Seasonality and habit persistence in a life cycle model of consumption

SUMMARY The allocation of non-durable consumers' expenditure over the four quarters of the year reflects choices made by consumers, implying a utility function exhibiting seasonality. Incorporating such a function, the life cycle model fails to explain fully the observed dynamics in UK seasonally unadjusted data. Seasonal habit persistence effects are introduced and found to be significant. Although this generalized model stands up well to testing in a univariate context, it fails to exclude a predictive role for lagged income changes. Seasonality is inherent in many economic time series, but it is rarely considered to be an issue of interest in econometric model-building. Rather than explicitly investigating the economics underlying seasonal variation, modellers typically try to remove its effects either by using seasonally adjusted data or by including seasonal intercept dummies in their equations. The former practice is criticized by Wallis (1974), who shows that it may distort the relationship between variab le the latter assumes that the function simply shifts up or down with the season. However, except for degrees of freedom considerations, there seems to be no a priori reason why the same parameters or functional form should apply at different times of the year: indeed, Gersovitz and MacKinnon (1978) argue that realistic utility, cost or production functions result in models in which seasonal influences cannot be separated out in a simple fashion. In the present study seasonality is treated as one of the features to be explained within the economic model. Thus, in the particular application to non-durable consumers' expenditure, it is argued that the allocation of expenditure over the year reflects seasonal tastes and hence seasonality in the underlying utility function: the implications of this are explicitly investigated. The framework for our analysis is the influential paper by Hall (1978), who derives a randomwalk model for real non-durable consumers' expenditure when rational expectations are combined with a life cycle theory of consumption. Hall does not investigate seasonality at all, and employs seasonally adjusted values in his empirical analysis: although subsequent US authors have examined many aspects of this theory, the type of data used has rarely been