Effects of educational assortative matching on the probability of divorce

Abstract

This study aims to measure the impact of different levels of tertiary education within a couple in the likelihood of divorce in Brazil. Brazil’s expansion of secondary and tertiary education over recent decades has reshaped the educational composition of couples, yet its implications for marital stability remain poorly understood. Using the nationally representative PNAD-Contínua rotating panel (2012-2019) with the help of the Ribas and Soares (2008) methodology for tracking individuals, we follow 452,447 married couples across five consecutive quarters. To isolate the effect of educational assortative matching, we estimate propensity scores with a rich set of demographic and economic covariates. A weighted linear-probability model on the matched sample yields an Average Treatment Effect on the Treated of –1.85 percentage points (p-value = 0.008). Considering the baseline divorce probability of 5.2% in our sample, this 1.9 percentage points reduction implies that couples whose schooling differs are about 37 % less likely to separate in the short-run than otherwise similar, education-matched couples. Because most mismatches should remain hypergamous (husband more educated), the result is consistent with gender-traditional norms that reward male economic advantage and dampen conflict. The short observation window, binary treatment definition, limitations on identifying divorce in the dataset and reliance on observable covariates restrict external validity, yet the finding challenges the common view that closing educational gaps will automatically strengthen marital stability. As gender parity in schooling widens, the protective effect of hypergamy may wane, underscoring the need for longer panels and gender-attitude measures in future research.