P104 Understanding Poverty

Centre for Development, Environment and Policy

Author: Colin Poulton

This version draws on two previous versions of the module authored by Colin Poulton and Stefanie Busse. The lead author for Unit 9 was Chris Joynes. The author wishes to express his gratitude to Laura Rodriguez Takeuchi and Amina Khan of Overseas Development Institute for excellent input into the scoping of this version.

ABOUT THIS MODULE

Within the Sustainable Development Goals (SDGs) the eradication of extreme poverty remains a central objective of international development efforts. As with the Millennium Development Goals before them, the SDGs recognise that poverty has multiple dimensions and that progress is needed on a number of fronts (economic, social and political) if poverty is to be effectively tackled. Government departments, international development agencies and non-governmental organisations (NGOs) are thus expected to design policies and to plan interventions with a clear understanding of how these will contribute to poverty reduction objectives in the areas concerned. This module is aimed at development practitioners – from government departments, international development agencies, NGOs or private business – who are involved in the design of policy or interventions to combat poverty in low- or middle-income countries. It aims to provide a sound understanding of the nature of poverty, its causes and consequences, of trends in poverty reduction across continents and regions of the world, and of debates as to the drivers of these trends.

STRUCTURE OF THE MODULE

The module is broadly structured in two parts. The first seven units provide a conceptual framework for understanding poverty and its causes (both proximate and deeper). The final three units review evidence on trends in poverty at global and regional level and the key factors accounting for these trends. Part I: The first two units consider what we mean by poverty and how poverty can be measured. The definition of poverty that we use makes a big difference to the number of people who are considered to be poor. Unit 1 examines money-metric definitions of poverty (based primarily on income or consumption levels), considering the strengths, weaknesses and practical measurement challenges associated with each. Basic numeracy skills are required to study this unit. Unit 2 examines multi-dimensional poverty measures in a similar way, along with definitions of poverty generated through participatory exercises with poor people. Unit 3 considers the relationships between economic growth, inequality and poverty reduction. The focus here is on macro-level dynamics, which shape the context in which local and household-level poverty dynamics play out. Unit 4 begins with the sustainable livelihoods framework and considers poverty dynamics at household level: how and why people fall into poverty and/or get stuck in it. Whilst Unit 4 focuses on household assets as an important determinant of poverty status, Unit 5 introduces social relationships and differentiation, and explores how these interact with economic factors to produce multidimensional poverty. The linkages between poverty and the environment are explored in Unit 6. This unit highlights the dependence of many poor households on natural capital, considers how population growth interacts with environmental management and poverty, and considers climate change as a huge evolving shock that will impact the livelihoods of millions of poor households. Finally, Unit 7 considers how the exercise of power affects poverty, given that powerlessness is often cited by poor people themselves as a key feature of the experience of being poor. It considers the conditions under which political leaders are likely to pursue pro-poor policies and also takes a critical look at the role of international development assistance (aid) in poverty reduction. Part II: Unit 8 presents available evidence on trends in monetary poverty and hunger across countries and continents. It also examines debate on the relative incidence of rural versus urban poverty. Projections by the World Bank indicate that the majority of the world’s extreme poor will live in rural areas until at least 2025. However, does the underlying definition of poverty understate the prevalence of urban poverty? Unit 9 considers drivers of observed trends in capability poverty. Finally, Unit 10 examines the SDGs. Drawing on a comparison with the Millennium Development Goals, it considers the process by which they were developed and the content of the goals themselves, so as to consider their potential to contribute to the eradication of extreme poverty in the world by 2030.

WHAT YOU WILL LEARN

Module Aims • To present the multiple dimensions of poverty and how they can be measured. • To explore both the proximate and deeper factors that trap people in poverty or assist them to escape poverty. • To compare trends in poverty reduction across continents and regions and to consider the key factors accounting for these trends. • To examine the impact of international interventions and initiatives on efforts to eradicate poverty.

Module Learning Outcomes By the end of this module, students should be able to: • demonstrate a rigorous and critical understanding of key concepts used in international poverty debates. This includes practical issues of measurement as well as definition • analyse both the proximate and deeper factors that trap people in poverty or assist them to escape poverty • critically assess the impact of international interventions and initiatives, such as international development assistance and the SDGs, on efforts to eradicate poverty.

ASSESSMENT

This module is assessed by: • an examined assignment (EA) worth 40% • a written examination worth 60%. Since the EA is an element of the formal examination process, please note the following: (a) The EA questions and submission date will be available on the Virtual Learning Environment (VLE). (b) The EA is submitted by uploading it to the VLE. (c) The EA is marked by the module tutor and students will receive a percentage mark and feedback. (d) Answers submitted must be entirely the student’s own work and not a product of collaboration. (e) Plagiarism is a breach of regulations. To ensure compliance with the specific University of London regulations, all students are advised to read the guidelines on referencing the work of other people. For more detailed information, see the FAQ on the VLE.

STUDY MATERIALS

There is no textbook for this module, but the following has been provided:  Collier, P. (2007) The Bottom Billion. Oxford, Oxford University Press. This is not a traditional textbook; rather, it provides a particular, high-profile, but also controversial analysis of the causes of poverty in low-income countries and how to eradicate it. You are invited to read selected chapters of the book as you work through the module and are encouraged to read others at some point during your studies. The selected chapters illustrate how economics, geography, governance and politics combine to trap large numbers of people in poverty or to help them escape from it. They also model an evidence-based and critical analysis of complex and often emotive questions. That said, you are encouraged to develop your own critical assessment of the arguments presented in the book and to develop your own understanding of what needs to be done to tackle poverty as you work through the module.

For each of the module units, the following are provided.

Key Study Materials Key readings are drawn mainly from the textbooks, relevant academic journals and internationally respected reports. They are provided to add breadth and depth to the unit materials and are required reading as they contain material on which you may be examined. Readings are supplied as digital copies and ebooks via the SOAS Online Library. For information on how to access the Library, please see the VLE. For some units, multimedia links have also been provided. You will be invited to access these as part of an exercise or activity within the unit, and to discuss their implications with other students and the tutor.

Further Study Materials These texts and multimedia are not always provided, but weblinks have been included where possible. Further Study Materials are NOT examinable; they are included to enable you to pursue your own areas of interest.

References Each unit contains a full list of all material cited in the text. All references cited in the unit text are listed at the end of the relevant units. However, this is primarily a matter of good academic practice: to show where points made in the text can be substantiated. Students are not expected to consult these references as part of their study of this module.

Self-Assessment Questions Often, you will find a set of Self-Assessment Questions at the end of each section within a unit. It is important that you work through all of these. Their purpose is threefold:

• to check your understanding of basic concepts and ideas • to verify your ability to execute technical procedures in practice • to develop your skills in interpreting the results of empirical analysis. Also, you will find additional Unit Self-Assessment Questions at the end of each unit, which aim to help you assess your broader understanding of the unit material. Answers to the Self-Assessment Questions are provided in the Answer Booklet.

In-text Questions

 This icon invites you to answer a question for which an answer is provided. Try not to look at the answer immediately; first write down what you think is a reasonable answer to the question before reading on. This is equivalent to lecturers asking a question of their class and using the answers as a springboard for further explanation.

In-text Activities

 This symbol invites you to halt and consider an issue or engage in a practical activity.

Key Terms and Concepts At the end of each unit you are provided with a list of Key Terms and Concepts which have been introduced in the unit. The first time these appear in the study guide they are Bold Italicised. Some key terms are very likely to be used in examination questions, and an explanation of the meaning of relevant key terms will nearly always gain you credit in your answers.

Acronyms and Abbreviations As you progress through the module you may need to check unfamiliar acronyms that are used. A full list of these is provided for you in your study guide.

TUTORIAL SUPPORT

There are two opportunities for receiving support from tutors during your study. These opportunities involve: (a) participating in the Virtual Learning Environment (VLE) (b) completing the examined assignment (EA).

Virtual Learning Environment (VLE) The Virtual Learning Environment provides an opportunity for you to interact with other students and tutors. A discussion forum is provided through which you can post questions regarding any study topic that you have difficulty with, or for which you require further clarification. You can also discuss more general issues on the News Forum within the CeDEP Programme Area.

INDICATIVE STUDY CALENDAR

Part/unit Unit title Study time (hours)

PART I CONCEPTUAL FRAMEWORK

Unit 1 Money-metric measures of poverty 15

Unit 2 Multidimensional poverty 10

Unit 3 Growth, inequality and poverty 15

Unit 4 Poverty dynamics 15

Unit 5 Social differentiation and poverty 15

Unit 6 Poverty and the environment 15

Unit 7 Power, politics, governance and aid 15

PART II TRENDS IN POVERTY REDUCTION

Unit 8 Trends in monetary poverty and hunger 15

Unit 9 Explaining trends in capability poverty 10

Unit 10 The Sustainable Development Goals 10

Examined Assignment 15 Check the VLE for submission deadline

Examination entry July

Revision and examination preparation Jul—Sep

End-of-module examination late Sep— early Oct

Unit Information 2 Unit Overview 2 Unit Aim 2 Unit Learning Outcomes 2

Key Study Materials 3

1.0 Income or consumption poverty 4 Section Overview 4 Section Learning Outcomes 4 1.1 Income-based measures of poverty 4 1.2 Consumption-based measures of poverty 7 1.3 Assets 10 Section 1 Self-Assessment Questions 12

2.0 Poverty lines 13 Section Overview 13 Section Learning Outcomes 13 2.1 National poverty lines 13 2.2 The international poverty line 16 2.3 Adult equivalence scales 20 Section 2 Self-Assessment Questions 22

3.0 Foster–Greer–Thorbecke poverty indicators 23 Section Overview 23 Section Learning Outcome 23

3.1 P0, P1 and P2 23

Unit Summary 27

Unit Self-Assessment Questions 28

Key Terms and Concepts 29

Further Study Materials 30

References 32

Understanding Poverty Unit 1

UNIT INFORMATION

Unit Overview

If we are seeking to ‘understand’ poverty, then a good place to start is with definitions. These, in turn, permit empirical analysis of factors that contribute to poverty and interventions that might reduce it. Defining poverty might seem a straightforward exercise, but in fact there are many different ways of defining poverty and each of these has its own measurement challenges. In this unit, we consider money-metric definitions based on consumption levels or income. Historically, these have been the dominant way of defining and measuring poverty, due in part to their adoption by the World Bank and other major donors. Thus, Sustainable Development Goal 1 (SDG 1) includes a target to ‘eradicate extreme poverty’, defined in money-metric terms, much as Millennium Development Goal 1 (MDG 1) incorporated a monetary poverty target. However, as we will see in this unit and the next, such approaches also have their critics. They are thus increasingly being complemented by multi-dimensional poverty measures, not just within academia, but at national and international policy levels.

Unit Aim

• To provide a detailed and critical knowledge of money-metric approaches to defining and measuring poverty.

Unit Learning Outcomes

By the end of this unit, students should be able to: • discuss the strengths and weaknesses of money-metric definitions of poverty • explain the advantages and disadvantages of measuring poverty in terms of consumption as opposed to income • explain and critically assess the concepts of the international poverty line, purchasing power parity and the Foster–Greer–Thorbecke poverty measures.

KEY STUDY MATERIALS

 Deaton, A. (2006) Measuring poverty. In: Banerji, A., Bénabou, R. & Mookherjee, D. (Eds.) Understanding Poverty. Oxford and New York, Oxford University Press. pp. 3– 15. This chapter covers much of the ground of this unit in a clear and accessible way. Deaton writes freely as only a highly respected authority can, with the result that the reading includes numerous personal insights and opinions alongside its simple presentation of basic concepts. The one downside is that its examples, although useful, are now becoming a little dated.

 Ruggeri Laderchi, C., Saith, R. & Stewart, F. (2003) Does it matter that we do not agree on the definition of poverty? A comparison of four approaches. Oxford Development Studies, 31 (3), 243–274. This paper considers the requirements of a satisfactory poverty measure, then reviews and critiques four broad approaches to the identification and measurement of poverty: monetary, capability, social exclusion, and participatory approaches. Finally, it investigates whether different measures of poverty identify the same people as poor and concludes that different approaches identify different people as being poor. Much subsequent work has explored this same question and reached the same conclusion. Therefore, please note the conclusion and don’t feel that you have to spend too long on the specific examples. Monetary approaches to defining and measuring poverty are reviewed and critiqued on pp. 247–253. Sections 3.2–3.4 (pp. 253–262) are most directly relevant to Unit 2 of this module. However, you may wish to read them quickly here, so as to more fully appreciate what monetary approaches to defining poverty do not cover.

1.0 INCOME OR CONSUMPTION POVERTY

Section Overview

Money-metric approaches to defining and measuring poverty focus on economic dimensions of well-being or the standard of living. Thus, Webster's dictionary defines poverty as ‘the state of one who lacks a usual or socially acceptable amount of money or material possessions’. This definition of poverty focuses on the ability to purchase or consume goods and services. In order to measure and compare poverty, you need an indicator which allows you to determine who is poor and who isn’t. In money-metric approaches, income or (more commonly) consumption levels are used as the indicator of well-being. This section will discuss the advantages and disadvantages of using income- as opposed to consumption-based definitions of poverty, including the challenges in collecting the relevant data. It will also briefly consider the use of physical asset holdings as an indicator of poverty.

Section Learning Outcomes

By the end of this section, students should be able to: • explain the advantages and disadvantages of measuring poverty in terms of consumption as opposed to income • discuss the use of asset holdings as an indicator of poverty.

1.1 Income-based measures of poverty

As you work through this and subsequent sections, it may help if you have in your mind a mental image of a less well-to-do household (rural or urban). Think about them for a minute. Where do they live? How many people are there in the household (adults and children)? Are they always all there? What do they do to earn a living? In this unit and the next, we are going to think about some of the conceptual and practical questions surrounding how someone – perhaps a government programme, an international development agency or a non-governmental organisation (NGO) – decides whether or not the members of this household should be classed as poor. Income-based definitions of poverty take the income of an individual or household over a defined period as the key indicator of their well-being during that period.

 Can you think of strengths and weaknesses of this approach?

Income is an important determinant of what an individual or household can consume at any given time. Income provides the means to buy food and clothes, rent housing if necessary, and get around. It allows the household to pay school fees for its children or buy medicines (assuming the income is high enough!). People with high incomes have money left over to save and have a good time.

Clearly, income is very important. Recognising this, an increasing number of countries provide various forms of benefits (pension, unemployment benefits, carer’s allowance, sick pay) to people who are unable to earn an income for one reason or another. However, people do not always live off their current income. For example, in times of income shortage they may draw on savings or borrow and when their income recovers or increases they may save or repay loans. This is called consumption smoothing, as efforts are made to ensure that consumption fluctuates less than income. This reminds us that income is not an end in itself. Rather, it enables people to consume. Therefore, current consumption is generally regarded as a better (more direct) indicator of welfare than current income.

Collecting data on incomes

There are also practical difficulties with measuring income, especially when it is the income of less well-to-do households in poor economies. Such information is usually collected through some form of questionnaire survey.

 Can you think what some of the difficulties might be?

The module author’s household receives the vast majority of its income from the salaries of the course author and his wife. These salaries are received on a regular basis and vary very little from month to month. By contrast, only a minority of people in low- income economies enjoy regular and predictable incomes from wages or salaries. Instead, they are hired as casual workers (when work is available), sell crops to generate income or engage in petty trading or other small business enterprises. In the case of agricultural production, petty trading and other small business enterprises, the revenue generated has first to cover the costs of production. It is the net income that is really available to provide income for the individual or household to live off and also ideally to enable new investment. In the absence of detailed record keeping, it can be difficult to remember how much revenue has been received over a specified period. Then, the relevant production costs need to be subtracted. There are also questions as to what should be counted as income. For example, say that two farmers enjoy exactly the same harvest of rice. One keeps all her rice to be eaten at home, whereas the other sells some of hers to buy flour to make chapatis. Are their incomes the same or different? The answer is that they are the same, but the first farmer receives her income as ‘non- monetised’ income or income ‘in kind’. However, if this is true for crop harvests, what about housing? Suppose your neighbour goes to take a job in a nearby town. Instead of selling his house, he rents it out and uses the rent that he receives to pay for his own accommodation in the town. Is his income higher than yours? Again, it is not, but this time we need to recognise that people receive imputed income from owner-occupied dwellings. There are many reasons why incomes may be under-reported through household surveys. Some have already been noted above.

 Can you think of others?

In addition to the reasons already mentioned: • Respondents may be reluctant to disclose their full incomes for fear that the information might be passed to the tax authorities. • The questionnaire may not prompt the interviewer to ask about all possible income sources. This is particularly likely where non-monetised income sources are concerned. For example, Cavendish (1999) explored the importance of income from the utilisation of environmental resources (eg wild foods and medicines, fuelwood, timber) to a sample of around 200 households in southern Zimbabwe in the mid-1990s. Some of these products were gathered and sold by the households concerned, but others were used entirely within the household. Cavendish estimated that the value of these products amounted to around 35% of the total income of the sampled households in the two years of his survey. However, data on income from such sources is often not collected by official surveys. • The questionnaire seeks to discover household income, yet all the questions are directed to a single respondent and he or she does not know about all the income streams received by other members of the household.

 Might you expect income data collected through household surveys to contain any systematic biases across occupation groups? If so, what might these be? Answer There are two major (potential) biases that we should note. Firstly, it appears that wealthier households are more likely to consciously under-report their incomes than poor households. This means that income surveys may under- estimate the extent of inequality within a population, even if such deliberate under-reporting makes little impact on the estimates of poverty. Secondly, as already noted, the problems of assembling reliable data are most severe where respondents obtain their incomes from agricultural production, petty trading or other small business enterprises — where incomes can be irregular and record keeping may be weak or non-existent.

In countries where many people in the population receive a large share of their income from wages or salaries, it may be easier to collect data on incomes than it is to measure consumption. These countries tend to be middle or high-income countries. In such countries, poverty estimates may be based on income data. By contrast, in most low- income countries, consumption data are preferred when it comes to estimating the number of people living in poverty.

Household versus individual data

Most surveys – whether of incomes or consumption – collect data for respondent households, rather than for individuals. They then estimate the incomes (or consumption) of individual members by dividing the recorded household income (or consumption) by the total number of members. Members may be weighted by age and sometimes also by gender. Apart from the convenience of only having to talk to one household member, a major justification for collecting income data at household level is that much income is pooled; in other words, it is available for general household needs irrespective of who actually earned it. However, this is rarely true for all income – and there are differences in the extent to which income is pooled across both households (depending on the relationships and level of trust between household members) and cultures. A much- cited ‘stylised fact’ is that women put in much of the labour for the cultivation of cash crops, but it is the men who take the crops for sale. The men then keep most of the money, spending some of it on beer, cigarettes or even marrying another wife. Thus, estimates of income or consumption that are based on household-level data ignore inequalities of income or consumption within households. As a result, it is highly likely that poverty is underestimated, with some women and children who live in poverty due to unequal intra-household relationships overlooked by official poverty figures. Practically speaking, the maintenance of separate ‘accounts’ within a household also means that any given respondent (even the household head) may only have a partial picture of the income that has come into the household (ie to all household members) during the period that is being investigated by a survey.

1.2 Consumption-based measures of poverty

It was argued above that consumption is a better indicator of well-being than income: incomes may fluctuate, but consumption tends to be more constant as households can draw on borrowing or savings in times of income shortage. Many statistical surveys of well-being and poverty, therefore, attempt to track consumption rather than incomes. Over the past 25 years in many countries a large investment has been made in periodic, nationally representative Living Standards Measurement Surveys (LSMS), with particular support from the World Bank. Some of the initial impetus for this investment came from debates over the impact of structural adjustment. Then there was a desire to see how countries were performing in relation to Millennium Development Goal 1. More generally, many governments want to know what impact the policies they are pursuing are having on well-being, poverty and inequality within their populations.

Collecting data on consumption

Consumption is measured by calculating expenditure on all goods and services, plus non-monetised consumption (of own produced crops, housing etc), over a specified period. Commonly, this only includes consumption of so-called private goods (goods which, once consumed by one person or household, are not available for consumption by another). Some argue that a better consumption indicator would be one that was broadened to include consumption of free public goods, such as education or sanitation.

 What benefits can you see from adopting this broader definition? What problems could you envisage with it?

It has also been pointed out that low consumption is not necessarily a reliable indicator of poverty. Social and cultural factors can also play a role in explaining low consumption, with some households and cultures choosing to save more than others. A big challenge with a consumption-based approach to poverty is measurement and, specifically, recall. • Think back over everything that you have eaten over the past 24 hours. How easy is it to remember? • How accurately could you remember what members of your household have eaten over the past week? Is there someone else in your household who would remember this better than you and, if so, why? • Could you remember all expenditure that you have incurred over the past week? Whilst people with lower consumption levels have less to remember, these questions still illustrate the difficulties of recall. For this reason, consumption sections of major surveys typically restrict questions to consumption over the last 24 hours or last week. Can you think of any further difficulties that this might present? Particularly in agrarian economies, consumption can be highly seasonal. Although we have argued that households have ways of smoothing consumption, so that it fluctuates less than income, financial markets are imperfect, so poor households generally cannot fully compensate for fluctuations in incomes through savings and/or credit. Even salaried workers may delay purchases of some items until they receive their pay cheque at the end of the month. More dramatically, in many agrarian economies there is a phenomenon called the ‘hungry period’ or ‘lean season’, just before the harvest, when stocks from the previous harvest are running low or are exhausted. In such months, poor households may cut the number of meals that they eat in a day from three to two or even one – even though this is a busy time of year for agricultural production activities. A major implication of this is that, if recall is limited to 24 hours, a week or even a month (for larger items), the resulting measure of consumption poverty will depend heavily on when the interview takes place. Was it at the beginning of the month or towards the end, before or after harvest? Large-scale surveys can get round many of the problems caused by limited recall simply because they cover so many households that interviews take place over the best part of a year. However, even this may not avoid all problems (see 1.2.1).

1.2.1 Seasonality and consumption poverty in Malawi In Malawi, the main periodic, national living standards survey is the Integrated Household Survey (IHS). IHSs were undertaken in 1998, 2004 and 2010—2011. Whilst these IHSs have generated much useful data, in the earlier rounds some observers were puzzled by the findings relating poverty incidence to other variables, which sometimes conflicted with common perceptions of, for example, living standards in different areas. Various explanations were put forward for this, but an important one was as follows: enumerators working in teams moved progressively around different areas, so they visited different areas at different times of the year. Although the sample design provided a sample across all seasons in each region, it was not possible to design the sample so that households in a specific city, for example, were sampled across all seasons. There was thus a danger that estimates of household expenditure and of poverty incidence would be biased by the time of interview. Indeed, if month of interview was introduced into regression analyses that sought to explain observed consumption, it consistently came out as a significant explanatory factor, even when more common (and fundamental) variables such as landholding size and participation in wage employment were included. Analysis that failed to control for this effect was thus likely to generate misleading results.

Source: Dorward, A (personal communication)

Deaton (2006) reports an experimental survey in India in which:

‘a randomly selected half of the households got a thirty-day reporting period [for food consumption – the normal survey practice in India, but not elsewhere], while the others got a seven-day reporting period. On average, households reported about 30% more food purchases on the seven-day questionnaire and only about 18% more on all expenditures including food – not such a large difference in itself, but enough to cut the measured number of poor in India by half!’ Source: Deaton (2006) p. 14.

In 2015, the World Bank is going to switch to using poverty data for India that is based on the 7-day recall period (Cruz et al, 2015: p. 12). The impact of this change will be dramatic, as over 100 million fewer Indians will be classified as extremely poor as a result (Ferreira et al, 2015). This equates to a fall of more than 10% in the total number of people in the world counted as extremely poor – all because of an apparently innocuous change in survey methodology. Similarly, Beegle et al (2012) report the consequences of adopting several variants of a consumption questionnaire in Tanzania. Using a $1.25 per day poverty line, their estimate of headcount poverty in their sample was 63% where a 14-day recall period was used for food consumption, but fell to 55% where a 7-day recall period was used. If a household diary was used (ie food consumption of all family members was recorded, to minimise recall problems), the estimate of headcount poverty varied from 56% to 60%, depending on the frequency of monitoring visits by enumerators. If just one person administered a personal diary, the estimate fell to 48%. Furthermore, the impacts of these changes varied by household size and literacy levels.

 To ensure that you have understood how recall periods affect the estimated poverty rate, explain in your own words why reducing the recall period for food consumption from 14 to 7 days might reduce the estimated headcount poverty rate. Answer The longer the recall period, the more likely it is that a respondent will forget particular items of food that they have consumed (or purchased in the case of an expenditure survey). Hence, their recorded consumption (expenditure) will be under-estimated, making them look poorer than they are.

A final issue with consumption measurement concerns a comparison of consumption across large and small households. This is dealt with in Section 2.3.

1.3 Assets

In addition to income, the Webster's dictionary definition of poverty (cited earlier) also highlights the importance of ‘material possessions’. Given that much thinking on the livelihoods of poor people highlights the importance of ‘assets’, is there a case for defining poverty in relation to asset holdings, rather than income or consumption?

 You might want to consider your own answer to this question before proceeding with the rest of this section.

Physical assets – such as houses, furniture, bicycles, radios and agricultural implements – are relatively easy to identify and, for a poor household, to count as part of a household survey. Such asset holdings influence current consumption in a number of ways, many – but not all – of which should be captured by consumption surveys. For example, use of productive assets, such as ploughing equipment, sewing machines and vehicles, should translate into higher income and higher recorded consumption. Similarly, if the imputed value of housing (the biggest asset that most home owners have) is well recorded, it should be higher for a household in a large, permanent house than for a household in a small mud and thatch building. However, the extra well-being that comes from having food and drink stored in a fridge – as opposed to being at room temperature and prone to go off – may not be captured by consumption data. Asset holdings may also provide a source of savings that can be drawn upon during periods of low income. However, if a household sells productive assets to support short-term consumption levels, it may compromise its future income and hence consumption possibilities by doing so. One major difficulty with defining poverty in relation to physical asset holdings is that the relationship between physical (and especially productive) asset holdings and well- being fluctuates over time. The reason for this is that other factors (eg economic upswings and downturns, price fluctuations) affect the return that individuals and

© SOAS CeDEP 10 Understanding Poverty Unit 1 households realise from their assets. Hence, a household with a reasonable stock of assets could experience low consumption in a given period because key market prices were unfavourable in that period, but then bounce back when prices recovered. Looking at this positively, focusing on a household’s asset endowment might give a better picture of the household’s medium-term consumption possibilities than current consumption does. This, however, depends on identifying a stable and/or predictable relationship between asset holdings and medium-term consumption (or other indicators of well-being). As yet, little work has been done on this. Another major practical difficulty in constructing an indicator of well-being based on physical asset holdings is how to value them. The value of most assets depreciates over time, so the value will be dependent on age, condition, how much second-hand items can be sold for locally, etc. Valuation is thus extremely data demanding. Finally, thinking on livelihoods does not focus solely on physical assets, but also on human, social, and other forms of capital. These are equally important in determining one’s livelihood possibilities and in influencing well-being. Thus, focusing on physical assets may fail to capture the full range of capabilities desirable for a fulfilled life. By contrast, multi-dimensional approaches to defining and measuring poverty do reflect the importance of multiple forms of capital to people’s well-being.

Section 1 Self-Assessment Questions

1 For each of the challenges below, identify whether they are associated with the collection of data on income and/or consumption using household questionnaires in low-income countries.

Challenge Income Consumption Seasonality

Limited recall

Intra-household distribution Desire to avoid tax

Valuing own produced food

2 Fill in the missing words/phrases. A household’s asset endowment might give a better picture of the household’s medium- term consumption possibilities than ______consumption does, but the relationship between asset endowment and current consumption tends to be ______.

2.0 POVERTY LINES

Section Overview

Once you have identified an appropriate indicator of well-being, if you want to compare poverty across households, regions or countries you need a way of determining who qualifies as poor as opposed to non-poor. Poverty lines can be defined as cut-off points. People who fall below the cut-off point or threshold are considered poor. This is inevitably somewhat arbitrary and there are debates about how a poverty line should be chosen. This section considers national poverty lines and then the particular challenges associated with the construction of an international poverty line.

Section Learning Outcomes

By the end of this section, students should be able to: • summarise the steps and challenges associated with setting a national poverty line • explain and critically assess the concepts of the international poverty line and purchasing power parity.

2.1 National poverty lines

Most poverty lines are country-specific. The factors that are taken into account when setting a poverty line vary considerably from country to country. This is entirely appropriate, as a poverty line should aim to capture what it means to be poor in a particular culture and context. (Moreover, this is true whether one is talking about a money-metric or a multi-dimensional poverty line). However, there are common elements to many poverty lines. First, we need to distinguish between absolute and relative poverty lines. • Relative poverty lines have mainly been developed for industrialised or middle-income countries. They are defined in relation to the average income or consumption in a country; in the United Kingdom (UK), for example, the poverty line is set at 60% of the country’s median income. Relative poverty lines recognise that the relative position of individuals or households in society is an important aspect of their welfare. This is because the consumption of certain goods or services and/or participation in certain activities is considered ‘normal’ in a given society. Poor households are unable to engage in this ‘normal’ behaviour within society, because of their lack of income. Note that the understanding of ‘normal’ will vary from country to country and indeed over time within a given country. • Absolute poverty lines are anchored in some supposedly absolute standard of what households require in order to meet their basic needs. Most low-income countries use an absolute poverty line, as does the USA.

2.1.1 The meaning of relative poverty The concept of relative poverty recognises that poverty is a social phenomenon. Townsend, an academic who has been pivotal in defining poverty and social exclusion in Europe, expressed poverty in 1979 as being a complex phenomenon composed of both absolute ‘lacks’ of goods and services and an individual (or household’s) relative place in society. ‘Individuals, families and groups in the population can be said to be in poverty when they lack the resources to obtain the types of diets, participate in the activities and have the living conditions and amenities which are customary, or at least widely encouraged and approved, in the societies to which they belong. Their resources are so seriously below those commanded by the average individual or family that they are, in effect, excluded from ordinary living patterns, customs and activities.’

Source: Townsend (1979) p. 31.

In the remainder of this section we will focus on absolute poverty lines. The basic steps in setting an absolute poverty line are to • determine a collection (or ‘basket’) of essential items that allow people to satisfy certain basic needs • estimate the cost of buying that basket of essential items. Absolute poverty lines are commonly one of two types. A given country may use both in its poverty monitoring.

Food poverty line

This line is established by determining the cost of a food basket that satisfies some set of minimal nutritional requirements. Households with consumption levels below this line are unable to meet basic nutritional requirements, even if they spend all their available income on food. This line is, therefore, considered to capture ‘extreme poverty’ (or some similar term). There are, of course, challenges in setting this line. For example, it is difficult to determine nutritional requirements, as the minimum food or calorie intake varies not only according to age and gender (see below), but also according to activity. Who do you think is likely to have a higher nutritional requirement – an agricultural labourer or a taxi-driver? How one satisfies these requirements might also depend on the nutritional characteristics of the local staple food, which in turn will affect the cost involved.

Non-food poverty line

This is often simply called the ‘poverty line’. It is set higher than the food poverty line, so more people are classified as poor using this definition. In addition to food items to satisfy basic nutritional requirements, it is assumed that the household has to incur other basic expenditures, for example, on housing, lighting, clothing and transportation, in order to survive. Defining essential non-food items is as difficult as determining basic nutritional requirements. In practice, therefore, the total expenditures of those who just satisfy basic nutritional requirements are examined. It can then be seen how much

© SOAS CeDEP 14 Understanding Poverty Unit 1 they spend on non-food items relative to food expenditure and this becomes the benchmark for expenditure on essential non-food items in the poverty line. Note that this benchmark can vary over time, as economies and societies change and new technologies – for example, mobile phones – spread even amongst the poor. Thus, in 2015 the Rwandan government is finalising its latest estimates of poverty in the country and has realised that the poor now spend a greater share of income on non- food expenditure than they did at the time of the last survey round. It is, therefore, not appropriate simply to adjust the poverty line to account for inflation since that last round. Instead, a new poverty line is being estimated from the latest survey data. This introduces an element of relativity even into absolute poverty lines.

 In addition to setting two poverty lines, some countries have two variants for each, one for urban and the other for rural areas. Can you think why this might be?

Defining need is only the first step. Once you have done this, the second step is to estimate how much it will cost to satisfy that need. Costs of living, especially food and housing, tend to be higher in urban than in rural areas. World Bank country poverty assessments routinely calculate separate poverty lines for urban and rural areas, with the former typically around 30% higher, although the differential is higher in poorer countries (Chen & Ravallion, 2007). Ideally, of course, you would calculate different lines for multiple areas of a country, as costs of living are not uniform across all urban or rural areas. However, as this is not realistic, having an urban–rural distinction is a useful and feasible step. The choice of poverty line is important not just in giving a reasonable assessment of the incidence of poverty in a country at a given point in time. Observed trends in poverty over time may also depend on the poverty line used. Thus, Chen and Ravallion (2008) reported falling global poverty during 1981–2005 using a US$1.25 per day international poverty line (see below for a discussion of this line), but a smaller fall – and in some sub-periods a rise – when a US$2 per day poverty line was used. These differences may simply reflect different dynamics occurring in different sectors of an economy. However, there could also be a distortionary effect: it is sometimes alleged that governments who know that their performance on poverty reduction is being monitored using a particular poverty line may prioritise policies that are likely to benefit those whose incomes are close to this line, but worry less about people whose incomes are already a little bit higher (or quite a bit lower). One criticism of almost any poverty line – including multi-dimensional ones – is that it conveys the impression that the lives of people whose consumption levels are just below the line (who are ‘poor’) are qualitatively different from the lives of people just above the line (who are ‘non-poor’). In practice, this is unlikely to be the case. Indeed, the lives of people just above the line are likely to have much more in common with the lives of people just below the line than they do with the lives of people whose consumption levels are well above the line.

 Consider the following four households: HH1 has average consumption of 50 kwacha per person per day; HH2 has average consumption of 95 kwacha per person per day; HH3 of 105 kwacha per person per day and HH4, 250 kwacha per person per day. The national poverty line is 105 kwacha per person per day. (a) Which of these households are poor, according to the official definition? (b) Consider HH2: in consumption terms, which of the other three households does it most closely resemble? (c) Answer the same question for HH3. Answer (a) HH1 and HH2 are poor according to the official definition, whereas HH3 and HH4 are ‘non-poor’. (b) In consumption terms, ‘poor’ HH2 has most in common with ‘non-poor’ HH3. The average consumption of HH2 is almost twice that of the other ‘poor’ HH1, but is 90% that of HH3. (c) Similarly, ‘non-poor’ HH3 has most in common with ‘poor’ HH2. The average consumption of HH3 is only around 40% of that of the other ‘non-poor’ HH4.

2.2 The international poverty line

A limitation of national poverty lines is that they do not allow for direct cross-country comparisons or aggregations, as some lines are higher than others. For this reason the World Bank has developed the so-called ‘international poverty line’ (IPL), which is the average of the poverty lines of 15 of the poorest countries in the world (Chen & Ravallion, 2008; Ferreira et al, 2015). These poverty lines are made comparable by converting them to US dollar terms using purchasing power parity (PPP) exchange rates, which are explained below. When the IPL was first set, it had a value of US$1.08 per person per day measured using 1993 PPP exchange rates. In 2008, it was revised, based on a greatly expanded data set of national poverty lines and a new (2005) set of PPP exchange rates (Chen & Ravallion, 2008). The revised line was US$1.25 per person per day at 2005 PPP exchange rates. Anyone whose consumption (or, where appropriate, income), adjusted for cost of living differences across countries (the notion of PPP), was below US$1.25 per person per day was, therefore, consuming less than the minimum amount considered essential to satisfy basic requirements in the world’s poorest countries. In 2015, the World Bank has once again re-estimated the IPL on the basis of a new set of (2011) estimates of PPP. The new line is US$1.90 per person per day at 2011 PPP exchange rates and is discussed in detail by Ferreira et al (2015). The IPL is translated into the local currency units (LCU) of a country using the most recent PPP exchange rate. As PPP exchange rates are only calculated infrequently, in the intervening periods the local currency equivalent of the IPL is updated from year to year using the domestic consumer price index (CPI).

From 2008 until 2015, therefore, the local currency equivalent of the IPL would be calculated as follows:

IPLcurrent LCU = 1.25  PPP2005  [CPIcurrent/CPI2005] where

PPP2005 = the 2005 PPP exchange rate of the country in question

CPIcurrent/CPI2005 = the ratio of the current value of the CPI index to its value in 2005.

The IPL was used to quantify progress towards Millennium Development Goal 1 – to halve the number of people who live on less than a dollar per day between 1990 and 2015 – and will similarly be used to monitor progress towards Sustainable Development Goal 1 (SDG 1). Periodic LSMS or other nationally representative household survey data are used to assess the proportion of the population living on less than the IPL and hence also trends in this figure. Note that national poverty assessments, such as those published by a National Statistical Office, typically assess the proportion of the population living below the national poverty line, which may not correspond to the IPL. However, the basic data used for both calculations are the same.

Purchasing power parity

In the early 2000s, sustained criticism of the IPL and its use by the World Bank in monitoring progress towards MDG 1 came from Reddy and Pogge, amongst others. Many of their criticisms focused on the PPP component of their IPL. Thus, before we look at the criticisms in detail, we need to understand the concept of PPP. Reddy and Pogge (2005) state that PPP exchange rates are commonly understood as:

‘the number of units of a country’s currency necessary to purchase the ‘same amount’ of commodities as can be purchased for one unit of the base country’s currency at the base country’s prices’ Source: Reddy and Pogge (2005) p. 9.

The prices of goods and services across countries are compared periodically through multi-country surveys conducted by the International Comparison Programme (ICP). As you may have noted from the previous pages, one ICP survey was conducted in 1993, the next in 2005 and the most recent in 2011. The USA is commonly used as the base country for PPP estimates. Thus, if a similar basket of goods is identified in country X and in the USA, the PPP conversion factor or ‘exchange rate’ for country X can be calculated as:

Local cost of basket (in LCU)/Cost of basket in USA (in US$)

Note that, when we use the term basket here, it is not the same basket as is used to determine a national poverty line. Rather, ‘basket’ simply means a collection of items,

© SOAS CeDEP 17 Understanding Poverty Unit 1 as when a shopper goes to a market or supermarket and puts several items that they want to buy in their shopping basket. Different baskets can be used for different purposes! It is important to understand why PPP exchange rates are different from more conventional exchange rates. In the short run, the latter are determined primarily by capital flows. However, there is a notion that their long-term equilibrium level is that which causes imports and exports of goods and services to balance. Thus, if a country is importing more than it exports, its exchange rate will depreciate so as to make imports more expensive in LCU terms and to provide an incentive for exporters to increase the volumes of goods and services that they export. In this world of free trade (and assuming that transport and other transaction costs are minimal), the US$ exchange rate is equal to the local cost of a traded good (or a basket of traded goods) in LCU, divided by the cost of that good (or basket of goods) in the USA (in US$). This looks identical to the formula that we have just quoted for the PPP exchange rate above! The difference between conventional exchange rates and PPP exchange rates occurs because not all goods are traded internationally. If you create a representative basket of goods to reflect the consumption patterns of people in any given place, it will almost certainly include a combination of both ‘tradable’ goods and services (those which are traded across national boundaries) and ‘non-tradable’ goods and services (which are not). Whilst the price of tradable goods will tend to be similar across developed and developing economies (for the reasons just explained), non-tradable goods and services tend to be cheaper in low-income economies. The main reason for this is that labour costs (a major component of production costs) are lower in low-income economies.

 Consider the following hypothetical basket of goods.

Within this basket, the cooking oil is the only fully tradable item. In our stylised example, therefore, let’s assume that the conventional exchange rate is US$1 = 5 LCU. Meanwhile, two of the other items have tradable elements (eg the wheat used in bread making is traded and there is an international market for managers of mobile phone operations), but neither these items nor the green vegetables are themselves traded internationally. What is the PPP exchange rate?

Answer PPP exchange rate = 18 / 5.5 = 3.27.

Note that, because non-tradable goods and services tend to be cheaper in low-income economies than in the USA or other higher income economies, the PPP exchange rate is lower (in terms of LCU per US$) than the conventional exchange rate.

As countries grow and incomes rise, the proportion of goods in any representative basket that are tradable also tends to rise. Thus, the PPP exchange rate tends to rise as a proportion of the official exchange rate over time (Chen & Ravallion, 2008). One of the main criticisms advanced by Reddy and Pogge (2005) was that the 1993 PPP estimates used by the World Bank up to that time had been calculated based on prices that were representative across the economy as a whole, rather than on prices that would apply to a typical basket of goods and services consumed by poor households. Perhaps surprisingly, Reddy and Pogge claimed that the poor consume fewer non-tradable goods and services than wealthier households, because they consume a large share of staple foods which are often at least partially tradable. (Think of rice and wheat as the dominant staples in Asia.) They argued that, if PPP conversion factors were calculated on the basis of the data collected by ICP for the ‘all foods’ or ‘breads and cereals’ categories, rather than for all goods and services, they came out 30% or more higher than the estimates actually used in converting the IPL into local currencies. Hence, poverty levels were underestimated by the World Bank’s approach. In its 2005 survey, the ICP made a deliberate effort to collect data on the cross-country costs of a basket of goods that was of particular importance to the poor. The resulting PPP estimates were then incorporated by the World Bank into revised estimates of the number of poor households across countries and regions. This switch to 2005 PPPs added 400 million (40%) to the official estimate of the number of poor people in the world (Chen & Ravallion, 2008), which shows how important methodological ‘details’ such as the PPP exchange rate can be in poverty calculations. It may also provide some support to Reddy and Pogge’s original argument. In defence of their work, Chen and Ravallion would respond by saying (correctly) that the quality of the PPP estimates is improving over time.

 When new PPP estimates are released, the IPL itself is also changed. Thus, originally it was US$1.08 per person per day (at 1993 PPP exchange rates) and was then changed to US$1.25 per person per day (at 2005 PPP exchange rates). These changes are intended to ensure that the purchasing power of the IPL in terms of the goods consumed by the poor in the poorest countries stays the same, despite changes to PPP exchange rates. Therefore, the total number of people living below the poverty line in the 15 countries used to construct the IPL should not be affected by changes to PPP exchange rates (Chen & Ravallion, 2008). Why, therefore, did the total number of the poor in the world increase so much with the incorporation of the 2005 PPP exchange rates into global poverty estimates? Answer Because changes to PPP exchange rates in other countries containing large numbers of poor people, eg China, were different from the average of the changes in the 15 countries whose national poverty lines are used in the estimation of the IPL.

So far, we have not questioned the assumption that it is possible to identify a basket of goods that is of particular importance to the (global) poor. However, a moment’s reflection on the diets of (poor) people in different continents shows that there is no single basket that satisfies this condition. For example, the poor in one place may

© SOAS CeDEP 19 Understanding Poverty Unit 1 consume rice as their dominant staple, whilst those in another place consume maize and those in yet another place consume yams or potatoes. This adds to the complexities of comparing prices across countries and hence of calculating a consistent set of PPP exchange rates for use in poverty analysis. In the 2005 ICP round, the chosen solution to this problem was to compile a basket for each region or sub-region of the world and then to use a set of countries that were at the boundaries of two regions or sub-regions (so-called ‘ring’ countries) to check that the prices of different regional baskets were as consistent with each other as possible. Some, however, have criticised the workings of this approach and argue that this led to inconsistencies in the 2005 PPP estimates – one of the reasons why they produced such big changes in the global poverty estimates. In 2011, therefore, the ICP collected price data from each country for as many items as possible from a ‘global core list’ of 618 items (Ferreira et al, 2015: p. 8).

 The 2011 PPP exchange rates are available from the World Bank website (World Bank, 2015). Use this to look at the PPP exchange rate for your country or a country you are familiar with. Convert the (new) IPL into local currency terms for this country by multiplying the PPP exchange rate by the dollar value of the IPL, ie 1.9. Now imagine trying to live on that amount of money per day. Remember that this includes the value of own produced food that you consume — not just what you buy. Do you think this gives a realistic picture of what constitutes poverty in the country? If not, how can you explain this based on the information provided in this section?

2.3 Adult equivalence scales

Having established a poverty line and collected the necessary data, an important step in preparing data for analysis is to decide how to convert information on household consumption into an estimate of consumption for individuals within the household. (Note that, in this section, we assume that consumption is allocated in a fair manner within the household, based on some indicator of need. However, in reality, men, especially household heads, often account for an ‘unfair’ share of consumption within their households). Two issues have to be addressed. The first is that consumption ‘needs’ vary across individuals. Thus, we noted above that minimum food or calorie intake varies according to age and gender. Hence, if total household consumption is simply divided by the number of household members, irrespective of age, this will bias upwards the estimated welfare of households composed mainly of adults relative to households with a large proportion of children (as the latter will tend to have lower recorded consumption per head, but in practice it will go further). One way in which this problem can be handled in poverty assessment is through the use of adult equivalence scales. Thus, the number of ‘adult male equivalents’ in each household is calculated, based on a system of weightings. These weightings are commonly based on scales produced by nutritionists, although there is no necessary

© SOAS CeDEP 20 Understanding Poverty Unit 1 reason why the weightings for consumption of non-food items should be the same as for food. An internet search will reveal that different scales are used by different people and authorities. One example is presented in the table in 2.3.1.

2.3.1 An example of an adult equivalence scale

Category Age (years) Average energy allowance Equivalence scale per day (kcal) Infants 0—0.5 650 0.22

0.5—1.0 850 0.29

Children 1—3 1300 0.45

4—6 1800 0.62

7—10 2000 0.69

Males 11—14 2500 0.86

15—18 3000 1.03

19—25 2900 1.00

26—50 2900 1.00

51+ 2300 0.79

Females 11—14 2200 0.76

15—18 2200 0.76

19—25 2200 0.76 26—50 2200 0.76 51+ 1900 0.66

Source: National Academy Press (1989)

A second issue that may arise is the argument that there are ‘economies of scale’ in certain kinds of consumption. Thus, a household with just two members will need a stove for cooking. However, a household of five might survive with the same type of stove. Thus, it is cheaper for larger households to meet their consumption needs than it is for smaller ones. Deaton and Zaidi (2002) suggest that this problem be addressed by adjusting household size (in ‘adult equivalent’ members) by a further factor that reflects the share of durable items with these ‘public good-type’ characteristics in total household consumption. They observe that, in poor rural areas where housing costs are low and most consumption consists of food, little adjustment for economies of scale in consumption is required. By contrast, such adjustment may be more important in urban areas where housing costs feature much more prominently in total expenditure. However, they recognise that there is little agreement on this issue, with the result that it may often be ignored by analysts.

Section 2 Self-Assessment Questions

3 No-one in the USA could live on less than US$1.90 per day. Therefore, why is the IPL set at US$1.90 per day at (2011) PPP rates?

4 Briefly summarise the main challenges arising from using a monetary indicator to measure progress towards SDG 1.

3.0 FOSTER–GREER–THORBECKE POVERTY INDICATORS

Section Overview

Once a poverty line has been identified and consumption data assembled, the next task is to assess how many people fall below that line and how many lie above it. However, some assessments of poverty and poverty trends may also go one step beyond this. Foster, Greer and Thorbecke (1984) derive three measures of poverty – incidence, depth and severity – using the same basic data. This section explains these three measures. They have most commonly been used with monetary poverty data, but can be applied more generally, for example with data on calorific intake, anthropometric data or with certain multi-dimensional poverty indicators. All that is required is data from a household survey and a threshold or benchmark line against which the indicator for each household can be compared. Please note: some students see either the formula or the spreadsheet exercise below and decide to skip this section. You should not do this if you are serious about obtaining a masters-level qualification in Poverty Reduction! If you struggle with either the formula or the spreadsheet exercise, please ask a friend or even a teenage maths student to help you.

Section Learning Outcome

By the end of this section, students should be able to: • explain and make use of the Foster–Greer–Thorbecke poverty measures.

3.1 P0, P1 and P2

The three Foster–Greer–Thorbecke measures are known as P0, P1 and P2. They are all calculated using the following formula:

a 1 q zy P  i a  nzi1  where P = the poverty measure (a = 0, 1 or 2) n = the number of people in the population q = the number of people below the relevant poverty line z = the relevant poverty line yi = the income of person i, a person below the poverty line.

The Σ sign means that you calculate the expression in brackets for every person from i to q (ie all the people below the relevant poverty line) and then sum the resulting

© SOAS CeDEP 23 Understanding Poverty Unit 1 figures. The expression in brackets captures how far the expenditure of poor person i falls below the poverty line, as a proportion of that poverty line. 1/n means that you then divide the total by the total number of households in the population.

In maths, X0 = 1, so for the P0 measure, you can effectively ignore the expression in brackets. Thus, P0 = q/n. In other words, P0 simply tells you the proportion of the population whose consumption falls below the poverty line. It is otherwise known as the poverty headcount ratio or the incidence of poverty. This most basic of poverty measures is a simple concept to grasp and, therefore, it is widely used in political debate. However, it does not take notice of how poor the poor are. If, between two years, the poor become less poor, but still have consumption expenditures below the poverty line, the headcount ratio will not change. As already noted, some argue that a focus on the poverty headcount ratio encourages politicians to pursue policies that lift the consumption levels of those who are close to the poverty line, so that they cease to be ‘poor’, whilst neglecting those whose consumption levels are lower (the extreme poor), who are more difficult to help.

The poverty gap index (P1) is intended to address this incentive problem, by also incorporating the depth of poverty of those who are poor. For each member i of the population who lives below the poverty line, it measures the shortfall of their expenditures below the poverty line, as a proportion of that poverty line. Unlike the headcount ratio, therefore, the poverty gap index will change if the poor become less poor, even if their expenditures remain below the poverty line.

However, P1 does not simply measure the average depth of poverty of those who are poor. Instead, it is a composite measure that combines the poverty headcount with the depth of poverty of those who are poor. This makes it more difficult to grasp or explain intuitively than P0. As a result, whilst it occurs quite frequently in academic literature, it is much less commonly used in political debate than P0.

Finally, in the poverty severity index (P2) the term in brackets in the formula is squared, which gives extra weight to the depth of deprivation of the poorest people. Hence, unlike the headcount ratio and the poverty gap index, the poverty severity index will pick up changes in the distribution of income within the group of poor.

However, P2 is one step more complicated than P1. Therefore, it is also rarely used in political debate.

Recognising that the P1 and P2 measures tend to be neglected, because they are not intuitive, the World Bank is now exploring alternative ways of conveying the depth of poverty. Cruz et al (2015) report a person-equivalent poverty measure first developed by Castleman et al (2015). This allows progress in poverty reduction – both incidence and depth – to be tracked in relation to a base year. Say, for example, that you have consumption data for your country of interest for 2010 and, in that year, poverty incidence was 25% and the average poverty gap for those who were poor was 40 pesos, ie their consumption per day was, on average, 40 pesos below the poverty line. In this case, the person-equivalent poverty measure counts how many person-equivalents there are living below the poverty line at a poverty gap of 40 pesos. Thus, a person living 60 pesos below the poverty line counts as 1.5 person- equivalents and a person living 20 pesos below the poverty line counts as 0.5 person-

© SOAS CeDEP 24 Understanding Poverty Unit 1 equivalents. When the next national living standards survey is conducted, you count all these up (adjusting the gap for inflation in the intervening period) and compare the answer with that obtained in 2010. Thus, even if the headcount ratio is unchanged, but the average depth of poverty has reduced, the person-equivalent poverty measure will show an improvement. By contrast, if the headcount ratio has fallen slightly, but the depth of poverty of those remaining in poverty has increased, the person-equivalent poverty measure may show little, if any, improvement.

Exercise: Poverty measurements

In the spreadsheet, Unit 1 Poverty Lines Exercise, (which is provided on your e-study guide) are consumption estimates (in current pesos per month) for 50 hypothetical individuals, in each of two years (1 and 2) when a household living standards measurement survey was undertaken. The spreadsheet also gives the local poverty lines for each of these two years. There are two lines: • a lower one, indicating extreme poverty (this is set as 90 pesos per person per month in year 1) • a 50% higher one (set as 135 pesos per person per month in year 1).

 For this hypothetical population of 50 people, for each of the two years and each of the two poverty lines, calculate the following Foster–Greer–Thorbecke measures:

— the incidence of poverty (P0) as a fraction or percentage of the total population)

— the depth of poverty (P1)

— the severity of poverty (P2).

Hints

0 X = 1, so for the P0 measure, you can ignore the expression in brackets. Thus,

P0 = q/n.

For P1, for each person below the poverty line, calculate the expression in brackets. Then sum these results and divide the total by n.

For P2, square the expression in brackets for each person below the poverty line before summing the results and dividing the total by n. It is fine to do all your calculations in pesos per month. (1) Display your answers in the following table:

Poverty rates by poverty line, year and poverty measure

(2) If you only consider the lower poverty line and the incidence of poverty (P0) measure, what appears to be the trend in poverty within this population between years 1 and 2? (3) If you consider the other poverty measures and the higher poverty line, how does this picture change? (4) Look in detail at the data. How do you explain the difference in results generated by the different poverty lines and measures?

Answers (1) The poverty rates are as follows: Poverty rates by poverty line, year and poverty measure

(2) If you only consider the lower poverty line and the incidence of poverty (P0) measure, it appears that poverty within this population is falling (from 24% in year 1 to 16% in year 2, ie a considerable fall).

(3) If instead you use the higher poverty line and the incidence of poverty (P0) measure, poverty apparently rises between year 1 (44%) and year 2 (52%). Thus, within this population the choice of line affects even the direction of change in

poverty over time. Similarly, if the P1 and P2 measures of poverty are used, poverty is seen to get worse between year 1 and year 2.

(4) Several people (nos 6, 24, 39, 42 and 46) who were close to the poverty line in

year 1 manage to rise above it in year 2. Hence, the incidence of poverty (P0) measure falls. However, several of the poorest people in year 1 (nos 8, 15, 19 and 33) experienced real declines in consumption between year 1 and year 2, which increases the depth and severity of poverty. Meanwhile, a number of people who were just above the higher poverty line in year 1 (nos 17, 31, 45, 47 and 50) fall

below it in year 2, such that the incidence of poverty (P0) measure rises if the higher poverty line is used. This is a similar phenomenon to that observed by Chen and Ravallion (2008).

UNIT SUMMARY

Defining poverty might seem a straightforward exercise but in fact there are many different ways of defining and measuring poverty. The dominant approach is to define poverty as a shortfall in incomes or consumption. This approach differentiates the poor from the non-poor by establishing a monetary poverty line, below which people are considered poor. The international dollar per day line is an example of such a line. In practice, all poverty lines are somewhat arbitrary. For example, ‘poor’ households living close to the poverty line may have much more in common with ‘non-poor’ households living just above the line than they do with very poor households living well below it. The unit has noted a range of practical problems that are encountered when trying to measure either income or consumption in low-income countries. There are also significant challenges in applying an IPL, not least those associated with estimating PPP. One conclusion that you could draw from all this is that it is a waste of time and effort trying to measure accurately the number or proportion of people in poverty. This would be an unhelpful conclusion to draw, as we need to have some way of assessing whether policies to reduce poverty are having any effect. However, we should be wary of the apparent precision embodied in official poverty statistics: for example, the headcount poverty rate in country X fell from 52.9% in 2010 to 43.4% in 2015. Rather, given the unavoidable uncertainties surrounding the figures, you can be more confident that figures are identifying some sort of real effect where: (a) the size of the change appears to be substantial; (b) it is sustained over time and (c) it is corroborated by other pieces of information. Finally, the unit introduced the three poverty measures originally presented by Foster,

Greer and Thorbecke (1984). One of these, the poverty headcount (P0) is widely understood and used in debate. The other two incorporate consideration of the depth of poverty of the poor alongside the headcount rate. This is intended to draw attention to measures that alleviate the poverty of the worst off. However, the P1 and P2 measures are more difficult to understand and communicate than the headcount measure, so are much less frequently used in public debate.

UNIT SELF-ASSESSMENT QUESTIONS

1 What are the advantages of measuring money-metric poverty using consumption rather than income as the main indicator?

2 Briefly summarise the problems associated with measuring the extent of poverty using the headcount index.

KEY TERMS AND CONCEPTS consumer price index An index number measuring the price of a typical basket of consumer goods and services purchased by households within a population. The index tends to rise over time due to inflation. deprivation What an individual is not having or lacking, eg education. money-metric Monetary value of incomes, goods, and services.

FURTHER STUDY MATERIALS

Ferreira, F., Chen, S., Dabalen, A., Dikhanov, Y., Hamadeh, N., Jolliffe, D., Narayan, A., Prydz, E., Revenga, A., Sangraula, P., Serajuddin U. & Yoshida N. (2015) A Global Count of the Extreme Poor in 2012. Data Issues, Methodology and Initial Results. Washington DC, The World Bank. Policy Research Working Paper 7432. Available from: http://www-wds.worldbank.org/external/default/WDSContentServer/ WDSP/IB/2015/10/14/090224b083144b10/2_0/Rendered/PDF/A0global0count00an d0initial0results.pdf This paper represents the World Bank’s most up-to-date presentation of the number of people in the world living below the international (monetary) poverty line, plus past trends in this figure. It incorporates the outputs of the 2011 survey round of the International Comparison Programme into global poverty estimates, entailing a ‘new’ international poverty line of US$1.90 consumption per person per day at 2011 purchasing power parity (PPP) exchange rates. The paper is a technical read, but sheds light on the statistical issues and debates underlying the official figures on the number of people living in extreme monetary poverty in the world. After a helpful introduction and historical review of global poverty measurement using PPPs, Section 3 discusses methodological issues encountered in computing global poverty estimates. This section generally reinforces what you will learn from the main unit text, but don’t get bogged down in the minutiae of the calculations and be prepared simply to move on if you find the going tough. (For example, you may need strong concentration to work your way through the logic of Section 3.3!) Section 4 then describes some debates surrounding the appropriate level for the 2011 international poverty line. Again, you don’t need to worry about all the counter- proposals, but it will help if you can understand why US$1.90 per person per day at 2011 PPPs is considered an equivalent line to US$1.25 per person per day at 2005 PPPs. Finally, Section 5 describes some results, which we will return to in Unit 8.

Haughton, J. & Khandker, S. (2009) Handbook on Poverty and Inequality. Washington DC, The World Bank. Available from: http://hdl.handle.net/10986/11985 Many of the concepts discussed in this unit are discussed in more depth in Chapters 2, 3 and 4 of this book. Therefore, it is suggested that you use the book as additional reference material, consulting it if you need more explanation on a particular topic or want to check whether you have understood something correctly or not.

World Bank (n.d.) How is Poverty Measured? [Video]. Washington DC, The World Bank. Duration 3:13 minutes. Available from: http://www.worldbank.org/en/news/video/2013/09/09/how-is- poverty-measured This very simple video presents the basic approach to, and a justification for, measuring poverty using a money-metric approach. It is intended as a mass communication tool, not as an academic defence of the approach. However, as you listen, note down (a) any points that you consider to be well made and (b) any places where the narrative seems to you to be weaker.

REFERENCES

Beegle, K., De Weerdt, J., Friedman, J. & Gibson, J. (2012) Methods of household consumption measurement through surveys: experimental results from Tanzania. Journal of Development Economics, 98 (1), 3–18. Castleman, T., Foster, J. & Smith, S. (2015) Person-equivalent Headcount Measures of Poverty. Washington DC, George Washington University, Elliot School of International Affairs. Institute for International Economic Policy Working Paper Series 2015-10. Cavendish, W. (1999) Empirical Regularities in the Poverty–Environment Relationship of African Rural Households. Oxford, The Centre for the Study of African Economies. Working Paper 99– 21. Available from: http://www.csae.ox.ac.uk/workingpapers/pdfs/9921text.pdf [Accessed 21 August 2015] Chen, S. & Ravallion, M. (2007) Absolute poverty measures for the developing world, 1981–2004. Proceedings of the National Academy of Sciences, 104 (43) 16757–16762. Chen, S. & Ravallion, M. (2008) The Developing World is Poorer Than we Thought, but no Less Successful in the Fight Against Poverty. Washington DC, The World Bank. Policy Research Working Paper No 4703. Cruz, M., Foster, J., Quillin, B. & Schellekens, P. (2015) Ending Extreme Poverty and Sharing Prosperity: Progress and Policies. Washington DC, The World Bank. Policy Research Note 15-03. Deaton, A. & Zaidi, S. (2002) Guidelines for Constructing Consumption Aggregates for Welfare Analysis. Washington DC, The World Bank. Living Standards Measurement Study (LSMS), Working Paper No 135. Available from: http://go.worldbank.org/6791D5FAT0 [Accessed 26 May 2015] Deaton, A. (2006) Measuring poverty. In: Banerji, A., Bénabou, R. & Mookherjee, D. (Eds.) Understanding Poverty. Oxford and New York, Oxford University Press. pp. 3–15. Ferreira, F., Chen, S., Dabalen, A., Dikhanov, Y., Hamadeh, N., Jolliffe, D., Narayan, A., Prydz, E., Revenga, A., Sangraula, P., Serajuddin U. & Yoshida N. (2015) A Global Count of the Extreme Poor in 2012. Data Issues, Methodology and Initial Results. Washington DC, The World Bank. Policy Research Working Paper 7432. Foster, J.E., Greer, J. & Thorbecke, E. (1984) A class of decomposable poverty indices. Econometrica, 52, 761–766. National Academy Press (1989) Recommended Dietary Allowances. 10th edition. Washington DC. Reddy, S.G. & Pogge, T.W. (2005) How Not to Count the Poor. Version 6.2. Available from: http:// citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.587.1531&rep=rep1&type=pdf [Accessed 26 May 2015] Townsend, P. (1979) Poverty in the United Kingdom. London, Penguin Books. World Bank (2015) Data. PPP Conversion Factor, GDP (LCU per international $). [Online]. Washington DC, The World Bank. Available from: http://data.worldbank.org/indicator/ PA.NUS.PPP [Accessed 27 October 2015]