V zvezi s propadom Silicon Valley Bank (SVB) je najbolj očitno dejstvo, da je bila, milo rečeno, banka slabo regulirana. Njeni upravi je uspelo od zakonodajalca izposlovati (2015), da čeprav njena bilančna vsota presega 50 milijard dolarjev, pa ne predstavlja sistemskega tveganja za bančni sistem, ker je majhna in ker ima najbolj varne naložbe (no, prejšnji vikend je najvišja finančna trojka v ZDA propad SVB razglasila za sistemsko tveganje). Zdaj pa vprašanje: če se SVB ne bi izmuznila regulatornemu nadzoru, bi FED s stres testi lahko zaznal njena tveganja?

Thomas Philippon, profesor na New York University, pravi, da ne. Iz preprostega razloga, ker obstoječi stres testi niso sposobni zaznati tveganj, ki so jim banke podvržene danes. Namreč scenariji šokov v obstoječih stres testih so narejeni za negativne šoke v povpraševanju: padec BDP, povečanje brezposelnosti, povečanje obrestnih spreadov, pri čemer pa se obresti na državne obveznice znižajo. Toda zdaj imamo opravka z negativnim ponudbenim šokom in pozitivnim šokom v povpraševanju. Posledica obojega je povišana inflacija, na katero se centralne banke odzovejo z dvigom obrestnih mer, kar pa vpliva na povečanje obrestnih mer državnih obveznic in posledično zniža njihove cene. To pa povzroči, da “zavarovanja” depozitov v banki z državnimi obveznicami postanejo premajhna. Natanko to, kar se je zgodilo v SVB.

1. SVB part 3: If only SVB had been forced to run stress tests on interest rates… right?
Well, it’s a bit more complicated than that.
In 2010 Dodd-Frank introduced tough requirements for banks with more than 50b in assets.
Then Barney Frank became a de facto bank lobbyist…

2. ..and he (and others) helped push the limit to 250b. Becker (ex CEO of SVB) told the senate banking committee in 2015: “Because SVB’s business model does not pose systemic risk, imposing the numerous Dodd-Frank requirements that were designed for the largest banks..

3. ..would place an outsized burden on us.”
The increase to 250b allowed Signature and SVB to grow without more oversight.
That much is clear. What is less clear is whether supervision would have worked.
This brings us to stress tests.

4. Stress testing has become (arguably) the most important tool for regulating big banks today.
But the value of the test depends on two key features:
– what are the stress scenarios?
– how are losses estimated?

5. Let’s start with the losses.
Post 2008-GFC, much of the focus has been on credit losses.
This time it is about interest rate risk.
The problem is that banks can hide this risk by claiming assets as held to maturity (HTM) or, to a lesser extent, available for sale (AFS).

6. Under HTM banks don’t have to recognizes losses. Under AFS they do for capital but not for current earnings (I’m simplifying here).
So even if you run a stress test, if the bank is allowed to claim most of its long term assets as HTM, you may not catch the risk.

7. Regulators should not have allowed SVB to claim so much of its assets as HTM.
Suppose they did that. The next question was: what scenario?
The problem is that the stress scenarios of the past years all assumed .. lower interest rates!!

8. why? because these scenarios mimic more or less the 2008 GFC, which is a negative demand shock so:
– GDP falls
– unemployment increases
– spreads and volatility increase
– but safe rates go .. down!
That’s what a demand shock does.

9. Today we have post pandemic supply shocks coupled with strong demand.
So risk is more like the 1970s, including higher interest rates.
Why was the Fed no stressing banks on higher rates in 2021 ands 2022?
Because it only has *one* severely adverse scenario.

10. And it has used that scenario to mimic a 2008 GFC style crisis, not a duration mismatch crisis.
Look at Table 3A here, the 10year rate is 1% under the stress scenario.
federalreserve.gov/publications/2…

11. There is no reason to have only one stress scenario. In this paper with Cecilia Parlatore we study the optimal design of stress scenarios. We find that you don’t need many scenarios, but having more than one is definitely a good idea.
pages.stern.nyu.edu/~tphilipp/pape…

12. For 2023 the Fed has (finally) introduced an exploratory scenario:
“For the first time, this year’s stress test will feature an additional exploratory market shock to the trading books of the largest and most complex banks”

Federal Reserve Board releases hypothetical scenarios for its 2023 bank stress tests The Federal Reserve Board on Thursday released the hypothetical scenarios for its annual stress test, which helps ensure that large banks are able to lend to h https://www.federalreserve.gov/newsevents/pressreleases/bcreg20230209a.htm

13. And, you guessed it, that scenario is the one that we are discussing: “The exploratory market shock is characterized by a recession with inflationary pressures.. Treasury rates increase as short-term rates rise sharply, while longer-term rates increase to a lesser extent.”

14. That’s nice, but a bit late. So bottom line:
A. All banks with >50b assets have to run stress tests. The compliance cost BS has to stop. It’s just a bunch of codes to run.
B. We need at least 2 stress scenarios (3 IMHO) with opposite assumptions on demand/supply.

Vir: Thomas Philipon (via twitter)