Stress testing, home working and the move to €STR

Funds turn to stress-testing in fast-forward and reverse

Buy-side risk survey: Covid-19 is changing the way investors think about stress tests

Moonshots shelved: banks spend on home-working tech

Dealers made success of remote working switch – now they’re investing in its future, and pausing grander ambitions

Non-cleared euro swaps market wrestles with discount rate switch

Buy-siders prepare for valuation change from move to €STR

COMMENTARY: Worse than worst-case

Risk.net this week took a deep dive into the issues of scenario development and stress-testing on the buy side, as a spin-off from our recent investor survey. Frequency and severity, as ever, were the key differentiators between institutions with good and bad stress-testing programmes – and the ECB found something similar on the banking side as well, with many banks in the EU using scenarios that were not severe enough to be effective.

It’s human nature not to want to undergo a test that you fear you may fail. Unfortunately, this caution means that stress test scenarios have a chronic tendency to be too mild. Institutions which stress-tested against the 2008 financial crisis would have been prepared for the market shock that followed worldwide lockdowns in March – though the onset would still have been unexpectedly fast. Those which regarded the 2008 crash as a once-in-a-lifetime event, meanwhile, would have suffered more severe consequences.

There are a couple of interesting conclusions from this. First, it underscores the value of stress-testing even against the wrong scenario. An institution which decided to test specifically against a pandemic scenario would probably have come up with something like the 2003 Sars outbreak – serious, but regional, and without global effects. Relying on that would have left it ill-prepared for March 2020; using the 2008 crisis, which in cause and development was different, would have left them in a better position. Getting away from a focus on the chain of causation – what the information security expert Bruce Schneier called “movie plot threat analysis” – is an advantage of reverse stress-testing as well. Here, firms assume the shock is bad enough to break their business, and work backwards to probe flaws in their practices and processes.

Second, and more worryingly, there have now been two 2008-size global financial crises in quick succession. Whatever your personal priors for the frequency of global crises, they should probably have increased in the last six months or so. It’d be brave to work on the basis that we had two in 12 years but we’re unlikely to have any more in the next half-century.

In 2008, risk managers may have been able to get away with saying “we never thought something like this could happen” and even in 2020 “we never thought something like this could happen again”. But it’s going to be harder to be taken seriously when you’re standing in the wreckage left by the great financial crisis of, say, 2031 (caused by who knows what natural or manmade chain of catastrophes), saying “we never thought something like this could happen again again.”

And, of course, there’s no good reason why the crises of 2008 and 2020 set the theoretical upper limit for how bad a crisis can be. Designers of stress test scenarios should cultivate a wide-ranging pessimism in preparation.

STAT OF THE WEEK

Goldman Sachs breached its trading risk limits on 16 occasions over the first six months of the year. The firm temporarily increased its firm-wide value-at-risk tolerance in Q2 to compensate – the first increase to its VAR limit, and first limit breaches, since 2015 

QUOTE OF THE WEEK

“We expected we would have a liquidity crunch one way or the other. We just did not know how big it would be, when it would be, and what would trigger it” – Nasreen Kasenally, UBS AM