Cash Holdings and Bank Compensation

1. INTRODUCTION Executive pay in banks and the possible incentives it provides for excessive risk taking have been the focus of considerable attention in the wake of the financial crisis. A particular concern is that traditionally compensation has been designed to align management's interests with those of equity holders but not those of creditors or other stakeholders such as taxpayers. From a regulatory perspective, the challenge is to modify compensation design in a way that continues to encourage value creation even as it discourages excessive risk taking that could lead to bank failures. In this article, we offer a simple set of guidelines for this purpose. Our approach, which relies on the use of cash rather than debt or equity as compensation, offers a framework for thinking about the role of cash in a bank's capital structure and for identifying a lower bound on the amount of cash that banks should be required to hold to help reduce the risk of systemic crises. The simplicity and transparency of a cash requirement--as well as the ease with which such a requirement could be made operational--are key. Our objective is to draw on the various properties of cash as part of a bank's assets to furnish us with a benchmark level of cash holdings that is optimal from a regulatory standpoint. Distilled to its basics, our approach is to use cash compensation in banks as a contingent asset of the banks. We propose that incentive compensation in banks involve a substantial cash component; that this component be deferred and placed in an escrow account with a vesting schedule; and that ownership of the account revert to the bank in "stressed" times (subject to creditors' forfeitures), allowing the bank to access this cash to pay down its debt or otherwise bolster its assets. Importantly we do not pin down the absolute size of cash holdings but determine this sum in relation to the bank's equity levels and other parameters; inter alia, as the equity cushion decreases, our proposed cash holding requirement increases. As an alternative to holding more cash, banks can choose to deleverage to bring down the minimum required cash holdings. For "typical" numbers for U.S. banks, we find a cash requirement of around 18 to 25 percent of equity value. However, empirical analysis suggests that the numbers are highly variable depending on the actual asset mix used by a bank at a given point in time; for instance, looking at the years immediately preceding the crisis, we find that cash requirements for many U.S. financial institutions (including those like Fannie Mae and Freddie Mac that would later fail) often exceeded 50 to 60 percent even by late 2006 and early 2007. (1) There is an important, if obvious, caveat to our proposal. Since our analysis focuses on avoiding bank failures in stressed times, the cash holdings we derive will necessarily be more than those required in "normal" times. We regard this as the natural cost of a strategy that aims to reduce the costs of financial system disruption stemming from bank failures. Our proposal is outlined in Section 2; a discussion of its empirical properties follows in Section 3. Section 4 examines the use of deferred cash in compensation and its role in promoting financial stability relative to that of other instruments, such as inside debt, deferred equity, and contingent capital. The model underlying the proposal is presented in Section 5. 2. THE PROPOSAL In Section 5 we derive our minimum cash holding rule in a simple model. We find that a bank's minimum cash C holding must satisfy (1) C [greater than or equal to] (1 - q)D - qE(1 - MES), or, equivalently, that (2) [C/E] [greater than or equal to] (1 - q) [D/E] - q(l - MES), where D is the amount of the banks debt, 1 - q is the potential loss in asset value that would result from a liquidation in stressed times, E is the equilibrium value of the bank's equity (assuming implementation of our proposal), and MES is the marginal expected shortfall of bank equity conditional on the banks being stressed at the time. …