A generalization to QUAIDS

Previous research shows that consumers’ response to price and income changes is heterogeneous. In addition, evidence from national data often does not support the classical assumption of one commodity-one price. This paper introduces a data coherent generalization to the quadratic form of the almost ideal demand system (g-QUAIDS) that incorporates the sources of heterogeneity in the demand function and allows for regional price variation. Aggregation over consumers imposes a linearization to the g-QUAIDS that requires a new set of price indices. The results from an empirical study by using microdata from the Household Income and Expenditure Survey of Iran highlight the impact of aggregation bias in relation to the level of aggregation. An investigation of the predictive power of linear versus nonlinear g-QUAIDS in different aggregation levels provides practical recommendations for consumer demand analysis.