Were Silicon Valley Bank's risk managers asleep at the wheel? Or did they do the best they could with the market conditions they were given? I explain what led to Silicon Valley Bank's collapse and let you decide.
After markets closed Friday, I was greeted by a Financial Times Headline: "Why did Silicon Valley Bank Fail?" with the subtitle, "SVB had grown to about $209bn in assets with a client base concentrated among tech and healthcare start-ups".
Economist Frances Coppola told Al Jazeera on March 11 that fallout from this failure would be, "concentrated in the tech sector," and therefore, "would affect that industry in Silicon Valley, London, and potentially Doha… but will be contained". Contained is a word you never want to hear from someone with a PhD in Economics during bank failures. Fed Chair Ben Bernanke infamously called the subprime mortgage problem "contained" in March and May 2007.
We need not understand SVB's particular client base to understand why the bank failed. The bank was sitting on a powder keg of interest rate risk (explained below), and once you understand that, its failure is cut and dry. By now, you've probably seen several memes or news articles comparing this failure to the 2008 contagion. While I too made a similar comparison above, this bank failure more closely resembles the Savings & Loan crisis of the late 1980s.
First, let's understand the business of banking. Banks and Non-Bank Financial Institutions (NBFI's) make money by borrowing short and lending long. They borrow short by taking deposits, which are effectively loans made by savers to the bank. Deposits have an almost infinitely short time to maturity since they can be pulled by the saver at any time. Then, they lend, usually for terms of one to 30 years. The source of their profit is the typical difference in short rates and long rates, known as the upward sloping "Yield Curve". This business model leaves all banks susceptible to failure at any time if enough of the bank's depositors demand cash for their accounts or transfer their balances to other banks.
Banks can "lend long" by making direct loans to households and businesses, by buying bonds, or both. Suppose a bank buys a billion dollars of newly issued US Agency Mortgage Backed Securities ("MBS"). These are bonds backed by bundles of mortgage loans, insured by the Government Sponsored Enterprises Fannie Mae, Freddie Mac, and Ginnie Mae. Even though buying a bond looks different from making a loan, the bank has effectively "lent long" to US homebuyers.
About 55 percent of Silicon Valley Bank's Assets were in bonds, specifically, the safest and most liquid bonds in the world. SVB invested in $92 billion of US Agency Mortgage Backed Securities ("MBS") and $16 billion of US Treasury Bonds ("Treasuries"). Their Loans and Leases, in which we would find any potentially bad loans to venture capital firms in the tech space, amounted to 35 percent of total assets.
Here is the asset side of their balance sheet reproduced from FDIC call reports. The amounts I reference are highlighted.
Source: Silicon Valley Bank FDIC Call Report Q4 2022
Loss allowances for the entire book of loans and leases, as of 4th quarter 2022, were $636 million. That's about 0.3 percent of total assets and 0.9 percent of the loan book. SVB's stock tanked 63 percent in a week, not because of its loans to Venture Capitalists, but because of the unrealized losses on their ultra safe bond investments. The bank was sitting on $17.7 billion of unrealized losses on Treasuries and MBS. How did this happen? It wasn't a 2008 situation where a bunch of the underlying mortgages behind the MBS went bad. This was Interest rate risk.
The biggest risk to the market value of a bond is what interest rates look like right now. Here's an example to understand why:
Suppose the current short term interest rate you would get in a money market fund is 0.05 percent, and I offer you an MBS issued by government-backed Ginnie Mae that will pay 1.5 percent annual interest for 30 years. How much would you pay for that 1.5 percent bond under these market conditions?
This is what the market looked like in July 2021. Ginnie Mae mortgage bonds paying 1.5 percent were selling at a premium price of about 101 cents on the dollar. Fast forward 21 months, and the bonds are fairing much worse. The "short rate" investors can get is now closer to 4.5 percent, and so these 1.5 percent bonds are not going to sell without a steep discount. The Market Value of a Ginnie Mae 1.5 is now about 77 cents on the dollar, a 24 percent drop.
All bonds are susceptible to this dynamic, but the lower the interest rate on the bond is, the more value it will lose when interest rates rise. For instance, Ginnie Mae 3.5's have also fallen in value, but with a price of 92 cents on the dollar, the discount is much smaller.
As the bank's Deposit base rose 82 percent from 2020 to 2021, from $107 to $195 billion, SVB loaded up on Treasuries and MBS just as the rates on such bonds dropped to all time low levels. As long as rates stayed low, this strategy looked good. But a huge, unanticipated move in rates, perhaps caused by the Fed jacking up the short rate by 4.5 percent in 12 months, would tank the value of their assets.
SVB isn't the only bank that made moves like this. US commercial banks bought $1.8 trillion of bonds in 2020 and 2021, bringing the share of bank credit going to bonds to an all time high of 35 percent.
Source: Federal Reserve Board of Governors.
When short sellers saw the shaky ground under SVB's balance sheet, and larger depositors began to understand that this would be a problem, the bank's stock price fell 63 percent and the bank lost $42 billion in deposits.
So, who's to blame - SVB, the Fed, or Both?
These all time low bond rates didn't just fall out of the sky. Let's go back to March 2020 for a moment.
During the COVID financial panic, the global bond market seized up. The stability of the two biggest bond markets in the world - US Treasuries and US Agency MBS - was threatened. In response, the Federal Reserve did what central banks always do when faced with this situation: it bought trillions of dollars of these bonds, and cut the short interest rate to zero. The Fed bought $3.2 trillion of Treasury Bonds and $1.4 trillion of Agency MBS.
When the Fed places an infinite bid for Treasuries and MBS, and drops its policy rate to zero, it pulls the interest rates on bonds down. These interventions had two major ramifications in the MBS market specifically.
First, it drove the interest rates on MBS and the underlying mortgages to all time lows: between 1.5 and 3 percent. Second, those all time low rates drove a record breaking surge in refinances, about $5.5 trillion worth. When a trillion dollars of mortgages are refinanced at an interest rate of 2 percent, a trillion dollars of 2 percent MBS will be issued, but that's not all that happens.
The old mortgages, paying higher interest rates, are part of existing mortgage backed securities issued some number of years prior. Those loans are prepaid, and those higher interest rate MBS cease to exist. Prior to the pandemic, there were about $11 trillion of Agency MBS outstanding. The Fed induced a refinance boom that replaced half of those bonds, which were paying pre-pandemic interest rates anywhere from 3.5 to 6 percent, with new MBS paying 1.5 to 3 percent.
The Fed also did its damnedest to convince market participants that rates were going to be low for the foreseeable future. In June of 2020, Fed Chair Jerome Powell told the world that the board was, "not even thinking about thinking about raising rates". Every communication the Fed gave to the public sent the same message. For instance, here's the Fed's March 2022 projection for the future level of the short rate.
Source: Federal Reserve Board of Governors, you may want to zoom in.
Just looks like a bunch of dots, I know. What the Fed is telling the public with this chart is that less than half of board members saw the short interest rate breaching 3 percent in the long run.
But SVB made huge mistakes as well. Financial Institutions use derivative contracts called Interest Rate Swaps to protect themselves from scenarios exactly like this. It's called "hedging". If a bank holds $100 billion of fixed rate bonds, and "pays fixed" on $100 billion of interest rate swaps, it is fully protected from the effect of rising rates on its bond holdings. In other words, the bank is "hedged". The downside is that all of the interest income the bank receives from the bonds will be paid out on the swaps, instead of contributing to profits.
About $28 billion of SVB's bond investments were held with the intention to sell (as opposed to holding until the bonds paid off), and it would've been prudent to hedge the rate risk on these holdings at least. Instead, the bank only offset $550 million of its interest rate risk with swaps.
Several banks will face failure, or come close to it, for having the same investment strategy as Silicon Valley Bank. What's the source of their failure? Are these recklessly managed, bad banks? Were they just doing what they could within the market conditions the Federal Reserve gave them? Or was it a bit of both? I leave that to you to decide.