Measuring poverty with noisy and corrected estimates of annual consumption: Evidence from Nigeria

Abstract

An opportunity to improve measurement and modelling of poverty in Africa arises from recent intra‐year panel surveys that observe household consumption in post‐planting and post‐harvest periods. Observing the same household twice lets an intra‐year correlation be estimated, which can be used to form a corrected estimate of annual consumption. The usual approach surveys consumption for just one short period, like a week or month, and extrapolates to an annual total. This may adequately estimate mean annual consumption for samples spread over a year but overstates dispersion. The resulting noise in consumption estimates inflates measures of poverty and inequality and creates misclassification errors that bias logit and probit models of poverty determinants. This study uses data from the 2012/2013 Nigeria General Household Survey panel to show effects on poverty measures of using annual estimates extrapolated from short‐period surveys. With the corrected extrapolation method that uses intra‐year correlations to adjust for inflated variances, Nigeria's poverty headcount rate falls by one half. Hence, much of the poverty measured in cross‐sectional surveys is transient poverty, for which different policy interventions are needed than for alleviating chronic poverty.