Counting the World's Poor: Problems and Possible Solutions

Abstract

The World Bank prepares and publishes estimates of the number of poor people in the world. These numbers, particularly the count of people living below $1 a day, are widely quoted and used by the Bank, the press, and by political leaders around the world. They are the raw material in the debate on whether or not world growth reduces world poverty. This paper discusses how the poverty estimates are constructed, and asks whether they can bear the burden placed on them. One specific difficulty is the use of purchasing power parity (PPP) exchange rates, whose revision induces large changes in poverty estimates for the same countries in the same years. Another area of dispute is the discrepancy in many countries between national accounts statistics, which are used to compute growth rates, and survey estimates, which are used to compute poverty estimates. To a considerable extent, the failure of world poverty to fall in the face of world growth is a failure of household survey data to be consistent with national income data. The details of survey design are also important. In India, changing the reference period for reporting consumption removes around 200 million people, a sixth of the world total, if not from poverty, at least from the poverty counts.

Keywords: poverty, purchasing-power-parity, surveys, national accounts

JEL Classification: I3, O1