1. Mountjoy (2022): Community Colleges and Upward Mobility
Citation: Mountjoy, J. (2022). Community colleges and upward mobility. Quarterly Journal of Economics, 137(4), 2029-2099.
Research question: Do community colleges expand educational attainment and earnings for the marginal student who gains access?
Identification strategy: Mountjoy [2022] exploits variation in access to community colleges created by the historical expansion of the community college system across U.S. counties, using the local density of community college campuses as an instrument for attendance. The strategy identifies effects for students who attend community college because of geographic proximity— the marginal student who would not have enrolled otherwise.
A key innovation is the careful accounting for the "diversion" effect: some students who would have attended four-year universities instead attend community colleges, reducing their educational attainment and earnings. The paper disentangles the effect on students who are diverted downward from the effect on students who are induced upward (those who would have obtained no post-secondary education otherwise).
Key results: For students who would otherwise obtain no post-secondary education (the "upward" compliers), community college attendance increases credential completion rates by approximately 25 percentage points and earnings by approximately 22% relative to the control mean. For students diverted from four-year colleges, attainment and earnings fall modestly.
The net effect of community colleges on the aggregate population is modestly positive, driven by the large earnings gains for upward-mobility students outweighing the small costs for diverted students.
Takeaway: Community colleges generate large upward mobility for students who would otherwise not access post-secondary education, but the aggregate net effect depends critically on the mix of induction and diversion effects.
2. Chetty, Friedman, Saez, Turner, and Yagan (2020): Income Segregation at Elite Colleges
Citation: Chetty, R., Friedman, J. N., Saez, E., Turner, N., and Yagan, D. (2020). Income segregation and intergenerational mobility across colleges in the United States. Quarterly Journal of Economics, 135(3), 1567-1633.
Research question: Do elite colleges provide upward mobility for students from lower-income families, or do they primarily serve students who would have succeeded regardless?
Identification strategy: Using IRS income tax records linked to college attendance records from the College Board and Department of Education, Chetty et al. [2020] construct a new statistic— the "mobility rate" for each college: the share of students who come from the bottom income quintile and reach the top income quintile after graduation. This is a descriptive but powerful measure of institutional mobility.
To move toward causal identification, the paper also exploits quasi-experimental variation from application rules: students who apply to selective colleges but are rejected (and thus attend less selective institutions) provide a comparison group for students who just make the cutoff and are admitted. This regression discontinuity-like variation, while not perfectly clean, supports the causal interpretation of selectivity effects.
Key results: Elite colleges (Ivy Plus) admit a share of students from the top 1% income distribution that is dramatically disproportionate to their share in the college-age population: roughly 16% of Ivy Plus students come from the top 1%, compared to 0.5% from the bottom quintile. This extreme income segregation at elite institutions reflects both differential application rates and admissions preferences.
The paper estimates that conditional on family income, attending an Ivy Plus college increases the probability of reaching the top 1% income distribution by approximately 60% relative to the typical selective college. The Ivy Plus advantage is substantially attenuated when controlling for high school characteristics, suggesting that peer effects and networking— not just signalling— contribute to the premium.
Takeaway: Elite US colleges are highly income-segregated, provide meaningful earnings premia for comparable students, and currently fail to serve as major engines of upward mobility for low-income students.
3. Zimmerman (2019): The Returns to College Admission for Academically Marginal Students
Citation: Zimmerman, S. D. (2019). The returns to college admission for academically marginal students. Journal of Political Economy, 127(6), 2431-2502.
Research question: What are the returns to four-year college admission for students who are on the margin of meeting admission cutoffs?
Identification strategy: Zimmerman [2019] exploits a sharp admissions cutoff in Florida's state university system: students with high school GPA at or above 3.0 are admitted to four-year universities; those below are not. This regression discontinuity design identifies the effect of four-year college admission for students on the GPA margin.
The design is sharp because the GPA cutoff creates a discontinuity in admission probability— students just above 3.0 are significantly more likely to enrol in four-year universities than students just below, conditional on applying. Using administrative data linking high school records to college enrollment, completion, and earnings, the paper traces effects over 10 years post-graduation.
Key results: For male students at the GPA cutoff, four-year college admission increases 6-year graduation rates by approximately 18 percentage points and earnings 10 years after the expected graduation year by approximately 22%. Female students show smaller and less precisely estimated effects, potentially reflecting differential labour market returns to credentials.
The earnings gains are concentrated among students from lower-income families and students who attended low-quality high schools, consistent with four-year college providing the largest human capital gains precisely for those with the fewest alternatives.
Takeaway: Marginal admission to four-year college generates large earnings gains for male students near the admission cutoff, particularly those from disadvantaged backgrounds.
4. Card (1995): Using Geographic Variation in College Proximity to Estimate the Return to Schooling
Citation: Card, D. (1995). Using geographic variation in college proximity to estimate the return to schooling. In L. N. Christofides, E. K. Grant, and R. Swidinsky (Eds.), Aspects of Labour Market Behaviour: Essays in Honour of John Vanderkamp. University of Toronto Press, pp. 201-222.
Research question: What is the causal return to a year of schooling?
Identification strategy: Card [1995] instruments years of schooling with a binary indicator for whether a four-year college was located in the county of the respondent's residence at age 14. The instrument exploits the observation that individuals who grew up near a college faced lower costs of college attendance (lower transport and living costs), leading to higher college attendance rates.
The key identifying assumption is that college proximity is unrelated to earnings potential conditional on observed controls (parental education, race, urban/rural residence). Card provides balance checks showing that proximity is conditionally uncorrelated with observables that predict earnings, supporting the exclusion restriction.
Key results: OLS estimates of the return to schooling yield approximately 7-8% per year of education. The IV estimate using college proximity is approximately 13-16%— again higher than OLS, consistent with Angrist and Krueger's (1991) finding that IV returns tend to exceed OLS returns for plausibly low-ability compliers.
Card interprets the result as consistent with heterogeneous returns: the marginal students who attend college because they grew up near one (the compliers) may face higher financial returns to education because their foregone income is low (they would have otherwise worked in low-wage jobs), making the cost-benefit calculation strongly positive.
Takeaway: Marginal college attendees— those induced to attend by geographic proximity— face high private returns to schooling, consistent with credit constraints binding for low-income students.
Synthesis: What We Know About Returns to Education
Across these four papers, a consistent pattern emerges: IV-based estimates of education returns tend to exceed OLS, and effects are largest for marginal students from disadvantaged backgrounds. This is consistent with a model where (a) ability bias in OLS is smaller than previously thought, (b) returns to education are genuinely higher for constrained students who would not otherwise invest, and (c) credit constraints and geographic access play meaningful roles in educational investment decisions. Community colleges and regional four-year colleges serve as the key access points for upward mobility, while elite college attendance remains heavily segregated by family income.
References
- Card, D. (1995). Using geographic variation in college proximity to estimate the return to schooling. In L. N. Christofides, E. K. Grant, and R. Swidinsky (Eds.), Aspects of Labour Market Behaviour: Essays in Honour of John Vanderkamp. University of Toronto Press, pp. 201-222.
- Chetty, R., Friedman, J. N., Saez, E., Turner, N., and Yagan, D. (2020). Income segregation and intergenerational mobility across colleges in the United States. Quarterly Journal of Economics, 135(3), 1567-1633.
- Mountjoy, J. (2022). Community colleges and upward mobility. Quarterly Journal of Economics, 137(4), 2029-2099.
- Zimmerman, S. D. (2019). The returns to college admission for academically marginal students. Journal of Political Economy, 127(6), 2431-2502.