Aid and income : another time-series perspective

In a recent article, Nowak-Lehmann, Dreher, Herzer, Klasen, and Martínez-Zarzoso (2012) (henceforth NDHKM) conclude that foreign aid has not had a significant effect on income, based on evidence from panel data potentially covering 131 countries over the period 1960-2006. The present study provides a replication of the empirical results reported by NDHKM. We uncover that NDHKM relied on a regression model that included a log transformation of variables that are not strictly positive. This led to a non-random omission of a large proportion of observations. Furthermore, we show that NDHKM's use of co-integrated regressions is not a suitable empirical strategy for estimating the causal effect of aid on income. Given the nature of the variables and the question under investigation, a Vector Autoregressive (VAR) model can arguably better address the inherent endogeneity problem in the aid-growth relationship. Evidence from a panel VAR model estimated on the dataset of NDHKM, suggests a positive and statistically significant long-run effect of aid on income.