A Coda on Regional Banks’ ‘Unrealized Losses’

Editors’ Note: This article mentions a few individual stocks in the course of discussing a broader theme we wish to illustrate. Please note that MarketMinder doesn’t make individual security recommendations.

Remember March 2023, when Silicon Valley Bank failed and many feared its problems would spread, wiping out America’s regional banks? The panic faded so quickly—and so much has happened since then—that it feels far longer than 22 months ago. But it got an interesting capstone last week, courtesy of a Wall Street Journal writeup of some banks’ earnings results and forecasts. We think it highlights the benefits of the accounting rule tweak that helped end 2007 – 2009’s global financial crisis.

Back in early 2023, Silicon Valley Bank was in deep trouble. Not because its loan portfolio was suddenly imploding, but because it had a lot of US Treasury bonds on its books and the vast majority of its deposits were industry-concentrated and uninsured. So as long-term rates rose and bond prices fell, these big depositors got antsy about the bank’s health and liquidity. In need of capital, the bank released plans to sell Treasurys at a loss, which ratcheted up concerns about liquidity, leading to more deposit flight and, ultimately, driving the bank into receivership. Eventually, First Citizens bought its ashes and continued operating the business, but for a time it was unclear whether or when large depositors would get their money back.

As another regional bank—New York’s Signature Bank—failed the same weekend, investors and depositors alike grew concerned about a run on regional banks (or, in industry terms, contagion). Analysts worked overtime trying to suss out which banks were at risk of failure, with many pointing to unrealized losses on US Treasury bond portfolios as a metric. But broad meltdown didn’t happen. Giant Swiss bank Credit Suisse failed—a victim of its own, unrelated and very long-running problems—and West Coast lender First Republic collapsed, echoing Silicon Valley Bank’s demise. But these institutions were the exception, not the rule, and life went on for the vast majority of regional banks.

Interestingly, US Treasury bonds haven’t exactly had a stonking huge rally since then. This isn’t a case of a temporarily distressed portfolio holding rebounding. Rather, banks were sufficiently diversified and liquid to ride out the storm. Which brings us to the Journal’s reporting Thursday: “Banks reporting fourth-quarter 2024 earnings are expecting a bump to their core banking income from those [US Treasury bond] portfolios. As older, lower-yielding bonds mature, they can be reinvested at today’s far higher yields on both bonds and cash. That can be a boost to net interest income, which is a measure of how much banks earn in yield on cash, securities and loans versus what they pay out on deposits and other funding.”^[i]

Yes, banks just let their Treasury holdings do what they are designed to do: pay interest until maturity, at which point Uncle Sam repays them and the bank buys new bonds at the going rate. The temporary market-value declines in 2023? Paper “losses.” But because the banks didn’t sell, those losses were just theoretical.

Once upon a time, those theoretical losses would have had a nasty effect on banks’ earnings. The 2007 – 2009 global financial crisis was largely a byproduct of an accounting rule, FAS 157, that required banks to mark all assets on their balance sheet at “fair” or market value. This is easy for a liquid security with easily determined pricing, like a stock or a Treasury bond. It is harder for illiquid, complex mortgage-backed securities that rarely trade. So when some financial institutions got in trouble and had to dump these complex securities at fire-sale prices to raise cash in a hurry, every bank that owned something similar had to write that asset’s value down according to that sale price. These mark-to-market “losses” rippled through the Financials sector, creating a vicious cycle of writedowns, fire-sales and failures.

The frustrating thing for all involved—and keen observers—was that in the vast majority of cases, banks never intended to sell these assets. The intermittent price movement was meaningless to their long-term plans, which entailed holding to maturity and collecting interest along the way. We have seen, over the years, that this strategy would have worked fine—the Fed, which ended up with Bear Stearns’ supposedly toxic assets, made a lot of money on them. But accounting rules at the time distorted the picture, translating about $200 billion worth of loan losses into nearly $2 trillion of exaggerated, unnecessary writedowns.^[ii]

A big reason Silicon Valley Bank’s troubles didn’t have the same effect is that, following the crisis and the obligatory enquiries, regulators clarified the rule. Instead of forcing banks to mark all assets to market, it let banks designate assets they don’t intend to sell as “hold to maturity” assets; those they intend to trade, “available for sale.” The latter are, sensibly, marked to market. The former, sensibly, aren’t. Unrealized losses are still disclosed on them—in footnotes. But they are valued at cost on balance sheets. Rate swings don’t affect them. The upshot: Silicon Valley Bank’s labeling change was what sealed its fate—it reclassified Treasurys as available for sale, crystalizing losses and sparking the bank run.

But other banks didn’t have to, and the amended rules enabled them to tough it out. At the time, we saw some grousing, just as we did during and after 2007 – 2009.^[iii] There has always been a school of thought arguing the mark-to-market value is more accurate than cost, so it is the correct price. But markets are volatile, and we think it is telling that no one really makes this argument when asset prices are above face value. No one thinks mass write-ups would be a great idea! They would see this as artificially inflating capital.

And we think they would be right. If a bank doesn’t plan to sell a bond at its higher trading value, but rather just hold on and collect interest until it is repaid in full, then the bank isn’t reaping a gain and shouldn’t report one. Similarly, if the bank doesn’t plan to sell a bond when its trading value is lower and can ride out the temporary declines, then the bank isn’t really taking a loss and therefore needn’t take an exaggerated, pointless paper loss. Mark-to-market is an accurate valuation only if the bank plans to sell. If not, then market value is likely to be meaningless, as this news demonstrates.