We study whether policy can shift gendered beliefs, norms, and labor market outcomes by exploiting a major expansion of earmarked paternity leave in Denmark. The reform generated large first-stage effects, substantially reallocating leave from mothers to fathers. Using a regression discontinuity design combined with new survey data linked to administrative records, we show that the reform makes parents more supportive of paternity leave, shifts gender-role beliefs in a progressive direction, and reduces perceived differences in childcare ability. The reform also narrows gender gaps in earnings and hours worked. The earnings gap falls by 34pp in the first year following childbirth (during leave) and by 2.8pp in the second year (after leave). These results demonstrate that policy can meaningfully influence beliefs, norms, and gender inequality. On the other hand, earmarking restricts families’ ability to allocate leave freely and lowers leave satisfaction, highlighting a central trade-off inherent in paternalistic policies.