Stress testing needs to reverse to move forward

Receive free Banks updates

We’ll send you a myFT Daily Digest email rounding up the latest Banks news every morning.

When Silicon Valley Bank imploded this spring, it sparked recriminations about the Federal Reserve’s stress tests — or the practice of running models to test whether banks have enough capital and liquidity to withstand shocks, such as a recession.

Such tests became mandatory for banks following the 2008 financial crisis. But, like any model, the tests are only as good as their inputs.

And with SVB, they failed. After all, as Patrick Honohan, the former governor of the Central Bank of Ireland, has noted, “none [of the tests] examined higher interest rates” — ie the factor that ultimately killed SVB.

Worse, the Basel regulatory framework only requires banks to protect themselves for a scenario where they lose deposits at a rate of 10 per cent a day, or considerably less. This is because this has historically been the pattern of bank runs.

At SVB, however, the rise of digitised finance enabled its depositors to remove almost a quarter of the bank’s assets in just a few hours, and another half was slated to leave, before the Fed shut it down. The past is not always a good guide to the future — as any rookie financial adviser should know.

So is there any solution to the flaws demonstrated by stress tests? Perhaps, if institutions use so-called “reverse stress tests”.

This week Michael Barr, the Fed’s vice-chair who oversees supervision, took part in a debate with John Williams, New York Fed president, about risk management (I moderated). During this, he suggested that reversing the stress test model might be one tool that could improve supervision.

“Instead of thinking of a stressful scenario and then seeing how it would play out through, say, the balance sheet of a firm, you look at a bank and you say, well, what would it take to really break this institution?” Barr suggested.

Consider this, if you like, akin to “white hat” hacking in the cyber security world. Rather than tracking an institution’s strengths, reverse stress tests ask, as Barr says, “what are the different ways this institution could die?”. It is all about identifying an Achilles heel.

How the Fed — or the banks — might actually adopt this innovation in the future is unclear, since Barr says the debate is still “pretty nascent”. But if the practice flies, it could be a very good thing. Indeed, I consider it one of the smartest ideas to emerge in finance for a long time.

That is partly because reverse stress tests would force financiers to become more imaginative about the future. This matters deeply right now given that markets are being hit by shocks that have not been seen for many decades (if ever), be that high inflation, protectionism, war or rapid digitisation.

The other big merit of reverse stress testing is that this process could also force financiers to think more about qualitative issues, alongside quantitative ones, when it comes to judging risk. This sounds like an entirely obvious thing to do. But before 2008, there was such reverence for models — and the concept of “rational” economic actors — that soft metrics, such as culture, tended to be ignored.

Thankfully, the Fed’s approach started to shift after 2008, not least because the crisis exposed the “flaw” of using an excessively model-based approach, to paraphrase Alan Greenspan, the former Fed chair.

And in recent years the New York Fed, under Williams, has held extensive seminars on banking culture, in an effort to import useful lessons from social scientists, non-financial sectors and jurisdictions such as the UK and Netherlands (which are arguably further advanced in terms of using cultural metrics).

Thus the seminar at which Barr and Williams spoke this week also featured a senior doctor, who explained to financiers that the medical world is trying to replace its traditional “blame” culture (ie one that responds to disasters by punishing individuals) with a focus on systemic failure analysis instead.

A nuclear energy executive then outlined how the managers of nuclear plants start each day with an audit of the previous day’s mistakes, to prevent a stealthy “drift” (or decline) in risk management standards. These are just two examples of practices bankers could emulate.

But notwithstanding this drive to embrace social science, there is still far more that the Fed — and the banks — could do to look at qualitative metrics. Just think, once again, about SVB. The supervisors apparently did not spot ahead of time how the bank’s internal culture, which aped the behaviour of its tech clients in Silicon Valley, had fostered dangerously high levels of optimism and a mania for growth.

Nor did Fed officials pay enough attention to their own internal culture problem. As Barr told me during the seminar, the central bank’s bureaucratic systems “make it difficult for the institution to act quickly with respect to supervision”. In plain English, its culture is too stodgy.

Can this be fixed? Not easily. But conducting reverse stress tests on banks could certainly help. So might another step: doing the same exercise for the Fed itself. Either way, let us all hope that the sheer shock of SVB’s collapse has now given Barr and others the political capital they need to innovate. If so, the dramas of this spring might yet prove a useful warning on the dangers of inertia and mere box-ticking.

gillian.tett@ft.com