The Impact of Audit Committee Financial Expertise on Management Discussion and Analysis (MD&A) Tone

Abstract

This study examines whether financial expertise of audit committees affects the quality of textual information conveyed through the management discussion and analysis (MD&A) section of corporate annual reports. Our empirical results reveal that audit committee financial expertise, particularly that which is directly connected to accounting, curtails managerial opportunism in the form of upward management of MD&A tone. In addition, we find that the effect of financial expertise is more pronounced when the audit committee is more powerful or when audit committee members face higher litigation risks. Overall, this study highlights the importance of audit committee financial expertise in improving the quality of qualitative disclosures.

Acknowledgements

The authors are grateful for insightful comments from Hervé Stolowy (editor), two anonymous reviewers, Jun-Koo Kang, Soo Young Kwon, Byungjin Kwak, Choong-Yuel Yoo, Jong-Hag Choi, and the seminar participants at the 2017 Korean Accounting Association Annual Winter Conference.

ORCID

Jaywon Lee http://orcid.org/0000-0003-4633-7939

Notes

1 These studies explore the informativeness of qualitative disclosures in corporate SEC filings (Campbell, Chen, Dhaliwal, Lu, & Steele, 2014; Feldman et al., 2010; Li, 2008, 2010; Mayew, Sethuraman, & Venkatachalam, 2015; Muslu, Radhakrishnan, Subramanyam, & Lim, 2015), media articles (Tetlock, 2007; Tetlock, Saar-Tsechansky, & Macskassy, 2008), earnings press releases (Davis, Piger, & Sedor, 2012), analyst reports (De Franco, Hope, Vyas, & Zhou, 2015), conference calls (Larcker & Zakolyukina, 2012; Price, Doran, Peterson, & Bliss, 2012), and other sources.

2 Similarly, Braiotta, Gazzaway, Colson, and Ramamoorti (2010) argue that audit committees should pay more attention to the SEC’s regulations regarding MD&A disclosures and assess managers’ compliance with SEC requirements.

3 Following prior studies that distinguish between the different types of financial expertise (e.g. DeFond, Hann, & Hu, 2005), we use the term ‘accounting financial expertise’ to refer to the expertise, which is directly connected to accounting.

4 In a similar vein, DeFond et al. (2005) examine the stock market reaction to the appointment of audit committee experts. They find that the stock market reacts positively only when the appointed audit committee member has accounting financial expertise.

5 The IBM audit committee charter is available at https://www.ibm.com/investor/att/pdf/auditcomcharter.pdf.

6 The Credit Suisse audit committee charter is available at https://www.credit-suisse.com/media/assets/corporate/docs/about-us/governance/standard-and-policies/audit-charter-en.pdf.

7 This enables us to assure that our classification of MD&A tone matches with its corresponding audit committee financial expertise. Our results remain unchanged if we exclude all firm-years in which a company filed Form 10-K/A.

8 See Loughran and McDonald (2016) for a comprehensive review of textual analysis research in accounting.

9 Earlier studies employ word lists from various sources such as the General Inquirer, Diction Word Lists, and Linguistic Inquiry and Word Count to gauge the tone of textual information (Li, 2010; Tetlock, 2007; Tetlock et al., 2008). However, Loughran and McDonald (2011) show that applying these word lists in the context of financial communications may cause misclassification. For example, the General Inquirer classifies words such as cost, depreciation, and liability as having negative tone, although they are part of common terminology in annual reports. Adjusting the list because of these misclassified words, Loughran and McDonald (2011) provide a more appropriate dictionary for analysis of financial communications.

10 In the word list of Loughran and McDonald (2011), words such as ‘litigation’, ‘unpaid’, and ‘misconduct’ are classified as having negative tone and words such as ‘efficient’, ‘improve’, and ‘profitable’ are classified as having positive tone. The full list is available at http://www3.nd.edu/~mcdonald/WordLists.html.

11 Li (2010) and Huang et al. (2014) use the number of company’s business and geographic segments as their measures of business complexity. However, we use NumMiss as our primary measure of complexity because the inclusion of segment variables substantially reduces our sample size. When we use the segment variables, the number of sample firm-years used for normal MD&A tone estimation drops to 21,423. Nonetheless, the untabulated results using this smaller sample show that our results remain robust even if we include the segment variables as a measure of business complexity.

12 To estimate the analysts’ consensus forecast variable, we consider each analyst’s most recent one-year-ahead EPS forecasts up to two months after the fiscal year-end date.

13 In addition, we estimate a pooled regression using the full sample after controlling for industry and year fixed effects. We find that our inferences remain unchanged.

14 Following the SEC’s categorization of expertise, some prior studies also consider the financial expertise resulting from audit committee members’ extensive experience as financial statement users. However, we do not consider this type of financial expertise in our main analysis for the following reasons. First, recent studies raise concerns over the validity of measuring user financial expertise (Cohen et al., 2014; Krishnan & Lee, 2009; Krishnan & Visvanathan, 2008). Second, we mainly focus on accounting and supervisory financial expertise to add to the stream of literature which compares the differing effects of these two types of expertise (e.g., Krishnan & Visvanathan, 2008). In a later section, we discuss this issue further and show that our results remain robust even when we control for the audit committee’s experience as financial statement users.

15 We define an audit committee member as having accounting financial expertise if s/he currently serves or previously served in one of the following positions: chief financial officer, principal financial officer, financial executive, president of finance, chief accounting officer, principal accounting officer, accounting officer, certified public accountant, audit partner, controller, head of accounting, head of finance, VP of finance, VP of accounting, or chartered accountant. For the supervisory expertise variable, the chief executive officer, chief executive, group chief executive, principal executive officer, president, and group president are considered to provide supervision over financial statement preparation.

16 For example, the NYSE requires all listed firms to discuss the content of their MD&A sections with external auditors (Cohen et al., 2007). Glendening et al. (2014) document that auditors’ negative attitude toward CAE decreases the likelihood of their client firms’ inclusion of CAE in the MD&A section.

17 For example, Hoitash et al. (2009) report that the proportion of accounting financial experts (supervisory financial experts) ranges from 0.20 (0.33) to 0.22 (0.41), depending on the sample composition. In their sample, Badolato, Donelson, and Ege (2014) also show that the proportion of accounting financial expertise and supervisory financial expertise ranges from 0.270 to 0.294 and from 0.281 to 0.321, respectively.

18 An average firm has five directors on its audit committee. This implies that a 1 standard deviation increase in AFEpercent is equivalent to having one more accounting financial expert in an average firm’s audit committee.

19 Alternatively, our results could be interpreted as audit committee financial experts using the MD&A section as a tool for downward expectation management. Specifically, audit committee members may use their expertise to amplify the negative aspects of a firm’s business, increasing the likelihood of meeting or beating expectations in the future. To address this issue, we divide the final sample into two subsamples with positive and negative AbTone and estimate equation (2) separately for each. The untabulated results reveal that the coefficient on AFEpercent is significant and negative (coeff = −0.030, p < .10) for the positive AbTone subsample, while it lacks statistical significance for the negative AbTone subsample. This provides evidence that downward expectation management does not drive our main results. However, these results should be interpreted with caution, as the partitioned sample analyses may bias the estimation of coefficients.

20 Beck and Mauldin (2014) compare the tenure between audit committee members and the CFO. However, we focus on the relative tenure of audit committee members compared to the CEO, as the CEO is primarily responsible for the content of the MD&A section.

21 Constructing rank variables allows us to mitigate the concerns attributable to differences in average tenure of CEOs and audit committee members (Beck & Mauldin, 2014).

22 Following Kim and Skinner (2012), we estimate the litigation probability by running a logistics regression model as follows: $Pr\left(Lawsui{t}_t\right)={\beta }_0+{\beta }_{1}FP{S}_t+{\beta }_{2}LogAsset{s}_{t-1}+{\beta }_{3}SalesGrowt{h}_{t-1}+{\beta }_{4}MktRetur{n}_t+{\beta }_{5}RetSkewnes{s}_t+{\beta }_{6}RetST{D}_t+{\beta }_{7}Turnove{r}_t+\epsilon ,$ where the dependent variable Lawsuit is set to 1 if the company has experienced any class action lawsuit during year t, and 0 otherwise. For the independent variables, FPS_t is set to 1 if the company is in the bio-technology industry (SIC 2833-2836 and 8731-8734), computer industry (SIC 3570-3577 and 7370-7374), electronics industry (SIC 3600-3674), or retail industry (5200-5961), and 0 otherwise. LogAsset_t-1 is the natural logarithm of total assets at the end of year t-1. Sales Growth_t-1 is year t-1 total sales less year t-2 total sales deflated by the end of year t-2 total assets. MktReturn_t is the market-adjusted annual stock return for year t. If a company is being sued, the accumulation of stock returns ends with the lawsuit class period end month. If a company is not being sued, the accumulation of stock returns ends with the year-end. RetSkewness_t is the skewness of monthly stock returns for year t. RetSTD_t is the standard deviation of the firm’s monthly stock return during year t. Turnover_t is the trading volume accumulated over a twelve months period, deflated by end of year t-1 shares outstanding. If a company is not being sued, the accumulation period ends with the year-end.

23 Alternatively, we apply a caliper distance of 1% or 0.1%. The untabulated results reveal that the findings remain unchanged.

24 When we compare the mean values of covariates between the treatment and control groups, we find statistically significant mean differences for four out of thirteen covariates: FirmSize, AnalystFollow, NumDirector, and Outside.

25 The untabulated results show that none of the covariates are significantly different between the treatment and control groups.

26 Additionally, we reestimate the results using an alternative definition of accounting financial expertise. Prior studies find that CEOs can influence the process of appointment of a new director to the corporate board (Shivdasani & Yermack, 1999). If this is the case, it is probable that the decrease in AbTone and the appointment of an audit committee financial expert are simultaneously affected by the incumbent CEO’s characteristics. To address this issue, we consider an audit committee member to have financial expertise only when s/he joined as a member of the audit committee before the incumbent CEO was appointed. Under this alternative definition, the selection of an accounting financial expert as an audit committee member is less likely to be affected by the incumbent CEO’s characteristics, thereby mitigating the potential endogeneity problem. The untabulated results show that our main results do not change when we apply this alternative definition.

27 Section 407 of the SOX requires all companies, other than small business companies, to comply with the disclosure requirements in annual reports with fiscal year-end dates on or after 15 July 2003 (15 December 2003 for small business companies). To allow sufficient time for market participants to respond to the new requirements of the SOX 407, we use 1 January 2004 as the cutoff date for the pre- and post-SOX 407 periods.