Implications of Pareto efficiency for two-agent (household) choice

We study when two-member household choice behavior is compatible with Pareto optimality. We ask when an external observer of household choices, who does not know the individuals' preferences, can rationalize the choices as being Pareto-optimal. Our main contribution is to reduce the problem of rationalization to a graph-coloring problem. As a result, we obtain simple tests for Pareto optimal choice behavior. In addition to the tests, and using our graph-theoretic representation, we show that Pareto rationalization is equivalent to a system of quadratic equations being solvable.