The Effects of Career and Technical Education: Evidence from the Connecticut Technical High School System

9,284 male applications within the bandwidth for graduation; 5,652 for quarterly earnings; 3,612 for age-23-plus earnings. Full applicant pool 25,072 male and 20,983 female applications across 16 schools

Connecticut eighth-grade applicants to the 16-school stand-alone Connecticut Technical High School System, cohorts 2006-2011 (2006-2013 for schooling outcomes); labour market observed 1.5-8.5 years after high school, to roughly age 25-26.

Fuzzy regression discontinuity on a school-by-year admissions score, bandwidth 10 (robust at 6 and 8), with school-by-year and town fixed effects and clustered standard errors; passes McCrary density and covariate balance tests; weighted-average treatment effect across schools and cohorts. Four limits that bound how far the number travels. (1) The estimand is the marginal applicant to an oversubscribed selective technical high school — a student who applied and wanted in, not a randomly assigned adolescent. (2) CTHSS schools spend more per pupil and have lower student-teacher ratios than the counterfactual schools, so the treatment is a bundle. (3) The authors' own heterogeneity result shows the effect is 12% larger where the counterfactual high school offers one SD FEWER CTE electives, so a meaningful share of the estimate is "this school versus that school". (4) The earnings horizon stops in the mid-twenties, so it cannot speak to the lifecycle-decay question at all. Dougherty co-authors this paper, the Massachusetts RD, and the systematic review that grades both.

For males: on-time high-school graduation +10.0 percentage points (SE 3.3) against an 80% control mean; total post-high-school earnings +44.3% (SE 9.3) on a $63,000 base; average quarterly earnings +32.6% (SE 6.7) on a $5,400 base, reported as 35% in the published version; quarters with earnings +1.12 (SE 0.49) on a base of 10. The earnings gain persists: +34.5% (SE 10.1) restricting to ages 23 and above, 33% at ages 23-25 in the published version. College is the cost: -0.48 semesters enrolled (SE 0.19), fading to -0.30 (ns) and to zero on ever-enrolled by 23. Tenth-grade test scores +0.18 SD and ninth-grade attendance +2pp on a 93% base. Two- and three-digit industry fixed effects explain less than one-third of the earnings gain, so most of it is not industry-specific placement. Every estimate for females is small and statistically insignificant.

Low. Assignment is by an admissions-score cutoff among applicants, so students just above and just below the threshold have the same expected distribution of heritable traits. The residual concern is generalisation, not confounding.

Directionally replicates Dougherty (2018, Massachusetts RD) on high-school graduation with a near-identical magnitude (+10pp against +7-10pp) using the same identification family on a different state system, and matches Hemelt, Lenard & Paeplow's +8pp Wake County lottery. The 32-35% earnings result is NOT independently replicated — no other RD or lottery study has produced a comparable administrative-earnings estimate — and Dougherty is an author on two of the three graduation studies, so the replication is partial rather than fully independent.