Selection Bias in Aid Effectiveness: Evidence From Aid for Trade and Foreign Direct Investment

ABSTRACT

This paper examines whether Aid for Trade promotes foreign direct investment using bilateral panel data over 2005–2019. Conventional estimators show mixed results across donors, with no significant effects for Japanese aid. To address potential selection bias, we employ a Bartik‐style instrumental variable exploiting donor sectoral specialization and recipient revealed needs. The 2SLS estimates reveal substantial heterogeneity. Only Japanese AfT demonstrates strong causal effects, while other donors show no significant relationships when instrumented. The contrast between conventional and IV estimates suggests Japan systematically targets recipients with lower FDI potential, creating negative selection bias that masks true effects in conventional estimators. These findings highlight the importance of addressing endogeneity in aid effectiveness research and suggest donor strategies shape both aid allocation and outcomes.