Portal de Periódicos da CAPES

What do we Learn About Consumer Demand Patterns from Micro Data

1993; American Economic Association; Volume: 83; Issue: 3 Linguagem: Inglês

1944-7981

Richard Blundell, Panos Pashardes, Guglielmo Weber,

Energy, Environment, and Transportation Policies

The aim of this paper is to assess the importance of using micro-level data in the econometric analysis of consumer demand. To do this we utilize a time series of repeated cross sections covering some 4,000 households in each of 15 years. Employing a number of different procedures, we conclude that aggregate data alone are unlikely to produce reliable estimates of structural price and income coefficients. However, once certain aggregation factors as well as trend and seasonal components are included, an aggregate model is not necessarily outperformed across all demand equations in terms of forecasting ability. (JEL D12, C52, C31) The purpose of this paper is to develop a complete consumer demand system based on a time series of individual household data and to use it to measure the biases introduced into the study of consumer de- mand behavior when aggregate data are used in place of the appropriate microeco- nomic data. We assess the suitability of ag- gregate data through the impact on income and price elasticities and by evaluating the ability of both micro- and aggregate-based models to forecast aggregate consumer de- mand. The biases, introduced by the use of aggregate data, depend upon the way that household characteristics interact with in- come and price effects and on departures of demand systems from linearity. We explore the structure of microeconomic demand sys- tems and the role of household characteris- tics in the behavior of consumer demand both for the light this may shed on the pattern of future demands and for the im- plications this behavior has for issues of aggregation. Consumer demand patterns typically found in micro data sets vary considerably across households with different household characteristics and with different levels of income. We model this variability by making intercept and slope parameters in the budget-share equations of our demand sys- tem depend on household characteristics and by allowing for nonlinear total log- expenditure terms. In fact, in theory- consistent demand systems it is total ex- penditure rather than disposable income that is allocated across goods. We find that this general framework leads to a well- specified data-coherent demand system, which is a quadratic extension of the popu- lar almost ideal model of Angus S. Deaton and John Muellbauer (1980). The micro- level estimates are shown to be sensitive to the treatment of endogeneity of total expen- diture and to the specification of interaction terms with household characteristics. In ad- dition, this specification is also shown to possess many attractive features for the evaluation of aggregate models.