The whole point of banking is that you borrow money from your depositors at low interest rates, but at a variable term length (The depositor can always withdraw), and you lend money to borrowers at high interest rates, but at a fixed term length (The bank can't just call in your mortgage tomorrow.)
Borrow short, lend long. The latter necessitates 'locking money up'.
A well-managed bank will properly manage the risk of the short loans getting called.
A poorly-managed bank will go all-in on getting short loans from people who are likely all going to call them in at the same time (startups), while putting their entire lending portfolio into lending long in an environment where long-term loans are dropping in value.
I know you are generalizing but there are times the yield curve is the other way and it is better to lend short and borrow long. But a well managed bank takes care to duration(not maturity) match their assets and liabilities while also taking into account liquidity needs and buffers. Also, the above applies to use of their own capital as well.
Demand deposits can be immediately be recalled, so where exactly do you suggest they park it? In the central bank -- no bueno they've denied banking license for narrow banking. Margin lending that allows recall at any moment? I can think of some options but frankly I'd rather have my money in a bank that over-extends themselves on treasuries than most the alternatives. At least I'd most likely get 90+% of my money back eventually.
Only retroactively in a bank run are you really able to see just what duration and what amounts were the limit.
It's not that they shouldn't have bought treasuries, it's that they shouldn't have bought such long dated treasuries, and if they did, they should have hedged against interest rates, and if they didn't, they should have realized the loss when it was smaller. But they did none of those things and it was fatal to them.
The Fed kept making it clear that it was raising rates, and it seems like SVB just slipped quietly into that good night without lifting a finger to save itself. Which is bizarre and confusing and there must be more to the story (and details are coming out, like the risk manager role remaining open for nine months), but it does seem like crazy risks were taken. But not in pursuit of additional gains, like we are used to seeing, but it's looking more like negligence or a misunderstanding of their position.
You can `s/treasuries/mortgage-backed securities/g` into my comment and it doesn't change much, but my understanding is that they had a lot of treasuries (not to the exclusion of having MBSs).
> To fund the redemptions, on Wednesday Silicon Valley Bank sold a $21bn bond portfolio consisting mostly of US Treasuries.
All your link shows is that Guardian, Reuters, and others have equally as bad reporting as commenters here, just parroting each other constantly...
SVB's actual announcement says "Additionally, earlier today, SVB completed the sale of substantially of its available for sale securities portfolio. SVB sold approximately $21 billion of securities, which will result in an after tax loss of approximately $1.8 billion in the first quarter of 2023."
I haven't seen any evidence that there were substantial Treasuries sold, I just see MBS on their balance sheet.
So put it where? Narrow banking is illegal by virtue of denying the banking license. You're basically left with what, something like recallable loans/margin? What are the other options?
Put it in bonds of whatever duration the bank chooses, but require sufficient equity that the shareholders will bear the loss and not the depositors?
Interest rates didn't increase in a single step. If the SVB had been forced to recognize their losses on a continuous MTM basis, then they'd have been forced to raise capital (or liquidate if they couldn't) by late 2022, when they were undercapitalized but not insolvent. The shareholders might still have been zeroed, but the depositors would have been fine.
In fact, the SVB designated those bonds as held-to-maturity, which allowed them to avoid reporting the loss, leaving them adequately capitalized for regulatory purposes despite being MTM insolvent. That accounting treatment doesn't change the actual economics though, so they still blew up.
>Put it in bonds of whatever duration the bank chooses, but require sufficient equity that the shareholders will bear the loss and not the depositors?
But that's literally what they did. They put it in 10 year treasuries that they had to sell for 87 cents on the dollar because every "thought leader" in Silicon Valley had the same idea at the same time and triggered a bank run on their own bank.
Everybody who has deposits will get 100 percent of their money back and everybody who holds equity in SVB will be (mostly) wiped out.
Correct. And they failed at it. Swap bonds for magic beans and see if it’s defensible. We know the bonds were worth less, but it doesn’t obviate failing to manage the half of the business that isn’t people handing you money
Correct. I think you are being blinded by the word "bonds" when evaluating the proportions you assign to mismanagement vs. being blindsided. I feel inviting swapping in a word that is anything except bonds may be a way to continue this interesting conversation