Foreign aid and taxation, revisited∗

Existing research into the relationship between foreign aid and taxation in recipient countries has both conceptual and econometric failings. Empirical results based upon observed tax revenues conflate the effect of aid upon tax structures with the effect upon flows of taxable economic activity, most notably the level of imports. The use of variables normalized by GDP is also likely to generate spurious correlations. Additionally, standard panel estimators applied to a heterogeneous dynamic data generating process will recover biased estimates of average effects. However, I show that the more suitable Pooled Mean Group estimator produces unstable estimates, using available data. Monte-Carlo simulation reveals this instability emerges when the assumption of a homogeneous long-run relationship between aid and taxation, is violated.