Bank Runs and Systemic Risk
As the Federal Reserve engineered an increase in interest rates, financial institutions saw losses emerge on their fixed income portfolios. Similarly, financial institutions saw their funding costs increase dramatically. After thirteen years of near zero short-term interest rates, some financial institutions failed to adjust to the new environment. Specifically, several medium-sized banks were grossly negligent in matching their asset and liability durations in this new environment. The business model for any bank is to lend long and borrow short: that is, to invest in assets that are of longer duration than your interest-paying liabilities, primarily bank deposits. If the duration risk of your assets spooks the holders of your liabilities (the depositors), you face a run on the bank. That is what is happened in mid-March.
If an individual’s total deposit exposure to a bank is below $250,000, then the Federal Deposit Insurance Corporation insures that individual’s bank account(s). Balances above $250,000 per individual, technically are uninsured by the FDIC and face the risk of a haircut and/or a delay in repayment. However, the Boards of the FDIC and the Federal Reserve can agree to insure depositors with outstanding balances above $250,000 if they determine that not doing so could result in a systemic risk to the banking system. The Boards made this determination on Silicon Valley Bank and Signature Bank. All depositors of these two seized banks, whether insured by the FDIC or uninsured, will be made whole. On the surface, the determination to insure the excess deposits seems odd: neither bank was defined as a “systemically important financial institution”, and, in fact both banks lobbied to avoid that designation; and both banks carried significant deposit balances that were uninsured by the FDIC (80% of all deposits at Signature and over 90% at SVB). The “systemic exception” was made to contain contagion among depositors at other banks. It will be politically difficult to not invoke this “exception” in cases of future bank seizures of similar size. Please see the FDIC website for questions of account insurance eligibility: https://edie.fdic.gov/fdic_info.html#06
On March 12th, 1933 Franklin D. Roosevelt gave a 13-minute radio address explaining to the American people the government’s imposition of a “bank holiday” to stem a run on banking institutions. The fireside chat said in part:
First of all, let me state the simple fact that when you deposit money in a bank the bank does not put the money into a safe deposit vault. It invests your money in many different forms of credit: bonds, commercial paper, mortgages and many other kinds of loans. In other words, the bank puts your money to work to keep the wheels of industry and of agriculture turning around. A comparatively small part of the money you put into the bank is kept in currency—an amount which in normal times is wholly sufficient to cover the cash needs of the average citizen. In other words, the total amount of all the currency in the country is only a small fraction of the total deposits in all of the banks. What, then, happened during the last few days of February and the first few days of March? Because of undermined confidence on the part of the public, there was a general rush by a large portion of our population to turn bank deposits into currency or gold. A rush so great that the soundest banks could not get enough currency to meet the demand. The reason for this was that on the spur of the moment it was, of course, impossible to sell perfectly sound assets of a bank and convert them into cash except at panic prices far below their real value.
President Roosevelt, of course, was explaining the dynamics of a fractional reserve banking system under the pressure of a systemic liquidity squeeze. Today, the world’s financial system is still based upon a fractional reserve banking model, and after 90-years of regulation and re-regulation, the system is still subject to liquidity pressure during crises of confidence among depositors.
What follows is an overview of the domestic banking system that is insured by the FDIC, as well as some of the idiosyncratic risks faced (and embraced) by several of the recently seized or at-risk banks. It is not required reading, unless it interests you. The U.S. banking system, at year-end 2022, held $22.9 trillion of assets. Balancing these assets are $20.7 trillion of liabilities and $2.2 trillion of bank capital. The majority of the banking liabilities consist of bank deposits. Of the $17.6 trillion of deposits in the U.S. banking system, $10.1 trillion (or 57%) are insured by the FDIC. About 10% of bank deposits are time deposits, which are relatively easy to match with assets of similar duration, while most of the rest are demand deposits, which are more difficult to match. Banks hold about $3.0 trillion of cash assets to meet deposit withdrawals, which is 17% of total deposits or 40% of non-insured assets (which are obviously more prone to abrupt withdrawals). Bank deposits, aided by government payments, grew 20.9% in 2020 and 11.8% in 2021. In 2022, bank deposits declined 0.7%. Through February 2023, bank deposits are shrinking at an annualized rate of 6.4%. We expect that rate of decline to increase in March, given the current sentiment.
To fully understand the issues spooking depositors currently, one needs to understand accounting and regulatory rules that are imposed on banks. Bank loans and leases are carried on the balance sheet net of an allowance for losses. They are not marked-to-market unless they are delinquent or in default. Securities purchased by banks are valued at market prices…sometimes. When a bank purchases a Treasury note, Agency debt, a corporate bond or a mortgage-backed security (MBS), it must make one of three defining classifications before it is added to the balance sheet as an asset. If it is designated as a trading security, it is marked to market on the balance sheet and any decline in value is shown in the income statement as a loss. If the security is designated as available-for-sale (AFS), it is marked to market on the balance sheet, but any loss or gain is not recorded on the income statement, but rather noted on the statement of Accumulated Other Comprehensive Income (AOCI). Finally, if the asset is designated as held-to-maturity (HTM), it is not marked to market on the balance sheet, nor is a loss noted on the AOCI statement. However, HTM securities are usually footnoted with marked-to-market prices in the company’s quarterly audited financial statements. A bank can only designate a security as HTM if it has both the intent and ability to hold that asset to maturity.
The reason for the myriad of designations is because bank managers and bank regulators agreed that recognizing quarterly fluctuations in market values of operating assets could misrepresent the volatility of a bank’s operations. There are strict rules regarding AFS and HTM securities. If a bank moves a security from AFS to HTM, it must be actually held to maturity. If a security is designated HTM, a bank cannot hedge its interest rate risk. If a HTM security is sold before maturity, not only must any loss be recognized upon that sale, but the entire HTM portfolio must be re-designated as available-for-sale, and all mark-to-market losses recognized. For purposes of measuring capital adequacy, regulators do not mark unrealized losses on HTM assets to market. So, in an environment of rising interest rates, a bank could show capital ratios that easily exceed regulatory minimums, but if forced to sell held-to-maturity assets at market rates, be determinedly insolvent. That is what happened with the first two banks seized by the FDIC, Silicon Valley Bank and Signature. The following chart shows the banking industry’s $620 billion of unrealized losses at year-end 2022. The unrealized loss represents about 28% of total bank capital.
Following the unraveling of Sam Bankman-Fried’s cryptocurrency FTX platform, a former S&L from San Diego came under pressure. Silvergate Capital had repositioned itself to service clients in the crypto- and digital services industry. It suffered an $8.1 billion run by its depositors in December of 2022, more than 60% of its deposits. The firm entered into a voluntary liquidation process to pay off its remaining depositors and close the bank. Surprisingly, it seems to have been able to liquidate its deposits at par, and the FDIC never seized the bank. Although Silvergate avoided receivership, the velocity of its demise looks to have shaken depositors at two other banks.
Signature Bank was the only large on-shore bank other than Silvergate Capital that pursued cryptocurrency and digital asset accounts. While the bank did not deal in cryptocurrencies directly, it did process dollar transactions for crypto clients. Its deposit base grew quickly from $40 billion in 2019 to $106 billion at year-end 2021. Total assets were $110 billion in December, with $88 billion of deposits and $8 billion of equity. However, over $80 billion of its deposits carried no FDIC insurance, making them prone to liquidation. The company lost $17.5 billion of deposits in 2022. On March 10th, Signature borrowed $4.5 billion from the Federal Home Loan Board of New York, but that amount was swamped by $18.0 billion of deposit withdrawals. As more withdrawal requests piled up over the weekend, the Fed denied requests to advance more funds and regulators seized the bank. Following the demise of Silvergate and Signature, there is no large or medium-sized on-shore bank that facilitates cryptocurrency companies. The FDIC has estimated that the seizure of Signature Bank will cost $2.5 billion. We doubt that the Federal Reserve likes regulated banking mixed with cryptocurrencies.
Silicon Valley Bank had a great reputation that it did not deserve, judging by its recent managerial decisions. SVB created a niche by serving venture capital firms, venture capital-backed companies, technology and life science start-ups and young lifecycle companies. The bank also catered to founders, executives, investors and employees of technology companies. The company understood the V.C. world and excelled in attracting assets. On the other hand, it did not seem to understand how to effectively invest the deposits it collected. Silicon Valley Bank flashed two of the primary warning flags that bank regulators look for: 1) high deposit growth and 2) undiversified funding sources. In 2021, SVB’s deposit base grew to $189.2 billion from $102.0 billion the year before (up 85.5%). In the first quarter of 2022, its deposit base peaked at $198.0 billion, and began to decline. Almost all of the growth in deposits was invested in held-to-maturity long-term MBS that paid very low rates. Silicon Valley Bank was supposed to be an expert in the venture capital funding and investment cycle. Like most private equity funds, the money that flows into venture capital funds is cyclical, experiencing booms and busts. For V.C. funds, 2021 was a boom year. Even though the bank acknowledged that bank balances for start-up companies were three-times its normal level (three years’ worth of spending versus a more normal funded burn-rate of one year), it failed to match its investments with its liability profile. By the fourth quarter of 2022, venture fundraising hit a nine-year low, down 65% from the fourth quarter of 2021. Rather than increasing its cash and AFS securities, SVB decreased these balances by $6.8 billion. As venture-backed companies continued spending, they withdrew funds from their deposits. New deposits from venture-backed companies dried up.
Overall, deposits fell more than SVB had cash on hand, and so it was forced to start selling its AFS portfolio. Silicon Valley Bank sold its AFS securities at a loss and let its interest rate hedges roll-off, probably to save money. As interest continued to rise, the AFS portfolio continued to decline, both because of sales and unrealized losses. But the really negative action was happening in the much larger held-to-maturity portfolio of mortgage-backed securities. The year-end 2022 HTM MBS portfolio had a carrying value of $91.3 billion, of which $86.0 billion was due after ten years. Remember, the carrying value of this portfolio is the number used to determine regulatory capital. The actual fair value (marked-to-market) of the HTM portfolio was $76.1 billion. If losses on both the AFS and HTM portfolios were recorded at fair value, Silicon Valley Bank would have burned through virtually all of its actual capital (SVB had a $15.1 billion unrealized loss on its HTM portfolio at year-end versus $16.3 billion of equity). When withdrawals began in earnest, management’s answer was to sell the available-for-sale portfolio for a loss, and raise $1.75 billion of new equity to plug the hole. Goldman Sachs bought the AFS portfolio, but failed to complete the equity sale. Once rumors leaked about what SVB was trying to do, venture capital funds instructed their investees to withdraw their deposits. With an unbelievable 93% of all its deposits exceeding the FDIC insurance limit, there was no reason for depositors to stay with SWB. On Thursday March 9th, depositors requested $42.1 billion of withdrawals, more than 24% of year-end deposits. This was more than SVB could pay out, so California regulators seized the bank. Silicon Valley Bank spent more than 39 years building its franchise, and lost it in less than 36 hours.
Mark Twain said that a rumor can get half-way around the world before the truth can put its shoes on. After the initial run on Silicon Valley Bank started, many regional banks saw deposit withdrawal requests, especially if they had a high level of uninsured deposits. San Francisco’s First Republic Bank was one of the hardest hit. The run on First Republic began before the FDIC had declared a “systemic exception” and backstopped the uninsured deposits of Signature Bank and Silicon Valley Bank. Monday March 13th came and the run on First Republic’s deposits continued. The bank had a more traditional business model than Silicon Valley Bank or Silvergate, holding a much higher percentage of its assets in loans and mortgages than securities, and as such, did not have a similar exposure to unrealized security losses. However, the chink in FRC’s armor was a close to 70% exposure to uninsured deposits, and a depositor base that was both wealthy and willing to move funds quickly. The fair value of First Republic’s loan book was below its carrying value, and this was enough to encourage withdrawals. Even after J.P. Morgan assembled a consortium of band to provide an uninsured $30.0 billion deposit to the bank, withdrawals continued. As we write this, the bank is in limbo, not yet seized but not rescued either. Either way, it is a shadow of its former self.
Deposits are fleeing regional banks and are heading toward the “too big to fail” money center banks. Although these banks also have HTM losses in their security portfolios, they have more stable deposit bases and funding alternatives that make it likely that their HTM portfolios will actually be ‘held to maturity’. The best of these look to be J.P. Morgan and Wells Fargo. Finally, Charles Schwab operates a bank essentially for the convenience of its investment and brokerage clients. Although the subsidiary bank holds an ugly portfolio of AFS and HTM securities, the bank has around 80% of its deposit assets eligible for FDIC insurance. Additionally, most of the accounts are sweep accounts that are tied to a Schwab brokerage, investment or custodial account, which should make them sticky. The parent company has very good access to many forms of liquidity, and if Schwab needs to add capital to the bank, we are confident that it would, rather than risking its brokerage franchise.
The surge in interest rates has almost every bank regretting some of the business they put on their books in the last two years. The banks that grew their deposits the fastest over that period are probably most at risk, as are the banks that had an idiosyncratic depositor base that turned out to be pro-cyclical when management thought it was counter-cyclical. Money moves faster these days, making a depositor-funded banking model less attractive than in previous decades. Some banks will fail, but most will earn their way out of problematic assets, as happens during every recession. Depositors will likely be bailed out, in almost all circumstances. The system will continue.
Your Team at Baxter Investment Management