Abstract: | Over the brief interview periods typical of budget surveys (such as the U.K. Family Expenditure Survey) households' recorded expenditures are liable to be misleading indicators of their underlying consumption. Most starkly, a household may not be observed to purchase commodities that it nevertheless consumes. This paper considers the implications of such measurement errors for the estimation of demand systems, and derives a consistent estimator for the special case of linear Engel curves. |