With First Citizens bank taking over the US assets of SVB in the United States, and HSBC having taken over the UK assets, Silicon Valley Bank has now disappeared for good.
So, with unmistakeable nostalgia, we are going to try to answer a simple question here:
Was SVB's business model fundamentally broken?
1. Low cost of funding has a cost...
I have to admit, I didn't look at SVB's balance sheet until 2023. As a competitor we saw them quite often, and my impression was that their credit process was pretty decent.
What we did see was that SVB was generally more aggressive than us on pricing. This seemed to be largely driven by a very low cost of funding - one which we could not match:
That's a cost of funding of 0.57%: lower than the US government.
However, as has been well publicised, the low percentage of SVB insured deposits (less than 3%) seems to have precipitated the run on the bank, so relying on such cheap uninsured deposits certainly is not without downside risks.
2. Trigger - the HTM losses
The trigger for the run on the bank was the rights issue they had to organise in order to plug a hole in the sale of their hold-to-maturity losses. The fact the HTM losses were more than 100% of their core tier-1 capital put them in a particularly bad spot, but others have done this. SVB's particular crime seems to have been to be reliant on a small ecosystem.
3. SVB's assets
The SVB loan book was barely 40% of its balance sheet. It had grown deposits at a rate that exceeded its ability to make loans. Even more surprising was its venture debt book - the area we saw them most - was in fact tiny, as I mentioned in an earlier post.
$6.7bn Venture Debt book (Investor Dependent - Early Stage and Growth Stage in SVB terminology)
That's 9% of the loan book, or barely 3% of the overall balance sheet. As we will see later, it is also where - unsurprisingly - many of the issues were; the rest of the book was mostly quite robust.
4. Overall a good book
This certainly aren't the kind of losses that has anyone going on a bank run, so whilst it no doubt has some issues - it certainly does not seem to point to a "fundamentally broken business model".
5. Issues principally in 4 areas
As we can see above, the issues were concentrated in its Innovation C&I book, CRE and the venture debt. The Global fund banking book was impressively resilient - though we are aware there had a few issues that aren't here (losses already taken and settled intra-year).
We're going to look at the venture debt book in particular here, since that's our bread and butter - but the C&I and CRE books do have their own sets of problem.
5. Worsening Venture Debt quality
Whilst overall the loan book was healthy, the venture debt book was showing some cracks, with Early-Stage debt in particular difficulty: over 15% of loans were past-due in some for or another. To be fair, this is not surprising given the state of the tech markets, particularly in 2022.
The question for many of these companies will now be - will they be able to refinance their debt now SVB is gone.
6. Some extend & pretend
This somewhat complex chart aims to look at one tell-tale sign of worsening risk situation: extend & pretend. If a bank's borrower gets into trouble, it is often tempting to extend the loan and pretend everything is fine. This however shows up in a worsening quality of older loans.
We see this clearly in the 2022 accounts for SVB: older loans show significant deterioration - with the amount of criticized & NPL loan over 2 years old being nearly double what they were historically. The cost? At fair value, this probably means at least a 10% impairment for venture debt loans of 2y or more will be needed, i.e. at least $200mm.
Given the prices First Citizen and HSBC paid for SVB, these costs - together with any issues in the C&I and CRE book - are easily absorbable. There is no doubt both HSBC and First Citizen have made great deals.
Conclusion
So was SVB's business model fundamentally broken?
In its over-reliance of cheap deposit from a single sector, yes
In its amateurish handling of treasury matters, yes
In its loan book overall, no - but there were areas of stress
In its venture debt book, no - but there were significant stresses, and it was almost certainly under-pricing the risk
The latter two may seem like relatively benign offenses. However, if - as seems likely - the poor return of part of the loan book led management to rely increasingly on Held-To-Maturity securities to earn a spread, then this was far from benign. The mispricing of the loan book risk may well have been a significant contributor to the bankruptcy of SVB.
This of course leaves us with an open question: will the buyers of the SVB book continue to provide credit to the tech ecosystem?
Comments