We believe that CECL will inadvertently force some community banks to make suboptimal lending decisions and accelerate community bank consolidation, while, at the same time, allow others to differentiate their business models. In this article, we consider some of the secondary effects of CECL on community bank CRE lending decisions and specifically the average life of newly originated loans.
After the financial crises, many participants believed that the overstatement of asset values and delayed recognition of credit losses was a weakness in GAAP. CECL requires that allowance for losses be estimated based on cash flows not expected to be received. This estimate needs to be quantified and can be based on past events, historical loss experience, current and reasonable forecasts, borrower credit strength, and forecast of the economic cycle.
CECL requires that losses are measured for the life of loan (LOL) and the rule will speed up the recognition of losses and will require banks to set aside reserves for lifetime expected losses on the day of loan origination. However, it is not a given that the new loss accounting methodology will increase community bank reserves. Analytical and research firm Trepp indicates that using commercial real estate data and current, expected credit loss modeling, CECL allowance for CRE loans at banks may be as low as 0.42%. This is a level lower than current allowance at community banks. For banks over $1Bn in assets, the current ALLL level for CRE loans is 1.04%. However, the biggest impact of CECL, in our opinion, is the possibility that banks shorten loan commitments to lower expected loss modeling under CECL, which must be measured for LOL.
Many community banks will not have the data to properly forecast CECL for various LOL scenarios. The result is that many banks will apply a relative linear allowance for remaining term of a loan. This is where CECL can skew decision making at community banks’ potentially resulting in a negative impact on performance. While longer loans have a higher expected loss rate, that relationship is not linear, and banks must measure remaining term and return on equity (ROE) tradeoffs carefully. As can be seen from a typical community bank below, the slope of the cumulative POD line is not the same.
In fact, breaking it down to the marginal probability of default, you can see where default probability starts low the first year and then builds, usually peaking between year’s three and seven (median of five years) depending on the profile of the credit.
One problem with CECL stems from the non-economic underpinnings where credit cost of the loan is recognized upfront, but revenue of the loan is recognized over the life of the loan. This may create an incentive to shorten commitment to lower loss expectations - however, this would be an incorrect conclusion for the following reasons:
First, the relationship between the remaining term and expected loss rates is not linear. Expected loss goes up only marginally as the term of the loan increases. The table below shows Trepp’s modeling of expected loss on CRE loans under two remaining term assumptions.
This research shows that doubling the term of the loan increases CECL loss expectation by only 49%. An economic model that incorporates the slightly higher expected losses, but substantially higher lifetime revenue would show higher ROE on longer loan commitments. Further, the most profitable period of an average CRE loan is approximately past five years post-inception when amortization accelerates principal reduction, the expected loss is lower due to property/business appreciation, and revenue continues to accrue for the lender. Unfortunately, banks that measure accounting income and emphasize immediate period revenue would prefer the shorter loan using GAAP, and CECL will only exacerbate this problem.
Second, longer commitments lead to more relationship banking. Longer commitments increase expected life of a loan which is highly correlated to a longer cumulative lifetime of the customer. Longer customer life means less competition, more payment history, more cross-sell opportunities and higher lifetime value of revenue. However, banks that cannot measure the benefits and ROE improvement from relationship deals will favor shorter loans. In other words, CECL will skew some bank’s preference for shorter loans because of the lower expected loss resulting in higher immediate GAAP income at the expensive of lower LOL economic returns.
Third, shorter loans tend to be self-selecting. Borrowers that cannot obtain longer commitments from various lending sources are less desirable credits. In the long-run, shorter loans often have higher credit risk and lower risk-adjusted economic return.
Fourth, in isolation, lenders may argue that they can roll over their short commitments on the same loan taking lower CECL allowance on each commitment rather than commit to longer remaining term. This would lead to higher GAAP income.
Unfortunately, borrowers believe, and many banks support the notion that a longer-term loan is always riskier. As such, many banks and borrowers price shorter loans and lower margins. Ironically, this usually exacerbates the risk as many banks unknowingly not only receive lower pricing but create a liquidity event forcing the borrower to refinance. By refinancing, the borrower often incurs additional transaction costs while also being tempted to take cash out and increase leverage.
In an improving and stable economy, this additional risk is likely overwhelmed by the improving cash flow of the company or project. However, with even just a slight downturn, the risk is minimized, expected losses increase, and lenders find that they cannot find an exit strategy for their credit. Ultimately, unless the borrower’s situation demands shorter debt, there is little benefit to lenders to provide borrowers with shorter loans and the additional refinancing triggers. Again, CECL may only serve to obfuscate actual economic returns with a mismatch between expected losses and revenue.
We are all for better quantifying risk and believe CECL is a huge improvement in risk management. However, bankers should be aware of the potential economic distortion for community banks where the cost of expected losses are recognized upfront and revenue is recognized as received. This distortion may lead to shorter loan commitments unless community banks adjust for the distortion by comparing cost and revenue for the same loan over the entire expected life of that credit relationship. When deciding on a CECL model, it is helpful to implement one with the sophistication to handle non-linear probabilities of default.
Submitted by Chris Nichols on October 24, 2018