| Sumario: | The possibility that real dividends may contain enough information to explain observed stock price variability when they are used to evaluate the intertemporal marginal rate of substitution (IMRS) between present and future real per capita consumption is examined. U.S. data from 1889 to 1994 on the Standard & Poor's stock price index, dividends, January production price index, commercial paper rate, and real per capita consumption of nondurables and services are used, and both dividend-based and consumption-based capital asset pricing models are considered. Estimation results suggest that the dividend-based capital asset pricing model (DCAPM) offers a superior explanation of the data than the consumption-based capital pricing model (CCAPM). In particular, the estimates of the relative risk aversion coefficient in the DCAPM are found to be much smaller than the corresponding ones in the CCAPM. In addition, it is shown that a DCAPM generates IMRS that satisfy the mean-variance bounds obtained by Hansen and Jagannathan (1991) with parameter values of approximately 2.5 or less if sampling error is allowed for.
|
|---|