Thoughts on Systemic Risk Regulation
Notes prepared for panel presentation at the 2015 Institute of International Finance North America Summit, New York.
Cost of Much Higher Capital Requirements
It is a great pleasure to be back with the IIF community and an honor to participate in such a distinguished panel. I will begin by summarizing results of recent research I have completed on a central question in banking regulation: How much does it cost to increase equity capital and reduce leverage? Higher capital requirements, especially for systemically important financial institutions (SIFIs), are a key part of Basel III regulatory reforms. Some academics have argued that capital requirements should be far higher than the Basel III targets. In particular, Admati and Hellwig (2013) propose that equity capital be increased to 25 percent of total assets, not risk-weighted assets (RWA). In comparison, for SIFIs under Basel III the ratio is set at 10 to 12 percent of RWA, and typically the risk-weighted amounts are only about one-half of the total asset amounts.
A key argument in proposals for far higher capital is the idea that raising equity capital and reducing debt will not be costly. Nobel Prize–winning economists Franco Modigliani and Merton Miller proposed almost 60 years ago that capital structure has no influence on capital costs. The M&M proposition is that the reduction in risk achieved by reducing leverage will lower the risk premium the market demands in the rate of return on equity by exactly enough to leave the average cost of capital unchanged, even though debt capital has a lower unit cost.
My study recently published by the Peterson Institute conducts a statistical test of the Modigliani-Miller theorem (Cline 2015). The M&M equations imply a straight-forward test, with the unit cost of equity on the left-hand side and the ratio of debt to equity on the right-hand side. Using data for the 50 largest US banks for 2002–13, I test whether the cost of equity capital is indeed higher for more highly leveraged banks, and by how much. One set of tests measures equity capital cost by the earnings yield of shares, in other words the inverse of the price-earnings ratio. Another set of tests examines earnings relative to the book value of shareholder equity. Averaging the two sets of results, I find that higher leverage does indeed increase the unit cost of capital. However, the influence is not nearly large enough to offset the increase in average capital cost resulting from a shift from lower-cost debt to higher-cost equity. I find that the Modigliani-Miller effect would offset only about 45 percent of the increase in capital costs. If required equity capital were increased by 15 percent of total assets (not risk-weighted) to get to the Admati-Hellwig range, the consequence would be to boost the average cost of capital by 62 basis points. If one considers the long-term consequence of less capital formation because of higher capital cost, and after considering nonbank sources of capital, the effect over 30 years would be equivalent to the loss of about one-third of one year's worth of GDP. Seeking greater systemic safety by raising capital requirements far higher than Basel III might still be worthwhile if the result would be far lower incidence of financial crises, but in reaching the optimal capital ratio one should not assume that additional capital would come at no cost.
Other empirical tests seem to be more supportive of Modigliani-Miller, but they have all used statistical tests relating leverage to the stock-market beta ratio in the framework of the capital assets pricing model. Unfortunately, it is well-known that that model overstates how sensitive stock prices should be to risk as measured by the beta. For their part, Admati and Hellwig acknowledge that distortions such as tax deductibility of interest might mean there are costs to the individual banks from reducing leverage, but they argue that from society's viewpoint the higher capital requirements should be imposed. But proper measurement of the welfare effects should instead take account of the practical impact on capital formation and growth, given the reality of tax structures.
Dodd-Frank and the Slide from Bagehot Lender-of-Last Resort to Fast-Track Bankruptcy
Among the broader issues that are the subject of this panel, my main concern about regulatory reform is that Dodd-Frank may have unduly compromised the scope for lender-of-last-resort action by the Federal Reserve and the Federal Deposit Insurance Corporation (FDIC), and created a regulatory bias toward fast-track bankruptcy at fire-sale prices. In a 2013 study with my colleague Joseph Gagnon, we concluded that the major interventions in the 2008 financial crisis had appropriately followed a central banking principle that dates back to the 1870s: Walter Bagehot's dictum that in a banking panic, the central bank should lend without limit to banks that are solvent (Cline and Gagnon 2013). Otherwise a liquidity crisis can unnecessarily transform into a solvency crisis. It was the need for this lender of last resort capacity that led to the creation of the Federal Reserve System in 1913. In my view the Bagehot principle should be applied on a basis that takes account of the medium-term value of assets in arriving at a solvency judgment, not the fire-sale prices that occur in the midst of a panic.
In our study, we found that all the firms that received special assistance, namely Bear Stearns, Fannie and Freddie, AIG, and by implication Citi and Bank of America, were indeed solvent. In contrast, the one key firm that did not receive assistance, Lehman Brothers, was insolvent, and had a balance-sheet hole on the order of $100 billion to $200 billion. As it turned out, the US government made a profit on the special financial support it provided, demonstrating solvency ex-post. Ben Bernanke had studied the Great Depression and concluded that the central mistake was to allow banks to fail excessively, in other words, to fail to exercise sufficiently the lender-of-last-resort function. Legislation that came out of that era included section 13(3) of Federal Reserve Act, which provides for emergency lending to individual firms "in unusual and exigent circumstances." In other words, authority to "Do whatever it takes." In an environment of political backlash against what seemed to be bailing out the banks, however, the Dodd-Frank legislation curbed this authority. The Fed can no longer lend to an individual company outside a program with "broad-based" eligibility. The special vehicles such as Maiden Lane set up to provide a $29 billion asset guarantee to JP Morgan in exchange for its takeover of Bear Stearns would probably not be possible now. Neither would the support of nearly $200 billion to AIG, or the Fed guarantees of about $300 billion asset-backed securities for Citigroup and about $120 billion for Bank of America, in exchange for preferred shares and warrants. For its part, the FDIC can no longer make loans under "Open bank assistance," like its lending to Continental Illinois in 1984. The FDIC can now only assist a bank that has been closed.
Under Dodd-Frank, there has instead been a shift toward prompt resolution. In the "Single Point of Entry" structure, the holding company is in effect put into accelerated bankruptcy, with shareholders wiped out and debt converted into equity. The subsidiaries of the holding company are meant to be insulated and can receive FDIC lending from the Orderly Liquidation Fund. This arrangement solves the problem of a lack of orderly resolution for a Lehman-type situation. However, if the alternative of lender-of-last-resort lending has been compromised, the overall result could be an unnecessary acceleration of bankruptcy, and liquidation of assets at fire-sale prices. Regardless of promises about solvent subsidiaries after shareholders and creditors of the parent holding company are bailed in, it would seem unlikely that the subsidiaries would be viewed as strong credit risks when the holding company is in liquidation.
Overall, Dodd-Frank brought a clear plus for resolution, to deal with a Lehman crisis, but arguably a substantial and perhaps more important minus for averting unnecessary bankruptcies by curbing the scope for traditional lender-of-last-resort action.1
Role of TLAC
On the specific instrument of Total Loss Absorbing Capacity (TLAC), I view the new requirements as a form of quasi-equity on the cheap. The basic idea is that medium-term debt can be forced to convert to equity in a workout, so it plays a role similar to that of shareholder equity while being cheaper. For the SIFIs, the current policy track is to use contingent convertible (coco) and other medium-term debt to in effect double the capital ratio from 12 percent of RWA to 24 percent, in the United States. Does 24 percent TLAC solve the problem? A first point to recognize is that TLAC is against RWA, not total assets, so the SIFI target for TLAC really amounts to about 12 percent of total assets (and the formal leverage target is double the Basel III amount of 3 percent, or only 6 percent of total assets). So TLAC presumably does not satisfy those who call for far more equity capital. A concern about TLAC, expressed by my colleague Avinash Persaud, is that the most likely candidates to hold cocos and other crammable debt are the hedge funds, but they are precisely the ones quick enough on their feet to run in the face of a brewing financial crisis (Persaud 2014). A plunge in prices of such debt in an incipient financial crisis would hardly help stabilize markets.
It is important to keep in mind that it took 80 years for a new financial crisis to appear comparable in risk to that of the Great Depression. In building the financial infrastructure, the question is where to draw the line in the trade-off between spending for insurance against extreme but rare storms and investing for output capacity growth. My sense is that the reforms under way from Basel III and the SIFI increments already go much of the way toward the right balance. However, I worry that there has already been too much of a restriction of the Lender of Last Resort authority of the Fed and FDIC, and a corresponding slide toward unnecessary preemptive bankruptcies with more damage than expected in the Single Point of Entry model of safe subsidiaries but bankrupt holding companies under the Orderly Resolution Authority of the FDIC.
Admati, Anat, and Martin Hellwig. 2013. The Bankers' New Clothes: What's Wrong with Banking and What to Do About It. Princeton: Princeton University Press.
Cline, William R. 2015. Testing the Modigliani-Miller Theorem of Capital Structure Irrelevance for Banks. Working Paper 15-8. Washington: Peterson Institute for International Economics (April).
Cline, William R., and Joseph E. Gagnon. 2013. Lehman Died, Bagehot Lives: Why Did the Fed and Treasury Let a Major Wall Street Bank Fail? Policy Brief 13-21. Washington: Peterson Institute for International Economics (September).
Kohn, Donald. 2011. Will the Federal Reserve Be Able to Serve as Lender of Last Resort in the Next Financial Crisis? Federal Reserve Bank of Boston, 56th Economic Conference, Boston, October 18-19.
Persaud, Avinash D. 2014. Why Bail-In Securities Are Fool's Gold. Policy Brief 14-23. Washington: Peterson Institute for International Economics (November).
1. For further discussion, see Kohn (2011).