The effect of corporate governance and managers on the value of companies has received great attention in the recent public debate. In the academic research, this increased attention has been associated with an effort to develop finer conceptual frameworks and analytical techniques to assess how governance and financial characteristics influence corporate policies and profitability. While theoretical models represent a successful approach under specific hypotheses, the econometric analysis of corporate governance and managerial characteristics has proven to be extremely challenging. Because governance and managerial characteristics are equilibrium outcomes largely determined by the firm itself, it is methodologically difficult to separate out their determinants from their consequences to infer causal effects. Since its infancy the empirical corporate governance and corporate finance research has faced this problem, which is often responsible for mixed empirical results. In my dissertation, I adopt a common methodological framework developed in the "program evaluation" literature to shed new light on the effects of governance and managerial characteristics on a variety of corporate policies and, ultimately, firm performance. In particular, I estimate a class of difference-in-differences models deriving the empirical identifications from policy changes that generate "quasi-natural experiments".