This study investigates CEO risk management ability. Using CEO education as a proxy for ability I examine the relationship between CEO education and various types of risk: (1) market risk, (2) credit risk, and (3) operational risk. Propensity score methods are used as a way to deal with the endogenous matching problem which exists in the executive compensation literature. These methods are proposed as an alternative to the managerial fixed effects approaches such as "spell fixed effects" and the mover dummy variable method (MDV). While the managerial fixed effects methods would fail when the explanatory variables of interest are time-invariant, it is possible to capture this variation in managerial effects by using propensity score methods. I find that the effect on the various types of risks varies by the type of risk and by the type and quality of education. Firms with CEOs that have law degrees are associated with fewer operational risk events. While firms with CEOs that have MBA degrees from top business schools are able to manage credit risk better than their peers. Overall, the quality of CEO education matters, and in many cases it is associated with a simultaneous reduction in firm risk and increase in firm value.