Estimation of household demand systems using unit value data

Many budget surveys present the interesting feature that for a wide range of goods they contain information, not only on expenditures, but also on quantities consumed. This allows the computation of individual unit values for the spending of each household on any good for which this is true. The variation in these unit values will usually result from both geographical variation in prices and from household choice regarding the quality of the goods purchased. Important links exist between quantity and unit value choices. Furthermore, if there is sufficiently precise information on geographical location of households then these links can be used in the estimation of consumer price reactions using unit value data. We develop a method which has the main advantage over alternatives of allowing us to combine appealing budget share specifications with a model of quality choice in a way which is fully consistent with demand theory. We present an application using data from the Czech Family Budget Survey. The demand system covers six categories of food, plus clothing and footwear and is estimated conditionally on expenditures on several other good categories, on durable ownership and on labour market status.