2SLS Estimates of Years of Schooling on Mortality and Health Outcomes
| Outcome | Mortality (1) | Hospitalizations (2) | Self-Reported Health (3) |
|---|---|---|---|
| Panel A: Full Sample | |||
| Years of schooling | 0.033* | −2.274 | −0.011 |
| [0.020] | [2.344] | [0.010] | |
| Sample size | 3,240 | 3,240 | 3,240 |
| Panel B: Seven-Day Donut Sample | |||
| Years of schooling | 0.048 | −4.336 | −0.020 |
| [0.030] | [3.635] | [0.015] | |
| Sample size | 3,114 | 3,114 | 3,114 |
| Panel C: 14-Day Donut Sample | |||
| Years of schooling | 0.045 | −5.796 | −0.027 |
| [0.038] | [4.665] | [0.020] | |
| Sample size | 2,988 | 2,988 | 2,988 |
↵Notes: Individual observations for cohorts born from mid-1944 to mid-1953 in the 1992 census collapsed to the day-of-birth level. Outcome variables are defined as in Table 1. 2SLS regressions instrument for years of schooling with the interaction of AFTER*TREAT, control for AFTER, and include saturated controls for day of birth interacted with AFTER and TREAT. Columns 1–8 report standard 2SLS estimates, while Columns 9–12 use report estimates from two-sample 2SLS. All specifications use a bandwidth of 180 days. Heteroskedasticity-robust standard errors are in parentheses. Significance: *p < 0.10, **p < 0.05, ***p < 0.01.