Why Bounce Back Loan Scheme loss estimates are still likely to be too high

In October 2020, my blog - The tale of how the British Business Bank forced Rishi Sunak to launch the Bounce Back Loans Scheme - questioned estimates by the Department for Business, Energy and Industrial Strategy (BEIS) that expected losses would be in the range 35% to 60% of amounts lent.


BEIS said at the time that these ranges were 'based on historic losses observed in prior programmes'. I suggested that the estimates seemed extraordinarily pessimistic, possibly because it was unclear what prior programmes were being considered.


In March 2021, BEIS estimated that 37% of Bounce Back Loans wouldn't be repaid, including both loans taken out that met the original scheme criteria, and those that did not (either due to fraud or misunderstanding). That was right at the bottom of the previous range but still made for gloomy headlines.


The small print around the March 2021 estimate revealed:

  • The fraud estimate was based on sample data from November 2020, 4 or 5 months out of date at the time. It included all 'possible' frauds, rather than applying any probability factor to cases that had not been confirmed either way.

  • Any misrepresentation of a firm's position, for example a firm overstating its revenue in order to borrow more than it should under the scheme rules (lending was limited to 25% of turnover) was reported as fraud. Every fraud case - no matter what its nature - would result in a 100% loss, it was assumed.

  • For all loans not classified as 'fraudulent', default forecasts were based on a model, that took into account borrower credit reference information (at the height of the pandemic, of course). This model followed IFRS 9 accounting principles, the relatively new international accounting standard that led to banks having to 'assume the worst' for many loans that are not being paid, far more so than was the case with the old accounting rules.

Many UK banks have already significantly reduced their IFRS 9 loss provisions in the last six months or so, as loans recover more strongly than their IFRS 9 models predicted. Even the International Accounting Standards Board, which came up with the IFRS 9 modelling approach, said in March 2020 that users 'may need to adjust their approaches to forecasting and determining when lifetime losses should be recognised to reflect the current environment'.

Then in December 2021, the National Audit Office provided updated information to suggest that around 7% of loans were, at that stage, in arrears. Obviously that means that 93% of loans were either being paid on time or - in many cases - no payments were yet due. Many loans that go into arrears recover in part or full. True, it's difficult to predict the future performance of loans that haven't yet started to be repaid. But the 7% figure still seems a long way from the 37% estimate that remains the official estimate.


On Wednesday, apparently in response to the resignation of Lord Agnew of Oulton, the counter-fraud minister, over the emergency business support programmes, Chancellor Rishi Sunak tweeted that the 'vast majority of people did the right thing' and that the Treasury is not 'writing it off' [the reported amounts lost due to fraud].


No doubt mistakes were made, either in the design of the scheme (see my earlier blog) or by some banks in carrying out even basic checks on applicants. However I wonder if Lord Agnew's decision to resign does justice to the facts that, in extraordinarily difficult circumstances:


  • A radically different scheme to support SMEs was put in place rapidly by policy makers and banks to replace the one that was so clearly failing - almost certainly saving huge numbers of businesses and jobs.

  • The vast majority of small business owners who took out the loans did so entirely appropriately. The idea that SMEs are out to 'cheat' the system does a massive injustice to UK entrepreneurs and could not be further from the reality of how the business community worked tirelessly to support the economy through the pandemic.

  • Even some who either didn't understand the application criteria or deliberately falsified parts of their applications would surely have had, and will continue to have, every intention of repaying the loans. Yet the official estimates assume that every one of these loans will be 100% lost. Theft cases - where the applicant had no intention to ever repay a penny - are shocking and it's unclear how they got through the basic checks that should have been in place, but seem likely to remain at a tiny proportion of total applications.

  • The loan loss estimates have come from a financial model that is based on IFRS 9, the new accounting rules that are widely seen to create overly pessimistic loss estimates.

Recent Posts

See All