Absolute Income Estimates for Households in Low- and Middle-Income Countries

Mark Brewster, Harvard University
June Y. T. Po, Harvard University

Estimation of permanent income from household surveys has been used for poverty surveillance and the examination of causal processes at the household-level in low- and middle- income countries. While previous approaches are appropriate for within-country comparisons, we estimate internationally and temporally comparable household permanent income. We use a latent variable approach to estimate an absolute measure of wealth from survey data detailing asset ownership. We validate the approach using survey data from three countries in the Living Standards Measure Study that has both expenditure and asset information. The wealth estimates have correlations of 0.75, 0.70 and 0.68 with household expenditure reported in LSMS South Africa 1994, Tanzania 2004 and Peru 1994, respectively. Moreover, adding demographic covariates increases the association with expenditure minimally, suggesting robustness in our estimates (R2 ratio = 0.89). Egypt 1988 and 2005 Demographic and Health Surveys were used to demonstrate trends in household wealth across time.