Determinants Of Long-Run Economic Growth: A Cross-Section Empirical Study

Cross-Section empirical study of economic growth flourished due to the development of new growth theory and seminal empirical work from Barro(1991) and Kormendi and Meguire(1985) . Though developing rapidly, empirical study can't avoid the following problems: parameter heterogeneity, nonlinearities, outliers, model uncertainty, endogeneity, measurement error, error correlation and regional spillovers, etcetera. It is a huge challenge to address all of these problems in a single framework. Based on the work by Sala-i-Martin et al.(2004), our paper found out three core determinant variables: initial income, investment price and primary schooling. Then, based on Levine and Renelt(1992) and Sala-i-Martin(1996,1997), we used EBA (Extreme Bound Analysis) to reexamine the robustness of many variables, and found out robust partial correlation between long-run economic growth and 47 out of 130 variables, including gross fixed capital formation, government efficiency, intelligent quality, expropriation risk, life expectancy, democracy, bureaucratic quality, infant mortality, and so on. We also addressed nonlinearities, endogeneity, measurement error, error correlation and regional spillover when addressing model uncertainty.Furthermore, we reexamined twenty-two variables by single regression when holding three core variables, and found that many variables have confident effect on economic growth, such as financial depth, aid, external debt, public investment, private investment, the ownership of media, and so on.