Events

Professor Aditya Kuvalekar
Associate Professor
Economics and Public Policy
Indian School of Business

How should one compare assortativity across populations whose type distributions differ? Comparisons across countries, cohorts or regions are confounded: a society can look more sorted simply because its type distributions make sorting easier. We take an axiomatic approach and provide two families of orders. The first ranks matchings by how far apart partners’ types are, using the expected value of a concave, nonincreasing function of the gap; it strictly extends the supermodular order. The second starts from the observation that positive assortative matching is the most assortative matching the type distributions allow, so all such matchings are equally assortative, and measures how far a matching falls short of it. With two types, this yields the marriage literature’s minimum-distance index, which had no characterization. Applied to intergenerational mobility, the rank-rank slope of Chetty et al. belongs to the second family only when the societies compared share the same distributions of parents’ and children’s wealth.