Analyzing Matching Patterns in Marriage: Theory and Application to Italian Data

Social scientists have long been interested in marital homogamy and its relationship with inequality. Yet, measuring homogamy is not straightforward, particularly when we are inter- ested in studying sorting on multiple traits. In this paper, we compare different statistical methods that have been used in the demographic, sociological and economic literature. We show that Separate Extreme Value (SEV) models not only generate a matching function with several desirable theoretical properties, but they are also particularly suited for the study of multidimensional sorting. We use small-scale survey data to study sorting among parents of children attending schools in Naples. Our findings show that homogamy is pervasive: not only men and women sort on age, education, height and physical characteristics, but they also look for partners that share similar health-related behavioral traits and risk attitude. We also show that marital patterns are well explained by a low number of dimensions, the most important being human capital. Moreover, children of parents with a high human capital en- dowment perform better at school, although they report lower levels of subjective well-being and perceived quality of relationship with their parents.