The current expected credit losses accounting standard looks to be in trouble. The mood during a recent meeting of the House Subcommittee on Financial Institutions and Consumer Credit seemed to reflect a bipartisan vote of no-confidence in the new accounting rules.
CECL was implemented primarily to force banks to maintain countercyclical reserves. All thorough analyses of the effect of the new rules have shown, to differing degrees, that allowances will continue to be procyclical after CECL comes into force during 2020.
It is against this backdrop that Rep. Carolyn Maloney, D-N.Y., asked the most pertinent question of the hearing: Is it possible to design a loan loss accounting standard that will generate truly countercyclical reserves?
In short, it is.
There are a number of government and market-driven functions in the economy that naturally act in a countercyclical manner. During a boom, absent a change in policy, tax collections rise as companies record higher profits, sell more goods and services and increase payrolls. These forces then reverse themselves in a recession, and tax receipts decline. Countercyclicality results because the shift toward a higher deficit in recession acts to cushion the economy and thus lessens the depth of the downturn.
This type of phenomenon is known as an “automatic stabilizer.” Such features are desirable because they smooth the performance of the economy without requiring any human intervention. The CECL rules, in contrast, require bankers to project the trajectory of the economy and then infer its impact on lifetime loan losses. Historical analysis shows that even with perfect foresight regarding the economic outlook, resultant reserves would only have been modestly impacted.
For true countercyclicality, proportional allowances need to be high when lending growth is high and low when growth is low or falling. So why not cut to the chase and make reserves a function of observed industry loan growth rates?
Such an approach could be applied to incurred loss allowances, though not ideally so. Suppose that for a commercial loan portfolio, a bank is holding $10 million in reserve, calculated using the existing methodology. Based on Federal Deposit Insurance Corp. data, we can discern that the long run annual growth rate of such assets is about 4.4% and that the current (Q3) growth rate is slightly north of this at 6.3%. The series achieved a boomtime average of 11.4% between early 2005 and mid-2007 and suffered a 7.7% annual recession decline from late 2008 to the end of 2010.
Under the proposed regime, the bank’s $10 million reserves would now be marked slightly higher — let’s assume around $11 million — given that current industry growth is only slightly elevated. Had the same set of loans been on the books of this bank in 2006, however, reserves would have been much higher. Regulators would be able to control the parameters of the formula but could easily have required that the bank hold $30 million at the time instead of the hypothetical $10 million baseline. Similarly, falling industry balances during 2009 would precipitate a decline in reserves for this set of loans, possibly to only $4-5 million. This would release funds to be used to generate new business for the bank and thus spark life in the struggling economy.
This change is simple, easily implemented and the mechanism would act as an automatic stabilizer. If particular lending products were growing very quickly, proportional reserves would increase to offset the higher level of risk developing in the market.
This type of approach could create some anomalies. Safe banks, for example, would be asked to increase their reserves during a boom even if the makeup of their portfolio was not changing. This phenomenon is true under CECL as well, though in a far less transparent manner.
The other point to note is that the proposed methodology provides no new information to investors. Under the current system, someone could easily look up the relevant growth rates from public data sources and make precisely the same adjustments to existing published allowances.
To rectify this, some form of loss modeling is required. While CECL relies on point-in-time modeling of credit losses, incorporating a forward-looking view of the economy, the new method could be based on simpler, through-the-cycle estimates of credit loss. These estimates form the basis of most credit scores that are routinely used by banks in their assessment of loan applications.
The loss estimates generated would then be factored by industry growth rates to ensure the countercyclicality of resultant allowances.
The Financial Accounting Standards Board set out to fix a specific problem — the procyclicality of loss allowances — and then proceeded to propose a complex solution that loses sight of the original aim.
If CECL is to be rethought with this aim in mind, policymakers should learn from successful countercyclical features in the economy and design the new loan loss accounting system accordingly.This post was originally published here