- UID
- 600097
- 在线时间
- 小时
- 注册时间
- 2011-1-24
- 最后登录
- 1970-1-1
- 主题
- 帖子
- 性别
- 保密
|
沙发

楼主 |
发表于 2011-8-14 21:12:39
|
只看该作者
【越障4-1】今天来个学术论文。。大家尽量看。。回忆大概框架就好了~
Is investment decision-making influenced byperceptions relating to auditors’ client dependence and amount of audit fees? ArnoldSchneider Collegeof Management, 800 W. Peachtree St., Georgia Institute of Technology, Atlanta,GA 30308 USA http://www.sciencedirect.com/science/article/pii/S0882611010000593 AbstractTheobjectives of this study are to examine whether investing decisions areaffected by knowledge about the auditor's revenue dependence on a client andwhether the amount spent by a company on audit fees affects decisions to investin the company. A behavioral experiment is conducted where risk assessments andinvesting decisions are made for four hypothetical investing scenarios. Thestudy finds that investing decisions are affected by knowledge about anauditor's revenue dependence on a client, but are not affected by knowledgeabout the size of a client's audit fees. Keywords: Investing;Client dependence; Audit fees; Auditor independence ArticleOutlineReferences Investorsoften rely on audited financial statements when making decisions to invest insecurities of companies. Prior research has documented that audit reportsgenerally influence the attractiveness of securities as investments (e.g., (Chow and Rice, 1982), (Loudder et al., 1992) and (Chen et al., 2000)). Investment decisions may beaffected by the perception of the auditor's independence from its client.1 Afactor which may impair this independence is the amount of audit fees receivedby the audit firm from a particular client in relation to its total revenues.If this proportion is high, there are concerns that the auditor may appease theclient because of the amount of revenues received from that client. Theseconcerns may reduce investors’ confidence in the audits that accompany thefinancial statements of the client. The main objective of this research is toinvestigate whether investing decisions are affected by perceptions of theauditor's revenue dependence on a client. The amountof audit fees that a client incurs in relation to its size may also affectinvestors’ perceptions of the audit. Greater amounts spent on auditing shouldresult in a higher level of audit effort, which can be perceived to beassociated with more reliable financial statements. A secondary objective ofthis research is to investigate whether the amount spent by a company onauditing affects decisions about investing in the company. Both ofthese issues have been examined in prior research studies. While Schneider (2010) investigated both of these issuestogether in a study on lending decisions, no research on investing decisionshas done so. These two issues have different implications for an audit and thefailure to treat them separately can produce effects that offset one another.Specifically, the presence of high audit fees cannot only create a perceptionof a high level of audit effort, but it can also lead to a belief that theauditor may be economically dependent on the client, which gives a perceptionof a lack of auditor independence. So, this study makes a unique contributionto the literature on investing decisions by separating these two issues (clientdependence and audit fee size) using a controlled experiment where twodifferent variables are employed to manipulate client dependence and the sizeof audit fees. In this research, the unit of analysis is the individual investordecision-making about equity investments as opposed to a market-based analysis.Individual investors merit study because: (1) they represent a significantsegment of the investment community and potentially impact stock prices (e.g.,DeLong, Shleifer, Summers, andWaldman (1991)); (2) the Securities andExchange Commission (SEC) is concerned about the welfare of individualinvestors, and; (3) knowledge about individual investor decision-making aids inthe development of theoretical models relating to investment markets (e.g., Daniel, Hirshleifer, andSubrahmanyam (1998)). In this study, four investing scenarios are constructed by havingtwo levels of revenue dependence (high vs. low percentages of client revenue inrelation to total audit firm revenue) as well as two levels of audit fees paid(high vs. low amounts of audit fees as a percentage of total company revenues).Risk assessments and two types of investing decisions made by participantsreveal that investing decisions are affected by the knowledge about theauditor's revenue dependence on a client. However, this research finds noevidence that investing decisions are affected by knowledge about the size of aclient's audit fees in relation to its total revenues. 1. Development of hypothesesRegulators (e.g., Wallman (1996))have expressed concern that an auditor may become economically dependent on aclient if the audit firm receives a significant portion of its total fees fromthe client. As Gul(1991: 164) states,“This may reduce the audit firm's ability to withstand pressure as a result ofan audit conflict, in turn impairing perceived independence”. Consequently,those who rely on audited financial statements might perceive that the auditorwill be inclined to resolve conflicts in favor of the client. Investors,therefore, might hesitate to invest in an audit client that receives a cleanaudit opinion if auditor independence is perceived to be impaired because ofeconomic dependence on that client. Researchstudies have examined various issues relating to client dependence and theresults have been inconclusive with respect to impairment of auditor independence. Chan,Lin, and Mo (2006) documentthat local auditors who have greater economic dependence on local clients tendto report more favorably on these clients. Hoitash, Markelevich, andBarragato (2007) examine fees paid to auditors andfind a significant positive relation between size-adjusted and abnormal totalfees paid to the auditor and two metrics involving accruals that were used tomeasure auditor independence. This suggests that auditors are reluctant tochallenge their important clients. The findings of Ahmed,Duellman, and Abdel-Meguid (2006) alsoreveal that auditors are more likely to grant concessions to influentialclients, but only those having relatively weak governance mechanisms. Unlike theabove studies, Craswell,Stokes, and Laughton (2002) obtainresults revealing that the level of auditor fee dependence has no impact on thelikelihood of an auditor issuing an unqualified audit opinion. Chungand Kallapur (2003) alsoshow that audit opinions are not associated with fees received from clients.Moreover, Reynolds and Francis (2000) find that larger clients of BigFive audit firms have lower levels of accruals compared to smaller clients andpotentially financially distressed larger clients are more likely to receivegoing concern opinions. These results lead them to conclude that auditorreputation protection concerns dominate economic dependence in makingdecisions. Hunt and Lulseged (2007) obtain similar findings fornon-Big Five audit firms, while Gaver and Paterson (2007) show similar results in a studyinvolving insurance companies. Li (2009) compares data after theSarbanes–Oxley Act of 2002 (SOX) to data prior to SOX and finds differences.Li's results reveal no relation between client importance and going concernopinions in 2001, but this relationship in 2003 was positive, in effectsupporting the findings of Reynolds and Francis (2000). On the other hand, the resultsof Ahmedet al. (2006) revealthat, for firms with weak governance, SOX has not been successful in mitigatingthe adverse effects of client importance on auditor independence. Even if anauditor's independence is not actually impaired in cases of influentialclients, investors who rely on audited financial statements may still perceivethat the auditor's independence is impaired. Research studying the perceptionsof financial statement users in lending and investing contexts generally hasfound that auditor independence is perceived to be impaired when there iseconomic dependence on a client. In a lendingscenario, Gul (1991) indicates that when a client'sfees are a significant portion of the audit firm's total revenue, lenders’perceptions of the auditor's ability to withstand management pressure areadversely impacted. Studies by (Firth,1980), (Bartlett, 1993) and (Lowe and Pany, 1995) also obtain similar findings ofnegative lenders’ perceptions under conditions of economic dependence. Inan investment setting, Pany and Reckers (1980) show that economic importance ofclients does not lead to impairment of independence as perceived by investors.On the other hand, Khurana and Raman (2006) find a significant positiveassociation between auditor fees (as a proportion of the revenues of either theaudit firm or the practice office through which the audit was conducted) andthe ex ante cost of equity capital, suggesting that client dependence isperceived negatively by investors. Asfor accountants’ perceptions about client dependence, prior studies generallysupport the claim that client importance adversely impacts auditor independence(e.g., (Firth, 1980), (Bartlett,1993), (Teoh and Lim, 1996) and (Beattieet al., 1999)).Research studies involving corporate finance directors (Beattie et al., 1999), financial analysts (Firth(1980)),and jurors (Brandon and Mueller, 2006) also report consistent findingsof perceived audit independence impairment when client fees represent a highproportion of total audit firm revenues. Whilethe evidence from studies dealing with accruals and audit opinions is mixed,the studies dealing with perceptions of financial statement users in bothlending and investing contexts have generally found that auditor independenceis perceived to be impaired when auditors are economically dependent onclients. This leads to the following hypothesis:
H1
Investing decisions will be adversely impacted when a company's auditorreceives a large percentage of its total revenues from that company. Theamount of audit fees paid by a company might also send a signal to investorsabout the audit. Large fees paid to auditors may increase the effort exerted byauditors (Hoitashet al., 2007).In turn, greater audit effort increases the probability that an auditor detectsproblems. There has been very little research on the implications of auditeffort. Using analytical methodologies, (Dye,1995) and (Hillegeist,1999) bothshow that greater audit effort increases the likelihood of detecting earningsthat are overstated. In a game theory setting, Hansenand Watts (1997) demonstratethat managers are less likely to manage earnings when audit testing costs arelow, but the study does not purport to measure audit effort. Caramanisand Lennox (2008)empiricallyshow that when audit hours are lower, companies are more likely to manageearnings upwards. Aperception of greater audit effort would seem to imply increased scrutiny ofthe client company. The increased scrutiny should produce a higher amount ofinvestor confidence in the presence of clean audit opinions. Furthermore, thehigher amount of audit fees may send a signal to investors that the companytakes the auditing activity seriously and that is why they are spending so muchmoney on the audit. Accordingly, investors should be more likely to makepositive investing decisions when companies spend greater amounts on audits.This leads to the following hypothesis:
H2
Investing decisions will be positively impacted when companies spend greaterproportions of their revenues on audits. 2.Variables and participants2.1.Independent and dependent variablesEconomicdependence (ED) and audit fee size (AF) are the two independent variables inthe between-subjects experimental design. Each of these variables ismanipulated with two levels. For the ED variable, participants were given oneof the following two scenarios: (1) the client's audit fee represented 1% ofits CPA firm's total local office revenue (no economic dependence; i.e.,ED = no), or (2) the client's audit fee represented 60% of its CPAfirm's total local office revenue (economic dependence exists; i.e.,ED = yes). These percentages are similar to ones used by (Brandon and Mueller, 2006) and (Schneider,2010). These figures are also consistentwith guidelines issued by the Institute of Chartered Accountants of England andWales, which state that the size of audit fee of a major client should notexceed 15% of the total fees to avoid impairment of auditor independence (Teoh and Lim, 1996).The descriptions in the case scenarios use local office rather than the totalfirm because, as Reynolds and Francis (2000) point out, the individual practiceoffice is the locus of contracting between the client and the firm. Thescenarios also described the CPA firms as non-Big Four because themanipulations of the independent variables are more realistic for those firms. For the AFvariable, participants were informed either that the audit fee for its mostrecent year was $2.5 million (AF = high) or $270,000(AF = low). The AF = high case mentioned that the size ofthe audit fee was not due to anything problematic, but rather that the companyjust desires thorough audits. This explanation was made to ensure thatparticipants unambiguously interpreted the high audit fee as a positive aspectof audit effort rather than as a negative perception about the company.Participants were given the client's income statement which showed revenuestotaling $100 million, so the AF = low case represents 0.27% ofrevenues, while the AF = high case represents 2.5% of revenues. Bothof these percentages are slightly higher than the range of .01% to 2% found by Francis (2004) ina sample of actual audit fees from companies that had average revenues of about$1.5 billion. Since these very large companies would probably have moreeconomies of scale in audit fees incurred than a company with revenues of$100 million, the slightly higher percentages of revenues for this studywould appear to be reasonable. Four scenarios result from these independentvariable manipulations and each of them is described in Appendix A. Theremaining information for each of the four cases was identical for allparticipants. They were instructed to evaluate a hypothetical electronicscompany for investment consideration. Participants were given information aboutthe company's products and services, three years of financial data, consensusanalyst forecast and recommendation, and historical stock prices for thecompany. There werethree dependent variables elicited from the participants: (1) an assessment ofthe risk associated with investing in the common stock of the company, (2) theprobability that they would invest at least 10% of a $1000 endowed amount inthe common stock of the company,2 and(3) the probability that they would retain this investment for at least oneyear.3 Althoughrisk assessments and investing probabilities would seem to be highly related,investors might have very different risk tolerances, and so while some mightinvest in risky companies, others might not. Moreover, decisions aboutpurchasing investments may be quite different than decisions about selling offinvestments. Hence, each of the three dependent variables may be capturingdifferent types of judgments. After giving responses to these three dependentvariables, participants then provided importance ratings for a set of eightfactors related to the investing decisions for the case scenario. Finally,demographic and manipulation check information were obtained from theparticipants. 2.2.Questionnaire respondentsA total of89 alumni and current graduate students at a U.S. university participated inthe study.4 Forthe alumni, emails with questionnaires attached were sent to recent graduates,who then returned completed questionnaires via email. For the current students,several business school faculty distributed questionnaires to their classes forstudents to complete on their own time and then collected the completedquestionnaires at subsequent class sessions. Theparticipants average 5.9 years of work experience, have an average age of29.1 years, 41% have a masters degree or higher, and 86% are male.Eighty-two percent of the respondents have sometime in the past purchased orsold stock investments. The distribution of respondents among the fourtreatment groups appears in Table 1. 3. Dataanalysis and findingsManipulationchecks were first conducted on each of the two independent variables to makesure that respondents distinguished between the two levels of each of theindependent variables. At the end of the questionnaire, all participants wereasked to rate the audit fee as a percentage of total revenues of the localaudit firm office (ED) and the size of the audit fee (AF) on a scale of1 = very low to 10 = very high. The average rating for EDwas 8.39 for the ‘yes’ group and it was 5.16 for the ‘no’ group. Thisdifference is statistically significant (t = 7.70;p < .001). The average rating for AF was 7.41 for the ‘high’ groupand it was 5.37 for the ‘low’ group. This difference is also significant(t = 5.28; p < .001). Thus, the manipulation checks weresatisfied, so it can be concluded that for each of the two independentvariables, participants distinguished between the two levels that weremanipulated. The datafor the entire set of respondents reveal a mean risk assessment for theinvestment of 5.57 (1 = very low risk; 10 = very highrisk), a .30 mean probability of investing in the company, and a .62 meanprobability of retaining the investment for at least one year. Table 2 showsthe cell means for each of the four treatments—the risk assessment cell meansare presented in Panel A, the investing probability cell means are presented inPanel B, and the retention probability cell means are presented in Panel C. Since the three dependentvariables are correlated with one another, a MANOVA was first performed. Theresults reveal that the ED variable is statistically significant(Lawley–Hotelling statistic = .174; p = .004), while the AFvariable is not statistically significant (Lawley–Hotelling statistic = .010;p = .841). The MANOVA was repeated with an indicator variabledistinguishing between the current students and alumni. The indicator variablewas not statistically significant and results for ED and AF were unchanged.These results are corroborated by separate ANOVAs for each of the independentvariables, as discussed next. Asexpected, the risk assessment variable for the economic dependence case(ED = yes) was higher than for the economic independence case(ED = no)—5.84 vs. 5.31. However, as shown in Table 3,Panel A, a 2 × 2 ANOVA reveals that this difference is notstatistically significant (F = 1.47; p = .230). For theprobability of investing as the dependent variable, the results are not only inthe expected direction (.20 for the economic dependence case and .40 for theeconomic independence case), but as shown in Table 3,Panel B, a 2 × 2 ANOVA is statistically significant for this variable(F = 8.04; p = .006). Likewise, the investment retentionprobabilities are in the expected direction (.53 for the economic dependencecase and .71 for the economic independence case) and Table 3,Panel C shows that this difference is statistically significant(F = 9.92; p = .002). These ANOVA were repeated with theage of the participants and their years of work experience as covariates, aswell an indicator variable distinguishing between current students and alumni,as but the results were unchanged. So, while economic dependence did not have asignificant effect on risk assessments, it did have a significant effect ondecisions relating to purchasing and retaining investments. Using the amount of audit fees (AF) as the independent variable, a2 × 2 ANOVA was also conducted for each of the three dependentvariables. As shown in Table 3, none of the three results for AF isstatistically significant. These ANOVA were repeated with the age of therespondents and their years of work experience as covariates, as well anindicator variable distinguishing between current students and alumni, but theresults are unchanged. Hence, the amount of audit fees paid had no impact onrisk assessments, investing probabilities, or probabilities of retaininginvestments. None of the interaction terms are significant at the .05 level. Asupplementary analysis was conducted which excludes the 16 participants withoutinvesting experience. ANOVA tables for each of the three dependent variablesare presented in Table 4. Results for both ED and AF remainunchanged—for ED, the investing and investment retention probabilities arestatistically significant, and for AF, none of the three dependent variablesare statistically significant. Again, none of the interaction terms aresignificant at the .05 level. Hence, for participants who have some previousexperience with investing, economic dependence appears to have an impact oninvesting decisions and retention decisions, but size of audit fees does notappear to influence risk assessments, investing decisions, or retentiondecisions.
Participants also rated theimportance of various factors in making their risk assessments and investmentdecisions, with a scale ranging from 1 = no importance to10 = very important. The average ratings appear in Table 5. The most important factors appear to bethe financial data and historical stock prices. The least important are thesize of the audit fee and standard industrial classification of the company.These importance ratings are consistent with one of the findings of thisresearch—that audit fee size does not impact investing decisions. The ratingsalso reveal that despite the finding that an auditor's revenue dependence on aclient affects investment decisions, this factor is not nearly as important asother factors such as financial data and historical stock prices. In fact,audit fee size and audit fee as a percentage of total firm revenues are two ofthe three lowest influential factors as perceived by the participants. 4.Summary and concluding remarksTheresearch findings reveal that investing decisions are affected by knowledgeabout an auditor's revenue dependence on a client. The data for the full set ofparticipants showed that probabilities of investing and probabilities ofretaining investments are lower when the company's audit fees represent a highproportion of its audit firm's total revenues. This also held true whenanalyzing data from respondents with prior investing experience. These findingssuggest that investors should be provided with information about the audit feescompanies pay in relation to the total revenues earned by their audit firms becausethis information influences investors’ probabilities of investing in thesecompanies as well as probabilities of retaining such investments. The resultsalso have implications for companies, auditors and regulators. Companieswishing to attract and retain investors should consider their choice ofexternal auditors, i.e., they may want to avoid engaging auditors who might beperceived as economically dependent on them. Likewise, audit firms that areconcerned with perceptions of their independence in the investment communitymay want to be careful about accepting clients that would create perceptions ofrevenue dependence on that client. As well, regulators may wish to considerways of influencing audit firms’ client acceptance decisions to avoid situationswhere revenue dependence on a client might impair perceived auditorindependence. Incontrast to the results for revenue dependence, this research found no evidencethat knowledge about the size of a company's audit fees affects riskassessments of investments, probabilities of investing, or probabilities ofretaining investments. Hence, information that more money has been spent onauditing, which presumably results in more audit effort, does not appear tolead to perceptions of more reliable financial statements by investors. The usuallimitations of behavioral experiments apply to this one as well. Most notably,investors often obtain more information about a company than was provided inthe research questionnaire. Another limitation is that the research wasconducted under one company setting only and therefore is not necessarilygeneralizable to other settings. Therefore, future research should investigatethe impact of audit fees and client revenue dependence with company scenarioshaving alternative characteristics as to industry, competitive environment,risk, financial performance, and historical stock prices. As well, economicincentives such as losing money from poor investing decisions are absent in thestudy. The participants did not actually put any capital at risk in thisexperiment. Future research can go beyond examining the investing intentionsdone in this study by capturing investment decisions in an experimental marketsetting where the participants’ compensation is based on the quality of theirinvestment decisions. There mayhave also been a potential confounding in the manipulation of economicdependence which could affect the interpretation of the results. Although theaudit firms were described to all participants as non-Big Four CPA firms, thedifferent sizes portrayed may have sent different signals about audit quality.Specifically, the sizes of the audit firms in the ED = no versionwere $250 million and $27 million while the sizes of the audit firmsin the ED = yes version were $4.17 million and $450,000. So, analternative explanation of the findings could be that investment decisions areaffected by auditor size (perhaps a proxy for audit quality), i.e., forcompanies audited by smaller audit firms, investment decisions were adverselyimpacted, as compared to companies audited by larger audit firms. Futureresearch should test the effects of audit firm size separately from economicdependence. Anotherlimitation relates to the interpretation of the audit fee manipulation. Toensure that participants did not ascribe the high audit fee manipulation to ahigh risk environment, they were told that the audit fee size was not due toanything problematic, but instead the company just desired thorough audits.Since this ambiguity might exist only for those participants in the high auditfee manipulation, participants in the low audit fee manipulation were not toldanything about the company desiring thorough audits. This information about thecompany desiring thorough audits may have also sent a signal about thecorporate governance, and in line with this type of culture, one might expectthe company to have higher quality financial statements. So, one cannot ruleout the possibility that the audit fee manipulation may have also manipulated “toneat the top.” Future research should investigate this alternative explanation. |
|