Auditor tenure and perceived credibility of financial reporting PDF

Preview

Summary

Auditing: A Journal of Practice & Theory American Accounting Association Vol. 32, No. 1 DOI: 10.2308/ajpt-50308 February 2013 pp. 183–202 Audit Partner Tenure and Cost of Equity Capital Masoud Azizkhani, Gary S. Monroe, and Greg Shailer SUMMARY: We examine whether audit engagement partner tenure...

Description

Auditing: A Journal of Practice & Theory Vol. 32, No. 1 February 2013 pp. 183–202

American Accounting Association DOI: 10.2308/ajpt-50308

Audit Partner Tenure and Cost of Equity Capital Masoud Azizkhani, Gary S. Monroe, and Greg Shailer SUMMARY: We examine whether audit engagement partner tenure and rotation affect investors’ perceptions, as proxied by the ex ante cost of equity capital. We find that partner tenure has a nonlinear relation with the ex ante cost of equity capital for non-Big 4 audit engagements prior to the introduction of partner rotation requirements, and that the imputed gains from partner tenure appear similar to the imputed gains of having a Big 4 auditor. Consistent with the tenure results, we also find that partner rotation is associated with increased ex ante cost of equity capital. Our results are very robust to a variety of sensitivity tests and raise important questions for future research. It is not known to what extent investors or analysts are aware of the audit partner’s identity or pay attention to audit partner tenure; if investors or analysts do not consider partner tenure, future research may identify omitted variables that have the same nonlinear relationship with the ex ante cost of capital that we observe for non-Big 4 audit partner tenure. Keywords: audit partner tenure; audit partner rotation; financial reporting credibility; cost of capital. JEL Classifications: M42; M48. Data Availability: The data used are from public sources identified in the manuscript.

Masoud Azizkhani is a Lecturer at The Australian National University and an Assistant Professor at Ilam University, Gary S. Monroe is a Professor at The University of New South Wales, and Greg Shailer is an Associate Professor at The Australian National University. We are grateful to the anonymous referees and Robert Knechel for their constructive feedback on various versions of this paper. We also thank Jean Bedard, Jeff Cohen, Neil Fargher, Dominic Gasbarro, Michael Martin, Ted Mock, Terry O’Neil, Renee Radich, Roger Simnett, Philip Sinnadurai, Tom Smith, Stephen Taylor, Seng Teh, Farshid Vahid, Mark Wilson, Sue Wright, the participants at the 2007 AAA Midyear Audit Conference, the 2007 International Symposium on Audit Research, the 2007 Accounting and Finance Association of Australia and New Zealand Annual Conference, the 2006 Australian National Center for Audit and Assurance Research Symposium at Australian National University, and seminar participants at The University of New South Wales and Macquarie University for their useful comments on earlier drafts. We thank Elizabeth Carson at The University of New South Wales for her generous sharing of auditor opinions data. Editor’s note: Accepted by W. Robert Knechel.

Submitted: January 2010 Accepted: September 2012 Published Online: September 2012

183

Azizkhani, Monroe, and Shailer

184

INTRODUCTION

S

uccessive proposals by the PCAOB (2009, 2011) would require an audit firm to identify the name of the engagement partner in the audit report. The proposals suggest that this provides useful information to investors by allowing investors to identify audit partner tenure and rotation. Professional accounting bodies and regulators have long been concerned that long auditor-client relationships impair audit quality (e.g., AICPA 1978, 1992; ICAEW 2002; U.S. House of Representatives [Sarbanes-Oxley Act] 2002; General Accounting Office 2003; CGAA 2003). However, the PCAOB noted that other commenters, generally accounting firms and associations, do not believe disclosure of partner identity will provide meaningful information to investors (PCAOB 2011). These opposing views are not well informed because of the limited research available regarding audit partner tenure and rotation effects.1 This study investigates the usefulness of disclosing partner identity by exploring whether audit partner tenure and partner rotation are informative to investors, as indicated by the ex ante cost of equity capital. We focus on the cost of equity capital as an investor response because of its expected association with the credibility of audited financial reports.2 We focus on audit partner tenure and rotation because of the PCAOB proposal and because existing regulatory requirements regarding auditor tenure and rotation, focus on the audit engagement partner rather than the audit firm. Boone et al. (2008) examine the effect of audit firm tenure on the cost of equity capital, but the effect of audit-engagement partner tenure on the cost of equity capital has not been investigated. Bamber and Bamber (2009) argue against naively extrapolating the effects of audit firm tenure to audit partner tenure because the costs and benefits of changes may be substantially different. We try to disentangle the effects of audit partner tenure from audit firm tenure and audit firm switching effects. By comparing partner and firm effects within our data set, we assess whether the observed audit firm effects in the extant literature proxy for, dominate, or are incremental to partner effects. We use Australian data because audit reports in Australia have identified audit engagement partners for many years.3 The Australian setting is comparable to other developed Anglo-American jurisdictions, such as the U.S. and U.K.4 We find that partner tenure has a nonlinear relation with the ex ante cost of capital for non-Big 4 engagements prior to the introduction of partner rotation requirements, and that partner rotation is associated with increased ex ante cost of capital. We find the results are robust to a variety of 1

2

3

4

Bamber and Iyer (2007) report a significant association between individual auditor’s tenure and their identification with their clients. Manry et al. (2008) find that audit quality ( proxied by discretionary accruals) increases with audit partner tenure over seven years for smaller clients. Using Taiwanese data for audits prior to the mandatory rotation of audit partners, Chi and Huang (2005) and Chen et al. (2008) find that audit quality ( proxied by abnormal accruals) increases with audit partner tenure. Using Australian data, Carey and Simnett (2006) find that the probability of a going-concern qualification and just beating or missing earnings benchmarks are negatively associated with partner tenure, but they do not find a significant relation between partner tenure and abnormal accruals. Chi et al. (2009) report a positive association between audit quality and audit partner rotation after rotation became mandatory in Taiwan. In Australia, Fargher et al. (2008) find abnormal accruals tend to be smaller in years following partner rotation, while discretionary accruals are higher following an audit firm switch. We use the ex ante cost of equity capital because it has an expected association with the credibility of audited financial reports, excludes the irrelevant time-series property of earnings effects, and provides a more direct measure of information risk (Kothari 2001; Boone et al. 2008). The Corporations Act 2001 (Cth), section 324(10) (Commonwealth of Australia 2001) requires the audit report to be signed in both the audit firm’s and engagement partner’s names. Audit reports also identify partners in Taiwan; Chi and Huang (2005) and Chen et al. (2008) find that audit quality (abnormal discretionary accruals) increases with partner tenure. However, the Taiwan audit setting has some problematic differences to many other countries. Taiwan has a dual signature requirement, but does not distinguish who is the main audit partner, and the tenure of the two partners may differ. It is also suggests that the enforcement of laws is weaker in Taiwan than in the U.S. or Australia (Chen at al. 2008).

Auditing: A Journal of Practice & Theory February 2013

Audit Partner Tenure and Cost of Equity Capital

185

sensitivity tests, but it is not known to what extent investors or analysts are aware of the audit partner’s identity or pay attention to audit partner tenure and we cannot rule out that some omitted variable is driving our results. We discuss possible explanations for our results, but do not establish causal effects, and we argue that further research is needed to explain the detected relations. Future research could investigate whether analysts and investors are aware of and use partner identity and tenure, or identify omitted variables that have the same nonlinear relationship with the ex ante cost of capital that we observe for non-Big 4 audit partner tenure. The remainder of this paper is organized as follows. We describe our research method and data in the second section. Sample selection and descriptive statistics are in the third section. This is followed by the presentation of our main results in the fourth section. Our discussion and additional analyses are in the fifth section. In the sixth section, our conclusion emphasizes opportunities for future research. RESEARCH METHOD AND DATA We examine the relation between client-specific ex ante cost of equity capital and audit partner tenure using Model (1); and engagement partner rotation using Model (2): Rp ¼ f ðPT; Big4; LEV; B=P; Beta; Growth; Size; VAR; ROA; Loss; GoingConcern; QualifiedOther; EQ; Mills; Industry; YearÞ;

ð1Þ

DRp ¼ f ðPR; AFT; AFS; Big4; DLEV; DB=P; DBeta; DGrowth; DSize; DVAR; DROA; Loss; GoingConcern; QualifiedOther; Mills; Industry; YearÞ; ð2Þ where: Rp ¼ client-specific ex ante cost of equity capital estimated by the PEG model, expressed as a percentage, estimated at first analyst forecast following release of the audited financial statements; PT ¼ engagement partner tenure (PT), measured as the number of consecutive years that the audit report has been signed by the same audit partner;5 PR ¼ partner rotation, which is a dummy variable equal to 1 if the engagement partner changed in year t without a change in audit firm, otherwise 0; AFT ¼ audit firm tenure (AFT), measured as the number of consecutive years that the firm has retained the audit firm; AFS ¼ audit firm switch, which is a dummy variable equal to 1 if the audit firm changed in year t, otherwise 0; Big4 ¼ dummy variable equal to 1 if the audit firm is a Big 4 accounting firm in year t, otherwise 0;6 LEV ¼ financial leverage measured by the ratio of total debt to total assets at the end of the fiscal year; B/P ¼ ratio of book value of equity to market value of equity at the end of the fiscal year; Beta ¼ share beta (systematic risk) calculated over 28 months to the fiscal year-end; 5

6

Consistent with Johnson et al. (2002), we treat the auditor-client relationship as continuing when name changes merely reflect mergers. During our study period, Price Waterhouse and Coopers & Lybrand merged, and most Arthur Andersen partners joined Ernst & Young after the collapse of Arthur Andersen. During the period covered by this study, mergers of accounting firms meant the Big N audit firms change from the Big 6 to Big 4. For convenience, we refer to the Big 4 throughout this paper.

Auditing: A Journal of Practice & Theory February 2013

Azizkhani, Monroe, and Shailer

186

Growth ¼ earnings growth measured as the difference between the mean analysts’ earnings forecasts for four- and three-years ahead divided by the mean of three-years ahead earnings forecasts;7 Size ¼ size measured by the natural logarithm of the market value of common equity at the end of the fiscal year; VAR ¼ earnings variability measured by the standard deviation of analysts’ earnings forecasts available on I/B/E/S International during the fiscal year-end month; ROA ¼ return on assets calculated as the ratio of earnings before interest and tax divided by total assets; Loss ¼ 1 if the firm reported a loss, otherwise 0; GoingConcern ¼ 1 if the company received a going concern qualification in year t, otherwise 0; QualifiedOther ¼ 1 if the company received a qualified audit report for a reason other than going concern in year t, otherwise 0; EQ ¼ earnings quality measured as the absolute value of residuals using the Dechow and Dichev (2002) approach; Mills ¼ inverse Mills ratio based on Heckman (1979) to control for Big 4 selection bias; Industry ¼ dichotomous indicator variables based on two-digit SIC codes (24 industries) used by the ASX; Year ¼ dichotomous indicator variables to control for fiscal year; and D ¼ change in the corresponding variable from year t1 to year t. Ex Ante Cost of Capital We estimate client-specific ex ante cost of equity capital using the price-earnings ratio divided by the short-term earnings growth rate, which is the PEG model formulated by Easton (2004). This measure dominates the alternative measures as a firm-specific estimate of ex ante cost of equity capital (Botosan and Plumlee 2005).8 This is consistent with prior use of the PEG model to examine the relation between the cost of equity capital and earnings attributes (Francis et al. 2004, 2005), and to measure financial reporting credibility in relation to audit firm type (Khurana and Raman 2004; Azizkhani et al. 2010), auditors’ provision of non-audit services (Khurana and Raman 2006), and audit firm tenure (Boone et al. 2008).9 Auditor Tenure We test raw, log transform, and quadratic terms for auditor tenure. We find the quadratic terms provide the best fitting models, and our reported results use the quadratic terms.10 We also interact tenure with auditor type (Big 4 versus non-Big 4) because partner-tenure effects may be stronger 7

8

9

10

As part of our sensitivity tests, we examine whether the results are sensitive to short-term versus long-term measures of growth. To test this, we also measure expected growth as the difference between the mean analysts’ earnings forecasts for three- and two-years ahead divided by the mean of two-year ahead earnings forecasts and two- and one-years ahead divided by the mean of one-year ahead earnings forecasts. We also measure realized growth as the difference between actual earnings per share for the current and previous year, divided by actual earnings per share for the previous year. Our results are not affected by the alternative measures of growth. Botosan and Plumlee (2005) compare different models for estimating ex ante cost of equity capital, and show that the target price method and the PEG model dominate other alternatives in terms of providing estimates that are correlated with all other risk factors (such as stock beta) in a consistent and theoretically expected direction. Our approach does not imply analysts specifically incorporate partner identity or tenure into their forecasting models. As discussed later, awareness of partner identity or tenure may influence the effective pricing of forecasts by conditioning financial statement credibility. The quadratic relation is consistent with recent evidence (Chi and Huang 2005; Boone et al. 2008; Davis et al. 2009).

Auditing: A Journal of Practice & Theory February 2013

Audit Partner Tenure and Cost of Equity Capital

187

(weaker) for auditors from smaller (Big 4) firms because learning effects and the conjectured economic dependence and reduced independence may be less for Big 4 auditors, who have more resources and larger client bases. Controlling for Big 4 Selection Bias We use two-stage regressions (based on Heckman 1979) to address the potential for Big 4 selection bias (Raghunandan and Rama 1999; Johnstone 2000). Our first-stage uses the following logistic regression model for auditor selection from Chaney et al. (2004):11 Big4 ¼ f ðTAssets; Aturn; DA; CA=TA; QR; ROA; LossÞ;

ð3Þ

where: TAssets ¼ the natural log of year-end total assets; Aturn ¼ the natural log of asset turnover, calculated as sales divided by total assets; DA ¼ the ratio of long-term debt to total assets; CA/TA ¼ current assets divided by total assets; and QR ¼ current assets minus inventory, divided by current liabilities. ROA and Loss are as defined earlier. To conserve space, we do not report these first-stage results.12 Other Control Variables Leverage (LEV) is prevalent in bankruptcy risk models and previous studies indicate that the perceived risk associated with leverage is positively related to the cost of equity capital (e.g., Modigliani and Miller 1958; Fama and French 1992; Gebhardt et al. 2001). Prior research consistently documents a positive relation between book-to-price ratios (B/P) and cost of equity capital as proxied by average realized returns (e.g., Fama and French 1992; Khurana and Raman 2004; Boone et al. 2008). A firm’s cost of equity capital is positively associated with its systematic risk or Beta (Sharpe 1964; Botosan and Plumlee 2005). Earnings from growth opportunities (Growth) are riskier than normal earnings, suggesting a positive relation between growth and equity risk (Beaver et al. 1970). Consistent with prior empirical research (Berk 1995; Khurana and Raman 2004; Botosan and Plumlee 2005; Boone et al. 2008), we control for firm size, as measured by market capitalization. We include VAR because Barth et al. (1999) report that stable and increasing earnings result in lower risk premiums. We include ROA because reported earnings are expected to directly affect the cost of capital through both investors’ expectations of returns (Gebhardt et al. 2001; Gode and Mohanram 2003) and it has implications for the financial health of the firm, as signaled by its inclusion in bankruptcy risk models. We use Loss and GoingConcern, in addition to LEV and ROA, as indicators of bankruptcy risk. Earnings and leverage measures are long-established components in failure prediction models (e.g., Altman 1968; Grice and Ingram 2001), and going concern qualifications in audit reports are a direct warning of this risk. We control for audit qualifications that do not indicate going concern issues in case these qualifications proxy for other omitted conditions. Francis et al. (2005) show that firms with poor accrual quality have a higher cost of capital than firms with good accrual quality.13 11

12

13

Chaney et al. (2004) also include foreign sales as a percentage of total sales to control for the likelihood that exporting firms are more likely to hire a Big 4 auditor. We do not include this variable because the data are not available. The auditor selection model is estimated using the sample of 2,499 firm-year observations, which includes firms for which we are missing partner tenure and other data. Guay et al. (1996) and Subramanyam (1996) also suggest a relation between earnings quality and cost of equity capital.

Auditing: A Journal of Practice & Theory February 2013

Azizkhani, Monroe, and Shailer

188

Following Francis et al. (2004), we utilize the Dechow and Dichev (2002) measure of earnings quality.14 We control for the year and client-industry fixed effects. In Model (2) we control for audit firm switches (AFS) to benchmark the relation between partner rotation and DRp. SAMPLE SELECTION AND DESCRIPTIVE STATISTICS Sample Selection and Data Our sample is selected from all Australian-domiciled companies listed on the Australian Stock Exchange during 1995–2005 that are included in the I/B/E/S International database. We exclude cases that do not meet the PEG model constraint that eps2 . eps1 . 0 and that have extreme estimates of Rp.15 We determined audit partner tenure and rotations, and firm tenure and switches, by searching the published audit reports for engagement partners’ and audit firms’ names back to the later of when the firm first listed or 1950 (the earliest year for which audit report data are available). Data for the control variables were extracted from I/B/E/S International, Connect 4, Australian Graduate School of Management Annual Reports Files, the CRIF database, and Aspect Financial Analysis databases. We obtained 2,346 firm-year observations for which we can estimate the ex ante cost of capital and identify audit firm and partner tenure. Missing control va...