Proposals for a new loan-loss accounting standard, published on November 5 by the International Accounting Standards Board (IASB), are "potentially too restrictive", says US comptroller of the currency John Dugan.
Speaking to Risk, Dugan applauded accounting standard-setters for moving in the right direction, but was concerned by elements of the IASB's proposals and with the direction signalled by its US counterpart, the Financial Accounting Standards Board (FASB).
Under the current incurred loss model, banks can only set aside money as reserves once evidence of an imminent loss has emerged. Politicians and regulators have attacked these rules for delaying recognition of losses and helping to magnify the scale of the crisis. A working group on loss reserves, co-chaired by Dugan, contributed to a Financial Stability Board report on pro-cyclicality, which was prepared for leaders of the Group of 20 (G-20) countries prior to a summit in London in April. The report called for new accounting standards to be more forward-looking and the G-20 summit endorsed those calls.
The IASB's proposals address those criticisms by shifting to an expected loss approach, in which banks estimate how much money they will lose on a loan and use that to arrive at a credit-adjusted interest rate. In other words, as revenues roll in, some proportion will gradually build up as a reserve (see pages 31-33).
The speed at which reserves will build up depends largely on how banks calculate expected loss, and Dugan says the proposal may not do enough to tackle pro-cyclicality. "If expected loss is interpreted in a way that constrains it so you don't get to look forward much more than you do under the incurred loss model, we will not have made much progress - and there are words used in the proposal that could be construed as overly constraining," he says.
"If expected loss is interpreted in a way that constrains it so you don't get to look forward then we will not have made much progress"
Under the proposal, banks will have to calculate expected loss over the life of the loan, but the standard-setter also says the loss estimate must be tied directly to each loan - banks cannot build in an assumption of economic stress. In the language of credit modelling, it is a distinction between point-in-time and through-the-cycle measures. "This is the issue people are asking about: what do you mean when you say expected loss? I think this is moving in the right direction. We just hope the details are as promising as the conceptual change appears to be," says Dugan.
Spain's dynamic provisioning system had been touted by some as an alternative to both incurred and expected loss - with the OCC said by some sources to be particularly keen. The system works by separating loans into several homogenous categories then using a formula based on historical performance to inform banks how much money to hold in a counter-cyclical loss reserve. In theory, the reserves build up in good times and are depleted when losses hit. However, standard-setters argued the approach didn't reflect the performance of individual loans. Dugan recognises those objections, but doesn't appear to have stopped hoping for some kind of compromise.
"We think expected loss is a better impairment model than incurred loss. We also think the Spanish model is kind of an expected loss model - but is not tailored to the particular circumstances of an individual firm in a way accountants and investors would like," he says.
Dugan also criticises elements of the plans put forward by the FASB. Although the standard-setter won't publish proposals for some months, it has also announced an intention to make loss recognition and reserving more forward looking. "That's the good news," he says. He is less enthused, however, about plans to simultaneously change the valuation framework for loans, by having them marked-to-market rather than carried at amortised cost. "I think that is not a good idea. One of the lessons from the crisis is that illiquid instruments like traditional loans get valued very badly when markets seize up - there is tremendous uncertainty in forecasts. We would have thought the prudent way to approach this from a valuation and accuracy perspective was to make sure we kept the amortised cost model."
The IASB's eight-month comment period should be lively, and Dugan says the standards remain malleable. "It is not too late for change"