This paper analyses the interplay of policy reform and entrepreneurship in a model where investment decisions and policy outcomes are both subject to uncertainty. The production costs of nontraditional activities are unknown and can only be discovered by entrepreneurs who make sunk investments. The policymaker has access to two strategies: 'policy tinkering', which corresponds to a new draw from a pre-existing policy regime, and 'institutional reform', which corresponds to a draw from a different regime and imposes an adjustment cost on incumbent firms. Tinkering and institutional reform both have their respective advantages. Institutional reform works best in settings where entrepreneurial activity is weak, while it is likely to produce disappointing outcomes where the cost discovery process is vibrant. Cross-country evidence that strongly supports such a conditional relationship is presented.