Growth equations: a quantile regression exploration

Abstract

Listen

Translate article

Estimation of the regression quantiles for the growth equation using broadly constituted samples shows evidence of unconditional income convergence for countries in the upper tail of the conditional growth distribution but not for countries in the lower tail. This finding is in contrast with results obtained from ordinary least squares estimation that show no evidence of convergence. The estimated regression quantiles for the conditional growth equation shows that the effect of control variables on the GDP growth rate varies significantly along the conditional growth distribution. These findings are suggestive of the potential information gains associated with the estimation of the entire conditional growth distribution, as opposed to the conditional mean only.