Messinis, George (1999) Habit Modification in Consumption: Theory and Evidence. PhD thesis, The University of Melbourne.

Abstract

Since Brown (1952), habit formation models of consumption have assumed that utility is additively separable in durables and non-durables, and memory loss is a unidimensional process. This thesis dispenses with these assumptions to extend the theory of habit formation and introduce habit modification in consumption. Two alternative representative-agent models are proposed: the first allows for a complementarity between durables and non-durables, the second emphasises the role of credit. The thesis concentrates on the latter model where the contractual nature of household credit produces a beneficial externality: credit erases the memory of ‘customary’ consumption and, thus, depletes the habit stock. The effect, however, is asymmetric since it is the commitment to a debt contract that induces a change in consumer behaviour. The model predicts the following: (i) past experience in debt motivates the consumer towards a more forward-looking behaviour; (ii) ‘excess sensitivity’ and ‘excess smoothness’ of consumption are likely to be time-varying, and (iii) increased access to credit will lead to higher levels of consumption in the long-run. Thus, under perfect capital markets, the model suggests that consumer behaviour will approximate that predicted by Hall (1978). The thesis also undertakes an empirical investigation to test the hypothesis of habit modification via credit. It utilises quarterly U.S. data for consumption expenditure on non-durables and services over the period 1959-1997. The study takes a broader view to consider several measures of credit; these include total household credit, consumer credit and interest paid on consumer debt. Testing begins with GMM estimation to see whether the model is consistent with the data in the short-run. The analysis is extended further to examine the validity of the hypothesis in the long-run. Residual-based tests and the Johansen procedure are applied to test whether credit cointegrates with consumption expenditure. Finally, recent econometric techniques are adopted to allow for an unknown structural break in the long-run relation between consumption, disposable income and credit. Overall, the data are consistent with the habit modification hypothesis in both the short-run and the long-run. More precisely, the evidence can be summarised as follows. First, lagged increases in consumer credit help predict changes in consumption. Second, the habit modification effect is asymmetric as predicted by the theory. Third, higher levels of household indebtedness seem to have contributed to a weaker habit effect over the period 1975-1997. Fourth, a long-run relationship exists between consumption expenditure, disposable income and household credit. Fifth, the long-run relation has undergone a structural shift in 1987. Last, it is unclear whether the asymmetry in habit modification extends to the long-run.

Search Google Scholar

Repository staff login