“Bring back burn downs!”
This is a frequent request via phone or email since Silicon Valley Bank and Signature Bank failed in early March. Burn downs are statistical models generated during the 2007-08 financial crisis that recorded credit cycle losses on banks’ loan portfolios and determined what was left over after loans, reserves and capital were “burned down” for credit losses.
As a survivor of the financial crisis era, burn down models were not a pleasant experience for either the bank or the research analyst. Massive valuation deterioration and losses exceeded most banks’ capital and reserves, due to excess leverage from the credit extended on land and other construction properties. Weak earnings meant bank cash flows were little help to the credit recognition process. It was a terrible feedback loop that I prefer to forget. Perhaps forgetting is not realistic, since memory and learning from mistakes should be an analyst’s top skills.
Loan Loss Modeling
When the Covid-19 shutdown first unfolded in March 2020, Janney’s research team developed a new credit risk model to address how many loans could go bad and what losses should be assigned to these possible problem loans. We tried to be thoughtful: Not all loans would go bad. We argued that out of 100 loans, there were 10 to 15 that would have problems; many of those could be addressed in advance and often at small losses. Janney’s 2020 loss expectations were 2% to 3% across-the-cycle loss rates for banks. Fortunately, the Janney model proved fairly accurate, and actual losses were significantly less than we thought.
In 2023, we have a new credit cycle unfolding with three failed banks, a presumed recession starting soon and explosive predictions of commercial real estate loan losses galore. Hotel worries from 2020 have been replaced by office CRE in 2023 — credit mayhem is here! What should banks do to get ahead of the credit risk recognition cycle? How do banks ease investors’ fears of a charge-off surge and reduction in tangible book value per share? Bank management teams should take each issue seriously and address them immediately.
First, bank executives must speak clearly and plainly about their wide open disclosure on pass versus non-pass risk ratings that are standard in quarterly and episodic public filings, such as 8-Ks. The Federal Deposit Insurance Corp. still redacts these disclosures in call reports — public banks have a chance to provide investors with real data.
Given these disclosures, criticized loans are better understood in 2023 than in 2008. The 2.5% to 3% of non-pass rated loans at banks today are a fraction of criticized loans in 2009 and 2010. We encourage companies to disclose their criticized levels from 2009 and prior, which hit highs in the range of 10% to 12%. Investors may appreciate how bad it was then and how fortunate banks are today facing modest credit issues.
Next, we encourage bank managers to explain their long-run experience that loan defaults are limited, which can help combat investor negativity toward problem loans. Begin with statistical proof that supports how most loans are good and will repay as agreed. Then, share how your company mitigates potential problems in advance of a default or rating change to non-pass, including special mention or substandard rating.
Most investors have no idea how often loans encounter issues that are addressed by management long before becoming a charge-off. This leads to more reserves assigned to loans, broader loan categories and ultimately reduced loss exposure.
Finally — and perhaps most importantly — preach and shout about underwriting that features less leverage today compared to past cycles. Today’s loans have more equity, lower loan-to-value ratios and less dirt (little to no raw land) versus the similar loans made 15 years ago in the run up to the financial crisis. It is imperative that banks make explicit comparisons between a 2023 loan and a 2008 loan: how each could be resolved and what loss expectation occurs from each credit cycle. To analogize, investors and analysts tend to be from Missouri: “Show us!”
The bottom line is that credit issues in 2023-24 do exist. Banks must take ownership of current credit problems they may encounter and determine how they can be swiftly resolved and at what cost. Loss content should be far lower; providing specific data and examples will be critical to repairing the lost confidence from the recent bank failures.