The great growth debate: A statistical look at Mankiw, Romer, and Weil, versus Islam

This paper takes a purely statistical look at two of the most important empirical growth papers authored by Mankiw et al. [1992] and Islam [1995]. MRW claim that the Solow model is justified only when human capital is added to the regression, while Islam claims that cross-country heterogeneity is the actual culprit. In a statistical sense, the author of this study finds that Islam was correct in the fact that mean heterogeneity does exist in MRW's data. However, after statistical adequacy is achieved, human capital continues to maintain its role as a significant determinant of growth even though the estimates are not robust for one of the two cross-country samples investigated. On the other hand, though Islam's models were not without statistical problems, they continue to maintain their traditional form and his estimates are robust to respecification. This paper also exemplifies the need for objective statistical testing methods in applied work.