A practical approach for modelbased poverty prediction

Abstract

The objective of this report is to provide practical guidance for producing poverty estimates based on ”light” household surveys. Mathiassen (2005) outlines the theoretical model. A household budget survey is used to estimate a statistical consumption model where a small set of variables are linked to consumption and poverty. These indicators are then collected through light surveys in years where no household budget survey is made available. By combining the light survey indicators and the parameters from the consumption model, poverty rates and their standard errors can be predicted. The report takes the reader through each step of the procedure, from preparing and utilizing the survey datasets, selecting good indicators and predicting the poverty rates, to evaluating the predictions. The SPSS syntax generated by the INE workshops is available at: www.ssb.no/en/int.