On the consistency of data with bargaining theories

We develop observable restrictions of well known theories of bargaining over money. We suppose that we observe a finite data set of bargaining outcomes, including data on allocations and disagreement points, but no information on utility functions. We ask when a given theory could generate the data. We show that if the disagreement point is fixed and symmetric, the Nash, utilitarian, and egalitarian max-min bargaining solutions are all observationally equivalent. Data compatible with these theories are, in turn, characterized by the property of co-monotonicity of bargaining outcomes. We establish different tests for each of the theories under consideration in the case in which the disagreement point can be variable. Our results are readily applicable, outside of the bargaining framework, to testing the tax code for compliance with the principle of equal loss.