Poverty measurement and policy: concepts and applications

Transcription

1 October 2014 PED401. Applications and Cases in International Development Teaching Notes 1 Poverty measurement and policy: concepts and applications Poverty reduction is one of the principal objectives of development. Some would argue it is the objective of development, though, as we shall see, the latter formulation is more controversial than it may sound. If it s a principal objective, we need to understand what it means, how to measure it, and how it relates to development patterns and policy choices. These notes explores issues of concepts and measurement, and will look at some examples of application from Vietnam and Guatemala. It is intended as a map of the issues rather than an in-depth case. The following case will turn more explicitly to strategic policy approaches to poverty reduction, with a focus on ongoing debates in India. Concepts We outline here some of the analytical tools, looking first at links between poverty and choice-theoretic concepts, second, how to link this to one approach to measurement of welfare changes across the distribution, third, outlining an alternative conceptualization from Amartya Sen s capability approach, and fourth, looking at implications of multiple dimensions for measurement. Poverty and choice-theoretic concepts Let s start with an indirect utility function v (p, w) = max u (x) such that px w This relates wealth (w), the choice of a vector of goods (x) at a price vector (p) Now suppose society has some way of agreeing that utility below some level is unacceptable. Then, in terms of the expenditure function 1 These notes were prepared by Michael Walton and are solely for teaching purposes. Thanks to Diana Zamora for inputs.

2 2 z = e ( p, u 0 ) Where u 0 is the utility level at the poverty line. Note that prices are central to this definition. If unacceptable levels of utility are fixed over space and time, then this is absolute, expenditure poverty, and allows application of standard techniques of welfare evaluation. How might the unacceptable level of utility and expenditure be arrived at? The standard practical approach in contemporary poverty measurement is to start from a nutritional minimum requirement, calculate the cost of this based on actual food consumption patterns of households who barely reach this minimum, and then add an allowance for non-food consumption, based on the actual shares of food and non-food amongst the poor. This can be further adapted to allow for both different requirements for children, adults and the aged (know as equivalence scales ), and for economies of scale in consumption for larger households. (If you note some potential for fuzziness in implementation here, you re right.) Note also that most analysts prefer expenditure to income, both because expenditure is conceptually closer to a measure of permanent income, and because it is typically measured with less error. 2 It is useful to show the poverty line in relation to the overall distribution of incomes or expenditures. This is done in Figure 1, that displays a typical stylized frequency distribution with respect to per capita household expenditures, with much greater densities around the middle of the distribution, and a long tail, owing to small numbers of households with high expenditure levels. The poor are those to the left of the poverty line. Figure 1. Poverty and the overall frequency distribution of household per capita expenditures. 2 There is a lot more to say on poverty measurement: these notes are only intended to bring out a few points. See Deaton (1997) chapter 3 for a good survey.

3 3 Increases in expenditure, due to growth that affects the whole distribution, can be represented as a right-ward shift in the curve, leading (typically) to a reduction in those in poverty (Figure 2). It is clear that changes in poverty are only one part of overall distributional dynamics. Figure 2 Growth in expenditures of the overall distribution and effects on poverty This presents a picture on changes, but says nothing about how to add them up across households to give us an overall welfare measure. Just counting individuals below the poverty line, or headcount poverty, is by far the most commonly used. It happens to be a pretty dumb welfare measure. Think of the following: below the line, increments to expenditure, say from desperately poor to moderately poor, have zero impacts on the headcount measure. By contrast, marginal increases in income that happen to cross the poverty line do affect the measure. And increments to expenditure above the line, say from just not-poor to the middle class, have zero effects. For calculating measures of poverty (below the line), a valuable, and much-used, set of indices was developed by Foster, Greer and Thorbecke in their (FGT) class of measures, where the poverty index, P α is given by P α = N -1 ((z w i )/z) α for y i z (otherwise 0) Where P α is the poverty measure, α is a poverty aversion factor, that measures the extent to which society is averse to expenditure levels below the poverty line, and N is the total population. This leads to the following commonly used poverty indices:

4 4 α = 0: poverty headcount α = 1: poverty gap the average shortfall from the poverty line expressed as a proportion of the poverty line α = 2: squared poverty gap which gives greater weights for larger individual shortfalls from the poverty line. The relationship between individual welfare and social welfare of the various values for α are given in Figure 3. Figure 3. The contribution to social welfare of changes in individual expenditure with alternative values of α in the FGT class of poverty measures The P 1 and P 2 measures are a considerable improvement over the headcount, but still give precisely zero weight to welfare above the poverty line. This is clearly still inconsistent with how virtually all societies value income. It would be nuts for a rich country. And, as you have seen, even people in the top parts of the distribution in poor countries are quite poor by global standards for which standards in rich countries are surely the appropriate standards. What does this imply? A more sensible approach would aggregate across individuals using a general social welfare function with positive, but declining weights with respect to income across the whole distribution. Furthermore, national poverty lines are typically higher for richer countries, even when they are held constant over time within a country (as they are in the United States, for example, where it is now about $16

5 5 per person per day for a four-person household). A focus on absolute expenditure poverty is a politically useful way of focusing government and donor attention on the most deprived, who should have highest priority for policy, but is no substitute for a coherent overall approach. Measuring welfare changes across the distribution compensating variation in practice As noted, embedding poverty concepts in a choice-theoretic approach allows standard measures of welfare comparison to be used. Friedman and Levinsohn (2002) develop an empirically implementable application of the compensating variation concept to explore ex ante impacts of large price changes in Indonesia. We want to know how much households would need to be compensated for price changes to restore the same level of utility. Take the expenditure function e(p,u). Then a first approximation of the compensation required is given by the first order Taylor expansion: e x. p But this only applies if the expenditure function is linear. If it is concave, the cost of restoring pre-crisis utility levels will be reduced to the extent households substitute away from goods whose prices rise disproportionately. This implies taking account of the full set of compensated derivatives of demand (s) from the Slutsky matrix. A second-order Taylor expansion provides a further approximation: e x. p + ½ p'.s. p Now this can be applied across the whole distribution of households, with results organized for whatever partition of the data that we are interested in. Friedman and Levinsohn (2002) analyze these proxies for the compensating variation across the distribution of initial expenditures with a rural and urban partition. As just noted, it is useful to look at the whole distribution, that provides us a basis for applying whatever interpersonal ranking we wish, whether these are based on a measure of social welfare or political salience. An exclusive focus on the poor (with your preferred value for α) is just one option. To illustrate an empirical application, Figure 4 reproduces results from the Friedman- Levinsohn study. They implement the formulae given above, based on actual price changes between January 1997 and October 1998 in Indonesia, the initial consumption patterns of households, own-production of food and housing and (for the second formula) the estimated own- and cross-price elasticities of demand. The compensating variation is expressed in terms of the ratio to initial expenditure levels it is the amount needed to

6 6 compensate households for the change in prices, that could then be compared with actual changes. This is calculated with and without substitution effects, using the approximation in the second formula (i.e. with a second-order Taylor expansion) for the latter. Allowing for substitution gives a substantially reduced estimate of CVs across the distribution, though the authors argue this is a lower bound, on the grounds that elasticities from large price changes will be lower than for the marginal changes estimated from the crosssection data. Impacts are lower in rural areas, where some households were protected by own rice production. How to think about policy responses and targeting with heterogeneous impacts will be the subject of a future case focused on Indonesia. Figure 4. Estimates of the compensating variation for rice price increases in Indonesia, contribution Source: Friedman and Levinsohn (2002) Capabilities and multiple dimensions When most economists refer to poverty, they usually mean expenditure or income poverty. Yet income poverty is only one aspect of deprivation. Both poor and non-poor people refer to many features of deprivation, including lack of health, lack of skills,

7 7 insecurity, lack of dignity, and lack of power (see Narayan et al. 2001, for a compendium of field-based research on the views of the poor people on what it means to be poor). This has led to arguments that well-being in general, and poverty in particular, has to be analyzed in terms of multiple dimensions. An influential way of exploring this is in the conceptual work of Amartya Sen on capabilities, or the freedom that people have to lead the kind of lives they value, and have reason to value (see Sen, 1983). This starts from the view that what is of value are functionings, or what people do, as opposed to the commodities they consume. These may range from quite basic considerations, such as being adequately nourished or being free from avoidable disease to more complex ones, such as being able to take part in the life of the community or the ability to appear in public without shame. 3 Income, or the consumption of commodities, may facilitate functionings, but are not, mostly, of intrinsic value. Capabilities then refer to the potential functionings that a person can undertake. It is an ex ante or opportunity concept. This potential will be shaped both by opportunities for private consumption (determined by private wealth and prices of commodities, as above), the public provisioning of services, and the political and social institutional context in which a person lives. Note that an approach based on opportunities is robust to changes in preferences. Preferences change in the course of development, for economic and socio-cultural reasons. Yet the expansion of capabilities, or opportunity sets, is, at least in principle, measurable independent of preferences. Note then there are two core ideas here: that of a focus on what people do; and the idea of choice over possibilities. However, a difficulty with an ex ante concept is measurement, since we actually observe outcomes, not possibilities. Sen himself recommends using outcomes (for example of health and education status) as a way into to assessing capabilities. This takes us to the broader issue of multiple dimensions of well-being and deprivation. Multi-dimensional measures Sen s conceptualization of well-being and so of deprivation includes various aspects of living. This is an example of a broader class of approaches that are multidimensional in character. The issue is actually not whether there are different aspects of well-being, but over how and whether to aggregate these into a synthetic measure. After all, the choice theoretic approach, outlined above, involves obtaining a money metric equivalent of a utility level, that is in turn derived from consumption of a range of goods. In principle these could include the utility derived, say, from leading a healthy life. 3 This last comes directly from Adam Smith. Sen used the term freedom to refer to the concept of capabilities in his later synthesis (Sen, 1999).

8 8 Multi-dimensional approaches involve explicit treatment of alternative aspects of wellbeing. This has a long history. Within development discourse, the ILO in the 1970s argued that development should be organized around realization of a set of basic needs, especially the satisfaction of food, shelter and clothing. The Millennium Development Goals fall directly within this tradition. The latest international attempt is in the preparation for the post-2015 development agenda. While the process within the United Nations General Assembly is ongoing, the report of the High-Level panel has been delivered, and Annex 1 provides their (illustrative) suggestions for goals and targets (United Nations, 2013). Look at this in light of concepts of well-being and deprivation. There have also been various attempts to capture multiple dimensions in an index. This may seem ironic, given that the core idea is that dimensions are multiple. But this has been driven by a concern that the discourse in developmental and political debates is driven excessively by income poverty (and, as we note below, this is thought to have implications for strategy.) The Human Development Index was developed by the UNDP in 1990, as part of their first Human Development Report: it involves equal weights to GNI per capita, life expectancy and a composite measure of actual and expected years of schooling in the population. This was much criticized for conceptual incoherence, especially as it is an ad hoc mixing of different kinds of data, with no particular rationale for the weighting. However, it has been successful in both development and political discourse. It is, for example, calculated at the municipality level in Brazil. A more interesting recent approach, developed by the Oxford Poverty and Human Development Initiative (and also used by the UNDP), is the Multidimensional Poverty Index (MPI). Unlike the HDI, this is built directly from real household-level data, as opposed to societal characteristics, especially from the Demographic and Health Surveys (DHS). Like the HDI it develops indicators for education, health and living standards, but calculates these at the household level. This allows it to calculate the extent to which households have multiple deprivations (that they call the intensity of poverty). It uses measures of assets as proxies for the standard of living (the DHS doesn t have income or expenditure data; and a case can be made that assets directly measure something of interest for living standards.) It uses specific cutoffs for each indicator, and counts a household as poor if it falls below the cutoff on one third of the weighted indicators with one third being arbitrary (but transparent). Figure 5 summarizes the composition of the index.

9 9 Figure 5. Dimensions and indicators in the Multidimensional Poverty Index Source: OPHI It is again easy to criticize this index, at the level of individual indicators and the aggregation. For example, measures of education don t take account of quality. There is no rationale for the weighting scheme. And there are many missing dimensions. However it has the major advantage of building from actual household data, is transparent, makes the important addition of measuring multiple deprivations, and a simple, transparent weighting is probably the best that can be done. The MPI is also attracting broader attention related both to the post-2015 development agenda debates, and the broader interest in measures of well-being beyond income. For an example from last year, the following was reported by OPHI: At last week s lively and stimulating side-event on Multidimensional poverty measurement in the post-2015 development context, the governments of Colombia, Mexico, Chile, the Philippines and Nigeria, along with the World Bank, UNDP and the OECD, all called upon the UN to adopt a new multidimensional poverty measure, the MPI2015+, to track progress toward the new goals adopted after The countries and institutions gathered together: * Recommended that the UN should adopt a new multidimensional poverty measure to complement the $1.25 a day income poverty measure and urged the UN to adopt a new multidimensional framework for poverty reduction worldwide, as part of a new post-2015 development strategy. * Called for the breaking apart of silos on poverty reduction, and for countries and international institutions to adopt a multidimensional, multisectoral approach to poverty reduction, underpinned by multidimensional poverty measures. * Shared concrete ways in which national Multidimensional Poverty Indexes (MPIs), whose indicators reflect national plans and country priorities, are being used for policy coordination, targeting, monitoring, and making visible real progress that income poverty measures overlook.

10 10 * Were united in their calls for better data and better measures of poverty. At the national level, the governments of Colombia, Mexico, the Philippines, Nigeria and Chile called on other nations to embrace new multidimensional poverty measures due to their effectiveness for policy change. They stressed that it is essential to complement any international MPI with nationally adapted indices for poverty reduction. Source: OPHI release, September 2013 As we heard from Sabina Alkire last month, she (and OPHI) are represented on the new UN committee that will be making recommendations on future survey designs. Other measurement issues The following notes are in the interest of provide a quick map of other issues. Missing dimensions: insecurity, disempowerment, humiliation etc An issue with all of the above is a bias toward the measurable. This leaves out aspects of well-being and deprivation that are both of considerable instrinsic value, and play significant instrumental roles in the dynamics of change. We highlight three here. Insecurity. This can include both the risk of physical violence and of adverse economic developments. Risk and uncertainty is, of course, a major domain of economic work with much more to come on this. Insecurity is relevant here as an intrinsic aspect of deprivation part of what it means to be deprived. It is also of great interest for understanding change and the extent of potential deprivation. A significantly larger part of the population may be at risk of falling into poverty because of adverse shocks than those counted as poor at a point in time. Since economic insecurity is typically defined in relation to changes in incomes, it is can be analyzed through use of short-run panel data, where this exists. This will be illustrated later in the semester for the case of Indonesia. Disempowerment. The concept of agency is already a central feature of the capabilities approach, in the sense of the ability to advance goals that one values and has reason to value. This is related to the (often looser) concept of empowerment, or its deprivation. As noted below, this came into the mainstream of development discourse on poverty from the late 1990s on. Humiliation and (in)dignity. The experience of humiliation, or, disrespect, can also be considered to be an intrinsic aspect of deprivation. As already noted above, these have a long tradition, at least from Adam Smith, who considered the ability to go about without shame to be a feature of a decent life. It also relates to a very different philosophical tradition of the nature well-being and deprivation built around the concept of recognition, notably in the work of the philosopher Axel Honneth.

11 11 These areas all raise questions for measurement. It can be practically or conceptually difficult to get measures that can be built from surveys and are also comparable though there is a variety of work on getting such measures. They are of interest both because of the intrinsic and instrumental roles, and precisely because they will often have a different coverage in the population from the more easily measurable income and other dimensions. Subjective measures A very different category of measure concerns asking people how they consider their lives to be. This is a fertile area, that does use systematic surveys, and that can then be compared with other measures of well-being. A good example is the use of Gallup surveys. These includes questions around emotional affect, often referred to as happiness. They also include questions of the type: Where do you stand on a ladder of life on a scale of 1 to 10 from the worst possible to the best possible life? These produce different answers. On a cross-country comparison, both are related to economic measures of well-being, with the latter having a stronger relationship across the range of incomes see Deaton (2013) chapters 1 and 2 for an engaging discussion. A note on the practical preparation of a poverty line While poverty lines are general conceptually dubious, we can t live without them. This raises the question of how you construct them. The core approach is to develop a method of identifying minimum requirements for decent living. As noted above, for low income countries, the most popular method is to anchor this on nutritional requirements, and build up the cost of this on the basis of a typical poor households diet and share of spending on food and non-food (as you can see, this involves iteration). For middle income and richer countries, specific calculations are often made for other necessities though this requires forming a view on what a necessity is. One method of doing this is asking people what they think and an example of this is given in Annex 2 for South Africa. And what about inequality? There is, course, a massive tradition of work on inequality, that is mainly for another time. However, three points are worth highlighting in relation to this survey of poverty concepts. First, the kind of measures of poverty described above can be thought of particular forms of inequality measures since they give specific social welfare weights across the population distribution, that just happen to be zero above a poverty line whether this is in the space of expenditure, education etc. (So, as soon as you reflect, they are particularly weird weights!) Second, the opportunity concept that is central to the capabilities approach to deprivation is related to the deep tradition in philosophical (and societal) thinking around equality of

12 12 opportunity. When we wrote the World Bank s World Development Report on Equity and Development, we also chose an ex ante concept of equity, in terms of equality of opportunity, drawing on these traditions. Third, as Deaton (2013) discusses in his book on The Great Escape, this escape from deprivation, that has been a remarkable feature of the recent historical period, typically affects some people before others both within countries, and, especially, those people lucky enough to live in countries that have gone through major transitions in economic and social welfare. This is a classic Pareto-improving change, and this form of increased inequality is surely a welfare improvement even if it is an illustration of inequality of opportunity, since most people have no say in the country in which they live. Bigger concerns over inequality arise if inequality itself leads to the worse outcomes (as with extractive institutions, in the Acemoglu-Robinson account!). Some debates The measurement of well-being and deprivation sometimes seems as contentious as the choice of strategy. The poverty numbers are almost always politically salient, and hit newspapers. Often debates on measurement are proxies for other debates on interpretation of development processes. Here are three debates. Income poverty versus human development. One debate relates to the contrasting conceptualisations of well-being described above, between the choice-theoretic, money metric approach, and one based around capabilities. This entered general development discourse around 1990 in the contrast between the World Bank s emphasis on income/expenditure poverty and the UNDP s on deprivations related to human development, defined in relation to capabilities, albeit with the deeply imperfect proxy of the HDI. (Amartya Sen was heavily involved in the HDRs, especially in the early years; however he had to be persuaded by Mahbub ul Haq, then HDR director, that the HDI made sense on political grounds.) Now defenders of income poverty would argue (then and now) that they have no problem with explicit treatment of, and intrinsically valuing, education, health and other aspects of well-being, but that is no substitute for a focus on a household s economic resources. (The 1990 WDR on poverty put a great deal of emphasis on education and health, see World Bank, 1990). However, beyond measurement, the debate also reflected distinct views on alternative ways of tackling deprivation: Drèze and Sen (1991) contrasted alternative development paths around their capacity to take public action to provide security (also broadly linked to human capabilities), in terms of what they called support-led security, growth-mediated security and unaimed opulence with the last typifying strategies that go for growth without effective provisioning of social and other basic services. Similarly, when analyzing India s development choices they say:

13 13 We consider...not only the facilities offered important as they are by well-functioning markets and beneficial exchanges, but also the fundamental role of human capabilities, and their dependence on basic education, health services, ownership patterns, social stratification, gender relations, and the opportunity of social cooperation as well as political protest and opposition. Drèze and Sen (1999). We will discuss these strategic choices in the following session on current policy debates in India. Empowerment, security and opportunity. Another debate unfolded within the process of preparation of the World Bank s 2000 WDR on poverty. By this time, mainstream development discourse had increasingly embraced issues of participation, communitybased development, social capital as keys to poverty reduction. (We ll discuss these in greater depth next semester). Participatory appraisal was also in vogue in which investigators went into poor communities and facilitated a variety of group-based and other interactive processes to elicit interpretations of the nature of sources of deprivation from poor people themselves. (What a radical thought!) Issues of insecurity, humiliation and disempowerment were major themes, alongside concerns over work, material resources and services. Ravi Kanbur (of Cornell University), was the initial director of the report, and he and his team developed a diagnosis and associated strategy that put issues of empowerment at the centre, with a framing of development strategy around the trilogy of empowerment, security and opportunity. Now this framing, and the associated analysis, was interpreted as downplaying the central role of growth as a poverty reduction strategy though here growth was something of a watchword for a set of beliefs around market-based, globally integrated, development paths. There were vivid debates within the World Bank, with substantial pushback from core parts of the economist community, and noises of concern from outside notably from the US Treasury, and pro-market and Ministry of Finance types in Latin America and India. Ravi Kanbur resigned in due course, though the final product was for the most part a presentational re-balancing than a substantive shift in position. In framing the approach the sequence was changed to opportunity, empowerment and security.. Where to draw the line? In either an income-poverty or other approach, the question arises of where to draw the line. As argued above, an important conclusion from any serious focus on poverty is that the poverty line is usually a nonsensical concept. This is in two senses: there are rarely discontinuities in well-being; and what is meant to be deprived depends on the social context the norms and expectations of participation, of work and what is meant to have a decent life in a particular society. However, poverty lines clearly do serve important political and administrative purposes. The choice of a line in a particular society is then also an essentially political process. But this can be informed by analysis and not only in the narrow sense of specifying particular nutritional or other needs, but also in drawing the implications of alternative lines. Lant

14 14 Pritchett has emphasized the implications of the dominant focus in the development community on lines (for income poverty, education, sanitation etc.), that are appropriate for very poor countries, and highly inappropriate for middle income and rich countries. Of course, any sensible welfare function will give rising weights to improvements in well-being the more deprived a person is (that s the point of the P 2, or higher order, weighting schemes for example), but both aspirations and the acceptable minimum requirements for decent living rise with the development process. Guatemala and Vietnam We can illustrate some of the concepts introduced above using data from Guatemala and Vietnam. These also provide examples of World Bank poverty assessments that became part of that organization s arsenal of analytical work after 1990 (and both are unusually rich pieces of work.) Both countries emerged from episodes of major conflict in the relatively recent past. In Guatemala s case this involved extensive violent repression by an authoritarian regime, especially during a civil war that lasted between 1960 and Guatemala has a long history of oligarchic rule, with profound divisions between a small elite of largely Spanish descent, and an indigenous majority. Indigenous groups suffered severe, institutionalized discrimination, and were victims of extensive violence during the war. The 1996 Peace Accords formally ended the war, and the country has been a (messy) democracy since. Vietnam, of course, also suffered a major internal war, with the US a major actor on one side. The war ended in 1975, with the communist government becoming the unitary government of the whole of the country. It has remained under one-party rule ever since, but has opened up to international trade and investment in the past two decades. Figure 6 provides the assessment of authority trends for the two countries from the Polity IV data base (that is based on the assessment of political scientists.) Figure 6. Trends in the degree of autocracy and democracy in Guatemala and Vietnam Source: Polity IV.

15 15 The contrasting histories have left different legacies in terms of the measured degree of inequality. Figure 7 provides one measure the Gini index of income inequality (that is intuitively an aggregate index of the distance in incomes between people in the population). It is only one way of measuring inequality, but it illustrates that Guatemala is highly unequal by international standards, while Vietnam is relatively equal. Figure 7. Estimates of the Gini index of income inequality, Guatemala and Vietnam Vietnam Guatemala Source: CIA Both countries have experience aggregate growth in the past two decades. In Guatemala s case, much of this was recovery from earlier declines, and recent growth has been erratic in part due to the 2007 Lehman s crisis. By contrast, Vietnam has experienced East Asian style growth from a very low base. By 2012 it had experienced substantial convergence, but still remained significantly poorer than Guatemala. Figure 7. GDP per capita in Guatemala and Vietnam, (USS$ 2005 at PPP) Vietnam Guatemala Source: World Development Indicators.

16 16 What has this pattern meant for poverty? Figure 8 provides a first look, using the poverty incidence concept (P 0 ), and a standardized poverty line of US1.25 per person per day adjusted for differences in purchasing power. Two things are noteworthy. Poverty fell dramatically in both countries, from an initial level of around half the population, but much more systematically in Vietnam. And by the end of the period, income poverty with this line was almost the same in the both countries, despite lower average incomes in Vietnam reflecting the much lower inequality. In a mechanistic sense, Vietnam has experienced relatively inclusive growth (in the mechanical sense of the poor participating in income increases, with only modest rises in inequality). Figure 8. Incidence of poverty for a line of US1.25 at PPP for Guatemala and Vietnam, % 50% 40% 30% 20% 10% 0% Vietnam Guatemala For Vietnam, Table 1 illustrates the results of using alternative poverty lines, and of the poverty gap (P 1 ) and squared poverty gap (P 2 ) measures. The large reduction in the P 2 measure is particular noteworthy, since this indicates that the gains in poverty reach down to the more severely poor households.

17 17 Table 1. Measures of the incidence, depth and severity of poverty in Vietnam with alternative poverty lines. Source: World Bank Let s now look at measures using the Multidimensional Poverty Index, as prepared by OPHI. These are provided in Figures 9 and 10 for Guatemala and Figures 11 and 12 in Vietnam. While the years don t always line up, owing to different data sources, several patterns emerge. For Guatemala the measured overall extent of deprivation on the MPI is comparable to the US1.25 (at PPP) a day, but substantially lower than the national poverty line. For Vietnam, the MPI is strikingly low at 4% of the population; while the OPHI table is still citing the 2002 poverty incidence, we saw in Table 1 that poverty incidence had fallen to less than 17% by 2008, but still much higher than the MPI measures. In the case of Vietnam schooling, health status, and assets ownership have advanced relatively fast. For (much) more background, see the poverty assessments of Guatemala and Vietnam in World Bank (2003, 2012). Questions You are an adviser in a central development ministry in either Guatemala or Vietnam. (1) Take one of three perspectives of poverty absolute expenditure poverty, restricted capabilities (as proxied by the multi-dimensional approach), or poverty as disempowerment and argue what this means for diagnosis of the country s poverty problem. (2) Articulate two policy directions that this diagnosis implies, that you would recommend to the government. (And note that increasing growth, or improving education is not a policy! You need to specify an area of action for the government.)

18 18 Figure 9 Alternative measures of deprivation, Guatemala Figure10 Deprivations in each indicator of the MPI, Guatemala Figure 11 Alternative measures of deprivation, Vietnam

19 19 Figure 12 Deprivations in each indicator of the MPI, Vietnam References Deaton, Angus The Analysis of Household Surveys. Baltimore and London: the Johns Hopkins Press The Great Escape: Health, Wealth, and the Origins of Inequality. Princeton University Press, Drèze, Jean, and Amartya Sen "Hunger and public action." OUP Catalogue "India: Economic development and social opportunity." OUP Catalogue. Friedman, Jed and James Levinsohn The Distributional Impacts of Indonesia s Financial Crisis on Household Welfare: A Rapid Response Methodology. The World Bank Economic Review, 16:3, Narayan, Deepa, Robert Chambers, Meera Naul Shah and Patti Petesch Voices of the Poor Crying out Change. New York: published for the World Bank, Oxford University Press. Oxford Poverty and Human Development Initiative (OPHI) Country reports on Guatemala and Vietnam. Downloaded September 2013 from Sen, Amartya Commodities and Capabilities. Amsterdam: North-Holland.

20 Development as freedom United Nations A new global partnership: Eradicate poverty and transform economies through sustainable development. The report of the high-level panel of eminent persons on the post-2015 development agenda. New York: the United Nations. World Bank World Development Report Poverty. New York: Oxford University Press World Development Report 2000/01. Attacking Poverty. New York: Oxford University Press Poverty in Guatemala. Washington DC: the World Bank Well Begun, Not Yet Done: Vietnam s Remarkable Progress on Poverty Reduction and Emerging Challenges. Hanoi: the World Bank.

21 Annex 1. The (illustrative) list of goals and targets for the post 2015 development agenda 21

22 22 Annex 2 The (illustrative) list of goals and targets for the post 2015 MDGs, contd Source: United Nations, 2013

23 23 Annex 2. What households say are the main causes of poverty, Guatemala (Table numbers refer to Annex in World Bank, 2003)

24 24 Annex 2. Socially perceived necessities: an example from South Africa In order to construct a direct definition of relative poverty that falls within the parameters of an acceptable standard of living, the definition process can be broken down into five stages. First, a list of possible necessities for an acceptable standard of living is developed; second, the list of possible necessities is incorporated into a survey to explore which items are defined as necessary by a representative sample of the society; third, certain items are identified as socially perceived necessities based on selected criteria; fourth, a poverty threshold is determined (e.g. how many socially perceived necessities need to be lacking in order to be classified as poor ); and finally a decision is then made about whether (and if so, how) to cost out an income level below which people are likely to be deprived based on this definition. Initially, a series of focus groups were undertaken across South Africa, to explore what possessions, services and activities people regarded as essential that each and every person in South Africa should have, have access to, or be able to do, in order to have an acceptable standard of living. Following on from these focus groups, a pilot module was included in a the 2005 national survey to obtain a nationally representative definition of necessities. Finally, a module was included in the 2006 national survey which again included the definitional questions but additionally included measurement questions to ascertain who had and did not have the socially perceived necessities. Main results are below:

25 25