Statistical models for zero expenditures in household budgets

Household budget surveys are an important source of data on consumer's expenditure, but for certain commodities, most notably tobacco and alcohol, the estimate from the household survey is less than the known consumption total. A model is considered in which the standard tobit specification is supplemented by the operation of a simple binary censor. The tobit model is basically a linear regression model in which non-positive observations of the dependent variable are replaced by zero. Results from the UK Family Expenditure Survey for the fiscal year 1973-1974 and the 1969-1970 Socio-economic Survey of Sri Lanka are used. Most models of misreporting accept that more zero purchases for certain goods are reported than would be the case if all zero purchases represented genuine non-consumption. The tests of this hypothesis all show that there are too few, not too many, zero purchases. Only for tobacco is this not the case; it is true for all the other goods considered in both surveys.