For years, bank loan loss reserve processes under generally accepted accounting principles have been fraught with subjectivity despite attempts under the financial accounting standards FAS 5 and FAS 114 to provide clear guidance to banks.
Recently, Comptroller of the Currency Thomas Curry endorsed a Financial Accounting Standards Board proposal to move away from an incurred-loss approach for estimating loan loss reserves to one utilizing expected losses. His endorsement further heightens bankers' anxieties that their reserve process will need to radically change while significantly raising reserves in the process.
Moving to an expected loss framework is a step in the right direction – away from the incurred loss method. But make no mistake: It comes with a great deal of subjectivity, model risk and confusion for investors trying to draw comparisons of loss reserves across institutions.
A major criticism of the incurred loss method was inherent procyclicality introduced in the reserving process that precluded banks from raising reserves during good times in anticipation of higher losses over time, while having to scramble during the crisis and raise reserve levels to catch up with mounting losses. The incurred loss approach ties the hands of those charged with setting loan loss reserves (an area I spent much focus on during my tenure at several banks) since a loss may only be recognized as a loss if it is "probable," which typically requires a high degree of confidence in that outcome.
By contrast, an expected loss method allows banks to directly take into account factors that drive default and loss over the life of the loan, including the introduction of loss forecasts which are subject to volatility, particularly if estimates are being derived over the life of the loan.
In theory the expected loss outcome will tend to raise the allowance for loan and lease losses as new loans are originated compared with the incurred loss method. Over the cycle this would dampen the extremes in loan loss reserving.
However, without more specificity in how expected losses are to be calculated, the plan leaves significant discretion to management to establish assumptions on critical drivers of loss events and exposes the industry to significant model risk, as an overhaul in the methodology is required if the FASB proposal is implemented.
On the issue of divergent model assumptions, take, for example, two banks in the same market originating similar mortgage products. One bank could make the argument that during the next few years, home price appreciation, which can stem defaults, will decline from an average of 5% per year to half that rate, based on the bank's internal analysis. Further, this bank estimates that every 1% decline in home prices leads to a 0.1% higher default rate on its loan portfolio, all else equal. The other bank assumes home price appreciation rates of 5% based on its model's results, and adds 0.05% to its default rate for every 1% decline in house prices.
Which assumption and which model is right? Probably neither. But my point is that two banks are making very different assumptions on similar loans that will have very different reserves. The expected-loss approach thus introduces the possibility that model results fall short of providing investors with clarity and accuracy around loss reserves.
In addition, the OCC will continue to have its hands full in understanding the impact of thousands of different loss reserve methodologies and assumptions across the national banking system. It is impossible to draw any consistency across reserve levels for banks today given the diversity of reserve methodologies and assumptions deployed in these calculations.
The Federal Reserve's bank stress testing program (the Comprehensive Capital Analysis and Review), although it comes with significant regulatory burden, might provide some insights into how greater consistency could be obtained in developing loan loss reserve estimates across the industry. CCAR outlines in great detail the key assumptions, factors and methodology that are to be used in a stress test. An expected loss estimate also relies on assumptions, inputs and models. So if maintaining consistency and alignment is important for estimating losses under a severely stressed scenario, why wouldn't it be important for losses in a more probable environment? Investors should welcome a move toward CCAR-like standardization for reserving, as it would permit more direct comparison across institutions of expected losses in bank portfolios.
And there's nothing like peer pressure to induce the right behavior. Currently, conducting meaningful peer benchmarking comparisons of reserve levels and ratios is more art than science. For example, when a borrower hits hard times, some banks take longer than others to tally the loss. While achieving an accurate portrayal of expected credit losses in bank portfolios is crucial to understanding the bank's risk profile, it is equally important that policies be designed to promote consistency in the application of methods and assumptions across institutions as much as is reasonably possible.
In the meantime, banks large and small will need to get prepared to meet the challenges of changing their reserve methodologies and processes if the FASB proposal is implemented. Regardless, such a change will not eliminate the alchemy of the loan loss reserve process.
Clifford Rossi is the Professor-of-the-Practice at the Robert H. Smith School of Business at the University of Maryland.