Imperfect Accounting and Reporting Bias

• Published date: 01 September 2017
• Date: 01 September 2017
• DOI: http://doi.org/10.1111/1475-679X.12170

DOI: 10.1111/1475-679X.12170

Journal of Accounting Research

Vol. 55 No. 4 September 2017

Printed in U.S.A.

Imperfect Accounting

and Reporting Bias

VIVIAN W. FANG,

∗ALLEN H. HUANG,

†AND WENYU WANG

‡

Received 7 August 2015; accepted 21 December 2016

ABSTRACT

Errors and bias are both inherent features of accounting. In theory, while

errors discourage bias by lowering the value relevance of accounting, they can

also facilitate bias by providing camouflage. Consistent with theory, we find

a hump-shaped relation between a firm’s propensity to engage in intentional

misstatement and the prevalence of unintentional misstatements in the firm’s

∗University of Minnesota; †Hong Kong University of Science and Technology; ‡Indiana

University.

Accepted by Haresh Sapra. We are indebted to an anonymous referee for valuable com-

ments and suggestions, Mirko Heinle and Gaoqing Zhang for numerous helpful discussions

on modeling, Alex Edmans for sharing the scaled wealth-performance sensitivity measure,

Andy Leone for sharing data on the GAO restatement classification, Simi Kedia and Rebecca

Files for sharing data on the GAO misstatement periods, and Rick Mergenthaler for sharing

data on the rule-based characteristics of accounting standards. The paper has also benefited

from comments and suggestions by Robert Bushman, Qi Chen, Daniel Cohen, Carlos Corona,

Hemang Desai, Alex Edwards, Rich Frankel, Frank Gigler,Charles Ham, Michael Iselin, Chan-

dra Kanodia, Steve Karolyi, Jon Karpoff, Allison Koester, Lian Fen Lee, Andy Leone, Meng

Li, Pierre Liang, Thomas Lys, Stan Markov, Xiumin Martin, Brian Miller, Dhananjay Nanda,

Tom Ruchti, Iman Sheibany, Joel Waldfogel, Michael Weisbach, Andy Winton, Guochang

Zhang, Yuan Zhang, and Yun Zhang, as well as participants at the 2016 MIT Asia Confer-

ence in Accounting; 2016 FARS Midyear Meeting; 2015 SMU Accounting Symposium; 2015

Conference on Investor Protection, Corporate Governance, and Fraud Prevention; 2015 Min-

nesota Empirical Accounting Research Conference; and seminar participants at Carnegie Mel-

lon University, City University of Hong Kong, Southern Methodist University, University of

Miami, University of Minnesota (Strategic Management & Entrepreneurship), and Washing-

ton University at St. Louis. We thank Yi Jiang, Reeyarn Li, Jianghua Shen, and Chao Tang

for excellent research assistance. An online appendix to this paper can be downloaded at

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

919

Copyright C, University of Chicago on behalf of the Accounting Research Center,2017

920 V.W.FANG,A.H.HUANG,AND W.WANG

industry for the whole economy and a majority of the industries. The result is

robust to using firms’ number of items in financial statements and exposure

to complex accounting rules as alternative proxies for errors and to using

the restatement amount in net income to quantify the magnitude of bias and

errors. To directly test for the two effects of errors, we show that when errors

are more prevalent, the market reacts less to firms’ earnings surprises and

bias is more difficult to detect. Our results highlight the imperfectness of

accounting, advance understanding of firms’ reporting incentives, and shed

light on accounting standard setting.

JEL codes: G32; G34; G38; M40; M41; M48; M53

Keywords: accounting errors; reporting bias; fraud; accounting regulation;

earnings response coefficient; fraud detection; textual analysis

1. Introduction

Accounting is a complex process that necessitates professional knowledge

and substantial judgment. Setting aside possible bias, accounting is still

imperfect and errors commonly occur. A 2007 Wall Street Journal article

reports a record high of 1,420 financial restatements in 2006 involving

nearly 10% of U.S. public companies, the majority of which are due to

small companies correcting errors with no apparent intention to misreport

(Posen [2007]). Hennes, Leone, and Miller [2008] (hereafter, HLM)

classify 73.6% of U.S. Government Accountability Office (GAO) restate-

ments as unintentional misapplications of generally accepted accounting

principles (GAAP) and only 26.4% of them as intentional misapplications.

Many more errors probably do not get corrected, but nonetheless affect

reported accounting numbers.

One possible implication of reporting errors is that they shape firms’ in-

centives for bias, or even fraud.1To see why, consider the world of Becker

[1968], in which a firm manager commits fraud only if her benefits of ma-

nipulating earnings outweigh the costs. With accounting being imperfect in

practice, both benefits and costs of the manager are likely to be functions

of the error rate in the firm’s financial reporting, because errors affect mar-

ket participants’ perceptions of accounting’s information value, as well as

their ability to discern the “correct” accounting numbers and detect fraud.

For this reason, we expect reporting errors to yield significant effects on re-

porting bias. We develop this idea analytically and empirically in this paper.

We begin by building a one-period reporting game, extending Fischer

and Verecchia [2000] (hereafter, FV). In the game, a risk-neutral firm

manager makes a potentially biased earnings report to a risk-neutral market

1Throughout the paper, we use “errors” to refer to unintentional misapplications of GAAP

and “bias” to refer to intentional misapplications of GAAP. Reporting bias rises to “fraud”

when it results in a material misstatement that violates securities laws (such as section 17a of

the Securities Act of 1933 or section 10 of the Securities Exchange Act of 1934) or related

securities regulations (such as Securities and Exchange Commission (SEC) Rule 10b-5).

IMPERFECT ACCOUNTING AND REPORTING BIAS 921

after observing the firm’s earnings. The earnings that the manager observes

consist of the firm’s true earnings and a noise term. This noise term is our

theoretical construct of interest, which we interpret as accounting errors.

The manager reports to the market her observed earnings plus her bias of

choice. The market forms a rational expectation of the firm’s true earnings

based on the firm’s reported earnings and its prior beliefs.

We modify FV by allowing the noise term to affect the manager’s cost of

bias, which aims to capture the potential “camouflage effect” of account-

ing errors on bias. This effect arises because accounting noise, by making

fraud detection more difficult, potentially lowers the manager’s costs of

manipulating earnings and facilitates opportunistic reporting.2There is,

however, an offsetting effect. This effect is first described in FV: As noise

in the accounting process increases, earnings become less value relevant

to the market, which reduces the manager’s benefits of biasing earnings

and dampens her incentives to do so. We thus refer to this as the “value

relevance–reducing effect.” The properties of the errors’ two effects deter-

mine the shape of the relation between the manager’s propensity to bias

earnings and errors’ variance.

We observe a hump-shaped association between a firm’s probability of

engaging in an intentional misstatement in a quarter (our proxy for bias

propensity) and the percentage of its industry peers engaging in unin-

tentional misstatements in the same quarter (our primary proxy for er-

rors’ variance) for a broad sample of U.S. firms from 1996Q1 to 2005Q4

and a majority of industries in the sample. The latter proxy assumes that

a larger variance makes possible extreme realizations of errors, which

are more likely to be detected. Both proxies are constructed following

HLM’s approach, which classifies the misstatements covered by the GAO

database into intentional and unintentional ones based on a combination

of searches for fraud-related keywords, regulatory enforcement actions, and

investigations. Thus, in this approach, bias and errors are defined as what

the “policeman” says, consistent with our model, in which bias is subject

to costs while errors are not. A hump-shaped association between bias and

errors is consistent with the two counteracting effects that we model. The

turning point of the observed association is high compared to the average

error rate of the respective sample, which indicates that the camouflage ef-

fect likely outweighs the value relevance–reducing effect for a majority of

our sample firms.

2A similar camouflage effect is observed in U.S. tax law practices, that is, widespread errors

in tax returns make it difficult to detect fraud. In FY2012, the IRS reported investigations of

2,987 tax frauds, which account for only 0.0015% of the 198 million taxpayers in the United

States. In an article titled “Negligence Versus Tax Fraud: How the IRS Tells the Difference,”

Nolo Press notes that the percentage of Americans convicted of tax crimes is strikingly small

compared to the 17% of noncompliant taxpayers estimated by the IRS. The article adds that,

because tax auditors are aware of the complexity of U.S. tax law, they “expect to find a few

errors in every tax return and do not routinely suspect [tax fraud]” (available at http://www.

nolo.com/legal-encyclopedia/negligence-versus-tax-fraud-irs-difference-29962.html).