We study the roles of quantity and quality of public education in an OLG model where the working adult cares for her child's education as well as for her elderly parent's consumption, and spends time providing assistance to her parent. First, we identify the optimal quality-quantity mix that a parent chooses as a function of policy (school fees, income tax, pension). Then, we discuss the socially efficient determination of the policy tools. We find that school fees tend to reduce education quantity by boosting working time for both kids and adults, and generally favour the generation of working adults (even though they are the ones paying them), possibly at the expense of a negative impact on the well-being of the young and of the elderly. The opposite is true if education is mostly financed through the income tax. Overall quality is only moderately sensitive to changes in the policy mix.

We estimate the causal effect of per-pupil expenditure on special education identification rates in U.S. school districts. While prior work has examined how categorical funding formula design shapes identification decisions, no study has estimated whether overall district resource levels affect identification rates. Using an unbalanced panel of approximately 11,600 districts from 1991 to 2017, we instrument for per-pupil expenditure with the plausibly exogenous timing of court-ordered and legislative school finance reforms. An event-study analysis confirms these reforms increase per-pupil expenditure with no evidence of differential pre-trends. Our preferred IV estimate indicates a $1,000 increase in per-pupil expenditure raises the share of students identified for special education by approximately 1.1 percentage points, or 8 percent relative to the sample mean. Effects are larger in districts with higher proportions of Black residents, college-educated adults, and higher median incomes. These findings suggest resource constraints meaningfully shape local special education identification decisions.

For decades in many countries, women have attended university in greater shares than men. Yet, men are more likely to enroll in programs with higher returns to education, specifically, those related to STEM. In this paper we explore the effects of increasing tuition on overall enrollment by gender and on the selection of STEM programs by women. We find women consistently enrolled at higher rates than men between 1991 and 2020, with the gap widening over the period from 10 to 16 percentage points. Men were more likely to register in STEM fields. This STEM gap has remained stable in traditional STEM disciplines suggesting systematic gender differences in incentives and behavior, reflecting factors such as men’s stronger engagement with higher-paying non-university jobs, higher expected returns to traditional STEM fields for men, narrower earnings dispersion for women across fields, and gender differences in cost sensitivity and risk aversion.

Gender peer effects in schools are well documented, but it remains unclear whether they reflect classmates’ characteristics, such as gender composition, or endogenous peer effects, where students respond to their peers’ achievement and effort. We study gender-specific endogenous peer effects in mathematics achievement in the final year of compulsory school (grade 9), using Swedish register data with classroom identifiers. To identify endogenous peer effects, we build on methods that use variation in peer-group size. A practical limitation of this approach is that identification is strongest in smaller groups, whereas real-world classrooms are usually much larger. This mismatch may explain why previous applications have often produced weak or statistically insignificant estimates. By exploiting uneven gender composition across classrooms, we obtain smaller effective peer groups and estimate the model using conditional maximum likelihood. We find substantial endogenous peer effects in mathematics achievement: approximately 0.48 standard deviations for girls and 0.20 for boys.