Developing and Testing Poverty Assessment Tools: Results From Accuracy Tests in Peru

This report presents the results of the accuracy tests of poverty indicators in Peru. It presents:

• An overview of the design of the field research for the accuracy test, and the computation of the applicable poverty line.
• Results on selected poverty indicators from nine regression models.

Further, the paper investigates poverty outreach of six microfinance institutions in Peru and finds that 20-22% of clients of publicly owned banks, rural savings and loan banks, as well as non-government organizations belong to very poor households.The paper concludes that:

• The nine regression models show satisfactory levels of overall accuracy, i.e. the percentage of household correctly predicted as being very poor or not very poor;
• All models estimated by Ordinary Least Squares (OLS) had lower accuracies among the very poor compared to the accuracies among the not very poor;
• Inaccuracies in prediction are not equally distributed over all expenditure percentiles but are systematically higher among the very poor compared to the not very poor.