Consumer cohorts and purchases of nonalcoholic beverages

Age, period and cohort (APC) variables are included in a demand system that is used to estimate Norwegian purchases of nonalcoholic beverages. To take account of censoring, a two-step method is used. In the first step, the probabilities of purchasing milk, carbonated soft drinks and other soft drinks are estimated by probit models. The APC variables are highly significant. Older cohorts have higher probabilities of purchasing milk and lower probabilities of purchasing carbonated soft drinks than younger cohorts. In the second step, the probability density functions and the cumulative density function are used to correct for censoring. In the corrected demand system, there are positive cohort and negative age effects for milk. These effects suggest that the replacement of older by younger cohorts, in an increasingly older population, will result in reduced per capita purchases of milk. For carbonated soft drinks, there are no cohort or negative age effects, while there are positive age but no cohort effects for other soft drinks. Copyright Springer-Verlag Berlin Heidelberg 2014