Risk Management

Better Loss Model May Have Prevented Bank Failures: GAO

The accounting model banks use to set reserves for loan losses contributed to the large number of community bank failures during the financial cris...
Vincent RyanJune 18, 2013

Weak loan underwriting and risky funding sources resulting from aggressive growth strategies caused the bulk of community bank failures from 2008 to 2011, according to a Government Accountability Office report released last week. But the Generally Accepted Accounting Principles model banks use for recognizing credit losses — which was based on historical loss rates that were unusually low pre-crisis — was also partly to blame.

The current “incurred loss” loan model requires evidence of a potential credit loss before a loan asset can be written down. But by following the incurred loss model from 2008 to 2011, banks recognized loan losses too late and didn’t have enough reserves set aside to absorb a huge wave of defaults, the GAO said.

In its findings, which included an analysis that found fair-value accounting contributed little or nothing to  failures of banks of all sizes, the GAO implicitly backed a Financial Accounting Standards Board proposal that would institute a loan-loss provisioning model that is forward-looking instead of backward-looking. It focuses on “expected” future losses. The GAO said the FASB’s proposed current expected credit loss model (CECL) would “establish a means of recognizing potential losses earlier on the loans [banks] underwrite and could incentivize prudent risk management practices.”

Many major U.S. banks have lined up against the CECL proposal, which FASB says could cause some of them to increase their loan loss reserves by as much as 50 percent.

In a June 10 comment letter to FASB, John H. James, corporate controller at Bank of America, said: “We believe that the immediate recognition of all credit losses expected over the contractual life of a loan, without any evidence of credit deterioration, lacks conceptual foundation and would result in the premature recognition of credit losses.”

James said the bank has “serious concerns about the reliability of estimates for credit losses that are expected to occur beyond the period for which forecasts are reasonable,” and those concerns “are not resolved by reverting to unadjusted historical averages or assuming that economic conditions will remain stable, as has been suggested.”

In another comment letter, Randy Mayor, CFO of Home Bancshares, a Conway, Ark. bank with $4.2 billion in assets, echoed  BofA’s statement. He said the CECL “would require very speculative forecasting on all loans including the economic cycle.” He also wrote said the model would “create a guessing game method which could lead to the perception of earnings management” by banks. (By tapping excess loan loss reserves in a poor earnings quarter, a bank can avoid spikes in reported financial performance.)

Bank of America believes the existing incurred-loss model is still the way to go, James said, but it could be enhanced in two ways to ensure more timely loss recognition: First, by letting credit losses be recognized earlier than U.S. GAAP currently permits. That would be accomplished by lowering the recognition threshold for a potential credit loss from “probable” to “more likely than not.” Second, the current model could be improved by requiring that the measurement of credit losses be “informed by reasonable and supportable forecasts of economic conditions in the forseeable future.”

While inadequate loan-loss reserves was a problem at financial institutions of all sizes during the crisis, it particularly hurt community banks that had gorged on commercial real estate loans. Between January 2008 and December 2011, 85 percent of the 414 FDIC-insured financial institutions that failed were small banks with less than $1 billion in assets, the GAO said. In the 10 states that had 10 or more bank failures, the concentration of commercial real estate (CRE) loans on community banks’ balance sheets was particularly high.

Some of this buildup in CRE loans occurred at banks that “had been chartered for less than 10 years at the time of failure and appeared in many cases to have deviated from their approved business plans,” the GAO said.

The ratio of total CRE loans to total risk-based capital at banks in the 10 states leaped to 535 to 1 in June 2008.  The ratio of a subset of CRE loans — acquisition, development and construction (ADC) loans — also jumped. ADC loans in particular ate into banks’ regulatory capital and earnings. The level of nonperforming ADC loans hit 11 percent in June 2008 and 46 percent by June 2011. Net charge-offs of ADC assets rose to 2 percent by June 2008 and then 12 percent in June 2011.

If community banks had been forced to increase loan-loss allowances earlier than they did, the GAO said, they wouldn’t have needed to try to raise capital as their loan books were rapidly deteriorating, “when they were least able to do so.”

The GAO uncovered two other facts about the prevalence of community bank failures during the financial crisis. One was that fair-value accounting contributed little to bank failures, because held for investment (HFI) loans, which were not subject to fair-value accounting, made up two-thirds of defunct banks’ assets. 

The other fact that GAO was that that in no state did a rash of bank failures crush the local economy. A high number of bank failures didn’t materially affect workers’ income levels or the unemployment rate, but did correlate significantly with real estate prices. “The impact of bank failures on a state’s economy … is less likely to appear in the overall labor market or in the broader economy,” the GAO concluded.