The relative importance of signaling and human capital accumulation in individual education decisions has long been debated by economists. We exploit a nationwide policy experiment in China to provide evidence that a model emphasizing the value of credentials predicts educational attainment decisions far better than a standard model of incremental human capital accumulation. Using a battery of new and old data sets, we exploit the gradual roll-out of a policy which adds a year to the length of primary school in China, and so increases the cost of all educational credentials by one year of forgone wages. We show this increase in cost does not significantly affect post-primary years of education or credential attainment and our very small and relatively precise estimates of the policy's effect on individuals' wages allow us to reject even modest effects more commonly found in the literature on returns to education. We use these results and provide additional evidence to argue that a country’s educational credential structure plays a major role in shaping educational attainment decisions and is thus an important consideration in design of public policy.