Silicon Valley Bank
The systemic view
Welp, that’s what I get for trying to take a vacation: a blizzard and a proper bank run. I’ll have to deal with the former this weekend; in today’s post, Soon Parted catches up on the latter.
Start with the balance sheet
The proximate issue, by now widely reported, is that Silicon Valley Bank built up a large securities portfolio, funded by deposits from tech and crypto start-ups. SVB also held a substantial loan portfolio. Based on the bank’s financial statements as of December 31, I made this schematic balance sheet:
This version of the balance sheet shows the securities portfolio at amortized cost. Amortized cost is a fine measure of value for securities that are held until maturity. In the low-interest-rate environment of the last several years, SVB and other banks could make a profit by accepting deposits at zero interest and buying securities, even if those securities didn’t pay much interest either.
The problem
SVB neglected the inconvenient possibility that its depositors would want their money back. The bank had some $14B of cash on hand at year-end; to cover withdrawals beyond that, it would have been obliged to sell securities. This is easy enough to do. But interest rates and bond yields have risen sharply over the past year, as the Fed has raised rates and sold assets. As a result, the market value of SVB’s portfolio is quite a bit below its amortized cost.
Those not in the habit of thinking about bond pricing can see it, very roughly, like this: the owner of a bond receives a quarterly interest payment, fixed at the time the bond was issued. Newly issued bonds, of which there is no shortage, offer a higher interest payment, reflecting today’s interest rates. So to attract a buyer for one of the old bonds, a seller must lower the asking price to compensate.
SVB’s balance sheet reflects these now-lower bond prices as “fair value,” noted but not counted in the balance sheet totals themselves. When depositors got spooked, asking for their money back suddenly and en masse, the bank had to reckon with market prices. We can re-write the balance sheet:
At market prices, the bank was insolvent—its equity was negative; liabilities were worth more than assets. Insolvency alone did not stop the bank from operating, but it did spark fear, leading depositors to pull out their money in a classic bank run, culminating in the bank’s closure on March 10.
It’s easy, and not incorrect, to fault SVB for poor management of its bond portfolio. But those securities would have fallen in value no matter who was holding them. Under the circumstances, it is SVB’s equity holders that must absorb the loss, but if it had not been them, it would have been someone else.
The system
Observing the run on SVB’s deposits, as well as those of similarly positioned Signature Bank, regulators made two interventions. First, to stop the bleeding, the FDIC chose to guarantee all of the banks’ deposits, not just those already covered by normal deposit insurance. This move has been widely panned as a bailout of moneyed interests. It’s hard to disagree, but I will spend my energy looking at the big picture.
This we can see in the second intervention: the Federal Reserve created a liquidity facility to try to stop other banks from falling into the same trap as SVB and Signature. The mechanics of the bank term funding program (BTFP) are outlined in the T accounts below.
BTFP is for banks with eligible assets (I show Treasuries, but agency bonds and agency mortgage-backed securities are also eligible) with unrealized losses, i.e. those in the same situation as SVB (1). Banks can bring these securities to the BTFP, where the Fed will accept them at face value as collateral for a one-year loan of money (2 and 3). The transaction can be represented like this:
These T accounts show the economic substance of the BTFP, which is comparable to a one-year repo loan. I write it this way to clarify what the Fed is doing. But note well that the collateral that the Fed accepts through the program will not be reported on its formal balance sheet, which responds to different considerations.
BTFP works by giving cash for securities, above market prices. This depends on the fact that the central bank doesn’t have to mark its securities to their market price; in fact the Fed always records its security holdings at face value. By funding the underwater bonds, the Fed is trying to buy time. BTFP has been created at a scale of $25 billion, so small potatoes. But credit limits can be increased overnight.
Considerations and implications
There is a bigger picture here in terms of monetary policy. Though everyone talks about rates, the Fed has also been selling securities, shrinking its balance sheet for the better part of a year and adding to downward pressure on bond prices.
In other words, with SVB, the Fed’s balance-sheet contraction has finally triggered a liquidity crisis. A smallish one, but enough to call the policy of contraction into question. With BTFP, the central bank is now offering to expand again, in a limited way. Whatever happens next, the underlying problem is not gone: unrealized losses on bank security portfolios are reckoned at $600 billion. To the extent that these losses must eventually be realized, their equity will be reduced. Banks will have to raise new capital, which is likely to tighten financial conditions.
The Fed’s case for its high-level policies (raising rates and shrinking its balance sheet) is based on the high rate of inflation of US prices. The inflation rate has slowed, but is still well above the formal target, so it would be difficult for the central bank to make a broad change in course unless either the SVB crisis spreads, or prices stop rising.
Otherwise, the Fed’s only option is to buy time to allow the overhang of unrealized losses to be worked off. That could happen slowly, which is what the Fed wants, or it could happen quickly, and hopefully not while I’m on vacation.