Seattle/Denver Income Maintenance Experiment (SIME/DIME)

Summary

SIME/DIME was the largest and most ambitious of the US income-maintenance experiments — and the one whose results most directly killed the political momentum for guaranteed income. The labor-supply findings broadly confirmed New Jersey: work effort declined, but by moderate amounts that would not have made a national program infeasible. The unexpected and consequential result was a sharp increase in marital dissolution among treated couples. This was inconsistent with the intuitive theory that financial security would stabilize marriages, and it landed in the middle of a political debate where guaranteed-income skeptics seized on it as evidence that the program weakened the family. Subsequent re-analyses — by Glen Cain, Douglas Wissoker, and others — found the divorce-rate finding was sensitive to model specification and may not have been a true treatment effect. But by then the political question was settled. SIME/DIME stands as a case study in the political fragility of social-policy experimentation: even rigorous evidence can be undone by a single counterintuitive finding, and the subsequent technical refutation often arrives too late to matter. Methodologically, the experiment remains a high-water mark for ambitious long-duration RCTs in social policy, and its labor-supply estimates continue to inform modern UBI debates.

Research question

"What are the long-run effects of a negative income tax on labor supply, family stability, and human-capital investment — and do effects differ for single-parent vs. two-parent families and across guarantee/tax-rate combinations?"

Methodology

Result

Mixed

Husbands reduced hours by ~7%; wives by ~17%; female heads of single-parent households by ~12%. Marital dissolution rates were notably higher in the treatment arm — roughly 36–42% higher than control among black and white families — a finding that surprised researchers and became politically controversial. Effects on educational enrollment were positive (modestly increased schooling among adolescents in treatment families).

Evidence strength

Strong

Randomized controlled trial with large sample.

Replication status

Partially replicated