Some million thresholds: Nonlinearity and cross-country growth regressions
This paper examines the robustness of the determinants of economic growth in cross-country regressions allowing for nonlinearity in the specification of the data generating process. The nonlinearity is modelled as regime-dependent parameter heterogeneity, where the regime is determined by the level of the explanatory variable whose robustness we aim to measure. Using a generalization of the procedure in Sala-i-Martin (American Economic Review, 1997), strong evidence of nonlinearity is found for practically all of the variables that are robustly correlated to growth in the linear setting, including those variables which are usually included in most cross-county growth regressions.
|Date of creation:||04 Jun 2003|
|Contact details of provider:|| Postal: 2 Dean Trench Street, Westminster, SW1P 3HE|
Phone: +44 20 3137 6301
Web page: http://www.res.org.uk/society/annualconf.asp
More information through EDIRC