Modeling Collective Rationality : An Nonparametric Test on Experimental Data

We provide a first nonparametric (revealed preference) test of the collective consumption model on the basis of experimental data. By using nonparametric testing tools and experimental data, we avoid the usual problems associated with parametric tests (e.g. non-verifiable parametric structure) and the use of 'real life' data sets (e.g. preference heterogeneity). In addition, our collective rationality test complements the existing nonparametric-experimental evidence on individual rationality. Focusing on dyads, we find that all observed consumption choices are consistent with the nonparametric collective rationality conditions. In fact, the consistency results for the parsimonious 'egoistic' collective consumption model (as a tool for describing dyads' choice behavior) are closely similar to those for the individual rationality model (as a tool for describing individuals' choice behavior). This suggest that for simple consumption decision settings, such as that considered in our experiment, the egoistic model may be useful for practical analysis. Still, our results also suggest that the more general collective consumption model, which accounts for consumption externalities and public consumption, can be useful even for modeling such simple decision settings. In fact, we can interpret that the appropriate model specification also depends on the specific dyad type (e.g. friends or partners; gender composition) and choice setting (e.g. public consumption or not) at hand