Banks rely on risk managers to prevent their employees from making high risk low value investments. Why can't the CEOs directly incentivize their employees to choose the most pro table investment? I show that having a separate risk manager is more pro table for banks and is also socially efficient. This is because there is conflict between proving incentive to choose the most profitable investment and providing incentives to exert effort on those investments. Hence, if the tasks are split between a risk managers who approves the investments and a loan officer (or trader) who exerts effort, then both optimal investment choice and optimal effort can be achieved. I further examine some reasons for risk management failure wherein a CEO may ignore the risk manager when the latter is risk averse and suggests safe investments. As is usually the case before a financial crisis, my model predicts that the CEO is more likely to ignore the risk manager when the risky investments are yielding higher profits.