As the end of the year approaches, banks around the world are scrambling to update estimates of likely bad debt losses, a crucial number for bank investors and savings watchers.
Calculating these losses still requires some judgment on the part of the banks, but this is meant to be largely a scientific exercise. This year it will rely more on guesswork. Models for estimating loan losses have collapsed during the pandemic, and there is no silver bullet to answer regulators’ calls for banks to do better.
Bank risk management departments knew they were in trouble as soon as the pandemic hit.
Estimates of loan losses rely heavily on models that draw on past experience, and the environment for Covid-19 was completely new. We had never ordered plastic surgeons to close their doors before, risk managers said at the time. Likewise, they said they never had dentists whose cash flow was gone because people were too afraid to visit them.
As the crisis progressed, new complexities in assessing loan losses arose. The stop-start nature of the lockdowns was a problem, but the biggest challenge came from the unprecedented government support being deployed to cushion the global economic fallout.
The result was immediate loan losses well below the tens of billions originally forecast. This has allowed banks to revoke some of the original loan loss charges recorded, but the backing has masked the true financial health of many of their borrowers.
Now the crisis is looming to provide more clarity as the end of the year approaches, as the number of coronavirus cases rises across much of the world.
“We all feel a little more optimistic about Covid but we haven’t seen the end of it yet. . . we haven’t seen the long-term economic scars, ”says the accounting director of a major UK bank, adding that some government support is still ongoing and the outlook for the pandemic is uncertain.
European banks have attempted to correct the uncertainty with “post-model adjustments” (PMAs), which add an additional layer of expected credit losses to the number spat out by their regular analysis. The United States uses different terminology – Q factors – to do essentially the same thing.
At the onset of the pandemic, these adjustments and provisions reduced expected loan losses, as banks offset the likely impact of government support against their models’ darker predictions.
Later, as this state aid flattered corporate and household balance sheets, models spat unrealistic loan losses and LDCs were increasingly called upon to add a layer of reality. At the end of June 2021, those adjustments resulted in a 20% increase in banks ‘total provisions, according to a letter from the Bank of England to banks’ CFOs.
At the end of September, that proportion was even higher, reaching its highest level since the UK introduced new loan loss accounting in 2018, a person familiar with the situation said. They added that across Britain’s big seven banks, LDCs stood at £ 7.3bn, or 24% of their total expected loan losses.
Having such a level of PMA is problematic. Bank models must be approved by regulators, who want consistency across the market. However, LDCs are more ad hoc. “It’s under intense regulatory scrutiny, but it’s really the company’s decision at the end of the day,” says the chief accounting officer. A PwC memo described the European version of the adjustments as “among management’s most difficult subjective judgments.”
The BoE warned in its late September letter that LDCs were only a temporary solution. He wants UK banks to do a lot of work on revamping the model, so that LDCs are not as needed in the future, by recognizing ‘changes in credit risk in a timely manner’ and sharing more data . The European Central Bank has not disclosed the LDC level of its banks, but it is monitoring the adequacy of the loan loss modeling.
The banks themselves say they would like to operate with better models, but it is not an easy task. “When you step back and try to reshape it, there are two very weird recessions in your data pool,” adds the chief accounting officer, referring to the financial crisis and the pandemic. In between, there weren’t a lot of defaults.
Then there are the practical difficulties of reworking their models while the pandemic is still ongoing. “The real problem is trying to figure out what the new normal is,” said the chief accountant.