This study provides a critical analysis of the growth regressions in Burnsi
de and Dollar [2000]. First, we analyse the relationship between aid and go
vernment expenditure in a modified neo-classical growth model. One of the m
ain results of the analysis is that while good policies spur growth, they m
ay at the same time reduce the effectiveness of foreign aid. Second, we sho
w that the econometric results in Burnside and Dollar emphasising the cruci
al role of interaction between aid and good policies in the growth process
are fragile, as they are extremely data dependent. Finally, we demonstrate
that the Burnside and Dollar data lend support to the idea that the associa
tion between aid and growth can be approximated by decreasing returns to ai
d. This finding conforms well to regression results in other recent studies
.