Monitoring of poverty requires timely household budget data. However, such data are not available as frequently as needed for policy purposes. Recently, statistical methods have emerged to predict poverty overtime by combining detailed household consumption and expenditure data with more frequent data collected from other surveys. In this paper we compare poverty predictions for Mozambique using different source data to test the robustness of the predicted poverty statistics. A critical element in this exercise of predicting poverty overtime is the stability of the parameters that determine household consumption. We find that the assumption of stable consumption determinants does not hold for Mozambique during the time period examined. We also examine what drives the resulting predicted poverty statistics. The paper then considers the policy implications of these findings for Mozambique and other developing countries.
Faculty of Social Sciences; Poverty prediction; household survey; Mozambique