated with the fragrance of the plant. High levels of THCV are believed to bring on the psychoactive effect more quickly and dissipate it more quickly (Cervantes, 2002). • Cannabichromene (CBC) may make up to 20% of the cannabinoids profile and may enhance the effect of THC (Ashton, 2001).

The Δ9 THC content of dried cannabis (often referred to as potency) depends on a number of factors. These factors include: the genetic make-up of the plant (Ashton, 2001); the growing conditions and techniques (Clarke and Watson, 2002; Hall and Pacula, 2003) with the unfertilised female plant (sinsemilla - without seeds) producing more resin (Hall and Swift, 2000; Cervantes, 2002; ElSohly, 2008); the part of the plant that is used; and finally how it is prepared and stored as the plant degrades over time (McLaren et al., 2008).

The flowering tops (‘buds’) have the highest THC concentration (Hall and Pacula, 2003) with the leaves, stems and seeds having a lower concentration of THC (McLaren et al., 2008). Plant variety is a major factor in potency, as for example, ‘hemp’ which is grown for its use as a fibre contains very low amounts of THC. Cross-breeding and genetic modification have produced hybrid sub-species with high levels of THC and varying amounts of other compounds (Adams and Martin, 1996; Hall and Swift, 2000). Hydroponic, or other methods of growing cannabis indoors under artificial conditions, is thought to produce higher concentrations of THC than cannabis that is grown naturally (Adams and Martin, 1996; Poulsen and Sutherland, 2000), however, this assertion is debated by others (El Sohly, 2002).

The common preparations of cannabis are marijuana, hashish and hash oil (Hall and Pacula, 2003). Marijuana is the dried flowering tops and leaves and is usually smoked in a hand-rolled cigarette (referred to as joints, reefers, blunts, spliffs) or using a water pipe (bong). When smoked, absorption of THC passes through the lungs into the

53 Chapter 4

bloodstream, and reaches the brain within seconds (Ashton, 2001). Smoking is the most common route of administration in Australia. The leaves can also be processed into a cannabis resin which is referred to as hashish (hash) (Cervantes, 2002) and oil can be extracted from this resin (hash oil). Hashish and hash oil maybe added to foods, cooked and eaten (Hall and Pacula, 2003). When cannabis is consumed orally, the bioavailability is lower as it passes through the liver before reaching the brain with the effect taking from 30 minutes to two hours (Ashton, 2001).

The terms joint and spliff appear to be used interchangeably both within Australia and internationally and it appears they both may, or may not, contain tobacco with the cannabis. This exchange from a users’ chat site illustrates the challenges in defining many of these terms.

Cocacrazy: “My bad, I figured spliffs was a common term. All weed = joint. Mix w/ tobacco = spliff Just to clarify. Also from what I've heard in this thread and from other threads it seems like Europeans mostly smoke spliffs and Americans smoke joints. Is this true?”

Ranunky: “They are both the same thing where I'm from so I like them both. A spliff/joint is tobacco+weed in skins.”

stonedandrolling89: “Joint = cannabis rolled in cigarette paper”

Ostrich: “Eh, i prefer spliffs, but we just call them joints cos here it's pretty much expected that there's tobacco in the joint”

stonedandrolling89: “always here people saying "joints are so much better than blunts" but i thought they were the same... so whats the difference?”

zigzag| dta: “Blunt - cannabis rolled in tobacco leaf “

Bluelighter: “its a cigar. you break it open and roll the weed inside the empty cigar.” (Bluelight, 2010)

In Australia, most (65.4%) of those who have used cannabis in the last year (recent use) state they prefer to use the ‘heads’ of the cannabis plant, and 37.9% prefer the leaf with ‘hydro’ cannabis being preferred by 41.4%. Just 11% report having ever used resin

54 Chapter 4

including hash, and fewer than 10% report having ever used oil (Australian Institute of Health and Welfare, 2008a). For the remainder of this thesis, all mentions of cannabis will be heads or leaf consumed by smoking unless otherwise specified. All estimates of consumption in this thesis will be based on joints which have at least four to six times more cannabis than a bong (Mackenzie et al., 2010). As will be discussed below, failure to account for different amounts of cannabis in a joint or bong will result in different estimates of consumption. Terms such spliffs, bongs or reefers will only be used when directly quoting other studies.

There is one further issue, that of potency. Although it is likely that the potency (Δ9 THC content) of cannabis is higher now as a result of genetic selection and targeted growing practices than it was 20 to 30 years ago, the extent of the increase in potency remains widely debated. Claims of increased potency of cannabis are not new, with claims of increased potency appearing in the 1970s (McLaren et al., 2008), with seizures in the UK in 1979 ranging from 0.2% to 17% THC (McLaren et al., 2008). Studies of a small number of seizures (~1000 to 2000 per year) of cannabis in the US have found THC concentration has increased from 2% in 1980 to 4.5% in 1997, to 8.5% in 2006 (McLaren et al., 2008). Closer examination of the data shows that for domestic samples the per cent of THC rose from 2.78% to 4.78% over the this period (ElSohly, 2008) suggesting that it is non-domestic cannabis which is of higher potency. In a review of the data on this issue, McLaren and colleagues found evidence of potency increasing in the US, an increase and subsequent decrease in the Netherlands, and no change in New Zealand (McLaren et al., 2008) with stable levels in most of Europe (King et al., 2005). A wide search of grey and peer review literature failed to find any published Australian data. Although there may have been some testing for potency by police laboratories, the actual potency of the cannabis consumed in Australia remains essentially unknown. For the remainder of this thesis it will be taken as given that the potency of cannabis used in Australia remains unknown although it is likely higher now than in the 1960s and 1970s.

4.3 Current cannabis consumption

While the rates of recent use (use within the previous 12 months of the question being asked) have declined, particularly since 1998, 9.1% of Australians aged 14 and over report using cannabis at least once in the previous year (see Figure 3) (Australian

55 Chapter 4

Institute of Health and Welfare, 2008b). The prevalence of use differs across age and gender, as does the frequency of use. From the public health perspective the actual number of cannabis users may be of less importance than the small number of regular users who continue to consume large amounts of cannabis at harmful levels (Roxburgh et al., 2010).

Figure 3: Rates of recent cannabis use in Australia in those aged 14+

20 17.9

16 12.7 13.1 12.9 12 11.3 9.1# 8

4 % use recent 0 1993 1995 1998 2001 2004 2007

Source: National Drug Strategy Household Survey (Australian Institute of Health and Welfare, 2008b): # significantly different at 95% level

A review of studies that assessed the size of the cannabis market in various jurisdictions was undertaken. Medline, EMBASE, CINHAL, Google Scholar, and EconLit were searched for the terms: cannabis or marijuana, and demand or market size or supply. In addition, the references from papers located were subsequently examined to retrieve other literature, in particular grey literature, where this issue has been addressed. Brief summaries of these studies are provided below.

Early work in Australia by Clements and colleagues used a combination of information from the early National Drug Strategy Household Surveys (NDSHS), surveys of university students, and “guesstimates” to estimate annual consumption by frequency of consumption (Clements and Daryal, 1999). While these estimates improved on estimates by previous researchers who failed to recognise that consumption would be expected to vary by frequency of use, Clements and Daryal relied heavily on estimates from a student population and self-reported guesstimates with no validation.

56 Chapter 4

Wilkins et al. (2002) estimated the total consumption for 15–45 year olds in New Zealand using data from the 1998 NZ National Drug Survey on the frequency of consumption and the reported number of joints consumed on a ‘typical’ occasion. Average consumption per occasion of use in New Zealand is reported at 0.81 joints for men and 0.61 joints for women with no reported variation by age, or frequency of use (Wilkins et al., 2002). Estimates of consumption are based on 0.5 grams of cannabis in an average joint.

A French study used data on cannabis use from three surveys: a general population survey for those aged 18–64; the European School Survey Project data for those aged 15–16; and a survey that targeted 17 year olds (Survey on Behaviours and Health) to examine the value of the French market for cannabis (Legleye et al., 2008). This paper reports that those aged 15–17 consume about 1 gram per month if not regular consumers, and up to 10 grams if regular but not daily consumers, and 27 grams per month if daily consumers; amounts for older age groups were similar except that regular but non-daily users consumed 5 grams per month (Legleye et al., 2008). This work used 0.33 grams of cannabis in an average joint.

A UK study estimated the size and value of the United Kingdom drug market. They used data from the Australian 2001 NDSHS on the number of joints and bongs of cannabis consumed on a day of consumption (Pudney et al., 2006) and frequency of use data from UK arrestee, school and household surveys. The Australian 2001 NDSHS reports that those who consumed cannabis in the past week indicated they consumed on average 5.5 bongs or joints, those who used in the past month consumed 2.2 while those who only used in the past year consumed 2.04 on each occasion. Recognising the overestimate given a joint usually has at least four to six times more cannabis than a bong (Mackenzie et al., 2010), the final UK study did not use the 2001 NDSHS data directly but inferred that intensive users consume 1.2 grams (±.3 mg) per day while less intensive users consume 40 to 50% of that amount.

There are several US studies, in particular a series titled “What America’s users spend on illegal drugs” which include estimates of cannabis consumed (Rhodes et al., 1997; Rhodes et al., 2000). Notably this series uses a fixed consumption of 18.7 joints per month per user and 0.39 grams per joint. The numbers of users were obtained from the National Household Survey on Drug Abuse (NHSDA). There are a number of

57 Chapter 4

methodologies used and discrepancies in quantities consumed in these reports (Kilmer and Liccardo Pacula, 2009). Some of the differences across the series are with the population used, and whether consumption is based on last month’s consumption, whether there are adjustments for lower frequency use, and whether or not adjustments occur for under-reporting (Kilmer and Liccardo Pacula, 2009).

More recently, the size of the cannabis market has been estimated as part of approximating the size of the global illicit drug market (Kilmer and Liccardo Pacula, 2009). Kilmer and Liccardo Pacula (2009) state “that the lack of information about typical quantities consumed on a use day (for cannabis and other drugs) severely limits the accuracy of demand side estimates” (p 12). The global estimates were based on a set of minimalistic assumptions including: only two types of users (those who have used in the past month and those who have used in the past year but not in the past month); an average number of days of consumption for the two groups, the number of joints per use day generated from American data; and 0.4 grams of cannabis per joint (range 0.3 to 0.5) plus prevalence data from each country.

It is evident upon reviewing these studies that every study addressed the lack of availability of data in a slightly different manner using national data where possible but often researchers needed to resort to using data from elsewhere. While assumptions are still required, the availability of Australian data means the primary estimates for NSW are based totally on available local data.

4.3.1 Measuring cannabis consumption

Understanding current use patterns in some detail and accurately assessing current use is challenging as a result of its illegality. When estimating the value and size of the demand for a legal commodity one can often turn to data collected on production, sales volume, or taxation but when examining an illicit market this type of information is not available.

There are two regular surveys which ask about cannabis use in detail and are designed to cover all or segments of the NSW population. They are the National Drug Strategy Household Survey (NDSHS) and the Secondary School Survey. The NDSHS is an Australian-wide population-based survey undertaken every three years. In 2007, a total of 23,356 people aged 14 and over were surveyed either by the drop and collect method

58 Chapter 4

or by computer-assisted telephone interviews with a 49.3% overall response rate (Australian Institute of Health and Welfare, 2008b). Households are selected by a multistage, stratified area, random sample design with sample sizes selected to result in reliable estimates for various strata. Subsequently each respondent’s data had a weight attached to it to overcome any imbalances in the design and data collection (Australian Institute of Health and Welfare, 2008b). It is these data which are used throughout much of this chapter. One limitation of these surveys is that any person who was itinerant, in prison, in hospital or resided in a university residence would have been ineligible for this survey and if those persons consume cannabis at a rate different from the rest of the population, the rates may be inaccurate. Also, as these data are self- reported, recall may be faulty or the person answering the question may deliberately lie but this is true of all surveys of this type.

Figure 4: Percentage of NSW students who reported using cannabis at least once in past twelve months

50 45 40 12 to 15 year olds 16 to 17 year olds 12 to 17 year olds 35 30 25 20 % students 15 10 5 0 1996 1999 2002 2005 2008 1996 1999 2002 2005 2008 1996 1999 2002 2005 2008

Source: Secondary School Survey 2005, 2008

Another regularly conducted survey is the Secondary School Survey. In the most recent survey (2008) the target was 122 schools representing the government, Catholic and independent school systems covering Years 7 to 12 (Centre for Epidemiology and Research, 2009). The downward trends in reported cannabis use amongst students is similar to that of the population overall (see Figure 4). However, several questions about frequency of use, quantity consumed on day of consumption and policy questions are either different from the NDSHS or not included. Additionally, it is widely understood that this survey likely under-represents cannabis consumption because those who may not be in school (either because they are away on that day or they have left

59 Chapter 4

school permanently) may have higher than average consumption. Due to the lack of comparable questions between the NDSHS and Secondary School Survey and the likelihood that the Secondary School Survey would not capture those missing from the NDSHS, only the data from the 2007 NDSHS were used.

Estimation of the total cannabis consumption (C) for the whole population of consumers required information on the prevalence of use, frequency and intensity of consumption is required. The total consumption can be defined as:

C = Σ[Popk,s * Prevk,s,f * Jf *gm]

Where k= age, s= sex, Pop is the population, Prev is the prevalence of cannabis use; f is the frequency of that use, J is the average number of joints smoked on each day of use and gm is the amount of cannabis in a joint measured in grams. In the work which follows the age categories are 14–19, 20–29, 30–39, and 40 plus; the population is that of NSW aged 14 and over, frequency of use is either daily, once a week or more, monthly, every few months, once or twice a year, not in the past year.

4.3.2 Estimating the number of users and quantity consumed

Three separate steps are required to estimate C. First, the prevalence and frequency of use are estimated from the NDSHS, and then in two steps the amount consumed, on days when cannabis is consumed, is estimated. The decision was made to estimate consumption per day and then quantity per joint as this provides transparency in assumptions whereas the assumptions which have occurred in other estimates of grams per day are not so obvious.

In 2007, of the population of NSW aged 14 and over, 8.6% stated they had used cannabis recently (within the past twelve months). A third of those who use cannabis recently report only using once or twice a year but 40.8% report using cannabis either daily or once a week or more (see Table 6). As a first step in estimating consumption, the frequency categories in Table 6, are converted into number of days used per year, where ‘every day’ is equated to 365 days per year; ‘once a week or more’ is equated to 3.5 times per week or 182 days on average, about once a month becomes 12 days per year, every few months six times per year and once or twice a year is two times a year. The most contentious of these frequencies, given the potential impact on overall

60 Chapter 4

consumption, is the assumption that once a week or more but not every day equates to the midpoint of on average 3.5 days per week. No information was found to assist with this decision so it was decided to use the midpoint. Both the current consumption and the projected consumption with legalisation will be affected by this assumption.

Table 6: Frequency of cannabis use in NSW by age (males and females combined)* Days Total by % of per 14–19 20–29 30–39 40+ Frequency of use category Total year Every day 365 18.3% 8,607 28,769 19,340 23,367 80,083 Once a week or 182 22.5% more 16,970 26,017 24,533 31,311 98,831 About once a 12 11.8% month 12,604 19,094 6,435 13,486 51,619 Every few months 6 14.7% 6,564 24,853 17,387 15,791 64,595 Once or twice a 2 32.7% year 16,217 47,466 39,478 40,212 143,373 Total by age 100 60,962 146,199 107,173 124,167 438,501 Source: Table generated from National Drug Strategy Household Survey data 2007; * numbers may not sum to totals due to rounding.

The next step is to multiply the days used per year by the number who consume then ascertain the quantity of cannabis, on average, that is consumed on each day of use. (Data were generated separately for males and females, and then summed; see Appendix-Chapter 4). As indicated above, this involves two steps, first identifying the number of joints (or joint equivalents) consumed each day, and then estimating the average amount of cannabis per joint. Unlike tobacco or alcohol there is no standardisation in the amount of the drug in a single joint. Table 7 provides a summary of the findings on the number of joints or grams of cannabis on a ‘use’ day from the various studies discussed above. Not all studies reported their data in similar format and thus are not directly comparable.

The primary estimate (column one, Table 7) was generated from the 2007 Australian NDSHS data. This analysis used reported number of joints used on any day when cannabis was consumed, by frequency of consumption. The means and the 95% confidence intervals are presented in Table 7. The average number of joints decreases as frequency of use decreases. The Rhodes and Pudney estimates were used as a sensitivity analysis.

61 Chapter 4

Table 7: Summary of the amount of cannabis consumed per occasion of use Australia NZ US Australia UK Australia. France Use NDSHS Wilkins Rhodes NDSHS Pudney Clements Legleye days 2007 2002 1994 2001 2006 1999 2008 per Joints & Grams Grams Joints per day of use Grams per year cones per per day per year Mean 95% CI M/F^ M/F^ day of use of use month 365 3.50 2.4, 4.6 0.81/0.61 2.53/1.79 5.54 1.2 520 27 182 1.82 1.6, 2.1 0.81/0.61 2.41/1.79 5.54 1.2 364 5 12 1.70 1.3, 2.1 0.81/0.61 1.5/1.31 2.20 0.6 48 5 6 1.33 1.2, 1.5 0.81/0.61 1.5/1.31 2.04 0.6 3 1 2 1.08 1.0, 1.2 0.81/0.61 1.5/1.31 2.04 0.6 1 1 ^M/F = male / female

Of those studies that report joints or bongs per occasion of use, Wilkins (2002) reports considerably lower estimates than the others, and they are constant over frequency of use although they do vary by males and females. Rhodes et al. (1997) also reports the number of joints for males and females but also varies consumption by frequency of use. As already discussed, a previous estimate from the 2001 NDSHS did not account for differences between cones and joints so it is not surprising these estimates are larger than the other estimates. Among the three studies which report in grams, when they are all converted to annual consumption, the Clements estimates are considerably higher than the others and appear similar to the NDSHS 2001 results.

The next step is to determine the quantity of cannabis in each joint. There are a range of estimates in the literature from 0.2 gm (Humphries and Joyce, 1982) in (Pudney et al., 2006); 0.33 gm (Legleye et al., 2008); 0.39 (Rhodes et al., 1997); 0.4 gm (Kilmer and Liccardo Pacula, 2009); 0.5 gm (Wilkins et al., 2002; Wilkins et al., 2005; Gettman, 2007). In an ongoing Australian study using Timeline Follow Back methods with a group of frequent cannabis users, researchers using ‘fake’ cannabis asked participants to indicate the quantity of cannabis per joint; this was then weighed and recorded. The average amount of cannabis per joint was 0.37 grams at baseline (n=39) (Mackenzie et al., 2010). Based on these results, 0.37 gm of cannabis per joint is used as the main estimate in future analysis; this is also the mid-range of other studies reported. For the purposes of sensitivity analysis 0.25 gm and 0.50 grams are used (Wilkins et al., 2002; Kilmer and Liccardo Pacula, 2009).

To summarise, the number of users and the frequency with which they consumed were estimated from the NDSHS 2007 data. Five estimates for number of joints used on a

62 Chapter 4

day of use were obtained, three from the NDSHS 2007 data (using the mean and the upper and lower confidence intervals), one from Pudney (2006) and the other from Rhodes (1994). In addition three estimates of the amount of cannabis in a joint were used 0.25, 0.37 and 0.50 grams.

Table 8 presents a summary of the estimates in total kilograms of cannabis consumed varying both the amount of cannabis per joint (rows A, B and C) and the number of joints used per use day (columns 2 to 6). The first thing to note is the variation across estimations methods for the number of joints. Examining the 0.37 grams per joint (B row), there is a 31% difference between Pudney and Rhodes, but only an 8% difference between the Australian NDSHS (2007) mean results and that of Pudney. The variation between the estimates using the upper and lower confidence intervals is larger than the variation between the 2007 mean and the other two estimates. The estimates with 0.25 gm per joint and 0.50 gm per joint along with the 95% confidence intervals provide possible ranges with results from Pudney and Rhodes falling within that range.

Table 8: Total annual consumption of cannabis in NSW: Summary of results Various assumptions on joints per occasion Source of data 2007 NDSHS US Rhodes Pudney* (1997) Low 95% CI Mean High 95% CI Grams per joint Totals kilograms A. 0.25 gm 25,165 34,211 43,260

B. 0.37 gm 37,245 50,632 64,025 57,393 47,497 C. 0.50 gm 50,331 68,422 86,520

*Pudney estimate combines frequency and amount of cannabis in a joint thus only a single estimate is obtained; it appears as if 0.4 gm per joint was assumed.

One way of assessing the validity is to convert the main estimate of 50,632 kilograms estimate into grams per user per year (115 grams) and compare it to what others have found. The 115 grams per user per year is higher than the range in the literature; Rhodes (1997) reports 84 grams for the US, and Pudney (2006) reports 110 grams for the UK. By way of comparison, Kilmer and Pacula’s (2009) estimates of global consumption found that in the US, Australia and Western/Central Europe a cannabis user would consume on average 96.2 grams per year.

For the remainder of this chapter the main estimate will be 50,632 kg (range 25,165 to 86,520 kg) from Table 8. This range encompasses the low 95% CI with 0.25 gm per

63 Chapter 4

joint, and the high CI with 0.5 grams per joint. The next section examines how cannabis use might change with legalisation.

4.4 Cannabis consumption under legalisation

In order to estimate potential changes in harms (for example, in health status, educational achievement, and demand for treatment) or benefits related to cannabis for the CBA, an estimate of change in prevalence and consumption of cannabis is required. This change in cannabis use will be applied to the harms and benefits in later chapters.

Most reforms in cannabis policy have moved from prohibition with strict criminal penalties to some lesser form of enforcement or liberalising of existing laws while cannabis remains illegal (Room et al., 2008). Therefore, there are no direct estimates of change in consumption associated with legalisation. But there are lessons from liberalisation. Two recent literature reviews have specifically addressed the impact of policy changes on the consumption of cannabis. One was conducted by a group of drug policy researchers (Room et al., 2008) and another by an economist from RAND (Pacula, 2010a). Both are relevant to this discussion. The first report examines a broad literature including policy, criminology and epidemiology and will be discussed first. The second review focuses on the economic literature.

After reviewing available evidence on the liberalisation of cannabis policies in a number of countries, including the United States, Australia, the United Kingdom, Italy, Switzerland and the Netherlands, Room and colleagues (2008) concluded that liberalisation had little or no impact on cannabis use. They concluded that in the United States, those states that had introduced reforms had not experienced an increase in cannabis use among adults or adolescents, while in Australia neither the movement from prohibition with strict criminal penalties to prohibition with civil penalties or prohibition with cautioning has led to a statistical increase in use of cannabis. Similar conclusions were reached about Switzerland, in those cantons that have more lenient policies, and in the United Kingdom when cannabis was reclassified from a Class B drug to a Class C in 2004. Further, in Italy, a country which has gone from penalisation to depenalisation and back to penalisation, there was no evidence that depenalisation has increased cannabis use although it was suggested that more research is required.

64 Chapter 4

Room and colleagues (2008) also concluded that the de facto legalisation of cannabis in the Netherlands “did not in of itself, lead to increases in population levels of cannabis use ...” In the early years of de facto legalisation there were reports of an increase in uptake of cannabis among young people which led to a subsequent increase in the legal age from 16 to 18 years of age to access the coffee shops. But overall, it does not appear as if de facto legalisation led to an increase in cannabis use for those aged 18 and over.

Pacula (2010), with issues of social acceptability and youth initiation in mind, reviewed the economic literature on cannabis. She found that although factors such as the impact of decriminalisation are equivocal, the threat of fines or jail and increased enforcement activities by police have the impact of decreasing initiation of cannabis use in youth. One study included in the review specifically examined the likely impact of a decrease in perceived harms from cannabis and a change in the social norms towards cannabis (Pacula et al., 2001). This study reported that a 10% decrease in perceived harms would generate a 28.7% increase in annual prevalence of cannabis use among youth. This occurs over and above any effect of changes in the monetary price or perceived risk. All of these factors, along with the evidence that youth appear to be very sensitive to the price of cannabis (a reduction of price by 10% will lead to a 3–5% increase in new users) led her to suggest that should social norms become more accepting of cannabis, cannabis initiation and use would increase among youth even though it would not be legal for this group to consume cannabis.

Pacula’s review also found that regular users are sensitive to changes in the monetary price of cannabis, with a 10% decline in price leading to a 2.5% increase in rates of use among regular users. Combining this with the results from models of quit behaviour, which demonstrated that youth who initiate at an earlier age are less likely to quit using cannabis, and the evidence that cannabis prices are correlated with the decision to initiate early, one can infer from these combined studies that lower cannabis prices might extend the duration of the typical use career.

While acknowledging that considerable work needs to be done in this area, Pacula, referring to proposed law changes in California, concludes that “the entire aetiology of marijuana use could change in response to changes in consumption, as there is evidence that initiation, escalation and duration of use will be impacted by elements of this policy

65 Chapter 4

change and in each case the effect is reinforcing: more new initiates, more regular users, and people using for longer periods of time.”(Pacula, 2010a) p17).

Others have taken a direct approach in trying to understand whether the social acceptability of cannabis consumption would shift with legalisation. Daryal (1999) undertook a survey of 327 first year economics students at the University of Western Australia. Based on a 86% response rate, if cannabis was legalised and price held constant, Daryal found consumption among males who are daily users would increase by about 21.3%; consumption by weekly, monthly and occasional users would increase by about 7–9%; while consumption by those who had ‘never used’ would increase by less than 1% and ‘no longer a user’ by 2.5% (Daryal, 1999). Overall it was argued that consumption would increase by about 4%. This led to the conclusion that legalisation had essentially no effect on encouraging non-users to take up cannabis however the authors point out approximately 25% of the respondents were international students from countries where cannabis is strictly illegal, with serious consequences if found in possession of cannabis. If caught, these international students may face major financial impositions and may be unlikely to risk cannabis use (Daryal, 2002). This sample was not a representative sample but once it was reweighted to more accurately reflect the population the increase in cannabis consumption was approximately 13% (Clements and Daryal, 1999).

Additional work in this area was conducted by Weatherburn and Jones (2001) who surveyed a random sample of 579 persons aged 18–29 in NSW on their current and past cannabis use, reasons for not currently using, and their expected behaviours should cannabis become legal. Among those who had never used, or not used in the past 12 months, 47% and 52% respectively reported they currently did not use as they did not like it (see Table 9). Among the ‘never used’ 78% stated they had not used because of its illegal status (three separate reasons) or prohibitive status (two reasons) compared to 37% of those who had not used over past the 12 months. Health concerns were given for 41% of the never used group and for 25% of those who had not used in past 12 months. Very few found cannabis difficult to purchase or too expensive.

66 Chapter 4

Table 9: Reasons for never/no longer using cannabis Not used in past Category** Reason Never used 12 months % % Personal Wouldn’t /don’t like 47 52 Social Friends or family wouldn’t like 21 16 Crim/prohib Cannabis illegal 41 25 Crim/prohib Afraid would get caught by police 10 5 Crim/prohib Afraid may go to gaol 10 4 Crim/prohib Drug testing in sport 4 0 Crim/prohib Drug testing in work place 13 3 Health Worried about health 41 25 Price Too expensive 7 5 Ease Too difficult to purchase 5 1 Source: Table 2 Weatherburn and Jones, 2001; **Column 1 labels and clustering assessment added by MS.

When asked whether they would use cannabis more frequently if it were legal, 63.6% of the whole sample said they would not. Of those who were current users 32% said they would definitely use more as compared to 5.3% of those who had never used and 12.7% of those who had used but not in the past 12 months. Additionally, of those who used weekly or more frequently, 53.5% said they would use more, and among those who consumed less than weekly, 23.4% would consume more. It was reported that rural and younger respondents were more likely to state they would increase use as did those with children.

Both Daryal (1999) and Weatherburn and Jones (2001) provide insights into what might occur if cannabis were legalised. Neither of these studies are representative of the general population; Daryal only uses university students and the Weatherburn sample is limited to 18 to 29 year olds. In an attempt to overcome these limitations, data from all the NDSHS 2007 respondents were analysed, using the question “If marijuana/cannabis were legal to use, would you ...? (Australian Institute of Health and Welfare, 2008b):

• Not use it, even if it were legal and available • Try it • Use it about as often as you do now • Use it more often than you do now • Use it less often than you do now • Don’t know.

67 Chapter 4

Most (95% of males and 97% of females), who had not used cannabis in the past year reported they would not use cannabis even if it were legal, while among recent users 78% said they would use it about as often as they do now (Australian Institute of Health and Welfare, 2008b). Less than 1% said they would use less. Overall 7.1% would use it more often but when the data were analysed across current use categories there was considerable variation (see Table 10). In the ‘never used’ and ‘not in the last 12 months’ categories 12.6% and 10.1% of those aged 14–19 said they would use cannabis if it were legalised.

Table 10: Per cent who stated they would start or increase cannabis use by age and current use category if cannabis was legal

30.0%

25.0%

20.0%

15.0%

10.0%

5.0%

0.0% Age 14-19 Age 20-29 Age 30-39 Age 40+ All Use daily 0.0% 9.9% 2.1% 5.4% 5.5% Weekly or more freq 27.5% 5.1% 2.8% 5.4% 7.2% About once a month 11.6% 5.9% 13.4% 5.2% 8.4% Every few months 14.5% 9.5% 4.1% 5.0% 8.3% Once or twice a year 19.7% 10.2% 8.5% 7.6% 10.6% Never use 12.6% 8.3% 2.7% 1.7% 4.0% Not used past 12 months 10.1% 8.2% 3.9% 6.1% 6.0%

Source: NDSHS 2007 data

Across the older age categories there was a lower percentage who said they would increase their cannabis use, notably 8.2% of those aged 20–29 said they would start to consume more cannabis if it were legalised. In the 14–19 year age group, of those who currently use cannabis daily none said they would consume more cannabis, but 27.5% who are currently consuming at least weekly said they would use more, as did 11.6% of those who currently use monthly and almost 20% who currently use only once or twice a year. Of those who are aged 30–39 years old and who are consuming monthly 13.4%

68 Chapter 4

suggest that they will consume more. The percentage increases for the older age groups were smaller, dropping to approximately 5% for the oldest age category.

In light of the two Australian studies (Daryal, 1999; Weatherburn and Jones, 2001) and the NDSHS data, all revealing an indication of increased use rates with legalisation it was determined that new estimates of use could be calculated. As the NDSHS data were available for both males and females (see Appendix–Chapter 4), and by age categories and frequency of use, it was decided to use these data to estimate potential increases in consumption.

4.4.1 Estimation of consumption under legalisation

The next step was to apportion new users to ‘frequency of use’ groups and to address what ‘more’ among current consumers meant. In order to assess overall consumption it was necessary to allocate those who indicated they would change to a new category. Two methods, which provide a potential range for total consumption, were used. In Method 1 all of those who indicated they would increase consumption were shifted up one frequency category i.e. those who were using once every few months were moved to monthly; those who were using monthly moved to weekly and so on. Those who were already using daily remained at daily. In the first method, those who were new or returning users moved to consuming once every few months.

Method 2 may provide a better longer term estimate as it takes into account that some new and returning users would become regular users whereas in Method 1 they are all only infrequent users (twice a year). In Method 2, current users were treated as described above but new and returning users were allocated according to the base distribution of daily, weekly, etc. The prevalence was the same for both methods, only the distribution varied. Table 11 provides the population rates of cannabis consumption by age group for both the current system and that under the assumption of legalisation. While the population prevalence rate increases from 8.6% to 12.4%, the rates for the younger age categories increase by substantially more, from 11% to 23% for those aged 14 to19 and from 15.4% to 23% for those aged 20 to 29.

Consumption was then estimated for both methods following the identical steps as for generating the original estimates. These were: 1) to generate days of use by multiplying frequency of use by number of users in each category; 2) then to multiply days by

69 Chapter 4

number of joints used per day of use; 3) then multiply numbers from the second step by grams per joint; 4) and then sum and convert into kilograms. As done previously, males and females were estimated separately, and estimates were generated for the main estimate using the Australian NDSHS data and 0.37 grams per joint. The results are given in Table 11.

Table 11: Prevalence of use in NSW by age group with and without legalisation Age category Status quo Legalisation New & Users Current Total returning 14–19 11.0% 12.3% 23.3% 20–29 15.4% 7.8% 23.2% 30–39 10.8% 3.4% 14.2% 40+ 3.9% 2.7% 6.6% Overall 8.6% 3.8% 12.4% Number 438,501 261,844 700,345 Source: NDSHS 2007 data

Under Method 1 the total consumption increases by 8% (from 50,632 to 54,665 kilograms per year), while it increases by 55.9% (from 50,632 to 78,940 kilograms per year) using Method 2. As well as providing main estimates for Methods 1 and 2 (in bold), Table 12 incorporates many of the previous assumptions. The total consumption for each method varies by the amount of cannabis per joint, and by the mean, and low and high confidence intervals on amount consumed on a day of consumption (as per Table 8).

Table 12: Estimating the total consumption of cannabis consumed under legalisation framework Original Method 1 Method 2

Average Total CI Lo Total CI Hi CI Lo Total CI Hi /joint 46,73 A. 0 .25 gm 34,211 27,143 36,936 39,220 53,338 67,461 2 69,16 B. 0.37 gm 50,632 40,171 54,665 58,045 78,940 99,842 4 93,46 C. 0.50 gm 68,422 54,285 73,872 78,439 106,676 134,922 5 Source: NDSHS 2007 data, McKenzie (2010), NCPIC (2010);

The difference of 24,275 kilograms in annual consumption between the two methods can be accounted for by the difference in distribution of use days. This is apparent in Figure 5 where the total number of users (700,345) is distributed according to age and use categories. The impact of distributing new and returning users according to the

70 Chapter 4

previous distribution is apparent; combining this effect with the amount consumed generates the difference in total consumption.

Figure 5: Estimated number who consume cannabis by age category and frequency of use (Methods 1 and 2)

400,000 400,000 Method 1 Method 2

300,000 300,000 14-19 20-29 14-19 20-29 30-39 40+ 30-39 40+ 200,000 200,000

100,000 100,000

0 0

Another way of illustrating the effect is to compare the average consumption. In the base estimate, the average annual consumption per consumer was 115 grams, in Method 1 it is 78 grams, and Method 2 it is 112 grams. It does not seem a realistic assumption that under legalisation that the average annual consumption would decline from 115 grams per year to 78 grams. Nor does it seem realistic that all new users would consume cannabis only a once or twice a year. Therefore, Method 2 (mean estimate and 0.37 grams per joint) will be used as the main estimate for modelling the impacts of legalisation in Chapters 8 and 9.

4.5 Impact of price changes and using price elasticity to estimate demand

The assumption of this chapter, and indeed this thesis, is that the price of cannabis will be held constant. This is a strong and contentious assumption, and will only be possible if the market is highly regulated and these regulations are enforced. This issue is further discussed in Chapter 6.

71 Chapter 4

4.5.1 Price elasticity literature

Despite having made the assumption that prices would remain constant under the public health framework, it is worth considering what might occur if the money price does not remain constant. One would normally turn to the price elasticity of demand to estimate how a price change might impact upon the demand for that good. Price elasticity is the per cent change in consumption associated with a 1% change in price. When the price elasticity is greater than the absolute value of one, it is considered as relatively price responsive and is referred to as elastic. When the absolute value of the price elasticity is between 0 and 1, this is considered price inelastic, and if equal to 0, this is perfectly inelastic.

Price elasticities (PE) are useful for projecting the demand for a good as a result of a price change whether that price change is as a result of taxation, the supplier’s decision to increase the price or a relative price change. Estimating price elasticities requires data on prices, and quantity consumed at each price. A meta-analysis of 132 demand for alcohol studies which produced over 1000 price elasticities provides some insights into issues to consider when contemplating using price elasticities (Gallet, 2007). This meta-analysis found a number of factors impacted on the price elasticity for alcohol: the model and method of analysis; whether the paper was published in a highly-ranked journal; the type or form of the data; the country; the type of alcohol beverages, and drinker’s age (Gallet, 2007). Some of these factors, such as age, type of alcohol and country are understandable but other factors such as differences in the model, form of the data or whether the paper was published in a highly-ranked journal are rather problematic for those wanting to use the results.

Estimating PE for cannabis brings additional challenges, as not only are there no transaction data, the quality and potency of the cannabis are unknown to both the purchaser and the analyst. In the absence of these data how is it that we have price elasticities on cannabis? There are two key ways these estimates are derived: using behavioural economics, and by analysing large cross-sectional or longitudinal survey data (Chalmers et al., 2010). Behavioural economic methods use experimental techniques with hypothetical drug purchases to assess the impact of price changes on demand. This research has predominantly not included cannabis (Petry and Bickel, 1998; Bretteville-Jensen, 1999; Petry, 2000; Bretteville-Jensen and Biorn, 2003;

72 Chapter 4

Chalmers et al., 2010) and those which have (Goudie et al., 2007; Cole et al., 2008), have focused on polydrug users, and the impact of perceived quality of cannabis on prices, with prices at or around the current street price. Neither examines directly the impact of changes in the price of cannabis.

The other method uses large data sets such as the Australian NDSHS, the US National Educational Longitudinal Survey or the Monitoring the Future Survey. These analyses are cross-sectional, i.e. they do not examine individuals’ responses to price changes, but rather compare consumption across some geographical area while adjusting for a suite of factors. For example, Cameron and Williams use multiple years of the NDSHS with the Australian states and territories as their unit of comparison (Cameron and Williams, 2001), while Pacula and colleagues (2003) use a sample from the National Educational Longitudinal Survey with their unit of comparison the 50 US states (Pacula et al., 2003a). In most of these studies the prices are constant within the geographical area of interest, with prices being obtained from a variety of sources (undercover police operations, surveys etc.) but not from the individual whose consumption behaviour is being analysed.

Both of these methods of estimating elasticities have limitations but even if they result in valid price elasticities, they are only relevant around a small price change because the elasticities are calculated based on the slope of the demand curve, and the slope varies as you move along the curve. This raises two further issues: 1) the actual demand curve for cannabis is unknown (Kilmer et al., 2010; Pacula, 2010a); and 2) it is expected that without regulations to hold the price constant, changes in retail price due to legalisation of cannabis are likely to be large thus the price elasticity around current price is unlikely to be usable. As cannabis has not yet been legalised this second point is in part speculative, but defendable as the price of cannabis has already declined sharply with the introduction of legalised medical use of cannabis in California (Caulkins, 2010b), and the cost to grow and market cannabis is so low that without regulation the market forces would result in much lower prices (See Chapter 6 and Caulkins, 2010b).

Kilmer et al. (2010) using the assumption of a ten-fold drop in price should legalisation occur without price controls in California projected a 76% increase in consumption. They stress that these estimates are very tenuous given the lack of data on which they are based.

73 Chapter 4

4.6 Estimating the retail value of cannabis

A retail price is necessary in order to estimate the value of the cannabis consumed. There are a number of caveats around any price. Current prices are based on perceptions of quality, cannabis strains, hydroponically grown versus bush, often estimated weights, the relationship between buyer and seller, location, and quantity purchased and likely other factors. The price of cannabis used throughout this research is the median street price in NSW for a gram of cannabis ($20) obtained from the NSW Illicit Drug Reporting System (IDRS), and the Ecstasy and Related Drug Reporting System reports (EDRS) (Phillips and Burns, 2008; Scott and Burns, 2008). The IDRS and the EDRS are drug monitoring systems which collect price, purity, availability and patterns of use of heroin, methamphetamine, cocaine, ecstasy and cannabis. These data are collected annually in major cities in Australia from drug users who represent a sentinel user population, from experts who work with drug users, and from existing data bases such as customs and seizure data.

The median price per gram of cannabis in NSW has been reported as stable or possibly having declined over recent years (Clements, 2004; Australian Crime Commission, 2008) although there are regional variations. There are reports that cannabis is sold at significantly higher prices in isolated northern communities. It is also probable that there is volume driven differences in pricing. The IDRS and EDRS reported street price for a gram of cannabis within NSW ranging between $10 and $300 per gram for hydro and $10 and $90 per gram for bush cannabis in 2007.

Table 13: Street value of cannabis Totals Low CI (0.25 gm per Mean (0.37 gm per High CI (0.5 gm per joint) joint) joint) Original consumption kg 25,165 50,632 86,520 Value @ $20 /gm $503,300,000 $1,012,640,000 $1,730,400,000 Consumption after legalisation (assumes new users distributed across all use frequencies– Method 2) kg 39,220 78,940 134,922 Value @ $20 /gm $784,393,738 $1,578,800,000 $2,698,431,882 Difference between original value and new estimates with legalisation Difference in kg 14,055 28,308 48,402 Difference in value $281,093,738 $566,160,000 $968,031,883 Source: (Stafford and Burns, 2009)

74 Chapter 4

Using the median street price of $20 per gram of cannabis, the range for the total street value of cannabis consumed annually in NSW is $1,012.6 million (range $503.3 million to $1,730.4 million) and increases to $1,578.8 million (range $784.4 million to $2,698.4 million) post legalisation (see Table 13). The range takes into account both variations in the amount of cannabis per joint and the number of joints used on a given occasion.

4.7 Discussion

In this chapter the literature on estimating cannabis consumption was reviewed. After assessing the strengths and limitations of the various methods and data sources in these papers, a method was derived for estimating current cannabis consumption in NSW for one year that reflected not only what was learned from the previous studies but also the data available. Others have generated such estimates (Wilkins et al., 2002; Clements and Zhao, 2005; Pudney et al., 2006; Legleye et al., 2008; Kilmer and Liccardo Pacula, 2009) for their respective countries and for the world (Kilmer and Liccardo Pacula, 2009). While the results of the previous studies may reflect what occurs in their country (Wilkins et al., 2002; Pudney et al., 2006; Legleye et al., 2008), several were constrained by lack of data, (Clements and Zhao, 2005; Pudney et al., 2006) or the need to generalise across countries (Kilmer and Liccardo Pacula, 2009). The estimates herein have built upon the methods and assumptions used by others but use only Australian data in the results.

International evidence suggests that to date the current liberalisation of cannabis policies has resulted in little or no increase in cannabis use. It does appear however, that price may matter, particularly for youth, when deciding whether to initiate cannabis use. Price may also matter in the quantity that current regular users consume suggesting that not only do lower prices lead to an earlier uptake, but potentially a longer cannabis using career with heavier use. Research suggests that those who start using at a very young age often use longer (Pacula, 2010a). Changes in the social acceptability of cannabis through legalisation may also have a significant impact on not only the prevalence of use but also on the frequency of use and the duration of the cannabis– using career. Currently many cannabis users begin use in their late adolescent years and often cease regular use in their late twenties or early thirties (Australian Institute of Health and Welfare, 2008a). Changes in the social acceptability and legality of cannabis may lower the risk of stigma and lead to longer use careers.

75 Chapter 4

Two estimates were generated for the potential increase in the total consumption under legalisation holding money price constant. In both methods those who state they will increase use are shifted up one use category from where they are currently (i.e. from monthly to weekly). Method 1 consumption estimate was 54,665 kilograms per year (an 8% increase), while Method 2 results in 78,940 kilograms per year (a 55.9% increase). The difference between the two methods is in how those who are not currently using but state they will start using are dealt with. One method is a more conservative estimate: those currently not using who indicate they will start using are placed in the lowest category ‘use a couple of time a year’. The other method, which allocates new users according to the existing distribution of users, likely reflects more closely what will happen. Whether the lower or upper estimate of quantity consumed is correct, both estimates start with an additional 3.8% of the population of NSW aged 14 and over indicating they will begin to use cannabis.

The primary estimate of consumption resulted in an annual average consumption per user of 115 grams which is higher than most other estimates, including the 96.2 grams per year global estimate of Kilmer and Pacula (2010). This difference may reflect the additional precision of the estimates in this current research, with additional categories of frequencies and Australian data for the number of joints consumed on use day. Or it may not be statistically different from the other published estimates (84 to 145 grams per user per year) which are within the range of the high and low confidence intervals estimated here.

There are a number of limitations and caveats to these estimations. In particular they rely on the validity of the survey-based data. These data rely on the recall and honesty of the respondents and that the population who are recruited, and who chose to respond to the survey are representative of the sample. Self-report surveys have been found to be reliable means of eliciting information about drug use (McElrath, Dunham and Cromwell, 1995 in Weatherburn (2001) although others report high rates of under- reporting by younger cannabis users (Kilmer and Pacula, 2009).

The questions included in surveys most often ask about last occasion, last month or usual behaviours and from these data a number of assumptions must be made to estimate the frequency and quantities used. In particular these results rely heavily on the response to the question regarding “what would you do if cannabis was legalised?”

76 Chapter 4

and assumptions around the distribution of new users and frequency of use. Throughout this analysis, a midpoint (3.5 days per week) was used as an average number of days of consumption for those who ‘do not use daily’ but use ‘weekly or more’, if an average of 2 times a week was used the total consumption changes from 50,632 kg to 45,431 kg; similarly if daily is not every day but 300 days a year, the total decreases to 43,895 kg. While these assumptions may have an effect, they are held constant for the base case and both legalisation methods. And the consumption estimates per user lie within the range of other estimates of consumption found in the literature.

In an attempt to generate a legalised and regulated policy option for cannabis that addresses the public health concerns related to increased cannabis consumption, and the additional uncertainties around the impact of price changes, all future modelling of costs and benefits will be estimated assuming the price is maintained at its current level.

77 Chapter 5

Chapter 5 : Cannabis and its impact on the CJS

5.1 Introduction

A key component of understanding the costs and benefits of the current cannabis laws is determining the resources expended by the criminal justice system in enforcing those laws. This chapter outlines in some detail these costs to government. Previous assessments of the criminal system as it pertains to drug use have taken a macro (top- down) approach to costing police resources (Mayhew, 2003; Baker and Goh, 2004; Bates, 2004; Miron, 2005; Moore, 2005; Drug and Alcohol Office, 2007; Collins and Lapsley, 2008; Kilmer et al., 2010) or analysed only a small portion of the police resources (Brooks et al.; May et al., 2002; May et al., 2007). A significant addition to drug policy research is the estimation of the costs of police time using a survey of police officers developed for this thesis. The results of this survey, police encounter data and unit record court data are then combined to provide a detailed estimate of costs to the CJS. Detailed analyses have not previously been applied to this issue and prior to making informed decisions such data are needed (Weiner et al., 1987). As an analogy, considerable effort has occurred in the health care sector developing more precise cost estimates of activities such as case-weighted costs for acute care, emergency and subacute care. These methods allow for a more precise attribution of costs to activities and outcomes.

The structure of this chapter is to first layout potential relationships (pathways, causality) between crime and cannabis use. This is followed by a more detailed description of the cannabis laws in NSW, and the components of the criminal justice system. This is then followed by detailed descriptions and analyses of the data (utilisation and resources) in each component. The last section of the chapter is a discussion of the total costs to the criminal justice system, highlighting the new knowledge this chapter adds to the debate and how these findings are used in Chapter 8.

5.1.1 Pathways

Cannabis is illegal, and one substantial cost of the current policy is enforcing the laws. In order to account for these expenditures, plus estimate costs under a future legalised and regulated policy a clear understanding of the pathways between the two is required. One way of exploring the relationship between criminal behaviour and cannabis is to 78 Chapter 5

consider four pathways between drugs and crime as developed by Pernanen (2002). Three of Pernanen’s pathways follow Goldstein’s tripartite framework (1985), and the fourth (number one in the list) recognises specific substance-based laws. The four pathways are 1) the substance-defined pathway (drug offences); 2) the illegal system model (gang related); 3) the pharmacological or intoxication pathway (offences committed as a result of intoxication); and 4) the economic-compulsive pathway (theft to purchase drugs). A review of the pathways as they pertain to cannabis was conducted. As a result of this review only those offences that are in the first pathway, cannabis offences, will be included. The interested reader is referred to Appendix– Chapter 5.

The costs of addressing issues of illicit system pathway and cannabis are not included in this work for several reasons—the first is that there were no data, but importantly even if there were data it is not clear how the marginal cost of dealing with cannabis would be handled given that the predominant focus of gangs are other drugs, such as heroin, methamphetamines and cocaine. Additionally as will be discussed in Chapter 6, the move to a legalised market will not in of itself ensure that any control gangs have over cannabis will be relinquished. The review of the economic compulsive pathway does not suggest any strong relationship between cannabis use and acquisitive crime. And while there may be some evidence of an association between higher rates of cannabis use by juveniles with increased rates of delinquency and potentially economic– compulsive offences, the results are not conclusive. This pattern does not appear to exist among adults.

5.2 Documenting existing laws and regulations pertaining to cannabis in NSW There are both federal and state laws pertaining to cannabis. Foremost, the importation, exportation, trafficking, possession and use of cannabis across the Australian border are all illegal activities. The most serious offences, involving imports or exports of commercial quantities (100kg and above for cannabis, 50kg and above for cannabis resin and 2kg and above for cannabinoids), can result in penalties of up to life imprisonment. There are federal offences targeting commercial cultivation of cannabis, domestic trafficking of cannabis (i.e. sale and distribution within Australia), and possession of cannabis. Federal laws extend to the possession of cannabis for personal

79 Chapter 5

use (carrying a maximum penalty of two years imprisonment) but domestic cannabis offences are generally enforced by state and territory law enforcement agencies whilst the Australian Government law enforcement agencies target border importation (McLaren and Mattick, 2006). There are few cannabis seizures by border protection, and most seizures are seeds (Willis, 2010).

Australia’s cannabis legislation differs between states and territories, although all legislative approaches are prohibitionist, meaning that any activities involving cannabis remain illegal. In NSW, the possession of any amount of cannabis is a criminal offence under the Drug Misuse and Trafficking Act No. 226, 1985 (NSW Parliament, 1985). Cannabis offences can attract serious penalties such as imprisonment although there is the option of issuing a caution, or diversion into treatment or education if the offence is minor. Since 2000, NSW police officers have been able to issue a formal caution to adults in possession of up to 15 grams of cannabis under the diversion program known as the Cannabis Cautioning Scheme (Baker and Goh, 2004; NSW Police Force, 2007).

In 2005, as a result of an increase in the of number of hydroponic cannabis operations detected by police in NSW, the perceived greater productivity of enhanced–indoor grown cannabis plants, the theft of electricity from the electricity grid by hydroponic growers and increased risks to children living in this environment, a number of amendments to the Drug Use and Misuse Bill were passed by the NSW Parliament (Tebutt, 2006). Commercial quantity was set at five times lower for indoor growing compared to outdoor grown cannabis, with commercial quantity for indoor cultivation was set at five to 49 plants. However, the Act gives recognition that home growers may cultivate this amount for their own use, thus for prosecution at this commercial level an intent to sell plants must also be demonstrated.

There is an additional aggregated offence in recognition of occasions when indoor cultivation occurs in the presence of children; there are additional risks for children in this environment (fire, electrocution, extreme heat, dangerous chemicals, insecticides and fumes and airborne bacteria). Finally, in recognition of the widespread practice among organisers of hydroponic cannabis operations to steal electricity from the electricity grid the Drug Use and Misuse Bill also amended the Electricity Supply Act 1995 increasing the maximum penalties associated with this practice (Tebutt, 2006).

80 Chapter 5

Juveniles are dealt with under the Young Offenders Act 1997 (New South Wales Consolidated Acts, 1997 (as at 27 November 2008). The intent of the Young Offenders Act is to guide young offenders who commit certain types of offences, including drug offences, away from court proceedings by directing them to alternative forms of intervention such as cautioning or warnings (NSW Police Force, 2009c).

5.3 The criminal justice system (CJS) as it pertains to cannabis Within the NSW CJS there are several sectors of importance: the New South Wales Police Force, the courts (Magistrates or Local, District and Children’s), the prosecutors and Legal Aid, the Department of Juvenile Justice and the Department of Corrective Services plus the Magistrates Early Referral into Treatment (MERIT). MERIT is a diversionary court program. Although there may be variation as to how suspected cannabis offenders are dealt with by the police, courts and corrections services, Figure 6 provides a simplified flow chart for some of the more common possibilities for offenders who were aged 18 years and over when the offence occurred. The process is similar for juveniles, with additional diversionary steps available to police and courts (this is described further below).

5.4 Measuring and valuing the resources used in the CJS

Key to assessing the economic implications of introducing a new policy is documenting the resource implications of the current laws. In addition to identifying which types and numbers of crimes to include, the costs attributable to those crimes must be identified. This can include costs to the justice system, victims costs, precautionary expenditures by business and individuals, avoidance behaviours, prevention and rehabilitation programs, over deterrence (activities not undertaken by innocent people for fear of being accused of criminal activity), criminal justice costs, burden imposed on incarcerated offenders and their families, and loss of quality of life (Brand and Price, 2001; Mayhew, 2003; Cohen, 2007). In this chapter, the focus is on the criminal justice system as well as the direct personal costs to the potential offender and their family. For example, legal costs are included, as are lost time at work for a family member to attend juvenile conferencing. Lost productivity due to incarceration is included in Chapter 8. All total costs are presented for a full year in AUD 2006-07 unless otherwise indicated.

81 Chapter 5

Figure 6: The substance-defined crime pathway for NSW (a simplified flow chart)

Population (≥ 18)

Does not commit cannabis Commit a cannabis offence offence

Not detected by police Detected by police

Possess/use ≤ 15 gm Possess/use ≤ 30 gm Cultivate Supply

Formal caution Formal caution Charged Not charged 1st 2nd

Plead guilty Plead not guilty

Sentenced Guilty - no conviction Found guilty & Not guilty recorded sentenced

Once the various components of the criminal justice system and the frequency of cannabis encounters with police, courts, and corrective services are quantified, the next step is to attach a value to the resources. The approach taken in this work is to cost each sector separately, but first a review of the existing literature as it pertains to documenting the costs of enforcing cannabis laws.

82 Chapter 5

5.4.1 Literature review: Issues when estimating CJS costs

Whether the objective is to estimate the costs of drug use to society (Collins and Lapsley, 2008); the social costs of cannabis harms (Moore, 2007); the costs of crime (Brand and Price, 2001; Mayhew, 2003; Rolling, 2008); the budgetary implications of cannabis policy (Miron, 2005; Kilmer et al., 2010); the impact of cannabis cautioning (Brooks et al.; Baker and Goh, 2004; Crime Research Centre, 2007) or to evaluate specific policies (May et al., 2002; May et al., 2007) it is important to identify the resources expended on various types of activities and the implications of changes in those activities.

Cohen argues that tallying the costs of crime prevention programs are relatively straightforward, at least in comparison to quantifying the intangible benefits (Cohen et al., 2004). This may be true but nonetheless this is not a straightforward exercise. For example, in policing, it is easy to understand that fewer police resources might be used when dealing with one simple cannabis offence than would an investigation into a significant importation of methamphetamines. However, a program to manage cannabis offenders may extend across a number of jurisdictions such as police, courts, corrective services and health and thus involve substantial amount of resources. Consideration must be given to the direct CJS expenditures, plus the use of other public resources and the use of other facilities which may be free or discounted to the program (Roman, 2004). These costs can come from a variety of sources such as budget papers, annual reports, data collection and key informant interviews but the data needs to be gathered in such a way that comparisons can be made to the counterfactual or alternative program (Roman, 2004; Drummond et al., 2005).

There are two main methods of estimating costs—micro costing (bottom-up) and macro costing (top-down) methods. Regardless of what method (or a combination of the two methods) is used the researcher must understand the context, and processes within the systems being evaluated.

Both methods have flaws. Using police resources as an example, the simplest form of macro costing would involve simply dividing the total annual police expenditures by the number of offences (or offenders) in which the police were involved. This would result in an average cost per known offence (offender); one could then multiply the number of offences of interest (i.e. cannabis offences) by the average cost to obtain an estimate of 83 Chapter 5

the total police expenditure on cannabis offences. Simple adjustments can be made to this process, for example by weighting the denominator for the expected resource intensity of various offences (Moore, 2005); or excluding 50% of the possess/use cannabis offences on the basis that they would have been detected while investigating other offences (Miron, 2005). However, without some measurement of actual cost there is no way of knowing whether any of these adjustments improve cost estimates.

Other adjustments exclude some proportion of total police expenditures to take into account that available statistics do not reflect all police activities such as the time spent on traffic and safety management and community policing (Mayhew, 2003; Moore, 2005; Rollings, 2008; Caulkins, 2010a). Some have excluded 30% (Rollings, 2008); or 10% of expenditures (Moore, 2005) while others argue for including only one third of police expenditures on arresting offenders (Caulkins, 2010a), but once again it is not known which, if any of these decisions, is correct.

Micro costing methods, on the other hand, rely on being able to identify all the resources required to handle a given activity—for example, the resources to manage a cannabis caution would include identifying all the steps in the procedure, which personnel are directly and indirectly involved, other resources used—and then applying salary costs/prices to all resources identified. With respect to the cannabis caution this might include time to seize and weigh the cannabis, check the identity of the potential offender, complete all necessary paper work, secure and/or destroy cannabis (see Baker and Goh, 2003) and then summing to obtain a total. For a relatively straight-forward police encounter such as a cannabis caution, such a process may not be too complex, but when many months of detective work or investigation is required, involving large teams, resulting in numerous arrests or charges, and multiple court hearings it becomes increasingly difficult to ensure all necessary costs are included.

5.4.2 Top-down approaches

Collins and Lapsley (2008) use a top-down approach to allocate individual state and national data on police expenditure, according to the proportions of detainee hours in police custody by offence (2002 National Policy Custody Survey) and the fractions attributable to illicit drugs and alcohol from the Drug Use Monitoring in Australia (DUMA) and Drug Use Careers of Offenders (DUCO) data. Information on

84 Chapter 5

dependency (DUMA) and intoxication of offenders (DUCO) is collected from offenders in the police station and combined with information on type of criminal charge. Collins and Lapsley (2008) do not apportion police costs between drug types.

Brand and Price developed total and average costs for various costs of crime in the UK (Brand and Price, 2000). Police costs were apportioned using a top-down method. The total police budget was split into crime and non-crime related components, and then the crime component was split by offence types based on the proportion of other CJS resources spent on each resource, but costs were not estimated for drug type. The CJS costs were estimated based on a computer model of flows and costs. A similar study in Australia to estimate costs of individual crimes did not apportion police or other CJS costs to different crime types but rather presents police costs only as a total (Mayhew, 2003; Rollings, 2008).

Moore (2005) approached the problem of allocating police expenditures attributable to all illicit drugs by combining the annual police expenditure data with data from a National Police Custody Survey of all police services (Taylor and Bareja, 2005). This survey reported on every occasion, throughout the month of October 2002, when a person was lodged in a police cell. Results of this study include the average number of hours a person was detained by 14 main offence types. These data (frequency and average duration by offence type) were then used to apportion the total police budget (minus 10% for those activities not represented by this type of police statistics). This method (referred to as Police Allocation – Method A from here on) assumed there was a direct relationship between the number of hours an offender was held in police custody and the total amount of police resources that were devoted to that offender. This method sought to overcome one of the criticisms: that of assuming each offence consumed the same resources.

An alternative approach (referred to as Method B) used offence type, the number of offenders in each of the 14 main offence types, and the proportion of offenders in each offence type that once found guilty received a penalty of incarceration to apportion the expenditures (Australian Government Productivity Commission, 2007). Rather than using time in police cells as a weighting for allocation of expenditures (as in Method A) this method used the probability of jail as an indicator of the severity of crime. Here the assumption was that the more severe the type of crime the more resources may be

85 Chapter 5

expended on average, on detecting, and arresting and processing offenders. Once costs were estimated for all illicit drugs using this method, costs were apportioned to cannabis based on proportion of cannabis offences.

Another method generated the percentage that cannabis offences (minus 50% of possess/use offences) were of the total offences (Miron, 2005). This percentage of total arrests attributable to cannabis (minus 50% of possess use offences) was then multiplied by the total police expenditures to estimate the amount that would be freed up if cannabis were legalised. The decision to use 50% of the possess/use offences was based on the assumption that a significant number will be under arrest for other offences and cannabis will be detected only while investigating these other offences. This will be referred to as Method C. Miron uses 50%, but cites a range of concurrent offences from 33% to 90.5%. An alternative assumption of cannabis being detected and then being charged with other offences was not considered by Miron.

Before moving to exploring alternative methods for estimating police costs it is worth examining the range of expenditures obtained from the top-down methods A, B and C as described above. To demonstrate, data from NSW on offences and expenditures by the police department for 2006 are used to provide three estimates for costs for policing related to illicit drugs. There were 1,315,676 recorded offences of which 24,271 were illicit drug offences. Of these, 13,775 were recorded cannabis offences (Bureau of Crime Statistics and Research, 2006b) (see Table 14).

Table 14: Cannabis offences Offences % of total

Total offences NSW 2006 1,315,676 100% Illicit drugs offences 24,271 1.8% Cannabis offences* 13,775 1.0% NSW Police Force expenditure $2,108 (millions 2007 AUD) Source: BOCSAR Police 2006; *Includes cannabis cautions

Table 15 presents results from applying the various assumptions described above to the data on all illicit drug offences in Table 14. The total recurrent expenditure on police, less payroll tax and revenue from own sources, was obtained from Productivity Commission documents (Australian Government Productivity Commission, 2007). Offence data and the court outcome data used in Method B were obtained from NSW

86 Chapter 5

Court data (Bureau of Crime Statistics and Research, 2006a). The final column is the average cost per offence for each of the methods.

Table 15: Implications of different assumptions when allocating police expenditure Police expenditure Police expenditure Method to apportion police costs millions (1 year, NSW) per offence Method A – using time in police cells as weight $88.1 $6,394 Method B – using rate of incarceration and $81.1 $5,884 frequency of offences as a weight Method C) –using only 50% of possess/use $29.1 $889 offences

What is immediately apparent from Table 15 is the substantial variation in the expenditures across the three methods of estimation. They range from an annual expenditure of $29.1 million to $88.1 million. Without additional information there is no way of knowing which, if any, is the most appropriate indicator of police costs related to cannabis offences.

5.4.3 Micro costing approaches

Recognising the flaws in top-down methods, a group of British researchers undertook micro costing of police activity for cannabis offences (May et al., 2002). By using custody records they ascertained that police officers in Britain took on average 3.5 hours each to deal with a cannabis offence (two officers) plus time to prepare for prosecution. This was determined to be comparable to the five hours per officer per case (again two officers) that an arrest for a cannabis offence was estimated to take as reported to the UK Home Affairs Select Committee on Drug Policy (May et al., 2002). Full costs were estimated by dividing total police expenditures by total FTE police officers and then by average number of hours per year to obtain an average cost per police officer per hour. This hourly cost was then multiplied by the average number of hours per case for a unit cost of £500 per case in 1999. This average cost included stop and search, arrest, conveying to station, booking in, repeat search, placing arrestee in cell, arresting officer compiling necessary files, and completes recording, and the taking of photographs and fingerprints. Notably they do not adjust downward for concurrent offences (May et al., 2002).

In a subsequent study examining the consequences of reclassification of cannabis these same researchers observed police activities, and among a sample surveyed, reported that

87 Chapter 5

the time to administer a street warning for cannabis, including confiscate and destroy the cannabis and complete all necessary paperwork ranged from 45 minutes to two hours (May et al 2007). An estimate of the number of policing hours and resources freed up for use elsewhere following the introduction of street warnings for cannabis was estimated using an average of 80 minutes per warning (which included activities from stop and search through to securing cannabis in the property storeroom), an average hourly cost, and the number of police involved.

In support of work being undertaken to assess the potential taxation revenue for California should cannabis be legalised, Caulkins (2010b) used a number of sources of data to assess police costs. He distinguished between misdemeanours and felony arrests; costs for booking from a 1987 Miami Dade County study, and a CBA on crime prevention activities in Washington State by Aos (2006) and (2001) which produced costs per arrest for drug and property crimes together and one for misdemeanours. These costs of policing of cannabis were about one-third of those produced by Miron (2005) for California.

5.5 NSW Police Force For the purposes of this analysis there are two components of policing to be considered. Specifically, there are the activities which occur within the Local Area Commands (LACs), and those within the NSW Drug Squad. Typically police activities related to cannabis at the level of the LACs include routine policing, time taken to question, search, transport suspected cannabis offenders, complete documentation, prepare for and attend court, secure and finally destroy the cannabis. Drug squad activities include investigation, surveillance and subsequent destruction of bush cannabis crops; investigation of hydroponic operations and their subsequent dismantling; and the accompanying documentation and court time.

Other policing activities, such as investigative activities by police in the LAC, or the use of dog squad for detecting drugs at dance parties, pubs, or concerts are not included. These activities when they occur are not targeted at cannabis alone and should the laws for cannabis change there would be no savings from either of these activities they would be ongoing to detect other drugs, such as methamphetamine and ecstasy (Key Informant interviews, November 2008).

88 Chapter 5

The following sections describe the data which were used to quantify the types of offences, and subsequently police resources expended on enforcing cannabis laws and regulations.

5.5.1 Police data Detailed data on each individual who had an encounter with police for a suspected cannabis offence were not available. In its stead, aggregated data from the NSW Police Force’s Computerised Operational Policing System (COPS) provided by the Bureau of Crime Statistics and Research (BOCSAR) are used. When a criminal incident or a group of criminal incidents are reported to, or detected by police an event is created on the COPS database. The variables included are two age groups (less than 18, and 18 and older), offence type (possess/use; supply, cultivate) and 13 methods of proceeding against the offender.

5.5.2 Methods of police proceeding As discussed previously there is a hierarchy of methods for managing those who are suspected of a cannabis offence. Moreover the options available to police are different for adult versus juvenile offenders. The various options are summarised in Table 16 with more detailed descriptions below.

The various methods of action by police will have very different resource implications. In addition to the nine categories listed in the table, the COPs data provided by BOCSAR included ‘other drug caution’, ‘criminal infringement notice’, ‘infringement notice’ and ‘missing/unknown’. The first three are not meant to be used for cannabis offences and in fact there were only a total of seven offences in these categories. In this study they are included in the cautioning numbers. There were 211 in the ‘missing/ unknown’ category, and these were combined with the ‘not proceeded against/ legal process nfc’. This leaves nine categories.

89 Chapter 5

Table 16: Definitions of methods by which police may proceed against offender Variable Description Charged More serious offences/offenders are dealt with by way of a bail court attendance

CAN No bail court attendance required

Future court Used for less serious offences; may or may not be transferred to the attendance/ field CAN police station Cannabis caution Cannabis Cautioning Scheme gives police the discretion to formally caution rather than charge adults detected for minor offences Youth conference A youth conference is a face-to-face meeting with offenders, victims and support persons, offending behaviour discussed, and plan is negotiated. Can be initiated at any point of contact – police, pre court, or post court. Youth Caution Applies to young offenders as per Young Offenders Act. Youth are brought to LAC, and the caution is a formal procedure delivered at a later date. Warning Juvenile/adult Legal process nfc When the legal process is missing or invalid (out of date). Not proceeded against Includes those not proceeded against 'deceased, reported, no formal action' and 'other'. Source: (Bureau of Crime Statistics and Research, 2006a)

Adults

Charged and held for bail hearing—those who are suspected of a cannabis offence may be held for a bail hearing once charged with a cannabis offence. This decision is made by the police/police prosecutor and will be a function of the situation, the offence, other current offences and past criminal record. Police activities would include transporting the offender to the police station, interviewing the individual, fingerprinting, securing the cannabis, completing paperwork, obtaining and securing valuables, and transporting to cells. Offenders will be held overnight in the police cells if brought to the police station after 3 pm Monday through Friday. On weekends, suspects in the greater area of Sydney are transported to Parramatta Court, while in regional areas the Court Registrar may elect to hear the matter. If the Court Registrar is not available or deems the case not suitable for them to handle, the offender is held until the next magistrates hearing.

A CAN applies when individuals are transferred to the police station, and are interviewed as above, but are not held for a bail hearing. The resource use will be similar to above with the exception of time in the police cells, and the bail hearing time and costs.

A Field Can or Future Court Appearance Notice is used for less serious offences/ offenders. After the offender’s identity is confirmed, questioning is complete, and the 90 Chapter 5

cannabis is secured, a summons to appear in court is issued. These may be issued in the field or at a police station.

The Cannabis Cautioning Scheme was first introduced in NSW in April 2000 as a method to manage less serious cannabis offences. The Cannabis Cautioning Scheme gives the police the discretion to caution rather than charge adults who can be adequately identified, who possess no more than 15 grams of dried cannabis, must not be in possession of cannabis resin, oil or living plants, have no prior convictions for drug use, sexual or violent offences. They must also admit to the offence and agree to sign the caution notice (Baker and Goh, 2004).

This process is presented as being a formal and more transparent way of dealing with minor cannabis offenders compared to informal warnings and less burdensome than being charged with a cannabis offence. Only two cautions per person are permitted (Baker and Goh, 2004). The police officer warns the offender of the potential health and legal consequences of cannabis use and contact number for the Alcohol and Drug Information Service (ADIS) is provided. Persons who receive a second and final caution are required to contact ADIS for a mandatory education session about their cannabis use (Baker and Goh, 2004; NSW Police Force, 2007).

A final method is the informal warning. Here the only information recorded is that the cannabis was detected, a verbal warning given. The name of the offender is not recorded. This has been predominantly replaced by the cannabis caution. For the purposes of this work, for adults only, warnings will be included with cannabis cautions.

Juveniles (age <18)

The methods and costs of proceeding against juveniles are included here although under legalisation it will not be legal for juveniles to possess or consume cannabis under the legalised–regulated system. Thus these costs will appear in both policies in the CBA. Juveniles are dealt with under the Young Offenders Act 1997 (New South Wales Consolidated Acts, 1997 (as at 27 November 2008). The alternatives to court include warnings, cautions, and youth conferencing with the intent to make young offenders accept responsibility for their actions, to avoid the cost and time of a court appearance and to steer young offenders away from detention (New South Wales Consolidated

91 Chapter 5

Acts, 1997 (as at 27 November 2008) (NSW Police Force, 2010b). Importantly however, to qualify for each of these non-court alternatives, the youth must admit the offence in the presence of a responsible adult (New South Wales Consolidated Acts, 1997 (as at 27 November 2008).

A warning may be given to a juvenile for any cannabis offence if the police (or other official) believe that it is in the interest of justice and the offender to deal with the matter in this way but a warning is not to be used if the offence involved violence. A previous similar offence does not preclude the use of warnings. A warning is to be given by the investigating officer and may be given at any place, including a place where the youth is found, and may be given to more than one juvenile at a time. Parents may be notified either in writing, verbally or in person of the warning. Anecdotal discussions with police suggest parents are frequently not notified. An official record of the warning must be made with details to be destroyed as soon as practicable after the individual reaches age 21 (New South Wales Consolidated Acts, 1997 (as at 27 November 2008)). Not all criminal offences may be dealt with by a warning, however, in relation to a cannabis offence, if the quantity is a small amount it may be dealt with by a warning.

Under the Juvenile Justice Act, a caution may be given to a juvenile, although the juvenile caution is a more formal process than for adults. It involves admission of the offence and agreement to the caution by the juvenile. As with adults, not all offenders or offences are entitled to a caution. For example, serious drug offences, offences which cause harm to others and offenders who are determined by the investigating officials to not be suitable for cautions are dealt with in other ways. Prior to a referral to the specialist youth officer who will subsequently organise the caution, an explanation of the offence, and the entitlement to legal advice, the entitlement to proceed to court and the process of the caution must be given to the juvenile by the investigating officer. For drug offences, information on treatment and counselling services must also be provided. This explanation should, if practicable, occur in the presence of a person responsible for the child, a legal practitioner, or if the juvenile is older than 16, an adult of their choosing (New South Wales Consolidated Acts, 1997 (as at 27 November 2008)).

92 Chapter 5

In NSW, via the Young Offenders Act, youth justice conferences were established to promote responsibility by the juvenile for his or her own behaviour, and to strengthen the family or family group, and to provide developmental and support services and to have due regard to the interests of any victim. These conferences are very resource intensive and not used principally for cannabis offences. In the 2006 COPs data there were only 17 occurrences of youth justice conferences for cannabis offences. The decision to hold a youth justice conference can be made by the prosecution, the courts, or by a NSW Police Specialist Youth Officer. When a referral for a juvenile conference is made each youth is assessed for suitability. Once the decision to hold a conference is made, a convenor is appointed. Convenors are recruited by the Department of Juvenile Justice, subsequently trained, paid and resourced to run the conferences. Considerable resources are used in preparing for each conference.

Responsibilities of the convenor include organising the date, time and location of the conference and who will be attending. The conference must be held at an agreed upon location but not at a police station, a court house or any office of the Department of Juvenile Justice. Persons attending include a parent or other responsible adult, a supervising officer if the young person is under probation or community service order; extended members of the family; lay advocate; lay advisor; interpreter, convenor; investigating officer from the NSW Police Force; youth liaison person from NSW police; victim or representative plus a support person (New South Wales Consolidated Acts, 1997 (as at 27 November 2008)). As a result of the conference a plan, which may include drug treatment, is then agreed upon and monitored (New South Wales Consolidated Acts, 1997 (as at 27 November 2008)).

As with those over the age of 18, juveniles who are suspected of a cannabis offence may be charged and held for bail, charged and not held for bail, or issued with a summons. This decision is made by the police/police prosecutor and will be a function of the situation, the offence and the current and past criminal record. Police resources would include transporting the offender to the police station, interviewing the individual, securing the cannabis, completing paperwork, obtaining and securing valuables, and transporting to cells. In addition time waiting for parents or legal guardians must be factored in.

93 Chapter 5

If the offence is drug trafficking the juvenile offender must be referred to the court however, the Director of Public Prosecutor can refer back to the police for a caution and the Children’s Court can give a formal caution (Department of Juvenile Justice, 2007).

5.5.3 Cannabis offence data There were a total of 14,082 persons of interest where cannabis was detected. Eighty- nine per cent were for possess/use, 3.9% for dealing or trafficking in cannabis and 7.1% for cultivation (Table 17). Most offenders were 18 years of age or over (88.8%) when the alleged offence was committed, 10.6% were less than 18 years of age while age was undefined for less than 1% of the sample. In subsequent analyses, those for whom age was undefined were allocated to either the adult or juvenile categories based on the existing distribution.

Table 17: Frequency of cannabis offences by age groups Type of offence Age category Frequency Percentage overall Possession and/or 10–17 1,432 10.2% use of cannabis 18+ 11,023 78.3% Age unknown 73 0.5% Subtotal 12,528 89.0% Dealing, trafficking 1017 28 0.2% in cannabis 18+ 516 3.7% Age unknown 8 0.1% Subtotal 552 3.9% Cultivating 10–17 32 0.2% cannabis 18+ 961 6.8% Age unknown 9 0.1% Subtotal 1002 7.1% All offences 10–17 1,492 10.6% 18+ 12,500 88.8% Age unknown 90 0.6% Total All ages* 14,082 100.0%

Source: 2006 COPS Data, BOCSAR. *There are slightly more cases included here than in Table 14 due to different counting rules. These data include all encounters where cannabis was detected even if subsequently ‘‘Not proceeded against/ legal process nfc’. In the following tables, the methods of proceeding against individuals by police, described above have been collapsed into five categories for ease of presentation. ‘Proceed to court’ includes being given a summons to appear in court, being taken to the police station, charged and held for a bail hearing or charged and not held for bail hearing. In the costing analysis, as the resource implications are different across methods of proceeding, the actual categories are retained.

94 Chapter 5

From Table 18 it is apparent there is considerable variation between how the police proceed with those less than 18 years of age and those 18 years of age and older. For example, only 13.4% of those less than 18 years of age proceed to court for possess/use offences compared to 49.6% of adults. Most (79.8%) of those aged 18 of age and older with cultivation offences proceed to court compared to 34.4% of those less than 18 years of age. Of those with possess/use offences 39.6% of juveniles and 25.5% of adults receive a caution while 20% of juvenile offenders received a caution for dealing/trafficking as do 31.3% of those with a cultivation offence whereas less than 1% of those aged 18 and over were cautioned for either.

Table 18: Methods of proceeding against for cannabis offences Caution/ Age Proceed Youth Informal #Not Categories youth Total group to court^ conference warning proceed caution** 193 17 571 163 497 1,441 10–17 Possession 13.4% 1.2% 39.6% 11.3% 34.5% 100.0% and/or use of cannabis 5497 0 2827 127 2632 11,083 18+ 49.6% 0.0% 25.5% 1.1% 23.7% 100.0% 11 1 6 2 10 30 Dealing, 10–17 36.7% 3.3% 20.0% 6.7% 33.3% 100.0% trafficking 461 0 2 1 58 522 in cannabis 18+ 88.3% 0.0% 0.4% 0.2% 11.1% 100.0% 11 0 10 0 11 32 10–17 Cultivating 34.4% 0.0% 31.3% 0.0% 34.4% 100.0% cannabis 774 0 2 2 192 970 18+ 79.8% 0.0% 0.2% 0.2% 19.8% 100.0% 215 18 587 165 518 1,503 10–17 14.3% 1.2% 39.1% 11.0% 34.5% 100.0% Totals 6732 0 2831 130 2882 12,575 18+ 53.5% 0.0% 22.5% 1.0% 22.9% 100.0% 6947 18 3418 295 3400 14,078 Grand Total All 49.3% 0.1% 24.3% 2.1% 24.2% 100.0% Source: 2006 COPS data, BOCSAR; ^ includes charged, CAN, and future court attendance; ** includes caution, youth caution, other drug caution, infringement notice; # includes not proceeded against, legal process unknown or missing/ unknown data.

5.5.4 Survey of police

Given the range of police costs obtained from the top-down costing methods, the different approaches to dealing with cannabis offenders (warnings, cautions, through to charge and hold for bail hearing), and differences in how adolescents and adults are

95 Chapter 5

dealt with by police, a decision was made to undertake a micro-approach in estimating the police costs of enforcing cannabis offences in NSW.

Subsequent to a discussion with members of the Drug Coordination Unit of the NSW Police Force, the decision was made to undertake a survey using a convenience sample of police officers in three Local Area Commands. The Local Area Commands (LACs) were selected to obtain one urban, one outer metropolitan, and one large rural LAC. The Commanders in three Local Area Commands were invited to participate in the study. Assistance with organisation and implementation of the survey was provided by individuals at the Drug Coordination Unit who sent a letter to the commanding officers, on behalf of the study, inviting the LAC to participate in the study. Once permission was received, MS negotiated with each LAC as to the time and method for distribution of the survey to serving police officers in each LAC. Ethics permission was granted (HREC 07293) by the University of New South Wales Human Research Ethics Committee. A copy of the survey and Participant Information Sheet are in the Appendices. The survey was completed anonymously.

The survey was designed to collect information on the average time taken to complete various activities as well as the number of police personnel involved in each activity. The structure of the questionnaire was based on information gleaned from the literature review of costing police activities (Brooks et al.; May et al., 2002; Baker and Goh, 2004; Crime Research Centre, 2007; May et al., 2007) and the various methods of proceeding by police (charge, caution etc.) (Goh and Moffat, 2008). There were separate survey questions for adults and juveniles reflecting differences in types of proceedings. Refinements to the survey, including modifications to the structure and wording, were made following consultation with current and past serving police officers.

The LACs were from the centre of Sydney (A), the outer western area of Sydney (B), and a large metropolitan non-Sydney area (C). In two of the LACs (A and B), all police officers in the LAC when MS visited were requested to anonymously complete the survey; of these, one visit was conducted in the morning, another over the day/evening shift change. In the third LAC, the preference of the Commander was for the survey to be distributed at shift change by a commanding officer over a number of days; the surveys to be completed anonymously and then returned to the shift commander, and

96 Chapter 5

then mailed to MS. No data on the individuals other than rank was recorded on the survey.

A total of 98 surveys were distributed across the LACs, and 94 were returned with useable information on the time it took to manage adult cannabis offenders. Two types of information were requested—the average time taken for each officer and the number of ‘other’ people involved. Not all officers had encountered all situations, thus not all sections were completed. Eighty-nine of the 94 surveys had information on managing juveniles, some were not completed because the officer had not dealt with juvenile offenders and others may have been missed as this was on the back page of the survey. The questions on the number of ‘others’ involved in the activity were less well answered, with only 70 officers completing those questions for the adult section and 56 for the juvenile section.

Once the surveys were completed, and the data entered, basic descriptive analyses were conducted including the average (and maximum and minimum) time overall officers to complete each of the methods of proceeding against cannabis offenders. Prior to the estimation of the overall average costs, the time to complete various activities was estimated for each LAC and the means compared across LACs. As there were no significant statistical differences between the LAC for any of the types of activities the overall average will be used in the remainder of the analysis (see Table 19).

Table 19: Comparing across Local Area Command: time in minutes for completion of various activities per officer Adults - Mean time in minutes Juveniles - Mean time in minutes

Groups N Caution Summons No Bail Juv. Caution No Bail bail warning bail LAC 1 14 76 318 408 418 77 176 404 412

LAC 2 49 108 408 539 546 102 183 751 759 LAC 3 31 114 451 599 608 91 218 562 569 F test 2.31 1.77 2.07 2.01 0.99 1.69 0.57 0.57 p value 0.10 0.18 0.13 0.14 0.38 0.19 0.56 0.56

Once the time taken to complete each type of procedure was calculated, this was multiplied by the average number of police encounters in each type of procedure. It had been the intent to calculate this for every officer and then average the overall time but due to the poor response to the number of ‘other’ officers involved this was not possible. Respondents were also asked how long it took to prepare for court, and average time spent at court on a day they were required to attend court for both those

97 Chapter 5

who plead guilty and those who did not. These times were then multiplied accordingly to either the proportions who plead guilty or not (Baker and Goh, 2004). All those who plead not guilty were presumed to have a court case.

Table 20 provides the average time in minutes for the four main forms of encounters for adults and four for juveniles (excludes conferencing and is only the police time for cautioning). It appears that although the Future Court Attendance notice may be issued in the field without transporting the offender to the LAC, many police officers transport offenders to the LAC. The time taken for juveniles for cautioning at 7.8 hours per encounter is more than twice that of adults. This is because, unless a warning is given, youth are returned to the LAC and police must wait for a parent or guardian to attend the LAC.

Table 20: Police survey: Total time per encounter for each type of police encounter Avg. No. Hours per encounter All Minutes per officer of officers includes all officers Adults Average CI L CL H Avg. Low High

Caution 106.4 95.1 117.9 2.05 3.7 3.2 4.0 Future court 408.0 367.1 448.9 2.17 16.9 13.3 16.2 attendance Not held for bail 539.1 485.0 593.3 2.19 22.2 17.7 21.7 Bail 547.4 492.7 602.0 2.19 22.5 18.0 22.0 Juveniles Warnings 107.4 93.6 121.1 1.97 3.5 3.1 4.0 Caution 226.1 205.0 247.1 2.08 7.8 7.1 8.6 No bail 724.9 463.0 986.8 2.16 26.1 16.7 35.5 Bail 734.6 470.0 999.3 2.15 26.4 16.9 35.9

The next step was to multiply the total numbers of hours per method of proceeding (Table 20) by the frequencies of each method (Table 18), and then convert hours into full time equivalents (FTE). The cost per FTE for 2006/2007 in NSW was obtained from the Report on Government Services (Productivity Commission, 2008) to which 28% on-costs were added to cover annual leave, long service leave and other benefits. The total costs per FTE were then multiplied by total FTEs to result in the total cost to police for these offences (Table 21). Also included are the additional costs for juvenile cautioning and juvenile conferencing. The methods of attributing costs for juvenile conferencing and additional costs for juvenile cautioning are described in Appendix– Chapter 5. As no additional information was available on those who were ‘not proceeded against’ the time taken to deal with adult cautions was applied to this category.

98 Chapter 5

The overall estimate of expenditure by police on cannabis offences is $8.96 million (Table 21); of this $8.2 million (91%) was spent on offenders aged 18 and over. Examining the data by type of offence; 84% ($7.5 million) of the police resources was spent on those who were identified as possessing or using cannabis.

Table 21: Total FTEs and costs by age category, offence type and method of proceeding Age 18+ Possess/use Deal Cultivate Total FTE $ FTE $ FTE $ FTE $ Charged 21.2 2,309,776 4.2 453,886 2.8 303,422 28.1 3,067,084 CAN 8.9 971,353 0.2 23,350 1.0 110,899 10.1 1,105,602 Summons 6.1 663,626 0.3 30,590 1.7 181,792 8.0 876,008 Future ct appear 15.1 1,644,186 0.2 16,994 1.6 177,590 16.8 1,838,771 Cannabis 5.8 629,400 0.0 1,061 0.0 669 5.8 631,130 caution** Warnings 0.3 28,325 0.0 446 0.3 28,771 Not proceeded # 5.4 586,968 0.1 13,020 0.4 42,765 5.9 642,752 Total adults 62.6 6,833,634 4.9 538,901 7.5 817,583 75.0 8,190,117 Age < 18 Possess/Use Deal Cultivate Total FTE $ FTE $ FTE $ FTE $ Charged 1.3 142,780 0.1 12,297 0.0 4,083 1.3 159,160 CAN 0.4 48,152 0.0 0 0.0 1,352 0.5 49,504 Summons 0.3 35,688 0.0 1,699 0.0 5,145 0.4 42,532 Future ct appear 0.1 8,506 0.0 0 0.0 850 0.1 9,356 Youth conference 0.1 8,003 0.1 5,516 0.0 0 0.1 13,519 Caution** 2.8 305,179 0.0 3,847 0.1 5,812 2.8 314,838 Warnings 0.3 34,532 0.1 13,382 0.0 0 0.3 47,914 Not proceeded# 1.0 105,344 0.0 2,039 0.2 22,958 1.2 130,341 Total juveniles 6.3 688,184 0.2 38,780 0.4 40,200 6.7 767,164 Overall total 69 7,521,818 5.2 577,681 7.9 857,783 81.7 8,957,282 ** includes caution, youth caution, other drug caution, infringement notice; # includes not proceeded against, legal process unknown or missing/ unknown data.

The low and high CIs from Table 20 were used to estimate a low and high range for total police costs at $7.5 million and $10.4 million per year.

5.5.4.1 Other police activities

The costs above include only those police resources used once the person comes to the attention of the police and are either cautioned, warned, or arrested. In addition to this, there are activities such as those conducted primarily by the NSW Police Drug Squad targeting indoor hydroponic setups or outdoor bush cannabis. The police time spent patrolling and doing usual activities is not included, as police would be undertaking these activities regardless of cannabis policy, but time spent investigating specific cannabis-related offences are included. 99 Chapter 5

The following information is based on a conversation with a member of the NSW Drug Squad in November 2008 and subsequent information provided in January 2009.

Hydroponic operations

Information on hydroponic operations is generally provided by a source or contact, not through detailed police investigations. A typical investigation may be approximately two working days by two officers to investigate the information and to obtain a search warrant. Once the warrant was obtained, several police would raid the suspected site. The police operation would involve six to eight police officers plus forensic and fingerprint experts, plus personnel from an energy company to disconnect the pirated electricity if necessary. A day would be required to write the report. Once costed, these estimates were then multiplied by the average number of hydroponic set-ups removed in a year (Table 22).

Outdoor (bush cannabis)

Outdoor cannabis sites, also referred to as bush cannabis, can be discovered by police through information provided by a source, or through the cannabis eradication program (CEP). The former is less predictable and may lead to ongoing investigations lasting up to six months involving several officers who work the case on an intermittent basis. The CEP program typically operates five times a year, on each occasion for five days. It involves Air Wing (helicopter) operating approximately six hours per day, 20 personnel per day and the Dog Squad. Specific well known growing locations are targeted.

Costs of the CEP program include the expenditures on helicopter, police personnel, and accommodation. As budget information for this program was not available, publically available costs of rental and landing costs for an equivalent type helicopter were obtained. As above, a standard cost per police officer including on-cost and overheads was used to cost police time. Additional costs of travel and food were also included (Police Association of NSW, 2009) (see Table 22).

The total estimate of these costs for these programs was $1,134,320 (AUD 2007). This does not include costs pertaining to discovery of cannabis as part of activities other than these described. For example, raids of methamphetamine laboratories, or syndicate activities where cannabis leaf was detected alongside of other drugs are not included.

100 Chapter 5

Table 22: Other police costs Other police costs Annual estimate of costs Cannabis eradication program $647,392 Detection, removal hydro operations $486,928 Cost of police cells (held for bail) $545,598 Total $1,679,919

Other associated costs not included elsewhere are the costs pertaining to holding the accused for a bail hearing. Those offenders who were charged prior to 3 pm Monday to Friday may have a same day bail hearing before the magistrate, others will be held overnight, while those in rural areas may be held over a weekend. In Sydney those who appear before a magistrate may need to be transported to Parramatta Court. As actual information on the number of nights offenders were held in jail was not available, it was assumed that 100% of those charged (and held for bail) were held for one night in police cells awaiting a bail hearing. Some may have been held for two or more nights while others were not held overnight. The cost is estimated at $545,598.

5.6 Magistrates early referral into treatment (MERIT) Operating alongside police and the traditional court responses is the Magistrates Early Referral into Treatment (MERIT) program. MERIT is a pre-plea three-month program designed for offenders with a demonstrable drug problem, who are eligible for bail, and who show potential for treatment and rehabilitation (Matruglio, 2008). An average of 709 persons with cannabis as their nominated drug of dependence use the MERIT program each year and of those who nominated cannabis as their problem drug, only 35% had cannabis as their principle criminal offence (Martire and Larney, 2009). This program is meant to allow defendants an opportunity to break the drug crime cycle by allowing them to focus on treating their drug problem separately from their legal matters.

Once identified as a potential MERIT candidate the participant is assessed for mental health status, drug use history, social factors, criminal record, and a risk assessment. The magistrate then reviews the assessment, and if eligible the participants may receive individual or group counselling, or attend external treatments such as detoxification or residential rehabilitation as well as support and case management from a caseworker. Additionally the court will make the defendant's involvement in MERIT a condition of bail necessitating close case management. If the defendant fails to attend treatment,

101 Chapter 5

commits further offences, or does not comply with bail conditions, the magistrate is notified and appropriate sanctions are applied.

As MERIT is a voluntary program, the offender's progress in treatment may be considered during the final hearing and sentence. In 2005, the most common sentence for those who completed MERIT was a bond with (21.5%) or without supervision (19.3%). For non-completers the most common sentence was a fine (24%) or imprisonment (23.5%) (Matruglio, 2008).

MERIT started in five local courts in 2000 and now operates in over 60 local courts in NSW (Chief Magistrates Office., 2008). Current expenditure on MERIT is not available but an economic evaluation of one pilot program (Lismore, 2000) estimated cost per accepted offender. The average cost per accepted MERIT participant, excluding rehabilitation and detoxification which are captured elsewhere (Chapter 7) was $4960 in 2006/07 AUD. The total annual cost was estimated at $3,516,640 AUD (709 * $4960).

5.7 New South Wales courts As in other studies (May et al., 2007) the full costs of court are included even if there are concurrent offences, as each charge is prosecuted (Registrar, NSW Children’s Court, personal communication, August 2008). Four courts operate in NSW, the Supreme, District, Local and Children’s Court. Serious criminal offences are generally dealt with in the District or Supreme Court while less serious offences are usually dealt with in the Local Court (Parliament of NSW, 2008).

If an accused person pleads ‘guilty’, they proceed to being sentenced by the court. If the accused pleads not guilty, they proceed to trial. In NSW, most criminal offences are dealt with summarily (without jury) by a magistrate. Sometimes – if the prosecution consents – defendants in the District or Supreme Courts elect to be tried before a judge alone but most criminal trials for serious crimes are heard before a judge and jury.

Local (Magistrate) Courts—This court deals with criminal and summary prosecutions; committal hearings; civil matters; juvenile prosecutions and care matters; as well as mental health issues. In NSW, the appointed magistrates have legal qualifications. Local Courts utilise the registry and their staff for administrative and clerical support; and often explaining court processes to court users who have no legal representation (Chief Magistrates Office., 2008). 102 Chapter 5

District Court—this is the middle level court and consists of the Chief Judge and other Judges as appointed. It is a trial court and may hear certain appeals. The NSW Drug Court, which operates as a District Court is an alternative method of dealing with drug dependent repeat offenders by engaging them in long-term treatment in an attempt to break the ongoing cycle of drug dependency and crime.

Children’s Court—deals with matters related to the care and protection of children and also criminal cases concerning children (Statistical Services Unit, 2007). In NSW it is a specialist court with 12 Children’s Magistrates and five Children’s Registrars (Brand and Price, 2001).

5.7.1 Court data In 2006, a total of 10,221 (6.7%) people attended court charged with illicit drug offences in NSW out of a total of 147,701 persons who were charged and appeared in any NSW Court (Statistical Services Unit, 2007). Data on all persons attending a NSW Court for a cannabis offence, whose charges were finalised in 2006, were obtained from BOCSAR. These data are collected from the courts and from a Reoffending Database (Statistical Services Unit, 2007).

This unit record data was for 6,026 unique individuals, who had a total of 6,646 unique appearances (5.8% had more than one appearance for a finalised cannabis offence in 2006). Cannabis offences accounted for 4.19% of all court appearances and 60.5% of all illicit drug offences. The court data for those charged with cannabis offences relates to criminal matters finalised.

Variables in the data set included unique personal identifiers, age, type of cannabis offence, type of court in which the offence was finalised (Local Court, District Court, Children’s Court), type of penalty for cannabis offence, flag for cannabis offence as principal offence, number of concurrent charges, and principal offence and penalty if not the cannabis offence. The data on each individual included the type of penalty given but it did not include the quantum of the penalty. It was necessary to use an average length of sentence for all illicit drugs by offence and court type (Statistical Services Unit, 2007). As the quantum of the penalty may be greater for other illicit drug offences (i.e. heroin, methamphetamine) compared to cannabis offences, the cost of the cannabis penalties may be overestimated. However, as 60.5% of all illicit drug

103 Chapter 5

offences that appear in court were for cannabis offences it is not expected the overestimation will be large.

In the unit record court data there were nine categories of cannabis offences that were subsequently collapsed to four groups (possess/use; dealing/supply; cultivate/import; other). Initially there were three categories of dealing (non-commercial, commercial and unknown quantity) and while analysis of the first two categories might have proven informative, most cases were classified in the ‘unknown’ category. Possess and use offences were grouped together as were cultivate and import (there was only one of the latter). The final grouping was ‘other’ which only had a few cases and most likely pertains to implements.

Table 23: Characteristics of cannabis offences by court type (2006) Children District Drug Local Variable Overall F p Court Court Court Court N (%) 268 210 16 5,952 6446 (4.2%) (3.3%) (0.2%) (92%) Mean age (SD) 16.56 36.07 30.31 32.89 32.31 235.8 <.001 (1.28) (11.82) (4.39) (10.23) (10.60) Age range^ 12-19^ 19-65 23-36 18-80 12-80 % Cannabis 53.0% 63.0% 0.0% 70.0% 68.7% 25.0 <.001 principal offence % With concurrent 58.6% 47.1% 100.0% 41.5% 42.6% 18.2 <.001 charges Mean no. of 1.81 0.89 12.38 1.04 1.01 36.8 <.001 concurrent offences (2.86) (1.32) (11.78) (2.22) (2.37) (SD) Cannabis offences (principal & non-principal) Possess/use % 88.4% 13.3% 93.8% 80.6% 79% Deal/supply % 9.0% 42.9% 0% 6.7% 8% Cultivate/import 2.2% 43.3% 0% 12.0% 13% Other 0.4% 0.5% 6.3% 0.8% 1% ^The Children’s Court is for juveniles, i.e. those less than 18 years of age but a delay in time to court may result in some being over 18 years of age by the time the case is finalised.

Most (92%) of the cases were heard in the Local Court (Table 23). There were 16 cases from the Drug Court, and although this court is formally part of the District Court, these cases are presented separately as they appear to be different from the others in that none of the Drug Court cases have a cannabis offence as a principal offence; they have a large number of concurrent offences (mean 12.38); and they appear younger (30.3 years of age) on average than those from the other adult courts. The age range for those

104 Chapter 5

whose cases were heard in the Children’s Court was 12–19 years of age with an average age of 16.56.

With the exception of those attending Drug Court, more than 50% of cases in each court had a cannabis offence as a principal offence. In the Local Court, 70% of persons had cannabis as a principal offence while only 53% of those attending Children’s Court did.

Not unexpectedly the characteristics of the individuals attending the various courts are different. Overall 42.6% had concurrent offences, with those attending Local Court having the least at 41.5%. The average for Children’s Court, District Court and Drug Court were all higher than the average. This is not surprising for the District and Drug Courts as they are more likely to hear more serious matters with persons who have more severe offences. Children’s Court handles both less and more serious offences involving juveniles.

Possess/use offences are the most common cannabis offence in Children’s (88.4%) and in Local Courts (80.6%), as well as the Drug Court (93.8%). The District Court has 42.9% for supply and 43.3% for cultivate offences. Possess/use offences are the most frequent for all age groups (Table 24). In the younger age groups possess/use offences make up 80% to 90% of offences however the proportion of higher severity offences generally increases with age.

Table 24: Frequency and distribution of cannabis offence by age category (principal and non-principal) Age <18 18&19 20–29 30–39 40– 49 50–60 60+ Total

Frequency of 210 433 2241 1919 1220 357 65 6445 cannabis offences Possess/use % 88.6% 89.4% 82.2% 79.6% 73.1% 59.1% 49.2% 78.8% Deal/supply % 9.0% 5.5% 7.9% 6.3% 9.6% 12.6% 9.2% 7.9% Cultivate/import % 1.9% 4.2% 8.8% 13.4% 16.9% 27.7% 41.5% 12.6% Other % 0.5% 0.9% 1.0% 0.7% 0.4% 0.6% 0.0% 0.7% The preceding tables and figures refer to all offences by offenders (n= 6646). Of the 6026 unique individuals 71% have had a previous non-cannabis conviction (see Table 25). Those with a previous offence are younger, less likely to have cannabis as their principal offence, and more likely to have a possess/use offence than those who have not have a court appearance for a non-cannabis offence in the past ten years. Those without a previous offence are more likely than the previous offenders to be charged with deal/supply or cultivate offences.

105 Chapter 5

Table 25: Characteristics of offenders with and without previous non-cannabis offences within past ten years No previous non- Previous non- Total cannabis conviction cannabis conviction N 1,742 (38.9%) 4,282 (71.1%) 6,026 Age (mean) SD 33.10 (12.89) 31.90 (9.5) 32.3 Cannabis as a principal offence 75.6% 65.3% 68.3% Current cannabis offence Possess/use 1,207 (69.2%) 3,560 (83.1%) 4,767 (79.1%) Deal supply 180 (10.3%) 264 (6.2%) 444 (7.4%) Cultivate/import 339 (19.4%) 430 (10.0%) 769 (12.8%) Other 17 (1%) 29 (0.7%) 46 (0.8%)

5.7.2 Outcomes of court appearance There are several possible outcomes for any court appearance each with different resource implications (see Table 26). Overall, 92.8% either plead guilty or were found guilty by the courts (ranging from 82.9% in the District Court to 93.3% in the Children’s Court).

Table 26: Outcome of court appearance by type of court (all cannabis offences) Court outcomes Children's Court District Court Drug Court Local Court N % N % N % N %

Dismissed/withdrawn/ 3 1.1% 9 4.3% 0 0.0% 44 0.7% no bill Guilty 241 89.9% 174 82.9% 14 87.5% 5,552 93.3% Not guilty 20 7.5% 2 1.0% 2 12.5% 350 5.9% Otherwise disposed of 0 0.0% 25 11.9% 0 0.0% 6 0.1% Unknown 4 1.5% 0 0.0% 0 0.0% 0 0.0% Total 268 100% 210 100% 16 100% 5952 100%

5.7.3 Cost of court–related activity for cannabis Average costs for cases finalised in each court are sourced from the Annual Report on Government Spending. Court Administration (Chapter Seven) provides expenditure, lodgement and finalisation data for all courts (Supreme, District, Magistrate and Children’s Court) in Australia (Australian Government Productivity Commission, 2007). Data are collected for this report from each state and territory on recurrent expenditure, lodgements and finalisations and by activity (i.e. criminal, civil and family). Recurrent expenditures include costs associated with the judiciary, court and probate registries, sheriff and bailiff’s offices, court accommodation and other

106 Chapter 5

overheads. The components of the expenditure include salary and non-salary expenditure, court administration agency and contract expenditure (Australian Government Productivity Commission, 2007).

The expenditures are reported against finalisations which represent the completion of matters in the court system. The cost per finalisation is derived by dividing the total net recurrent expenditure within each court for each year by the total number of finalisations for the same period. The report acknowledges that the cost per finalisation indicator results may include some arbitrary allocation of expenditures between criminal and civil jurisdictions, and the use of an average obscures that some finalisations take only a short time and require few resources, whereas other finalisations may be resource intensive and involve complicated trials (Australian Government Productivity Commission, 2007). In addition the net expenditure is calculated by deducting paid court fees (but not fines) from total expenditure.

Table 27: Estimates of total court costs by offence type Children’s District and Drug Local Court Total Court Court N 268 236 5952 6446 Possess/use $226,098 $259,333 $2,601,058 $3,086,489 Deal/supply $22,896 $542,790 $214,632 $780,318 Cultivate/import $5,724 $548,821 $385,904 $940,449 Other $954 $12,062 $24,390 $37,406 Total $255,672 $1,363,006 $3,225,984 $4,844,662 Per cent 5.3% 28.1% 66.6% 100%

The average cost per finalisation for each NSW Court is as follows: Local Court $542, District Court $6,031 and Children’s Court $954 (all in 2006/07 AUD) (Australian Government Productivity Commission, 2007). The average court costs for Drug Court participants have been reported at an average cost of $22,000 per participant (BOCSAR Report, 2008). Importantly, if court appearances for cannabis offences are less complex than the average offence these average costs per case may overestimate the true costs.

The total court costs for all cannabis offences are estimated at $4.8 million. Reflecting its high costs the District Court accounts for 28% of the court costs for 4% of the cases.

107 Chapter 5

5.8 Other costs: prosecution and Legal Aid

In addition to the police, court and penalty costs there are the costs of prosecuting and defending a case. Unit costs for prosecution and Legal Aid were not available. For the purpose of this work the costs are estimated in two ways. Both estimates start with the budgeted expenditure for prosecution from NSW Treasury documents, Criminal Prosecutions (18.1) and Net Expenses Legal Aid, Criminal Law Services (17) (Attorney General and Minister for Justice., 2008). As there were no existing data on what may have been expended on cannabis offences, two methods of allocating a portion of these costs to cannabis offences were used. The first used the percentage (0.5%) of the total police recurrent expenditures ($2,107,958,000) that was spent enforcing cannabis laws (see Table 28). This assumed that the distribution of prosecution resources and Legal Aid would be similar to that of police expenditures. The second method was to use the proportion of cannabis offences to total offences. Due to the uncertainty in these estimates, the midpoint was used for the main estimates, and the two original numbers were used as the potential range. The main estimates for prosecution was, therefore, $2.0 million and for Legal Aid $1.2 million.

Table 28: Estimates of DPP and Legal Aid costs Main Method 1 Method 2 Category of cost Total budget estimate (0.5%) (4.4%) midpoint DPP net expenses $82,456,217 $412,281 $3,593,342 $2,002,811 Criminal Prosecutions Legal Aid net expenses $50,940,301 $254,702 $2,219,917 $1,237,309 Criminal Law Services

5.9 Corrective Services Corrective Services NSW is the department which is responsible for the supervision and program management of adult offenders in custody and in the community. This includes fulltime and periodic incarceration in corrections faculties, home detention, community service orders, and probation and parole (Department of Corrective Services, 2008).

5.9.1 Penalties Detailed descriptions of each of the types of penalties can be found in the Appendix– Chapter 5. Table 29 provides the frequency of each type of penalty by type of court, for those who either plead guilty or were found guilty of a cannabis offense. 108 Chapter 5

Notably 64% of those who attended Local Court for whom the outcome was guilty received a fine for their cannabis offence, as compared to 27% for Children’s Court and 3% for District Court. The most common penalty for District/Drug Courts combined was imprisonment (55%) compared to 5% for the Local Court and 2% for Children’s Court. The most common outcome for Children’s Court was dismissed with caution (33%).

Table 29: Penalties for cannabis offence by type of court Children’s District and Local Court Court Drug Court Penalty type N % N % N % Adults only

Bond with supervision (adult) 328 6% 7 4%

Bond without conviction 269 5% 0 0%

Bond without supervision (adult) 422 8% 5 3%

Community service order (adult) 86 2% 4 2%

Home detention 7 0% 2 1%

Imprisonment (adult) 295 5% 104 55%

No conviction recorded 281 5% 4 2%

Periodic detention 32 1% 14 7%

Suspended sentence without 81 1% 16 9% supervision (adult) Adults and juveniles 0% 0%

Fine 64 27% 3549 64% 5 3% Nominal sentence 5 2% 109 2% 11 6% Suspended sentence with supervision 1 0% 93 2% 16 9% (adult) Juveniles only

Bond with supervision (juvenile) 23 10%

Bond without supervision (juvenile) 27 11%

Community service order (juvenile) 5 2%

Dismissed after YJC (juvenile) 8 3%

Dismissed with caution (juvenile) 80 33%

Juvenile control order (juvenile) 4 2%

Probation with supervision 16 7%

Probation without supervision 6 2%

Suspended control order with 2 1% supervision (juvenile) Total 241 100% 5,552 100% 188 100%

5.9.2 Costs related to penalties The cost for penalties were obtained from various sources as indicated in Table 30 and adjusted by CPI as necessary. For those offences where, as part of the sentence, there is

109 Chapter 5

‘no supervision’—the cost of the penalty is zero. All non-custodial penalties with supervision are allocated the same cost per day for the average length of the sentence; this also is applied to Community Service Orders based on seven hours per day. An average cost per day of $11.60 was used for community corrections (PC Chapter 8, Corrective Services, 2008).

Table 30: Costs of penalties Penalty Amount Per Source Bond with supervision (adult) $11.61 Day (Productivity Commission, 2008) Bond with supervision (juvenile) $11.60 Day (Productivity Commission, 2008) Bond without conviction $0.00

Bond without supervision (adult) $0.00

Bond without supervision $0.00 (juvenile) Community service order (adult) $11.60 Day (Productivity Commission, 2008) Community service order $11.60 Day (Productivity Commission, 2008) (juvenile) Dismissed after YJC (juvenile) $5,047 See Appendix-Chapter 5

Dismissed with caution (juvenile) $90.47 Occasion

Type of offence & Fine $332-$ 851 (Statistical Services Unit, 2007) court Home detention $62.81 Day (Henderson, 2006) (Department of Corrective Imprisonment (adult) $201.70 Day Services, 2008) Juvenile control order (juvenile) $1,196 Day (Productivity Commission, 2008) No conviction recorded $0.00

No penalty $0.00

Nominal sentence $0.00

Periodic sentence $186.60 Day (Productivity Commission, 2008) Probation with supervision $11.60 Day (Productivity Commission, 2008) Probation without supervision $0.00

Suspended control order with $11.60 Day (Productivity Commission, 2008) supervision (juvenile) Suspended sentence with $11.60 Day (Productivity Commission, 2008) supervision (adult) Suspended sentence without $0.00 supervision (adult) The costs for each penalty was calculated for each individual in the court data based on the penalty received; the results are summarised in Table 31. (Fines are considered as a cost to the individual, see Table 34). The 226 persons attending District Court account for more than half of the total (60.5%) costs. Twenty-eight per cent of the total costs are attributed to those with possess/use cannabis offences, with the remainder split between supply and cultivate offences. The majority of the possess/use penalty costs are attributed to those who appear in Local Courts, while most supply and cultivate penalty costs are attributed to the 226 individuals who appear in the District Court.

110 Chapter 5

Table 31: Total cost of penalties for cannabis offences Children's Type of offence District Court Local Court Total penalties Court N 268 226 5952 6446 Possess/use $192,999 $2,152,850 $5,196,977 $7,542,826 % Posses/use 79% 13% 50% 28% Deal/supply $50,952 $7,068,145 $2,909,567 $10,028,664 % Deal/supply 21% 43% 28% 37% Cultivate /import $981 $7,115,196 $2,272,714 $9,388,891 % Cultivate/import 0% 44% 22% 35% Other $90 $0 $21,298 $21,388 % Other 0% 0% 0% 0% Total $245,023 $16,336,192 $10,400,556 $26,981,771 Per cent of total 0.9% 60.5% 38.5% 100.0%

5.10 Total cost to the CJS for enforcing cannabis laws Once all the costs are summed (see Table 32), the costs of the penalties comprise more than half (54.8%) of the total expenditures. The costs of policing make up the next largest amount at 21.7% of the total. The costs expended on enforcing cannabis laws for adults are considerably more than that for juveniles, in part reflecting the longer sentences for some adults.

Table 32: Total government expenditures on CJS to enforce cannabis laws Age <18 Age 18+ Total % of total

Police (Table 21 and Table 22) $813,893 9,870,036^ $10,683,929 21.7% MERIT (P 101) N/A $3,516,640 $3,516,640 7.1% DPP (Table 28) $171,535 $1,831,277 $2,002,812 4.1% Legal Aid (Table 28) $105,972 $1,131,337 $1,237,309 2.5% Courts (Table 29) $255,672 $4,588,990 $4,844,662 9.8% Penalties (Table 31) $245,023 $26,736,748 $26,981,771 54.8% Total $1,592,094 $47,675,028 $49,267,123 100.0% ^ additional costs for detection and eradication added to costs for those 18+

Several values are explored in the sensitivity analyses below including the high and low ranges of expenditures for the police, prosecution and Legal Aid and length of sentence. As actual data on the actual quantum of the sentence was not known the average length of sentence for all illicit drug offences was used. Although cannabis made up more than 60% of the offences, the sentences for other drugs such as heroin and amphetamines may be significantly longer than for cannabis, thus affecting the average. The impact of this was explored by decreasing the length of every incarceration by 50%. This resulted

111 Chapter 5

in a 30.6% decrease in costs. In percentage terms, the ranges in the other factors have less impact (Table 33).

Table 33: Sensitivity analyses: varying key inputs Sensitivity Total % Change analyses Original total $49,267,123

Police costs (age 18+)

Lower range $7,521,818 $47,784,931 -3.0% Higher range $10,418,632 $50,681,745 2.9% DPP

Lower range $412,281 $47,676,592 -3.2% Higher range $3,593,342 $50,857,653 3.2% Legal Aid

Lower range $254,702 $48,284,516 -2.0% Higher range $2,219,917 $50,249,731 2.0% Decrease days incarcerated by 50% $11,922,900 $34,208,252 -30.6%

5.11 Costs to the individual

Up to now the costs have been limited to those borne by the government but there are a number of costs borne by the individual as a result of the enforcement of existing cannabis laws. This includes the costs of the fines, the value of parents’ time attending the police station when the youth is arrested, cautioned, required to attend court or juvenile conferencing, and costs of a solicitor. The value of own time for adult offenders while incarcerated is further discussed in Chapter 8 and the personal costs of stigma related to a criminal record in Chapter 7.

The value of fines given by the courts for one year is $22,821 for juveniles and $1,240,847 for adults (see Table 34).

Not all offenders would access Legal Aid, and in lieu of not knowing what proportion might not use Legal Aid and for which cases, it is assumed that the expenditure on a private solicitor would at least equate to the Legal Aid expenditure of $1,237,309 which does not seem unreasonable, at least as a minimum given some of the cases in the District Courts maybe of some length.

112 Chapter 5

Table 34: Total fines for cannabis offence by type of cannabis offence and court type Children’s District & Local Court Total Court Drug Court Number who received 66 3549 5 3618 fines Possess/use $20,584 $1,075,016 $1,000 $1,096,600 Deal/supply $1573 $9,361 $0 $10,934 Cultivate/import $664 $145,798 $0 $146,462 Other $0 $9,672 $0 $9,672 Total $22,821 $1,239,847 $1,000 $1,263,668

From Table 18, Table 23 and Table 29 there are a total of 258 youth who attended court, 26 received youth conferencing, and 587 who received a youth caution. Assuming that court and youth conferencing require four hours of a parents time and cautioning two hours this equates to 2,508 hours, and applying the average hourly wage of $23.04 (Australian Bureau of Statistics, 2006a), this is $57,712. The direct personal costs including fines, value of lost time to parents, and defence attorney costs results in $2.48 million.

Table 35: Personal costs Juveniles Adults Fines $22,821 $1,240,847 Value of parents’ lost work time $57,712 Defence attorney $1,237,309 Total $80,533 $2,478,156

5.12 Discussion

In this chapter the costs of enforcing the current cannabis laws in NSW are estimated. In doing so this chapter extended the knowledge on costs related to cannabis offences in NSW. Previous estimates of costs of enforcing cannabis laws in NSW were limited to an evaluation of the cannabis cautioning program (Baker and Goh, 2004) and only to those offences which were suitable for that program and did not include full costing of police time. The $10.6 million estimate of police expenditures for enforcing cannabis laws is considerably lower than the $29.2 to $88.1 million estimates obtained using top- down methods. Despite some adjustments to account for complexity top down methods result in much larger estimates, a point that has been made by others (Caulkins, 2010a) suggesting the average time police take to deal with a cannabis offence is much less than the overall average offence. Methods for dealing with such issues have been well established in health through the development of case mix groupings such as Diagnostic Related Grouping (Weiner et al., 1987). DRGs which are used for funding, and for

113 Chapter 5

research and evaluation purposes, while not without flaws reflect the resource implications of the diagnosis, the severity of illness, age, any complications or co- morbidities of the patient. The cost weights attached to every DRG are useful for understanding the relativity of the costs. The closest in magnitude of the top-down estimates, $29.2 million (Miron’s method) excludes 50% of the cannabis possess/use offences. This is based on the assumption that 50% of offenders were detected only after being detained for another offence and thus the marginal cost was zero. In the data used here, 39.6% of adult possess/use police encounters received a cannabis caution suggesting that this group of offenders had no concurrent offences otherwise the offender would not qualify for the caution (Baker and Goh, 2004). For the remainder of the offences, it is not known which offence resulted in the police encounter.

The police costs estimated here were based on a pragmatic sample of NSW Police Force members and as such may not be truly representative. Although not a statistically representative sample of police, officers do move frequently between LAC and the data collected would reflect their various experiences. The LACs were selected for size to try to ensure a larger number of officers available to complete the survey, and also to include those areas where officers would have had a higher probability of having encountered cannabis offenders and managed them using a wide range of methods. Though this is a limitation, given the wide disparity of costs in the literature, and the extent to which the costs of policing cannabis is an argument for legalisation (Miron, 2005; Rolles, 2009; Wood et al., 2010) it is important that we continue to refine the methods so that the measurement of these costs are as valid as possible. This study attempted to construct high-quality estimates of policing costs, building on work by others (May et al., 2002; Baker and Goh, 2004; May et al., 2007).

These estimates are for one year only, and assume that police activities are stable, something which may or may not be true for cannabis. Police may actively target specific offenders thus shifting the relative nature of the activities.

Another limitation is that the costs of courts, prosecution and Legal Aid were estimated using either an average cost (Productivity Commission, 2008), or top-down methods. As they comprise 10.2%, 4.2% and 2.6% respectively (courts, prosecution, Legal Aid) of the total any over or underestimate of the true costs is unlikely to have a significant impact on the total. Police expenditures and penalty costs account for 21.6% and 54.6%

114 Chapter 5

of the total respectively as such the impact of variations in these costs will be explored in the final model.

The other important caveat pertains to the costing of incarceration. As the length of the sentence for cannabis offences was not available, the average lengths of sentences for all illicit drugs, by each type of offence, by court type were used to estimate costs. This may result in an overestimate of the costs, however as discussed previously, 60.5% of all illicit drug offences that appear in court were for cannabis offences suggesting this effect may not be a significant issue.

Notwithstanding the caveats these data should be useful to police, policy makers and researchers as a method for understanding the implications of the current NSW cannabis laws. These data will be included in the final results of the model in Chapter 9

The next chapter sets out the regulatory framework for the legalised system as an alternative to the existing laws. First the framework is described and justified, and this is followed by an estimate of the costs to government, growers and consumers of such a regulatory framework.

115 Chapter 6

Chapter 6 : The public health perspective: a legalised–regulated framework

The premise of this chapter, which sets out one framework for legalising cannabis and the cost implications, is that if a legalised–regulated market for cannabis is to be introduced it should occur in such a way as to not to actively promote the increased use of cannabis. The legalised regulatory framework developed here aims to minimise personal and social harms related to cannabis (Haden, 2004; Rolles, 2009; Wood et al., 2010). As such, this framework does not treat cannabis as if it were a harmless substance but rather as a psychoactive drug with the potential for causing dependence in some users, impacting decision-making and ability to drive while under the influence and with the potential for negative health and social effects.

It will be demonstrated throughout this chapter that there is much to learn from other substances, particularly alcohol and tobacco, in how governments can use regulation to influence substance use. From the public health perspective, there are gains from regulating the cannabis industry to avoid the negative effects of advertising, competition, oversupply and low prices while still providing access to cannabis. Evidence from California where the introduction of medical cannabis has resulted in a significant drop in prices demonstrates that if these controls are absent the ability to limit prices or supply will be diminished (Kilmer et al., 2010).

In Chapter 2 a variety of policies for cannabis were discussed, including those currently operating in Australian and other international jurisdictions. Chapter 2 also contains a discussion of the potential value of a public health approach that would attempt to minimise the overall harms, not just those pertaining to the impact of a criminal record or the costs to the criminal justice system.

There is no single framework for regulation and legalisation of currently illicit drugs, but many have suggested potential elements for such regulation (Kleiman, 1992; McDonald et al., 1994; MacCoun and Reuter, 2001b; MacCoun and Reuter, 2001a; Haden, 2002; Haden, 2004; Rolles et al., 2006; Haden, 2008). Elements typically include regulating some or all of: the growers, the distributors and sellers and the consumers. Furthermore, the regulations for cannabis need to reflect the characteristics and potential harms of the drug. It is because of the multitude of potential regulations

116 Chapter 6

and uncertainties around enforcing them that some suggest “the cost of regulation is a choice variable, not an outcome variable” (Kilmer et al., 2010). That is, the costs of enforcing a minimal set of regulations may be minor whereas the economic costs of implementing and enforcing restrictive regulations may be significant.

In order to construct a legalised but highly regulated framework for cannabis in the NSW context and to estimate the resource implications of such a framework, the remainder of this chapter will draw on the proposed regulations from the literature, and actual regulations in Australia for other potentially harmful products such as tobacco, alcohol, prescription medications, guns, growing of poppies for production of morphine as well as those for the growing of hemp. While one outcome for such a strategy would be increased revenue to government from the sales taxes (GST) and excise taxes collected from the sale of a legal product, the objective here is not to maximise government revenues but rather to use governments’ abilities to regulate and set prices in order to minimise harms from cannabis. Recalling from Chapter 2 that no other country in the world has legalised cultivation and supply of cannabis, or any other drug currently considered illicit in Australia, such a framework and the resource implications herein are susceptible to assumptions and available data.

In constructing a regulatory framework there are a number of issues (for example, who can sell and who can buy cannabis, the necessity or not of licences and for whom, standards etc.). Currently there is a range of suppliers of cannabis to the market, from those growing for themselves and/or friends; those who are supplementing other income by growing a few cannabis plants through to large–scale growers (Willis, 2010). This chapter attempts to develop a feasible framework while addressing the public health objectives of harm minimisation.

6.1 Dealing with regulation

Economists often argue that regulation is costly, and rightly so, but not regulating can also be costly. Regulations play a central role in maintaining social, environmental and economic standards (Banks, 2003). Without regulations outcomes such as poor food quality standards, health hazards, environmental failures, building and bridge collapses, excessive prices, and chaos on roads may well ensue.

117 Chapter 6

Even with the best of intentions when trying to create a set of regulations, there are those who will gain and those who will lose. Regulations developed in an attempt to appease all interest groups may actually result in pleasing no one or result in such weak regulations that they serve no purpose (Grabosky, 1995; Banks, 2003). Equally, if the regulations (or laws) are not seen as valid or necessary, and on occasion even if they are, compliance with regulations will be low unless enforcement occurs (Grabosky, 1995). Alternatively if the punishment is seen to be too severe, enforcement may not occur (Grabosky, 1995). Some examples may clarify these points. Taking the last point first, there were several examples in Chapter 2 such as in Russia where reportedly those required to enforce the existing cannabis laws found the process and outcomes to be too onerous to the individual and to the system. The decision was then made to issue warnings or require offenders to obtain treatment rather than arrest someone for possession (Schreck, 2004). If the laws are perceived as unenforceable, or not reasonable, they continue to be broken. Other examples are: drink driving prior to the introduction of random breath testing, the use of mobile phones while driving even when illegal, and speeding.

The process of developing regulations requires in-depth policy analysis, as well as the ability to examine potential regulations with a critical eye (Grabosky, 1995). Regulations may also result in unexpected behavioural changes (Banks, 2003) and even with careful thought this may occur in the regulation and price control of cannabis markets. One expected outcome might be a shift by some younger cannabis users from cannabis to ecstasy when cannabis loses its ‘forbidden fruit’ effect. This might result in increased harms from the use of ecstasy or the contaminants of ecstasy tablets.

Regulation and its enforcement are costly, both in terms of direct expense and opportunity costs (Grabosky, 1995). It has been estimated that at the federal level in Australia in 2001/02 there were 30,000 staff and an associated expenditures of $4.5 million (2001/02 AUD) involved in explicit regulatory functions (Banks, 2003). Additionally, compliance with regulations poses both a monetary and time cost on individuals and businesses. Again in Australia, in 1994 the Productivity Commission estimated compliance with regulations imposed costs of $11 billion with around 85% borne by small and medium sized businesses (Banks, 2003). Although these data are somewhat dated, and there have been changes in regulations, the point is made that

118 Chapter 6

there is a substantial burden to society and to firms that can be attributed to regulating. So much so, firms may leave the regulated industry.

All of these factors need to be considered when constructing a regulatory framework. There may well be a trade-off between efficiency and public health. When constructing regulations care should be taken to ensure that the regulations: • do good • are better than the alternative • state what they are going to do • be clear and concise • be consistent with other laws • be administered by accountable bodies • be robust to errors (Banks, 2003).

This next section describes the proposed regulatory framework including the rationale for the selected components. This is then followed by estimates of the potential economic implications of such a regulatory framework. These estimates of the resource implications will include implementing, complying with, and enforcing such a regulatory framework. As will be evident there are many uncertainties in these estimates, and these must be taken into consideration when considering their implications.

Before continuing, a couple of points need to be made. First and foremost, this is not written in legal language. I am not legally trained. This chapter is an attempt to move forward from the many suggestions that cannabis should be legalised and to think through a ‘reasonable’ model. Every attempt has been made to be consistent, to recognise the economic implications in terms of the impact on supply or demand decisions. It is but one of many possible frameworks. The other point is that this work, as discussed earlier, does not include the important issue of International Laws and Treaties nor does it attempt to quantify the political and social costs of holding the debate on this topic. Should a decision be made to undertake such reform under the current International Treaties, there would no doubt be significant ramifications of such an action. Finally, the details in the sections that follow are to enable economic costing to be undertaken for the purposes of the CBA.

119 Chapter 6

6.2 The regulatory framework

The goal is to present a regulatory framework that is one alternative to the current criminalisation of cannabis use, cultivation and supply. The intent is to provide one viable option that does not actively promote or glamorise the use of cannabis, but neither does it make the product so difficult or expensive to obtain that consumers prefer to continue purchasing their cannabis from illegal drug dealers.

As a result of the large health burden of tobacco and alcohol, much work has been done to inform the public of their risks and to increase the social unacceptability of tobacco smoking (Cornelius et al., 1995; Callard et al., 2005; Collins and Lapsley, 2008; Freeman et al., 2008; Hotham et al., 2008; NSW Health, 2008; Cancer Council Victoria., 2010; Laslett et al., 2010a) and to examine measures to limit binge drinking and violent behaviours from excessive alcohol consumption (Babor et al., 2003; New South Wales Government, 2006; Jones et al., 2009; NSW Police Force, 2009b; Communities Office of Liquor, 2010). This framework draws on this literature.

The regulations are divided into two categories, those which are intended for the purchaser/consumer of cannabis and those intended for the grower/ distributer and retailer. Government departments (agencies), either NSW or federal, would have the responsibility of enforcing these regulations.

6.2.1 The purchaser/consumer

The introduction of a legalised–regulated framework for cannabis could at the same time respect an individual’s civil liberties and their right to purchase (Nadelmann, 1989) but also introduce certain responsibilities for the consumer.

i. Positive licences for consumption are proposed by (Nadelmann, 1989) and included in MacCoun and Reuter (2001) as one potential element of a regulated framework (Rolles, 2009). Here it is proposed that in order to purchase cannabis, cannabis seeds, or cannabis implements in NSW such a licence would be required and to minimise ambiguity it will be referred to as purchase–licence from here-on. This purchase-licence would serve two purposes. Firstly it would provide a structure to require the applicant to demonstrate that they were aware of the regulations and laws, and the potential harms of cannabis (Leitzel, 2008).

120 Chapter 6

The applicant would need to be aware of the acute cognitive effect of cannabis, the possibility of dependence, lung damage and mental health problems. Additionally, information on where to obtain treatment if necessary would be provided.

Secondly, such a licence might go some way to ameliorating ‘cannabis tourism’, such as that which is creating challenges in parts of the border regions of the Netherlands with the cannabis coffee shops (Buxton et al., 2008). The purchase- licence would be available to any resident aged 21 years of age (see below) residing in the state of NSW who passed the knowledge test and would require photo identification. Requiring a licence is not unusual; licences are required for driving a motor vehicle and the possession of a gun (NSW Police Force, 2010b). As with driver’s licences, it would need to be renewed every five years, and could be revoked by courts for such activities as providing cannabis to those less than 21 years of age, selling black-market cannabis or other activities which contravened the cannabis regulations. ii. The legal age of consumption would be 21 years of age. It would be illegal to sell or otherwise provide cannabis to those under the age of 21. Those found doing so would be subject to criminal prosecution and possible revocation of a purchase-licence. This is similar to NSW alcohol laws which prohibit a person supplying liquor to minors on premises, unless the person is a parent or guardian of the minor; it is an offence subject to a maximum fine of $11,000 or 12 months imprisonment or both, or an on-the spot fine of $1,100 (NSW Police Force, 2009b). In NSW, a minor who obtains, consumes or carries away liquor may be subject to a fine up to $2,200 (NSW Police Force, 2009b).

Most countries have age restrictions for purchasing alcohol and tobacco, driving cars and using guns. Scientific evidence suggests that there is likely to be a lower probability of damage to the still-developing brain if frequent drug consumption is avoided at least until brain development has ceased (Lubman et al., 2007), which occurs at or around 25 years of age. Additionally, the increased vulnerability of adolescents to experimentation with drugs of abuse, and the tendency to dependence is another reason to delay onset of drug use (Volkow and Li, 2004). Although the current legal age for the purchase of

121 Chapter 6

alcohol in Australia is 18 year of age and having a different age for cannabis would be a challenge to implement and enforce, there appears to be scientific evidence to support this decision. Currently the age for legal consumption of alcohol in the United States is 21, and this has been found to reduce alcohol consumption among young people and decrease road traffic crashes (Babor et al., 2003). iii. The proposed legalised–regulated framework would permit cultivation of 10 plants at one time for personal use without a growers’ permit, similar to being permitted to manufacture one’s own alcohol without paying excise or licence provided a still is not used. To operate a still for the production of alcohol (even for personal use) necessitates an excise manufacturing licence and the payment of excise taxes (Australian Taxation Office, 2009a). Cannabis can be a relatively easy plant to grow, thus its common nickname ‘weed’. Licensing every home grower would be challenging to enforce and not realistic. The choice of 10 plants was an arbitrary number. As several crops a year can be grown, permitting up to 10 plants would allow up to 30 plants per year. Some hypothesise that the ability to purchase cannabis legally would decrease the number of growers, and those remaining would likely be hobbyists or

connoisseurs (Rolles, 2009). iv. As with alcohol, driving or operating heavy equipment under the influence of cannabis would be illegal. Elsewhere (see Chapter 8) the current evidence on the relationship between cannabis use and road accidents and the potential impact on mortality and morbidity relating to motor vehicle accidents under legalisation is discussed. Challenges remain in determining the relationship between the impact on motor skills and judgment and the results of tests for THC using oral fluids (Drummer et al., 2007) as they cannot easily be compared to blood test results and they are more variable than the RBTs used for alcohol. Under a legalised–regulated framework, ongoing research would be required to improve understanding between the level of cannabis detected and impact on

mental and motor skills. v. There will be limits on where cannabis can be consumed. Cannabis smoke is not without its health harms; it has twice the tar content as tobacco and low

122 Chapter 6

levels of THC have been found in passive smokers (Mørland et al., 1985; Westin and Slørdal, 2009 ; Röhrich et al., 2010). Although THC acquired from second- hand exposure has been low in adults (Röhrich et al., 2010), THC found in the urine of two-year-old children has be attributed to passive exposure to cannabis (Bleeker and Malcolm, 2002). Cannabis will not be able to be consumed in any workplace, in public buildings, parks, pubs and bars, restaurants, on public transport, within any educational facility, outdoors within 100 meters of school property, or in personal cars when children 16 years of age or younger are present. Totally smoke-free workplaces have been demonstrated to be associated with 3.8% reductions in prevalence of smoking and in 3.1 fewer cigarettes smoked per day per continuing smoker in the United States, Australia, Canada, and Germany (Cancer Council Victoria, 2010). This demonstrated negative impact on the social acceptability of smoking may also have the same impact on the uptake and continuation of cannabis smoking. It is feasible the effect may be even stronger as cannabis alone is generally less addictive than tobacco (van Dijk, 1998). Smoking in cars is prohibited in six Australian jurisdictions as it is in several Canadian provinces and US states (NSW Health, 2009). Some argue that bans on smoking in public places may result in increased exposure in the home, however surveys in Ireland where smoking has been banned in all workplaces have found an increase in the number of smoke-free households (Cancer Council Victoria., 2010). vi. This framework will not deal with the medical use of cannabis. If cannabis is to be used as a medical treatment there are regulatory structures such as the Therapeutic Goods Administration and the Pharmaceutical Benefits Advisory Committee which would oversee and regulate medicinal cannabis.

6.2.2 The grower vii. A permit to grow cannabis would be required for all growers, other than those growing 10 or fewer plants for personal use. The initial permit would require an application form to be completed, require the applicant to be at least 21 years of age, require that applicants have a criminal check and no criminal record. Permits would not be transferable between individuals or firms. The number of permits will be restricted and contracts to supply cannabis to the agency responsible for distribution and retail will need to be negotiated. The licence

123 Chapter 6

costs will not be prohibitive. Currently permits are required to grow hemp for fibre and oil, and poppies for morphine in Tasmania. Contracts and licences to grow poppies are negotiated based on world demand for the products (Department of Justice, 2009). The tobacco growing industry (now defunct in Australia) had such a system of licensing and inspections (Geis, 2005). viii. As with poppies grown for morphine, growers will only be permitted to sell their cannabis product through pre-negotiated forward contracts (Department of Justice, 2009) to a non-profit enterprise. The price received by the grower will be a percentage of the retail price. As discussed in Chapter 4 allowing a free market supply without controlled prices would likely lead to oversupply and downward pressure on prices.

ix. Growers would be required to keep records of the seed type planted, and take measures to provide security against theft of crop (Department of Justice, 2009; Rolles, 2009). Cannabis can be grown outdoors or indoors but for the purposes of this model, costs will be based on indoor (greenhouse) cultivation.

The costs of growing in a greenhouse setting are higher than for a field crop but the cost of maintaining security is less. The requirements for poppy growing in Tasmania include a minimum of five wire fences with the top strand being barbed wire; or four plain wires, providing top wire is electrified and there are properly fitted gates and panels. Additionally, farmers need to ensure that poppy seeds are not accidentally spread by the wind. Since hemp growing has been permitted there have been a number of thefts from fields presumably from those who mistakenly believe it is cannabis and as such it is expected considerable security around growing operations would be required. Security for indoor cultivation would be somewhat easier but the grower would be required to demonstrate adequate provision of security when negotiating their contract.

x. As with other agricultural enterprises, commercial growers and employees will be required to obtain pesticide, fungicide and herbicide certification. For this in NSW, initial training is two days with a refresher course required every five years. Growers will be required to document all application of pesticides, fungicides and herbicides to their crop (NSW Department of Primary Industry, 2005).

124 Chapter 6

xi. Cannabis growers, as with other employers, will be required to meet all NSW Horticultural Award conditions, including Occupational Health and Safety training, provision of protective clothing, superannuation, annual leave, and other employee benefits, and pay at least the minimum award wage for farm employees (Industrial Relations, 2010). In addition, other business requirements such as submitting taxes to the Australian Taxation Office on behalf of their employees and filing quarterly BAS submissions would be required. xii. Growers may grow any potency of cannabis. Sales of the crop must be accompanied by results of tests indicating the THC potency, as well as other key components.

Traditionally, outdoor cannabis has been Cannabis Sativa and the indoor varieties have been Cannabis Indicus but either can be grown in both growing locations. There can be differences in potency (Bleeker and Malcolm, 2002). As with alcohol, in a real market cannabis could be priced according to potency with higher levels of potency resulting in higher prices.

6.2.3 Distributor / retailer xiii. Distribution and retail sales of cannabis will be through a monopoly government-run enterprise. Models suggested by others include a single non- profit private non-government organisation (NGO) or a cooperative (King County Bar Association, 2005). A cooperative may have as its objective to maximise the profits for its members and not focus on harm minimisation principles while a private NGO may have objectives that differ from the goal of minimising the consumption of cannabis as well as minimising other harms. Government monopolies of off-licence sales of alcohol exist in several provinces in Canada, in some Scandinavian countries (Babor et al., 2003) and in some US states (Kilmer et al., 2010). The decision as to which type of non- profit organisation is less important than ensuring it is a non-profit. The conversion of an existing legal industry to such a structure would normally require compensation to business owners for loss of the value of their firms, but if such a structure was introduced when cannabis was initially legalised, the assumption can be made that no financial compensation is warranted.

125 Chapter 6

It is proposed here that a government-owned and run monopoly would be responsible for the distribution and selling of cannabis with the overall objective of seeking to maximise public health objectives through the use of plain paper packaging, constraints on labelling, limited hours of operations, no advertising or promotion, and continued enunciation of the potential harms of cannabis use. Controlling the availability of alcohol has been demonstrated to reduce alcohol problems (Babor et al., 2003). Limits on advertising and promotion and use of warnings on cigarette packages have resulted in a decrease in smoking rates (Cancer Council Victoria., 2010).

Callard et al (2005) in proposing alternative business regimes for tobacco suggests that “(t)he rational social choice is to select an institutional form that facilitates, rather than hinders, the achievement of our public health outcomes and other social goals.” Corporations are required by law to act in the best interest of their shareholders, which most often will not coincide with public heath objectives (Callard et al., 2005). There is substantial evidence that large tobacco companies and alcohol distributors use their resources to promote the use of their products often to the detriment of individual consumers and society overall (Babor et al., 2003; Cancer Council Victoria, 2010). Current efforts to combat proposed decreases in hours of operation of bars and pubs, and density of pubs and the proposed move to plain paper packaging on cigarettes provide ample evidence of why if public health objectives are important they should be addressed at the onset.

Advertising and promotion is normal behaviour for a profit making firm, but what is also of concern are other behaviours used in the promotion of the tobacco products. Examples include implying that light cigarettes are less harmful, actively increasing the addictiveness of their product, and using their economic influence to oppose government’s attempts to regulate controls (Callard et al., 2005). To avoid this behaviour, and to protect against a new industry attempting to advertise and promote their product, for the purpose of this exercise a government non-profit enterprise will purchase cannabis in the dried form from the licensed growers at a predetermined price. This non-profit enterprise would then be responsible for the distribution and sale of the product.

126 Chapter 6

xiv. The density of retail outlets would be limited. Control of the number of alcohol outlets has been demonstrated to have an effect on consumption (Rolles, 2009). The intent is not to limit access such that it is a burden to purchase cannabis but also not to have a store selling cannabis on every street. The actual location of the premises would be subject to the same community and council consultation and scrutiny as currently exists for the establishment of a new alcohol outlet (Communities Office of Liquor Gaming and Racing, 2010). This process can be a time and resource intensive exercise. xv. Only cannabis and cannabis implements would be sold in these shops and cannabis would be sold for off-premises consumption only. One rationale being to discourage unplanned purchases and make it easier to prevent those less than 21 years of age from making purchases. Such a method of selling would require the consumer to make specific visits to the shops (Chapman and Freeman, 2009). Some might argue for Dutch ‘coffee shops’ but as the objective is to take a public health approach to minimise harms, the current laws that ban cigarette smoking in workplaces, including pubs and clubs (NSW Health, 2008) avoids problems with secondary smoke (Röhrich et al., 2010). xvi. Products would be sold in plain paper packaging (Rolles, 2009) with appropriate health and pregnancy warnings and these warnings would occupy 50% of the front and back of the package (World Health Organization, 2005; Freeman et al., 2008). Studies demonstrate that required health warnings are more easily recalled and appear more serious when presented in plain packaging (Freeman et al., 2008). Anti-smoking researchers report that tobacco manufacturers recognise the importance of packaging and that “once exposed to innovative packaging especially young adults see their current packaging as dated and boring” (Chapman and Freeman, 2010).

In a move against the legislation on plain paper packaging for tobacco proposed in Australia by the independent Senator Fielding 2 (2009) tobacco companies have indicated that they will challenge the decision in the courts using constitutional and trademarks arguments, or sue for compensation (Chapman and Freeman, 2010). In the cannabis context as there are no existing trademarks,

2 The Senate Inquiry on this Bill was due to report in August 2010 but as a Federal election was called this report has not been submitted. . 127 Chapter 6

or packaging, or brand advertising in existence this should not result in additional costs within the CBA model. xvii. As well as health warnings, labels would be required to contain information on whether the cannabis was hydroponic or bush, the type of cannabis, its potency, and the weight of dried cannabis. Cannabis seed catalogues currently advertise their seeds according to a variety of characteristics such as time to harvest, THC potency and other ingredients. There is much research to be done to understand the relationship between all the ingredients of cannabis (THC, cannabidiol, cannabinol etc.) and the impact on the consumer, and in time other information may be required on the label. This type of research is difficult, if not impossible, while cannabis is illegal but would be a currently unquantifiable benefit of legalisation. Food product labels are currently required to list ingredients as well as their nutritional value (World Health Organization, 2005; Gosten, 2009; Food Standards Australian New Zealand, 2010) and alcohol containers must state the percentage of alcohol. xviii. Packaging would need to be child-proof (Rolles, 2009).

xix. The sale price will be determined by an independent regulatory board. The prices will include an excise tax and GST components (Rolles, 2009). The rates of taxation, thus prices, per gram may vary by potency, similar to the price variations in beer, where low alcohol beer has a lower excise tax (Australian Taxation Office., 2010). But as indicated previously, for this model prices are held constant.

xx. Staff working in cannabis shops will be required to receive a responsible service of cannabis training provided by a recognised trainer. The role of the currently run alcohol courses are to ensure that all staff understands their obligations when selling alcohol. Currently RSA training courses are provided by a large number of trainers who must be approved by the Casino Gaming and Liquor Authority (Communities Office of Liquor, 2010). These RSA courses could be modified for staff selling cannabis and would include knowledge on how to access treatment. And as with pharmacists and those who serve alcohol they would be required to refuse service to those who are deemed under the influence (Rolles, 2009). The Liquor Promotion Guidelines (2009) are designed to ensure

128 Chapter 6

licensees sell and supply alcohol responsibly and include a risk assessment guide to help licensees decide if a particular activity or promotion is appropriate. Such guidelines would need to be developed for cannabis. xxi. To ensure everyone has equitable access to purchasing cannabis, sales by internet/fax will be permitted but must conform, where applicable, to existing regulations for internet sales of alcohol (Communities Office of Liquor Gaming and Racing, 2010). Sales would only be permissible to NSW addresses and require sighting of the purchase–licence upon delivery. Sales would occur through a website operated by the same government agency (or non-profit organisation) responsible for over-the-counter sales. The website must include notification that it is illegal to sell to anyone less than 21 years of age. Upon delivery of the cannabis, confirmation must be obtained that the person receiving the order is at least 21 years of age. Adults cannot ask children to accept deliveries; there would be an on the spot penalty of $220 or court fine of $2,220. As above there would be no advertising, promotion, sponsorship of seeds or cultivars to consumers on the website.

6.2.4 Government(s)

Many of the responsibilities attributed to government, such as determination and collection of GST and excise taxes are the mandate of the Commonwealth of Australia (Geis, 2005). Achieving cooperation and sharing of responsibilities across multiple levels of government increases the complexity and costs of these activities. In this next section, while it is recognised that different jurisdictions may actually have carriage of determining and enforcing regulations and setting the excise tax level there is no attempt to delineate different governments’ responsibilities (all government costs are included regardless if incurred at the federal, state or municipal level).

Governments’ responsibilities lie in two areas: establishing and maintaining a regulatory agency and then enforcing those regulations established by that agency. It is proposed a single government body be established to set policy and to ensure a coordinated approach. It is proposed that such a body be modelled on such entities as the Wool Marketing Board, the Wheat Board or the Independent Pricing and Regulatory Tribunal (IPART) while having an additional responsibility to address public health

129 Chapter 6

concerns. IPART which exists in NSW is the independent economic regulator in NSW that oversees regulation in the electricity, gas, water and transport industries (Independent Pricing and Regulatory Tribunal, 2010). The cannabis regulatory agency would be responsible for:

• establishing the infrastructure for consumer and grower licences • overseeing security matters • determining the number of field officers • liaising with NSW Police re enforcement of regulations • determining the quantity of cannabis to be supplied (purchased from licensed growers) • determining the number of growers, determine the price to growers which would include a return on the growers’ investment • setting the retail price for cannabis • establishing the enforcement strategy • collecting and collating statistical information to assist with setting the price and quantity to be supplied • determining areas for necessary research and make recommendations to funding bodies.

The composition of the board would need careful consideration but in order that informed decisions are made, it should include public health experts, consumer and growers’ representatives, as well as police, cannabis researchers and legal experts.

The activity of pricing cannabis would be similar to that of the Independent Pricing and Regulatory Tribunal (IPART) (Independent Pricing and Regulatory Tribunal, 2010). For this CBA, the price to the consumer will be maintained at the current average street price per gram in Sydney (Black et al., 2007; Phillips and Burns, 2008) which is approximately $20 gram. As discussed elsewhere (see Chapter 4) this is to minimise the increase in use driven by an otherwise expected drop in money price related to legalisation (Hall and Babor, 2000; Kilmer and Liccardo Pacula, 2009).

130 Chapter 6

6.3 Costs of production, distribution and sale of cannabis

This next section is laid out as follows: i) a discussion of the costs of producing cannabis under an illegal system; ii) estimates of the costs of growing, and selling under a legal framework; and iii) estimates of the costs of the regulatory system itself.

6.3.1 Growing cannabis under an illegal system

The illegality of cannabis means it is grown in circumstances that a legal crop would not be, such as in cupboard, or bedrooms, or behind fake walls of a house with the walls lined with plastic and windows blocked off to hide the lights; in crawl spaces under houses; or in large sheds. In Figure 7, a photograph taken during a police raid on an illicit cannabis growing operation, the lighting, the plastic sheeting on the ceiling and fans for circulating air are evident and provide an illustration of the complexity of some growing operations.

Figure 7: Example of an illegal growing operation

(NSW Police Force, 2010a)

Alternatively, illicit cannabis may be grown outdoors in a private garden, elsewhere on private property or in a state forest or other remote bushland. Some of these situations may not exist if growing cannabis were legal, although, as evidence by illicit sales of tobacco, the simple act of legalising cannabis is unlikely to remove the black market (Australian Customs Media., 2007; Australian Customs Media., 2008; Australian Taxation Office). Issues pertaining to the black market are discussed further below.

131 Chapter 6

As with any plant a growing medium and light are required. The growing medium may be either soil, liquid or a combination of the two. Plants grown indoors are often grown in a liquid medium with the addition of irrigation. This can lead to high levels of moisture, often resulting in infestation of moulds, mildew and fungal growth (Cervantes, 2002) not only in the plants but in the walls and ceilings. This is not only harmful to the plant and potentially the consumer, but also to those working in this environment. Growers counter this problem by lining walls and ceiling with plastic sheeting, the use of fans and the introduction of fresh air into the room (Cervantes, 2002).

The form of lighting is conditional on where the cannabis is grown and much has been written about cannabis and lighting. Outdoor-grown cannabis relies on sunlight and as such has a longer growing cycle. Indoor-grown cannabis uses artificial lighting. The level of, and sophistication of management of the lighting can affect the growth rates of the cannabis plant as growth of the plant can be ‘forced’ by the use of artificial lights. This involves finely controlled lighting with light meters, and timers, and changing the types of light emissions at the different stages of growth (Cervantes, 2002). Other considerations are the density of plants per light, wattage of the light and then provision of the necessary plant nutrients to accompany this forced growth. Once the grower turns to the use of high intensity lighting then fans and air conditioners must be introduced so the plants do not overheat under the lights. All of these draw large amounts of electricity which is expensive. To counter this, growers may turn to theft of electricity which requires illegally tapping into the electricity grid, a somewhat dangerous exercise.

Estimation of costs for growing cannabis in either an illegal or a legally-regulated framework is a challenge. Information was obtained from a wide range of sources including: the police, books on cannabis cultivation, the internet, personal conversations with those familiar with growing cannabis, and academic and other publications. While useful information was garnered, it remains a challenge to collate these data to arrive at useful resource estimates. Key pieces of information required include growing location, method of cultivation, density of planting, quantity and type of nutrients, rates of application of herbicides or fungicides, the amount of pruning undertaken and the source of light.

132 Chapter 6

In what follows, one setting for growing illicit cannabis is described in more detail in an attempt to make clearer the complexities of trying to estimate the resource implications of growing cannabis

6.3.1.1 One example of indoor growing

A spreadsheet of costs obtained by police from a confiscated computer removed from a cannabis-growing operation suggested that the setup costs in a three bedroom suburban Sydney house was approximately $105,000 including annual running costs of food (plant nutrients) and fuel of $3800 (Anonymous, 2010). This example may or not be generalisable. The costs of an illegal operation are summarised in Table 36.

Table 36: One estimate of costs and income for a single growing operation Item Value Source Set up costs (not depreciated or $105,300 NSW Police Force annualised) Consumables* $6,580 NSW Police Force Annual labour costs ($25/hour) $65,001 Caulkins 2010 Harvest labour costs ($200/plant) $921,600 Personal communication Rent (three BR house) $20,800 Sydney Total expenses $1,119,281

Harvest (64 lights, 1.5 lb per light, 4 Estimated from Brion & 145 kg crops per year) Associates, 2010 Cost per kg $7,720

Street price $20 per gram (Black et al., 2007) Total street value $2,902,991

Income minus expenses $1,869,511

*Includes fuel for the generator which this grower used rather than tapping into electricity grid.

The spreadsheet obtained from the police did not include the actual size (square metres) of the house; the labour costs throughout the growing period; the harvesting or drying costs or the intensity of planting or output. Some of the additional information is obtained from several sources including Brion and Associates (2010) who prepared an economic analysis for a proposed commercial cannabis operation in California (Brion and Associates and Nilsson Consulting, 2010).

Assuming 1,024 square feet of growing space, and one person to maintain the operation until harvest, and 64 lights with productivity from each light of 1.5 pounds per crop (Brion and Associates and Nilsson Consulting, 2010; Brion, 2010) the resulting production could be 36.4 kilograms per harvest or 145.5 kilograms per year with four 133 Chapter 6

harvests. At a street value of $20 per gram this equates to $2.9 million dollars. Even accounting for additional harvesting costs of an estimated 4-hours per plant to harvest the buds, with payment either in kind or $200 per half day (Personal communication BB, 2010), if the grower sells the product at the current street price, s/he may achieve an income of $1.8 million. This assumes all set-up costs occur in one year. If, as normally is done these costs are spread over multiple years, the profit would be larger although with risk of detection these growing operations may move frequently and incur greater costs than a normal business operation. Also this assumes the grower sells the dried cannabis directly to the consumer; if there was an intermediary the profits would be shared. The costs of employing sellers or street dealers are not included.

As pointed out by Caulkins (2010) in his attempt to undertake a similar analysis in California these estimates are fraught with challenges, and require considerable speculation. However, even with the caveats around density of planting, use of nutrients, prevalence of pests and diseases discussed earlier, these data combined with production estimates from Brion and Associates (2010) provide some indication of the potential profits that are currently available from these operations.

6.3.2 Estimating the resource implications of the various regulations

The next section describes the estimated costs of complying with and enforcing the regulations under the legalised–regulated model. Where there is potential evidence of the frequency and costs of non-compliance these are also included. This section then concludes with a discussion of the costs to government of such a regulatory structure and the issues of the black market. The costs of implementing and enforcing the regulations are based on best data available and where a logical range is available this is used for the sensitivity analysis. However, where no potential data to generate a range is available, 25% plus or minus is used in the sensitivity analysis. The choice of 25% is arbitrary but will provide more indication of the potential impact of variation than a narrower range.

134 Chapter 6

6.3.3 Individuals

6.3.3.1 Costs to the individual as a response to the regulations

i. Purchase–licence: For the purposes of this exercise the cost of obtaining a cannabis- user’s licence is $48 per year plus a five-year photo card at $45 per card. These costs are based on the current cost of a drivers licence in NSW (Roads and Traffic Authority, 2010). From Chapter 4, the base estimate of users under the legalised framework was approximately 690,000. Approximately 140,000 (20%) of these would renew their photo card annually. Every year there will be new users but there will also be others who do not renew. Not included here are the costs of the time for the consumers to study for, and undertake, a short comprehension test on the potential harms of cannabis and methods of reducing these harms. The total cost to the consumers is $31,122,000 annually and assumes cost recovery for government. ii. Compliance with the law that cannabis consumption is legally permitted for those over the age of 21 years: As with every set of regulations there will be non- compliance and it will be up to the regulatory agency to determine the level of enforcement. To estimate the potential personal costs of these infractions, the current number of persons who were issued an infringement notice by police for consuming alcohol underage or with supplying alcohol to someone under age were used. Data were provided by BOCSAR and the fines are the on-the-spot fines that may be issued for these infractions (Communities Office of Liquor, 2010). The average, over the three years of 2006–07 to 2008–09, of the number of infringement notices was 1,428 per year for those less than 18, and 132 to adults for supplying. These numbers were then multiplied by the value of the on-the-spot fine for a total of $1.5 million and $435,000 respectively. As the age of legal consumption in this model is 21, the number of infringement notices would likely be larger than presented here. iii. As with alcohol, driving under the influence of cannabis would be illegal under legalised–regulated model. In the most recent year (2009) for which data are available 480 drivers tested positive for cannabis (Daley, 2010). In this CBA it is assumed, in the absence of any data that the number of drivers detected will double due to both more people consuming cannabis and more police resources targeting

135 Chapter 6

these behaviours. The costs to the individual of these fines are estimated to be $960,000. v. NSW Health has introduced laws limiting where tobacco can be consumed, and there are on-the-spot penalties in place but no data were located on how many infringement notices have been issued. For the purposes of this work an estimate of 1,000 per year was assumed and existing penalties ($250) for tobacco were used. The overall monetary impact of this assumption is not large.

Table 37: Costs to the individual as a result of the regulation

Cost N Total Range Source

Licence Annual licence $48 546,000 $26,208,000 $19,656,000 – (Roads and Traffic required to $32,760,000 Authority, 2010) purchase cannabis website/ 25% plus and minus 5 year photo card $45 109,200 $4,914,000 $3,685,500 – (Roads and Traffic (pro-rated for 1 $6,142,500 Authority, 2010) 25% year) plus and minus Sub-total: licence $31,122,000 $23,341,500 – costs $38,902,500 Non-compliance Fines N Total Source/ Range Found in $1,100 1,428 $1,570,800 1,178,100– (Communities Office possession of 1,963,500 of Liquor Gaming and cannabis & not 21 Racing); BOCSAR/ years of age or over 25% plus and minus Penalties for $3,300 132 $435,600 $480,000 – (Communities Office selling, or 1,200,000 of Liquor Gaming and providing to Racing); BOSCAR anyone less than 21 /25% plus and minus years of age Additional offences $2,200 480 $960,000 187,500 – (Roads and Traffic of driving / 312,500 Authority, 2009) / 25% machinery under plus and minus influence (mid range fine) Disregarding $250 1,000 $250,000 $187,500 – (New South Wales restrictions as to $312,500 Government, 2006; where cannabis can Cancer Council be consumed Victoria., 2010) / 25% plus and minus Sub-total: cost of $3,216,400 $2,172,300 – non-compliance 4,020,500 Estimate of total $34,338,400 $25,513,800 – consumer costs 42,923,000

136 Chapter 6

Summing across these categories the potential costs to individuals for compliance with licensing and non-compliance with laws was estimated at $34.3 million (90% of which is the cost of the annual licence and photo card).

6.4 Grower

6.4.1.1 Estimating the cannabis production

As illustrated in the discussion of costs for illegal production of cannabis, estimating the cost of growing cannabis has several difficulties. Legal cannabis could be grown in greenhouses or large sheds, as a field crop, or as personal crop at a private residence much like tomatoes. Both the costs of growing and the amount of cannabis bud harvested would likely be lower for an outdoor crop (Caulkins, 2010b; Knight et al., 2010) although when the costs of security against theft are included the cost differential between indoors and outdoors may not be so large.

Information obtained from growers, police, and some of the many how-to-grow cannabis books suggest that a well tended, trimmed, and highly fed plant can grow eight to ten feet either indoors or outdoors, resulting in many buds, whereas more commercially grown crops with plants in closer proximity, and not tended as carefully, will have fewer buds. What is not known with any certainty is the magnitude of the interaction between each of these variables in terms of productivity and costs. Little research is available with which to try to estimate the costs of production or the potential yield. Recently, as part of a project to explore the impact on excise taxes should cannabis be legalised in California (Kilmer et al., 2010) an attempt was made to estimate the cost of cannabis cultivation (Caulkins, 2010b). Caulkins estimated growing costs in three situations—field, greenhouse, and five foot by five foot home grow set-up. Through some necessarily notable assumptions, growing costs were estimated at $30 per pound as a field crop; $150 per pound in a greenhouse; and $300 per pound indoors. Harvesting and drying costs were extra.

The difference in costs between indoor and greenhouse growing in the Caulkins (2010) estimates were that greenhouse operations relied heavily on sunlight, using few grow lights and electricity but the actual impact on productivity is not known. After reviewing information from Brion and Caulkins, plus other information obtained through on-line sources and private conversations, applying Australian costs – and

137 Chapter 6

thereby making just as many assumptions as Caulkins, the costs of growing cannabis in a large greenhouse with additional lighting in Australia were estimated. A single greenhouse can produce a number of crops per year. Brion (2010) uses six crops in her estimation of what might be grown in a factory-based growing operation in California. In Brion’s model, to achieve six crops, the plants are started elsewhere and then transplanted. Others assume three or four crops per year but none appear to allow for any crop failures.

The remainder of this section will conservatively assume that four crops per year would be grown in one acre greenhouses. In this model the greenhouse is staffed 24 hours a day, seven days a week (as per Brion), thus mitigating the need for extra security. Costs related to complying with regulations, such as testing for potency, transporting product, negotiating contracts, and legal fees are discussed in the next section. The data in Table 38 and Table 39 draws on Caulkins (2010) and Brion (2010) but is supplemented by Australian wages, prices for electricity and water as well as information from a well known cultivation guide for indoor cannabis (Cervantes, 2002). Additionally, the data are converted into metric measurements using an online converter (http://www.onlineconversion.com/weight_common.htm). US costs were converted to 2007 Australian dollars.

Table 38: Estimating electricity costs for indoor growing Cervantes Applied Caulkins Sources

Square feet 45,000 45,000 Brion Watts/sq foot 25 40 Cervantes/ Caulkins Total watts/hour (sq ft * watts/sq ft) 1,125,000 1,800,000 Cervantes Hours per light/crop 1,440 1,440 Cervantes Total watts (hours * total watts per 1,620,000,000 2,592,000,000 hour) Total kWh (divide watts by 1000) 1,620,000 2,592,000

Cost per kWh 0.16 0.16 Integral Energy, NSW Electricity costs per crop $259,200 $414,720

One of the most costly resources for cultivation of indoor cannabis is the lighting and the energy used by these lights. Using estimates from Brion (2010) of one light per 16 square feet, and Cervantes (2010) recommendation of 400 watts for sixteen square feet (25 W per square foot), and information on number of hours at various growing stages from Cervantes it was estimated that a total of 1.6 million kWh are required per crop. At $0.16 per kWh, the total cost per crop is $259,200. Caulkins (2010) estimated that

138 Chapter 6

for a 25 square foot room 40 watts of lighting are required which results in $414,720 per crop, holding all other variables constant. These calculations are presented in Table 38. Obviously, the use of solar panels or wind energy may decrease the long term energy costs for the grower but these were not factored into this analysis.

The other major cost variable is labour. Again using estimates from Brion, who estimated a 100,000 square foot cannabis growing operation, would require 167 fulltime employees to staff the facility 24 hours a day, and seven days a week. This converts to approximately 75 employees for a one-acre greenhouse. The salary costs were estimated per crop and equate to 18.75 FTE per crop.

Table 39: Summary of cost of growing in greenhouse Greenhouse Description Value Data Source Growing 1 acre area @8.34 /sq ft Set-up Greenhouse $34,875 amortized over 5 (Caulkins, 2010b) years & per crop $1,544 per Annual rent for 3 Land 3 acres rural (Elders Real Estate, 2010) crop acres rural property Operating Labour (includes $985,528 n=18.75 per crop; (Cervantes, 2002; Brion costs planting, tending, 24 hour 7 days wk, and Associates and cultivation and $626.6 /wk ; 28% Nilsson Consulting, 2010) drying) on-cost (Industrial Relations, 2010) Nutrients $98,448 $61.6/kg (Caulkins, 2010b) Lights $5,624 2,812 lights (Caulkins, 2010b) (Brion and Associates and 1,620,000 kWh @ Nilsson Consulting, 2010); Electricity $259,200 $0.16 per kWh (Cervantes, 2002); /NSW prices (Cervantes, 2002) / NSW Water $3,620 45 litres /plant prices Total indoor growing costs per $1,388,839 acre per crop Sensitivity analysis Per cent change

Change lights to 40 watts per sq $1,544,359 11% ft Decreasing employees to 15 FTE $1,191,733 -14%

As well as being responsible for planting, tending to the plants, ensuring the lighting and nutrients are maintained at appropriate levels, this staff would also harvest the cannabis buds which in this model is done manually, and takes approximately three to four hours per plant (Confidential source, BB, 2010). There are examples of mechanised harvesting equipment but there is some suggestion that such equipment impacts on the quality of the harvest. It has been estimated that approximately 50,000

139 Chapter 6

plants could be grown in a one-acre greenhouse (Brion, 2010). It was assumed that the one-acre building was situated on a three-acre rural-zoned property. This would allow space for other buildings such as for drying, offices and parking as per Brion (2010).

To test the impact on the total costs of some of these assumptions, the use of electricity and number of employees were varied. When the watts per square foot increase from 25 to (Cervantes) to 40 (Caulkins), the overall costs for one crop in a one acre indoor growing operation increase by 11%. If the number of employees drops from 18.75 fulltime equivalent (FTE) per crop to 15 (a 20% decrease), the overall costs decrease by 15%.

6.4.1.2 Complying with regulations – the grower

When operating in the legal setting the grower may not have the risks of operating in an illegal environment but does face other costs. The legal grower is subject to the regulations established for cannabis as well as regulations for farmers, employers and other businesses in the state of NSW. Costs of complying with such regulations are estimated based on an average greenhouse operation of one acre under cultivation, as described above, with additional space for offices, drying and packaging, etc.

In order to comply with proposed regulations, cannabis sent by the growers to the distributors must be tested for potency. This is to enable potency information to be contained on the packaging. Potency (amount of THC) is an area where additional research is required as little is known about the actual potency of cannabis consumed (McLaren et al., 2008; Knight et al., 2010). It is known however, that THC can vary within a plant and between plants even when grown under identical conditions (Knight et al., 2010). Costs of testing were confidentially sourced from one laboratory at $165 for the first test and $50 for every test thereafter if more than 10 were done (Confidential source LL, 2010). At this cost structure the range of potential costs if testing was done on every 50 kg to 100 kg, ranges from $95,000 to $45,000 per grower. An arbitrary choice of one test per every 100 kg of dried cannabis bud was used for this model. Additional research is required to ascertain the correct quantity of tests per kilogram required to assure consumers of the level of potency in the cannabis they purchase. Testing for pesticides was also included.

140 Chapter 6

In order to estimate the compliance and regulatory costs, the number of growers needs to be estimated. This requires knowing the demand for cannabis and average productivity of a crop. Toonen et al. (2006) examined cannabis plants confiscated by the Netherlands police and estimated an average yield per hydroponically indoor-grown plant 33.7g (1.2 oz) for a median sized ‘grow room’ of 259 plants. In assessing the size of the Canadian market, Bouchard (2007) used a combination of police arrest and seizure data, as well as interviews with growers. The yield of indoor hydroponic plants was estimated at between 25.5g (0.9 oz) per plant for a large crop (over 100 plants), and 36.9g (1.3 oz) per plant for a small or medium crop (under 100 plants) (Bouchard, 2007). Caulkins used the Toonen information, and estimated production of 2,041 kilograms of dried cannabis bud per crop per a one acre greenhouse (Caulkins, 2010b). Brion used information from a number of California growers, and combined lighting and nutrition factors to estimate that a one acre greenhouse could produce 1,595 kilograms of dried cannabis bud per crop (Brion and Associates and Nilsson Consulting, 2010).

Table 40: Additional annual costs per grower as result of regulations Total annual Category Per Cost per Source cost Staff training for $350 per person herbicide, pesticide use; every five years; http://www.dpiw.tas. requires training update Staff =75 assume a 50% $26,250 gov.au/ every five years, initial turn over in staff two days and one day each year refresher $165 for first $3600 to 4400 Previous assumptions Per 100 kg if sample and then per year ($900 and confidential all one plant $50 for every Testing level of THC to $1,100 per quote from testing type sample if N > crop) laboratory 10

$6400 to Previous assumptions Per 100 kg if $175 per sample 14,000 per and confidential all one plant Testing for pesticides if > 10 samples year ($2600 to quote from testing type 3500 per crop) laboratory

Legal contracts / Annual $1,000 Estimate solicitor (Department of Licence Annual $500 $500 Justice, 2009) $41,750 Total cost per one acre ($36,000 to operation $45,000)

Recently, a research group in New Zealand grew three cycles of six cannabis plants in experimental conditions in a 4.32 metre by 3.48 metre greenhouse. They harvested an

141 Chapter 6

average of 680 grams per plant with a range from 340 to 1,300 grams (Knight et al., 2010). Despite the larger amount of cannabis per plant that other estimates, this equates to 1,009 kilograms per acre which is lower than other sources. However, the paper does not indicate whether the six plants filled the greenhouse to capacity. Therefore using these data to estimate production for one acre may be a substantial underestimate of what might be produced in such a greenhouse. Alternatively, it may be that the additional space allowed the plant to grow more buds. Given this uncertainty only the previously described production estimates of Caulkins (2010b) and Brion (2010) will be used below.

In order to determine the potential number of one-acre greenhouses required to service the NSW market under legalisation the estimates of the number of kilograms per acre of dried cannabis bud are combined with the various cannabis consumption estimates from Chapter 4 (Table 12). The results are presented in Table 41 (underlined numbers indicate main estimates). Estimates of the number of greenhouses required are provided for all 18 estimates of consumption (Table 12 in Chapter 4) for both Brion’s and Caulkins’ production estimates. A worked example for the main estimate (78,790 kg.) from using Brion’s production is presented below.

78,790 kg of cannabis / (1,595 kg per crop * 4 crops per year) = 12.4 one-acre greenhouses

Table 41: Number of greenhouses required Kilograms per Average gm acre of dried Method 1 Method 2 per joint cannabis bud CI Low Mean CI Hi CI Lo Mean CI Hi

0.25 gm 4 6 7 6 8 11 1595 (B) * 4 0.37 gm 6 9 11 9 12 16 crops 0.5 gm 9 12 15 12 17 21

2041 (C) * 4 0.25 gm 3 5 6 5 7 8 crops 0.37 gm 5 7 8 7 10 12 0.5 gm 7 9 11 10 13 17 B= Brion assumptions, C= Caulkins assumptions

Depending on the various assumptions the number of greenhouses ranges from three to 21. The main estimate requires 10 to 12 one acre greenhouses. This assumes new consumers are distributed across rates of use according to existing distribution, and joints have on average 0.37 grams of cannabis. Even though the difference between

142 Chapter 6

Brion’s and Caulkins’ yield appears quite large, in term of estimating the number of growers required (given a one-acre greenhouse) there is little difference.

The estimation of the number of greenhouses rely on all previous assumptions but also on several other key factors such as: 1) no cannabis is home grown for personal use; 2) all cannabis which is sold will be grown by legal commercial growers; and 3) only NSW residents will purchase cannabis. Regarding the first factor, it is likely that many of those who are currently growing for their own consumption will continue to do so and others may take up this practice, although if the commercial product is seen to be of high quality and easily attainable at a reasonable price, home growing may diminish (Rolles, 2009). However, if prices are perceived as high and quality poor, cannabis may be less likely to be purchased in the retail market (Decorte, 2010). This is what appears to have happened in Canada with cannabis grown for medicinal purposes. The quality and potency of cannabis that was grown from seeds provided by the US Drug Enforcement Agency in a disused mine shaft were perceived to be low (Lucas, 2008). This led many medical cannabis users to turn to the black-market cannabis or to grow their own (Lucas, 2008).

Another assumption, that those in the existing black market will stop supplying cannabis if cannabis becomes legal is an assumption which while untested is improbable. This will be discussed further below in the context of the tobacco black market. A third assumption is that cannabis will only be sold to and for NSW residents. Despite the licensing system proposed, it is likely that some unquantifiable amount of on-selling by NSW residents will occur and moreover would be very difficult to stop. Failure of the first two assumptions would lead to a lower demand for legally grown cannabis while failure of the last one would lead to an increase in demand. For the purposes of this exercise analyses will be based on 11 (range 3 to 21) one-acre greenhouses required to supply the demand.

Table 42 combines the annualised cost of cultivation and production estimates per greenhouse to provide a cost per kilogram for both Brion and Caulkins estimates of production of cannabis. The main estimates for cost of production per kilogram are $547 and 700, with a range from $411 to $873. Costs of production are also provided allowing for one crop failure per year.

143 Chapter 6

Table 42: Cost of producing dried cannabis per greenhouse Sensitivity analysis Sensitivity analysis Main estimate low high Cultivation costs per $4,421,532 $3,316,148 $5,526,916 greenhouse per year Compliance costs per year $41,750 $36,000 $45,000 Total costs per one acre $4,463,282 $3,352,148 $5,571,916 operation Kilograms per acre (B) 6,380

Kilograms per acre (C ) 8,164

Cost per kilogram (B) $700 $525 $873 Cost per kilogram(C) $547 $411 $682 Assume one crop failure per year Cost per kilogram (B) $933 $701 $1,164 Cost per kilogram (C) $695 $522 $867 (B) Refers to Brion and (C) to Caulkins estimates of production

6.4.1.3 Return on investment for growers

The cost per kilogram described above does not include one key item, that of the return on investment for growers. A wage cost for the growers’ time might be included in the wage estimates above but this does not provide a return on their investment. In considering what a fair rate of return might be for growers, assessing the consequences of getting the decision wrong in the face of maintaining a retail price (including all taxes) at the current street price of $20 per gram is necessary. The difference between the growing costs and the revenue at the retail level is striking. The estimates of the annual production of cannabis bud in one greenhouse at 6,380 kg to 8,164 kg per year would provide retail revenues of $127,600,000 to $163,280,000. This is a 29- to 37- fold increase over costs of production.

If the price paid to the legal growers is too low they might be tempted to sell some of their product to the black market, and existing black-market growers who do not obtain a legal growers licence may not leave the market. It has been reported that legal tobacco farmers earned up to $10,000 per bale from selling to a black-market distributor and $800 per bale to a legal buyer, a 12.5 fold difference. At the same time consumers could buy one kilogram of chop-chop (finely cut unbranded black-market tobacco) for $80 to $100 compared to $320 for legal roll your own tobacco, a 3.2 to 4 times difference (Geis, 2005).

Black-market tobacco sellers have previously sold tobacco at about 6% to 8% of the black-market price which would equate to a $1.60 per gram for cannabis (Geis, 2005). 144 Chapter 6

Kilmer and colleagues (2010) estimated that a 10% drop in the price of cannabis in California, would result in an increase in consumption of 5.4%. This increase would be over and above the increase projected in Chapter 4. It is not feasible to estimate the effect on demand should prices decline from $20 to $1.60 (a 92% drop) as price elasticities are only valid in the immediate area of the price change and further, we do not know the shape of the demand curve (Kilmer et al., 2010).

Although not directly comparable, dairy farmers receive $0.48 to $.50 per litre for milk which is retailed at $1.10 to $1.65 per litre; a two- to three-and-half-fold difference (Australian Competition and Consumer Commission, 2008). In a submission to the ACCC, the Victorian Farmers Federation report that the retail mark-up on vegetables was about two to two-and-half times (Lloyd and Ford, 2008). If the assumption is made, that relationship between the farm gate price for cannabis and the retail price is similar to that for milk, the grower would receive approximately $5.50 to $8.80 per gram, which on average equates to $35 to $72 million per year per one acre greenhouse. It is this range that will be used in the summary of costs for the legalised–regulated policy option.

Table 43: Potential payments to growers by price and annual production of dried cannabis Annual production Annual production 6380 kg per acre 8164 kg per acre Payment to grower @$5.50 per gram $35,090,000 $44,902,000 Payment to grower @ $8.80per gram $56,144,000 $71,843,200

Total of 11 growers @ $5.50 per gram $385,990,000 $493,922,000 Total of 11 growers @ $8.80 per gram $617,584,000 $790,275,200

6.4.2 Estimating revenues

Not all cannabis consumed will be purchased at the cannabis shops. Under this model individuals are permitted to grow 10 cannabis plants at one time for personal use without a growers’ permit. Currently 7.2% of those who consume cannabis report that they usually obtain their cannabis by growing it themselves (Australian Institute of Health and Welfare, 2008b). In addition there will be the leakage of legal growers diverting to the black market or currently illegal growers continuing to operate. Data from tobacco provides estimates of the potential magnitude of this leakage.

145 Chapter 6

The legal tobacco growing industry no longer exists in Australia but when it did it was a highly regulated industry, with tariffs on imports and a significant excise tax on the sale of cigarettes. Both of these measures, introduced to control supply and decrease consumption, created distortions in the market. A review of the industry in 1994 found that the industry was inefficient, non-competitive and on the brink of collapse as a result of the structure, controls and supports. Over the subsequent years the number of growers dropped from 600 in 1994 to 346 at the end of 1995 and the industry was completely phased out by 2006 (Geis, 2005). This put downward pressure on prices the growers received from tobacco companies for their product (Cancer Council Victoria., 2010) while increases in the excise taxes on tobacco subsequently raised the price of cigarettes in the market place. This led to increased incentives to both sell and purchase illegal tobacco.

In 2001, the National Audit Office estimated that there was between $99 million and $220 million of revenue leakage from illicit tobacco. Further in 2007, it was estimated that of all tobacco consumed 6.4% was purchased illegally, resulting in a loss of $450 million in taxes (Cancer Council Victoria., 2010). Although there is some debate over the validity of the numbers, the tobacco industry has even higher estimates of government losses at $600 million (8.5%) (Geis, 2005). Other estimates suggest that one of every 17 cigarettes smoked in Australia contains chop-chop (Cancer Council Victoria., 2010).

One might think that the black-market supply of tobacco may have diminished without a legal tobacco growing industry as has been the situation since 2006 but it appears this is not the case. For example, in 2009 the Australian Taxation Office (ATO) reportedly seized a kiln, and around nine tonnes of stripped tobacco plants and four tonnes of tobacco leaf from greenhouses in NSW. This crop had an estimated excise value of $1.2 million (Australian Taxation Office). The ATO also reported seizures of 12 tonnes of leaf and six tonnes of cut tobacco in 2006, and nine tonnes of tobacco in 2008 (Australian Taxation Office). In addition, in 2007, Australian Customs and Border Protection Service reportedly foiled some 40 separate attempts to smuggle cigarettes and tobacco into Australia, mostly in sea cargo containers through Port Botany in Sydney. Australian Customs and Border Protection suggested they successfully prevented potential revenue evasion of approximately $214 million over the three years 2006–2008 (Australian Customs and Border Protection Service., 2009). 146 Chapter 6

In summary, although tobacco is a legal substance, regardless of whether it was legally permitted to be grown in Australia or not, there is considerable black-market tobacco available. This is despite substantial resources expended on detection and enforcement. Geis (2005) reports the Excise Tax Unit allocated about 75 per cent of its investigative resources to violations involving chop-chop. The penalties for illicit transport, storage, manufacture and sale of tobacco can extend to loss of licence for legal growers, fines of up to $55,000 and two years imprisonment (Geis, 2005). In addition there are also Penalty Infringement Notices (PINS) for the possession or sale of tobacco on which an excise tax has not been paid – these can be imposed on the spot and are up to $2200.

There is little to suggest that the estimate of 6.4% to 8.5% of tobacco consumed being from black-market sources will not also occur in cannabis. For the purpose of these estimates, the total revenue to government will be adjusted downward by 15% (range 10 to 20%) to account for the black market and home growing.

6.5 The distributor and retailer

In this model of legalised–regulated use and supply of cannabis, the decision was made that only cannabis and implements for using cannabis would be available for sale in shops that sold cannabis. The opportunity costs of this decision are significant. If cannabis sales were permitted alongside alcohol, the marginal costs of selling cannabis would be minimal but more resources may be required to enforce regulations. As a key objective of constructing such a restricted legalised model, was to put forward one way of allowing cannabis to be sold legally, whilst attempting to ensure it is sold under conditions that ensure public health objectives can be met, this model of retailing is maintained despite its potentially high opportunity costs.

6.5.1 Operational costs and complying with regulations

In considering the potential number and distribution of retail outlets for cannabis the data for tobacco and alcohol outlets were taken into account. There are approximately 35,000 retail outlets for tobacco in Australia (Cancer Council Victoria., 2010), with approximately 30% (10,500) likely to be in NSW, and 1,600 outlets for packaged alcohol (off-licence) sales in NSW but there is considerable clustering of outlets in some locations. Rather than replicate the alcohol model, another approach is to use the

147 Chapter 6

existing structure of the Local Government Councils and Shires of which there are 165 in NSW (Local Government and Shires Associations of New South Wales, 2010).

Table 45 presents potential costs for retailing cannabis under a variety of assumptions. Notably, all these assumptions retain the initial assumption of cannabis and cannabis implements-only shops. The base case (Model 2) assumptions are that all outlets will be open 12 hours per day, and for safety, each outlet always has at least two persons in the shop at all times. Those cells shaded in grey indicate where assumptions have changed. Model 1 has one outlet in every council or shire (n=165), Model 2 has two (n=330), Models 3, 4 and 5 have 500 outlets over the state and in Model 6 there are 1000 retail outlets in the State and the final model has 1600 outlets which is approximately equivalent to the number of off-licence retail outlets. Other than Models 2 and 5, all outlets are open every day of the year with the exception of Christmas Day and Good Friday as is the current regulation for hotels and pubs (Communities Office of Liquor Gaming and Racing, 2010). Models 2 and 5 close one day each week. Each shop has a manager; there are eight regional managers across the state and an overall general manager. The wages are NSW Public Employees Award 2008, and on-costs (superannuation, leave, long service leave etc.) are included in the final costs. Also included are an annual rental cost for a shop and security costs including cameras and fees to a security company. The total cost of operating retail outlets under these various assumptions ranges from $47.7 million to $450.1 million annually. The latter is as a result of having similar number of cannabis outlets as there currently are alcohol off- licence sales outlets.

Model 2 with 330 outlets around the state, two staff in the shop at all times, open 12 hours a day, but closed one day a week is used as the base model. This is approximately 2,000 consumers per shop as compared to approximately 2,700 persons per retail outlet for packaged alcohol (80% of NSW adult population consumes alcohol, 1,600 outlets) although there are other locations to purchase alcohol such as bars, pubs and restaurants. The operating costs for Model 2 are $82.9 million.

It would not be compulsory that each council or shire must have two outlets; some might have more either because of distance or population while some may choose to have only one or none (Wouters et al., 2010). The placement of these shops would be subject to community input as are new alcohol outlets. The current cost estimates for

148 Chapter 6

those undertaking a community impact statement for a new alcohol point of sale location is about $15,000 (Department of Arts Sport and Recreation, 2009). If the same community consultation process were followed for 330 outlets, this would require approximately $4.95 million (range $2.4 to $24 million) to ensure the selected locations met all necessary guidelines and restrictions. This would include not being located near a school. This cost would be a one-off cost for each new location.

Table 44: Unit cost for estimating costs of retail outlets Unit costs Classification

Shop staff $36,229 NSW Admin clerk level 3 Shop supervisor $46,320 Area supervisor Regional manager $66,166 Regional field manger General manager $110,000 Senior manager Solicitor $76,896 Solicitor grade 3 Security $10,000 Rent $15,000 Electricity / utilities $2,000

The costs of packaging, labelling and transporting the cannabis have not been estimated as it is unclear what these costs might be. The cost of warnings and content information, as well as child-proof containers would not be an inconsequential cost but estimates were not available.

149

Table 45: Cost estimates for retail outlets Model 1 Model 2 (base) Model 3 Model 4 Model 5 Model 6 Model 7 Number of outlets 165 330 500 500 500 1000 1600 Opening hours 12 12 12 12 12 12 12 Days open per year 363 312 363 363 312 363 363 Minimum staff in shop 2 2 2 4 2 2 2 Total hours needed per shop 8712 7488 8712 17,424 7488 8712 8712 Sales staff per shop 5.19 4.46 5.19 10.37 4.46 5.19 5.19 Supervisors 165 330 500 500 500 1000 1600 Sales staff 691 1141 2093 4686 1729 4186 6697 Total shop staff 856 1471 2593 5186 2229 5186 8297 Operational staff Regional managers 8 8 8 8 8 8 8 General manager 1 1 1 1 1 1 1 Admin staff /internet orders 10 10 10 10 10 10 10 Legal staff (negotiate with growers) 2 2 2 2 2 2 2 Costs Summary Rent/ utilities $2,805,000 $5,610,000 $8,500,000 $8,500,000 $8,500,000 $17,000,000 $27,200,000 Security (camera equip /alarm) $1,650,000 $3,300,000 $5,000,000 $5,000,000 $5,000,000 $10,000,000 $16,000,000 Total operation costs $4,455,000 $8,910,000 $13,500,000 $13,500,000 $13,500,000 $27,000,000 $43,200,000 Staff costs (with 28% on-costs) Shop staff $32,027,264 $52,905,105 $97,052,315 $217,291,191 $80,159,250 $194,104,631 $310,567,409 Shop supervisor $9,782,784 $19,565,568 $29,644,800 $29,644,800 $29,644,800 $59,289,600 $94,863,360 Regional manager $677,540 $677,540 $677,540 $677,540 $677,540 $677,540 $677,540 General manager $140,800 $140,800 $140,800 $140,800 $140,800 $140,800 $140,800 Solicitor grade 3 $196,854 $196,854 $196,854 $196,854 $196,854 $196,854 $196,854 Admin staff /internet orders $463,731 $463,731 $463,731 $463,731 $463,731 $463,731 $463,731 Staff costs $43,288,973 $73,949,598 $128,176,040 $248,414,916 $111,282,975 $254,873,156 $406,909,694 Total $47,743,973 $82,859,598 $141,676,040 $261,914,916 $124,782,975 $281,873,156 $450,109,694

150 Chapter 6

6.6 Cost to government of enforcing these regulations

The costs of enforcing the proposed regulations are not known although some research on tobacco control is applicable. The Australian National Expert Advisory Committee on Tobacco has recommended expenditures for tobacco control in Australia (Ministerial Council on Drug Strategy, 2005) and these expenditures are broken down into categories such as enforcement, research, programs to motivate quitting, and assisting high need groups (Cancer Council Victoria., 2010). The Australian estimates are based on a set of guidelines produced by the US Centre for Disease Control (CDC) that were developed to assist with budgetary decisions in tobacco control (Centers for Disease Control and Prevention, 2007).

Table 46: Costs to government of enforcing regulations Regulation Low Mid High Low Mid High Source /law Per capita NSW recommended Enforcing (Cancer Council underage laws $0.23 $0.60 $0.96 $1,016,161 $2,614,142 $4,188,267 Victoria., 2010); (both use and NSW Pop, CPI supply) Enforcing restrictions as (Cancer Council to where $0.09 $0.31 $0.54 $375,536 $1,348,421 $2,334,624 Victoria., 2010); cannabis can NSW Pop, CPI be consumed Total $1,391,697 $3,962,562 $6,522,891

The recommended national per capita expenditure data were provided for individual expenditure categories (enforcement, health, cessation programs etc.) but only the overall per capita was provided for each jurisdiction. A ratio of the NSW per capita amount ($5.70) to the national per capita amount ($6.90) was multiplied by each of the expenditure categories to estimate the amount for NSW. The per capita amounts were then multiplied by the NSW population and finally the original 2004 estimates were converted into 2007 AUD (ABS, 2008), and finally adjusted downward for the expected prevalence of cannabis use in the population (12.4%) compared to that of tobacco (19.4%) (Australian Institute of Health and Welfare, 2008b).

Table 46 provides estimates of the costs of enforcing regulations. The amount recommended to enforce underage laws is $2,614,142 (range of $1,391,697 to

151 Chapter 6

$4,188,267). The amount recommended to enforce restrictions as to where cannabis can be consumed is $1,348,421 (range $375,536 to $2,334,624).

The MERIT program, as described in Chapter 5, is a pre-plea three-month program designed for offenders with a demonstrable drug problem, who are eligible for bail, and who show potential for treatment and rehabilitation (Matruglio, 2008). The need for MERIT would not disappear with legalisation. Only 35% of those who used MERIT and nominated cannabis as their main drug had a cannabis offence as their principal offence. The remainder (65%) of those offenders who nominated cannabis as their drug of concern had other principal offences such as unlawful entry with intent /burglary, traffic offences, or acts committed with intent to cause injury. Therefore 65% of the previous expenditures are included under the legalised–regulated model for the CBA ($2,461,648).

Driving a vehicle under the influence of cannabis is already illegal. Police currently have the power to conduct roadside testing on any driver for THC (the active component of cannabis) (Roads and Traffic Authority, 2008). The rationale for these laws and the testing relates to the acute negative impact on cognitive abilities from cannabis (Hall and Pacula, 2003). The purpose of a testing program is two-fold; first to work as a deterrent against driving under the influence of a drug and secondly to detect drug affected drivers (Tay-Teo, 2009).

A detailed description of the NSW protocol for drug testing was not available therefore data from a Victorian trial of drug driving testing in 2004–05 were used (Drummer et al., 2007). Resource implications were extrapolated from an evaluation of the Victorian Roadside Testing trial which was conducted with one ‘drug bus’, two police vehicles, and drug test devices, storage, transport and forensic testing of the samples (Tay-Teo, 2009) (p 168). In 2004–05 AUD, the Victorian Police report the cost of the testing at $1.3 with an additional $1.0 million for staff for the twelve months of the trial (Tay-Teo, 2009).

The resource implications of a drug-testing program for NSW were not available, with the exception of information from a press release in October 2009 by the Honourable Tony Kelly, Police Minister who announced the government would obtained a further five vans on top of the existing three for drugs testing at a cost to refit of $140,000 (Kelly, 2009).

In using these data there are two issues that need to be addressed: 1) what proportion of the $2.3 million are marginal costs and; 2) given the recent move to increase to eight

152 Chapter 6 drug testing vans (Kelly, 2009) would the NSW Police Force add more resources to random drug detection if the cannabis were legalised? The cost of additional vans may not be as straightforward as multiplying the $2.3 million for one ‘drug bus’ in Victoria by five (the current number of five vans in NSW) or eight (the number forecast by Minister Kelly) (Kelly, 2009). This is because it is unclear what proportion of the Victorian expenditure was for the purchase of the drug bus and laboratory equipment and staff in administrative roles versus police and analysts time. The budget for the Traffic and Commuter Services division which is responsible for drug and alcohol testing, along with other traffic offences was $213.4 million (Minister for Police and Minister for the Illawara, 2008) in 2008/09 therefore $11.34 million expenditure on a single program would not be unfeasible but at 5.4% of the expenditure it does appear excessive given 37,000 drug tests had occurred over three years compared to the 4,204,525 random alcohol tests conducted in one year (NSW Police Force, 2009a). The other question is, given the recent expansion of drug-driving testing, whether any additional expansion would be necessary should cannabis become legalised. Unfortunately there are no data with which to address these questions. Therefore, the same value of drug driving testing ($11.34 million) will be included for both the status quo and the legalised model.

6.7 Cost to government of a regulatory agency

Another important cost to government pertains to establishing the regulations and determining how they will be enforced. There is a precedent, in the Tasmanian Poppy Advisory and Control Board which in order to be compliant with the 1961 United Nations Single Convention on Narcotic Drugs regulates the control and supervision of poppy growing and production in Tasmania (Department of Justice, 2009). The budgeted costs for this regulatory agency are used as the basis for the costs of operating the cannabis regulatory board. As described on their website, the role of the Poppy Advisory Board is to:

• process applications for licences • to advise on all matters relating to the cultivation, production and transport of poppies and poppy material • to collect and collate statistical information and prepare reports

153 Chapter 6

• to liaise with Australian Government Departments to fulfil Australia's obligations under the International Drug Conventions and to oversee security matters for Tasmanian crops • to maintain strict controls over all aspects of growing through to processing.

Field officers of the agency regularly patrol crops and liaise with the Tasmanian Police Poppy Task Force to ensure compliance. During the growing and harvesting season, crops are regularly monitored by Poppy Advisory and Control Board field officers and any illegal activity is investigated by Tasmania Police Poppy Task Force Officers. The costs in the 2008 Tasmanian Budget were $690,000 to cover the activities described above with approximately 13,000 hectares under cultivation.

Although fewer hectares will be required for cannabis cultivation in NSW, the cannabis regulatory agency will have the responsibility for determining the supply, negotiating contracts and setting the price paid to grower as well as field supervision. Given these responsibilities, the size of the state both in population and geography and the fact that cannabis is grown year round, it is expected the costs for the regulatory agency would be higher than for the Poppy Advisory and Control Board in the small geographical area of Tasmania. No precise estimate can be made so an additional 50% is placed on the $690,000, resulting in an annual expenditure of $1,042,500. Given that this figure is small relative to some others any error is unlikely to have a significant impact.

Table 47: Estimate of annual costs for regulatory agency Main estimate Range: low Range: high Annual cost for regulatory $1,042,500 $690,000 $1,390,000 agency

A final category of costs are programs provided by the state to convey consumer information on potential harms of cannabis use. Examples of messages might include: to avoid adding tobacco to cannabis; the dangers of driving under the influence of cannabis; and the symptoms of cannabis dependence. Other campaigns may involve encouraging moderate use, and information on where to seek treatment. Once again recommended expenditures for tobacco control in Australia are used (Ministerial Council on Drug Strategy, 2005) but adjusted for the relative prevalence.

Table 48: Consumer information campaigns Main estimate Range: lower Range: high

154 Chapter 6

Consumer information $209,378 $162,390 $240,490 State-wide campaigns to promote Quit messages $11,819,461 $5,428,014 $17,959,328 Addressing social determinants $449,535 $309,170 $562,620 Total $12,478,374 $5,899,574 $18,762,437 Source: (Cancer Council Victoria., 2010)

6.8 Summary

Table 49 presents a summary of the costs of complying with the regulatory structure and Table 50 summaries the potential of revenues and expenditures. All costs are converted to total annual costs.

The costs for the individuals and growers to comply (and for individuals some costs for some not complying) with regulations is estimated at $36.37 million while the costs of running the retail enterprise, the regulatory agency, drug-driving testing and providing consumer information is $134.9 million. The costs of the retail shops are included as this is the opportunity cost of this legalised–regulated model. The overall costs of the regulatory framework are $171.3 million.

Table 49: Summary of costs of complying with and enforcing regulations (range) Annual Base estimate Range: low Range: high Individuals (N=546,000)

Annual licence and five yearly photo card $31,122,000 $23,341,500 $38,902,500 (Table 37) Cost of failing to comply regulations (Table 37) $3,216,400 $2172300 $4,020,500 Growers (N=11)

Growers compliance costs $459,250 $396,000 $495,000 Retail costs

Operating cannabis shops (Table 45) $82,859,598 $47,743,973 $450,109,694 Regulatory agency

Operational costs (Table 47) $1,042,500 $690,000 $1,390,000 Enforcing compliance at the level of the $6,424,211 $3,237,933 $9,599,952 individual including MERIT Drug driving testing $11,500,000 $2,300,000 $18,400,000 Consumer information and quit campaigns (Table 48) $12,478,374 $5,899,574 $18,762,437 Payments to growers (Table 43) $617,584,000 $385,990,000 $790,275,200

The data in Table 50 provides the expected total revenue of $1,360 million (range $1,263- $1,437 million) under the assumption that 15% of cannabis will be self-grown or supplied by the black market.

Table 50: Potential revenue and expenses Base estimate Range: low Range: high

155 Chapter 6

Revenue

Market value all cannabis $1,599,840,000 $1,403,600,000 $1,796,080,000 Per cent of retail cannabis home -15% -10% -20% grown or sold on black market Total value of cannabis sold at retail $1,359,864,000 $1,263,240,000 $1,436,864,000 outlets Expenses

Payment to growers $617,584,000 $385,990,000 $790,275,200 Cost of regulatory agency and $83,902,098 $48,433,973 $451,499,694 retail outlets Total expenses $701,486,098 $434,423,973 $1,241,774,894 Potential surplus $658,377,902 $828,816,027 $195,089,106

The total direct expenses of operating the monopoly cannabis market are comprised of the payment to growers, the cost of the regulatory agency, and the cost to operate the retail market at a total of $701 million (range $434 to $1,241 million). The difference between expenses and revenues is estimated at $638 million (range $195 to $828 million).

6.9 Discussion

This chapter has begun the process of trying to identify the potential costs of one legalised but regulated framework for cannabis. Many have recommended that such a framework be developed (Nadelmann, 1992; McDonald et al., 1994; van Dijk, 1998; Hall and Babor, 2000; MacCoun and Reuter, 2001a; Haden, 2002; Englesman, 2003; Hall and Pacula, 2003; Rolles et al., 2006; Haden, 2008; Rollings, 2008; Room et al., 2008; van den Brink, 2008; Hall and Degenhardt, 2010), and several have outlined potential characteristics (National Task Force on Cannabis Regulation, 1982; Englesman, 2003; Haden, 2008; Rollings, 2008) but to date there have been no apparent attempts to try to quantify the costs of such an endeavour.

In an attempt to learn from the tobacco and alcohol control, this framework has incorporated many regulations some may find oppressive but each choice was made with the intent to minimise population and social harms. For example, the requirement that individuals obtain a licence in order to purchase cannabis requires the individual to demonstrate knowledge of the potential harms of cannabis, and where to seek help for dependence should this be of concern. Additionally, if they are convicted of certain activities such as supplying cannabis to those less than 21 years of age individuals may lose their cannabis licence for a period of time. Finally, the licence may lessen cannabis tourism. The presumption was made that this would be an objective of the regulatory

156 Chapter 6 policy. To simplify the model, different prices for different potencies have not been used but evidence from alcohol policy (Babor et al., 2003; Goudie et al., 2007) suggests that if legalisation were to be introduced differential pricing should be considered to encourage use of lower potency cannabis.

Decisions as to what to include in the model, while affected by existing and proposed regulations for alcohol, tobacco, and poppy growing, were made in an attempt to meet the criteria for good regulation outlined earlier (Banks, 2003). That is, regulations should be better than the alternative, be clear and concise, be consistent with other laws, and be administered by responsible organisations and be robust to errors. There is one obvious inconsistency with existing laws. This is the decision to make cannabis legally available only to those of 21 years of age. This is not consistent with alcohol laws but is consistent with the literature that suggests limiting alcohol consumption to those 21 years of age limits long-term harms (Babor et al., 2003). The choice of 21 was a deliberate decision and based on the theory that the brain is still developing until at least this age (Lubman et al., 2007). Other evidence also suggests that later cannabis initiation is less harmful (Chen et al., 2009; van Ours and Willams, 2009).

Another requirement of the model which some may find objectionable, but which again was introduced deliberately, was the decision to have a limited number of legal commercial growers and to conduct the distribution and retail operations through a non- profit structure, in this instance a government monopoly. Such a structure may be resisted by the Australian Consumer and Competition Commission which has as one objective to ensure there is competition in agricultural industries (Australian Competition and Consumer Commission, 2008). However, given the potential harms, and the documented efforts of the tobacco and alcohol industries to ensure their sales are not restricted, and given legal responsibilities of company boards to attempt to maximise profits for their shareholders (Caetano, 2006; Snyder et al., 2006; Freeman et al., 2008; Anderson et al., 2009; Chapman and Freeman, 2009; Cancer Council Victoria., 2010) a regulated monopoly was the preferred model. It must be acknowledged that not all business costs such as the transportation and packaging costs have been included but these are expected to be minor relative to revenues and other costs.

It is noteworthy that the net revenue to government from this model is in order of $658 million for NSW. This compares to the total revenue to government in 2004/05 from

157 Chapter 6 alcohol taxation (excise taxes, custom duties, GST and wine equalisation tax) of $5,112.5 million and from tobacco taxes (excise tax, custom duties, and GST) it was $6,675.4 million. Some might argue that governments may use the revenue from cannabis sales (whether directly or through taxes) as source of income, such as recently proposed in California (Kilmer et al., 2010). The development of a regulatory arms-length regulatory agency to set the prices and payments to growers may go some way to prevent this occurring.

Finally, it is clear that without regulation of the cannabis supply chain, an oversupply and subsequent price decrease would likely occur given the potential profit. But implementing a regulatory framework has its own set of problems (Geis, 2005). The large potential revenues imply that there will be considerable pressures placed on developing the regulations and policies around such an endeavour. For ease of calculation, the greenhouse size was set at one acre, but there may be a larger number of smaller operations each producing a range of product. However, as the complexity increases, the enforcement of regulations and deterrence of the black-market cannabis will become more difficult and increase in expense. The setting of the price generates further issues. If it is too low the legal growers sell into the black market, and if it is too high the intrinsic value of the licence increases which creates its own challenges. Additionally, there is no guarantee that the move to a legal market would eradicate the black market.

The lessons from this chapter are glaring. There is still much to learn about the actual production costs of cannabis but whatever the costs it is unlikely they will ever approach the potential revenues if prices are maintained at or close to current street prices. This leaves open the question as to where the bulk of the revenues end up (government, growers, or the black market).

When considering implementing a regulated–legalised framework for cannabis, given its current illicit status, there is a unique opportunity for the development of a policy framework in a controlled fashion. Forefront in decision-making should be the trade-off between non-profit monopoly control of the retail market and the reality that corporations are required by law to act in the best interest of their shareholders, which most often will not coincide with public heath objectives (Callard et al., 2005).

158 Chapter 6

It is apparent that such an exercise is not without its challenges and should any authority intend to legalise and regulate cannabis there is considerable scope for additional work on ensuring the policy is based on scientific evidence.

159 Chapter 7

Chapter 7 : Contingent valuation: quantifying stigma from a criminal offence

7.1 Introduction

The previous two chapters have focused on the costs to the criminal justice system under the current cannabis policy and the costs to consumers, growers, retailers and government under a legalised–regulated policy. Chapter 3 provided the welfare economic theory for the CBA and outlined the various costs and benefits, both public and private to be included in the CBA. In this chapter contingent valuation (CV) is used to value one of the potential benefits of moving to a legalised policy. This is the perceived loss in utility related to the stigma from a criminal record for possession and use of cannabis. Stigma has been described as that which “occurs when a person possesses an attribute or a status (a ‘stigma’) that makes that person less desirable or acceptable in other people’s eyes ...” (Lloyd, 2010).

This chapter adds to the literature in two ways; it is a rare use of contingent valuation being undertaken as part of a traditional CBA (Borghi, 2007) and it will add to the policy relevant data for assessing responses to cannabis policy options. CV studies are conducted widely, for example, in health care (Diener et al., 1998; Olsen and Smith, 2001); in crime prevention (Ludgwig and Cook; Cohen et al., 2004; Cohen, 2007), to examine the value of intangible effects of crime (Atkinson et al., 2005) and improved street lighting (Willis et al., 2005). Despite the widespread use there remains methodological concerns that often lead to the results being challenged (Diener et al., 1998; Smith, 2006; Frew, 2010b; McIntosh, 2010).

Early concerns regarding the application of CV to environmental issues led to the development of the Report of the NOAA Panel on Contingent Valuation (Arrow et al., 1993). This report made several recommendations on methods to be used in the conduct of a CV (McIntosh, 2010) but the applicability of these recommendations to evaluation of health and health care was not clear (Diener et al., 1998). As a consequence CV methods within health (and elsewhere) have varied considerably.

Whatever the methods, the survey must be acceptable to respondents and the results must be internally consistent and reproducible (Golafshani, 2003). The methods must also be assessed for validity. That is, does the research measure what it was intended to

160 Chapter 7

measure (criterion validity) (Golafshani, 2003); is the measurement accurate; and can inferences be made from the results to the theory on which the survey was based (construct validity)? Ensuring that each of these areas of potential concern is addressed is important in any CV study (Frew, 2010a; McIntosh, 2010).

The hypothetical nature of the question is often criticised, although it is this quality that allows valuation of benefits where other studies cannot (McIntosh, 2010) as often the respondents are being asked to value a new good or service (Olsen and Smith, 2001; Frew, 2010a). The hypothetical nature of the studies may lead respondents to overstate their ability to pay (Johannesson et al., 1996) in an attempt to please, or because they recognise it is a hypothetical question (McIntosh, 2010). Alternatively they fail to recognise the budgetary impact such a decision may have. A related criticism of CV is the relationship between willingness and ability to pay. Those who can, often choose to pay more which may subsequently bias against low income groups or programs targeted at these groups. However, if societal valuations are of interest, the use of a representative sample will go some way to address this latter issue (Johannesson et al., 1996).

The remainder of this chapter provides the background, relevant economic theory, the methods, descriptive analysis, estimates of the societal value of the stigma, and regression analysis examining the characteristics that impact on the amount respondents are willing to pay (WTP).

7.2 Background

It has been argued that the stigma associated with a criminal record is a central concept in relation to cannabis policy. Stigma can arguably operate to decrease the likelihood of use. Some suggest that the personal costs of a criminal record are a justifiable

161 Chapter 7

consequence of being found in possession of cannabis and act as a deterrent for the use of an illegal substance (Lott, 1992; Ahern et al., 2007). Still others debate drug use itself, making the point that unlike stigma attached to mental illness there is a value to having negative stigma related to drug use, and that in fact those drugs with higher stigma (heroin) have a lower prevalence of use than those with low stigma (cannabis) (McKeganey, 2010). Further, it is argued that the probability of criminal sanction for use is so low, and most health outcomes are widely debated, particularly with cannabis, that the fear of stigma may be the only major deterrent of illicit drug use (McKeganey, 2010).

Conversely, stigma for minor offences can be seen as a negative consequence that could have far-reaching implications for an individual’s future, disproportionate to the original offence. A key factor in the provision of discretionary powers to police and prosecutors, or the complete removal of the criminal offence for possession of a small amount of cannabis from the legal statutes (European Monitoring Centre for Drugs and Drug Addiction, 2007; Room et al., 2008) is the perceived negative personal costs to the individual (Lenton and Heale, 2000; Weatherburn and Jones, 2001; Weatherburn et al., 2003; Hughes and Stevens, 2007; van Laar and van Ooyen-Houben, 2009).

There are two intersecting sets of research. One is the stigmatisation of an individual because they are a ‘drug user’ and the other is the stigmatisation as a consequence of a criminal record. In a review of research on the stigmatisation of drug users, it was argued that stigmatisation occurs whenever there is an attribute (stigma) that makes one less acceptable in someone else’s view (Lloyd, 2010). The point is made that this stigmatisation only becomes serious when the stigma obscures the rest of a person’s identity and that this often occurs with drug use. Although this chapter deals with stigma from a criminal record, research into stigmatisation from drug use is also germane, as those who receive a criminal record for a cannabis offence may be labelled both a drug user and a criminal thus compounding the effect (Lloyd, 2010).

Ahern et al. (2007) examined the impact of stigma on the health of drug users and found that those who perceive that they are being discriminated against or feel stigmatised have poorer health status (Ahern et al., 2007). This poorer mental and physical health status persists even after adjusting for a number of potentially confounding factors. Lloyd (2010) concluded that studies examining drug users and stigma demonstrate that

162 Chapter 7

stigmatisation has a profound negative impact on problem drug users. Additionally, those in treatment for their drug use, who feel stigmatised and discriminated against, are more likely to drop out of treatment (Ahern et al., 2007).

Research into the consequences of a criminal conviction can include social stigma— embarrassment and disapproval of family (Lenton et al., 1999 ) and friends (Rasmusen, 1996; Lenton and Heale, 2000); economic stigma such as negative impact on employment and future wages, loss of professional registration, denial of some types of business licences, negative impact on ability to hold public office and lost wages if imprisoned (Lott, 1992; Waldfogel, 1994; Rasmusen, 1996; Lenton et al., 1999 ; Lenton and Heale, 2000; Pager, 2003; Funk, 2004); and potential restrictions on foreign travel (Lenton and Heale, 2000).

What remains unknown, and what this chapter sets out to do, is to quantify the societal value that is placed on not obtaining a criminal record and potential stigmatisation from a criminal record for a cannabis offence.

7.3 Methods

CV is a survey technique that can be used to value policies or programs, or outcomes from those policies, which are not valued by the market (Drummond et al., 2005). The theoretical underpinnings lie in utility maximisation with an income constraint. Given the budgetary constraint of the individual, the utility function may take the form of

U = f(X, Snc, Y) > U0=f(X, Scr, Y) where X is the bundle of goods consumed, Scr is stigma (i.e. a person found guilty of a cannabis offence has a criminal record; Snc is stigma (or the lack of stigma) with a payment to avoid the criminal record, and Y is household income. The compensating variation C, the measure of the utility change is

U = f(X, Snc, Y-C) => U0=f(X, Scr, Y) where C is the amount of money paid, to reduce stigma from Scr to Snc but will leave the person just as well off as before.

163 Chapter 7

7.3.1 Survey

In order to ensure acceptability and content validity a CV study requires the development of a hypothetical scenario, describing the good or service, that is reasonable and acceptable to respondents. Respondents are then asked what the maximum they would be willing to pay for the good described, or alternatively, minimum amount they would need to accept in order to be deprived of the good. Respondents may be totally unfamiliar with the context therefore the scenario must be clear, succinct, and provide all necessary information in order for a decision to be made, while at the same time not overburdening the respondent with too much information (Smith, 2003). It must also be realistic enough that the respondent takes it seriously and believes that it could happen (Smith, 2003; Willis et al., 2005) but not elicit strong preferences (Smith, 2003). This requires neutral language and piloting of the survey to ensure it is readable and unambiguous (Frew, 2010a).

The scenario in this work described a realistic situation where the respondent (or in alternative versions of the questionnaire, someone close to them) was detected by police with a small amount of cannabis. A small amount of cannabis was described as no more than sufficient quantity to make approximately 24 joints. The respondents were informed the person detected with the cannabis had no previous criminal record, and had committed no other offences at the time. A copy of the questionnaire is included in the Appendix–Chapter 7. The scenario briefly described the process of being arrested, having to attend court, and some of the potential consequences of a criminal record. Consequences included a potential negative impact on employment opportunities, a decreased ability to travel to some countries, as well as the social stigma of having a criminal record. Respondents were informed in the survey that approximately three in every 100 people who use cannabis in any one year are detected by police.

7.3.1.1 Payment vehicle

Part of getting respondents to take the survey seriously, is ensuring the payment vehicle is reasonable and suits the context of the study (Smith, 2003). The payment vehicle is the type of payment being asked of the respondent (Frew, 2010a). Inappropriate payment vehicles may result in biased responses (Smith, 2006; Frew, 2010a; McIntosh, 2010). This may occur when respondents are asked to consider a payment for a good or service for which they do not normally pay for directly, e.g. health care in a public 164 Chapter 7

system (Smith, 2006) or the vehicle does not fit into the scenario. Payment options often used in CV studies include direct out-of-pocket payments, changes in taxation, insurance or voluntary payments. A tax-based payment is often use in health care but was not relevant for this survey. Here, the question was framed in terms of a civil penalty (fine) to be paid to avoid a criminal record. This also precluded respondents from thinking the payment was a bribe to police or court officials. The range provided to the respondents in this survey, as presented below, was the existing range of fines for cannabis possession/use in Australia.

7.3.1.2 Format of elicitation

In determining the elicitation method there are two main issues: starting bias (Smith, 2006) and method of elicitation (Smith, 2006; Frew, 2010a). Starting bias, which is when the first value respondents see affects their choice, is different from strategic bias (Frew, 2010a). Strategic bias, when respondents deliberately overstate their WTP for strategic purposes, is unlikely to occur in this context.

CV questions can be asked using an open-ended question, payment card method, bidding game, single bounded dichotomous choice/referendum method, double bounded dichotomous choice format or one-and-a-half bound distribution (Ableson, 2000; Pearce et al., 2006). Open-ended questions ask the person directly what they are WTP (Frew, 2010a) but they often result in a large number of non-responses (Pellegrini and Jeanrenaud; Smith, 2000), zeros, outliers and unreliable responses (Pearce et al., 2006) particularly if the subject has not thought about the issue. Additionally, in a normal market we are not used to stating the price of a desired purchase. Iterative bidding, similar to haggling in markets is often claimed to capture the highest price that consumers are willing to pay (Frew, 2010a). Other frequently used methods are closed- end and dichotomous choice questions, where respondents are asked yes or no whether they are willing to pay a given amount. Another method, the payment card provides respondents a range of values from which to choose.

In this study, a decision was made to use a payment card format with a follow-up question for those who selected the maximum amount. Following this all respondents were informed what percentage of their weekly income they had selected (having previously been asked about household income) and all were permitted to change their

165 Chapter 7

answer. Although some maintain the dichotomous choice method provides realistic responses, and is more realistic, it was felt that the simplicity of a payment card was preferable as respondents would have just completed nine discrete choice profiles. It was felt that to use the more complex bidding/trade-off questions such as the bidding game format might result in a high rate of non-response.

Smith explores whether offering the payment card values in either a random order, moving from high to low values, or from low to high values when using a payment card make a difference in the WTP responses (Smith, 2006). He found there was no statistical difference in results between high-low and random presentation, or between low-high and high-low, but there were a statistical differences between low-high and random values on the payment card. Although not statistically greater, the high-low version consistently gave higher results than the others. This current survey used both high-low and low-high presentation of the values to explore whether there was starting- point bias. Additionally, as some respondents may know with some certainty that they would never use or be found with cannabis and thus find the question posed unreasonable, half of the respondents were asked what they would be WTP if the person detected with cannabis was a loved one. There are mixed findings in the literature about WTP for one’s self versus others, with some detecting that WTP for one’s self is more important (Smith, 2007) while others find no difference (Schwarzinger et al., 2009).

In summary, WTP was elicited using four versions of a payment card. In two versions, the respondents were asked to consider what they might pay in terms of a fine if they were found by police with a small amount of cannabis in their possession; one of these started with zero and went to $2,500 and the other started at $2,500 and went to zero. In the other two versions, the respondent was asked what fine they would pay if ‘someone who was close to them’ (a child, partner or sibling) was found in possession of a small amount of cannabis; again there were two versions of the card—one low to high and the other high to low.

The values for the fines were based on the actual fine structure in Australia for the possession/use of a small amount of cannabis and were $0, $50, $150, $250, $500, $750, $1,000, $1,250, $1,500, $2,000 and $2,500.

166 Chapter 7

Prior to the final study being undertaken, the survey was piloted among 60 drug and alcohol researchers, health economists, and PhD students and subsequently modified to improve the ease of readability. Pilot respondents did not report any difficulty in either understanding the question or in making a choice of an amount they were WTP, neither did they report that they found the survey unrealistic.

7.3.1.3 The sample

This study was conducted alongside a larger internet panel survey measuring community preferences for different types of cannabis policies (see Chapter 10). The larger survey consisted of nine complex discrete choice tasks; the CV question was the tenth task for the respondents. The sample was recruited from a panel of 90,000 Australians through a commercial survey company (SSI). The survey was completed anonymously online. Recruitment was stratified by ‘ever used cannabis’ with an attempt to approximate the population distribution of cannabis use (~35%) and age and gender distribution of the general population over the age of 18.

7.3.1.4 The mode of administration

Research elsewhere suggests that face-to-face administration of CV surveys may provide the ‘best’ results but the method of elicitation plays a role in choice of mode of administration. Elicitation methods such as bidding game, single bounded dichotomous choice, double bounded dichotomous choice format or one-and-a-half bound distribution are best done face to face or over the telephone (Frew, 2010a). As this survey was included alongside a complex discrete choice survey, and involved questions about illicit drug use it was felt that anonymous administration might avoid answers that were approval seeking (Smith, 2003). Administering a face-to-face survey required resources that were beyond this study and telephone was not suitable for the DCE.

The administration of the survey involved random panel members being notified of the study and their participation requested. Respondents then logged onto a website to complete the survey. As various strata (cannabis use, age, and gender) were completed, access to the survey was denied to those strata. In addition, a predefined algorithm was used to identify those who completed the whole survey too quickly; thus those who did not take sufficient time to read the background or the questions could be excluded.

167 Chapter 7

Demographics such as age, gender, educational status, household income, employment status, marital status, number of children were collected. Additional questions on previous cannabis use, whether the person had previously attended court, whether they believed cannabis had any health benefits, whether they felt cannabis was addictive or not and where the thought they were on a 10-point left–right political scale. The questions on cannabis use are those which are used in the National Drug Strategy Household Survey (Australian Institute of Health and Welfare, 2008b); the questions on attitudes regarding cannabis use have previously been used in research conducted among individuals with private health insurance; the socio-demographic questions were those used in the 2006 Census (Australian Bureau of Statistics, 2006b) and the question on political views was obtained from the Australian Survey on Social Attitudes (Phillips et al., 2007).

Data were collected over a ten-day period in December 2009 (2nd to 12th). Respondents who completed the survey had their name added to a draw for a prize and a small donation was made to a charity for each person who completed the survey. Ethics permission was granted by the University of New South Wales HREC.

7.3.2 Analyses

Two sets of analyses were conducted (descriptive and regression analyses). The descriptive analyses include the mean and median WTP and the total societal valuation of stigma. This includes examining differences in the mean WTP from three potential responses. The three responses are the initial indication of WTP, the follow-up question for those who selected the maximum amount in the first question, and the final response where respondents were permitted to change their answer after being informed what percentage of their weekly income they had selected. Additionally for the final response a 1% trim was applied to exclude outliers. Normally the mean WTP is preferred over the median when using the results in a policy context (Ableson, 2000 ; Pearce et al., 2006), however, if the data are skewed or there are some large outliers the median may be a better predictor of what the public are willing to pay. Both are reported. As the WTP responses are expected to be positively skewed non-parametric tests were used. A Kruskal-Wallis test was done to test the null hypothesis that there were no differences across the four versions of the survey. The Wilcoxon Rank Sum test was used to test

168 Chapter 7

the null hypothesis that there were no differences between the high-low and low-high versions and between paying for self versus others.

Regression analyses were done to examine the impact that the various characteristics and attributes may have had on the WTP. Ex ante it was expected that household income would be a strong predictor of WTP all else being constant. Household income was adjusted for family size using an OECD modified scale (OECD Project on Income Distribution and Poverty, 2005). It was also expected that as cannabis use differs by age and gender (males and younger age groups consume at a higher rate), that those groups might recognise their increased risk, and all else constant be willing to pay more to avoid a criminal record. Similarly, those who were employed, either full time or part time, or those who were currently looking for work recognised the risk they may put their employment status under may also choose to pay more.

It was anticipated that those who thought cannabis was likely to be addictive might be WTP more as would those who thought cannabis might have health benefits. Ex ante, the expectations on whether those who were politically left-leaning were likely to be WTP more than those who were on the right or in the middle of the political scale were unclear (McKnight, 2005).

An initial log linear OLS regression analysis was estimated with all the relevant demographics, characteristics and attitudes. The dependent variable, the WTP, was log transformed as is often done with skewed data, and one was added to all values due to zeros in the responses, i.e. Ln (WTP +1) (Wooldridge, 2009; Frew, 2010b). Two other versions, Ln (WTP +0.1) and Ln (WTP +10) were also analysed. Tests for misspecification (Ramsey RESET) and Breusch-Pagan test for heteroskedasticity were performed. Two quadratics, one each for age and income were included in the final model to account for misspecification and weighted least squares used to account for the heteroskedasticity (Wooldridge, 2009).

7.4 Results

A total of 1,670 persons logged on to undertake the survey over a period of one week (see Figure 8). Of those, 350 belonged to stratum already full and could not access the survey beyond the screening questions. A total of 222 people who started the survey did not complete it; they either logged off before completion or the internet became

169 Chapter 7

inactive. Of the 1,022 valid survey responses, 147 did not complete the household income question and as it was expected that household income would be an important explanatory variable, those respondent were excluded leaving a total sample of 875 for the WTP analysis.

Figure 8: Survey respondents

Total logged on N = 1670

Completed survey Did not complete Stratum full (not N = 1098 survey N = 222 required) N= 350

Completed survey Completed survey and met time criteria but did not meet time N = 1022 criteria N = 78

Income not stated N = 147

Completed income question N = 875

Comparisons of characteristics (age, gender, employment status, previous cannabis use) did not reveal any differences between those who stated their income (n=875) and those who did not (n=147).

7.4.1 Demographics, characteristics and attitudes

There were no apparent differences in demographics across the four versions (high-low; low-high; self and other) of the survey. The average household income was $59,000 (AUD) with an average household size of 2.25 (SD 1.21) persons. The sample was 48% female and the average age was 46.3 (SD 15.19). Within the sample, 24.3% had a tertiary education, 20.7% completed less than Year 12, 12.4% completed Year 12 and the remainder had a diploma or qualifications from a technical college. Sixty-five point six per cent of the sample was married or in a de facto relationship.

Over the whole sample, 41.7% had ever used cannabis, 11.4% reported using within the past year, with the average frequency of consumption being 7.7 occasions in the past

170 Chapter 7

year. Eighty-six point three per cent indicated that they believed cannabis was always or usually addictive, and 26.4% believe there are some health benefits from the use of cannabis.

Table 51: Characteristics and demographics (n=875) About self About loved one Total

Characteristic Lo-hi Hi-lo Lo-hi Hi-lo ALL n=217 n=230 n=218 N= 210 N=875

Age – mean (SD) 47.4 46.8 47.1 45.5 46.3 (15.10) (15.1) (15.48) (15.15) (15.19) Males % 56 54 50 48 52 Believe cannabis has health benefits % 28.1 25.2 27.0 25.2 26.4 Believe cannabis addictive % 87.6 84.5 83.9 89.5 86.3 Used cannabis in past year % 11.1 11.7 11.0 11.9 11.4 Used cannabis ever % 43.3 42.6 37.8 43.3 41.7 No. of occasions used past year – 7.7 8.0 8.7 6.0 7.6 mean (SD) (44.4) (43.64) (45.01) (37.46) (42.74) Employment status# FT/PT employed % 51.7 53.4 51.3 53.3 52.6 Retired/ pension/ looking for 28.2 33.0 34.4 29.3 32.1 employment Full-time student % 5.53 2.61 3.67 3.33 3.8 Education^ Education less than Year 12 % 24.8 20.4 18.0 20.0 20.7 Education equal Year 12 % 19.4 18.3 21.6 12.4 19.7 University education % 23.5 25.6 22.9 25.2 24.3 Ever attended court % 18.4 23.0 17.4 17.6 19.2 Married/de facto % 69.1 64.5 64.2 64.8 65.6 Number of children 0.68 0.63 0.65 0.67 .65 (1.09) (1.05) (1.0) (.96) (1.03) With children living in household % 35.9 33.5 32.1 40.0 35.31 Household income mean (SD) 58,162 58,710 59,543 59,951 59,079. (40,179) (46,513) (46,746) (42,260) (43,999) #Excluded category is homemaker; ^Excluded category is TAFE/technical college, diploma

There was some apparent differences between the survey respondents’ age distribution population (Figure 9) compared to the 2006 Census data (Australian Bureau of Statistics, 2006b). There are clearly fewer males aged 18 to 29 and more males 60+ in the sample than the population and there are fewer females in the 60+ category. Other age groups are similar.

171 Chapter 7

Figure 9: Comparisons of age

18.0% 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% 18-29 30-39 40-49 50-59 60+

Sample -Male Pop- Male Sample- Female Pop- Female

Source: 2006 Census data, ABS

Examining educational attainment, the sample is more highly educated than the overall population which is not surprising as it is an internet-based survey (see Table 52). This is also true of the panel in general. Educational status was not significantly correlated with WTP, nor was it significant in early regression models and therefore was excluded from the final models, as was a state variable. States and territories were grouped using dummy variable =1 if it had a civil penalty for cannabis program and zero otherwise.

Table 52: Educational attainment Sample Population

Left school before finishing year 10 6% 16% Finished year 10 (School Certificate equivalent) 14% 35% Finished year 12 or higher 79% 49% Source: 2006 Census data, ABS

The distribution of income between the sample and the population (Figure 10) is similar. In terms of employment, the results are similar to the labour force data which reports an unemployment rate of 5.2% (age 16+) compared to 6.3% of the survey respondents reportedly looking for work (Australian Bureau of Statistics, 2010).

172 Chapter 7

Figure 10: Distribution of weekly income

120%

100%

80%

60%

40% survey 20% Population

Source: Population data sourced from 2006 Census data, ABS

In summary, while the survey sample does differ slightly in some characteristics, income distribution among the respondents which was expected ex ante to be a key explanatory variable is similar to the population distribution.

7.4.2 Societal WTP

Every survey respondent had at least one opportunity to alter their initial response to the WTP question and those who originally selected the maximum, $2,500, had two opportunities. The latter group were provided an opportunity to state their maximum willingness to pay, and then all respondents were informed of the percentage of their weekly income they had selected. All respondents were then informed they could revise their amount if they wished. At the end of the final revision, 121 respondents had changed the amount they were willing to pay.

Figure 11 demonstrates the distribution of final responses, with those in blue (shaded darker) falling within the responses in the original list, with the remainder over and above the original values given. In the ‘more’ category the amounts range from $20,000 to $250,000.

173 Chapter 7

Figure 11: Distribution of revised responses to WTP question

160 140 120 100 80 60 40 Frequency of responses 20 0

Final WPT $

The mean willingness to pay for the payment card only (first question) was $897 (SD 865). After the final revision, the mean willingness to pay was $1,780 (SD 9204) with a range from $0 to $250,000 and was highly skewed (Table 53). A 1% trim was applied. Trims are often used in the presence of outliers that may be of dubious reliability (Chilton et al., 2004). The mean WTP of the trimmed data was $1,230 (SD 1945), with a range of $0 to $10,000. Notably the median was $500 for each of the three comparisons. The trimmed data were subsequently used to estimate the total willingness to pay excluding 1% of the responses which may have been of doubtful reliability.

Table 53: Mean WTP for payment card, final payment and 1% trim Payment card Open payment Open payment 1% trim N 875 875 866 Mean $898 $1,780 $1,231 Std. deviation 865 9,204 1,945 Median $500 $500 $500 Range $0 to $2,500 $0 to $250,000 $0 to$10,000 Skewness 0.83 23.04 2.97

Across the four versions of the survey, the mean willingness to pay to avoid stigma over the 866 respondents ranged from $1,112 in the group that was asked about WTP for others and who had the low-high version, to $1,322 in the group asked about self who also had the low-high version (Table 54). Although the trimmed data are less skewed,

174 Chapter 7

some skewness remains as evident in the differences between the mean and medians. The median is $500 in all but one group. The group asked about ‘self’, with the ‘high- low version’, had a median of $750.

Table 54: WTP to avoid a criminal record by scenario (trimmed data) Scenario about loved All Scenario about self one Lo-hi Hi-lo Lo-hi Hi-lo

N 866 214 229 215 208 Mean $1,231 $1,322 $1,303 $1,112 $1,179 Std. deviation 1,945 2,256 1,839 1,820 1,839 Median 500 500 750 500 500 Skewness 2.97 2.79 2.94 2.90 3.14

The Kruskal Wallis test rejected the null hypothesis of no differences in WTP across the four groups (χ2=8.498, p=0.04). A Wilcoxon Rank Sum test (results in Table 55) suggests a difference in WTP between those presented with a low-high payment card relative to those who were presented with the high-low payment card, with the high-low version producing significantly higher amounts. This suggests that starting bias may be present. While not significant at the 5% of significance, the comparison between amounts WTP for self versus a loved one is approaching significance (p = 0.07).

Table 55: Wilcoxon Rank Sum (WTP) – 1% Trim Z score p Comparison Payment card: comparing self versus loved ones 1.81 0.07 (pooled all high and all low) Payment card low to high (pooled all self and all 2.332 0.02 loved ones)

7.4.2.1 Valuation for the cost benefit analysis

Whilst the mean/median results are of interest in and of themselves, the results can additionally be used to generate a total societal figure for the value of stigma. This figure can then be used as a measure of benefit in the CBA (see Chapter 9). In order to arrive at a total value of stigma avoided, when criminal sanctions are not imposed for possess/use offences, the results of the WTP must be aggregated (Borghi, 2007). The population to which this population is applied is a critical issue. There are a number of potential populations to consider and they include: the whole population of cannabis users who bear the risk; those cannabis users who were detected by police for a possession/use offence; those who went to court; or finally just those who were detected

175 Chapter 7

for a cannabis offence and did not have a prior criminal (non-drug) offence. As is illustrated in Table 56, the total value of the societal WTP to avoid stigma is very different depending on which population is chosen, and also whether the mean or the median is used.

Table 56: Total societal willingness to pay options 1% trimmed mean Total value median Population (one year) N $1231 $500 $ $ Cannabis- using population 438,501 $539,794,731 $219,250,500 Possess/use offenders (includes 12,254 $15,084,674 $6,127,000 cautioning) Cannabis offenders who go to court 6,026 $7,418,006 $3,013,000 Cannabis offenders minus those with prior 1,076 $1,324,556 $538,000 & concurrent offences

In this study, which was conducted to quantify one benefit of moving from a policy where cannabis use is illegal (but with a cannabis cautioning program) to a legalised policy, the most appropriate population is those who actually attend court for a cannabis offence (n=6,026). This is the group who are faced with the potential stigma, whereas those who are not detected or are cautioned do not receive a criminal record. Thus this is a measure of the value of the stigma avoided in moving from one policy to another. When the number who attended court in one year for cannabis offence was multiplied by the mean WTP, the result is $7.4 million. Both the mean and the median are reported (Table 56), but as the mean reflects the range of values chosen by the sample, it is used in the analysis (Drummond et al., 2005).

7.4.2.2 Comparing WTP responses among those who changed their minds and those who did not

Before examining the regression analysis the characteristics of those who changed their answers and those who did not are presented. One hundred and twenty-one respondents chose to change their answers from payment card to the open-ended question; of these 18 decreased the amount and 103 increased it. Table 57 presents those characteristics which are different between the two groups. The final mean WTP for the group who chose to change their response was $8881 (Std dev 23,557) while the group that did not change had a mean of $639 (Std dev 621). Those who did change their mind were on average less likely to have recently used cannabis in the past, were more likely to be married, have a higher household income, be employed either full time or part time, and

176 Chapter 7

not believe that cannabis has any health benefits. The group who changed were also more likely to classify themselves right-leaning on a left-to-right political scale than those who did not change.

Table 57: Differences between those who changed their answers and those who did not Did not change Did change

N= 764 N=121

Mean SD Mean SD p

Final WTP ($) 639 621 8,881 23,557 <0.001 Cannabis: ever used % 43 34 0.045

Married/de facto % 64 78 0.001

Household income $ 38,768 29,872 47,395 35,281 0.012 Believe cannabis has benefits % 27 20 0.050

Full- or part- time employed % 51 62 0.019

Right on left/right political scale % 18 30 0.008

There were no significant differences in age, gender, recent use of cannabis, other employment status, or number of children.

7.4.3 Regression analyses

The results of the regression analyses are provided in Table 58 and Table 59. As previously indicated the dependent variable, the WTP, has been log transformed as is often done with skewed data, and one was added to all values due to zeros in the responses i.e. Ln (WTP +1) (Wooldridge, 2009; Frew, 2010b). Two other versions, Ln (WTP +0.1) and Ln (WTP +10) were also analysed with no change in the signs, levels of significance, or substantive changes in the coefficients. There were no indications that the zeros were protest responses, as the respondents completed this and all remaining questions. It is quite reasonable that some people would be willing to pay nothing so all data remained in the analysis.

There are five models. The first three use the final response (untrimmed) and demonstrate the impact of misspecification and heteroskedasticity. The Ramsay RESET test in Model 1 (OLS) suggested misspecification, however, after introducing quadratics for age and income this was resolved (p=0.313). The Breusch-Pagan test suggested heteroskedasticity (p<0.001). A weighted least squares model with weights proportional to income and income squared was subsequently used (Wooldridge, 2009).

177 Chapter 7

Table 58: Regression analysis using data from the final revised stated WTP3 OLS Model 1 OLS Model 2 WLS Model 3 Open-ended Open-ended Open-ended N= 875 Coeff Sig Std Coeff Sig Std Coeff Sig Std error Error Error Dependent variable = Ln (final WTP+1) Household inc 0.000013 *** 0.00003 0.00004 *** .00001 0.00004 *** 0.0000 1 Income2 -2.1E-10 *** 4E-11 -1.8E-10 *** 5.0E-11 Female 0.347 ** 0.159 0.267 0.167 0.196 0.153 Married/defacto 0.303 * 0.166 0.243 0.165 0.255 0.155 Can^: used -0.206 0.242 -0.630 ** 0.310 -0.590 ** 0.273 recently Can^: health -0.357 ** 0.170 -0.335 ** 0.169 -0.307 ** 0.157 benefits Can^: believe 0.435 ** 0.215 0.397 ** 0.212 0.492 ** 0.194 addictive Hi_lo 0.161 0.146 0.163 0.144 0.197 0.133 Employ FT/PT 0.601 ** 0.256 0.404 0.258 0.557 ** 0.261 Retired 0.281 0.313 0.252 0.311 0.537 * 0.316 FT Student 0.505 0.458 0.404 0.455 0.578 0.441 Looking for 0.399 0.376 0.392 0.374 0.555 0.389 work No. of children 0.074 0.084 0.100 0.084 0.068 0.082 Right on 0.354 * 0.187 0.294 0.186 0.289 * 0.169 political scale Age -0.010 0.007 -0.037 0.032 -0.004 0.030 Age2 0.000 0.000 0.000 0.000 Female*use 0.907 * 0.460 0.928 ** 0.419 Constant 4.618 *** 0.470 4.742 *** 0.750 3.934 *** 0.715 R2 0.09 0.12 0.12 F 5.91 6.67 6.74 p <0.000 <0.001 <0.0001 Breusch-Pagan, 2.82 <.001 p Ramsey 0.001 RESET, p *** p<.0001 *** p<.05 * p<.10 ^can = cannabis

Before examining individual coefficients, it is worth noting that in re-specifying the model and correcting for heteroskedasticity there were no sign changes. The main economic variable, the household income adjusted for family size (OECD modified scale) remained highly significant across all models. The explanatory power of the models, R2=0.12 is low but this is not unusual in cross-sectional data (Wooldridge, 2009).

3 In an earlier model, educational classifications and other political views (left, centre and missing) were included but never approached significance and for the sake of parsimony were not included in the final models. 178 Chapter 7

Table 59 presents the results for data from the original payment card data (Model 4) and the final data with a 1% trim applied (Model 5). Model 3 is also re-presented for ease of comparison. Although there were significant changes in the mean WTP between three sets of data, there are only a few differences across the various regression models. There are no sign changes, and the income effect remains large. One variable, high-low dummy was not significant in the model using the open-ended WTP data (Model 3) but was significant in the payment card (Model 4) and returns to significance at the 90% level using the trimmed data.

Table 59: Comparing payment card, open-ended and trimmed data Model 5: WLS opened- Model 4: WLS payment Model 3: WLS open-ended ended & 1% trim card (n=875) (n=875) (n=866) Coeff Sig Std error Coeff Sig Std Error Coeff Sig Std error Dependent Ln (WTP Payment card+ 1) Ln (open-ended WTP + 1) Ln (TRIM open-ended variable WTP +1) Inc_ 0.00003 *** 0.00001 0.00004 *** 0.00001 0.00004 *** 00.00 household Inc_HH2 -1.8E- -1.94E- -1.7E-10 *** 4.5E-11 *** 5.0E-11 *** 4.91E-11 10 10 Sex_female 0.173 0.136 0.196 0.153 0.186 0.152 Married 0.202 0.138 0.255 0.155 0.218 0.153 Used -0.426 * 0.240 -0.590 ** 0.273 -0.564 ** 0.270 recently Health -0.278 ** 0.140 -0.307 ** 0.157 -0.288 * 0.155 benefits Addictive 0.503 ** 0.171 0.492 ** 0.194 0.523 ** 0.192 Hi_lo 0.274 ** 0.118 0.197 0.133 0.223 * 0.131 Emp FT/PT 0.647 ** 0.237 0.557 ** 0.261 0.409 * 0.258 Retired 0.577 ** 0.287 0.537 * 0.316 0.42 0.311 FT student 0.693 * 0.393 0.578 0.441 0.520 0.434 Looking for 0.571 0.354 0.555 0.389 0.534 0.382 work No. of 0.120 0.074 0.068 0.082 0.070 0.081 children Age -0.015 0.027 -0.004 0.030 -0.014 0.030 2 Age 0.000 0.000 0.000 0.000 0.000 0.000 Female*use 0.673 * 0.370 0.928 ** 0.419 0.852 ** 0.414 Right on the political 0.241 0.150 0.289 * 0.169 0.242 0.168 scale Constant 4.084 0.638 3.934 0.715 4.066 0.70

2 R 0.12 0.12 0.12 F 6.78 6.74 6.91 p <0.0001 <0.0001 <0.0001

The willingness to pay is always significantly and positively related to whether the respondent believes cannabis is usually or always addictive, and whether the respondent 179 Chapter 7

is employed full time or part time. The coefficients on the various employment levels become non-significant in Model 5, the trimmed-ended response. Those who believe there are health benefits and those who had used cannabis recently are WTP less than those who have not. Although highly correlated with WTP, age was not significant, neither was the number of children in the household.

In Model 5, the household income coefficient indicates a 0.004% increase in WTP for every additional dollar of income, up to a turning point of approximately $103,800 per annum, at which time the rate per additional dollar of income WTP begins to decrease (Wooldridge, 2009). The turning point for income across the three WLS models in Table 59 is between $88,000 and $110,000, with the payment card the lowest.

The coefficient on gender was significant in Model 1, but when the interaction between use and females was added it was no longer significant. As there are differences in the rates of cannabis use between males and females at the population level an interaction term (female* use) was constructed to examine whether these differences extended to differences in WTP. The significant coefficient on this interaction term suggests that there are differences in WTP. Using the trimmed results, relative to males who have not used cannabis in the past 12 months, females who have not used are willing to pay 18.6% more, and females who have used cannabis are willing to pay (18.6-56.4+85.2) = 47.4% more, while males who have used cannabis in the past 12 months would be willing to pay 56.4% less.

7.5 Discussion

The results in this chapter suggest that individuals are able to place a value on avoiding the stigma associated with a criminal charge related to cannabis possession/use. The results indicate respondents are willing to pay a mean of $1,231 (median value of $500) to avoid stigmatisation by family, friends and neighbours due to a criminal record for either themselves or a loved one. This stigma may potentially impact on the individual’s ability to obtain or maintain employment, on future earnings, on the likelihood of seeking treatment for dependence and on ability to travel overseas, as well as restricting professional and business activities (Lott, 1992; Waldfogel, 1994; Rasmusen, 1996; Lenton and Heale, 2000; Pager, 2003; Funk, 2004).

180 Chapter 7

Results from the regression analyses suggest that respondents who worked either full or part time, or believed cannabis was addictive were willing to pay more, as were those with higher household incomes even when adjusted for family size. This was not an unexpected finding. Theory would suggest that those who have more at risk and those who are able to pay more will pay more. Intuitively one might expect those who use cannabis or believe it has health benefits might be willing to pay more. Perhaps surprisingly, they were willing to pay less. This may be because they have different social norms for cannabis use; many in their social groups may consume cannabis so they are less concerned with the potential stigma. Additionally they may be aware of the low risk of detection and believe they are not at risk at being detected. Although there was no significant difference between males and females WTP, there were notable differences between males and females when recent use was interacted with gender. Females who have used cannabis recently were willing to pay more than their male counterparts who had not.

The decisions to conduct the survey online and to use a payment card were pragmatic given the resources available. The payment card method, with plausible monetary values for the fine (Ableson, 2000) while minimising outliers, is susceptible to biases due to the starting/ending points and this was evident here. Other formats such as bidding games are also subject to starting-point bias (Johannesson et al., 1996; Ableson, 2000; Pearce et al., 2006). Bounded dichotomous choice requires the respondent to say only yes or no to one amount however, this can be an inefficient method of obtaining estimates (Pearce et al., 2006). Double bounded, and one-and-a half bounded are variations of bounded dichotomous choice but they do not all correspond to the same WTP distribution (Pearce et al., 2006) and impose assumptions on to the shape of the demand curve (Johannesson et al., 1996). While all have their limitations, the payment card method was assessed to be a suitable method for this study.

Allowing individuals to give an open-ended response did result in outliers that affected the mean and the distribution around the mean. The application of a 1% trim point moderated this somewhat. Personally administered surveys may have led to fewer outlier bids (Smith, 2000) but were not feasible. It is interesting that only 121 of the respondents changed their bids when provided the opportunity, and while this may reflect that their initial choice was their true choice it is also possible their responses

181 Chapter 7

may have been affected by the payment card values. This would be true in all of the methods, with the exception of a single open-ended question.

The use of commercial market research firms and internet panels are frequently used in this type of work (Bartels et al., 2006; Champ and Welsh, 2006; Louviere and Lancsar, 2009) although it is recognised that biases may be introduced as not everyone has access to the internet. It is also feasible that the hypothetical nature of this study may have resulted in ‘yea saying’ or unconsidered responses although as this was the last of a series of questions on cannabis policy and cannabis use respondents would have had opportunity to reflect on these issues.

The theoretical validity of contingent valuation results can sometimes be tested by examining whether the results conform to predictions and assessing whether expected relationships hold between the demographic and attitudinal information collected (Pearce et al., 2006). This study did find the expected strong relationship between WTP and household income, and between WTP and whether or not the respondent was employed.

It is not possible to test for convergent validity of the value of stigma as there is no evidence of such a study having been conducted previously. Nonetheless, these results can be placed in the context of what happens in three jurisdictions in Australia where civil fines (no criminal record) are issued for detection of a small amount of cannabis. In Western Australia, after four years of operation, only 66.3% of the Cannabis Infringement Notices (CINs) were finalised either by paying a fine or attending a treatment facility. Failure to pay the CIN resulted in the loss of one’s driver’s licence. In South Australia only about 55% of offences were expiated (Hunter, 2001). In South Australia, when matters are forwarded to court due to lack of payment, this may result in a criminal record. A similar pattern exists in the Australian Capital Territory (ACT) where the highest rate of expiation was 67.8% between 1994 and 1996, and more recently the rate dropped to 48.6% (Drug and Alcohol Office, 2007). The average civil infringement notice (civil fine) for possession or use of cannabis is between $100 and $200 compared to a societal WTP of $1,230, with only 8% not indicating the would be WTP zero.

182 Chapter 7

What is not clear is whether differences between the stated WTP in this study and results from these programs arise because of overstating of WTP to avoid stigma in the survey, or differences between respondents and participants in the civil infringement programs. The participants of these programs are not the same as the respondents of the survey. Cannabis users are younger and more likely to be male and disadvantaged. In Western Australia, 81.1% of those issued with a CIN were aged between 18 and 34 (Drug and Alcohol Office, 2007) whereas the contingent valuation survey was relatively representative of the population. There were higher rates of non-payment among the Indigenous population, females, those issued with multiple CINs, and those who live in some of the rural and remote areas. This final point is important as attending treatment is an alternative to paying the fines in WA but a lack of access to treatment facilities for those who were unable to afford to pay the fines was an impediment (Drug and Alcohol Office, 2007). Moreover, many of those in Western Australia who failed to expiate had a previous criminal record; were already disqualified drivers; or had a number of unpaid fines, meaning they may have had little to gain by paying the fine.

It is debatable which population should be used to estimate the aggregate societal total WTP for ‘stigma avoided’. For the CBA, the appropriate population is those who will receive a criminal record under one policy but not the other as they are the group who would potentially be stigmatised. This results in a total of $7.4 million using the mean WTP. It might be argued that this may be an underestimate of the total value for two reasons. Firstly, respondents may be willing to pay that amount regardless of the risk of arrest suggesting that the total societal WTP is the mean WTP multiplied by the population of NSW. Secondly, only one value per offender has been included and there is evidence in the literature that family members also bear some stigma when a family member receives a criminal record (Chilton et al., 2004). If asked, additional family members may also be willing to pay.

In conclusion, this study has quantified one of the personal (dis)benefits of moving from one policy to another. The total social value of $7.4 million has been estimated using the assumption that only those who avoid a criminal record are the valid population for aggregation purposes. Despite the limitations of the method discussed this appears to be a valid measure to include in the CBA.

183 Chapter 8

Chapter 8 : Other benefits, harms and costs related to cannabis policies

In previous chapters, the resource implications on the criminal justice system of enforcing the existing cannabis laws were estimated, as were potential costs of operating and enforcing one model of legalisation and regulation of cannabis. Some personal costs and (dis)benefits such as the value of stigma from a criminal record have been estimated. Additionally, the frequency and quantity of cannabis consumed has been estimated for both the current policy and a legalised–regulated model. In this chapter, several other potential harms and benefits are addressed. In order to quantify these harms and benefits, and to obtain a monetary valuation, this chapter uses secondary data and information from recent literature reviews as well as from the previous chapters. Included in this chapter is an assessment of the potential increase in demand for health care resources related to treatment for cannabis use disorder, low birth weight newborns, accidents and mental health. Additionally, the value of the wellbeing from using cannabis and the potential negative impact on educational attainment is estimated.

8.1 Health effects and resource implications

8.1.1 Overview

This section considers the health care resource implications and associated costs as a result of morbidity and mortality related to cannabis use. A comprehensive review of mortality due to cannabis for the Global Burden of Disease project concluded that cannabis use was not associated with elevated total mortality in the population (Calabria et al., 2008). Evidence remains inconclusive as to the relationship between cannabis use and respiratory and oral cancers (Calabria et al., 2008; Hall, 2009), and suicide (Calabria et al., 2008). There are however, indications of an association between cannabis use and traffic accidents (Calabria et al., 2008); between cannabis use and schizophrenia (Degenhardt et al., 2000), and between cannabis use in pregnant women and low birth weight newborns (Burns et al., 2006). Those for which there is evidence of effect will be discussed further below. The projections for the legalised–regulated model are based on the increased consumption in current users and the increased number of users estimated in Chapter 4.

184 Chapter 8

The potential costs and benefits of the medical use of cannabis are not included in this work. Despite several jurisdictions having introduced laws permitting the medical use of cannabis, a search of the Cochrane Database produced only four completed reviews examining the therapeutic effects of cannabis (Phillips et al., 2001; Krishnan et al., 2007; Mills et al., 2007; Curtis et al., 2009) with one in progress (Lutge, 2005). The overarching finding of all Cochrane Reviews is a failure to confirm that there is any scientific evidence of cannabis or cannabinoids as an effective therapeutic medication. For example, the review of whether cannabinoids are effective in improving behaviour in dementia found no evidence of effectiveness and completed studies were poorly presented with insufficient data to draw any valuable conclusions (Krishnan et al., 2007). In two reviews of treatments for ataxia or tremor in people with multiple sclerosis, there was not enough evidence to suggest that any treatment including oral medications containing cannabis provided sustained improvement in ataxia or tremor (Shakespeare et al., 2003; Mills et al., 2007). A review conducted on whether THC was effective in treating Tourette’s concluded there was not sufficient evidence to support the use of cannabinoids in treating tics and obsessive compulsive behaviour in people with Tourette's syndrome (Curtis et al., 2009). A review of the use of cannabinoids for suppressing nausea and vomiting during chemotherapy in children and adolescents found that it was probably effective but produced frequent unpleasant side effects (Phillips et al., 2001).

A Cochrane Review of the effectiveness of cannabis in the treatment of wasting syndrome in patients with AIDS and cancer is underway (Lutge, 2005). The protocol for the assessment indicates that the positive and negative effects of cannabis, such as immune suppression, psychic discomfort and respiratory changes, as well as appropriate dosages will be considered.

An unquantifiable benefit of legalisation would be an improvement in ease with which research on the therapeutic effects of cannabis could be conducted. Although such research may discover therapeutic effects of cannabis in the future, based on current evidence there does not as yet appear to be any.

185 Chapter 8

8.1.2 Cannabis use disorder

Despite often being perceived as a benign drug, with a lower risk of dependence than other drugs (van Dijk, 1998; Room et al., 2008), cannabis dependence is the most common form of illicit drug dependence (Hall, 2009) and after alcohol it is the next most frequent drug for which treatment is sought (Australian Institute of Health and Welfare, 2008c). Cannabis dependence is characterised by marked distress resulting from a cluster of problems that reflect impaired control over one’s cannabis use despite harms arising from its use (Room et al., 2008). Individuals attempting to withdraw from consuming cannabis may suffer from symptoms of withdrawal including decreased appetite, irritability, anxiety, insomnia, and depression (Room et al., 2008; Hall, 2009).

In somewhat dated data, about 2% of the adult Australian population met the criteria for cannabis dependence and abuse in the previous year (Swift et al., 2001) with 1.5% meeting the criteria for cannabis dependence and 0.7% the criteria for cannabis abuse (Teesson et al., 2002). In the USA, a lifetime prevalence of 4% (Anthony et al., 1994) is reported, with a risk of dependence of between 7% to 9% in those who have ever used (Anthony et al., 1994; Perkonigg et al., 2008), and between one in two and one in three among daily users (Hall and Pacula, 2003).

Analysis of the most recent Australian National Mental Health and Wellbeing Survey (2007) found 1% of the Australian population aged 16 years and older met the criteria for a cannabis use disorder (CUD) in the previous 12 months (Teesson et al.), where CUD is either cannabis abuse or cannabis dependence. The National Mental Health and Wellbeing Survey (NSMHWB) data is a stratified sample of persons aged 16–85 years living in private dwellings in Australia. The overall response rate was 60% and there were 8,841 fully responding participants. CUD was assessed using a modified version of the World Mental Health Survey Initiative version of the Composite International Diagnostic Interview (WMH-CIDI 3.0). CUDs were assessed in those who had used cannabis more than five times in the past 12 months. Population rates of CUD vary from 0.2% among those 45 years and older through to 2.8% for those aged 16 – 24.

In order to estimate the additional numbers with CUD, it was first necessary to obtain the current number for NSW. This was done by multiplying the age and gender population rates of CUD for Australia (Teesson et al.) by the NSW population (Column 186 Chapter 8

2 in Table 60). Standard errors were used to generate 95% confidence intervals. The data for these calculations are found in Appendix-Chapter 8. A rate of CUD (Column 4) was then estimated among those in NSW who used cannabis on more than five occasions in the past 12 months (Column 3). This rate was then applied to the total number of cannabis users who consumed cannabis on more than five occasions in one year from Chapter 4. The results of this final step are in the last two columns of Table 60.

Table 60: Projected increase in cannabis use disorder Use more Rates of 4 Projected CUD Projected CUD Table CUD . than 5 CUD if use Method 1 Method 2 occasions > 5 times N N % N N

14-19 10,579 44,745 27.7% 12,579 22,755 20-29 19,064 98,733 25.8% 20,899 28,873 30-39 13,740 67,695 27.3% 14,455 17,965 40+ 9,315 83,955 13.7% 9,924 14,585 Total 295,128 57,857 84,177 52,698 95% CL 28,139 – 87,540 41,618 – 126,674

Additional CUD 5,158 31,479

95% CL 2,704-7,607 16,183- 46,740

The current number with CUD is estimated at 52,698 persons. The projected increase for Method 1, the more conservative estimate of increased consumption (Chapter 4), is an additional 5,158 (95% CI 2,704-7,607) people with CUD, while Method 2 results in 31,479 (95% CI 16,183- 46,740) additional people with CUD.

8.1.3 Treatment for dependence

The demand for treatment of CUD has been increasing internationally and while diversion from the criminal justice system to treatment may account for some of this increase, this pattern exists even in the Netherlands where cannabis use has been de facto decriminalised since 1976 (Room et al., 2008). Data from the Australian National Minimum Data Set, a nationally standardised set of data collected from drug treatment agencies (Figure 12), demonstrates that the number of episodes of treatment for cannabis has increased from 23,800 to 31,800 over the period of seven years. However, this 33% increase is overshadowed by the 56% increase in those seeking treatment for

4 As the age groups were not identical in NMHWS survey, averages based on proportions were used to construct the rates (see Appendix-Chapter 8) 187 Chapter 8

alcohol. This would suggest court-mandated treatment is not likely to be the driver for any increase in treatment for cannabis in NSW.

Figure 12: Australian trends in treatment by principal drug of concern

70,000

60,000

50,000

40,000 Alcohol 30,000 Amphetamines 20,000 Cannabis 10,000 Heroin

Source: (Australian Institute of Health and Welfare, 2009) (Table 4.2)

The current annual costs of providing treatment for cannabis use disorders in NSW have been estimated at $6,777,688 AUD 2007 (Ngui and Shanahan, 2010). These results are part of a separate study which was conducted to examine the health consequences and health care costs from the use and misuse of cannabis in NSW. There were no comprehensive data on health care resource use related to cannabis in Australia. Detailed methods and results are fully described in Ngui and Shanahan (2010). Health consequences and costs presented for the status quo used in this chapter are sourced from this monograph; estimates for the legalised–regulated model were estimated separately for this thesis.

Ngui and Shanahan (2010) used a number of sources of data, and estimated costs of treatment provided in drug and alcohol clinics, cannabis clinics, residential rehabilitation, withdrawal management, general practitioner consultations and referrals (see Table 61). Not included were privately provided counselling services and treatment provided through the Magistrates Early Referral into Treatment (MERIT) program. MERIT costs are included in the CBA under CJS (Chapter 5). Privately provided counselling was not included as no data on the private sector were available. A limitation of the Ngui and Shanahan (2010) estimates are the lack of data on the

188 Chapter 8

unique individuals receiving treatment for cannabis use disorders which raises challenges for estimating costs with an increase in the number of people with CUD.

Table 61: Costs of treatment for cannabis use disorders in NSW 2007 Episodes/ Cost per Cannabis treatment Cost (2007 $) separations occasion Residential rehabilitation (episodes) 431 $2,898,684 $6,725 Hospital (separations) 902 $1,307,610 $1,450 Withdrawal management (episodes) 1,127 $1,083,124 $961 Counselling (episodes) 2,451 $1,072,308 $437 GP (consultations) 3,018 $217,170 $72 Assessment only (episodes) 1,727 $163,674 $95 Information and education only (episodes) 113 $35,098 $310 Total costs $6,777,668 Source: (Ngui and Shanahan, Table 26)

As neither the number of persons receiving treatment, nor the rate of treatment among those with CUD was available, the rate of episodes of each type of treatment relative to CUD was used in the legalised–regulated model. For example, the rates of residential rehabilitation were calculated by dividing the number of residential rehabilitation episodes (Table 61) into the estimate of those with CUD (Table 60) 431/52,669 = 0.8%. This relies on the assumption that the association between CUD and treatment, and types of treatment remain constant, and that any increased demand for treatment would be met. To estimate the additional treatment episodes, these rates were multiplied by the additional number with CUD (Table 60). These episodes were then multiplied by the average costs for each type of treatment from Table 61. Again estimates were calculated for both Methods 1 and 2 and the 95% CI around the number with CUD.

The results of these calculations suggest that the additional costs for treatment are $663,331 with 95% CI 328,570-924,228 for Method 1 and $4,048,047 with 95% CI $1,966,291-$5,679,046 for Method 2 (see Table 62).

189 Chapter 8

Table 62: Additional episodes and costs of treatment: two estimates for CUD Additional episodes Additional costs

Method Method Method Method Rates 1 2 1 2 Residential rehabilitation 0.8% 42 257 $283,706 $1,731,347 (episodes) Hospital (separations) 1.7% 88 539 $128,019 $781,248 Withdrawal management 2.1% 110 673 $106,010 $646,937 (detoxification) (episodes) Counselling (episodes) 4.7% 240 1,464 $104,839 $639,793 GP (consultations) 5.7% 295 1,803 $21,269 $129,797 Assessment only (episodes) 3.3% 169 1,032 $16,059 $98,001 Information and education only 0.2% 11 67 $3,429 $20,924 (episodes) Total $663,331 $4,048,047

$328,570- $1,966,291- 95% CI $924,228 $5,679,046

8.1.4 Mental health

One of the most widely debated potential harms from cannabis use is that of schizophrenia and psychosis. Three hypotheses about the relationship between cannabis and psychosis have been put forward: i) heavy cannabis use causes cannabis psychosis; ii) cannabis may precipitate an episode of schizophrenia and iii) cannabis use may exacerbate schizophrenia (Hall and Degenhardt, 2004). In reviewing the literature, Hall and Degenhardt (2004) conclude that the evidence for cannabis use ‘causing’ cannabis psychosis is weak, but that the evidence for cannabis precipitating schizophrenia in vulnerable individuals is increasing. This appears to be confirmed by others (Fergusson et al., 2005; Moore et al., 2007; Cohen et al., 2008; Ferguson et al., 2008; McLaren et al., 2010). Others argue against this by pointing out biases in selection, in unmeasured confounding, in publication bias, and the fact that inferences cannot be made from observational data (Werb et al., 2010). The point is also made that the population-level rates of psychosis are not correlated with the increased rates of cannabis use in recent decades (Werb et al., 2010).

Despite this ongoing debate the weight of evidence appears to suggest some relationship between cannabis use and schizophrenia and/or psychotic disorders. Using data from the most recent systematic review, Ngui and Shanahan (2010) estimated the current costs in NSW of mental health care for those with schizophrenia and/or psychotic disorders who are using cannabis. Those people who use cannabis daily appear to have a 109% higher risk of developing psychotic symptoms compared to those who have

190 Chapter 8

never used (odds ratio (OR) of 2.09, 95% confidence interval (CI): 1.54, 2.84), while those who had ever used have a 41% higher risk (OR: 1.41, 1.20, 1.65) than those who have never used (Moore et al., 2007).

A brief description of the methods used by Ngui and Shanahan is provided below (additional detail in (Ngui and Shanahan, 2010). The number of persons who had used cannabis, by frequency of use, and who were treated for and/or diagnosed with schizophrenia/psychotic disorders in the last 12 months in NSW, was obtained from the NDSHS 2007. A population attributable fraction (PAF) was calculated using the odds ratios (ORs) obtained from a meta-analysis and applied to the NDSHS data (Moore et al., 2007). Two different PAF were calculated, one for the daily use and one for weekly, monthly, etc. Two respective PAFs were calculated and applied accordingly to the number of persons. The resulting number (916 cases) was then multiplied by the costs of treatment and diagnosis. The number of persons by treatment/diagnosis status was then multiplied accordingly by the different costs. Ngui and Shanahan obtained costs per case from three studies that have estimated the cost of treating schizophrenia (Access Economics, 2002; Carr et al., 2003b; Andrews and Tolkien II Team, 2006). Adjustments were made according to whether they had been diagnosed only; diagnosed and received treatment; or were receiving ongoing treatment. The average annual costs ranged from $3,139 to $18,554. The total annual costs were estimated to be $6.2 million 2007 AUD (NSW).

The next step is to estimate the number who might acquire schizophrenia or psychosis with any increase in cannabis consumption. It has already been established that the relationship is not a one-to-one relationship with some disputing any causality (Werb et al., 2010). Analyses by Hickman and colleagues (2009) provide data which were useful to address this question if one accepts that there is some causal pathway. They conducted an analysis of the rates of use of cannabis, the incidence of schizophrenia and psychosis by age and by gender using data from Wales and England. They then incorporated information on risk of schizophrenia and psychosis from a meta-analysis and estimated the number of cannabis users that must be prevented (NNP) in order to prevent one case of schizophrenia and one case of psychosis.

The data from Hickman were combined with the numbers of estimated new users in each age category from Chapter 4. Analysis was done for both heavy and light

191 Chapter 8

cannabis users, for males and females across age groups from 16 to 39. New heavy users are those who use daily or weekly, and light users are those who use monthly or less. This classification does not reflect that someone who uses monthly or less may use very heavily (similar to binge drinking) on each occasion of use. The results are found in Table 63.

Table 63: Number of new heavy and light cannabis users Age 14–19 Age 20–29 Age 30–39

Males

New heavy users 18,374 21,494 9,040 New light users 16,928 24,964 12,821 Females

New heavy users 11,559 8,656 5,147 New light users 22,207 29,272 11,824

The median NNP schizophrenia and psychosis numbers with the 25th and 75th percentiles used for sensitivity analysis (Hickman et al., 2010) were applied to the number of new users to provide estimates of the numbers of new cases of psychosis and schizophrenia by age, gender, level of use and diagnosis.

The total number of additional persons with schizophrenia is estimated at 24 (range 18 – 32) and with psychosis is 55 (range 34–66) persons. See Table 64 for details and Table 65 for a summary.

192

Table 64: Number of additional persons with schizophrenia and psychosis Males 14–19 20–29 30–39 14–19 20–29 30–39 Females 14–19 20–29 30–39 14–19 20–29 30–39 Schizophrenia Schizophrenia Heavy Use NNP New cases Heavy Use NNP New cases Median 3,130 3,331 4,496 5.87 6.45 2.01 Median 7,079 6,967 9,588 1.63 1.24 0.54 th 25 percentile 2,472 2,709 3,547 7.43 7.93 2.55 25th percentile 5,316 5,401 7,206 2.17 1.60 0.71 th 75 percentile 4,029 4,202 5,799 4.56 5.12 1.56 75th percentile 9,894 9,231 13,124 1.17 0.94 0.39

Light use NNP New cases Light use NNP New cases Median 11,701 12,431 16,888 1.45 2.01 0.76 Median 27,059 26,217 36,192 0.82 1.12 0.33 th 25 percentile 8,204 8,866 11,695 2.06 2.82 1.10 25th percentile 17,782 18,061 24,378 1.25 1.62 0.49 th 75 percentile 18,083 18,910 26,017 0.94 1.32 0.49 75th percentile 44,098 41,302 58,570 0.50 0.71 0.20 Total 7.32 8.46 2.77 Total 1.43 1.92 2.19

Psychosis Psychosis Heavy use NNP New cases Heavy Use NNP New cases Median 1,541 1,546 2,464 11.92 13.90 3.67 Median 2,154 2,654 3,199 5.37 3.26 1.61 th 25 percentile 1,256 1,272 1,982 14.63 16.90 4.56 25th percentile 1,705 2,147 2,533 6.78 4.03 2.03 th 75 percentile 1,946 1,903 3,117 9.44 11.30 2.90 75th percentile 2,742 3,369 4,097 4.22 2.57 1.26

Light use NNP New cases Light use NNP New cases Median 5,752 5,721 9,159 2.94 4.36 1.40 Median 8,019 9,872 11,876 2.77 2.97 1.00 th 25 percentile 4,121 4,148 6,486 4.11 6.02 1.98 25th percentile 5,691 7,040 8,354 3.90 4.16 1.42 th 75 percentile 8,721 8,664 14,005 1.94 2.88 0.92 75th percentile 12,225 15,090 18,408 1.82 1.94 0.64 Total 14.86 18.26 5.07 Total 8.14 6.23 2.60

Source: NNP Columns (Hickman et al., 2010)

193 Chapter 8

Table 65: Schizophrenia and psychosis: additional health care costs New cases Costs

Males Females Total Average Total

Schizophrenia – median 19 6 24 $18,554 $447,005 25th percentile 14 4 18 $332,039

75th percentile 24 8 32 $588,826

Psychosis - median 38 17 55 $7,391 $407,713 25th percentile 29 4 34 $249,630

75th percentile 48 18 66 $485,647

Total 79 $854,718

Range 52– 98 $581,669 –$1,074,473

Once the number of new cases was estimated, resource implications were assessed using the annual average costs of providing health care for those additional persons with psychosis and schizophrenia (Ngui and Shanahan, 2010). The additional health care costs were $854,718 (range $581,669 –$1,074,473).

8.1.5 Low birth weight newborns

Although early animal studies appeared to suggest that very high doses of cannabis in mice, rats and rabbits cause birth malformations, these birth defects most often occur following the use of crude marijuana extract rather than THC suggesting it was other cannabinoids, not THC, which had the teratogenic effects (Room et al., 2008). Nevertheless, recent studies have found that cannabis use during pregnancy is associated with low birth weights (Hatch and Bracken, 1986; English et al., 1997; Fergusson et al., 2002; Hurd et al., 2005; Burns et al., 2006).

Burns et al. (2006) merged data from the NSW Midwives Data Collection with five years of NSW Hospital data to assess drug use on obstetric and birth characteristics in those women with an ICD-10AM diagnosis for cannabis and other illicit drugs compared to all obstetric admissions. Twenty-nine per cent of births to women with a cannabis use ICD-10AM diagnosis were classified as low birth weight (LWB) compared to 9.6% for non-drug exposed. From these data, and using hospital cost weights Ngui and Shanahan (2010) estimated the current costs of LBW newborns as a result of cannabis use at $1.6 million. Additional details are found in Ngui and Shanahan (2010).

Again, to estimate potential additional hospital costs for LBW newborns related to an increase in cannabis use required several assumptions. While the ICD-10AM codes 194 Chapter 8

used in the Burns et al. (2006) paper do not necessarily reflect cannabis use disorders (CUD), it seems unlikely that LBW will occur as a result of casual cannabis use therefore the data used in this analysis are restricted to the estimate of additional women aged 14 to 39 (as per Burns) with CUD.

Table 66: Additional low birth weight newborns and related health care costs Current Method 1 Method 2 Women aged 14–39 with CUD 9,687 11,049 17,301 CI lower 2,739 3,151 5,344 CI Higher 35,177 28,843 43,956 LBW newborns 90 103 161 CI lower 25 29 50 CI Higher 327 268 408 Rate 0.9%

Average cost $17,893

Costs $1,610,410 $1,836,860 $2,876,365 CI lower $455,311 $523,801 S888,490 CI higher $5,848,158 $4,795,266 $7,307,690 Under Method 1 the additional hospital costs for low birth weight newborns attributable to cannabis increases by 14.8% while under Method 2, the increase is 78.6%. The estimates are also presented for the 95% CI around the CUD. An important assumption here is that the relationship between pregnancy and cannabis use remains constant. This may not be true, as those women who are currently not using cannabis because it is illegal may also be more likely to desist in using cannabis when planning to become pregnant. Any additional lifetime costs to care for LBW newborns are not included.

8.1.6 Accidents

There have been equivocal findings of the effect on cannabis on driving (Room et al., 2008). Some argue that the effects of cannabis on cognitive functioning may result in the tendency to drive more slowly and take fewer risks than alcohol-intoxicated drivers (Sewell et al., 2009). Alternatively, other studies suggest that high levels of THC have increased the risk of culpable driving (Grotenhermen et al., 2007; Calabria et al., 2008) particularly when alcohol and cannabis are used together (Fergusson and Horwood, 2001b; Laumon et al., 2005; Pacula, 2010b). A recent review of the literature concluded that cannabis has a collision-enhancing effect (Mann et al., 2008). Given that of those in NSW who had used cannabis in the past 12 months, 29% report having driven a motor vehicle under its influence (Australian Institute of Health and Welfare,

195 Chapter 8

2008a) this is an area worth exploring in the context of a policy that may result in increased number of cannabis users.

Table 67: Number of persons in accidents attributable to cannabis use, NSW, 2007 Driver* Passenger* Pedestrian* Total*

Baseline - current (from Ngui, 2010)

Fatalities 5 1 1 7 Serious injuries 62 20 7 90 Permanently injured & needing LT care 3 1 0 5 Minor injuries 236 78 28 341 Total recorded accidents 306 100 37 443 Method 1 Fatal accidents 5.8 1.2 1.2 8.2 Serious injuries 72.2 23.3 8.2 104.8 Permanently injured & needing LT care 3.5 1.2 0.0 5.8 Minor injuries 274.8 90.8 32.6 397.0 Total recorded accidents 356.3 116.4 43.1 515.8 Method 2

Fatal accidents 8.1 1.6 1.6 11.3 Serious injuries 100.2 32.3 11.3 145.4 Permanently injured & needing LT care 4.8 1.6 0.0 8.1 Minor injuries 381.3 126.0 45.2 551.0 Total recorded accidents 494.4 161.6 59.8 715.8 Source: (RTA, 2000; RTA, 2008); Biecheler et al. (2008); ABS (1990); Ngui and Shanahan 2010; Chapter 4; *totals may not sum due to rounding

The next step is to estimate the potential impact of increased use of cannabis. Crancer and Crancer (2010) analysed California’s Fatality Analysis Reporting System (FARS) and found that 5.5% of passenger vehicle fatalities in 2008 occurred when one or both drivers had positive tests for cannabis. These authors projected a cannabis use rate of 20% in California if the Proposition 19 vote for legalisation had been affirmative, and that this would have resulted in 836 persons killed in traffic fatalities annually. This is a fatality rate 3.6 times the 230 fatalities in 2008 (Crancer and Crancer, 2010). Pacula (2010b) takes a more conservative approach in estimating potential fatalities in California. She used a 58% increase in prevalence of cannabis use and the rate of increase in fatalities rising in proportion to prevalence on a base of 126 fatalities a year. Combining these data resulted in an increase in fatalities from 35 to 91 per year depending upon hypothesised price, non price and price elasticity effect (Pacula, 2010b).

196 Chapter 8

Assuming that the number of accidents increases with not only the increase in prevalence of use, but also with the frequency of use in those who use, the two methods of estimating frequency of use from Chapter 4 were utilised to estimate potential cannabis-related accidents (Table 67). Here the number of total days of consumption in each of the two methods was used. That is, the probability of a fatality given the number of days of cannabis consumption (current) was applied to the two estimates of projected changes in days of use. This was then repeated for all other injuries. With these assumptions, the increases in fatalities range from one to four, and serious injuries increase by ten to fifty in a year.

Table 68: Additional health care costs related to traffic accidents Injured & Serious Minor Fatal long term Total injuries injuries care Method 1

Driver $3,663 $190,308 $3,303 $64,275 $261,549 Passenger $733 $61,390 $1,101 $21,243 $84,467 Pedestrian $733 $21,486 $0 $7,626 $29,845 Total $5,128 $273,185 $4,404 $93,144 $375,861 Method 2

Driver $13,724 $713,060 $12,377 $240,829 $979,990 Passenger $2,745 $230,019 $4,126 $79,596 $316,486 Pedestrian $2,745 $80,507 $0 $28,573 $111,825 Total $19,214 $1,023,586 $16,502 $348,998 $1,408,300

Using average costs from Ngui and Shanahan (2010) the additional health care costs under legalisation were calculated. Results are in Table 68, where the additional costs ranging from $375,861 to $1,408,300.

Table 69: Estimates of losses related to fatal accidents. Additional Fatalities VOSL VOSLY fatalities Current 7 $44,100,000 $3,034,059

Method 1 8 0.8 $4,979,832 $342,610 Method 2 11 4.0 $24,967,316 $1,717,739 Source: (Access Economics, 2008; Hensher et al., 2009)

The additional direct health care costs are relatively low but in a CBA the economic value of a life must also be considered. Environmental and food safety economics often use the value of a statistical life (VOSL) in assessing environmental and regulatory policies (Johnson and Adamowicz, 2010). The US Environmental Protection Agency uses a mean willingness to pay of $7000 to avoid a 0.001 risk of death, with a resulting

197 Chapter 8

VOSL of $7 million to be used in US regulatory CBAs (Johnson and Adamowicz, 2010). A review conducted by Access Economics located 17 Australian studies which quantified a VOSL (Access Economics., 2008). The mean of these 17 VOSL was $5.7 million (range $.9 to $28.4 million 2006 AUD) with a mean value of a statistical life year (VOSLY) of A$433,437 and a median of A$119,589. Hensher and colleagues (2009) used discrete choice and WTP methods to estimate a VOSL of $6.3 million (2007 AUD) in a study examining injury and mortality avoidance measures in traffic.

If Hensher’s values for a VOSL are applied to the fatalities, the additional loss in Method 1 is $4,979,832, and $24,967,316 under Method 2. However, as this study is quantifying only annual costs and benefits, it is more appropriate to use the VOSLY of $433,437 which results in an additional loss of $342,610 to $1,717,739. This highlights one of the drawbacks of only using annual costs in a CBA. Using a life year value does not capture the loss of the remainder of the life which would have been, and also does not capture value of previous lives lost.

8.1.7 Respiratory cancers and other lung disease

The current epidemiologic research on whether cannabis causes lung cancer has mixed findings (Hall, 2009). A systematic review of the association between cannabis use and lung cancer found that while there were a number of reasons to suspect that cannabis smoke would cause lung cancer, the evidence was mixed (Mehra et al., 2006). Elsewhere it is suggested that there may be a dose-related effect of cannabis use on lung or oral cancers (Aldington et al., 2008; Room et al., 2008).

Also of concern is approximately two-thirds of Australian cannabis users report mixing tobacco with their cannabis (Australian Institute of Health and Welfare, 2008a) as the tobacco ensures it burns smoothly (Patton et al., 2005). In recent years when most sales of cannabis are of bud material with only a little leaf material, tobacco is often used to extend the amount of material in the joint. Analysis of data from a Victorian longitudinal study found that weekly or greater cannabis use during teenage years predicted initiation of, and dependence on tobacco in early adult years (Patton et al., 2005). Unlike other studies, (Amos et al., 2004), the Victorian study did not find that cannabis use predicted lower rates of cessation of tobacco but the study acknowledges that might change as the sample ages (Patton et al., 2005). If legalisation of cannabis

198 Chapter 8

were to increase the frequency of use and the average length of cannabis using careers as well as the number who consume cannabis, the increase in harms may come not from cannabis per se but rather from the tobacco consumed with cannabis.

In the 2001 and 2004 NDSHS data, only 5.5% of those who reported smoking cannabis in the past week had never smoked tobacco and 58.9% were current daily tobacco smokers (Williams and Skeels, 2006a). Although it is not known with certainty to what extent new or returning cannabis smokers would choose to mix tobacco with their cannabis or the extent to which mixing cannabis with tobacco results in a tobacco dependency, given the well documented harms of tobacco it is worth considering the potential impact.

Table 70: Estimates of new tobacco smokers under legalisation % Current Legalisation Source Total cannabis consumers 438,501 700,345 Chapter 4

(Clements and Zhao, Consume tobacco & cannabis 56.3% 246,876 394,294 2009) (Clements and Zhao, Tobacco only 23.7% 103,925 165,982 2009) Excess tobacco smokers 32.6% 142,951 228,312 Calculated among cannabis users Potentially new tobacco 85,361 smokers

A variety of data sources were used to estimate both costs and (dis)benefits but as the data are not directly comparable with other methods these results are only included in the CBA sensitivity analyses. The conditional probabilities of cannabis smokers who also smoke tobacco was 56.3% in pooled 1995, 1998, and 2001 NDSHS data (Clements and Zhao, 2009) with a population rate of tobacco smoking of 23.7%. The excess (above the population rate) of tobacco smokers among cannabis smokers is estimated at 32.6%. If these probabilities are applied to the number of current and projected cannabis users, an increase of 83,361 additional tobacco smokers may result (see Table 70).

Information from Collins and Lapsley (2008) on health care costs, costs related to fires from smoking and number of deaths per year from smoking were converted to per smoker values (see Table 71). All dollar values were converted to 2006/07 values. Again the VOSLY of $433,437 used (Access Economics., 2008).

199 Chapter 8

Table 71: Potential costs related to additional tobacco smokers Tobacco Costs Source

Annual health care costs related to (Collins and Lapsley, $337,960,410 smoking (Australia) 2008)Table 32, Annual fires and property damage (Collins and Lapsley, $161,868,601 related to smoking (Australia) 2008)Table 32, Total costs (Australia) $499,829,010

Deaths due to tobacco (annual) 14,901 Table 22 Value of statistical life year lost (2006) $433,437 (Access Economics., 2008) (Australian Institute of Health Number of tobacco smokers (2004) 3,129,800 and Welfare, 2005) Annual health and fire costs per smoker (total costs/smokers) $160 Estimated Rate of deaths (annual) per smoker 0.48% Estimated Additional smokers 85,361 Table 70 Additional deaths 406 .48% * 85,361 Additional (dis)benefits $176,150,672 Deaths * VOSLY Additional health and fire costs $13,632,150 Smokers times $160

The potential additional costs for health care and fires equate to $13.6 million and the additional (dis)benefit is a loss of $176.1 million.

8.2 Cognitive impairment and impact of cannabis use on educational attainment and earnings of youth

Cannabis acutely impairs cognitive performance (Hall and Pacula, 2003) but whether there is lasting cognitive impairment of the still-developing brain when cannabis use begins in adolescent years remains open to debate. As cannabis legalisation would only apply to those aged 21 and over, any negative consequences and long term sequelae to those under this age can be considered a spill over effect (as outlined in Table 5 in Chapter 3).

Attempting to identify the quantum of any negative effect of long-term use is a challenge as many studies exploring these issues have not controlled adequately for pre- existing characteristics (Hall and Pacula, 2003) or not used sufficiently subtle measures of cognition (Solowij, 1998 in (Room et al., 2008). Although there appears to be evidence of short-term memory deficits and impaired attention spans in those with long- term use (greater than 10 years) debate continues as to whether these deficits are related to drug effects, abstinence effects, changes in the brain or cumulative THC exposure (Room et al., 2008). Even more unclear are the related societal costs and (dis)benefits.

200 Chapter 8

The issue of the impact of heavy cannabis use on the still developing adolescent brain and any subsequent negative impact on educational attainment also remains contentious although there is some evidence that early initiation (in adolescence) is correlated with heavier and longer period of use (Pudney, 2001). Although population surveys often find an association between cannabis use and poor educational attainment and early school leaving, the issue of causality has not yet been resolved (Pacula et al., 2003b; Room et al., 2008; McCaffrey et al., 2010). There are several hypotheses which have been explored and these include: i) cannabis is a contributory cause of poor school performance and/or early school leaving; ii) cannabis use is a consequence of poor educational attainment; and iii) cannabis use and poor educational attainment are the result of common factors. Identifying what, if any, these key common factor/s might be has proven difficult as adolescence is normally a period of change and development (McCaffrey et al., 2010).

Three recently published papers (van Ours and Willams, 2009; Horwood et al., 2010; McCaffrey et al., 2010) epitomise the challenges and debate in this area. One study, using cross-sectional Australian NDSHS data, began with the premise that cannabis use in high school reduces educational attainment and set out to explore gender and age effects of initiation of cannabis on educational attainment (van Ours and Willams, 2009). Van Ours and Williams (2009) found that boys who start using cannabis at age 13 reduce their educational attainment by 1.1 years; and girls by 1.9 years, while starting cannabis use at age 15 leads to a decrease of educational attainment of 0.8 years for boys and 1.3 years for girls. Their interpretation is that cannabis caused this diminished educational attainment. However, cross-sectional studies often have limitations as a basis for inferring causal relations (Macleod et al., 2004) as pre-existing behaviours cannot be accounted for.

The second study, a meta-analysis of longitudinal data from three Australasian cohort studies (one each from New Zealand, Victoria, and Queensland) was conducted in response to criticisms in the literature that previous cohort study results may be a function of the sample and methods (Horwood et al., 2010). The results from the meta- analysis of data from three longitudinal studies suggest that early use of cannabis accounts for 17% of the overall rate of failing to complete high school, 5% of the overall rate of failure to attend university and 3% of the overall rate to complete a university degree compared to those who do not initiate cannabis use (Horwood et al., 201 Chapter 8

2010). These authors, while making the point that caution still needs to be applied when asserting a causal relationship, argue that these data support the interpretation that cannabis use causes a negative impact on educational attainment.

The third study, by McCaffrey and colleagues (2010), reviewed the existing evidence and methods and concluded the existing literature did not support a causal relationship between cannabis use and psychosocial harms (including educational attainment). They questioned previous methods of accounting for omitted variables and the use of instrumental variables (IV) to account for endogeneity. McCaffrey and colleagues (2010) set out to explore whether there was a causal relationship between heavy and persistent cannabis use and school dropout rates through the use of a longitudinal data set with multiple data collection points and propensity score methods to account for differences in characteristics of non-users and users prior to uptake of cannabis (McCaffrey et al., 2010). While this paper found that there was a relationship between cannabis use and early school leaving, after they included a number of demographics, family characteristics, and pre-existing differences between users and non-users the odds ratio decreased from 5.595 to 2.406. The inclusion of levels of academic achievement prior to cannabis use into the model further decreased the odds ratio to 1.883. The further addition of a variable for smoking cigarettes in Years 8, 9 and 10, with or without the prior academic achievements, decreased the odds ratio to 1.2 which is statistically insignificant. The impact of adding the tobacco smoking variable may indicate that: 1) cannabis use causes tobacco smoking and the introduction of smoking introduces a negative bias; 2) students who smoke cannabis and tobacco are different from those who smoke cannabis only; 3) cannabis use is influenced by other factors and there is an omitted variable bias (McCaffrey et al., 2010). The authors argue that the first is not feasible and further test the other two explanations. They find that peer associations might be correlated with some omitted variable that is correlated with baseline differences between cannabis users and non-cannabis users. They conclude that the findings do not support a causal model in which cannabis use results in dropout through cognitive impairment.

Although the findings of these three studies are not dissimilar the interpretation and implications of the findings are different. All studies found that use of cannabis, particularly heavy cannabis use in high school was associated with poorer educational attainment. The challenge remains as to how to interpret and use these findings in the 202 Chapter 8

context of a potential increase in new users. If McCaffery et al. (2010) are correct and the dropout rate is not due to the impact on cognition but rather some other construct yet to be identified then the impact on educational attainment of an increase in number of users may be negligible. Alternatively, if van Ours and Williams (2009) and Horwood, Ferguson et al. (2010) are correct the sum of the losses to individuals may be substantial if cannabis use is initiated in adolescence.

In order to provide an estimate for the CBA, the numbers from van Ours and Williams (2009) are used to estimate the potential maximum negative impact. Multiplying the 9.4% of males and 8.2% of females who start using cannabis before the age of 15 (van Ours and Willams, 2009) by the estimate of new cannabis users from Chapter 4 results in an estimate of an additional 22,386 persons who initiate cannabis by age 15 (see Table 72). This is then multiplied by the average years of education lost for males and females, again from van Ours and Williams (2009). The average return for an additional year of schooling of 10% (range 8% to 12%) of average earnings (Leigh and Ryan, 2008; van Ours and Willams, 2009) was multiplied by the average annual earnings which is in turn multiplied by the average years of education lost i.e. [( 9.4% * 148,801) * 0.8) * (10% * average annual wage)]. As the work in this thesis is assuming a state of equilibrium has been reached, and all measures are for only one year, only annual earnings were used.

Table 72: Estimates of the value of a loss of educational attainment*

% of Avg. Annual users Additional New initiate years Total wage lost Total wages start by users by 15 education years lost lost age 15 lost 10% Males 9.4% 148,801 13,987 0.8 11,190 $5,249 $58,734,248 Females 8.2% 102,420 8,398 1.3 10,918 $5,249 $66,992,449 Total 251,221 22,386 22,108 $125,726,697 *totals may not sum due to rounding

As a sensitivity analysis, both the rate of return and number of years of educational attainment lost were varied. If the McCaffrey et al. (2010) interpretation is correct there would be no lost wages. If the loss in educational attainment is 0.2 years (Pacula et al., 2003b) the value of the educational attainment loss as a result of initiation of cannabis range is $24.99 million. The results from Harwood et al. (2009) were not used as their findings were not reported in a compatible manner. When the return rate of additional

203 Chapter 8

years of education varies from 8% to 12% (Leigh and Ryan, 2008), the results are in a range from $100.6 million to $150 million per year.

8.3 Utility gained from cannabis use

In consumption theory, the consumer chooses a vector of goods and services, given her or his income, the prices and her individual preferences. Typically cost benefit analyses in the illicit drug and alcohol fields have not recognised any positive utility from the purchase and consumption use of tobacco, alcohol or illicit drugs (Zarkin et al., 2000; French et al., 2002b) but others argue it is necessary to recognise positive utility otherwise the intervention is overvalued (Weimer et al., 2009; Vining and Weimer, 2010) or in this case the move to the legalisation of cannabis would be undervalued.

Moderate alcohol consumption is seen as socially and legally acceptable and enjoyable but moderate cannabis consumption given its illegal status has not been perceived in that light. Despite this there are currently an estimated 129 to 190 million people worldwide consuming cannabis (Wood et al., 2010). Use of cannabis for pleasure, medical, and spiritual purposes is not a new phenomenon. For example, cannabis sativa has been found in burial pits of the Han Yangling Mausoleum built more than 2000 years ago (Yang et al., 2009); its use has been documented in Central Europe in 5000 BC; and medicinal use of cannabis dates back to at least 2350 BC in Central Asia and over 2000 years in Sub-Saharan Africa (Russo, 2007). Translations of ancient historical documents provide descriptions of the use of cannabis for both pleasure and ceremony in Asia and the Middle East (Russo, 2007).

Much has been written in the literature on the harms and negative consequences of cannabis use and several sections of this thesis have addressed these, including cannabis use disorders (CUD), cannabis use among adolescents, educational attainment and potential negative impact on the adolescent brain, and the potential for psychosis among those who choose to consume cannabis. These are obviously important issues but many cannabis users do not suffer from any of these negative consequences, most will not acquire a CUD, and evidence to date suggests most will stop consuming or decrease their consumption of cannabis when they get married, have families, or the demands of their jobs increase. This behaviour change may be due to developing an awareness of

204 Chapter 8

risks to employment status because of its illegal status, or be due to family commitments or lifestyle changes or just the process of maturation.

Cannabis users when asked why they consume cannabis provide a number of reasons such as it helps them to relax, to get intoxicated, to socialise, and it enhances some activities, lessens boredom, and aids in sleep (Boys et al., 2001). Thus to the consumer, the consumption of this drug provides some utility, such that they are willing to pay for its use although some, particularly adolescents, may be myopic in assessing their own risks (Vining and Weimer, 2010).

The challenge is how best to estimate the value of social consumption while acknowledging some consumption is myopic and fails to take account of future risks (Weimer et al., 2009) and some is as result of dependence. The Australian Productivity Commission addressed this issue with gambling by estimating the consumer surplus for the 97.9% of recreational gamblers, and then asserting an estimate of the loss due to the 2.1% of the problem gamblers (Australian Productivity Commission, 1999; Vining and Weimer, 2010). Another approach to estimating the offset for addiction to tobacco use was undertaken by Weimer et al. (2009) who used data from a WTP study for tobacco cessation programs and data on demand for tobacco to calculate a consumer surplus with and without addiction. The ratio of the consumer surpluses is, it is argued, the percentage loss in consumer surplus which should be included in the CBA (Weimer et al., 2009).

Graphically, it is easy to consider the consumer surplus with a demand curve for non- dependent persons (DR), with DA representing the additional demand when those who are dependent are included. In Figure 13, which builds on Figure 1 in Weimer et al.

(2009), the consumption for the non-dependent population would be QR, with a consumer surplus of PRaPC. When consumption moves to QA from QR, the additional expenditure is Pc times (QA minus QR) but the consumer value is only QRacQA. The consumer surplus for those who are dependent is equal to PRaPc - abc.

205 Chapter 8

Figure 13: Demand curve for cannabis

Price

PL DL

D PA A

PR D R Pc a b d

c e

QR QA QL Quantity

Legalisation (a demand shock) would cause an outward shift in the demand curve (DL) because the risk is lower, social acceptability is higher even while price (and supply) is

maintained constant at Pc. As above, the consumer surplus would be PLdPc minus some amount greater than abc to account for the addictive nature of the good. However, and importantly, the shape of the demand curve for cannabis is not known (Kilmer et al., 2010) for either the current status or the projected legalisation model, thus the traditional consumer surplus cannot be estimated.

However, it may be possible to proxy the additional value gained from the policy change. By first estimating the value of cannabis consumed under the status quo which

equates to Pc * QA and the additional cannabis consumed Pc * (QL – QA) under legalisation. From each of these amounts, the amount of cannabis consumed by both those who are less than 21 years of age and those with a cannabis use disorder (CUD) are subtracted. This assumes that these two groups of consumers receive no benefit from the cannabis consumed (thus possibly underestimating benefits).

Application of the above assumptions results in 41% to 50.2% of cannabis consumption attributable to those aged less than 21 years, or over 21 years of age but with a CUD (see Table 73). Using the price of $20 per gram the value of the cannabis consumed by those considered to be moderate consumers is between $579.1 million and $786.9

206 Chapter 8

million, with the difference between the status quo valued at between $66.4 and $207.7 million.

Table 73: Estimating the value of pleasurable consumption % of total Value of Cannabis N≥21 with Consumption N <21 consumption pleasurable consumption CUD excluded excluded consumption Current 14,149 7,528 21,677 42.8% $579,102,601 Method 1 14,304 8,085 22,389 41.0% $645,529,030 Method 2 28,643 10,958 39,601 50.2% $786,790,148 Additional $66,426,429 to $207,687,547 value

By using the value of quantity consumed as a proxy for consumer surplus, the true consumer surplus may be over- or underestimated but without knowing the shape of the demand curve it is not possible to know the true value. However, if the value of consumption to the consumer is ignored completely in a CBA the overall benefit of policy change would be undervalued.

8.4 Productivity

A final component in this chapter is productivity. When considering the impact on productivity as a result of a policy change, there are several considerations: the individuals’ lost opportunity to earn wages if incarcerated (Cohen, 2000) under the status quo option; the potential societal impact in terms of lost productivity if an increase of cannabis use impacts negatively on productivity; and the gain in employment opportunities in a legalised industry. Early studies on the impact of drug use on wages resulted in the counterintuitive finding that cannabis use resulted in an increase in wages, this reflects that those with higher incomes can purchase more cannabis (Williams and Skeels, 2006b). Subsequent studies explored for a dose effect. In a review of the literature Williams and Skeels (2006b) note “it is simply uncertain as to whether there are negative labour market consequences of drug use in general, and cannabis use in particular”. While there is no evidence that cannabis use does impact productivity, there would be losses in earnings to those individuals who were incarcerated for cannabis offences (Cohen, 2000), and data from Chapter 5 are used to provide an estimate of the magnitude of these personal losses. In Chapter 6, it was estimated that the legalised–regulated model would generate approximately 900 employment opportunities in the greenhouse operations, and 1500 in the retail sector. The additional benefits of adding these positions to the economy have not been 207 Chapter 8

estimated. It is uncertain what the net impact would be in a near full employment economy given many of these jobs would be transfers from the black economy.

Using the data for those who were incarcerated (N=399) and the average time sentenced for each offence type and court, and the average minimum hourly wage ($13.47 as of July, 2007), the total in lost wages is estimated at $8,737,868 for one year. If the average weekly wage is used the total is $14,485,513. As unemployment rates tend to be higher among a drug-using population and the stigma from a criminal record may have a negative impact on employment and future wages (Lott, 1992; Waldfogel, 1994; Rasmusen, 1996; Lenton et al., 1999 ; Lenton and Heale, 2000; Pager, 2003; Funk, 2004) the minimum wage rate will be used in the CBA with the upper range used in the sensitivity analysis. These estimates include only potential lost wages and do not include any valuation of the loss of personal time while incarcerated.

8.5 Discussion

This chapter has quantified a number of the harms and benefits for both the current and legalised–regulated cannabis policies. The results from this chapter will be used in Chapter 9 where the relative impact of each of the costs and benefits will be assessed. Included here are costs to the health care system (treatment for cannabis use disorder, for psychosis and schizophrenia related to cannabis use, for low birth weight newborns as consequence of maternal cannabis consumption, and motor vehicle accidents as a result of driving under the influence of cannabis); (dis)benefits to the individual (loss of educational attainment and earnings for youth); and benefits (wellbeing from cannabis use). In addition, the potential harms related to the increased consumption of tobacco through the mixing of cannabis and tobacco was discussed.

Common to all topics included in this chapter was the distillation of the published and grey literatures using review articles or meta-analyses where possible. The estimates rely on the base rates of consumption and the subsequent estimates of use under legalisation estimated in Chapter 4; if these assumptions do not hold then the estimates in this chapter also do not hold.

What was apparent in undertaking this chapter, and an associated study on examining the use of health care resources as a result of cannabis use (Ngui and Shanahan, 2010) was that there is little consensus on the direct relationship between cannabis use and

208 Chapter 8

many of the potential consequences. For example, the debate on whether cannabis causes schizophrenia is ongoing (see for example, (Hall and Pacula, 2003; Hall and Degenhardt, 2004; Kalant, 2004; Macleod et al., 2004; Cohen et al., 2008; Degenhardt et al., 2008; Hall, 2009; Hall and Degenhardt, 2010; Hickman et al., 2010; Macleod and Hickman, 2010) with no agreement about whether cannabis causes schizophrenia or those with schizophrenia self-medicate with cannabis. Rather than enter this debate, the data from two reviews were used to estimate potential costs (Macleod et al., 2004; Hickman et al., 2010). A similar approach was undertaken to examine the potential impact on educational attainment of those who start using cannabis at a young age. Here the disagreement is whether cannabis causes poor educational attainment or the two are as a result of some unknown third factor (van Ours and Willams, 2009; Horwood et al., 2010; McCaffrey et al., 2010).

Although many of the inclusions in this chapter are still subject to debate, there are other areas where a relationship between cannabis and a potential harm have been suggested but on balance the evidence appears not to suggest a relationship. This includes such areas as: depression, cardiac disease and gastrointestinal effects. Future research in these areas might prove otherwise.

There are a number of limitations particularly with the precision with which these data were estimated, and these will be further discussed in Chapter 9. While every research question requires different data, it is highly evident that in order to quantify the resource implications of different policy alternatives for government, improved methods of documenting the resource implications are needed.

209 Chapter 9

Chapter 9 : Results of the CBA

9.1 Introduction

This chapter brings together the costs and benefits of the two policy alternatives but before presenting the results the key principles of a CBA are restated. The costs and benefits are valued in terms of the claims they make on and the gains they provide to the community as a whole so the perspective is wider than that of any particular interest group (Commonwealth of Australia, 2006). CBA has its foundations in welfare economics where social welfare is comprised of the utilities of each individual within the society and individuals are the best judges of their own welfare (Drummond et al., 2005). Also key is Pareto optimality which is defined as no person should be worse off under an alternative program compared to prior to its introduction. As satisfying every individual with any new policy would be virtually impossible, the potential compensation test (PCT) has been devised. The PCT allows a determination of whether there could hypothetically be compensation from the winners to the losers rather than actual compensation occurring (Tsuchiya and Williams, 2001). If the PCT is passed then the policy is determined to be allocatively efficient.

Given social welfare is comprised of the utilities of every individual member of society, coercive policies or restrictions on the individuals choice to consume cannabis imply some loss of individual welfare compared to an alternative allowing more freedom of choice (Godfrey, 2006). From a welfarist perspective, the status quo (illegal with cautioning) would only be preferred to legalisation with regulation if the external costs associated with cannabis consumption are lower and more than compensate for the loss of welfare from the restriction of individual choice.

In conducting a CBA, (dis)benefits normally only accrue to consumers, producers, or third parties through externalities. If a policy change affects government revenues indirectly through changes in tax revenues, welfare payments, or subsidies they normally are ignored in a CBA as they are considered transfer payments (Ableson, 2000). This is because an increase in taxation revenue to government is a loss to the taxpayer and these transfers do not affect resource use or overall net economic benefit (Ableson, 2000). However, when moving from a situation where the revenue from cannabis sales will accrue to governments rather than being captured by the illicit drug

210 Chapter 9

market it could be argued that some of these revenues are a gain to society. The question then becomes, what proportion of government revenues are a net gain to society? If the assumption is all, then this assumes that none of the earnings of illicit cannabis dealers result in revenue transfers to government or gains to others (non-drug dealers). This might be the situation if all the earnings were transferred offshore. If, on the other hand, these illicit earnings are used to purchase groceries, cars, houses, etc. then one could argue that not all of the revenue to government from cannabis sales should be included on the benefit stream.

Thus there are two arguments against including all net revenues from cannabis sales on the benefit side: the principle that the benefits in a CBA are the sum of utilities, and government does not accrue utilities; and not all the illicit drug revenues are lost to society. Alternatively, moving from an illegal cannabis market to a legal market there are substantial direct revenues to government and ignoring them completely does not seem justifiable. Particularly, as although governments do not accrue utilities they do accrue votes and this directly impacts on public policy. As a result, the NSB will be calculated for the current policy (status quo) and the legalised–regulated model with and without potential net revenues included.

9.2 Methods for dealing with uncertainty

Uncertainty is introduced in any economic model. This uncertainty has to do with the choice of model; the choice of variables in the model; and the quantification of inputs and outcomes in the model. The choice in this thesis was a static equilibrium–state model. While a more dynamic model would have enabled feedback loops, assessment of long term outcomes, and lifetime costs and outcomes, the necessary data on probabilities, transition rates and outcomes were not available, nor was it feasible within this work to obtain the necessary estimates.

Data limitations include not only the uncertainties surrounding the introduction of a legalised–regulated alternative but also the considerable knowledge gaps around what currently occurs. For example, there is a lack of available data on who gets what treatment, the duration of that treatment and the outcomes of treatment are largely unknown. As a result, a decision was made to use a static model with a range of costs and benefits rather than a dynamic model with fewer variables.

211 Chapter 9

Having constructed a static model there are methods which can be used to explore the impact of some of the key assumptions in the model. The use of standard deviations, confidence intervals and micro simulation can be used to explore variability in the data. Structural uncertainties can be assessed using one-way sensitivity analysis and scenario analysis and probabilistic sensitivity analysis can be used to explore parameter uncertainty (Gray et al., 2011). However, probabilistic sensitivity analyses require knowledge of distributions and are not undertaken here.

A range is presented for each key variable in the model. Where possible the range was generated statistically and 95% confidence intervals are used (i.e. policing costs; the numbers of persons with CUD; and the increase in psychosis and schizophrenia under legalisation). In other instances the range was constructed by varying a cost variable (i.e. average wages versus minimum wage in estimating wages lost while incarcerated; or the average cost of a residential rehabilitation day, the cost weight for low birth weight newborns), or by varying the risk ratio (MVA). For health-related consequences in the legalised–regulated model the main result was generated using Method 2 (the distribution of new or returning cannabis users was the same as the existing distribution). The range around the main estimate was generated from Method 1 for the low range and a combination of Method 2 plus the high range from the status quo. Where no other information was available to generate the range an arbitrary plausible range (Gray et al., 2011), 25% plus/ minus the main estimate, was used.

9.3 Results of the CBA

Before presenting the various components and their individual impact, the net social benefit (NSB) and the benefit/cost ratio (BCR) with and without the net revenues from the retail operations are discussed. As indicated previously, all costs and benefits are in 2006/07 AUD and are for one year only.

The total costs are estimated at $80.1 million (range $54.2 to $113.4) for the status quo and $90.7 million ($53.8 to $128.9) for the legalised–regulated alternative (see Table 74). The total costs are comprised of costs to government, personal costs and compliance costs for the grower (additional detail is found in Table 75. The cost to government for the status quo is $77.6 million (range $52.4 to $110.2) and for the legalised–regulated alternative it is $55.9 million (ranges $27.9 to $85.4). This includes

212 Chapter 9

the costs to the criminal justice system, health care system, the costs of enforcing regulations, prevention programs and the costs of operating the regulatory agency. Costs to the individual are primarily the payment of fines (civil and criminal) and the cannabis licence costs with the latter being the cause of the large difference between the $2.6 million expenditure under the status quo and the $34.3 million under the legalised option.

Table 74: Summary of total costs, total benefits and net revenues for both models Status quo Legalised – regulated

Main Low High Main High Low range Summing up estimate range range estimate range (millions) (millions) millions millions (millions) (millions) Total costs $80.1 $54.2 $113.4 $90.72 $53.8 $128.8 Total benefits/ $362.7 $282.1 $513.0 $318.8 $222.4 $394.2 (dis)benefits Net government $659.4 $829.5 $196.5 revenue Net social benefit in millions AUD (median, 10th and 90th percentile) Median 10th 90th Median 10th 90th

Costs and revenues $293.5 $207.9 $379.0 $228.7 $154.2 $314.5 from retail excluded Costs and revenues $752.4 $408.7 $1077.4 from retail included Benefit cost ratio Costs and revenues 5:1 4:1 from retail excluded Costs and revenues 11:1 from retail included Retail benefits included but lost wages and cost of 1.3:1 retail operation included in costs

The total benefits are estimated at $362.7 million (range $282.1 to $513.0 million) for the status quo and $318 million (range $222 to $394 million) for the legalised–regulated alternative. This includes lost wages from incarceration, the value of stigma from a criminal record, impact of lost educational attainment, and an estimate of the value gained from the use of cannabis (again detail is found in Table 75).

The benefit cost ratios (BCR) are presented in the last three rows of Table 74. As discussed in Chapter 3, the BCR is very sensitive as to how costs are defined (Ableson, 2000; Commonwealth of Australia, 2006). For example, if payments to growers and

213 Chapter 9

costs of running the retail operation are deducted from total revenues prior to including as the benefits, the BCR for the legalised–regulated alternative is 11:1. Alternatively, if the retail costs and payments to the growers are included in the cost category rather than being removed from the revenues prior to being included in the benefit stream, the BCR becomes 1.3:1. As the NSB is the preferred method of reporting results of a CBA (Ableson, 2000; Drummond et al., 2005) and given the ambiguity around the BCR only the NSB will be discussed in the remainder of the chapter.

The median NSB and the 10th and 90th percentiles were estimated through the use of a Monto Carlo simulation, using 1000 random draws. The median NSB for the status quo is $293.5 million ($207.9 to $379.0) and $228.7 million ($154.2 to $314.5) for the legalised–regulated model. The results are presented in Figure 14. The addition of government revenues minus expenditures results in a NSB larger than the other two results. Notably the uncertainty around the results also increases. An important caveat, as discussed above, is that this assumes all revenues which go to government are considered new revenues, i.e. no portion of the revenue from the illicit cannabis market returns to government as revenue or benefits others.

Figure 14: NSB main estimates from Monte Carlo simulation $1,200

$1,000

$800

90th percentile $600 10th percentile

Millions $400 50th percentile

$200

$0 Status quo Leg-reg Leg-reg with revenue

Table 75 provides a summary of the various components with references to the original tables. It is apparent that the government expenditures on criminal justice, enforcing regulations, consumer information and health care including treatment for CUD are

214 Chapter 9

greater for the status quo than for the legalised–regulated option; however the personal costs are greater for the legalised–regulated option.

Table 75: Summary of costs and benefits* Status quo Legalised–regulated

Main Low High Main Low High Expenditures Table Millions $ Millions $ Police (including drug 32 22.14 11.66 31.88 11.50 2.30 18.40 driving testing) costs Court 29 4.84 3.63 6.06 Prosecution and Legal Aid 28 3.24 0.67 5.81 Penalties / Corrections 32 26.98 20.24 33.73 MERIT 32 3.52 2.64 4.40 2.46 1.85 3.08 Regulatory agency 47 1.04 0.69 1.39

Enforcing regulations 46 3.96 1.39 6.52 Consumer information, quit 48 12.48 5.90 18.76 campaigns and support Treatment (CUD) 61,62 6.78 6.23 7.50 10.83 7.44 12.46 Schizophrenia/ psychosis 65 6.22 4.67 8.11 7.07 5.25 9.18 Low birth weight newborns 66 1.61 0.46 5.85 2.88 0.52 7.31 Motor vehicle accidents 68 2.31 2.19 6.93 3.72 2.56 8.33 Total government costs 77.64 52.38 110.26 55.93 27.91 85.43 Fines (current) 36 1.26 0.95 1.58 Parents lost work time 36 0.06 0.04 0.07 Defence attorney 36 1.24 0.93 1.55 Personal Licence costs 37 31.12 23.34 38.90 Fines under regulatory 37 3.22 2.17 4.02 structure Total personal costs 2.56 1.92 3.20 34.34 25.51 42.92 Growers - compliance 40 0.46 0.40 0.50 costs Total costs $80.1 $54.2 $113.4 $90.72 $53.8 $128.8

Dis(benefits)

Lost wages - incarcerated p 209 -$8.74 -$6.55 -$14.49 Stigma 56 -$7.42 -$1.32 -$15.08 VOSLY (accidents) 69 -$3.03 -$2.28 -$3.79 -4.75 -3.38 -5.94 Wellbeing value 74 579.10 434.33 $723.88 645.5 484.2 786.8 Education attainment 72 -197.3 -$157.8 -$236.7 -323.0 -258.4 -386.7 Total Benefits $362.7 $282.1 $513.0 $318 $222 $394

Retail operation Payments to growers 43 -617.5 -790.2 -385.9

Operating shops 45 -82.86 -450.1 -47.74

Total revenues 50 1,360 1,437 1,263

Potential net revenue $659.5 $196.5 $829.5

*due to rounding not all totals may not sum exactly

Within the model the assumption is that 15% (range 10% to 20%) of potential revenue to government will not be collected either because of home growing or the persistence of the black market. Others have projected larger losses as a result of the black market 215 Chapter 9

particularly if attempts are made to retain prices at current levels (Kilmer et al., 2010). This would lead to greater uncertainty around the results for the legalised–regulated option.

9.4 Sensitivity analyses

While the previous results provide a range around the main estimate, it is also useful to explore the impact of individual variables on the results. Figure 15 presents the impact on the NSB when the two largest (dis)benefits (decreased educational attainment and gain in wellbeing) are varied. The results are presented in groups of three, each benefit is varied for each of the current (SQ), the legalised–regulated (LR) and legalised– regulated including revenues (LRR). For each option, education (dis)benefit is decreased to 50%, and then to zero; and wellbeing is decreased to 50% and then to zero. Finally the results are presented excluding both education and wellbeing. The costs are held constant in the simulation with only the ranges of the benefits allowed to vary. The median and the ranges are generated using Monte Carlo simulations with 1000 random draws.

Not surprisingly, the NSB increases when only half the negative value for educational attainment is included, or removed completely. When the value for the wellbeing gained from cannabis is decreased to 50% or removed completely the NSB for both the status quo and legalised–regulated alternative become negative as they do when the values for both educational attainment and wellbeing are excluded. That is, only when the value attributed to personal wellbeing from cannabis is decreased do the results indicate that neither are an efficient use of societies’ resources. In the final set of data only the legalised–regulated alternative with revenues has a positive NSB (at $500 million). This assumes that all revenues to government are new, that there is no loss in educational attainment to youth related to cannabis use, and no utility gained by those aged 21 years and older who have moderate consumption of cannabis.

216 Chapter 9

Figure 15: Examining impact on the NSB of various assumptions regarding (dis)benefits (millions $)

2,000

1,500

1,000

500 Millions

-500

90th percentile 10th percentile 50th percentile

217 Chapter 9

Next, the impact on total government expenditures of using the low and high ranges for those cost categories which made up 5% or more of the expenditures were explored. The cost of drug driving testing had the largest impact Figure 16, but as the same values are included in both alternatives (due to lack of data), they are excluded from further sensitivity analyses. For the status quo alternative, varying the penalties received (i.e. length of incarceration) by the low or high range results in a potential decrease or increase in total government expenditures of 8.6% whereas varying the costs attributed to providing health for schizophrenia and CUD, as well as policing costs have less than a 5% impact on total costs to government. In the legalised–regulated model using the low and high range for providing consumer information (advertising and QUIT Line information) would have an 11.8% impact on government expenditures. The high ranges for MVA and low birth weight newborns have an impact between 5% and 10% while the remainder are less than 5%.

Figure 16: Percentage change in government expenditures when key resources are varied

Legalised-regulated

Motor vehicle accidents Low birth weight newborns Schizophrenia CUD treatment Drug driving testing Consumer info/QUIT lines Enforcing regulations

Status quo Schizophrenia CUD treatment Drug driving testing Penalities Police

-20.0% -15.0% -10.0% -5.0% 0.0% 5.0% 10.0% 15.0%

Only the purchase–licence costs and those cost categories which accounted for more than 5% of government expenditure were assessed for their impact on the NSB. Here the main estimate was used as the base for the comparison, and only the low–high ranges for one variable at a time was allowed to vary in the simulation. This resulted in

218 Chapter 9

very narrow ranges. The median NSB and the percentage variation from the original are presented in Table 76. Only the costs pertaining to penalties (incarceration) in the status quo alternative, and the fines and provision of consumer information in the legalised–regulated alternative have more than a 5% impact on the original NSB.

Table 76: Sensitivity analysis of varying the costs Original 50th Variation from NSB percentile original NSB

$ millions $ millions % Status quo original 293.5

Penalties 315.7 6.4% Legalised– regulated original 228.7

Fines 220.7 -6.0% Consumer info/QUIT Lines 222.4 -5.3% Purchase–licence 226.3 -3.6% Low birth weight newborns 230.0 2.1% MVA 231.8 1.3% CUD 232.0 1.2% Enforcing regulations 236.3 -0.7% The final sensitivity analysis explores the potential impact of additional people becoming tobacco smokers under the legalised–regulated option. In Figure 17 the original NSB for the legalised–regulated alternative is presented alongside the results when the additional negative impact of tobacco smokers included. When the negative impact of tobacco is included the result is a considerably lower NSB. As discussed in Chapter 8 only the impact of additional smokers was estimated therefore these results were not included in the main results.

Figure 17: The impact of the additional (dis)benefits of tobacco smoking on the legalised–regulated option

350

300

250

200 90th 150 10th

Millions 100 50th

0 Tobacco Legalised-regulated -50

219 Chapter 9

9.5 Discussion

This chapter brought together the costs and benefits estimated in the previous chapters into the net social benefit which was greater than zero for both policy options. This would suggest that the PCT test has been met, that is, those who might be made worse off under the alternative program could compensate the losers. This would also suggest that under the assumptions in the main estimates, both policies are assessed to be allocatively efficient. The sensitivity analyses demonstrated under what conditions this test may not hold. For example, when the wellbeing from cannabis measure is decreased to 50% of the original estimate, or removed completely neither the status quo nor the legalised–regulated options have an NSB that is greater than zero. This holds true even when the potential negative impact of education loss is treated as zero. The addition of the revenues to government increases the NSB, making the legalised– regulated option highly attractive.

The sensitivity analyses also demonstrated that, at least within the ranges estimated for this CBA, variation of most cost components are unlikely to appreciably impact the final results. Two categories of costs, the penalties received under the status quo and the provision of consumer information and running a program such as QUIT Line, may be sufficiently large as to have an appreciable effect on the NSB if funded at the recommended level for tobacco (Cancer Council Victoria, 2010).

When considering the results of this study there are a number of limitations and challenges to consider. While many have already been alluded to in the discussions of the individual chapters, it is useful to discuss them here in the context of the results. Considering the data and assumptions used to populate the model, it needs to be acknowledged that the task of conducting this CBA required many assumptions, and required data from a multitude of sources. Conducting a CBA where one policy is highly speculative and the good involved is illegal meant that data needed to be drawn from a variety of sources such as self-report survey data; peer-reviewed and grey literatures; police, court and treatment data; government documents such as annual reports and budget papers; policy documents on tobacco and alcohol; and finally, key informant interviews. Notwithstanding the multiple sources, methods and assumptions, every attempt was made to be consistent in assumptions and, where possible to use data from Australia.

220 Chapter 9

The quantification of resource use was an ongoing challenge; both in terms of identifying the factors which impacted on the resource consumption and in determining the appropriate price. For some activities, a top-down costing (prosecution, regulatory costs) or available costs (average cost of court) method was used whereas micro costing methods were used for others (treatment and policing). For comparability, where micro costing was used, every attempt was made to include overhead and on-costs (pension, annual leave and other benefits).

Accurately measuring the quantity of cannabis consumed by a population is challenging, as a result of its illegality. While current cannabis consumption has been estimated by others (Wilkins et al., 2002; Pudney et al., 2006; Legleye et al., 2008; Kilmer and Liccardo Pacula, 2009; Clements and Zhao, 2005) and surveys have been conducted exploring respondents’ intent should cannabis become legalised (Daryal, 1999; Weatherburn and Jones, 2001; NDSHS, 2007), Chapter 4 provides a first estimate from combining these data. It was estimated that a total of 50,632 (range 34,211 to 68,422) kilograms of cannabis is currently being consumed per year in NSW. This increases to 78,940 (range 39,220 to 134,922) kilograms for the legalised–regulated option. This increase in consumption in NSW is as a result of an estimated increase of prevalence of use from the current 8.6% to a prevalence of 12.4% in those aged 14 and older. The overall increase in prevalence is similar to that estimated by Weatherburn and Jones (2003) and projected by MacCoun when price is held constant (MacCoun, 2010). The pattern of potentially larger increases in prevalence and frequency of consumption by younger groups was evident as suggested by Pacula (2010). However, the average quantity consumed (115 grams per year) per person who consumes falls within the range of other estimates (Kilmer and Liccardo Pacula, 2009).

The estimates of additional cannabis users, as well as the data on frequency and days of use were applied when estimating the additional numbers: in treatment for CUD; who may be involved in accidents as a result of cannabis use and driving under the influence; and who may develop schizophrenia or psychosis. Under the assumption of an increase in prevalence to 12.4%, there is a potential increase of 32,000 persons who may develop a CUD, 79 additional persons who would develop schizophrenia or psychosis and 160 additional low birth weight newborns, and four additional road fatalities. The respective increase in health care costs for each of these categories was: 60% for treatment for CUD, 14% for schizophrenia or psychosis; 79% for LBW newborns, and 221 Chapter 9

61% as a result of accidents. Combined, this resulted in an increase in the health care costs from $16.9 million to $24.5 million (a 45% increase).

The costs to the criminal justice system of the current system were quantified in Chapter 5. A survey of over 90 police officers, in three LACs, was conducted to quantify the police time taken to manage those accused of various cannabis offences. The cost of this time was then combined with other activities such as crop surveillance, and the detection and removal of hydroponic operations. Costs to the criminal justice system were estimated at approximately $60 million for the status quo, while the cost to government of enforcing the legalised–regulated option was estimated at $19 million. Notably neither estimate included the costs of combating gang-related activities, which based on the reported illicit sales of tobacco (Geis, 2005) would likely continue to some degree under legalisation. This chapter illustrated the importance of understanding the actual resource implications of key inputs. Total costs to the whole criminal justice system (police, courts and penalties) of enforcing the current policy, while still substantial at $60 million, were less than the costs of only the police activities estimated elsewhere at $80 to $88 million (Moore, 2007; Miron, 2005; see Chapter 5 for further discussion of this issue). A substantial portion (44%) of the CJS expenditures under the status quo is the cost of penalties (primarily incarceration) received for convictions for a cannabis offence; 75% of these costs are attributed to those with a past criminal history or concurrent offences. While each offence for which a person is convicted results in its own penalty, the prior and other criminal offences may also impact upon the penalty given by the court. Taken together with the estimate of police costs, this would suggest that previous estimates of potential savings from the CJS may be over estimates.

The resource implications of a highly regulated–legalised alternative were estimated in Chapter 6. In this chapter the regulations and their resource implications were drawn from the more developed alcohol and tobacco literature but again required assumptions regarding the resource implications of enforcing compliance and the costs of the failure to comply. The basis for each regulation was to minimise the direct harms from the cannabis legislation (i.e. criminal record, stigma, cannabis tourism) while also attempting to minimise the harms from increased cannabis use (i.e. CUD, MVA, schizophrenia and psychosis). Such a model which reflects many of the characteristics identified elsewhere as one way of legalising cannabis (Nadelmann, 1992; Haden, 2008; Rolles, 2009) results in significant opportunity costs. These included the costs of 222 Chapter 9

licensing and enforcing regulations to decrease the harms and these costs are substantial. For example, the costs for the purchase–licence program are estimated at $31 million, and the costs of providing consumer information, education and Quit Line support was estimated at $12.5 million.

A number of additional costs and benefits were estimated in Chapter 8. There are two categories in particular, where considerable uncertainty remains and their inclusion does impact on the final results. The two categories are the utility (or wellbeing) gain to the non-cannabis dependent adult who chooses to consume cannabis and that of the potential loss in education attainment in those who start consuming cannabis as an adolescent. The main limitation in estimating the wellbeing from cannabis use is that the shape and slope of the demand curve are not known (Kilmer et al., 2010), thus the consumer surplus cannot be estimated.

While some are of the view that consumption of drugs should not be valued, without some valuation of the benefit that individuals receive from consumption, a true estimate of costs and benefits cannot be determined and consumption decisions will not be properly understood (Weimer et al., 2009). As a proxy for consumer surplus the value of the cannabis consumed, minus the amount consumed by those who were less than 21 years of age or those who had a CUD, was estimated for both the status quo and the legalised–regulated options. Without evidence on the shape of the demand curve the extent of the over- or underestimation is not known, however, without some estimation of the value individuals place on consuming cannabis, the CBA would be incomplete.

The uncertainty around the potential loss of educational attainment is a different issue. Here the debate is whether there is a negative impact on the educational attainment of adolescents or not (van Ours and Willams, 2009; Horwood et al., 2010; McCaffrey et al., 2010). The main results in this study have assumed there is a negative impact (van Ours and Willams, 2009; Horwood et al., 2010) but zero impact is explored in the sensitivity analyses.

Those who advocate for legalisation of cannabis often argue that the harm, in particular stigma, that results from a criminal record due to possession/or use of a small amount of cannabis is not defensible. Contingent valuation methods were used to estimate a societal valuation of the potential stigma avoided should legalisation be introduced. In

223 Chapter 9

Chapter 5, it was demonstrated that of those adult offenders detected with a small amount of cannabis, 26% received a cannabis caution and thus no criminal record. The mean value of stigma obtained from the survey was applied to the remaining 74% who attended court for a cannabis offence. This resulted in a $7.2 million benefit from stigma avoided in moving to the legalised-regulated option. A limitation of the CV study; similar to that of other CV studies, is the uncertainty around whether the amount stated is a true representation of the individuals’ valuations.

9.5.1 Caveats and limitations

As discussed earlier in this chapter a static equilibrium-state model was employed in this thesis. A more dynamic model would have enabled feedback loops, assessment of long term outcomes, and lifetime costs and outcomes. However, as the necessary data on probabilities, transition rates and outcomes were not available a simpler but potentially broader model was used. This permitted a wide range of costs and benefits to be included.

Considerable effort was made to be inclusive in estimating the costs and benefits for the CBA. However, in delineating the boundaries of this research the impact of legalisation on the consumption of other illicit drugs and alcohol were deliberately excluded. This included the gateway effect of cannabis into other ‘harder’ drugs, and whether cannabis was potentially a substitute for or complement to alcohol. The literature has mixed findings as to whether cannabis is a complement to or a substitute for alcohol (Cameron and Williams, 2001; Williams, 2004), plus the impact likely varies by age and consumption patterns. If cannabis is a substitute, the potential impact on the alcohol industry may be significant. In 2004/05 the total Australian revenue to government for alcohol tax was $5,112.5 million (excise taxes, custom duties, GST and wine equalisation) (Collins and Lapsley, 2008). The NSW portion is approximately $1,527 million (this is taxes only). The total net revenue from cannabis (after paying growers, retail outlets costs etc.) in NSW was estimated at $659.5 million. Although not directly comparable to taxes on alcohol, the expenditure on cannabis will be of a sufficient magnitude that it may be of concern to alcohol producers. The potential impact on alcohol producers was not estimated. A substitution effect might result in a lower burden from alcohol in terms of health and policing but if cannabis is a complement to alcohol the harms may increase.

224 Chapter 9

The issue of a gateway effect arises as most illicit drug users consume cannabis even if it is not their principal drug. There are often three main explanations offered. Those who consume cannabis are more likely to use hard drugs because: 1) they have the opportunity to obtain them through their illicit dealer; 2) the use of cannabis increases the propensity to use other drugs; and 3) those who use cannabis have a tendency to use other drugs for a third unknown reason (Room et al., 2008). Room and colleagues (2008) point out that opportunity does play a role in other drug use, but that social environment, peer groups and delinquent behaviours do not explain all the relationships between cannabis and other drug use. The question remains unanswered as to whether increasing the rate of cannabis use would lead to an increase in the use of heroin, cocaine and methamphetamines.

Reflecting back to Chapter 3, there were several other categories of costs and benefits that were identified as not being included. The direction of impact on the results of at least two of these categories, the change in search time and impact on workplace productivity are uncertain. Regarding productivity, there does not appear to be any evidence on the impact of the use of cannabis on productivity in the workplace. In the legalised–regulated model the search time necessary to purchase would become known to the individual, but as it is proposed, with only a limited number of shops where cannabis may be purchased, the travel and purchase time may actually increase. It is quite likely that additional information on potency, and assurances that the product is free from contaminants at point of purchase would, all else equal, be a benefit.

The current prevalence rates and the potential increase in use under legalisation depend on the NDSHS data. Thus it relies on the validity of the survey; the representativeness of population who are recruited; the recall of the respondents; and their honesty. Population surveys often miss hard to reach populations, however well conducted self- report surveys have been found to be a reliable means of eliciting information about drug use, particularly for cannabis (Katz et al., 1997; Mc Elrath, Dunham and Cromwell, 1995 in Weatherburn, 2001).

Evidence from the de facto legalisation of cannabis use in the Netherlands would suggest that any decision to legalise cannabis would result in considerable pressure from the International Narcotics Control Board. In addition, political pressure against legalisation from other countries such as the United States would be expected (van Dijk,

225 Chapter 9

1998). There would also likely be substantial domestic political pressures. One only needs to examine the political and community debate around the introduction of the Medically Supervised Injecting Centre in Sydney and the proposed Heroin Treatment Trial in the ACT to comprehend the potential debate that would likely occur (Lawrence et al., 2000; van Beek, 2004; Drug Free Australia, 2010).

The potential impact on families is again uncertain. If prevalence and consumption of cannabis both increase under a legalised–regulated alternative, one might expect that distress to families may increase. However, removal of the criminal status may lessen the negative impact. In 2006, a new law was introduced in NSW allowing authorities to remove children from homes where cannabis was being grown (Tebutt, 2006). In part this was because of the risks of dangerous chemicals, insecticides, fire, fungal growth and fumes. A benefit of legalisation would hopefully be the end to such dangerous growing practices.

While this research focused only on NSW, the introduction of a policy such as legalisation in one jurisdiction would create its own set of problems. These were not dealt with directly. The rationale for including only one state was to simplify the analyses and to include comparable resources, consumption patterns, treatment and diversion programs, policing activities and benefits. It was not feasible to include all jurisdictions.

The most significant assumption was holding the retail price constant. This assumption may appear improbable for two reasons—firstly, a government may not want to impose a price on cannabis either directly or through taxes that is so much above the cost of growing and distribution. This is a matter of policy and political will, and there are substantive public health reasons to support such an outcome. The second factor—the extent of the black market is not predictable given the available information, although data from tobacco is useful. Others in the context of the United States have argued that it will not be possible to maintain current prices (Kilmer et al., 2010) as it would require anti-competitive actions by government but there are precedents for government- controlled monopolies in alcohol, and utilities elsewhere (Babor, 2003). Traditional arguments of inefficiencies of government monopolies, while potentially true are unlikely to be a serious issue in the model developed here as there are substantial gaps between the production costs and retail prices.

226 Chapter 9

9.5.2 Conclusion

This chapter has brought together the estimates of the multiple and often contradictory costs and benefits of two cannabis policy options. It appears that there is little difference between the NSB for the status quo in NSW and the NSB for a legalised– regulated alternative if the potential transfers of revenue to government are excluded. These results suggest that policy alternatives are similar in their efficiency in use of society’s resources, notwithstanding all the caveats outlined above. As one objective of a CBA is to value all the costs and benefits with the broad purpose of assisting with social decision-making (Boardman et al., 2001) these results leave the decision-maker without a clear decision. Considering only these results, and not the potential government revenues, there are two ways forward—maintain the status quo as there is no real evidence that change is better; or legalise and regulate cannabis as there is no real reason not to. Both of these arguments can be made, and it may fall to politics and ideology to drive any final decision.

It will be tempting to turn to the potential revenues and to use them as an argument for legalisation. But it must be reiterated that there is considerable uncertainty around these potential revenues, largely because of the uncertainty as to what would actually happen to the black market (Kilmer et al., 2010) and the ability to maintain the price at the current street price. Finally, the legalised–regulated alternative was built on the premise of a government monopoly, price control at current prices, licensed consumers, contracted suppliers, and considerable other regulations. Without these, the costs and benefits might appear very different as the market may be oversupplied, prices may fall and demand may increase even further.

Many have made the point that additional research is required as a prerequisite to the rational consideration of public policies (Strang et al., 2000; MacCoun and Reuter, 2001b; Hall and Pacula, 2003; Hall and Lynsky, 2009). The results of this CBA of two cannabis policies, one of which is a modelled legalised option, will start to redress some of the evidence gaps that arise when making public policy in this area. Moreover, by starting to sum across the various domains the trade-offs become increasingly apparent.

227 Chapter 10

Chapter 10 : A discrete choice experiment: policy options for cannabis

10.1 Introduction

In the previous chapters, costs and benefits of two legislative policy options for managing cannabis in NSW were estimated and then summed. The conversion of resource use and benefits into a common monetary metric allowed the summation of the relative costs and benefits and an assessment of which policy is the most efficient from a societal perspective. While such a method appears to be transparent, and can be useful when there are multiple and conflicting outcomes of a policy (Elvik, 2001; Vining and Weimer, 2010) it relies on the analyst’s ability to quantify those benefits and inputs thought to be important.

The policies and their potential outcomes explored in this thesis are complex and contentious. The complexity of the policies and the often conflicting outcomes can expose such methods to the criticism of having overlooked significant factors. It can be argued that assessing their value to society involves, alongside those factors already included, consideration of liberty, human rights and justice, none of which are easily quantified in monetary terms. The discussion in the previous chapter underscored the point that it is unlikely all the costs and benefits can ever be included. Nor, as evidenced by the discussion of the valuation of the potential benefit from the use of cannabis, will there likely be agreement on how potential harms and benefits should be valued. Thus even a carefully constructed CBA may not fully represent society’s true preferences.

Failure to identify or attach a value to important utility-bearing inputs or benefits in any CBA may result in: 1) the overall conclusion being correct, that is the relative net social benefit does not change, but the magnitude of the difference between the two policies is incorrect; 2) the incorrect option is determined to be the most efficient choice; or 3) no perceptible problem, as the factors omitted are either insubstantial or they cancel each other out.

The public hold disparate views on the most appropriate status for cannabis. What is not apparent is which characteristics of the various policy options are important particularly as many are unaware of the legal status of cannabis in their jurisdiction

228 Chapter 10

(Pacula et al., 2004; Chanteloup et al., 2005; MacCoun et al., 2009). In a review of Australian surveys that asked respondents for their opinion on legalisation and decriminalisation of cannabis, Mathews-Simmons and Ritter (2009) report that support for such policies appears to have peaked in 1998 and has subsequently declined. However, they also note that there are increasing numbers of respondents who have no firm opinions, and notably when more information is provided in the survey questions about specific policies, respondents are more likely to express their support for those policies (Matthew-Simmons et al., 2008).

Research suggests that there is at least some connection between public opinion and public policy decisions particularly when the issue has salience (Monroe, 1998; Burstein, 2003; Matthew-Simmons et al., 2008). It is this connection which makes it worthwhile to examine societal preferences for cannabis policies, and assess whether the general public hold views that are consistent with current policy and what characteristics or consequences of the policy are important. Understanding potentially acceptable trade-offs will assist in framing future debate.

Discrete choice experiments (DCEs) are another economic tool which can begin to explore which factors of cannabis policy are important to society, as well as be used to compare and contrast with the results of the CBA from the previous chapter. DCEs rely on welfare theory, with social welfare comprised of the welfare (or utilities) of each individual and individuals assumed to be the best judge of their own welfare (Drummond et al., 2005). DCEs further draw on Lancaster’s economic theory of value (1966, 1971) which assumes that individuals derive utility not from a good itself but rather from the attributes of that good. This attribute-based methodology permits the inclusion of attributes from a number of sectors (i.e. health, consumption, criminal justice).

The use of DCEs is one way of valuing individuals’ preferences for goods or services not normally traded in a market (Louviere et al., 2000; Amaya-Amaya et al., 2008). The results can be decomposed into their component parts to examine the impact of the various attributes on the results, as well as the relationship between various socio- demographic characteristics. The different attributes can be examined to explore for preference heterogeneity.

229 Chapter 10

A recent comprehensive review of DCE papers in health economics, has identified that DCEs have been used for a range of purposes (De Bekker-Grob et al., 2010 ) with the largest category of papers evaluating patient experiences (35%), while 33% of the papers reviewed explored trade-offs between health outcomes and patients’ experiences. This review suggests fewer papers (5%) explored priority setting or policy setting preferences across a general population. Most papers in health have elicited preferences from patient groups on health or non-health outcomes (Ratcliffe and Buxton, 1999; Cunningham et al., 2008; Gerard et al., 2008b; Louviere and Lancsar, 2009; Negrin et al., 2009); from service providers on preferences for working conditions, or remuneration, or prescribing decisions (Mark and Swait, 2003; Scott et al., 2007; Louviere and Lancsar, 2009) or clinicians or policy makers on policy choices (Baltussena et al., 2006). Some have broadened the application to explore societal and community preferences (Green and Gerard, 2009) similar to how DCEs are implemented in the transport and environmental fields (Hensher, 1993; Hanley et al., 1998; Garrod et al., 2002; Carlsson et al., 2003; Caussade et al., 2005; Green and Gerard, 2009).

This chapter uses DCEs as an alternative method of assessing societal preferences for different cannabis policies. The feasibility of using this approach to elicit preferences for different cannabis policies while varying the health and criminal justice consequences, as well as the rates of use of cannabis, was first explored through a pilot study as part of this thesis (Shanahan, 2009). Having determined the feasibility of the methods for addressing this question, a study with respondents from the general public was conducted to explore preferences for criminalisation of cannabis, cannabis cautioning, civil penalties, and the legalisation and regulation of cannabis. This study appears to be the first use of DCEs in the illicit drugs policy field.

The aims of the research in this chapter are: 1) to assess preferences for the four policy options in the context of the trade-offs between health harms, criminal justice expenditures, rates of cannabis use and the location of purchase; and 2) to explore the potential policy relevance of the results.

Ethics approval for this study was received from the UNSW HREC and ratified by Southampton University.

230 Chapter 10

10.2 Theory

DCEs are a form of experiment which can be used to estimate the relative utility of alternative goods or services (Amaya-Amaya et al., 2008). DCEs rely on an individual’s knowledge or perceptions of their own preferences, and on their ability to make trade-offs between alternatives in the presence of constraints in terms of money, time, availability and so on. DCEs extend beyond consumer theory in three ways: i) traditional consumer theory assumes homogenous goods while Lancaster’s view is that it is the characteristics (attributes) of the good that determine the utility they provide. It is the variation of these characteristics that is one of the fundamentals of a DCE; ii) discrete choice theory deals with a finite and mutually exclusive set of alternatives as opposed to an infinite set, as in consumer theory; and iii) consumer theory assumes deterministic behaviour but choice theory introduces the concept that individual choice behaviour is intrinsically probabilistic (random).

It is the attributes (characteristics) of the good or policy which are important, as it is the attributes which drive decision-making (Hensher et al., 2005). Preferences are elicited by providing a series of profiles (choices) to respondents. Each profile is comprised of a set of attributes describing the good, a level for each attribute with levels varying across the profiles. Each choice made implies that it provides a higher level of utility than the alternative. Individuals have a concept of the value (indirect utility) for each choice but the researcher cannot know all the factors that affect someone else’s choice therefore the utility is comprised of the knowable part and the random or unknowable parts. The random part may be due to unobserved attributes, unobserved preference variation, specification error or measurement error, or inter-individual differences in utilities as a result of variation in tastes (Manski 1977 in (Ben Akiva and Lerman, 1985; Amaya-Amaya et al., 2008).

Such a method is useful when the market, which often plays the role of equating the price and the benefits gained from a product or service, does not fulfil that role. This is often the case with policies and services within the public sector where DCEs are frequently used.

231 Chapter 10

10.3 Methods

The methods for developing the survey and conducting the experiment draw heavily on two publications, chapters 2 and 4 of Ryan, Gerard, and Amaya-Amaya’s Using Discrete Choice Experiments to Value Health and Health Care (2008) and Lancsar and Louviere’s (2008) Conducting Discrete Choice Experiments to Inform Healthcare Decision Making: A User’s to Guide.

10.3.1 Model specification

The utility function can be presented as follows:

(1)

�� ,��1…..,� where individual i will choose alternative j if, and only if, that alternative maximises their utility amongst all J alternatives. The Utility U for individual i is conditional on

choice j, and is decomposed into explainable or systematic Vij and non-explainable or

random component εij . Vij can be further broken down into Xjk, a vector of attributes of the good or policy, and Z, a vector of N characteristics of the individual i, and β and γ

are the respective coefficients to be estimated for K attributes, with γn coefficients indicating the impact that the personal characteristics have on choice (Green and

Gerard, 2009).

(2) � � �� ∑��1 �� ∑��1 �� where yij is equal to 1 if alternative j is chosen, and 0 otherwise and 1 is the choice if and only if

Vij + εij > Vim + εim for all j ≠ m which rearranges to

Vij - Vim > εim- εij.

Utilities are not observed, but by observing the choices, utilities can be estimated

(Hensher et al., 2005). Additionally (εim - εij) is not observed directly and so it is only possible to make observations up to a probability of occurrence with some distribution or density function. It is the choice of this distribution which affects interpretation of the probabilities (Amaya-Amaya et al., 2008). Different density functions for the

232 Chapter 10

unobserved part of the utility εij lead to different families of probabilistic discrete choice models.

If the distribution of εij is independent and identically distributed (IID) as extreme value type 1 (Gumbell) distribution then a multinomial logit model is estimable (Louviere et al., 2000; Amaya-Amaya et al., 2008):

�� 1� � � �� , ��1, … . � ∑��1 � where μ is a scale factor which is not estimable separate from tastes (Amaya-Amaya et al., 2008) and is assumed equal to one. It is this model that is used as a starting point for most DCE as it is simple, and allows exploration of the data set, ensuring it has been entered correctly, is clean and is providing meaningful results (Hensher et al., 2005). However, if heterogeneity across individuals is expected the resulting parameters are inefficient and biased, and mixed logit may be preferable (Amaya-Amaya et al., 2008).

10.3.2 Survey

This next section lays out the various steps in the development and in conducting the survey. This includes the rationale and process for selecting for the attributes and their levels, the mathematical design, the scenario development (the story and background) and the sampling strategy.

10.3.2.1 Attributes and levels

As described above, attributes are the characteristics of the good or in this case the policy, which Lancaster argued determines the utility, and levels are the different measures each attribute may be given in the survey. It is the combination of the attributes and their levels that result in the choices made by consumers. Careful attribute and level selection is essential (Lancsar and Louviere, 2008). Too few, or exclusion of important attributes, will result in misspecification while too many attributes will make the experiment difficult to design and cumbersome for the respondent to complete.

Nine key attributes of the policy and its consequences were documented during the literature reviews for the other chapters in this thesis, in particular chapters 2, 4, 5 and 8.

233 Chapter 10

Following the literature reviews, consultations were held with 10 key informants selected for their experience in the fields of cannabis policy (2), criminal justice (3), school-based programs (2), treatment (4) and/or their previous and current cannabis consumption (3). The consultations focused on eliciting from each person their views on important factors to consider when planning cannabis policy. Upon completion of the interviews, there was considerable congruence between the important factors identified by the key informants and those selected from the literature. In arriving at five attributes a number of simplifications occurred. Two potential health-related attributes (cannabis dependence, and cannabis related-mental illness) were combined with other cannabis related-health consequences into a single total health-related expenditure. Two other factors (the price of cannabis and the overall probability of being arrested) were held constant in the survey and respondents were informed of this in the scenario. Respondents were informed that the probability of arrest would remain constant in those policies where cannabis was still illegal, as respondents may otherwise infer that specific programs may result in a policy targeting cannabis users. The issue of price was seen as important, as some might expect under legalisation without government interventions, such as excise taxes, that the price of cannabis would decline. The final excluded factor, ‘knowledge of the purity’ of cannabis was felt to be highly correlated with location of purchase and would also logically only occur when cannabis was legal, and when purchased from a legal shop. As such, the levels would be conditional on the legal status of cannabis and correlated with ‘purchase in cannabis shop’.

Once the attributes were selected, the descriptions of the attributes were modified to ensure they were comprehensive but written in a language that was succinct and not too technical. The survey was then pre-piloted to ensure readability and then a pilot study of 60 colleagues was conducted and analysed (Shanahan, 2009). The pilot study was conducted to ascertain whether such a study was feasible, whether an experimental design was achievable, and then the results were analysed to obtain information for improving the experimental design. Additionally, the data were checked to determine whether trading between attributes occurred. If levels are perceived as unreasonable or as not different trading may not occur. Few difficulties in completing the pilot survey were reported and the results were intuitive (Shanahan, 2009). Following the pilot study, modifications were made to the introduction and instructions because the some

234 Chapter 10

pilot study respondents indicated these to be too detailed Also, as little trading occurred between the two level ‘rates of use among young people’ attribute, the range was widened and the number of levels was increased from two to four.

Attribute levels can be qualitative or quantitative (Hensher et al., 2005; Amaya-Amaya et al., 2008; Lancsar and Louviere, 2008). In this study, three attributes were quantitative (‘health harms’, ‘criminal justice expenditures’ and ‘use’). To provide the opportunity for trading there must be at least two levels for each attribute. Additional levels add complexity to the design, but with three or more levels linearity does not have to be assumed (Lancsar and Louviere, 2008). The actual value or description for each level needs to appear credible to the respondent to ensure face validity; however, there must also be sufficient variation to ensure trading occurs between levels.

Table 77: Attributes and levels Attribute Definition Levels Policy The legal status of Arrested and go to court cannabis Civil fine or attend educational session for less than 15 grams but police may choose to charge; arrest for more than 15 grams Cannabis caution possible but police may charge for less than 15 grams; arrest for more than 15 gramsb No offence – legally traded good; Crime if provide to persons <18 years of age

Location The purchase location Cannabis shop sells to only someone of age 18 and over Illegal drug dealer sells to anyoneb

Health harms The annual cost to treat $10 million health problems from $20 millionb cannabis use $30 million $40 million

Criminal justice The cost to enforce the $20 million expenditure cannabis laws $40 million $60 millionb $80 million

Use Number of 14 to 29 3 in 100b year olds using 8 in 100 cannabis at least once a 13 in 100 month 18 in 100 b – indicates the levels for the status quo option

In the final survey five attributes were selected, four had four levels and one had two levels (see Table 77). Two attributes reflect policy options and three are potential outcomes from the policy. The five attributes selected were: 1) the legal status of

235 Chapter 10

cannabis; 2) the location of purchase; 3) the annual expenditures on health care as a result of cannabis use, both for treatment of cannabis dependence and abuse, and subsequent mental health and physical health problems; 4) the rate of monthly or more frequent use of cannabis among 14 to 29 year olds; and 5) the annual expenditure by the criminal justice system on enforcing cannabis laws.

10.3.2.2 Detailed description of attributes and levels

Legal status of cannabis

The four levels of the legal status policy attribute used in this study are criminal status, criminal but with cannabis cautioning (currently in four Australian jurisdictions), criminal but with civil penalty (currently in four Australian jurisdictions), and legalisation (see Chapters 2 and 6 for additional information). In this study, criminal status is defined as when every person found with any amount of cannabis is charged and goes to court, and if found guilty receives a criminal record. A cannabis caution is issued by police for possession of a small amount of cannabis (less than 15 grams) and no criminal offence is recorded when an individual is cautioned. A caution involves the police removing the cannabis from the individual, warning them of the dangers of cannabis and that it is illegal, and recording their details. A civil penalty or Infringement Notice (fine) is issued when police detect someone with a small amount of cannabis (less than 15 grams). If the fine is paid no criminal conviction is recorded. Under both cautioning and civil penalty other cannabis offences may result in a criminal charge. The final of the four policy choices is legalisation. Here cannabis is a legally traded good for those aged 18 and over.

Location of purchase

The two levels for this attribute are: those 18 years of age and older may purchase in a legal shop or anyone can purchase from an illicit drug dealer. One argument offered by those who are pro-legalisation is that if cannabis were purchased legally from a cannabis shop consumers would have less contact with illicit drug dealers thus there would be less likelihood of coming into contact with other illicit drugs. There is some evidence from the Netherlands to support this argument (Reinarman, 2009). Normal trade laws would require labelling of quantity, type of cannabis, and potency if cannabis were to be sold as a legal good.

236 Chapter 10

Rates of use of cannabis

The rate of current monthly use of cannabis among 14–29 year olds of 13.8% was estimated from the 2007 National Drug Strategy Household Survey (Australian Institute of Health and Welfare, 2008a). Four levels, equidistant from 3% through to 18% were used. A wide range was used to try to ascertain if low use rates were preferred, and whether expected higher rates with legalisation would be acceptable. Eighteen per cent is similar to the projected prevalence under legalisation in Chapter 4 for this age group.

The annual cost to treat health problems from cannabis use

There are several potential health harms associated with cannabis use particularly when young people use cannabis heavily (see Chapter 8). Additionally, a number of people who use cannabis will become dependent with estimates of approximately 1 in 10 people who use cannabis becoming dependent (Teesson et al., 2002). The current expenditures for treatment of dependence and other cannabis-related illnesses ($20 million) NSW were estimated elsewhere. See Chapter 8 and Ngui and Shanahan (2010) for additional details. The other three levels were based on potential impact on costs of treatment and other health consequences with sufficient differences to maximise trading. The expenditures on health harms related to cannabis were $10, $20, $30 and $40 million per year.

The cost to enforce the cannabis laws

The ‘status quo’ expenditure for enforcing cannabis laws in NSW is an estimate of the costs of policing, courts, penalties, prosecution and court-imposed and court-provided treatment ($60 million per year). The status quo costs are based on those estimated in Chapter 5. As with health care expenditure the other levels were chosen around the status quo to reflect potential range under different assumptions while ensuring sufficient variation between levels. The expenditures to enforce cannabis laws were $20, $40, $60 and $80 million per year.

10.3.2.3 Alternatives

Alternatives are the combinations of attributes and levels which make up the profiles presented to the respondents (Lancsar and Louviere, 2008). There are three somewhat interrelated decisions about alternatives that must be made. These are determining: 1) how many alternatives to include; 2) whether to use labelled or unlabelled alternatives;

237 Chapter 10

and 3) whether to use an opt-out or status quo alternative (Amaya-Amaya et al., 2008; Lancsar and Louviere, 2008).

Published DCEs vary in the number of alternatives offered to respondents, from a dichotomous choice between a new treatment and the status quo through to multiple alternatives with or without a reference alternative (Rolfe and Bennett, 2009). It is typically argued that designs with more items are more complex (Swait and Adamowicz, 2001a, b in (Hensher, 2006) thus posing more of a burden resulting in less reliable information. Others argue that it is not necessarily the quantity of the information but rather the relevance of the information that is important (Hensher, 2006; Hensher, 2009). Controlled experiments designed to explore this issue have somewhat conflicting findings. One study found that varying the number of alternatives between two and three resulted in more robust results from the models with three alternatives compared to those with two alternatives (Rolfe and Bennett, 2009). However, Adamowicz and colleagues (2005) found that those faced with three alternatives were more likely to choose the status quo than those faced with two alternatives. Another study found that increasing the number of alternatives to four improved the probability of respondents finding a preferred option but beyond four the higher levels of complexity were problematic (Caussade et al., 2005). In light of this evidence and the likelihood that the profiles in this study were at least moderately complex the decision was made to use three alternatives, one of which is a fixed-level status quo option.

Many early studies used forced choice designs which did not allow an opt-out alternative (Amaya-Amaya et al., 2008); however, this ‘conditional choice’ may require unrealistic choices and may result in biased results (Rolfe and Bennett, 2009). Status quo, pivot or no-choice options can be used to avoid a forced choice (Rolfe and Bennett, 2009). Starmer (2000) argues that the reference alternative has an important role and additionally may be crucial to explaining real economic behaviour as it may provide the memory content necessary to make the DCE choices meaningful. Alternatively, others demonstrate preferences for the status quo, suggesting consumers value what they know (endowment effect) (Salkeld et al., 2000) or indicate serial non-participation (Rolfe and Bennett, 2009). In this study it is expected that most survey respondents will not be aware of the existing laws and policies for cannabis in their jurisdiction (MacCoun et al., 2009) and without the context provided by the status quo, many may find the choices lack relevancy. 238 Chapter 10

Figure 18: Example of one profile Attributes Policy A Policy B Current Policy

If an adult is found with cannabis No offence – Civil penalty Cannabis caution — less than 15 grams legally traded possible but possible but good; crime if police may police may provided to choose to charge. choose to charge. — more than 15 grams persons <18 Arrested and go to Arrested and go to years of age court court. Cost to treat health problems $40 million $30 million $20 million from cannabis Costs to enforce cannabis laws $20 million $40 million $60 million

Number of 14 to 29 year olds 13 in 100 8 in 100 8 in 100 using cannabis at least once a month Where cannabis is purchased Cannabis shop Illicit drug dealer Illicit drug dealer sells only to sells to anyone sells to anyone someone of age 18 If you could choose a policy for NSW, which policy would you    choose?

The final decision is whether or not to use labelled or unlabelled alternatives. The decision to use unlabelled alternatives should be made in the context of the research problem (Louviere et al., 2000; Hensher, 2006; Amaya-Amaya et al., 2008; Lancsar and Louviere, 2008, 2009 #805). While the use of labelled alternatives can be useful in conveying information it can also be problematic. The connotation of labels, such as criminalisation, depenalisation, cannabis cautioning or legalisation, could well result in respondents ignoring the actual attributes and levels presented (Hensher et al., 2005; Amaya-Amaya et al., 2008) and choosing for example, ‘criminalisation’ on every choice without considering the levels of the attributes because she/he may believe that criminalisation is the only way to manage cannabis. This may then lead to a correlation between the individual responses and the label. Therefore the decision was made to use unlabelled alternatives, with the exception that the status quo alternative is presented as the “current policy”.

Another benefit of the use of unlabelled alternatives is that it is not necessary to identify and include all possible alternatives (Hensher et al., 2005 ) but rather the choice of attribute levels can reflect the variation in alternatives. Additionally generic experiments are more likely to be independent and identically distributed (IID) than labelled experiments. This is because, as described above, the perceptions around the alternative’s name may be correlated with the attributes in the experiment which then 239 Chapter 10

leads to a failure of IID (Hensher et al., 2005). In an unlabelled experiment, by definition, the alternatives are interchangeable so unique utility functions for each alternative are meaningless.

In summary, in this study, the current NSW cannabis policy was presented as the reference or status quo alternative with the two unlabelled policy options. The status quo option also provides a reference for the respondents who may not be aware of existing policy. Figure 18 provides an example of one profile.

10.3.2.4 Design

The full factorial, of all combinations of levels and attributes and the two alternatives, is 130,816 possible choice sets [((44 * 21) * ((44 * 21) -1))/2]. Although blocked designs are increasingly being used to cover the full factorial (Lancsar and Louviere, 2008; Green and Gerard, 2009) given this number of possible choice sets this was not practicable so a fractional factorial design was chosen. Typically, in the situation when neither a full factorial nor a blocked design with all possible combinations are feasible, researchers have used some form of main effects design usually ignoring interactions (Louviere et al., 2000; Hensher, 2006).

Orthogonality

Orthogonal main effects plans (OMEP) have traditionally been used in DCEs when full factorials were not possible (Lancsar and Louviere, 2008). Orthogonality was typically defined as when the levels of each attribute vary independently of each other and for any two attributes all combinations of pairs of levels appear with proportional frequency (Huber and Zwerina 1996 in (Hensher et al., 2005). In earlier DCEs an orthogonal design was usually obtained from a book of designs or software such as SPEED (Ryan et al., 2008; Louviere and Lancsar, 2009). Researchers then often randomly selected one choice set to be the comparator; randomly split the choice sets into two; selected preferred choice sets; or removed a row from the design which to the researchers appeared to provide a behaviourally unrealistic scenario. The problem is that each of these actions destroyed orthogonality of the design (Hensher et al., 2005; Street et al., 2005; Rose and Bliemer, 2008).

Orthogonal designs, at least in relation to linear models, allow for an independent determination of each attributes’ contribution on the dependent variable and they 240 Chapter 10

maximise the power of the design to detect statistically significant relationships. Logistic regressions estimate the parameters of a linear model derived from the data and transform this into a non-linear S-shaped model (Hutcheson and Sofroniou, 1999). But while orthogonality is an important criterion to determine independent effects in linear models, discrete choice models are not linear (Train, 2003). Others make the point that it is not the design which should be orthogonal but rather the differences in attribute levels that should be orthogonal (Hensher, 1993), while other researchers have begun to question the relevance of orthogonal designs when applied to DCEs (e.g., Huber and Zwerina, 1996; Kanninen, 2002; Kessels et al., 2006; Sándor and Wedel 2005).

There are other challenges to orthogonality. For example, if a no-choice alternative or a status quo option is used it will impact upon respondents’ choices and must be incorporated in the design. Other often overlooked threats to orthogonality include differences in response rates for different blocks when a blocked design is used, the post–hoc addition of demographic characteristics, and transitioning between design and actual levels of the attributes particularly if equidistant spacing is not used.

There are currently two approaches which have moved beyond the traditional OMEP – the optimal orthogonal design also known as D-optimal designs (Street et al., 2008) and D-efficient designs (Rose and Bliemer, 2008; Rose and Bliemer, 2009). The argument for a D-optimal design approach is that respondents are forced to trade on all attributes in the experiment, whilst the orthogonality of the design ensures that the independent influence each attribute has upon choice can be determined. Optimality under this definition differs from that of D-efficient designs, in that D-optimal designs attempt to maximise attribute level differences whereas D-efficient designs attempt to minimise the elements that are likely to be contained within the variance covariance matrix of the models (Rose and Bliemer, 2008). If any information about the parameters is available to assist with the D-efficient design, then efficient designs will always outperform orthogonal designs as efficient designs use the knowledge of the prior parameters to optimise the design (Rose and Bliemer, 2009).

Complicating the design for this project are the use of a defined status quo option and the constraint that one level (purchase in a shop) of the attribute ‘location of purchase’ could only reasonably occur in combination with the fourth level of the policy option (legalisation). This made it more difficult to find a suitable orthogonal, balanced

241 Chapter 10

design. Therefore the newly available NGENE software (version March 2009) was used to solve this complexity. It is able to search until it finds the most D-efficient design possible from a variety of designs. NGENE can be used to derive a variety of designs (including orthogonal, optimal orthogonal designs or efficient designs).

One of the criteria needed to find the most D-efficient designs are prior information from previous studies for each attribute as a starting point for the design. As no previous study similar to this had been conducted there were no priors. This can pose a problem although Rose and Bleimer (2009) make the point that D-optimal orthogonal designs start with zero priors. They suggest that an efficient design with zero priors is a reasonable option if none are available from the literature. As a first step, to generate the design for the pilot study, the assumptions were made that, all else being equal, most respondents prefer lower amounts of monthly cannabis use among 14 to 19 year olds, lower expenditures on criminal justice system and lower expenditures on health care. In the absence of existing priors, these priors were set at -0.1. The priors for ‘location of purchase’ and ‘policy’ alternatives were set at zero.

The priors for the design of the final study were the coefficients from the analysis of the pilot study (Shanahan, 2009). The final design was thirty-six rows blocked into four versions and included a blocking variable (Lancsar and Louviere, 2008). The final design can be found in the Appendix–Chapter 10.

10.3.2.5 The sample and mode of administration

As this study was attempting to elicit societal preferences for cannabis policy, the desire was to achieve a sample representative of the population in terms of age, gender, and previous cannabis use. Different survey administration modes were considered. Research suggests that face-to-face administration of DCE surveys may provide the ‘best’ results, and may permit confirmation about some characteristics of the respondent, but as this was a population-based survey involving questions about illicit drug use it was felt that anonymous administration might avoid answers that were approval seeking (Smith, 2003; Champ and Welsh, 2006). In addition, face-to-face surveys are expensive to implement as a result of required training and interviewer costs (Champ and Welsh, 2006) and as such required resources that were beyond this study. Telephone surveys were not felt to be suitable due to the complexity of questions in the

242 Chapter 10

survey. The use of commercial market research firms and internet panels are frequently used in this type of work (Bartels et al., 2006; Champ and Welsh, 2006; Louviere and Lancsar, 2009) although it is recognised that biases may be introduced, as not everyone has access to the internet.

The sample was recruited from an internet panel of 90,000 Australians held by a commercial market research firm (SSI). Initially, random individuals from the panel were invited by email to participate in the survey, but as predefined quotas for various age and gender stratum were filled, more targeted invitations to fill the age and gender stratum were issued until a sample of 1000 completed surveys were obtained. The blocks were randomly allocated to respondents. Additionally, the order of presentation of the nine profiles within each of the blocks was random.

A check of consistency of responses was not conducted as it is unclear how to define inconsistency in the context of this study. Moreover, recent recommendations suggest researchers should consider leaving apparent inconsistent choices in the analysis (Louviere and Lancsar, 2009).

10.3.2.6 Socio-demographics and attitudes

Data were collected on a range of characteristics and attitudes which ex ante were expected to impact on policy choices. It was anticipated that those who had used cannabis recently would prefer a less restrictive policy, as would those who indicated they believed that cannabis has some heath benefits. Also, as the rate of cannabis consumption is on average higher in younger age groups and in males (Australian Institute of Health and Welfare, 2008a) it was expected that both those who are younger or male may prefer a more lenient cannabis policy (legalisation or civil penalties), and be more accepting of a higher rate of use among young people. Conversely, those who believe cannabis is usually or always addictive might be expected to have a preference for lower rates of use and for a more restrictive policy. Ex ante there were no clear expectations between the choice of policy attribute and education and between policy choice and employment status. Conversely, it was expected that those who indicated they were left on a left–right scale would gain higher utility from less restrictive cannabis policy whereas those who indicated they were on the right would prefer a more restrictive policy and lower rates of use.

243 Chapter 10

10.3.3 Analyses

10.3.3.1 Model estimation

The random utility model was estimated first with a multinomial logit (MNL) model and followed by a mixed logit (ML) model using the NLOGIT software package. ML is also referred to in some literatures as a random parameters model or mixed multinomial model (Hensher et al., 2005). MNL was used as the starting point for analyses (Hensher et al., 2005; Amaya-Amaya et al., 2008) as it is relatively easy to estimate and can be used to confirm that the data are clean, and that sensible results can be obtained from the data (Hensher et al., 2005). The expectation, confirmed in the pilot study, was that there would be considerable preference heterogeneity across the sample and the ML would improve the model. ML models improve on the ability to account for variation by allowing for variation of the attribute levels over the respondents (Hensher et al., 2005; Hole, 2008; Kjaer and Gryrd-Hansen, 2008; Train, 2009; Scuffham et al., 2010). ML models assume errors are independent across alternatives and individuals but not identically distributed across individuals (Amaya- Amaya et al., 2008).

The mixed logit (ML) utility function becomes

� � �� ∑��1 �� ∑��1 ��

� �� ∑��1 ��

� �� where individuals i=1…I; face a choice among J alternatives in each of S choice situations and is the full set of explanatory variables (both the characteristics of the

individual and �the attributes in the choice experiment). Still assuming the individual makes the choice which gives them the highest utility, the utility associated with each alternative j, as evaluated by each individual i in choice situation s, is represented by a discrete choice model through a utility expression above (Hensher et al., 2005). Imposing that the unobserved is IID (across individuals, alternatives and choice

situations) does not allow for�� error components of different alternatives to be correlated. To account for this we need to introduce into the utility function, through

244 Chapter 10

the βi some elements that may be heteroskedastic. This may be done through the use of random terms (Amaya-Amaya et al., 2008). In a ML model, the parameters of the observed variables are allowed to vary over respondents according to a predefined distribution (Amaya-Amaya et al., 2008) with a number of distributions to choose from (normal, logarithmic, triangular, and uniform) (Hensher et al., 2005). This allows the

estimation of the unconditional probability over all possible values of βi. The variables determined to be random are estimated with a mean, and some distribution.

If the mean of the standard deviation on each random parameter is not significantly different from zero, then the dispersion around the mean is equal to zero and it need not be treated as a random parameter (Hensher et al., 2005). The default strategy, if there is no prior information on which variables to treat as random parameters is to start with all attributes as random. The estimated standard deviations are examined and if the mean of the standard deviation is not significant it is not treated as random; this is repeated until the most parsimonious model is obtained.

In the final model the legal status of cannabis and the monthly rate of cannabis use in 14 – 29year olds were treated as random parameters with a normal distribution (mean of zero and a variance of one).

10.3.3.2 Coding and imposing linearity

Qualitative attributes (i.e. the legal status of cannabis, and location of purchase) may be coded either using dummy or effects coding; and quantitative attributes can be coded either by effects/dummy coding or their actual levels can be retained (Gerard et al., 2008b). To avoid the confounding between the level chosen as the base and the constant (Gerard et al., 2008b) dummy coding was not used. Using the actual levels of quantitative attributes presumes linearity. A Wald test was used to assess linearity (Hensher et al., 2005) of ‘rates of use’, ‘expenditures on CJS’ and ‘health care expenditures as a result of cannabis use’. The assumption of linearity was rejected for rates of use and this variable was effects coded. The two expenditure variables were treated as linear.

Effects coding, as with dummy coding, requires that one level be excluded in order to avoid dummy variable trap. Effects coding uses L-1 variables, which are coded as a series of 0, 1, and –1s. When the level of interest is present it is coded 1, when the

245 Chapter 10

excluded level is present it is coded –1, and zero otherwise. The effects coding for the legal status of cannabis is demonstrated in Table 78 where cannabis cautioning is the base or excluded level. Effects coding permits separate examination of the base case from a status quo or null option (Bech and Gryrd-Hansen, 2005). When effects codes are used, the constant term reflects the utility associated with the status quo (Bech and Gryrd-Hansen, 2005). All demographics and characteristics, as well as the legal status of cannabis, monthly rates of use by 14–29 year olds and location of purchase were effects coded.

Table 78: Policy attribute: example of effects coding Coding Choice Variable: Variable: Variable: Crim CivPen LegReg Criminal status 1 0 0 Civil penalty 0 1 0 Cannabis cautioning –1 –1 –1 Legalised and regulated 0 0 1

The use of ML permits the incorporation of the multiple observations by each respondent, taking into account the potential for correlated responses across observations (Hensher et al., 2005).

The estimation of the population probabilities in ML requires simulation. Halton sequencing has been demonstrated to provide more efficiet results than other options and is frequently used to simulate the log likelihood function (Hensher et al., 2005; Hall et al., 2006; Howard and Salkeld, 2008). A small number of draws (n=10–50) were used while models were being tested and then final models were run with a larger number of draws (500 to 1000) to test for stability.

10.3.3.3 Alternative specific constant

Typically alternative specific constants (ASC) are not used in a generic DCE as they do not provide any useful information (Hensher et al., 2005) unless testing for order bias. However, in the presence of a status quo option a constant should be included to assess whether there are characteristics of the status quo other than those included that might affect the utility (Bech and Gryrd-Hansen, 2005). In estimating this model a single constant was used in the utility functions for the two unlabelled equations.

246 Chapter 10

10.3.3.4 Policy analyses

Preference heterogeneity can be expressed by differences in the strength of the preference or by a difference in the direction of the preference (Eberth et al., 2009). One way of demonstrating the difference in preferences is by examining the proportion of the population who may have a negative or positive coefficient. This can be calculated using the mean and the standard deviation of the random parameters (Train, 2000; Eberth et al., 2009; Train, 2009).

Additionally, the interaction between an attribute (policy or rates of use attributes) and a covariate (i.e. age, gender, beliefs about cannabis, previous use of cannabis) permits examination of whether there is any preference heterogeneity around a given level of that attribute with respect to that covariate. One is then able to assess whether individuals with different demographics or beliefs have a preference for a particular level of an attribute. If the interaction is not statistically significant it does not imply no heterogeneity only that none is revealed (Hensher et al., 2005).

The parameter estimates can also be used to derive utilities for each of the alternatives by combining the coefficients with the relevant value for that variable (Hensher and Green, 2003). Scenarios were constructed, and utilities were derived containing levels of attributes which may be of interest to policy makers. Although the use of the standard deviations on the parameters would permit more detailed analyses at this time only the means were used in the scenarios.

In summary, in the final model, each respondent is faced with a choice of three alternatives (two unlabelled, one status quo), over nine scenarios with the choice recorded as a binary with ‘1’ representing an alternative chosen and ‘0’ where the alternative is not chosen. All discrete choice models require that one of the alternatives in the choice set must be a base (Mark and Swait, 2003); in this study the status quo alternative provided the base. The main analysis was a mixed logit model. The assumption of a linearly additive utility function results in:

Ualt = B0 + B1-3 policy + B4CJS expenditure + B5 health expenditure + B6-9 use + B10

purchase location + Bi characteristics + Bj interactions + .

��

247 Chapter 10

10.4 Results

A total of 1,670 persons logged on to undertake the survey over a ten day period in December 2009 (2nd to 12th). After the first 100 respondents had completed the survey the data were examined to ensure trading was occurring, and surveys were being completed. Of the 1,670 persons who logged on to undertake the survey, 350 (21%) belonged to stratum already full and were not permitted to complete the survey beyond the cannabis use, age and gender questions (screening questions). A total of 222 (16.7%) of the remainder did not complete the survey, they either logged off before completion or the internet became inactive. At total of 78 (5.9%) of the respondents who completed the survey were excluded as they did not meet the predefined time criteria. Data from 1,020 respondents with a total of 9,780 completed profiles were available for analysis.

Figure 19: Rates of cannabis use among the survey respondents and the population 50% Survey NDSHS 40% 30% 20% 10% 0% 18+ 14+ 19+ 18+ 14+ 19+

Ever used cannabis Used cannabis recently

NDSHS – National Drug Strategy Household Survey 2007

The blocks were randomly allocated to respondents, as were the nine profiles within each of the blocks. Although attempts were made to achieve a sample representative of the population, males aged 18–24, and populations over age 65 are underrepresented and, as illustrated in Figure 19, the rate of cannabis use among the sample was slightly higher than the population. As this survey had few respondents of the oldest age groups this is perhaps not surprising. See Chapter 7 for further discussion of population and differences within the sample.

Basis demographics and characteristics are found in Table 79. There appears to be no significant differences across the four blocks. Some differences between the sample and the population are found but as in Chapter 7 no attempt will be made to weight the

248 Chapter 10

sample. As the income question was only completed by 86% of the sample, in order to use the complete sample this variable was not used in the DCE analysis.

Table 79: Basic descriptive information Block Block Block Block All F p 1 2 3 4 N 1022 253 256 258 255

46.33 Age (SD) 47.37 46.78 45.97 45.25 .918 .432 (15.42) Male 531 .52 .55 .53 .50 .50 .514 .672 Cannabis use

Ever used 412 40.3% 42% 42% 36% 42% .720 .540 Recent use 115 11.3% 12% 11% 10% 12% .172 .915 Married/de facto 649 63.5% 67% 64% 62% 61% .776 .507 Number of 61.84 .65 .58 .61 .63 .200 .896 children (SD) (1.01) Health benefits– 273 26.7% 29.2% 24.6% 27.5% 25.5% .562 .640 Yes Cannabis sometimes 867 84.8% 85.4% 83.6% 83.7% 86.7% .425 .735 /always addictive Education

Less than year 12 210 20.5% 24.9% 19.5% 18.2% 19.6% 1.365 .252 Complete year 12 189 18.5% 19.4% 18.4% 20.2% 16.1% .529 .662 University 260 25.4% 24.1% 27.0% 25.2% 25.5% .184 .907 TAFE/Diploma 363 35.5% 31.6% 35.2% 36.4% 38.8%

Employment

Full or part time 528 51.7% 51.8% 54.3% 50.0% 50.6% .371 .774 Not working/ 328 32.1% 28.9% 33.6% 33.7% 32.2% .903 .439 pension /retired Full-time student 46 4.5% 5.5% 2.7% 4.7% 5.1% .598 .617 Homemaker 120 11.7% 13.8% 9.4% 11.6% 12.2%

Political scale

Left 147 14.4% 14.6% 18.0% 13.2% 11.8% 1.47 .221 Middle 489 47.8% 50.6% 43.4% 48.8% 48.6%

Right 205 20.1% 20.9% 19.5% 20.2% 19.6% 0.067 .977 Missing values 181 17.7% 13.8% 19.1% 17.8% 20.0% 1.296 .275 Ever attended 198 19.4% 17.8% 22.7% 18.2% 18.8% 0.814 .486 court

Overall the distribution of choices for the three alternatives was very similar, with 33% selecting the status quo with the remainder spilt between the two unlabelled alternatives – 32% Alternative 1 and 35% Alternative 2. This would suggest that across the alternatives there was no apparent first choice or status quo selection bias. In this sample, 3.7% only ever selected the status quo option and 16.6% never selected it (see Table 80).

249 Chapter 10

Table 80: Did people only choose or never choose status quo? Choices by individuals N Percentage Only choose status quo 38 3.7% Never choose status quo 170 16.6% Choose both 814 79.6% Total 1022 100%

Respondents revealed that overall their preferred mean annual public expenditure on health care related to cannabis use was $23.3 million, and for the criminal justice system it was $51.1 million. Their preferred rate of monthly cannabis use was approximately 10% of the 14–19 year-old population.

Table 81: Mean for attributes by alternative Alternatives 1 All Attributes Status quo & 2 alternatives Heath care expenditure Million $ 25.0 20.0 23.3 CJS expenditure Million $ 46.7 60.0 51.1 Use among 14 to 29 year Rate per 100 10.5 8.00 9.67 olds Location of purchase Illegal = –1 –0.67 –1 –0.78

10.4.1 Multinomial logit (MNL) results

In an ASC only model, the ASC for Policy A was insignificant ruling out order bias (Hensher et al., 2005) but the ASC for the status quo alternative was significant. A single constant (the same for both generic choices) included in all models, was positive and significant suggesting that, all else being equal, there was a negative preference against the status quo.

The MNL model was used to establish that the data were in a usable form but the explanatory power of the MNL model was low with a pseudo R2= 0.05. The coefficients on all levels of attributes were significant in the MNL model but the fit was poor. Given the diversity of the population’s view on cannabis policies it was expected that including interactions between attributes and socio-demographic characteristics and allowing for variation of the attribute levels over the respondents would result in a better fitting model.

10.4.2 Mixed logit results (ML)

In the move to ML, a model with no interactions was firstly compared to the MNL model. The fit of the model improved as evidenced by a pseudo R2 = 0.19 compared to

250 Chapter 10

2 2 a pseudo R = 0.05 for the MNL, and the log-likelihood ratio test with Χ 7df= 2945, chi- squared critical value= 14.07 (Hensher and Green, 2003). In the move from MNL to ML all the coefficients retained the same sign and all attributes remained significant. The age coefficient becomes non-significant in the ML model, and the coefficient on ‘believe there are benefits from cannabis’ (Yes = 1) is not significant. The coefficients on the standard deviations of the random parameters distribution (policy and use) were significant suggesting there is heterogeneity around the mean. The standard deviations on the other attributes were not significant and they are not treated as random in the final models. As with the MNL model the constant was significant and positive suggesting that, all else being equal, there are characteristics of the status quo which are not captured that negatively affects the utility of the status quo.

With the addition of the interactions the fit of the model improved as measured by a pseudo R2= 0.22. Again the log-likelihood function and AIC also improved. When compared to the previous ML model with no interactions, the log-likelihood ratio test 2 resulted in a Χ 48df = 485 with a chi-squared critical value of 65 which is consistent with the hypothesis that the model with the interactions explain the data better than the model without the interactions. The sign on the legalisation attribute changes from negative to positive when the interactions are added, and the coefficient on one level of ‘use’ becomes insignificant.

Interpreting the coefficients

The coefficients on the levels for the legal status of cannabis attribute were all significant as were the derived standard deviations. This indicates that not only is there a significant difference between the base level (cannabis cautioning) and each of the other levels of the legal status but there is some dispersion of preferences in the sample population. Relative to cannabis cautioning, criminal status is not preferred but both civil penalty (fine) and legalisation are preferred to cautioning.

Not surprisingly the signs on the coefficients for ‘use of cannabis by those aged 14 to 29 monthly or more often’ suggest that less cannabis use is preferred to more among this general population sample. However, there appears to be no significant difference in preferences between the rates of 13 out of 100 and 8 out of 100 young people

251 Chapter 10

consuming cannabis monthly. The mean standard deviations on all ‘use’ coefficients were significant suggesting that again there is some heterogeneity in the sample.

The coefficient on Sex (male = 1, female = –1) is negative and significant suggesting that females prefer the status quo relative to other alternatives. In contrast, those who do not perceive that there are any benefits from cannabis do not prefer the status quo. Over the whole sample the coefficients on whether the respondent felt that cannabis was addictive (always, sometimes =1) and whether the respondent had used cannabis within the past 12 months are not statistically significant. That is there is no difference in preferences for or against the status quo option, depending on these variables.

252 Chapter 10

Table 82: Main results MNL ML – No interactions ML – with interactions

Coeff p Std E Coeff p Std E Coeff p Std E

Random parameters

Constant 1.277 *** 0.103 1.982 *** 0.240 1.916 *** 0.237 Policy alternative

Criminal -0.321 *** 0.030 -0.538 *** 0.075 -1.447 *** 0.258 SD of coeff 1.726 *** 0.082 1.494 *** 0.074

Civil fine 0.346 *** 0.033 0.884 *** 0.059 0.746 *** 0.182 SD of coeff 0.876 *** 0.071 0.789 *** 0.070

Legalised -0.13 *** 0.034 -0.293 *** 0.088 0.506 *** 0.256 SD of coeff 2.134 *** 0.084 1.845 *** 0.076

Monthly cannabis use in 14 to 29 year olds

3 in 100 0.38 *** 0.032 0.602 *** 0.049 0.620 *** 0.153 SD of coeff 0.400 *** 0.075 0.375 *** 0.079

13 in 100 -0.121 *** 0.031 -0.192 *** 0.043 -0.113 0.150 SD of coeff 0.251 ** 0.100 0.254 ** 0.101

18 in 100 -0.337 *** 0.034 -0.552 *** 0.051 -0.880 *** 0.165 SD of coeff 0.404 *** 0.089 0.329 ** 0.112

Non-random parameters

Health expend -0.015 *** 0.001 -0.028 *** 0.002 -0.028 *** 0.002 CJS expend -0.016 *** 0.001 -0.026 *** 0.001 -0.026 *** 0.001 Location of purchase 0.256 *** 0.028 0.443 *** 0.043 0.444 *** 0.043 Sex (Female = –1) -0.124 *** 0.024 -0.183 ** 0.065 -0.193 *** 0.063 Age -0.005 ** 0.002 -0.006 0.005 -0.006 0.005

EDU < Yr 12 -0.124 ** 0.044 -0.221 * 0.119 -0.259 * 0.117 EDU = Yr 12 0.105 ** 0.044 0.159 0.122 0.182 0.119

EDU= Uni -0.075 * 0.042 -0.097 0.111 -0.094 0.110

Cannabis: Use recently -0.094 ** 0.039 -0.053 0.104 0.023 0.107

Can: benefits -0.174 *** 0.027 -0.206 ** 0.073 -0.171 ** 0.072 Can: addictive 0.035 0.033 0.037 0.089 0.042 0.089

Never married 0.053 * 0.031 0.093 0.086 0.081 0.085

Interactions

Crim*Benefit -0.584 *** 0.085

Crim*Addictive 0.171 * 0.099

Crim*Use recently -0.833 *** 0.148

Crim*Left view -0.668 *** 0.158

Crim*Right view 0.471 *** 0.154

Civil fine*Age -0.008 ** 0.003

Civil fine*Benefits -0.182 ** 0.061

Civil fine*Addictive 0.162 ** 0.074

Civil fine*Use recently -0.378 *** 0.098

Legal*Age 0.013 ** 0.005

Legal*Benefits 0.861 *** 0.083

Legal*Addictive -0.275 ** 0.100

Legal*Use recently 1.092 *** 0.126

Legal*Left views 0.766 *** 0.155

Legal*Right views -0.498 *** 0.162

Legal*Missing views -0.358 ** 0.150

Use 18 in100*Age 0.011 *** 0.003

Use 18 in100*Benefits 0.113 ** 0.054

LLF -9611 -8139 -7896

^^Pseudo R2 0.05 0.19 0.22

AIC 2.11 1.79 1.75

^^^LL ratio test 978 2945 485

Df 17 7 48

Chi 2 critical 27.59 14.07 65.17

^^ relative to constant model ^^^relative to previous model; *** p<.001; ** p <.05; *p <0.1

253 Chapter 10

The coefficients on the health expenditure and criminal justice expenditure attributes are both negative, reflecting less expenditure is preferred to more; both coefficients are highly significant. A WALD test for linearity suggested that assumption of linearity held for CJS (Χ2 = 231, p <.000) and for Health (Χ2 = 377, p <.000) but not so for rates of use of cannabis (Χ2= .26, p = 0.61). The coefficient on location of purchase was positive and highly significant reflecting strong preference for cannabis to be purchased at a legal shop.

The standard deviations of the random parameters were all statistically significant suggesting that there was heterogeneity of preferences in the population for these attributes. With the assumptions of normal distributions, this allows for the distribution of the coefficients to be examined using the following Prob (β >0) = Prob (z> -mean/std dev) (Train, 2000; Eberth et al., 2009; Train, 2009). Calculating these for each of the random parameters provides the percentage of the respondents who have a positive preference for each level. The distribution of the coefficients for criminal status level of the policy attribute is illustrated in Figure 20, where 84% do not prefer this option, however 16% do prefer criminal status. While on average civil fines are preferred, the distribution of the coefficients suggests that 82.8% prefer it to cannabis cautioning while 60.8% prefer legalisation.

Figure 20: Distribution of coefficients in the sample for the criminal status option

Standard 16% prefer deviation criminal status 1.726

-1.447 0

The distribution of the coefficients suggest that 95.1% of the sample prefer only 3 out of 100 young people aged 14 to 29 consuming cannabis monthly relative to 8 out of 100;

254 Chapter 10

and 99.6% do not prefer 18 out of 100 young people aged 14 to 29 consuming cannabis monthly.

Comparing the relative size of coefficients is not meaningful across the all of coefficients as they are measured in different units (i.e. effect codes and dollar values). However, as rates of use and policy attributes are both effects coded, comparing the size of the coefficients between these coefficients is valid. The two largest coefficients, the level of criminal status in the policy attribute and the highest rate of use of cannabis, are both negative; reinforcing the strong negative tendencies for these two levels.

10.4.2.1 Interactions

Only those interactions between the random parameters and the personal characteristics that are statistically significant are discussed below. (The complete set of results is in the Appendix–Chapter 10). In Table 83 the coefficients for the random parameters with their level of significance are across the top of the table and the three left-hand columns include the coefficients for the significant interactions between random parameters and personal characteristic and their level of significance. The remaining values are the combined effect of the attribute and the interaction. By examining the data in this manner it can be assessed how each of the characteristics impact on the coefficient, whether it increases or decreases the strength of the preference, and whether it changes the sign.

Criminal status (CRIM)

Assessing the interaction between the preferences for or against treating all cannabis offences as a crime, several characteristics provide some significant but not unexpected findings. The coefficient on this variable is negative, that is, it is not preferred relative to cannabis cautioning. Both those who believe there are some health benefits from cannabis and those who have used cannabis in the past 12 months dislike criminal policy relative to cannabis cautioning even more than the average (coefficient becomes more negative). This is also true for those whose views are left leaning (relative to the middle) on a political scale of 0 to 10, where 0 to 3 is classified as left. Those who believe cannabis is always or usually addictive, and those who state their political views are right leaning (relative to the middle) on the political scale where 8 to 10 are

255 Chapter 10

classified as right have positive preferences for the criminalisation policy (becomes less negative).

In summary, for those who believe in health benefits, have used cannabis recently or are left leaning, the disutility of criminalising cannabis is stronger than the mean whereas for those who believe cannabis is addictive, or are right leaning, their utility for this form of cannabis policy is increased relative to cannabis cautioning.

Table 83: Significant interactions and random parameters Attributes and levels

Crim Civil Legal Use: 18 in 100

Coeff -1.447 0.746 0.506 -0.880 Interaction

Signif *** *** * ***

Crim* Benefit -0.584 *** -2.031

Crim*Addict 0.171 * -1.276

Crim* Recent use -0.833 *** -2.280

Crim* Left -0.668 *** -2.114

Crim* Right 0.471 ** -0.976

Civil* Age -0.008 ** 0.737

Civil* Benefits -0.182 ** 0.564

Civil* Addict 0.162 ** 0.908

Civil* Use -0.378 *** 0.367

Legal* Age 0.013 ** 0.520

Legal* Benefit 0.861 *** 1.367

Legal* Addict -0.275 ** 0.231

Legal* Recent use 1.092 *** 1.598

Legal* Left 0.766 *** 1.273

Legal* Right -0.498 ** 0.009

Use: 18 in 100* Age 0.128 ** -0.752

Use: 18 in 100 * 0.113 ** -0.767 Benefit *** p <0 .001; ** p < 0.05; * p <0.1

The findings for the civil penalty as applied to possess/ use offences (and otherwise, criminal offence for all other cannabis offences) are somewhat different. The coefficient on this attribute level was positive. Those who are on average older have, on average, less preference for civil penalty relative to cannabis cautioning (becomes less positive) as do those who think cannabis has health benefits (becomes less positive). Those who believe cannabis is addictive have a preference for this policy over cannabis cautioning (becomes more positive). This is in contrast to those who have used

256 Chapter 10

cannabis in the past 12 months who do not prefer civil penalty to cannabis cautioning (becomes less positive).

Examining the preferences for the legalising cannabis option, those who are older, believe there are health benefits from cannabis, are left on the political scale or have used cannabis within the past 12 months on average have a greater preference for this policy relative to cannabis cautioning (becomes more positive). The preferences for the legalisation option by those who believe cannabis can be, or is addictive, or those who report being right leaning are negative (becomes less negative).

Although there is demonstrated heterogeneity in the rate of use attribute as evidenced by the significance of the mean of the standard deviations, the interactions included in this model do not provide much explanation for this heterogeneity. While not surprising, the younger population have a stronger preference for the highest rate of use relative to the average although the strength of preference is not strong enough to change the coefficient to positive. Those who believe cannabis has some health benefits also have a preference for the highest rate of use.

10.4.3 Utility scores

As an unlabelled experiment was used the estimated utility functions provide generic results. Scenarios were constructed for each of the policy alternatives to estimate an indirect utility function (V) for each policy option, e.g. V(crim) = constant + CoeffCrim * 1

+ CoeffUse 8 in 100 *1+ CoeffHealth * $10 mill+ CoeffCJS * $80. This was repeated for each of the policy options including cannabis cautioning and the status quo (see the first three columns in Table 84). Status quo and cannabis cautioning differ only by the inclusion of the constant in the cannabis cautioning option utility function and not in the status quo. This estimation does not account for the known heterogeneity. In column two, the levels selected were those which were thought to best represent the level for that policy option while column three contains the calculated utility score. In the fourth and fifth columns one level is changed in each row, and the resultant utility scores are provided. These utility scores are ordinal and as such are only relative to the other scores presented here.

257 Chapter 10

Table 84: Relative utilities scores –constructing scenarios for sensitivity analyses Resultant Legal status Description of level of each Utility Level utility of cannabis utility function score (V) changed score(V) Legal, Use 18/100; health expenditure $40 million, CJS Use 8/100 1.119 Legal and expenditure $20 million, 0.351 Health $60M –0.205 regulated purchase illegal dealer, with CJS $10M 0.613 constant Civil penalty Civil pen, Use 13/100; health for expenditure $20 million, CJS Use 8/100 0.465 possession, expenditure $60 million, -0.021 Use 18 /100 –0.788 criminal purchase illegal dealer, with CJS $40 M 0.502 otherwise constant Cannabis Cautioning, Use 13/100; caution for health expenditure $20 Use8/100 –0.086 possession, million, CJS expenditure $60 -0.571 Use 18/100 –1.339 criminal million; purchase legal shop, CJS $40 M –0.048 otherwise with constant Crim, Use 8/100; health expenditure $10 million, CJS Use 3/100 –1.725 Criminal expenditure $80 million, -1.972 CJS $60M –1.449 purchase illegal dealer, with Health $30M –2.005 constant Cautioning, Use 13/100; health expenditure $20 Use8/100 –2.001 Status quo million, CJS expenditure $60 -2.487 Use 18/100 –3.254 million; purchase illegal CJS $40 –1.964 dealer, no constant Concentrating on columns one to three first, the differences between the scores reflect the impact of not only the policy attribute, but also the other attributes such as rates of use of cannabis, location of purchase and expenditures on health and criminal justice. With these assumptions, the least preferred option is the status quo (cautioning for possession but otherwise criminal). The score for cannabis cautioning separate from the status quo also results in a negative estimate but far less so. This suggests that it is some unidentified attributes of the status quo that is affecting these results, not cannabis cautioning per se. The criminal status is also strongly negative. In contrast, legal status is positive in the scenario presented here.

The next two columns explore the impact of changing the assumptions on some levels for each of the utility functions. If for example, the rate of use was 8 per 100 young people the utility score increases to 1.119 whereas if the health expenditure related to cannabis harms was $60 million annually, the score becomes negative. None of the changes in levels result in cannabis cautioning, status quo or the criminal status policies becoming positive, however civil penalty becomes positive when the expenditure on CJS decreases to $40 million or when use is 8 per 100 young persons.

258 Chapter 10

10.5 Discussion

This chapter and the original pilot study demonstrated that it is feasible to use DCEs to elicit preferences for cannabis policy from the general public. Respondents did appear to engage in the survey, with the majority of respondents trading between one of the unlabelled policy options and the status quo for NSW. The sample of 1022 respondents from an existing Australian internet panel was slightly more educated than the general population, and had fewer younger males (19 to 29 year olds) and more older males (60+) than the general population. In other characteristics such as cannabis use, employment status and income, the sample is broadly representative of the population.

This study adds to the methodological literature through the use of the status quo option. Many DCEs use forced choice designs (Amaya-Amaya et al., 2008); however, this ‘conditional choice’ may require unrealistic choices and may result in biased results (Rolfe and Bennett, 2009). The use of a status quo option not only avoids this potential bias (Rolfe and Bennett, 2009) it also functions as a reference to real policy. This was an important consideration as it was expected that most survey respondents would not be aware of the existing laws and policies for cannabis in their jurisdiction (MacCoun et al., 2009). Without the context provided by the status quo, many may have found the choices lacked relevancy.

The ML model with normally distributed random parameters and interactions provided a better fit than either the MNL model or the ML model without interactions. All the coefficients on the attributes were statistically significant with the exception of the coefficient on the rate of 13 per 100 young people using cannabis which was not found to be statistically different from 8 per 100 (the base rate). Overall, there is a preference for lower use by those who are aged 14 to 29 years, with the coefficient on the lowest rate (3 per 100) having a positive coefficient and the highest rate (18 per 100) has a negative coefficient.

Criminal status (as a level in policy attribute) was not preferred with 84% preferring cannabis cautioning, which leaves 16% preferring the criminal status. While it appears clear that criminal status is not preferred, the minority of respondents who are in favour of this policy may be of a sufficient size to affect debate on this issue. In contrast, the coefficient on civil penalties was positive with 82.8% preferring this policy option to cannabis caution. Preferences for legalisation were not so clear with 60.8% preferring 259 Chapter 10

legalisation to cannabis cautioning suggesting that if this debate was held, there could be a considerable number opposed to legalisation.

Although not directly comparable, recent surveys asking respondents their views on cannabis policies have found that almost half (45%) were opposed to the legalisation of cannabis, which is not too different from these results. What is different is that those surveys found there was strong support for increased penalties for sale or supply of cannabis (Australian Institute of Health and Welfare, 2008a) whereas the results of this current research suggest that the criminalisation of cannabis is not a preferred option. One of the apparent conflicting results is a majority preference for a legalised policy option but also strong preference for a lower rate of cannabis use among young people. This is unlikely to occur if the prevalence of alcohol and tobacco consumption and the results from Chapter 4 are any indication of what might happen if cannabis were to be legalised.

The coefficients on health and criminal justice expenditure were both negative suggesting that this sample regarded these expenditures as normal goods; less expenditure is preferred. The overall mean expenditure for health care related to cannabis was $23.3 million, which is slightly more than the $20 million for the status quo alternative while the mean expenditure for CJS was $51.1 million, which is 14% less than the $60 million in the status quo alternative.

The attribute coefficient on location of purchase was positive and highly significant. This is despite the fact that it could only be chosen when the legalised option appeared. Although collinearity between the policy and the location of purchase attributes cannot be ruled out, this finding would suggest a strong desire that cannabis should be purchased in a legal market, away from the illicit drug dealers, and with the knowledge of the potency of the product. Separation of cannabis from other illicit markets and knowledge of the potency of the product are reasons often given for legalising cannabis (van Dijk, 1998; Wodak et al., 2002; Rolles, 2009).

The significant and positive constant on the unlabelled utility functions reflects a negative view of the status quo. While it was necessary to label the status quo, the apparent negative connotations attached to this label reaffirmed the choice to use otherwise unlabelled alternatives. Words like legalisation and criminalisation would no

260 Chapter 10

doubt have driven choices either for or against regardless of the levels of the attributes. When scenarios were constructed for each of the policy levels, the utility score for cannabis cautioning, which is comprised of the status quo plus the constant, was considerably less negative than that of the status quo. This suggests that there is some unexplained variance for the status quo. This difference could not be explained by jurisdiction of residence. A dummy variable was constructed to assess whether preferences were different for respondents residing in a jurisdiction with a civil fine program versus cannabis cautioning. Location of residence was not included in the final model as it never approached significance in earlier models. As most people are not familiar with the cannabis laws in their jurisdiction (Pacula et al., 2004; Fetherston and Lenton, 2005) the lack of significant finding is not surprising.

The interactions between personal characteristics and the random parameters further demonstrated that different groups had different preferences. Those who believe cannabis has some health benefits and those who had consumed cannabis recently, dislike the criminal status and civil penalty policy more than the average and they had a stronger preference for the legalisation policy. On the other hand, those who believe cannabis is usually addictive had a greater preference for the criminal status, and civil penalty policy but less than an average preference for legalisation.

One surprising finding may be the significance of the political left–right scale. There are some clear findings, those on the right have increased preferences for criminalisation and those on the left tend to prefer legalisation. These findings can be argued both as predicable and as unpredictable. On one hand, if the right are those who argue for freedom of choice as do many economists, including Friedman and 500 other economists in a letter written to the US President encouraging him to legalise cannabis (Marijuana Policy Project, 2006) then the findings are not as one expected. Alternatively, if those who indicate they are on the right are concerned with family values and moral regulation (McKnight, 2005) then the strong positive association between the criminalisation policy and being right on the political scale is consistent.

As in any study there are limitations. A number of constraints were placed on the designs. These included the use of the status quo option; the inclusion of the attribute for location of purchase which could only appear with the ‘legalise’ level in the policy attribute; and the lack of priors. These constraints combined meant it was not feasible,

261 Chapter 10

using NGENE or other readily available methods for obtaining designs (SPSS, published, online sources) to generate a design which included interactions between the attributes. While this is obviously a limitation of this study, the coefficients from this study would provide reliable estimates for a future design which might result in a design with interactions. In addition advances in the design software could also enhance the likelihood of finding an improved design. Additional information on each question may have permitted Best Worst Scaling, but as the scenarios were already complex, any attempt to collect even more information might have resulted in a high rate of non completers. Comments from respondents confirmed that they felt the scenarios were complex and repetitive.

In constructing the survey a decision was made to combine the various health harms and burden on the health care system into one attribute expressed in millions of dollars. The burden to the criminal justice system was also expressed in total expenditures. The use of a single attribute for health consequences may not have captured the full dimensions of cannabis use and health concerns. Future work could explore the use of multiple attributes, such as rates of psychosis, accidents, and cannabis dependence.

The unexplained variance around the status quo and the low pseudo R2 suggests that there may be important attributes or personal characteristics missing. For example, work in Chapter 8 and 9 suggest that the impact of cannabis use on education attainment may be an important attribute. Alternatively, although the fit improved both with the addition of the interactions and the move to ML, the model may still be misspecified.

10.6 Conclusion

This chapter is the first known use of DCE in the drugs policy field and applies current practice in the field. It demonstrates that it is a feasible method for eliciting preferences for policies for cannabis and potentially other illicit drugs.

This DCE study using a community sample with random parameters and mixed logit demonstrated that respondents do hold disparate views on the most appropriate policy for cannabis. A considerable amount of the variation in views can be explained by respondent characteristics and attitudes. While the results suggest that there may be an improved model for these data, and more work is necessary on determining the best set

262 Chapter 10

of alternatives, two issues are abundantly clear. There is a distinct lack of preference for completely criminalising cannabis offences and a moderate level of preference for complete legalisation of cannabis. Despite this there is a strong preference for lower rates of cannabis use among young people.

In bringing together complex and inter-sectoral policy issues through the DCE framework the results of this work provide policy analysts with another tool to systematically explore trade-offs between attributes and their levels.

263 Chapter 11

Chapter 11 : Discussion

I cannot admit any easy distinction between promise and achievement; for the admission of promise is recognition of something already there; and every real achievement, in spite of the brevity of life, should be a promise of something further. Not, of course, that this continuing promise is anything but disconcerting to the majority even of the most sympathetic readers; we must all be ready to risk the imputation of having gone too far….. Of course one can ‘go too far’ and except in directions in which we can go too far there is no interest in going at all; and only those who will risk going too far can possibly find out just how far one can go. (Eliot, 1931)

The interest in legalising cannabis waxes and wanes but recently there appears to be a resurgence in interest worldwide (Rolles, 2009; Brion and Associates and Nilsson Consulting, 2010; Macleod and Hickman, 2010; Pease-Watkin, 2010; Werb et al., 2010; Wood et al., 2010). The proposal to legalise cannabis in California at the November 2010 election was narrowly defeated with 53.5% voting against Proposition 19 (California Secretary of State Debra Bowen, 2010). Recently a Bill to make it legal to sell cannabis to those over the age of 21 in Washington State was introduced into the state legislature. It is not expected to pass. One of the drivers for the apparent resurgence of interest, at least in the US, is the potential taxation revenue from cannabis sales. However, as Kilmer and colleagues (2010) point out it is not possible to determine with any certainty the potential magnitude of these revenues.

In general, the debate over legalising cannabis has been hampered both by the lack of evidence and a preferred model of legalisation, although recently several advocates have recommended legalising cannabis within a public health framework (Haden, 2002; Haden, 2008; Rolles, 2009; Wood et al., 2010).

This thesis informs the discussion on cannabis policy within the public health framework, by employing two separate methods to assess the social valuation and societal preferences of two policy options in the context of NSW. Many assumptions were required and include assumptions pertaining to: the frequency of consumption; the likely uptake of cannabis if legalised; health harms; educational attainment; and the frequency of motor vehicle accidents related to cannabis consumption. Additionally, it is essential to bear in mind when considering the overall results that the context is

264 Chapter 11

specific to NSW (or at most Australia). The existing patterns of enforcing laws against cannabis use, of providing treatment for cannabis use disorder, of treating low birth weight newborns and rates of cannabis use are all specific to this work. These activities do not take place separately from the socioeconomic conditions, disposable income, the availability of health care and existing preferences and attitudes in NSW. The results as they are presented are a function of all of these assumptions, the legal and geographical context and most importantly the legalised–regulated alternative constructed for this analysis. Any major change in any of these characteristics may lead to a different cost structure, and different harms or benefits.

Some may suggest given all the assumptions that this work has “gone too far” and others will argue it has “not gone far” enough. Rather than trying to address this issue directly, it is useful to consider what economics brings to the policy-setting agenda. Some argue that economics can provide ‘the answer’ but an economic evaluation only provides input into the solution. Economic methods are often portrayed as non- emotive, rigorous, and comprehensive and thus can provide the answer to challenging problems such as those which arise from illicit drug use. This is not necessarily so. Economic analysis can, and should be rigorous, systematic, clear in its perspective and comprehensive but it does not, nor should it be expected to provide a single right answer (Drummond et al., 2005). This is especially true with complex questions such as what to do about drug policy in general, or cannabis policy specifically. Economic analyses can “aid decision makers – not replace them – by providing insight into the objective dimensions of inherently subjective poky decisions” (Warner, 1991). Further it provides information on the trade-offs involved in making different kinds of social investment (McIntosh, 2010).

In undertaking a CBA it was necessary to sum the costs and the benefits, and to quantify the benefits in a monetary measure, and herein is the challenge. Often so-called CBA are simply a summary of the costs of the programs where important benefits or harms are not included due to difficulty in measurement (Drummond et al., 2005). As a result, such studies are limited in their usefulness.

In the complex area of illicit drugs policy the list of potential inputs, harms and benefits is long (MacCoun and Reuter, 2001a; Hall and Pacula, 2003; Godfrey, 2006). Even once the analyst has determined which benefits and harms to include, estimating the

265 Chapter 11

impact of a policy change and then valuing the change in monetary terms has many challenges. Chapters 4 to 9 provide examples of these challenges where some costs and benefits are estimated with more precision and certainty than others. For example, in order to address the oft-cited claim that there are potentially large savings to the criminal justice system if cannabis were to be legalised, a micro costing of police time was undertaken and demonstrates that while considerable resources are expended (approximately $10 million), it is not as large as others would suggest. A contingent valuation study was conducted to quantify stigma as a result of a criminal record and revealed this is a costly burden ($7.42 million). Other estimates, such as the impact on prevalence rates should cannabis legalisation occur and many of the health consequences were estimated with less precision. With other inputs, such as the rates of schizophrenia and the loss of educational attainment in adolescents due to cannabis use, there continues to be disagreement over whether a causal relationship exists or the extent to which it exists (Fergusson et al., 2002; Fergusson et al., 2005; van Ours and Willams, 2009; Hickman et al., 2010; Horwood et al., 2010; Macleod and Hickman, 2010; McCaffrey et al., 2010). Thus throughout the conduct of the CBA many decisions about what was included and excluded were required – the aforementioned value judgement of the analyst comes into play. No matter how diligent the analyst this is always the case, however, in a good economic analysis all decisions are transparent and defendable.

If the net social benefit (NSB) of the CBA is greater than zero, the policy/program being evaluated is said to be a worthwhile social benefit, and if multiple policies have a NSB greater than zero, the NSB that is the largest is said to be preferred. As reported in Chapter 9, the NSB for NSW for the status quo cannabis policy was $293.5 million (range $207.9 to $379.0) and the NSB for the legalised–regulated model was $228.7 million ($154.2 to $314.5). Both are greater than zero, and given the uncertainty in the estimates, the ranges suggest that these policy alternatives are similar in their overall efficiency of use of society’s resources, notwithstanding all the caveats. However, certain constructs such as individual liberty and human rights were not included. These are difficult if not impossible to value directly. For those who argue that individual liberty should not be infringed upon by the state this omission would be a significant oversight.

266 Chapter 11

Discrete choice experiments, (DCEs), another method of assessing societal preferences are valuable when outcomes or benefits are difficult to value (Drummond et al., 2005). Through the assumption that individuals derive utility not from the policy itself but rather from the attributes of that policy, DCEs allow analyses of the component parts of the policy (Gerard et al., 2008a). It is then possible to begin to understand which characteristics of a policy are important. Additionally, preference heterogeneity between the components and various socio-demographic characteristics can be examined through the use of mixed logit and random parameters. The results from the DCE (Chapter 10) demonstrate that individual respondents hold disparate views on the most appropriate policy for cannabis and some of that variation can be explained by characteristics and attitudes. Relative to the status quo (cannabis cautioning for use/possession and criminal otherwise), there was a distinct lack of preference for completely criminalising cannabis offences. Again relative to the status quo, approximately 60% preferred legalisation and when given the choice between the status quo and civil penalty for use/possession, 83% preferred a civil penalty policy.

The issue of individual liberty and human rights were not addressed directly in the DCE but when respondents selected between the different alternatives their views may have impacted their personal choices. The fit of the model improved with the addition of personal characteristics and the use of a mixed logit model, but there is scope for additional improvements to the model. Understanding individuals’ views towards drug use, liberty and human rights may improve the explanatory power of the model.

Considering these two sets of results together suggests some ambiguity. First, if the revenue to government is excluded, the results from the CBA suggest there is no clear difference between the two policy options and the results from the DCE suggests that the majority prefer civil penalties for possession of a small amount of cannabis and criminal otherwise. However, the DCE also suggests that 64% of the respondents prefer legalisation to cannabis cautioning and when the transfers to government (revenues) are included in the CBA the result is the largest NSB. These results suggest several choices: maintain the status quo, move to civil penalties or legalise and regulate cannabis. Analysis of the Western Australia civil penalty scheme has demonstrated the prevalence of cannabis use has not increased and there are some cost savings to the criminal justice system with the introduction of a civil penalties program (Crime Research Centre, 2007) but the wider costs and benefits have not been assessed for this 267 Chapter 11

policy so the overall societal impact is not known. However, it is unlikely they will be significantly different from the status quo given cannabis use has not increased.

Maintaining the status quo will result in: police and criminal justice resources being consumed enforcing cannabis laws; individuals becoming stigmatised both socially and economically from criminal records from cannabis offences; and individuals not legally permitted to consume a drug which has been consumed since the dawn of recorded history (Hill and Newlin 2002 in Babor et al., 2010). Conversely, if the decision was made to adopt legalisation there will be a vigorous and potentially uninformed debate. Possibly, the most damaging argument against legalisation is an issue not dealt with in this thesis. The existing United Nations Convention Against Illicit Traffic of 1988 stipulates that all signatories, of which Australia is one, establish a criminal offence for the possession of drugs, including cannabis, for the purposes of trafficking, and for the possession for personal consumption (European Monitoring Centre for Drugs and Drug Addiction, 2007). Moreover, there will be an unpredictable, but a possible 35% increase in the prevalence of cannabis use with associated harms. On the positive side those who wish to consume cannabis and who are at least 21 years of age may do so legally with the knowledge that the cannabis they consume is uncontaminated and of a known potency. These are just some of the trade-offs and up to now the lack of quantification of these trade-offs have been sufficient to maintain the status quo.

Some will argue that the addition of government revenues (minus the costs of running the retail operation and the payments to the growers) results in a NSB larger than the other two results. Technically government revenues are not benefits. If revenues are treated as benefits in the move from an illicit drug market there remains considerable uncertainty around the potential quantum of government revenues (See Chapter 9 and Kilmer et al., 2010). This uncertainty is driven by the lack of information on actual consumption, the actual price, and the retail model. The central reason for the uncertainty is the lack of information on the resources required to dismantle or at least inhibit the current black market.

The legalised–regulated alternative was built on the premise of a government monopoly, price control at current prices, licensed consumers, contracted suppliers and considerable other regulations. Without these controls the costs and benefits might be very different as the market may be oversupplied, and prices may fall with a subsequent

268 Chapter 11

additional increase in demand. Anti-competition laws may prohibit the introduction of a monopoly arrangement in some countries but there are examples of government monopolies in Australia such as the national broadband network which is currently under construction. It would take strong leadership and the support of agencies such as the Australian Competition and Consumer Commission (ACCC) to ensure such a model was permitted with the primary motivation of promoting public health. The tobacco and alcohol industries provide many examples of why the responsibility of retailing cannabis should be retained by government or a non-profit organisation (Callard et al., 2005; World Health Organization, 2005; Freeman et al., 2008; Cancer Council Victoria, 2010). For those interested in a legalised–regulated model a detailed study of existing anti-competition laws is necessary. Every nation will have its own laws and a detailed understanding of what might be permitted is required prior to discussion of a regulatory structure being developed to mediate public health harms.

Importantly, legalising cannabis will not resolve all problems relating to cannabis as “(p)olicy can minimize the damage drugs cause and influence what sort of drug problems exist, but it does not allow a society to choose to be completely free of drug problems” (Kleiman, 1992 in Babor et al., 2010). This brings us to an obvious and significant exclusion in the analyses in this thesis – there are no estimates on the impact on consumption of alcohol or other illicit drugs should cannabis become legalised. It was beyond the scope of this work. Currently the literature is in disagreement as to whether alcohol and cannabis are substitutes or complements and, even if this was known with certainty, it is not known whether the current relationship would persist if cannabis was legalised.

Additionally, more information on the relationship between tobacco use and cannabis is essential. Over 60% of Australians who consume cannabis mix it with tobacco. Given the potential health harms, including high rates of dependence from tobacco, this is an important area of research. Questions include, do those who start using cannabis without previously smoking tobacco add tobacco to their joints at the same rate as those who also smoke tobacco? Does tobacco in a joint result in tobacco dependence? And, as tobacco is often used to improve burning and to extend the cannabis in a joint (Knight et al., 2010; van Beurden et al., date unknown) is there something else less harmful that might be substituted? Answers to these questions would improve not only the economic model but also be key inputs into public health messages. 269 Chapter 11

The question of whether cannabis has a causal impact on lifetime educational attainment has yet to be definitively resolved. Several studies suggest that cannabis use has a significant and negative impact on education while others suggest another factor is the cause (van Ours and Willams, 2009; Horwood et al., 2010; McCaffrey et al., 2010). All three of these studies, using different methods and data sources, suggest that those who consume cannabis as adolescents have on average lower educational attainment. The studies differ on whether cannabis is the cause. Given the acute effects of cannabis on the brain, and the potentially lasting negative impact on the developing brain, it is highly conceivable that frequent use of cannabis will negatively impact educational achievement as well as social development. For these reasons the legal age of consumption in this model was 21 and considerable resources were allocated to enforcing those regulations.

There are studies which document the relationship between cannabis and negative health outcomes but much remains unknown about the relationship between the dose and duration of use and health outcomes. For example, does the risk of acquiring a negative health outcome decrease or disappear once consumption ceases? Or what is the cumulative effect of long-term cannabis use which becomes more common with legalisation? What are the long-term health care resource implications of longer cannabis careers? Additional research is also required on effectiveness of medical cannabis. This research would be easier to conduct if cannabis were legalised. This is an unquantified benefit of legalisation.

Finally, having undertaken a static economic analysis, there is considerable scope for a more sophisticated model. Such a model should include information on cycling in and out of use, the lifetime impact on educational attainment, and forgone wages, life time health burden including repeated treatments for CUD, and lifetime impact of schizophrenia where it is attributable to cannabis use. For such a dynamic model to be estimated considerable information is required.

There is also scope for refinement of the discrete choice tool which can be used to clarify preferences through the inclusion of several attributes such as educational attainment, impact on life time earnings, and refinement of health harms.

270 Chapter 11

This thesis set out to explore, using economic tools, whether there was a difference in the costs and benefits of different policies for cannabis. The policies and their potential outcomes investigated in this thesis are complex and contentious. The complexity of the policies and the often conflicting outcomes can expose such methods to the criticism of having overlooked significant factors. The discussions in previous chapters underscored the point that it is unlikely all the costs and benefits can ever be included. Nor will there likely be agreement on how potential harms and benefits should be valued. There is, however, still research to be done, some of which cannot be undertaken until a jurisdiction legalises cannabis. This thesis has used several methods not widely used in drug policy analyses, including the contingent valuation study in Chapter 7 and a discrete choice experiment in Chapter 10 to quantify societal benefits and explore preferences for different cannabis policies.

Based on this evaluation, the legalised–regulated option with a framework which included significant regulatory costs plus the imposition of a purchase-licence, and a 35% increase in prevalence of consumption, does not appear to have an NSB different from that of the status quo. Additionally, there is likely to be some gain to government from revenue transfer. Perhaps now is the time that the burden of proof should be shifted to those who want to maintain the status quo (Boyum and Reuter, 2005).

271 References

References

(2009). Plain Tobacco Packaging (Removing Branding from Cigarette Packs) Bill 2009.

Ableson, P. (2000). Public Economics Principles and Practice. Sydney, Applied Economics.

Access Economics. (2002). Schizophrenia: costs - an analysis of the burden of schizophrenia and related suicide in Australia, SANE Australia.

Access Economics. (2008). The health of nations: the value of a statistical life. Canberra, Australian Safety and Compensation Council.

Adams, I and Martin, B. (1996). Cannabis pharmacology and toxicology in animals and humans. Addiction 91: 1585-1614.

Adler, M and Posner, E. (2006). New Foundations of Cost-Benefit Analysis. Cambridge, Massachusetts, Harvard University Press.

Ahern, J, Stuber, J and Galea, S. (2007). Stigma, discrimination and the health of illicit drug users. Drug and Alcohol Dependence. 88: 188-196.

Aldington, S, Harwood, M, Cox, B, Weatherall, M, Beckert, L, Hansell, A, Pritchard, A, Robinson, G and Beasley, R. (2008). Cannabis use and risk of lung cancer: a case-control study. Eur Respiriratory Journal. 31: 280-286.

Amaya-Amaya, M, Gerard, K and Ryan, M. (2008). Discrete Choice experiments in a nutshell. Using Discrete Choice Experiments to Value Health and Health Care Ryan, M, Gerard, K and Amaya-Amaya, M., Dordrecht, The Netherlands, Springer.

Amos, A, Wiltshire, S, Bostock, Y, Haw, S and McNeill, A. (2004). 'You can’t go without a fag . . . you need it for your hash’—a qualitative exploration of smoking, cannabis and young people. Addiction. 99: 77-81

Anderson, P, de Bruijn, A, Angus, K, Gordon, R and Hastings, G. (2009). Impact of Alcohol Advertising and Media Exposure on Adolescent Alcohol Use: A Systematic Review of Longitudinal Studies. Alcohol and Alcoholism .44: 229- 243.

Andrews, G and Tolkien II Team. (2006). Tolkien II: a needs-based, costed stepped- care model for mental health services. Sydney, World Health Organisation Collaborating Centre for Classification in Mental Health.

Angel Gutierrez, M. (2009). Mexico Senate OKs bill to legalize drug possession. Reuters.

272 References

Anonymous. (2010). Cost Sheet for Cannabis Growing Operation. Sydney.

Anthony, J, Warner, L and Kessler, R. (1994). Comparative epidemiology of dependence on tobacco, alcohol, controlled substances, and inhalants: Basic findings from the National Comorbidity Survey. Experimental and Clinical Psychopharmacology. 2: 44-268.

Arrow, K, Solow, R, Portney, P, Leamer, E, Radner, R and Schuman, H. (1993). Report of the NOAA Panel on Contingent Valuation.

Ashton, C. (2001). Pharmacology and effects of cannabis a brief review. British Journal of Psychiatry. 178: 100-106.

Atkinson, G, Healley, A and Mourato, S. (2005). Valuing the cost of violent crime: a stated preference approach. Oxford Economic Papers. 57: 559-585.

Attorney General and Minister for Justice. (2008). Budget Papers 2007-08. Sydney, NSW Government

Australian Bureau of Statistics. (2008). Consumer Price Inde Australia. Statistics, Australian Bureau of Statistics. Canberra, Commonwealth Government. 6401.0.

Australian Bureau of Statistics. (1990). 4120.0 - Disability and Handicap, Australia, 1988. Retrieved 1 September, 2009, from http://www.abs.gov.au/ausstats/[email protected]/b06660592430724fca2568b5007b86 19/1618215db8a5c6abca25692d0083019d!OpenDocument.

Australian Bureau of Statistics. (2006a). Employee Earnings and Hours, Australia, May 2006 Canberra, Australian Bureau of Statistics. Commonwealth Government.

Australian Bureau of Statistics. (2006b). NSW 2006 Census Data. Canberra, Australian Bureau of Statistics. Commonwealth Government.

Australian Bureau of Statistics. (2008). National Survey of Mental Health and Wellbeing: Summary of Results. Australian Bureau of Statistics, Canberra. Commonwealth Government.

Australian Bureau of Statistics. (2010). 6202.0 - Labour Force, Australia Canberra Commonwealth Government of Australia.

Australian Competition and Consumer Commission. (2008). Report of the ACCC inquiry into the competitiveness of retail prices for standard groceries. Canberra, Australian Competition and Consumer Commission.

Australian Crime Commission. (2007). Illicit Drug Data Report 2005-06: Cannabis. Canberra, Commonwealth of Australia.

Australian Crime Commission. (2008). Illicit Drug Data Report 2007-08: Cannabis. Canberra, Commonwealth of Australia.

Australian Customs and Border Protection Service. (2009). Annual Report 2008–09. Canberra, Australian Commonwealth Government.

273 References

Australian Customs Media. (2007). Customs stems unprecedented flood of smuggled cigarettes and tobacco 2008.

Australian Customs Media. (2008). Customs smokes out tobacco smugglers. Canberra, Australian Commonwealth Governement.

Australian Government Productivity Commission. (2007). Report on Government Services 2006 Court Administration. Canberra

Australian Institute of Health and Welfare. (2005). 2004 National Drug Strategy Household Survey. Detailed findings. Drug Statistics. Canberra, Australian Institute of Health and Welfare.

Australian Institute of Health and Welfare. (2008a). 2007 National Drug Strategy Household Survey. Detailed Findings Drug Statistics Series No 22. Canberra, Australian Institute of Health and Welfare.

Australian Institute of Health and Welfare. (2008b). 2007 National Drug Strategy Household Survey. First Results Drug Statistics Series No 20. Canberra, Australian Institute of Health and Welfare.

Australian Institute of Health and Welfare. (2008c). Alcohol and other drug treatment services NMDS specifications 2006-07:data dictionary, collection guidelines and validation processes. Canberra, Australian Institute of Health and Welfare. 54.

Australian Institute of Health and Welfare. (2009). Alcohol and other drug treatment services in Australia 2007-08: report on the national minimum data set. Drug Treatment Series Number 9 Cat. no. HSE 73. Canberra, AIHW.

Australian Productivity Commission. (1999). Australia's gambling industries.Inquiries Report. Canberra.

Australian Taxation Office. (2009a). Frequently asked questions for distillation equipment (stills). Feb, 2009. from http://www.ato.gov.au/businesses/content.asp?doc=/content/66216.htm&page=2 &H2=&pc=001/003/043/003/004&mnu=12124&mfp=001&st=&cy=1.

Australian Taxation Office (2009b) Illicit tobacco crops seized in NSW Southern Highlands. Canberra,

Australian Taxation Office. (2010, July 1, 2010). Excise Tarriff Working Pages. from http://law.ato.gov.au/atolaw/view.htm?Docid=PAC/BL030002/1&PiT=9999123 1235958. Australian Commonwealth Government.

Babor, T, Caetano, R, Casswell, S, Edwards, G, Giesbrecht, N, Graham, K, Grube, J, Gruenewald, P, Hill, L, Holder, H, Homel, R, Osterberg, E, Rehm, J, Room, R and Rossow, I. (2003). Alcohol: No Ordinary Commodity. Oxford, Oxford University Press.

274 References

Babor, T, Caulkins, J, Edwards, G, Fischer, B, Foxcroft, D, Humphreys, K, Obot, I, Rehm, J, Reuter, P, Room, R, Rossow, I and Strang, J. (2010). Drug Policy and the Public Good. Oxford, Oxford University Press.

Baker, J. (1998). Juveniles in Crime - Part 1: Participation rates and risk factors. Sydney, NSW Bureau of Crime Statistics and Research. Sydney, NSW Government.

Baker, J and Goh, D. (2004). The cannabis cautioning scheme three years on: An implementation and outcome evaluation. Sydney, Bureau of Crime Statistics and Research. Sydney, NSW Government.

Ballotta, D, Bergeron, H and Hughes, B. (2008). Cannabis control in Europe. A Cannabis Reader: Global Issues and Local Experiences. Rödner Sznitman, S, Olsson, B and Room, R. Lisbon, European Monitoring Centre for Drugs and Drug Addiction. Monograph Series 8.

Baltussena, R, Stolka , E, Chisholm, D and Aikins, M. (2006). Towards a multi-criteria approach for priority setting: an application to Ghana. Health Economics. 15: 689-696.

Banks, G. (2003). The good, the bad and the ugly: economic perspectives on regulation in Australia. Conference of Economists, Business Symposium. Hyatt Hotel, Canberra.

Bartels, R, Fiebig, D and va Soest, A. (2006). Consumers and experts: an econometric analysis of the demand for water heaters. Empirical Economics. 31: 369-391.

Bates, S. (2004). The Economic Implications of Marijuana Legalization in Alaska. Fairbanks, Alaskans for Rights & Revenues.

Baum, F. (2008). The New Public Health. Oxford, Oxford University Press.

Bech, M and Gryrd-Hansen, D. (2005). Effects coding in discrete choice experiments. Health Economics. 14: 1079-1083.

Ben Akiva, M and Lerman, S. (1985). Discrete Choice Analysis: Theory and Applications to Travel Demand. Cambridge, MIT Press.

Bennett, T, Holloway, K and Farrington, D. (2008). The statistical association between drug misuse and crime: a meta-analysis. Aggression and Violent Behavior. 13: 107-118.

Biecheler, M, Peytavin, J, The SAM Group., Facy, F and Martineau, H. (2008). SAM survey on “drugs and fatal accidents”: search of substances consumed and comparison between drivers involved under the influence of alcohol or cannabis. Traffic Injury Prevention. 9: 11-21.

Black, E, Roxburgh, A, Degenhardt, L, Bruno, R, Campbell, G, de Graaff, B, Fetherston, J, Kinner, S, Moon, C, Quinn, B, Richardson, M, Sindicich, N and White, N. (2007). Australian Drug Trends 2007 Findings from the Illicit Drug

275 References

Reporting System (IDRS). Australian Drug Trends Series No. 1. Sydney, NDARC.

Bleeker, A and Malcolm, A. (2002). Mulling it Over: Health Information for People who Use Cannabis. Manly, Manly Drug Education and Counselling Centre.

Bluelight. (2010). Cannabis discussion: spliffs versus joints. vBulletin® v3.7.4, Retrieved December 21, 2010, from http://www.bluelight.ru/vb/archive/index.php/t-426925.html.

Boadway, R and Bruce, N. (1984). Welfare Economic. Oxford, Basil Blackwell Publisher Limited.

Boardman, A, Greenbery, D, Vining, A and Weimer, D. (2001). Cost-Benefit Analysis: Concepts and Practice. 2nd Edition Upper Saddle River, NJ, Prentice Hall.

Bolesa, S and Miottoa, K. (2003). Substance abuse and violence. A review of the literature. Aggression and Violent Behavior 8: 155-174.

Borghi, J. (2007). Aggregation rules for cost-benefit analysis: a health economics perspective. Health Economics. 17: 863-875.

Bouchard, M. (2007). A capture-recapture derived method to estimate cannabis production in industrialized countries. First Annual Conference of the International Society for the Study of Drug Policy. Oslo, Norway.

Boys, A, Marsden, J and Strang, J. (2001). Understanding reasons for drug use amongst young people: a functional perspective. Health Education Research. 16: 457- 469.

Boyum, D and Reuter, P. (2005). An Analytic Assessment of U.S. Drug Policy. Washington, DC, AEI Press.

Brand, S and Price, R. (2000). The economic and social costs of crime. Home Office. London, Home Office. Research Study 217.

Brand, S and Price, R. (2001). The economic and social costs of crime. Home Office. London.

Brazier, J, Ratcliffe, J, Salomon, J and Tsuchiya, A. (2007). Measuring and Valuing Health Benefits for Economic Evaluation. Oxford, Oxford University Press.

Bretteville-Jensen, A. (1999). Gender, heroin consumption and economic behaviour. Health Economics and Econometrics. 8: 379-389.

Bretteville-Jensen, A and Biorn, E. (2003). Heroin consumption, prices and addiction: Evidence from self-reported panel data. Scandinavian Journal of Economics. 105: 661-679.

Brion and Associates and Nilsson Consulting. (2010). Industrial Medical Cannabis Cultivation & Manufacturing Facility Economic Analysis. City of Oakland, Alameda County.

276 References

Brion, J. (2010). Big Oakland grow = big bucks. Oakland. Personal communication

Brooks, A, Stothard, C, Moss, J, Christie, J and Ali, R. (1999). Costs Associated with the Operation of the Cannabis Expiation Notice Scheme in South Australia. DASC Monograph. Adelaide, Drug and Alcohol Services Council.

Brouwer, W, Culyer, A, van Exel, N and Rutten, F. (2008). Welfarism vs. extra- welfarism. Journal of Health Economics. 27: 325-338.

Bureau of Crime Statistics and Research. (2006a). Court Statistics. Sydney, NSW Government Law Link.

Bureau of Crime Statistics and Research. (2006b). Recorded Crime Statistics, NSW Goverment Law Link Sydney.

Burns, L, Mattick, R and Cooke, M. (2006). The use of record linkage to examine illicit drug use in pregnancy. Addiction. 101: 873-882.

Burstein, P. (2003). The impact of public opinion on public policy: a review and an agenda. Political Research Quarterly .56: 29-40.

Buxton, J, Haydon, M and Mathias, R. (2008). The control and regulation of currently illegal drugs. International Encylopedia of Public Health. 2: 7-16.

Caetano, R. (2006). A ‘perfect storm’ in developing countries: economic growth and the alcohol industry. Addiction.101: 149-152.

Calabria, B, Degenhardt, L, Hall, W and Lynskey, M. (2008). Cannabis-related mortality. Illicit drugs discussion paper. No 3. Group, Global Burden of Disease Mental Disorders and Illicit Drug Use Expert Group. Sydney, NDARC, Sydney. University of New South Wales..

California Secretary of State Debra Bowen. (2010). Official declaration of the vote results on November 2, 2010 General Election - State Ballot Measures. from http://www.sos.ca.gov/elections/sov/2010-general/07-for-against.pdf.

Callard, C, Thompson, D and Collishaw, N. (2005). Transforming the tobacco market: why the supply of cigarettes should be transferred from for-profit corporations to non-profit enterprises with a public health mandate. Tobacco Control .14: 278- 283.

Cameron, L and Williams, J. (2001). Cannabis, alcohol and cigarettes: substitutes or complements? The Economic Record. 77: 19-34.

Cancer Council Victoria. (2010). Tobacco in Australia: facts and issues a comprehensive online resource. Melbourne.

Cancer Council Victoria. (2010). Tobacco in Australia: Facts and issues a comprehensive online resource. Melbourne.

Carlos Hildago, J. (2009). Argentina decriminalizes personal drug consumption. CATO@Liberty. Accessed 2010.

277 References

Carlsson, F, Frykblom, P and Liljenstolpe, C. (2003). Valuing wetland attributes – an application of choice experiments. Ecological Economics. 47: 95-103.

Carr, VJ, Neil, AL, Halpin, SA, Holmes, S and Lewin, TJ. (2003a). Costs of schizophrenia and other psychoses in urban Australia: findings from the low prevalence (psychotic) disorders study. Australian New Zealand Journal of Psychiatry. 37: 31-40.

Cartwright, W. (2000). Cost-benefit analysis of drug treatment services: review of the literature. Journal of Mental Health Policy and Economics. 3: 11-26.

Caulkins, J. (2010a). Cost of marijuana prohibition on the California justice system. Working Paper Santa Monica, RAND Drug Policy Research Centre. WR-763- RC.

Caulkins, J. (2010b). Estimated cost of production for legalised cannabis. Working Paper Series. Santa Monica, RAND Drug Policy Research Centre. WR-764-RC.

Caulkins, J and Kleiman, M. (2007). Drug Policy. Understanding America: The Anatomy of an Exceptional Nation. Schuck, P and Wilson, J. New York, Public Affairs.

Caussade, S, de Dios Ortuzar, J, Rizzi, L and Hensher, D. (2005). Assessing the influence of design dimensions on stated choice experiments. Transportation. Research Part B 39: 621-640.

Centers for Disease Control and Prevention. (2007). Best Practices for Comprehensive Tobacco Control Programs (reprinted with corrections). Atlanta, U.S. Department of Health and Human Services.

Centre for Epidemiology and Research. (2009). New South Wales School Students Health Behaviours Survey: 2008 Report. Sydney, NSW Department of Health.

Cervantes, J. (2002). Indoor Marijuana Horticulture. The Indoor Bible. Self published

Chabrol, H and Saint-Martin, C. (2009). Cannabis use and delinquint behaviors in high- school students Addictive Behaviors. 34: 187-189.

Chalmers, J, Bradford, D and Jones, C. (2010). The effect of methamphetamine and heroin price on polydrug use: A behavioural economics analysis in Sydney, Australia. International Journal of Drug Policy. 21: 381-389.

Champ, P and Welsh, M. (2006). Survey methodologies for stated choice studies. Valuing Environmental Amenities Stated Choice Methods. A Common Sense Approach to Theory and Practice. Bateman, I. Dordrecht, The Netherlands, Springer.

Chanteloup, F, Lenton, S, Barratt, M and Fetherston, F. (2005). Effects of the Western Australian Cannabis Infringement Notice Scheme on regular cannabis users regarding attitudes, use, and drug market factors – Baseline, Year 1. Perth, NDRI.

278 References

Chapman, S and Freeman, B. (2009). Regulating the tobacco retail environment: beyond reducing sales to minors. Tobacco Control. 18: 496-501.

Chapman, S and Freeman, B. (2010). The cancer emperor’s new clothes: personal view. British Medical Journal. 340: 1035.

Chen, C, Storr, C and Anthony, J. (2009). Early-onset drug use and risk for drug dependence problems. Addictive Behaviors. 34: 319-322.

Chief Magistrates Office. (2008). Local Court of New South Wales Annual Review 2007. Sydney Attorneys General Department, NSW Government.

Chikritzhs, T, Dietze, P, Allsop, S, Daube, M, Hall, W and Kypri, K. (2009). The “alcopops” tax: heading in the right direction. Medical Journal of Australia. 190: 294-295.

Chilton, S, Covey, J, Jones-Lee, M, Loomes, G and Metcalf, H. (2004). Valuation of health benefits associated with reductions in air pollution. Final report. London, Department for Environment, Food and Rural Affairs.

Clarke, R and Watson, D, Eds. (2002). Botany of natural cannabis medicines in cannabis and cannabinoids: pharmacology, toxicology, and therapeutic potential. New York, Haworth Press.

Clements, K. (2004). Three facts about marijuana prices. The Australian Journal of Agricultural and Resource Economics. 48: 271-300.

Clements, K and Daryal, M. (1999). The economics of marijuana consumption. unpublished, University of Western Australia. Perth.

Clements, K and Zhao, X. (2005). Economic Aspects of Marijuana, The University of Western Australia.

Clements, K and Zhao, X. (2009). Economics and Marijuana. Consumption Pricing and Legalisation. Cambridge, Cambridge University Press

Cohen, M. (2000). Measuring the costs and benefits of crime and justice. Criminal Justice Review.4: 263 -315.

Cohen, M. (2007). Valuing crime control benefits using stated preferences approaches. Nashville, Vanderbilt University, Owen Graduate School of Management.

Cohen, M, Rust, R, Steen, S and Tidd, S. (2004). Willingness-to-pay for crime control programs. Criminology. 42: 89-109.

Cohen, M, Solowij, N and Carr, V. (2008). Cannabis, cannabinoids and schizophrenia: integration of the evidence. Australian New Zealand Journal of Psychiatry 42: 357-368.

Cole, J, Goudie, A, Field, M, Loverseed, A and Charton, S. (2008). The effects of perceived quality on the behavioural economics of alcohol, methamphetamine,

279 References

cannabis, cocaine, and ecstasy purchases Drug and Alcohol Dependence 94: 183-190.

Collins, D and Lapsley, H. (2008). The costs of tobacco, alcohol and illicit drug abuse to Australian society in 2004/05. Monograph Series. Strategy, ND. Canbera Australian Government Department of Health and Ageing.

Commonwealth of Australia. (2006). Handbook of Cost Benefit Analysis. Canberra

Communities Office of Liquor Gaming and Racing. (2010). Underage drinking offences. Retrieved November 28, 2010, from http://www.olgr.nsw.gov.au/youth_fines.asp.

Communities Office of Liquor Gaming and Racing. (2010). Selling alcohol online, by phone, fax or mail order. Retrieved August 2, 2010, from http://www.olgr.nsw.gov.au/liquor_home.asp.

Communities Office of Liquor, Gaming and Racing. (2010). RSA - Responsble Service of Alcohol. Retrieved August 2, 2010, from http://www.olgr.nsw.gov.au/courses_rsa.asp.

Cornelius, M, Taylor, P, Geva, D and Day, N. (1995). Prenatal tobacco and marijuana use among adolescents: effects on offspring gestational age, growth, and morphology. Pediatrics 95: 738-743.

Crancer, A and Crancer, A. (2010). The Involvement of Marijuana in California Fatal Motor Vehicle Crashes 1998 -2008. Unpublished. http://druggeddriving.org/pdfs/CAMJStudyJune2010.pdf

Crime Research Centre. (2007). WA Diversion Program – Evaluation Framework (POP/STIR/IDP) Final Report for the Drug and Alcohol Office. Perth University of Western Australia.

Crime., Aio. (2004). Young people and crime. Sentencing options by state and territory. Canberra, Australian Government.

Cunningham, C, Deal, K, Rimas, H, Buchman, D, Gold, M, Sdai-Jarvie, K and Boyle, M. (2008). Modeling the information preferences of parents of children with mental health problems: a discrete choice conjoint experiment. Journal of Abnormal Child Pyschology 36: 1123-1138.

Curtis, A, Clarke, C and Rickards, H. (2009). Cannabinoids for Tourette's Syndrome. Cochrane Database of Systematic Reviews. Issue 4. Art. No.: CD006565.

Daley, M. (2010). Minister warns drug drivers you will be caught. Police, New Southth Wales Minster for Police. Sydney, NSW Government.

Daryal, M. (1999). Prices, legalisation and marijuana consumption. Perth, Economic Research Centre, Department of Economics,The University of Western Australia.

280 References

Daryal, M. (2002). Prices, Legalisation and Marijuana Consumption. Perth, Economics Department, University of Western Australia.

De Bekker-Grob, E, Ryan, M and Gerard, K. (2010 ). Discrete choice experments in health economics: a review of the literature. Health Economics Online. http://onlinelibrary.wiley.com/doi/10.1002/hec.1697/abstract

Decorte, T. (2010). Small scale domestic cannabis cultivation: an anonymous Web survey among 659 cannabis cultivators in Belgium. Contemporary Drug Problems 37: 341-370.

Degenhardt, L, Calabria, B, Hall, W. and Lynskey, M. (2008). Overview of injuries/diseases to be included in the comparative risk assessment for regular cannabis use Illicit Drugs Discussion Paper No. 5 Global Burden of Disease Mental Disorders and Illicit Drug Use Expert Group. Sydney, National Drug and Alcohol Research Centre, University of New South Wales.

Degenhardt, L, Lynsky, M and Hall, W. (2000). Cohort trends in the age of initiation of drug use in Australia. Australian and New Zealand Journal of Public Health. 24: 421-426.

Department of Arts Sport and Recreation. (2009). Review of Operations Department of the arts, sport and recreation. Annual report 2007–08. Sydney, Government of New South Wales.

Department of Corrective Services. (2008). Annual Report 2007/08. Sydney, NSW Government.

Department of Justice. (2009). Becoming a poppy grower. Retrieved August 8, 2010, from http://www.justice.tas.gov.au/poppy/becoming_a_grower.

Department of Juvenile Justice. (2007). NSW Department of Juvenile Justice Annual Report 2006-07. Sydney NSW Government.

Department of Juvenile Justice. (2008). Department of Juvenile Justice Financials. from http://www.treasury.nsw.gov.au/__data/assets/file/0017/11474/bp3_14jjust_f.rtf.

Diener, A, O'Brien, B and Gafni, A. (1998). Health care contingent valuation studies: a review and classification of the literature. Health Economics 7: 313-326.

Donnelly, N, Hall, W and Christie, P. (1999). Effects of the cannabis expiation notice scheme on levels and patterns of cannabis use in South Australia: evidence from the National Drug Strategy Household Surveys 1985-1995. Monograph Series National Drug Strategy. Canberra, Department of Health and Aging.

Drug and Alcohol Office. (2007). Statutory Review of Cannabis Control Act 2003. Report to the Minister for Health. Technical Report. Perth, Western Australia, Drug and Alcohol Office .

281 References

Drug Free Australia. (2010). Analysis of KPMG evaluation of the Sydney Medically Supervised Injecting Centre. Sydney, Drug Free Australia.

Drug Policy Alliance. (2010). Californians head to polls for historic vote on marijuana legalization. Retrieved December 11, 2010, from http://www.drugpolicy.org/dpaction/pressrelease/pr110210b.cfm.

Drummer, O, Gerostamoulous, D, Chu, M, Swann, P, Boorman, M and Cairns, I. (2007). Drugs in oral fluids in randomly selected drivers. Forensic Science International. 107: 105-110.

Drummond, C. (2002). Cannabis control: costs outweigh the benefits: against. British Medical Journal. 324: 107-108.

Drummond, M, Sculpher, M, Torrance, G, O'Brien, B and Stoddart, G. (2005). Methods for the Economic Evaluation of Health Care Programmes, Third Edition. Oxford, Oxford Medical Publications.

Drummond, M, Weatherly, H, Claxton, K, Cookson, R, Ferguson, B, Godfrey, C, Rice, N, Sculpher, M and Sowden, A. (2007). Assessing the Challenges of Applying Standard Methods of Economic Evaluation to Public Health Interventions. York, Public Health Research Consortium, University of York.

Eberth, B, Watson, V, Ryan, M, Hughes, J and Barnett, G. (2009). Does one size fit all? Investigating heterogeneity in men's preferences for benign prostatic hyperplasia treatment using mixed logit analysis. Medical Decision Making 29: 707-715.

El Sohly, M. (2002). Chemical constituents of cannabis Cannabis and Cannabinoids: Pharmacology Toxicology, and Therapeutic Potential. Grotenhermen, F and E., R. New York: 27–36.

El Sohly, M. (2008). Quarterly report potency monitoring project. University of Missiippi, National Institute on Drug Abuse.

Elders Real Estate. (2010). Elders Rural Property Album. Retrieved 2010, from http://www.eldersrealestate.com.au/rural.php.

Elliot, TS. (1929). Preface to Transit of Venus: Poems Crosb, H. Paris, Black Sun Press.

Crosby, H. (1931) Transit of Venus: Poems. Paris, Black Sun Press. Preface by T.S. Eliot. Double check the year, as it was originally published in 1929 and then published again as a collection in 1931, from what I can see

Elvik, R. (2001). Cost-benefit analysis of road safety measures: applicability and controversies. Accident Analysis and Prevention. 33: 9-17.

Englesman, E. (2003). Cannabis Control: the model of the WHO tobacco control treaty. The International Journal of Drug Policy. 14: 217-219.

English, D, Hulse, G, Milne, E, Holman, C and Bower, C. (1997). Maternal cannabis use and birth weight: a meta-analysis. Addiction. 92: 1553-1560.

282 References

European Monitoring Centre for Drugs and Drug Addiction. (2007, March 2007). European Legal Database on Drugs. 2010, from http://eldd.emcdda.europa.eu/html.cfm/index5769EN.html.

Fazey, C. (2003). The Commission on Narcotic Drugs and the United Nations International Drug Control Programme: politics, policies and prospect for change. The International Journal of Drug Policy. 14: 155-169.

Fergusson, D, Poulton, R, Smith, P and Boden, J. (2008). Cannabis and psychosis. British Medical Journal. 332: 172-175.

Fergusson, D and Horwood, L. (2001a). Cannabis use and traffic accidents in a birth cohort of young adults. Accident Analysis & Prevention. 33: 703-711.

Fergusson, D, Horwood, L and Ridder, E. (2005). Tests of causal linkages between cannabis use and psychotic symptoms. Addiction. 100: 354-366.

Fergusson, D, Horwood, L and Swain-Campbell, N. (2002). Cannabis use and psychological adjustments in adolescence and young adulthood. Addiction. 97: 1123-1135.

Fergusson, D, Swain-Campbell, N and Horwood, L. (2003). Arrests and convictions for cannabis related offences in a New Zealand birth cohort. Drug and Alcohol Dependence. 70: 53-63.

Fergusson, DM and Horwood, LJ. (2001b). Cannabis use and traffic accidents in a birth cohort of young adults. Accident Analysis & Prevention. 33: 703-711.

Fetherston, J and Lenton, S. (2005). Effects of the Western Australian Cannabis Infrigment Scheme on public attitutes, knowledge, and use. Perth, National Drug Research Institute.

Flynn, T, Louviere, J, Peters, T and Coast, J. (2007). Best-worst scaling: What it can do for health care research and how to do it. Journal of Health Economics. 26: 171- 189.

Food Standards Australian New Zealand. (2010). Food Standards Code. Retrieved August 2, 2010 from http://www.foodstandards.gov.au/

Freeman, B, Chapman, S and Rimmer, M. (2008). The case for plain packaging of tobacco products. Addiction. 103: 580-590.

French, M, Roebuck, M, Dennis, L, Diamond, G, Godley, S, Tims, F, Webb, C and Herrell, J. (2002a). The economic cost of outpatient marijuna treatment for adolecents: findings from a multi-site field experiment. Addiction. Suppl 1: 84- 97.

French, M, Salome, H, Sindelar, J and McLellan, A. (2002b). Benefit-cost analysis of addiction treatment: methodological guidelines and empirical application using the DATCAP and ASI. Health Services Research. 37: 433-455.

283 References

Frew, E. (2010a). Benefit assessment for cost-benefit analysis studies in health care using contingent valuation methods. Applied Methods of Cost-Benefit Analysis in Health Care. McIntosh, E, Clarke, P, Frew, E and Louviere, J. Oxford, Oxford University Press.

Frew, E. (2010b). Benefit assessment for cost-benefit analysis studies in health care: A guide to carrying out a stated preference willingness to pay survey in health care Applied Methods of Cost-Benefit Analysis in Health Care. McIntosh, E, Clarke, P, Frew, E and Louviere, J. Oxford, Oxford University Press.

Friedman, M. (1972). Prohibition and Drugs. Newsweek: 104.

Funk, P. (2004). On the elective use of stigma as a crime-deterrent. European Economic Review. 48: 715-728.

Gallet, G. (2007). The demand for alcohol: a meta-analysis of elasticities. Australian Journal of Agricultural and Resource Economics. 41: 121-135.

Garrod, G, Scarpa, R and Willis, K. (2002). Estimating the benefits of traffic calming on through routes: a choice experiment approach. Journal of Transport Economics and Policy. 36: 211-231.

Geis, G. (2005). Chop-chop: The illegal cigarette market in Australia. Working Paper. Canberra, Centre for Tax System Integrity. 48.

Gerard, K, Ryan, M and Amaya-Amaya, M. (2008a). Introduction. Using Discrete Choice Experiments to Value Health and Health Care Ryan, M, Gerard, K and Amaya-Amaya, M. Dordrecht, The Netherlands, Springer.

Gerard, K, Shanahan, M and Louviere, J. (2008b). Using discrete choice modelling to investigate breast cancer screening. Using Discrete Choice Experiments to Value Health and Health Care. Ryan, M, Gerard, K and Amaya-Amaya, M. Dordrecht, The Netherlands, Springer.

Gettman, J. (2007). Lost taxes and other consequences of marijuana laws. The Bulletin of Cannabis Reform. Issue 4. http://www.drugscience.org/Archive/bcr4/bcr4_index.html

Gieringer, D (2009) Benefits of marijuana legalization in California.

Godfrey, C. (2006). Evidenced based illicit drug policy: the potential contribution of economic evaluation techniques. De Economist. 154: 563-580.

Goh, D and Moffat, S. (2008). New South Wales Recorded Crime Statistics 2007. Statistical Report Series Retrieved December 12, 2008, http://www.bocsar.nsw.gov.au/

Golafshani, N. (2003). Understanding reliablity and validity in qualitative research. The Qualitative Report. 8: 597-607.

Gosten, L. (2009). FDA Regulation of Tobacco. Journal of American Medical Association. 302: 14591460.

284 References

Goudie, A, Sumnall, H, Field, M, Clayton, H and Cole, J. (2007). The effects of price and perceived quality on the behavioural economics of alcohol, amphetamine, cannabis, cocaine and ecstasy purchases. Drug and Alcohol Dependence. 89: 107-115.

Grabosky, P. (1995). Counterproductive regulation. International Journal of Sociology and Law. 23: 347-369.

Gray, A, Clarke, P, Wolstenholme, J and Wordsworth, S. (2011). Applied Methods of Cost-Effectiveness Analysis in Health Care. Oxford, Oxford University Press

Green, C and Gerard, K. (2009). Exploring the social value of health-care interventions: a stated preference discrete choice experiment. Health Economics. 18: 951-976.

Grillo, I. (2009). Mexico's New Drug Law May Set an Example. Time. Mexico City

Grotenhermen, F, Leson, G, Berghaus, G, Drummer, O, Kruger, H, Longo, M, Moskowitz, H, Perrine, B, Ramaekers, J, Smiley, A and Tunbridge, R. (2007). Developing limits for driving under cannabis. Addiction. 102: 1910-1917.

Gyrd-Hansen, D. (2005). Willingness to pay for a QALY: theoretical and methodological issues. Pharmacoeconomics. 23: 423-432.

Haden, M. (2002). Illicit IV Drugs: A public health approach. Canadian Journal of Public Health. 93: 431-434.

Haden, M. (2004). Regulation of illegal drugs: an exploration of public health tools. International Journal of Drug Policy. 15: 225-230.

Haden, M. (2008). Controlling illegal stimulants: a regulated market model. Harm Reduction Journal 5: 1477-7517.

Hall, J, Fiebig, D, King, MT., Hossain, I and Louviere, J. (2006). What influences participation in genetic carrier testing? Results from a discrete choice experiment. Journal of Health Economics 25: 520-537.

Hall, W. (2009). The adverse health effects of cannabis use: What are they, and what are their implications for policy? International Journal of Drug Policy. 20: 458- 466.

Hall, W and Babor, T. (2000). Cannabis use and public health: assessing the burden. Addiction. 95: 485-490.

Hall, W and Degenhardt, L. (2004). Cannabis use and psychotic disorders: an update. Drug & Alcohol Review 23: 433-443.

Hall, W and Degenhardt, L. (2010). What are the policy implciations of the evidence on cannabis use and psychosis? Addiction.. 105: 1332-1333.

Hall, W and Lynsky, M. (2009). The challenges in developing a rational cannabis policy. Current Opinion Psychiatry. 22.

285 References

Hall, W and Pacula, R. (2003). Cannabis Use and Dependence. Public Health and Public Policy. Cambridge, Cambridge University Press.

Hall, W and Swift, W. (2000). The THC content of cannabis in Australia: evidence and implications. Australian New Zealand Journal of Public Health 24: 503-508.

Hanley, N, Wright, R and Adamowicz, V. (1998). Using choice experiments to value the environment. Environmental and Resource Economics 11: 413-428.

Hart, C. (2005). Increasing treatment options for cannabis dependence: a review of potential pharmacotherapies. Drug and Alcohol Dependence 80: 147-159.

Hatch, E and Bracken, M. (1986). Effect of marijuana use in pregnancy on fetal growth. American Journal of Epidemiology 124: 986-993.

Health Officer Council of British Columbia. (2005). A public health approach to drug control in Canada. Discussion paper.

Henderson, M. (2006). Benchmarking Study of Home Detention Programs in Australia and New Zealand. Melbourne, National Corrections Advisory Group.

Hensher, D. (1993). Stated preference analysis of travel choices: the state of practice. Transportation 21: 107-133.

Hensher, D. (2006). How do respondents process stated choice experiments? Attribute considerations under varying information load. . Journal of Applied Econometrics 21: 861-878.

Hensher, D. (2009). Attribute processing and heuristics in choice analysis. First International Conference on Choice Analysis. Harrogate, UK

Hensher, D and Green, W. (2003). The mixed logit: The state of practice. Transportation 30: 133-176.

Hensher, D, Rose, J, de Dios Ortúzar, J and Rizzi, L. (2009). Estimating the willingness to pay and value of risk reduction for car occupants in the road environment. Transportation Research Part A 43 692-707.

Hensher, D, Rose, J and Greene, W. (2005). Applied Choice Analysis. A Primer Cambridge Cambridge University Press.

Hickman, M, Vickerman, P, Macleod, J, Lewis, G, Zammit, S, Kirkbride, J and Jones, P. (2010). If cannabis caused schizophrenia—how many cannabis users may need to be prevented in order to prevent one case of schizophrenia? England and Wales calculations. Addiction. 104 1856-1861.

Hillig, K. (2005). Genetic evidence for speciation in Cannabis (Cannabaceae). Genetic Resources and Crop Evolution: 161-180.

Hoaken, P and Stewart, S. (2003). Drugs of abuse and the elicitation of human aggressive behavior. Addictive Behaviors. 28: 1533-1554.

286 References

Hole, A. (2008). Modelling heterogeniety in patients' preferences for the attributes of a general practitioner appointment. Journal of Health Economics. 27: 1078-1094.

Horwood, L, Fergusson, D, Hayatbakhshm, M, Najman, J, Coffey, C, Patton, G, Silns, E and Hutcheson, D. (2010). Cannabis use and educational achievement: findings from three Australasian cohort studies. Drug and Alcohol Dependence. 110: 247-253.

Hotham, E, Ali, R, White, J and Robinson, J. (2008). Pregnancy-related changes in tobacco, alcohol and cannabis use reported by antenatal patients at two public hospitals in South Australia. Aust N Z J Obstet Gynaecol. 48: 248-254.

Howard, K and Salkeld, G. (2008). Does attribute framing in discrete choice experiments influence willingness to pay? Results from a discrete choice experiment in screening for colorectal cancer. International Society for Pharmacoeconomics and Outcomes Research. 12: 354-363.

Hughes, C and Stevens, A. (2007). The Effects of Decriminalization of Drug Use in Portugal, The Beckley Foundation Drug Policy Programme Briefing Paper 14.

Hunter, N. (2001). Cannabis expiation notice (CENs) in South Australia, 1997 to 2000. Information Bulletin. Adelaide, Office of Crime Statistics, Attorney General’s Department. No 27.

Hurd, YL, Wang, X, Anderson, V, Beck, O, Minkoff, H and Dow-Edwards, D. (2005). Marijuana impairs growth in mid-gestation fetuses. Neurotoxicol Terato.l 27: 221-229.

Hurley, J. (1998). Welfarism, extra-welfarism, and evaluative economIc analysis in the health sector. Health, Health Care and Health Economics. Perspectives on Distribution. Barer, M, Getzen, T, and Stoddart, G., Chichester, Wiley.

Hutcheson, G and Sofroniou, N. (1999). The Mutivariate Social Scientist. London, Sage

Independent Pricing and Regulatory Tribunal. (2010). Legislation. from http://www.ipart.nsw.gov.au/welcome.asp.

Industrial Relations. (2010). NSW Horticultural Industry (State) Award. Sydney, NSW Government. SERIAL C2523.

International Narcotics Control Board. (2009). United Nations Office on Drugs and Crime World Drug Report 2009 New York, Nations Office of Drugs and Crime: 167.

Johannesson, M and Jonsson, B. (1991). Economic evaluation in health care: Is there a role for cost-benefit analysis? Health Policy. 17: 1-23.

Johannesson, M, Jonsson, B and Karlsson, G. (1996). Outcome measurement in economic evaluation Health Economics. 5: 279-296.

Johnson, F and Adamowicz, V. (2010). Valuation and cost-benefit analysis in health and environmental economics. Applied Methods of Cost-Benefit Analysis in

287 References

Health Care. McIntosh, E, Clarke, P, Frew, E and Louviere, J. Oxford, Oxford University Press.

Jones, C, Kypri, K, Moffatt , S, Borzycki, C and Price, B. (2009). The impact of restricted alcohol availability on alcohol-related violence in Newcastle, NSW. Crime and Justice Bulletin. Sydney, Bureau of Crime Statistics and Research. No. !37.

Kalant, H. (2004). Adverse effects of cannabis on health: an update of the literature since 1996. Prog Neuropsychopharmacol Biol Psychiatry. 28: 849-863.

Katz, C, Webb, V, Gartin, P and Marshall, C. (1997). The validity of self-reported marijuana and cocaine use Journal of Criminal Justice. 25: 31-41.

Kelly, T. (2009). New drug testing van operational: Press release. Retrieved December 1, 2010, from http://www.police.nsw.gov.au/about_us/?a=160214.

Kerley, G. (2004). Drug Law and Enforcment in Australia and the Netherlands. A Case Study of Comparative Criminal Law. Faculty of Law, Australian National University, Canberra, Unpublished.

Kilmer, B, Caulkins, J, Liccardo Pacula, R, MacCoun RJ. and Reuter, P. (2010). Altered State? Assessing how marijuan legalization in California could influence marijuana consumption and public markets. Santa Monica, RAND Drug Policy Research Centre.

Kilmer, B and Liccardo Pacula, R. (2009). Estimating the size of the global drug market. A demand side approach. Report 2. Santa Monica, RAND Corporation.

King County Bar Association. (2005). Effective drug control: Toward a new legal framework. Drug Policy Project. Seatle, King County Bar Association.

King, L, Carpentier, C and Griffiths, P. (2005). Cannabis potency in Europe. Addiction. 100: 884-886.

Kisely, S. (2005). A tale of two jurisidictions. Can Australia and Canada learn from each other's experience with cannabis control? Australian and New Zealand Journal of Psychiatry. 39: 154-160.

Kjaer, T and Gryrd-Hansen, D. (2008). Preference heterogeneity and choice of cardiac rehabilitation program: Results from a discrete choice experiment. Health Policy 2008: 124-132.

Kleiman, M. (1992). Against Excess. New York, Harper Collins.

Knight, G, Hansen, S, Connor, M, Poulsen, H, McGovern, C and Stacey, J. (2010). The results of an experimental indoor hydroponic Cannabis growing study, using the ‘Screen of Green’ (ScrOG) method—Yield, tetrahydrocannabinol (THC) and DNA analysis. Forensic Science International. 202 36-44.

Krishnan, S, Cairns, R and Howard, R. (2007). Cannabinoids for the treatment of dementia. Cochrane Database of Systematic Reviews. 2007 Issue 2.

288 References

Lancsar, E and Louviere, J. (2008). Conducting Discrete Choice Experiments to Inform Healthcare Decision Making. A Users Guide. Pharmacoeconomics. 26.

Laslett, A-M, Catalano, P, Chikritzhs, T, Dale, C, Doran, C, Ferris, J, Jainullabudeen, T, Livingston, M, Matthews, S, Mugavin, J, Room, R, Schlotterlein, M, Wilkinson, C and (2010a). The Range and Magnitude of Alcohol’s Harm to Others. Fitzroy, Victoria: , AER Centre for Alcohol Policy Research, Turning Point Alcohol and Drug Centre, Eastern Health.

Laslett, A-M, Catalano., R, Chikritzhs, T, Dale, C, Doran, C, Ferris, J, Jainullabudee, T, Livingston, M, Matthews, S, Mugavin, J, Room, R, Schlotterlein, M and Wilkinson, C. (2010b). The range and magnitude of alcohol's harms to others Fitzroy, Victoria, AER Centre for Alcohol Policy Research, Turning Point Alcohol and Drug Centre, Eastern Health.

Laumon, B, Gadegbeku, B, Martin, J and Biecheler, M. (2005). Cannabis intoxication and fatal road crashes in France: population based case-control study. British Journal of Medicine. 331: 1371.

Law Commission. (2010). Controlling and Regulating Drugs. Issues Paper 16. Wellington. http://www.lawcom.govt.nz/project/review-misuse-drugs-act- 1975?quicktabs_23=issues_paper

Lawrence, G, Bammer, G and Chapman, S. (2000). ‘Sending the wrong signal’: Analysis of print media reportage of the ACT heroin prescription trial proposal, August 1997. Australian and New Zealand Journal of Public Health. 24: 254- 264.

Legleye, S, Ben Lakhdar, C and Spilka, S. (2008). Two ways of estimating the euro value of the illicit market for cannabis in France. Drug and Alcohol Review. 27: 466-472.

Leigh, A and Ryan, C. (2008). Estimating returns to education using different natural experiment techniques. Economics of Education Revie. 27: 149-160.

Leitzel, J. (2008). Regulating Vice Misguided Prohibitions and Realistic Controls. Cambridge, Cambridge University Press

Lenton, S, Christie, P, Humeniuk, R, Brooks, A, Bennett, M and Heale, P. (1999 ). Infringement versus Conviction: the Social Impact of a Minor Cannabis Offence Under a Civil Penalties System and Strict Prohibition in Two Australian States. Canberra, Commonwealth Department of Health and Aged Care.

Lenton, S and Heale, P. (2000). Arrest, court and social impacts of conviction for a minor cannabis offence under strict prohibition. Contemporary Drug Problems. 27: 807-832.

Lin, V and Gibson, B. (2003). Evidence-Based Health Policy. Problems and Possibilities. Oxford, Oxford University Press.

Lloyd, K and Ford, G. (2008). Submission to ACCC Inquiry into the competitiveness of retail prices for standard groceries. Melbourne, Victorian Farmers Federation.

289 References

Lloyd, L. (2010). Sinning and Sinned Against: The Stigmatisation of Problem Drug Users. London University of York.

Local Government and Shires Associations of New South Wales. (2010). About the Association. Retrieved Sept 8, 2010, from http://www.lgsa.org.au/www/html/88-about-the-associations.asp.

Lott, J. (1992). An attempt at measuring the total monetary penalty from drug convictions: The importance of an individual's reputation. Journal of Legal Studies. 21: 159-187.

Louviere, J, Hensher, D and Swait, J, Eds. (2000). Stated Choice Methods Analysis and Application. Cambridge, Cambridge University Press.

Louviere, J and Lancsar, E. (2009). Choice experiments in health: the good, the bad, the ugly and toward a brighter future. Health Economics, Policy and Law. 4: 527- 546.

Lubman, D, Yücel, M and Hall, W. (2007). Substance use and the adolescent brain: a toxic combination? Journal of Psychopharmacology 21: 792-794.

Lucas, P. (2008). Regulating compassion: an overview of Canada's federal medical cannabis policy and practice. Harm Reduction Journal. 5:5: 1-13.

Ludgwig, J and Cook, P. (2001). The benefits of reducing gun violence: evidence from contingent-valuation survey data. Journal of Risk and Uncertainty. 22: 207-226.

Lutge, E. (2005). The medical use of cannabis for reducing morbidity and mortality in patients with HIV/AIDS. Cochrane Database of Systematic Reviews: Protocols. 2005 Issue 2.

MacCoun, R. (2010). Estimating the non-price effects of legalization on cannabis consumption. Working Papers. Santa Monica, RAND. WR-767-RC.

MacCoun, R, Pacula, R, Chriqui, J, Harris, K and Reuter, P. (2009). Do citizens know whether their state has decriminalised marijuana? Assessing the perceptual component of deterrence theory. Review of Law & Economics. 5: 347-371.

MacCoun, R and Reuter, P. (2001a). Drug War Herisies Learning from Other Vices, Times and Places. Cambridge, UK, Cambridge University Press

MacCoun, R and Reuter, P. (2001b). Evaluating alternative cannabis regimes. The British Journal of Psychiatry. 178: 123-128.

MacCoun, R, Reuter, P and T., S. (1996). Assessing alternative drug control schemes. Journal of Policy Analysis and Management. 15: 330-352.

Mackenzie, J, Norberg, M, Lim-Howe, N and Copeland, J. (2010). Quantifying Cannabis Users: Is there an Average Sized Joint and Cone? National Drug and Alcohol Research Centre Symposium. Sciencia, UNSW.

290 References

Macleod, J and Hickman, M. (2010). How ideology shapes the evidence and the policy: what do we know about cannabis use and what should we do? Addiction. Online

Macleod, J, Oakes, R, Copell0, A, Egger, MK, Hickman, M, Oppenkowski, T, Strokes- Lampard, H and Davey Smith, G. (2004). Psychological and social sequelae of cannabis and other illicit drug use by young people: a systematic review of longitudinal, general population studies. The Lancet. 363: 1579-1588.

Mann, RE, Adlaf, E, Zhao, J, Stoduto, G, Ialomiteanu, A, Smart, RG and Asbridge, M. (2007). Cannabis use and self-reported collisions in a representative sample of adult drivers. Journal of Safety Research 38: 669-674.

Mann, RE, Stoduto, G, MacDonald, S and Brands, B, Eds. (2008). Cannabis use and driving: implications for public health and transport policy. A cannabis reader: Global issues and local experiences. Lisbon, European Monitoring Centre for Drugs and Addiction.

Marijuana Policy Project (2006) An Open Letter to the President, Congress, Governors, and State Legislatures.

Mark, T and Swait, J. (2003). Using stated preference modelling to forecast the effect of medication attributes on prescriptions of alcoholism medicaiotns. Value in Health. 6: 474-482.

Martire, K and Larney, S. (2009). Principal drug of concern: An analysis of MERIT and RAD client characteristics and outcomes Crime Prevention Issues Sydney, Attorney General's Department, NSW Government. 7.

Matruglio, T. (2008). Magistrates Early Referral into Treatment: An overview of the MERIT program from July 2000 to December 2007. Crime Prevention Issues Sydney, Crime Prevention Division, Attorney Generals Department. No. 2

Matthew-Simmons, F, Love, S and Ritter, A. (2008). A review of Australian public opinion surveys on illicit drugs, DPMP Monograph Series. Sydney, National Drug and Alcohol Research Centre.

May, T, Duffy, M, Warburton, H and Hough, M. (2007). Policing cannabis as a Class C drug, An arresting change? Foundation, JR. York, ICPR, School of Law, King's College.

May, T, Warburton, H, Turnbull, P and Hough, M. (2002). Times they are a changing. York, Joseph Rountree Foundation.

Mayhew, P. (2003). Counting the costs of crime in Australia: technical report. Technical and Background Paper Series, Australian Institute of Criminology.

McCaffrey, D, Pacula, RL, Han, B and Ellickson, P. (2010). Marijuana use and high school dropout: the influence of unobservables. Health Economics. 19: 1281- 1299.

291 References

McDonald, D, Moore, R, Norberry J., Wardlaw, G, Ballenden, N and AIC. (1994). Legislative options for cannabis in Australia. National Drug Strategy. Canberra Commonwealth Department of Human Series and Health.

McIntosh, E. (2010). Introduction. Applied Methods of Cost-Benefit Analysis in Health Care. McIntosh, E, Clarke, P, Frew, E and Lourviere, J. Oxford, Oxford University Press.

McKeganey, N (2010) Bad stigma ... Good Stigma? Drink and Drug News.

McKnight, D. (2005). Beyond Right and Left New Politics and the Culture Wars Sydeny, Allen & Unwin.

McLaren, J and Mattick, R. (2006). Cannabis in Australia. Use, supply, harms, and responses. Monograph Series No. 57. Canbera Drug Strategy Branch, Australian Government Department of Health and Ageing.

McLaren, J, Silns, E, Hutcheson, D, Mattick, R and Hall, W. (2010). Assessing the evidence for a causal link between cannabis and psychosis: A review of cohort studies. International Journal of Drug Policy. 10: 10-19.

McLaren, J, Swift, W, Dillon, P and Allsop, S. (2008). Cannabis potency and contamination: a review of the literature. Addiction. 103: 1100-1109.

Mehra, R, Moore, B, Crothers, K, Tetrault, J and Fiellin, D. (2006). The Association Between Marijuana Smoking and Lung Cancer Archives of Internal Medicine. 166: 1359-1367.

Mills, R, Yap, L and Young, C. (2007). Treatment for ataxia in multiple sclerosis. Cochrane Database of Systematic Reviews 2007 Issue 1.

Minister for Police and Minister for the Illawara. (2008). NSW Police Budget Paper No. 3 - 2008-09. Police, N. Sydney, NSW Governement

Ministerial Council on Drug Strategy. (2005). Australian National Tobacco Strategy 2004–2009: Guide to Planning and Investing in Tobacco Control. Canberra

Miron, J. (2005). The budgetary implications of Marijuana prohibition. Cambridge, MA, Harvard University.

Mishan, E and Quah, E. (2007). Cost benefit analysis, 5th Edition. Abingdon, Routledge.

Monroe, A. (1998). Public Opinion and public policy, 1980–1993. Public Opinion Quarterly. 62: 6-29.

Moore, T. (2005). What is Australia's 'Drug Budget'? The policy mox of illicit drug- related government spending in Australia. Drug Policy Modelling Project Monograph Series. Melbourne, Vic., Turning Point Alcohol and Drug Centre Inc.

292 References

Moore, T. (2007). Working estimates of the social costs of a gram and per users for cannabis, cocaine, opiates, and amphetamines. DPMP Monograph Series. Monograph No 14. Sydney, National Drug and Alcohol Research Centre.

Moore, TH, Zammit, S, Lingford-Hughes, A, Barnes, TR, Jones, PB and Burke, M. (2007). Cannabis use and risk of psychotic or affective mental health outcomes: a systematic review. Lancet. 370: 319-328.

Mørland, J, Bugge, A, Skuterud, B, Steen, A, Wethe, G and Kjeldsen, T. (1985). Cannabinoids in blood and urine after passive inhalation of Cannabis smoke. Journal of Forensic Sciences. 30: 997-1002

Nadelmann, E. (1989). Drug Prohibition in the United States: Costs, consequences, and alternatives. Science. 245: 939-947.

Nadelmann, E. (1992). Thinking seriously about alternatives to drug prohibition. Deadalus. 121: 85-132.

National Task Force on Cannabis Regulation. (1982). The regulation and taxation of cannabis commerce. Washington, DC.

Negrin, M, Pinilla, J and CJ., L. (2009). Willingness to pay for alternative policies for patients with alzheimer's disease. Health Economics. 3: 257-275.

New South Wales Consolidated Acts. (1997 (as at 27 November 2008)). Young Offenders Act. Sydney, New South Wales Government.

New South Wales Consolidated Acts. (1997) as at 27 November 2008. Young Offenders Act. Sydney, New South Wales Government.

New South Wales Government. (2006). Alcohol and other drugs in the workplace. Guide to developing a workplace alcohol and other drugs policy. Sydney, New South Wales Government.

Ngui, R and Shanahan, M. (2010). Cannabis use disorder treatment and associated heath care costs: NSW 2007. Drug Policy Modelling Program Sydney, NDARC. Monograph 20. Sydney, University of New South Wales.

New South Wales Department of Primary Industry. (2005). Farm Chemicals. Sydney, NSW Government.

New South Wales Health. (2008). Information for Tobacco Retailers and Consumers. Sydney, NSW Government.

New South Wales Health. (2009). Smoke Free Cars. Sydney, NSW Government. Retrieved August 2, 2010. http://www.health.nsw.gov.au/publichealth/healthpromotion/tobacco/smoke_fre e_cars.asp

New South Wales Parliament. (1985). Drug Misuse and Trafficking Act 1985 No 226 Sydney, NSW Government.

293 References

New South Wales Police Force. (2007, Feb. 2007). Cannabis Cautioning Scheme. Sydney, NSW Government. Retrieved June 2008, from http://www.police.nsw.gov.au/community_issues/drugs/related_information/can nabis_cautioning_scheme.

New South Wales Police Force. (2009a). Annual Report 2008-09. Sydney, NSW Police Force

New South Wales Police Force (2009b) Liquor Laws and Under 18's. Sydney, New South Wales Police Force.

New South Wales Police Force. (2009c). Thin Blue Line Information Section Young Offenders Act. Retrieved November 2009, from http://www.policensw.com/info/gen/y2.html.

New South Wales Police Force. (2010a). Indicators of plantation. Retrieved December 2010, from http://www.police.nsw.gov.au/community_issues/drugs/cannabis_plantations

New South Wales Police Force. (2010b). Licences and Genuine Reasons. Retrieved August 5, 2010, from http://www.police.nsw.gov.au/__data/assets/pdf_file/0016/131164/fact_sheet_in formation_on_obtaining_nsw_licence.pdf.

Nutt, D. (2009). Estimating drug harms: a risky business? Eve Saville Lecture 2009, Centre for Crime and Justice Studies, King’s College London.

OECD Project on Income Distribution and Poverty. (2005). What are equivalence scales? Retrieved May 15, 2010, from http://www.oecd.org/dataoecd/61/52/35411111.pdf.

Olsen, J and Smith, R. (2001). Theory vesus practice: a review of willingness-to-pay in health and health care. Health Economics. 10: 39-52.

Pacula, RL. (2010a). Examining the Impact of Marijuana Legalization on Marijuana Consumption: Insights from the Economics Literature. Working Paper Santa Monica, RAND Drug Policy Research Centre

Pacula, RL, Chriqui, J and King, J. (2003a). Marijuna decriminalization: what does it mean in the United States? . Cambridge, National Bureau of Economic Research

Pacula, RL, Chriqui, J and King, J. (2004). Marijuana decriminalization. What does it mean in the United States? Santa Monica, RAND. WR-126.

Pacula, RL, Grossman, M, Chaloupka, F, O'Malley, P, Johnston, L and Farrelly, M. (2001). Marijuana and Youth. Risky Behavior Among Youth: An Economic Analysis Gruber, J. Chicago, IL University of Chicago Press.

Pacula, RL, Ringel, J and Ross, K. (2003b). Does marijuana use impair human capital formation. Santa Monica, RAND. WR-121.

294 References

Pacula, RL. (2005). Marijuana use and policy: what do we know and have yet to learn? NBER Reporter. Santa Monica, RAND.

Pacula, RL. (2010b). Examining the impact of Marijuana legalisation on harms associated with marijuana use. Working Paper. Santa Monica, Rand. WR-769.

Pacula, RL and Kilmer, B. (2003). Marijuana and crime: is there a connection beyond prohibition? NBER Working Paper Series. Cambridge, National Bureau of Economic Research.

Pager, D. (2003). The mark of a criminal record. American Journal of Sociology. 108: 937-975.

Parliament of NSW. (2008). The Judiciary. Retrieved December 16, 2008. http://www.parliament.nsw.gov.au/prod/web/common.nsf/key/ResourcesSystem CourtsTheJudiciary.

Patton, G, Coffey, C, Carlin, J, Sawyer, S and Lynskey, M. (2005). Reverse Gateways? Frequent cannabis use as a predictor of tobacco initiation and nicotine dependence. Addiction. 100: 1518-1525.

Payne, J. (2006). A discrete time survival study of drug use and property offending: implications for early intervention and treatment. Technical and Background Paper. Canberra Australian Institute of Criminology.

Pearce, D, Atkinson, G and Mourato, S. (2006). Cost benefit analysis and the environment. Recent developments, OECD.

Pease-Watkin, C. (2010). Colombian police allowed to seize 'personal doses' Columbia News, Columbia Reports.

Pellegrini, S and Jeanrenaud, C. Willingness to pay of the Swiss population for a public health programme against alcohol dependence. Neuchatel, Switzerland, University of Neuchatel.

Perkonigg, A, Goodwin, RD, Fiedler, A, Behrendt, S, Beesdo, K, Lieb, R and Wittchen, HU. (2008). The natural course of cannabis use, abuse and dependence during the first decades of life. Addiction. 103: 439-449; discussion 450-431.

Pernanen, K, Cousineau, M, Brochu, S and F., S. (2002). Proportion of crimes associated with alcohol and orther drugs in Canada. Toronto, Canadian Centre on Substance Abuse.

Petry, N. (2000). Effects of increasing income on polydrug use: A comparison of heroin, cocaine and alcohol abusers. Addiction. 95: 705-717.

Petry, N and Bickel, W. (1998). Polydrug abuse in heroin addicts: A behavioural economic analysis. Addiction. 93: 321-335.

Phillips, B and Burns, L. (2008). NSW Drug Trends 2008 Findings from the Illicit Drug Reporting System (IDRS) Australian Drug Trends Series. Sydney, National Drug and Alcohol Reseach Centre, University of New South Wales.

295 References

Phillips, R, Gopaul, S, Gibson, F, Houghton, E, Craig, J, Light, K and Pizer, B. (2001). Antiemetic medication for prevention and treatment of chemotherapy induced nausea and vomiting in childhood. Cochrane Database of Systematic Reviews Issue 9. Art. No.: CD007786.

Phillips, T, Tranter, B, Mitchell, D, Clark, J and Reed, K. (2007). Australian Survey of Social Attitudes, 2007. Canberra, The Australian National University, ACSPRI Centre for Social Research.

Police Association of NSW. (2009). Crown Employees Police Officers Award Sydney, Police Association of NSW.

Pollack, H and Reuter, P. (2007). The implications of recent findings on the link between cannabis and psychosis. Addiction. 102: 173-176.

Poulsen, H and Sutherland, G. (2000). The potency of cannabis in New Zealand from 1976 to 1996. Sci Justice 40: 171-176.

Productivity Commission. (2008). Report on Government Services 2008. Canberra, Australian Commonwealth Government.

Pudney, S. (2001). Keeping off the grass? An econometric model of cannabis consumption in Britian. Journal of Applied Econometrics 19: 435-453.

Pudney, S, Badillo, C, Bryan, M, Burton, J, Conti, G and Iacovou, N. (2006). Estimating the size of the UK illicit drug market. Measuring different aspects of problem drug use: methodological developments. Nicola Singleton, N, Rosemary Murray, R and Tinsley, T, Home Office Online Report 16/06.

Rasmusen, E. (1996). Stigma and self-fulfilling expectations of criminality Journal of Law and Economics. 39: 519-544.

Ratcliffe, J and Buxton, M. (1999). Patients' preferences regarding the process and outcomes of life-saving technology. International Journal of Technology Assessment in Health Care. 15: 340-351.

Reinarman, C. (2009). Cannabis policies and user practices: Market separation, price, potency, and accessibility in Amsterdam and San Francisco. International Journal of Drug Policy. Volume 20: 28-37.

Rhodes, W, Langenbahn, S, Kling, R and Scheiman, P. (1997). What America’s users spend on illegal drugs, 1988-1995. Rockville MD, Office of National Drug Control Policy.

Rhodes, W, Layne, M, Johnston, P and Hozik, L. (2000). What America’s Users Spend on Illegal Drugs 1988-1998. Cambridge, Office of National Drug Control Policy.

Roads and Traffic Authority. (2008). Roadside drug testing. Retrieved December 1, 2010, from http://www.rta.nsw.gov.au/roadsafety/alcoholdrugs/drugdriving/roadsidetesting. html.

296 References

Roads and Traffic Authority. (2009). Drug and alcohol offences. Retrieved November 29, 2010, from http://www.rta.nsw.gov.au/rulesregulations/penalties/serioustrafficoffences/alco holanddrugs.html.

Roads and Traffic Authority. (2010, February 2010). Licensing. Retrieved Sept 9, 2010, 2010, from http://www.rta.nsw.gov.au/licensing/.

Röhrich, J, Schimmel, I, Zörntlein, S, Becker, J, Drobnik, S, Kaufmann, T, Kunt, zV and Urban, R. (2010). Concentrations of delta9-tetrahydrocannabinol and 11- nor-9-carboxytetrahydrocannabinol in blood and urine after passive exposure to Cannabis smoke in a coffee shop. Journal of Analytical Toxicology. 34: 196- 203.

Rolfe, J and Bennett, J. (2009). The impact of offering two versus three alternatives in choice modelling experments. Ecological Economics. 68: 1140-1148.

Rolles, S. (2009). After the war on drugs: Blueprint for regulations. Bristol, Transform Drug Policy Foundation.

Rolles, S, Kushilick, D and Jay, M. (2006). After the war on drugs options for control. Bristol, Transform Drug Policy Foundation.

Rolling, K. (2008). Counting the costs of crime in Australia: a 2005 update. Research and Public Policy Series. Canberra, Australian Institute of Criminology.

Roman, J. (2004). Can cost-benefit analysis answer criminal justice policy quetions, and if so, how? Journal of Contemporary Criminal Justice 20: 257-275.

Room, R, Fischer, B, Hall, W, Lenton, S and Reuter, P. (2008). Cannabis Policy: Moving beyond stalemate. Global Cannabis Commission Foundation, TB. Oxford, Beckely Foundation.

Rose, J and Bliemer, M. (2008). Constructing Efficient Stated Choice Experimental Designs, Institute of Transport and Logistics Studies, University of Sydney.

Rose, J and Bliemer, M. (2009). NGENE 1.0 User Manual & Reference guide Choice Metrics.

Roxburgh, A, Hall, W, Degenhardt, L, McLaren, J, Black, E, Copeland, J and Mattick, R. (2010). The epidemiology of cannabis use and cannabis-related harm in Australia 1993–2007. Addiction. 105: 1071–1079.

Roads and Traffic Authority. (2000). Road traffic crashes in New South Wales 1999. Haymarket, NSW, Road and Traffic Authority.

Roads and Traffic Authority. (2008). Road traffic crashes in New South Wales: statistical statement for the year ended 31 December 2007. Haymarket, NSW, Road and Traffic Authority.

Russo, E. (2007). History of Cannabis and Its Preparations in Saga, Science, and Sobriquet. Chemistry & Biodiversity. 4:1614–1648

297 References

Ryan, M, Watson, V and Gerard, K. (2008). Practical issues in conducting a discrete choice experiment. Using Discrete Choice Experiments to Value Health and Health Care Ryan, M, Gerard, K and Amaya-Amaya, M. Dordrecht, The Netherlands, Springer.

Salkeld, G, Ryan, M and Short, L. (2000). The veil of experience: Do consumers prefer what they know best? Health Economics. 9: 267-270.

Schreck, C (2004) No More Jail Terms for Drug Possession Russia. Moscow Times.

Schwarzinger, M, Carrat, G and Luchini, S. (2009). "If you have the flu symptoms, your asymptomatic spouse may better answer the willingness-to-pay question" Evidence from a double-bounded dichotomous choice model with heterogeneous anchoring. Journal of Health Economics. 29: 873-884.

Scott, A, Bond, C, Inch, J and Grant, A. (2007). Preferences of Community Pharmacists for Extended Roles in Primary Care: A Survey and Discrete Choice Experiment Pharmacoeconomics.: 783-792.

Scott, L and Burns, L. (2008). NSW Trends in Ecstasy and related drug markets 2008 Findings from the Ecstasy and Related Drugs Reporting System (EDRS) Australian Drug Trends Series. Sydney, NDARC.

Scuffham, P, Whitty, J, Taylor, M and Saxby, R. (2010). Health system choice. A pilot discrete-choice experiment eliciting the preferences of British and Australian citizens. Applied Health Econ Health Policy. 8: 89-97.

Sewell, RA, Poling, J and Sofuoglu, M. (2009). The effect of cannabis compared with alcohol on driving. Am J Addict. 18: 185-193.

Shakespeare, D, Boggild, M and Young, C. (2003). Anti-spasticity agents for multiple sclerosis. Cochrane Database of Systematic Reviews Issue 4: Art. No.: CD001332.

Shanahan, M. (2009). Using DCE to measure preferences for cannabis polcies: A pilot study. Australian Health Economics. Conference. Hobart.

Shanahan, M, Degenhardt, L and Hall, W. (2004). Estimating the economic consequences of a reduced heroin supply in Australia 2000-2003. NDARC Technical Report Sydney, NDARC, UNSW. 195.

Sindelar, J, Jofre-Bonet, M, French, M and McLellan, A. (2004). Cost-effectiveness analysis of addiction treatment: paradoxes of multiple outcomes. Drug and Alcohol Dependence. 73: 41-50.

Smith, R. (2000). The discrete-choice willingness-to-pay question format in health economics. Medical Decision Making. 20:194-206.

Smith, R. (2003). Construction of the contingent valuation market in health care: a critical assessment. Health Economics. 12: 609-628.

298 References

Smith, R. (2006). It's not just what you do, it's the way that you do it: the effect of different payment card formats and survey administration on willingness to pay for health gain. Health Economics. 15: 281-293.

Smith, R. (2007). Use, option and externality values: are contingent valuation studies in health care mis-specified? Health Economics. 16: 861-869.

Snyder, L, Fleming Milici, F, Slater, M, Sun, H and Strizhakova, Y. (2006). Effects of Alcohol Advertising Exposure on Drinking Among Youth Arch Pediatr Adolesc Med. 160: 18-24.

Stafford, J and Burns, L. (2009). Australian drug trends 2009: findings from the Illicit Drug Reporting System (IDRS) Sydney, NDARC, UNSW.

Statistical Services Unit. (2007). New South Wales Criminal Court Statistics 2006. Statistical Report Series Sydney, NSW Bureau of Crime Statistics and Research, Attorney General's Department.

Strang, J, Witton, J and Hall, W. (2000). Improving the quality of the cannabis debate: defining the different domains. British Medical Journal 320: 108-110.

Street, D, Burgess, L and Louviere, J. (2005). Quick and easy choice sets: Constructing optimal and nearly optimal stated choice experments. Intern. Journal of Research in Marketing 22: 459-470.

Street, D, Burgess, L, Viney, R and Lourviere, J. (2008). Designing discrete choice experiments for health care. Using Discrete Choice Experiments to Value Health and Health Care Ryan, M, Gerard, K and Amaya-Amaya, M. Dordrecht, Springer

Swift, W, Hall, W and Teesson, M. (2001). Cannabis use and dependence among Australian adults: results from the National Survey of Mental Health and Wellbeing. Addiction. 96: 737-748.

Tay-Teo, K. (2009). Cannabis prevention in Australia evidence from economic evaluations of three preventive programs Centre for Health Policy, Programs and Economics, School of Population Health Melbourne, University of Melbourne. Doctor of Philosophy.

Taylor, N and Bareja, M. (2005). 2002 National police custody survey. Technical and background paper, Canberra, Australian Institute of Criminology.

Tebutt, C. (2006). Drug Misuse and Trafficking Amendment (Hydroponic Cultivation) Bill Extract from NSW Legislative Assembly Hansard and Papers. DoH. Sydney, NSW Legislative Assembly Hansard, NSW Government.

Teesson, M, Lynsky, M, Manor, B and Baillie, A. (2002). The structure of cannabis dependence in the community. Drug and Alcohol Dependence. 68: 255-262.

299 References

Teesson, M, Swift, W, Hall, W, Groves, R, Slade, T and Mills, K. Prevalence, correlates and comorbidity of DSM-IV Cannabis Use Disorders in Australia: Findings of the 2007 National Survey of Mental Health and Wellbeing In preparation.

Tonry, M and Wilson, J. (1990). Drugs and Crime Criminal Justice Review. 13: 574.

Train, K. (2000). Mixed Logit. Qualitataive Choice Analysis Workshop. University of California at Berkley: 135-166.

Train, K. (2009). Discrete Choice Methods with Simulation, Second edition. Cambridge, Cambridge University Press.

Tsuchiya, A and Williams, A. (2001). Welfare economics and economic evaluation . Economic Evaluation in Health Care. Merging Theory With Practice Drummond, M and McGuire, A. Oxford, Oxford University Press van Beek, I. (2004). In the eye of the needle: Diary of a medically supervised injecting centre. Sydney, Allen&Unwin. van Beurden, E, Zask, A, Passey, M and Kia, A. (Unknown). The Mull Hypothesis: is cannabis use contributing to high tobacco use prevelence among young North Coast males? NSW Public Health Bulletin, NSW Health 19. van den Brink, W. (2008). Forum: decriminalisation of cannabis. Current Opinions in Psychiatry. 21: 122-126. van Dijk, J. (1998). The narrow margins of the Dutch drug policy: a cost benefit analysis. European Journal on Criminal Policy and Research. 6: 369-393. van Laar, M and van Ooyen-Houben, M. (2009). Evaluation Dutch Drug Policy, Trimbos instituut, Netherlands Institute of Mental Health and Addiction. van Ours, J and Willams, J. (2009). Why parents worry: Initiation into cannabis use by youth and their educational attainment. Journal of Health Economics. 28: 132- 142. van Ree, E. (1999). Drugs as a human right. International Journal of Drug Policy. 10: 89-98.

Victims of Crime Bureau., Victims Services., Attorney General’s Department of NSW., Criminal Law Review Division., Attorney General’s Department of NSW and NSW Sentencing Council., (2004) Sentencing Information Package. Sydeny, New South Wales Sentencing Council.

Vining, A and Weimer, D. (2010). An assessment of important issues concerning the application of benefit-cost analysis to social policy. Journal of Benefit-Cost Analyses. 1: 1-37.

Volkow, N and Li, T. (2004). Drug addiction: the neurobiology of behaviour gone awry. Nature Reviews Neuroscience. 5: 963-970.

300 References

Waldfogel, J. (1994). Does conviction have a persistent effect on income and employment? International Review of Law and Economics. 14: 103-119.

Warner, K. (1991). Legalizing Drugs: Lessons from (and about) Economics. The Milbank Quarterly. 69: 641-661.

Weatherburn, D and Jones, C. (2001). Does prohibition deter cannabis use? Crime and Justice Bulletin. Sydney, Bureau of Crime Statistics and Research, New South Wales Attorney's General Department.

Weatherburn, D, Jones, C and Donnelly, N. (2003). Prohibition and Cannabis Use in Australia: A Survey of 18- to 29-year-olds. The Australian and New Zealand Journal of Criminology. 36: 77-93.

Weimer, D, Vining, A and Thomas, R. (2009). Cost-benefit analysis involving addictive goods: Contingent valuation to estimate willingness-to pay for smoking cessation. Health Economics. 18: 181-202.

Weiner, S, Maxwell, J, Sapolsky, H, Dunn, D and Hsiao, W. (1987). Economic incentives and organizational realities: managing hospitals under DRGs. Milbank Quarterly. 65: 463-487.

Werb, D, Fischer, B and Wood, E. (2010). Cannabis policy: Time to move beyond the psychosis debate. International Journal of Drug Policy. 21: 261-264.

Westin, A and Slørdal, L. (2009). Passive inhalation of cannabis smoke--is it detectable? Tidsskrift for Den Norske Laegeforening. 129: 109-113.

Wilkins, C, Bhatta, K and Casswell, S. (2002). A `demand side’ estimate of the dollar value of the cannabis black market in New Zealand. Drug and Alcohol Review. 21: 145-151.

Wilkins, C, Reilly, J, Pledger, M and Casswell, S. (2005). Estimating the dollar value of the illicit market for cannabis in New Zealand. Drug and Alcohol Review. 24: 227-234.

Williams, J. (2004). The effects of price and policy on marijuana use: what can be learned from the Australian experience? Health Economics. 13:123-137.

Williams, J and Mahmoudi, P. (2004). Economic relationships between alcohol and cannabis revisited. Economic Record. 80: 36-49.

Williams, J and Skeels, C. (2006a). The impact of cannabis & cigarette use on health. Research Paper No. 969. Melbourne, University of Melbourne.

Williams, J and Skeels, C. (2006b). The impact of cannabis use and cigarette use on health. Research Paper 969, University of Melbourne, Department of Economics.

Willis, K. (2010). Cannabis supply into and within Australia. AIC Bulletins, from http://ncpic.org.au/ncpic/publications/aic-bulletins/article/cannabis-supply-into- and-within-australia.

301 References

Willis, K, Powe, N and Garrod, G. (2005). Estimating the value of improved street lighting: A factor analytical discrete choice approach. Urban Studies. 42: 2289- 2303.

Wilson, J. (1990). Against the Legalization of Drugs Commentary. 89: 21-28.

Wodak, A, Reinarman, C and Cohen, P. (2002). Cannabis control: costs outweigh the benefits: For. British Medical Journal. 324: 105-106.

Wood, E, Werb, D, Fischer, B, Hart, C, Wodak, A, Bastos, F, Montaner, J and Kerr, T. (2010). Tools for debate: US Federal Government data on cannabis prohibition. Vancouver, International Centre for Science in Drug Policy.

Wooldridge, J. (2009). Introductory econometrics. A Modern Approach. Mason, OH, South-Western Cengage Learning.

World Health Organization. (2005). WHO Framework Convention on Tobacco Control. Geneva, Switzerland, WHO Document Production Services.

Wouters, M, Benschop, A and Korf, D. (2010). Local politics and retail cannabis markets: The case of the Dutch coffeeshops. International Journal of Drug Policy. 21: 315-320.

Yang, X, Liu, C, Zhang, J, Yang, W, Zhang, X and Lű, H. (2009). Yangling Mausoleum and their significance to Early Western Han agriculture. Chinese Science Bulletin. 54: 1738-1743.

Zarkin, G, Cates, S and Bala, M. (2000). Estimating the willingness to pay for drug abuse treatment: A pilot study Journal of Substance Abuse Treatment. 18: 149- 159.

Zerbe Jr, R, Bauman, Y and Finkle, A. (2006). An aggregate measure for benefit-cost analysis. Ecological Econoimcs. 58: 449-461.

Zerbe Jr, R and Bellas, A. (2006). A Primer for Benefit-Cost Analysis. Cheltenham, UK, Edward Elgar.

302 Appendix

Appendix Table of Contents

Appendix – Chapter 2 Page An overview of the existing legal status of cannabis in various countries..... 304 Appendix–- Chapter 4 Estimating current rates of use males ...... 314 Estimating current rates of use females...... 315 Estimating new consumption under legalisation males...... 317 Estimating new consumption under legalisation females ...... 318 Appendix– Chapter 5 Cannabis – crime pathways ...... 319 Police Survey...... 323 Description of penalties received for cannabis ...... 326

Appendix – Chapter 7 Contingent valuation survey...... 332 Appendix– Chapter 8 CUD / Dependence data...... 335 Appendix– Chapter 10 DCE design from NGENE ...... 337 DCE survey ...... 339 Mixed logit code ...... 346 Final model results (all interactions) ...... 347

303 Appendix

Appendix – Chapter 2

An overview of the existing legal status of cannabis in various countries

Despite its classification as a narcotic drug by the United Nations how cannabis is controlled or managed varies considerably and remains highly controversial (European Monitoring Centre for Drugs and Drug Addiction, 2007). The European Union provides a case study of the differences in the interpretation and implementation of the UN Conventions although most countries appear to avoid imprisonment for the personal use of cannabis (European Monitoring Centre for Drugs and Drug Addiction, 2007). In the section that follows a brief overview is provided for some countries in Europe, North America, South America, New Zealand and Australia. Searches in the English language grey and published literature located considerable descriptive information on the laws and policies for the European Union countries, United States, Canada, and Australia, and for some for South American countries but little for most Asian and African countries. The “European Legal Database on Drugs”(European Monitoring Centre for Drugs and Drug Addiction, 2007) and a recent publication titled “The Global Cannabis Commission Report”(Room et al., 2008) are the sources of much of the descriptive information which follows .

Europe

Austria

Austria has been a long time proponent of treatment for drug use (European Monitoring Centre for Drugs and Drug Addiction, 2007). The possession of cannabis is illegal, and anyone who possesses, produces, or supplies drugs to somebody else can be sentenced to up to 6 months of prison or be required to pay a fine. However, it appears that prosecution will not occur if the possession is small, not for trafficking purposes and does not involve a minor. The penalty is up to 5 years incarceration for simple trafficking and up to 25 years for leader of a gang that is trafficking drugs (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Belgium

Prior to 2001 there was a lack of clarity, and an inconsistency in the consequences for the possession of cannabis for personal use in Belgium (Room et al., 2008). In 2001, standardisation occurred with the introduction of civil penalties (fines) of 75-125 Є for

304 Appendix

first time possession. If there were subsequent offences, the fines increased to 130-250 Є. In addition there are additional penalties which may involve incarceration and substantial fines if the offence occurs in the presence of minors, in school grounds, or in the military. But, if there are no other no signs of dependency or other circumstances a warning may be given if the offender attends treatment (Room, 2008; EMCDDA, 2009). The penalties for supply and trafficking range from several months to twenty- five years imprisonment depending upon quantity and whether there is gang involvement.

Britain

As a result of the reclassification of cannabis was from Class B to a Class C drug in 2000 there were fewer criminal controls for cannabis (Room et al., 2008). Directions to the police were to not arrest someone in possession of cannabis unless there were other circumstances present. However, it was reclassified as a Class B drug in 2008 following considerable variation and implementation of the new laws (Drug and Alcohol Office, 2007). The penalties for cannabis possession are up to three months imprisonment and/or a fine of up to £2 500 however there are other sanctions such as a formal warning (no permanent record), and a caution (entered onto national police database) (European Monitoring Centre for Drugs and Drug Addiction, 2007). Trafficking offences may result in imprisonment of up to 14 years (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Czech Republic

The unauthorized handling of narcotic drugs is a criminal offence in the Czech Republic but drug use per se is not a criminal offence. Possession for personal use was not a criminal offence before 1999; subsequently there have been several changes to the laws. Currently, anyone found with a small amount of cannabis for personal use, which is defined for operational purposes by the Public Prosecutor as 20 cannabis cigarettes with more than 1.5% delta 9THC, is subject to administrative sanction. This sanction is either a fine or a warning administered at the police station. Notably, there may be a requirement for a judicial decision as to whether the detected cannabis falls into the accepted amount and if deemed more than the acceptable amount criminal proceedings may result. Supply and trafficking may result in imprisonment of up to 15 years if it

305 Appendix

occurs within organised crime, the fine will be above 164,000 Є (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Croatia

When it gained independence, Croatia ratified all the International Conventions. Use of drugs in public is punishable with a fine however if the offender is found to be addicted and is committing an offence as a result of this, the court will require treatment. Administrative fines for possession range from 140 Є – 14 000 Є while the consequences for supply and trafficking may be incarceration from three to fifteen years (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Denmark

In Demark, although drug use itself is not a crime, it is a criminal offence to import, export, sell, purchase, supply, receive, manufacture, process or possess drugs even a small amount of drugs. Consequences are a fine or imprisonment of up to 2 years, and while there is no differentiation between drug types in the law, the Courts do differentiate when imposing a penalty. Penalties for a small amount of cannabis for own use tends to be a lesser fine. Sentences may also be suspended on condition the offender attends treatment for drug abuse (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Finland

Finland was the first Nordic country to criminalise drug use, and although there have been a number of revisions to the laws, the use of drugs remain a criminal offence. While the maximum penalty for an ordinary drug offence is two years imprisonment, and ten years for an aggravated drug offence, possession for own use is punishable by a fine or maximum of six months imprisonment. Most often a fine is issued, or prosecution and punishment may be waived if the offence is to be considered insignificant or if the suspect has sought treatment (European Monitoring Centre for Drugs and Drug Addiction, 2007).

France

As elsewhere, cannabis possession is a criminal offence, with no legal distinction between different narcotic substances. However, judicial authorities have the right to

306 Appendix

decide how to deal with offences. Drug users arrested could be held in the police station up to 48 hours but most only spend a few hours before being released. Personal use offences are mostly dealt with by therapeutic alternatives or by a warning not involving treatment, unless there is a previous history of offences (European Monitoring Centre for Drugs and Drug Addiction, 2007). Import, export, transportation, possession, supply, delivery, or acquisition attract penalties of up to 10 years imprisonment and up to €7 600 000 fine (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Germany

Similar to other countries, German law does not define use of a narcotic drug as a criminal offence but does consider possession, without authorisation as an offence. The police are required to report every offence to the public prosecutor's office, where a decision may be made not to prosecute if they consider the offender's guilt to be minor, or feel there is no public interest in the offence and the narcotics were only intended for the offender's own use in small quantities (European Monitoring Centre for Drugs and Drug Addiction, 2007). Minor offences may be punishable with a fine. Additionally with the approval of the court a sentence may be suspended if the person seeks treatment. Serious offences of trafficking may result in imprisonment of up to fifteen years.

Greece

In Greece, a distinction is made between ‘an addicted’ and a ‘non-addicted’ person when considering possession for personal use. Those deemed non-addicted may be arrested and sentenced from ten days to five years or receive a fine. Or, for a first sentence, they may be required to follow a counselling program as an alternative to incarceration. Those who are deemed addicted receive a more lenient sentence —then are “declared non punishable” and are required to attend treatment but treatment is often not available so they are just released (European Monitoring Centre for Drugs and Drug Addiction, 2007). Those found in possession of trafficable amounts may be sentenced between one to ten years depending on circumstances (European Monitoring Centre for Drugs and Drug Addiction, 2007).

307 Appendix

Italy

The offence of ‘using’ illicit drugs is not found in the Italian criminal code. Possession, acquisition, or import for personal consumption remain prohibited and are subject to administrative sanctions. Sanctions for the possession of cannabis are less than for other illicit drugs. First offences usually result in a warning from police, but for repeated offences or repeated failures to attend treatment, sanctions may include prohibiting the person from leaving their place of residence without authorisation; requiring them to present at police station at least weekly; having their driving licence suspended; confiscation of their passport or in the case of non-citizens their resident permit; or requiring community service at least one working day a week (European Monitoring Centre for Drugs and Drug Addiction, 2007). Penalties for trafficking larger amounts of cannabis may result in imprisonment of 2 to 6 years or fines of €5000 to €77000.

Netherlands

The 'main' drug law in the Netherlands is the Opium Act (European Monitoring Centre for Drugs and Drug Addiction, 2007) in which Dutch drug policy is strictly prohibitionist in nature. The sale and supply of listed drugs (including cannabis) are illegal, and include possession, manufacture and cultivation (Kerley, 2004). Possession is subject to imprisonment for a maximum of six months. However, arresting and criminalising users possessing small quantities for personal use of any drug is not regarded as a priority for law enforcement. This is specified mainly by guidelines issued by the Office of the Public Prosecutor (European Monitoring Centre for Drugs and Drug Addiction, 2007). The behaviours of use and possession of small amounts of cannabis are not criminalised in the Netherlands although drug use is not permitted in schools or on public transportation. The policy around cannabis was undertaken in part as an attempt to separate it from the market for ‘hard’ drugs and is often referred to as de facto legalisation. ‘Coffee shops’ have emerged as the 'official/unofficial' sales channel for cannabis, albeit under strict conditions (European Monitoring Centre for Drugs and Drug Addiction, 2007). It is policy that the possession and sale of five or less grams of cannabis will not be prosecuted but there has been a significant amount of international pressure on the Netherlands to change its policies.

308 Appendix

Along with separating the markets, arguments for this regulatory structure were to encourage those who need treatment to seek it in a non-threatening environment and to allow law enforcement resources to be prioritised elsewhere (European Monitoring Centre for Drugs and Drug Addiction, 2007). The maximum penalty for importing or exporting any quantity of cannabis is four years imprisonment and/or a €67 000 fine (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Norway

Here the objective of a society free from drug abuse is sought through a multidisciplinary approach involving prevention, treatment and rehabilitation and enforcement of the supply side and control measures. In 1984, use and possession of small amount of drugs changed from a misdemeanour to a crime but if the drug is being possessed for a short term, i.e. ‘current use’ then the consequence is less severe. If however, the drug is being ‘stored’ or there is a substantial quantity then there may be fines or imprisonment (European Monitoring Centre for Drugs and Drug Addiction, 2007). Incarceration may be up to ten years for supply and trafficking.

Portugal

Prior to 2001 drug use was a criminal offence punishable by a fine, or up to three months imprisonment for less than 3 doses or up to a year for more. After 2001, drugs remain illegal but the consequences (sanctions) changed (European Monitoring Centre for Drugs and Drug Addiction, 2007). For possession of a modest amount of drug, with no other evidence of serious offences, offenders are referred to the Commission for the Dissuasion of Drug Addiction (CDA), where treatment is offered. These CDAs are meant to support dependent users in terms of treatment but can impose fines, community service, and suspension of professional practice (Hughes and Stevens, 2007). Other offences such as supply or trafficking attract a sentence between four and twelve years (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Spain

Spanish law on drugs refers directly to the 1961 Single Convention on Narcotics Drugs and in the 1971 Convention on Psychotropic Substances. The laws on trafficking and supply have been developed in response to the frequent trafficking through Spain to other European countries and have some of the most severe penalties in Europe

309 Appendix

(European Monitoring Centre for Drugs and Drug Addiction, 2007). On the other hand, drugs use and possession for personal use do not constitute a criminal offence under Spanish law although consumption in public is an order offence which is subject to administrative sanctions or fines from €300 to €30 000. Fines may be suspended if the person attends a drug treatment program.

Russia

In Russia, as of 2004, the possession of no more than 10 times the amount of a "single dose" is considered an administrative infraction rather than a criminal offense. Punishment would be a fine of no more than 40,000 rubles (~$1,380) or community service. Ten times a single dose of cannabis is said to be 20 grams. The possession of larger amounts or trafficking will likely result in incarceration (Schreck, 2004).

Sweden

Sweden has strongly pursued the goal of a drug free society through the use of increased penalties for drug offences, police powers to urine-test suspected drug users and maintaining a large number of treatment places to which suspected drug users may be coerced to attend (Hall and Babor, 2000). Use or possession of cannabis for personal use is considered a minor offence punished by a fine with the amount contingent on income. More serious offences may receive a custodial sentence.

Switzerland

The possession and use of cannabis is illegal in Switzerland and remains so despite several attempts to introduce into the legislature exemptions for personal use (Room et al., 2008). The enforcement of the federal laws occur differentially in the different cantons however increasingly it appears that they laws are being enforced more rigorously (European Monitoring Centre for Drugs and Drug Addiction, 2007).

Americas

Argentina

Following a challenge to the countries drug laws by a 19 year old who was convicted of possession of 2 grams of cannabis, and sentenced to 45 days in prison, the Argentinean Supreme Court ruled it unconstitutional to prosecute offences involving the use of

310 Appendix

cannabis in private. This ruling was ratified by the government (Carlos Hildago, 2009). Supply and cultivation remain illegal.

Brazil

The possibility of a jail sentence for possession of drugs was removed by legislation and was replaced by treatment, or community service, or fines, or suspension of a driver’s licence. This law which also increased the penalties for suppliers was ruled unconstitutional in 2008 (Room et al., 2008). Penalties for trafficking may be between eight and twenty years in prison (Room et al., 2008).

Canada

In Canada, cannabis is prohibited where even possession / use of small amounts may result in criminal charges, fines or even incarceration (Room et al., 2008). In Canada the Federal Prosecution System is responsible for prosecution (Kisely, 2005). Since 1996, conditional sentencing became available for some drug offences, where if the accused pleads guilty s/he may receive treatment or community service (Room et al., 2008). Although cannabis possession offences have increased, the rates of prosecution by the Federal Prosecution System have declined to as low as 35% of offences in some areas of the country (Kisely, 2005).

Columbia

As elsewhere, cannabis is illegal in Columbia but the Columbian Supreme Court has ruled that personal possession small quantities (< 20gm) of any psychoactive drug is legal (Room et al., 2008); this was subsequently limited to only in the home. In December 2009, the Congress prohibited the possession and carrying of drugs and more recently, while waiting for new laws to be passed, have stated that the police can seize all drugs. Currently, the political situation is in flux with the current legislature passing laws to change the Columbian Constitution to ensure possession of all drugs are illegal (Pease-Watkin, 2010).

Mexico

The Mexican Senate approved a bill decriminalising possession of small amounts of narcotics for personal use (Angel Gutierrez, 2009) which was subsequently passed by the lower house with little protest in August 2009 (Grillo, 2009). Thus, anyone caught

311 Appendix

with two or three joints can no longer be arrested, fined or imprisoned. Police will give them the address of the nearest rehab clinic and advise them to get ‘clean’. This law essentially legalises what police were already doing as resources are overextended in an attempt to deal with the ongoing drug wars in Mexico.

United States

The overarching federal law on the legal status of cannabis in the US is total prohibition (Hall and Pacula, 2003) although many every state has its own drug legislation. Many states have legislation which decreases the penalties for possession of a small amount of cannabis (Pacula, 2005). There is a range of laws and consequences from de-penalising the possession of cannabis for personal use, treating it as misdemeanour or regarding it as civil offence (Room et al., 2008). Still other states allow those convicted of a non- violent drug offence to participate and complete drug treatment programs which if not completed may result in incarceration (Room et al., 2008).

California may be the closest to legalisation through the permission of medicinal use of cannabis as almost anyone can get medical letter from a willing doctor enabling them to obtain cannabis for medicinal purposes – a quasi back door legalisation that has seen the price of cannabis fall.

California currently has two cannabis related bills (SB1449 and AB2254) before the legislature plus a proposition to the voters to Regulate, Control and Tax Cannabis Act 2010 (RCTC) (Kilmer et al., 2010). SB1449 would reduce the penalty for possessing less than one ounce of cannabis to an infraction similar to a parking violation; this bill has passed the Senate and is being considered by the full Californian Assembly (Kilmer et al., 2010). AB2254 and the RCTC would legalise cannabis; with AB2254 subjecting cannabis to a $50 per ounce tax while the RCTC would legalise up to 1 ounce of cannabis, and cultivation of plants for own use in a five foot by five foot space (Kilmer et al., 2010). Cannabis would remain illegal under the US Federal Law.

Asia and Africa

Information on the drug laws for most of Asia and Africa is not readily available. Search of the peer review literature and internet searches resulted in locating only blogs and Wikipedia as sources.

312 Appendix

India

Cannabis use is illegal in India with the exception of regulated bhang . Bhang is usually in lassi drinks, or is sold in government regulated shops. Other use is illegal and subject to incarceration

Indonesia

Cannabis use and possession is illegal with a maximum sentence of four years in prison. In some circumstances the sentence may be lower with the addition of a fine.

Pakistan

While cannabis grows wild in Pakistan, and it is often sold and used openly in parts of Northern Pakistan it is officially a criminal offence subject to incarceration .

Oceania

New Zealand

The legal status of cannabis in New Zealand falls under the Misuse of Drugs Act 1975 which is administered by the Ministry of Health but enforced by Police and Customs (Report on Health Committee, 2003). Cannabis resin is listed as a Class B drug, while cannabis leaf, fruit and seed is a Class C drug (moderate risk). The penalties are fine of up to $500, up to 3-month prison sentence or both for possession for personal use. There are other options, for example, in recent analysis of encounters with police for cannabis possession of less than 28 grams, 28% resulted in a warning, 6% were sent to youth aid, and 3% were offered diversion (Wilkins, 2010). Supply or importation may result in imprisonment of up to 8 years. There have been many public campaigns to decriminalise cannabis but so far none have succeeded (Ministerial Committee on Drug Policy, 2007).

313 Appendix

Appendix – Chapter 4

Estimating current rates of use

The data in the tables below were used to generate current and future consumption. The tables in the body of thesis provide data for males and females combined. The tables below are the numbers who currently consume cannabis, the number of total days where consumption occurs for all those who consume, and the total joints consumed on those days and finally the total grams of cannabis consumed. Data are presented by age category and by frequency of consuming.

Males

Table A1: Current use patterns males NSW 2007 [A] Once/ Approx once Every few 1 or 2 times Age Every day week or a month months per year more 14-19 5,716 10,709 7,202 4,402 5,434 20-29 21,393 22,025 12,042 14,705 27,159 30-39 17,030 16,904 3,805 9,007 27,853 40+ 17,996 27,070 11,018 8,455 21,958 Source: NSDSH 2007 data

Table A2: Days cannabis consumed [B] Once/ Approx 1 or 2 Every few Every day week or once a times per months more month year Days of 365 182 12 6 2 consumption

Table A3: Estimating total days of cannabis consumption by males [C= B * A] Approx 1 or 2 Once/ week Every few Age Every day once a times per or more months month year 14-19 2,086,340 1,949,038 86,424 26,412 8,151 20-29 7,808,445 4,008,550 144,504 88,230 40,739 30-39 6,215,950 3,076,528 45,660 54,042 41,780 40+ 6,568,540 4,926,740 132,214 50,730 32,937 All 22,679,640 13,961,038 408,814 219,420 123,608 Source: NSDSH 2007

Table A4: Average number of joints per use day by frequency [D] Frequency of use Joints 95% CI Every day 3.50 2.4 4.6 Once / wk or more 1.82 1.6 2.1 Approx. once / month 1.70 1.3 2.1 Every few months 1.33 1.2 1.5 1 or 2 times per year 1.08 1.0 1.2 Source: NSDSH 2007

314 Appendix

Table A5: Total number of joints consumed in one year [E = C*D]. Once/ Approx 1 or 2 Every few Age Every day week or once a times per months more month year 14–19 7,297,900 3,553,789 147,088 35,131 8,781 20–29 27,313,503 7,309,012 245,937 117,357 43,888 30–39 21,743,045 5,609,604 77,711 71,882 45,009 40+ 22,976,385 8,983,199 225,020 67,477 35,483 Source: NSDSH 2007

Table A6: Total cannabis consumed in one year (males, NSW, 2007) [F= E * .37 gm] Once/ Approx 1 or 2 Every few Total use in Age Every day week or once a times per months 12 months more month year 14–19 2,700,223 1,314,902 54,423 12,999 3,249 4,085,795 20–29 10,105,996 2,704,334 90,997 43,422 16,238 12,960,988 30–39 8,044,926 2,075,554 28,753 26,597 16,653 10,192,483 40+ ,501,262 3,323,784 83,257 24,967 13,129 11,946,398 All 29,352,408 9,418,574 257,430 107,984 49,270 39,185,664 Source: NSDSH 2007

Females

Table A7: Current use patterns females NSW [A] Approx Once/ week Every few 1 or 2 times Age Every day once a or more months per year month 14–19 2,891 6,261 5,402 2,162 10,783 20–29 7,376 3,992 7,052 10,148 20,307 30–39 2,310 7,629 2,630 8,380 11,625 40+ 5,371 4,241 2,468 7,336 18,254

Table A8: Defining days of cannabis consumption [B] Once/ Approx 1 or 2 Every few Every day week or once a times per months more month year Days of 365 182 12 6 2 consumption

Table A9: Estimating total days of cannabis consumption by females [C= B * A] Once/ week Approx once Every few 1 or 2 times Every day or more a month months per year 14–19 1,055,215 1,139,502 64,824 12,972 16,175 20–29 2,692,240 726,544 84,624 60,888 30,461 30–39 843,150 1,388,478 31,560 50,280 17,438 40+ 1,960,415 771,920 29,617 44,017 27,381

315 Appendix

Table A10: D= Joints per use day Frequency of use Joints 95% CI Every day 3.50 2.4 4.6 Once / wk or more 1.82 1.6 2.1 Approx. once / month 1.70 1.3 2.1 Every few months 1.33 1.2 1.5 1 or 2 times per year 1.08 1.0 1.51.2 Source: NSDSH 2007 data

Table A11: Joints consumed by females [E = C*D] Once/ week Approx once Every few 1 or 2 times Age Every day or more a month months per year 14–19 3,691,083 2,077,717 110,326 17,254 17,425 20–29 9,417,305 1,324,748 144,025 80,988 32,815 30–39 2,949,291 2,531,689 53,713 66,879 18,785 40+ 6,857,422 1,407,485 50,406 58,548 29,498 Source: NSDSH 2007 data

Table A12: Total cannabis consumed in one year (females, NSW, 2007) [F= E*.37] gm Once/ Approx 1 or 2 Every few Total use in Age Every day week or once a times per months 12 months more month year 14–19 1,365,701 768,755 40,821 6,384 6,447 2,188,108 20–29 3,484,403 490,157 53,289 29,966 12,142 4,069,956 30–39 1,091,238 936,725 19,874 24,745 6,951 2,079,532 40+ 2,537,246 520,770 18,650 21,663 10,914 3,109,243 All 8,478,587 2,716,407 132,634 82,758 36,454 11,446,839 Source: NSDSH 2007 data

Results from Tables A6 and A12 were summed for total estimates. As part of the sensitivity analysis the above work was replicated using 95% the confidence intervals from the Australian NDSHS estimates, as well as using the estimates from Rhodes (1997) and Pudney (2006), and with 0.25gm and 0.5gm per joint.

316 Appendix

Estimating new consumption under legalisation

Question L20 from the NDSHS which asks respondents the following questions was used to estimate possible increase in prevalence and then consumption. If Marijuana/Cannabis were legal to use, would you? Not use it, even if it were legal and available; Try it; Use it about as often as you do now; Use it more often than you do now; Use it less often than you do now; Don’t know (AIHW, 2008)

Males The data below presents the estimates for both Method 1 and Method 2 as described in Chapter 4. Males and females were again estimated separately.

Table A13: Percentage of males that indicated they will either start using or increase use by current use status Current Once a About Once or Every Every few Not in last use week or once a twice a Never day months 12 months status more month year 14-19 0.0% 27.1% 3.0% 12.5% 10.3% 11.6% 12.7% 20-29 12.4% 4.1% 8.2% 8.5% 9.8% 10.2% 10.2% 30-39 1.4% 2.1% 11.3% 1.9% 10.9% 4.9% 3.7% 40+ 3.6% 6.5% 6.5% 6.1% 10.8% 5.1% 2.0%

Table A14: Total consumers males: Method 1 Once a About Once or Every Every few week or once a twice a day months more month year 14-19 8,621 10,928 7,751 4,962 36,503 20-29 22,305 23,014 13,297 17,379 65,137 30-39 17,383 17,333 3,978 12,053 45,467 40+ 19,752 27,785 11,533 10,819 61,134

Table A15: Total consumers males: Method 2 Once a About Once or Every Every few week or once a twice a day months more month year 14-19 12,790 21,360 14,442 9,508 10,415 20-29 29,741 29,690 17,238 22,115 35,741 30-39 20,561 20,505 4,872 13,025 32,946 40+ 27,424 39,721 16,357 13,897 31,759

317 Appendix

Females Table A16 Percentage of females that indicated they will either start using or increase use by current use status Once a About Every Once or Current use Every Not in last week or once a few twice a Never status day 12 months more month months year 14-19 0% 26% 29% 17% 31% 9% 12% 20-29 3% 2% 0% 10% 5% 6% 6% 30-39 3% 4% 16% 7% 5% 3% 2% 40+ 9% 1% 2% 3% 3% 3% 1%

Table A17: Total consumers females: Method 1 Once a About Every few Once or Every day week or once a months twice a year more month 14-19 4,504 7,829 5,765 5,456 38,730 20-29 7,462 3,992 8,066 11,197 44,878 30-39 2,607 8,059 3,238 8,915 21,833 40+ 5,426 4,290 2,684 7,934 45,392

Table A15: Total consumers females: Method 2 Once a About Every few Once or twice Every day week or once a months a year more month 14-19 7,078 13,633 11,536 6,617 22,401 20-29 12,825 7,198 12,947 19,315 34,517 30-39 3,569 11,517 4,194 13,253 17,012 40+ 11,130 8,744 5,379 16,290 37,083

318 Appendix

Appendix–Chapter 5

Cannabis – crime pathways

This following section provides more detail on Pernanen’s hypothesised cannabis-crime pathways. Three of Pernanen’s pathways follow Goldstein’s tripartite framework (1985), and the fourth (first in the list) recognises specific substance-based laws. They are 1) the substance-defined pathway; 2) the illegal system model; 3) The pharmacological or intoxication pathway; and 4) The economic-compulsive pathway. After a comprehensive review only the first pathway was included. The interested reader is referred to Appendix 5.

The substance-defined pathway—these are actions which are defined as illegal through legislation which regulates drug use. For example, in Australia the possession, supply and cultivation of any amount of cannabis are all illegal.

Illegal system model—these are activities that are illegal even if not related to illicit drugs but are offences committed in the course of selling drugs, collecting drug debts, and conflicts over drug territory (Pernanen et al., 2002). They are often gang-related and may be related to drugs or to any other activity (gambling, racing, etc) where the lack of a legal market is filled by an illegal market. While there is a black market for cannabis, MacCoun and Reuter (2001, p 362) make the point that these black markets are less violent than other drug markets. Pacula and Kilmer (2003) cite both the ONDCP (2001) and Goldstein (1985) as sources suggesting that violence in the cannabis market is rare, in part due to the fact that most transactions are not visible as they occur in private homes or other off-street locations. Only 5% of Australians indicate they purchased the majority of their cannabis sales outdoors (Australian Institute of Health and Welfare, 2008a) thus competitors are unable to see transactions.

In Australia it is likely that organised crime groups, including outlaw motorcycle gangs, are involved in the cultivation and distribution of cannabis. According to the Australian Crime Commission, there has also been a noticeable increase in the involvement of Vietnamese crime groups in cannabis cultivation and supply in recent years (Australian Crime Commission, 2008). In discussions with members of the NSW Drug Squad, the suggestion was that the focus of gangs was on other drugs (not cannabis). On the other

319 Appendix

hand, others suggest that cannabis may provide a constant base income for organised crime gangs.

The pharmacological or intoxication pathway—here the hypothesis is that the drug itself makes, or helps make, the drug user engage in a crime because of the acute psychoactive effects of the substance. In alcohol research this model is often referred to as the dis-inhibition model (Pernanen et al., 2002). Nash Parker developed the theory of selective dis-inhibition where ingestion of a drug may result in a chemical reaction that while not increasing tendency to violence it inhibits the normative values that might otherwise inhibit violence (Payne, 2006). Therefore, in exploring changes to cannabis laws, which may lead to increases in use, the extent to which violent or other crimes might rise in association with increased use must be assessed. This literature which is reviewed below is not substantive.

Fattah (1971), after critically reviewing the relationship between cannabis and crime, reported that most researchers agreed that cannabis did not change the personality structure, and that there was little agreement on the causal relationship between cannabis and crime (Pernanen et al., 2002). Subsequently, Abel (1977) examined the evidence dealing with the alleged relationship between cannabis and violence and determined that there was a consensus that cannabis did not precipitate violence in the majority of those using it sporadically or chronically.

In general, reviews have concluded that violent behaviour is usually either decreased or unaffected by cannabis use (Fergusson et al., 2003). Whilst cannabis has been shown to inhibit aggressive behaviours there is some evidence that long term use of cannabis may change the nervous system such that social communications are disrupted; an effect that may increase one’s involvement in altercations that escalate to violence (Reiss & Roth, 1993 in (Bolesa and Miottoa, 2003). In particular, there is some suggestion of a link between violent offences and cannabis use among juveniles.

Analysis of data from a New Zealand birth cohort (Ferguson and Horwood, 1997) considered four measures of delinquency and found a dose response which persisted even after adjustment for covariates and also persisted after adjustment for drug use and criminal behaviour in user’s peer group. Hierarchical regression analysis of self-report data from French high school students found that after controlling for confounding

320 Appendix

variables (such as alcohol misuse, psychopathic and borderline personality disorders) cannabis users were no more likely than non-cannabis users to have delinquent behaviours, however there again was a dose response effect amongst young cannabis users. While cannabis use did not predict criminal behaviours, once adjusting for confound behaviours, more frequent cannabis users were more likely to have delinquent behaviours than infrequent users (Chabrol and Saint-Martin, 2009). Notably neither of these studies was able to identify causal relationships.

One NSW survey of secondary students found after adjusting for demographic and developmental characteristics, as well as alcohol and other illicit drugs, that students who were frequent users of cannabis were 2.3 times more likely to report having participated in assault (Baker, 1998). The definition of assault used in this study was “attacked them with the intent to hurt them outside of sport”. While the authors attempted to capture only the most serious offences, they acknowledged that it was quite likely that the crimes were quite broad-ranging and included many minor offences as well as more serious ones. It appears that there may be some evidence that frequent cannabis use in youth may result in increased delinquent or aggressive behaviour but these data do not provide sufficient evidence to estimate any potential impact on the CJS.

Turning to adults, analysis of arrest data from the United States for adults found that cannabis users are no more likely to be arrested for violent crime (Pacula and Kilmer, 2003). Others suggest that the greatest risk of violence from a user of cannabis is within the first week of an acute period of abstinence for those who are heavy users, a so-called cannabis withdrawal syndrome (Hoaken and Stewart, 2003).

The economic-compulsive pathway—refers to acquisitive crime committed by dependent users for the purpose of obtaining money to purchase illicit drugs and alcohol (Pernanen et al., 2002). There is little evidence that dependent adult cannabis users commit crimes to pay for their cannabis (MacCoun and Reuter, 2001a). Further support was provided by Pacula (2003) who examined American crime data (ADAM – arrestee drug abuse monitoring data & UCR – uniform crime reports) to assess the causal association between cannabis use and crime. Although finding that self-reported cannabis users are more likely than non-cannabis users to be arrested for property and income-producing crime even after adjusting for age, gender, race and county

321 Appendix

characteristics, price of cannabis, alcohol and cocaine, use of cannabis, recent use of alcohol and past crime, the authors indicate caution in drawing any strong conclusions from these findings. The reason given was when this analysis was repeated using only those whose urine tests indicated cannabis use, the association between acquisitive crime and cannabis use was not present (Pacula and Kilmer, 2003).

Bennett et al. (2008), reports conflicting findings from a meta-analysis of studies examining the relationship between cannabis use and different types of crime. Ten studies in this meta-analysis specifically examined the relationship between cannabis and crime where the odds of cannabis users offending are about 1.5 times greater than non-cannabis users (Bennett et al., 2008). When the individual studies were examined there was no significant relationship reported between cannabis use and prostitution (n=3 studies); between cannabis use and shoplifting (n=1 study); or between cannabis use and cannabis supply (n=1 study). There were conflicting findings between cannabis use and property crime (one found no relationship and one a positive relationship); while three studies found a positive relationship between all types of crime and cannabis. Bennett and colleagues suggest these apparent conflicts are due to different methodologies, different comparators as well as differences in the types of criminal behaviours examined.

Analysis of a New Zealand cohort at age 21, examined the relationship between cannabis use, arrest and conviction of cannabis offences, other offences and demographic characteristics and found that frequent use of cannabis, being Maori, male, having had a previous arrest for cannabis and having other offences all increased the risk of arrest for a cannabis offence but the rates of acquiring a criminal record solely due to cannabis was very low. Only 0.7% acquired a criminal record in this way suggesting that cannabis-related offences add to the existing criminal records of those who are in contact with the system. Plus, the overall rates of arrest were low (Fergusson et al., 2003).

322 Appendix

PARTICIPANT INFORMATION STATEMENT AND CONSENT FORM Approval No 07293

Project Title: Developing a model to assess the economic consequences of cannabis policy options Purpose of the study You are invited to participate in survey which is part of a project to estimate the economic consequences of cannabis policy options. As part of this study we are estimating the time it takes police to manage simple cannabis offences. Once collected, these data will be used to estimate the average time and costs related to these activities. We are collecting information from police at three Local Area Commands. The study is being conducted by Marian Shanahan (Health Economist, PhD Candidate), and Associate Professor Alison Ritter (Director) from the Drug Policy Modelling Program, National Drug and Alcohol Research Centre, The University of NSW Description of study If you decide to participate we ask that you complete the attached survey. The survey will take between 10 and 15 minutes. We are asking you to estimate the average time it usually takes you to complete the activities listed when dealing with either an Adult or Juvenile cannabis offender (there are two separate forms). Upon completion, please place the survey in the envelope provided. There are no identifiable risks from participating in this study. We cannot and do not guarantee or promise that you will receive any benefits from this study. Your consent Completing this survey is completely voluntary and you are not under any obligation to consent to complete the survey. Submitting a completed survey is an indication of your consent to participate in the study. You can withdraw any time prior to submitting your completed survey. Once you have submitted your survey anonymously, your responses cannot be withdrawn. Your decision whether or not to participate will not prejudice your future relations with the University of New South Wales. Confidentiality and disclosure of information This is an anonymous survey. Your name will not be individually identified in any publication arising from this research. These data once aggregated may be included in a peer reviewed paper and monograph. We may also present the results at an academic conference. Complaints Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052 AUSTRALIA (phone 9385 4234, fax 9385 6648, email [email protected]). Any complaint you make will be investigated promptly and you will be informed of the outcome. Further information If you have any questions, please feel free to contact Marian Shanahan on (02) 9385 0229.

323 Appendix

Survey of time taken to complete various police activities related to cannabis offences Local Area Command ______Rank ______Please take 10 minutes to complete this survey now, and thank you in advance for doing so. We are interested in the average amount of time it takes to complete each of the activities listed below in the context of a possess/ use, supply or cultivate cannabis offence where the police officer detects the offence through routine policing. Please think about the last time you were involved in any of these activities with an individual who has come to the attention of the police for a cannabis offence. There are separate questions for Adults and for Juveniles. You may not have been involved in all of these activities in which case leave section blank. Avg. Time in number & minutes (on type of Step ADULTS ONLY average per additional officer) personnel involved Common to all offences 1 Initial search and questioning of the offender 2 Determine identity /carry out CNI 3 Compile file and complete necessary paper work, COPS event & COPS intelligence report 4 Secure / weigh cannabis at police station 5 Destroy cannabis at police station

Specific to Cannabis Caution (in addition to steps 1- 5) 6 Weigh / secure cannabis (on street) 7 Write up and obtain signature on caution 8 Convey offender to station (only if you do this routinely)

Specific to Field CAN / Future Court Attendance Notice (in addition to steps 1- 5) 9 Weigh / secure cannabis (on street) 10 Write up and obtain signature on court attendance notice 11 Convey offender to station (only if you do this routinely) 12 Issue court notice

Specific to A No Bail Court Attendance required (in addition to steps 1- 5) 13 Convey to police station 14 Booking / repeat search 15 Questioning of offender / lay charges 16 Photograph, fingerprints 17 Time contacting/ waiting for support person/ legal rep 18 Return possessions to individual 19 Release offender from police custody

Specific to Bail court attendance required – (as in No Bail Attendance Steps 1- 5 and 13 -17) plus 20 Place individual in cells

Preparation for court 21 Statement of facts 22 Gathering additional evidence (interviewing, testing etc)

324 Appendix

Attendance at court 23 Defendant pleads guilty 24 Defendant pleads not guilty

Avg. number & Time in type of minutes (on Step JUVENILES ONLY additional average per police officer ) personnel involved Costs and time specific to Juvenile Conferencing will be estimated elsewhere Common to all offences 1 Initial search of the offender and questioning 2 Determine identity /Carry out CNI check 3 Secure / weigh cannabis at police station 4 Destroy cannabis at police station Specific to Warning (in addition to steps 1-4 above) 5 Issue warning 6 Complete COPS event 7 Letter to parents / guardian Specific to Juvenile caution for cannabis (in addition to steps 1-4 above) 8 Return to police station 9 Contact responsible adult & wait for arrival 10 Plan /organise caution 11 Prepare & deliver caution (if involved/ may occur at separate time) 12 Compile file and complete necessary paper work, COPS event & COPS intelligence report Specific to No Bail Court attendance required (in addition to steps 1- 4 above) 13 Return to police station / repeat search 14 Weigh / secure cannabis 15 Contact responsible adult & wait for arrival 16 Charge 17 Compile file and complete necessary paper work, COPS event & COPS intelligence report 18 Photograph, fingerprints 19 Referral to Juvenile Justice (if appropriate) 20 Return possessions to individual 21 Release individual Specific to Bail court attendance required – as in No Bail Court attendance, steps 1-5, and 12-18 plus 22 Place individual in cells Preparation for court 23 Statement of facts 24 Gathering additional evidence (interviewing, testing etc) Attendance at court 25 Defendant pleads guilty 26 Defendant pleads not guilty

Additional comments …. ______

325 Appendix

Penalties received for cannabis offences

Corrective Services

Corrective Services Department is the department which is responsible for the supervision and program management of adult offenders in custody and in the community. This includes fulltime and periodical incarceration in corrections faculties, home detention, community service orders, and probation and parole (Department of Corrective Services., 2008).

Penalties – Adults

When an adult is either found guilty or pleads guilty there is a wide range in the penalties that can be specified by the judge or magistrate. The length and severity of the penalty or size of the fine are a function of the court, the offence and offender. There is likely to be a penalty specified for each offence for which the person is found guilty, although there is a Principal Offence, which is defined as that offence which receives the most serious penalty according to a set of rules (Statistical Services Unit., 2007). If there were two or more offences which received the same penalty, the offence which received the largest quantum is the principal offence (fine, length of sentence, longest parole). Penalties may be concurrent or cumulative. On some occasions the court may convict the offender and order that no sentence be imposed. This is a section10a order(Victims of Crime Bureau. et al., 2004; Statistical Services Unit., 2007). The main sentences likely to be given for a cannabis offence are:

Imprisonment

The sentence given specifies the term of the sentence, provides both the whole term and the non-parole period. The non-parole period is the minimum amount of time an offender must remain incarcerated. Unless there are special circumstances, the non- parole period is at least three-quarters of the term of the sentence (Victims of Crime Bureau. et al., 2004). Another form of imprisonment is Periodic imprisonment where the offender is held in custody periodically i.e. weekends for six months (Statistical Services Unit., 2007).

326 Appendix

Home detention

This is a full-time imprisonment in the home of up to 18 months that is not available to those currently or previously convicted of certain serious offences including offences which involve violence such as murder, sexual offences, as well as certain serious drug offences. This is seen as a lesser sentence than incarceration. Conditions can be imposed, which permit the offender to leave home for certain approved periods, for example, for the purpose of employment or receiving medical treatment. Compliance with the order is supervised by way of electronic monitoring and random visits by a Probation and Parole Officer (Victims of Crime Bureau. et al., 2004).

Suspended sentence

A court may impose a sentence of imprisonment of up to 2 years and then suspend the sentence on the condition that the offender enters into a good behaviour bond(Statistical Services Unit., 2007).

Deferred sentence

The court can postpone passing sentence on an offender for a period of up to 12 months from the date of conviction to allow the offender to be assessed for rehabilitation; or to demonstrate that rehabilitation has taken place; or for any other purpose. At the end of the period the court will sentence the offender after considering any additional information or developments that have occurred(Victims of Crime Bureau. et al., 2004).

Community service order

When this is imposed as a principal penalty, the offender is required to perform supervised unpaid work for the community up to a maximum of 500 hours can be imposed on any one occasion. The Community Offender Services are responsible for supervising (Victims of Crime Bureau. et al., 2004).

Bonds

There are several types of bonds. Under the conditions of a bond the court may record a conviction (Section 9) or elect not to record a conviction (Section 10, see below) but they basically require the offender to be on good behaviour for a certain length of time. Other conditions include probation where the offender is subject to the supervision and control of the Community Offender Services for a specific period of time, attend for

327 Appendix

drug or alcohol treatment, and appear before the court if called on to do so for any breach of the bond. (Victims of Crime Bureau. et al., 2004; Statistical Services Unit., 2007)

Bond without conviction /No conviction recorded /Conditional discharge

These are commonly referred to as a Section 10 and previously known as a Section 556A order. The court may elect not to proceed to a conviction but either discharge the offender with no penalty, with some nominal penalty, or with a good behaviour bond, or with an agreement to enter into an agreement to participate in and comply with an intervention program designed to promote the offender’s treatment or rehabilitation. If an offender breaches the condition to be of good behaviour he or she can be convicted and sentenced for the original offence (Victims of Crime Bureau. et al., 2004; Statistical Services Unit., 2007).

Fine

This is a monetary penalty which is expressed in penalty units. The maximum penalty units for each offence are set out in legislation. Currently one penalty unit is $110 (NSW Victim Crime Bureau et al., 2004).

Nominal sentence

This was previously referred to as the “Rising of the Court’ where the offender remains in court until its next adjournment. The offender is convicted but there is no formal sentence.

Dismissal

The offender is found guilty but the charges are dismissed without a conviction being recorded (Victims of Crime Bureau. et al., 2004).

Department of Juvenile Justice

The role of the Department of Juvenile Justice is to supervise and manage the care of juvenile offenders in detention and in the community. There are eight juvenile detention centres and one short term emergency centre in NSW. Centres provide a range of services, with the goal of reintegration into the community. Community services include assessments for courts, providing court ordered supervision such good behaviour bonds, probation, Community Service Orders (CSOs), parole orders; 328 Appendix

providing specific problem oriented programs, casework management, and youth conferencing(Department of Juvenile Justice.).

Penalties Juveniles

There are some differences in penalties given in Children’s Court reflecting the age of the offender as per the Children (Criminal Proceedings) Act 1987 (s 33) (Crime., 2004)

Juvenile Detention

A sentence to juvenile detention is also referred to as a Control Order. The sentence includes fixed term and offender is incarcerated in a juvenile facility.

Probation with or without a bond

Probation may occur with or without with Department of Juvenile Justice supervision and is for a designated period of time.

Good behaviour Bond

This can be with or without a recorded conviction and with or without conditions or supervision

Dismissed with caution

This is a caution the court determines the police could have given; record of the caution is kept and there is a limit on the number of cautions

Proven but dismissed without conviction and with caution – as above

Youth conference

The Court may refer the juvenile to a Youth Conference if police have not. See police procedures above. On being informed of successful completion, the court dismisses the charge.

Refer to Drug Court

The young offender must admit offence and have drug related offending and they are sentenced on completion of program.

329 Appendix

Costs related to penalties

For those offences where, as part of the sentence, there is “no supervision” the cost of the penalty is zero. All non-custodial penalties with supervision are allocated the same cost per day for the average length of the sentence; this also is applied to Community Service Orders based on 7 hours per day. The average cost per day for community corrections in NSW is $11.60 (PC Chapter 8, Corrective Services, 2008).

Juvenile conferencing description and additional information

In NSW, via the Young Offenders Act, Youth Justice Conferences were established to promote responsibility by the juvenile for his or her own behaviour, and to strengthen the family or family group, and to provide developmental and support services and to have due regard to the interests of any victim. These conferences are very resource intensive and not used principally for cannabis offences. In the 2006 police COPs data there were only 17 occurrences of Youth Justice Conferences for cannabis offences.

A conference can only be held if the young person admits to the offence after being given the opportunity to receive legal advice, and agrees to attend the conference, and if the offence and offender is deemed suitable for a conference. The decision to hold a Youth Justice Conference can be made by DPP, the courts, or by a NSW Police Specialist Youth Officer.

When a referral for a Juvenile Conference is made each youth is assessed for suitability by Juvenile Justice Department staff. Once the decision to hold a Conference is made, a convenor is appointed. Conveyors are recruited by Department of Juvenile Justice, subsequently trained, paid and resourced to run the conferences. Considerable resources are used in preparing for each conference including contacting various stakeholders, victims, police, and families; and notably it is not uncommon for a youth to not turn up for the scheduled conference.

Once the conference is held and a management plan is agreed to there are resources used to supervise the youth to ensure that the plan is carried out.

Responsibilities of the conveyor include organising the date, time and location of the conference and who will be attending. The convenor must consult with the person/s that made the referral, the subject of the conference, a responsible adult and any victims.

330 Appendix

The conference must be held at an agreed upon location but not at a police station, a court house or any office of the Department of Juvenile Justice. Persons attending include a responsible adult, supervising officer if the young person is under probation or community service order; extended members of the family; lay advocate; lay advisor; interpreter, convenor; investigating officer from the NSW Police; youth liaison person from NSW police (may); victim or representative plus a support person(New South Wales Consolidated Acts., 1997 (as at 27 November 2008). As a result of the conference a plan, which may include drug treatment, is then agreed upon and monitored.

Actual resources for individual Juvenile Justice Conferencing were not available therefore data from NSW Treasury papers were used to estimate an average cost of a conference. The annual budget data for Juvenile Justice Conferencing and number of conferences facilitated for the financial year 2007/08 were combined and the result was an average cost per conference of $5385 (or $5047in 2006/07 dollars). This does not apportion any expenditure to those referrals that do not proceed to conferences nor does it include police costs of arresting and processing the defendant. (http://www.treasury.nsw.gov.au/__data/assets/file/0017/11474/bp3_14jjust_f.rtf).

331 Appendix

Appendix – Chapter 7

Contingent valuation survey

Only the survey questions pertaining to the contingent valuation question are included here. The remainder of the questions including demographics and attitudes are included in the Chapter 10 Appendix.

There are four versions of the survey. Version 1 and Version 2 are identical except for the ordering of the fine values. Version 3 and Version 4 are identical except for the ordering of the Fine values. The WTP scenario in Version 1 and Version 2 starts with “You”. The WTP scenario in Version 3 and Version 4 starts with “Someone who is close to you”. Version 1 and Version 4 are presented.

Now we are asking a different question - please carefully read the hypothetical scenario below and answer the hypothetical question posed.

Scenario

You have been detected by the police in possession of 12 grams of cannabis. This is an amount of cannabis that is sufficient to make approximately 24 joints. This is the only offence you have ever committed. The police can charge you with a criminal offence. This would require you to attend Court, where if you plead guilty, or are found guilty by the Magistrate, would result in a criminal record. Having a criminal record may limit employment opportunities, ability to travel overseas to some countries, as well as carry the stigma of having a criminal record.

An alternative response by police may be to issue a fine. With this fine there would be no criminal record. Failure to pay the fine within 60 days results in additional interest charges, and if the fine is not paid, this will lead to the loss of drivers licence.

Thinking of the scenario you just read, which of the amounts listed below best describes the maximum fine you are willing to pay, to avoid court and a possible criminal record for possession of cannabis?

332 Appendix

CONSIDERING YOUR ABLITY TO PAY PLEASE SELECT THE MAXIMUM FINE YOU WOULD BE WILLING TO PAY TO AVOID A CRIMINAL RECORD

Fine $0 $50 $150 $250 $500 $750 $1000 $1250 $1500 $2000 $2500

For those who selected $2500 only

What is the maximum fine you would be willing to pay to avoid court and a possible criminal record? $______

For everyone

You chose $_X which is approximately __XX __% of your weekly income. Would you like to change your answer? Yes____ No____

If yes, what is the maximum fine you are willing to pay to avoid court and a possible criminal record? ______

Version 4

Now we are asking a different question - please carefully read the hypothetical scenario below and answer the hypothetical question posed.

Scenario

Someone who is close to you (i.e. a child, partner or sibling) has been detected by the police in possession of 12 grams of cannabis. This is an amount of cannabis that is sufficient to make approximately 24 joints. This is the only offence this person has ever committed. The police can charge them with a criminal offence. This would require attendance at Court, where if this person pled guilty, or was found guilty by the Magistrate would result in a criminal record. Having a criminal record may limit

333 Appendix

employment opportunities, ability to travel overseas to some countries, as well as carry the stigma of having a criminal record.

Thinking of the scenario you just read, which of the amounts listed below best describes the maximum fine you would be willing to pay, to avoid court and a possible criminal record for possession of cannabis for this person?

CONSIDERING YOUR ABLITY TO PAY PLEASE SELECT THE MAXIMUM FINE YOU WOULD BE WILLING TO PAY TO AVOID A CRIMINAL RECORD

Fine $2500 $2000 $1500 $1250 $1000 $750 $500 $250 $150 $50

For those who selected $2500 only

What is the maximum fine you would be willing to pay to avoid court and a possible criminal record? $______

For Everyone

You chose $_X which is approximately __XX __% of your weekly income. Would you like to change your answer? Yes____ No____

If yes, what is the maximum fine you are willing to pay to avoid court and a possible criminal record? ______

334 Appendix

Appendix – Chapter 8

Table A16: Australian prevalence of 12-month DSM-IV cannabis use disorder by sex and age. Males Females

Sample Weighted Sample Weighted Age % SE % SE Freq Pop Freq Pop

16–24 3.4 0.7 30 44,329 1.4 0.5 12 17,868 25–34 2.9 0.9 15 41,637 0.4 0.3 4 5,587 35–44 1.9 0.8 9 28,423 0.4 0.2 6 6,670 45–54 0.4 0.2 5 6,176 0.2 0.1 3 3,119 Overall 1.5 0.3 59 120,564 0.4 0.1 25 33,244 Source: NMHWB 2010 Survey (Teesson et al.);

The prevalence data in Table A16 were used to estimate numbers with CUD in NSW. It was necessary to convert the rates from those in Table A16 to match those used elsewhere. This was done as follows: [(1/3 of those using cannabis in the ages 14-19 year olds were given the rate of 1%) + 2/3 of those using cannabis in ages 14-19 year olds were given the rate of for those 16–24) + (of those 20 to 29 using cannabis 50% were given the rate of the 16–24 year olds + 50% were given the rate of the 25–34 year olds) and so on. All those 40+ were given the rate of the 45 to 54 year olds.

Once the numbers for those currently estimated were calculated, a ratio constructed of current CUD to the number currently using on more than 5 occasions. The ratio was then applied to estimated numbers consuming under the legalisation–regulation option (Method 1 and Method 2).

Table A17: Estimated rate of CUD in those who use cannabis on > 5 occasions/ year

Age categories Rates of CUD

14–19 27.7% 20–29 25.8% 30–39 27.3% 40+ 13.7% Source: estimated from NMHWB 2010 Survey (Teesson et al.) and NDSHS data

335 Appendix

Table A18: Projected increase in cannabis use disorder in NSW Use more Rates of CUD Projected CUD Projected CUD Table than 5 CUD if use Current Method 1 Method 2 occasions > 5 times N N % N N

14–19 10,579 44,745 27.7% 12,579 22,755 20–29 19,064 98,733 25.8% 20,899 28,873 30–39 13,740 67,695 27.3% 14,455 17,965 40+ 9,315 83,955 13.7% 9,924 14,585 Total 295,128 57,857 84,177 52,699 95% CL 28,139 – 87,540 41,618 – 126,674

Additional CUD 5,158 31,479

95% CL 2,704-7,607 16,183- 46,740

Source: estimated from NMHWB 2010 Survey (Teesson et al.) and NDSHS data

336 Appendix

Appendix – Chapter 10

Design from NGENE for final survey

Choice alt1.a alt1.b alt1.c alt1.d alt1.e alt2.a alt2.b alt2.c alt2.d alt2.e alt3.a1 alt3.b1 alt3.c1 alt3.d1 alt3.e1 Block 5 2 0 18 30 80 1 0 18 10 60 2 0 8 20 60 1 10 2 0 8 40 60 1 0 18 30 20 2 0 8 20 60 1 15 0 0 3 10 80 3 0 8 40 40 2 0 8 20 60 1 19 1 0 18 30 20 0 0 8 20 80 2 0 8 20 60 1 20 3 1 8 30 40 0 0 8 10 20 2 0 8 20 60 1 23 1 0 18 40 80 1 0 13 40 60 2 0 8 20 60 1 24 1 0 13 10 60 3 1 13 30 40 2 0 8 20 60 1 34 0 0 13 10 20 3 0 3 30 60 2 0 8 20 60 1 36 2 0 13 10 20 3 1 3 30 60 2 0 8 20 60 1 6 1 0 13 40 40 2 0 8 10 60 2 0 8 20 60 2 11 0 0 13 30 80 2 0 13 20 80 2 0 8 20 60 2 14 3 1 8 10 60 2 0 13 40 20 2 0 8 20 60 2 17 3 0 13 30 20 0 0 18 30 40 2 0 8 20 60 2 18 3 0 3 20 40 1 0 13 20 40 2 0 8 20 60 2 21 0 0 8 20 20 2 0 8 30 60 2 0 8 20 60 2 27 3 1 13 20 40 1 0 3 30 40 2 0 8 20 60 2 30 2 0 8 30 40 3 1 13 20 20 2 0 8 20 60 2 35 1 0 8 20 20 3 1 8 30 40 2 0 8 20 60 2 1 3 1 8 20 60 1 0 3 40 80 2 0 8 20 60 3 2 1 0 8 40 60 3 0 8 20 20 2 0 8 20 60 3 3 0 0 3 40 40 3 0 18 10 40 2 0 8 20 60 3

337 Appendix

12 3 0 8 10 60 0 0 3 40 60 2 0 8 20 60 3 13 3 1 3 40 20 2 0 3 20 80 2 0 8 20 60 3 25 2 0 18 30 20 0 0 18 10 40 2 0 8 20 60 3 26 3 0 3 20 40 0 0 8 40 20 2 0 8 20 60 3 31 0 0 3 40 60 3 0 18 10 20 2 0 8 20 60 3 33 2 0 3 40 40 2 0 13 10 40 2 0 8 20 60 3 4 1 0 18 40 40 3 1 3 20 60 2 0 8 20 60 4 7 3 0 3 10 60 0 0 3 40 60 2 0 8 20 60 4 8 3 1 18 10 20 3 0 8 40 60 2 0 8 20 60 4 9 0 0 18 20 40 2 0 18 30 40 2 0 8 20 60 4 16 3 0 18 30 20 0 0 3 20 80 2 0 8 20 60 4 22 1 0 3 20 80 1 0 18 20 20 2 0 8 20 60 4 28 2 0 18 30 60 1 0 18 10 80 2 0 8 20 60 4 29 2 0 13 20 80 3 1 13 10 20 2 0 8 20 60 4 32 0 0 13 10 60 2 0 13 40 20 2 0 8 20 60 4

338 Appendix

THE UNIVERSITY OF NEW SOUTH WALES PARTICIPANT INFORMATION STATEMENT

HREC 09127

Internet Based Survey on Cannabis (Marijuana) Policy

Measuring preferences for cannabis policies

Purpose of the study

We are carrying out a survey on options for cannabis (marijuana, pot) policies and we would like your views on some alternatives. The questionnaire asks about a limited set of policies and we realise you may prefer other features in real life; however, it is important to us to have your views on these alternatives. We will use the results to advise policy makers.

Description of study

If you decide to participate, the survey will take about 15 minutes to complete. We are asking you to complete a number of survey questions which ask you to make a choice between different policies. We also ask some questions about you including questions about previous cannabis use.

Confidentiality and disclosure of information

This is an anonymous survey. Your name will not be individually identified in any publication arising from this research. It is completely up to you whether you participate in the study. If you do not wish to complete the survey after you have begun, you are free to stop at any point. Your decision not to participate will not prejudice your future relations with the University of New South Wales. We are not asking you to sign a consent form as your decision to participate implies consent.

These data once aggregated may be included in a peer reviewed paper. We may also present the results at an academic conference. Any information that is obtained in connection with this study and that can be identified with you will remain confidential and will be disclosed only with your permission except as required by law.

To participate in this study, you must be 18 years or older. There are no identifiable risks from participating in this study. We cannot and do not guarantee or promise that you will receive any benefits from this study.

339 Appendix

This study is being conducted by Marian Shanahan, and Assoc Prof Alison Ritter from the Drug Policy Modelling Program, NDARC, The University of New South Wales and Dr. Karen Gerard, University of Southampton, UK.

Complaints

Complaints may be directed to the Ethics Secretariat, The University of New South Wales, SYDNEY 2052 AUSTRALIA (phone 9385 4234, fax 9385 6648, email [email protected]). Any complaint you make will be investigated promptly and you will be informed of the outcome.

Further information

If you have any questions, please feel free to contact Marian Shanahan on (02) 9385 0229.

Survey Screening question (Target 33% yes; 60% no ) 1. Have you ever used cannabis / marijuana? Yes No

First we would like to ask some questions about you. Demographics and Characteristics 2. Are you: Male  Female  3. What is your current age in years ? ______

4. What is your Marital Status? Never Married (Single)  Married / Defacto  Separated/Divorced/Widowed  Prefer Not to State/other 

5. Are there any dependent children in your household? No  Yes  _____ 6. Number of dependent children

7. What is your employment status (Select one)? Employed Full-time   Employed Part-time (Less than 35 Hours)  Retired/Unable to Work/Disabled  Full time student  Not Working/Looking for Work 

340 Appendix

Housewife/Househusband

8. What level of education have you completed(Select one)? Left school before finishing year 10 9.  Finished year 10 (School Certificate equivalent) 10.  Finished year 12 (completed HSC equivalent) 11.  Completed a TAFE/ Diploma 12.  Completed a university degree 13.  Other (Please specify ) ______) 14. 

15. What is your gross household income, before tax or other deductions, from all sources? Please include any pensions and allowances, and income from interest or dividends. $1 - $199 per week ($1- $10,399 per year)  $200 - $299 per week ($10,400 - $15,599 per year)  $300 - $399 per week ($15,600 - $20,799 per year)  $400 - $499 per week ($20,800 - $25,999 per year)  $500 - $599 per week ($26,000 - $31,199 per year)  $600 - $699 per week ($31,200 - $36,399 per year)  $700 - $799 per week ($36,400 - $41,599 per year)  $800 - $999 per week ($41,600 - $51,999 per year)  $1,000 - $1,499 per week ($52,000 - $77,999 per year)  $1,500 - $1,999 per week ($78,000 - $103,999 per year)  $2,000 -$2,499 per week ($104,000 -$129,999 per year)  $2,500 - $3,499 per week ($130,000 - $181,999 per year)  $3,500 or more per week ($182,000 or more per year)  Prefer Not to State  Don’t Know 

341 Appendix

Preferences for cannabis policies In this section of the survey you are being asked to choose between hypothetical policy options for cannabis (marijuana/pot/grass/ganga). There are nine hypothetical choices. First we provide you with an example of the type of questions we will be asking in this section, and then before we give you the questions some information on the various features of the policies are provided.

THIS IS AN EXAMPLE Description Option A Option B Current Policy If an adult is found with Civil fine or cannabis attend — less than 15 grams No offence - educational Cannabis caution legally traded session but police possible but good; Crime if may choose to police may provided to persons charge. choose to charge. <18 years of age. Arrested and go Arrested and go — more than 15 grams to court. to court. Where is cannabis Illicit drug dealer Illicit drug dealer Cannabis shop purchased? sells to anyone sells to anyone Cost to State to treat health problems from $30 million $5 million $20 million cannabis Costs to State to enforce cannabis laws $60 million $40 million $60 million Number of 14 to 29 year olds using cannabis at 3 in 100 13 in 100 8 in 100 least once a month If you could choose a policy, which policy  X  would you chose?

In this study there are four alternatives for cannabis policy. They are:

• All cannabis offences are considered criminal, that is, if a person is found with any amount of cannabis they can be arrested and receive a criminal record if found guilty.

• Those found with a small amount (less than 15 grams) of cannabis receive a civil fine but if the fine is not paid, the offender may lose their driver’s licence. If a person is found with larger amounts of cannabis, they will likely be arrested and receive a criminal record if found guilty.

342 Appendix

• Those who are found with a small amount of cannabis may be ‘cautioned’ by police. If a person is found with larger amounts of cannabis, they will likely be arrested and receive a criminal record if found guilty.

• Cannabis use, possession and supply of any amount of cannabis are legal and regulated for all those 18 years of age and over.

In the questionnaire there are two locations where cannabis may be purchased — either from a cannabis shop where the quality of the cannabis is known and only cannabis may be purchased or alternatively it may be bought from an illicit drug dealer where other drugs may also be obtained.

In Australia, currently, about 3 in every 100 people who use cannabis in any one year are detected by police.

There are many things to consider when choosing a policy option.

Possible benefits, and costs savings arising from the various policy options have to be weighed against the costs of enforcing the policies: police costs; courts; jails; and ensuring those under 18 are not able to access cannabis. Many individuals report a sense of pleasure from the occasional use of cannabis but different policies impose different costs on cannabis users (for example, getting a criminal record, paying a fine, the stigma of a criminal record, or infringements of civil liberties).

Does the policy have a reasonable chance of reducing the harms associated with cannabis use, such as: risk of addiction (about 1 in 5 who use cannabis more regularly will become addicted); adverse impacts on adolescent mental health and educational attainment; or risk of motor vehicle accidents?

Does the policy reduce the costs associated with adverse impacts of use, such as costs to the health care system and costs of reduced participation in education and the workplace?

Please consider each hypothetical policy and its outcomes carefully and then indicate which option you most prefer — each choice presents a different set of cannabis policy options. We appreciate these may not seem ideal but telling us which option is the most

343 Appendix

preferred will provide valuable information. Remember there are no right or wrong answers.

There are 9 scenarios – Please complete ALL of them and the following questions. (There were 36 in total survey questions, 9 each across the four versions and are not included here)

Cannabis and you 16. Have you used cannabis / marijuana in the last 12 months? (if No please go to Q 18) Yes  No (Go to Q12) 

17. In the last 12 months, how often did you use cannabis? Every day   Once a week or more  About once a month  Every few months  Once or twice a year 18. Do you think there any benefits to a person’s health from cannabis? Yes  No  Don’t know 

19. If YES, indicate below [SELECT ALL THAT APPLY] Relieves stress  Helps people with AIDS  Makes you feel good  Improves concentration  Helps asthma  Improves appetite  Increases sex drive  Relieves nausea  Relieves pain  Helps glaucoma  Relieves depression  Other -Please specify ______

20. Do you think cannabis is addictive? Always  Sometimes  Never  Don’t know  I do not wish to answer this question 

General Questions 21. Have you been present at a criminal court proceeding in any capacity within the past decade or so? Yes, only once  344 Appendix

Yes, more than once  No  Don't know  22. In politics, people sometimes talk about the 'left' and the 'right'. 23. Where would you place yourself on a scale from 0 to 10, where 0 means the left and 10 means the right? Please circle your choice.             Left0 1 2 3 4 5 6 7 8 9 Right10 Can’t choose This is the end of the survey – THANK YOU

345 Appendix

Commands for Mixed logit regression analysis Nlogit ;lhs = choice, cset, altij ;choices = PolA, PolB, NSWpol ;prob=prFMXL1000 ;utility= utFMXL1000 ;Halton ;rpl = age, SEXF0, BENEEF, addicef, CANUYREF, LEFTEF, RGHTEF, MISSLFRT ;fcn= ConA[n],pol0(n), pol1(n),pol3(n),USE0_EFF(n),USE2_EFF(n),USE3_EFF(n) ;pts=1000 ;parameters ;pwt ;pds=9 ;Model: U(PolA) = ConA+ pol0*Pol0 + pol1* Pol1+ pol3* Pol3 + HEAL_ACT*HEAL_ACT +CJS_ACT*CJS_ACT + USE0_EFF*USE0_EFF + USE2_EFF*USE2_EFF + USE3_EFF*USE3_EFF + BLOC*BLOC + SEXF0*SEXF0 + AGE*AGE + EDULT12*EDULT12 + EDUEQ12*EDUEQ12 + EDUUNI*EDUUNI + CANUYREF*CANUYREF + BENEEF*BENEEF + ADDICEF*ADDICEF + MS_NEV*MS_NEV

U(PolB) = ConA+ pol0*Pol0 + pol1* Pol1+ pol3* Pol3 + HEAL_ACT*HEAL_ACT +CJS_ACT*CJS_ACT + USE0_EFF*USE0_EFF + USE2_EFF*USE2_EFF + USE3_EFF*USE3_EFF + BLOC*BLOC + SEXF0*SEXF0 + AGE*AGE + EDULT12*EDULT12 + EDUEQ12*EDUEQ12 + EDUUNI*EDUUNI + CANUYREF*CANUYREF + BENEEF*BENEEF + ADDICEF*ADDICEF + MS_NEV*MS_NEV

U(NSWpol) = SEXF0 *SEXF0 + age*age+ EDULT12 *EDULT12 + EDUEQ12 *EDUEQ12 +...

346 Appendix

Table of final results with all interactions Standard Variable Coefficient Error b/St.Er. P[Z>z] CONSTANT 1.9156604 0.236894 8.087 0.0000 Random parameters CRIMINALISATION -1.44684 0.257554 -5.618 0.0000 CIVIL PENALTY 0.7455895 0.182017 4.096 0.0000 LEGALISATION| 0.5063752 0.255573 1.981 0.0476 USE 3/100 AGE 14 TO 29 0.6199762 0.153377 4.042 0.0001 USE 13/100 AGE 14 TO 29 -0.112754 0.150404 -0.75 0.4535 USE 18/100 AGE 14 TO 29 -0.880137 0.1652 -5.328 0.0000 Non random parameters HEALTH EXPEND -0.027799 0.002227 -12.481 0.0000

CJS EXPEND -0.026148 0.001303 -20.067 0.0000 LOCATION OF PURCHASE (LEGAL) 0.4442658 0.043259 10.27 0.0000 SEX FEMALE =0 -0.193479 0.06348 -3.048 0.0023 AGE -0.006174 0.00462 -1.336 0.1815 EDUCATION < Y12| -0.258787 0.116531 -2.221 0.0264 EDUCATION = Y12 0.1815071 0.118895 1.527 0.1269 EDUCATION – COMPLETE DEGREE -0.0942 0.110122 -0.855 0.3923 USED CANNABIS RECENTLY 0.0226605 0.106704 0.212 0.8318 BELIEVE BENEFITS FROM CANNABIS -0.171342 0.072495 -2.363 0.0181 BELIEVE CANNABIS ADDICTIVE 0.0422294 0.088509 0.477 0.6333 NOT MARRIED/DEFACTO 0.0809051 0.084712 0.955 0.3395 Interactions with random parameters CRIM:AGE -0.00393 0.00460 -0.85400 0.3932 CRIM:SEX 0.03582 0.06970 0.51400 0.6073 CRIM:BEN -0.58404 0.08514 -6.86000 0.0000 CRIM:ADD 0.17102 0.09929 1.72200 0.0850 CRIM:CAN -0.83306 0.14809 -5.62500 0.0000 CRIM:LEF -0.66751 0.15796 -4.22600 0.0000 CRIM:RGH 0.47117 0.15357 3.06800 0.0022 CRIM:MIS 0.22593 0.14344 1.57500 0.1152 CIV PEN:AGE -0.00846 0.00341 -2.47900 0.0132 CIV PEN:SEX -0.06477 0.05139 -1.26100 0.2075 CIV PEN:BENIFIT -0.18202 0.06053 -3.00700 0.0026 CIV PEN:ADDICT| 0.16241 0.07356 2.20800 0.0272 CIV PEN:USE CANNABIS| -0.37838 0.09777 -3.87000 0.0001 CIV PEN:LEFT_POLIT 0.06177 0.11087 0.55700 0.5774 CIV PEN:RGH_POLIT -0.10477 0.11068 -0.94700 0.3438 CIV PEN:MISS_POLIT 0.02892 0.10592 0.27300 0.7848 LEGAL:AGE 0.01339 0.00491 2.72600 0.0064 LEGAL:SEX 0.00477 0.07247 0.06600 0.9475 LEGAL:BENEFIT 0.86089 0.08336 10.32700 0.0000 347 Appendix

LEGAL:ADDICT| -0.27512 0.09962 -2.76200 0.0057 LEGAL:USED CANNABIS| 1.09199 0.12641 8.63900 0.0000 LEGAL:LEF 0.76629 0.15462 4.95600 0.0000 LEGAL:RGH -0.49771 0.16175 -3.07700 0.0021 LEGAL:MIS -0.35760 0.15027 -2.38000 0.0173 USE 3/100 AGE 14 TO 29:AGE -0.00378 0.00287 -1.31500 0.1884 USE 3/100 AGE 14 TO 29:SEX -0.06973 0.04354 -1.60200 0.1093 USE 3/100 AGE 14 TO 29:BENNEFIT -0.04586 0.04957 -0.92500 0.3549 USE 3/100 AGE 14 TO 29:ADDICT 0.06619 0.06076 1.08900 0.2760 USE 3/100 AGE 14 TO 29:USED CANNABIS| -0.07132 0.08090 -0.88200 0.3780 USE 3/100 AGE 14 TO 29:LEFT_POLIT| 0.08143 0.09505 0.85700 0.3916 USE 3/100 AGE 14 TO 29:RGH_POLIT| -0.15390 0.09661 -1.59300 0.1112 USE 3/100 AGE 14 TO 29:MISS_POL 0.03990 0.09044 0.44100 0.6591 USE 13/100 AGE 14 TO 29:AGE -0.00147 0.00291 -0.50400 0.6142 USE 13/100 AGE 14 TO 29:SEX 0.00998 0.04328 0.23100 0.8176 USE 13/100 AGE 14 TO 29:BENNEFIT -0.03618 0.04989 -0.72500 0.4683 USE 13/100 AGE 14 TO 29:ADDICTIVE 0.02965 0.06048 0.49000 0.6240 USE 13/100 AGE 14 TO 29:USED CANNABIS 0.04796 0.07545 0.63600 0.5250 USE 13/100 AGE 14 TO 29:LEFFT_POLIT -0.14876 0.09043 -1.64500 0.1000 USE 13/100 AGE 14 TO 29:RGH 0.13847 0.09248 1.49700 0.1343 USE 13/100 AGE 14 TO 29:MISS_POLIT 0.02983 0.08907 0.33500 0.7377 USE 18/100 AGE 14 TO 29:AGE 0.01092 0.00314 3.47900 0.0005 USE 18/100 AGE 14 TO 29:SEX -0.01128 0.04716 -0.23900 0.8110 USE 18/100 AGE 14 TO 29:BEN 0.11253 0.05429 2.07300 0.0382 USE 18/100 AGE 14 TO 29:ADD -0.08463 0.06784 -1.24800 0.2122 USE 18/100 AGE 14 TO 29:USED CANNABIS -0.01006 0.08402 -0.12000 0.9047 USE 18/100 AGE 14 TO 29:LEF 0.04199 0.10443 0.40200 0.6876 USE 18/100 AGE 14 TO 29:RGH_POLIT 0.07080 0.10477 0.67600 0.4992 USE 18/100 AGE 14 TO 29:MISS_POLIT -0.00810 0.10031 -0.08100 0.9357 Standard Deviations of random parameters NsCONA 1.62713 0.06618 24.58600 0.0000 NsCRIM 1.49368 0.07416 20.14000 0.0000 NsCIV PEN | 0.78887 0.06965 11.32600 0.0000

348 Appendix

NsLEGAL | 1.84525 0.07613 24.23900 0.0000 NsUSE 3/100 AGE 14 TO 29_E 0.37501 0.07853 4.77500 0.0000 NsUSE 13/100 AGE 14 TO 29_E 0.25389 0.10128 2.50700 0.0122 NsUSE 18/100 AGE 14 TO 29_E 0.32875 0.11174 2.94200 0.0033

349