2003
Some comments on country-to-country poverty comparisons

Andrea Vigorito

Due to the highly controversial nature of poverty studies, some of the problems arising when establishing international measures are the same as those that are faced when countries establish national poverty lines. The World Bank has advocated making these comparisons according to consumption or income, and in particular, has established a threshold of one dollar per day per person, based on 1985 purchasing power parity. Although it might be useful to resort to income based measures, these on their own are insufficient, as concepts of poverty are becoming more complex and multidimensional. There is now a wide consensus regarding the fact that access to health and education is just as important as income and that in the future, the consensus will probably include empowerment and participation in citizen life.

Introduction

The purpose of this article is to highlight some problems that appear when comparing poverty rates among countries and particularly when trying to establish a common basis on which to make these comparisons. Many of these problems have been widely addressed by experts on the subject. In particular, we will examine whether the threshold proposed by the World Bank of a purchasing power parity (PPP) of one 1985 US dollar—the most commonly disseminated international poverty line at present—is a suitable indicator of poverty in various countries.

This discussion has been arranged into four sections. Section I will discuss the pertinence of establishing an international poverty line. Section II will go on to analyse the context in which it is advisable to make comparisons of well-being among countries and the discussion around the multiple dimensions of poverty will be reviewed. Section III examines how far it is possible to address poverty incidence using a single measure and questions the limits of the income space to evaluate people’s well-being. Since several national studies show that this is not the case, Section IV suggests that an effort should be made to choose a wide array of indicators to rank countries with respect to many different criteria and dismisses the idea of uni-dimensionality. In particular, the data-gathering efforts of Social Watch could be used to strengthen efforts in this direction.

I.       Why establish an international poverty line?

Setting national poverty lines plays an important role in formulating economic and social policies. In this respect, Atkinson (1993) proposed considering the existence of poverty lines as institutions and reported on the relevant role they play in the assessment of the social performance of a country through time. But is it valid to extend this reasoning to an international scale and consequently, is the establishment of a world poverty line going to help assess efforts made to reduce poverty, as proposed for example by the Millennium Development Goals? Without entering into a discussion of the relevance of a particular goal regarding poverty reduction, problems arising when establishing a single, international poverty line will be reviewed. Due to the highly controversial nature of poverty studies, some of these problems are the same as those that are faced when countries establish national poverty lines.

Comparison of poverty among countries can serve many purposes, from assessing the living conditions of people in different regions or countries to the allocation of resources from international financial aid. However, the concentration of efforts to identify the poor often is made without much discussion about the concepts implicit in comparing individual well-being, since most studies focus either on contributing to policy controversies or on the actual design of policies.

Explicit or not, making international comparisons of deprivation among countries requires the establishment of various criteria as a starting point. In particular, it requires deciding whether it is necessary and possible to establish a common poverty line against which all countries can be compared, and determining its characteristics.

Kanbur (2001) has argued very convincingly that this idea of counting the world’s poor on the basis of a common line could be seen as “the finance ministry approach”, which reflects the concern of many institutions about designing policies to struggle against poverty. However, this concern is not perceived in the same way by civil society and leads to confrontations because the perspectives and time horizons for comparison vary. From the standpoint of international bodies, and particularly international financial institutions, establishing international rankings is relevant. However, classifications made on the basis of a single poverty threshold necessarily simplify reality and disregard many factors that are important when seen from a national point of view or from a civil society perspective.

In this respect, the World Bank (WB) has advocated making these comparisons according to consumption or income, and in particular, has established a threshold of one dollar per day per person, based on 1985 purchasing power parity.[1] Supporting this point of view, Ravallion (2002) wrote that the use of national lines to make international comparisons leads to treating differently people or households who have similar real rates of consumption. He also recognises that this extreme line of poverty of the WB is conservative, as people who are considered poor by national standards are not considered so when using this poverty line. Even when the use of comparative poverty lines is agreed upon, should these be absolute or relative? In this respect, the question is if an international measure of poverty should consider absolute lack of income or should also consider inequality of income. Ravallion provides evidence that the WB has prepared relative poverty measures that still rely on a common international poverty line.

Furthermore, Ravallion also indicates that when analysing regions or countries, the WB experts do not use these international estimates, but rely on national poverty lines, something which can be noticed in the country reports prepared by this institution.

One of the criticisms of the WB poverty line is that it does not correspond to a capabilities or basic needs basket (see for example, Pogge and Reddy, 2002). But even setting an international basic food basket would be a very difficult task, especially considering the diversity among regions in meeting their caloric and nutritional needs. Any criteria of this sort would certainly lead to establishing baskets of different monetary values in different countries. Turning the food basket into a poverty line would also generate new problems as the relative prices of non-food goods vary significantly from one country to another.

In Latin America, the Economic Commission for Latin America and the Caribbean (ECLAC) chose a middle of the way path, as its poverty estimates are made on the basis of national basic food baskets, which are established on the basis of expenditure surveys or come from standard baskets according to the country, and are then multiplied by a common Orshansky coefficient to obtain regional poverty lines (ECLAC, 2000). This methodology attempting to harmonise national habits with international comparability criteria has also been questioned regionally because of the apparently arbitrary setting of the Engel coefficient at 0.5, although this value arises from previous research by the organisation.

The WB threshold would probably appear low in middle-income regions where poverty and income inequality are high, as is the case in Latin America, labelled as the most unequal region in the world, particularly in regard to its per capita GDP. Although it is probable that the proportion of people living under this threshold in Latin America is smaller than in vast zones of Asia and Africa, there are nevertheless problems of severe malnutrition in the region and the cost of the basic food baskets established by many countries exceeds the WB threshold. Although an extremely low threshold will identify regions and countries where deprivation is extreme, it will often overlook other places with less acute, but still problematic, poverty. To recognise these less extreme cases we must use additional measures or indexes.

Furthermore, the establishment of a common threshold in terms of income or expenditure disregards the vast differences from country to country of the cost of and access to various services. Thus, if access to public health, housing and education varies from country to country, unsatisfied needs will also vary considerably. Gardiner et al. (1995), in their comparative study of relatively similar countries in the European Union, particularly the United Kingdom and France, show the serious problems arising in attempts to compare income and analyse the impacts of the different health and housing systems on these comparisons. Their results show that estimating these costs correctly led to different rankings among the European countries they considered. This observation must be kept in mind when performing comparisons solely according to income or expenditure.

The assessment of poverty in many countries has evolved toward a broader conception of poverty than the mere lack of income, including the comprehensive—but sometimes vague—idea of social exclusion. The efforts made by the United Nations Development Programme (UNDP), crystallised in the Human Development Index, also point to broadening the dimensions used to evaluate the performance of the different countries (see for example, UNDP, 2002). As concepts of poverty widen and become more complex, why is there still an interest in simplifying indexes and rankings? Although it is true that there is a strong link between social exclusion and poverty, the concept of social exclusion can be valid among groups whose income is not significantly different.

In order to obtain more illuminating comparisons, it might be useful to evaluate simultaneously international poverty rankings and the evolution of poverty measured by national thresholds. Criteria for measuring poverty vary from region to region. While in all the countries of the Americas absolute poverty lines are used, in Europe poverty is measured more often through relative poverty lines that arise from the idea that after having accomplished a reasonable satisfaction of basic needs, the main concern is with gaining access to the society’s resources.

Income and consumption are difficult variables to measure, particularly in developing countries and the WB has contributed to the development of corrective measurement instruments and methodologies. The quality of income data varies significantly from country to country as household surveys use different criteria, cover different areas, and include different sources of income. The populations of different countries also have different propensions towards under-reporting.

Furthermore, using current income alone is a very important source of error, particularly among the poor, whose income is very erratic; in many countries, all or nearly all of poor people’s income is in kind. Although it is therefore recommended that consumption be used as a more reliable indicator, in practice in many countries income is still used, because the countries do not gather periodic data on consumption.

Besides, there is an ongoing debate on the accuracy of the WB methodology used to set the international poverty lines. Pogge and Reddy (2002) question various aspects of these estimates and, in particular, question the way purchasing power parity (PPP) indicators are used. These authors criticise the fact that in the construction of these indexes, the WB uses the price of all the consumer goods, not only those consumed by the poor. This is derived from the fact that relative prices vary considerably from country to country, particularly the prices of tradable and non-tradable goods. While the former do not vary significantly from country to country, the latter do. According to Pogge and Reddy (2002), PPP is estimated weighing the prices incorporated into the index by average consumption. Thus, international comparisons should be based on goods and services consumed by the poor and not on any other type of goods. As a result of this methodological option, and due to recent changes in the estimation procedure, poverty rates are significantly underestimated and poverty trends are misleading. These authors consider that it would be possible to make more accurate international comparisons of poverty by correcting the estimation of PPP. A greater effort is needed in this direction.

In his reply to Pogge and Reddy, Ravallion (2002) acknowledges that PPP estimations present serious problems and that they should try to reflect consumption in lower income households in the distribution, but he does not consider that Reddy and Pogge’s methodological proposals would lead to obtaining better estimates of poverty. In particular, he considers that the PPP calculation methodology proposed by Reddy and Pogge has serious problems.

II.     The multiple dimensions of poverty

Kanbur (2001) points out that, in contrast to twenty-five years ago, there is now a wide consensus regarding the fact that access to health and education is just as important as income and that in the future, the consensus will probably include empowerment and participation in citizen life.

For this reason, the income method needs to be complemented by other dimensions, addressing the quality of life of households or individuals, given that not all individuals have the same rate of converting income into what Sen (1992) has called functionings. This concept refers to the insufficiency of equalising opportunities alone as a mechanism to achieve equity and eradicate poverty. Thus functionings are individual’s abilities to take advantage of opportunities in different areas (being well nourished, access to health services, housing, etc.).

From this approach, it appears that the relationship between income and capabilities is parametrically variable among communities, families and even among individuals in the same family. This is based on two elements. In the first place, it varies according to differences in ages, gender, social roles, geographical location of the household, and other variables over which the individual may or may not have control. Secondly, there may be individual difficulties in converting income into functionings; it is probable that people with different needs and abilities require different amounts of income to achieve the same goals, and inequality may be more intense than what is perceived by looking at the space at income alone.

These considerations point to the fact that the identification of the poor exclusively by the income method leaves out dimensions that may be very relevant when defining the individual’s access to available resources, especially when attempting to make international comparisons.

“If we go on to analyse poverty, the identification of a minimum combination of basic capabilities may be a good way of setting out the problem of assessing and measuring poverty. It may lead to very different results from those obtained when concentrating on the inadequacy of income as a criterion to identify the poor. The conversion of income into basic capabilities may vary greatly among individuals and also among the various societies, so that the possibility of achieving minimally acceptable levels of basic capability may be associated with different levels of minimally adequate income. The standpoint of poverty concentrated on income, based on the specification of income in a ‘poverty line’ that does not vary among individuals, may be very mistaken in the identification and assessment of poverty.” (Sen, 1996, p. 68)

However, this does not mean discarding the idea of establishing an income threshold. In fact:

“As income is not desired for itself, any notion of poverty based on income must refer directly or indirectly to those basic purposes that promote income in its function as a means [to an end]. In fact, in studies on poverty referring to developing countries, the income in the ‘poverty line’ is frequently derived in an explicit way from reference to nutrition standards. Once it is recognised that the relationship between income and capabilities varies among communities and individuals in the same community, it will be considered that minimally acceptable levels of capabilities are variable: they will depend on personal and social characteristics. However, while minimum capabilities can be achieved by means of strengthening the level of income (given the other personal and social characteristics on which capabilities depend) it will be possible (for the specified social and personal characteristics) to identify a minimum adequate income to achieve minimally acceptable levels of capability. Once this correspondence has been established, it will not matter whether poverty is defined in terms of a failure of basic capability or as a failure in obtaining the corresponding minimally adequate income.” (Sen, 1995, p. 69)

Furthermore, in his work comparing India and China, Sen makes clear how two countries having very similar GDP have very wide differences in terms of basic capabilities for survival and education, a fact that also alerts us to the limitations of uni-dimensional comparisons.

III.    Why use income?

The analysis of poverty and inequality made by economists has mainly focused on income and consumption and paid scant attention to other ideas about poverty until very late in the twentieth century. In her analysis of the origins of poverty studies, Ruggeri-Laderchi (2000) suggests that this lack of conceptualisation is typical of poverty studies prepared by economists since the birth of this field of study in Great Britain in the nineteenth century. According to her interpretation, the lack of interest in alternative definitions of poverty is related to the positivist vision predominating in the first poverty analyses, in particular in the studies by Rowntree and Booth, who were more concerned with estimating poverty and finding mechanisms to reduce it than with questioning ideas about the nature of poverty.

For many users and producers of poverty studies, income incorporates other dimensions of welfare, or at least it is considered that the lack of income is sufficiently correlated with deprivation in other dimensions to enable it to serve as a good summary. Furthermore, in regions where poverty and inequality are at extreme levels and worsening, it may be understandable that imperfect approaches such as poverty lines based on consumption and comparisons of welfare based on income are considered by many as sufficiently good shortcuts to deal with the most acute problems.

However, empirical studies carried out by Ruggeri-Laderchi for Peru and Chile make it evident that the other dimensions of poverty are not consistent with income in all cases. The same conclusion can be drawn from the myriad of studies that map basic needs and income poverty or even from the UNDP’s Human Development Index.

Consideration should also be given to the fact that economists find income a very attractive variable because it is conceptually easier to relate to a standard analysis of economic change, making it possible to link evolution of poverty and inequality with the rest of the economy, in particular with the evolution of the labour market. This possibility makes easier the derivation of policy conclusions from poverty analysis (Rius and Vigorito, 2000).

Additionally, the disciplinary imperative of quantifiability is satisfied by income, which can be treated as a continuous variable. This difference between income and other variables that can be used to quantify poverty is very important. The sophistication of techniques available to apply to basic income indicators also helps to give the impression of objectivity that seems so essential to legitimise economic analysis in many areas (Rius and Vigorito, 2000).

IV.     Toward multidimensional poverty assessments

The preceding paragraphs emphasise the need to consider the multiplicity of spaces where the needs of the world’s population manifest themselves when trying to assess poverty levels. Although it might be useful to resort to income based measures, these on their own are insufficient, as concepts of poverty are becoming more complex and multidimensional, as Kanbur (2001) maintains. In turn, the studies showing that the classification of countries differs according to different types of needs measured, also argue for the use of a multiplicity of indicators to assess comparative performance. Very often these results contradict the wishes of policy-makers, for whom uni-dimensional classifications are simpler and therefore more attractive.

Although the attempt at making international comparisons in the space of income should not be abandoned, classifications of countries cannot consider income exclusively. Including other indicators gives a truer picture than using a single, universal measure of absolute poverty. The efforts made by Social Watch in collecting and publishing information should, in conjunction with other sources of international data, help to generate richer classifications, resulting in the systematisation and production of new, more valuable, indicators. Thus, the dimensions of access to drinking water and sanitation, malnutrition, life expectancy at birth, and distribution of income, considered together, offer a good starting point for describing the evolution and current level of living conditions in developing countries.

Bibliography

A. B. Atkinson (1993), The institution of an official poverty line and economic policy, WSP/98, STICERD, London School of Economics.

A. Rius and A. Vigorito (2000), “Sen meets the marketplace of ideas: The Capability Approach and poverty research in Latin America”, paper presented at the Seminar An operalisation of Sen’s capabilities approach, Cambridge University.

A. Sen (1992), Inequality re-examined, Cambridge University Press.

A. Sen (1996), “Vidas y capacidades” in M. Nussbaum and A. Sen (eds.) La calidad de vida, Fondo de Cultura Económica, Mexico DF.

A. Sen (1999), Development as freedom, New York: Anchor Books, Random House.

C. Ruggeri-Laderchi (1999), The many dimensions of deprivation in Peru: theoretical debates and empirical evidence, Working Paper No. 29, Queen Elizabeth House, University of Oxford.

C. Ruggeri-Laderchi (2000), The monetary approach to poverty: a survey of concepts and methods, Working Paper No. 58, Queen Elizabeth House, University of Oxford.

Economic Commission for Latin America and the Caribbean (ECLAC) (2000), Panorama Social de América Latina. Santiago, Chile.

K. Gardiner, J. Hills, J. Falkingham, V. Lechene and H. Sutherland (1995), The effects of differences in housing and health care systems on international comparisons of income distribution, WSP/110, STICERD, London School of Economics.

M. Evans, S. Paugam and J. Prèlis (1995), Chunnel vision: poverty, social exclusion and the debate on social welfare in France and Britain, Discussion Paper WSP/115, STICERD, London School of Economics.

M. Ravallion (2002), How not to count the poor? A response to Reddy and Pogge, mimeo, www.socialanalysis.org.

R. Kanbur (2001), Economic policy, distribution and poverty: the nature of disagreements, working paper, Cornell University.

T. Pogge and S. Reddy (2002), How not to count the poor, mimeo, www.socialanalysis.org.

United Nations Development Programme (UNDP) (2002), Human Development Report 2002.

World Bank (1999), World Development Report 1999/2000.

World Bank (2000), World Development Indicators 2000.

Note:

[1] The methodology for derivation of the World Bank’s international poverty line is described in various documents, among others, World Bank (2000). Basically, it is derived from a study of national poverty lines in over thirty countries, taking those corresponding to the poorest countries.

Andrea Vigorito holds a Master’s degree on Economics from the London School of Economics. At present, she works as a researcher and professor at the Institute of Economics, Faculty of Economic Sciences, University of the Republic (Uruguay). Her main areas of research are focused on poverty and economic inequality.