Abstract
A univariate time series analysis of the consumption of beer, wine and spirits in the UK over the period 1964–1995 is presented. The analysis shows that the consumption of beer and wine exhibits stochastic seasonality while the consumption of spirits exhibits deterministic seasonality. Moreover, the three series are found to have stochastic trends. Analysis of the out-of-sample forecasting power of the various models reveals that the model with stochastic trend and seasonality is superior to other models. The results cast doubt on the validity and soundness of the practice of modelling the consumption of alcoholic beverages by assuming deterministic trend and seasonality.