A Sensitivity Analysis of Cross-Country Growth Regressions

A vast amount of literature uses cross-country regressions to find empirical links between policy indicators and long-run average growth rates. The authors study whether the conclusions from existing studies are robust or fragile when small changes in the list of independent variables occur. They find that although"policy"appears to be importantly related to growth, there is no strong independent relationship between growth and almost every existing policy indicator. They also find that very few macroeconomic variables are robustly correlated with cross-country growth rates. They clarify the conditions under which one finds convergence of per capita output levels and confirm the positive correlation between the share of investment in GDP and long-run growth. They conclude that all findings using the share of exports in GDP could be obtained almost identically using the total trade or import share and also that few commonly used fiscal indicators are robustly correlated with growth. Finally, the authors highlight the importance of considering alternative specifications in cross-country growth regressions.