Musing on the Middle Kingdom and More:

SVB's Avoidable Collapse, Part 2

Oct 30, 2024

Here is Part 2 of my account of the decisions and dynamics that brought about SVB's collapse. It's been really interesting to look back at this part of my life, especially as I'm so engaged in getting my book about starting the China bank out in public. Only 15 more days until publication (pub) day! One of the things I'm learning is how important it is to get the book pre-ordered, so if you could go to Bookshop.org or Amazon or Barnes & Noble and place a pre-order now, it would be great! You'd get the book hot off the presses and the publisher would have a better idea of how many copies to print.

As always, thanks so much for reading this. Feel free to comment. If you’re receiving this in horror, please click on unsubscribe at the bottom of the newsletter.

And now, back to SVB and its demise....

The Avoidable Collapse

Part Two

Since Silicon Valley Bank (SVB) collapsed, I’ve spent a lot of time trying to figure out what exactly happened.

Here's what it looks like to me: Starting a few years ago, and continuing well into 2022, the venture capital market had gotten “frothy.” Let me explain what this means. In 2010, the last full year I was SVB’s CEO, venture capitalists in the U.S. invested only about $30 billion. By 2021 that amount had risen to well over $300 billion, or 10 times 2010’s total. And that’s just in the U.S. alone; globally the total had more than doubled. Why? For at least three reasons: 1) interest rates were too low; 2) the entire world had become fascinated with innovation; and 3) central banks around the world were flooding their economies with liquidity due to COVID.

SVB’s client companies had gotten their share of this money and deposited it with our bank. Our total deposits had ballooned over the period, reaching more than $200 billion in 2021. There was no way that SVB could lend out this amount of money, which is what banks normally do with deposits. Companies were flush with cash and didn’t need to borrow. Our loan growth couldn’t keep pace with growth in deposits.

But unused deposits (that is, cash sitting in a bank’s vaults) are a significant drain on return on equity (ROE), so the bank had to do something with the money. And this is where things went wrong. Remember, our expertise was lending, not investing in bonds. We should have stuck with lending, and left investing in bonds to other players in the financial markets. That’s what we had done in the past, when I was still at the Bank. However, management chose to invest in bonds on their own. But, rather than putting these excess deposits into low-yielding, very short-term bonds, where they would be immediately accessible if needed while earning something more than nothing, SVB decided to put a significant chunk of the money into higher-yielding, longer-term bonds. This reduced the drain on ROE—but meant the money would be tied up (or only accessible at a cost) if the bank needed it early. It’s like this: You get a better rate on a longer-term government bond than on cash in your checking account, but your money is tied up for the period of the bond. To buy a longer-term bond is to gamble that rates aren’t going up and that you won’t need the money in that period.

It seems the management team chose to gamble. Worse yet, they failed to hedge the “bet.” I’m not exactly sure how much that would have cost, or why they chose not to, but apparently they didn’t.

Of course, if interest rates in the U.S. had remained low, this gamble might have worked out well. But nobody can accurately predict interest rates. If a guess proves right, it’s pure coincidence. Thus, investing in longer-term bonds had the potential to be a terrible mistake. Like gambling, you could win, or you could lose.

In point of fact, SVB’s guess looked like it was a winner until March 2022, when the Fed started fighting inflation. And then rates rose with a vengeance. Even that change wouldn’t have made any real difference, except on paper, if none of the depositors had wanted their money back until the bonds had fully matured. But the higher rates took some of the froth out of the venture market and some of SVB’s clients needed their money, so the decision to invest excess deposits in longer-term bonds without adequate hedges was indeed a mistake. As it turned out, a fatal one.

By the end of the day on Friday, March 10, SVB was no more. Since then, the regulators have been breaking it up into pieces and auctioning them off to the most qualified and highest bidders. Even as I write this, not all the parts have been sold off. A few may never be sold. The bank had been designed as a group of mutually complementary products and services. Each and every product and service was synergistic with all the others, and all of them were tailor-made for technology companies. The value of the whole was, we believed, far greater than the sum of the parts. We were a full-service shop, so to speak.

Moreover, our culture also made the bank uniquely powerful. All our 8,500 employees were committed not just to their own success, but to the success of their clients and all their co-workers as well, no matter where in the organization they worked. And the compensation program was structured in a way that encouraged cooperation. That is, in my view, a rarity in the financial services industry.

In shuttering SVB, the regulators destroyed an important player in our country’s innovation space. SVB knew the innovation industry and the venture capital industry in a way that no other institution does. We’d grown up with it.

First Citizens, the North Carolina bank that bought most of the commercial banking assets, has collected the bulk of the employees of the old SVB. The rest have been scattered to the four winds, going to Stifel, J.P. Morgan, HSBC, and a few others. Of these groups, the one at First Citizens, under the leadership of Marc Cadieux (who had been at the old SVB for over 30 years), stands the highest chance of success, in my assessment.

But I doubt that anyone will ever be able to replicate the old SVB in its former entirety. It may take decades to create a replacement. And it might never happen again.