How the last-ditch effort to save Silicon Valley Bank failed
wsj.com
wsj.com
Matt Levine: Silicon Valley Bank Ran Out of Money - https://news.ycombinator.com/item?id=35265027
Having to fire sale bonds (to meet withdrawal obligations) that are selling at a discount because any potential buyers have better yielding instruments to buy results in solvency concerns from the relative excess losses.
As long as the bonds yield more at maturity than you paid for them (along with the cost of the money you used to pay for them, which is very cheap for most large banks), then holding them to maturity is guaranteed to pay off. Whereas we have no idea how cheap they could ultimately become on the spot market at any given time that the owner may need to fire sale them.
Why should depositors leave more money than they need for short-term working capital in the SVB at 0% when a Treasury bill pays 4% with less risk? When all rates were roughly zero, it was easy to just lazily leave everything in the bank; but when rates increased, the reward for leaving and the risk of staying (because the bank is now mark-to-market insolvent, and thus particularly vulnerable to a run) both increased.
When interest rates increased, the NPV of the SVB's assets decreased. Someone had to take that loss. Their managers and shareholders presumably hoped that would be the depositors, by leaving their money in a risky bank earning interest below the risk-free rate, slowly accepting the loss over time. The depositors had no economic incentive to do that though, and they didn't.
For emphasis, the SVB's problem isn't that their bonds are trading at irrationally low prices due to liquidity problems (i.e., a "fire sale"). The market price for their bonds has behaved exactly like a textbook NPV model, and similar assets trade with normal tight spreads; the price is just lower than they wanted. Hold-to-maturity accounting allowed them to ignore that for regulatory purposes, but accounting doesn't change reality.
> Their managers and shareholders presumably hoped that would be the depositors, by leaving their money in a risky bank earning interest below the risk-free rate, slowly accepting the loss over time. The depositors had no economic incentive to do that though, and they didn't.
That's why the grandparent comment mentioned a large bank buying SVB out. Large banks have diverse, non-business deposits, many of which tend to sit, and not chase yields. People in more precarious positions keep emergency funds in the bank, not in CDs or bonds that they can't easily cash out on a moment's notice, or stocks that are risky. People in even more precarious positions are constantly paying money to the bank at a steep interest to cover their credit card payments. This is really cheap money to the bank (and more easily liquidatable at par than low-yielding bonds), though it has to cover those in the most precarious positions who default.
I guess a lot of people hoped that a different large bank would buy the SVB or its assets at a price that would have maintained the accounting fiction that it was solvent. Such banks had no economic incentive to do that though, and once again they didn't.
Hypothetically because they are martyrs who also want whatever goodwill SVB created in its clients (which now become the clients of the other bank). But yeah, it's a stretch.
But that's pretty risky for a bank.
The mark-to-market hole at the SVB was pretty deep though, and their depositor base had just made itself look particularly unattractive. So maybe the FDIC will eventually manage to find a buyer, but it indeed looks pretty hard.
Quite true.
If you hold that bond to maturity in 5 years the US government will give you your $100 in full though.
The Fed has now started a program that lets banks use their treasury bonds as collateral against cash loans - if this was going a few weeks ago it would have saved SVB.
As in, does the program let you borrow cash from the government at a lower rate than the bonds are paying? How could that work?
The program is undercollateralized though, so if a bank fails with such a loan outstanding then the Fed loses money. That part is a subsidy.
The banks share (and bond) capital is on the line as well.
It's still a subsidy though. For example, if you lend my just-barely-solvent bank $100 unsecured, then I can bet it double or nothing on a fair coin flip. If I win then I keep all the gain, so it's +$50 EV to me. If I lose, then my bank fails and you take the loss, so it's -$50 EV to you. Banks are obviously regulated in many ways to discourage that kind of risk-taking, but this policy change slightly weakens that overall edifice.
I think the program might not work like you think? Beyond the initial decision to create the program, no one at the Fed exercises significant discretion. It's open to essentially all American banks, most of which are private businesses with the usual profit motives, whose managers decide when to borrow and how to use the proceeds. If such a bank gets in trouble, then its managers now have slightly greater incentive and ability to try gambling their way out, since their shareholders get all the upside and the Fed gets some of the downside.
https://www.federalreserve.gov/newsevents/pressreleases/file...
one-year overnight index swap rate plus 10 basis points - so call it (currently) 4.7%
What does this say about the rest of the banks in a similar situation wrt their balance sheet?
Are you sure the limits on your account would apply to e.g. a wire transaction? Of course there are limits on ATM withdrawals and some other transfer methods, but I'm not sure they'd apply if you wanted to send a wire.
Not many people (at least, people who don't deal with large sums) have experience with doing a wire transfer. Sometimes they're a bitch to deal with - but it's what you'd use if you are doing very large sums (like, in the order of 6 or 7 figures or more).
In most of Europe, wire transfers are really easy to do, and where I live, most banks now support instant wire transfers (10 seconds) under some limit (in my case, 15 000 Eur). Doing them using QR codes and smart banking apps is really straightforward and user-friendly, so people do them routinely even for small sums.
that's not the wire transfer i am referring to - it's the one where you'd use a SWIFT code to designate the bank.
Between EU countries, you need SWIFT, but within a single country, you usually don't (I think the exception is Poland, where SWIFT is the default...?).
In the US the only limits are via the ATM or Debit card. If you write a check, go in person, or send a wire, you can withdraw as much as you want, with no time delay or restriction.
> Couldn’t they just use one of their terms or conditions to delay the withdrawals and give themselves ample time to react?
That's simply not a thing.
That said, even wiring 6-7 figures isn’t a big deal as long as it’s normal for your account and there are some basic security things setup.
I have had some call backs from their anti-fraud department after putting in a wire request to make sure I wasn't falling for some scam.
sure, customers will periodically get mad and take their business elsewhere, but it's a fever dream to think that elsewhere is any different.
If the funds are in a DDA (Demand Deposit Account a/k/a standard US checking account), the bank is legally obligated to honor outflow transfers of any size, at any time.
(Not physical cash withdrawals, although they have to make timely arrangements to satisfy these too)
Banks can make DDAs unattractive for large balances (e.g. no interest), but I'm not aware of any limitation on a customer's right to access their funds held in a DDA.
If banks could bend these rules, bank runs would never happen.
they should have legally obligated SVB to remain solvent!
They did!
When SVB failed to meet those regulatory obligations, the bank was seized and the depositors were made whole.
So the system worked, right?
Without these changes, the SVB's depositors would have had access to maybe 50% or more of their uninsured money immediately, and maybe 90% or more eventually. They'd maybe have been made whole eventually; but the FDIC's inability to find a buyer over the weekend suggests that the SVB's assets weren't obviously greater than its liabilities to depositors, so maybe not.
The SVB's depositors have been made whole now only because regulators intervened with emergency policy changes to rescue them. That might have been a good idea, since it stopped contagion; or it might have been a bad idea, since it encouraged future risk-taking in anticipation of a similar ad hoc rescue. It certainly wasn't any kind of rules-based system working, though.
The SVB had been insolvent on a mark-to-market or NPV basis since around September. Accounting rules on bonds they intended to hold to maturity allowed them to ignore that, but didn't change economic reality.
I'm no economist, but I think the (1-year) window for banks to borrow against the full face value of government bonds and MBS assets is interesting and probably reasonable. These are not risky investments, just illiquid. The Fed will get their money back. Providing liquidity to the system is part of the Fed's job.
> These are not risky investments, just illiquid.
It's not a question of liquidity. Similar assets trade with tight spreads, at prices very closely predicted by a textbook NPV model. The price just went down when interest rates went up, exactly as expected. There's no significant uncertainty in the price. Waiting won't make it go up, except in the same sense that waiting turns $100 in Treasury bills into $104 a year from now.
The Fed is making an undercollateralized loan. If a bank fails with such a loan outstanding, then the Fed will lose money. Interest rate risk is as real as credit risk or any other risk, and this would be a real economic loss.
From what I understand (could be wrong) the discouragement to risk taking is that the risk takers were wiped out in this case, only the depositors kept their money.
I don't think these are necessarily bad decisions--it's reasonable to make an example of the worst offender, and then help the rest survive to prevent systemic contagion. It's absolutely not "the system worked" though, except to the extent that the system is regulators making stuff up on the fly.
so it was a perfect storm, for them.
It sounds like you are referring to something like an ATM withdrawal limit or online transfer limit which is just a risk / fraud limiting measure for those particular channels rather than a limitation on the account itself.
When you have treasury access you can wire as much money as you want practically instantly just like with SVB. I bank with one of the big 5.
Nothing is stopping you from going into your bank and wiring money away it's just cash that they don't want to give out without multi-day delays.
Also, it's not a cash withdrawal. We just wired the money to other accounts. Cash would require tons of paperwork.
But the fake money never gets printed. It just lives in bits and bytes. You can move it around as much as you want.
Fortunately, it's as good as real money. It's just not as good as real wealth. Real wealth is in ammunition. Never goes down in price, skyrockets in price when the world gets more dangerous, and can be used to protect itself. That's where America has you beat, because we can turn fake money into ammunition whenever we want, as fast as we want.
Not to mention, it's a consumable resource that keeps well in storage but requires a high level of technology to produce. And even under non-apocalyptic conditions, it's about half as efficient as silver coins by weight and volume.
People downvoting this are politically opposed to guns because that's the fashionable thing to be in SV, but hackers have a strong gun culture. Always have, always will. Because we're practically all either anarchists or military.
Are you sure they're not downvoting you for being completely off topic while ranting about a hobbyhorse?
The banks have corporate customers sign a number of complicated forms to put limits on things like who can wire, who can sign, who can weite cheques etc.
There are further controls in the online corporate banking
"The Fed needed a test trade to be run before the actual transfer could occur. That took time and the Fed didn’t extend its own daily deadline of 4 p.m. PT for collateral transfers to help SVB. Time ran out on the bankers and SVB couldn’t get the money that day."
Though, maybe they should never been in that situation to begin with.
I don't know if "the experts" all agree with "this expert" but he pointed out that in buying all the mortgage backed securities, SVB got to put them in a "hold to maturity" account that is not "marked to market" (the value in the account does not match their decreased market value, so on the books it looks like a lot of value in there, but in reality it's not)
at the same time, as the coupons on these bonds pay out and the bank collects them, that money just shows up on the bank's income statement, not landing in the hold to maturity account itself as part of the value of the asset. This rings all the "hey, look we are profitable" and "hey don't I get a bonus" bells, twisting incentives for the bank and its officers (and other banks) to keep do this, lather rinse repeat.
recipe for disaster when interest rates go up.
If the accounting auditors are doing their job then the bank managers don't get to cook the books. The accounting rules are set by the accounting industry. If the financial statements don't reflect what's happening, it's the accountants fault. So, it's incorrect to imply that bankers are doing something nefarious. At this point no one has suggested an Enron or FTX type scandal.
What's more likely is just the incompetence with liquidity management, incompetence with the crisis handling.
no, what's more likely (if what has been said between the two of us comports with reality) is that the incentives for the officers of the bank are in conflict with the healthy management of the bank.
Bankers running a bank into the ground cannot be blamed on regulators or the accounting profession which only "review" what has happened, they never have their fingers on the levers (although additional accounting and banking regulations may be required to avoid this type of problem in the future)
I think the more subtle distinction would be making short term risk adjusted incentive identical to long term risk adjusted incentive, so people don't mistakenly think gambling is a good idea.
The accountants don’t just review. As i mentioned, they established the accounting rules for the financial statements - which incentives are measured against.
The accounting is so far off that financial analysts spend a significant amount of time un-doing the obsfuscation the accounting does. And if they can’t see it, stuff like SVB happens
My understanding was that realising losses on swaps used to hedge their MBS portfolio led them to require more regulatory capital and when they mishandled communication during the raise, it failed leaving them essentially dead in the water. The next day all the VCs who had been in conversations about the raise told their portfolio cos to move funds out, those founders told their friends and SVB was a dead bank.
Banking on the West Coast is hard, even harder in further flung areas like Hawaii.
The banks open in normal business hours but cant get a single thing done if its after 4pm in New York City. So the further west you go, the fewer amount of hours you have to get anything moved.
Often, overregulation by Hawaiian authorities is cited which is a just criticism.
So it is for crypto as well, nonzero but very limited
That is truly fascinating and something I hadn’t noticed when visiting. I wonder what implications this has for capital intensive construction projects there.
Presumably the builders still finance through major institutions but need to pay out all local labor via bank accounts from institutions on the island? So the local authorities have essentially taxed banking generally?
I needed a deposit for a sublet there, and needed lots of cash for that. So when it posted the next business day, I walked to the teller and asked for cash. It wasn’t a problem except for everyone else that didn't imagine and was trying to use fintech apps and learned about arbitrary transaction amount limits.
For regulatory purposes, the SVB was permitted to ignore those losses in bonds that it intended to hold to maturity. Accounting doesn't change reality, though. If the SVB's assets are actually worth more than their liabilities to depositors in any economically meaningful sense, then why do you think the FDIC was unable to find a buyer?