Actionable insights on equities, fixed-income, macros and personal finance Start 14-Days Free Trial
Actionable investing insights Get Free Trial

What we should not learn from the SVB crisis


Silicon Valley Bank has gone under. And the US government has rescued the depositors. What just happened?

SVB was 40 years old. It was conservative in a bunch of ways – it took risks but only with a small portion of depositors’ money. When you give money to a bank it typically lends the money out to earn interest, pays you a part and keeps the rest. Simple enough?

Apparently not. Because you could earn that interest in 10 years, but your customer may want money back tomorrow. Then what? Oh, easy, you say. Maybe 10% of my customers want money back within the next year, so I’ll keep 10% of the money in “short term” loans or something. And a similar calculation for two years, three years and so on.

Congratulations, because you have just done something called Asset Liability Management. This is the English equivalent of “Jab Mangega, Tab Milega”. (When you ask, you will receive) But it doesn’t have the same score in Scrabble so the “ALM” version won.

Regardless, here’s how even this “ALM” did crazy things with SVB.

The Problem with SVB was not “general” ALM. It was their customers.

Silicon Valley Bank served startups, most of which were funded by Venture Capital companies. The bank would also lend, as “venture debt”, more money to these funded companies. Backed by the shares that the company had  just raised money, with more equity “warrants” that allowed them to buy equity if they wanted, later. This model worked well in some companies where the returns were humongous as those companies achieved scale; the equity warrants captured significant upside. But remember, most startups fail, so it would have lost in many others; but the wins outshone the losses and SVB was fine.

Until recently. After Covid, a massive number of startups got funded large sums of money. In two years, from $60 billion of deposits, the bank saw inflows of another $120 billion. This massive amount of money was now in the bank as checking account deposits. They had to pay out zero. And they’d get the business because they were so friendly to startups.

The obvious thing was: invest that deposit into something earning decent returns and you’re done. And then, they had the right thought process – don’t give all that money out in loans, invest a good chunk in things that will be guaranteed to return your money. The stuff that is guaranteed payback in the US is government backed securities – Treasuries, T-Bills, and a strange beast called Agency Mortgage Based Securities.

There’s an “Agency” in the US, Fannie Mae, which is a quasi-government-backed institution. It buys mosells securities linked to mortgage loan pools, which is basically a bunch of housing loans. MBS pays out higher interest rates than the government, but is kinda-sorta-government-backed so it has the same sense of security. Usually, investing in this would win  you things like Banker of the Year Award (SVB did get one a few days before it was bankrupt) but there is much more to this game than awards.

SVB invested $80 billion (half its deposits) into MBS. Because US Treasuries were yielding 0.25%. And MBS yields were at 1.5%. To get this higher yield, and given this was “safe”, SVB piled on.

And then inflation hit the US in 2022. A few things happened:

  • Interest rates went up. The Fed raised rates slowly, but surely, from 0% to 4.5% in a year. When the risk free rate goes up so much, what happens to a bond? It drops in value. The most “safe” securities in the US – 20 year treasury bonds – saw a price drop of as much as 30% in a year. So yes, all bond prices went down.
  • MBS was adversely affected. The Fed was itself the biggest buyer of MBS, in the Quantitative Easing program. It stopped buying. The stopping of buying means a massive drop in demand for MBS.
  • And then, mortgages in the US are fixed-rate for a term. So, if rates fall, you pay back a mortgage and refinance it at lower. If rates rise, you continue to pay and don’t refinance. So when you look at the past you might say oh, people pre-pay their mortgages, which start at 30 year terms, within 10 years and refinance them. An MBS holder gets paid back principal when rates fall (means you don’t get the higher rates) and gets a fixed rate coupon with no prepayment when rates go up (means you don’t get the higher rates). As people began to realize this, they didn’t want to buy MBS, so demand fell even more.

Yields on MBS went to 7% – a much higher gap with treasuries than before. Basically anyone holding MBS – a long term product – saw their bonds fall 20% or more in value.

This is all ok, Deepak, you’re thinking. So what?

Well, just as interest rates were going up, SVB itself realised it has  problem with its customer base. Most of its depositors were startups. They didn’t need much loans (other than that venture debt that SVB gave some). But they raised money because they were not profitable. Meaning, they would burn cash. In the days of easy money, they would just go raise more cash when they needed it. But the days of easy money had become the days of 4.5%+ money which is rocket-science-difficult money. They couldn’t raise, so they burnt into their deposits.

SVB was losing deposits, and at the same time the investments it owned had lost 20% or more, today. Of course, they could hold for another 10 years and the investments would become “whole”. But their customers, the startups, didn’t have 10 years.

For some strange reason, they knew this in all of 2022 (when both losses and withdrawals were limited) and didn’t move a finger. In 2023, it suddenly dawned upon them that oh, these withdrawals could become a problem. So they sold whatever they could – a $21 billion portfolio called “Available for Sale” – and booked a $2 billion loss. To make up for the loss, they talked to a bunch of PE funds and others to raise that much capital back.

So far so good, you think. Everyone does this in rough times. Except at exactly this time, another, much smaller bank called Silvergate had just gone belly up for exactly the same reason. They had deposits from Crypto companies, and Crypto companies, instead of extolling the virtues of Crypto, started to spend their US dollars, which Silvergate had also invested in long term securities, and poof, it just collapsed. Imagine, with this background, you do something like:

SVB: Hey, I just lost a ton of money selling government guaranteed bonds, so please invest in us

Others: Wait, why?

SVB: Haha, these startups, they seem to actually not be able to raise money, so we’re bleeding deposits, but it’s all fine, just give us some more capital and we’re fine

The VCs: Excuse me. You don’t have money to pay depositors?

SVB: No, what we’re saying is, we do. It was locked in with MBS earlier, and we sold it now and have $20 billion, which is like gazomba huge types. So yeah, relax, don’t panic.

Everyone: Don’t panic?

SVB: Yeah.

Instantly, everyone panics.

The VCs, yelling: Get the hell out of SVB! Everyone! They don’t have money! You will lose everything!

(Sound of $42 billion being demanded from SVB in one whoosh)

In the banking system, perception is reality. If you tell people not to panic, they panic. And withdraw.

SVB then goes belly up because it doesn’t have the money to pay. It had $20 billion – in India that’s 160,000 crores, which if you had in your bank account, your Relationship Manager would call you all day from different phone numbers. And yet, they ran out of money.

It was some bad “ALM”, yes. But it was also a combination of extremely concentrated customer profiles (startups and VCs), very bad messaging timing and hubris, and some well-respected investors causing a bank run by publicly saying we’re telling people to exit.

The end result: FDIC took over SVB, fired top management, and promised they would pay the “insured” $250,000 per account in a few days, with some more time for the rest. The same VCs who were all very big on no-government-interference decided it was time for government interference, threatened to bring down the entire world, and then the FDIC said: okay dudes, all the deposits SVB are guaranteed. Oh, and this other bank, called Signature, is also going bust? Fine, guarantee their deposits too, in full!

(They intend to pay any losses entirely through the insurance collections that banks pay for each account. So the US tax payer isn’t paying.)

You might ask, why were startups parking money at zero, when interest rates had risen to as high as 5% recently? The VCs will have us believe that startups are techies, they don’t know finance. Then hire a CFO, no? Or get a virtual shared CFO or whatever the current sharding mechanism is called. In fact it would have made SVB realize much earlier that oh gawd,  deposits are vanishing.

The core problem was:

  • The $250,000 insured limit would be fine for most banks, but this bank had rich VCs and loaded startups with far more.
  • This customer profile needed money out during a time of rising interest rates
  • The bank knew, but chose to ignore it
  • The customers were vocal about leaving all at the same time

The recognition of any of the problems earlier might not have led them down this path this fast.

The Wrong Lessons From The SVB Crisis

You don’t waste a good crisis. You learn from it. (And indeed, you exploit it) So here’s some things people have decided to learn, which is entirely wrong:

a) SVB should have invested only in short term government securities. All of you banks who have short term obligations, should do that too.

b) The bonds that SVB held were so risky because they lost value.  But if SVB didn’t sell there would be no loss if held to maturity!

c) Startups who raised money aren’t at fault for not knowing that their bank could go under, so why should they suffer?

It’s not having wrong ALM, it’s staying with wrong ALM

Smug people on twitter have pontificated that this is just bad ALM, oh come on, the first thing you learn in banking. Hey, it was also the first paragraph or two of this post, too.

It’s pretty darn easy to say that SVB should have kept all their money in short term bonds. But I will say that decision was perfectly fine at the time. In fact, nearly every bank has this as a problem – they all buy long term government bonds. However, by mid-2022, it was clear – even from SVB’s own conference call – that the tide had turned in interest rates. Rates would go up, bond prices were starting to fall.

It was at this time – mid-2022 – that they should have “adjusted” their portfolios. Sure, it meant taking a loss, but it’s always better to take a loss earlier than to wait till later. The idea is to constantly adjust your portfolio to evolving needs. Our customer base needs money and can’t raise more, and we will bleed deposits, so let’s start slowly taking the losses we should. Heck, or even buy interest rate hedges to cover our losses in case rates go up further.

In India, the Franklin Templeton situation in 2020 was similar – they had illiquid longer term investments, and investors could withdraw when they wanted. Unfortunately, there, they simply couldn’t sell some securities (no one wanted them because of credit risk) and so they had to freeze withdrawals; SEBI then got in, and made SBI mutual fund take over the further sale of all bonds in the closed funds. Here too there was a problem in that it might not even have been bad lending (indeed, most people have got their money back since) but it was the maturity of the underlying bonds that was a problem. Franklin could have also adjusted constantly but they were the biggest or only lender to many companies, and no one else would buy those bonds, which led them to a crisis.

You would expect banks to do constant adjustment, at least with liquid (and safe) bonds. And banks in the US have to reveal all such losses, so it’s not like they didn’t know.

The lesson really is: you can make a mistake, but you shouldn’t stay wrong. Regularly adjust a portfolio that you may have to liquidate any time. Especially when your customers can demand their money any time. In one sense, it’s also relevant for us as a PMS – if we invest in illiquid stocks, we have a big risk that we can’t exit in a steep fall. So we choose only liquid stocks, even if it means avoiding a lot of companies that might have made us profit.

The corollary is: if you have money that’s locked in for a long time you can then invest longer term. Banks give fixed deposits that are “callable” – meaning, the customer can say close my FD and give me money right now, minus a penalty. Even if I have a 10 year FD, I can withdraw it when I need it; it’s not really a long term instrument for a bank. However, if a bank issued long term bonds, and I bought it, I cannot redeem early. I can sell in the market, but cannot redeem with the bank. So the bank can invest that money for the long term and do actual ALM. But banks, even in India, don’t issue long term bonds for the most part – something they should quickly learn from this crisis too.

The lesson for us in the stock market is simple: in stocks, don’t have money that you need to use in the next few years. At any time, if there is a need for money in the next year or two, and you don’t have other sources, take out money from stocks. One of the most favoured reasons to sell is: when you need the money. The flip side: if you don’t need the money, and you like what you’ve invested in, stay for the longer term, because you can.

On that note: You don’t have to be “loyal” to your stocks. The stock doesn’t know you own it.

The risk isn’t only that loans lose money when they don’t pay back

One myth is, and I’m someone who also said it, that SVB invested in “good bonds”. It was just that they were long term in nature. But they would pay back, guaranteed.

Let me give you an example of banks in India, who don’t drop their lending rates when interest rates fall. They complain that people who lock in long term money using fixed deposits have to be paid higher deposit rates, so they can’t cut their lending rates. This makes no sense because

a) FDs mature all the time, and they’re renewed at lower rates.

b) There are ways to “hedge” your interest rate, using swaps. You can trade it to say you will pay floating rates, and get fixed rates. Since your deposit means you have to give fixed rates, you can convert the deposit into a floating rate using that swap. Banks can hedge our part of their portfolio in that way, so this excuse too isn’t valid.

Now if you “hedge” the wrong way, you will still lose money. An exporter that hedges his USD invoices faces an opportunity cost risk when the dollar rises heavily – but the exporter will only be paid at the lower exchange rate because of the hedge. And because he’s hedged, a customer might decide to go with a cheaper competitor who hasn’t hedged (who will get greater rupee revenue).  The problem wasn’t that his customers didn’t pay, but it was because his “risk management” didn’t play out.

A bank has a similar problem. We say “bad lending” because we hear of subprime borrowers, shady promoters, or unprofitable companies. But in many cases, it’s pristine borrowers – like the government – and you still lose money. When interest rates go up, bonds lose money, even if it’s bond issued by the government. It’s how the system works and is called “interest rate risk” or “market risk” – both terms are related when it comes to bonds. The concept is: if you’re losing money, find out what you’re losing (marked to market, or MTM) and then provision for that loss accordingly.

If MTM losses are too steep, it will mean you need help, especially if your customers can withdraw at any time. A bank must knock the regulators’ doors early. But there’s ego, reputation and bonuses – each of which make banks delay till the last minute.

Bank regulations allow banks to not take MTM losses on certain bonds, called “Held to maturity”. In Warren Buffet terminology, stuff they want to hold forever, at least until the issuer pays them back. (This is fine unless too many depositors ask for their money back earlier) To avoid stupidity, RBI only allows 23% of deposits to be marked HTM, and this will fall to 19% in the next few years. The US Fed has no such upper limits, so SVB could pretend they had no losses on 50% of deposits. In fact, the US gave a special privilege to SVB to not even mark losses on bonds marked as ‘Available for sale” (AFS). The Fed is quite incredible in the lack of regulation here, but I expect a lot will change soon.

But it’s not all hunky dory in India: The Home Loan Tenure Extension Example

A recent survey of home loan borrowers by Livemint shows that in certain cases, banks have increased loan tenures to as much as 50 years. For the same monthly EMI, when interest rates go up, borrowers cannot increase their monthly outflow. So banks tend to increase the tenure of a loan instead. But at some point this makes no sense – how can a borrower pay EMI till he’s 75?

If you raise rates from 7% to 10% in two years (this is exactly what’s happened in India from 2021 to 2023), the formula breaks. The original monthly EMI doesn’t even service the interest on the loan, leave alone the principal!

Then what happens? The bank tells you to raise the EMI. At this time, the risk managers at the bank are worried, so they reset the term back to, say, 20 years. And your EMI “suddenly” jumps up by 25%. Some people might just have to stop paying or attempt to sell their property if the loan EMI ends up being too much, and there will be delays, defaults etc. This could entirely be avoided.

The way to avoid it was to increase the EMI slowly every little time the RBI increased rates. But it’s also that banks should lend more carefully against property, leaving enough buffers in homeowners’ hands to deal with such increases. Giving loans when you take 50% of a person’s take home income as a monthly installment, is downright crazy – any increase and the borrower could be in trouble.

It’s not a bad loan – the house might eventually give enough money back when sold – but it’s a huge systemic risk. When every bank does this “extend the tenure” thing, then every bank has to do the “increase the EMI” thing at the same time, causing stress. Then borrowers who have a tough time repaying, try to sell houses at the same time. More supply, less demand, and the prices fall, hurting banks, all at the same time.

We often ignore the fact that risks exist even when everything looks “safe”. You don’t know what you don’t know, sure. But in too many cases, like the tenure-extension example above, you know but pretend to not know because oh I might take a loss today.

The risk is often in what you know, but choose to ignore because it looks “safe”.

Startups (and everyone, really) must manage liquidity better

India’s just had a Yes Bank, two years ago. Depositors were rescued in 15 days. Which is a good thing, but it made a lot of people including startups very very worried. Imagine that you can’t access money for a couple weeks, even if you have it in your bank.

Since then, of course, most startups that have raised a lot of money in India tend to keep money in different banks. Given that interest rate cycle is rising, it’s very important to do this:

  • have at least two bank accounts that are operating.
  • Due to RBI rules, companies that borrow from banks may not be allowed to have more than one operating account (from which regular payments are made) but in such cases, attempt to split borrowing from two banks so that you can still have two operating accounts.
  • You can keep money in very short term liquid instruments (liquid funds, ultra short term funds) instead of in a bank FD
  • Reduce institutional risk (more than one mutual fund, more than one bank account)
  • Monitor the “safe” risk regularly – for mutual funds, it’s about what they invest in, which is available once a month. This applies to startups because individuals may not have the time or skill to do it.
  • For banks, Indian banking is safe for depositors as there is almost an implicit guarantee that RBI will rescue depositors. Leverage rules are tight too, so banks don’t go overboard. However, many NBFCs (who issue bonds or take deposits) are overleveraged, and depositing too much in them can result in trouble.

In general, diversify, even at the cost of returns in debt instruments. Safety trumps performance. In fact, safety *is* performance.

Note: At Capitalmind Wealth PMS, we have debt portfolios for both short and long term to help avoid the pitfalls that we’ve seen earlier. Not the best return, but we navigated through the 2018 and 2020 debt crises reasonably well, if I may say so myself. Look us up if this interests you!

This post is finally coming to an end

The point about this post wasn’t as much to tell people that SVB was right (it wasn’t) or that the world’s banking system is in trouble (it might not be) but to build a thought process around risk.

In the lines that I heard in the movie, “The International”, and I replace the word “Destiny” with “Risk”:

Risk often finds you on the path you take to avoid it.


Like our content? Join Capitalmind Premium.

  • Equity, fixed income, macro and personal finance research
  • Model equity and fixed-income portfolios
  • Exclusive apps, tutorials, and member community
Subscribe Now Or start with a free-trial