Lehman Brothers’ (LEH) annual financial services conference took place this week in New York; all ears were tuned in to what presenting banks have to say about credit quality.
And why not? Credit costs will be the major driver of industry earnings at this point in the credit cycle. Unfortunately, on the available evidence, many bank investors simply don’t understand the industry’s credit-management practices and related accounting. Which means that a lot of the sell-side’s credit-related takeaways from the conference figure to be roughly 180 degrees wrong. In particular, there will be plenty of commentary in coming days about “loan loss reserve adequacy” that will be misguided.
This happens every cycle. Back in 1989 and 1990, I wrote a number of reports with my then-partner, Frank DeSantis, that tried to explain loan quality-related accounting. From what I’ve been hearing lately, investors are as misinformed now as they were back then. Here, then, is a quick primer.
First off, the most glaring and misleading mistakes investors make have to do with the way they try to judge the adequacy of an institution’s loan loss reserve. Too often, they put bald sets of numbers from different banks side by side, then conclude that one bank is “underreserved” compared to another. This in turn leads analysts to overestimate future loan loss provisions, underestimate future earnings, and overestimate future capital needs. They make a hash of things, in other words.
To set the record straight, let’s start with some basics. Nonperforming assets typically fall into three categories: nonaccrual loans, restructured loans, and other real estate owned [OREO]. A banks puts a loan on nonaccrual status (and stops accruing interest on it) once it becomes 90 days past due as to principal or interest, unless the loan is both well-secured and in the process of collection. However, loans can be placed on nonaccrual earlier should management believe principal and interest will not likely be repaid. Restructured loans are loans whose terms have been changed such that they are at below-market rates for the risk taken. And OREO is the current estimated market value of collateral the bank has acquired through repossession.
OK? For an illustration, the table below shows the mix of nonperforming assets at one bank, Zions Bancorp (ZION), as of June 30:
The Zions mix is typical. As you can see, restructured loans make up a very small percentage of nonperforming assets. When I quiz a bank’s management about nonperformers, I like to ask how much total cash was received on nonaccrual and restructured loans in the prior quarter. The answer provides some indication of the relative health of these problem assets. I also try to find out the original balances of nonaccrual loans. This can reveal two pieces of information: how aggressive the bank has been in writing down the problem loans, and the severity of its underwriting mistakes.
Now let’s move to chargeoffs. Recall that a defaulted loan is charged off against the bank’s loan loss reserve, which in turn is built via the addition of loan loss provisions each quarter. (The loan loss provision is the expense item, not the chargeoff.) Consumer loans typically get charged off automatically once they become 120 days or 180 days past due, depending on the type of loans and the bank’s policies.
Nonaccrual commercial loans, by contrast, aren’t as homogenous as consumer loans are, and are accounted for differently. If the loans are large enough, they fall under an accounting rule called FAS 114. (The size of loan that qualifies for FAS 114 varies by institution. At Zions, the cutoff for is $500,000, while at Wells Fargo, it’s above $3 million.)
What’s FAS 114? It requires banks to determine the amount of impairment via one of three methods:
- The present value of expected cash flows,
- The observable fair value of the loans, or
- The fair value of the collateral serving the loan.
The amount of the impairment should take into account the estimated costs associated the disposition of the credit/collateral.
But banks vary in how they apply FAS 114. Zions, for instance, establishes specific reserves at the time it takes the impairment, while other banks, such as First Horizon (FHN), charge their loans down by any impairment and reserves nothing. One method is not more conservative than the other but—and this is important--investors should know the difference when they compare one institution to another.
To see how this works, let’s go back to Zions. It has $570 million in nonaccrual loans, of which $467 million are deemed impaired. But based on the bank’s recent analysis and appraisals, only $164 million of the $467 million has a collateral shortfall, for which $35 million in loan loss reserves have been established.
First Horizon, meanwhile, has $770 million in nonperforming loans, of which $373 million are considered FAS 114 impaired. Only $39 million of these have associated loss reserves (of $5 million). The remaining $334 million of impaired loans have already been written down by $91 million and are carried at 62% of current appraised value.
Obviously, First Horizon’s approach has recognized chargeoffs faster than Zions will. Over time, the income statement impact should be the same, however. But if you try to compare one company’s reserve adequacy to the other’s by looking at each one’s ratio of loan loss reserve to nonperforming loans at any given time, though, you’ll end up with a misleading result. First Horizon writes down its impairment early on, while and Zions reserves for it and will write it off later.
That apples-to-oranges mismatch of loss accounting practices is one reason analysts often misinterpret loss data. But it’s not the only one. They also get hung up on other semi-bogus numbers, as well. For instance, the two most misleading measures of loss reserve adequacy are 1) the ratio of loss reserves to nonperforming assets, and 2) the ratio of reserves to nonperforming loan percentages. Sure enough, these two metrics are the favorites of many analysts and journalists.
Why are the numbers misleading? First, they are overly mechanistic. The reserve is built through backward-looking formulas that take into account the institution’s loan mix, its loan grades, and its historic loss rate by loan type for all of its loans, not just the nonperforming loans.
Second, any additional writedown or loss on the disposition of OREO does not come out of the loan loss reserve. Instead it shows up as an expense. Remember the titles: “loan loss reserve” and “other real estate owned.” Once a bank possesses the property or other collateral, that property is no longer a loan. It’s an asset. Any further value decline, therefore, cannot come out of the loan loss reserve.
To better understand how a bank determines the appropriate size of its loan loss reserve. let’s look at First Horizon’s (excellent) disclosure in its 10-K. (As an aside, I encourage all banks to adopt disclosures similar to the ones First Horizon first began providing 20 years ago.)
This table tells you a lot! First, the company classifies residential construction loans it has made to individuals as retail loans, while most other institutions classify them as commercial construction loans.
Second, notice that $118 million of the loan loss reserve, 35%, is associated with First Horizon’s retail loans. This is determined by using recent loss experience, by loan category, and establishing a reserve for the estimated next 12 months of losses.
Third, First Horizon grades all its commercial loans on a 16-point scale, with “1” being the equivalent of AAA-rated credits. Most banks use a nine-point rating system for their commercial loans but, like First Horizon, establish reserves based on historic loss experience by loan grade and type. Notice that $167 million of the loss reserve, almost half, is set aside for fully performing commercial loans graded one through twelve!
So a combination of a reserve for retail loans and a reserve for performing commercial loans make up over 80% of First Horizon’s loan loss reserve. Why, then, would anyone measure a bank’s reserve adequacy by comparing one raw number, its total loan loss reserve, to another, its nonperforming loans? That’s not just simple--it’s simpleminded! Some bank managements make matters worse by pretending the number is important and include it in their press releases and presentations. I say, stop it! Don’t succumb to Wall Street’s shortcuts! Instead, provide more thorough disclosure so investors can make better-informed judgments.
Finally, the true driver of reserve-building is the downgrading of the entire commercial loan portfolio that occurs in a credit cycle. This can be seen in the following table which compares First Horizon’s commercial loans by grade and associated reserves at year-end 2007 with year-end 2006.
First Horizon increased its loan loss reserves by $126 million last year, but that rise was not primarily the result of an increase in nonaccrual loans. Rather, it was driven by the downward shift in the grades of its performing commercial loans and higher frequency and severity assumptions used to reserve for future retail construction loan losses. The loan loss reserve for performing credits grades ten through 13 rose by $50 million as a result of the increase in the buckets due to downgrades.
Investors need to stop simplistically looking at the reserve as a percentage of nonperforming loans, and should demand a level of disclosure along the lines that First Horizon provides. Investors can then determine an institution’s trends in credit quality, by looking at changes in loan grades of the entire portfolio.
Tom Brown is head of Bankstocks.com
Disclosure: See declared holdings for fund managed by author: Second Curve Capital