Photographer: Cem Ozdel/Anadolu Agency/Getty Images
Photographer: Cem Ozdel/Anadolu Agency/Getty Images

The New York Fed has released a big package of papers on "the economics of large and complex banks and their resolution in the event of failure," which you can read here, or read about here. I suppose the intra-Bloomberg View betting was mostly on "Evidence from the Bond Market on Banks' 'Too-Big-to-Fail' Subsidy," and I suppose I lose. That evidence shows "that the spreads of bonds issued by the largest banks are, on average, 40.6 basis points below the smaller banks’ bond spreads, after controlling for bond characteristics, including the credit rating, maturity and amount of issue, as well as conditions in the bond market at the time of issue," so there you go, fine, whatever.1

But there is much other goodness that you could dip into at your leisure. For instance: Does bigness in banking lead to operating cost savings? (Yep! Adding $1 billion of assets saves around $1 to $2 million a year in non-interest expense; add in that TBTF subsidy and you've got plenty of incentives to get bigger.) Are too-big-to-fail banks riskier? (Also yep; additional government support increases impaired loans.) Or, further afield, would the New York Fed characterize the Lehman Brothers bankruptcy as a success? (Not particularly!)

Capital Requirements

Perhaps most notable is the New York Fed researchers' focus on bail-in debt as a solution to bank failure. Here, for instance, is a paper by their lawyer, Joseph H. Sommer, which I enjoyed, though, full disclosure, it is a little lawyery.2 But the model is useful. It starts with a notion of "financial liabilities," which is related to, though broader than, concepts like "information-insensitive debt" or "safe assets":

Financial liabilities are those that only financial firms are in the business of incurring. They include things like bank deposits, derivative contracts, insurance policies, and repos. Corporate debt or trade credits are not financial liabilities. This is also true of the corporate debt or trade credit of financial firms. ...

Financial liabilities are more than claims to a future stream of income. Yes, they are that, but they are something else, too. Some of them, such as bank deposits, are also sources of liquidity. Others -- derivatives or insurance policies -- shift risk. These liabilities are also credit instruments. But unlike, say, corporate bonds, their credit nature is incidental to -- but inherent in -- their liquidity or risk-shifting functions. Liquidity and risk shifting are valuable in themselves -- valuable beyond the face value of the financial liability. This is not a conjecture. It is a revealed preference. Insurance policies cost more than their net present value. Liquid debts pay less well than illiquid ones.

These financial liabilities are problematic. One, because nice financial liabilities you've got there, would be a real shame if someone were to go and break them. Any time any company fails, someone (equity holders at least, and often creditors) loses money, but when a bank fails, the damage is not just the lost money but also the lost insurance or liquidity or whatever function of those liabilities. If the value of financial liabilities goes beyond their face value, then so does the potential loss.

Two, because financial liabilities tend not to create great monitoring incentives to discipline banks' risk-taking: "The creditors of banks and insurers are widely dispersed, and not in the business of lending money. (This is inherent in the definition of 'financial liability.')" Three, because they're run-prone: If you're not there to take credit risk, you're more likely to move your money to a safer bank at the first sign of trouble.

And, finally, because financial liabilities are an incentive to increase leverage. Sommer describes financial liabilities as "products," that is, as things that banks sell in their business, rather than as debt that they raise. And they are that -- but they're also debt. They're thus almost by definition cheap debt: Customers want the product (liquidity, risk-shifting), and they pay for the product by, in effect, lending money to banks at below-market rates. Like most companies, banks want to sell as much of their products as they profitably can, but since their products are also their liabilities, that tends to create a lot of leverage.

That last paragraph might be relevant to your intuitions on questions like "why are banks so highly levered?" or "do banks have a too-big-to-fail subsidy?" but let's move right along.

The idea of a bail-in requirement is to clearly separate a bank's financial liabilities -- deposits, derivatives -- from its non-financial liabilities, and to treat them differently as a matter of explicit law rather than approximate custom. A bank would need to issue holding-company-level debt, and, if it ran into trouble, the debt would be written down (or poofed into equity). This would re-equitize the holding company, which tends not to have any financial liabilities in Sommer's sense, and it would in turn recapitalize the operating subsidiaries, which tend to be all about financial liabilities.3

This is more or less the thinking behind the Fed's "living wills" for big banks, and it makes sense to me. The loose way in which I have often thought about bank failures is:

  • A bank has assets on one side of its balance sheet, and claims (deposits, bonds, preferred stock, common stock, whatever) on the other.
  • Sometimes the assets go down in value.
  • When that happens, some of the claims go down in value, meaning that some claimants (shareholders, bondholders, depositors) lose money.
  • When the somebody who loses money is sympathetic or politically powerful or otherwise someone you don't want to lose money, you have a problem.

The shareholders are rarely (not never!) sympathetic; the depositors usually (not always!) are. Other categories of claimants are all over the place.

Sommer's framework does a good job of separating whom you should care about -- people who hold "financial liabilities," which have social value beyond just being an investment -- from whom you shouldn't -- people who just own bank bonds as a credit investment and are compensated for taking that credit risk. And then it does a good job of actually putting the losses on the credit investors, who sort of deserve them, and sparing the financial claimants, who don't.

On the other hand, my lazy framework might map better to existing reality. Here is an article about how Spanish banks might owe money to investors who lost money on their junior debt, because those investors were apparently confused about how junior their junior debt was. Here are some owners of Lloyds bail-in debt standing up for themselves. Just because the Fed and the law thinks of bank holding company debt as bail-in-able -- and has a good rationale for that -- doesn't mean its holders will, or that they won't find a sympathetic ear when they're actually bailed in.

A sort of companion piece to Sommer's is "What Makes Large Bank Failures So Messy and What to Do About It?"4 It quantifies the run risk of uninsured financial liabilities, and then makes the somewhat intriguing argument that a requirement of X amount of common equity plus Y amount of long-term debt is better bank-safety regulation than a requirement of X+Y common equity, because the long-term debt "functions as 'capital in resolution' that serves to stall runs by holders of uninsured liabilities." I confess that I do not understand this argument but maybe you will.5

Anyway, there is much to ponder here, but the message of the whole package can perhaps be summed up like this. On the one hand, there are big banks, and they are important, and it is important that they not "fail" in a way that impacts their non-bond creditors, and the way to do that is to put the burden of failure quickly and explicitly on their bond creditors. On the other hand, the market thinks that the bonds of those big banks are extra-safe, because the government would never put the burden of those banks' failure onto their bondholders. The New York Fed's task is to change the market's mind about that, so that if a big bank ever is actually bailed in, it won't come as a shock.

1 Is that $83 billion a year? Nope, nope it is not. I ballpark JPM's long-term debt at $272 billion, BAC's at $204 billion, C's at $178 billion, GS's at $168 billion, and WFC's at $153 billion, for a total of a little under $1 trillion of debt. A 40.6 basis point subsidy on those banks' bonds works out to about $4 billion a year. And I insist that a 40.6 basis point spread on long-term bonds does not translate into a 40.6 basis point spread on everything, because -- well, because of the rest of this post, is why.

2 The abstract begins, "All men are created equal, but all liabilities are not," which is worrying, and the number of Catskills jokes quoted in the paper is not zero.

3 Sommer:

By a stroke of luck, the Bank Holding Company Act encourages the parent entity of a financial firm to be pretty much a pure holding company. This means that the parent entity does not rely on financial liabilities. Futhermore, the parent is the cheapest source of funding in the organization. Therefore, the parent can downstream this cheap debt to the subsidiaries. This means that the third-party liabilities of the subsidiaries are mostly financial.

By the way "financial liabilities" is an awful term of art; it kind of already means something, and that something is kind of the opposite of what Sommer means. Anyway!

4 It begins, "This paper uses 'messy' repeatedly, so we should be clear at the outset what we mean by that term," which is more my speed.

5 It seems to rely on the idea that a bank will not be put into resolution until its common equity capital reaches zero, and that once it is put into resolution it will not experience any further losses? Neither of these assumptions strikes me as plausible but perhaps I am mis-reading.

Here, meanwhile, is an enjoyable paper about trust-preferred securities as regulatory arbitrage, finding that "banks with TPS will be riskier than other banks with the same amount of regulatory capital, and therefore, more adversely affected by the credit crisis." Trust preferreds are more or less the pre-2007 version of bail-in debt, though without the explicit resolution mechanism.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Tobin Harshaw at tharshaw@bloomberg.net.