A Direct Measure of Inefficiency within Couples: Tax Optimization in French cohabiting couples

The literature on household modelling usually assumes the efficiency of household decisions while testable conditions often depend on auxiliary assumptions or are not suited to detect inefficient behavior. In this paper, we suggest a direct measure of inefficiency in cohabiting couples regarding a decision repeated each year: tax filling. In France, cohabiting couples with children are registered as two separate tax units but must allocate each child to only one of these units for the purpose of tax rebates. Using tax registers and simulations for the years 2013 and 2014, we find that around 25% of all of cohabitant couples do not allocate children optimally in their tax returns, which is a direct evidence of inefficiency. We discuss several pathways: cognitive aspects (transaction costs, `simple rule’ bias and inertia) and non-cooperative behavior (related to the lack of binding agreement, or potential asymmetry of information, between partners). We find traces of heuristics (like equal split, when the number of children is even) while transitions point to a large degree of inertia (confirming the existence of both efficient and inefficient types). Inefficient couples tend to separate more and to marry less in the subsequent period.