In two recent memos, I took strong exception to the model-driven, aggregated-data based assessments that pass for cost-benefit analyses in all too many U.S. banking rules. Now comes the inter-agency decision to finalize a three-year transition period before recognition of the capital impact of current expected credit loss (CECL) accounting. Dispensing with even a fig leaf of cost-benefit analysis, the rule dismisses concerns about CECL’s effect by smacking banks for complaining only now about capital impact after years in which they should have identified any problems with FASB’s standard. In short, if banks don’t call a foul early in the game, regulators win. To be sure, the agencies are right – banks have a long history of taking rules seriously only after regulators go to a lot of trouble crafting a policy they then come to love. But that banks are tardy doesn’t necessarily make them wrong. They might be with regard to CECL, but the lack of analytical rigor in the rule ensures that we’ll now only find out the hard way.
Let me make it clear at the start that I am no fan of incurred-loss reserving, having said so when FASB first proposed CECL. Incurred-loss reserving was designed to prevent the “cookie-jar” accounting epitomized at Freddie Mac and summarily sanctioned by the SEC. That earnings management is bad does not, though, make incurred-loss provisioning good.
As the great financial crisis expensively proved, loan-loss reserves needed upon loss can be loan-loss reserves well beyond a stressed company’s capital capacity. CECL is thus an important discipline on earnings incentives that would otherwise incline companies to forego reserving until it might be too late. Banks can be slapped around by examiners for insufficient reserves; nonbank lenders on which the market increasingly depends have no such disciplinary authority and thus should be subject to transparent, forward-looking reserving along FASB’s new lines.
However, not only are banks subject to provisioning discipline, but they also come under complex, costly capital rules. Nonbanks don’t. Thus, while CECL will cost nonbanks some earnings and will surely create a useful risk buffer, banks are already required to establish capital cushions for unexpected loss. The key question for CECL at banks thus is not whether CECL makes sense – it does – but how it integrates with regulatory-capital requirements to ensure that an array of policy objectives are not unnecessarily undermined.
First, CECL is for expected loss, but the distinction between the expected and unexpected is at best ambiguous, especially given the extent to which risk-based capital standards establish probability of default and loss given default based on established historical precedent. Current capital rules fudge this a bit by allowing limited amounts of loss provisions to count towards Tier 2 capital, but all the new rule does is describe when and how reserves count and then promise to think some more about this complex interaction should it at some time to come turn out to be a problem.
Further, risk-based capital raises CECL complexities not only because it may well double-count expected loss, but also because by doing so it’s procyclical. Regulators know this well – that’s why stress tests were instituted after the crisis – and Fed literature demonstrates it quite convincingly. While front-loading risk buffers makes banks safer, it also cuts deeply into what’s left of bank profitability on higher-risk, longer-term exposures. How much pain results in how much safety given the interplay between CECL and capital could be discerned from a marginal, qualitative cost-benefit analysis, but none is to be found in the CECL rulemaking process.
Third, bank regulators continue to expect banks just to buck up in the face of costly rules, but banks instead realign their strategies to head to more profitable lines of business. If history proves precedent – and it’s likely to do so – then CECL in concert with risk-based capital rules will, stress-testing notwithstanding, make it far less likely that banks will make the growth-generating loans urgently needed in slow- or no-growth macroeconomic conditions. Indeed, banks may well simply abandon equality-enhancing loans such as long-term, low-cost student loans, low-down payment mortgages, and small-business loans. If any of these loans are made, then they’ll likely only be for sale into a secondary market, significantly increasing market dependence on the GSEs and Ginnie Mae in mortgage finance and crippling other sectors without government-backed securitization channels.
Look as hard as you like at the final rule, you’ll find no mention of procyclicality or the transformation of key consumer-finance sectors into securitized businesses or none at all. The final rule is instead almost entirely procedural and descriptive. Where comment letters posed critical questions e.g., regarding the treatment of reserves in regulatory capital or stress tests – substance is put off for another day.
What’s even worse is that nowhere do the agencies say why they think it necessary to do nothing more than ease banks into CECL’s pain. Is the rationale for doing so the importance of ensuring alignment between U.S. and global rules? Fear that U.S. latitude would lead other nations to back out of their less-onerous version of CECL? Faith in the importance of aligning regulatory-capital standards with accounting dictates? The agencies decision to stand firm on CECL could well be right, but they sure don’t make it easy to agree with them that it is.