How Important are Human Capital, Physical Capital and Total Factor Productivity for Determining Economic Growth in the United States, 1840 - 2000 Scott Baier, Sean Mulholland, Robert Tamura, and Chad Turner 1 November 2005 Abstract This paper presents new data on physical capital at the state level for the United States from 1840 - 2000. After combining this new data with recently created state level human capital and income data, we are able to use standard growth accounting techniques to estimate the contribution of aggregate input growth and total factor productivity (TFP) growth on income growth across the states of the United States. In existing growth accounting literature, limitations on data availability typically confine this type of analysis to cross-country comparisons for relatively short periods of time. TFP, or the residual component, is found to be large relative to the portion of output growth that is explained by measured aggregate input growth. Often the explanation for this result appeals to institutional heterogeneity across countries. The data used in this paper includes measures of human capital, physical capital and output growth from 1840 – 2000, a period that is longer than typical for the existing literature. In addition, as our data is across states of the United States instead of across countries, one would expect less institutional heterogeneity in this study than in others in the literature, and therefore expect a larger fraction of income growth to be explained with measured input growth. We find that that 64% of output growth from 1840 – 2000 is accounted for by input growth. We conduct a variance decomposition to examine the role that aggregate input growth and TFP growth have in explaining the cross-sectional variance of income growth across states. As the results are sensitive to the treatment of the observed correlation between aggregate input growth and TFP growth, we construct an upper and lower bound. We find the variance of TFP can explain between 48% and 88% of the variance of income growth, leaving between 12% and 52% to be explained by aggregate input growth. However, these results are sensitive to initial conditions. Preliminary: Do not quote without permission. 1 Baier and Tamura are affiliated with Clemson University, Mulholland with Moravian College, and Turner with Nicholls State University. Baier and Tamura are also affiliated with the Atlanta Federal Reserve Bank. All views expressed here are the authors’ and not necessarily those of the Federal Reserve Bank of Atlanta or the Federal Reserve System.