For the Investor: Benefits of the “CECL” model and “vintage” disclosures

For the Investor: Benefits of the “CECL” model and “vintage” disclosures

By Hal Schroeder, Board Member, FASB.


“Banks don’t make bad loans in bad times; they make bad loans in good times.”

William J. McDonough, former banker and at the time President and CEO of the Federal Reserve of New York, spoke those words in 1999, well before the 2007–2009 Financial Crisis. Some of you may have been with me in the audience that night to hear his comments.

During the seven years between McDonough’s prophetic words and the financial crisis, publicly available data from the Federal Deposit Insurance Corporation (FDIC) show that loans in the U.S. commercial banking system increased 85 percent, while reserves to cover credit losses increased only 21 percent.

The volume of pre-crisis lending activity is notable because some stakeholders have argued that investors were “caught by surprise” by rising credit risk, and that accounting rules need to change accordingly.

There were many surprises for investors during the financial crisis. But credit risk should not have been one of them. In fact, a central change proposed in the FASB’s Credit Impairment project is intended to better align the accounting rules to more accurately reflect how investors view the economics of lending and its inherent credit risks.

My reasoning behind the need for better alignment is simple. Again using publicly available FDIC data, consider that loan loss reserves of all U.S. commercial banks fell to 1.15 percent (of gross loans) in 2006—just one year ahead of the financial crisis. That marked a low point, seen only twice before in the last seven decades. (To be clear, this decline is more reflective of a major shortcoming of current Generally Accepted Accounting Principles or GAAP, than an intentional under-reserving by banks.)

This common ratio, using numbers reported under GAAP, in my view is often misinterpreted as an indication that credit quality is getting better. Contrast this perception with how the markets valued bank stocks leading up to the financial crisis.

From 2001 to 2005, bank stocks in the Standard & Poor’s 500 typically traded on a price-to-book basis at a roughly one-multiple discount—for example, a market price of two times book value per share for banks, compared to three times for all non-financials in the S&P 500. This valuation gap, or discount, began to notably expand six quarters before the financial crisis, and had doubled by its beginning in 2007. A reasonable explanation for the sharp devaluation is that investors had begun to question GAAP book values of banks. This was because they believed loan loss reserves were understated for the increasing credit risk.

So if investors were able to compensate for the increased credit risk not reflected in GAAP numbers, why is there need for an accounting change?

My response is that investors make educated guesses about the inherent credit losses in loan portfolios. But rather than guessing, it would be far more efficient for management to make a more-informed estimate of those expected losses.

The proposed Credit Impairment standard would enable more timely recognition of credit losses for all loans and debt securities held at amortized costs. This would be accomplished by:

  1. Recording an allowance that represents management’s estimate of all contractual cash flows not expected to be collected; otherwise known as the Current Expected Credit Loss (or CECL) model. That estimate would use more forward-looking information and avoid the inherent bias of the solely “rearview mirror” approach embedded in current GAAP.
  2. Providing investors with management’s best estimate of expected losses at the point when credits are originated, then updating those estimates as more information becomes available.
  3. Improving disclosures so changes in credit risk would be more transparent. Such disclosures provide investors with information needed to better understand management estimates and, when an investor’s views may differ, to develop their own alternative estimates.

Focusing now on the proposed disclosures, I believe one of the most significant improvements is shedding light on how banks determine loan loss reserves.

Since CECL is an “origination” model—that is, expected credit losses are booked when loans are made—investors want to know what the total originations were for each period. So the Board initially proposed a “roll forward” that detailed how loan balances changed from period to period. However, some banks responded that detailing how much money was lent and collected each period would be operationally challenging—given current systems.

In addressing this challenge, the Board developed an alternative, referred to as a “vintage” disclosure, based on existing credit-quality disclosures. Current GAAP requires banks to disclose “credit-quality indicators” for each class of loans. The new requirement would further disaggregate those disclosed amounts by year of origination (or vintage).

In my view, a big win for investors would be the ability to derive, for each class of loans, a roll forward of the loan balances and related loan loss reserve by vintage. I say “derive” because, in an effort to balance costs and benefits, investors may not get everything they would like to have. For example, there are some key data points they may have to estimate, such as pay downs and charge-offs by vintage.

Fortunately there are often good public sources, including securitizations of comparable loans that can be used to fill in data holes. However, filling those holes shifts costs toward investors and may diminish accuracy, thereby having some impact on the usefulness of the disclosures. That said, any derived values should sum up to known amounts disclosed in the financial statements. Those known amounts would provide investors with needed guard rails to ensure any derived amounts are within a reasonable range. While the derived amounts may not pass muster in the current audit environment, they should be sufficiently reliable to assess relevant credit trends.

The bottom line is that investors will be able to estimate key numbers they’ve said are important:

  1. The amount of originations for each period, by class of loans.
  2. The original Day-One estimate of expected credit loss (aka provision), as well as subsequent changes to the original estimate.
  3. A split of the current-period provision between the portion related to current-period originations and that related to estimate changes for loans originated in prior periods.

If a thoughtful investor puts in the effort, he or she will be rewarded with a far better understanding of how well management assesses credit risks and how credit risk is changing over time.

We continue to welcome your thoughts on the Credit Impairment project—especially on the disclosure requirement to disaggregate credit-quality disclosures by vintages. Your feedback will help us finalize the standard by the second half of this year.