La déclaration des enfants par les couples non mariés est-elle fiscalement optimale ?

Since 1995, fiscal law requires that un-married couples declare each of their children on only one of their two separate fiscal declarations. Thereby, their choice can be the result of an optimization, since the reduction each child implies depends on the family income. Fiscal simulation models generally assume that such an optimization takes place. It is interesting to see whether this assumption is supported by existing data. Such an examination can also contribute to forecasting what would be the impact of a full individualization of personal income tax, or contribute to the understanding of collective decision processes within households. Using the data from INSEE's 1997 Fiscal Revenue Survey, we provide an empirical study of this optimization, which leads to some stylized facts about household behavior. We observe that the "fiscal optimum" is reached by about three out of four un-married couples. This certainly does not result from a random choice. Indeed, the latter would lead to far more non-optimal couples. The losses that non-optimal couples incur are relatively weak. Three types of explanations can be given for this non-optimality : complex household structures (a former union or a large number of children), complex earning structures (non-wage earnings), or, mechanically, the level of earnings themselves