Design of Auto Insurance Systems in Ontario

The Design of Auto Insurance Systems: Research and Implications for Ontario

Sharon Tennyson Mary Kelly Anne Kleffner Associate Professor Associate Professor Associate Professor Department of Policy School of Business & Haskayne School of Analysis and Management Economics Business Cornell University Wilfrid Laurier University University of Calgary

FINAL REPORT TO INSURANCE BUREAU OF CANADA August 23, 2012

Summary

This study aims to provide guidance in auto insurance system design and considerations for achieving a well‐functioning automobile insurance system in Ontario. Such a system is characterized by a number of desirable features, including fair compensation, deterrence, efficiency, and fairness.

However, because system features designed to achieve one of these features may make it harder to achieve some others (e.g., compensation and deterrence, efficiency and fairness), trade‐offs are often required to balance conflicting objectives. Different stakeholders may also have conflicting objectives; for these reasons, establishing whether an auto insurance system is functioning well is not a simple task.

The report describes a variety of combinations of the basic design features of auto insurance systems are used in Canada and elsewhere, and compares the performance of auto insurance systems in

Canada based on the measures that characterize a well‐functioning system. This comparison demonstrates that Ontario’s auto insurance system performance is worse than those in other provinces on many dimensions and much worse in the areas of affordability and sustainability. However, although the Ontario automobile insurance system exhibits a number of problems, the evidence does not suggest that the design of the system is fundamentally flawed.

Background research on Ontario’s insurance system and the history of insurance reforms is presented to shed light on the factors that explain this poor performance. That analysis yields several key findings. First, reform policies in Ontario appear to have affected outcomes in the short run, but in many cases inflationary trends quickly reappeared. Second, Ontario claim frequency and severity are higher than those in other provinces in almost all cases and all years, and the increases over time in both claim frequency and claim severity have been substantially greater in Ontario than in other provinces.

This is despite having the lowest fatal accident rate and one of the lowest rates of accidents causing injury of all provinces. A final insight obtained from this analysis is that the claims growth in Ontario i

stems primarily from the Greater Toronto Area (GTA) rather than reflecting province‐wide problems.

Inflationary trends are highest in the GTA, due to increases in both claim frequency and claim severity.

Thus, while the auto insurance product is sustainable across most of Ontario, premiums in the GTA are insufficient to cover the cost of auto insurance claims in the GTA.

Based on this review and on analysis of problems and solutions in other insurance systems, key issues in the design of a well‐functioning auto insurance system are discussed. The overall conclusions of the report are as follows:

 Insurance provides many benefits to individuals and to society, but also may increase incentives to drive, reduce driving precautions, and lead to benefit‐seeking or even outright fraudulent claiming. The specific design features of an insurance system affect these behavioural incentives. Incentive problems are greatest when insurance benefits are generous and pricing is socialized so that premiums do not reflect individual risk of loss. A well‐functioning insurance system must recognize and attempt to counteract these incentive problems.  When insurance availability and affordability to all consumers are important objectives, adjusting the insurance system to address cost problems is difficult because many solutions will reduce availability and/or affordability for some. In jurisdictions in which driving is not considered a right or an economic necessity, the public policy problems in auto insurance are less complicated.  Auto insurance does not operate in a vacuum: it interfaces with the private and public health systems, road safety and driver licensing mechanisms, the legal system, demographics and economics of a region. Recognizing this, policy makers should consider problems and solutions that lie outside of the insurance system along with those internal to the system when addressing auto insurance issues.  In the aggregate, rising costs are the source of rising insurance premiums. Because the link between cost increases and premium increases is not transparent, there is often pressure to reduce premiums without recognition that the underlying problem is one of cost. However, suppressing premiums does not lead to cost reductions and may even exacerbate cost increases. Insurer or public policy responses aimed squarely at underlying cost and incentive problems are more effective and more permanent responses.

 Because of the threshold, excessive claiming behaviour is more problematic in partial no‐fault systems than in pure no‐fault systems because benefit overuse arises due to attempts to breach the tort threshold. The tort threshold, the magnitude of additional benefits available if the threshold is breached, and the penalties for fraudulent claims should be constructed so that the “lottery” aspects of partial no‐fault are minimized.

These considerations suggest that the following principles should be observed when enacting auto insurance reform:

 It is important that policy makers clearly articulate the objectives of the insurance system and the trade‐offs that must be made, recognizing that trade‐offs exist between different design options.  Generous first‐party insurance, and partial no‐fault systems, lead to benefit‐seeking behaviours which must be monitored and controlled. Claimants have incentives to over‐claim when first‐party benefits are generous. Measures to reduce fraudulent claiming and reduce claims costs should be established at the industry level with support of government enforcement mechanisms. These measures could include harsh penalties for fraudulent healthcare providers, better use of analytics to detect fraud, and consumer education.  If affordability is addressed via premium subsidies these should be based on income and not on risk: full premium subsidies should be available to low risk and low income individuals. Subsidies available to low income but high risk individuals should reflect the underlying poor driving record or accident experience.  The provision of insurance reduces incentives to take care. Incentives for safe driving and responsible claiming are better aligned if premiums paid by drivers are commensurate with their risks. Low‐risk drivers should be provided with opportunities to distinguish themselves from high‐ risk drivers in the rate classification system. High‐risk drivers should be identified by a system of experience rating that incorporates both driving record and at‐fault claims.  Cost controls on healthcare services are important to prevent cost shifting to the auto insurance system.  The long run success of changes to system design to achieve policy goals, whether it is compensation levels, or benefit provision or pricing structure, are impacted by consumer knowledge of insurance and societal beliefs about fairness and societal attitudes towards excessive or fraudulent claiming behaviours.

iii

 Government insurance providers have often been more successful than private insurers in containing claims costs due to their greater control over compensation and indemnification schedules. However, the trade‐off involves potential under‐indemnification if benefits to those most seriously injured are capped (as, for example, in Quebec), and the loss of consumer choice of products and providers. Moreover, the full costs of the auto insurance system are not reflected in premiums if the government insurer runs a deficit.

Specific additional considerations with respect to Ontario auto insurance include the need to realign premium structure to reduce cross‐subsidization across the province. Although in theory premiums are risk‐based, decades of prior approval rate regulation have created cross‐subsidies across driving record classes and across territories. It is also imperative to develop a distinct pricing mechanism that is fair to new drivers and maintains incentives for safe driving.

Contents

Summary ...... i

1 Introduction ...... 1

2 Background on Automobile Insurance ...... 4

2.1 Economics of Automobile Insurance ...... 4

2.2 Objectives of an Automobile Insurance System ...... 7

3 Design Considerations of an Automobile Insurance System ...... 9

3.1 Mandatory Insurance Coverage ...... 10

3.1.1 Residual Markets ...... 13

3.2 Public versus Private Insurance ...... 15

3.2.1 Public Intervention in Private Insurance Markets...... 19

3.3 Tort versus No‐fault Compensation Systems ...... 21

3.4 Pricing Mechanisms ...... 27

4 Automobile Insurance Systems Design and Performance ...... 36

4.1 Insurance Systems in Canada ...... 36

4.2 Comparing System Performance ...... 40

5 Automobile Insurance in Ontario ...... 57

5.1 The Current Product ...... 57

5.2 The History of Auto Insurance Reform in Ontario ...... 59

5.2.1 Bill 68 (1990) ...... 63

5.2.2 Bill 164 (1994) ...... 65

5.2.3 Bill 59 (1996) ...... 66

5.2.4 Bill 198 (2003) ...... 67

5.2.5 Elimination of Designated Assessment Centres (2006) ...... 69

5.2.6 Reforms of 2010 ...... 69

5.3 Ontario Claim and Cost Trends ...... 70

5.3.1 Bodily Injury Liability Claims ...... 71

5.3.2 Compulsory Property Damage Claims ...... 78

5.3.3 Statutory Accident Benefits ...... 83

5.4 System Performance ...... 89

5.5 Summary and Interpretation ...... 95

6 Experience from Other Jurisdictions ...... 99

6.1 Insurance Systems in Other Countries ...... 99

6.2 Summary of Case Studies ...... 101

7 Policy Considerations for a Well‐Functioning Insurance Market ...... 105

7.1 Policy Considerations for Ontario ...... 108

7.1.1 Pricing Mechanism Design ...... 110

7.1.2 Benefit Provisions ...... 113

7.1.3 Claims Cost Controls ...... 114

7.1.4 Consumer Education ...... 118

7.1.5 Larger Policy Issues ...... 120

8 References ...... 122

9 Appendix: Case Studies of Other Jurisdictions ...... 128

9.1 Alberta, Canada ...... 128

9.2 Massachusetts, U.S.A...... 132

9.3 Michigan, U.S.A...... 138

vii

1 Introduction

Automobiles and driving have a fundamental role in our society. Although driving is not a “right” according to various courts of law in Canada1, car ownership and driving lead to better employment outcomes and higher wages (Raphael and Rice, 2002). The costs associated with automobile accidents are equally important to society. In 2009 there were 2,011 fatal collisions and 123,192 additional collisions causing personal injury across Canada. Over 2,200 individuals died and 11,450 suffered serious injuries in car collisions (Transport Canada, 2011). It is estimated that the cost of traffic accidents is about 1.9 percent of GDP in Canada (International Transport Forum, 2011). As the primary mechanism for funding these accident costs, the automobile insurance system is an important concern. In addition, the social importance of the automobile and the extent of automobile accident compensation in North American society generate public interest in automobile insurance.

There are many different automobile insurance systems worldwide and 13 different (provincial) insurance systems across Canada alone. There are significant differences in the insurance product and regulatory regimes across the provinces. Market conduct regulation of insurance is a provincial (and territorial) responsibility. As such, each province has its own rules for underwriting and marketing insurance policies, and its own regulatory approach to product and price oversight. Generally, products and prices are monitored and controlled, rating and underwriting are regulated, and insurers face extensive filing requirements.

Regulation of the automobile insurance system may be imposed to achieve a number of different objectives. In Canada an important objective of regulation is to achieve social goals due to the social risk created by driving. Drivers are exposed to risk of significant loss from automobile accidents, and in turn

1 See Regina v Tri‐M Systems (B.C.S.C.), note 24: http://sense.bc.ca/disc/stead2.htm. 1

impose accident risk and accident costs on others. The purchase of automobile insurance helps to internalize the costs of accidents through payment of policy premiums that reflect the expected costs of driving. When a driver chooses to drive uninsured, accident costs are shifted to other drivers or to society as a whole. The cost to an individual from driving uninsured is small, as the majority of personal injury costs are shifted to society through taxes and/or levies on insurance policies.

The potential for drivers to shift accident costs to others is one reason for government regulation of automobile insurance. The existence of publicly funded, universal medical care in Canada makes it easy for drivers to shift this cost to others, creating an additional reason to make the purchase of auto insurance compulsory. The cost of providing medical care to those injured in automobile accidents is substantial. A potential for overutilization of medical care exists if insurance benefits are generous and there is a separation between the provision of care and the financing of care. Thus, there may be a role for governments to optimize the use of the medical system. The division of public and private responsibilities for the provision of auto insurance may be chosen from a wide spectrum of arrangements and is a critical design element of the automobile insurance system.

Since the early 1990s Ontario has made a number of significant changes to the automobile insurance product in an effort to control costs and improve compensation. Changes began with the introduction of no‐fault insurance in 1990 and include the more recent reductions to statutory accident benefits. Despite a series of regulatory interventions over time, the automobile insurance system in

Ontario has consistently been problematic for the government and for the industry. This report aims to contribute to the process of creating a more sustainable insurance system for the province.

Rather than focusing solely on concerns in Ontario, this report uses insights from theoretical and comparative analysis of auto insurance systems to identify principles for designing a sustainable

automobile insurance system. The analysis reviews insurance theory and evidence on the experiences of other jurisdictions to identify practices that work well for the insurance industry, consumers, and society.

Section 2 of the report provides the conceptual framework for automobile insurance and identifies the objectives of a well‐functioning automobile insurance system. These are comprised of availability, affordability, service quality, cost efficiency, sustainability, fairness, informed consumers, and adequate competition. In Section 3 the key policy levers that are available when designing an insurance system are described. These are compulsory versus optional insurance; public versus private provision; tort versus no‐fault compensation; and pricing schemes. We summarize the evidence from academic research regarding the strengths, weaknesses, and overall effectiveness of each design feature as employed in existing insurance systems.

Because an auto insurance system combines choices regarding each of the individual system elements, there exists a wide range of automobile insurance systems. The systems employed in Canada are documented in Section 4. We compare the performance of the auto insurance systems in each of the provinces in Canada, defining metrics to measure the objectives outlined in Section 2. Section 5 reviews the history of auto insurance reform in Ontario and examines the statistical evidence on the historical performance of Ontario’s system using the lens of this analysis. We consider the performance of Ontario in relation to other provinces, and we also compare the performance of different geographical regions of

Ontario. In Section 6 we briefly discuss auto insurance systems and auto insurance reforms in jurisdictions outside of Ontario. Case studies provided in the Appendix support the discussion in Section 6. We conclude with policy considerations for a well‐functioning insurance market and specific considerations for the Ontario insurance market in Section 7.

2 Background on Automobile Insurance

2.1 Economics of Automobile Insurance

Insurance is a mechanism by which one can protect against the risk of a future contingent loss.

The buyer pays a fixed sum of money, the premium, to an insurer in exchange for the insurer’s promise to compensate the insured for a potential financial loss. Because individuals are typically risk averse, they prefer to purchase insurance instead of facing the risk of the loss. The insurer pools together the premiums of many individuals to pay for the future losses that will be incurred. At its most basic level, each insurance buyer is charged a premium that reflects the expected costs that he or she imposes on the insurance system, and the premiums paid by all buyers are pooled to pay the losses of those who experience a loss.

Automobile insurance provides compensation to parties who suffer injuries or incur damages due to automobile accidents and losses. Given the variety of losses which may be incurred, automobile insurance is actually a package of insurance coverages that provides protection for both damage to property and injuries to persons. Because there is the potential for causing losses or damages to others as well as to one’s self, the package includes both liability (“third‐party”) insurance and first‐party insurance.

First‐party insurance covers costs of one's own automobile‐related property or injury losses. Liability insurance covers damages to other persons or their property that arise in an automobile accident with the insured, and for which the insured is held responsible. The function of liability insurance differs in some respects from that of first‐party insurance. In addition to protecting the insured driver from the unexpected costs associated with automobile accidents that are his fault, liability insurance provides protection to other drivers by assuring that funds will be available for their compensation in such a circumstance.

Insurance has many economic benefits. Insurance reduces uncertainty for risk‐averse drivers by providing protection against losses from automobile accidents. Insurance also increases the productivity of financial resources by removing the need for individual drivers to hold large contingency funds (Curak et. al, 2000). Thus, insurance provides great benefits to individuals and to society. However, the cost of insurance and the potential solvency of the insurance provider is impacted by moral hazard in both risk‐ taking and claiming behaviours, and by adverse selection.

Moral hazard arises from the fact that in the presence of insurance, individuals have incentives to take less care than they would otherwise. In auto insurance, for example, individuals may drive with less care or be less careful when choosing a location to park a vehicle, because they know they are insured.

Moral hazard exists if this change in behaviour increases the likelihood or size of a loss, and if insurers cannot directly observe the actions of the insured or if it is prohibitively costly to do so. Because insurance protects drivers from immediately bearing the full costs of accidents that they cause, the effects of insurance on incentives for safe driving must be carefully considered.

Most insurance contracts are contracts of indemnity – their purpose is to put the insured into the same financial condition as before the loss. However, the nature of the insurance product creates incentives to report fictitious losses or exaggerated losses, since receiving insurance payments for a loss that was not experienced leads to financial gain for the claimant. This negative consequence of insurance is intrinsic to all insurance systems, and the design of the insurance contract, insurance contract law, claims handling practices and even the selection of insurance customers must take this into account

(Baker 1994). Fraudulent claiming is often an opportunistic response to the insurance contract. That is, individuals may decide whether to file an insurance claim, and the amount for which to file, based on the expected gains from filing relative to the costs of filing. Preventing and detecting this kind of “claims

abuse” is an important facet of insurance management in both private and public insurance systems.

Fraudulent claiming may also result from organized criminal activity, however.

Moral hazard can be reduced if insured individuals are required to pay part of any loss themselves, either by having to incur a deductible or by having to pay all costs beyond a certain threshold amount. Insurance companies also use experience rating (basing future premiums on the past losses of the insured, or in the case of auto insurance, past losses and driving record) to mitigate both moral hazard and excessive claiming practices. Moreover, insurance theory indicates that if fault is accurately determined, charging each driver a premium equal to the driver’s expected losses is the best way to preserve safe driving incentives (Shavell, 1986).

Adverse selection also reduces the ability to spread risks through insurance. Adverse selection exists when insureds have private information about their risk levels that insurers cannot observe, or when insurers are prohibited from using such information to distinguish between risk types. If insurance is offered at one standard premium to a heterogeneous group of individuals, it will be overpriced for those who are low risk and underpriced for those who are high risk. This will result in low‐risk individuals buying less insurance or even no insurance if possible, while high‐risk individuals will be more likely to buy insurance and will want to purchase more of it. The resulting insurance pool will be dominated by high‐ risk individuals. In practice insurance providers use both risk classification and experience rating to distinguish between risk types, and governments may require individuals to purchase insurance to overcome the tendency for low‐risk individuals to opt out.

In summary, automobile insurance is a useful tool for risk‐averse individuals to transfer their risk of loss associated with driving, as well as to assure that those injured by at‐fault drivers are compensated for their losses. Adverse selection and moral hazard create barriers to offering insurance because they reduce the ability to pool risk, increase the costs of claims, and raise the costs of managing the insurance

pool. The design of an automobile insurance system has important implications for the extent of moral hazard and adverse selection problems and, more generally, will affect the extent to which incentive misalignments reduce the benefits and raise the costs of insurance. Such problems are a part of the trade‐offs which must be considered in designing the automobile insurance system.

2.2 Objectives of an Automobile Insurance System

The discussion of automobile insurance highlights the benefits and costs of insurance, and raises a number of important issues that must be considered when designing an automobile insurance system.

The following set of specific objectives summarizes key attributes that are relevant to these issues and that account for the interests of various stakeholders (policyholders, insurers, regulators, taxpayers).

These objectives are widely held to be hallmarks of a well‐functioning insurance system, and may be used to determine the health of an insurance system:2

 Availability: Consumers have timely access to a variety of auto insurance products that meet their

needs. When automobile insurance is compulsory, insurance availability may determine access to

driving.

 Affordability: Consumers can acquire auto insurance at appropriate prices. Income differences across

consumers may affect perceptions of affordability.

 Service Quality: Consumers are satisfied with their products and services.

 Cost Efficiency: Products and services are provided at least possible cost.

2 See, for example, the objectives developed by the automobile insurance working group in the Province of Alberta in 2007; Cummins and Tennyson (2002); Tennyson (2012). 7

 Sustainability: Products, service quality, costs, and prices are at levels which are viable over the long‐

term.

 Fairness: Consumers benefit from fair treatment in the insurance system, both in the pricing of the

insurance product and in the claims settlement process.

 Informed Consumers: Consumers understand the automobile insurance product and make informed

decisions about purchasing and claiming.

If insurance is produced and sold by firms in the private market, the additional objective of competition must be met in order to assure that the operation of the market leads to a well‐functioning insurance system:

 Competition: A number of insurers actively compete to serve consumers’ automobile insurance

needs.

3 Design Considerations of an Automobile Insurance System

The ability to achieve the objectives listed above is a function of a number of different parameters that define an automobile insurance system. This section describes the primary characteristics of the auto insurance system that will impact the achievement of outcomes. Four design considerations are of particular importance:

 Compulsory versus optional insurance

 Private insurance versus government provider

 The role of liability in determining compensation

 The pricing mechanism

It is important to recognize that specific design parameters are better at achieving certain objectives and are less effective in achieving others. In the concluding section of this report we will discuss the trade‐offs that policy‐makers face in designing an effective automobile insurance system. For example, although consumers are inherently interested in affordable insurance, insurers must be able to cover their costs and earn a fair rate of return. Those who are injured desire a system with generous benefits, but this creates affordability issues. Without explicitly recognizing these trade‐offs automobile insurance reforms often address one problem but create another. Further, auto insurance does not operate in a vacuum: it interfaces with the private and public health systems, road safety and driver licensing mechanisms, and the legal system. The availability of public transportation, household wealth levels, societal acceptance of fraud, and the importance society puts on holding at‐fault drivers accountable also impact the auto insurance system. Any attempts at reform should recognize this interrelationship.

3.1 Mandatory Insurance Coverage

The economic rationale for compulsory auto insurance arises from the same rationale that gives rise to government intervention in insurance markets: part or most of the costs of accidents fall on other parties. Due to potentially inadequate resources, in the absence of liability insurance an at‐fault driver may not be held financially responsible for damage caused to others. In the presence of both universal health insurance and private medical benefits, injuries to one’s self or to others will be paid by taxpayers and/or the insurance company. Some costs may also fall upon employers for employees who have disability and health benefits from work.

Because (uninsured) drivers do not bear the full cost of their driving, they will drive if the marginal individual benefit to them of driving is greater than the marginal individual cost. Compulsory insurance, if priced appropriately, is a mechanism that forces people to internalize the full cost of their driving and thereby give up driving (and reduce associated costs) if the marginal benefit is not greater than the marginal cost. A second argument for compulsory insurance relates to improving safety incentives. Liability insurance that is priced in relation to driving risk can encourage greater safety because drivers know that by driving more safely their premiums will go down, or by having an accident their premiums will go up.

The main argument against compulsory insurance is the regressive impact on the distribution of income. Compulsory insurance laws are more likely to affect low‐income individuals since they are the ones most likely to go without coverage. Without insurance, their costs would have fallen on (primarily) the publicly funded health insurance system. This income shift from low‐income drivers to high‐income drivers is a common criticism of compulsory insurance laws. Mechanisms for reducing the regressive impacts include rate regulation or some degree of social pricing that subsidizes the cost of coverage.

A second consideration with compulsory insurance is the ability to enforce compliance. Greater numbers of uninsured drivers increase the cost of insurance for insured drivers, potentially increasing the problem of affordability; this in turn may result in more uninsured drivers (Smith and Wright, 1992).

Evidence on the extent of this problem is mixed, since some estimates show no relationship between uninsured driving rates and the cost of automobile insurance (Ma and Schmit, 2000; Jaffee and Russell,

1998). However, Smith and Wright (1992) find a positive relationship between uninsured driving and average premiums using data from 27 large U.S. cities. Using state‐level data from the U.S., Tennyson and

Weiss (2011) also find a positive relationship, with estimates suggesting that a 10 percent increase in uninsured driving leads to a 2 percent increase in automobile insurance premiums. These authors also find some evidence of a feedback effect whereby higher premiums lead to higher rates of uninsured driving.

In practice, mandatory insurance coverage typically includes liability coverage, personal injury protection or first‐party accident benefits (more common in Canada than the United States), and uninsured motorist coverage.3 Legal requirements generally specify an amount of insurance which must be purchased, with coverage amounts above this limit available but optional. The argument for mandating third‐party liability coverage is the potential for parties injured by negligent drivers to be uncompensated and forced to bear (at least a portion of) their own injury costs. In the presence of universal healthcare, accident costs that are covered would be shifted on to taxpayers, but would still be borne by others than the negligent party. In other words, negligent drivers can shift the cost of accidents to others.

3 A few jurisdictions require coverage for first‐party property damage to one’s own automobile. 11

Although mandatory liability insurance helps to ensure that injured parties are compensated by negligent drivers, the time and expense associated with settling claims and the uncertainty associated with the value of the settlement is a disadvantage to relying on this means alone for compensation, especially for injured persons. This is an argument for some level of mandatory first‐party accident benefits that are paid without reference to fault (no‐fault insurance systems). In Canada, first‐party accident benefits are compulsory in every province except Newfoundland and Labrador (and in

Newfoundland and Labrador over 70 percent of insureds purchase first‐party accident benefits). Such benefits are designed to achieve quicker and more accurate compensation, lower settlement costs (by eliminating many liability claims), and potentially quicker recovery of health (evidence shows that access to appropriate treatment soon after injury leads to better results).

In an attempt to reduce liability insurance transactions costs, some jurisdictions mandate compulsory first‐party property damage coverage for not‐at‐fault damage. As noted by Kelly, Isotupa and

Kleffner (2009), key benefits of first‐party as opposed to third‐party recovery for vehicular property damage are the elimination of the adversarial relationship that often results from two insurers having to negotiate the fair amount to be paid for a claim involving their own insureds, the elimination of the majority of subrogation costs for property damage claims, and a better matching of premiums to the actual potential loss.

Despite the fact that some level of auto insurance is mandatory throughout most of North

America, a substantial percentage of motorists drive uninsured. In the United States, the Insurance

Research Council (IRC) estimated the percentage of uninsured motorists in 2009 to be 13.8 percent overall, and found that the uninsured motorist problem varied greatly from state to state, from 4.5 percent in Massachusetts to 28 percent in Mississippi (Insurance Information Institute, 2012). Estimates for Canada are less widely available, but using methods similar to the IRC suggests uninsured driving rates

of 5 percent to 15 percent across the provinces. Because of this, all jurisdictions in Canada and most in the United States mandate that drivers buy uninsured motorist coverage to protect against injury caused by uninsured at‐fault motorists.

3.1.1 Residual Markets

The problem of uninsured drivers raises another important issue related to compulsory insurance: If auto insurance is mandatory for driving, but private insurers are not bound to offer insurance to every driver, then what is to become of those drivers who cannot obtain insurance in the market? Most jurisdictions, and all jurisdictions which mandate insurance purchase, provide some alternative market or mechanism through which such drivers – generally high‐risk drivers – may obtain insurance. This is often termed the “involuntary” or “residual” market. Residual markets, which usually provide insurance at subsidized prices, deal with the problem of drivers who are unable to find an insurer willing to provide them with insurance. Associated losses from the residual market pool are usually funded by loss‐sharing among private insurers or through general tax revenues.

The need for residual markets in automobile insurance is generally low, except when restrictions on market pricing lead to shortages of insurance supply in the market. In a market in which regulation restricts pricing, insurers will tend to reject drivers for which there are large differences between expected loss costs and the allowable premium charges. These rejected risks must then receive insurance through the residual market. In this way, regulatory restrictions on pricing raise the percentage of drivers insured through the residual market (Ippolito, 1979; Grabowski, Viscusi, and Evans, 1989; Harrington,

2002). The size of the residual market is generally considered to be an important measure of how well private auto insurance markets are functioning: “Healthy auto insurance markets, where coverage is widely available and affordable, are characterized by keen competition among companies, widespread

choice for consumers, and small (i.e. less than 2% of written vehicles) residual markets” (Facility

Association, 2003; p. 1).

In the public insurance markets in Canada, government‐run insurers provide coverage to all drivers, regardless of their risk. In the private market provinces of Alberta, Ontario, New Brunswick, Nova

Scotia, Prince Edward Island, Newfoundland and Labrador, and the three territories, insurers have created an unincorporated non‐profit organization, the Facility Association, which serves as a mechanism to provide insurance to those who cannot obtain insurance through the voluntary market. The operations of the Facility Association differ between each jurisdiction. The Facility Association operates a residual market (FARM) in each private market province, and it administers risk sharing pools in Ontario, Alberta,

New Brunswick, and Nova Scotia.

An approach which may be used as a support to the residual market is to mandate that insurers must accept all applicants. This take‐all‐comers requirement has been used in several U.S. states, and has been in force in the province of Ontario since 1993 and in Alberta since 2004. Under this requirement, insurers are legally obligated to offer coverage to those requesting it regardless of age, gender, marital status or any other characteristic. Alberta’s take‐all‐comers rule is fairly stringent in that the rules for the risk to be written within the FARM (Facility Association Residual Market) are mandated by the province and are not based on each insurer’s underwriting requirements. Under Ontario’s take‐all‐comers regulations, insurers must cover all potential insureds who meet their underwriting requirements, but firms are allowed to use rating variables such as territory, vehicle types, driving records and driver classes, accident history, and past convictions to define those risks that they will underwrite.

If residual market subsidies are financed through higher premiums for drivers in the voluntary market, this results in shifting of costs from high‐risk to low‐risk drivers. Regulations that subsidize high‐ risk drivers will dampen safety incentives for high‐risk drivers. If residual market deficits are passed on to

drivers in the voluntary market through higher premiums, the link between insurance premiums and driving behaviour is further weakened.4 This will distort driver safety incentives in ways that may lead to reduced precautions for all drivers and thus higher claim frequency and higher loss costs (Harrington and

Doerpinghaus, 1993; Weiss, Tennyson and Regan, 2010).

3.2 Public versus Private Insurance

As indicated above, the mandatory nature of automobile insurance stems from the reliance on automobiles, the social importance of automobile compensation, and the risks that driving imposes on others. Thus, an important question is what level of government involvement is desirable in the mandatory automobile insurance market. The government may act as the supplier of insurance, through public insurance or a public‐private partnership; or when insurance is privately supplied the role of government may include defining the mandatory product (such as mandatory limits and benefits provided), and regulating rates, underwriting practices, and claims resolution.

Arguments in favour of public insurance include potential economies of scale and lower transaction costs. This leads proponents of government‐run automobile insurance to argue that a government‐run monopoly will lead to lower prices and increased insurance benefits, relative to premiums paid.

Fronsko (2011) notes that the goal of private insurers is maximization of firm value whereas public insurers likely have objectives that are aligned with legislative objectives. Typical objectives for a government auto insurer, for example, include road safety programs and road redesign aimed at reducing car crashes, and affordability and accessibility of auto insurance – sometimes at the expense of actuarially

4 Harrington (1990) shows that there are subsidies from the voluntary to the residual market in a number of states, while Jaffee and Russell (1998) show that residual market prices in California during the 1980s were lower than those in the voluntary market for a large number of insured drivers, indicating cross‐subsidization. 15

fair premium setting. Another key objective is long term viability, although this is a difficult concept to measure if there is no transparent accounting separation between the government insurer and the government itself.

Useful examples of monopoly public insurers are the provincial workers’ compensation (WC) boards in Canada and the four U.S. states that have monopolistic state WC funds. There are also 21 U.S. states that have quasi‐public workers’ compensation funds where private insurers compete with the state provider. A 2009 report found that in the U.S. the 25 public and quasi‐public plans perform better financially than the private market in a number of performance categories, and at least as well when it comes to the bottom line. The state funds were also shown to be more effective at preventing losses and improving safety. Further, although the public programs tend to have higher losses than the WC insurance industry as a whole, they offset those losses with lower expenses, higher investment returns, bigger dividends to employers, and better injury prevention efforts (Jablonowski, 2009).

According to Jablonowski (2009) the key to state fund success stems from a dedicated approach to loss prevention and control.5 Given that they typically work with other state agencies and share responsibility for health and safety, they are able to incorporate state‐of‐the‐art loss prevention initiatives with financial rewards tied to the insured’s loss performance. Although public WC funds have higher loss ratios, they also have lower expenses. This is due in part to the fact that they do not pay taxes or licensing fees, but also results from efficiencies achieved through use of technology and coordination of loss prevention, safety and health efforts with other state agencies.

5 In Canada the potential benefits of a public auto insurer and the levers that the government is able to pull in order to reduce automobile accident costs is illustrated by Quebec’s coroner calling on Transport Canada to require manufacturers to equip vehicles with a system for detecting drowsiness and loss of driver alertness. The coroner has also recommended that Quebec’s auto insurer, the Société de l’assurance automobile du Québec (SAAQ), undertake a public education campaign about driving while tired. The coroner recommends that the campaign should caution drivers on the use of cruise control, noting that this might be lulling drivers into “a false sense of security.” 16

In the context of automobile insurance, the monopoly insurer in Saskatchewan, Saskatchewan

Government Insurance (SGI), provides another example of how costs can be controlled. SGI has been able to control costs both by being strict on what they will pay but also being very proactive on treatment and getting people better. The Early Intervention program requires an injured person to be committed to a specially designed treatment program in order to achieve the best recovery. Because it is a Crown insurer initiative, SGI is able to ensure consistent messaging and reap the benefits from a single, focused approach. Given that disability income benefits are generous (90 percent of net income), an important goal is to ensure that people are back to work as soon as possible. The incentives of the insurer and the injured party are aligned because the injured party only continues to receive benefits if she or he is fully committed to the treatment plan.

Another potential advantage of a public insurer is the ability to negotiate regulated prices (or set maximum prices) for treatment from healthcare practitioners. The provincial WC boards in Canada are able to effectively control costs associated with treatment by using a set fee schedule. The absence of such a schedule in private insurance contributes significantly to higher costs. For example, in Michigan higher costs as compared with other states are attributed in large part to charges for medical services.

The Insurance Institute of Michigan (2010) reports that reimbursement costs for no‐fault claims are roughly three times the reimbursement costs for workers’ compensation and four times the reimbursement costs for Medicare for the same procedure. Similarly, in Canada private automobile insurers in some instances pay two to three times the amounts that a provincial WC Board pays for the same treatment.

Differences in prices charged to private auto insurers as compared with public insurers may be due to cost‐shifting as well as to differences in cost control mechanisms. Cost‐shifting occurs when healthcare service providers increase charges to private payers as a means of recouping revenue when

public reimbursements are reduced. A report by the Insurance Research Council (IRC, 2010) documents healthcare cost‐shifting from U.S. hospitals to auto injury claims, as a result of low reimbursement rates from government health insurance programs (Medicare and Medicaid) and the costs of treating uninsured patients. The Insurance Research Council estimates that U.S. auto insurers pay additional hospital charges of more than $1.2 billion per year as a result of this practice.

There are very few academic studies on the specific costs and benefits of private versus public auto insurance, perhaps because of the lack of transparency of government entities. Kennedy and Mehr

(1977) compare the performance of the public insurer in Manitoba with Alberta’s private insurance marketplace. They find that private insurers clearly outperform the public insurer in terms of product and service, but that the government insurer can offer insurance at a much lower cost. This price advantage however is artificial: they note that premiums charged in Manitoba were inadequate and that deficits were offset by tax dollars.

Quebec’s public auto insurance fund has also run a persistent deficit. Data from the Société de l’assurance automobile du Québec (SAAQ) show that that annual revenues were less than expenditures in every year between 1982 and 2004. By 2004 SAAQ reported a $450 million premium shortfall and a $617 million shortfall in assets, attributed to rising compensation costs and the failure to index insurance rates to inflation (SAAQ, 2006). Despite legislative changes and changes to the rate structure which were designed to remedy this shortfall, SAAQ reported an overall deficit of $1.6 billion in 2010 (SAAQ, 2011).

Although the recent deficit may be attributed to the 2008 financial crisis, the deficits demonstrate the ability of government insurers to maintain operations despite persistent shortfalls; and provide further evidence that premiums charged by government‐run insurance programs may not reflect underlying claims costs.

Opponents to public auto insurance raise the following issues that arise when there is a monopoly provider: potentially poor customer service, lack of competition, and lack of consumer choice.

Government monopolies also have little incentive to operate efficiently, and may lead to lower rates of innovation. These static and dynamic inefficiencies arise due to the lack of a profit incentive and protection from competition. A related argument is that lack of transparency and centralized decision‐ making will lead to politicized decisions that redistribute benefits for political reasons (e.g., Besley,

Gouveia and Dreze, 1994).6 Such concerns have recently been raised regarding the Insurance Corporation of British Columbia (ICBC), British Columbia’s public automobile insurer. In the 2012 rate review (BC

Utilities Commission, 2012), several observers charged that ICBC’s management compensation and the number of highly‐compensated employees are excessive, and that information disclosure regarding

ICBC’s operating expenses is inadequate.

3.2.1 Public Intervention in Private Insurance Markets

Due to the differences in public and private objectives – in public insurance affordability and accessibility are typically given higher priority than service quality, consumer choice and fairness – private auto insurance markets are often regulated to achieve social objectives. In order to achieve the objectives established, the extent of government regulation must be determined. The regulator attempts to balance the need for a financially stable and competitive auto insurance sector with the need to ensure that consumers pay affordable premiums, receive fair benefits, high service quality and timely compensation.

The economics of policy analysis deem government intervention to be desirable only when (i) actual or potential failures of market competition exist; (ii) these failures could or do lead to meaningful

6 A large literature in economics examines inefficiencies in government‐run enterprises, and the results of these studies form part of the basis for the privatization movement in developed and developing economies in the latter part of the 20th century (e.g., Stiglitz, 1998; Schleifer, 1998). 19

economic inefficiencies; and (iii) government intervention can ameliorate the inefficiency. Failures of competition are most likely to arise when one or a few market players control a large share of the market, externalities arise from market contracting, free‐rider concerns exist, or there is imperfect information.

Failures of market competition are often evidenced by markets with only a few large sellers, high seller profits, lack of price variation, or consumer lock‐in.

In privately‐run automobile insurance systems sellers of diverse size and characteristics7 operate in most markets; automobile insurance profits tend to be modest in comparison to other industries; and price competition is readily apparent. Thus, there is no clear economic rationale for regulating prices in automobile insurance markets as long as the general legal system prevents collusion among firms in the economy.

Skipper (1998) notes that imperfect information may be the most significant impediment to efficient contracting in insurance markets. However, Baltensperger et al. (2008) argue that because insurers recognize these information asymmetries as a cost of doing business, they have developed mechanisms to control adverse selection and moral hazard using both contract design and pricing mechanisms. Therefore regulation of this aspect of the industry is not essential.

Nonetheless, information asymmetries suggest the need for other forms of insurance regulation.

Insurance is the sale of an intangible promise: to pay financial compensation for losses experienced as a result of some future event occurring. The quality of that promise will necessarily be affected by insurer decisions that occur after the contract is signed, including investment decisions, capital holdings, and decisions regarding claims payments. These can’t be perfectly observed and credibly committed to at the

7 In 2009, there were 124 private insurers selling auto insurance in Canada. Ninety‐five companies belonged to a larger group, 25 insurers were incorporated outside of Canada, and 30 were mutual insurers. The largest private insurance company sold $1.75 billion in auto insurance in 2009, whereas the average premium revenue per company was $130 million (Source: Author calculation from A.M. Best data). 20

time of insurance purchase. The difficulty inherent in monitoring the strength of the insurer provides insight as to why the solvency of insurers is monitored by regulators.

Information asymmetries may also give rise to the potential for insurer misrepresentation of products or product quality. The complexity of contractual terms and pricing may make it difficult for consumers to become fully informed about price and quality of insurance products. Regulatory oversight may be needed to assure that consumers are treated fairly by insurers and that beneficial exchanges occur. The emerging field of psychology and economics8 provides additional reasons for market conduct regulation by providing evidence that consumers make biased and error‐prone decisions. For example, it has been suggested that auto insurance must be mandatory because drivers often underestimate both the frequency and severity of their potential losses. However, mandatory purchase creates additional concerns that insurer market conduct must be regulated. Thus, government oversight may be needed to assure that consumers are treated fairly, and most governments regulate market conduct in insurance

(Tennyson, 2010).

3.3 Tort versus No‐fault Compensation Systems

A fundamental characteristic of automobile insurance systems that varies across jurisdictions is the scope of liability for automobile accident damages. Compensation is either based on tort or is no‐ fault. In a tort system an at‐fault driver is held liable for injuries and damages caused to third parties.

Under a pure no‐fault approach, there is no tort liability for automobile accidents and compensation is made through first‐party personal injury benefits. Many no‐fault systems are modified or partial no‐fault, whereby tort liability is limited but not eliminated. In these jurisdictions persons injured in an automobile accident have the right to sue at‐fault parties only if their injury or loss meets a defined threshold.

8 See Camerer et al., (2003); Thaler and Sunstein, (2008); and Barr et al., (2008a), (2008b). 21

One of the advantages of a tort system is deterrence of negligent behaviour. However, a primary disadvantage is that it can be relatively slow and inefficient in providing compensation due to the necessity of establishing negligence. No‐fault automobile insurance was developed as an alternative to the tort system and was intended to improve the efficiency of compensation to injured parties.

Specifically, the advantages of no‐fault compared to tort are faster settlement of claims, lower claims settlement costs, and better matching of compensation to economic loss since benefits are available to both not‐at‐fault and at‐fault parties. However, two disadvantages of no‐fault are often cited. First, moral hazard is increased. Greater first‐party benefits for at‐fault drivers and the elimination of tort removes the personal liability associated with being at fault and may reduce the incentive to take care. Second, restrictions on the right to sue are viewed as inherently incompatible with basic common law principles.

No‐fault does not imply that fault is not assigned in an automobile accident, but merely that injured parties are compensated for injuries regardless of fault, typically by their own insurer. By focusing on first‐party compensation which does not rely on establishing fault, loss adjustment expenses can be reduced and injured parties can be compensated more quickly. In addition, there is better matching of compensation to economic loss because benefits are available to both not‐at‐fault and at‐fault parties, but compensation for non‐economic loss is reduced. Furthermore, no‐fault insurance provides compensation in more situations, for example, single vehicle accidents. In exchange for these guaranteed benefits, no‐fault places some or total restriction on the ability to sue others for damages and requires purchase of first‐party insurance coverage for the insured’s own bodily injury losses.

As indicated above, no‐fault systems for automobile compensation can be either pure or partial.

In a pure no‐fault system persons injured in an automobile accident may not sue an at‐fault party under any circumstances. Injured parties receive compensation for economic losses according to a schedule of benefits. Most jurisdictions in North America which have no‐fault insurance have a partial no‐fault system

whereby a person may sue an at‐fault party only if their injury meets a defined threshold of severity. This threshold can be verbal and/or monetary. For example, in Ontario an injured person may sue for pain and suffering only if the person “has died; or has sustained permanent serious disfigurement or permanent serious impairment of an important physical, mental or psychological function” (Insurance Act, 1990).

Because it is defined in terms of a description of the circumstances which permit a lawsuit, this is known as a verbal threshold. Monetary tort thresholds permit lawsuits when an injured person’s medical expenses exceed a certain dollar amount. Monetary thresholds are generally less desirable because they can become ineffective over time due to medical cost inflation, and can create incentives to overstate medical expenses in order to meet the threshold.

Some jurisdictions (Saskatchewan in Canada, and Kentucky, New Jersey, and Pennsylvania in the

United States) have a choice system in which insureds can choose to be covered under either a no‐fault or a tort system. In Saskatchewan, the default is the no‐fault system, and only about 5 percent of drivers choose the tort system. There is a limited right to sue for pain and suffering in Saskatchewan’s modified no‐fault regime,9 but a person may sue for economic losses that exceed no‐fault benefits. However, the no‐fault benefits are generous. In 2012, the cap on medical and rehabilitation coverage is over $6 million and income replacement is up to $82,804 a year with no time limit beyond normal retirement age. As such, this modified no‐fault scheme functions more like a pure no‐fault scheme.

As noted above, a potential disadvantage of no‐fault insurance is greater moral hazard because of increased first‐party benefits for at‐fault drivers. In addition, some argue that the elimination of tort

9 The right to sue is limited to situations in which the at‐fault driver is convicted of impaired driving or is convicted of deliberately using his or her vehicle to cause harm. 23

removes the personal liability associated with being at fault10 and may reduce the incentive to take care.

Fronsko (2011) argues that this moral hazard is mitigated in practice through road safety legislation and enforcement which penalizes those that drive unsafely. He also notes self‐preservation and societal attitudes guard against a decrease in road safety caused by the presence of first‐party compensation for automobile accidents. To the extent that this is not the case, no‐fault insurance may lead to higher accident rates, which will increase insurance costs.

The empirical evidence on this question is mixed. Kochanowski and Young (1985) find that the adoption of no‐fault auto insurance in the U.S. does not impact fatality rates. Similarly, using the frequency of property damage liability claims (the number of claims divided by insured exposures) as a proxy for accident rates, Loughran (2001) shows that overall accident rates are no higher in no‐fault states than in tort states over the period 1976 to 1998.

In contrast, Devlin (1992) finds that fatal accidents increased by 9.62 percent after the introduction of no‐fault in Quebec. Cummins, Phillips and Weiss (2001) find that fatality rates were 2.7 to

17.3 percent higher in states with no‐fault insurance for the period 1968 to 1994. Moreover, they find that among no‐fault states fatality rates were positively associated with the stringency of the tort threshold: states in which it is easier to sue negligent drivers had lower rates of fatal auto accidents. They also find that greater use of experience rating mechanisms mitigates some of the moral hazard that arises from no‐fault insurance. In an Australian study, McEwin (1989) finds that abolishing tort liability increased fatality rates by 16 percent, suggesting that no‐fault is associated with less road safety.

10 However, the presence of mandatory automobile insurance with relatively low deductibles and relatively high limits of liability carried by most drivers in Canada in effect removes personal liability of at‐fault drivers. 24

Because a higher fraction of injuries receive compensation under no‐fault insurance, insurance costs may be higher than under tort even if there is no increase in accidents. However, per‐injury compensation should be lower and transactions costs of compensation should also be lower. For losses below the threshold, greater reliance on first‐party benefits should result in a reduction in transaction costs because fewer liability claims need to be resolved (given the caveat that legal‐based transaction costs are not replaced with other administrative costs in assessing the extent of injury). Cost savings also arise due to reduction in the amounts paid out for non‐economic losses. The trade‐off between liability costs saved and accident benefit costs incurred will depend on the level of accident benefits mandated and the threshold set for tort claims. Thus, the predicted net effect of no‐fault on insurance costs is ambiguous, and will depend on specific features of the system and on responsiveness of behaviours to incentive changes.

The stringency or level of the threshold in a no‐fault system will be a primary determinant of whether no‐fault is able to reduce costs and improve compensation. The impact of no‐fault depends on both the threshold that determines the right to sue and the level of first‐party benefits available to those who are injured. Threshold no‐fault compensation systems increase the marginal benefits of building up bodily injury claims in order to exceed the threshold for general damages eligibility (Weisberg and Derrig,

1991, 1992; Cummins and Tennyson, 1992, 1996). Moreover, because no‐fault claims do not face the scrutiny of the legal system, and because no‐fault states have placed a priority on paying benefits quickly, exaggeration and fraud have become a problem in many no‐fault states. Cole, et al. (forthcoming) note that fraud is higher in U.S. states with (partial) no‐fault insurance. States have responded by decreasing the number of days that accident victims have to report accidents and submit medical bills; many states have also enacted stiff penalties for fraudulent activities.

Further evidence on the importance of the threshold is provided by Browne and Puelz (1996).

They find that low monetary, but not verbal, thresholds lead to higher bodily injury liability claim costs.

More recently Cole, et al. (forthcoming) show that in general no‐fault laws are associated with both lower liability losses and total losses compared with tort states, and “add‐on” measures (mandatory first‐party benefits) also are related to lower loss levels. However, the level of losses is influenced greatly by the specific provisions in the no‐fault statutes. In particular, states with a hard‐to‐pierce threshold have lower total losses and lower liability losses while states that replace a higher percentage of lost wages have higher total losses. They also find that states with choice systems (similar to Saskatchewan) have higher losses and liability losses. As demonstrated by other studies, it is not the presence of no‐fault laws alone that dictates whether costs are higher or lower, but the specific provisions of the no‐fault laws in place.

Several other studies of U.S. no‐fault systems conclude that no‐fault results in an increase in the cost of auto insurance (e.g., O’Connell and Joost, 1986; Derrig, Weisberg and Chen, 1994; Cummins, Phillips and

Weiss, 2001).

In the U.S., 12 states continue to use the no‐fault approach to automobile accident compensation, but many constituents there believe that the promise of no‐fault has never borne fruit.

Indeed, the Rand Institute for Civil Justice in 2010 published a retrospective study of no‐fault insurance with the objective to “help explain why no‐fault was adopted, flourished, and then lost some of its political luster as a policy option” (Anderson, Heaton and Carroll, 2010; p 3). It concluded that the main shortcoming of no‐fault insurance was that it has not lowered automobile insurance costs in the way that was first envisioned. During the time period studied, the 12 no‐fault states in the U.S. (Michigan, Florida,

Hawaii, Kansas, Kentucky, Massachusetts, Minnesota, New Jersey, New York, North Dakota, Pennsylvania, and Utah) were consistently among the highest‐cost states for automobile insurance. In addition, four states that enacted no‐fault in the 1970s later repealed it due to high costs.

A study by Kelly, Kleffner and Tomlinson (2010) examines loss costs per vehicle for provinces in

Canada. They find that government‐run pure no‐fault insurance has lower average costs per insured vehicle than either private modified no‐fault or private tort insurance, suggesting that the increase in costs due to generous first‐party accident benefits are outweighed by the savings that stem from lower claim settlement costs and the elimination of non‐economic damages. An important caveat, however, is that differences in the statutory automobile insurance product across provinces mean that costs are not directly comparable.11

3.4 Pricing Mechanisms

The previous three considerations discussed – mandatory insurance, private versus public insurance, and the role of liability in compensation – together all influence insurance costs and premiums.

Moreover, these design characteristics also impact the importance of and ability to achieve different objectives. For example, when insurance is mandatory, affordability and fairness are both deemed to be essential to a well‐functioning auto insurance system. Here we discuss insurance pricing and the importance of pricing mechanisms that are sustainable and provide fair, affordable insurance to policyholders.

Insurance prices are established prior to knowing what the true cost of the product is. Rates need to be adequate to ensure insurer solvency but not excessive due to affordability concerns. Artificially keeping rates low will affect both insurers’ ability to pay claims and insureds’ incentives to drive safely.

However if insurance is compulsory, affordability is a legitimate concern because access to an automobile has been shown to promote economic stability (Raphael and Rice, 2002). Therefore, effective regulation

11 British Columbia, which administers government‐run tort insurance, was not included in the analysis due to lack of data. However, auto insurance premiums in British Columbia are among the highest in Canada, and claim cost increases have necessitated an 11.2 percent rate increase for 2012 (BC Utilities Commission, 2012), suggesting that government‐run insurance does not necessarily lead to lower insurance costs. 27

must ensure rates are adequate in order to not put insurer solvency at risk and to not undermine competition in the market, while also providing affordability for policyholders.

Pricing adequacy in the aggregate is important for the long run sustainability of the insurance system. If automobile insurance premiums in total are not sufficient to cover the full economic costs of the insurance system, drivers receive a subsidy from others in the economy. Because the use of automobiles creates a variety of externalities to society, including accident risks but also pollution and use of scarce non‐renewable resources (Parry, Walls and Harrington, 2007), public policies should not be designed to provide a subsidy to this activity in the aggregate.

In private sector insurance systems, pricing inadequacy will distort insurance supply, reducing competition and insurance availability. If insurer returns are below a competitive rate of return, every dollar of capacity that an insurer devotes to the regulated market represents a lost opportunity to earn a higher return on that dollar in unregulated markets or in other ventures. This creates pressure on insurers to reduce insurance offerings or to exit the market, resulting in potential availability problems for consumers.

Insurance pricing can be achieved in different ways: the two extreme options are risk‐based insurance prices that reflect individual expected claim costs, and social pricing that reflects average claim costs. In a risk‐based pricing model premiums are based on the characteristics of the insured that are known to impact expected losses or expenses. Social pricing places less emphasis on individual rating factors and more emphasis on the risk profile of the community, typically through a territorial rating variable. In practice, most automobile insurance systems fall somewhere in the middle. For example, in the public insurance provinces prices reflect the risk profile of the insured through factors such as vehicle type, usage and the insured’s driving record but typically do not reflect individual characteristics that

cannot be controlled by the insured – such as age and gender – even if they have been shown to be actuarially relevant.12

Social pricing schemes, however, tend to lead to higher rates of accidents and claims and thus to higher average insurance costs. Evidence from Quebec after the 1978 introduction of no‐fault with prices based only on vehicle type confirm that accident and claims rates were higher than otherwise predicted

(Dionne, 2002). Case studies of U.S. state auto insurance markets with regulated pricing mechanisms that severely restrict premium differences across driver groups also see higher claims rates and accident costs

(Tennyson, Weiss and Regan, 2002; Grace, Klein and Phillips, 2002; Harrington, 1991, 2002; Derrig and

Tennyson, 2011).

Under a risk‐based pricing scheme consumers who have higher expected claim costs are charged higher insurance premiums to reflect the greater risk they impose on the insurance system. Risk‐based pricing that reflects true costs (differences in expected claim costs across policyholders) provides proper safety incentives for policyholders to take care. If insurance premiums reflect the expected marginal costs of coverage, consumers have appropriate information on which to base their decisions to purchase insurance and to drive (Harrington and Doerpinghaus, 1993). Thus, an important theoretical principle of insurance pricing is that the premiums charged should reflect the expected value of a driver’s insured losses. If insurance premiums cannot fully adjust to a driver’s choices regarding safe driving, for example because these are not perfectly observed or because they vary over time, insurance ownership may lower a driver’s incentives to take care. Economic theory also predicts that the safety investments of drivers

12 In Saskatchewan driving record is used but not a driver's age, gender, or place of residence. In Manitoba rates are based on where the insured lives, vehicle use, vehicle type, and driving record. In British Columbia premiums vary with location, rate class, and claims history. 29

may be diminished when premiums do not reflect past accident experience or driving record (Shavell,

1982, among others).

One practical difficulty in implementing risk‐based insurance pricing is the difficulty in measuring a driver’s expected losses. In practice, insurers must set premiums based on observable characteristics that they find to be correlated with loss experience. This results in grouping drivers into categories and charging the same base premium to all drivers within a category. Insurers also adjust premiums over time based on an individual’s observed loss and driving experience, since losses provide a signal of the expected value of future losses (Boyer and Dionne, 1989). These mechanisms – underwriting (sorting by observable characteristics) and experience rating (premium changes based on observed losses and driving record) – help to link auto insurance premiums to drivers’ risk, but are not perfect.

Providing incentives for safe driving is an important advantage of risk‐based pricing models.

However, collecting and analyzing data for the purpose of estimating individual expected claim costs is also costly; thus there is a limit to how much risk classification is beneficial. Finger (2006) notes that risk classification systems become more complex as competition among insurers increases. To the extent that more complex classification systems can better match premiums with expected losses, this will result in less adverse selection and moral hazard among insureds. However, it is also possible that competition leads to excessive risk classification from society’s perspective. This can occur if classification is costly to undertake (Crocker and Snow, 1986; Harrington and Doerpinghaus, 1993).

Another practical difficulty with risk‐based pricing of automobile insurance is that it can make insurance unaffordable for some drivers, particularly high‐risk and/or low‐income consumers. Although risk‐based pricing is fair in that it charges the highest‐risk policyholders the highest price, this can compromise the objective of achieving affordability for all drivers. However, unless driving is considered a right rather than a privilege, it is difficult to argue that the most unsafe drivers should be subsidized by

other drivers. Nonetheless, risk‐based pricing reduces access to insurance and to driving for certain populations, and in practice this may include some who are not high risk. This occurs because premiums are based on driver categories and imperfectly linked to individual risk, and because some determinants of risk (such as location) are not specific to the driver. For this reason risk‐based pricing often faces social objections.

Powers (2010) provides an argument for social pricing for mandatory insurance. He notes that if insurance is mandatory, risk classification is not essential to prevent market failure. In addition, for individuals (as opposed to corporations), risk classification likely does not provide the right financial incentives for policyholders to reduce their risk levels. As well, normative social justice arguments do not support the use of risk classification systems when the likelihood of a loss is small. He argues against the unfairness of individuals being compelled to accept premiums based upon classifications such as age, gender and territory, characteristics which cannot be changed through reasonable risk control efforts.

Due to potential affordability problems especially given the mandatory nature of the product, and fairness concerns as they relate to individual characteristics that are not controlled by the insured, many jurisdictions employ some degree of social pricing in order to address the perceived shortcomings of risk‐ based pricing. Thus, in practice auto insurance pricing uses a combination of both risk‐based and social pricing, with various types of regulation in place to address affordability, fairness and sustainability.

Restricting underwriting practices or rating variables is a common social pricing mechanism used in various provinces across Canada. In Alberta driving record, driver training, age, gender, marital status, territory, type and use of vehicle, and years licensed are all valid risk classification variables, but the premium grid sets a maximum premium for basic coverage for any driving record. In Newfoundland and

Labrador rating based on age, gender, and marital status was eliminated in 2005. In Nova Scotia insurers may not use age or marital status as of 2004, and it has been recommended that gender also be

eliminated. In contrast, Ontario has fewer risk classification restrictions and doesn’t prohibit gender, age, or marital status. However, several rating criteria may not be used to determine automobile insurance rates: credit history; bankruptcy; employment status; ownership of a credit card; how long the driver has lived in his or her current home; accidents occurring on or after September 1, 2010 for which the insured is less than 25 percent at‐fault; not‐at‐fault accidents; whether the vehicle is owned or leased; and whether there was a period of time in which the insured had no automobile insurance coverage.

The use of age and gender as risk classification variables to price auto insurance is a contentious issue. In private Canadian insurance marketplaces, a set of legal precedents has resulted in auto insurers being allowed to use age as a basis for pricing. The landmark case on this point is Zurich Insurance Co. v.

Ontario (Human Rights Commission (1992) 16 C.H.R.R. 3D/255 S.C.C.). In that ruling, the Supreme Court declared that the insurance industry could continue to use criteria such as age and marital status as a bona fide means of assessing risk, but that the industry could not do so indefinitely because of the discriminatory tendencies of the practice.

The correlated relationship between age and accident frequency is well known: younger and older drivers have higher crash risk. Nicoletta (2002) reports that in the year 2000, persons aged 16 to 24 accounted for 15 percent of Canada’s population, 13 percent of total driving licenses, and only 7 percent of the total kilometres driven. For elderly drivers, Dellinger, Kresnow, White and Sehgal (2004) find that fatal and non‐fatal risks to others increase with a driver’s age, particularly for drivers over 75. This relationship can be seen Figure 1 below provided by Kelly, Nielson, and Snoddon (2012). The U‐shape curve shows the greater crash risk at both the older and younger driving ages, revealing the disproportionate percentage of young and old drivers killed in traffic accidents in Canada from 1999 to

2009.

Figure 1 – Percent of Accidents Leading to Fatalities by Age

Relative Accident Frequency by Driver Age

2.500

2.000

1.500

1.000

0.500

% Driver Fatalities / % Drivers By Age / % Driver Fatalities Driver Age 2009 2008 2007 2006 2005 2004 2003 2001 2002 2000 1999

Source: Kelly, Nielson and Snoddon (2012)

Government restrictions on the use of certain category variables (such as age and gender), and/or restrictions on price differences across categories, have the unintended consequence of reducing links between insurance pricing and driver risk. This may alter incentives to drive and to purchase insurance, and may reduce incentives for road safety. The evidence from Canada supports the idea that these incentive effects are important. Pricing restrictions on individual rate differences lead to more high‐risk drivers on the road. Accident data in British Columbia show a higher proportion of young drivers who are the principal owners of vehicles; injury collision rates among young drivers are higher than in other provinces; and the rates of injury per million passenger kilometers are consistently higher than found in provinces where risk‐based pricing is used (Leadbetter and Kovacs, 2004). Higher rates of injury per million passenger kilometres in government‐run provinces than in private provinces result in potentially higher costs per insured vehicle.

To achieve affordability objectives, all Canadian provinces with private auto insurance markets use prior approval rate regulation. In Canada, prior approval rate guidelines typically contain provisions on how rates or risk classification differentials can be modified. For example, in Ontario, “insurers should cap differential changes at +/‐ 10% from the current differential in the direction of the coverage indication” (Financial Services Commission of Ontario, 2011; p 7). The dampening of differentials flattens the premium structure and creates cross‐subsidies across risk classification groups similar to social pricing models. Prior approval regulation also weakens the links between claims experience and insurance premiums, creating greater premium volatility (Leadbetter, Voll and Wieder, 2008) and as noted with social pricing, increases incentives for claiming.

Experience rating of premiums based on a driver’s history of accidents, claims, or moving traffic violations may help to counteract the negative incentive effects of social pricing schemes. All jurisdictions in Canada use some degree of experience rating that is based on a driver’s at‐fault claims history. Insurers may also levy premium surcharges based on an insured’s driving record. In attempting to balance the objectives of affordability and incentives for safe driving, Alberta established maximum benchmark premiums for any driving record in response to dissatisfaction from new drivers automatically being classified as high‐risk drivers and paying high premiums. In Alberta, drivers move up and down a rate grid based on whether their driving experience produced an at‐fault claim or a claim‐free year. After six claim‐ free years premiums reach the maximum discount of 65 percent. This mechanism maintains incentives for safe driving and achieves affordable premiums for new drivers. As long as the grid prices are sustainable such that the benchmark premium and the incremental amounts charged for at‐fault accidents are sufficient to cover the claim costs of high‐risk drivers, this pricing model is able to meet the stated objectives.

In Ontario the experience rating mechanism assigns drivers to a driving record class based on the number of years without an at‐fault claim. By law not‐at‐fault claims may not be used to modify premiums and therefore do not affect the driving record class assignment. A discount is earned for each additional year of driving that is free of at‐fault claims. Drivers with six or more consecutive years without an at‐fault claim are placed in class 6. Some insurers offer an additional discount for class 6 drivers with

10 or more consecutive years of driving that are free of at‐fault claims. Insureds with more than one minor driving conviction in the preceding three years are usually not eligible for discounts. This mechanism of assigning drivers to a class provides incentives for safer driving.

Government insurance systems may also incorporate accident experience or driver demerits in determining individual premiums; such pricing reforms were introduced in Quebec after 1992. In jurisdictions with private insurance markets, government may mandate that insurers employ a government‐determined bonus pricing scheme. This was the case in several European countries

(including France, Belgium, and Germany) prior to European integration and deregulation of insurance pricing in the single market (Dionne, 2002). Empirical studies demonstrate that such pricing schemes are effective in increasing road safety and reducing insurance claiming (Dionne, 2002), and that drivers’ past experience is predictive of future claiming behavior (Boyer and Dionne, 1987). Thus, the use of experience rating in premium determination should be an important part of social pricing schemes. More generally, pricing models should be designed to maintain the incentives for safe driving, treat different policyholders fairly based on their risk characteristics, and provide affordable insurance and a fair rate of return to insurers to ensure solvency and a competitive market.

4 Automobile Insurance Systems Design and Performance

An automobile insurance system combines all of the choice components discussed in the previous section. The functioning of the system will depend on both the benefits and drawbacks of each design element, and the effects of pairing the different elements together. The choice of different combinations of design features produces great potential for variety among end‐use automobile insurance systems. We first summarize the different system designs that exist in jurisdictions in Canada. This is followed by an examination of the performance of Canadian auto insurance systems.

4.1 Insurance Systems in Canada

Table 1 illustrates the spectrum of different automobile insurance system characteristics across

Canada’s provinces. Three provinces operate government‐run auto insurance for mandatory coverage

(British Columbia, Saskatchewan, and Manitoba). One province (Quebec) employs a combination of government insurance for auto bodily injury risks and private insurance for property damage risks. In the remaining provinces and territories auto insurance is sold by private companies, with these jurisdictions imposing varying levels of regulation of pricing, practices and products. However, even in the provinces with private insurance, the automobile insurance policy itself is a statutory document.

Table 1 – Canadian Automobile Insurance Plans

British Alberta Saskatchewan Manitoba Ontario Quebec New Nova Prince Newfound Columbia Brunswick Scotia Edward ‐land and Island Labrador Mandatory BI/PD, BI/PD, AB, BI/PD AB, UA, FPF BI/PD AB, BI/PD, AB, PD, DC‐PD, BI/PD, AB, BI/PD, AB, BI/PD, AB, BI/PD, UA Coverages1 AB, UM, UA UA, FPF UA, DC‐PD AB, UA UA UA UA UA, DC‐ PD Admin Public w/ Private Public w/ private Public w/ Private Public for BI; Private Private Private Private private for optional private for Private for for coverage optional PD optional coverage coverage Regime Tort Tort Choice – No‐fault Pure No‐ Modified Pure No‐ Tort Tort Tort Tort default fault No‐fault2 fault3 Compulsory $200,000 $200,000 $200,000 $200,000 $200,000 $50,0004 $200,000 $500,000 $200,000 $200,000 Min. Liability Right to Sue Yes Yes6 No (No‐fault); No Yes – verbal No Yes6 Yes6 Yes6 Yes8 for Non‐ Yes (Tort)5 threshold2,8 economic Loss Right to Sue Yes Yes Yes7 (No‐fault); No Yes7 No Yes Yes Yes Yes for Yes8 (Tort) Economic Loss Medical $150,000 $50,000 $6,250,817 (No‐ No limit $50,000/ No time or $50,000, $50,000, $25,000, $25,000, and Rehab fault); $3500 if amount limit time limit time limit time limit time limit Payments $24,440 for non‐ minor/ 4 years 4 years 4 years 4 years per Person catastrophic, and $1 million if $183,308 for catastrophic catastrophic (Tort) Funeral $2,500 $5,000 $9,376 (No‐fault) $7,753 $6,000 $4,142 $2,500 $2,500 $1,000 $1,000 Expenses $6,110 (Tort)

British Alberta Saskatchewan Manitoba Ontario Quebec New Nova Prince Newfound Columbia Brunswick Scotia Edward ‐land and Island Labrador Maximum $300/wk $400/wk 90% net income; $85,500/yr 70% of net 90% net $250/wk $250/wk $140/wk $140/wk Disability (max), (max), 80% max income gross (max), 90% wages to wages; max (max) (max), 80% (max) (max) Benefits 75% of of gross $82,804/yr (No‐ of net max income of gross gross wages fault); wages $400/wk gross wages wages $368/wk (Tort) $55,000/yr Time Period 2 years 2 years No limit (No‐ No limit 2 years, 3 years 2 years 2 years 2 years 2 years partial, total fault); longer if partial, partial, partial, partial, to partial, lifetime disability 2 years partial, unable to lifetime total lifetime lifetime age 65 if lifetime total lifetime total pursue any disability total total totally total disability (Tort) suitable disability disability disabled disability occupation Not‐at‐fault First‐ Third‐party Third‐party Third‐party First‐party First‐party First‐party Third‐party Third‐party Third‐party PD Recovery party since 2004

Table adapted from Kelly, Kleffner and Tomlinson (2010). Sources: Provincial auto insurance boards 1 BI/PD – Third‐party liability for bodily injury and property damage; PD – third‐party liability for property damage only; AB – accident benefits which is first‐party bodily injury coverage that applies regardless of fault; UA – uninsured auto protection generally provides BI coverage only to the extent that the insured is not at fault; UM – underinsured motorist; DC‐PD – direct compensation property damage which is first‐party compensation for auto damage to the extent that the insured is not at fault; FPF – first‐party all‐perils (similar to collision and comprehensive coverages). No‐fault provinces have a BI/PD component to cover accidents that happen either outside the province, or that involve motorists from outside the province. 2 The right to sue for pain and suffering only exists if the injured person dies or sustains “permanent and serious” disfigurement and/or impairment of important physical, mental or psychological function. A dollar threshold for monetary damage exists for losses in excess of statutory accident benefits (Insurance Act, 1990). 3 Lawsuits are not permitted with respect to injuries sustained in automobile accidents in Quebec. Victims are compensated by their government insurer for their injuries whether or not the accident occurs in Quebec 4 Applies to property damage claims in Quebec and personal injury and property damage claims outside Quebec. 5 Under the no‐fault option, insureds may sue for non‐economic losses in three specific incidents: the at‐fault driver was impaired, the at‐fault driver was convicted of using an automobile to deliberately injure the party, or there exists a third party such as a car manufacturer whose negligence was implicated in the accident. An insured may also sue for expenses not covered by the no‐fault benefits provided. A list of benefits, more detailed than the summary provided in this table, can be found at http://www.sgi.sk.ca/sgi_pub/vehicle_insurance/coverage_information/pdf/guide_nofault.pdf. 6 Cap exists on general damages for “minor injuries.” 7 Monetary loss must exceed first‐party benefits. 8 Award is subject to a deductible.

Manitoba’s public insurance system and Quebec’s mixed system determine compensation on a no‐fault basis; Saskatchewan’s public system allows choice between no‐fault and tort with no‐fault as the default option; and British Columbia’s public system is tort‐based. Five of the six private insurance provinces – Alberta, New Brunswick, Nova Scotia, Prince Edward Island, and Newfoundland and Labrador

– are tort‐based, with only Ontario using a (partial) no‐fault approach.13 Automobile liability insurance is compulsory in every province and first‐party accident benefits are compulsory in all provinces except

Newfoundland and Labrador.

Rate controls also occur in all provinces. The provinces with public systems (and Quebec) utilize government‐determined social pricing schemes which greatly limit price variation across drivers, while most of the private systems impose less prescriptive prior‐approval regulation of rates. The one exception is Alberta, which uses a grid system for primarily high‐risk drivers that is more akin to social pricing, although it applies to less than 10 percent of drivers. Affordability and accessibility concerns have arisen in all provinces with private insurance systems, with the exception of Prince Edward Island. In response, regulators in these provinces have imposed risk classification restrictions and have periodically mandated rate freezes.

This brief comparison illustrates that the key insurance system elements combine in various ways across the provinces, but it also highlights that Ontario’s system is unique within Canada. Ontario is the only province with a privately administered (partial) no‐fault system; all other private systems operate on a tort basis. Compulsory first‐party injury insurance and risk‐based pricing modified by government rate

13 The three territories have tort‐based compensation mechanisms and employ actuarial‐based pricing methodologies to the extent possible. We do not analyze the territories here because of their geographical location (the three territories lie above the 60th parallel of latitude and geographically consist of sub‐arctic and tundra regions) and small population (the population density of the territories is 0.039 persons per square kilometre). The Northwest Territories is the most populous territory with a population of 41,462. Almost half of the inhabitants (19,234 people) live in the capital city of Yellowknife. 39

regulation are common to all of the privately administered systems, however. Given the differences, it is instructive to compare the performance of auto insurance systems across the provinces.

4.2 Comparing System Performance

The performance of different auto insurance systems can be compared using the objectives of auto insurance systems outlined in an earlier section. Such a comparison requires defining and obtaining observable measures that represent each desirable outcome. This is difficult in some cases  for example, measuring whether consumers are informed  because measures are simply hard to construct or there is a lack of data. In other cases measures can be defined but are likely to reflect two or more outcomes of interest. For example, the extent of uninsured driving is a measure of insurance availability but is also affected by the affordability of insurance. Finally, in many instances performance measures are available for private insurance systems but not for public systems, either due to lack of public reporting (e.g., insurance costs for total bodily injury) or to differences in system design (e.g., measures of uninsured driving). Subject to these limitations, we propose evaluating the success of a province in meeting the objectives of an auto insurance system using the measures discussed in the following paragraphs.

Table 2 compares the performance of the provincial auto insurance regimes in terms of the measures described below. Data for provinces with public insurance systems are limited, but are presented where obtainable.

Table 2 – Performance of Provincial Auto Insurance Regimes

Administration Private Public Public Private Mix Regime Modified Tort No‐fault Tort Choice/ No‐fault No‐fault No‐fault Default Province ON AB NB NS PE NL QC1 BC SK MB Availability Uninsured Auto Claim Rate 15.3% Not 5.35% 4.69% 10.38% 6.55% Not Not Not Not (2009)2 reported reported reported reported reported FARM (residual market) Size Not Not Not Not 0.1% 0.3% 1.9% 0.8% 2.1% 4.1% (2011)3 applicable applicable applicable applicable Risk Sharing Pool (2011)4 Not Not Not Not Not Not 2.6% 6.6% 1.7% 1.9% applicable applicable applicable applicable applicable applicable % GIO Complaints Labelled as 4.64% 0 0 4.17% 0 0 5.56% 7.69% 0 0 “Availability” (2011)5 Affordability Average Premium (2010)6 $1,433 $1,087 $813 $807 $758 $986 $719 $1,118 $856 $949 % GIO Complaints Labelled as 2.63% 3.82% 7.14% 0 0 0 8.33% 0 0 0 “Affordability” (2011)5 Service Quality % GIO Complaints per 10,000 5 1.013 0.757 0.146 0.159 0.137 0.431 0.123 0.026 0.009 0.0016 Population (2011) Cost Efficiency – Private Insurers Market Share of Direct Sellers 7 20.14% 22.30% 15.28% 12.40% 18.20% 30.02%  0.04% 1.47% 0.16% (2009) Market Share of Commodity 8 16.36% 15.91% 10.77% 14.02% 15.60% 4.00%  1.11% 0.07% 0.05% Sellers (2009) Market Share of National 9 29.33% 40.37% 32.36% 44.15% 53.80% 33.01%  1.38% 2.75% 0.79% Insurers (2009) Sustainability Average Loss Ratio Not 10 84.10% 68.87% 58.15% 54.63% 49.85% 70.31%  82.11% 76.97% (2006–2010) reported 10‐Year Growth in Losses per 52.71% 4.07% ‐41.57% ‐32.46% ‐23.22% 12.73%  5.12% Not ‐22.66%

Administration Private Public Public Private Mix Regime Modified Tort No‐fault Tort Choice/ No‐fault No‐fault No‐fault Default Province ON AB NB NS PE NL QC1 BC SK MB Vehicle (2000–2010)11 reported 10‐Year Growth in Premium Not per Vehicle (2000–2010)12 65.85% 34.04% 9.09% 31.65% 25.99% 36.52% 14.13% 3.37% reported 37.60% 10‐Year Growth in Average Not Claim Severity (2000–2010)13 152.36% 15.41% ‐2.39% ‐7.13% 7.38% 11.37%  ‐8.80% reported ‐11.17% Injury Accidents per 100,000 Licensed Drivers (2009)14 678.3 701.7 700.2 1109.2 787.5 647.3 856.6 648.4 943.5 929.6 Fatal Accidents per 100,000 Licensed Drivers Driver (2009) 5.9 12.8 12.1 10.6 12.4 8.8 10.2 12.1 21.2 11.1 Fairness CV of Premiums Across 104.5 21.0 Territories (2009)15 CV of Loss Ratios Across 19.7 6.8 Territories (2009)16 CV for BI/PD Loss Ratios by 12.9 9.6 13.2 8.6 19.1 14.3 Driving Record Class (2009) Competition – Private Insurers Insurers with Positive 85 73 63 60 47 51  44 37 39 Writings (2009) Market Share of Largest Four 31.11% 44.26% 38.62% 39.05% 47.40% 56.27%  3.60% 3.28% 0.54% Insurers (2009) Herfindahl‐Hirschman Index 17 427 687 636 638 781 1055  1673 1654 859 (2009) Sources: General Insurance Statistical Agency (for private market provinces), Facility Association, General Insurance OmbudService, Manitoba Public Insurance, Saskatchewan General Insurance, Insurance Corporation of British Columbia, AM Best Ltd. and Transport Canada Entries for which data exists but are not publicly available are labelled “Not Reported”. If a particular data element is not relevant to a particular province, the entry is labelled “Not Applicable”. 1 In 2006, the Quebec government altered the mechanism for both reporting bodily injury losses and for pricing the expected losses in the auto insurance premium. Thus QC insurance figures are not comparable to other provinces and are labelled with a “”. We also do not report competition among private insurers in Quebec, as many insurers are provincially regulated and are not included in AM Best Ltd. 2 The uninsured auto claim rate is calculated as the number of uninsured auto claims divided by the number of bodily injury liability claims. In some provinces, property damage to

Administration Private Public Public Private Mix Regime Modified Tort No‐fault Tort Choice/ No‐fault No‐fault No‐fault Default Province ON AB NB NS PE NL QC1 BC SK MB vehicles can be claimed in specific situations only. 3 Rules for admitting drivers into the Facility Association residual market differ by province. Alberta and Ontario have very stringent rules, and in contrast, there are few rules in Prince Edward Island. Thus results are not directly comparable across provinces. 4 The risk sharing pools in Ontario and Alberta are for “grey risks” and the risk sharing pools in New Brunswick and Nova Scotia are for new drivers. 5 The General Insurance OmbudService is a dispute resolution service that is available to all Canadians who have concerns with their auto, home, or business insurance. They classify complaints received in over 20 different categories, relating to affordability, availability, billing problems, claims issues, policy concerns, and cancellations. We report just the percentage of complaints listed as “availability” concerns, acknowledging that likely understates concerns as policy cancellation, non‐renewal complaints, and inability to purchase a coverage or endorsement may also be a sign of availability issues. Issues of affordability are similarly understated using the percentage of complaints listed as “affordability” concerns, because issues with policy rating, policy renewal, and premium increases may also be indicative of affordability concerns. 6 The average premium is calculated as total earned premiums for both mandatory and optional coverages divided by number of earned vehicles in the province. British Columbia does not separate out private passenger vehicles from other vehicles in their publicly available data. 7 An insurer is classified as a direct seller if it sells insurance via its own sales force or exclusive agents, face‐to‐face to consumers. 8 An insurer is classified as a commodity seller if it sells insurance over the Internet, by telephone, or by direct mail. 9 An insurer is classified as a national insurer if it has positive direct written premiums in auto insurance in all 10 provinces. 10 We first calculate the annual loss ratio (total incurred losses and adjustment expenses divided by earned premiums) for each of the five years and then calculate the average the annual loss ratios. 11 The 2000 and 2010 loss costs are calculated as total incurred losses and adjustment expenses divided by number of earned vehicles for each of the two years. The growth rate is calculated as (2010 loss cost – 2000 loss cost)/(2000 loss cost). 12 The 2000 and 2010 premiums are calculated as total earned premiums divided by number of earned vehicles for each of the two years. The growth rate is calculated as (2010 premium – 2000 premium)/(2000 premium). 13 The 2000 and 2010 claims costs are calculated as total incurred losses and adjustment expenses divided by number of reported claims for each of the two years. The growth rate is calculated as (2010 claims cost – 2000 claims cost)/(2000 claims cost). 14. From Transport Canada, “fatal collisions” include all reported motor vehicle crashes that resulted in at least one death, where death occurred within 30 days of the collision, except in Quebec (eight days) and “personal injury collisions” include all reported motor vehicle crashes which resulted in at least one injury but not death within 30 days of the collision, except in Quebec (eight days). 15 The CV describes the dispersion of a variable in a manner that does not depend on the variable’s mean, with a higher CV indicating higher dispersion. A CV of zero would imply that there is no variation in loss ratios across territories. CV is the coefficient of variation which is calculated as 100 times the standard deviation of a variable divided by the mean of the variable. A small CV for premiums would imply that premiums are socially fair – there is little variation by location across the provinces. 16 We first calculated the loss ratio (total incurred losses and adjustment expenses divided by earned premiums) for all coverages for the statistical reporting territories in Ontario and Alberta. We then calculated the weighted mean loss ratio and weighted standard deviation of loss ratios for each province. Territorial weights are the number of earned vehicles in each territory divided by the total number of earned vehicles in each province. The CV for TPL loss ratios by driving class uses third‐party liability incurred losses and earned premiums for the seven driving record classes (class 0 through to class 6) in each province. A small CV implies that premiums are actuarially fair. 17 The Herfindahl‐Hirschman Index for each province is 10,000 times the sum of the squares of direct written auto insurance premiums for each firm operating within the province divided by the square of total direct written auto insurance premiums for the province. A higher number implies greater concentration and less competition.

Availability: The rate of uninsured driving in a province is a good indicator of consumer participation in the automobile insurance market. Uninsured driving may arise because of problems of insurance availability, if drivers choose to drive uninsured because they have difficulty finding an insurance agent or company, or if they find the application process confusing or burdensome. Problems of insurance affordability may also be a cause of uninsured driving if drivers feel that they cannot afford insurance. Uninsured driving is typically measured in the literature by the number of insurance claims filed under uninsured auto coverage divided by the number of bodily injury liability claims. Some of the differences in uninsured auto claims rates may arise because of provincial differences in both the ability to access the tort system and in the level of first‐party accident benefit coverage available.

The percentage of cars that are insured through the residual automobile insurance market is another commonly used measure of the availability of insurance in private insurance markets. Markets in which insurers are reluctant to insure some drivers at prevailing or allowed premium levels will see larger percentages of drivers who must obtain insurance through this involuntary market mechanism. As above, differences in the eligibility criteria for placing a risk in the residual market between provinces will also generate differences between the provinces. We account for both the percentage of cars insured through the residual market and the percentage of cars insured through the provincial risk sharing pools. The risk sharing pools in both Ontario and Alberta which are open to “grey risks” were established in response to the introduction of take‐all‐comers rules in these provinces. The risk sharing pools in New Brunswick and

Nova Scotia are for new drivers only, and as such are not directly comparable to the risk sharing pool mechanisms in Ontario and Alberta.

We also record complaints made with the General Insurance OmbudService (GIO). The GIO’s mandate is to provide a cost‐free, independent, and impartial process to resolve complaints about general insurance products and its services are available in all provinces – those with private market and

those with government‐run insurers. The GIO classifies complaints into more than 20 categories as can be seen in Table 3.

In Table 3, we capture the percentage of complaints categorized as “availability” complaints. This measure understates true availability concerns, as we do not include other complaint categories such as policy non‐renewals or policy cancellations as part of this measure.

Using the uninsured auto claim rate, Ontario has the highest reported rate of uninsured driving among the private market provinces. However, because of the threshold to access the tort system, the claim frequency for bodily injury liability claims in Ontario is much lower than the other private market provinces. Thus the high uninsured auto claim rate in Ontario could also arise because of the lower bodily injury claim frequency. Insurers in all private market provinces make some use of the residual market mechanisms available in each province. Because of stringent regulations on the use of FARM, the combination of the FARM and risk sharing pool in Ontario (2.7 percent of total vehicles) is the best measure of the residual market in this province. Thus Ontario ranks fourth out of six provinces with respect to this variable. Nova Scotia (0.8 percent), New Brunswick (1.9 percent) and Prince Edward Island

(2.1 percent) have lower use of the residual market mechanism while Newfoundland and Labrador (4.1 percent) and Alberta (6.9 percent) have higher usage.

Table 3 – Summary of GIO Complaint Data 2011 for All Provinces

Number of Complaints Complaint Heading BC AB SK MB ON QC NB NS PE NL Affordability 12 38 9 1 Availability 1 67 6 1 Billing problem (broker related) 5 6 1 1 Billing problem (company related) 5 30 2 Claims, accident benefits, disputes 8 57 1 1 Claims, accident benefits, questions/entitlement 12 66 1 1 Claims, delay 1 31 55 9 1 1 3 Claims, dispute over liability/fault rules 1 38 229 9 1 6 2 Claims, other 3 37 35 4 2 1 1 Claims, repairs/settlement unsatisfactory 2 64 179 22 2 3 1 Claims, rights/entitlement 51 1 78 26 2 6 1 4 New business, misquotes 3 32 No category provided 1 Non‐renewal, accident/claim frequency 1 43 Non‐renewal, driving violations 13 Non‐renewal, other 27 Policy cancellation, non‐disclosure on application 40 1 1 Policy cancellation, non‐payment of premium 97 3 1 1 Policy cancellation, other 14 78 4 1 Policy cancellation, refund calculation 2 37 1 1 1 Policy rating 1 13 123 4 1 2 Policy renewal, premium increase 59 2 1 Question, coverage 2 15 27 1 1 Question, underwriting 1 4 13 1 2 Refusal to provide a coverage or endorsement 16 1 Source: Private communication with General Insurance OmbudService

The FARM and risk sharing pool market shares indicate some concern in terms of consumers’ ability to purchase insurance in the voluntary market. Further problems of availability are indicated by the percentage of GIO complaints related to availability for Ontario auto insurance. We note that there are also availability complaints in the government auto insurance provinces of Quebec and British Columbia, where the government mandate is to ensure availability to all drivers.

Affordability: Insurance affordability is measured using the annual average premium expenditure

(for all coverages) per insured car in a province. This average does not capture the price of insurance for drivers in all territories of a province, and it does not measure affordability of insurance relative to household income. However, since the average cost of claims is the primary determinant of premiums, this average measure is relevant for understanding the overall performance of the insurance system. It should also be borne in mind, however, that differences in average premiums across provinces fundamentally reflect differences in insurance benefits provided.

Our second measure of affordability is the percentage of complaints that the GIO categorizes as

“affordability” complaints as provided in Table 3. As with the above measure for availability, we acknowledge that counting only the category “affordability,” and not categories on policy rating or policy renewal, will underestimate concerns.

Ontario has the highest average premiums of all provinces, with the average Ontario premium being more than 30 percent higher than Alberta and more than 70 percent higher than premiums in the

Atlantic provinces. Ontario premiums are 28 percent higher than British Columbia premiums, and British

Columbia has generous first‐party benefits and unlimited access to the tort system. As noted previously, comparison across provinces is problematic due to the different benefit provisions in each province, yet the fact that the 10‐year growth in premiums in Ontario is 65.85 percent is an indicator of affordability problems. This is more than double the average growth rate of 27.46 percent in the other private market

provinces. Even though only 2.58 percent of GIO complaints relate to affordability, and Alberta has a higher rate of affordability complaints per 10,000 population than Ontario (0.33 versus 0.30), the explosive growth in premiums is of grave concern.

Service quality: Measuring insurance service quality is difficult, both because of the diversity of insurer services provided and the lack of available summary statistics measuring quality. Researchers have often used consumer complaints as a rough measure of insurer service quality, and we provide some evidence on complaints here. We were unable to obtain data on the number of complaints made to insurers or to provincial regulators. Thus, the complaint data reported here is the number of consumer complaints recorded by the GIO. We normalize the number of complaints from each province to the provincial population.

The number of complaints per 10,000 population (1.013) is noticeably higher in Ontario than other provinces, the next highest being Alberta (0.757) and the other provinces being much lower. To examine where concerns exist, we again examine the complaint data in Table 3. We focus here on

Alberta, Ontario, and Quebec.

For all three provinces, the largest fraction of complaints arises from the claims settlement process, despite all provinces having very different settlement processes. In Ontario, 48 percent of complaints (5.44 complaints per 10,000 population), in Alberta, 77 percent of complaints (6.61 complaints per 10,000 population) and in Quebec 66 percent of complaints (0.9 complaints per 10,000 population) are related to the claims settlement process. We observe that that the tort process of Alberta gives rise to the highest proportion of complaints by population, the pure no‐fault system of Quebec gives rise to the lowest proportion and the partial no‐fault system in Ontario falls in the middle. This is not surprising as claims outcomes are more predictable in a no‐fault environment.

However issues of non‐renewal and policy cancellations are of more concern in Ontario (23 percent of all complaints) than in either Quebec (8 percent) or Alberta (5 percent). The issues of policy cancellation and non‐renewal in Ontario are further indications of availability and affordability concerns.

Cost efficiency: To minimize the cost of service provision, the most efficient sellers who are best able to meet consumers’ desires must be present and successful in the market. Following the U.S. literature, we measure the percentage of overall market share held by national insurers (insurers who write in all 10 provinces) and direct writers. We use the term direct writers to capture those insurers that sell insurance face‐to‐face to their customers through exclusive agents or their own sales staff only.

Following from Kelly and Kleffner (2006), we also measure the market share held by commodity writers.

Commodity writers are insurers who sell auto insurance directly to consumers through indirect means such as affinity groups, the Internet, and over the phone.

The market share of commodity sellers is higher in Ontario than other private provinces, indicating that consumers have a cost‐efficient option for purchasing their insurance. However, the market share of national insurers is lower than other provinces: 23.3 percent compared to 40 percent in

Alberta, 44 percent in Nova Scotia, and 53 percent in Prince Edward Island. Despite the concerns about the auto insurance product because of the size of the auto insurance market, there are many more insurers in Ontario than in other provinces. This can also be seen in the Herfindahl‐Hirschman Index (HHI) for Ontario compared to the other provinces.

Sustainability: In both private and public insurance systems, long run sustainability requires that premium revenues cover losses and operating expenses (after taking into account investment income earned and taxes paid). The ratio of earned premiums to incurred claims and loss adjustment expenditures (the loss ratio) is therefore a rough measure of premium adequacy. Because the loss ratio does not account for investment income, taxes, or other expenses which are not related to the

settlement of claims such as underwriting and general overhead expenses, there is no specific value of the loss ratio which can be used as a target for premium adequacy.14 Moreover, because accidents and claims expenditures are stochastic and evolve over time, the loss ratio from a single year may be a poor measure of premium adequacy. Nonetheless, the average value of the loss ratio over time provides some evidence of premium adequacy (or excess).

Other measures of insurance system sustainability are the rates of growth in premiums and claims over time. High rates of inflation in premiums will lead to problems of affordability over the long term, and are indicative of underlying problems with cost growth (in claims or other expenses). We use the total percentage growth in premiums per insured vehicle for the decade 2000–2010 to measure premium inflation across provinces. We also capture the growth in average claim severity for the same period.

A final indicator of the potential for sustainability concerns is the automobile accident rate in a province. High underlying rates of accidents lead to higher costs of claims, which pose a threat to sustainability. Moreover, a well‐functioning insurance system should not measurably decrease incentives for road safety. We use two variables to capture the automobile accident rate in each province – the number of accidents with personal injury (but no fatalities) and the number of fatal accidents. Transport

Canada collects these accident statistics from police reports, and as such these statistics are distinct from insurance claiming behaviour.

Table 2 clearly indicates that the Ontario auto insurance system is not sustainable. The five‐year average loss ratio was 84.10 percent which is significantly higher than other private market provinces,

14 The expense ratio for the entire U.S. property/casualty insurance market in 2011 was 29 percent (Insurance Information Institute, 2012). MSA Research Inc. (2012) reported expense ratios of 30 percent for personal and multi‐ line insurance companies in Canada. Using this as a benchmark implies that without accounting for taxes and investment earnings, insurers with loss ratios greater than 70 percent would lose money. 50

whose average loss ratio was 60.36 percent. The Ontario loss ratio also exceeds that of the government insurers in both Manitoba and British Columbia. With loss ratios this high, enough premiums are not being collected to cover all expenses and as such insurer equity must be used to cover operating losses.

This creates concerns that firms heavily involved in underwriting Ontario auto insurance could suffer from solvency concerns in the future. Indeed, Leadbetter and Stodolak (2009) in a study on Canadian insolvencies note that the majority of Canadian insurer insolvencies from 1990 to 2007 were Ontario insurers who were primarily writers of auto insurance.

As well, the 10‐year growth in average losses per insured vehicle was 52.71 percent in Ontario compared with 4.07 percent in Alberta and negative growth in New Brunswick, Nova Scotia, and Prince

Edward Island. Therefore it is not surprising that the 10‐year premium growth rate in Ontario (65.85 percent) is also significantly higher than the other private market provinces’ 10‐year average premium growth rate of 27.50 percent. The growth in both losses and premiums is clearly driven by increases in severity, as the 10‐year growth rate in average claim severity in Ontario is 152.36 percent. Insurance

Bureau of Canada notes that the number of catastrophic injury insurance claims is rising faster than other claims, even though hospitalizations from motor vehicle accidents are falling. From 2004 to 2010 the total number of no‐fault accident benefit claims rose 28 percent but the number of large claims has more than doubled.15 Severity growth is much flatter across all other provinces, ranging from ‐7.13 percent in Nova

Scotia to 15.41 percent in Alberta.

15 http://www.canadianunderwriter.ca/news/claims‐costs‐for‐ontario‐auto‐insurers‐still‐out‐of‐control‐ ibc/1001413525/. 51

Figure 2 – Number of Auto Accidents in Ontario: 1999 to 2009

Number of Fatal And Personal Injury Accidents - Ontario 900 70,000 800 60,000 700 50,000 600 500 40,000 400 30,000 300 20,000 200 100 10,000

Number of Fatal Accidents Fatal Number of 0 0 Number of Personal Injury Accidents Injury Personal Number of

Fatal Personal Injury

Source: Ontario Ministry of Transportation, 2009

This is indeed puzzling as vehicles and roads have gotten safer over the past decade, and in 2009,

Ontario had the second lowest injury accident rate and the lowest fatal accident rate among all of the provinces. And as can be seen in Figure 2, even with a growing number of road users there has been an overall decline in the number of auto accidents causing injury or death over the past decade.

Fairness: In general, measures of fairness may be conceived of horizontally (similar individuals receive similar outcomes) or vertically (individuals in different circumstances receive outcomes that reflect their circumstances). These different conceptions may lead to different judgments regarding the fairness of a system. Similarly, with regard to insurance pricing there are two dramatically opposing views of fairness. The concept of fairness underlying social pricing requires that differences in premiums charged to individuals in different locations or of different ages or genders should be minimal – only those characteristics that are controllable should be used in pricing of insurance. Another definition of

social pricing is that prices should reflect the underlying ability to pay.16 Risk‐based or actuarial pricing relies on a different metric of fairness – namely, that differences in premiums charged to different individuals should vary based on differences in their expected accident probabilities and claims costs. The first notion implies that within‐province variation in premium charges should be small; the second implies that within‐province variation in premium charges relative to claims expenditures should be small. We compare both of these measures across provinces.

For Ontario and Alberta we collect premium and loss data by GISA territories for total coverages, and for all private market provinces we collect third‐party liability coverage by driving record class. As discussed in Section 3.4, the experience rating mechanism in all private market provinces assigns drivers to a driving record class based on the number of years of driving that are free of at‐fault claims. Drivers with zero years of no at‐fault claims are placed in driving record class 0, drivers with one year of no at‐ fault claims are placed in driving record class 1, and so on up to driving record class 5. Drivers with 6 or more consecutive years of no at‐fault claims are placed in driving record class 6.

We measure fairness in social pricing (horizontal fairness) by the coefficient of variation (CV) of premiums across the statistical territories in the provinces of Ontario and Alberta. The CV describes the dispersion of a variable and is calculated as one hundred times the standard deviation of the variable divided by the mean. If social pricing exists, there should be little difference in the premiums charged across territories and therefore the CV should be relatively close to zero. As can be seen in Table 2, significantly more horizontal equity or fairness exists across the four statistical territories in Alberta than the nineteen statistical territories in Ontario. In 2009, the CV for Alberta was 21.0 compared to a CV of

104.5 for Ontario.

16 Many social insurance products such as healthcare for the elderly or poor, or employment insurance are priced in this manner. 53

Vertical fairness reflects the extent to which individuals pay premiums commensurate with their risk. If individuals in each statistical territory are paying a premium commensurate with their risk levels, then the loss ratios across different territories should be fairly similar, and thus the CV of loss ratios should be relatively close to zero. With respect to vertical fairness across territories, we note that Alberta has fairer premiums than Ontario. This is surprising given the observation about greater horizontal fairness in Alberta. This implies that although premiums vary greatly across the different statistical territories in Ontario, losses vary even more and premiums have not been set in Ontario to capture the differences in loss experience across the different statistical territories.

We next examine fairness in risk‐based pricing using the CV of loss ratios across driving record class for all private market provinces. Fairness in risk‐based pricing would result in similar loss ratios across each driving record class and therefore CV values relatively close to zero. Overall, because of the stochastic nature of losses, if premiums are actuarially fair, then larger provinces should have smaller CVs.

This is reasonable because if some driving record classes have very few drivers, then there is less predictability in incurred losses (for example in Prince Edward Island, there were just over 1000 drivers in driving record class zero compared to almost 60,000 drivers in driving record class zero in Ontario). This relationship between provincial size and the CV exists for the Atlantic provinces. However, the CV of

Alberta loss ratios by driving record class (9.6) is higher than for the next smallest province Nova Scotia

(8.6), reflecting the cross‐subsidization of the grid pricing mechanism. In addition, Ontario’s CV of loss ratios by driving record is 12.9 and is closer in size to that of New Brunswick. Thus it appears that there are more cross‐subsidies and less fairness across driving record class in Ontario than in most provinces.

Competition: A competitive market requires a large number of firms with no single firm dominating the market through a large share of business. Competition is measured by the number of insurers active in automobile insurance sales in a province; the combined market share of the largest four

private insurers in the market (the four firm concentration ratio); and the Herfindahl‐Hirschman Index of market concentration (HHI). The HHI is calculated as the sum of the squared values of each firm’s market share. By squaring the market shares before adding them, firms with larger market shares are weighted more heavily in the index. This leads to a higher HHI when there is a smaller number of firms in the market and when market shares are unevenly distributed across firms. By convention the HHI is multiplied by 10,000: a value near zero indicates many small firms of equal size, while a pure monopoly yields a value of 10,000.

Lack of competition is not a problem in the Ontario market. There are more insurers writing auto in Ontario than any other province and there is a lower concentration of the top four insurers (31.1 percent in Ontario as compared to 44.3 percent in Alberta and 56.3 percent in Newfoundland and

Labrador). The HHI of market shares is lower in Ontario than in other markets with private automobile insurance: 427 in Ontario versus 687 in Alberta and 1055 in Newfoundland and Labrador. This heavy competition in Ontario is to be expected. Tennyson (1997) found similar results with respect to California

– it was one of the most stringently regulated states for auto insurance, but its large size made it very attractive to insurers.

Overall there exist concerns with respect to affordability and availability as evidenced both by the percentage of affordability and availability complaints as noted in Table 2 and the further analysis of complaint data given in Table 3. There also appears to be great variability in loss experience across the province. Ontario has greater variability in premiums across territories than does Alberta, and even greater variability in loss ratios. This suggests that the differences in premiums across different territories are not sufficient to cover the differences in losses incurred in different regions of the province.

An examination of the sustainability criteria highlights the fact that the Ontario automobile insurance system is not sustainable. Ontario clearly has the most expensive insurance product in Canada

and the costs of this system continue to grow. Most troubling is that the growth in claims costs is at odds with improvement in road safety that has been evidenced in the province over the last decade.

The next section of the report examines these issues in more detail. We look at the change in the

Ontario auto insurance product since the introduction of no‐fault insurance in 1990. We examine the claims experience of Ontario compared to other provinces over this time period and we examine the claims experience within different territories of the province.

5 Automobile Insurance in Ontario

To provide greater insight into the factors that underlie the explosive claims and premium growth of Ontario’s auto insurance system, in this section we review the auto insurance system in Ontario. We start with an overview of the current product and then review the history of auto insurance reform in

Ontario. We conclude this section with a discussion of claims costs and trends for the province, paying particular attention to any differences between the four largest metropolitan areas and the rest of the province. Statistics Canada lists the four largest Census Metropolitan Areas in Ontario as

Toronto/Mississauga, Ottawa/Gatineau17, Hamilton/Burlington, and Kitchener‐Waterloo/Cambridge. To examine the claims costs trends, we use the statistical territories developed by the General Insurance

Statistical Agency and define the Greater Toronto Area (GTA) as statistical territories 710 (Oshawa,

Aurora, Newmarket, and Orangeville) and 717 (Metropolitan Toronto and Markham, Richmond Hill,

Vaughan, and Peel), Ottawa is statistical territory 711, Hamilton/Burlington is statistical territory 704

(Halton and Hamilton‐Wentworth) and Kitchener‐Waterloo/Cambridge is statistical territory 706

(Brantford, Guelph, Kitchener‐Waterloo, and Cambridge). The GTA is by far the largest metropolitan area in Ontario (and Canada). In 2009 (the latest year for which we have data for different geographic territories), the GTA accounted for 37 percent of earned vehicles and 45 percent of total premiums collected in the province.

5.1 The Current Product

As in other jurisdictions in Canada, auto insurance is a mandatory product in Ontario. There must be proof of insurance before license places can be attached to a vehicle or registration can be renewed.

17 Statistics Canada categorizes the combined area of Ottawa, Ontario, and Gatineau, Quebec as a single Census Metropolitan Area. 57

Those who operate a vehicle without insurance are guilty of an offense and can be fined between $5,000 and $25,000 for a first offense and between $10,000 and $50,000 for subsequent offenses.18

Currently, drivers in Ontario are required to purchase four mandatory auto insurance coverages: a minimum ($200,000) of third‐party liability coverage for both bodily injury and property damage,19 first‐ party (no‐fault) statutory accident benefits, first‐party coverage for not‐at‐fault damage to one’s own automobile (direct compensation/property damage) and first‐party coverage against damage caused by at‐fault uninsured drivers. Optional coverage includes at‐fault collision damage for the vehicle, comprehensive insurance for non‐collision losses, underinsured motorist coverage as well as additional limits for liability and additional coverage for first‐party accident benefits.

Statutory accident benefits are relatively generous. A unique aspect of the Ontario system is that the maximum amount of accident benefits available is a function of the level of impairment of the individual, and individuals with catastrophic impairment are entitled to up to $1 million in medical and rehabilitation benefits and $1 million for attendant care benefits. The current standard auto insurance policy, effective since September 1, 2010, is summarized in Table 4. The current policy provides $50,000 for medical and rehabilitation benefits (including assessment costs) and $36,000 for attendant care benefits. Additional limits can be purchased for both non‐catastrophic and for catastrophic injuries.

Finally the tort deductible associated with court awarded compensation for pain and suffering may be reduced to $20,000 from $30,000, and for Family Law Act claims to $10,000 from $15,000.

18 http://www.mto.gov.on.ca/english/dandv/vehicle/register.shtml 19 The Financial Services Commission of Ontario reports that over 99 percent of Ontario drivers purchase more than the mandatory minimum. 58

Table 4 – Statutory Accidents Benefits in Ontario

Coverage Current Standard Auto Policy Options Available to Increase Benefits

Medical, rehab, and attendant care $50,000 for medical and rehab $100,000 or $1,100,000 for medical benefits for non‐catastrophic benefits, including assessment costs; and rehab benefits, including injuries $36,000 for attendant care benefits assessment costs; $72,000 or $1,072,000 for attendant care benefits

Medical, rehab, and attendant care $1,000,000 for medical and rehab An additional $1,000,000 for benefits for catastrophic injuries benefits including assessment costs; medical, rehabilitation, and $1,000,000 for attendant care attendant care benefits including benefits assessment costs

Caregiver benefit Up to $250 per week for the first Up to $250 per week for the first dependant plus $50 for each dependant plus $50 for each additional dependant; available only additional dependant; available for for catastrophic injuries all injuries

Housekeeping and home Up to $100 per week, available only Up to $100 per week, available for maintenance expenses for catastrophic injuries all injuries

Income replacement benefit 70 percent of gross income up to Weekly limit can be increased to $400 per week $600, $800, or $1000 per week

Dependant care benefit Not provided Up to $75 per week for the first dependant and $25 per week for each additional dependant to a maximum of $150 per week

Death and funeral benefits $25,000 lump sum to an eligible $50,000 lump sum to an eligible spouse; $10,000 lump sum to each spouse; $20,000 lump sum to each dependant; maximum $6,000 dependant; maximum $8,000 funeral benefits funeral benefits

Indexation benefit – applicable to Not provided Annual adjustment according to the income replacement benefit, non‐ CPI for Canada earner benefit, caregiver benefit, attendant care benefit, or medical and rehab benefit

Source: Financial Services Commission of Ontario (www.fsco.gov.on.ca)

5.2 The History of Auto Insurance Reform in Ontario

Since 1989 significant reform of the Ontario automobile system has occurred on six separate occasions, the most recent going into effect on September 1, 2010. A summary of key changes and the

motivation for reforms is given in Table 5. Here we detail the major changes in the auto insurance product and the regulation of said product over the past 25 years.

As noted by the Auditor General of Ontario (2011), each legislative reform was enacted primarily to contain rising accident benefit costs (and their impact on premiums), and each reform has provided only a temporary reduction in premiums (and losses) as illustrated in Figure 3 below.

Figure 3 – The Impact of Auto Insurance Reforms on Average Premiums

Source: Ontario Auto Insurance Anti‐Fraud Task Force, 2011

Table 5 – A Summary of Auto Insurance Reform in Ontario from 1990 to 2010

Year Bill Rationale for Reform Wording for Verbal Right to Sue for Right to Sue First‐party Accident Benefit Threshold Non‐economic for Economic Reforms Losses Losses 1990 Bill 68 Improve accessibility and Sustained permanent and Yes, if injury meets Yes, if they $500,000 limit on medical benefits, affordability, correct serious disfigurement or verbal threshold exceed first‐ disability benefits for both shortcomings in tort sustained permanent serious party benefits employed and unemployed system. impairment of an important bodily function 1994 Bill 164 Cover more injured Sustained permanent and Yes, if injury meets No $1 million limit on medical benefits, claimants, move closer to serious disfigurement or verbal threshold, increased disability benefits, pure no‐fault coverage. sustained permanent serious $10,000 deductible introduction of caregiver and impairment of an important applies housekeeping benefits bodily function 1996 Bill 59 Reduce costs associated Sustained permanent serious Yes, if injury meets Yes, if they Introduction of two‐tiered benefits: with high first‐party disfigurement or serious verbal threshold, exceed first‐ $100,000 limit on medical benefits accident benefit claims. impairment of an important $15,000 deductible party benefits for most injuries, but $1 million physical, mental or applies limit for catastrophic impairments psychological function 2003 Bill 198 Improve affordability and Introduced definitions for Yes, if injury meets No change Small change in benefits, but accessibility by “serious,” “ important,” and verbal threshold, from Bill 59 expansion of the definition of decreasing assessment “permanent” to tighten $30,000 deductible “catastrophic” costs. threshold applies 2006 Elimination Reduce costs by No change from Bill 198 No change from Bill No change No change from Bill 198 of DACs eliminating Designated 198 from Bill 198 Assessment Centres (DACs). 2010 Bill 16 Reduce costs associated No change from Bill 198 No change from Bill No change Sharp reduction in benefits for with high first‐party 198 from Bill 198 “non‐catastrophic” impairments; accident benefit claims. $2,000 cap on fees and expenses Reduce expenses in for conducting an assessment; assessment process. restriction on caregiver benefits and housekeeping for catastrophic injuries only; caps on minor injury claims

For example, the 2003 reforms resulted in a decrease in average premiums of 16 percent over the period 2003 to 2007. The average premium fell from approximately $1,499 per vehicle in November 2003 to $1,260 in June of 2007. However, the removal of the Designated Assessment Centres in 2006 increased the number of assessments per file. Cameron (2011) noted that the cost of assessments on a per vehicle basis increased from $39 in 2004 to $141 in 2009. The impact of the 2010 reforms cannot be fully assessed yet, but McGillivray (2011) comments that slower growth of accident benefit costs in the second half of 2010 was evidence that the reforms were having the desired effect.

The report of the Auditor General of Ontario (2011) in its review of the Financial Services

Commission of Ontario (FSCO) provides a concise summary of the spiralling costs of auto insurance claims in Ontario. From 1999 to 2009 the number of people killed or injured in auto accidents in Ontario declined by 25 percent. However from 2005 to 2010, the total cost of injury claims rose 150 percent even though the number of injury claims in the same period increased only 30 percent. The number of personal injury claims in 2009 was almost 75,000, 20 percent higher than the number of people who reported having been injured in an automobile accident that year.20 Across the province in 2010, there were approximately 584,000 auto insurance claims, with claims costs totalling $8.7 billion: 52 percent of costs were for accident benefit claims, 23 percent were for third‐party liability and the remaining amounts were for property damage to automobiles (both not‐at‐fault and at‐fault).

In 2005, statutory accident benefits accounted for one‐third of auto insurance claims costs; by

2010, they accounted for one‐half of all auto insurance claims costs in Canada. Claims costs for the other lines of auto insurance grew 16 percent during the same five‐year period, while Statutory Accident

20 The Auditor General’s report does not provide references for the data it presents. However, road safety statistics are typically reported through the Ministry of Transport, whereas General Insurance Statistical Agency (GISA) collects data on auto accident claims. 62

Benefit claims grew by 92 percent. The high cost of pre‐treatment examinations and assessments

(roughly one‐third of every dollar spent on accident benefit claims) was inconsistent with assessment costs in other provinces. Toronto was identified as particularly problematic. Between 2008 and 2009, statutory accident benefits payments rose 37 percent in the Greater Toronto Area (GTA), compared to 23 percent in other Ontario cities and just 14 percent in rural areas. The Ontario Auto Insurance Anti‐Fraud

Task Force (2011) notes that although the GTA accounts for only one‐third of insured vehicles, it accounted for more than 80 percent of the growth in accident benefit claims.

To provide greater insight into the intended effects and actual consequences of the various reforms, we briefly summarize the reforms and discuss what the anticipated outcome would be (in terms of frequency and severity of claims) if the reforms were successful. We then examine claim frequency, claim severity, and loss adjustment expenses over time in light of these predicted effects.

5.2.1 Bill 68 (1990)

After years of rising auto insurance prices and concerns about affordability and availability of insurance, auto insurance reform was the main election issue in the 1987 provincial election (Baetz, 1993) and the Liberal government’s win was partially due to their promise to effect rate reductions. In February

1998, the Liberal government introduced the Ontario Automobile Insurance Board Act, establishing both the Board as the rate control agency and introducing controls on the rate classification process used by auto insurers. As chronicled by Lascher (1999), the Liberals’ ability to reduce premiums was damaged by the December 1988 government‐commissioned Mercer Report which recommended that to restore profitability to the sector, auto insurance premiums should be increased by 35 to 40 percent. Devlin

(1993) concludes that given the political environment and the presence of strict rate regulation, the insurance industry was forced to lobby for no‐fault auto insurance provisions.21

On September 15, 1989, the Liberal government announced the proposal for the adoption of the

Ontario Motorist Protection Plan (OMPP) which despite vigorous objection from various political and grassroots parties took effect on June 22, 1990, and remained in effect until December 31, 1993. This reform was based on the recommendation by the Ontario Task Force on Insurance lead by David Slater in

May 1986. The task force recommended that the Ontario government adopt a pure no‐fault system of compensation for automobile accidents, similar to existing schemes in Quebec and Saskatchewan, due to the shortcomings of the tort system. Key shortcomings listed by the task force were the high administrative cost of providing insurance and the unpredictability of compensation under the tort model. It was recommended that the new auto insurance regime impose mandatory first‐party coverage and eliminate liability between drivers, but still allow for liability claims from pedestrians. The report cautioned that introducing no‐fault insurance would increase moral hazard as first‐party benefits would increase substantially.

The OMPP adopted was a partial no‐fault scheme with a verbal threshold. The verbal threshold meant that insured parties had access to legal recovery for non‐economic damages if they sustained permanent and serious disfigurement or sustained permanent serious impairment of an important bodily function. Those injured would receive benefits as defined by the no‐fault benefits schedule. In this schedule, the benefits for many groups of people – unemployed, caregivers and students – were minimal.

The weekly benefits to employed persons were much more generous: 80 percent of gross weekly income

21 No‐fault was not a novel idea in Ontario. Because of soaring auto insurance premiums since the early 1980s, many government‐sponsored commissions have examined ways to contain costs and improve coverage. The Slater Report in 1986 recommended a pure no‐fault system. Two years later, the Osborne Report strongly supported a pure tort system but with expanded mandatory first‐party accident benefits. 64

up to a maximum of $600 a week for three years. If the verbal threshold was met, the claimant had full right to sue for recovery of monetary losses that exceeded the first‐party benefits. Collateral benefits such as payment received under an employee’s disability or supplemental health plan were deducted from any damages awarded to the plaintiff. OMPP also introduced direct compensation for not‐at‐fault property damage to the insured’s vehicle. It was estimated that the introduction of OMPP should reduce insurance costs by $500 million per year.

These reforms were expected to reduce the frequency of bodily injury liability claims, but increase average severity of these claims due to the stringent liability threshold. The average severity of accident benefit claims would increase as first‐party benefits were more generous. However it is unclear whether the overall average severity of accident benefits plus bodily injury liability claims would be higher or lower than prior to the reforms depending on whether first‐party benefits were more generous than what was paid under tort. With respect to damage to automobiles, because of the new direct compensation – property damage (DC‐PD) coverage, a decrease in third‐party property damage claims was anticipated, and a corresponding increase in first‐party DC‐PD claims. Overall, cost savings were expected due to a reduction in loss adjustment expenses associated with shifting claims from third‐party to first‐party compensation.

5.2.2 Bill 164 (1994)

The newly elected NDP party introduced Bill 164 in January 1994 (and this bill remained in place until October 31, 1996). The key modification was a change in the verbal threshold to encompass an insured’s serious disfigurement or serious impairment of an important physical, mental, or psychological function. This is a lower threshold than under OMPP, due to the absence of a “permanent” requirement and the inclusion of non‐physical injuries, allowing more injured persons to sue for non‐economic damages.

The right to sue for economic loss was eliminated in its entirety, but first‐party benefits were greatly increased. For example, weekly income replacement benefits were set to 90 percent of net income to a maximum of $1,000 per week. Weekly benefits were also introduced for caregivers, students and the unemployed. In addition the upper limit for medical and rehabilitation costs was significantly increased to $1 million. Collateral benefits were no longer deducted from a claimant’s tort award.

However, a $10,000 deductible was introduced for non‐economic damages (those whose losses were assessed to be less than $10,000 were not allowed to recover).

These reforms were expected to increase the frequency of bodily injury liability claims arising from the lowered eligibility threshold. However, because the right to sue for economic loss was eliminated, it is unclear whether bodily injury liability claim severity would increase under Bill 164. In the absence of moral hazard, a change in the frequency of accident benefit claims should not be expected, but increased benefits could attract more claims, leading to higher claim frequency. Overall, the increased first‐party benefits should lead to higher claim severity, but the reforms should have no expected impact on property damage claim frequency or severity.

5.2.3 Bill 59 (1996)

The Conservative government introduced further changes in 1996. On November 1, 1996, Bill 59, the Auto Insurance Rate Stability Act became law. This bill restored the right to sue for economic damages if losses exceeded the compensation available under statutory accident benefits. The verbal threshold was tightened from Bill 164, re‐introducing the word “permanent” into the threshold for both disfigurement and impairment. The deductible for non‐economic damages was increased to $15,000 and was applied before any split in liability. Statutory accident benefits were reduced from Bill 164; for example, weekly income benefits were capped at 80 percent of net income up to $400 a week. Collateral benefits were deducted from tort awards for monetary losses only for pre‐trial collateral benefits.

Most importantly, Bill 59 introduced a two‐tiered schedule of payments based on the level of injury of the claimants. Individuals who are catastrophically impaired (as defined in the Statutory Accident

Benefits Schedule) had access to a higher level of medical and rehabilitation benefits ($1 million instead of $100,000), and only those catastrophically injured could claim future care costs. To contain costs, medical providers could not start therapy until they had approval from the insurance company for the treatment plan.

The two‐tiered system was introduced to contain claims costs by reducing the average severity of losses paid under the Statutory Accident Benefits Schedule. However, the two‐tiered system also would increase loss adjustment expenses because it creates a greater need for assessment for benefits purposes. The reforms were not expected to impact the frequency of accident benefit claims unless the reduced benefits also reduced moral hazard in claiming. The expected impact on bodily injury liability claim frequency is ambiguous. Because the verbal threshold was tightened again, this should decrease the number of bodily injury liability claims; however the reintroduction of the right to sue for economic damages combined with the reduction in statutory benefits could lead to an increase in liability claims and an increase in liability claim severity. No change in frequency or severity of property damage claims was expected.

5.2.4 Bill 198 (2003)

The Liberals won the 2003 election, having campaigned on a promise to freeze soaring insurance rates immediately if elected and then reduce them by an average of 10 percent. Shortly after being sworn in on October 23, Ontario's new Liberal government froze auto premiums effective that day. On

November 26, the government introduced the Automobile Insurance Rate Stabilization Act which froze insurers' rates at levels approved on or before October 23 and prevented further approvals from taking place for 90 days.

Reforms were introduced as part of the “Keeping the Promise for a Strong Economy Act.”

Although the wording of the verbal threshold was unaffected, the revised Insurance Act provided definitions for the threshold to reduce moral hazard. Serious was defined as “substantially interfere with a person’s ability to continue regular employment, interfere with most of the person’s activities of daily living.” Importance was defined as “important to most of the person’s usual activities of daily living” and permanent was defined to mean “expected to continue without substantial improvement when sustained by persons in similar situations.” The definition of catastrophic impairment was slightly expanded.

The most significant reforms were associated with reducing expenses associated with the assessment process, restricting fees that could be charged by healthcare professionals, and reducing excessive or abusive claiming behaviour associated with accident benefits claims. Other reforms included an increase in the deductible for non‐economic damages, and the removal of the need to have a catastrophic injury to claim future care costs. Pre‐trial collateral benefits were deducted from awards on a matching basis (i.e., lost income for lost income) but future losses were not reduced by collateral benefits.

The major focus of these reforms was to decrease assessment costs and costs arising from medical treatments. If successful, the reforms would lead to lower loss adjustment expenses for accident benefit claims, and a decrease in the severity of these claims. The frequency of accident benefit claims would not be impacted. The guidance given to the terms in the verbal threshold was meant to decrease the number of bodily injury liability claims, and the increase in the deductible for non‐economic damages was expected to decrease the average severity of liability claims. No change in the frequency or severity of property damage claims was expected.

5.2.5 Elimination of Designated Assessment Centres (2006)

Designated Assessment Centres (DACs) were created in 1994 to provide neutral opinions on injuries and treatment plans in Ontario auto insurance. On March 1, 2006, the Liberal government, acting on an election promise, eliminated the DACs and new procedures were put in place. Insurers and the government insisted that the DACs had outlived their purpose – they were expensive and their existence did not improve health outcomes. The medical profession argued that the DACs provided consumer protection and minimized delays until treatment. They argued that removing the DACs would actually increase costs in the system. Thus, the expected impact of the removal of DACs is unclear. If the insurance industry and government were correct, then loss adjustment costs would decrease and there would be no other impact on accident benefit costs. If the medical profession was correct, then the removal of

DACs would result in poor health outcomes, higher accident benefit claims, and increased loss adjustment expenses.

5.2.6 Reforms of 2010

The 2003 changes in the Insurance Act required FSCO to undertake a review of the effectiveness of auto insurance in Ontario at five‐year intervals and make recommendations to the Ministry of Finance.

2008 marked the first statutory review, and the results of the review led to changes being enacted in

2010.

Reforms included the implementation of a cap of $3,500 on minor injury medical‐rehabilitation expenses, the development of standardized guidelines for treatment of minor injuries, including assessment costs in coverage limits, and stricter monitoring of fees charged by healthcare providers and assessors. The new policy provides $50,000 for medical and rehabilitation benefits (including assessment costs), compared to $100,000 (excluding assessment costs) under the previous policy, and $36,000 for attendant care benefits, compared to $72,000 previously, if the insured is not catastrophically injured. 69

There was no change in benefits for those who are catastrophically injured. More consumer choices

(higher limits and varying deductibles) for accident benefits, DC‐PD, and recovery for non‐monetary losses under tort were added.

These reforms should decrease the severity of accident benefit claims as first‐party benefits were reduced, but frequency should be unaffected. However, it is possible that there will be an increase in the frequency of bodily injury liability claims due to the reduction in benefits for non‐catastrophic injuries.

These reforms should produce no change in frequency or severity of property damage claims. However, if moral hazard is reduced because of the decreased benefit for non‐catastrophic injuries, then a reduction in claims for non‐catastrophic injuries is expected.

5.3 Ontario Claim and Cost Trends

Using data provided by the General Insurance Statistical Agency (GISA), we examine trends in

Ontario claiming behaviour and insurance costs. These data will provide insight into the effects of the reforms and will highlight the key drivers of auto insurance cost growth. We analyze trends both in claim frequency (number of claims divided by number of insured vehicles) and severity (total loss and loss adjustment costs divided by number of claims) for each of the compulsory automobile insurance coverages. We first look at provincial level data for the years 1991 through 2010. Data for Ontario are displayed along with data for the private auto insurance markets of Alberta, New Brunswick, and Nova

Scotia to facilitate comparison of trends in Ontario relative to external market factors (such as countrywide trends that reflect changes in cost of services and the number of road users). Premium and claim severity data are converted to constant 2010 dollars using the Consumer Price Index (CPI) for each province in order to net out general price changes within and between provinces from auto insurance cost growth. We examine GISA data of premiums and claims history by statistical territory from 2000 to

2009 to further distinguish between the experience of the largest urban centres within Ontario (the GTA,

Ottawa, Hamilton/Burlington, Kitchener‐Waterloo/Cambridge) and the remainder of the province.

5.3.1 Bodily Injury Liability Claims

Figure 4A displays annual average claim frequency for bodily injury liability (BI) claims. The sharp decrease in claim frequency in the tort provinces is driven by two effects: the increased safety of vehicles which has resulted in less severe injuries for those involved in accidents and success in automobile insurance reform in each of the tort provinces. Countrywide there has been a decrease in the number of fatal collisions, and the number of collisions causing bodily injury. This is reflected in the falling BI claims frequencies in Alberta, New Brunswick, and Nova Scotia, but not Ontario, although there is a slight decrease in frequency rates since 2000.

Figure 4A – Number of BI Claims per 100 Vehicles – Provincial Comparisons

BI Claim Frequency 1.40

1.20

1.00

0.80

0.60

0.40

0.20

0.00

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Relative to the pure tort provinces, the Ontario data show the expected low frequency of claims due to the verbal threshold, but a constant upward trend in claims over the period. Overall, in Ontario 71

average BI claim frequency increased 150 percent during this time period, suggesting that the auto insurance reforms have been generally ineffective by this measure. The one exception is over the period

2003–2006 when there was a decrease in BI claim frequency, suggesting that the strengthening of definitions in the verbal threshold had the desired effects.

Figure 4B displays annual average claim frequency for BI claims for the major metropolitan centres versus the rest of Ontario. In every year, the frequency of claims in the GTA exceeds the frequency of claims in the rest of the province. The volatility of claim frequency in the GTA is also much greater. It is also evident that the tightening of the verbal threshold in 2003 had a greater impact on claims in the GTA than in the rest of the province. However, the impact was short lived as claims frequency increased dramatically in the following years.

Figure 4B – Ontario Comparison Number of BI Claims per 100 Vehicles

BI Accident Frequency 0.30

0.25

0.20

0.15

0.10

0.05

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

Higher population density has also been shown to be correlated with fewer serious accidents

(Kochanowski and Young, 1985; and Kelly, Kleffner and Tomlinson, 2010), especially if it is associated with

more inner‐city and less rural interstate driving (Cummins, Phillips and Weiss, 2001). Because of the threshold for tort claims in Ontario, BI claims arise from accidents severe enough to pierce the threshold, which would arguably imply that jurisdictions with a higher frequency of bodily injury liability claims have more serious accidents. Except for Ottawa, the results of this graph are opposite to the expected. We had anticipated that less urban and rural areas (i.e., the rest of Ontario) should have the highest rate of liability claims and the jurisdictions with the most congested roads (i.e., the GTA) should have a much lower frequency of BI claims.

Figure 5A shows trends in annual average BI claim severity (cost per claim) for Ontario versus the other provinces, and Figure 5B provides the GTA versus the rest of the province. As expected, the verbal threshold in Ontario leads to high average claim severity relative to other provinces. We also observe large variations in average severity over time, suggesting that the auto insurance reforms did affect claim severity. From 1991 to 1993 bodily injury liability severity trended upward, implying that in real terms, even with the strict verbal threshold, claims costs were growing. A sharp drop in average severity is observed in 1994 with only a slight uptick in costs through 1996. The removal of the right to sue for economic loss greatly reduced the size of claims payments. However, settlements for non‐economic damages continued to grow slightly. The right to sue for economic losses was restored in 1996, and average claim severity rose in every year between 1997 and 2002, before experiencing a slight drop in

2003.

Figure 5A – Average BI Costs per Claim in Constant 2010 $ – Provincial Comparisons

BI Claim Severity in Constant 2010 $ $200,000 $180,000 $160,000 $140,000 $120,000 $100,000 $80,000 $60,000 $40,000 $20,000 $0

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

The auto reforms in 2003 had an immediate, but short‐lived impact of reducing claims costs.

However, in 2004 and 2005 severity rose rapidly before trending downward from 2006–2009. In 2010 BI severity increased again, although in real terms average BI claim severity was 3.25 percent less in 2010 than in 1991. This growth in liability severity is concerning. Opponents of no‐fault insurance argue that the adversarial process of the tort system is more effective in containing claims costs. Yet in Ontario we observe a lack of cost containment for even those claims settled through the tort process. This has not been the experience in the other three provinces, which saw much lower average claim severity, slower growth in severity over time, and reductions in claims severity after insurance reforms during the 2003 to

2005 time period.

Figure 5B – Ontario Comparison Average BI Costs per Claim in Constant 2010 $

BI Claims Severity $250,000

$200,000

$150,000

$100,000

$50,000

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

Figure 5B depicts the severity of accidents within different parts of Ontario. The GTA has on average the lowest severity of all claims, and the smaller towns and rural areas that create the “rest of

Ontario” have the highest claim frequencies. Severities for other cities generally fluctuate between these two boundaries, and it is evident that the various reforms have not had a uniform impact on the size of claims over the various regions in Ontario. The random fluctuations in claim severity over time do not contradict the hypothesis that the adversarial nature of the tort process is efficient in minimizing the impact of excessive claims behaviour, at least at an individual claims level. Overall, these claim severity results are more consistent with expectations from the literature. However when combined with the results from Figure 4B, it is likely that there are many nuisance claims filed in the larger urban centres.

Because different insurance compensation rules may lead to shifting of costs between BI and first‐party accident benefits (AB) coverages, it is possible that the increases in BI claim severity are mirrored by corresponding reductions in AB severity. Figure 6A and Figure 6B explore this hypothesis, plotting the average injury cost per insured vehicle for the study period. 75

Figure 6A – Average AB+BI Costs per Insured Vehicle in Constant 2010 $ – Provincial Comparisons

Average AB+BI Loss per Vehicle 2010 $ $1,100 $1,000 $900 $800 $700 $600 $500 $400 $300 $200 $100 $0

Alberta New Brunswick Nova Scotia Ontario

Source: Authors’ calculations from GISA data

The 1994 reforms in Ontario (more generous first‐party benefits and a reduction in the tort threshold) resulted in an increase in the injury claims costs per vehicle, as expected. There was a 15 percent drop in injury costs from 1996 to 1997 after the introduction of the two‐tiered benefit schedule in Bill 59. However injury costs per vehicle rose steadily until reforms were enacted again in 2003. From

2002 to 2004, injury costs per vehicle fell from $690 (constant 2010 dollars) to $555, yet claims costs began rising steadily in 2005. Fundamentally, the average cost per injury in Ontario is much higher than what is paid in the tort provinces of Alberta, New Brunswick, and Nova Scotia.

Figure 6B – Ontario Comparison Average AB+BI Costs per Insured Vehicle in Constant 2010 $

Average AB+BI Loss per Vehicle 2010 $ $1,600 $1,400 $1,200 $1,000 $800 $600 $400 $200 $0

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

Figure 6B graphs the accident benefits and BI loss per vehicle for the four largest urban areas and the remainder of Ontario. Except for the GTA, there has been very little change in claims levels from 2000 to 2009. The average annual combined growth rate for the regions is less than 3 percent. It is possible that there is some expansion of benefits over the nine years. Additionally, if price inflation in the medical sector exceeds the general level of price expansion, then we would also see growth over time. If we examine the non‐GTA regions only, we see a decrease in overall costs in 2003, indicating that those reforms were successful. There is a slight increase again in 2006 when the DACs were eliminated.

Although average claims costs increased overall between 2008 and 2009, only in Ottawa do 2009 claims exceed claims levels in 2002 and 2003.

Despite the experience in the remainder of Ontario, claims costs are growing exponentially in the

GTA and have been consistently higher than in the other regions. The 2003 reforms provided short term relief to claims growth, but by 2007 claims were again at 2002 levels. The average loss per vehicle grew

by 14 percent between 2007 and 2008 and 26 percent between 2008 and 2009. However, according to the Ontario Road Safety Annual Reports for 2008 and 2009 produced by the Ministry of Transport, there were 7.7 percent fewer accidents, 18 percent fewer fatal accidents, and an increase of 3 percent in accidents in which individuals were reported to be injured. Thus it does not appear that the rapid claims growth is caused by a dramatic increase in the number of serious accidents.

5.3.2 Compulsory Property Damage Claims

The next set of figures show comparable data for property damage (PD) claims under compulsory coverages (third‐party liability and DC‐PD combined) for Ontario versus other provinces. The data in

Figure 7A show that PD claim frequency is higher in Ontario than in other provinces, but follows a generally downward trend during the study period. The PD claim frequency rate was nearly 30 percent lower in 2010 than 1991. Although PD claim frequencies are falling in all provinces, the fact that Ontario’s claim frequency is the highest in almost every year is puzzling, since statistics collected by Transport

Canada suggest that there are significantly relatively fewer accidents on Ontario roads than in the other provinces. In 2009 (2003), the percentage of road fatalities per 100,000 population was 4.1 (6.9) in

Ontario, 7.7 (7.5) in Nova Scotia, 8.8 (12.4) in New Brunswick, and 9.6 (12.2) in Alberta.

Figure 7B shows the provincial distribution of direct compensation property damage claims only, and as such is not directly comparable to Figure 7A. This graph is consistent with expectations: there is a greater frequency of claims in more congested urban areas, and a lower frequency in rural areas. The greatest decline in frequency is in the GTA.

Figure 7A – Number of PD Claims per 100 Vehicles – Provincial Comparisons

PD Claim Frequency 5.00 4.50 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.00

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Figure 7B – Ontario Comparison Number of PD claims per 100 Vehicles

DC-PD Claim Frequency 6.00

5.00

4.00

3.00

2.00

1.00

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

The declining property damage claim frequency suggests that the increase in BI claim frequency is not due to an increase in accident rates. In fact, comparing trends in Figure 4A to those in Figure 7A

reveals that the number of BI claims per PD claim (both DC‐PD and PD liability) in Ontario increased nearly 250 percent between 1991 and 2010. This is displayed in Figure 8A. Again, because fewer claims pierce the threshold in Ontario, liability claim frequencies are much lower in Ontario. Also evident from this graph is that the auto insurance reforms of 2003–2005 in New Brunswick, Nova Scotia, and Alberta were largely successful in containing growing BI claims.

Figure 8A – Number of BI Claims per 100 PD Claims – Provincial Comparisons

Ratio of BI to PD Claim Frequency 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Values calculated in Figure 8B are the ratios of accident benefit claims to DC‐PD claims, and as such examine all accidents that give rise to claims. Consistent with earlier graphs, the GTA has the highest ratio of AB to DC‐PD claims, although there is evidence that the 2003 reforms curtailed the number of personal injury claims in the GTA. It is difficult to reconcile why the average rate of personal injury claims in the GTA is significantly higher than the rest of the province. The higher ratio in the GTA is even more difficult to interpret when the ratio of accidents with personal injury (including fatality) to total accidents using Ministry of Transportation collision data (2009) is 0.23. For Ottawa, the corresponding ratio is 0.20 and for the combined regions of Waterloo and Wellington (similar to GISA statistical territory of 80

Brantford, Guelph, Kitchener‐Waterloo, and Cambridge) the ratio is 0.21. The Ministry of Transportation ratios are expected to be lower as they will account for all accidents reported to the police, but the data in Figure 8B would not include (in the denominator) single vehicle accidents which give rise to property damage only. There may be other reasons related to claiming behaviour that would also give rise to slight differences between the two sets of ratios. However, it is difficult to reconcile the differences in ratios for the GTA.

Figure 8B – Ontario Comparison Number of BI Claims per 100 PD Claims

Ratio of AB to DC-PD Claim Frequency 0.60

0.50

0.40

0.30

0.20

0.10

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

Figure 9A reveals that property damage claim severity increased at a rate slightly higher than the general inflation rate in each province, but the increases in Ontario are similar to those in other provinces. These data show that the auto insurance reforms had little or no effect, as expected. Although there are some variations in the trends over time, the time patterns are largely consistent with those in other provinces. Similarly, as shown Figure 9B, claims in the GTA were slightly higher than the rest of

Ontario, but that could also reflect a higher cost of repairs, or a higher quality of car.

Figure 9A – Average PD costs per Claim in Constant 2010 $ – Provincial Comparisons

PD Claim Severity in Constant 2010$ $6,000

$5,000

$4,000

$3,000

$2,000

$1,000

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Figure 9B – Ontario Comparison Average DC‐PD costs per Claim in Constant 2010 $

DC-PD Claims Severity $6,000

$5,000

$4,000

$3,000

$2,000

$1,000

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

5.3.3 Statutory Accident Benefits

We now turn to analysis of trends in claims for statutory accident benefits (AB). To provide greater insights we present the data separately for medical costs benefits and non‐medical benefits such as income replacement and attendant care. Data for the territorial comparisons within the province are not separated by medical and non‐medical benefits and thus are not presented here.

Figure 10 and Figure 11 display provincial trends in medical cost claim frequency and severity, respectively. Figure 10 shows that for much of the study period, the frequency of claims for medical benefits in Ontario was about the same as that in other provinces. Again, this is problematic given the lower accident rates in Ontario compared with other provinces. Claim frequency increases through 1994, declines in 1995–1997, increases again from 1998–2002, begins to decline in 2003 then declines sharply in 2004–2008. In 2009 claim frequency again begins to rise. Notably, mandatory first‐party accident benefit coverage was in place over the entire time frame and as such, absent moral hazard, there should be no relationship between claim frequency and specifics of the auto insurance product.

Figure 10 – Number of Medical Benefits Claims per 100 Vehicles – Provincial Comparisons

Medical Benefits Claim Frequency 1.20

1.00

0.80

0.60

0.40

0.20

0.00

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data 83

Figure 11 shows trends in average medical coverage claim severity over time. The data show the costs of these claims in Ontario are much higher than other provinces, and that Ontario has experienced dramatic increases in costs over time both relative to general inflation and relative to other provinces. It is reasonable, however, to expect that Ontario has higher claims severity than the tort provinces, both because first‐party coverage limits are higher in Ontario but also, if the verbal threshold is effective, fewer claims will be settled under the tort mechanism and injured parties will be compensated primarily through first‐party accident benefits.

Figure 11 – Average Medical and Rehabilitation Expenses Costs per Claim in Constant 2010 $ – Provincial Comparisons

Medical Benefits Claim Severity in 2010 $ $70,000

$60,000

$50,000

$40,000

$30,000

$20,000

$10,000

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

The effects of auto insurance reforms are nonetheless evident in the data. For example, there is a notable increase in claim severity in the 1993–1996 period, with a marked decline in 1997. Since Bill 164 in 1994 enhanced accident benefit coverage and eliminated the right to sue for economic loss, claim severity was expected to grow. In November 1996, Bill 59 introduced a two‐tiered schedule of accident benefit payments, and fewer claimants had access to the more generous benefits. 84

Beginning in 1997, average claim severity trended gradually upward until 2003 when the rate of increase relative to CPI became more rapid. Between 2003 and 2009 average medical coverage claim severity increased 33 percent in real terms. However, there is no inflection point observed after 2006, contrary to the suggestion that elimination of the DACs led to higher claim costs. The reforms of 2003 were associated with reducing the assessment costs and tightening the verbal threshold. The rapid growth in claim severity would indicate that these reforms were largely unsuccessful. In 2010 average claim severity declined in real terms. Thus preliminary results indicate that lowering benefits available to those that are not catastrophically injured had a real impact on claims costs.

Figure 12 and Figure 13 display the between‐province trends for income replacement, attendant care, and other non‐medical benefits. The frequency of non‐medical benefits claims declines markedly over time, from 0.14 per 100 vehicles in 1991 to 0.08 per 100 vehicles in 2010. Comparing the trends in

Figure 12 to those in Figure 7A for property damage claims reveals further that non‐medical benefits claims per property damage claim decreased by nearly 43 percent over the period. This suggests a real decline in these claims, over and above the decline in accidents over this period.

Moreover, the patterns over time suggest that declines occurred in response to insurance reforms. Non‐medical benefit claims declined from 1991–1993, spiked up in 1994 and then declined sharply until 1996. This growth in 1994 is consistent with the introduction of weekly benefits to a larger group of insureds (unemployed, caregivers, and students). The reduction in 1996 coincides with the introduction of the two‐tiered benefit schedule and the requirement that the claimant must be catastrophically injured to receive future care costs. The frequency of claims increased gradually after

1996 until 2003, when another steep decline was observed. Although those not catastrophically injured could claim future care costs after 2003, other initiatives aimed at reducing moral hazard associated with

accident benefit claims appear to have been successful. Non‐medical benefits claim frequency has been relatively constant since 2004.

Figure 12 – Number of Non‐Medical Benefits Claims per 100 Vehicles – Provincial Comparisons

Non-Medical Benefits Claim Frequency 0.70

0.60

0.50

0.40

0.30

0.20

0.10

0.00

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Figure 13 – Average Non‐Medical Benefits Costs per Claim in Constant 2010 $ – Provincial Comparisons

Non-Medical Benefits Claim Severity in 2010 $ $50,000

$40,000

$30,000

$20,000

$10,000

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Average claim severity for non‐medical benefits claims, seen in Figure 13, also appears to be sensitive to the auto insurance reforms. Claim severity increased gradually from 1991–1993 and at a more rapid pace from 1993–1996. In 1997, after the introduction of the two‐tiered schedule of payments in Bill 59, average claim severity fell in real terms back to about 1991 levels. However, inflation in claim severity outpaced general CPI increases each year, so that claim severity rose in real terms in every year since then. The Ontario Auto Insurance Anti‐Fraud Task Force (2011), notes that the biggest increases from 2006 to 2010 in these non‐medical benefits costs per insured vehicle have been in housekeeping awards (an increase of 178 percent) and attendant care (increase of 67 percent). Overall, during the period 1991–2010 average non‐medical claim severity increased nearly 110 percent in real terms.

We now examine the frequency of accident benefits claims in Figure 14 and the severity of claims in Figure 15 for the four largest urban areas compared to the rest of Ontario. As with the earlier analyses,

Figure 14 also shows that the claiming behaviour in the GTA is markedly different than the rest of the province. The frequency of accident benefit claims in the GTA is almost double that of the rest of the province. According to the Ontario Ministry of Transportation (2009), the number of people injured in accidents per 100 vehicles registered is 1.58 in the GTA compared to 0.78 in Ottawa and 1.87 in Waterloo and Wellington regions. However, the rate of accident benefit claims per 100 vehicles for both Ottawa and Waterloo‐Wellington is substantially lower than the number of individuals who are injured. In the

GTA, the rate is higher. The fatality rate is also lower in the GTA than in either Ottawa or Waterloo‐

Wellington.22 Thus again, the higher claims in the GTA do not seem to be associated with actual road conditions or accident rates. Therefore it is not surprising that reforms introduced in 2003 to reduce excessive claiming behaviour had more of an impact in the GTA than the rest of the province.

22 The fatality rate is often used as a measure of overall accident severity in a jurisdiction because there is no moral hazard associated in the claiming behaviour for fatalities. 87

Figure 14 – Ontario Comparison Accident Benefits Frequency

Accident Benefits Claim Frequency 3.00 2.50 2.00 1.50 1.00 0.50 -

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

Figure 15 graphs the average AB severity per claim for the four largest urban areas in Ontario and the remainder of the province. All territories exhibit an upward growth, even though overall inflation levels have been removed from the data. This is an indication that reforms have not been successful in containing first‐party benefit costs at the individual claims level. Before 2006, the levels of losses were fairly comparable across all regions (with Ottawa having the consistently lowest costs). However, after

2006 claims costs grew dramatically across the province, indicating that the DACs were serving a valuable role with respect to cost containment. The growth in the GTA is particularly explosive with accident benefit claims growing at an annual rate of 23 percent. Given that we expect less severe accidents in the

GTA and that there is no external indication that the severity of accidents was growing at this rate, this would indicate that either there is excessive claiming behaviour or outright fraudulent behaviour in the

GTA.

Figure 15 – Ontario Comparison Accident Benefits Costs per Claim in Constant 2010 $

Accident Benefits Severity $70,000 $60,000 $50,000 $40,000 $30,000 $20,000 $10,000 $0

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

5.4 System Performance

As we noted earlier, the issues of affordability and availability are crucial when auto insurance is compulsory. For this reason, auto insurance often becomes a politically charged issue. The criteria of both fairness and informed customers, regardless of the underlying auto insurance mechanism adopted, are also important criteria that are often overlooked, especially when there are more severe problems with the product.

Figure 16A shows the average premium per province for Ontario, Alberta, New Brunswick, and

Nova Scotia for 1991 to 2010 and Figure 16B shows the average premiums across the different territories in Ontario. Similar figures have been calculated by consumer groups and the public press to promote a

discussion about the auto insurance premiums across the country and within the province.23 These figures show that across the private market provinces drivers in Ontario, and especially drivers in the

GTA, pay the highest premiums. In the absence of unbiased information on costs, drivers may assume that high premiums are being driven by insurers earning higher profits. Without knowledge of the underlying claims costs, drivers in Ontario, and especially drivers in the GTA, believe that they are paying too much for auto insurance.

Figure 16A – Average Premium All Coverages in Constant 2010 $ – Provincial Comparisons

Average Premium in Constant 2010 $ $1,600 $1,400 $1,200 $1,000 $800 $600 $400 $200 $0

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

23 See, for example Arbitrage Magazine at http://www.arbitragemagazine.com/general/paying‐higher‐auto‐ insurance‐premiums/; the website Autos.Ca.Msn.Com at http://autos.ca.msn.com/editors‐picks/canadas‐highest‐ auto‐insurance‐rates; and the Consumers’ Association of Canada at http://www.consumer.ca/index.php4?id=1696. 90

Figure 16B – Ontario Comparison Average Premium All Coverages in Constant 2010 $

Total Premium per Vehicle $2,500

$2,000

$1,500

$1,000

$500

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

To determine whether premiums are too high requires comparing premiums to losses, however.

Figure 17A shows loss ratios, which are calculated as total incurred losses divided by total earned premiums, for Ontario, Alberta, New Brunswick, and Nova Scotia for 1991 to 2010. The figure demonstrates clearly that drivers in Ontario are not paying higher premiums so that insurers can earn greater profits. Although Ontario drivers pay the highest premiums, insurers earn the least amount of profit from this book of business.

The loss ratios for Ontario are too high to be sustainable: insurers cannot continue to offer the product at current pricing levels given the level of benefits paid out. Insurers, being private companies, must earn a fair rate of return or they will have to withdraw from the market. Figure 16A provides further insight on reforms in the other private market provinces. Reforms enacted in other provinces have generally been successful. Other provinces have experienced declining or relatively flat premiums (with respect to overall inflation) since 2003 and yet as shown in Figure 17A, insurers have also experienced

large declines in loss ratios. Reforms in other parts of the country have addressed affordability concerns without threatening the sustainability of insurers.

Figure 17A – Auto Loss Ratio by Province

Industry Auto Loss Ratio in Percent by Province 120

100

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Fairness, where consumers benefit from fair treatment both in the pricing of the insurance product and in the claims settlement process, is also a goal of a healthy insurance system. Because the product is different in each province, it is difficult to evaluate fairness in pricing across the country. It is more straightforward to consider fairness within the same jurisdiction. Figure 17B shows loss ratios for

Ontario only and provides a basis for discussing both sustainability and fairness. Most evident is that there are issues in terms of sustainability. Insurers cannot provide the rate of return required by their shareholders when loss ratios are in excess of 100 percent. Thus, as was illustrated in Figure 17A, there are province‐wide sustainability concerns.

The auto policy, and the benefit provisions provided, is the same in all territories across Ontario.

Thus fairness would imply that drivers pay the same for insurance, relative to their expected losses,

across the province. Figure 17B illustrates that this is not the case, especially in more recent years. From

2000 until 2005, lower loss ratios in Ottawa clearly indicate that this territory was in some part subsidizing the rest of the province. Since 2005, the entire province has been subsidizing the GTA. The growth in premiums in the GTA has not even come close to covering costs within this territory. This is clearly not vertically fair for drivers living outside of the GTA. As well, a fairness concern also arises due to low‐risk drivers in a high loss territory having to pay higher premiums due to where they live. Across territories, this pricing structure is not sustainable. Because of rate regulation, even in the presence of a take‐all‐comers rule, insurers will have an incentive to minimize their presence in the GTA in order to compete within the more profitable territories. This will create an availability issue in the GTA.

Figure 17B – Ontario Comparison of Auto Loss Ratios

Industry Auto Loss Ratio

120 100 80 60 40 20 0

Drivr Age GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

Figure 18A shows the combined impact of the trends in frequency and severity of Ontario auto insurance claims on loss costs per vehicle in constant dollar terms and compared to other provinces. The data show that Ontario auto insurance costs in 1991 were comparable to costs in other provinces (lower

than Alberta, 11 percent higher than New Brunswick, and 22 percent higher than Nova Scotia). By 2010 average auto insurance loss per vehicle was dramatically higher than in other provinces. In addition,

Ontario is the only province in which loss costs had increased in real terms. In 2010 Ontario’s average loss cost per car was 173 percent of Alberta’s average loss cost, 270 percent of New Brunswick’s, and 290 percent of Nova Scotia’s.

Figure 18A – Total Losses per Insured Vehicle in Constant 2010 $ – Provincial Comparisons

Losses per Vehicle in Constant 2010 $ $1,600 $1,400 $1,200 $1,000 $800 $600 $400 $200 $0

Ontario Alberta New Brunswick Nova Scotia

Source: Authors’ calculations from GISA data

Figure 18B graphs the total losses per vehicle for the different territories in Ontario. Except for the GTA, the results are encouraging: claims costs have been largely contained, and there has been little real growth in losses per vehicle since 2000. The story in the GTA is different. Losses fell sharply with the

2003 reforms, and since then have been increasing at a rate of roughly 12 percent a year.

Figure 18B – Ontario Comparison – Total Losses per Insured Vehicle in Constant 2010 $

Total Loss Per Vehicle $2,000 $1,800 $1,600 $1,400 $1,200 $1,000 $800 $600 $400 $200 $0

GTA Halton/Hamilton Ottawa KW Guelph Cambridge Rest of ON

Source: Authors’ calculations from GISA data

5.5 Summary and Interpretation

Ontario’s private market auto insurance system offers generous first‐party benefits and the right to sue for injuries sustained in an automobile accident when injuries exceed the verbal or monetary threshold. Insurance is mandatory for all drivers, and therefore consumers need to have access to an affordable product that suits their needs. Insurers must be able to provide the auto insurance product in a manner that is fair and efficient to consumers, and must sell insurance at prices that are viable in the long run. However, the analysis presented above and in Section 4 clearly indicates that these objectives for a well‐functioning insurance market are not being met in Ontario.

Even with a take‐all‐comers rule in place, we find some evidence in Table 2 and Table 3 that there are availability concerns. Given the mismatch between losses incurred and premiums charged as evidenced in the fairness measures in Table 2, and the graph of loss ratios (Figure 17B), we anticipate that these availability concerns, especially in the GTA, will increase in the future. Service quality is also an

issue, as evidenced in Table 3. Additionally, Insurance Bureau of Canada notes that currently there are

30,000 unresolved claims in Ontario awaiting dispute resolution services.24

A 2012 survey by InsurEye Inc., however, found that drivers in Quebec, Saskatchewan, and

Ontario were the most positive about the quality of auto insurance in their provinces. In spite of the high cost, Ontarians were in general happy with their claims experience. Drivers in British Columbia and

Manitoba had the lowest consumer satisfaction with auto insurance, and in particular drivers were unhappy with their limited choices for auto insurance.25

Auto insurance reforms over the past two decades in Ontario have been aimed at reducing the cost of insurance. But now the average premium in Ontario is over $1400, and as can be seen in both

Figure 3 and Figure 16, past reforms have been ineffective at reducing premiums. Reform policies in

Ontario appeared to affect outcomes in the short run, but in many cases inflationary trends quickly reappeared, especially in the GTA. Two exceptions to this are in bodily injury liability claim severity and disability income claim frequency. Trends in these cost elements appear to have been affected over the longer term by the auto insurance reforms.

The high premium cost in the GTA is particularly problematic; however it is important to emphasize that premiums are not being driven by insurer profits, as can be seen by the high loss ratios in

Figure 17B. There are many insurers operating competitively in Ontario. High premiums stem from the fact that Ontario claim frequency and severity are higher than in other provinces in almost all cases and all years, and especially across the GTA. The higher claim severity may be due to higher cost‐of‐living in

24http://www.canadianunderwriter.ca/news/claims‐costs‐for‐ontario‐auto‐insurers‐still‐out‐of‐control‐ ibc/1001413525/ 25http://www.canadianunderwriter.ca/news/quebec‐drivers‐most‐satisfied‐with‐auto‐insurance‐b‐c‐drivers‐the‐ least‐survey/1001353834/

the GTA and more generous benefits in the Ontario auto product, but this cannot be determined with certainty from the comparisons here. Moreover, the increases over time in both claim frequency and claim severity have been substantially greater in Ontario than in other provinces. An equally problematic observation is that compared with the other provinces, Ontario has a lower frequency of reported collisions, making the higher frequencies for first and third party bodily injury and property damage claims difficult to reconcile. These underlying claims pressures threaten the entire provincial auto insurance market.

A further insight obtained from this analysis is that the auto insurance product is sustainable across most of Ontario. Total losses per vehicle have been relatively flat across all parts of Ontario (except for the GTA). It is the GTA that is creating sustainability issues and the GTA is being subsidized by drivers in the rest of the province. The impact of decades of prior approval rate regulation has led to a supposedly risk‐based pricing system that no longer adequately separates insureds by their accident costs. To ensure affordability, changes to differentials for rating variables have been capped increasing cross‐subsidization across different risk profiles of insureds.

Most concerning is that there have been no successful cost containment measures within the

GTA from any of the previous product reforms. This contrasts with the experiences in the tort provinces, where there have been sustainable reductions in the frequency and severity of bodily injury liability claims, and therefore premiums, since enacting reforms. These trends highlight the reality that regulatory reforms may address today’s most easily observable problems while masking larger systemic issues. Some current proposals for solving Ontario’s auto insurance problems are susceptible to this criticism. In 2012, three bills were introduced by two GTA members of parliament in the Ontario legislature: Bill 43

“Insurance Amendment Act (Elements in Classifying Risks for Automobile Insurance)”; Bill 45 “Insurance

Amendment Act (Risk Classification Systems for Automobile Insurance)”; and Bill 71 “New Drivers'

Insurance Rate Reduction Act.” The aim of all these bills was to reduce the price of insurance in the GTA by removing various risk‐based rating variables such as territory, age, and gender of driver. The discussion of pricing mechanisms in Section 3.4 of this report, and the evidence presented here regarding higher claims costs in the GTA relative to the rest of Ontario, demonstrate that these proposals address only the symptoms of the problem and will not improve outcomes for the auto insurance system.

More thoughtful reforms are needed. Gambrill (2009) sums up the problems succinctly: “The system is too complex, nobody understands or wants to fill out the paperwork, there are too many health professionals’ assessments for minor injuries, most of the insurers’ money is propping up the system

(including cottage industries for assessment) rather than going to the claimants, the system disproportionately caters to minor injury claimants, etc. etc.” (p. 26).

6 Experience from Other Jurisdictions

As noted previously, different jurisdictions choose varying combinations of auto insurance design features which leads to important differences across auto insurance systems. Different jurisdictions also place different weights on the various objectives of automobile insurance. Nonetheless, all auto insurance systems pursue the same basic objectives and adjust the same set of decision variables to determine the final system design. Moreover, the effects of auto insurance design on the incentives and behaviours of participants is an issue that must be dealt with in all systems. Considering the experiences of other jurisdictions may therefore yield insights into beneficial or problematic system design features, and lessons regarding solutions to auto insurance problems. This section of the report describes the characteristics of auto insurance systems in a range of comparison countries, and summarizes lessons obtained from more detailed analysis of several of these auto insurance systems. The case studies of the selected systems are presented in the Appendix.

6.1 Insurance Systems in Other Countries

In the U.S., much like in Canada, auto insurance systems are determined at the state level. All states have privately administered systems, and the majority of states employ tort‐based compensation and optional first‐party injury insurance (with compulsory minimum liability insurance requirements). A few states employ partial no‐fault systems in which first‐party injury insurance is mandatory, but benefit limits are typically low. Michigan is the only no‐fault state with compulsory first‐party insurance which provides for unlimited medical benefits. Michigan is also unique among the states in providing for automobile accident property losses on a no‐fault basis. Several states allow drivers to choose between no‐fault and tort‐based compensation, and others allow optional purchase of first‐party injury benefits.

In most states, auto insurance pricing is largely risk‐based but about one‐third of states actively regulate rates. A few states have statutes or regulations that require rate differences across groups to be capped or leveled, or that restrict rating territories; but the restrictions are generally modest. A larger number of states place restrictions on the use of age and gender as rating factors (Weiss, Tennyson and

Regan, 2010). Historically, several U.S. states have experimented with regulated insurance pricing mechanisms that provide significant premium subsidies to some consumer groups. However, these systems proved unsustainable and the social pricing mechanisms were largely abandoned (Bartlett, Klein and Russell, 1999; Derrig and Tennyson, 2011).

Auto insurance systems in Europe are also privately administered tort‐based systems. Liability insurance for both bodily injury and property damage is compulsory, but first‐party injury insurance is optional. While the European Union (EU) single market initiatives have led to some standardization of systems across countries, allowable damages and injury compensation vary in accord with national tort laws (CEA, 2010). Minimum liability limits tend to be much higher in Europe than in North America, with compulsory per person injury liability in the range of €2.5 million and property damage liability of

€1 million (CEA, 2007). EU directives mandate that auto insurance pricing is risk‐based and not subject to government regulation, although prior to 1994 auto insurance prices were extensively regulated and often set by government or an industry association (CEA, 2010). The transition to market‐based pricing was unevenly paced across countries but by 2005 all EU members had deregulated auto insurance rates.

Like Canada and the U.S., Australia’s auto insurance systems vary by state. Third‐party liability insurance for bodily injury is mandatory in all provinces. Compulsory third‐party liability insurance is provided through a government monopoly in several states, including Victoria, South Australia, and

Western Australia. In these states third‐party liability insurance is funded through fees levied on vehicle registrations, and benefits are determined by the state. In South Australia and Western Australia

100

compensation is available through the government insurance fund only for not‐at‐fault drivers. Victoria’s government fund administers a no‐fault system. The fund compensates all injured parties for medical and rehabilitation expenses. Private lawsuits (against at‐fault drivers) for pain and suffering or loss of earnings are permitted in cases of serious injury, but damages are capped. New South Wales, in contrast, has a privately administered system of compulsory third‐party injury insurance in which claims are compensated on a no‐fault basis, with private lawsuits allowed only in cases of serious injury.26

New Zealand operates a government‐run pure no‐fault system of injury compensation, under which all litigation is barred. The Accident Compensation Corporation (ACC) provides personal injury insurance for all injuries – not just auto‐related injuries – which is funded through vehicle registration fees and gasoline taxes. Benefits cover a broad range of services including medical care and rehabilitation, home care and wage replacement, all subject to government‐determined compensation schedules.27

This brief consideration of auto insurance systems in other countries illustrates the variety of systems in use. However, it also illustrates the commonalities in auto insurance systems across the world.

Brief summaries of the experiences of Alberta and two U.S. states are presented in more detail in the

Appendix, in order to provide insights into problems and solutions that may be relevant for Ontario. The following section summarizes and draws implications from these cases.

6.2 Summary of Case Studies

The case studies in the Appendix recount the experiences of three different jurisdictions with auto insurance cost growth problems: Alberta in Canada, and Massachusetts and Michigan in the U.S.

26 This information was obtained from Australian state insurance regulatory websites. 27 The information was obtained from the website of the ACC. 101

Most generally, these case studies provide confirmatory evidence that cost problems are not unique to

Ontario or its auto insurance system. While all three case studies are based on private insurance systems,

Alberta has tort‐based auto insurance; Massachusetts has partial no‐fault with a low tort threshold and minimal first‐party accident benefits; Michigan has partial no‐fault but with a threshold based on serious permanent injury and unlimited first‐party accident benefits. In each of these jurisdictions, availability and affordability of insurance have been important objectives to decision makers; thus, their experiences with problems and solutions are relevant for Ontario.

Problems of affordability and accessibility in Alberta, especially for young drivers, led to imposition of regulatory restrictions on rates and classification variables. After the 2003 reforms in

Alberta, significant cost reductions were achieved in BI claim severity, in part due to the establishment of a cap on pain and suffering awards for minor injuries. At the same time accident benefits were increased and the government instituted treatment protocols for minor injuries. Despite the increase in accident benefits there was still a decline in total AB + BI costs as shown in Figure 6A, from $573 in 2002 to $318 in

2005, falling further to $254 in 2009. More recently, however, there is evidence that more claimants are not being treated within the set time periods/number of treatments established in the treatment protocols. A further concern is recent evidence that claimants are attempting to have claims fall outside the definition of minor injury. Concerns about cost inflation have returned in recent years.

To protect affordability and consumer access to compulsory insurance, Massachusetts introduced a strong set of social pricing criteria, strict control over insurance rates, and strict mandates regarding insurance provision and payment of claims. While successful in promoting access to insurance, the system worked less well in promoting affordability. High‐risk drivers received large subsidies to their insurance premiums, but the average cost of insurance rose more rapidly than in other states. Over the longer term the system’s emphasis on premium leveling and insurance availability severely diminished private

102

insurance supply, and reduced all participants’ incentives to hold the line on insurance costs. Organized fraud rings arose to take advantage of this situation. Coordinated anti‐fraud efforts and relaxation of rate regulations helped to reverse the adverse trends.

Michigan administers the most ambitious no‐fault auto insurance plan in the U.S., with strict limits on the right to sue for personal injuries and generous lifetime benefits to those injured in automobile accidents. The costs associated with catastrophic injuries (those exceeding $500,000) are spread over all drivers in the state through an annual premium surcharge. The compensation system works well, assuring that all injured persons receive adequate benefits and without the high legal costs and delays associated with the tort system. However, there have been persistent cost problems and residents of urban areas have faced especially high premiums. Experiments with urban rate restrictions were largely unsuccessful, leading to availability problems in those areas (Bartlett, Klein and Russell,

2001). Legislation to reform the system is currently being considered due to unsustainable growth in the costs of compensating catastrophic claims (Tennyson, 2010).

Each of these cases helps to illustrate problems of cost growth in auto insurance systems and the difficulties inherent in maintaining insurance affordability for all drivers. The primary determinant of insurance premiums is the cost of insurance claims. As costs grow, premiums must follow or insurance availability will suffer. In the case of personal injury costs, claim costs will grow with medical cost inflation, better medical treatments, increases in utilization, or cost‐shifting to the auto insurance system.

Cost of income replacement benefits will be increased by wage inflation or increases in work absence due to auto injuries. Increases In non‐economic damages can occur due to changing court standards for awards and eligibility, to claimant efforts to gain access to these damages, or to outright fraud. An important point to recognize is that some cost drivers fall outside of the auto insurance system, and therefore may be controlled only by benefit limits or changes in benefit eligibility.

103

However, another important point is that claiming behaviours are an important cost driver: changing benefits and benefits eligibility may produce changes in claiming behaviours that reduce or undermine the cost‐savings intended by the changes. This appears to be occurring in Alberta, as noted above. Derrig, Weisberg and Chen (1994) document a similar dynamic in Massachusetts after the tort threshold was increased to restrict liability claims eligibility to those with $2000 in medical costs rather than $500. These authors document a dramatic upward shift in the cost of personal injury claims after this change. Where prior to the threshold increase many injury claimants reported just over $500 in medical costs, after the threshold increase the modal medical cost increased to just over $2000. These trends provide evidence of benefit‐seeking behaviour on the part of claimants and/or providers. Simple attempts to “regulate away” cost increases by changing benefits or eligibility are often not successful for these reasons. Adjustments to monitoring and claims administration are also required and in some cases may be more important than product changes.

If the insurance system cannot be organized in a manner that aligns incentives of all participants toward cost reductions, this type of moral hazard in claiming remains an inherent problem. Moreover, claim participants will not often recognize the link between their claiming behavior and the rising costs of insurance. Claimants focus on individual benefits received and ignore the effect of their behaviour on the system as a whole. In the case of medical and other services, claimants may not even be aware of costs due to the distance between services received and payments made. In the aggregate, rising costs are passed on to consumers in the form of higher premiums, but the link between cost increases and premium increases is not transparent. This often leads to pressure to reduce premiums without recognition that the underlying problem is one of cost. However, providing premium subsidies does not lead to cost reductions and may even exacerbate cost increases. Insurer or public policy responses aimed squarely at the underlying cost or incentive issues will be more effective and more permanent than other responses. 104

7 Policy Considerations for a Well‐Functioning Insurance Market

The theoretical analysis of automobile insurance, the evaluation of Ontario’s system, and examination of other automobile insurance systems as given in the Appendix give rise to many policy options for designing an auto insurance system. In this concluding section of the report we first outline these options and highlight strengths and problems with the different system designs. We then outline some alternatives that can be used to mitigate the inherent weaknesses in different auto insurance systems. We wrap up with a brief discussion of how these choices might be applicable to the design of auto insurance in Ontario.

Table 6 provides a summary of the inherent strengths of, and problems with, different policy options for designing an auto insurance system. The policy choices available relate to compensation mechanism, both the level of first‐party accident benefits and the role of liability in compensation; pricing mechanism, social versus risk‐based; and provider, government or private market. We consider the product design first and then discuss the impact of pricing mechanisms and choice of provider on system outcomes.

105

Table 6 – Models of Insurance Systems

Compensation Main Problems Inherent Strengths Impact if there is Cross‐ Impact if Insurance is Provided Mechanism subsidization of Premiums by Government Monopoly Pure tort with no Mismatch of compensation with Safety incentives; Benefits are improved Benefits are access for all and to minimal losses; Market efficiencies accessibility and improved increased claims cost control. accident benefits Potential insufficient affordability. Inherent problems are lack of indemnification; Inherent problems are adverse consumer choice and lack of Extensive use of court system selection, reduced safety competition. (high claims adjustment incentives, and potential negative expenses) impact on claiming behaviour.

Pure tort with Claims cost control; Safety incentives; moderate to Unnecessary use of court system Compensation for all generous accident benefits

Pure no‐fault Moral hazard – both ex‐ante and Compensation for all with generous ex‐post; accident benefits Claims cost control concerns increase

Partial no‐fault Threshold issues; Compensation for all; with moderate Ex‐post moral hazard; Safety incentives (less than to generous Claims cost control pure tort); accident benefits Market efficiencies

Adverse selection (more high‐risk drivers on the road), ex‐ante moral hazard (all drivers take less care) and ex‐post moral hazard (excessive claiming after a loss) are created by all social pricing systems; but problems of ex‐post moral hazard tend to be greater when social pricing is combined with private market systems. The presence of generous first‐party benefits increases both ex‐ante and ex‐post moral hazard even in the presence of tort compensation. Private auto insurance systems with generous benefits and social pricing contain inherently more incentive conflicts than other systems.

106

Beginning at the top, pure tort systems with no to minimal accident benefits (most U.S. states) provide strong safety incentives for safe driving and efficiencies in both products and services associated with private insurance, but are weak in terms of matching compensation to losses and costly in terms of the expense incurred to determine liability and settle claims. The addition of moderate to generous accident benefits to a tort system (Alberta, New Brunswick, Nova Scotia, and Prince Edward Island) improves matching of compensation to losses and safety incentives, but creates a potential challenge in terms of claims cost control given that all parties are entitled to benefits.

The primary advantage of pure no‐fault systems with generous accident benefits (Quebec and

Manitoba) is compensation for all injured parties, but this comes at a cost in terms of an increase in both ex‐ante and ex‐post moral hazard. Insureds take less care on the road and when an accident happens, insureds may exaggerate their claims to receive greater compensation. As well, because of this incentive to exaggerate claims, the more generous are accident benefits, the more important and difficult claims cost control becomes. When claims costs are not controlled, affordability concerns arise.

A partial no‐fault system with moderate to generous accident benefits (Ontario) also achieves compensation for all and promotes safe driving choices, but ex‐post moral hazard and claims cost control remain major concerns. In theory, a partial no‐fault mechanism should reduce costs over a tort mechanism if claims can be settled more quickly outside the tort system and if the tort threshold is not eroded. However, in the Ontario system, the settlement process for first‐party accident benefit claims quickly became very cumbersome and a number of potential problems arose due to defining the tort threshold and establishing which claims fall within the threshold.

The fourth column of Table 6 examines the impact of introducing social pricing or cross‐ subsidization of premiums. Regardless of the compensation mechanism in place, the introduction of social pricing should improve accessibility and affordability for all. However, as noted earlier, cross‐

107

subsidization of premiums creates adverse selection problems, reduces incentives to drive safely and increases ex‐post moral hazard in claiming. Finally, the last column of Table 6 shows that if insurance is provided by a monopolistic government insurer, the greatest advantages are improved accessibility and more centralized tools to control claims costs. This comes at the cost of lack of consumer choice, lack of competition, and potentially lower service quality.

In general, pure tort systems are considered to be more effective in providing safety incentives while pure no‐fault systems with generous accident benefits are more effective in providing compensation. It is important for policy makers to clearly articulate the objectives of the system, recognizing that trade‐offs exist between different design options. Because of differences across the driving population in consumer knowledge of insurance, consumer attitudes towards fairness, and consumer attitudes towards irresponsible claiming behaviours, some design features will be more successful in some jurisdictions than in others.

7.1 Policy Considerations for Ontario

Insurance provides many benefits to individuals and to society. However, as discussed and illustrated here, insurance creates adverse behavioural incentives and as such may also lead to a greater propensity to drive, reduced precautions when driving, and benefit‐seeking or even outright fraud in claiming. Such incentive problems may be particularly prevalent when insurance benefits are generous, and when insurance pricing is socialized so that premiums do not reflect the risk an individual brings to the insurance pool. Well‐functioning insurance systems must recognize negative incentive effects and counteract them where they occur.

The most direct policy responses would be to raise the entry price of driving (counteracting the increased propensity to drive); penalize risk‐taking in driving (counteracting reduced precautions); and provide minimal insurance benefits and implement strictly risk‐based premiums (counteracting benefit‐ 108

seeking in claiming). However, these solutions will reduce access to driving and create insurance affordability problems for some consumers, and may leave many without adequate accident compensation. When insurance availability and affordability to all consumers are important objectives, as is often the case with automobile insurance, adjusting the insurance system to counteract negative behavioural incentives is more difficult. In jurisdictions in which driving is not considered a right or an economic necessity, the public policy issues in auto insurance are less complicated.

Ontario has a private market, partial no‐fault auto insurance system design with a high tort threshold. The benefits available under first‐party accident benefits are a function of the severity of the injuries incurred: most injuries are non‐catastrophic and claimants may claim up to $50,000 in medical payments, but those who are catastrophically injured are eligible for up to $1 million. Generous first‐ party benefits have been enshrined in legislation since the introduction of no‐fault insurance in 1989.

In Ontario and other private insurance markets in Canada, when insurance costs and availability become a concern, public insurance is often promoted as a solution to the problem. Our analysis does not indicate that the existing problems are a result of the private market structure, nor do we see in other private provinces the same problems as in Ontario. Further, government run insurance is not a panacea to cost problems. Government insurance programs face the same problems of moral hazard and compensation cost inflation that occur in private insurance markets, and benefit and premium structures are often adjusted in response. Provinces with government insurance do have advantages in responding to cost problems due to market‐wide control measures, but these conditions could be replicated in private market provinces while maintaining the private market structure that ensures market efficiencies as private insurers compete for customers.

Nonetheless, as is evident from the analysis presented in this report, the Ontario auto insurance system is not healthy – there are affordability concerns, the GIO receives many complaints regarding all

109

aspects of the product, insurers are struggling to make money, and there is a large amount of cross‐ subsidization across insureds, especially between the GTA and the rest of the province. Despite these problems, the evidence does not suggest that the design of the system is fundamentally flawed.

However, there are significant cost concerns associated with the GTA. Here we make recommendations regarding the elements of the system that need to be addressed in order to improve performance as measured by the objectives of a well‐functioning auto insurance system. Our recommendations encompass pricing mechanisms, benefit provisions, claims cost controls, consumer education, and larger societal solutions.

7.1.1 Pricing Mechanism Design

In a compulsory insurance system, the choice to drive, the amount of driving, and precautions taken in driving are affected by the price of insurance. Carefully designed pricing mechanisms support affordability and accessibility, align consumer incentives to support safe driving, and reduce excessive claiming. We discuss how different pricing mechanisms can be used to achieve these policy goals, recognizing that purely risk‐based pricing may not support affordability and that purely social pricing does not support safe driving and responsible claiming objectives.

Within the insurance system, relaxing mandatory insurance requirements for low income drivers is one approach to addressing affordability. Some U.S. jurisdictions have introduced so‐called “dollar‐a‐ day” policies which are low cost but provide limited liability insurance. The limited coverage has the effect of shifting costs onto other drivers who may need to supplement insurance protection with underinsured motorist coverage. If such limited policies are made available, it is important for incentive reasons to base eligibility solely on income. Another approach to reducing insurance premiums for some urban drivers is to allow miles‐based insurance pricing: since urban drivers generally drive fewer miles than other drivers, and since miles driven is a good unit of measure for risk exposure, these policies are

110

a way to offer lower costs to low‐risk drivers in high‐risk areas. This “pay‐as‐you‐drive” auto insurance contract is gaining popularity in Canada, as more insurers are offering these contracts.

Allowing the use of more rating variables, rather than restricting the number of rating variables as has been suggested in recent Ontario legislation to reduce the number of territories, is another way to provide premium reductions for low‐risk drivers in urban areas by allowing them to distinguish themselves from high‐risk drivers. For example, while controversial, insurance scoring in the U.S. has been shown to be an effective classification variable due to its correlation with claims experience.

Insurance scoring is an underwriting and risk classification tool that uses an insured’s credit, motor vehicle records and claims history. Currently, Ontario and the Atlantic provinces do not allow credit history to be used for auto insurance rating and underwriting, while Alberta and Quebec permit the use of credit history for property damage coverage in auto. In all provinces, regulations are silent on the use of credit scoring for homeowner’s insurance, therefore insurers are permitted to use it. Insurers argue, and evidence shows, that credit history, or “financial discipline,” is a good predictor of insurance experience in terms of driving habits, and restricting its use results in less accurate pricing.28

Adjusting insurance premiums based on claims experience or safe driving record allows low‐risk drivers to distinguish themselves from high‐risk drivers, and provides financial incentives for safe driving.

Basing premium adjustments on traffic violations, as well as at‐fault claims experience, provides a stronger signal of driving risk than only using at‐fault claims experience. However, if it is expensive for insurers to capture moving traffic violation data, they will only do so for those drivers that they suspect

28 It is not likely that insurance scores will be allowable for mandatory auto insurance in Ontario in the future because of the concerns that the insurance score itself might be discriminatory. For example, the Ontario Human Rights Commission (1999; p 27) noted that “Any newly proposed risk classification system, even if shown to be a better measure of risk, should at least not further contravene rights under Part 1 of the Code any more than any current classification system does.” The Commission further asserted that certain factors included in an insurance score, namely credit card ownership, bankruptcy status, employment status and stability, and residence status and stability would likely contravene Ontario’s Human Rights Code. 111

are higher risk.29 Thus another policy recommendation is to improve access to provincial driving records for insurers.

Despite the importance of risk‐based pricing, it is important to recognize that making premiums even more experience‐rated may also have its disadvantages. Finger (2006) notes that as premium discrepancies become greater between risk groups, drivers have a greater incentive to manipulate the risk classification variables. With respect to experience rating, we conjecture that determination of fault in an accident would become much more difficult, time consuming, and expensive if premiums were to increase even more after an at‐fault accident. Greater financial penalties would mean that insureds would question the inflexibility of the fault determination rules, thereby increasing the number of claims going to mediation. This concern and the difficulties in fault determination suggest that any strengthening of experience rating should focus on driving record rather than accidents.

In Ontario, the premium charged for new drivers is not distinguished from the experience rating mechanisms for higher‐risk drivers. It is not evident that this practice is fair or reasonable. Several provinces and many companies have developed mechanisms to promote affordability for new drivers, for example the grid in Alberta, risk sharing pools in the Atlantic provinces, and company level pricing models that retroactively reward new drivers for claims free experience. From a policy perspective, the point to emphasize is that mechanisms that promote affordability should not lessen incentives for safe driving since strengthening the responsiveness of changes in insurance premiums to driving or accident experience has been shown to reduce moral hazard and adverse selection without violating social pricing objectives.

29 It is easier and less costly for government insurers to include moving traffic violations in pricing because they have instant and costless access to this information. In private market provinces there may be a cost. For example, in Ontario, the cost of a statement of driving record is $12. 112

Fundamentally, using rate regulation to protect consumers from insurance premiums that adequately reflect underlying costs is counterproductive as this reduces the willingness of insurers to provide the product and it masks the problems that cause cost growth. Regulation of rates may lead to a higher proportion of high‐risk drivers in the insurance pool and has been shown to lead to higher claims rates and loss costs. In addition, artificially keeping insurance prices low reduces the incentives for insurers to police claim cost growth because such monitoring requires costly resources. Instead, a premium subsidy to low‐income (not high‐risk) drivers is a better policy lever to promote affordability.

In Ontario, two decades of prior approval rate regulation have weakened the relationship between premiums and expected losses, and have created significant cross‐subsidies between the GTA and the rest of the province. Evidence from the public market provinces of Quebec, British Columbia,

Saskatchewan, and Manitoba demonstrates that when premiums for higher‐risk drivers are subsidized, the unintended consequence is an increase in the number of high‐risk drivers on the road. Hence an important policy lever to maintain and potentially strengthen in Ontario is the risk‐based pricing mechanism. Although this may require specific measures to achieve affordability as noted above, flattening of the pricing structure is not optimal.

7.1.2 Benefit Provisions

Generous first‐party accident benefits should be part of the auto insurance system design when compensation for all injured parties, regardless of fault, is considered to be an important objective.

However, insurance systems with generous first‐party benefits are expensive and the presence of benefits increases both ex‐ante and ex‐post moral hazard. Thus an obvious solution to reducing costs and aligning incentives is to reduce the level of benefits provided. For example, some U.S. states have achieved affordability by not including first‐party accident benefits in the mandatory auto insurance policy.

113

However, depending on the objectives established this may not be a feasible alternative, and a removal of first‐party benefits is unlikely to occur in Canada. Given that the existence of these benefits is accepted as the norm, it is not politically feasible to remove them, although the level of benefits could be adjusted. In addition, the inclusion of first‐party accident benefits allows insureds to internalize a part of the cost of driving which is particularly important given that all Canadian residents have access to universal healthcare and many have access to supplemental health benefits.

Since generous benefits lead to greater benefit‐seeking behaviours, cost and utilization controls must be a part of the system. Over‐utilization of healthcare has been observed in many insurance contexts. The marginal cost of medical care is more likely to exceed the marginal benefit of additional medical care (improved health outcomes) when there is a separation between the benefit recipient, benefit provider, and who pays for the care as providers and recipients have no intrinsic motivation to minimize costs of claims. When costs of care are not considered, even small gains to a patient’s health or well‐being from more expensive tests, treatments, procedures, or facilities, may lead to decisions in their favour. Moral hazard also increases if insurance provides for generous wage replacement benefits as injured persons are less likely to return to work quickly if they receive generous benefits. If generous benefits are provided, then cost control mechanisms must be enacted to ensure affordability and reduce moral hazard.

7.1.3 Claims Cost Controls

No‐fault systems that provide generous auto insurance compensation benefits create incentives for ex‐post moral hazard and potentially fraudulent claiming. The experience in Ontario, and in particular the GTA, illustrates that generous accident benefits without sufficient controls in place create an environment that can lead to unsustainable claims growth. Thus it is essential to develop mechanisms for controlling claims costs.

114

To control costs, the Ontario no‐fault auto insurance policy has been designed with two hurdles or thresholds: the verbal and monetary thresholds after which the tort system may be accessed and the catastrophic injury definition which provides a higher level of first‐party accident benefits. A key concern with a threshold or injury definition in a policy is that over time these are seen as barriers for claimants to overcome in order to be eligible to receive greater benefits. The experience in Ontario indicates the difficulty of defining a threshold that will maintain its effectiveness over time. From a policy perspective, a few points are critical: clarity in defining the threshold; vigilance on the part of companies and government in enforcing the threshold; and when possible, avoiding the creation of a “lottery” incentive that will promote fraudulent claiming. If there is not a lot to be gained financially from piercing the threshold (i.e., if the claimant is able to access great medical benefits but these are based on indemnity and there is no “money in the pocket”) there is less incentive to attempt to do so.

Costs are controlled in many health insurance policies by transferring some of the treatment costs to the benefit recipient through the introduction of co‐pays or other cost transfer mechanisms.

These policy tools can be used to reduce over‐utilization as long as providers do not help beneficiaries to circumvent controls.

Excessive costs may also arise from the deliberate overuse of benefits or from the filing of fictitious or exaggerated claims, which will be more problematic if claimants receive money directly from the insurer to pay for treatment. When systems place a priority on paying benefits quickly, when there is not strict oversight of the benefit providers and when claims do not face the scrutiny of the legal system, over‐utilization and fraud become especially problematic. Monitors and protocols are needed to detect and remove ineligible beneficiaries, and periodic evaluations are important when benefits are provided over a long period of time. The Auto Insurance Anti‐Fraud Task Force will be making recommendations to the Ontario government in the fall of 2012 and we anticipate that this report will include numerous

115

strategies. One example would be that only licensed and approved medical practitioners are eligible to receive payments from insurance companies. This will reduce the number of exaggerated claims.

The use of treatment guidelines produces many positive outcomes. Well‐designed treatment guidelines ensure best practices in treatment allowing injured individuals to recover faster. If guidelines can be easily implemented, then this should also reduce the cost of handling claims. In addition, guidelines can be written to reduce ex‐post moral hazard – for example, requiring an insured to cover costs upfront if an insured requires medical treatment beyond what is recommended by the protocol.

Reimbursement by the insurer would happen only if the treatment were later deemed to be necessary.30

To address ex‐post moral hazard, insurers must have policies in place to verify the validity of claims and detect potential fraud. We observe, however, that controlling claims costs is more difficult for a private insurance market than for a government insurer since it is harder for individual insurers to effectively implement changes to control claims costs. Each individual insurer’s efforts to control benefit overuse, claim exaggeration, or fraud is generally not sufficient to reach the most socially desired outcome (Picard, 2008). An insurer will invest in these activities up to the point of maximizing profits; the insurer will not take into account the potential spill‐over effects that create a “claiming culture” when overuse, exaggeration, or fraud go undetected and undeterred. Collective action is needed, especially to mitigate organized insurance fraud.

30 In Alberta, for injuries that meet the diagnostic and treatment criteria under the Diagnostic and Treatment Protocols Regulation, auto insurers are priority payers for accident benefits for the first 90 days or up to 21 treatments. Beyond that, if the injured person has extended healthcare benefits elsewhere, he or she must first pay for the treatments incurred up front and the auto accident benefits pay the difference between what is covered under the extended health policy and the treatment costs. 116

Industry‐wide measures to help reduce fraud and control claims costs are required. The 2011

Report of the Office of the Auditor General of Ontario notes that the problem of fraud is worse in

Ontario than elsewhere in Canada. Fraud can arise from individuals or from providers. The more costly problem is due to providers (healthcare professionals, medical suppliers, and lawyers) who defraud the system. In Ontario, an infrastructure for fraud has developed. Clinics, paralegals, and healthcare providers understand how to manage (or manipulate) a claim to maximize the payout. The recent initiatives in Ontario, such as the Health Claims for Auto Insurance (HCAI) database to detect such potentially fraudulent activity, and new rules to ensure that consumers are actually receiving the treatments invoiced by healthcare clinics and other providers, will make it more difficult for providers to defraud the system.

Organized fraud is important to target due to its systematic and repeated nature. Establishing a centralized Insurance Fraud Bureau along with mandatory insurer reporting and criminal sanctions for perpetrators is one approach which has seen success in some U.S. states (Weisberg and Derrig, 1992).

Some jurisdictions such as Louisiana have also seen success from combining criminal justice approaches with media campaigns to educate consumers about the costs that fraud imposes on them, in terms of lower insurance availability and higher premiums. Fraud can also be minimized by requiring those that provide health‐care to auto accident victims to be licensed and regulated. Recommendations forthcoming from the Anti‐Fraud Task Force that can be initiated on an industry‐wide basis will be important levers to control claims costs.

Unfortunately, there is limited evidence that competitive insurance market mechanisms have been successful in the long run containment of claims costs using these control mechanisms. This problem highlights the need to focus policies to minimize incentives for abusive claiming practices, rather than simply restricting benefits to claimants.

117

Even in the absence of excessive claiming, the interplay between government mandated and private provision of medical services also leads to higher costs of auto insurance. Cost shifting takes the form of higher charges for services when provided as part of an auto accident claim than when those services are provided as part of a workers’ compensation claim or a general health claim. Higher per unit service charges will lead to higher claims costs for auto insurance even when services are utilized appropriately. Higher claims costs for auto insurance may also arise if certain treatments are not covered by government health insurance, such as services provided by physiotherapists, chiropractors, and psychologists.

7.1.4 Consumer Education

Insurance is a product that is not well understood by many consumers. Furthermore, the mandatory nature of the automobile insurance product creates a market where uninformed consumers who may not understand the product and see no differentiation across companies will purchase the insurance based only on lowest price without understanding the service and benefits provided. Having consumers that are making informed decisions about their auto insurance and are satisfied with their products and services will help to alleviate some of the current problems in the Ontario auto insurance system.

As noted earlier, in an auto insurance consumer satisfaction survey conducted by InsurEye Inc.,

Ontario ranked third amongst the ten provinces in terms of customer satisfaction. Ontarians were generally satisfied with their claims experience but unhappy with the value for money they receive for their auto insurance. The fact that, overall, people believe they do not receive adequate value for what they pay raises two important issues. First, there may be a lack of understanding of what benefits are provided by the mandatory product, which contributes to the perception that insurance is overpriced.

Consumers need accurate and timely information in terms of what is covered, options available, and 118

what to do when there is a claim. Second, individuals who pay what they perceive as too much for auto insurance are more likely to attempt to “recover” premiums at the time of making a claim. The incentive for recovery is exacerbated if premiums are not risk‐based. The more informed consumers are about the insurance they purchase, the less likely they will have unrealistic expectations that are not met.

The Insurance Information Institute lists the following four criteria for developing an effective and successful insurance education program.31 A successful education program needs to:

 Be inclusive with all participants working towards a common goal.  Be emotional and personal, encouraging individuals to become involved and take action.  Provide tools for individuals to make success possible.  Be long term and responsive to changes over time.

The importance of a continual and responsive education program has been echoed by Insurance

Bureau of Canada (O’Reilly, 2008). Insurance Bureau of Canada’s Restoring Consumer Confidence campaign increased awareness of injury prevention and road safety, and contributed to immediate but short lasting increases in consumer confidence. However, consumers have little trust in the insurance industry and have incomplete knowledge of insurance, which necessitates a continual and adapting education program.

In Ontario if an injured party and their insurance company disagree about the claimant’s entitlement to or amount of accident benefits, the claimant may apply for mediation. Mediation is a mandatory first step before going to court or arbitration. The 2011 Report of the Office of the Auditor

General of Ontario indicates that there has been a large increase in the number of applications for mediation, increasing by 135 percent over the last five years. Currently about half of all injury claims are

31 http://www.iii.org/national‐roundtable‐on‐insurance‐literacy‐best‐insurance‐education‐practices/ 119

in mediation. In addition, resolutions are taking 10 to 12 months to resolve rather than the legislated 60 days. The majority of mediation applications (80 percent) are from the GTA even though it accounts for just 45 percent of automobile accidents involving injury.

The cause of the increase in the number of applications for mediation is unknown, but it is reasonable to expect that if consumers are not well‐informed and do not have access to good information about their insurance coverage, this will contribute to the number of disputes. In addition,

Cameron (2012) explains that some of the increase in mediation applications is due to a FSCO bulletin

(FSCO Bulletin A 02/11, issued 22 March 2011) which directs auto insurers to more proactively challenge questionable claims.

7.1.5 Larger Policy Issues

As noted at the beginning of this report, automobiles and driving have a fundamental role in our society. Auto insurance exists to provide indemnification for road users who experience losses in auto accidents. Although beyond the scope of this research, there are other policy options available that will improve road safety and decrease reliance on the automobile. Both these measures  a decrease in auto accidents and reduction in dependence on the automobile  will contribute indirectly to the long run viability of auto insurance.

Because auto insurance is compulsory, the cost of insurance has become a mechanism that helps to keep high‐risk drivers off the road. However, the impact of this mechanism is uneven on the rich and the poor: unsafe but wealthy drivers can still afford the insurance, low‐income drivers may not.

Social equity indicates that other mechanisms, such as more effective driver training and licensing, are required. Graduated licensing of new drivers has been shown to be an effective mechanism for saving lives and reducing injuries. For example, the introduction of graduated licensing has resulted in a decline

120

of crash rates among sixteen‐year‐olds by 25 percent in Michigan and by 60 percent in Ohio (Sleet,

Schieber and Gilchrist, 2003). Similarly, the introduction of graduated delicensing of elderly drivers may also promote road safety.

In addition to comprehensive graduated driver licensing programs, Kissinger (2009) notes the following five effective regulatory and legislative solutions for improving traffic safety: installation of electronic stability control in cars; enforcement of seatbelt and helmet laws; reduction and monitoring of speed limits on roads; road redesign including roundabouts, rumble strips, and cable median barriers; and a reduction in impaired driving. Electronic stability controls are conjectured to decrease deaths in single vehicle accidents by up to 56 percent. A reduction in impaired driving can be implemented by changes in car design, specifically the installation of alcohol ignition interlocks. When such safety enhancements are made, all drivers benefit, and safer roads should lead to lower insurance premiums.

Finally, the need to drive to support economic stability may not be necessary in dense urban areas. Where other modes of transportation are practical, policy makers should consider solutions that lie outside of the insurance system – for example, developing better public transportation networks and encouraging bicycling or walking. Promoting economic development zones which cluster commercial and residential areas in closer proximity, or providing transportation subsidies to vulnerable groups based on income and location, are other possibilities.

121

8 References

Anderson, James M., Paul Heaton, and Stephen J. Carroll. “The U.S. Experience with No‐Fault Automobile Insurance: A Retrospective.” Rand Monographs. Santa Monica, CA: Rand Institute for Civil Justice, 2010. Baetz, M. “Auto Insurance in Ontario.” Laurier Institute for Business and Economic Studies Case Study G‐ 92‐130‐01, Wilfrid Laurier University, 1993. Baker, Tom. “Constructing the Insurance Relationship: Sales Stories, Claims Stories, and Insurance Contract Damages.” Texas Law Review 72 (1994): 1395. Baltensperger, Ernst, Peter Buomberger, Alessandro A. Iuppa, Benno Keller, and Arno Wicki. “Regulation and Intervention in the Insurance Industry – Fundamental Issues.” The Geneva Reports Risk and Insurance Research. Geneva, Switzerland: Geneva Association, 2008. Barr, Michael S., Sendhil Mullainathan, and Eldar Shafir. “Behaviorally Informed Financial Services Regulation.” Washington, DC: New America Foundation, 2008. ———. “Behaviorally Informed Home Mortgage Regulation.” Washington, DC: New America Foundation, 2008. Bartlett, Dwight K. III, Robert W. Klein and David T. Russell. “Attempts to Socialize Costs in Voluntary Insurance Markets: The Historical Record.” Journal of Insurance Regulation, 17 (1999). BC Utilities Commission. Revenue Requirements for Universal Compulsory Automobile Insurance for the Policy Year Commencing Effective February 1, 2012. Vancouver, 2012. Besley, Timothy, Miguel Gouveia, and Jacques Dreze. “Alternative Systems of Health Care Provision.” Economic Policy 9, no. 19 (1994): 199–258. Blackmon, B. Glenn, Jr., and Richard Zeckhauser. “Mispriced Equity: Regulated Rates for Auto Insurance in Massachusetts.” The American Economic Review 81, no. 2 (1991): 65–69. Boyer, Marcel and Georges Dionne. “Description and Analysis of the Quebec Automobile Insurance Plan.”Canadian Public Policy/Analyse de Politiques 13, no. 2 (1987): 181‐195. Boyer, Marcel, and Georges Dionne. “An Empirical Analysis of Moral Hazard and Experience Rating.” The Review of Economics and Statistics 71, no. 1 (1989): 128–134. Browne, Mark J., and Robert Puelz. “Statutory Rules, Attorney Involvement, and Automobile Liability Claims.” The Journal of Risk and Insurance 63, no. 1 (1996): 77–94. Camerer, Colin, Samuel Issacharoff, George Loewenstein, Ted O’Donoghue, and Matthew Rabin. “Regulation for Conservatives: Behavioral Economics and the Case for ‘Asymmetric Paternalism.’” University of Pennsylvania Law Review 151, no. 3 (2003): 1211–1254. Cameron, James I. “The Evolution of Auto Reforms.” Canadian Underwriter 78, no. 5 (2011): 58, 61. ———. “When the Dam Bursts . . . ….” Canadian Underwriter 79, no. 5 (2012): 12–14. Cole, Cassandra R., Kevin Eastman, David A. MacPherson, Patrick F. Maroney, and Kathleen A. McCullough. “The Impact of No‐Fault Legislation on Automobile Insurance.” North American Actuarial Journal, forthcoming.

122

Comité Européen des Assurance (CEA). The European Motor Insurance Market. CEA Statistics no. 32, 2007. Available online at http://www.insuranceeurope.eu/uploads/Modules/Publications/1201184486_the‐european‐ motor‐insurance‐market.pdf ———.The European Motor Insurance Market. CEA Statistics no. 38, 2010. Available online at http://www.insuranceeurope.eu/uploads/Modules/Publications/1283505667_cea‐ motor_web.pdf Crocker, Keith J., and Arthur Snow. “The Efficiency Effects of Categorical Discrimination in the Insurance Industry.” Journal of Political Economy 94, no. 2 (1986): 321–344. Cummins, J. David, and Sharon Tennyson. “Controlling Automobile Insurance Costs.” Journal of Economic Perspectives 6, no. 2 (1992): 95–115. Cummins, J. David, and Sharon Tennyson. “Moral Hazard in Insurance Claiming: Evidence from Automobile Insurance.” Journal of Risk and Uncertainty 12, no. 1 (1996): 29‐50. Cummins, J. David, Richard D. Phillips, and Mary A. Weiss. “The Incentive Effects of No‐Fault Automobile Insurance.” Journal of Law and Economics 44, no. 2 (2001): 427–464. Ćurak, M., S. Lončar and K. Poposki. “Insurance Sector Development and Economic Growth in Transition Countries.” International Research Journal of Finance and Economics 34 (2009): 29‐41. Dellinger, Ann M., Marcie‐jo Kresnow, Dionne D. White, and Meena Sehgal. “Risk to Self Versus Risk to Others: How Do Older Drivers Compare to Others on the Road?” American Journal of Preventive Medicine 26, no. 3 (2004): 217–221. Derrig, Richard A. “Price Regulation in US Automobile Insurance: A Case Study of Massachusetts Private Passenger Automobile Insurance 1978‐1990.” The Geneva Papers on Risk and Insurance, 18 (1993): 158‐173. Derrig, Richard A., and Sharon Tennyson. “The Impact of Rate Regulation on Claims: Evidence from Massachusetts Automobile Insurance.” Risk Management and Insurance Review 14, no. 2 (2011): 173-200. Derrig, Richard A., Herbert I. Weisberg, and Xiu Chen. “Behavioral Factors and Lotteries under No‐Fault with a Monetary Threshold: A Study of Massachusetts Automobile Claims.” The Journal of Risk and Insurance 61, no. 2 (1994): 245–275. Devlin, Rose Anne. “Automobile Insurance in Ontario: Public Policy and Private Interests.” Canadian Public Policy 19 (1993): 298–310. ———. “Liability versus No‐Fault Automobile Insurance Regimes: An Analysis of the Experience in Quebec.” In Contributions to Insurance Economics, edited by Georges Dionne, 499–520. Norwell, MA: Kluwer Academic Publishers, 1992. Dionne, Georges. “Insurance Regulation in Other Industrial Countries.” In Deregulating Property‐ Liability Insurance: Restoring Competition and Increasing Market Efficiency, edited by J. David Cummins, 361‐396. Washington, DC: AEI‐Brookings Joint Center for Regulatory Studies, 2002. The Dominion of Canada General Insurance Company. “2012 Annual Review of Automobile Insurance Premiums for Basic Coverage.” Submission to the Alberta Automobile Insurance Rate Board, May 31, 2012.

123

Facility Association. “Annual Review of Automobile Insurance Premiums for Basic Coverage.” Submission to the Alberta Automobile Insurance Rate Board, June 12, 2012. ———. “Considerations for Residual Market Regulation,” November 24, 2003. Financial Services Commission of Ontario. “Insurer Rights and Responsibilities to Challenge Questionable or Abusive Claims.” Bulletin A‐02/11, March 22, 2011. ———. “Technical Notes for Automobile Insurance Rates and Risk Classifications Filings.” March 2011. Finger, Robert J. “Risk Classification.” In Foundations of Casualty Actuarial Science, Chapter 6. Arlington, VA: Casualty Actuarial Society, 2006. Fronsko, Andrew. “Road Safety and Insurance Markets Overview.” Discussion Paper no. 2011‐25. Submission to the International Transport Forum, September 2011. Gambrill, David. “Ontario’s Ailing Auto Insurance.” Canadian Underwriter 76, no. 5 (2009): 24–33. Grabowski, Henry, W.Kip Viscusi, and William N. Evans. “Price‐Availability Tradeoffs of Automobile Insurance Regulation.” Journal of Risk and Insurance 56 (1989): 275‐299. Grace, Martin F., Robert W. Klein, and Richard D. Phillips. “Auto Insurance Reform: Salvation in South Carolina.” In Deregulating Property‐Liability Insurance: Restoring Competition and Increasing Market Efficiency, edited by J. David Cummins, 148–194. Washington, DC: AEI‐Brookings Joint Center for Regulatory Studies, 2002. Harrington, Scott E. “Effects Of Prior Approval Rate Regulation of Auto Insurance.” In Deregulating Property‐Liability Insurance: Restoring Competition and Increasing Market Efficiency, edited by J. David Cummins, 285‐314. Washington, DC: AEI‐Brookings Joint Center for Regulatory Studies, 2002.

Harrington, Scott E. “Auto Insurance in Michigan: Regulation, No‐Fault, and Affordability.” Journal of Insurance Regulation 10, no. 2 (1991): 144–183. ———. “The Relationship between Voluntary and Involuntary Market Rates and Rate Regulation in Automobile Insurance.” The Journal of Risk and Insurance 57, no. 1 (1990): 9–27. Harrington, Scott E., and Helen I. Doerpinghaus. “The Economics and Politics of Automobile Insurance Rate Classification.” The Journal of Risk and Insurance 60, no. 1 (1993): 59–84. Hoyt, Robert E., David B. Mustard, and Lawrence S. Powell. “The Effectiveness of State Legislation in Mitigating Moral Hazard: Evidence from Automobile Insurance.” Journal of Law and Economics 49, no. 2 (2006): 427–450. Insurance Act, R.S.O., c. I.8, Article 266. Ontario. 1990 Insurance Information Institute. 2011 Year End Results, April 16, 2012. Available online at www.iii.org/articles/2011‐year‐end‐results.html. ———. Compulsory Auto/Uninsured Motorists, May 5, 2012. Available online at www.iii.org/issues_updates/compulsory‐auto‐uninsured‐motorists.html. International Transport Forum. Road Safety Annual Report 2011, May 11, 2012. Available online at http://internationaltransportforum.org/irtadpublic/pdf/11IrtadReport.pdf.

124

Ippolito, Richard. “The Effects of Price Regulation in the Automobile Insurance Industry.” Journal of Law and Economics, 22, (1979): 55‐89. Jablonowski, Mark. “Workers’ Compensation State Funds: Evolution of a Competitive Force.” Conning Research and Consulting, 2009. Jaffee, Dwight, and Thomas Russell. “The Causes and Consequences of Rate Regulation in the Auto Insurance Industry.” In The Economics of Property‐Casualty Insurance, edited by David F. Bradford, 81–112. Chicago, IL: University of Chicago Press, 1998. Kelly, Mary, Anne E. Kleffner, and Maureen Tomlinson. “First‐party versus Third‐party Compensation for Automobile Accidents: Evidence from Canada.” Risk Management and Insurance Review 13, no. 1 (2010). Kelly, Mary, Norma L. Nielson, and Tracy Snoddon. “Aging Population and Driver Licensing: A Policy Perspective.” Unpublished Paper. Wilfrid Laurier University, 2012. Kelly, Mary, Sapna Isotupa, and Anne E. Kleffner. “The Impact of Adjuster Moral Hazard on Driving Records.” North American Actuarial Journal 13, no. 4 (2009): 425–437. Kennedy, Kenneth F., and Robert I. Mehr. “A Case Study in Private v. Public Enterprise: The Manitoba Experience with Automobile Insurance.” Journal of Risk and Insurance 44, no 3 (1977): 595–621 Kissinger, Peter. “Building a Safety Culture on Today’s Roadways.” ITE Journal 79, no. 8 (2009): 22–23. Kleffner, Anne E., and Norma L. Nielson. “Auto Insurance Reform for Canada’s Tort Provinces.” Insurance and Risk Management 72, no. 1 (2004): 91–121. Kochanowski, Paul S., and Madelyn V. Young. “Deterrent Aspects of No‐Fault Automobile Insurance: Some Empirical Findings.” The Journal of Risk and Insurance 52, no. 2 (1985): 269–288. Lascher, Edward L. The Politics of Automobile Insurance Reform: Ideas, Institutions and Public Policy in North America. Washington, DC: Georgetown University Press, 1999. Leadbetter, Darrell, and Paul Kovacs. “The Economic Burden of Motor Vehicle Collisions in British Columbia.” Insurance and Risk Management 72, no. 3 (2004): 465–489. Leadbetter, Darrell, and Peter Stodolak. “Why Insurers Fail: Inadequately Pricing the Promise of Insurance.” Toronto, ON: Property and Casualty Insurance Compensation Corporation, 2009. Leadbetter, Darrell, Jane Voll, and Erica Wieder. “The Effects of Rate Regulation on the Volatility of Auto Insurance Prices – Evidence from Canada.” Assurances et Gestion des Risques 76, no. 1 (2008): 21– 54. Loughran, David. “The Effect of No‐Fault Automobile Insurance on Driver Behavior and Automobile Accidents in the United States.” Rand Monographs. Santa Monica, CA: Rand Institute for Civil Justice, 2001. Ma, Yu Luen and Joan T. Schmit. “Factors Affecting the Relative Incidence of Uninsured Motorists Claims.” Journal of Risk and Insurance, 67, (2000): 281‐294. McEwin, R. Ian. “No‐Fault and Road Accidents: Some Australasian Evidence.” International Review of Law and Economics 9, no. 1 (1989): 13–24. McGillivray, Glenn. “Plowing On Through.” Canadian Underwriter 78, no. 6 (2011): 18–35. ———. “A White‐Knuckle Ride.” Canadian Underwriter 68, no. 5 (2001): 16–28.

125

Moulton, Alexis. “TMJ Injuries and the Soft Tissue Injury Cap in Alberta.” McLennan Ross LLP, January 2012. MSA Research Inc. MSA Quarterly Outlook Report Q4‐2011. Toronto, ON: MSA Research Inc., 2012. Nicoletta, John. Driving Characteristics of the Young and Aging Population. Statistics Canada Catalogue no. 53F0007XIE. Ottawa, ON: Statistics Canada, 2002. O’Connell, Jeffrey, and Robert H. Joost. “Giving Motorists a Choice between Fault and No‐Fault Insurance.” Virginia Law Review 72, no. 1 (1986): 61–89. Office of the Auditor General of Ontario. 2011 Annual Report. Toronto, ON: Auditor General, 2011. Oliver Wyman. “Actuarial Analysis for Industry‐Wide Adjustment Effective as of November 1, 2012.” Submission to the Alberta Auto Insurance Rate Board, June 1, 2012. Ontario Auto Insurance Anti‐Fraud Task Force. “Interim Report,” December 2011. Available online at www.fin.gov.on.ca/en/autoinsurance/interim‐report.pdf. Ontario Human Rights Commission. Human Rights in Insurance. Discussion Paper. Toronto, ON: HRC, 1999. Ontario Ministry of Transportation. Ontario Road Safety Annual Report 2009. Toronto, ON: MTO, 2009. ———. Ontario Road Safety Annual Report 2008. Toronto, ON: MTO, 2008. O’Reilly, Mary Lou. “Sharing the Podium with the Governator.” Canadian Underwriter 75, no. 12 (2008): 16, 18. Parry, Ian W. H., Margaret Walls, and Winston Harrington. “Automobile Externalities and Policies.” Journal of Economic Literature 45, no. 2 (2007): 373–399. Picard, Pierre. “Natural Disaster Insurance and the Equity‐Efficiency Trade‐Off.” The Journal of Risk and Insurance 75, no. 1 (2008): 17–38. Powers, Michael R. “Where Ignorance Is Bliss: The ‘dark corner’ of Risk Classification.” The Journal of Risk Finance 11, no. 4 (2010): 353–357. Raphael, Steven, and Lorien Rice. “Car Ownership, Employment, and Earnings.” Journal of Urban Economics 52, no. 1 (2002): 109–130. Rea, Samuel A., Jr. “Economic Analysis of Fault and No‐Fault Liability Systems.” Canadian Business Law Journal 12 (1987): 444–472. Regan, Laureen, Sharon Tennyson, and Mary Weiss. “The Relationship between Auto Insurance Rate Regulation and Insured Loss Costs: An Empirical Analysis.” Journal of Insurance Regulation 27, no. 1 (2008): 23–46. Regina v. Tri‐M Systems, B.C.S.C., 1998. Shavell, Steven M. “The Judgment Proof Problem.” International Review of Law and Economics 6, no. 1 (1986): 45–58. ———. “On Liability and Insurance.” The Bell Journal of Economics 13, no. 1 (1982): 120–132. Shleifer, Andrei. “State versus Private Ownership.” Journal of Economic Perspectives 12 (1998): 133–150. Skipper, Harold D., Jr. “Rationales for Government Intervention into Insurance Markets.” In International Risk and Insurance, edited by Harold D. Skipper, Jr., 236–260. Boston, MA: McGraw‐Hill, 1998.

126

Sleet, David A., Richard A. Schieber, and Julie Gilchrist, “Heath Promotion Policy and Politics: Lessons from Childhood Injury Prevention.” Health Promotion Practice 4, no.2 (2003): 103–108. Smith, Eric and Randall Wright, “Why Is Automobile Insurance in Philadelphia So Damn Expensive?” American Economic Review 82 (1992): 756‐72. Société de l’assurance Automobile du Québec (SAAQ). For a Sustainable and Equitable Plan. Quebec, 2006. ———. On the Right Path for a Sustainable and Equitable Plan. Quebec, 2011. Stigliz, Joseph. “The Private Uses of Public Interests: Incentives and Institutions.” Journal of Economic Perspectives 12 (1998): 3–22. Sulzenko‐Laurie B, Riis V, Grubisic E‐ A survey of injury claims data after introduction of injury care protocols in Alberta, Canada.J Occup Environ Med. 2010 Apr;52(4):450‐5. Tennyson, Sharon, Laureen Regan, and Mary A. Weiss. “Automobile Insurance Regulation: The Massachusetts Experience.” In Deregulating Property‐Liability Insurance: Restoring Competition and Increasing Market Efficiency, edited by J. David Cummins, 25‐80. Washington, DC: AEI‐ Brookings Joint Center for Regulatory Studies, 2002. Tennyson, Sharon and Mary A. Weiss. “State Regulations, Uninsured Driving and the Cost of Automobile Insurance.” Working paper, 2011. Thaler, Richard H., and Cass R. Sunstein. Nudge: Improving Decisions about Health, Wealth and Happiness. New Haven, CT: Yale University Press, 2008. Transport Canada. 2009 Canadian Motor Vehicle Traffic Collision Statistics. TP 3322, 2011. Available online at http://tc.gc.ca/eng/roadsafety/tp‐tp3322‐2009‐1173.htm. ———. 2003 Canadian Motor Vehicle Traffic Collision Statistics. TP 3322, 2004. Available online at www.tc.gc.ca/eng/roadsafety/tp‐tp3322‐2003‐menu‐630.htm. Weisberg, Herbert I., and Richard A. Derrig. “Fraud and Automobile Insurance: A Report on the Baseline Study of Bodily Injury Claims in Massachusetts.” Journal of Insurance Regulation 9 (1991): 497– 541. ———. “Massachusetts Automobile Bodily Injury Tort Reform.” Journal of Insurance Regulation 10 (1992): 384–440. Weiss, Mary A., Sharon Tennyson, and Laureen Regan. “The Effects of Regulated Premium Subsidies on Insurance Costs: An Empirical Analysis of Automobile Insurance.” The Journal of Risk and Insurance 77, no. 3 (2010): 597–624. Yelen, Suzanne. “Withdrawal Restrictions in the Automobile Insurance Market.” Yale Law Review 102 (1993): 1431–1455. Zurich Insurance Co. v. Ontario, C.H.R.R., 1992.

127

9 Appendix: Case Studies of Other Jurisdictions

Here we examine the automobile insurance systems in other countries and jurisdictions and describe what combinations of features of the automobile insurance system have worked well elsewhere, what features have led to poor system functioning, and what types of reforms have improved system functioning over the long term.

9.1 Alberta, Canada

In Alberta, private insurers compete to sell the mandatory automobile insurance product (third‐ party liability, first‐party accident benefits, and uninsured motorist coverage) as well as additional voluntary coverage. First‐party accident benefits for medical payments are $50,000 (as of 2004) and minimum liability limits are $200,000, although most people buy $500,000 or $1,000,000.

Figure 16A (in main text) shows the growth in average premiums in Alberta during the early

2000s. In 2001 the average premium in Alberta was $845 and increased to $1,123 in 2004. A main driver was the increase in bodily injury liability severity. Average BI claim severity increased from about

$49,000 in 2000 to almost $57,000 in 2002, a 16 percent increase. As a result of growing consumer dissatisfaction, especially due to high premiums for new drivers, Alberta instituted significant reforms to the automobile insurance system in 2004. The key provisions to the reforms included:32 (i) a cap on pain and suffering awards ($4,00033) for minor injuries resulting from a collision; (ii) treatment guidelines for minor injuries (prior approval from the insurance company was not required to begin treatment); and

(iii) a pricing grid that established the maximum premium insurers can charge for basic coverage for any driving record. Although there was a court challenge to the minor injury cap, the Alberta Court of Appeal

32 See www.ibc.ca/en/car_insurance/AB/Reforms.asp. 33 The cap is indexed and in 2012 is $4,641. 128

upheld the cap noting that it represented only one aspect of a wider legislative scheme that also included protocols for diagnosing and treating minor injuries and increased medical benefits available to minor injury claimants who still required treatment after preliminary medical treatment protocols were followed.34

The reforms helped to reduce the average premium from $1,182 in 2004 to $1,021 in 2007.

From GISA data, the total estimated savings over the period 2004–2007 for Albertans was $1.13 billion.

The short‐term impact of the reforms is demonstrated in Figure 5A (in main text) which shows that after increasing steadily from 1991–2003 average BI claim severity then experienced a relatively sharp decrease from 2003–2005 before levelling out. In addition, average AB+BI costs also declined, as shown in Figure 6A, from $573 in 2002 to $318 in 2005, falling further to $254 in 2009.

Increasing first‐party accident benefits is one potential way of reducing third‐party liability costs, in addition to the cap on pain and suffering. Figure 8A (in main text) shows the number of BI claims per

100 PD claims in Alberta, which after peaking in 1999 (0.3818 BI claims per 100 PD claims), then began to decline and flattened out over the period 2001–2003, then decreased significantly until 2009 (0.1810

BI claims per 100 PD claims) before increasing again in 2010 (0.2145 BI claims per 100 PD claims). More generous accident benefits also impact claim frequency for medical benefits, as shown in Figure 10 in the text. The number of medical claims per 100 vehicles decreased from 1999 to 2003, increased until

2006 and then decreased. Such an increase in the frequency of medical benefit claims most likely

34For example, the Minor Injury Regulation was brought in along with the Diagnostic and Treatment Protocols Regulation (DTPR), which provides for 10 or 21 treatment sessions in the first 90 days of the injury without the need to seek approval from the insurer. Assuming the treatment protocols in the DTPR are followed, after a 90‐day period is reached, an injured person can apply to an insurer for further payments, which the legislation increased from $10,000 to $50,000. 129

reflects easier access to benefits, since other provinces saw a decrease in this measure during the same period.

The introduction of injury care treatment protocols in Alberta has had a number of impacts on utilization of care as well as costs. Sulzenko, et al. (2010) note that over 40 percent of claims were still open at six months post‐injury. Currently an increasing number of injured persons end up extending their treatment past the 21 treatment/90 day period, raising questions about whether the protocols are appropriate based on more recent evidence of best practices in treatment. However, this is to be expected to some degree since the protocols are based on what works well for most people. Managing those claimants for whom the treatment under the protocols is not sufficient could become a more significant concern over time. However, from a cost containment standpoint since additional treatment has to be approved and the case reviewed, insurers have a built‐in control measure.

As well, there is evidence that claims are trending upward. The Oliver Wyman submission to the

Automobile Insurance Rate Board (AIRB) identifies a number of key issues in 2012 that are impacting total costs. Medical payment loss costs increased by 5 percent in 2011 and the frequency of medical expense claims has been increasing since 2010. The Dominion reported that there was a 6.7 percent increase in accident benefits over 2010/2011. In addition, accident benefit frequency increased 9.5 percent from 2010 to 2011, and 24.3 percent from 2009 to 2010. With respect to liability, BI frequency increased by 4 percent in 2010 and 3 percent in 2011 after a steady decline and an 11.9 percent increase in bodily injury liability claim costs (Oliver Wyman, 2012).

This recent evidence from Alberta suggests that the 2003 reforms are less effective in the long term in controlling claim cost growth. “We see that post reform trends tend to deteriorate after a period of initial improvement. … the expected increase in claim counts that typically follows after reforms have been in place for a number of years is now evident for all mandatory coverages.” (The Dominion of

130

Canada General Insurance Company, 2012; p. 4). Also noted by this insurance company is an increased propensity for injured parties to retain legal representation and attempt to look for areas that fall outside the cap.

The Alberta experience parallels what has been observed elsewhere regarding automobile insurance reform. Short‐term cost reductions are achieved but over time the effectiveness of the changes is eroded. A specific concern of insurers in Alberta recently is the impact on injury claims due to the decision in Sparrowhawk v. Zapoltinsky [2012 ABQB 34]. The question addressed was whether or not an injury to the temporomandibular joint (TMJ) represents a minor injury such that it would fall within the cap on soft tissue damages. The Court concluded that the plaintiff’s TMJ injury was not a minor injury and therefore did not fall within the cap (Moulton, 2012). This is expected to result in an increase in claims involving TMJ injuries. In addition, given that treatment is not provided under the Diagnostic and Treatment Protocols Regulation this could also cause an increase in the average size of claims.

Another important element of the Alberta reforms was the introduction of the premium grid.

Since the introduction of the risk sharing pool, the Facility Association has become one of the larger insurers in the province. Alberta has the fourth‐largest private passenger residual market share in North

America with a 6.9 percent market share (Facility Association, 2012), and the size of the residual market is a significant concern. The 2012 Facility Association submission to the AIRB indicates that results from the Grid pool were financially neutral, but the non‐grid risk sharing pool and the residual market are not financially sustainable and require subsidization from other consumers (Facility Association, 2012).

131

9.2 Massachusetts, U.S.A. 35

Social objectives are explicit in the automobile insurance laws of Massachusetts: to achieve universal insurance coverage and assure reasonable rates for all drivers, while maintaining insurer solvency. Massachusetts established compulsory auto insurance in 1927, and though insurance was sold by private insurers, premium rates for compulsory coverages were set by the state insurance commissioner. Partial no‐fault insurance was adopted in 1971, requiring drivers to purchase first‐party personal injury protection insurance and restricting the right to file an injury liability claim to those whose medical expenses exceeded $500 (raised to $2,000 in 1988). Maximum first‐party injury benefits were set at $2,000 (raised to $8,000 in 1988). Along with compulsory first‐party insurance, uninsured motorist coverage and property damage liability coverage also became compulsory.

From 1978 through 2007, affordability objectives were promoted by use of state‐set insurance rates established through an annual hearing process. State set rates varied by territory and driver class, with classes and territories determined by the state. All insurers were required to use the state‐set rates unless approval to charge lower rates was sought and granted by the insurance commissioner.

In many years, the insurance rates set by the state were below comparable rates in other states.

This was especially true in periods of high inflation such as the 1980s which were characterized by rapid growth in auto insurance loss costs (Cummins and Tennyson, 1992). Evidence of this can be seen by comparing the percentage by which (aggregate) premiums exceeded loss costs, defined as earned premiums minus incurred losses, divided by earned premiums. Figure 19 below displays the annual average premium markup for Massachusetts auto insurance versus the U.S. average for years 1980 through 1998.

35 This discussion draws heavily on Tennyson, Weiss and Regan (2002) and Derrig and Tennyson (2011). 132

Figure 19 – Premium Markup Over Losses: Massachusetts versus U.S.A.

Premium Markup Over Losses 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00

Massachusetts Countrywide

Source: Tennyson, Weiss and Regan, 2002

To promote universal coverage through affordable premiums for all drivers, age, gender, and marital status were prohibited as rate classification variables and the state permitted only 26 rating territories to be used (determined by the state). Rates were also systematically leveled across classes and territories and annual rate increases for a class or territory were limited in relation to the statewide average increase.

To protect insurance availability for all drivers, a take‐all‐comers rule was established, under which insurers could not deny insurance coverage to any applicant. Undesirable policies could be ceded to the residual market facility (Commonwealth Automobile Reinsurers, CAR), but policyholders insured through CAR were charged the same rates as in the voluntary market and losses from CAR were shared by all insurers in the state. Drivers were also strongly protected from cancellation: an insurer could not refuse to renew a policy for reasons other than fraud, failure to pay a premium, or other technicalities.

Finally, to assure payment of claims the state implemented legislation requiring prompt payment and imposing treble damages on insurers in cases of bad‐faith handling of insurance claims. 133

The regulatory efforts to assure affordability were frustrated by continued increases in claims costs. The figure below displays average loss costs per insured vehicle for Massachusetts as compared with the U.S. as a whole, for years 1973 through 1998; the vertical line in 1978 indicates the starting date of most of the Massachusetts regulations. Figure 20 documents a startling upward trend in

Massachusetts costs which began just at this time and continued through the early 1990s when reforms began to be implemented.

Figure 20 – Loss Costs per Insured Vehicle, Massachusetts versus Other States

Loss Costs per Insured Vehicle $600.00 Stringent regulation $500.00 Fraud Bureau $400.00 $300.00 $200.00 $100.00 $0.00

Massachusetts Other NE States All Other States

Source: Derrig and Tennyson, 2011

By the early 1990s it was clear that the system was no longer sustainable and regulatory reforms began to be considered. Initially reform came in piecemeal fashion, and included a higher tort threshold, mandatory insurer cost containment programs, stronger experience rating, greater pricing flexibility for insurers, and establishment of a centralized Insurance Fraud Bureau. Comprehensive reforms took longer to accomplish, but became effective in 2008 when a system of “managed competition” was established. Under managed competition firms may set their own rates and offer discounts and product

134

variations, subject to state prior approval and to restrictions on underwriting.36 In a case study of the regulatory reforms, Tennyson (2011) finds that the reforms led to a number of positive developments in the market without leading to increases in insurance prices or reductions in insurance availability.

Insurance premium expenditure growth has declined relative to previous trends; insurance availability has been maintained; insurer underwriting results have been maintained; and claims rates have remained at pre‐reform levels.

Several studies have examined the evolution of the Massachusetts auto insurance system over time, in an attempt to understand the problems that led to its undoing. A number of problems have been identified. First, as a result of the stringent suppression of rates in the 1980s, many insurers withdrew from the auto insurance market (Derrig, 1993; Yelen, 1993). In 1980, 62 auto insurers operated in the state; by 1998 this number had dropped to 38; and by 2006 only 19 auto insurers served the state. This reduced efficiency and undermined competition (Tennyson 1997), putting upward pressure on costs and premiums.

Second, the system of social pricing led to further reductions in insurance availability. Capping and tempering led to cross‐subsidies in insurance pricing: some drivers (low risks) paid more than their risk‐based rate and other drivers (high risks) paid less. For a private insurer, offering insurance to a subsidized driver was less profitable than offering insurance to a driver paying a surcharge. When rates are suppressed, a larger fraction of drivers receive subsidies. Insurers cede subsidized policies to the residual market, and the residual market grows. Figure 21 demonstrates that the percent of cars insured in the Massachusetts residual market was exceptionally high, in some years insuring the majority of cars in the state.

36 For example, age, sex, income, and credit history may not be used in underwriting and state‐determined rating territories remain in force. 135

Figure 21 – Residual Market Size, Massachusetts versus Other States

Percent of Cars Insured in Residual Market 80% 70% 60% 50% 40% 30% 20% 10% Residual Market Share 0%

Year Massachusetts Unregulated States Average Other Regulated States

Source: Tennyson, Weiss and Regan, 2002

Large residual markets are a hallmark of lack of insurance availability, and can lead to additional problems in the market. Because insurers do not bear the full costs of claims from policies insured in the residual market (due to sharing of residual market profits and losses among all insurers), incentives to control claims costs are greatly reduced when large portions of the market are involuntarily insured

(Blackmon and Zeckhauser, 1991). This permits moral hazard and fraud in claiming and puts upward pressure on costs and premiums. Further, social pricing coupled with an emphasis on compensation

(through take‐all‐comers laws, bad faith liability, and limits on cancellations) decreased insurers’ ability to control claim costs, and decreased drivers’ incentives to control accident and claims costs.

This led to significant moral hazard (and deliberate fraud) in claiming. As summarized in Derrig and Tennyson (2011, p. 180), “The prediction of fraudulent claim behavior was observed soon after the

1988 Reform Law provision that raised the monetary threshold to file a tort claim from $500 to $2,000 in claimed medical bills. Weisberg and Derrig (1992) document the increase in numbers and intensity of

136

medical provider visits with the result being a much larger‐than‐anticipated 1989 proportion of auto injury claims with medical bills in excess of $2,000, the new tort threshold. More recently, Derrig, et al.

(2007) discuss auto injury claims with the appearance of fraud and/or buildup37 both countrywide through the Insurance Research Council (IRC) Study of 2002 Claims and in Massachusetts through the developments in the town of Lawrence. In Lawrence, Insurance Fraud Bureau (IFB) activities reduced injury claims per 100 accidents from 141 in 2002 to 60 in 2004 and claim payments from $48.6 million in

2002 to $19.8 million in 2004. Granted, Lawrence was an exceptional case identified as far back as 1991

(Weisberg and Derrig, 1991), but reductions on a lesser scale in other towns have been realized by IFB efforts since 2002.”

Other studies of the Massachusetts market have examined and commented on the political evolution of the regulatory system (Yelen, 1993; Derrig, 1993; Tennyson, Weiss and Regan, 2002).

Tennyson, Weiss and Regan (2002, p. 77) conclude that “the various features of the system are logically linked when viewed from the perspective of the pursuit of social goals. A major objective of the auto insurance laws in the state is universal coverage of drivers at reasonable rates for all. The universal coverage aims are what lay behind the laws for mandatory insurance, mandatory offer rules and limitations on policy cancellation. With mandatory insurance laws in place, affordability of coverage becomes important and provides the rationale for restrictions on rate class and territory definitions, for rate subsidies and for the lack of residual market premium surcharges. In periods of rising costs the suppression of overall rate levels is one way to protect insurance affordability, and insurers’ losses are mitigated by the sharing of residual market losses. However, with the possibility of rate suppression, and due to the residual market sharing rules, comes the need to place exit restrictions on insurers. The

37 “Build‐up” is the term of art for excessive treatment for the injury (if any) sustained in an auto accident with treatment usually provided by a chiropractor or physical therapist. 137

complexity of this regulatory system thus reflects the difficulty of achieving its objectives within a market system.”

9.3 Michigan, U.S.A.38

Established in 1973, the system in Michigan is a private, compulsory no‐fault auto insurance system which contains elements of social pricing for catastrophic injuries. Michigan is the only no‐fault state in the U.S. to provide for unlimited medical benefits for auto‐related injuries. Injured persons receive unlimited lifetime medical benefits, including reasonable and necessary expenses for rehabilitation and long‐term care; loss‐of‐service benefits for up to three years to pay for help with tasks of everyday living such as laundry, cooking and cleaning;39 replacement of 85 percent of wages for up to three years after the accident; and survivor and funeral benefits. The costs of catastrophic auto‐related injuries are funded through annual assessments to insurers, who in turn incorporate these costs as a fixed assessment on the premiums charged to Michigan’s drivers. The current assessment is $175 per insured vehicle.

Michigan’s no‐fault auto insurance was designed with three public policy goals in mind: (1) to increase the benefits paid to injured persons, (2) to ensure prompt payment of benefits, and (3) to reduce the proportion of premium dollars paid out for administrative (i.e., legal) costs.40 The system has been highly successful in meeting these goals. Due to unlimited medical benefits, injured persons receive generous and certain compensation. And, due to the strong restrictions on lawsuits, Michigan sees a much smaller proportion of automobile injury claims settled through the liability system than in other states. Liability claims for additional compensation (including pain and suffering) are limited to

38 This discussion draws heavily on Tennyson (2011). 39 Loss‐of‐service benefits are subject to a daily maximum, and wage benefits are subject to a monthly maximum. 40 Michigan Insurance Workgroup Report, 2006. 138

situations involving individuals with a serious impairment of body function, individuals with serious permanent disfigurement, or death. Nearly 80 percent of automobile injury claims in Michigan are settled with no involvement of the liability system.

The drawback of the system is its expense. Concerns about the high costs of no‐fault arose soon after its inception, with calls for reform coming as early as 1989.41 But because costs were thought to reflect the generous benefits provided, only minor changes have been made over the years. Figure 22 provides a stark graphic representation of the resulting cost trends, displaying the average cost

(severity) per first‐party injury claims in Michigan in each year from 1980 through 2009. Costs grew slowly in the early years, but the rate of growth has increased over time. Between 1997 and 2009

Michigan’s average cost injury claim rose by 252 percent.

The bulk of the claims expenditure growth is due to catastrophic injury claims. Catastrophic claims may involve injury to the brain and/or spinal cord, and injuries that result in serious and permanent disability. People who are permanently disabled can require attendant or residential care for the remainder of their lifespans, in addition to medical care and rehabilitative services. These claims remain in the system year after year, leading to high claim payment expenses long after the accident has occurred. Research estimates suggest that catastrophic claims account for less than 2 percent of first‐ party injury claims by number, but for over 45 percent of loss costs (Tennyson, 2011).42

41 See Report on No‐Fault Auto Insurance Reform in Michigan, Senate Commerce and Technology Committee. 42 The Michigan Catastrophic Claims Association (MCCA) claim and loss data are from the MCCA FY ending June 2008, reported in the MCCA’s Annual Statement to OFIR. Total PIP losses for Michigan are from the NAIC 2007/2008 Auto Insurance Database Report. Total PIP claims for Michigan are estimated from 2007‐2008 Fast Track data. Calendar year data for 2007 and 2008 are averaged to estimate fiscal year losses and claims. 139

Figure 22 – Cost per Injury Claim in Michigan

Cost Per Injury Claim $35,000.00

$30,000.00

$25,000.00

$20,000.00

$15,000.00

$10,000.00

$5,000.00

$0.00

Source: Tennyson, 2011

The ultimate costs of catastrophic claims are difficult to predict and create a very risky exposure for insurers. To help insurers reduce uncertainty associated with these claims, the Michigan Catastrophic

Claims Association (MCCA) was created to act as a reinsurer for large no‐fault claims. The MCCA is a private, non‐profit, unincorporated association created by the Michigan state legislature in 1978.43 Every automobile insurer operating in the state is required to be a member. The MCCA reimburses insurers for the portion of each no‐fault claim that exceeds a specified amount, currently $500,000.44

In 1991, the MCCA was paying about $100 million in reimbursements each year. The annual payment amounts doubled to $200 million per year by 1997, and doubled again within another five

43 The MCCA was created by Public Act 136 of 1978, which added section 3104 to the Michigan Insurance Code. Its form and operations are governed by the enabling legislation. Information is taken from the MCCA 2010 Annual Statement, posted on the MCCA website http://www.michigancatastrophic.com. 44 Originally, insurers were required to pay the first $250,000 of every no‐fault claim, and the MCCA reimbursed them for any additional payments. Beginning in 2002, the per‐claim loss amount that insurers must retain has gradually increased [Public Act 3 (2001)]. 140

years. By 2010, annual reimbursements had grown eightfold and in 2011 MCCA paid out nearly $1 billion. Annual costs are clearly growing exponentially rather than linearly, due to the long‐term nature of claims that accumulate in the system. Between 1978 and 2011, 27,169 claims had been reported to the MCCA and approximately one‐half of these (13,522) remained open in 2011.45

As a result of the claim cost growth, insurance premiums in Michigan have increased more rapidly than those in other states, so that in recent years Michigan drivers have been paying relatively more for automobile insurance than they did in the past. In 1997 Michigan had the 18th highest costs of automobile insurance in the nation, but by 2007 the state had the 11th highest auto insurance costs.

Average premiums per car in Michigan rose by 30.5 percent during that period, more than twice as fast as the national average. Nonetheless, average insurance premiums in Michigan grew only half as fast as average loss costs, meaning that the premium increases are not due to higher insurer profit margins.

Auto insurance reform has become an important public policy concern in Michigan as it has grown clear that the current system is unsustainable. One new element in the current discussions is the recognition of excessive costs in the system. Insurers document that they pay far more for services than do other providers, and evidence of fraud and abuse in claims is mounting. One key problem is that benefit reviews, benefit limits, and cost controls are generally very weak in Michigan’s auto no‐fault system. Although wage and service benefits are subject to maximum reimbursement rates, medical and residential care expenses have no lifetime limits and medical reimbursement rates are uncapped. Fee schedules for medical services and reviews of service utilization are important components of cost controls in most government insurance programs, but are not a part of Michigan’s no‐fault system.

Coordinated fraud control systems are also not present.

45 Statistics are obtained from the MCCA website, http://www.michigancatastrophic.com. 141