There is little risk of a crisis in the banking sector, but that does not mean there aren’t badly run banks. QE and Basel III have made the banking sector significantly more liquid and resilient such that a replay of the GFC is very unlikely. However, individual banks under poor management can still be subject to bank runs. At a high level, a bank has short-dated liabilities and longer dated assets. A well managed bank holds enough assets to meet potential outflows, but also manages its liabilities to avoid substantial outflows. This post reviews the basic liquidity problem banks face, shows that SVB was particularly poorly run, and suggests that higher rates and QT will challenge bank liquidity in the coming years.
Asset Values
Banks don’t have to mark all their assets to market, but unrealized losses have a direct impact on their available liquidity. All assets on a bank’s balance sheet fluctuate in value over time depending on factors like interest rates and market conditions. Suppose a bank bought $100 in Treasuries that then traded at $90 as interest rates rose. The bank has a $10 unrealized loss, but if they held the Treasury to maturity the value would recover to $100. These fluctuations affect both loans and securities and can be managed through derivatives. A bank’s unrealized losses are opaque because of accounting choices and valuation difficulties. But usually those losses don’t matter, unless you really need to raise cash.
Banks have a wide range of options to raise cash, but all those options depend on the market value of a bank’s assets. A bank could sell an assets outright at market value, or borrow against it. There is a range of lenders that are willing to lend secured to a bank, ranging from private investors to Federal Home Loan Banks to the Fed’s discount window. But they will only lend against the asset’s market value (plus a hair cut). A significant amount of unrealized losses means that a bank has less potential cash to meet withdraws. An exception is the Fed’s new facility, which is willing to lend against the face value of Treasuries and Agencies. Note that medium and small banks usually hold very few Treasuries and Agencies.
Liability Management
Deposit liabilities are difficult to value because they can in theory be withdrawn at any time, but are in practice “sticky.” The account balance of a typical depositor fluctuates, but they usually keep a certain sum in the bank. Rather than an overnight loan, a deposit is more like a long dated zero coupon bond. The value of deposits are not marked to market, but in theory they become less valuable with higher interest rates. Higher rates can reduce the value of a bank’s assets, but they can also reduce the value of their liabilities. A bank that manages its liabilities well will not need to fire sell its assets to meet withdraws.
Banks have a number of tools to manage their liabilities. A bank can structure its liabilities by issuing CDs or other contractual borrowings that are not due until a set maturity. It can also diversify its depositor base across account types, as some depositors are more volatile than others. For example, retail depositors are protected by $250k in FDIC insurance so they tend to be very sticky. But large depositors are not fully covered and may withdraw their money suddenly in fear of losses. A bank whose deposit base is almost entirely uninsured is very vulnerable to bank runs.
Higher Rates Decrease Liquidity
The level of liquidity in the banking system is exceptionally high, but is set to steadily decline from higher rates and QT. Higher interest rates reduce the market value of bank assets, which reduces the amount of cash that a bank can raise. A secular upward swing in interest rates can potentially impose huge losses on both the securities and loan portfolio of banks. At the same time, QT is set to steadily withdraw a over a trillion in liquidity out of the banking system over the next two years. This is not a problem today, but it may be in the coming years.
SVB was not subject to the Basel III LCR and NSFR ratios as set out in its latest 10-K: “Because we are a Category IV organization with less than $250 billion in average total consolidated assets, less than $50 billion in average weighted short-term wholesale funding and less than $75 billion in cross jurisdictional activity, we currently are not subject to the Federal Reserve’s LCR or NSFR requirements, either on a full or reduced basis.”
I think it’s pretty clear they were deliberately keeping just under the $250bn threshold.
SVB was a VC firm masked as a bank.
Their loan industry clients , deposit client concentrations and their undisciplined interest rate mismatch scream of intentional ignorance. Effectively quantifying VC risk is much more complex than their interest rate mismatch and client concentration data. Imo they VCeed their biggest investor base: the FDIC, FED and Treasury and perhaps the taxpayers, time will tell.
Jaldeko,
Good point, as is Julien’s response that they were probably keeping under this limit. In fact, they actively lobbied to reduce the requirements for smaller banks like themselves. Either they are a systemic risk or they are not. If they are a systemic risk, they need to adhere to the higher requirements of the larger banks. If they are not a systemic risk, then why is Fed/FDIC saying they are and using that exception to cover uninsured depositors?
This is classic “privatize the gains, socialize the losses”. In the future, we have to regulate these smaller banks more (since effectively they are SIFIs based on current Fed/FDIC/Treasury declarations and actions). If, in the end, it will always come down to “we have to save them to prevent contagion”, then we should be charging higher insurance premiums to the banks to cover both insured and uninsured depositors. Said another way, we should raise the insured deposit limit and collect the premiums, because, in practice, we are treating these “uninsured” deposits as insured.
Good point, Jaldeko that SVB is not subject to liquidity constraint.
My question is how come SVB couldn’t borrow reserves from the federal funds market or repo market before fire-selling treasuries and thus making capital loss?
SVB must have suffered from the outflow of reserves as many depositors transfer their deposit from SVB account to other banks. And, the normal procedure is to go to money markets such as the FFM, repo, eurodollar, commercial paper market, or even the discount window to replace the loss of reserves despite higher interest rate in those markets. How come SVB could not find any lender, but instead sell treasuries at loss or raise money by selling shares? It does not make sense…. Please help me understand!
Very fair point. There are some stories flying around that SVB was actually borrowing from the Fed DW. But I haven’t seen anything on repo market borrowings, which would have been the natural thing to do given the huge amount of HQLA they had.
do the zero risk weightings on sovereign debt carry any responsibility in these situations?
Favorable risk based capital treatment does encourage banks to hold HQLA, but I don’t think it plays a big role here. SVB’s peer groups didn’t gorge on HQLA, and I don’t think medium sized banks are very constrained on that metric.
It makes sense for Banks to hedge the interest rate risk on their AFS holdings, but do they typically also hedge their HTM book?
Well they should do if its not really HTM even if its accounted for as HTM! There was a catastrophic lack of risk management at SVB. What’s more they did hedge at least some of their interest rate risk but these hedges were closed sometime in 2022. Given it was clear 2022 was going to be a hiking cycle that is outrageous really.
So if we breaking things, and the fed stops raising rates. And inflation does not immaculately drop, especially with the printer starting to Brrr again. Does that make stagflation for years the default scenario now? Higher rates, lower growth and long grinding reprice.
Yea!
More line they will increase the inflation target level from 2% to something higher (their measures of inflation are BS anyways & do not correspond to reality but yeah more Stagflation seems even more likely than it already was… what a bunch of idiots running the Fiat Banking Cartel these days…
Shouldn’t this be reversed?
“At a high level, a bank has short-dated liabilities and longer dated assets.”
Deposits = ST liability
Loans/Bonds = LT asset
Ah, you are right. My mistake, I read it wrong.
Does this matter?: “Dick Bove did an interview on CNBC Asia today. He pointed out that the “net worth” of the US Federal Reserve is -1.1 Trillion dollars due to essentially the same funding/investing mismatch that took down SVB. So the only way for the Fed to backstop the system is to print money.” ~ from “Bond King” Jeff Gundlach’s (of Doubleline) Twitter…
The Feds ZIRP created their warehouse of another ZIRP..Zombie Interest Rate Portfolio that has hundreds of billions of unrealized losses. Their printing press allows them to utilize magical accounting
Is it relevant that SVB was operating without a risk management officer for a significant period just before @##%##
Is plausible deniability relevant?
Perhaps not. Their bad decisions to load up on long-duration bonds was made 2-3 years ago when they DID have a risk officer. I get the feeling that risk officer was a /wsb trader.
How did you calculate the % of uninsured deposits? It’s not immediately clear based on the source FFIEC 31. Thanks
You could infer it from the 2022 and 2021 10k
2022 shows no risk hedges in place – basically all AFS securities had been sold to Goldman Sachs by year end 2022
Triggering the MTM losses and the need for more capital
In SVB case business model seems to have been always to rely on fresh deposits of cash proceeds from DeSpacs, IPOs and VC rounds
If these dried up then at the core the problem is that SVB’s unrealised MTMs on HTM high quality liquid assets. Having allocated too few HQLA to the liquidity portfolio and too much to the higher yielding HTM banking book, such negative MTM exceeded ALL of SVB’s CET 1 capital . The lesson here is 1) that social media and highly concentrated , highly connected depositor basis , means that deposit outflow assumptions about stickiness are not reliable . And 2) for counter party credit risk analysts that differentiating uninsured deposits is as irrelevant as diffferenriating HTM and AFS assets
Do we know what interest rate the BTFP is offering to the banks?
This seems like an unnecessary moral hazards as it completely removed the incentives for banks to prudently hedge duration risks, I’d argue that a bazooka bailing out all deposits > $250K is probably a better way to go to prevent Regional Bank collapse en-masse.
1yr OIS swap rate plus 10 bps
Joseph: How does the FDIC charge banks for the insurance fund? Is it based on total insured deposits? For instance, if a bank has lots of insured deposits and few uninsured deposits, do they pay more than a bank that’s the opposite?
This was a simple failure of interest rate risk management that led to a liquidity problem. The full suite of International Basel standards would not necessarily have helped here but, of all the risks a bank faces, interest rate risk in the banking book (IRRBB) is the least regulated and is largely only covered through stress testing. This gap exists for the reasons Joseph explains, i.e. if a bank will hold to maturity why should it recognize a loss? However SVB provides a new answer to that question.
When interest rates rise rapidly there are always financial accidents. It is always hard to predict in advance where those accidents will happen, but bank runs taking down uninsured deposits is a well known risk. I always draw down my deposits in my credit union to keep them below 250K. Treasury Direct is good for this. Perhaps, business banking customers should be offered supplemental FDIC insurance through their bank when they open their accounts, and if they decline they will not be bailed out in the event that the bank goes under.
SVB had unique circumstances due to the industry it supported, but the increase in rates and the inverse yield curve, and institutions holding longer-term assets with low yields, is indicative of what happened during the S&L crisis in 1986 when net interest margins got squeezed and those LT mortgage securities and CMO became at risk for default. Appears the Fed is not monitoring these banks’ assets/liability duration mismatch reports (Schedule H) to identify these at-risk banks? Since the govt backs both treasuries and 90% of the MBS issued, a major economic collapse would just be the govt bailing out the financial markets with more debt owed by the taxpayer. Seems the govt is more leveraged than some of these banks?
Holding an asset to maturity that is labeled HTM for accounting purposes does NOT guarantee that the bank will not incur a loss while holding that asset. That asset is financed by the cost of deposits. If the cost of deposits over the lifetime of the asset is greater than the yield of the asset, the bank will incur a loss.
So all mark to market accounting should be doing is trying to estimate the cost of funding an asset to its total return. If the difference is positive the bank is dissipating its capital while it holds that asset. That still represents a risk to depositors unless the bank can keep it’s funding costs below that implied by the yield curve.
“Banks can now bring this impaired collateral to the Fed and get cash to meet deposit outflows, but the Fed charges the short-term market rate, which is closer to 4% or 5%.”
BAGEHOT’S DICTUM: the central banks should lend early and ‘without limits’ to solvent firms at a ‘higher interest rate’ with ‘good collateral’. Discounting was made a penalty rate on January 6, 2003
But Volcker did the opposite.
And: “In 2002, the Federal Reserve began to set the discount rate above the federal funds rate, reversing its previous practice of keeping the discount rate below the funds rate.”
A universal guarantee on all bank deposits, like during the GFC, will reduce the supply of loan funds, will reduce the transaction’s velocity of funds, will reduce the real rate of interest, and thus will lower R-gDp and raise the Federal Deficit.
The FED’s Ph.Ds. don’t know a debit from a credit, a bank from a nonbank.