Competition for a limited number of seats in prestigious college generates a “rat-race” equilibrium effects, leading to increased household investment. I construct and estimate a dynamic tournament model where each household chooses the quality and hours of private tutoring as well as hours of student self-study. The model rationalizes the high amount of parental investments in secondary school despite their lower effects on academic achievement. The estimated model is first used to evaluate the impact of household investments on intergenerational persistence of earnings. I find that heterogeneity in parental investment significantly contributes to intergenerational earnings persistence. Removing child efforts amplifies this persistence by 30%, emphasizing the role of self-effort in moderating the intergenerational link. Secondly, by leveraging the number of competitors in the framework, I assess the heterogeneous impact of the rapidly shrinking cohort size on the demand for parental investment.

Using predicted wages from 18 CPS data waves between 1962 and 2019, we use local linear regression estimates of each spouses predicted wage as a function of his/her spouses wage and show how they have changed over time. We look further into issues such as how much of assortative mating on wages can be explained by assortative mating on education and when, in a couples lifetime, wages should be observed. We find strong evidence of assortative mating in both wages and education with assortative mating in education explaining much of the assortative mating in wages. We find that using wages at the time of marriage causes significant measurement problems as they do not reflect the true long-term expected variation in wages. Finally, we find that most of the assortative mating occurs at relatively low values of wages.

We propose easy econometric methods to test for positive assortative mating. Compared to existing methods such as by Siow (2015), our procedure avoids the cost associated with imposing positive assortative mating restrictions, and it allows for more general tests. The test does not require numerical optimization, and the test statistics can be calculated easily. We compare the power of the tests using simulation. The empirical application suggests that the null hypothesis of no assortative mating is rejected for selected years of CPS data that range from 1962 to 2019.

Compulsory public education is a common policy tool to achieve greater participation in education, particularly from marginalized groups. The COVID-19 pandemic brought unprecedented school closures around the globe. Households seek their own mitigation strategy against education loss via private substitutes. We use the COVID-19 response of South Korea as a laboratory for investigating the effects of pandemic-induced school closure on parental investment and educational outcomes. Specifically, we exploit the school-level variations in in-person learning days driven by government and local authorities. Combining administrative data and longitudinal survey, our fixed effects regression identifies that losing 10 additional days implies a 9.5% increase in private education expenditure. In addition, we build and estimate a structural model tournament to quantify the impact of school closure on educational inequalities. A counterfactual experiment suggests that a $450 voucher for private tutoring may compensate for the test score loss caused by a 20-day loss of in-person schooling. Our results shed light on how households make schooling decisions and have implications on the longer-run educational inequality induced by the pandemic. Methodologically, we propose an estimation algorithm that tightly links the natural experiment estimates and the structural model.