We analyze the simplest possible model of endogenous growth to account for the role of financial development. In our setting, financial development affects productivity and determines the amount of resources subtracted to capital investment. We show that under very general assumptions, the relation between economic growth and financial depth is nonmonotonic, and eventually bell-shaped. We empirically assess our results in a framework that allows to distinguish between long-run and short-run effects. We establish a cointegrating relation and derive the long-run elasticities of per capita gross domestic product (GDP) with respect to employment, the physical capital stock, and financial depth-relying on linear as well as nonlinear models for the finance-growth nexus. We employ the results of the first step estimation to specify an error-correction model and find that there is strong evidence for a nonlinear relationship between financial depth and per capita GDP, consistently with what was predicted by our theoretical model.