supported by NSF Crant SES 8821943 and by 2 Sloan Foundation Fellowship. 2

This paper offers theory and evidence on the connection between human capital variables and cross country variation in growth rates. Section 2 below presents the outline of a framework that organizes the subsequent In a model that allows for discussion. It conclusions can be simply stated. an explicit research and development activity designed to foster the creation of new goods, simple growth accounting relationships do not hold. In addition to the usual relationships between the rates of change of inputs and outputs suggested by growth accounting, there will be a role for the level of human capital variables in explaining the rate of growth of output and the rate of investment. In a regression equation that tries to estimate separate Toles for both investment and human capital variables in explaining the rate of growth, collinearity may cause the human capital variables not to enter in the equation They should still have explanatory power for investment. The empirical part of the paper (Section 3) focuses exclusively on the implication that the level of a human capital variable like literacy has a distinct explanatory role in cross country regressions for per capita income growth. The theoretical section serves only to motivate this hypothesis, and empirical section can be read independently. Tests of the implications for The conclusions from this investment are postponed for later inquiry. analysis can be summarized as follows: 1) There are results that can be interpreted to mean that the initial level of literacy and its rate of growth are positively related to per capita income growth. However, these results can more plausibly be interpreted to mean that there is substantial mismeasurement in the estimates of the level of income across countries that biases th