Just paste one or more regression outputs — combined into one table automatically.
Scroll down to see how it works
. reg lwage educ exper tenure, robust
(output omitted)
─────────────────────────────────────────
lwage │ Coef. Std.Err. P
────────────┼────────────────────────────
educ │ 0.0891 0.0063 0.000
exper │ 0.0041 0.0017 0.016
tenure │ 0.0143 0.0032 0.000
_cons │ 0.4580 0.0992 0.000
─────────────────────────────────────────
R-squared = 0.159 N = 1,526
. reg lwage educ exper tenure female married, robust
(output omitted)
─────────────────────────────────────────
lwage │ Coef. Std.Err. P
────────────┼────────────────────────────
educ │ 0.0854 0.0058 0.000
exper │ 0.0039 0.0016 0.014
tenure │ 0.0128 0.0030 0.000
female │ -0.2964 0.0358 0.000
married │ 0.0534 0.0301 0.076
_cons │ 0.6271 0.1044 0.000
─────────────────────────────────────────
R-squared = 0.213 N = 1,526
Copy & paste both outputs into tables.pub — they get combined into one table automatically.
Table 1: Robustness — IV, Heckman, and Quantile Estimates
| (1) IV | (2) Heckman | (3) Qreg (p50) | |
|---|---|---|---|
| Education | 0.0921*** | 0.0867*** | 0.0793*** |
| (0.0142) | (0.0068) | (0.0071) | |
| Experience | 0.0038** | 0.0044*** | 0.0036** |
| (0.0018) | (0.0016) | (0.0019) | |
| Female | −0.3012*** | −0.2841*** | −0.2756*** |
| (0.0372) | (0.0345) | (0.0401) | |
| Mills ratio (λ) | 0.142** | ||
| (0.068) | |||
| First-stage F | 42.1 | ||
| N | 1,526 | 2,104 | 1,526 |
| R² / Pseudo R² | 0.148 | — | 0.091 |
Notes: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses. IV instruments education with distance to nearest college.
Table 3: Heterogeneous Returns to Education
| (1) Men | (2) Women | (3) Age < 40 | (4) Age ≥ 40 | |
|---|---|---|---|---|
| Education | 0.0762*** | 0.0893*** | 0.0912*** | 0.0714*** |
| (0.0074) | (0.0081) | (0.0079) | (0.0085) | |
| Experience | 0.0048*** | 0.0031* | 0.0062*** | 0.0021 |
| (0.0019) | (0.0018) | (0.0024) | (0.0021) | |
| Urban | 0.1243*** | 0.1189*** | 0.1354*** | 0.1081*** |
| (0.0312) | (0.0298) | (0.0341) | (0.0287) | |
| N | 842 | 684 | 891 | 635 |
| R² | 0.187 | 0.204 | 0.195 | 0.176 |
Notes: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses. All specifications include tenure and a constant.
Journal of Labor Economics · Vol. 44, No. 2 · 2026
Sarah M. Chen and David A. Rodriguez
Department of Economics, University of Michigan
Abstract
This paper re-examines the returns to education and labor market experience using data from the Current Population Survey (2018–2024). We estimate Mincerian wage equations with progressively richer specifications. Our preferred specification yields a return to education of 8.2 percent per year of schooling. IV estimates using distance to college as an instrument suggest modest upward bias in OLS, consistent with positive selection.
3. Results
Table 1 reports robustness checks using alternative estimators. The IV specification instruments education with distance to the nearest college. The first-stage F-statistic of 42.1 exceeds conventional thresholds for weak instruments. The Heckman selection model addresses potential sample selection, yielding a positive and significant Mills ratio.
Table 1: Robustness — IV, Heckman, and Quantile Estimates
| (1) IV | (2) Heckman | (3) Qreg (p50) | |
|---|---|---|---|
| Education | 0.0921*** | 0.0867*** | 0.0793*** |
| (0.0142) | (0.0068) | (0.0071) | |
| Experience | 0.0038** | 0.0044*** | 0.0036** |
| (0.0018) | (0.0016) | (0.0019) | |
| Female | −0.3012*** | −0.2841*** | −0.2756*** |
| (0.0372) | (0.0345) | (0.0401) | |
| Mills ratio (λ) | 0.142** | ||
| (0.068) | |||
| First-stage F | 42.1 | ||
| N | 1,526 | 2,104 | 1,526 |
| R² / Pseudo R² | 0.148 | — | 0.091 |
Notes: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses. IV instruments education with distance to nearest college.
The quantile regression at the median yields a somewhat smaller education coefficient of 0.079, suggesting that returns are higher in the upper tail of the conditional wage distribution. Across all three estimators, the gender gap remains large and precisely estimated.
3.2 Progressive Specifications
Table 2 presents our OLS estimates with progressive covariate adjustment. Column (1) reports the baseline Mincerian specification. Columns (2)–(4) add demographic controls, industry fixed effects, and region fixed effects. The return to schooling declines modestly from 8.9 to 7.9 percent.
Table 2: OLS Estimates with Progressive Controls
| (1) Baseline | (2) +Demo. | (3) +Ind. FE | (4) +Region FE | |
|---|---|---|---|---|
| Education | 0.0891*** | 0.0854*** | 0.0823*** | 0.0792*** |
| (0.0063) | (0.0058) | (0.0055) | (0.0054) | |
| Experience | 0.0041** | 0.0039** | 0.0035** | 0.0033** |
| (0.0017) | (0.0016) | (0.0015) | (0.0015) | |
| Female | −0.2964*** | −0.2812*** | −0.2743*** | |
| (0.0358) | (0.0341) | (0.0338) | ||
| Married | 0.0534* | 0.0412 | 0.0389 | |
| (0.0301) | (0.0289) | (0.0287) | ||
| Industry FE | No | No | Yes | Yes |
| Region FE | No | No | No | Yes |
| N | 1,526 | 1,526 | 1,526 | 1,526 |
| R² | 0.159 | 0.213 | 0.241 | 0.258 |
Notes: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses.
The gender wage gap is substantial and persistent. The coefficient on Female implies that women earn approximately 27–30 log points less than men across all specifications. The marriage premium is positive but loses significance once we account for industry composition.
3.3 Heterogeneous Returns
Table 3 examines heterogeneity in the returns to education by gender and age. The returns to education are higher for women (0.089) than for men (0.076), and higher for workers under 40 (0.091) than for older workers (0.071). The urban premium is robust across all subgroups.
Table 3: Heterogeneous Returns to Education
| (1) Men | (2) Women | (3) Age < 40 | (4) Age ≥ 40 | |
|---|---|---|---|---|
| Education | 0.0762*** | 0.0893*** | 0.0912*** | 0.0714*** |
| (0.0074) | (0.0081) | (0.0079) | (0.0085) | |
| Experience | 0.0048*** | 0.0031* | 0.0062*** | 0.0021 |
| (0.0019) | (0.0018) | (0.0024) | (0.0021) | |
| Urban | 0.1243*** | 0.1189*** | 0.1354*** | 0.1081*** |
| (0.0312) | (0.0298) | (0.0341) | (0.0287) | |
| N | 842 | 684 | 891 | 635 |
| R² | 0.187 | 0.204 | 0.195 | 0.176 |
Notes: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses. All specifications include tenure and a constant.
4. Discussion
Our estimates align closely with the meta-analytic findings of Card (1999) and the more recent work of Autor (2014). The finding that returns are higher for women than men is consistent with Dougherty (2005), who argues that female college graduates benefit disproportionately from access to professional occupations.
References
Autor, D. H. (2014). Skills, education, and the rise of earnings inequality. Science, 344(6186), 843–851.
Card, D. (1999). The causal effect of education on earnings. Handbook of Labor Economics, 3, 1801–1863.
Dougherty, C. (2005). Why are the returns to schooling higher for women than for men? Journal of Human Resources, 40(4), 969–988.
Mincer, J. (1974). Schooling, Experience, and Earnings. Columbia University Press.
All tables generated with tables.pub — scroll to browse
Paste your output, get a table. Takes about ten seconds.