Perspective on Risk - May 19, 2023 (Bagehot is out of date)
The roof, the roof, the roof is on fire; Information, Signaling & Bank Runs; Federal Reserve’s May 2023 Supervision and Regulation Report; Crypto Runs; More
When Do You Put Out The Fire, And When Do You Let It Burn?
Let’s start with three quotes ascribed to Tim Geithner:
You shouldn’t refuse to deploy fire engines to a burning neighborhood in order to highlight the dangers of smoking in bed.
Once a run is underway, anything that increases the uncertainty of creditors about if and when they’ll get paid will exacerbate the run. Crisis responders who get obsessed with moral hazard and Old Testament justice make crises worse.
In a brutal financial crisis like ours, actions that seem reasonable – letting banks fail, forcing their creditors to absorb losses, balancing government budgets, avoiding moral hazard – only make the crisis worse (TFG)
So this, to me, begs the critical question of “when do you let the fire burn?” Tim would probably opine that you should let it burn when the spillovers and contagion are manageable, when you have the tools to contain the fire, and when it is not systemic. In other words, when you will lose the house, but not the neighborhood. n When do you let the forest fire clear out the accumulated dry tinder?
Logan Comments on Liquidity Provision
We’ve discussed Lorrie Logan’s perspective on crisis liquidity provision before. Now, as President of the Dallas Fed, she recently gave Remarks on liquidity provision and on the economic outlook and monetary policy. These are useful framing for this discussion. In the remarks she states:
One of the first lessons we are taught as central bankers is that to prevent stress and address crises should they arise, a central bank should stand ready to lend freely against good collateral.
When a central bank offers a liquidity backstop, it supports financial stability in two ways: by reassuring the public and by actually adding liquidity to the system.
In many cases, the mere availability of ample liquidity can reassure depositors, calm stresses and help private funding markets continue to function smoothly. The central bank may not even need to make any loans.
She also states, as a result of the current round of bank failures:
When phones can move money, and bank names can trend on apps on those same phones, bankers must look at contingent liquidity sources in a new light. The traditional ways to mitigate deposit flight, such as face-to-face conversations and phone calls to answer questions from depositors, may not be available in all instances. If you can’t borrow from a funding source with almost as much speed as your depositors can push buttons on their phones, it might not help you meet withdrawals.
Bagehot Is Out Of Date
Quite a while before the GFC, Walter Bagehot famously thought about the same questions. In 1873 he wrote the classic Lombard Street: A Description of the Money Market. This is one of the books in the central banking bible. Bagehot’s book delves into the functions of the money market, the role of the Bank of England, the behavior of banks during crises, and the concept of a lender of last resort. The first line I quoted of Logan’s remarks is a classic formulation of Bagehot.
Bagehot famously is known for the doctrine that the central bank should lend freely but at high interest rates and only to solvent institutions with good collateral. This specific phrase, however, is an interpretation of what Bagehot says, and is not explicitly in Lombard Street. He does state:
In a panic, the holders of the ultimate Bank reserve (whether one bank or many) should lend to all that bring good securities quickly, freely, and readily. By that policy they allay a panic; by every other policy they intensify it.
For years, economists have used Bagehot’s framing to consider revisions to how central banks operate. In particular, quite a bit of attention has been paid to the “time inconsistency” problem.
The concept of 'time inconsistency' comes from the field of economics and refers to a situation where a policy that seems optimal when initially set becomes suboptimal when the time to implement it actually arrives. In the context of Bagehot's lender of last resort doctrine, the time inconsistency problem emerges from the potential conflict between ex-ante (before the event) and ex-post (after the event) incentives for both the central bank and financial institutions.
During normal times, the central bank would like banks to act prudently and avoid excessive risk. But if it is known that banks will be bailed out in an emergency, this might lead the central bank to lend at lower rates or to institutions that wouldn't normally qualify for assistance. One can argue that the Fed is doing this now with their Bank Term Lending Program.
This is commonly referred to as “moral hazard;” the same Old Testament justice Tim Geithner suggested we ignore.
In 2016, former BoE Gov. Mervyn King gave a lecture titled Rethinking Crisis Liquidity Provision. Here, he states:
The problem is that the financial world today is very different from that which Bagehot had observed. In particular, Bagehot implicitly assumed that banks would always have a sufficient quantity of “good” collateral to cope with a panic. In that case, the central bank would merely be swapping money for “good” collateral – a pure liquidity crisis.
Nothing could have been further from the truth in 2007-08. Collateral against which central banks could safely lend was in short supply. Even in Bagehot’s time, Thomson Hankey, a former Governor of the Bank of England, had written well before Lombard Street appeared, “the banking community must be undeceived in the idea that promises to pay at a future date can be converted into an immediate payment without a supply of ready money adequate for that purpose”. The provision of cheap central bank insurance is an incentive to take excessive risks.
I’d like to suggest that the financial world is much different from that of the 1870s, and that along with the rethinking of deposit insurance guarantee we need to revisit Bagehot.
In particular, I think three changes to the financial system drive this need for rethinking.
First, there was no deposit insurance in Britain in 1873. Deposit insurance was introduced there in 1917. There is no mention of deposit insurance in Lombard Street. It should be noted that there was a bank examination regime in place that dates back to 1844.
Second, bank balance sheets are much more transparent that they were back in the 1870s. Mandating bank balance sheet transparency has been a tool used to reduce moral hazard by making it more difficult for banks to take excessive risk, as the “market discipline” exerted by uninsured depositors, and debt and equity holders would raise riskier firm’s cost of capital.
The third change is the speed with which information is transmitted, and with which money can be moved from a suspect bank. It is this third element that was most highlighted in the recent set of failures.
Speed of transmission affected the crisis in at least two ways. The principal way was how quickly uninsured commercial deposits were withdrawn. At its peak, SVB reportedly saw $1mm+ leave each minute. The second was that there was insufficient time for the Discount Window to evaluate collateral to lend against. The speed with which deposits flew was significantly faster than new funds could be obtained through the Discount Window.
The standard response to this problem has been the calls by Paul Tucker, Former BoE deputy calls for radical overhaul of bank funding, and Mervyn King, We Need A New Approach To Bank Regulation, to mandate changes to the timing, amount and quality of collateral pre-positioned at the Discount Window to facilitate easier funding during the crisis. Having only externally-rated bond collateral reduces the time to evaluate the collateral. Having a quantity sufficient to cover 100% of uninsured deposits insures liquidity. Both of these have drawbacks on credit creation, but more importantly both rely on a minimalist reading of Bagehot’s dictum of lending on good collateral.
Logan has similar thoughts in her remarks:
I strongly believe that every bank in Texas and every bank in our country should be fully set up at the discount window as part of its liquidity toolkit.
Bagehot commented on the level of “reserves” or liquidity insurance that a central bank needed to maintain to cover deposit flight:
That the amount of the liabilities of a bank is a principal element in determining the proper amount of its reserve is plainly true; but that it is the only element by which that amount is determined is plainly false. The intrinsic nature of these liabilities must be considered, as well as their numerical quantity. For example, no one would say that the same amount of reserve ought to be kept against acceptances which cannot be paid except at a certain day, and against deposits at call, which may be demanded at any moment.
Nor can you certainly determine the amount of reserve necessary to be kept against deposits unless you know something as to the nature of these deposits. If out of 3,000,000l. of money, one depositor has 1,000,000l. to his credit, and may draw it out when he pleases, a much larger reserve will be necessary against that liability of 1,000,000l. than against the remaining 2,000,000l. The intensity of the liability, so to say, is much greater; and therefore the provision in store must be much greater also.
So Bagehot does start with the implicit understanding that banks are being examined, and that it is the role of the central bank to a priori deal with weak banks. He then assumes that the banks that come to borrow are, for the most part, solvent but illiquid, and that the pace of outflows can be mitigated in sufficient time by a central bank lending against collateral.
Logan seems to admit that the Fed isn’t really following Bagehot’s dictates these days anyway.
When banks establish access to the discount window, they can also gain access to the Bank Term Funding Program (BTFP) … We lend through the BTFP based on the collateral’s par value, even if rising interest rates have reduced its market value.
Nice euphemism. She really should have added “below the loan value” to that last sentence. Doesn’t sound like Bagehot’s “well secured.”
Similarly, Bagehot’s “high rates” has gone bye the bye.
Discount window loans are priced a bit above prevailing market rates …
How to change things
So here is where I might change things.
If we accept that the central bank should have access to inside information on the banks even with strong transparency, the central bank ought to be able to evaluate the solvency of banks in advance of a crisis. Lending involves making credit decisions, and the central bank ought to be able to make a decision on the credit worthiness of a bank at least as well as it can on an individual piece of collateral.
The presumption in Bagehot is that the central bank is available to provide ‘on demand’ liquidity. This, in effect, gives the bank the option to draw liquidity from the central bank (again, subject to collateral)
But what if, instead, we made banks pay for the option to get liquidity from the central bank. In fact, what if we required banks to obtain committed central bank liquidity is a ‘to be determined’ amount.
The sizing of the committed liquidity would be related to the nature of the deposits, as proscribed by Bagehot. There could be differential rates for different types of deposits.
In theory, the moral hazard risk could be mitigated by differential market-based (or worse) pricing. One could also consider requiring private-sector participation in the first-loss position of any such option.
My conception puts the central bank less in the role of the fireman “putting foam on the runway” and more in the role of designing a safer system. Or going back to the first Logan line I quoted:
One of the first lessons we are taught as central bankers is that to prevent stress
and address crises should they arise, a central bank should stand ready to lend freelyagainst good collateral.to banks assessed as solvent.
We should cease too to be surprised at the sudden panics. During the period of reaction and adversity, just even at the last instant of prosperity, the whole structure is delicate. The peculiar essence of our banking system is an unprecedented trust between man and man: and when that trust is much weakened by hidden causes, a small accident may greatly hurt it, and a great accident for a moment may almost destroy it. - Walter Bagehot
Information, Signaling & Bank Runs
When all is said and done, the change that policymakers will need to struggle with is the rapidity with which wholesale deposits can now be withdrawn from banks. There will be a lot of analysis and policy proposals. Haelim Anderson and Adam Copeland wrote a helpful blog post putting the SVB and Signature Bank runs into historical perspective. Very simple, clear paper.
Banks Runs and Information (NY Fed Liberty Street)
In this post, we highlight how these [SVB, Signature, First Republic] banks, with their concentrated and uninsured deposit bases, look quite similar to the small rural banks of the 1930s, before the creation of deposit insurance. We argue that, as with those small banks in the early 1930s, managing the information around SVB and SB’s balance sheets is of first-order importance.
Deposit insurance plays a key role in the management of information about banks by taking away a depositor’s incentive to monitor a bank—making depositors information-insensitive. Wanting to maintain some market discipline, Congress capped the size of insured deposits, with the goal of creating two types of depositors: small, insured depositors and large, uninsured depositors. Because the larger depositors remain uninsured, they are likely to continue to monitor banks and thus restrict how much risk a bank places on its balance sheet.
… the recent bank runs and subsequent discussions of overall banking stability do not seem to reflect issues with changes to the overall coverage of deposit insurance.
Rather, we argue that SVB and SB are in the unusual position of looking like the small banks of the 1930s, before the FDIC offered deposit insurance. Both SVB and SB had a depositor base that looked local, in that depositors reportedly interacted with one another in their regular business dealings. SVB’s depositors were connected through venture capital networks and SB’s depositors were connected through law firm networks. Further, these depositors were largely uninsured and therefore sensitive to (negative) information about their bank’s balance sheet. Finally, the majority of SVB’s and SB’s depositors could be considered financially naïve, at least relative to the financial firms that drive the majority of uninsured deposits in the United States today.
Federal Reserve’s May 2023 Supervision and Regulation Report
Federal Reserve’s May 2023 Supervision and Regulation Report to Congress is out. Here is Barr’s remarks that accompanied the report.
Interestingly, they start off with this little defensive tidbit:
During 2022, Federal Reserve supervisors began preparing for the increased possibility of a more challenging economic environment for banks. In view of unprecedented growth of deposits during the pandemic and questions about how depositors would react to more adverse conditions, supervisors focused on assessing firms’ ability to manage risks related to liquidity. Supervisors also undertook additional examination work to evaluate interest rate risks and the impact on firms’ funding options.
Later on, in Supervisory Priorities, they note:
Anticipating a more challenging banking environment, the Federal Reserve focused examination work during 2022 on assessing the adequacy of bank risk management in addressing the impact of higher interest rates on liquidity, asset values, and credit quality. …
Work to date has included horizontal assessments of contingency funding plans at firms supervised by the Large Institution Supervision Coordinating Committee as well as of liquidity risk management at large and foreign banking organizations. In addition, the Federal Reserve has increased supervisory activities at community banking organizations (CBOs) and regional banking organizations (RBOs) with elevated interest rate risk exposures. For those CBOs and RBOs with significant securities depreciation or otherwise exhibiting elevated interest rate risk, the Federal Reserve is conducting targeted examinations to assess the adequacy of their liquidity and interest rate risk management.
Note the present tense “is” in the last sentence.
I actually don’t doubt that this 2022 focus is true. Examiners read the paper and saw what was coming. In many ways, the exams that the Fed chooses to conduct often have a ‘signaling value.’
I hadn’t read these reports before so I went back and took a look at the November report.
They had a boxed item titled Box 3. Effects of Securities Depreciation on Banks’ Capital and Liquidity Positions that pretty well hit things on the head:
Securities holdings at banks rose to a record high in 2022, largely driven by the deposit surge that followed the onset of the pandemic (figure A).1 Banks added nearly $2.3 trillion in securities from the start of 2020 to the end of 2021, when interest rates were low.
As interest rates increased in 2022, the fair value of securities held by banks fell significantly. (The fair value of securities is generally inversely related to interest rates.) Securities depreciation attributable to available-for-sale securities resulted in $224 billion of unrealized losses as of June 30, 2022 (figure B).2 Unrealized losses on available-for-sale securities are included in accumulated other comprehensive income, reducing banks’ tangible book value. Lower tangible book value can adversely affect stock price valuation in periods of stress or market participants’ capital assessments. For some large banks, these unrealized losses also reduce their regulatory capital.
In addition, liquidity regulations require certain large banks to hold a prescribed level of liquid assets. Eligible securities are included in the calculation of liquid assets at their fair value. As a result, these large banks saw their liquid asset buffers decline because of the loss in value of eligible securities. However, large banks continue to meet their regulatory liquidity requirements.
Barr’s remarks have a few interesting tidbits too. Relative to his earlier testimony, he comments more on capital requirements:
SVB's failure confirms the importance of strong levels of bank capital. While the proximate cause of SVB's failure was a liquidity run, the underlying issue was concern about its solvency—the bank's ability to absorb the losses on its securities and repay its depositors and other creditors. … Stronger capital will guard against the risks that we may not fully appreciate today and reduce the costs of bank failures.
With respect to capital, we need to evaluate whether our capital requirements appropriately measure the ability of banks to absorb losses. Had SVB been required to reflect declines in the face value of available-for-sale securities in its capital, it may have held more capital to cover these losses.
Today, for example, the Federal Reserve generally does not require additional capital or liquidity beyond regulatory requirements for a bank with inadequate capital planning, liquidity risk management, or governance and controls. I believe that needs to change in appropriate cases. Higher capital or liquidity requirements can serve as an important safeguard until risk controls improve, and they can focus management's attention on the most critical issues.
This last bit is more akin to the approach in the UK. I am a fan, if done correctly.
Crypto Runs
Economists from the FRB-Chicago have written A Retrospective on the Crypto Runs of 2022.
In this article, we describe the spectacular collapse of several crypto-asset platforms in 2022 following investment losses and widespread customer withdrawals.
One view of customers’ withdrawals from crypto-asset platforms in 2022 is that market discipline has effectively punished firms that took excessive risks and committed other abuses in attempting to pay high returns to customers.
Most of the financial instability associated with these failures has been confined to the crypto-asset ecosystem. However, the failure of FTX put severe pressure on Silvergate Bank and Signature Bank in late 2022, which both provided an array of banking services targeted at crypto-asset firms; these two banks suffered severe deposit withdrawals and closed or failed in 2023, contributing along with the failure of Silicon Valley Bank to the recent banking turmoil.
To understand the large size and rapid pace of the runs, we make the following four observations:
We note that customers of these crypto-asset platforms did not have access to protection akin to deposit insurance, and so they should have been expected to behave more like uninsured depositors.
Second, while the platforms had many retail clients, the runs were spearheaded by customers with large holdings, some of which were sophisticated institutional customers.
Third, the runs were purely electronic, with customers withdrawing funds through apps or web portals with ease.
Fourth, efforts to generate high returns caused these platforms to hold insufficient liquidity buffers.
This could have been written about SVB.
Bank Runs During Crypto Winter (Gorton, Zhang)
“Crypto Winter” refers to a systemic event that occurred in the cryptocurrency ecosystem—what we call “crypto space”—in 2022. Crypto space was wracked by plummeting crypto prices, the troubles of a large crypto hedge fund, and runs on many crypto lending platforms.
We begin with two observations.
First, despite mass marketing campaigns to the contrary, crypto lending platforms recreated banking all over again. Crypto lending platforms were vulnerable to runs because, like all banks, they borrowed short and lent long. This is the essence of banking, so we label these lending platforms “crypto banks.”
Second, crypto space was largely circular. Once crypto banks obtained deposits and investments, these firms borrowed, lent, and traded mostly with themselves. As a result, Crypto Winter did not cause the kind of financial turmoil that we witnessed in either 2008 or 2020, and it did not cause an economic recession.
Calls For Papers
Federal Reserve Stress Testing Research Conference will be held October 19-20, 2023. Paper or extended abstract submissions due by June 15, 2023. https://www.bostonfed.org/news-and-events/events/federal-reserve-stress-testing-research-conference/2023
Conference on Frictions in Real Estate and Infrastructure Investment will be held November 2-3, 2023. Paper submissions due by July 1, 2023. https://www.federalreserve.gov/conferences/frictions-in-real-estate-and-infrastructure-investment-conference-call-for-papers.htm
What I’m Listening To
Käärijä Cha Cha Cha (30 sec Youtube).
My Finns got robbed. Clearly should have won Eurovision.