I really thought I was done writing about the Hare, but then I realized it was directly relevant to the Silicon Valley Bank (SVB) disaster.
As a brief refresher, I wrote during the height of the insuretech boom how the Lemonades of the world tried to win the startup race by sprinting to the front. While we know normally this is a losing strategy (and it is still is from an underwriting perspective), it could prove an effective tactic during a time of cheap capital.
How is this relevant to SVB? Because SVB’s failure was directly tied to having bank relationships with too many Hares.
They saw a surge in bank deposits in ’20 and ’21 from Hares fresh off new fundraises. Because SVB doesn’t have much of a lending franchise, they invested these dollars in long term Treasuries, as we all are now aware.
Like their clients, SVB focused on what was best for today, rather than consider if it would make sense tomorrow. Therefore, they didn’t hedge the interest rate risk in those Treasury bonds.
Why not? Because, silly rabbit, earnings growth would be higher if they didn’t hedge! Grow fast and break things doesn’t leave time for contemplating risk management.
The Smell of Burning Cash
What happened next? All those hares started learning about cash burn and realized they had too many expenses and not enough margin. Since they could no longer rely on raising money, they had to start consuming what was in the bank account.
This meant SVB’s deposit balances would decline. Unfortunately, because they invested in long term Treasuries, they didn’t have natural maturities to provide liquidity as cash left the bank.
This is an important point that has been largely overlooked. There is a lot of commentary about bank runs happening because deposit rates are too low or because customers are worried about FDIC insurance.
But the initial strain on SVB’s deposits wasn’t because customers were voluntarily leaving the bank. No, they were running out of money themselves.
In other words, cash burn by Hares with their tail on fire began the collapse. If its clients had better business models, SVB would still be standing.
How To Run A Bank
Now, of course, SVB’s dumb decision to take on a significant duration mismatch only made things worse. What’s unfortunate is this was easily avoidable.
Way, way back in the day, I worked at a large regional bank managing these risks. One of my responsibilities was modeling the sensitivity of deposits to interest rates, economic conditions, whether the customer had other products with the bank, account size, etc.
They let a 22 year old do this. It’s not all that hard.
It is not terribly different than managing an insurer’s portfolio. You have estimates of when claims will be paid, but they may come sooner or later, so you have to plan for that.
In the same way, you can estimate the stickiness of deposits, but they may leave before or after you expect.
Thus, you have to hold a liquidity reserve to handle any unexpected spike in outflows and you have to match duration so your bonds mature at roughly the same time as your liabilities.
There are a few really basic things you can do to handle this. One is you barbell your investments.
So instead of buying 5 year Treasuries, you buy perhaps 1/2 three month treasuries and 1/2 ten year Treasuries. This ensures you have rolling maturities for cash flow and also that, if rates change, you can reinvest at the new rates.
Another option is to buy interest rate caps. So if your model says the bank’s capital would be offsides if rates went from 1% to 3%, then you buy caps that pay you if rates go over 3%. Now, you have reduced the unrealized loss risk and you also produce some cash to fund exiting depositors.
There are many other tactics available as well. This isn’t rocket science. Managing duration risk is much easier than managing credit risk.
Why was it missed? Because as we keep learning, Hares are really good at focusing on shiny things and really bad at focusing on key operational metrics that are needed to be financially healthy.
Correlation Can Imply Causation
SVB ignored that its deposits were more vulnerable because of its client base – having correlated clients can cause them all to have trouble at the same time.
Note, they didn’t get intro trouble only because they were startups, but specifically because they were Hares who would run out of cash.
Unfortunately, Hares went extinct due to financial climate change caused by rising interest rates.
If SVB had prepared for this possibility by diversifying or hedging, they may have been able to adapt and survive. But since they were a Hare themselves, of course they didn’t do this.
Worse, they doubled down betting low interest rates would last forever. This was the fatal error. They didn’t understand higher rates wouldn’t just mean their interest rate bet would fail, but that it would simultaneously expose their clients.
In investing, we call this a Texas hedge, which is another way of saying you weren’t hedged at all. You actually made the same bet twice unknowingly.
So the lesson is if you’re going to be a Hare, don’t also have Hares for clients. Two Hares are slower than one.
And if you are going to have Hares for clients, you need to be a really good operator who knows how to prepare for unexpected change – which SVB clearly wasn’t.
Once again, the Tortoise wins. Banks should be run deliberately and prudently.
If you are going to run a bank like a Hare, your bank will eventually get run.