April 8, 2011 Princeton Conference in Honor of Burt Malkiel Concocting CoCos George M. von Furstenberg Indiana University For comments and related materials: vonfurst@indiana.edu Writ Raghu Rajan 2010: • If the govt. is likely to bail out SIFIs, the solution to the problem of institutions becoming too SIF has to be found elsewhere than in stiffening the backbone of regulators or limiting their discretion. There are three possible solutions: (1) Prevent institutions from becoming systemically important. (2) Keep them from failing by creating additional private sector buffers. (3) Make it easier for the authorities to fail (resolve) them. Focusing on (2), he proposed a double-trigger for Contingently Convertible Debt Securities, cocos. The two events he required to trigger conversion include the bank’s designated regulatory capital ratio falling below a value that is set at or somewhat above the regulatory minimum. This trigger is the identifying characteristic of cocos and the only one that has been used in practice. Macro-Prudential Triggers: A Critique • • • • The likelihood that cocos with the single trigger just described will be converted is, of course, much greater during a financial crisis than in normal and more profitable times. Hence their conversion will have some countercyclical effect. However, requiring conversions to wait until the entire banking system is in deep trouble as judged, for instance, by the assessment of regulators based perhaps on aggregate bank losses (Rajan) causes dangerous delay, and then a bunching of conversions and a flood of newly-issued bank stocks, when the authorities finally decide to pull the second trigger. Ex ante, valuing cocos would become very difficult and involve speculations about the behavior of regulators which Rajan, in dismissing the first option, otherwise would not want to depend on. Wanting to preserve “debt discipline” over management by preventing cocos conversion except when the entire banking system is in trouble is dubious also because denying cocos conversion to a SIFI, or to a number of financial institutions that together constitute a SIFI, could bring on the very bankruptcies (and ensuing externalities) against which cocos are designed to provide a measure of selfinsurance. In addition, cocos should take the form of long-term, not short-term debt, so that mothballing cov-lite cocos debt may not do much for discipline. COCOS mandates?? Basel II International Standard Basel III International Standard SIFI-SWISS (TBTF Comm of Experts) Progressive Component Add-on for SIFIs not det. 6% Cocos (Trigger 5%) Buffer (Countercyclical buffer 0 – 2.5%) 3% Cocos (Trigger 7%) Common Equity: 5.5% Conservation Buffer 2.5% Minimum Source: Tier 2: 4% Addititonal Tier 1: 2% Tier 2: 2% Addititonal Tier 1: 1.5% Minimum Common Equity (old definition): Minimum Common Equity (new definition): 2% 4.5% Patrick Raaflaub CEO S.FINMA.S.A. Chargeable on min. from cocos above: 3.5% Minimum Common Equity: 4.5% Small Countries with “Too Big to Save” Banks “ Basel III Bern Swiss SIFI Policy Bern I“ Swiss SIFI Policy Total Capital: 10.5% 19% w. Countercyc. 10.5-13% ? Common Eq. Min.: 7% 10% The actual capital-ratio targets aimed at by financial institutions are the result of optimizing capital ratios subject to constraints, such as the regulatory minima set for such ratios, and the penalties for violating them. When the regulatory minimum is increased, the excess of the actual over the minimum capital ratio intended by financial institutions is reduced. A crude way of supporting this statement is contrasting the positive level of the capital ratio that is chosen when the minimum is zero, with equality of the actual and required levels when the minimum capital ratio is 100% or 1. Hence, at least in the relevant range, actual capital ratios will rise by less than any increase in the minimum capital ratios. Even so, the actual Swiss capital ratios, with the “Swiss finish”, are going to be far above those of Basel III. If losses of tangible common equity of the banking system amount to ca. 3% of total assets as later concluded, and the size of the tangible assets of the two largest banks is 30% of GDP in the U.S. and 500% in Switzerland, it would take 0.009 (0.9%) of GDP to recapitalize the two in the US and 0.15 (15%) in Switzerland. Because a Swiss TBTF back-stop is not very credible, Swiss capital ratios need to be higher. Cocos are meant to make the additions less painful. Critical Cocos Specifications • Trigger Definition and Level (Staggering)(Book or market) • Conversion Terms (Formula or Market-Value) • Relative Size of Issue (Adequacy) (Mandates or Market) Coco-Sceptics: Dilution, no Market? • Oswald Grüner, CEO of UBS, (FT 03/03/2011) has said that a contingent convertible is “a very dangerous instrument”. “As soon as you get near these trigger levels – you do not have to hit them – what do you think shareholders will do? They will get the hell out of that stock, so fast, because you know it will halve in value if it is triggered.” • Those negative views do not keep UBS from being a leading tout in the cocos market, having worked on LBG’s November 2009 and Rabobank’s March 2010 cocos, or cocos-like issues. Barclays Capital has suggested that the reg.-driven cocos market, by 2018, could be huge (€700 billion). • A supporting cast at UBS (CFO John Cryan) has faulted cocos mandates on the grounds that there is “no natural universe of buyers for something that is really quite an obscure and untried and untested instrument.” Cryan even wondered “if ever one of these did trigger whether that would not cause a systemic issue with that market.” More cocos trashing: Charles A.E. Goodhart, “The Squam Lake Report,” JEL, March 2011 49(1): 114-119 “Many, perhaps most, versions of CoCos, as defined by their respective trigger mechanisms and conversion terms, would not be effective, in some cases positively dangerous. The version that would be, in [his] view, socially best would impose such a large dilution on existing shareholders that they would have an incentive to issue new equity at the first hint of trouble, rather than wait for the conversion to be triggered.” P. 117 Goodhart, Central Banking Mag 21(1) August 2010, p. 30 (29-33) CoCos convertibility must be triggered by falls in market, not in accounting valuations, because accounting values adjust too slowly. • Comment: Stock prices become highly volatile in a financial crisis and may deviate from fundamental values for flash-crash as well as extended periods. Conversion and its terms and consequences become needlessly difficult to gauge ex ante. • Compromise: Stockholders by majority vote and the Board of Directors of financial institutions should be able to request an unscheduled forward-looking and up-todate regulatory determination of an institution’s capital ratio to see whether the trigger point has been reached. Such a determination could also be made a legal precondition (and protection) for conversion. More Thrashing from Goodhart The only [cash-flow] benefits that cocos conversion brings is cessation of interest payments. A stronger mechanism would be to require the authorities to ban all dividend payments. Comment: The second suggestion is extraneous and can be adopted with or without cocos. Basel III already includes a reduction in the % of earnings that may be paid on dividends or sbb’s. The prohibition would rise to 100% of earnings if capital ratios cover less than 25% of the sum of the compensation and countercyclical buffers. Response: By bringing regulatory capital back up to adequate or wellcapitalized levels, cocos conversion can lower collateral requirements imposed on a financial institution, raise the rating of its debt, and increase borrowing capacity, for instance on trade credit or CP, which could bring in cash. The counterparty risk on dealing with such an institution would be reduced. Micro- vs. Macro-Prudential • Should conversion be allowed only if an individual bank and the whole financial system were in crisis? • The CoCos pay-off function, incurring a large loss when all other assets are also doing badly, is highly unattractive. • More problems: Contagion from cocos conversion, hedging through shorting, death spirals? • A financial crisis is marked by idiosyncratic risk being overshadowed by systemic (highly correlated) risk. • Surely the deadweight losses of receivership, from not having CoCos on the balance sheet in a crisis, are even less attractive. • These concerns counsel against setting the conversion trigger by the market value of equity. Conversion Rates: At least two methods are available. 1. The number of shares obtained in the conversion is determined so that their value at a “recent average price (as further specified)” would be equal, at that price, to the face value of the cocos being converted. Use of the “recent average price” may be qualified by the phrase, “but not less than $x per share,” where x is a floor price. This proviso is designed to limit dilution by placing an upper bound on the number of shares to be issued in the conversion. Except when the floor price is binding, the number of shares obtained from the conversion thus is not known until the “recent average price” has been recorded. The cocos-like tier 2 Buffer Capital Notes (BCNs) when to be converted, “convert into Credit Suisse Group ordinary shares at their prevailing market price over a 30-day period preceding the notice of conversion, subject to a minimum price of USD 20” (Credit Suisse, 2011). On March 16, 2011, CS (ADR) was trading in the $41-$42 range in New York. The second, and ex ante calculable conversion method, to be favored 2. The second method focuses on the fair share (S) of new shares (NS), SNS, in the total number of equity shares outstanding after conversion, NS + ES, where ES is the number of existing shares prior to conversion. SNS would be determined by just two comparable statistics that allow NS to be expressed as b times ES. For instance, if the face value of cocos is equal to 4% of total (not risk-weighted) assets as reported in a company’s last Quarterly Report and the trigger level of the leverage ratio is 3% of such assets, then NS=(4/3)ES if the cocos are converted all at once. Then SNS = NS/(ES+NS) = 4/7, and the new shares account for a majority of the total number of shares outstanding because the cocos conversion contributed more to regulatory capital than the 3% that was left over. On the other hand, if cocos amounted to only 2% of total assets so that NS=(2/3)ES at the time of conversion, then the new shares would account for 2/5 or 40% of the total number of shares outstanding after conversion. By making the number of shares issued independent of their market value at conversion in this way, downward stock-price spirals and the resulting dilution of the value of the stakes of existing shareholders can be avoided. Dilution through CoCos Conversion? • • “The debt–overhang problem arises when even informed financiers refrain from injecting additional equity since the proceeds of the investment are primarily going to existing debt-holders rather than the new equity holders.” Brunnermeier, FCIC Testimony Feb. 26-27, 2010. Here, by contrast, new shareholders share the benefits from cocos conversion pro rata with existing shareholders. As a result, the dilution of existing shareholders is fully compensated. Senior debt-holders see less subordinated (cocos) debt and more common equity (from cocos conversion) below them so that the new capital structure does not enhance the value of their senior claim to that extent. Nevertheless, they too will share in the benefits that result from the large reduction in PCA risk, followed by receivership and large deadweight losses (of more than 30% of total assets in FDIC experience). While it is true that any dollar contributed to a company that ultimately goes bankrupt is a dollar contributed to its most senior claimants, the benefits of bankruptcy avoided are shared by junior and senior debt claimants and shareholders in line with their exposure to bankruptcy risk and its consequence of destroying recovery value. The increase in the market value of the debt remaining after cocos conversion, if charged against current earnings under MTM accounting for liabilities, should be disallowed for reasons explained by Bishof and Ebert 2011. Managing the Links between Financial Reporting and Regulatory Accounting “Amortized cost remains relevant for debt instruments (both loans and securities) not held for short-term profit-taking” (p. 170). “Cumulative Gains and Losses due to Changes in Own Credit Risk – if the fair-value option is applied for financial liabilities, all unrealized gains and losses that have resulted from changes in the fair value of liabilities that are due to changes in the bank’s own credit risk are derecognized from tier 1 capital” (p. 171) under Basel III.” Jannis Bischof and Michael Ebert, “The Mixed Accounting Model Under IAS 39: Current Impact on Bank Balance Sheets and Future Developments,” Journal of Financial Transformation , Cass-Capco Institute Paper Series on Risk # 31, Part 2, March 2011, 165-172. Deadweight Losses of Bankruptcy: Lehman Brothers Holdings Inc.& Other Debtors &Controlled Entities Millions of $ Date Total Assets Total Liabilities Stockholders Equity Total Liabilities and Stockholders' Equity Preferred Stock Common Stock and Addit. Paid-In Capital Retained Earnings and Other Stockholders' Eq. Total Common Equity Total Stockholders' Eq. May 31 2008 Sept. 14 2008 Marc 31 2010 639,432 626,240 253,175 613,156 560,396 307,286 26,276 65,845 -54,111 639,432 626,240 253,175 6,993 13,077 11,035 2,382 47,431 26,162 16,901 19,283 26,276 5,337 52,768 65,845 -91,307 -65,145 -54,111 3.65 (Sept. 12) 0.1 36.81 (May Stock Price ($) at Close 30) Assumption: (1) Market Value of Equity is (2) Market Value of Equity Falls Always Equal to its Book Value Value of Equity 1 Column: No. of Shares 2 to 0 Prior to Cocos Conversion Value Value per Share of Equity 3 4 No. of Value Shares per Share 5 6 A. Existing Shareholders Initially Very Well 700 Capitalized Level After Deterioration 300 up to Trigger Point Dilution upon -(4/7)300 Pulling Trigger Pro Rata Benefits from (3/7)400 Cocos Debt Cancellat'n Post-Conversion 300 Value 100 7 700 100 7 100 3 0 100 0 100 -1.7143 0 100 0 100 1.7143 100 1.7143 100 100 1.7143 (3/7)400 3 171.43 B. Value of Cocos Bonds or of New Shares from Cocos Conversion Initial and Pre-Convers. 400 Value of Cocos Transfer from Existing (4/7)300 133-1/3 Shareholders' Dilution Pro Rata Benefits from (4/7)400 133-1/3 Cocos Debt Cancellat'n Post-Conversion 400 133-1/3 Value 400 1.2857 0 133-1/3 1.7143 (4/7)400 3 0 133-1/3 1.7143 228.57 133-1/3 1.7143 C. Combined Value of all Equity and Cocos Debt Claims Outstanding Initially of which Equity Equity after Conversion 1,100 700 100 700 233-1/3 7 3 1,100 700 100 400 233-1/3 7 1.7143 LOSS OF RISK CAPITAL (= EQUITY CAPITAL) 1. The IMF’s Global Financial Stability Report (GFSR), April 2009, p. 36, estimates that by the end of 2008 the first two years of the 2007-2010 financial crisis had reduced the ratio of tangible common equity (TCE), defined as total equity less preferred shares and intangible assets or as core tier 1 capital, to total tangible assets (TA) to 3.7%. In the United States. reported writedowns of $510 billion exceeded capital of $391 billion raised during 2007-2008 by $119 billion. For 2009-2010, the IMF expected writedowns of $550 billion to exceed net retained earnings of $300 billion by $250 billion. To both raise the leverage ratio, TCE/TA, from a little over 3.7% to 4% and to keep it at 4% would require an equity infusion of $275 billion in 2009-2010 according to this April 2009 estimate. The TA value for the U.S. that is implicit in these calculations is $11,250 billion. All these and subsequent estimates for the U.S. do not include the GSEs. 2. [In the subsequent GFSR of November 2009, p. 15, the estimate of the capital infusion still required in the U.S. from the middle of 2009 to reach a 4% leverage ratio by the end of 2010:Q4 was only $130 billion. Thus it is as if over half of the $275 billion capital injection previously estimated for 2009-2010 was made in the first half of 2009. Realistically, data and baseline revisions are also likely to have played a part.] 3. To raise the leverage ratio to the “well-capitalized” level of 6%, the “approximate leverage of U.S. banks in the mid-1990s, prior to the buildup in leverage in the banking system that contributed to the crisis,” would require a total equity infusion $500 billion according to the April 2009 IMF estimate. 4. The two IMF estimates combined imply that the financial crisis required an additional equity cushion of at least 275/11, 250 or 2.4%,and at most 500/11,250 or 4.4% for the crisis to be overcome by the end of 2010. Leverage Ratios for the 40 Largest U.S. Banking Operations before the Crisis, June 30, 2005, and TARP Top 3 Next 37 All 40 Tier 1 Capital ($ billions) 223 240 463 Total Assets, TA ($ billions) 3970 3162 7132 Total-Asset Weight 0.5567 0.4433 1 Aggregate Leverage Ratio % 5.62% 7.59% 6.49 Average Leverage Ratio % 5.66% 8.12% 7.94 0.51% 3.01% 2.96 115 74 189 2.29% 1.85% 2.09% Standard Deviation TARP Funding ($ billions) as % of (1.082)^3 (TA), to adjust to mid-2008 Sources: Calculated from (i) R. Alton Gilbert, Keep the Leverage Ratio for Large Banks to Limit the Competitive Effects of Implementing Basel II Capital Requirements, Networks Financial Institute at Indiana State University, Policy Brief 2006-PB-01, January 2006, Table 5, and (ii) U.S. Treasury Department, Office of Financial Stability, Troubled Asset Relief Program (TARP), Transaction Report for the Period Ending Feb 23, 2011. How Big was the Hole? Risk Capital Rebuilt 2009-2010 • • The Federal Reserve, in a March 18, 2011 press release announcing completion of its Comprehensive Capital Analysis and Review of the capital plans of the 19 largest U.S. bank holding companies reported that from the end of 2008 through 2010, “common equity increased by more than 300 billion” at these 19 companies. Considering that some of the smaller banks also issued common equity in 2009-2010, the combined equity issue of approximately $325 billion was quite enough to meet the IMF’s 4% leverage target set in April 2009 for the end of 2010 under the then-assumed (somewhat too high) writeoff and (if anything too low) retained-earnings conditions. The Fed appears to have been satisfied that most of the 19 largest banks tested (one significant exception was Bank of America) had reached a level of capitalization sufficient to allow a resumption of (initially still modest) dividend payments or stock buybacks. The common equity raised has been sufficient to allow TARP funding to be repaid by the end of 2010 by all of the 19 banks, except Ally Bank, the inglorious successor to GMAC. In sum, it appears that the recent financial crisis left a hole of 2.4% to 4.4% of total assets, with additional indications pointing to 3% as the most representative round number in this range. TARP funding (under CPP and TIP) provided an amount equal to about 2% of total bank assets starting on 10/28/2008 and ending on 09/30/10.