Does Sentiment Explain Consumption?

Carroll, Fuhrer and Wilcox (1994) studied the capacity of consumer sentiment to help explain the behaviour of consumption in the US. Their study was important for at least two reasons. Firstly, since household consumption accounts for about two thirds of US GDP, fluctuations in consumption may result in significant changes in the state of the macroeconomy. It is therefore important to develop models to explain and forecast consumption. In this regard Carroll, Fuhrer and Wilcox (1994) drew attention to the often-neglected variable ‘consumer sentiment’ and its potential role in explaining variations in consumption expenditure. Secondly, their finding that sentiment does indeed have explanatory power for consumption in the US has implications for theories of aggregate consumption, because at least one way of accounting for these findings involves some violation of the simplest certainty equivalence versions of the life-cycle and permanent-income theories. The results of Carroll, Fuhrer and Wilcox (1994) raise important theoretical and empirical issues that deserve careful study. Our contribution to such study is twofold. Firstly, we develop a model that suggests theoretical reasons why consumer sentiment may influence consumption expenditure. Secondly, we consider empirically the question of an independent influence running from sentiment to consumption in the context of Australia using a carefully specified consumption function as the ‘test-bed’ for the analysis. We are motivated to do this because although the Carroll, Fuhrer and Wilcox (1994) analysis is stimulating, it is based on a relatively ad hoc specification of the aggregate consumption function, a specification that arguably suffers from omitted variable bias, something which may seriously weaken their findings. In this paper we find that consumer sentiment does have small independent explanatory power as far as aggregate consumption is concerned even when ‘standard’ macroeconomic variables are allowed to play their full role. JEL Classification: C51, D24. ________________________ This research was supported by MURG 20523147 and we are very grateful for that support. We would also like to thank Don Harding and Elizabeth Webster at the Institute of Applied Economic and Social Research within the University of Melbourne for helpful comments at the outset of our research and Michael Grady and Lynne Stoddart for research assistance during the 2 course of the project. The usual caveat concerning errors naturally applies. *email: [email protected], **email: [email protected].