EVALUATING THE SOCIAL PROTECTION PROGRAMS IN PAKISTAN: CAUSAL EVIDENCE FROM QUASI-EXPERIMENTAL APPROACH

Abstract

This paper evaluates the causal role of social protection programs in Pakistan, with special reference to Benazir Income Support Programme (BISP) and Ehsaas Programme using nationally representative household survey data and administrative records to address selection bias and endogeneity employing a multi-method causal inference methodology. In particular, the work applies regression discontinuity design (RDD) based on the Proxy Means Test threshold, instrumental variable (IV) estimation, propensity score matching (PSM), and doubly robust (DR) estimators, together with machine learning methods. Results reveal that program participation has a favorable and significant influence on household wellbeing while estimated treatment results show increases in per capita consumption of around 9-15% and decreases in the incidence of poverty and improvements in education and health outcomes. Heterogeneity study reveals larger affects among poorer and rural households. The results provide strong evidence on the effectiveness of social security programs and have substantial implications for policy design and targeting efficiency in developing economies.