Designing Regulatory Sandboxes Without Capture

Sandboxes work when entry criteria are public, cohorts rotate, and sunset clauses are real. When incumbents help write rules and never graduate, pilots become marketing devices.

We recommend separating the team that promotes innovation from the team that enforces baseline protections. That separation costs staff time but reduces the warm-glow bias toward friendly applicants.

Metrics should include consumer complaints, not only firm headcounts. A sandbox that generates slick demos yet hides complaint data is signaling marketing success, not policy learning.

Finally, publish lessons even when experiments fail. Failed pilots that disclose reasons save public money; silent failures invite repeated mistakes with prettier slide decks.