How Poor Is Poor?

February 3, 2005
ECONOMIC SCENE
U.N. Aims to Cut Poverty in Half as Experts Wonder How to Measure It
By ALAN B. KRUEGER

ONE of the United Nations' top goals is to cut in half the proportion of people living in extreme poverty by 2015, compared with 1990. The World Bank is responsible for keeping track. Accurately monitoring poverty is essential for knowing whether the goal is achieved and whether antipoverty strategies are working.

But measuring poverty is difficult for a particular country, let alone the world. The movie star Angelina Jolie challenged celebrities at the World Economic Forum in Davos, Switzerland, last week to know "absolutely what they're talking about" when it comes to poverty, yet even experts would have trouble meeting her standard. Knowing the extent of the rise or fall of worldwide poverty is difficult because poverty is not easy to define or measure.

First, establishing a poverty line - or level of consumption below which one is considered impoverished - involves an element of arbitrariness. For many poor families, not having enough money amounts to not having enough food. But there is no particular threshold level of income or expenditures above which people automatically become fully functioning, nourished members of society.

"Poverty lines are as much political as scientific constructions," said Angus Deaton, a Princeton economist and expert on economic development. In such places as different as the United States and India, the poverty line was initially set with reference to minimum standards of food consumption. Yet over time, Professor Deaton noted, the poverty lines in both countries were adjusted to keep pace with overall price inflation, not the price of food or the share of food in the average family's budget. Despite straying from its original conception, the poverty line survived because of its political and administrative usefulness.

The U.N. has set the line for extreme poverty at living on less than $1 a day. This threshold has obvious rhetorical appeal and surely qualifies as extreme poverty by any standard in developed countries; it is also not far off the poverty line used by many of the poorest countries themselves.

Once an international poverty line is set, it must be converted to local currencies. This is trickier than it sounds. Currency exchange rates are inappropriate because most of the items that the poor consume are not traded on world markets. Living expenses are much lower in rural India than in New York, but this fact is not fully captured if prices are converted with currency exchange rates.

To convert the $1 poverty line into foreign currencies, the World Bank uses indexes of "purchasing power parity." Simply put, these indexes reflect the cost of buying a standard bundle of goods in each country.

Although it is desirable to use purchasing indexes, they are not available for all countries and are skewed toward representing the purchases of the wealthiest households, not the poorest, when they are available. Another problem is that the bundle of goods that poor families actually buy varies from country to country because of differences in tastes and availability.

Thus, the $1 poverty line is best viewed as an approximation.

Once the poverty line is set in local currency, the consumption of a representative sample of households must be compared with the line to determine the percent of people getting by on less than $1 a day. (Each household's consumption is spread equally among its members, another leap of faith.)

Again, this is harder than it sounds. The World Bank typically relies on whatever government surveys that countries routinely produce.

But there is no uniform standard in the way countries collect and process their data, which is important because the poverty rate is sensitive to how consumption is measured.

Consider India, home to 33 percent of the world's poor - or 20 percent, depending on how the data are collected.

India was a pioneer in social surveys and has one of the best government statistical agencies in the world. Still, uncertainty shrouds the level of poverty in India. In one experiment, India's national survey organization asked half of the households it surveyed to report their spending on certain items over a 30-day period and half over a seven-day period. Households reported 30 percent higher food consumption per day in the shorter interval, enough to cut the poverty rate in half. It is not certain which measure is more accurate, although follow-up work points toward the longer interval.

Perhaps the best one can hope for is consistency of measurement within countries to detect changes in poverty over time. But continuing past practices can prolong the use of misleading poverty counts that are not comparable across countries. Clearly, there is a need for Latin American countries, which usually measure poverty by income rather than consumption, to collect reliable household consumption data because consumption is a better measure of living standards.

The herculean measurement problems aside, careful research by Shaohua Chen and Martin Ravallion of the World Bank indicates that much progress has been made toward the goal of halving poverty in China and India. But, they found, little progress has occurred in Latin America and Africa, and the former Soviet states are slipping into deeper poverty. Because China and India accounted for 60 percent of the world's poor in 1990, the goal of halving poverty may be achieved a decade from now, even while many regions see no progress.

Despite the progress in China and India, 18 percent of the world's population still somehow survives on less than $1 a day.

The United Nations has recently held a number of brainstorming sessions to gather proposals for the secretary general's report to the General Assembly on achieving the development goals, which will be delivered next month.

An essential prerequisite is to improve poverty statistics and ensure their integrity.

Although the process of setting a poverty line is necessarily political, the task of measuring poverty should be insulated from political influences. The World Bank, however, is an inherently political institution.

Yet no other international body currently has the expertise or resources to monitor worldwide poverty, so it is important for the next president of the World Bank to value and protect the impartiality of the statistical and research staff. The U.N. could also help by working with statistical agencies around the world to develop uniform standards for poverty surveys and then to ensure that their data are adequately documented and publicly archived. To this end, the U.N. could restart its Household Survey Capability Program, which supported statistical offices in developing countries in the 1980's.

This may not be a cause that celebrities are ready to line up for, but improving poverty data will put the world in a better position to monitor progress and evaluate poverty reduction strategies by the time the poverty line is moved up to $2 a day.


Alan B. Krueger is the Bendheim professor of economics and public affairs at Princeton University. E-mail: akrueger@princeton.edu



Copyright 2005 The New York Times Company |