Job Market Paper
Work in Progress
Disparate Impact of Social Safety Net Inclusion on Internal Migrants in Indonesia
Discussion of targeting the poor mostly focuses on the performance of specific methods on the general population, while the performance across groups are rarely examined. This project examines the performance of the proxy-means-testing (PMT) algorithms across internal migrants and locals in Indonesia. Using the same household expenditure sample and household characteristic variables, I reproduce the PMT algorithm adopted by Indonesian central government. I find that conditional on the same true household expenditure level, the PMT algorithm over-predicts internal migrants expenditure comparing to locals. Zooming in to the poor households the central government aims to target, whose expenditure lies in the bottom 25% of the population, the PMT algorithm's over-prediction results in an 8 percentage points increase of exclusion error for migrants than locals. Preliminary analysis indicates that among the PMT variables, household head education and occupation, and residential neighborhood characteristics contribute the most to the over-prediction bias. Future steps will explore whether local labor and residential markets function differently for locals and migrants and how these differences contribute to the PMT algorithm biases.