Revealed preference tests of collectively rational consumption behavior

To verify the empirical adequacy of a particular household consumption model, it is important to develop efficient tests did can be applied to real-world data. These tests check whether the observed household behavior is "rational" in the sense that it is consistent with the predictions of the model. In this talk, we present different approaches based on revealed preferences to test collective models of household consumption. Testing collective rationality is computationally difficult (NP-hard). In order to overcome this negative result, we introduce mixed-integer programming formulations which can be used for medium-sized datasets. Next, we propose simulated annealing heuristics, which allow for efficient testing of the collective model on large datasets. We present the results of computational experiments with our approaches.
Joint work with: Fabrice Talla Nobibon, Laurens Cherchye, Thomas Demuynck, Bram De Rock and Frits CR Spieksma.