Comparing Alternative Measures of Poverty: Assets-Based Wealth Index vs. Expenditures-Based Poverty Score

PDF document icon WP-11-123.pdf — PDF document, 466 kB (477,341 bytes)

Author(s): Foreit KGF, Schreiner M

Year: 2011

Comparing Alternative Measures of Poverty: Assets-Based Wealth Index vs. Expenditures-Based Poverty Score Abstract:

How comparable are different measures of poverty?  In the economic literature, poverty is often measured by income or expenditures. National health surveys, such as a Demographic and Health Survey (DHS), do not routinely collect data on household expenditures but rather estimate relative wealth by analyzing household assets and housing characteristics. This paper assesses the predictive validity of using DHS data to assign expenditures-based poverty rankings to households and individuals, using countries for which there are contemporaneous expenditures surveys and a DHS.  The estimates come from a three-step method. First, potential poverty indicators are matched between the expenditures survey and the DHS. Second, a poverty scorecard is constructed based on data from the expenditures survey, using only indicators that appear in both that survey and the DHS. Third, the scorecard is applied to the DHS to produce estimates of expenditures-based poverty.

Three countries were analyzed: Bangladesh, Ethiopia, and Malawi. After matching items between DHS and expenditures surveys, the estimated proportions of people living below the poverty line of $1.25 a day purchasing power parity (PPP) were comparable between the DHS and the expenditures surveys. When we compared individuals’ assets-based wealth quintile with their expenditure-based poverty scorecard quintile, the assets-based wealth quintiles correlated positively with expenditures-based poverty scores.  Marked differences between countries were also found.  Appreciable mismatch between the two poverty measures could be seen, most notably in the case of Ethiopia where more than a third of the individuals in each assets-based wealth index quintile were ranked two or more quintiles above or below that quintile on the expenditures-based poverty scorecard. 

We conclude that assets-based wealth indices do not consistently track well against expenditure-based measures of poverty and should not be used to identify individuals or households living below consumption- or expenditures-based poverty thresholds.  We also recommend that future DHS country surveys harmonize the selection, wording and response categories of questions on household assets and dwelling characteristics with national expenditure surveys, and vice-versa.