We consider the collective model of labor supply with marketable domestic production. We first show that, if domestic production is mistakenly ignored, the “collective” indirect utilities that are retrieved from observed behavior will be unbiased if and only if the profit function is additive. Otherwise, in the non-additive case, the direction and the size of the bias will depend on the complementarity/substitutability of spouses’ time inputs in the production process. We then show that, even if domestic labor supplies are not observed, valid welfare comparisons are possible. This identification result generalizes that in Chiappori (1992).