The most popular example of the fecklessness of Silicon Valley Bank is that it stupidly amassed a $124bn bond portfolio and then — even more madly — didn’t hedge against the swelling interest rate exposure.
But is that right? FT Alphaville dug into the balance sheets of SVB and Credit Suisse for an
incredibly sad geeky compare-and-contrast. The tl;dr is perhaps not quite as idiotic as a lot of people assume, but pretty dumb. Be warned, the following is acronym-heavy.
The first thing to remember is that SVB’s bond portfolio was basically in two different accounting buckets. At the end of 2022 it held $91.3bn in a “held-to-maturity” portfolio — bonds you plan to hold on to until they are repaid — and $26.1bn in an “available-for-sale” portfolio, which is marked to market.
Here’s a snapshot from SVB’s end-of-2022 financial accounts.
Let’s take the chunkier HTM portfolio first. Securities in the HTM basket can be carried at their nominal par value, because the assumption is that they are being held until they’re repaid in full.
As the table below shows, most of SVB’s $91.3bn HTM portfolio consisted of very-long-term, agency-guaranteed, mortgage-backed securities maturing in 10 years or more ($56.6bn to be exact).
The creditworthiness of this stuff is extremely high, but it’s also very sensitive to interest rates (for bond nerds, the average duration of the HTM portfolio was 6.2 years).
Because of rising rates the actual market value of the HTM portfolio was about $76bn at the end of 2022, according to someone who has seen the details of the portfolio and shared them with FTAV — an unrealised loss of $15.1bn.
Yes, SVB didn’t have any hedges on this bit. But doing so would arguably be nonsensical. Remember, the entire HTM portfolio is held at par, but the value of the hedge would obviously fluctuate with the market.
So if rates rise then a bank makes money on the hedge, but the bonds stay at par. If rates fall then they lose money on the hedge, but they can shift bonds from HTM to AfS and sell them at the higher price. That means it basically becomes a directional bet on interest rates that flows straight into the income statement, something that most banks abhor.
For example, Credit Suisse’s HTM portfolio of Treasuries maturing in 1-5 years stood at a pretty minimal $992mn at the end of 2022. The market value was about $949mn, but there doesn’t seem to be any hedge on here either despite the unrealised loss.
FTAV gathers that some big commercial banks often do hedge a bit of the interest rate risk anyway, just in case. But generally they just try to hold mostly shorter-term bonds to minimise the interest rate sensitivity.
That is something SVB definitely did not do — since ca 2018 they actually added a lot of duration by piling into 30-year MBS. But in practice, not hedging the interest rate risk on the HTM was probably not Silicon Valley Bank’s biggest mistake.
However, let’s turn to the AfS side. Unfortunately, here be dragons.
This is what SVB’s AfS portfolio looked like at the end of 2022. As you can see, it was mostly Treasuries. Remember, these are carried at fair value, ie marked to market.
That’s pretty big. For comparison, Credit Suisse held “trading assets” with a market value of $70.5bn at the end of 2022 — it constantly buys and sells securities of all sorts on behalf of clients — but its actual AfS portfolio (of mostly corporate debt) stood at $860mn.
The AfS bucket is definitely where most self-respecting banks lugging around a big portfolio of bonds will hedge their interest rate risk. Otherwise, the income statement would bounce around according to whatever the market does from one quarter to the next.
SVB seems to have been aware of danger. Here’s what CFO Daniel Beck told analysts in early 2021:
. . . We’re certainly positioning at this point for the potential for higher rates. So in the quarter, we put on close to $10 billion worth of swaps on that available-for-sale portfolio. And we’re going to continue to do more to protect against that, to mitigate the impact of potential further rate movement.
And at the end of 2021, SVB’s financial accounts indicate that on the AfS side it held $15.26bn of interest rate swaps to hedge against the impact of rising rates on its big bond portfolio. So what happened?
Well it looks that weakening profitability in 2022 as the tech world made SVB do something really dumb. In the first quarter, it unwound $5bn of AFS hedges to book a $204mn gain, and in the second quarter it dumped another $6bn of hedges to lock in a $313mn gain.
Or as the bank put it in a July 2022 presentation to investors, it was “shifting focus to managing downrate sensitivity”. (H/T the FT’s Antoine Gara for the below slide):
You can see the shift here in SVB’s 2022 annual report. By the end of last year it only had $563mn worth of hedges left on its books. For comparison, the notional value of Credit Suisse’s interest rate swap hedges was $135.7bn at the end of 2022.
Essentially, to juice its P&L in the short term, SVB ambled into 2023 almost completely unhedged — in effect, a massive multibillion-dollar bet that interest rates were approaching their peak.
Ironically, it was kinda right! The 10-year Treasury yield peaked at about 4.29 per cent in October last year, and after declining sharply in January only went as high as 4 per cent in early March (and it has since slid back below 3.5 per cent because of the mess unleashed by SVB).
However, the Achilles heel of SVB’s balance sheet was not the asset side, it was its liabilities. Specifically this:
Gormlessly, SVB had amassed a stupendous pile of uninsured deposits, almost entirely in just one industry that was burning through its deposits as VC funding dried up.
Deposits are typically considered very stable, sticky funding, but in SVB’s case it proved anything but. With money gushing out by last Friday and no way of selling unhedged HMT securities without realising an even bigger loss than the $1.8bn incurred when dumping most of the AfS portfolio on March 8, the FDIC had to come swooping in.
Bank balance sheets are a knotty business, and FTAV hopes we haven’t mangled anything here. But if we have, let us know in the comments.
Credit Suisse’s core problem clearly seems to be its stumbling business, and it has minimal exposure to higher rates. In contrast, SVB might narrowly be forgiven for not hedging more of its HTM book, but locking in low rates and leaving its AfS portfolio almost naked to bolster profits — despite a clearly unstable deposit base — looks like an asset-liability snafu that will become a cautionary tale for bank treasurers and regulators.