Roland G. Fryer · National Bureau of Economic Research 2012 · randomized field trial · n=84 schools

Vertical versus Horizontal Incentives in Education: Evidence from Randomized Trials

Cited 19 times in the scientific literature.

Level 2 - randomized trial

Randomized field trials (graded by design analogy for economics/education).

OpenAlex W2767075410 · doi:10.3386/w17752 · record verified 2026-08-31

What was done

Randomized field experiments were conducted across 84 urban public schools in two US cities to evaluate how incentive alignment affects student achievement. In Washington, DC, a horizontal incentive system rewarded students across multiple inputs (attendance, behavior, interim assessments, homework, and uniforms). In Houston, TX, a vertical incentive system rewarded multiple stakeholders (students, parents, and teachers) for a single input (math objectives).

What was found

Both models produced large and statistically significant gains on the specific outcomes directly incentivized, though no numerical point estimates or p-values were reported in the abstract. Horizontal incentives improved both math and reading test scores. Vertical incentives focused exclusively on math increased math achievement but led to decreased test scores in reading, science, and social studies.

Why it matters

This study demonstrates that while targeted incentives can boost performance in specific subjects, single-subject multi-agent incentives can distort effort away from untargeted academic areas.

Limits

The abstract omits exact numerical effect sizes, test score changes, and student/teacher sample sizes beyond the school count (84 schools). The two incentive models were tested in different cities with different target inputs rather than in a direct head-to-head factorial trial, which introduces potential contextual confounding. Long-term durability of gains was not reported.

Cited by