The Stochastic Implications of Permanent Income Hypothesis for US Speculative Traders: Implications for Consumption-Based Asset Pricing

Abstract

This paper examines the stochastic implications of permanent income hypothesis for speculative prices from a sample of economic data from 1967 to 2017 in the United States. One of the standard assumptions of the Consumption-Based Capital Asset Pricing Model (CCAPM)—the time separability of utility—is relaxed in the model specification of Mankiw and Shapiro (1985) and finds that the expected change in earnings has no obvious connection with stock price changes for monthly and yearly data. This finding, while accepting the excess sensitivity of consumption to income, suggests that the past consumption—unconstrained by expected change in income of that period—influences the utility of future consumption. Disposable income and consumption expenditure are highly autoregressive and non-stationary for monthly, quarterly, and yearly time series. The hypothesis that disposable income follows a random walk is clearly rejected for three-time horizons and the consumption is excessively sensitive to income for monthly and yearly data. The rejection of income follows a random walk due to liquidity constraint for quarterly data. The results of impulse response functions question the OLS/AR type of (univariate) regressions used to test the randomness of disposable income and the excess sensitivity. Equity price changes are, however, found to be completely independent from disposable income for frequent observations of income, which suggests that the use of consumption as a variable in capital asset pricing is a subjective assessment. Furthermore, the empirical evidence shows that the equity price changes cannot be effectively forecasted by the predictable change in disposable income.