Explaining Productivity Differences: the Importance of Black — White Schooling Differences

Using a newly created data set containing real output per worker, real physical capital per worker and human capital per worker for US states from 1840 to 2000, Turner et. al (2007b) analyzed the growth rates of aggregate inputs and total factor producvity (TFP). Given that instiutional differences across states are likely to be smaller than those observed across countries, the result that variation in TFP still accounts for the lion’s share of variation in output growth is somewhat surprising. We continue this line of work by documenting the importance of TFP in explaining cross sectional variation in the levels of log income. We also consider the possibility that one major institutional difference across states, slavery, might explain TFP differences across states. To this end, we create and present a new state level measure of years of schooling by race from 1840- 2000. Again, surprisingly, exploiting this series has very little impact on the upper bound of the fraction of income variation that can be explained by inputs. Thus as in Lucas (1988) and Prescott (1998), the mystery of TFP remains. We do, however, find support for external effects of physical and human capital, as suggested by Romer (1986,1990), Lucas (1988), and Tamura (2002,2006). Past cross-country studies indicate that variation in TFP accounts for a large fraction of the cross-sectional variation in output, whether levels or growth rates are analyzed. In searching for explanations of the TFP variation across countries, researchers often appeal to institutional heterogeneity as an explanation. Using a newly created data set containing real output per worker