Markov Property and Excess Sensitivity in Aggregate Consumption

Abstract

The hypothesis that dynamically optimizing economic agents try to maintain smooth consumption streams in the face of rather volatile income fluctuations has been extensively studied ever since Friedman(1957) put forward the permanent income hypothesis. Recently, with the advancement of time series econometrics and with the propagation of the rational expectations principle, this hypothesis has undergone major theoretical and empirical scrutiny. From these efforts several controversial issues have emerged that are conceptually important in understanding aggregate consumption behavior. Among these, we will study two closely related issues in this paper.

The first issue concerns stochastic properties of time series movements in aggregate consumption. This question was first raised by Hall (1978) who has argued that the most important implication of the permanent income hypothesis under rational expectations is that the best forecast of consumption in the next period, given current information, is a fixed function of current consumption only. Hall's own empirical results partially reject the Markov property of aggregate consumption, while using different data Mankiw (1982) could not reject the Markov property. Muellbauer (1983) also rejects the Markov property using U.K. time series data. What are the implications of these empirical results for the intertemporal optimizing behavior of consumers? In particular, do they reject the consumption smoothing hypothesis?