Hamid Mehran and Anjan Thakor
How does any firm decide on its capital structure—how much equity (capital) to use, how much debt? And what does research tell us about why banks have so much more financial leverage than other firms? How does this inform capital regulation? This post provides a fresh perspective on these questions, identifying the forces that shape the privately optimal capital structure choices of banks, the manner in which government safety nets distort these choices, and how capital regulation ought to be redesigned in light of these distortions. In particular, we discuss a novel approach (developed in Acharya, Mehran, and Thakor ) to capital regulation that involves a two-tier capital requirement as well as how such a requirement can enhance banking stability.
Why Do Banks Have High Leverage?
Academic corporate finance enters this debate with the famous Modigliani and Miller (1958) leverage-indifference theorem. In a world without frictions (no taxes, no bankruptcy costs, no deposit insurance or other safety nets), Modigliani and Miller show that the capital structure decision, for a given size of the firm and given asset portfolio composition, doesn’t affect firm value and is thus irrelevant.
The real world, of course, looks very different from that of Modigliani and Miller, particularly for banks. The standard argument against applying the Modigliani and Miller theorem to banking is that deposits are a factor of production in banking—banks not only use deposits to make loans but also to provide liquidity and transaction services to depositors. Thus, we should expect banks to be highly levered since deposits are a form of debt. However, with finite (constrained) core deposit supply, it isn’t obvious why banks cannot add large amounts of equity to the deposits they gather, which would allow them to gather all the deposits they can and still achieve high capital ratios.
A second popular argument is that banks prefer high leverage because debt interest payments are tax deductible but shareholder dividends aren’t. This is true, but it can’t explain why banks are more levered than nonfinancial firms that enjoy the same debt tax shield.
A third argument emphasizes the disciplining role of leverage. As leverage increases, the loss absorption capacity provided by equity in the event of bankruptcy shrinks, increasing creditors’ risk exposure and inducing them to monitor the bank more closely. As Calomiris and Kahn (1991) note, uninsured depositors who monitor the bank and suspect managerial inefficiency (either fraud or poor loan monitoring/risk management) can withdraw those deposits. Uninformed depositors, observing the withdrawals, may follow suit. This precipitates a full-scale bank run and may force liquidation of the bank. Fear of such a leverage-induced run represents market discipline that keeps managers efficient and honest.
While debt can help reduce certain agency problems in banks, too much debt can invite risk-taking. Jensen and Meckling (1976) argue that sufficiently high leverage makes bank managers and shareholders take more risk to maximize the value of their equity option on bank assets (so-called asset substitution). Coping with this moral hazard requires limits on leverage unless the discipline through liabilities suffices to offset the asset-substitution potential. “Run”-able demand deposits provide just that discipline, but in general asset-substitution moral hazard will dominate the value of discipline at sufficiently high levels of bank leverage (Diamond and Rajan 2000).
In our paper, we examine this tension between the disciplining role of debt and the risk-inducing role of debt. In particular, we analyze a model in which creditors incur a fixed cost to monitor managers for inefficiency or fraud and threaten to liquidate the bank (“run”), but at the same time the presence of leverage can give bank managers the incentive to invest in riskier projects. We show theoretically that the bank is caught between a rock and a hard place when choosing its privately optimal capital structure. If it doesn’t choose enough leverage, the bank’s creditors don’t have enough “skin in the game” to make monitoring worthwhile, so they can’t credibly threaten to discipline the bank for inefficiency or fraud. However, if leverage is too high, asset substitution occurs and the bank takes excessive risk at the creditors’ expense.
We show that the bank’s privately optimal leverage must be high enough to induce creditor discipline but low enough to ensure that the bank’s risk-taking isn’t excessive. Optimal leverage ensures that bankers are taking efficient risks, but not excessively risky gambles (for example, funding undercapitalized mortgages).
The Problem with Safety Nets
We go on to show that the privately optimal capital structure can break down in the presence of government safety nets because these safety nets would partially protect the bank’s creditors when the bank fails. Deposit insurance, as well as other safety-net initiatives like bailouts of some failing banks, turns overnight debt financing, which would ordinarily be very risk sensitive, into relatively risk-insensitive financing. A similar argument applies to undercapitalized over-the-counter derivative exposures of large financial firms to each other.
In addition, the government safety net includes the central bank as the lender of last resort via the discount window. The discount window complements deposit insurance, while deposit insurance allows banks to obtain cheaper funding and subsidizes the right side of the balance sheet; the discount window gives banks access to short-term liquidity from the central bank when the market is unwilling to provide it, thereby reducing the refinancing risk banks face on the left side of the balance sheet. The economic rationale for safety nets is to prevent a widescale collapse of the banking sector and avoid contagion.
However, it’s now becoming abundantly clear—both in theory and in practice—that these regulatory safety nets come at a substantial cost, not just ex post in terms of fiscal outlays, but also ex ante in terms of moral hazard. The most obvious moral hazard is that banks are encouraged to become more highly levered. Because creditors don’t face the same risk they would in the absence of the safety net, the credit-disciplining effect discussed earlier is dampened, and the pricing of bank debt becomes relatively insensitive to the amount of leverage. As a result, leverage appears “cheap” to banks even as they take on increasing amounts of leverage that make themselves riskier.
The presence of the safety net upsets the balance of a finely tuned capital structure as described above: enough equity capital to attenuate asset-substitution moral hazard, yet not so much to water down the market discipline provided by (uninsured) creditors. In addition to this bank-specific effect, we argue that risk-taking is highly correlated across banks. For example, banks may herd on similar asset classes for lending or investments.
Now, correlated risk-taking induces correlated failures, making it more likely that there will be government bailouts. We show that the mere anticipation of bailouts may cause banks to choose highly correlated, excessively risky projects. Creditors will not “punish” banks ex ante in the pricing of (uninsured) credit for the systemic risk in their portfolios because they anticipate being bailed out. All market discipline of debt is lost and banks end up choosing much higher leverage. This channel of moral hazard is interesting. It’s the bank’s creditors that get bailed out ex post, not the shareholders, but the ex ante lowering of the cost of bank debt due to this accrues to the bank’s shareholders. This increases the attractiveness of riskier gambles on the macroeconomy by bank shareholders.
One way to mitigate this correlation-induced systemic risk is through appropriate pricing of deposit insurance. However, since some guarantees are implicit, appropriate pricing of deposit insurance premiums may not suffice. We argue that to prevent the “expropriation” of taxpayer funds through excess leverage and correlated risk-taking by banks, the bank’s leverage ratio must stay below a particular upper bound. And at the same time, creditors shouldn’t perceive banks to be so safe that they don’t discipline bank asset choices via monitoring and timely pricing of credit risks (“run”). In our model, there are two ways to do this.
The Role of a Special Capital Account
One way is a regular core capital requirement that guarantees that the bank’s leverage never exceeds the upper bound so as to keep risk-shifting incentives in check. The other—more innovative part—is a special capital account (SCA) that’s built up by restricting dividend payouts. An important purpose of this SCA is to provide the bank with a readily available resource that can be tapped to refurbish the core capital account instantaneously, and automatically, when it’s diminished due to losses. Dividends are then restricted to ensure that the SCA is rebuilt back to its original level over time through earnings retention. The SCA can be thought of as “collateral” that the regulator asks the bank to put up in exchange for a “loan” represented by the deposit insurance put option, and this collateral is lost in the event that there’s an idiosyncratic failure of the bank (as opposed to a systemwide meltdown with bailouts).
The SCA has several noteworthy features. One is that capital must be invested in liquid securities like Treasuries in order to remove managerial discretion over the use of that capital preventing the dissipation of excess cash.
A second feature is that the SCA accrues to shareholders as long as the bank is solvent, but it accrues to the regulator—rather than the bank’s creditors—in case the bank is insolvent and there isn’t an industrywide rescue of banks. That is, in case of idiosyncratic failures, bank creditors are forced to take losses. The fact that creditors don’t benefit from the SCA in that case means that creditors have enough skin in the game to discipline banks. Regulators would need to be empowered to seize the special capital account in the event of bank insolvency, just as the Federal Deposit Insurance Corporation Improvement Act of 1991 instructs regulators to shut down sufficiently undercapitalized banks. Thus, the idea is a form of “capital preservation” whose goal is to ensure that the probability of the bank failing is minimized but also preserves market discipline.
A third feature is that the level of the SCA can be set so that it’s a countercyclical capital requirement. In other words, the SCA requirement can be set so that it’s higher during good market conditions and lower during bad market conditions. Moreover, banks should be allowed to come into compliance with this requirement by building up retained earnings, so that raising new equity in the market can be avoided. If poor earnings cause the core capital account to be depleted, transfers from the SCA to the core capital account are purely mechanical and devoid of regulatory discretion, so there’s no new information released to the market. This is in contrast to a voluntary issuance of equity by the bank, which may incur adverse-selection costs (as it would for any firm), and be interpreted on average by the market as negative news.
A fourth feature of the SCA is that if poor earnings cause the core capital account to be depleted, the transfers from the SCA to the core capital account are purely mechanical and devoid of regulatory discretion. That is, both the transfer from the SCA to the core capital account and the accompanying dividend restrictions would be mechanically triggered by prespecified rules. The absence of regulatory discretion here means that no information is communicated by the transfers, which means one need not worry about bad news being associated with the transfer and the dividend restrictions. Also, because the SCA is invested in Treasuries or other cash-like instruments, the bank always has a buyer of liquid assets that can be tapped in the event of a liquidity crunch.
The views expressed in this post are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Any errors or omissions are the responsibility of the authors.
Anjan Thakor is the John E. Simon Professor of Finance at the Olin Business School of Washington University in St. Louis.