SVB quick post mortem

There will be hundreds of articles, books and business school cases on SVB and AFS (available-for-sale) vs HTM (held-to-maturity) securities, so this is likely my last take.

From what we know so far the story is pretty simple. The “average” pitch to the tech community was something like: “I’ll give you $15m to buy a nice house in SF and a vineyard in Napa Valley, and you keep the $100m your company just raised here as deposits” (on which SVB paid almost no interest, 4bps at the end of 2021).

Then it “recycled” these $100m deposits in (rough %):

• $40m in loans to entrepreneurs (full breakdown is in the image below): mostly capital call lines of credit to companies/VCs, but also personal properties (“private bank” is 90% loans for personal residences) and nice hobbies (loans for “premium wine” were so big - $1.2b – to deserve their own category...). A very concentrated loan book: same founders, same VCs, same boards, …
• $60m in marketable securities: these companies were awash with cash and deposits were way higher than available loan opportunities

They could have bought T-bills and 1Y notes, but the yields were very low. So they went with 10Y treasuries and MBS for some extra juice, but according to the CFO last year the yield was still only 1.65%-1.75% (and fixed, not floating like most loans). Also, if they needed this extra return it means they were likely underpricing the loans to win business.

With VC funding slowing down in 2022 (and less IPOs, less secondary offerings, no SPACs, …), deposits were withdrawn - not added to - to pay for operating expenses. But rates were going up, so they got killed on the fixed rate bonds they had to sell (although HTM securities do not need to be marked-to-market until sold). For non-fixed income specialist: the duration of MBS lengthen when rates increase, as fewer people prepay their mortgages to refinance.

Maybe they thought inflation was temporary; maybe they didn’t believe that the Fed would keep hiking so fast. But they had no hedged in places, no swaps, nothing. After the 3rd, 4th, 5th, … hike they should have realised something was brewing: at the end of 2022 the HTM unrealized losses were already over $15bn, so virtually insolvent. (*) Not having a Chief Risk Officer for 9 months from April 2022 to January 2023 didn’t help…

A couple of counter examples:

• Most of Bank of America’s AFS securities are hedged to floating rate: "We have about $200bn of treasuries there & they're all swapped to floating precisely to insulate us"
• When asked why Interactive Brokers doesn't extend duration to generate higher returns, the CEO replied: "That's the kind of risk that we do not want to take. We will continue to invest in T-bills" (avg duration of its portfolio is 42 days)

It’s amazing how many companies/banks/funds still don’t understand that the objective of risk management is not the P&L (“get an extra 10bps of profits”) but rather survive to fight another day.

(*) Incidentally, Eurozone banks hedge much more of their HTM exposure thanks to… Italy! In the US, if you completely hedge your treasuries duration risk you get the short-term rates, so what’s the point? In EU, if you hedge with the bund/swaps, you still get a nice spread, because you can invest in a variety of underlying government bonds, especially the deep pool of higher yielding BTPs. (**)

(**) I know, it’s simplified, you then have credit risk instead of duration risk but you get the point.