Can We Prove a Bank Guilty of Creating Systemic Risk?

That is a new paper by Jon Danielsson, Kevin R. James, Marcela Valenzuela, and Ilknur Zer, forthcoming in the JMCB, here is the abstract:

Since increasing a bank’s capital requirement to improve the stability of the financial system imposes costs upon the bank, a regulator should ideally be able to prove beyond a reasonable doubt that banks classified as systemically risky really do create systemic risk before subjecting them to this capital punishment. Evaluating the performance of two leading systemic risk models, we show that estimation error alone prevents the reliable identification of the most systemically risky banks. We conclude that it will be a considerable challenge to develop a riskometer that is both sound and reliable enough to provide an adequate foundation for macroprudential policy.

Here is their broader piece on models of systemic risk.  Paul Krugman had a good argument today in his column on this topic:

What determines whether a firm is systemically important? There aren’t any cut-and-dried rules — there can’t be, because if there were, corporate lawyers would find ways to evade them. Instead, it’s a judgment call. But financial giants that don’t like being regulated are trying to use litigation to question those judgments.

Maybe, but I’m reluctant to define “following the law” as necessarily an attempt to game the system.  Would we use that same argument to show restrictions on leverage are counterproductive?  Quantifiable laws can indeed limit financial risk and of course Dodd-Frank, and many other financial regulations, is replete with such quantifiable laws.  Why is quantification suddenly so ineffective or even counterproductive?  We don’t from Dodd-Frank advocates hear much about this possibility in other contexts.  Furthermore, cannot the discretionary judgments of the regulators be gamed, and gamed all the more so?  Especially in generations two, three, and four of this process.  Especially in a world with a revolving regulatory door.  Especially in a world where regulators are looking for cover and, in subsequent iterations, may simply fall back on numerical standards in any case?  I say when in doubt, go with the rule of law, not the rule of men.

So I’m sticking with my previous view that the MetLife decision is more good news than bad.


"What determines whether a firm is systemically important? There aren’t any cut-and-dried rules" Congress and the elites can do anything.

"What determines whether a firm is systemically important?" The politicians, and quasi-politicians, on its payroll.

Ask Microsoft. If you contribute to K street, you're not a monopoly.

Or if you show the NSA your backdoors.

The government approves all bank mergers. Government regulates insurance companies. Where do they get off throwing stones? How does Canada do it? Maybe the ridiculous amount of regulations has created a barrier to competition by new, smaller firms. I have read estimates that you need $300 million in assets/deposits to be a profitable bank and cover the cost of regulation. And that was a few years ago.

Not only new, start-up banks, the American community bank is an "endangered species."

All Dodd-Frank did was create more bureaucracy and political correctness. The Office of Women and Inclusion? Pleeeeease.

I had to google that one. Even the way things are today I thought that was too stupid to be true.

Canada does it by only having a very few banks (5 ?)

They are also fairly conservative by nature. They tend to significantly exceed minimum regulatory requirements with regard to risk and capital. Canadian banks tend to pad their profits in a lot of ways that aren't popular, but a degree of willingness to legislate changes keeps them from going too far with it. There is a lot less dogmatism even from the right wing that banks must be permitted to do whatever they want in order for the best results to be achieved.

In competitive markets, profit margins are held down by the threat of new entrants. If the regulations make it such that no entrants are unable to gain traction, we should expect to see rising profits for established firms and ever greater concentration in that industry.

That sounds exactly like what we're seeing in the US finance sector, no?

Interesting. Of course, banks can just blame the regulations for the fee hikes.

Fine Mr. Krugman, if you think that Firms should have to have higher capital requirements somewhat arbitrarily, then have the Federal government pay for the higher reserves. Merely declare how much in excess you require and have the Treasury compensate the affected firm to the difference between what the average reserves earned and the average non-reserve assets earned.

The FED pays interest on reserves.

Yes, but at a lower rate than the banks can get on non-reserves. If the Fed once to designate certain banks as special, then pay the banks a higher rate for the additional required reserves.

Isn't the idea that banks should be covering the costs of their own risks, and/or regulated in ways which prevents them from creating risks that negatively affected people who are not sharing in the profits?

Without wanting to take a position on the degree to which such perspectives may be legitimate in different circumstances ... finance and banking is one of those sectors that leads to expressions like "capitalism for the poor, socialism for the rich."

So we should create certain policies to mitigate system risks and then create additional policies that create incentives to achieve that classification? Sound's like perfect DC behavior to me.

OK, but the government gets to claw back the money it previously paid out once the institution starts earning less than the risk-free rate on its non-reserves -- and the management team should be personally liable without limit for paying back the government in the event such losses materialize. If that was the rule, I suspect most institutions would gladly opt for higher capital requirements.

"OK, but the government gets to claw back the money it previously paid out once the institution starts earning less than the risk-free rate on its non-reserves "

Why would there be any money to claw back? The organization can easily assess the returns on it's non reserve assets quarterly (or yearly) and the Fed can pay the difference on the required reserves. There shouldn't be a payout above the recorded difference with periodic third party audits.

"Since increasing a bank’s capital requirement to improve the stability of the financial system imposes costs upon the bank"


This kind of backwards thinking is probably a sign of the times.

Allowing a bank to maintain an anemic - and frankly fraudulent - capital requirement imposes costs upon the consumers by reducing the stability of the financial system.

The bank does not incur "cost" by improving its capital reserves. It incurs costs by NOT maintaining adequate capital reserves, and those costs are either subsidized by the government or legal liability for those costs is waived entirely.

This might be slightly persuasive if bank failures--as opposed to failures by non-bank entities like Lehman, Fannie Mae, and Reserve Fund--had been a cause of consumer losses in the recent downturn.

Which is why Goldman Sachs elected to become a bank when the shit hit the fan

Both Goldman & Lehman owned deposit taking banks.

"The bank does not incur “cost” by improving its capital reserves. It incurs costs by NOT maintaining adequate capital reserves,..."

Then have the Federal government pay for the difference in the cost of reserves between the general reserve requirement and any higher standard. That seems fair enough.

You mean have the taxpayers pay. Have the taxpayers pay to maintain reserves on money that doesn't exist, that banks then lend out and profit from.

Sounds great.

"Have the taxpayers pay to maintain reserves on money that doesn’t exist". I don't understand what you mean when you say the "money that doesn't exist"?

I tie in with (Sigh) ...

On a related note: What about not policing systemic risks ex post but preventing the build-up ex ante? This has been proposed by Jonathan McMillan, for instance:

Externalizing the cost of risk-taking is the premise of incorporation. Fractional reserve banking is fraud. Both require the sanction of government policy to operate in a market economy. The state is not interested in maintaining a functioning market, beyond what it takes to satisfy special interests and keep them running.

Well, maybe in practice, but that's not how it's supposed to be. Incorporation allows people to pool risk, and this risk is supposed to be priced into their expectations of returns (e.g., higher interest rates if they take out a loan). Allowing people to evade personal liability for business losses allows them to take more risks, but this is still supposed to be priced into the cost of capital. The risk is not supposed to be externalized.

It's an involuntary association. The only agreement an incorporated company makes is with the government. It doesn't ask third parties whether they agree it's own liability or the liabilities of its shareholders should be limited. That risk becomes an external cost.

"Incorporation allows people to pool risk"

Interestingly, most of the biggest American investment banks did not start incorporating until the 1980s. Before, they were organized as partnerships and the partners seem to have been very careful about risking the company's money because it was also their own money.

You're thesis rests on the assumption that the marginal reduction in expected distress costs is always greater than the marginal cost of increasing capital. That's simply not true in all states of the world. Take the extreme limit, assume a bank held well in excess of 100% of very safe capital relative to liabilities. Since the expected probability of bank failure is 0%, increasing capital reserves will have no mathematical impact on distress costs. Yet the bank's capital costs will still rise.

Even accounting costs socialized by the legal system and government guarantees, capital reserves are subject to decreasing marginal returns. While equity capital costs are basically linear. While the privately optimal leverage will be higher than the socially optimal leverage, there is still a point where banks are taking on too little risk, not too much. Whether we are past that point or not is an empirical question. But the fact that liquidity is now barren in many critical financial instruments, like nearly all major corporate bonds, suggests that we've overshoot financial risk constraints post-2008.

"You’re thesis rests on the assumption that the marginal reduction in expected distress costs is always greater than the marginal cost of increasing capital."

Name one state in the world that outlaws fractional reserve banking.

Double sigh. In the paper we assume that a financial crisis imposes costs upon society and that increasing bank capital ratios lowers the probability of a crisis. On the other hand, increasing bank capital ratios also has a private and a social cost. That there is a private cost has got to be beyond dispute (why else is MetLife expending real resources here?). Numerous central bank studies (which we cite in the paper) also find that increasing bank capital levels has a social cost. The regulator in the paper acts to minimize the sum of the expected social cost of a crisis and the social cost of higher capital ratios. The net value of the policy options available depends upon the regulator's ability to identify systemically risky firms.

"Numerous central bank studies (which we cite in the paper) also find that increasing bank capital levels has a social cost."

Shocking, that a central bank allowed to engage in fractional reserve banking would find that maintaining adequate capital reserves would be "costly" to its business. I am shocked, I declare.

Depending on a regulator’s ability to identify risk is like handing a sextant to a farmer and asking him to navigate a ship across the ocean.

"Maybe, but I’m reluctant to define 'following the law' as necessarily an attempt to game the system."

This is an odd formulation. Isn't that what "gaming the system" often is - finding an activity that is technically (or at least straight-faced arguably) allowed, even though it was not meant to be allowed, and taking advantage of it?

I do support as clear of numerical limits as possible, however. Discretion leads to corruption.

Discretion will be used to punish their ideological enemies.

What constitutes "capital"?

Many pension funds were very well capitalized in 1999-2000. Especially the California public employee pensions.

Coal mining companies indemnity funds for cleaning up mines were well capitalized in 2010.

Credit insurers were well capitalized in 2006 and thus could easily insure hundreds of billions in mortgage securities.

Dodd-Frank and other actions has required Basel accords make major changes in the global definitions of capital.

Capital quality was defined by credit rating agencies who were paid by those issuing securities for sale who needed high ratings to sell them.

I'd like Tyler to clearly defined "capital" to prove that it's simple definition, and not an art (in artisan, craftsman, sense).

Until a clear definition of capital is made, setting capital ratios is subjective.

According to regulators in US, Banks have Tier 1 capital consisting of equity capital plus certain kinds of preferred stock. Tier 2 capital is Tier 1 capital plus a portion of the allowance for bad debt debts plus subordinated debt with a maturity in excess of 5 years. Basel has certain other requirements.

Of course, the adequacy of capital depends on the quality of and risk in assets.

Not to mention interest rate and market risk.

Which is why regulators used risk-based capital as one (of several) measures of the equity backstop.

Not all capital is created equal. As stated above, only certain types of capital are included in Tier 1 capital.

Yes, it is subjective. There is no if and or but about it. You establish many tiers of capital according to liquidity, market risk, etc., and assess them subjectively to arrive at a risk-adjusted capital level. There will always be an element of subjectivity about it.

RE: "Here is their broader piece on models of systemic risk."

Authors abstract says "During calm periods, the underlying risk forecast models produce similar risk readings; hence, model risk is typically negligible. However, the disagreement between the various candidate models increases significantly during market distress, further frustrating the reliability of risk readings."

Minsky, pure and simple. The decline in "model risk" may not be particularly important because they may all be equally wrong.

I am not sure, but I think Tyler is choosing his preferred mode of failure.

Perhaps everyone is. They know there will be more crises. They know there will be future bailouts. Commentators high and low just want to signal group membership as background music.

Part of that will be people who sing "no regulation" in the first act, exit the stage, and return for the final "I never wanted a bailout" chorus.

There were no significant deregulatory actions in the banking sector leading up to 2008.

That's just wordplay, Glass-Steagall not "significant" or whatever.

The minor removal of one provision of Glass-Steagall had no contributory effect whatsoever on the financial crises. It just so happens to be the ONLY deregulation move by the federal government leading up to 2008, which is why Progressives latch onto it. But it had nothing to do with the crises.

Would congressional ratification of a FSOC SIFI determination convert the determination from "the rule of man" and into "the rule of law"?

Would FSOC's SIFI be "the rule of law" if formulated using words that do not specifically identify a particular institution (e.g., SIFI "includes any company engaged in the business of insurance having consolidated assets in excess of $500B, other than (i) a bank (within the meaning of the BHCA) regulated by the OCC or the Fed, (ii) . . . .")?

Congress already granted FSOC substantial discretion in determining what is and is not a SIFI. The problem is that FSOC defined a standard, then didnt follow the standard or support its determination with reasoned evidence. In addition to the criteria explicitly stated by Congress, Congress granted FSOC the power to use its own judgment to determine what a SIFI is.

The judgment clearly stated that FSOC was free to change the rules, but it must do so in a reasoned way and communicate that to the industry.

Put another way, men make laws, but then they are bound by the laws they make. FSOC cant change the laws on the fly.

Yep, just curious the specifics of Tyler's concerns (and perhaps also my own). E.g., maybe even a FSOC SIFI designation wouldn't raise concerns if it became effective only after Congress ratified it. (Sure, that would raise other issues.) Or maybe we should be ok with an approach that requires FSOC to designate classes of institutions - not specific institutions - as SIFIs. (Sure, that might be a distinction without much a difference.)

This SIFI process does seem somehow unique. The agencies have the authority to designate which institutions will be subject to their requirements and then even the authority to devise pretty much whatever requirements they want for each such institution. And all without much in the way of cost/benefits. Separate and apart whether that's consistent with admin or constitutional law . . . is this the optimal framework?

This judgment demonstrates that agencies must consider costs and benefits, not only at the stage of prudential regulation but in SIFI designation itself.

The problem with these nonbank firms is that their systemic risks (if they exist) are idiosyncratic. A customer approach to regulation is likely the correct approach. But there is a rational fear that regulators cannot fashion prudential regulation any better than the market solution, I.e. government failure. Regulators can cause the very instability they intend to prevent, and this fact was part of the reasoning in the MetLife judgment.

Rules are subject to congressional review if they have a large enough impact. Otherwise I don't understand what ratification has to do with this. FSOC derives its power from Congress, and Congress can amend or take it away.

I'm not honestly sure how much this discussion matters, given how complex the rules and question at hand are. I work in a much less involved area of finance and not one of investment banks I've dealt with agree on what the tax laws say in my sector. Not one. Despite spending tens of millions on accountants and attorneys. The tax code I deal with is much shorter and less complicated than Dodd Frank. It's very hard to write simple rules in finance, and if you can't write rules simple enough for well intentioned and funded investors to comply with then you have a de facto discretionary system. The discretion simply lies with prosecutors and judges instead.

The law GRANTS discretion to FSOC. But FSOC has to make rules and then stick to them. These rules must have a rational basis. They must be consistent in their application. They cannot change rules without a reasoned explanation.

The arbitrary and capricious standard means that judges CANNOT substitute their judgment for the agency. The judge may only determine whether the agency had reasons for its actions and followed appropriate procedures. Under that standard, if a rule is arbitrary or capricious, the judge MUST set it aside. But otherwise agencies are afforded extraordinary deference. They don't even have to be RIGHT in their decisions. They only have to have a reasonable explanation in advance. But they are not allowed to come up with a reasonable explanation after the fact. Their explanation has to come BEFORE or AT THE TIME of their decision. They can even change their minds, provided they give a reasoned explanation for doing so.

This legal standard is not particularly onerous. The fact FSOC failed to meet it is concerning about the judgment and processes of FSOC.

Here is the law.

"(1) DETERMINATION.—The Council, on a nondelegable basis and by a vote of not fewer than 2⁄3 of the voting members then serving, including an affirmative vote by the Chairperson, may determine that a U.S. nonbank financial company shall be supervised by the Board of Governors and shall be subject to prudential standards, in accordance with this title, if the Council determines that material financial distress at the U.S. nonbank financial company, or the nature, scope, size, scale, concentration, interconnectedness, or mix of the activities of the U.S. nonbank financial company, could pose a threat to the financial stability of the United States.
(2) CONSIDERATIONS.—In making a determination under
paragraph (1), the Council shall consider—"

and then Dodd-Frank goes on to list factors (A) through (K) that must be considered.

- point 1 - if the agency skips one of the factors in the list (A) through (K) in its rulemaking record, it leaves itself open to the possibility of being successfully challenged if the standard is arbitrary and capricious.

- point 2 - FSOC does not need to provide "proof" in its determinations for nonbanks such as MetLife. But it does need to provide plausible evidentiary support for the proposition that firms that are highly leveraged can contribute to financial instability, or that markets that rely on funding collateralized by the underlying good itself can suffer bouts of instability, or that.... The courts will be deferential if the rulemaking record is robust, even if outside experts challenge with quality statistical counter-evidence.

You still need good management. A lot of banks started speculating in real estate the day after the examiners left. Many got caught up in the real estate frenzy. They made speculative bets instead of calculated decisions. Howmany banks failed because they bet the ranch on Fannie and Freddie and then they failed? History repeats itself. The regulators have to set the standards for the culture. If Bill Taylor (former head of FDIC) had still been alive, I doubt the subprime crisis would have got as far as it did. You need leadership that understands banking and is not afraid to call out politicians with good intentions. We have too many technocrats in charge of our regulatory agencies. They are counting angels on the head of pins instead of understanding business. The agencies have been also politicized over the last 8 years. Too many of the executives have political agendas and are promoted for political reasons. Wake up people. There is no substitute for knowledge, experience and technical expertise except unemployment.

Why can't you have both rules and discretion? E.g., have the hard limit, but also have a standard where you've basically saying, "if you're just trying to just find a way to get around the limit, then we're going to slap you."

Bank regulation is not a criminal punishment; next

They are if they are not applied to all equally.

Bank regulations are provided for in law, imposed by entities given their powers by the law, and the courts interpret the law. Does Cowen deny that current regulations are part of the "rule of law"? Do we all get to deny the legality of laws we don't like?

We generally try to formulate law so it is NOT subject to discretionary interpretation. If it is, then it's not good law. Law should not be something only an elite, privileged few may interpret for the rest of us heathens to understand and obey.

Interpretation? Of course. But discretionary applications in other regards happen all the time. DAs decide to prosecute or not to prosecute.

Discretionary enforcement and discretionary interpretation are two entirely distinct matters.

"Do we all get to deny the legality of laws we don’t like?"

No. We get to challenge them in court.

Exactly. Bring a suit to court and win that way. Don't insist that law you dislike isn't law because not enough math.

"a regulator should ideally be able to prove beyond a reasonable doubt"

This is a very unusual standard and is almost certainly not the legal standard under the law in question.

Almost universally, the burden of proof that applies to regulatory and civil action (i.e. decisions that don't result in criminal penalties of incarceration), is a "preponderance of the evidence" standard. This means that action is appropriate if the fact finder determines that something is more likely than not to be true, a much, much lower standard of proof than guilt beyond a reasonable doubt which is appropriate in the far more uncertain area of regulatory law.

Proof that it is more likely than not that a bank poses a systemic risk as that term is defined by the regulator (since we defer to regulators in interpreting regulations that they are charged with enforcing under the Chevron case), is a much more manageable task.

You are correct that 'reasonable doubt' is not the correct standard, but neither is 'preponderence of the evidence.'

The correct standard is arbitrary and capricious. That is, FSOC may make decisions upon its own discretion and experience. This discretion is granted deference UNLESS its decision is not based upon a reasoned analysis of relevant factors, and consistently applied.

FSOC failed because it changed its standards without explanation, and did not adequately address relevant factors, I.e. costs. One could argue about whether a particular analysis is reasoned, but one cannot defend an analysis that doesn't exist. FSOC never considered costs.

I suspect that the current regulatory emphasis on discouraging large financial institutions from being large will not contribute much to financial stability.

In a benign macro environment, even a large financial or interconnected financial institution could go under without much wider systemic impact (e.g. Barings, Refco, MF Global, to name a few examples). But when sentiment in the financial markets sours, even moderate losses at small institutions can cascade into systemic crises. In those situations, the problem isn't the size or interconnectedness on the institutions per se. It is the correlation. Consider market market funds during the 2008 crisis. A small loss on Lehman bonds at a single money market fund triggered a loss of confidence throughout the entire money markets, forcing the Fed to backstop an entire asset class! But it is hard to argue that any particular money market fund was large or systematic or especially interconnected. Quite the opposite. And yet they had a systematic crisis nonetheless. Why? Because a loss at even a small fund or institution can trigger a crisis if the market perceives that many other similar funds or institutions could have similar problems. It is the correlation that triggers the crisis, not necessarily the size or interconnectedness.

The unintended consequence of the current policies towards banks is to push much risk into places where the risk can't be easily perceived. Result: The Fed will be forced to bail out a whole new set of players (large bond market funds, perhaps?) that they never anticipated would need a backstop.

Once again, Consider:

There is NO system - no unitary function.

What bodies of authorities "regulate" are particular relationships and/or the determinations of the circumstances in which they transpire.

"Systemic" is inferred from the possibilities that any one or more set of relationships will (or can) affect other (not directly connected) relationships.

How is that inference made demonstrable.

More narrowly, there is no reason at all for the Fed to have oversight of a life insurance company like MetLife. There's no discretionary liquidity crunch possibility like there is for a bank (people don't choose to die en masse to collect the benefits), and no systemic effects. Seen from the point of view of the insurance industry (yes, I work in that industry, but not for Met), determining Met as systemically important displayed such an appalling ignorance of the economics of the life insurance industry that it had to be seen as an abuse of regulatory discretion. The only reason Met was targeted was because they took Treasury money in 2009.

That was one of MetLifes arguments which the judge never reached because she already had sufficient cause to grant the motion to dismiss.

Sometimes I am reminded that most economists are terrible lawyers (and vice versa). This is one of those times.

If you start from the major premise that every regulation must be able to satisfy the standard of the criminal law (beyond a reasonable doubt) as to every person affected by it, then indeed, almost no regulation can ever be justified. But that's not the law and never has been. Can you show beyond a reasonable doubt that prohibiting me from hunting out of season is necessary to preserve duck populations? Almost certainly not. Luckily, that's not the standard.

Often times.

While I welcome and expect people to become engaged in important issues beyond their areas of expertise, I usually expect them to admit some ignorance and defer somewhat to experts. Clearly, one should not fall into Appeals to Authority or ad hominems, but people should acknowledge the limits of their own understanding.

One needn't be a lawyer to appreciate what happened in this case. An educated layman can understand the judgment. As with most prominent legal cases, people appear concerned most about the personal ramifications of the outcome rather than the law and legal reasoning behind it. Mood affiliation much?

Seems like another legal aspect might be considered. Some business activities include conditions of Strickt Liability -- in other words even if they take all reasonable, and even unreasonable, precausions to prevent others from bening hurt the business will still be held liable for any harm. A similar view of banking insitutions greater than some defined size or engaged in a collection of defined activities could be faced with a similar standing. That might effectively leave the market to decide the organizations level of systemic impact and that assessment would be reflected in the market capitalization of the issues equity.

Could the FDIC act like a private insurer and refuse to insure a bank that is estimated to be too risky or too big.

Comments for this post are closed