Behavioural finance, alt data and risk-free rate problems

The week on Risk.net, June 27–July 3, 2020

7 days montage 030720

The scientists probing the human mind for an investing edge

Recent advances in behavioural finance could give rise to new quant models and strategies

Firms hone use of new data to pick Covid-19’s winners

Managers are using alternative data to assess the pandemic’s effect on individual stocks

Risk-free rates may fail liquidity test for hedge accounting

Experts fear trades referencing SOFR and €STR will not be eligible for hedging relief


COMMENTARY: Failing better

The Covid-19 pandemic has shown how bad a worldwide operational risk crisis can get. Underestimating the danger of such a crisis is a black mark for op risk managers. The topic didn’t even make it into this year’s Top 10 Op Risks, with almost no respondents to our survey mentioning it even though the outbreak had already broken out of China, and despite multiple near misses or less serious incidents (Sars, Mers, H5N1, Ebola) in previous years.

Why not? As Risk.net reports this week, one problem was a refusal to consider that a pandemic of this scale could ever happen. The organisations that modelled pandemics considered “regional, not global” outbreaks – damaging, but limited in scale and extent, along the lines of the 2003 Sars epidemic in East Asia. It’s a known problem of scenario planning. A desire for psychological comfort means that, when asked to consider worst-case scenarios, people tend instead to describe the worst scenario that they can reasonably expect to survive – which is not the same thing.

Also this week, Risk.net looked at how understanding this sort of psychological shortcoming could provide an investing edge – in part through incorporation in agent-based models – to produce a less rational and more realistic representation of human investors. Combined with the use of natural language processing and big data considering the chances of success for possible cures and vaccines for Covid-19, the prospect is one of radical improvements in risk management, forecasting, investing and many other areas.

The improvement for investors might only be temporary, though. What’s the advantage in having a model of the market that faithfully reproduces the behavioural quirks and idiocies of human investors, just as those quirks are being ruthlessly ironed out of the actual market by new trading software intended to outperform the irrational humans it replaces?

But that’s a problem for investors – for risk managers, the advantages could be less mixed. If software can help them look past the inherent limits of their imaginations and produce better and more realistic (though less comfortable) scenarios, the benefits will be widespread – and permanent.


STAT OF THE WEEK

A wild first quarter caused the number and size of margin shortfalls at Eurex to shoot up. An equity derivatives account had an uncovered exposure of €510 million ($573 million) at one point in the quarter, and a fixed income account a whopping €665 million deficit.

 

QUOTE OF THE WEEK

“The ideal scenario would have been to build an op resilience framework and then … a chance to test the framework. But no-one’s finished with the framework, so we got locked into crisis management” – Sam Lee, Sumitomo Mitsui

  • LinkedIn  
  • Save this article
  • Print this page  

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact [email protected] or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact [email protected] to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: