Unrecognized Expected Credit Losses and Bank Share Prices

• Published date: 01 June 2021
• Author: P. BARRETT WHEELER
• Date: 01 June 2021
• DOI: http://doi.org/10.1111/1475-679X.12353

DOI: 10.1111/1475-679X.12353

Journal of Accounting Research

Vol. 59 No. 3 June 2021

Printed in U.S.A.

Unrecognized Expected Credit

Losses and Bank Share Prices

P. BARRETT WHEELER∗

Received 22 January 2019; accepted 30 December 2020

ABSTRACT

Accounting for credit losses under U.S. GAAP is transitioning from an

incurred to an expected loss model. The model change was motivated by

concerns that reporting only incurred losses does not provide investors with

sufficient and timely information about banks’ credit risk. In this paper, I de-

velop a measure of lifetime expected credit losses using vintage analysis to ex-

amine whether stock prices reflect information about unrecognized expected

credit losses in an incurred loss regime. Consistent with investors being able

to obtain information about expected losses that are not recognized in the

financial statements, I find that unrecognized expected credit losses are nega-

tively associated with bank stock prices. The pricing of these losses is stronger

for larger banks, consistent with lower costs of obtaining this information

for banks with better information environments. I also find that recorded

∗A.B. Freeman School of Business, Tulane University

Accepted by Regina Wittenberg Moerman. This paper is based on my dissertation

developed at Indiana University. I thank the members of my dissertation committee, Leslie

Hodder (chair), Brian Miller, Greg Udell, and Jim Wahlen for their insights and encourage-

ment in the development of this paper. Further, I thank Aleksander Aleszczyk (discussant),

Gus De Franco, Gans Narayanamoorthy, Ramgopal Venkataraman(discussant), Teri Yohn, my

fellow doctoral students at Indiana University, three anonymous reviewers, and workshop par-

ticipants at Vanderbilt University,Tulane University, the University of Houston, George Wash-

ington University, the 2018 FARS Midyear Meeting, and the 2019 AAA Annual Meeting for

helpful feedback. Finally,this paper is dedicated in memory of my father, a lifelong accountant

and banker who provided feedback on early drafts of this paper and found it very interesting,

until it got into all the regressions. An online appendix to this paper can be downloaded at

http://research.chicagobooth.edu/arc/journal-of-accounting-research/onlinesupplements.

Email: pwheeler1@tulane.edu

805

© 2021 The Chookaszian Accounting Research Center at the University of Chicago Booth School of

Business

806 p. b. wheeler

allowances were less than estimated expected losses, on average, consistent

with concerns that implementing the expected loss model will adversely

impact regulatory capital adequacy.

JEL codes: G21, M41

Keywords: banking; allowance for loan losses; loan loss provisions; incurred

loss model; current expected credit loss model; standard setting

1. Introduction

Forecasting credit losses for the loan portfolio is a key concern for all bank

stakeholders. It is necessary for managers to properly price loans, for reg-

ulators to assess a bank’s ability to absorb losses, and for investors to make

informed capital allocation decisions. The financial crisis of 2007–2009

sparked a debate about the appropriateness of the “incurred loss model”

(ILM), under which credit losses are recorded only when it is probable

they have been incurred (e.g., White and Stovall [2013], Cumming [2015],

Haslett [2015], ABA [2016b]). Critics of the model argue that investors

need more information about all expected losses and that requiring banks

to reserve for all expected losses would have helped investors better under-

stand the risk in banks’ loan portfolios prior to the crisis.

In response to these critics’ calls for increased transparency, the FASB

issued a change to U.S. GAAP that requires banks to estimate and record

a reserve for lifetime expected credit losses on loans at the time of loan

origination and to incorporate forecasts of economic conditions into their

estimates.1The new model—referred to as the current expected credit loss

or “CECL” model—has sparked controversy. Widely acknowledged as the

most significant change in bank accounting in decades, bankers lobbied

against the standard, expressing concerns about the cost of implementa-

tion, increased subjectivity in loan loss estimates, an adverse impact on reg-

ulatory capital, and its potential to exacerbate lending procyclicality.

Although critics of the ILM argue that it prevents investors from under-

standing the risks in banks’ loan portfolios, little empirical evidence exists

on this issue. On the one hand, given the importance of credit losses to

bank performance, the benefits of obtaining information about expected

losses could outweigh the costs. On the other, the incomplete revelation hy-

pothesis (Bloomfield [2002]) predicts that information that is more costly

to extract from public data will be less fully revealed in price. Whether

investors understand the extent of unrecognized expected losses is an

1The new credit losses standard—ASU 2016-13 (ASU)—was issued on June 16, 2016, and is

effective for fiscal years and interim periods beginning after December 15, 2019, for business

entities that are SEC filers and not considered smaller reporting companies and for fiscal years

and interim periods beginning after December 15, 2022, for all other entities. However,banks

previously required to adopt in 2020 were given the option to defer adoption as part of the

Coronavirus Aid, Relief, and Economic Security (“CARES”) Act, signed into law on March 27,

2020.

unrecognized expected credit losses and bank share prices

807

empirical question. In this paper, I examine this question by developing a

bank-specific measure of lifetime expected credit losses using a method in-

tended to mimic one acceptable estimation method suggested by the FASB.

My method, based on vintage analysis, incorporates both historical lifetime

loss patterns and a forward-looking adjustment for recent changes in eco-

nomic conditions.2I calculate the magnitude and direction of differences

between my measure of expected credit losses and allowances recorded un-

der the ILM—which I refer to as the “expected-incurred difference”—and

examine whether bank stock prices reflect this difference.3

My tests, which extend research on the relation between loan loss al-

lowances and firm value (e.g., Beaver et al. [1989], Barth, Beaver, and Stin-

son [1991], Ahmed, Takeda, and Thomas [1999]), show that stock prices

are negatively associated with the expected-incurred difference in a pooled

sample from the fourth quarter of 2006 to the fourth quarter of 2016. This

finding suggests that investors can obtain information about unrecognized

expected losses under the ILM, which contradicts the notion that it pre-

vents investors from understanding lifetime expected losses.

I find that the association between the expected-incurred difference and

stock prices is stronger for larger banks, consistent with them having more

transparent information environments that reduce the costs of extracting

information about expected losses (e.g., Atiase [1980], Atiase [1985], Free-

man [1987], Ro [1988], Bhushan [1989]). Further, I find that the pricing

of expected losses is stronger for banks with assets greater than $10 bil-

lion after implementation of stress tests required by the Dodd-Frank Act of

2010. The Dodd-Frank Act requires banks above this threshold to produce

and report to federal regulators information about the impact of various

hypothetical economic scenarios on the performance of their loan port-

folios. Beginning in 2013, the Federal Reserve began publishing the re-

sults of these stress tests for banks with total assets greater than $50 billion

while requiring banks with assets between $10 billion and $50 billion to

submit company-run stress test results and publicly disclose a summary of

the results for the “severely adverse” scenario from 2015 to 2017. My results

suggest that the reporting of stress test results improved the information

environments for larger banks with respect to expected losses.

2The FASB does not require a specific credit loss estimation method under CECL. In

the ASU’s “Basis for Conclusions,” the board recognizes that different methods can result

in a range of acceptable estimates (see paragraph BC50). Vintage analysis is one acceptable

method for estimating lifetime losses discussed in the ASU. I do not consider other acceptable

measurement methods in this study.

3A positive expected-incurred difference indicates that a bank is understated relative to

expected losses, whereas a negative expected-incurred difference indicates overstate-

ment. The terms “understatements,” “overstatements,” “understated,” “overstated,” and

“unrecognized expected credit losses” in this paper refer to the level of recorded

allowances relative to estimates of expected losses. I do not attempt to measure understate-

ments relative to what allowances should be under the ILM, which is typically the focus of

prior research.