Income distribution and aggregate demand in the United States economy: An empirical investigation

This dissertation presents a neo-Kaleckian, open economy macroeconomic model that focuses on the simultaneous determination of the profit share of national income and the rate of capacity utilization, and tests this model using annual time-series data for the U.S. economy for the period 1955--2004. The model is summarized in two relationships, an investment-saving equilibrium condition ("IS curve"), which is the equation for capacity utilization, and a producer equilibrium condition ("PE curve"), which is the equation for the profit share. The slopes of both curves are ambiguous in theory, and the econometric analysis focuses primarily on determining these slopes. The two equations are estimated using both instrumental variables (IV) and ordinary least squares (OLS), due to ambiguous evidence about whether there is significant simultaneity bias in the OLS estimates. An unrestricted vector autoregression (VAR) model is also estimated as a sensitivity test and for comparison with previous studies. The data are converted to first differences of natural logarithms for stationarity reasons, and all equations are estimated using general-to-specific methods and rigorous specification tests following the London School of Economics approach to econometrics. Both the IV and OLS estimates (and a VAR model with five lags) show that the profit share has a significant, positive long-run effect on capacity utilization, but capacity utilization does not have a significant long-run effect on the profit share. However, in the short run, capacity utilization has a positive contemporaneous effect on the profit share, which is offset by a negative effect with a one-year lag. Another robust finding is that a measure of U.S. international labor cost competitiveness has a significant, positive long-run effect on the profit share. Some other results differ between the IV and OLS estimates, and may reflect simultaneity bias in the latter. There are no significant structural breaks in these relationships during the sample period, as has been argued by some previous authors. Impulse responses using the VAR model suggest that the dynamics of the system involve dampened oscillations. In addition, the slopes of the two curves are consistent with stability of the long-run equilibrium.