Abstract

Gender quotas remain controversial despite evidence of their effectiveness in reducing labor market gender inequality. We study how informational narratives about quotas affect support, and how effects depend on pre-existing causal beliefs about inequality. In a pre-registered survey experiment with 2,404 Italian workers and managers, we compare demand-side (discrimination, bias) versus supply-side (participation, confidence, role models) framings. All information increases unincentivized stated support, most strongly under demand-side narratives, but none affects the extensive margin of an incentivized donation, revealing a clear say–do gap. Conditional on donating, however, supply-side framing significantly raises amounts given. Open-ended responses show narratives reshape reasoning primarily among those with diffuse priors (generic cultural explanations). We formalize this in a simple model featuring misalignment costs and tail-driven effects: narrative success depends on the distribution of prior beliefs, which acts as a state variable determining optimal framing across contexts.