International Monetary Fund


Share This

Poverty, Disconnected

Finance & Development, December 2009, Volume 46, Number 4

Ravi Kanbur

PDF version

Why soup kitchens may be fuller than ever, even as official statistics suggest poverty is coming down


Children collecting water in

ECONOMISTS have long sought to improve on gross domestic product as a measure of growth and well-being. What is needed, many say, is a new way to gauge economic, environmental, and social sustainability.

For those at the bottom of the income pyramid, living on a dollar a day or less, such musings may seem both irrelevant and farfetched. But work by the Commission on the Measurement of Economic Performance and Social Progress—set up by the French government under the leadership of economists Joseph Stiglitz and Amartya Sen—represents the culmination of many years of effort to reduce reliance on per capita income growth or consumption.

Distributional indicators, such as poverty statistics constructed from household income and expenditure surveys, help spotlight the plight of the poor. In some countries, such as India, the announcement of official poverty figures is a major event with significant political and policy implications. And in the past two decades many countries have begun to conduct household surveys aimed at chronicling poverty, with the result that poverty statistics are more widely available across the globe.

What have we learned from the new data? Setting aside the effects of the crises of the late 2000s and looking back two decades from the mid-2000s, the broad facts can be classified into the following stylized patterns (Kanbur, forthcoming). Where there has been no economic growth, poverty has risen. This is true of many African and some Latin American countries. In a large number of countries, including the biggest ones, such as India and China, and even in some African countries, such as Ghana, there has been fast growth by historical standards, and poverty—the percentage of the population below the poverty line—has fallen, as measured by official data.

What is interesting, however, is the disconnect between the optimistic picture painted by these official data on poverty and the more pessimistic view of grassroots activists, civil society, and policymakers more generally. This disconnect does not, of course, lend itself to quantification in the way that official poverty figures are presented. Rather, the evidence is more indirect and qualitative. Examples include the findings of “participatory poverty appraisals” in Ghana and elsewhere, governmental concerns about social unrest in China, the Indian election results of 2004 (where, after a decade of falling poverty according to official data, the ruling party’s “India Shining” slogan was defeated by the opposition’s “Common Man” slogan), and, indeed, the general unease policymakers display when it comes to distributional issues—even in countries that have good performance on poverty numbers.

What is going on? Could it be that official poverty data are misleading? Can poverty on the ground rise when official data report it is falling? There are five reasons why there may well be a disconnect between the seemingly good quantitative evidence that poverty is falling and the widespread concern that things have not really improved.

The numbers game

Consider an economy in which the incidence of poverty has been falling 1 percentage point a year. This is a good rate of decline, especially for an African country. At this rate, depending on the initial poverty level, an economy would be well on its way to achieving the first Millennium Development Goal, which is focused on reducing the incidence of income poverty.

But suppose the population in this economy is growing 2 percent a year. In this case, although the proportion of those living below the poverty line is declining by 1 percentage point a year, the absolute number of poor people is increasing by 1 percentage point a year. This explains why soup kitchens are fuller than ever, there are more street children than ever, and there are more distressed farmers than ever, even though official “headline numbers” suggest declining poverty.

The disconnect is sharpest in economies where poverty incidence is declining relatively slowly and where the population is growing relatively quickly—as in many countries in Africa. But the tendency is present in all economies. Even in China, which has seen a spectacular decline in both the incidence of poverty and the absolute number of poor people in recent years, the rate of decline of poverty incidence is greater than the rate of decline of the number of poor people (Chakravarty, Kanbur, and Mukherjee, 2006).

Capturing the value of public services

Household surveys are excellent at capturing the market value of goods and services bought and sold. Expenditure data generated from respondents are the building block of poverty data in countries like India and Ghana. Over the years, these surveys have also become increasingly better at capturing the value of a number of nonmarket activities, such as production for home consumption.

However, household surveys are not good at capturing the value of public services such as health, education, and transportation. Conceptually, there is no particular difficulty in incorporating these into the standard money metric measures of well-being. Empirically, however, there are severe difficulties in estimating the value of these services for each household.

In any event, this is not the way official statistics are compiled. Of course, the surveys do collect information on the availability and quality of health care, education, water, sanitation, and other services. But there is no integration of the value of these into the income and expenditure measure of well-being from which the poverty rates are calculated.

Consider, then, an economy that is changing from relying primarily on public services to relying primarily on the private sector. Many people will argue that it is precisely such a transformation that will result in higher growth. The household survey data will capture the growing number of transactions in the expanded private sector, but they will not capture the corresponding decline in public services. And that is a problem, because no matter how inefficient, these services have at least some value to poor people.

Because the value of public services is not accounted for in standard household survey measures of well-being, standard official poverty statistics overstate the improvement in well-being throughout the population, including for those at the lower end of the income distribution scale. Hence, the statistics overstate the reduction in poverty resulting from the shift of more activities to the private sector.

Accounting for inequalities within households

Another defining feature of standard household income and expenditure surveys is that all money metric information is collected at the household level. The usual way of converting this information into measures that reflect individual well-being is to divide by household size and assign the per capita household income or consumption of the household to each individual in the household. But as we know, there can be great inequalities within households, with women and children receiving a much smaller share of total household consumption than men.

Accordingly, intrahousehold inequality information is suppressed. For example, an analysis of a specially designed nutrition survey in the Philippines showed that ignoring intrahousehold inequality understated true inequality and poverty by as much as 30 percent (Haddad and Kanbur, 1990).

These findings suggest that the poverty rate reflected in the official statistics is lower than what the true income distribution would show. But we don’t have the data to calculate these differences, leaving us with a disconnect between the (more optimistic) official poverty reduction narrative and reality on the ground.

Poor winners and poor losers

Consider a country where major structural changes are under way. In general, these changes will create winners and losers—in the short and long term. If the poor are all winners, or if there are some poor winners and no poor losers, poverty will decline. But measured poverty may also decline even if a significant number of the losers are poor, because their losses are outweighed by the gains of the other poor. The anguish of increasing poverty among some, perhaps a sizable number, of the poor will not be captured by the national-level decline in poverty. There will be a disconnect between those who focus on official statistics and those whose focus is losers among the poor.

Because national-level poverty data are calculated from snapshot surveys, we cannot test this logic directly. The available panel data do show a marked decline in well-being for a significant proportion of the population, which lends some weak support to the hypothesis. But the literature has not used these data to identify the effects of liberalization or global integration.

However, the observed increasing inequality in the periodic surveys that underpin national poverty data also supports this logic. Certainly, poverty reduction rates across regions within a country vary widely. In Ghana, for example, during the 1990s national poverty declined, but poverty in the north remained stagnant or increased for some measures. In Mexico in the late 1980s and early 1990s, declining poverty at the national level was not reflected in the poor south (Kanbur and Venables, 2007). In other countries, poverty measures with a stronger emphasis on the depth of poverty decreased less, indicating a greater problem among those living well below the poverty line compared with those living close to it (McKay and Aryeetey, 2007).

Death and poverty

All official poverty indices have one feature in common: holding all else constant, the death of a poor person reduces poverty. If a poor person dies, measured poverty goes down!

This does not sit well with our moral compass, but it is an inescapable feature of poverty indices, and the higher level of mortality among poor people means it is an ever-present issue in poverty numbers.

How can we get around this problem and still keep our statistics intact? One answer is to bring mortality rates, or life expectancy, explicitly into the picture (Kanbur and Mukherjeee, 2007). Doing so would allow us to counteract the fact that measured poverty would decline if HIV/AIDS increased mortality among the rural poor. In another, more positive example, reducing infant mortality among the poor will tend to increase measured poverty. Here also, a social evaluation must counteract the statistical effect by considering well-being in all its dimensions, including by looking at life expectancy.

Better measures, better outcomes

For all these reasons, poor people may stand to benefit substantially from a new approach such as the one proposed in October 2009 by the Commission on the Measurement of Economic Performance and Social Progress. The proposed approach would use household surveys more extensively to capture a fuller set of data and paint a much more accurate picture of the living conditions of the poor. It would help authorities in designing policies to help people escape poverty.

Still, this is not enough. Simply producing poverty statistics alongside per capita income will still yield poverty statistics that paint too rosy a picture, because they ignore many of the other issues highlighted in the Commission report—nonmarket services, gender inequalities within households, and non-income dimensions of well-being. There is plenty of work to be done.


Chakravarty, Satya, Ravi Kanbur, and Diganta Mukherjee, 2006, “Population Growth and Poverty Measurement,” Social Choice and Welfare, Vol. 26, No. 3, pp. 471–83.

Commission on the Measurement of Economic Performance and Social Progress, 2009, report; see

Haddad, Lawrence, and Ravi Kanbur, 1990, “How Serious Is the Neglect of Intra-Household Inequality?” Economic Journal, Vol. 100 (September), pp. 866–81.

Kanbur, Ravi, forthcoming, “Globalization, Growth and Distribution: Framing the Questions,” in Equity and Growth in a Globalizing World, ed. by Ravi Kanbur and A. Michael Spence (Washington: World Bank for the Commission on Growth and Development).

———, and Diganta Mukherjee, 2007, “Premature Mortality and Poverty Measurement,” Bulletin of Economic Research, Vol. 59, No. 4, pp. 339–59.

Kanbur, Ravi, and Anthony J. Venables, 2007, “Spatial Disparities and Economic Development, in Global Inequality, ed. by D. Held and A. Kaya (Cambridge: Polity Press), pp. 204–15.

McKay, Andrew, and Ernest Aryeetey, 2007, “Growth with Poverty Reduction, but Increased Spatial Inequality: Ghana over the 1990s, in Determinants of Pro-Poor Growth: Analytical Issues and Findings from Country Cases, ed. by Stephen Klasen, Michael Grimm, and Andy McKay (New York: Palgrave Macmillan).

Ravi Kanbur is a Professor of Economics at Cornell University.


Finance & Development Logo

Write to us

F&D welcomes comments and brief letters, a selection of which are posted under Letters to the Editor. Letters may be edited. Please send your letters to

Free Email Notification

Receive emails when we post new items of interest to you.
Subscribe or Modify your profile