A country's growth of output is identically equal to its ratio of investment to output and the productivity of investment. In "new" growth theory regressions, which include the investment ratio, all other included variables pick up why the productivity of investment differs between countries. This paper converts a "new" growth theory regression equation into productivity of investment equation which allows for the direct testing of the diminishing returns to capital hypothesis of neoclassical growth theory, and to identify the major determinants of differences in the productivity of investment using the general-to-specific model selection algorithm - Autometrics. Nineteen explanatory variables are considered, and export growth, property rights, latitude, and education turn out to be the most important. Eighty-four countries are taken over the period 1980-2011. There is no evidence of diminishing returns to capital across countries, so investment matters for long run growth.
"new" growth theory productivity of investment cross-country growth regressions