A Simpler Way to Clawback and RECOUP Losses
[Note, after publishing this post I corrected typos, elaborated on how the EDGE database could be used to create synthetic unlimited liability for executive shares compensation and revised the second figure to include banks that closed due to consolidations.]
Toward the end of my previous blogpost, I mentioned how the EDGE database could be used as an input to estimate “Synthetic Unlimited Liability” penalties for executives of distressed corporations, which rather than going through bankruptcy proceedings, impose costs on taxpayers. Here, I’ll discuss how it offers a simpler alternative to recent bank policy initiatives.
For instance, a Bank Reg Blog post last week summarizes the third attempt by regulators to propose a clawback rule to implement Section 956 of the Dodd-Frank Act, which was supposed to be done by 2011. Given the complexity of the proposed rule, which mostly rehashes the 2016 proposed rule, the Bank Reg Blog post provides a good summary, which I won’t repeat here, beyond mentioning that it covers entities with at least $1 billion in total assets.
There’s also a bill circulating in the Senate called the “Recovering Executive Compensation Obtained from Unaccountable Practices” or RECOUP Act of 2023, which aims to introduce executive compensation clawbacks for large failed banks with over $10 billion in assets. The RECOUP Act targets executive bonuses, sales of securities during the preceding 24 months, and increases the maximum civil penalty from $1 million to $3 million. What’s wrong with that?
As usual, in bank legislation and regulation, there’s a tendency to focus on nominal dollar thresholds when determining which banks qualify, which accordingly fails to account for the effects of regulatory creep from inflation. For instance, as with anti-money laundering rules that focus on daily $10,000 in daily transactions ($10,000 in 1970 equals almost $79,000 today!), the effects of inflation will sweep in more smaller banks as the real value of the “large” bank asset thresholds, such as $1 billion, $10 billion, $100 billion or $250 billion, erodes over time.
Similarly, as with the nominal deposit insurance limit of $250,000, the real value of protection will erode over time due to inflation ($250,000 in October 2008 equals almost $361,000 today), the $3 million civil penalty limit in the RECOUP Act will become less severe over time. The Act also focuses on just two years of securities sales. Why? I think there’s a simpler, more transparent way to do this through unlimited (or perhaps less onerous double) liability. If these ideas are new to you, I’ll begin by introducing single, double and unlimited liability for shares.
Single, Double and Unlimited Liability from a Contingent Claims Perspective
Harvard Professor Ben Esty showed how to think about single, double, triple and unlimited liability, which he called “contingent” liability with a simple graph (I’ll exclude triple liability from the remaining discussion for simplicity). The analysis draws from the “contingent claims analysis” literature, an area of finance in which people analyze the value of claims contingent on the state of nature. Such claims could include common equity shares, bonds and options.
To start, keep in mind that the market value of a bank’s assets should equal the sum of the face value of deposits and the bank’s market value of equity and the liability penalty. As in Esty’s paper, assume the face value of bank deposits equals $100 (could be millions, or billions but I’ll drop the extra zeroes here). I’ll assume as Esty does that that the par value penalty specified by shares of stock equals $10. Lastly, I’ll assume that bank assets can range from $40 in the worst state of the world to $140 in the best state of the world. The graph below depicts the three forms of contingent liability, separately, to show that as you increase the contingent liability penalty, the payoff to an executive transforms from one that looks like a hockey stick to a 45 degree line in the case of unlimited liability. Any dashed portions of the lines reflect what could turn into socialized losses.
Starting from the upper left, you see that as with common equity shares, single liability means the most you can lose is your entire investment, as when the stock price goes to zero (the stock price can’t be negative with single liability). Under double liability, not only can you lose the entire investment, but you could be made to pay the $10 par value, as when bank assets fall to $90, but the bank still owes depositors $100. Unlimited liability goes further. In this case, not only could you lose your entire investment, but you could be made to pay the $60 penalty, as when the assets fall to $40, but the bank still owes depositors $100. I’ll later discuss how well double liability worked for national banks prior to the Federal Reserve, but before then, I’ll introduce Synthetic Unlimited Liability.
Synthetic Unlimited Liability
Synthetic Unlimited Liability might provide a simpler alternative to the Section 956 proposal or the RECOUP Act. A key merit of using contingent claims analysis arises from the fact that it makes it possible to show how to recreate one type of financial claim with other existing ones, “synthetically”. The most famous is the paper by Fischer Black and Myron Scholes, which showed how you could recreate a risk-free bond, synthetically, using call and put stock options and existing shares of the same stock; or more generally, if you had any two of the three, you could recreate the third, synthetically.
My reason for proposing to create unlimited or contingent liability “synthetically” using existing common equity shares is that it does not require changing the liability of existing shares in executive compensation, and it could also apply to options compensation. The proposal I have requires specifying:
1) For equity compensation, given the fraction of a bank’s total common equity that an executive has received as part of deferred compensation, make the executive liable for that same fraction of any realized costs arising from the bank’s failure, if it doesn’t go through bankruptcy proceedings. You can estimate such costs ex ante using the methodology I use for the EDGE database. You could also extend this idea to any non-bank corporation that might seek bailouts instead of bankruptcy protection.
2) For every call option that a bank executive receives as part of deferred compensation, make that executive sell an equal number of put options with the same strike and maturity; this in effect is what option traders call a “synthetic long stock” position. Here too, you could also extend this idea to any non-bank corporation that might seek bailouts instead of bankruptcy protection.
Doing so will recreate the 45 degree diagonal line in the graph above for unlimited liability for stock and options compensation.
If you think the unlimited liability penalty is too severe, you can always introduce caps on the penalty:
1) For equity compensation, instead of the executive paying the fraction of any costs arising from the bank’s failure, if it doesn’t go through bankruptcy proceedings, change the cost figure to a fraction of the costs of failure (e.g., using the earlier numbers if the costs were $60 as in unlimited liability, then double liability would cap the cost at $10).
2) For option compensation, instead of making executives sell an equal number of put options as call options received, make them sell “bull put spreads”, by selling an equal number of put options with the same strike and maturity and buy an equal number of put options at a lower strike price but same maturity.
Doing so will recreate the hockey stick line in the graph above for double liability for stock and options compensation.
Now that I’ve introduced the “what instead”, I’ll get to the “why”? Contingent liability lowered the cost of banking crises by giving bank executives reasons to close banks early rather than continue operating.
When Double Liability Worked Well
Few of us were alive when national banks were subjected to double liability (some states also applied double, triple or unlimited liability to banks chartered in their jurisdiction). As Eugene White showed, that system worked pretty well from 1865 after the National Bank Act of 1864, which subjected bank shares, which were held by owner managers to double liability, until the creation of the Federal Reserve in 1913.
Before the Discount Window, White reports that about 80 percent of banks voluntarily liquidated. After the Discount Window, voluntary liquidations fell dramatically through the Great Depression. Using data from Table 9 on p. 58 of the 1940 Office of the Comptroller of the Currency’s (OCC’s) Annual Report, the figure below shows that after the Discount Window, the percentage of voluntary closures declined to as low as 30 percent by 1933.
What changed? White argues that the introduction of the Discount Window, from which troubled banks borrow from the Federal Reserve, gave weak banks a lifeline. Prior to the Discount Window, bank owner-managers tried to avoid the double liability penalty, but Discount Window lending now enticed bank owner-managers to gamble and hang on for dear life, but failure was typically inevitable. After the Great Depression, rather than blaming the Discount Window, officials blamed double liability, which was subsequently eliminated under the Federal Deposit Insurance Corporation (FDIC), which applied a heavier hand to troubled banks. The elimination of double liability is unfortunate, given that the administrative costs of crises has risen over time.
White reports that the administrative costs of bank failures from 1865-1913 equaled roughly $1 billion in 2009 US dollars. By the Great Depression, the costs of bank failures escalated to $39 billion in 2009 US dollars, and as I recently discussed, they’ve continued rising since then.
Conclusions
In a way, the Spring 2023 bank failures mirrors the post Discount Window through Great Depression period in that 25 percent voluntarily liquidated (i.e., Silvergate), while 75 percent failed (i.e., Silicon Valley Bank, Signature and then First Republic). It would be great if banks voluntarily liquidated more often, which would lower the administrative costs of failures. As such, Synthetic Unlimited Liability might also address the deficiencies with “Prompt and Correction Action.”