dao et al - 2012 - shareholder voting on auditor selection, audit fees, and audit quality

Saul 1996, 135 suggests that having shareholder ratification ofthe auditor is ‘‘more than a symbolic act’’ in the context of strengthening auditor independence.Accordingly, our objective

Vol 87, No 1 DOI: 10.2308/accr-101592012

pp 149–171

Shareholder Voting on Auditor Selection,

Audit Fees, and Audit Quality

Mai Dao University of Toledo

K Raghunandan Dasaratha V Rama Florida International University

ABSTRACT:The Advisory Committee on the Auditing Profession (ACAP), formed bythe U.S Department of the Treasury, has recommended that all public companies berequired to have shareholder ratification of auditor selection Using data from 1,382 firmsfor the year ending December 31, 2006, we find that audit fees are higher in firms withshareholder voting on auditor ratification We also find that firms that started having ashareholder vote pay higher fees than firms that stopped having a shareholder vote Inthe second part of our study, we find that in firms with shareholder voting on auditorselection (1) subsequent restatements are less likely and (2) abnormal accruals arelower Our findings are consistent with the experimental results inMayhew and Pike(2004), and provide empirical grounding for the debate about mandating shareholdervoting on auditor selection

Keywords: auditor selection; shareholder voting; audit fees; audit quality

I INTRODUCTION

T he objective of this study is to examine the association between shareholder involvement in

auditor selection and (1) audit fees and (2) audit quality Motivation for this study comesfrom the U.S Department of the Treasury’s (DoT 2008) Advisory Committee on theAuditing Profession (ACAP), which recently recommended that all public companies must have anannual shareholder ratification of the external auditor The ACAP (DoT 2008, VIII:20) justified thisrecommendation by stating that that the annual submission of auditor selection for ratificationenhances ‘‘competition in the audit industry.’’ The ACAP did not provide any empirical evidence insupport of the above recommendation, and research related to shareholder involvement in auditorselection is sparse

We gratefully acknowledge many useful comments and suggestions from Steve Kachelmeier (senior editor) and two anonymous reviewers We thank Abhijit Barua, Vishal Munsif, and Paul Tanyi for help with data collection and analysis Editor’s note: Accepted by Steven Kachelmeier.

Submitted: October 2009Accepted: June 2011Published Online: August 2011

The ACAP did not elaborate on exactly what benefits would accrue from any increasedcompetition among auditors However, the traditional view of the courts, legislators, and regulatorshas been that competition in any sector leads to higher quality and lower price (U.S Supreme Court1978; Federal Trade Commission 2003) In contrast, economic theory suggests that the joint effect

of competition on price and quality is not uniform; the effect of increased competition on price andquality will depend on the relative elasticities of price and quality (Dranove and Satterthwaite 1992;Kranton 2003) Even within the context of professional services, prior studies examining differenttypes of services have shown divergent results about the joint effects of enhanced competition onprice and quality (Kwoka 1984; Haas-Wilson 1986; Rizzo and Zeckhauser 1992; Kessler andMcClellan 2000)

While the ACAP recommends shareholder voting on auditor selection based on argumentsabout increased competition, the same recommendation can be arrived at by using a governance andaccountability framework We argue that any arrangement that strengthens the role of theshareholders in auditor selection also changes the incentives of the auditors and strengthens auditorindependence

The Sarbanes-Oxley Act (SOX, U.S House of Representatives 2002) made audit committeesformally responsible for the selection and compensation of the external auditor; however, evidenceindicates that even in the post-SOX era, managers exercise significant control over the hiring andfiring of auditors (KPMG 2004; Cohen et al 2010) If managers retain significant influence overauditor selection, then auditors will be more likely to go along with the preferences of managers(Saul 1996) Hence, institutional arrangements that strengthen the role of shareholders in auditorselection will also strengthen auditor independence

Mayhew and Pike (2004) examine, in a laboratory market setting, the effects of alternativeauditor hiring rules on competition in the audit market They find that, in their experimental setting,investor involvement in auditor selection leads to both higher audit quality and an increase in auditfees Mayhew and Pike (2004, 820) note that ‘‘further research into the institutional structures thatpromote audit committee independence or produce the types of incentives for auditor independencedocumented in our investor selection treatments is clearly warranted.’’

The ACAP recommendation represents a step in the direction toward increased shareholderinvolvement in auditor selection Saul (1996, 135) suggests that having shareholder ratification ofthe auditor is ‘‘more than a symbolic act’’ in the context of strengthening auditor independence.Accordingly, our objective is to empirically examine the association between shareholderinvolvement in auditor selection and both audit fees and audit quality, using archival data To theextent that increased shareholder involvement in auditor selection alters the governance andaccountability dynamics between auditors, management, and shareholders, we find that shareholdervoting on auditor selection leads to both higher audit fees and better audit quality

We use data from a sample of 1,382 companies included in the 2006 edition of the BoardAnalyst database We find that audit fees arehigher in firms that submitted auditor selection for ashareholder vote, after controlling for other factors associated with audit fees This finding isinconsistent with the argument generally advanced by legislators, regulators, and the judiciary, but

is consistent with the experimental evidence in Mayhew and Pike (2004)

As noted by Hermalin and Weisbach (2003), endogeneity is a pervasive problem in anygovernance-related archival-empirical research For example, it is possible that the same factors thatare associated with some firms voluntarily having a shareholder vote on auditor selection, sincesuch a vote is not currently legally required, are also associated with higher audit fees We perform

a Hausman test of endogeneity between audit fees and shareholder voting on auditor selection, andfind that there is support for the two being endogenous Accordingly, we perform two-stageregression analysis and find that the positive relationship between shareholder involvement inauditor selection and audit fees persists

We also perform a different type of test, relying on changes in firms’ policies relating toshareholder voting on auditor ratification Specifically, we focus on firms that changed from nothaving a shareholder vote on auditor selection in 2005 to having such a vote in 2006, and viceversa Because auditor-selection-related issues are supposed to be the prerogative of the auditcommittee after SOX, we require that the firms selected for such change analyses do not have achange in the composition of the audit committee With this additional restriction, we have 54 firmsthat newly initiated an auditor ratification vote in 2006 and 45 firms that stopped having an auditorratification vote in 2006, and did not have a change in the membership of the audit committee Forthis subset of 99 firms, we perform the audit fee regression in the ‘‘changes’’ form We find that theindicator variable measuring change in auditor ratification policy is positive and significant,indicating that firms that initiated an auditor ratification vote paid higher audit fees compared tofirms that stopped having an auditor ratification vote.

While audit pricing is directly observable, audit quality is a more difficult construct Thequantity or quality of audit work is not directly observable, so researchers have used differentmeasures for audit quality In this study, we use subsequent restatements as our measure for auditquality A restatement implies that the previously filed financial statements are unreliable and,hence, the presence (or absence) of a restatement can be viewed as a direct measure of audit quality

We restrict our analysis to restatements with a negative effect on financial statements We find thatfirms with a shareholder vote on auditor ratification during 2006 were less likely to have asubsequent restatement of fiscal year 2006 financial statements

Hennes et al (2008) point out that not all restatements are the same, and suggest using stockprice reactions as a direct measure of problem restatements Accordingly, we further restrict ouranalysis to subsequent restatements that also had a negative stock price reaction With this analysis

we also find that firms with a shareholder vote on auditor ratification were less likely to have a laterrestatement that elicits a negative market reaction

As with audit fees, we conduct Hausman tests of endogeneity but cannot reject the hypothesis

of exogeneity between subsequent restatements and shareholder involvement in auditor selection.Unlike with audit fees, we do not perform the analyses with only the subset of firms changingauditor ratification policies because only four firms from each of the change groups had asubsequent restatement

As additional analysis, we use clients’ abnormal accruals as another measure of audit quality.Although using accruals measures as a proxy for audit quality is subject to many limitations (Ball2009), previous studies have used clients’ abnormal accruals as a measure of audit quality given thepaucity of observable measures related to audit quality (e.g., DeFond and Subramanyam 1998;Francis and Krishnan 1999) Following this tradition, we examine the association betweenshareholder voting on auditor selection and abnormal accruals We find that performance-matchedabnormal current accruals are lower in firms that submit auditor selection for shareholderratification A Hausman test indicates the presence of endogeneity, but the association betweenshareholder voting and accruals quality persists after controlling for endogeneity

In summary, we examine an issue that is currently of interest to regulators and the accountingprofession but lacks relevant archival evidence Research methods involve trade-offs, and thebenefit of an experimental setting such as the one used by Mayhew and Pike (2004) is that itenables researchers to control and manipulate factors of research interest while ignoring otherexternal factors that are present in the real-world setting Conversely, the benefits from archivalresearch are the use of more natural settings that currently exist and greater external validity.Ideally, results from multiple methods should complement each other and there is greaterconfidence in the inferences when different research methods yield the same basic findings Theresults from our archival tests validate the experimental results of Mayhew and Pike (2004) in aclosely related setting—that increased shareholder involvement in auditor selection leads to an

increase in both audit fees and quality—but are inconsistent with the traditional arguments oflegislators, regulators, and the judiciary in the U.S that enhanced competition will lead to lowerprices and higher quality.

Section II discusses the background This is followed by a description of hypotheses, researchmethod, data, and discussion of results related to the two issues examined in this study: audit fees inSection III and audit quality in Section IV Section V concludes with a summary and discussion

II BACKGROUNDShareholder Voting: Competition Perspective

Even when there were eight large international auditing firms, legislators had expressedconcerns about the (lack of ) competition in the audit market (U.S Senate 1976, 1977; U.S House

of Representatives 1985) Such concerns have become more pronounced as the Big 8 have nowbecome the Big 4 since 2002 For example, Section 701 of SOX mandates the General AccountingOffice (GAO) to study and report back to Congress about the concentration of the audit market andthe problems resulting from limited competition among public accounting firms The GAO (2003,

Highlights) concluded that ‘‘the significant changes that have occurred in the profession may haveimplications for competition and public company choice, especially in certain industries, in thefuture ’’ In a follow-up report, the GAO (2008, Highlights) stated that ‘‘the loss of another large firmwould further reduce large companies’ auditor choice and could affect audit fee competitiveness.’’Thus, it is clear that the extent of competition in the market for audit services for publiccompanies (and the resultant impact on fees) continues to be an issue of significant interest tolegislators and regulators Such concerns led to the U.S Department of the Treasury forming, inOctober 2007, an Advisory Committee on the Auditing Profession (ACAP) The ACAP held aseries of meetings during 2007 and 2008, and issued a report in October 2008, proposing sixrecommendations related to concentration and competition in the audit market Recommendation

#5 suggests the adoption of ‘‘annual shareholder ratification of public company auditors by allpublic companies ’’ and notes as follows (DoT 2008, VIII:20–21):

The Committee believes shareholder ratification of auditor selection through the annualmeeting and proxy process can enhance the audit committee’s oversight to ensure that theauditor is suitable for the company’s size and financial reporting needs This may enhancecompetition in the audit industry The Committee also urges exchange self-regulatoryorganizations to adopt such a requirement as a listing standard

Competition, Price, and Quality

From an economic perspective, increased competition generally leads to higher quality and/orlower price However, if price and quality are jointly determined, then the effect of increasedcompetition on price and quality is not clear-cut; the effects of increased competition on price andquality will depend on the relative elasticities of price and quality (Dranove and Satterthwaite 1992;Kranton 2003)

Nevertheless, the U.S Supreme Court (1978, 695) noted in a case brought by the NationalSociety of Professional Engineers: ‘‘The Sherman Act reflects a legislative judgment that ultimatelycompetition will produce not only lower prices, but also better goods and services Theassumption that competition is the best method of allocating resources in a free market recognizesthat all elements of a bargain—quality, service, safety, and durability—and not just the immediatecost, are favorably affected by the free opportunity to select among alternative offers.’’ Similarly,the Federal Trade Commission (2003, 1) states that ‘‘vigorous competition forces producers to

minimize costs and prices and to increase quality.’’ Thus, the legislative, judicial, and executiveviews appear to be that increased competition will lead to lower prices and higher quality.Empirical evidence on the joint effects of competition on price and quality is sparse and mixed(Kranton 2003; Kallapur et al 2009) Prior research shows that, when the ban on advertising wasremoved (i.e., competition increased), the price of optometric services declined without an adverseimpact on quality (Kwoka 1984; Haas-Wilson 1986) However, in the case of physicians’ services,the removal of the ban on advertising led to higher prices and an increase in at least some measures

of quality (Rizzo and Zeckhauser 1992) Similarly, Kessler and McClellan (2000) show that in thecase of hospitals, prior to 1991, higher levels of competition led to both higher costs and improvedquality.1

Shareholder Voting: Governance and Accountability Perspective

While the ACAP recommendation is based on the effects of increased competition, there is analternative perspective that is useful in the context of shareholder voting on auditor ratification.Until the enactment of SOX, client management was responsible for selecting the auditor UnderSection 301 of SOX, audit committees are directly responsible for the appointment, compensation,and oversight of the auditor

Accordingly, Section 301 of SOX is a recognition of the governance and accountabilityperspective of auditor selection Specifically, if management is responsible for hiring or firing theauditor and negotiating the audit fee, then it is more likely that the auditor will go along with thewishes of management (Saul 1996) Mayhew and Pike (2004, 798) note that the auditor’s ‘‘financialdependence on the client depends heavily on the client’s ability to hire and fire the auditor As aresult, the control over hiring and firing the auditor serves as the core incentive for auditors tomaintain or compromise their independence.’’ By removing the authority to hire and fire the auditorfrom management, and vesting such authority with representatives of shareholders, Section 301 ofSOX sought to shift the locus of power in the auditor-manager relationship—and thereby enhanceauditor independence

However, as noted by Mayhew and Pike (2004), in many companies, the board of directors isdominated by management This enables management to retain significant influence over auditor-related decisions, including the hiring and firing of auditors In these circumstances, auditors’natural response is to be more likely to go along with managements’ preferred accounting choices.Along these lines, a survey of audit committee directors and executives by KPMG (2004, 5) foundthat, even after SOX, ‘‘69 percent of respondents said the chief executive officer or the chieffinancial officer had the most influence over the compensation of the external auditor, and only 27percent thought the audit committee had the most influence.’’ In a later study based on interviewswith 30 audit partners and managers from the Big 4 firms, Cohen et al (2010, 752) report that

‘‘management continues to be seen as a major corporate governance actor and, contrary to the intent

of SOX, often the driving force behind auditor appointments and terminations.’’

The ACAP noted that, even though SOX gives the audit committees the responsibilities related

to the appointment and compensation by the auditor, shareholders should have a say in auditorselection and that ‘‘ratification allows shareholders to voice a view on the audit committee’s work,including the reasonableness of audit fees and apparent conflicts of interest ’’ (DoT 2008, VIII:20).Thus, shareholder voting on auditor ratification, while perhaps not as effective as direct shareholderselection of the auditor, is a step toward increasing the role of shareholders in the auditor selectionprocess Reflecting this view, shareholder activists have continued to press companies to let

1

This is the so-called ‘‘Medical Arms Race’’ hypothesis, and this logic has been used by the courts in litigation involving hospital mergers See, for example, U.S vs Carilion Health System, 892 F 2d 1042.

shareholders vote on auditor selection because this is considered to be a good corporate governancepractice (Krishnan and Ye 2005; Institutional Shareholder Services [ISS] 2007a, 2007b).

Support for this governance-and-accountability-based perspective comes from the actions ofthe Financial Reporting Council (FRC) in the United Kingdom The FRC established the MarketParticipants Group (MPG) in October 2006 to provide advice on ‘‘possible actions that marketparticipants could take to mitigate the risks arising from the characteristics of the market for auditservices to public interest entities in the United Kingdom ’’ (FRC 2006, 1) The MPG issued itsreport in October 2007 with a set of recommendations; Recommendation #8 notes that the FRCshould amend the rules ‘‘to include a requirement for the provision of information relevant to theauditor re-selection decision,’’ while Recommendation #10 states that ‘‘investor groups, corporaterepresentatives, firms, and the FRC should promote good practices for shareholder engagement onauditor appointment and re-appointments ’’ (FRC 2007, 10) The MPG report noted that therecommendations were ‘‘directed at improving the accountability of boards for their auditorselection decisions’’ (FRC 2007, 10)

Finally, we interviewed three audit partners from three of the Big 4 audit firms and asubcommittee chair of the ACAP about shareholder involvement in auditor selection Of the auditpartners, one is a regional managing partner, the second is an office managing partner of a largemetropolitan office, and the third is a regional leader for audit practice Two of the audit partnersand the ACAP member noted that shareholder voting on auditor selection led to slightly higherrisks for the auditor To paraphrase, the partners and the ACAP member noted that ‘‘everyoneexpects the auditor to receive 98 or 99 percent approval from the shareholders, so even if you get 90

or 95 percent approval, there are bound to be questions from the audit committee; in addition, even

if remote, there is a non-zero probability of a significant proportion of shareholders voting againstthe auditor.’’ Two partners also discussed anecdotal evidence about Ernst & Young, which had 38percent of Sprint shareholders voting against ratification in 2003, resulting in unfavorable mediacoverage.2To the extent that perceptions affect auditors’ judgments and actions, we can expect bothaudit quality and price to be higher when there is a shareholder ratification vote on auditor selection

III SHAREHOLDER VOTING AND AUDIT FEESRegulators in the U.S and U.K are considering recommendations to require all publiccompanies to have a shareholder vote on auditor selection Some recent studies have examinedissues related to shareholder ratification of the auditor, including characteristics of firms submittingauditor selection for a shareholder vote (Krishnan and Ye 2005) and the determinants ofshareholder votes for ratification (e.g., Raghunandan 2003; Hermanson et al 2009) However, there

is little archival evidence related to the effect of having a shareholder vote on price and/or quality inthe market for audit services This study seeks to fill this void, given the calls from regulators torequire companies to have such a shareholder vote

As noted earlier, the experimental evidence in Mayhew and Pike (2004) and the governance/accountability perspective suggest that shareholder voting on auditor selection leads to higher auditfees and improved audit quality Nevertheless, given the paucity of evidence on this issue, andgiven the stated opinions of the judiciary and the executive that increased competition is expected toresult in lower price and better quality, we do not make a directional prediction for our firsthypothesis Formally stated, our first hypothesis is:

H1 (null form): There is no association between shareholder involvement in auditor selection

and audit fees

2

Following the shareholder vote, Ernst & Young was replaced as the auditor for Sprint.

We use the following regression model to test H1:

LogðAuditFeesÞ ¼ a0þ a1LogðTotalAssetsÞ þ a2InvRecTA þ a3SqrtSegments

þ a4Foreign þ a5CurrentRatio þ a6Leverage þ a7ROA

þ a8GC þ a9ICW þ a10Big4 þ a11Initial þ a12VOTE þ error: ð1ÞThe variables are defined as follows:

Log(AuditFees)¼ natural logarithm of audit fees;

Log(TotalAssets)¼ natural logarithm of total assets;

InvRecTA¼ ratio of assets in inventory plus accounts receivable to total assets;

SqrtSegments¼ square root of the number of business segments;

Foreign¼ 1 if foreign income is reported, 0 otherwise;

CurrentRatio¼ ratio of current assets to current liabilities;

Leverage¼ ratio of total liabilities to total assets;

ROA¼ ratio of income before extraordinary items to total assets;

GC¼ 1 if the firm receives a going-concern modified opinion, 0 otherwise;

ICW¼ 1 if the firm has an adverse SOX 404 opinion, 0 otherwise;

Big4¼ 1 if the auditor is a Big 4 audit firm, 0 otherwise;

Initial¼ 1 if the audit engagement is a first- or second-year audit, 0 otherwise;

VOTE¼ 1 if the firm submits auditor selection for shareholder ratification vote, 0 otherwise.3

Starting from Simunic (1980), prior research has used various measures related to client size,complexity, and risk as control variables in audit fee models We use the same audit fee model as in

Raghunandan and Rama (2006) Following a long tradition in the auditing literature, we use the logtransformed audit fees,Log(AuditFees), as the dependent variable Log(TotalAssets) is employed tomeasure client size, and we expect the coefficient of Log(TotalAssets) to be positive Threevariables—InvRecTA, SqrtSegments, and Foreign—proxy for client complexity, and we expectpositive coefficients for these three variables

Five variables, CurrentRatio, ROA, Leverage, GC, and ICW, control for client financialcondition and internal control; the coefficients of CurrentRatio and ROA are expected to benegative, while the coefficients ofLeverage, GC, and ICW are expected to be positive A large body

of literature has shown that there is a Big N audit fee premium, so we includeBig4 and expect thecoefficient of this variable to be positive Prior research also suggests that audit fees are discountedfor initial years of audit engagements (e.g., Simon and Francis 1988; Whisenant et al 2003);however, recent research suggests that initial audit engagements have an audit fee premium in thepost-SOX period (Huang et al 2009), so we include Initial in the model but do not make aprediction about the sign of the coefficient

The independent variable of interest isVOTE The governance/accountability perspective andexperimental results of Mayhew and Pike (2004) suggest that the coefficient of VOTE should bepositive

Data

We start our sample with all U.S companies that are included in the 2006 version of CorporateLibrary’s Board Analyst database As part of our analyses of audit quality, we examine subsequentrestatements; since it takes some time before restatements are discovered and disclosed, using data

3

Continuous variables (used in this and any subsequent regression) are winsorized at the 1st and 99th percentiles.

from fiscal year 2006 enables us to examine subsequent restatements over a three-year period (untilthe end of 2009) Section 404 of SOX first became applicable for fiscal years ending on or afterNovember 15, 2004, and auditors have noted that there has been a steep learning curve related toaudits in the post-Section 404 period Hence, we restrict the analysis to the 2,084 companies withfiscal years ending December 31, 2006.4Consistent with prior research related to audit fees andbecause firms in financial sectors have different financial statement reporting formats (and aresubject to additional regulations), we exclude 602 firms with SIC codes from 6000–6999.

We obtain data about audit fees, audit opinions, and auditor changes from the Audit Analyticsdatabase, while financial data are from the Compustat database and 10-K filings available on theSEC’s website We manually collect data about submission of auditor selection for shareholderratification voting from proxy filings available on the SEC’s website After eliminating 100observations with missing data, our final sample includes 1,382 firms

Descriptive Statistics

Panel A of Table 1 provides descriptive statistics about the variables used in the audit feeregression model The mean (median) audit fees for the sample firms are $2.86 ($1.53) million Themean and median values for the control variables are generally in line with those found in other studies(e.g., Ghosh and Lustgarten 2006; Raghunandan and Rama 2006) that examine audit fees in the post-SOX period.5Seventy-five percent of the sample firms submitted auditor selection for shareholderratification in 2006 As seen in Panels B and C of Table 1, firms that submitted auditor selection forshareholder ratification tend to be larger, are more likely to be audited by a Big 4 auditor, and are lesslikely to have adverse Section 404 reports, going-concern modified audit reports, or auditor changes

We test for multicollinearity between the independent variables by examining the correlationmatrix (untabulated) We find that all of the bivariate correlations are less than 0.50 We also findthat the variance inflation factors for the variables in the regression model are all less than 2.0,indicating that multicollinearity does not cause problems with our inferences

Audit Fee Regression Results

Table 2 presents the regression results for the model with Log(AuditFees) as the dependentvariable The overall model is significant, and the adjusted R2of 71 percent is in line with prioraudit fee studies All control variables in the model (except forLeverage and Initial) are significantand have the expected coefficient signs The variable of interest,VOTE, is positive and significant(p , 0.05) The magnitude of the coefficient of VOTE is 0.085, indicating that companies

4

There are two reasons why we restrict the analysis to clients with a 12/31 fiscal year-end First, the Big 4 firms noted in submissions to the SEC in 2005 and 2006 that SOX 404-related work had a significant impact on audit fees and that there is a steep learning curve effect for SOX 404-related work; hence, a firm with a 12/31/2006 fiscal year-end would be in the initial phase of the third year of SOX 404 work, while a firm with a 2/28/2006 (6/30/ 2007) fiscal year-end would still be in the second year (later phase of the third year) cycle of SOX 404 We do not want to contaminate our sample by mixing firms that would have been in different stages of the learning curve The second reason relates to the change from AS 2 to AS 5; the PCAOB approved the change in May 2007 to use a more top-down approach, and this had a pronounced effect on audit work and fees; the PCAOB provided permission for immediate application of the standard.

TABLE 1Descriptive Data(n¼ 1,382)Panel A: Sample Statistics

Variable Mean Standard Deviation 25th Percentile Median 75th Percentile

VOTE¼ 1(n¼ 1,036)

p-value for t-test(Mann-Whitney U test)

VOTE¼ 1(n¼ 1,036)

p-value(Chi-square test)

The p-values are two-tailed.

(continued on next page)

submitting auditor selection for shareholder ratification pay, on average, 9 percent higher audit fees.This result is consistent with the evidence in Mayhew and Pike (2004) that increased shareholderinvolvement in auditor selection is associated with higher audit fees.

Further Analyses: Endogeneity

Hermalin and Weisbach (2003) note that almost all research related to corporate governance issubject to the problems associated with endogeneity Thus, it is possible that the same factors thatare associated with greater shareholder involvement in auditor selection are also associated withhigher audit fees Krishnan and Ye (2005) find that audit committee characteristics are associatedwith the likelihood of shareholder voting on auditor ratification Prior research also shows that goodcorporate governance is associated with higher audit fees (e.g., Carcello et al 2002), possibly due tothe demand for higher quality or quantity of monitoring by auditors To the extent that goodcorporate governance—such as, an audit committee with a greater number or proportion ofaccounting experts—is associated with both greater shareholder involvement in auditor selectionand higher audit fees, the association documented earlier could reflect the same underlyinggovernance characteristics

Hence, we perform the following two-stage analysis In the first stage, in addition to the auditfee model discussed above, we use the following PROBIT regression model:

VOTE¼ c0þ b1LogðTotalAssetsÞ þ b2ACExpert þ b3DirectorVote þ b4Return

þ b5CEOCHR þ b6InsiderOwn þ b7Big4 þ b8Initial þ error: ð2ÞThe variablesVOTE, Log(TotalAssets), Big4, and Initial are as defined earlier Other variables aredefined as in Krishnan and Ye (2005), as follows:

ACExpert¼ proportion of audit committee directors who are accounting or auditing experts;DirectorVote¼ minimum percent of votes against the election of a director in the prior year;Return¼ prior-year common stock return minus the mean of two-digit SIC’s common stockreturn for the prior year;

CEOCHR¼ 1 if the CEO is also the Chair of the Board of Directors, 0 otherwise;

InsiderOwn¼ proportion of common stock held by officers and directors

TABLE 1 (continued)

This table provides descriptive statistics about variables used in the audit fee model The sample includes 1,382 firms that (1) are in the 2006 edition of the Corporate Library database, (2) are U.S firms, (3) have a December 31 fiscal year-end, (4) are in nonfinancial industries, and (5) have required data.

Variable Definitions:

Log(TotalAssets) ¼ natural logarithm of total assets;

InvRecTA ¼ ratio of assets in inventory plus accounts receivable to total assets;

SqrtSegments ¼ square root of the number of business segments;

Foreign ¼ 1 if foreign income is reported, 0 otherwise;

CurrentRatio ¼ ratio of current assets to current liabilities;

ROA ¼ ratio of income before extraordinary items to total assets;

Leverage ¼ ratio of total liabilities to total assets;

GC ¼ 1 if the firm receives a going-concern modified opinion, 0 otherwise;

ICW ¼ 1 if the firm has an adverse SOX 404 opinion, 0 otherwise;

Big4 ¼ 1 if the auditor is a Big 4 audit firm, 0 otherwise;

Initial ¼ 1 if the audit engagement is a first- or second-year audit, 0 otherwise; and

VOTE ¼ 1 if the firm submits auditor selection for ratification vote, 0 otherwise.

Our control variables are based on results from prior research Krishnan and Ye (2005) findthat the likelihood of having shareholder voting on auditor selection is (1) positively related toLog(TotalAssets), ACExpert, Return, and CEOCHR, and (2) negatively related to DirectorVote andInsiderOwn We include Big4 and Initial based on the evidence in Table 1 that firms with andwithout shareholder voting on auditor selection differ along these dimensions.6

TABLE 2Audit Fee Regression ResultsModel:

LogðAuditFeesÞ ¼ a0þ a1LogðTotalAssetsÞ þ a2InvRecTA þ a3SqrtSegments

þ a4Foreign þ a5CurrentRatio þ a6Leverage þ a7ROA þ a8GC

þ a9ICW þ a10Big4 þ a11Initial þ a12VOTE þ error:

p-values are two-tailed.

The sample includes 1,382 firms that (1) are in the 2006 edition of the Corporate Library database, (2) are US firms, (3) have a December 31 fiscal year-end, (4) are in nonfinancial industries, and (5) have all required data.

Variable Definitions:

Log(TotalAssets) ¼ natural logarithm of total assets;

InvRecTA ¼ ratio of assets in inventory plus accounts receivable to total assets;

SqrtSegments ¼ square root of the number of business segments;

Foreign ¼ 1 if foreign income is reported, 0 otherwise;

CurrentRatio ¼ ratio of current assets to current liabilities;

ROA ¼ ratio of income before extraordinary items to total assets;

Leverage ¼ ratio of total liabilities to total assets;

GC ¼ 1 if the firm receives a going-concern modified opinion, 0 otherwise;

ICW ¼ 1 if the firm has an adverse SOX 404 opinion, 0 otherwise;

Big4 ¼ 1 if the auditor is a Big 4 audit firm, 0 otherwise;

Initial ¼ 1 if the audit engagement is a first- or second-year audit, 0 otherwise; and

VOTE ¼ 1 if the firm submits auditor selection for ratification vote, 0 otherwise.

6 Table 1 shows that firms with and without auditor ratification also differ in terms of GC and ICW However, we do not include them in the model because these two variables are not known until the end of the year (after the decision to have a shareholder ratification vote).