Abstract:The effect of interest rates, capital goods prices, and taxes on the capital stock is an issue of central importance in economics, with implications for monetary policy, business cycle models, tax policy, economic development, growth, and other areas. For more than 30 years it has been difficult to obtain precise estimates of these effects, and there is little consensus in the profession on their magnitude, despite their importance for both theory and policy. In this paper, we therefore turn to panel data, specifically a newly constructed data set with more than 50 years of firm-level data on the capital stock and with detailed industry-specific data on the interest rate, the price of investment goods, and tax parameters. Using this rich panel data set, we implement recently developed tests for cointegration in panel data. These tests allow us to determine whether the long-run implications of Jorgensonian neoclassical, q, irreversibility, and (s,S) theories are supported by the data. Using the same data, we then use recently developed panel cointegration estimators to assess the quantitative effect of the interest rate, capital goods prices, and taxes on the capital stock. We would like to thank Mark Blanchette for excellent research assistance. Schaller thanks MIT for providing an excellent environment in which to begin this research and the SSHRC for financial support.
IntroductionAccording to neoclassical growth theory, the capital stock is one of the main determinants of the long-run standard of living. In some versions of endogenous growth theory, the capital stock plays an even more important role by influencing the rate of economic growth.According to standard economic theory, the long-run capital stock is determined by the interest rate, taxes, and capital goods prices. The quantitative magnitude of these effects is of crucial importance for many areas of economics, including monetary policy, business cycle models, tax policy, trade, economic development, and growth.Unfortunately, there is little consensus on the magnitude of these effects. For example, Chirinko (1993) concludes that "the response of investment to price variables tends to be small and unimportant relative to quantity variables," while Hassett and Hubbard (2002) conclude that the user cost elasticity is probably between -0.5 and -1.Caballero (1994, 1999) and Schaller (2006) argue that there are serious problems in obtaining unbiased estimates of user cost elasticity from short-run movements in investment, as the great majority of the previous literature has tried to do. Empirical researchers are trying to estimate the elasticity of the demand for capital, but the equilibrium quantity and price depend on both supply and demand. At business cycle frequencies, there are substantial movements in demand. If the supply curve for capital is upward sloping in the short run, as we believe most supply curves are, econometric methods that emphasize high-frequency fluctuations in the data will tend to pick up movements along this supply cur...