Research Database

This paper uses micro data to examine differences in the cyclical variability of employment, hours, and wages for skilled and unskilled workers. Contrary to conventional wisdom, we find that, at the aggregate level, skilled and unskilled workers are subject to essentially the same degree of cyclical variation in wages. That is, relative offer wage differentials between skilled and unskilled workers are acyclical. However, we do find important differences in the patterns of employment and hours variation for skilled vs. unskilled workers when a college degree is used as a proxy for skill. Workers with a college degree have little cyclical variation in employment or weekly hours, while uneducated workers have highly procyclical employment and hours. Thus, we find that the quality of labor input per manhour rises in recessions, thereby inducing a countercyclical bias in aggregate wage measures. We find substantial differences across industries in the cyclical variation of employment, hours, and wage differentials. We interpret these results as indicative of important inter-industry differences in labor contracting.

This study demonstrates that the U.S. equity premium has declined significantly during the last three decades. The study calculates the equity premium using a variation of a formula in the classic Gordon stock valuation model. The calculation includes the bond yield, the stock dividend yield, and the expected dividend growth rate, which in this formulation can change over time. The study calculates the premium for several measures of the aggregate U.S. stock portfolio and several assumptions about bond yields and stock dividends and gets basically the same result. The premium averaged about 7 percentage points during 1926–70 and only about 0.7 of a percentage point after that. This result is shown to be reasonable by demonstrating the roughly equal returns that investments in stocks and consol bonds of the same duration would have earned between 1982 and 1999, years when the equity premium is estimated to have been zero.

This paper documents several advantages associated with using state level consumption data to examine consumption behavior and especially to estimate the Intertemporal Elasticity of Substitution (IES). In contrast to the results of Hall (1988) and Campbell and Mankiw (1989), we provide substantial evidence indicating that the IES is significantly different from zero and probably close to one. Since the overidentifying restrictions of the standard Euler equation are generally rejected, we use these data to explore the nature of these rejections and evaluate an alternative specification of consumer behavior proposed by Campbell and Mankiw (1987, 1989, 1990). We take special care of examining the robustness of our results with respect to problems caused by the mismeasurement of the interest rate. In particular, we identify a common time component in expected consumption growth across states which, under the specifications of the theory, should reflect real interest rate movements. We find that the common time component closely matches the expected real return on Treasury bills as should be expected if the IES is different from zero and if the T-bill rate is an appropriate measure of interest rates.