How did Silicon Valley bank collapse? Economist Douglas Diamond explains

Small wonder that most Americans are stunned and confused by the sudden fall of Silicon Valley Bank. How did a cornerstone of the dynamic venture capital community, the nation’s 16th-largest lending institution that until recently enjoyed the growth worthy of the tech startups it served, fall so hard, so fast? Is its failure the legacy of poor practices specific to SVB, or is the Fed’s policy of drastically hiking rates that hammers the value of banks’ investments endangering fellow midsize lenders? Even if bad management destroyed SVB, why didn’t its top regulators, the Fed and the California banking authorities, see this runaway train wobbling on the tracks, and force the drivers to throttle back before it derailed?

I thought of just the expert to skirt the usual dense jargon and provide easy-to-grasp answers. He’s Douglas Diamond, professor at the University of Chicago’s Booth School of Business who shared the 2022 Nobel Prize for Economics with his research partner, Philip Dybig of Washington University in St. Louis, and former Fed chairman Ben Bernanke. The Diamond-Dybig research that captured the Nobel stressed that banks are inherently fragile and vulnerable to “runs,” because if customers exit en masse, the lenders may need to sell their bonds or loans, which would have fully paid off on maturity, at fire-sale prices. Hence, a panic can unnecessarily ruin an otherwise healthy bank. Diamond and Dybig emphasize that both sound regulation and prudent management that broadly diversifies the risk in both the loan and investment portfolios, and makeup of customers, are essential to instilling client confidence required to keep America’s banks out of harm’s way.

In October, just after receiving the prize, Diamond warned in a Fortune interview that the Fed’s policies of raising rates at a brutal, virtually unprecedented pace would trigger dangerously big losses in the bond portfolios of companies and banks that believed inflation-adjusted yields sitting at near-zero for years would stay there for years to come.

But in our hour-long interview on the SVB debacle, Diamond stated that though Fed policy hurt, it wasn’t the main reason for the implosion. Nor did SVB suffer the classic “sound bank wrecked by a stampede” scenario. Instead, SVB deployed just about every bad policy on both the assets and liabilities sides of its balance sheet. For Diamond, SVB is a case study in how setting a rickety structure to enable breakneck expansion created daunting risks that prudently run banks, despite the Fed’s huge run-up in rates, have avoided.

What make banks work, and how SVB broke the mold

Diamond described the template for how banks secure their customers’ trust, and protect themselves from a wave of withdrawals. “The papers that the Nobel Committee recognized explained how banks should be structured,” he explains. “On the asset side, banks make loans to lots of different types of people and businesses. Ideally, banks create safe assets out of risky ones by diversifying. They have diversified funding sources so that since depositors all don’t need their money on the same date, that diversification allows the bank to economize on what they hold in cash and liquid assets.” As for liabilities, he adds, it’s key that banks serve a wide, varied range of depositors. Having loads of retail customers is a boon. When rates on Treasuries jump, they’re less likely to empty their savings or money market accounts to get some extra yield than are corporate clients.

As Diamond notes, it’s crucial to understand the role of the two classes of investments on SVB’s balance sheet. The first grouping is called Available for Sale, or AFS. It consists of securities in the trading account banks are free to sell at any time. All bonds in the AFS designation must be “marked to market” at the end of each quarter. If a bank is holding Treasuries it bought early last year at extremely low yields, and rates jump, the prices of those bonds fall sharply, hitting the bank’s capital. The second investment category is Held to Maturity, or HTM. It comprises the fixed income securities that the bank intends to keep on its balance sheet until they’re redeemed at their full par value. Once each quarter, banks can shift securities between AFS and HTM—if they need to replenish their equity, they’ll transfer bonds from the long-term hold to the trading account. But if a bank transfers HTM securities that have an unrealized loss, that would raise liquidity but hit their book equity even harder. This is a quandary SVB faced before the deluge.

At the close of 2022, SVB counted $26 billion in AFS, virtually all in Treasuries and “agency” mortgaged backed securities issued mainly by GSE’s Fannie Mae and Freddie Mac. As Diamond points out, those AFS bonds were all highly liquid; they’d easily sell at full market price, and stood no danger of suffering a haircut if dumped fast. SVB’s balance sheet also contained $91 billion in HTM bonds, of which over 90% sat in agency-issued mortgage securities that also benefit from a deep, active market. Its $74 billion credit portfolio was highly concentrated, consisting primarily of loans to tech startups, as well as their founders and managers. Those companies and Silicon Valley bigwigs also were also their main depositors. It’s been reported, in fact, that SVB often placed covenants in its loan agreements requiring that a borrower keep its deposits at the bank.

SVB mismatched its investments to the deposits funding them

The bonds in AFS, the ones SVB would need to sell in an emergency, were generating a puny yield of just 1.79% as of mid-March. Clearly, it had purchased most of those securities well before rates started spiking big-time in the spring of 2022. The average maturity on the AFS portfolio was a substantial 3.6 years. At the end of 2022, nearly 90% of the HTM loans carried maturities of over 10 years, and the return on that bedrock portfolio was just 1.63%—once again, SVB had bought almost all those bonds way before rates exploded. Its loan portfolio was also garnering low returns of well under 4% after provisions for credit losses.

“Their investments were pretty long-term, and they were generating very low yields,” says Diamond. “They must have figured that scenario would work fine if every depositor stayed forever, and they kept accepting zero rates on checking accounts and sub-1% rates on money market funds. In that case, they could hold their bonds to maturity and get full value.” It’s clear that SVB’s strategy to “go out on the yield curve” to garner an extra 0.5%, say, on a five-year versus a one-year Treasury, was a mistake. The crunch came in 2022, when yields on five-year Treasuries competing with the ones they bought a just a year before jumped from under 1% at the start of the year to the mid-4% range by fall. Suddenly, SVB was forced to pay 4.5% on savings accounts, a multiple of what it offered a year before.

It appears that before mid-2022, SVB’s deposit base was extremely stable. Not only did it keep adding new customers at a rapid clip, but its existing clients kept their deposits in place. But when rates surged, depositors who’d parked billions in SVB’s checking and money markets pulled their cash in pursuit of the sumptuous yields on Treasuries. In 2022, SVB lost 8% of its deposits, and the exodus accelerated in January and February. On March 8, it issued an 8K stating that it had sold all of its AFS bonds to raise money and pay fleeing customers, and sought to refill its coffers via a $1.25 billion stock offering. The AFS sale raised $21 billion, causing a pre-tax loss of around $2.4 billion, or 11%.

For Diamond, SVB faced two fundamental problems of its own making. The first was the fall in the value of its bonds, which had long maturities compared with a deposit base potentially far less stable than those at a JPMorgan Chase or Bank of America. “The rise in rates hit their bonds and cut their capital down,” says Diamond. “They had to write down the AFS bonds whether they sold them or not. The management claimed it was a fire sale, but it wasn’t a fire sale. Those bonds were highly liquid. SVB didn’t take any discount for selling in a hurry.” Hence, he says, SVB was far from the traditional disaster case where a flight of deposits forces a bank to jettison hard-to-sell assets at distress prices.

Diamond posits that even before the 8K announcement ignited the run, SVB was close to insolvent, and rapidly heading for failure. “As their cost of ‘funding,’ meaning the interest they had to pay on deposits, kept rising to 4% and higher, they’re forced to pay that higher interest needed to keep their customers. But their bond portfolio, where the money comes from, is paying them less than 2%,” he says. “Put simply, you’re getting less than 2% on your assets and paying out, say, 5% on your liabilities.” That deficit of interest coming in and out meant that SVB was destined for big operating losses.

To cover those losses, SVB would need to raise cash by shifting its longer-maturity HTM securities to the trading account. But doing so would have pounded its capital even harder. And in a footnote to the 8K, SVB noted that if it marked its HTM securities to market, the adjustment would wipe out all of its book capital. “The downward adjustment on long-term bonds carrying low rates would be even steeper than on the three-to-five-year bonds in AFS,” says Diamond. Even if they were able to hold the HTM portfolio to maturity, the operating losses would eventually render the bank insolvent. “It appears that they were already dead or dying before the meltdown,” says Diamond.

It also astounds Diamond that although SVB was highly vulnerable to a rise in rates, it did little hedging to offset the obvious risk that yields would eventually jump from their historic lows. Rate hedging is practiced extensively by well-run institutions such as JPMorgan.

SVB’s failure to diversify the deposit base increased risks of a run

For Diamond, besides failing to match its investments’ maturities to its depositors’ quicksilver demands, SVB also violated the second tenet of sound banking: attracting a broad mix of customers. He points out that SVB had an extremely small proportion of retail clients to balance all the Silicon Valley startups and their wealthy founders. “SVB reportedly looked for companies that were getting new VC funding, and offered them large loans,” says Vivian Fang, an accounting professor at the University of Minnesota. “That’s how they grew their business so rapidly.” Adds Diamond, “Keep in mind that money had been flowing out for six months. These weren’t programmers or teachers leaving. They were CFOs. Almost all of their deposits were wholesale.” Once again, the CFOs were much quicker to pull cash from checking accounts and grab those big Treasury yields than regular folks would have been.

An incredible $48 billion in deposits departed in a single day. “It was the fastest bank run in history,” marvels Diamond. “The customers in the Silicon Valley community all talk to one another. When Peter Thiel and Y-Combinator, the startup hub, say to get your money out, when that happens, the run will be fast and complete.” 

On contagion, Diamond is concerned about lax regulation and the Fed’s super-tough policies

Of course, most midsize banks aren’t risking the funding mismatch and all-in-on-a-single-client approach that sank SVB. Still, Diamond worries that the Fed’s oversight of regional banks, in itself, is far too light to prevent further blowups. In the early years following the passage of the Dodd-Frank legislation in 2010, the central bank imposed the same tough annual tests on midsize lenders as the likes of Wells Fargo or Citigroup. But in 2018, the Trump administration successfully championed a regulatory relief bill that greatly reduced the frequency and severity of the stress exams for regionals. “I looked at the latest stress test, and the Fed was assessing how the banks would perform at rates from 0% to 2%, as if 2% was as high as they’d ever go. So almost any bank would pass. The standard should have been 0% to 7%.” (SVB was exempt from what would have been its last stress test in 2021 because its assets were still below the required level. It was not scheduled for testing in 2022 when its assets passed the threshold.) In SVB’s case, Diamond is surprised that the Fed and California Department of Financial Protection and Innovation didn’t see the red flags raised by SVB’s slender, restless clientele and holdings of low-yielding, long-duration bonds.

Obviously, the Fed will need to predict the midsize banks’ outlook using much higher rate assumptions in the future, a shift that could require lenders to hold far more capital. Since marking their AFS securities to market is already denting their equity, it’s conceivable that regionals will need to float equity to restore their capital. In turn, announcing you need to sell stock could send depositors for the exits. Of course, it’s the super-tough Fed policy that’s put banks in this difficult position. “When the Fed takes rates from 1% to 5% in a year, it shouldn’t be surprising if that causes trouble in the system,” says Diamond. “When we spoke after I received the Nobel, I talked mainly about how fast-rising rates would hurt companies. But the incredible speed of the hikes hurts banks a lot too.” Diamond believes that the Fed should be “much more slow and deliberate” in raising rates, in part to forestall more SVB-like shocks to the system.

Doug Diamond won his Nobel for, in the words of the National Bureau of Economic Research, providing “insights [that] form the basis of modern bank regulation.” For Diamond, the sound management practices and regulation that he extolled in his research, that make banks safe, was sorely lacking in the SVB catastrophe. Banks get in trouble when they veer from the Diamond model. We can only hope that SVB was a lone case, and that the Fed’s relentless march and weak regulation won’t produce a flurry of renegades that roil America’s credit markets just as our economy teeters on the brink of recession.  

SOURCE

No votes yet.
Please wait...

Leave a Reply

Your email address will not be published. Required fields are marked *