Rupak Ghose is the chief operating officer of Galytix, and a former financials research analyst at Credit Suisse.
During WWII, aeroplanes that had been in combat were assessed for where they should be reinforced. The thinking was to look for bullet holes. But the mathematician Abraham Wald realised that planes needed to be reinforced where there were none — those were the spots most likely to be taking lethal blows when planes were not returning.
The work of Wald has proved foundational for thinking around survivorship bias in areas like asset management. But it could have relevance to bank risk managers as they navigate how to assess dangers of future disasters.
There are very few banks that don’t return from battle like war planes. After all this is a highly regulated industry which is “too big to fail”. But the mountain of risk management data and processes that banks use is very reactionary.
We saw this with the reaction to the financial crisis. We saw this after the Libor and FX scandals. And more recently with prime brokerage following the collapse of Bill Hwang’s Archegos. But why spend billions of dollars and have armies of risk managers investigate the bullet holes, instead of what else could be out there?
For all its flaws, the stock market casino has a long history of investors pricing in the “known unknowns” and recalibrating valuations. Short sellers are adept at looking not just for deterioration of current metrics but non-linear factors that could bring down the plane — or in this case cause significant losses to banks.
By contrast, bank risk managers are used to relying on historical precedents. Artificial intelligence and cloud-based analytics are all the rage in banks, but don’t really solve the problem of reliance on rigid, model risk frameworks that rely on probabilities based on back-testing.
Risks can emerge from unlikely spots. Pandemics were not top of bank risk managers lists three years ago. Nor was the impact of a Russia invasion of Ukraine a year ago. The impact of inflation on creditworthiness is difficult to predict with accuracy. Which borrowers have the pricing power to pass on rising costs to customers and which do not is one known unknown. The impact of the recent Covid-19 outbreak in China is another.
Returning to the lessons from Wald, risk managers need to focus not just on where they can see existing risks but a wider field — including unknown unknowns.
A good example for banks is fraud. Banks have focused on where fraud has already caused bullet holes, in terms of regulatory fines and sanctions. But traditionally fraud has been left out of credit risk analysis of borrowers. How many bank risk managers were held accountable for lending to Wirecard and NMC Health? With a recession heading our way bank risk managers better be ready to spot the next Enron, WorldCom and Adelphi.
The carnage in the technology and crypto sectors shows that there are no prizes for reacting late. Credit rating agencies continue to be slow to catch corporate stress. They even rated Wirecard investment grade eight months after an FT investigation unearthed detailed evidence of fraud — only downgrading the company a few days before the bankruptcy. Other high-profile bankruptcies follow a similar pattern.
The ECB has warned that banks need to be ready for forward-looking indicators of credit deterioration, such as corporate governance, IT technical debt and climate risks.
One way for banks to be prepared for those dangers — where they don’t have a trail of bullet holes to analyse — is to use synthetic data. This has been advocated by regulators. But more data alone is not a panacea. Much of it remains siloed, labelled inaccurately and processed in batches rather than automatically.
To be honest, bank risk management jobs are a nightmare, a classic damned-by-any-choice situation. Either you are resented as a killjoy break on business, or as a blinkered supine apparatchik who makes a convenient scapegoat if something goes horribly wrong.
But bank risk managers still need to overhaul their approach to be more dynamic and forward-looking, before we head into the next inevitable financial, economic or political shitstorm. Platitudes and PR will only go so far.