Poverty analysis is generally based on the use of household economic survey data collected over a period of several months. This paper aims to estimate the poverty measurement bias due to the effect of the period of households’ interview. A new method is presented, based on the decomposition of the poverty severity index by estimating a composed error econometric model and using a semi-parametric estimation technique. Empirical validation for the Tunisian case shows that the bias is important and is not constant; it varies from month to month and year to year.
ISSN: 2241-0376 (Online)