Earlier this month the Federal banking agencies issued a draft rule designed to incorporate the impact of a change of accounting – the new GAAP accounting standard governing the allowance for loan losses – into the regulatory capital framework.1 The “current expected credit loss” methodology, or “CECL” has potentially significant economic consequences as it effectively raises bank CET1 capital requirements, yet the proposal is incomplete and appears to have overlooked several important issues. This blog post highlights three such considerations that should be analyzed by the agencies before moving ahead with the proposal, namely: (i) the procyclicality of CECL; (ii) the impact of CECL on stress testing and (iii) the size of the CECL “day one” impact for regulatory capital purposes.
We have observed previously that CECL represents the most significant rewrite of U.S. GAAP in the past 40 years. CECL is a sea change because U.S. GAAP currently requires banks to provide a reserve for loan losses when it is probable that a loss has been incurred and when the loss is reasonably estimable. In contrast, CECL requires banks to establish reserves immediately upon origination of a loan in an amount sufficient to offset all future expected losses arising over the entire life of the loan.
Absent some form of regulatory capital relief, the banking agencies have conceded that those higher loan loss reserves will reduce banks’ CET1 capital, even as risk stays the same. This was a consequence that FASB did not consider as part of its accounting standard; thus, any such impact would be the consequence of how the agencies choose to implement CECL.
CECL forces banks to recognize expected future losses immediately but does not allow them to recognize immediately the higher expected future interest earnings banks receive as compensation for risk. Banks are thus forced to recognize the risk but not the return, imparting a strong bias against normal risk-taking. The inevitable result will be a decrease in lending and other bank intermediation.
Worse still, CECL will magnify this effect during an economic downturn, as projected losses over the life of the loan will increase. Thus, for a given loan, its capital requirement will rise as the economy worsens, particularly for longer tenor loans. This could significantly reduce the availability of credit for mortgages, student loans, term small business lending and loans to non-prime consumer borrowers. This is a purely procyclical effect, utterly contrary to the agencies’ repeatedly stated goal of capital allowing banks to continue lending under stress. This effect is not as marked under current GAAP, as banks do not make loans where they foresee a loss as probable and estimable; thus, the capital cost of lending does not rise as significantly.
CECL was not intended by its supporters in the accounting community to operate in such a highly procyclical manner. Rather, they expected that CECL would force banks to record higher provisions in advance of a crisis (or decrease their reserves in advance of a boom cycle). Unfortunately, however, banks (and regulators) do not have perfect foresight – that is, the ability to forecast a recession and the subsequent recovery in advance of its occurrence and adjust their provisioning accordingly. Rather, banks and other market participants react once the change in the economy has occurred and is well underway. So how do the regulatory agencies propose to deal with the procyclical impacts of CECL because perfect foresight doesn’t exist? Quite simply, they don’t.
The CECL proposal also represents a large and curious step backwards for the banking agencies in their efforts to integrate the capital and stress testing frameworks. The CECL proposal was released in draft only two days after the agencies released another major regulatory capital proposal to integrate the existing Basel III capital framework and CCAR (the stress testing framework for U.S. banks above $50B in assets) by means of a new stress capital buffer (SCB). While the CECL proposal requests comments on how CECL may relate to stress testing, and the Fed has held two meetings with the industry on the proper approach to incorporate CECL into CCAR, both the CECL and the SCB proposals otherwise provide no indication of the agencies’ views in this area. Since the supervisory stress credit costs are a key component of the SCB and CECL reserves will be included in such stress credit costs beginning as early as 2020, banks are left in total darkness as to how the Fed will incorporate CECL into CCAR stress capital projections. Critical issues for consideration by the Fed (and the other agencies for purposes of their DFAST stress tests) include (i) the manner in which theoretical lifetime CECL front-loaded reserves should be included in the stress tests; (ii) how to determine the return to normalized long-run loss rates beyond the stress test’s nine-quarter planning horizon, and (iii) the appropriate life-of-loan loss rate for any new lending during the nine-quarter stress horizon, including whether banks are permitted to factor in improvements in loan underwriting during such period. Notwithstanding the total lack of any suggested approaches from the regulatory agencies, banks are nonetheless required to comment by June 25 on the SCB proposal (July 13 for the CECL proposal). While the banks may indeed learn how the Fed intends to include CECL in the stress test when the SCB rule is finalized, that would be inconsistent with the APA’s requirements for notice and comment, as banks would not have had an opportunity to comment on the approach chosen by the Fed.
The proposal also provides an optional three-year phase-in for the regulatory capital impact of CECL under U.S. GAAP beginning in 2020. The agencies readily concede they don’t know the quantum of this reduction in bank regulatory capital levels since they have neither conducted a quantitative impact study nor undertaken any other empirical work under any assumed economic conditions (since no can project the condition of the economy in 2020), proffering only that the impact “could be significant”.2 The banking agencies rejected a longer transition period beyond 2023 since any longer transition period “heightens the risk of capital increases coinciding with a potential future downturn in the business cycle.” 3
Finally, the proposal also claims that banks will have had four years by 2020 to prepare for CECL, since the FASB finalized CECL in 2016, and when coupled with the proposed three-year phase-in period it’s really a seven-year period of advance notice to the banking industry, which should be “a sufficient amount of time”.4 Unfortunately, this is disingenuous as many fundamental accounting questions remain unanswered, and must be settled by the FASB before banks and others can finalize models and subject them to their required model validation procedures to determine the impact of CECL upon implementation in 2020 and on an ongoing basis. These critical accounting issues include the ability to recognize recoveries on loans such as credit card receivables and the treatment of nonaccrual interest receivables. As a result, this purported “seven-year notice period” is illusory, and in all events does nothing to mitigate the ongoing procyclical nature of CECL: for the reasons stated above, post phase-in CECL reserves will be larger than current loan loss reserves and are likely to increase during a stress period, thereby exacerbating the downturn. And of course, the phase-in provides no relief for the CCAR impact of CECL.
Disclaimer: The views expressed in this post are those of the author(s) and do not necessarily reflect the position of The Clearing House or its membership, and are not intended to be, and should not be construed as, legal advice of any kind.
2 Remarks of Chairman Gruenberg, Federal Deposit Insurance Corporation, Transcript of Meeting from April 17, 2018, http://fdic.windrosemedia.com/.
3 CECL Proposal at 54.
4 Id. at 27.