Note on Assessment and Improvement of Tool Accuracy

This paper summarizes the key issues discussed during the workshop organized by the Developing Poverty Assessment Tools project. The participants identified a number of standards for evaluating the performance of poverty assessment tools.

The paper states that accuracy of a particular set of indicators can be assessed by comparing the poverty status predicted by each potential tool with the true poverty status as established by the benchmark (LSMS) data.

he note outlines key concepts that should be accounted for in poverty assessment tools:

• Accuracy criteria:
  • Total accuracy;
  • Poverty accuracy;
  • Non-poverty accuracy.
• Incidence figures:
  • Actual poverty incidence;
  • Predicted Poverty Incidence.
• Errors:
  • Undercoverage;
  • Leakage;
  • Poverty Incidence Error.

Finally, the paper presents alternative approaches to increase accuracy of the poverty assessment tools. These tools focus on finding indicators that correctly identify people at the low end of the income distribution and include:

• Two-step method: It predicts who the non very-poor are and then eliminates them from the analysis;
• Quantile regression method: Regressions are estimated through different points of the distribution, allowing the researchers to assess the relative importance of different variables;
• Linear probability: This method selects variables based on a linear model with a dependent variable with a binary value;
• Variance ratio method.