Incremental Information of Auditor Quality Baolei Qi Xi’an Jiaotong University Email: [email protected]Yi Si Xi’an Jiaotong University Email: [email protected]Gaoliang Tian Xi’an Jiaotong University Email: [email protected]Martin G. H. Wu* The University of Illinois at Urbana-Champaign Email: [email protected]January 12, 2015 * Corresponding author. The paper’s former title was “Do individual auditors affect clients’ real activities manipulation? Evidence of learning and auditor quality.” We appreciate the helpful comments received from Andrew M. Bauer, Timothy Bauer, Paul J. Beck, Joseph Carcello, Monica Causholli, Long Chen, Richard Crowley, Keith Czerney, Keith Decie, Paul Demere, Laura Li, Tracey Majors, Heather Pesch, Theodore Sougiannis, Anne Thompson, Donghui Wu, and the workshop participants at University of Illinois at Urbana- Champaign, Xi’an Jiaotong University, Fudan University, Hunan University, the 2014 American Accounting Association Annual Conference, and the 2014 Temple University Accounting Conference. Gaoliang Tian acknowledges financial supports from the National Science Foundation of China (Grant No. 71102095) and the Scientific Research Foundation of the Humanities and Social Planning Project of the Ministry of Education of China (Grant No. 13YJA630081).
50
Embed
Incremental Information of Auditor Quality · 2017-08-11 · impacts of different auditors as incremental information of auditor quality. This result suggests that the auditor-client
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
feedback and feed-forward effects, and real activities manipulation
JEL Classification: M49
Data Availability: All the data used in this study are publicly available.
~1~
1. Introduction
Audit clients often manage earnings through accounting accruals and real activities manipulation
(RAM), and earnings quality affects audit outcome. Hence, both accruals and RAM are useful proxies for
audit quality. Using abnormal accruals and modified audit opinions (MOD) as proxies, for example, Gul,
Wu, and Yang (2013) find evidence supporting audit quality varies across different auditors. Their
evidence suggests individual auditors have distinct impacts on the audit quality measured at the firm level.
But it is unclear whether individual auditors contribute incremental information beyond what the firm-
and city/office-level audit quality provides. If they do, what is the extent to which the incremental
information goes beyond the financial reporting responsibilities shared by the auditor and the client?
RAM usually refers to opportunistic reduction in clients’ discretionary expenditures (e.g., SG&A,
R&D, quality inspection, etc.), overproduction, and price discounts to boost sales temporarily. Although it
is not likely to violate the GAAP or GAAS, RAM may work as a substitute for or complement to the
accrual method. Hence, an increase in auditors’ scrutiny constrains accrual manipulation and in turn,
increases clients’ RAM, which suggests individual auditors either respond to or influence RAM or both.
We refer to the auditors’ responses and influences as the associations between the auditors and their
clients’ economics. Since clients’ RAM is part of the clients’ operating decisions, it will capture more
information contents about the clients than the clients’ accruals for financial reporting do.
We predict and find that clients’ RAM varies across different auditors. Specifically, we document
that auditors respond to or influence clients’ economic decisions through RAM.1 In multiple regressions
of RAM, we explicitly include control variables such as audit fees, going-concern opinions (GC), MOD,
etc. These controls are traditional proxies for audit quality at the firm level. We interpret the RAM
impacts of different auditors as incremental information of auditor quality. This result suggests that the
auditor-client relationships, in China at least, are the result of clients’ economics that is complex and far-
reaching, relative to the financial reporting responsibilities shared between the auditor and client.
1 We also perform the sensitivity analyses of the fixed effects of audit firms and individual auditors on Tobin’s Q,
which represents clients’ investment opportunities and hence economics. Our findings are broadly consistent.
~2~
Auditors unremittingly seek opportunities to improve their understanding of clients’ business
models to enhance audit quality. In a nonstrategic dynamic model, for example, Beck and Wu (2006)
analyze how auditors’ learning and selection of professional services are jointly affecting audit quality. A
large number of archival and experimental studies have documented a plethora of evidence supporting
that auditors develop audit judgments and decision-makings by specialization, leadership, and portfolio
choices (see Section 2 for the literature review). More recent studies also find that auditors react toward
clients’ RAM (Kim and Park 2014, Commerford, Hermanson, Houston, and Peters 2014a, 2014b, and
Greiner, Kohlbeck, and Smith 2013). “In particular, we expect high quality auditors to consider not only
whether the client’s accounting choices are in technical compliance with GAAP, but also how faithfully
the financial statements reflect the firm’s underlying economics,” summarize DeFond and Zhang (2014).
It is well-known that audit clients’ economic decisions affect auditors’ judgments, learning, and
selection of professional services (e.g., Wu 2006 considers oligopolistic markets for both auditing and
consulting services). It is intriguing, however, whether auditors can conversely influence clients’
economic decisions. Notice that the negative answer to the above question would cast grave doubts and
limitations on the value-generating ability of auditors, thus jeopardizing the economic demand for audits.2
But the positive confirmation has been at best the feedback effects of auditors’ approvals of accounting
policy choices on clients’ business operations. Four arguments supporting the feedback effects in the
literature are worthy of special mention.3 (1) Qualified audit opinions often carry severe consequences for
clients’ economics. GC options are frequently called “self-fulfilling prophesies” and hence precisely
pinpoint the feedback effects that accounting policy choices may have on audit clients’ economics.
2 An analogy is the debate of information disclosure in an exchange economy versus a production economy.
3 The traditional view is RAM is outside of auditors’ purview as auditors have neither incentives to affect nor means
to detect clients’ RAM. In the world’s fast-changing environments, investors start questioning the usefulness of the
traditional financial statement audits. “Regulators have rebuked three of the Big 4 firms for not fixing past problems
fast enough,” Rapoport (2013) reports, “Many investors rely on the audit report to help assure them a company’s
numbers are accurate. … Regulators and industry critics say investors need more information from auditors about
matters such as whether a company’s accounting is aggressive, and what auditors think are the most important
features of the company’s finances.” New rules for annual audit reports are currently under consideration in the U.S.
in order to enhance auditors’ feedback effects on clients’ economic decisions, not exclusively on clients’ accounting
policy choices. Furthermore, the Chinese audit market must also be significantly influenced by “Chinese relations,”
which may help to yield complex and far-reaching auditor-client relationships beyond the financial reporting
responsibilities between the auditors and clients.
~3~
(2) Audit reports often serve as one of the principal means that shareholders use to evaluate their
company’s financial position and managerial performance. For example, shareholders almost always
include earnings, which have the virtue of being audited, in executive compensation. And compensation
contracts will influence the company’s strategic decisions and economic outcomes. (3) The Sarbanes-
Oxley Act of 2002 (SOX) requires auditors to evaluate and attest to their audit clients’ “internal controls
for external financial reporting” (COSO 2013). The SOX has unprecedentedly bound together the clients’
business operating activities and financial reporting decisions, thereby causing them interlocking and
inseparable.4 (4) Finally, auditors are an incredible source of information for audit clients. Selection of
professional services will enhance auditors’ capabilities to compete in oligopolistic markets for auditing
and consulting services (Wu 2006).5
The above arguments are compelling for the feedback effects of auditors on clients’ economics,
but there lacks empirical evidence. The goal of this study, therefore, is to document the first evidence (to
the best of our knowledge) supporting individual auditors contribute incremental information beyond
what the firm-level audit quality provides. Our study is mostly close to Gul et al. (2013) for three reasons.
First, both studies use archival data from China because two auditors (partners or senior managers) of
each audit engagement are required to sign their names on audit reports. Second, many Chinese listed
companies used both accruals and RAM in order to meet the threshold levels set by the Chinese
governments for funds and operations (Li, Zheng, and Lian 2011). Constraints on accruals could have
driven RAM. Third, the decade-long fast-growing economy in China offers a unique opportunity for
auditors to interact closely with clients, thereby knitting very intricate auditor-client relations.
4 In Section 2, we will review some recent studies of the economic (financing, investing, and operating) impacts of
disclosures of “material weaknesses of internal control over external financial reporting.” 5 For example, Rapoport (2012) reports:
Consulting and other nonaudit lines of business are growing at rates far outpacing auditing. … At
PwC, ninety percent of advisory work is for nonaudit clients, said Dana Mcilwain, PwC’s U.S.
advisory leader. The Big Four also argue that consulting provides synergies even if they don’t
consult for and audit the same companies. Offering consulting gives them expertise they can draw
upon when related issues arise at their audit clients, they say. … “We believe the services we’re
in actually help us on the front of audit quality,” said John Ferraro, Ernst &Young’s global chief
operating officer.
~4~
More importantly, our study complements and extends Gul et al. (2013). We document the extent
to which RAM (as a complement to accruals) varies across different auditors. While accruals’ variation
shows that auditors have distinct impacts on audit quality, RAM’s variation will suggest auditors react to
or influence clients’ economic decisions differently. Therefore, the auditor-client relations are exhilarated
by more than merely the financial reporting responsibilities shared between the auditor and the client. Our
extension comes from the inclusion of explicit control variables such as audit fees, GC, MOD, etc., in
multiple regressions of RAM. As all of these control variables are proxies for audit quality at the firm
level, we refer to the association between auditors and RAM in the regressions as incremental information
of auditor quality.
The implication of our findings is straightforward. Regulatory rules (see, e.g., signature and
rotation of engagement partners) curbing free riders in auditing would help strengthen the auditor-client
relations and, hence, auditor quality. Hence, our study is a valuable addition to the list of recent studies
following the “Concept Release on Requiring the Engagement Partner to Sign the Audit Report” by
PCAOB (2009).6 These studies focus on auditor quality, rather than the firm- and city/office level of audit
quality.7 The identification of individual auditors does make the comparison between audit quality and
auditor quality possible. Since audit quality at the firm level is an average of auditor quality, identifying
and eliminating the “tail distribution” of the latter quality will naturally increase the former.
The remainder of our study proceeds as follows. In Section 2, we provide institutional
backgrounds and a literature review. We elaborate our research questions. In Sections 3 and 4, we discuss
research design, empirical findings, and regression robustness tests. Section 5 concludes the study.
6 See, e.g., Nelson and Tan (2005), DeFond and Francis (2005), Church, Davis, and McCracken (2008), Francis
(2011), Carcello and Li (2013), Knechel, Rouse, Schelleman (2013), and Gul et al. (2013). 7 Audit firm-level studies include Becker, DeFond, Jiambalvo, and Subramanyam (1998), Francis and Krishnan
(1999), and Lennox (1999), and audit office-level studies include Reynolds and Francis (2000), Krishnan (2003,
2005), Francis, Reichelt, and Wang (2005), and Reichelt and Wang (2010).
~5~
2. Institutional Background, Literature Review, and Research Questions
2.1 Emerging Market Economy of China and Literature Review
The establishment of the Shanghai and Shenzhen stock exchanges in the early 1990s is a sign that
China has started converting its central planning economy into a market economy. In China, shareholders
largely lack the necessary knowledge about monitoring and controlling their listed companies, however.
The securities regulations are also lax relative to the counterparts of the developed markets and
economies. The Chinese regulators are practicing the philosophy called “crossing river by feeling the
stones” as they are unsure whether regulatory policies are efficient (Aharony, Lee, and Wong 2000). As a
result, the Chinese stock markets exhibit two unique characteristics (Chen, Lee, and Li 2008). One is the
quota system designed to assure business stability, and the other is the large quantities of stock
transactions regulated by the Chinese governments. Both characteristics give a rise to the multiple
threshold levels set by the Chinese governments for the Chinese listed companies to meet in order to
continue business operations. Thus, many Chinese listed companies actively manage earnings (Chen and
Yuan 2004, and Chen et al. 2008).8
Companies often manage earnings through either the accrual method or the RAM method or both.
Generally speaking, the accrual method is easier to use relative to the RAM method (Roychowdhury
2006). While accruals are subject to GAAP and GAAS, RAM is not and usually has a negative impact on
future firm performance (Ewert and Wagenhofer 2005). A large number of studies have compared the
usages of the two methods, yet it is unclear that one method dominates the other. For example, Graham,
Harvey, and Rajgopal (2005) report the majority of the 400 CFOs surveyed are eager to use RAM. Cohen,
Dey, and Lys (2008) find that companies switch from the accrual method to the RAM method after the
enactment of SOX. Bartov and Cohen (2009) find a decline in the frequency of meeting or beating
8 A delisting policy of the China Securities Regulatory Commission (CSRC, 1998) has two special provisions:
Special Treatment (ST) and Particular Transfer (PT). ST means a listed company incurring losses in two consecutive
years will receive a “detention notice.” PT means a company incurring losses in more than two consecutive years
will be explicitly identified as such. As a result, ST and PT will impose significant operating and financing hurdles
on the company going forward. For example, a company must have its Returns on Equity greater than six (ten)
percent in three successive years in order to issue a Rights Offering (Seasonal Equity Offering). In China, the two
Offerings are the primary source of additional financing for all Chinese listed companies (Chen et al. 2008).
~6~
analysts’ earnings forecasts after SOX. Bhojraj, Hribar, Picconi, and McInnis (2009) find companies that
use both methods to beat earnings forecasts exhibit short-run stock price advantages at the expense of
worse operations and stock performances in subsequent three years. Gunny (2010) reports that companies
using RAM to meet thresholds will have operational performance advantages over those that do not.
Cohen and Zarowin (2010) find that companies using both methods during seasonal equity offerings will
use more RAM after SOX. Zang (2012) finds that executives make tradeoffs between these two methods.
Finally, Li et al. (2011) document that the Chinese listed companies use both methods in order to meet the
threshold levels set by the Chinese governments for funding and continuing operations.
Although RAM is part of clients’ business operations and hence subject to neither GAAP nor
GAAS, one cannot conclude auditors are irrelevant or unconcerned. In fact, both audit fees and qualified
audit opinions are always influenced by clients’ business operations.9 A large number of experimental and
archival studies has examined whether audit judgments and decisions are improved by specialization and
leadership in client industries (see, e.g., Simunic 1980, 1984, Craswell, Francis, and Taylor 1995,
Solomon, Shield, and Whittington 1999, Taylor 2000, Ferguson and Stokes 2002, Ferguson, Francis, and
Stokes 2003, Low 2004, and Wu 2006) and portfolio selection of clients (see, e.g., Simunic and Stein
1987, 1990, and Johnstone and Bedard 2003, 2004). GAAS, SOX, and COSO stipulate that auditors bear
undeniable responsibilities for assessing and influencing clients’ financial reporting efficiency and clients’
internal control quality (see, e.g., Nelson, Elliott, and Tarpley 2002, Nelson and Tan 2005, Knechel,
Rouse, and Schelleman 2009, and Francis 2011). More recently, studies begin examining the impacts of a
company’s financial reporting decisions (e.g., material weakness disclosures) on the company’s investing
and operating decisions. For example, Bratten, Causholli, and Myers (2014) consider whether the use of
fair value accounting by banks influences the relative costs and effectiveness of earnings management
tools over their managerial decisions. Cheng, Dhaliwal, and Zhang (2013) consider the investment
9 We had a candid conversation with more than 30 experienced partners of the international Big Four accounting
firms and many domestic auditing firms. They told us that clients’ abnormal operating decisions were a major
concern of theirs. They administered particular attentions to those clients receiving ST or PT (see the previous
footnote).
~7~
behaviors of companies that disclosed internal control weaknesses under SOX. Feng, Li, McVay, and
Skaife (2013) investigate whether ineffective internal control over external financial reporting impacts a
company’s operations. Bauer (2012) considers the relations between tax avoidance and weak internal
controls. Audits create value because they assure the creditability of accounting information, thereby
improving the efficiency of resource allocation (DeFond and Zhang 2014).
Statement on Auditing Standards 90 requires that auditors judge “the quality, not just the
acceptability, of the accounting principles applied in the financial statements.” No. 14 requires auditors to
evaluate potential bias in corporate executives’ judgments. Furthermore, the Chinese Auditing Standards
(CAS) requires CPAs in China to undertake analytical procedures in risk evaluation at the beginning of
auditing and reviewing processes, although it is optional for substantive testing. Analytical procedures are
used to compare both financial and non-financial data with other companies’ in the same industry, as well
as annual and seasonal data of the same company. Analytical procedures consist of investigating
unexpected or inconsistent fluctuations in financial and non-financial data.
Individual auditors failing to perform the audits according to the CAS will be penalized and
sanctioned in China (Firth, Mo, and Wong 2005, 2012). In 1994, the Chinese Securities Regulatory
Committee (CSRC) issued “Provisional Rules and Regulations on Administration of Stock Issues and
Trades” (PRASIT). The Securities Law (1999, 2005) provides the legal basis for CSRC to supervise all
listed companies and market intermediaries such as audit firms, engagement auditors, securities brokers,
lawyers, etc. Article 35 of PRASIT requires “audit firm and engagement auditor should provide the fair
audit report when fulfilling their duties.” Article 37 states: "the engagement partners will be sanctioned by
CSRC if their audit report contains fraudulent, serious misleading contents or misstatements. Penalties
include warnings, fines, suspensions, disqualifications, and imprisonments.” In 1998, CSRC sanctioned
Cheng Du Shu Du for failing to report its client's (PT Hong Guang) fictitious sales and inflated inventory,
and banned the two engagement partners for life. During 1998 and 2013, CSRC penalized 128 individual
partners (fined 62 and banned or disqualified 66).
~8~
2.2 Research Questions
Previous studies show that people, rather than organizations, make decisions. The personalities of
decision makers affect decision outcomes (Kachelmeier 2010, Bamber, Jiang, and Wang 2010, and
Dyreng, Hanlon, and Maydew 2010). In the framework of audit production (Knechel et al. 2009 and
Francis 2011), audit engagement partners put their labor into auditing processes to gather evidence,
thereby affecting audit results. Individual auditors’ effort, experience, risk preference, and incentive will
change their attitudes toward and in turn, their learning about clients’ business models. Therefore, the
identity disclosure of auditors in China provides a natural experimental setting to examine whether
individual auditors respond to and influence over clients’ economic decisions beyond they do at the
audit firm level.
We use the Chinese Stock Market and Accounting Research Database (CSMAR). In China, two
auditors of each audit engagement are required to sign their names on audit reports. We collect individual
auditors’ personal demographic characteristics such as age, gender, position in the audit firm, political
affiliation, education background, work experience, etc., see Subsection 3.3 for details. We also collect
audit fees relative to clients’ total assets (Table 1), GC, and MOD (Table 2). Finally, we compute clients’
RAM over the sample period between 2000 and 2012 and across the three categories of audit firms
(Table 3).
(Insert Tables 1, 2, 3 and Figures 1, 2A, and 2B, and 3 about here.)
Audit fees relative to clients’ total assets and qualified audit opinions are almost monotonically
decreasing (Figures 1, 2A, and 2B), but clients’ RAMs are not (Figure 3). The changes in clients’ RAM
cannot explain the declines in audit fees and qualified audit opinions. Thus, we predict that significant
changes in auditor-client relations must take place in China, which suggests that individual auditors rather
than audit firms respond to or influence clients’ economic decisions (for which clients’ RAM is a proxy).
Adopting the regression methodology of Bertrand and Schoar (2003), we assign an indicator
variable to each auditor. We investigate whether clients’ RAM varies across different auditors. More
~9~
specifically, we consider three related issues. (i) Model (1) considers whether clients’ RAM varies across
different auditors. (ii) Model (2) studies whether audit fees and qualified audit opinions help to explain
the variation in RAM. (iii) Model (11) considers whether auditors’ personal demographic characteristics
have additional power in explaining the fixed effect that individual auditors have on clients’ RAM.
3. Empirical Design
3.1 Multivariate Regression Models
Following Gul et al. (2013) and Bertrand and Schoar (2003, Model 1), we investigate empirically
the fixed effects of audit clients, audit firms, and individual auditors on clients’ unexpected changes in
real activities (RAM). The regression model is
𝑦𝑖𝑡 = 𝛼𝑡 + 𝛾𝑖 + λ𝐴𝐹 + λ𝐼𝐴 + 𝜷𝑿𝑖𝑡 + 𝜀�̃�𝑡. (1)
The dependent variable, 𝑦𝑖𝑡, represents clients’ RAM (see its definitions in the next subsection), its
subscripts represent Client 𝑖 and Year 𝑡. The independent variables include four indicator variables:
(1) 𝛼𝑡 is the coefficient on the Year indicator variable, and captures the year fixed effect;
(2) 𝛾𝑖 is the coefficient on the Client indicator variable, and captures the client fixed effect;
(3) λ𝐴𝐹 is the coefficient on the audit firm indicator variable, and captures the fixed effect of
audit firms (subscript AF means audit firms); and
(4) λ𝐼𝐴 is the coefficient on the individual auditor indicator variable, and captures the fixed effect
of individual auditors (subscript IA means individual auditors).
𝑿𝑖𝑡 is a vector of time-varying, client- and auditor-level control variables (see Table 4 for the definition
and descriptive statistics), and 𝜀�̃�𝑡 is the error term of the regression model.
It would be impossible to estimate the fixed effect of individual auditors if auditors changed
neither audit firms nor audit clients. For example, consider an auditor who was with the same firm and
clients. Then, the fixed effect of this particular auditor would be identical to the fixed effect of her audit
firm. Fortunately, we control and limit this problem in the following three ways. First, note that there are
two significant “waves” of changes in auditors during our sample period. The first wave refers to the
~10~
massive “influx” of auditors into China in the late 1990s, and the other refers to the “exodus” of auditors
ten years later. The majority of these auditor changes happened to the International Big 4 accounting
firms. Second, the “conversion” process from the central planning economy to a market economy in
China was a decade long. In this process, accounting firms in China also went through an unprecedented
level of mergers and acquisitions. Accounting firms changed their names almost yearly during the sample
period; individual auditors were constantly switching audit firms, branches, and clients. Third, and, most
importantly, we use the two selection criteria suggested by Gul et al. (2013) to construct our sample of
individual auditors (see details in Subsection 4.1).
In our sample, therefore, the average tenure (i.e., the number of consecutive years) of the auditor
with a client is three years and one month. The average tenure of the audit firm with a client is three years
and nine months (Panel B, Table 4). The average tenure of the audit office with the client will naturally
fall between the above two tenure numbers. As their gap is small, we assume that the audit office tenure is
not a significant explanatory variable and, hence, omit it from the regression. The city/office-level
findings of Gul et al. (2013, Panels C-E, Table 3) support this assumption. Then, we predict that audit
fees relative to clients’ total assets (RAFee) will contribute to the fixed effect of audit firms because audit
fees in China are set at the year’s beginning. We also predict that auditors’ propensity to issue qualified
audit opinions (GC and MOD) will contribute to the fixed effect of individual auditors. Thus, we extend
Model (1) by explicitly including RAFee, GC, and MOD as additional control variables,10