TESTING THE AUGMENTED SOLOW MODEL

 

Jonathan Temple 

Nuffield College

 

November 3, 1995

 

Abstract

This paper applies robustness ideas from the modern statistics literature to the study of the augmented Solow model. It also tests the model in other dimensions, including sensitivity to measurement error. The main findings are that the speed of conditional convergence is highly uncertain, that technology parameters obtained from the augmented Solow model cannot be trusted, and that the model does not work well when attention is restricted to either the OECD or developing countries. Not only that, the equation for steady state human capital is rejected by the data.