We extend the sensitivity analysis of cross-country growth regressions of L
evine and Renelt ( 1992) by introducing a semi-parametric formulation of th
eir regression function. Our results differ from theirs in how certain poli
cy variables affect growth rates. We find that distortion variables, such a
s the standard deviation of gross domestic credit and inflation and real ex
change rate distortions, have a robust negative effect on growth. JEL class
ification: O47, C14.