Transfers in the gravity equation: The case of foreign aid

This paper presents a theoretical gravity model of trade in which foreign aid is considered as a transfer instead of being part of the trade cost, as it has been previously done in the related literature. We argue that the usual specification leads to invalid out-of-sample predictions, biased coefficients and moreover it ignores heterogeneity. The proposed model is estimated for a sample of 188 countries over the period 1988-2013 using panel fixed effects and PPML techniques and the resulting trade elasticities with respect to aid are compared with those obtained from the traditional specification. The main results show that average effect of one additional US $ of aid is around 0.56$ of total imports according to our model, whereas with the alternative model an average effect of an implausible amount of 11$ of imports is obtained. In addition, a decomposed version of the model provides a new framework to disentangle the political effects of aid from the budget effects. While we consider the case of foreign aid, the modeling framework also applies to the study of other transfer, as for example remittances.