Single-Point-of-Entry: No Bank Left Behind

Last December the FDIC put out for comment a proposal for a Single-Point-of-Entry (SPOE) Strategy to implement its Orderly Liquidation Authority (OLA) under Title II of Dodd-Frank. Single-Point-of-Entry has gotten a lot of policy traction. The Treasury Secretary supports it and there’s huge buy-in from Wall Street.  And it’s an approach that is likely to ensure financial stability in the event that a systemically important financial institution gets into trouble.  There’s just one problem with it.  SPOE means “No Bank Left Behind”.  

Here’s the idea behind SPOE. SIFIs have complex corporate structures with lots of subs under a holding company and all sorts of liabilities at both holding company and subsidiary level. The SPOE strategy is based on splitting off the subs from the holding company in a single fell swoop. The HoldCo gets placed into FDIC OLA receivership and all of the subsidiaries (along with their liabilities and certain HoldCo liabilities) are transferred to an FDIC-run bridge holding company. The only liabilities that would remain with the old HoldCo would be the long-bonds (senior and subordinated).  These bonds would get repaid based on what the FDIC was able to sell/recover on the assets of the subsidiaries.  The FDIC’s bridge holding company (HoldCo2) would get financing from the FDIC, which would fund the loan through a line of credit with Treasury. So HoldCo2 loses money (let’s say because the subs holding lots of bad mortgages or dodgy MBS or bad derivative positions), the FDIC will be on the hook.  

So three key things to see here. (1) the equity and the long-bonds of the HoldCo would get wiped out or get pennies on the dollar in the FDIC receivership. And (2) all of the liabilities of the subsidiaries would be paid in full because (3) the FDIC (and ultimately Treasury) are taking on all of the credit risk of the bank’s operating subsidiaries. 

The fundamental operation of SPOE is that all creditors of the banks get bailed out.  SPOE involves the guarantee of all bank liabilities other than the long-bonds of the bank holding company. In other words, not only are deposits and the payment systems protected, but all of the derivative counterparties, all of the commercial paper creditors, all of the repo counterparties, and all of the securities lending counterparties get bailed out.  

Traditionally, when a bank fails the FDIC protects the insured deposits, and also guaranties the bank’s payment systems obligations and often the uninsured deposits. But other bank creditors, like derivative counterparties get only the protection they have bargained for—their collateral and guaranties. The same is true when a bank holding company is placed into bankruptcy. Unfortunately, it’s a slippery slope from protecting payment systems, to protecting liquidity (to ensure payments can get made), to protecting all types of short-term credit and then to giving away the farm (or bank). But didn’t we create the Fed and Federal Home Loan Banks to make sure that there was liquidity when times get tough? Shouldn’t that be enough? Why do we need to protect all types of short-term credit? Wasn't the lesson from 2008-2009 that we should be trying to discourage this sort of shadow banking?  

SPOE institutionalizes the 2008-2009 bailouts. It ensures that Wall Street will be rescued if a SIFI goes down. The result is to eliminate all credit risk for Wall Street when dealing with SIFIs, which ensures that the SIFIs will stay too-big-to-fail and that there will not be market discipline.

SPOE should also look very familiar to the sharp-eyed observe. It's nothing other than the much-criticized structure used in the Chrysler and GM bankruptcies.  The good assets and favored liabilities are transferred to a new, government-backed entity, while the disfavored liabilities remain with the old, liquidating entity.

To be sure, the problem with Chrysler and GM was that no one knew ex ante which liabilities would be favored and which would not be, so it was impossible to price for them. While SPOE does not have that problem, it suffers from the inverse problem—the long-bonds of the HoldCo are essentially co-cos, just with the conversion trigger being OLA, rather than some pre-receivership event. As a result, the HoldCo long bonds are going to have to have a pretty high yield, and there’ll have to be a lot of them to make the SPOE strategy effective. And who’s going to buy them? SIFIs aren’t going to be allowed to invest in this junk.  

I appreciate that SPOE ensures minimal disruption from the failure of a SIFI, and that seems to be the FDIC's goal with OLA. But there’s a cost to limiting post-failure disruption. That’s the enormous pre-failure market distortion that comes from guaranteeing all bank liabilities other than the long-bonds of the holding company. What happens to market discipline? Do we really want the government to be in the business of guaranteeing payment on all repos and derivative contracts of banks?  

I don’t think the FDIC’s proposal intends to be a “No Bank Left Behind” solution, but that’s the effect. If we’re going to throw in the towel on market discipline and put the public fisc behind all Wall Street dealings with SIFIs, we should just be honest and nationalize the SIFIs. If the public is going to take the downside risk, it should also get the upside benefit.

No one wants to have that conversation (and I don’t think it’s the right thing to do), so instead, we pretend that the public fisc really isn’t on the hook: the bridge institution will merely have a line of credit from the FDIC, which will have a line of credit with Treasury, but there won’t be any risk because it will all be well collateralized and the FDIC is really a mutual insurance fund. The structure of second-loss public capital backing first-loss private capital is a time-honored tradition in US bank regulation, going back to the 1863 National Bank Act. This is how National Banks were originally structured, how Federal Reserve Notes originally worked, and how the FDIC’s Deposit Insurance Fund works.  This structure is all about disguising the role of the government in the financial markets while ensuring that the benefit is still there. But it should be no surprise, then, when the government ends up on the hook.  

If the FDIC is going to be guaranteeing all bank liabilities, it needs to be charging full freight for the service. This means having much larger FDIC insurance premiums that reflect individual SIFI’s level of non-deposit liabilities. If there’s going to be a guarantee, it needs to be explicit and priced. The bitter has to go with the sweet. I suspect that if Wall Street has to pay full freight for a government guarantee, it won’t be so enamored with the SPOE strategy. For now, however, Wall Street thinks it sees a free lunch and is licking its lips over SPOE. I hope the FDIC wises up to this, and I hope that Congress makes clear that the full faith and credit of the United States is only given at market rates.  

Comments

[Warren]

Could this structure be improved by lowering and limiting the FDIC's liability? There are benefits to the SPOE strategy, related to just the effort to detail the procedures used to liquidate a large organization, in advance, so that all parties can know what to expect and adjust their contracts accordingly. The problems with risk ownership that seem to produce TBTF and moral hazards may all be rooted in Deposit Insurance. If banks had to compete for funding sources that were only insured up to $25,000 or even $100,000 you would move some risk ownership back and align their incentives a bit closer with the public's. The right $ amount to insure per depositor could be a little closer to the $ amount that average person keeps on deposit in these risk-free accounts. It's probably less than $250,000.

[anon]

Adam, didn't you, and most academics, support the administration on Chrysler and GM? Why the change of heart on government's heavy hand in this case?

[Ebenezer Scrooge]

The statute says that Wall Street has to pay full freight for the government guarantee: full stop. SPOE works on retrospective assessments, rather than ex ante premiums. But the industry does pay, if the senior debt isn't enough to absorb the loss. (Yes, senior debt acts as a CoCo, just like it does in Chapter 11.) This looks just like a bog-standard mutualized clearinghouse.

And generally, the senior parent debt should be enough, if the events of 2008 are any guide. There was ultimately no need to impair the senior debt of AIG, Citi, Bear Stearns and Wachovia. (That's not to say that senior debt shouldn't have been impaired if we had the legal technology to do so, just that senior debt impairment would have been sufficient.) Senior parent debt was sacrificed in WaMu, but the bank went through untouched. Only Lehman might have required risk mutualization under SPOE.

I disagree with Professor Levitin's argument that protecting bank debt creates moral hazard. This protection is mostly a priority scheme, and Modigliani-Miller seems to say that priority schemes are irrelevant. Moral hazard can only comes about if the entire senior debt stack is insufficient. This is pretty rare even for 2008 events (see above), and the market is very unlikely to be able to price such extreme tail events.

There is a pretty broad consensus in the insurance business that capital markets can't handle sufficiently extreme tail events. Is banking any different? Would Professor Levitin abolish FEMA because of the moral hazard it may create?

[Adam]

@ Ebenezer Scrooge

1. The Dodd-Frank statute also says “no bailouts.” If you believe that banks will pay full freight, perhaps you’d be interested in a bridge I have for sale… Anna Gelpern’s work has shown pretty convincingly that laws are routinely stretched or disregarded when a financial crisis is at hand. And 2008 is a nice illustration—using the Currency Exchange Stabilization Fund to guarantee money market mutual funds was a real eye roll, just as it was when that same Stabilization Fund was used in the Mexican bailout.

2. After-the-fact assessments are, by definition, not paying full freight. After-the-fact assessments cover losses, but not risk. This is the difference between indemnification and insurance. Insurance requires payment of up front premiums that are priced based on risk of loss. Indemnification is after-the-fact coverage of actual losses—no harm, no foul, essentially. One can conceive of indemnification as self-insuring, but that’s a real problem when dealing with limited liability actors that might be under-capitalized and have agency problems. Full risk internalization requires up-front assessments, not just coverage of actual losses. To illustrate, consider the US’s TARP investments. The US made money in the end. That implies no assessment for loss would be necessary. But it doesn’t compensate the US for the risk that the US took. For full freight coverage, it’s got to be pre-funded insurance assessments.

3. You're totally misreading Modigliani-Miller. M-M says capital structure is irrelevant _in a world with no bankruptcy costs_, no differential tax treatment, etc. Like the Coase theorem, it’s not a description of the real world; it’s meant to underscore why the assumptions matter. In other words, M-M says that priority _does_ matter when bankruptcy is possible. Here's we're dealing directly with a question of bankruptcy costs.

4. Insurance creates moral hazard only when the insured can control the risk assumed but is not incentivized to minimize the risk. That’s not a problem for FEMA. FEMA doesn’t create moral hazard because the risk it addresses—natural disasters—is not one the insured can control. Flood insurance is a different matter because one can choose whether to live in a flood plain or not.

5. The real weight of your argument is that equity plus the long-bonds (functionally subordinated debt) will be sufficient to absorb all losses under any foreseeable circumstance, so the possibility of the FDIC actually taking a loss isn't real. I hope that's true, but that depends on ensuring that there really is enough capital/sub-debt. And the banks will fight that because lots of equity and sub-debt is expensive. I don't trust that we will ever make the capital "levies" high enough for the 100-year flood, much less maintain those levies throughout good times. The political economy of bank regulation is just stacked against doing that. Today there's still some memory of 2008, but wait another five years and see how strongly 2008 resonates.

@ anon: GM/Chrysler were formally legal. But they were not ideal, and they perverted absolute priority. (fwiw, I've got a similar problem with the rights offerings that are becoming so common in Chapter 11 plans.) Beyond that, bankruptcy is a distributional exercise; it's not just procedural, as the Chicago school maintains. In terms of distribution, I'm much more concerned about a bankruptcy deal that harms labor, which cannot easily diversify investment and price for risk, than one that harms financial institutions, which should be diversified and pricing for risk. I don't like this ends justifies the means approach to bankruptcy, but it makes me more comfortable with Gm/Chrysler than with SPOE.

[anon]

Adam: got it. You like bankruptcy as a distributional exercise when you are doing the distributing to favored groups, even if we "pervert" the law to do it. When one borrows from Peter to pay Paul, you can always count on the support of Paul.

[Ebenezer Scrooge]

@Adam: Aha, a good rumble! I'll key my responses to your points:

1. Agreed: Laws were routinely stretched in the middle of the financial crisis and they always will be. But: The retrospective assessments occur after the crisis, not during. Looking at some of the megafines of the past few years, I don't see much shyness about Uncle sticking his palm out after the dust settles, although he does seem a bit shy about offering three hots and a cot to deserving individuals.
2. I don't understand this. Indemnification is the same thing as insurance. You're right that the fact that Uncle's large (but not 100%) recovery does not mean that there was no risk. But I don't see what this has to do with retrospective assessments.
3. I think you are totally misreading Ebenezer, and not thinking hard about M-M. The M-M seesaw means that there is nothing inherently efficient about a particular capital structure--you have to posit something beside the existence of the structure to argue for its efficiency. M-M puts the burden of explanation on the person who argues for a particular capital structure. Of course, if you believe in the finance theory of the past five years or so (Gorton), it's more efficient to put parent debt at risk ahead of subsidiary debt, because of the independent value of liquidity, risk-shifting, and the like. I argued something more modest: that there was no good argument against structural subordination of parent debt, even without belief in modern finance theory. Do you have such an argument?
4. Natural disasters are not risks that people can control? Building codes? Stilts? Backup generators? Buying insurance? Not buying a house in the valley? FEMA makes sense for a number of reasons: coordination, capital market imperfections, and just plain old paternalism. But inability to mitigate isn't one of them--most "inability to mitigate" arises from liquidity constraints on individuals and small businesses--just another capital market imperfection. I have few paternalistic feelings for megabanks, but systemic liquidity is a premier coordination problem, and capital market imperfections were patent in 2008.
5. Your point 5 is a good one. But if the banks do control our political economy as much as you think they do (and they might!), SPOE is probably the best we can get. Is there something consistent with your model of political economy that is better? It is better to protect against 50-year floods than have no protection at all.

(fwiw, I mostly agree with your response to anon, although I don't see absolute priority perverted any more in Chrysler/GM than it is in a bog-standard modern CH11. Uncle did little that a typical modern DIP lender doesn't do.)

[tw]

Adam - Do you have any thoughts on whether the adoption of one of Bankruptcy Code SIFI proposals floating around Congress would lessen the likelihood of a government bailout?