A latent variable model of consumer taste determination and taste change for complex carbohydrates

Despite evidence that aggregate consumption of complex carbohydrates has risen over the last decade, food consumption surveys suggest that fewer households are consuming less bread, pasta, rice, potatoes and corn. This paper estimates systems of complex carbohydrate demand using cross-sectional data from 1977-78 and 1987-88 in order to explain this paradox. Changes in demand that are not explained by changes in prices or income are explained by variations in taste. Because tastes cannot be directly observed, the paper uses a multiple indicator and multiple cause (MIMIC) model to construct a suitable proxy variable. In the MIMIC approach, the 'indicators' are residuals from a household demand function that includes prices and incomes as explanatory variables, while household and demographic proxies are 'cause' variables. The objectives in applying the MIMIC model are to determine the effect of consumer tastes on complex carbohydrate demand and, comparing cross-sectional survey data from two different periods, test the hypothesis that these tastes change over time. The data consist of US complex carbohydrate prices and expenditures, taken from the USDA Household Food Consumption Surveys in 1977-78 and 1987-88. The results show that structural changes in demand, or those that are not explained by changes in price, income, or the cause variables, led to an increase in complex carbohydrate demand of 5% from 1977-78 to 1987-88. However, changes in demand explained by the cause variables cause demand to fall by 9%.