We consider a household location choice model conditional on the job location of each spouse, and we question the optimality of the couple household location. We develop a procedure for testing the Pareto-optimality of household location which is robust to many specification errors as well as to the omission of relevant variables under quite general conditions.

The procedure allows both testing Pareto-optimality and estimating household and individual preferences under the null hypothesis (Pareto-optimality). We propose a two-step version of the estimation procedure, based on a minimum distance estimator. The procedure is consistent and easy to estimate compared to full maximum likelihood.

The methodology is applied to the case of Greater Paris region. The results show that Pareto-optimality is rejected in a simple model in which the value of time of each spouse and the spouses’ bargaining powers are constant. However, Pareto-optimality is no more rejected in a richer model in which the spouses’ value of time and bargaining powers depend on each spouse’s age, education, wage and nationality. The method proposed then allows disentangling the determinants of each spouse’s value of time and bargaining powers.