Notes from the Vault
Larry D. Wall
August 2013
Investors overwhelmingly prefer a change in loan loss accounting rules that could increase a bank's loan loss allowance by an estimated 50 percent, according to the Financial Accounting Standards Board (FASB) Feedback Summary released in early July. One of the more important changes by this FASB proposal on credit losses would be to replace the current "incurred loss" model, in which loan losses are not recognized in financial statements until it is probable (based on current information and events) that the loan is impaired. This standard was criticized by the Financial Crisis Advisory Group as delaying the recognition of loan losses and overstating asset values during the financial crisis. The proposed new standard (in paragraph 825-15-25-1) would require banks to hold a loan loss allowance equal to the expected loss over the life of the loan, using all currently available information, including "reasonable and supportable forecasts." This change would force earlier loss recognition, with banks required to start provisioning for a loan loss at the end of the period when the loan was made.
An example of the change
I show the impact of changing from "incurred loss" to "expected loss" with a simple example. Suppose a bank makes a $10 million loan during the first period for which it estimates a probability of default of 1 percent. If the loan goes into default, the expected loss is 40 percent, or equivalently the bank expects to recover 60 percent of the value of the loan. Then suppose something bad happens to the borrower and the probability of default increases after the end of the first period but before the end of the second period. Exactly how much of an increase cannot be determined precisely, but a reasonable estimate is that the new probability of default is somewhere in the range of 70 percent to 90 percent. The following table illustrates the effect of the change:
The bank does not create an allowance (sometimes called a reserve) at the end of the first reporting period for the loan under the "incurred loss" model. No reserve is permitted because a credit loss is not "probable." In contrast, under the "expected loss" model, the bank must increase the allowance for the full value of the expected losses over the life of the loan at the end of the first period. The result is that the bank will report provisions of $40,000 ($10 million x 1 percent x 40 percent), and reduce its pretax net income by $40,000 under the expected loss model relative to the incurred loss model.
At the end of the second reporting period, the bank still does not recognize any loss in the incurred loss approach as long as bank management estimates the likelihood of default to be less than probable. Current accounting standards (Accounting Standards Codification 310-10-350-18) state, "Probable is a higher level than more likely than not," but otherwise leaves the term ambiguous. Scott Taub, former deputy chief accountant at the U.S. Securities and Exchange Commission, says that in practice the term "probable" is generally interpreted as "something in the neighborhood of 75 to 80 percent." Applying these requirements for probability of default suggests the bank would report no loan loss allowance even if the probability of loss was 70 percent. However, if bank management determines that default is probable, the bank must increase its loan loss allowance by the amount of the impairment. In this case, the impairment is the difference between the recorded investment in the loan and the present value of the expected cash flows (as may be proxied by the observable market price of the loan). As the bank had not previously created an allowance for this loan and the market price of the loan would decrease by the amount of the expected loss, the bank reports an increase in its loan loss provision and a decrease in its pretax net income of $3.6 million. Thus, the difference between management determining there is a 70 percent probability of default and a 90 percent probability of default is a decrease of $3.6 million in pretax accounting net income for the second period.
The accounting for the loss is very different under the expected loss model. The new allowance for this loan will range from $2.8 million if bank management thinks the new probability of default is 70 percent to $3.6 million for a 90 percent probability of default, a $800,000 difference. Moreover, because the bank had created a $40,000 allowance when it made the loan, the loan loss provision (and reduction in pretax net income) is reduced to the range of $2.76 million to $3.56 million, depending upon the estimated probability of default.
This example highlights three important ways in which the proposal would change loan loss accounting. First, when the loan is made the bank recognizes no loan losses in the incurred loss model but recognizes an immediate loss in the expected loss model. As a result of this change plus other proposed changes in loan loss accounting, FASB Chairman Leslie Seidman said in an interview that many larger financial institutions have estimated a 50 percent increase in their loan loss allowance.
Second, the effect of relatively small changes in probability can have a huge effect on loan loss provisions in the incurred loss model, with any estimated probability of default below 75 percent to 80 percent resulting in no loss, whereas any estimated probability of default above this range would require a provision equal to the expected impairment. Third, changes in the estimated probability of default under the expected loss model will change a bank's required allowance, provision, and pretax net earnings. However, there is not the big threshold effect that exists between probable and less than probable in the incurred loss model.
The proposal's impact on banks
The proposed accounting change per se does not change the underlying economics of the loan. Borrowers still have the same legal obligation to repay the loan regardless of how the bank accounts for the loan. Moreover, contrary to occasional claims that banks set aside cash to cover their loan loss allowances, lenders are not obligated to hold any more cash. Rather, the significance of this proposed change in loan loss accounting arises from how banks, bank supervisors, and investors respond to changes in financial accounting.
An accounting change that forces higher loan losses would effectively represent an increase in banks' equity capital requirements under existing regulations. Existing capital regulations require banks to fund part of their assets with equity (shareholder funds), which includes earnings that have not been paid out to shareholders (retained earnings). An increase in banks' loan loss allowance will reduce pretax and after-tax earnings and therefore retained earnings (unless the bank also cuts its dividends). Thus, it is not too surprising that the FASB Feedback Summary reports that preparers (banks) generally oppose the change in loan loss requirements and bank supervisors generally support the increase in effective capital requirements (see my paper with University of South Carolina Professor Timothy W. Koch for a discussion of the supervisors' perspective on loan loss accounting).
The issue of how the proposed change would affect investors is more complicated. The expected loss model will generally result in an understatement of the value of new loans. As the late Emory University Professor George J. Benston and I noted in a paper, the interest rate charged on loans almost always fully covers the cost of the loan—including its expected credit losses. Thus, when banks make a loan, the expected value of the loan repayments exceeds the amount of the loan, implying that no loan loss provision is needed to fairly value the loan. Further, by recording part of the loss when the loan is made, the bank is able to smooth its earnings report by shifting part of its loss recognition to good times when many new loans are being made and away from bad times when fewer new loans are granted and losses are higher. This income smoothing reduces the volatility of reported net income and, thus, could contribute to investors underestimating the true risk of the firm.
The FASB Feedback Summary provides some explanation as to why "almost all investors" showed strong support for the proposed changes. Investors want to have reliable estimates of an individual bank's credit losses to evaluate the potential risk to the bank's earnings and capital, and to evaluate the appropriateness of its risk-based pricing model. Investors understand that management's estimates of loan losses necessarily contain a significant element of subjectivity; therefore, investors devote "vast amounts of time" in analyzing the credit quality of the portfolio.
Despite the subjectivity of management's estimate of loan losses, investors nevertheless find the reserves—or more accurately, the change in reserves—to be a valuable signal about credit quality information that is available only to management. Presumably, part of investors' frustration with the current rules is that the triggers for loss recognition increase the subjectivity of reported loan loss estimates in any given period and make it more difficult to infer management's private information. In contrast, investors get a more accurate picture of management's estimates, as small differences in their estimated probabilities of loss make only a small difference in the reported loan loss allowance.
Although FASB's proposal provides one method of requiring more timely recognition of losses that have been incurred on loans, there are other proposals that would have the same effect. For example, Benston's and my proposal to value loans at the lower of their cost or economic value is similar to FASB's in that both would require recognition of deterioration in loan values even if default is not probable. However, our proposal differs in that we would not have recorded any loss provision on fairly priced loans that had not deteriorated, whereas FASB requires some loan loss provision on all loans. The result is that relative to our proposal, FASB's proposal provides investors with a more complete measure of management's expectations for the total loan losses of the portfolio.
Conclusion
FASB's proposal to move away from the incurred loss model's "too little, too late" recognition of loan losses is good news for investors and supervisors (and the taxpayers backing banks' deposit insurance). The net benefits of going all the way to early loss recognition are less clear. Although many supervisors may hope each dollar of additional loan loss allowances will increase banks' ability to absorb one more dollar of losses, such a favorable outcome is unlikely. Instead, banks are likely to perceive the extra loss absorbing ability in their loan loss allowance as at least partially reducing their need to maintain a capital buffer above the regulatory requirements. Full recognition of expected credit losses will give investors better insight into management's private information, but it does so at the cost of understating assets in good times and mandated income smoothing. From the investors' perspective, FASB's survey suggests that most of the investors who were polled would prefer better loan loss estimates despite the costs.
Larry D. Wall is the executive director of the Center for Financial Innovation and Stability at the Atlanta Fed. The author thanks Paula Tkac for helpful comments on the paper. The views expressed here are the author's and not necessarily those of the Federal Reserve Bank of Atlanta or the Federal Reserve System. If you wish to comment on this post, please e-mail atl.nftv.mailbox@atl.frb.org. You can also read previous Notes from the Vault posts.