Several recent papers have shown the relevance of collective models for the empirical investigation of household labor supply and consumption. Yet the estimation of collective models in the presence of non-linear budget sets and participation decisions remains a daunting task. This paper compares collective and unitary models on the basis of simulated collective data with income taxation. We distinguish the cases of individual and joint taxation. Estimating the unitary model we obtain strikingly different ‘preference’ parameters depending on the type of taxation. We also obtain substantial differences between predicted adjustments to labor supply following a switch between tax regimes, and hence potentially wide-ranging definitions of revenue-neutral versions of tax reforms. Finally we discuss distortions affecting the welfare analysis of reforms on the basis of unitary estimates when the model generating the data is a collective model. The results suggest that increased efforts should be devoted to the estimation of collective models with taxation.


Pareto optimal allocation, tax reform, simulated data