Children and Demand: Direct and Non-Direct Effects

It is universally accepted that children have important effects on household demand patterns. This is usually attributed to the direct effect of children; for example children are food intensive. Alternative inferences are that the observed correlations between children and demand patterns are due to non-direct effects, such as fixed effects, state dependence or intra-household effects. These non-direct effects make the consistent estimation of direct effects problematic. We employ a French family expenditure survey that has a number of unusual features to explore the source of the correlation between children and demands. In a first set of tests, we use a sample of older households (over-55`s) for whom we have the details of their completed fertility and whether or not they currently have children living at home. We consider only those who do not have children currently living at home. If there are only direct effects then the demand patterns of those who have had children should be the same as those who never had children. We find that this is not the case. For the second set of tests, we use a sample of couples aged up to 55 and test for the exogeneity of children variables using background variables as instruments for children. We find that children are not exogenous for some goods. These two findings together cast doubt on the usual practice of identifying direct children effects with the coefficients on the children variables in demand equations.