A note on cross-country growth regressions

A large and growing literature searches for the determinants of economic growth, employing cross-country regression analysis. This approach implicitly assumes that countries possess similar structural characteristics (e.g. production technologies and institutional patterns). Structural differences, be they political, economic, social or other, between countries, therefore, do not condition the growth process. Or if they do, then the effects are randomly distributed with zero mean. If such factors do condition the growth process in a non-random way, then these studies are potentially flawed. The growth process is considered using pooled cross-section, time-series data with fixed- and random-effect econometric techniques-techniques that attempt to accommodate across-country and across-time structural differences. Strong support is found for the convergence hypothesis and for a positive effect of the investment share of real output on economic growth. In addition, it appears that increases in the government's share of output contribute to slower economic growth, but the absolute level of that share does not.