• No results found

Audit quality and non-GAAP earnings, related or not? : an analysis whether the presence of a big 4 auditor influences non-GAAP earnings

N/A
N/A
Protected

Academic year: 2021

Share "Audit quality and non-GAAP earnings, related or not? : an analysis whether the presence of a big 4 auditor influences non-GAAP earnings"

Copied!
43
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

MSc Accountancy & Control, variant Accountancy

Faculty of Economic and Business, University of Amsterdam

Master Thesis:

Audit Quality And Non-GAAP

Earnings, Related Or Not?

An Analysis Whether The Presence Of A Big 4

Auditor Influences Non-GAAP Earnings

Final Version

Candidate: M.W. Blokdijk Student Number: 10214534

Date: June 20, 2015

Supervisor: Dr. Réka Felleg

Word Count: 15,854

(2)

Statement of Originality

This document is written by student Marcel Blokdijk who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources others than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Abstract

I examine whether the presence of a Big 4 auditor, as a proxy for audit quality, affects the reporting of non-GAAP earnings in a way that firms audited by Big 4 auditors use less opportunism in non-GAAP earnings. Next to that I examine if the non-GAAP earning of firms audited by Big 4 auditors are more useful for investors. I first confirm that non-GAAP earnings are higher than GAAP earnings. Second, I find no statistical evidence that the difference between non-GAAP and GAAP earnings is higher for firms audited by non-Big 4 firms, however there is indicative evidence. Third, I find that firms audited by Big 4 auditors use less discretion in their non-GAAP earnings by excluding more nonrecurring and less recurring items to create their non-GAAP earnings. Fourth, I find that non-GAAP earnings of firms audited by Big 4 auditors contain more useful information for investors. These results confirm my expectation that audit quality has a positive effect on non-GAAP earnings.

Acknowledgement: I am grateful to Réka Felleg at the University of Amsterdam, for

(3)

Content

1. Introduction ……….. 4

2. Theory and Hypothesis Development ………... 8

2.1. Incentives to meet or beat earnings thresholds ………... 8

2.2. Non-GAAP earnings ………... 9

2.3. Audit Quality Differences ………... 12

2.4. Hypothesis Development ……… 12

3. Sample and Methodology ………. 15

3.1. Sample ………. 15

3.2. Methodology ……… 17

3.2.1. Proxy for non-GAAP earnings ……….. 17

3.2.2. Confirming non-GAAP earnings exceed GAAP earnings ……… 18

3.2.3. Meeting and Beating Analyst Forecast ……….. 19

3.2.3.1. Expected and Unexpected non-GAAP exclusions ………. 20

3.2.3.2. Controlling for other method of meeting and beating analyst forecast ………... 21

3.2.4. Usefulness of non-GAAP earnings ……… 22

3.2.4.1. Relative Usefulness ……… 22

3.2.4.2. Incremental Usefulness ……….. 22

3.2.4.3. Predictive Power ……… 23

4. Results ………. 25

4.1. Descriptive Statistics and Univariate Tests ……….. 25

4.2. Correlations ……….. 26

4.3. Non-GAAP earnings are higher than GAAP earnings ………. 29

4.4. Difference non-GAAP earnings and GAAP earnings between Big 4 and non-Big 4 audited firms ……… 29

4.5. Meeting and beating analyst forecasts with non-GAAP earnings ……… 31

4.6. Usefulness of non-GAAP earnings ………... 34

4.6.1. Relative Usefulness ………. 34

4.6.2. Incremental Usefulness ………... 35

4.6.3. Predictive Power ………. 36

4.6.4. Usefulness of non-GAAP earnings in context with other findings ….... 37

5. Conclusion ………... 38

(4)

1. Introduction

I am examining whether the presence of a Big 4 auditor, as a proxy of audit quality, influences the opportunistic reporting in non-GAAP earnings. Specifically, I examine whether firms audited by a non-Big 4 auditor are more likely to use non-GAAP earnings to meet or beat analyst forecasts than firms audited by a Big 4 auditor. In addition, I examine if the non-GAAP earnings from firms audited by a Big 4 auditor are more useful for investors than firms audited by a non-Big 4 auditor. To do that I examine the relative usefulness, the incremental value difference between non-GAAP earnings for firms audited by Big 4 auditors and non-Big 4 auditors and the predictive power of non-GAAP earnings to predict next year’s earnings per share.

My research is motivated by 1) the incentive of managers to meet or beat earnings thresholds and use non-GAAP earnings opportunistically to do this, as well as 2) the incentive of Big 4 auditors to protect their reputation of providing high quality audits.

The incentive of managers to meet or beat an earnings threshold like an analyst forecast is to realize a short-window gain. The reason for that is that lower earnings quality firms that meet or beat an analyst forecast have better stock returns than firms who miss an earnings threshold. Next to that, not meeting an analyst forecast causes a larger negative effect than the positive effect of meeting or beating an analyst forecast, which is another incentive to meet or beat analyst forecast (Bhojraj et al, 2009; Skinner and Sloan, 2002). Meeting and beating analyst forecasts can be done by using earnings management and expectation management. However, Doyle et al. (2013) show that the use of non-GAAP earnings is also a way to meet or beat analyst forecasts. Critics of non-GAAP earnings suggest that non-GAAP earnings are used by managers for creating a more positive image than it is in reality and thereby misleading investors. Firms do this by excluding items that should have been included or vice versa (Bhattacharya et al., 2003). Several studies provide empirical evidence for this: Black and Christensen (2009) show that firms exclude recurring items in the non-GAAP earnings to meet or beat strategic earnings targets. Van Raak (2005) provides evidence that non-GAAP earnings are used for opportunistic reasons. Next to that, he shows that GAAP earnings are more informative for investors than non-GAAP earnings. The misleading of investors through the use of non-GAAP earnings is investigated by Elliot (2006) and Frederickson and Miller (2004). They find that nonprofessional investors are misled by non-GAAP earnings, which causes them to assess the earnings performance to be higher and invest larger amount in the firms. It also affected nonprofessional investors’

(5)

stock price judgement: when non-GAAP earnings are disclosed at a higher value than GAAP earnings, they value that stock higher. Professional investors are not influenced by non-GAAP earnings because of their sophistication in calculating the value of a firm.

Auditors reduce the information asymmetry between managers and stakeholders by providing reasonable assurance about the information provided. Big 4 auditors are perceived to provide higher quality audits than non-Big 4 auditors (Khurana and Raman, 2004). Prior studies argue that Big 4 accounting firms have established brand name reputations for

providing high quality audits and therefore have incentive to protect their reputation (Francis, 2004).

Despite the fact that non-GAAP earnings are non-GAAP disclosures, auditors are still responsible for them (SAS 8). Under the provision of SAS 8 auditors are responsible for ensuring that a firm does not release overtly misleading information to investors. This means that auditors are required to review voluntary disclosures and prevent misleading or

opportunistic reporting of non-GAAP earnings in voluntary disclosure. If non-GAAP earnings are misleading for stakeholders they represent a fraud risk factor. The auditors’ brand name reputation is at risk when they are associated with an aggressive client, which is a client who is reporting non-GAAP earnings in an opportunistic way.

While under SAS 8 auditors are responsible for voluntary disclosures such as non-GAAP earnings, non-non-GAAP earnings represent an alternative measure of performance and this alternative measure is unaudited. Non-GAAP earnings are primarily constructed at the discretion of the managers of the firm (Entwistle et al., 2006). If non-GAAP earnings are unaudited it can be argued that auditors do not influence non-GAAP earnings.

Prior literature on this topic, Chen et al. (2012), examined whether audit firms wanted to be compensated for the extra fraud risk of a client that aggressively reports non-GAAP earnings. They found that more opportunistic non-GAAP earnings, that is when non-GAAP earnings are higher than GAAP earnings, are associated with higher audit fees. This means that there is evidence that audit firms react to the higher fraud risk factor of opportunistic non-GAAP earnings reporting. Next to that, investors perceive Big 4 audit firms to provide higher quality audits than non-Big 4 auditors. Therefore, because Big 4 audit firms have a better brand name reputation and have more to lose, the fraud risk factor can possibly have a greater negative effect on Big 4 audit firms than on non-Big 4 audit firms. If Big 4 audit firms react to firms reporting their non-GAAP earnings more opportunistically, the firms audited by Big 4 auditors are reporting less opportunistic non-GAAP earnings than firms audited by non-Big 4 auditors.

(6)

Whether the quality of an auditor influences non-GAAP earnings is not examined before in the literature about non-GAAP earnings. Therefore, in this study I provide empirical evidence on this subject. To do this I examine whether the presence of a Big 4 auditor

influences opportunistic reporting in non-GAAP earnings. I answer this with examining the following research question: Are Big 4 auditors associated with less opportunistic non-GAAP earnings of clients than non-Big 4 auditors and therefore do their clients provide more useful information for investors? I expect that when a firm is audited by a Big 4 auditor the firm reports less opportunistic non-GAAP earnings. In this paper I use IBES actual EPS as a proxy for non-GAAP earnings following Doyle et al. (2013). In my analysis I examine opportunism in 4 steps. First, in line with Chen et al. (2012) I confirm that non-GAAP earnings are higher than GAAP earnings. Second, I examine if there is a difference between firms audited by a Big 4 auditor and firms audited by a non-Big 4 auditor in the difference between GAAP and non-GAAP earning. Following Chen et al. (2012), non-GAAP earnings exceeding GAAP earnings is one sign for opportunism. Third, I examine whether firms audited by non-Big 4 auditors define non-GAAP earnings to meet or beat analyst forecast, following the paper of Doyle et al. (2013). Fourth, I use the method of van Raak (2005) to examine whether the non-GAAP earnings presented by firms audited by a Big 4 auditor are more useful for investors than when a firm is audited by a non-Big 4 auditor. I do this by examining the predictive ability of GAAP and GAAP earnings. When there is more opportunism in the non-GAAP earnings I expect the predictive ability to be lower and when there is less opportunism I expect the predictive ability to be higher.

I hypothesize and confirm that non-GAAP earning are higher than GAAP earnings in the full sample. I also hypothesize that the difference between GAAP earnings and non-GAAP earnings is bigger when a firm is audited by a non-Big 4 auditor. For that hypothesis I do not find statistical evidence, however the actual difference is higher. I also hypothesize and confirm that there is more opportunism in the non-GAAP earnings of firms audited by non-Big 4 auditors to meet or beat analyst forecast. Firms audited by big 4 auditors exclude more nonrecurring items and less recurring items than firms audited by non-Big 4 auditors. Furthermore, I hypothesize and conclude that the non-GAAP earnings of firms audited by Big 4 auditors contain more useful information than firms audited by non-Big 4 auditors. These results show that while non-GAAP earnings are not audited, audit quality influences GAAP earnings. When there is higher audit quality there is less opportunism in the non-GAAP earnings, and non-non-GAAP earnings are more useful for investors.

(7)

My paper contributes to the literature about non-GAAP earnings in several ways. I show that, although non-GAAP earnings are unaudited, audit quality affects non-GAAP earnings in a way that when the audit quality is higher non-GAAP earnings are reported less opportunistic and therefore are providing more information. Furthermore, I confirm the findings of Doyle et al. (2013) that non-GAAP earnings is a substitute for earnings management to meet or beat analyst forecast.

These results are useful for the business, managers and investors. Managers who disclose non-GAAP earnings while they know that the audit quality on their firm is low signal opportunism in their non-GAAP earnings. Investors who pick up this signal might use this to evaluate the firm and this could affect the decision making process of investors. However, this is not examined in the literature and therefore might be interesting to examine in future research. Next to that, my paper provides evidence that there is more opportunism in the non-GAAP earnings of firms audited by non-Big 4 auditors. This is also interesting for investors when they evaluate a firm.

My results are also useful for regulators. The results are post regulation G, which was implemented to decrease opportunism in non-GAAP earnings, however my paper indicates that there still is opportunism in non-GAAP earnings and that audit quality affects

opportunism in non-GAAP earnings. Next to that, I confirm the findings of Chen et al. (2012), that non-GAAP earnings are a substitute for earnings management to meet or beat analyst forecasts. My finding can be used by regulators as an incentive to improve the regulations about non-GAAP earnings.

The paper proceeds as follows. Section 2 discusses the prior relevant literature and develops my hypothesis. Section 3 contains the sample selection and methodology. Section 4 presents the results and in section 5 I conclude.

(8)

2. Theory and Hypothesis Development

2.1 Incentives to meet or beat earnings thresholds

Managers are willing to sacrifice long-term value to meet short-term earnings objectives. The majority of managers would not invest into a project with a positive net present value if the project leads to the firm missing the quarter consensus forecast. Managers are willing to take actions like decrease discretionary spending, including R&D and advertising expense to meet an earnings threshold (Bhojraj et al, 2009). Graham et al. (2005) provide evidence for this by interviewing CFOs. They report that the CFOs believe that meeting and beating analysts’ forecast on earnings per share builds credibility and helps to maintain or increase their firm’s share price and that they are willing to make economic sacrifices to meet and beat the analyst consensus.

To meet an earnings threshold managers are using earnings management: accruals or real earnings management. They also use expectation management, lowering the expectations of analysts which causes the forecast to be lower and easier to meet or beat (Koh et al., 2008). Another way to meet or beat an earnings threshold is disclosing non-GAAP earnings that do meet the earnings threshold (Doyle et al., 2013).

The literature about meeting and beating analyst forecast finds the following empirical evidence. Bhojraj et al. (2009) find that firms that use earnings management to meet an earnings threshold have a short term benefit for meeting or beating the threshold. Despite the lower earnings quality the firms have because of the earnings management, they have similar or better stock returns than firms who have high quality earnings but miss their thresholds. This suggests that firms have a short term benefit for meeting or beating earnings thresholds. Furthermore, Kasznik and McNichols (2002) find empirical evidence that meeting or beating an earnings threshold is beneficial for the market premium of the firm and that if the firm also does this in subsequent years the premium increases. Next to that, Skinner and Sloan (2002) show that the negative effect of missing an earnings threshold are larger than the positive effects of meeting an earnings threshold. Therefore managers have incentives to not miss an analyst forecast.

Keung et al. (2010) find different results, in their paper they find that in the 2000’s investors see zero and small earnings surprise as a warning for earnings management and therefore penalize firms that report them. Keung et al. also find evidence that this skepticism towards the zero and small earnings surprises is justified. They find that the relation between

(9)

future earnings surprises and current earnings surprises is more negative when firms report a zero or small earnings surprise.

Lopez and Rees (2002) examine market reactions of firms meeting or beating analyst forecasts. They find that when firms meet or beat analyst forecast the market reacts with significant higher stock prices. And if they meet or beat the analysts’ forecast consistently they reward it even more. However, when a firm fails to meet analysts’ forecast the market assesses a substantial penalty, unrelated to the magnitude of the forecast error.

Cheng and Warfield (2005) examine the incentive of managers to meet or beat analyst forecasts. When managers are compensated on equity basis, stock-based compensation and stock ownership, they have an incentive for meeting or beating analysts’ forecast, because the market rewards this with a higher share price, which is an incentive for managers to earn more money. McVay et al. (2006) provide more evidence that managers have incentive to meet or beat the benchmarks, they find that managers manage earnings to meet an earnings threshold. When they do this, they can sell their shares for a higher price.

2.2 Non-GAAP earnings

Non-GAAP earnings is an alternative measure of firm performance from the performance according to GAAP. Non-GAAP earnings differ from GAAP earnings by the extra exclusions that are made by managers. Managers argue that they use non-GAAP earnings to provide more information for investors, which leads to better decision making for investors, because the information that non-GAAP earnings provide contain more useful information than GAAP measures. Opponents of non-GAAP earnings argue that non-GAAP earnings are used to manage investors’ perception of the firms’ performance. Enabling, for example, the firm to beat analysts’ earnings forecast or to turn GAAP losses into non-GAAP profits. The

opponents are not concerned about misleading analysts but more about misleading nonprofessional investors (Entwistle, 2006).

If managers are right, non-GAAP earnings should contain more useful information than GAAP earnings. The literature finds mixed results. Brown and Sivakumar (2003) compare value relevance of GAAP earnings and non-GAAP earnings. They measure three things to determine the value relevance, predictive ability, valuation and information content. They find that non-GAAP earnings score better on all three measures. They conclude that the reason for this is that non-GAAP earnings have less transitory items than GAAP earnings. Another reason is that managers seek to provide value relevant information to the market place. On the other hand, van Raak (2005) finds evidence that GAAP earnings are more

(10)

useful for investors. He provides this evidence by measuring incremental information and predictive ability of non-GAAP earnings.

Johnson & Schwartz (2005) conclude that investors do not consider non-GAAP earnings to be informative. They find no evidence that non-GAAP earnings are misleading investors, but they neither find evidence that investors react to non-GAAP earnings. Which means that, in their research non-GAAP earnings are not useful for investors.

Lougee and Marquardt (2004) aim to find if the information in non-GAAP earnings systematically varies with GAAP earnings. In their research they find evidence that non-GAAP earnings have greater relative and incremental information when the information in the GAAP earnings is low, or when GAAP earnings contain earnings surprises that are positive. However, they find no evidence that there is greater relative or incremental information content in non-GAAP earnings when prior GAAP earnings information is high.

One of the first papers to examine non-GAAP earnings is Bradshaw and Sloan (2002). Their sample consisted of firms in the period 1985-1997. For this period they documented an increase in the use of non-GAAP earnings, and a growing difference between GAAP and non-GAAP earnings. This difference is caused by an increase of expenses that are excluded in the non-GAAP earnings. Next to that, they provide evidence suggesting that the increased attention on non-GAAP earnings is driven by reporting strategies of firm managers. These results can be caused by two reasons. First, the growing difference can be caused by an attempt by managers to create a higher valuation for their firm by reporting higher non-GAAP earnings, which is opportunism. The second reason is that the increased emphasis on non-GAAP earnings is caused by managers that want to remove transitory items out of the earnings. This would make non-GAAP earnings an improved measure for determining future cash flows and provide investors with extra information.

The literature about non-GAAP earnings after Bradshaw and Sloan (2002) examined which one of the interpretations was the base for non-GAAP earnings reporting. The greater part of the literature finds evidence that non-GAAP earnings are used for opportunistic reasons. Bhattacharya et al. (2003) find for the period 1998-2000 that the expenses excluded in non-GAAP earnings are mostly recurring and that non-GAAP earnings result in a profit more often than GAAP earnings. Next to that, they find that non-GAAP earnings tend to beat analysts’ forecasts a lot more than GAAP earnings. All these points are indicators that firms use non-GAAP earnings for opportunistic reasons. Black and Christensen (2009); Chen et al. (2012) and Doyle et al. (2013) contribute on the literature about meeting and beating analyst forecasts, by examining this is periods pre and post SOX. They find similar results as

(11)

Bhattacharya et al. (2003). Literature about items excluded in non-GAAP earnings, Doyle et al. (2003); van Raak (2005); Brown and Sivakumar (2003); Lougee and Marquardt (2004), provide evidence that the items excluded in non-GAAP earnings are recurring. Next to that, they show that information provided by non-GAAP earnings is not more useful than GAAP earnings. Most of the papers about non-GAAP earnings provide evidence that non-GAAP earnings are used for opportunistic reasons. On the other hand, the literature about non-GAAP earnings enhancing investors’ information is limited to a few papers. Bhattacharya et al. (2003), Bradshaw and Sloan (2002) Brown and Sivakumar (2003) provide evidence that the incremental value of non-GAAP earnings is higher than GAAP earnings. However, these papers are one of the first papers to examine this subject. More recent studies (van Raak, 2005; Christensen, 2007; Black and Christensen, 2009; Doyle et al, 2013) show that non-GAAP earnings are used for opportunistic reasons.

The Securities and Exchange Commission (SEC, 2003) also noticed that non-GAAP measures were mostly used for opportunistic reasons and therefore implemented regulation G in 2003. The goal of Regulation G 1is to prevent misleading reporting in non-GAAP

earnings. The SEC wants to prevent misleading with non-GAAP earnings by mandating that when a non-GAAP measure is disclosed a disclosure of the most directly comparable GAAP measure is provided. Also a reconciliation between the non-GAAP and GAAP measure should be disclosed.

After the regulatory intervention of the SEC the literature abut non-GAAP earnings, provide empirical evidence of the effects of Regulation G (Heflin and Hsu, 2008; Kolev et al., 2008). First, after Regulation G there is a decline in non-GAAP earnings disclosures. Second, after the implementation of Regulation G there is a decline in the magnitude of GAAP – non-GAAP differences. Third, there is a decline in the probability firms disclose

1“Regulation G includes the general disclosure requirement that a registrant, or a person acting on its behalf, shall not make public a non-GAAP financial measure that, taken together with the information accompanying that measure, contains an untrue statement of a material fact or omits to state a material fact necessary in order to make the presentation of the non-GAAP financial measure, in light of the circumstances under which it is presented, not misleading Whenever a company that is subject to Regulation G, or a person acting on its behalf, publicly discloses any material information that includes a non-GAAP financial measure, Regulation G requires the registrant to provide the following information as part of the disclosure or release of the non-GAAP financial measure: a presentation of the most directly comparable financial measure calculated and presented in accordance with GAAP. And a reconciliation (by schedule or other clearly understandable method), which shall be quantitative for historic measures and quantitative, to the extent available without unreasonable efforts, for

prospective measures, of the differences between the non-GAAP financial measure presented and the most directly comparable financial measure or measures calculated and presented in accordance with GAAP” (SEC, 2003).

(12)

non-GAAP earnings to meet or beat analyst forecast, Fourth, exclusions in non-GAAP earnings are, on average, of higher quality after Regulation G. This indicates that Regulation G reduces opportunism in non-GAAP earnings.

2.3 Auditor Quality Differences

Francis (2004) discusses that audit quality is associated with the following elements, auditor independence, auditor knowledge, earnings quality, legal regimes and the incentives they create, and audit firm size. For my paper I focus on audit firm size. Francis also discusses that the risk of loss of reputation is one factor that improves audit quality for Big 4 firms. This implies that Big 4 auditors have a brand name reputation for providing high quality audits and that they want to protect this. They want to protect it because a reputation of providing high quality audits brings benefits. For example, Choi et al. (2008) provide evidence that Big-4 firms ask a premium for their audit services, which is an incentive to protect your

reputation, because no client wants to pay a premium if they think that the audit firm is not providing high quality audits. Next to that, Greenwood et al. (2005) find that professional service firms must generate a superior reputation to succeed. This means Big-4 audit firms have incentive to protect their brand name reputation.

DeAngelo (1981) is the first to examine differences in audit firm size. The literature about audit quality discusses that firm size should not affect audit quality, because if a professional standard is maintained the quality of an audit should be equal. DeAngelo argues that larger accounting firms have higher quality audits. The main point she argues is that bigger audit firms are less dependent on the profits of one client than smaller audit firms. Because of this a Big-4 auditor is more independent, which is one of the key factors for audit quality.

2.4 Hypothesis Development

Prior literature provided evidence that non-GAAP earnings are higher than GAAP earnings (Bradshaw and Sloan, 2002; Brown & Sivakumar, 2003; van Raak, 2005; Black and Christensen, 2009; Doyle et al, 2013). Because non-GAAP earnings are higher than GAAP earnings we know that the exclusions made to create non-GAAP earnings are expenses. Bradshaw and Sloan (2002) also explain that this difference is caused by more expenses that are excluded in the non-GAAP earnings.

Despite that Heflin and Hsu (2008) and Kolev et al. (2008) provide evidence that after the implementation of Regulation G the non-GAAP earnings decrease, because of the stricter

(13)

rules by the SEC, non-GAAP earnings are still higher than GAAP earnings. Next to that, Chen et al. (2012) states that when non-GAAP earnings are higher than GAAP earnings this is a sign of opportunism in the non-GAAP earnings. Therefore H1 aims to confirm that in my sample non-GAAP earnings are higher than GAAP earnings. By doing this I confirm that the non-GAAP exclusions are expenses, and that in my time period non-GAAP earnings are used opportunistically. The hypothesis is formally expressed, as follows:

H1: Non-GAAP earnings are higher than GAAP earnings.

Chen et al. (2012) provides evidence that audit firms react to the higher fraud risk opportunistic clients have. Higher GAAP – non-GAAP earnings differences indicate more opportunism, which indicates a higher fraud risk factor. Francis (2004) explains that Big 4 audit firms have a better reputation than non-Big 4 auditor, because of that Big 4 auditors have more incentive to protect their reputation. To protect their reputation Big 4 auditors do not want to be associated with firms reporting their non-GAAP earnings opportunistically because they are responsible for them under the provision of SAS 8. I expect that firms audited by a non-Big 4 report less opportunistically. Therefore I examine whether the difference between GAAP and non-GAAP earnings is lower for firms audited by a Big 4 auditor than for firms audited by a non-Big 4 auditor. The hypothesis is formally expressed, as follows:

H2: The difference between GAAP and non-GAAP earnings is lower for firms audited by Big 4 auditors.

Managers are willing to take actions, which are not beneficial for the firm in the long-term, like decrease discretionary spending, including R&D and advertising expense to meet an earnings threshold (Bhojraj et al, 2009). The incentives of managers to meet short-term earnings objective is to avoid negative consequences (Skinner and Sloan, 2002) and to obtain short-term gains (Bhojraj et al., 2009).

One way to meet or beat an earnings threshold is to use the discretion offered in non-GAAP earnings (Doyle et al., 2013). Under the provision of SAS 8 auditors are responsible for the disclosed statements by their clients. Therefore, when firm managers use the

discretion in non-GAAP earnings opportunistically to meet or beat an earnings threshold, the auditor is responsible. Auditors associated with clients that report non-GAAP earnings

(14)

opportunistically are affected in their reputation. Since Big 4 auditors have a better reputation than non-Big 4 auditors, Big 4 auditors have a bigger incentive to protect their reputation because they have more to lose. Because of that I expect that firms audited by a Big 4 auditor use their discretion in non-GAAP earnings less to meet or beat an analyst forecasts.

Therefore, I examine whether firms audited by a non-Big 4 auditor use non-GAAP earnings more often to meet or beat analyst forecast than firms audited by a Big 4 auditor.

H3: Managers of firms audited by a Big 4 auditor use exclusions to increase non-GAAP earnings more often to meet or beat analysts’ forecast than firms audited by a Big 4 auditor.

Managers argue that they use non-GAAP earnings to provide more information for investors so they can make better decisions, because the information they provide contains more useful information than GAAP measures. Opponents of non-GAAP earnings argue that non-GAAP earnings are used to manage investors’ perception of the firms’ performance. Enabling, for example, the firm to beat analysts’ earnings forecast or to turn GAAP losses into non-GAAP profits. The opponents are not concerned about the misleading of analysts but more about the misleading of nonprofessional investors (Entwistle, 2006).

The prior literature gives mixed evidence on which measure is more useful for investors, non-GAAP or GAAP earnings. But when non-GAAP earnings are used to provide more information to investors the non-GAAP earnings should be more useful for investors, which is not the case when non-GAAP earnings are used for opportunistic reasons. When non-GAAP earnings are used for opportunistic reasons the GAAP earnings are more useful for investors.

Chen et al. (2012) explain that although non-GAAP earnings are voluntary, under the provision of SAS 8 auditors are responsible for the disclosed statements by their clients. Because of this auditors do not want to damage their reputation through clients that report opportunistic non-GAAP earnings. Since Big 4 auditors have a better reputation than non-Big 4 auditors, Big 4 auditors have a bigger incentive to protect their reputation because they have more to lose. Because of that, I expect that firms audited by a Big 4 report less

opportunistic non-GAAP earnings. When firms report less opportunistic non-GAAP earnings, the non-GAAP earnings are used to provide more information to investors. Therefore, I examine whether the non-GAAP earnings disclosed by firms audited by a Big 4 auditor

(15)

provide more informative value than the non-GAAP earnings disclosed by firms audited by non-Big 4 auditor. The hypothesis is formally expressed, as follows:

H4: Non-GAAP earnings are more useful for investors when firms are audited by Big 4 auditors than when they are audited by non-Big 4 auditors.

(16)

3. Sample and Methodology

3.1 Sample

To test my hypotheses, a sample of US firms from which data is available about analyst forecasts, fundamental firm characteristics and share prices is necessary. I use IBES Summary History (IBES) to obtain analyst forecast and IBES Actual EPS, Compustat to obtain data about fundamental firm values and CRSP to obtain data about the stock prices.

The information is obtained for the fiscal year 2005-2012. The reason for this sample is that the rules for non-GAAP earnings changed in 2003 due to regulation G. Therefore data is obtained starting from the fiscal year 2004. Some variables for the equations to answer the hypothesis needed data from the prior year, thus the sample starts at 2005. Information is obtained until 2012.

Data about financial institutions are dropped from the sample because financial institutions have different rules for disclosing earnings.

I start with the complete databases from Compustat for the fiscal years 2004-2012. This results in 124,842 firm year observations for 18,319 unique firms. After that firm years from which not sufficient data is available to calculate the required variables are dropped from the sample. Then the datasets from IBES, Compustat and CRSP are merged and 20,066 firm years are left from 4,482 unique firms. After that, I calculate the discretionary accruals, discretionary cash flows, discretionary expenses and discretionary production costs. The final sample contains 9,052 firm year observations for 2,072 unique firms. In table 1 the data process is shown.

Table 1

Firm

years Unique Firms Deleted firm years Complete Compustat data

(2004-2012) 124,842 18,319

Deleting firm years with insufficient

data 79,974 13,497 44,868

After merging with IBES 26,075 5,452 53,899

After merging with CRSP 26,075 5,452 0

After calculating variable Sales

Growth 20,066 4,482 6,009

After calculating accrual bases and

real earnings management 9,052 2,072 11,014

(17)

3.2 Methodology

3.2.1 Proxy for non-GAAP earnings

Studies that examine non-GAAP earnings proxy for non-GAAP earnings in two ways: using actuals reported by services as IBES or the method where data about the non-GAAP earnings is hand-collected. The benefit of using the hand-collected method is that it directly addresses the adjustments to the GAAP earnings made by managers were the actuals method addresses the adjustments analysts make. Another benefit from the hand-collected data method is that the sample using this method only contains firms that actually report non-GAAP earnings. When the actuals method is used, the sample also contains firms that do not report non-GAAP earnings, although analyst report actual EPS for those firms (Jennings and Marques, 2011).

Doyle et al. (2013, p. 43) discuss why the IBES Actual EPS and the hand-collected EPS are such a close match. In their paper they discuss the 5 stages how non-GAAP earnings and corresponding analyst forecasts are generated. First, to make their forecast analysts are aware of the exclusions that are going to be made for the anticipated unusual items. However, they do not know the amount for which the exclusions are going to be made. Second, the analysts decide whether they include or exclude the exclusions they are aware of in their earnings forecast. Third forecast providers, such as IBES, survey all analysts’ estimates to establish whether the majority includes or excludes an item. They follow a majority policy, where the accounting basis of each company is determined by what the majority of analyst do. And for each firm the consensus mean estimate from forecast data providers is made publicly. Fourth, the managers observe whether their own firms’ economic performance, relative to the analyst consensus forecast. At this point the manager choose either (1) use opportunism to meet or beat the analyst forecast and/or (2) create a new type of exlusion that increases non-GAAP earnings. The managers report this in the press release. At last, the analysts and forecast data providers observe the earnings announcement and the non-GAAP exclusions that are made. They adjust if there are differences between the exclusions made by analysts and by the managers. Thus the IBES actual earnings numbers already includes any ex-post adjustments made by analyst to undo any managerial opportunism.

I use the actuals method. The proxy for non-GAAP earnings that I use is IBES Actual EPS. This method has got several benefits. By using this method a bigger sample can be examined, and this paper can be compared directly to other studies that use the same method (Doyle et al., 2013). Next to that, Doyle et al. (2003) argue that because of the close relation management and analyst have, it is difficult to believe that the two parties are not focused on

(18)

the same earnings definition. Otherwise the question arises, what would it mean to beat analyst forecasts? Furthermore, Bhattacharya et al. (2002) show that 65 percent of the time IBES Actual EPS and the non-GAAP EPS reported by managers in press releases is exactly the same. The remaining difference between hand-collected non-GAAP earnings and IBES Actual EPS is just 1 percent. Doyle et al. (2003) do a random test on 50 press releases to check this, in 48 cases IBES Actual EPS and hand-collected EPS are a perfect match.

3.2.2 Confirming non-GAAP earnings exceed GAAP earnings

To answer H1, I confirm that the non-GAAP EPS are higher than the GAAP EPS, the GAAP EPS are the earnings per share before extraordinary items and discontinued operations by Compustat. I do this by performing a t-test on the means of the non-GAAP EPS and GAAP EPS for the complete sample. To do this I test the following equation:

µNon-GAAP EPS = µGAAP EPS (1)

Where µNon-GAAP EPS is the mean of the non-GAAP earnings per share and

µGAAP EPS is the mean of the GAAP earnings per share.

To answer H1 I compare the means using a t-test between the means of non-GAAP EPS and GAAP EPS. When the non-GAAP EPS are higher and the means are significantly different it implies that there is more opportunism in the non-GAAP EPS.

To examine H2 I run a T-Test to examine whether the differences between the means differ when a firm is audited by a Big-4 auditor or a non-Big-4 auditor.

µNon-GAAP EPSBig 4 = µGAAP EPSBig 4 (2)

µNon-GAAP EPSNon-Big 4= µGAAP EPSNon-Big 4 (3)

µExclusions non-Big 4 ≠ µExclusions Big 4 (4)

Where µNon-GAAP EPSBig4is the mean of the Non-GAAP earnings per share for

firms audited by a Big 4 auditor. µGAAP EPSBig 4is the mean of the GAAP earnings per share

for firms audited by a Big 4 auditor. µNon-GAAP EPSnon-Big 4is the mean of the non-GAAP

earnings per share for firms audited by a non-Big 4 auditor. µGAAP EPSNon-Big 4is the mean

of the GAAP earnings per share for firms audited by a non-Big 4 auditor. µExclusions non-Big 4

(19)

non-Big 4 auditor. µExclusions Big 4 is the difference between non-GAAP earnings and GAAP

earnings for firms audited by a Big 4 auditor.

To answer the H2, I first examine if for both groups the non-GAAP earnings are higher than the GAAP earnings (2) and (3). Then I compare if the difference between the non-GAAP and GAAP earning is bigger when a firms is audited by a non-Big 4 auditor (4).

3.2.3 Meeting and Beating Analyst forecasts

To answer the third hypothesis I use a method used by Doyle et al. (2013). I compare the propensity to meet or beat analyst earnings forecast for firms that report

income-increasing non-GAAP exclusions and the effect of the presence of a Big 4 auditor on this. I focus on firms where non-GAAP earnings exceed GAAP earnings, this means that my main independent variable is, Pos Excl Use, which is equal to one if management has non-GAAP earnings that are greater than GAAP earnings and otherwise is equal to zero. I do this with the following equation:

MBEt= γ0 + γ1 Big 4 + γ2 Pos Excl Usei,t + γ3Big 4 * Pos Excl Use + γ4 Book-to-Marketi,t+ γ5Sales Growthi,t + γ6Ln Sizei,t+ γ7Profitablei,t+ γ8ROAi,t+ υi,,t (5)

Where MBE, the main dependent variable, is an indicator variable that is equal to one when Surprise is greater than or equal to zero. Surprise is defined as the IBES Actual EPS figure minus the median consensus analyst forecast from IBES. Big 4 is equal to 1 when a firm is audited by a Big 4 auditor (Compustat auditor score 4, 5, 6, 7). When a firm is audited by a non-Big 4 auditor Big 4 is equal to 0. Pos Excl Use is defined by using the IBES Actual EPS. If IBES actual EPS exceeds the GAAP EPS number, then Pos Excl Use¸ is equal to one.

Big 4 * Pos Excl Use is the interaction term between Big 4 and Pos Excl Use. Book-to-Market

is measured as the book value of equity divided by the market value of equity at the end of the fiscal year. Sales Growth is the quarterly change in revenue over the same quarter in the prior year. Profitable is an indicator variable which is equal to one if IBES Actual EPS is positive and zero otherwise. ROA controls for firm performance and is equal to IBES Actual

EPS scaled by total assets per share (Doyle et al., 2013).

I use a multivariate logistic regression to investigate the effect of Pos Excl Use on the propensity to meet or beat the analyst forecast.

The Pos Excl Use variable is designed to find the decision by managers to use exclusions that increase the income. If management is using opportunism in the exclusion of

(20)

expenses from GAAP earnings to meet or beat analyst forecast, I expect a positive relation between Pos Excl Use and MBE.

To answer my hypothesis I add an interaction term between Big 4 and the decisions by managers to use income-increasing exclusions, this is the interaction variable Big 4 * Pos

Excl Use. When the interaction term is negative and significant, it means that firms audited

by Big 4 auditors use income-increasing non-GAAP exclusions less often to meet or beat analyst forecasts than firms audited by a non-Big 4 auditor. Which indicates that firms audited by non-Big 4 auditors use non-GAAP earnings more opportunistically. When the interaction term is positive this indicates that firms audited by Big 4 auditors use non-GAAP earnings more opportunistically. I expect that the coefficient is negative and therefore that firms audited by a non-Big 4 auditor use more opportunism in their non-GAAP earnings.

3.2.3.1 Expected and Unexpected non-GAAP exclusions

Following Doyle et al. (2013) I am going to test the exclusions that are unexpected or unanticipated by analysts. This is because exclusions that are expected by analysts should be incorporated into their forecasts and therefore do not increase the chance of meeting or beating the forecasts. I do this with the following equation

MBEt= γ0 + γ1 Big 4 + γ2 Pos Other Excl Usei,t + γ3 Big 4 * Pos Other Excl Use + γ4 Pos Special Item Use + γ5Book-to-Marketi,t+ γ6Sales Growthi,t + γ7LnSizei,t+ γ8

Profitablei,t+ γ9ROAi,t+ υi,, (6)

Where Pos Other Excl Use is equal to 1 if other exclusions are greater than zero and otherwise 0. To proxy for expected exclusions, exclusions are divided into special items and other exclusions. Where Special Items are defined as operating income per share minus GAAP EPS before extraordinary items. The remaining total exclusions that are not captured by special items are used as proxy for unexpected exclusions. Other exclusions is defined as Exclusions minus Special Items. This means that other exclusions are items that the

management of a firm has excluded in non-GAAP earnings that have not been identified as nonrecurring items by Compustat. Big 4 * Pos Other Excl Use is the interaction term between

Big 4 and Pos Other Excl Use. Pos Special Item Use scores 1 if Special items are greater than

(21)

To examine the effect of a Big 4 auditor I add the interaction term Big 4 * Pos Other

Excl in the equation. When the interaction term Big 4 * Pos Other Excl is significant and

scores negative it means that firms audited by a Big 4 auditor use less other exclusions in their positive exclusions. And when the interaction term is positive it means that firms audited by Big 4 auditors use more other exclusions in their positive exclusions. Prior literature found evidence that the other exclusions are behaving as if they are recurring operating expenses that management strategically excluded from GAAP earnings, reporting them as being non-recurring and transitory (Doyle et al., 2003; Heflin and Hsu, 2008; Kolev et al., 2008; Doyle et al., 2013). Therefore, a negative interaction term means that there is more opportunism in the non-GAAP earnings of firms audited by a non-Big 4 auditor than for firms audited by a Big 4 auditor. I expect the interaction term to be negative, which indicates that firms audited by non-Big 4 auditors use more opportunism in their non-GAAP earnings.

3.2.3.2 Controlling for other methods of meeting and beating analyst forecasts

In the third hypothesis, I propose that managers of firms audited by non-Big 4

auditors use exclusions in non-GAAP earnings more opportunistically to meet or beat analyst forecasts. To rule out that the results for this hypothesis are driven by another form of

earnings management namely: discretionary accruals, discretionary production, discretionary expenses and discretionary cash flow. I examine if discretionary accruals, discretionary production, discretionary expenses and discretionary cash flow are correlated with exclusion use (Doyle et al., 2013). Therefore, I add a proxy for accruals manipulation in the model to provide evidence that the results are not driven by accrual manipulation. To do this I estimate the following regression equations:

MBEt= γ0 + γ1 Big 4 + γ2Pos Excl Usei,t + γ3Big 4 * Pos Excl Use + γ4Pos Disc Acci,t + γ5Pos Disc CFO + γ6Pos Disc Prod + γ7Pos Disc Exp + γ8Book-to-Market + γ9Sales Growthi,t+ γ10LnSizei,t + γ11Profitablei,t + γ12ROAi,t + υi,,t (7) MBEt= γ0 + γ1 Big 4 + γ2 Pos Other Excl Usei,t + γ3 Big 4 * Pos Other Excl Use + γ4 Pos Special Item Use + γ5Pos Disc Acci,t + γ6Pos Disc CFO + γ7Pos Disc Prod + γ8 Pos Disc Exp + γ9Book-to-Market + γ10Sales Growthi,t+ γ11LnSizei,t + γ12

(22)

Where Pos Disc Acc is equal to one when the discretionary accruals is higher than zero, if the discretionary accruals score otherwise Pos Disc Acc is equal to 0. The variable

Pos Disc CFO is equal to 1 when the discretionary cash flows are higher than zero, otherwise Pos Disc CFO is equal to 0. The variable Pos Disc Prod is equal to 1 when the discretionary

production costs are higher than zero, otherwise Pos Disc Prod is equal to 0. The variable

Pos Disc Exp is equal to 1 when discretionary expenses are higher than zero, otherwise Pos Disc Exp is equal to 0 (Doyle et al., 2013).

I control for accrual manipulation using annual performance-adjusted discretionary accruals using the performance-adjusted Modified-Jones model, as calculated by Cahan et al. (2011). Discretionary Cash flows, discretionary production costs and discretionary expenses are calculated following the method of Roychowdhury (2006) (Doyle et al., 2013).

3.2.4 Usefulness of non-GAAP earnings 3.2.4.1 Relative Usefulness

To examine the fourth hypothesis I use the regression model used by Jennings et al. (2001) and van Raak (2005). The model is used to compare the ability of non-GAAP earnings to explain observed share prices for firms audited by a Big 4 auditor and firms audited by non-Big 4 auditors. The following regressions is used:

Pi= γ0 + γ1Non-GAAP EPSi + γ2 Big 4 + γ3Big 4 * Non-GAAP EPS + ei (9) Piis the company i’s closing stock price on the last day of the third month after the end of its fiscal year. Following Jennings and van Raak, I chose for this period to ensure that current financial statements were publicly available for virtually all companies in the sample.

Big 4 * Non-GAAP EPS is the interaction term between Big 4 and Non-GAAP EPS.

To assess whether a Big 4 auditor influences the effects the usefulness of non-GAAP earnings I added an interaction term in the equation, Big 4 * Non-GAAP EPS. When Big 4 *

Non-GAAP EPS is significant and positive this means that when a firms is audited by a Big 4

company the non-GAAP earnings are more useful for investors because they explain the stock price better. If the interaction term scores negative the non-GAAP earnings of firms audited by a non-Big 4 auditor explain the stock price better. The goal of non-GAAP earnings is to provide more information for investors, thus the more the non-GAAP earnings explain the stock price the better.

(23)

3.2.4.2 Incremental Usefulness

To provide further evidence on which earnings measurement is the most useful for users of financial statements I use the formula of van Raak (2005) to calculate the

incremental utility for the difference the two groups. The difference between the two earnings measurements is equal to the components that are left out of the GAAP earnings to create the non-GAAP earnings. The variable Exclusions is non-GAAP EPS – GAAP EPS. The formula used is:

Pi = γ0 + γ1Non-GAAP EPSi + γ2 Exclusions + γ3 Big 4 + γ4 Big 4 * Exclusions + e4i (10) Where Exclusions is the difference between non-GAAP EPS and GAAP EPS. Big 4 *

Exclusions is the interaction term between Non-GAAP EPS and Big 4.

To assess whether a Big 4 auditor effects the incremental utility of the non-GAAP earnings I added the interaction term Big 4 * Exclusions. When the interaction term is negative it means that the exclusions in the non-GAAP earnings contain less incremental value when a firm is audited by a Big 4 auditor. When the exclusions contain incremental value, it means that they are useful for investors. Therefore, when the interaction term is negative the incremental value of the non-GAAP earnings is higher for firms audited by a Big 4 auditor. And when the interaction term is positive the incremental value of the non-GAAP earnings is higher for firms audited by a non-Big 4 auditor.

3.2.4.3 Predictive Power

Following van Raak (2005), I examine the difference in prediction power of the non-GAAP EPS for firms audited by a Big 4 auditor and firms audited by a non-Big 4 auditor on next year’s EPS. The formula used is:

EPSit = γ0 + d1Non-GAAP EPSit-1+ γ2Big 4 + γ3 Big 4 * Non-GAAP EPSit-1+ e4i (11) Where EPSitis the earnings per share in the current fiscal year, Non-GAAP EPSit-1is the

non-GAAP earnings per share in the prior year. Big 4 * Non-non-GAAP EPS it-1is the interaction term

(24)

To assess whether there is a difference between firms audited by a Big 4 auditor and firms audited by a non-Big 4 auditor I added an interaction to the equation, Big 4 *

Non-GAAP EPS it-1. When the interaction term scores is significant and positive it means that the

non-GAAP earnings of the year before predict the earnings per share of next year better when a firm is audited by a Big 4 auditor. The goal of non-GAAP earnings is to provide more information for investors, thus the more the non-GAAP earnings predict the earnings per share of next year the better. Therefore, when the interaction term is positive non-GAAP earnings provide more useful information for investors when a firm is audited by a Big 4 auditor. When the interaction term is negative it is the non-GAAP earnings of firms audited by non-Big 4 auditors provide more useful information for investors.

(25)

4. Results

4.1 Descriptive Statistics and Univariate Tests

Table 2 presents descriptive statistics for the complete sample. The aim of this paper is to examine the effect of audit quality on non-GAAP earnings, where Big 4 audit firms versus non-Big 4 audit firms is a proxy for audit quality, I compare the firms that are audited by a Big 4 auditor with firms audited by a non-Big 4 auditor. Of the 9,051 firm-year

observations from 2005-2012, 7,702 are audited by a Big 4 auditor and 1,349 are audited by a non-Big 4 auditor.

The dependent variable is MBE. Firms audited by Big 4 auditors meet or beat the analysts’ forecast 54,5% of the time and firms audited by non-Big 4 43,7% of the time. H3 predicts that firms audited by non-Big 4 auditors use non-GAAP earnings more often to meet or beat analysts’ forecast than firms audited by a Big 4 auditor. The results found in this univariate test is contradicting to that prediction. This contradicting result can be caused by other firm characteristics which cannot be controlled for in this univariate test. Since this paper is the first to examine the relation between Big 4 auditors and using non-GAAP earnings to meet or beat analysts’ forecast, there is no prior evidence to compare this result with. However, this results can be explained by prior literature. Keung et al. (2010) find that investors, after the implementation of SOX, see just meeting or beating analyst forecasts as a red flag. Which means that firms have less incentive to manage earnings to just meet or beat earnings. Next to that, Heflin and Hsu (2008) and Kolev et al. (2008) find that after the implementation of Regulation G in 2003 the non-GAAP earnings are used less

opportunistically and provide more information for investors. Because of that, the exclusions made to create the non-GAAP earnings can make the non-GAAP earnings meet or beat the consensus forecast, but this can be caused by exclusions that contain no information for investors.

For the variable Exclusions, which represents the difference between non-GAAP earnings and GAAP earnings, is the mean higher for firms audited by a non-Big 4 auditor. Although this difference is not significant, the variable indicates that the difference between GAAP and non-GAAP earnings are in actual fact higher for firms audited by a non-Big 4 auditor. Chen et al. (2012) discuss that a bigger positive difference between GAAP and non-GAAP earnings indicate more opportunism in the non-non-GAAP earnings. However, this results is not significant. This is caused because there is no difference between exclusions for firms audited by Big 4 auditors and firms audited by non-Big 4 auditors. This paper provides evidence for this question, which is not earlier answered in the prior literature.

(26)

The control variables Profitable and ROA indicate that firms audited by Big 4 auditors are more profitable than firms audited by non-Big 4 auditors. Next to that the Book-to-Market ratio is lower for firms audited by Big 4 auditors, which indicates that their share price is valued higher than firms audited by non-Big 4 auditors. Overall the control variables show that the economic conditions for firms audited by non-Big 4 auditors are worse than for firms audited by Big 4 auditors. These results are consistent with the findings of Lawrence et al (2011). They find that there are differences in the client characteristics of Big 4 and non-Big 4 clients. Their results show that clients of Big 4 auditors are bigger in size and more profitable than clients of non-Big 4 auditors. The fact that clients of non-Big 4 auditors are less profitable than clients of Big 4 auditors might be an incentive to report more

opportunistic non-GAAP earnings.

4.2 Correlations

In table 3 a spearman correlation table is included. It shows that MBE is positively correlated with Big 4 which is opposite of what is expected in H3. Pos Excl Use, Pos Other

Excl Use are positively correlated with MBE which is expected in H3. Next to that Pos special items use is negatively correlated with MBE this is an interesting correlation. The

reason for that is that exclusions are separated into other exclusions and special items. The positive correlation of other exclusions use indicates that when a firm uses positive other exclusions, they meet or beat analyst forecasts more often. But when they use special items they tend to meet or beat analyst forecast less often, since other exclusions are items not flagged as nonrecurring. This indicates that using opportunism in the exclusions, by using other exclusions, helps meeting or beating analyst forecasts. This is an interesting result for H3. However, this is a univariate test, so the results can be caused by interactions.

Next to that, the correlation of Big 4 with Positive Other Exclusion is negative and the correlation with Positive Special Item Use is positive. This indicates that firms audited by a Big 4 auditor use more special items in their exclusions and less other exclusions. This result also suggests that firms audited by Big 4 auditors exclude more nonrecurring items and less items that are not flagged as nonrecurring. Which indicates that there is more opportunism in the exclusions of firms audited by non-Big 4 auditors. However, this is a univariate test.

(27)

Table 2, Descriptive statistics

The full sample consists of 9,051 firms-year observations from 2005-2012. The variables are defined as follows: Big 4 is when a firm is audited by a Big 4 company. Non-Big 4 is when a firm is audited by a non-Big 4 auditor. MBE is an indicator variable equal to one if the earnings surprise is greater than or equal to zero. Pos Excl Use is an indicator variable equal to one if Exclusions is greater than zero. Pos

Other Excl Use is an indicator variable equal to one if Other Excl is greater than zero. Pos Special Items Use is an indicator variable equal to

one if Special items are greater than zero. Non-GAAP EPS is the IBES reported actual earnings per share. GAAP EPS is the basic income per share before extraordinary items. Surprise is equal to GAAP EPS less the consensus earnings forecast. Exclusions is equal to

Non-GAAP EPS minus Non-GAAP EPS. Other Excl is equal to Exclusions less Special items. Special Items is equal to operating income per share

minus GAAP EPS. Book-to-Market is measured as the book value of equity divided by the market value of equity. Sales Growth is equal to sales divided by Sales y-1. Ln Size is equal to the natural log of price multiplied by shares outstanding. Profitable is an indicator variable that is if equal to one if Non-GAAP EPS is greater than or equal to zero. ROA is equal to Non-GAAP EPS divided by total assets per share.

Pos Disc Acc is equal to one if discretionary accruals are greater than zero. Pos Disc Prod is equal to one if discretionary production costs

are greater than zero. Pos Disc Exp is equal to one if discretionary expenses are greater than zero. Pos Disc CFO is equal to one if discretionary cash flows are greater than zero. All continuous variables are winsorized at the 1% and the 99% levels.

Full Sample Big 4 non-Big 4 Variable N Mean Std. Dev. N Mean Std. Dev. Difference of mean t-Test p-Value MBE 7702 0.545 0.497 1349 0.437 0.496 -7.393 0.0000 Pos Excl Use 7702 0.461 0.498 1349 0.435 0.495 -1.780 0.0751 Pos Other Excl Use 7702 0.458 0.499 1349 0.473 0.498 1.044 0.2964 Pos Special Items

use 7702 0.601 0.489 1349 0.593 0.491 -0.578 0.5627 Non-GAAP EPS 7702 1.184 1.526 1349 0.429 0.969 -17.533 0.0000 GAAP EPS 7702 1.119 2.032 1349 0.320 1.286 -13.958 0.0000 Surprise 7702 -0.025 0.331 1349 -0.073 0.298 -5.020 0.0000 Exclusions 7702 0.065 1.436 1349 0.098 0.927 0.810 0.4180 Other Exclusions 7702 -0.155 1.011 1349 -0.006 0.686 5.215 0.0000 Special Items 7702 0.219 0.733 1349 0.102 0.462 -5.631 0.0000 Book-To-Market 7702 0.526 0.423 1349 0.668 0.561 10.732 0.0000 Sales Growth 7702 1.109 0.244 1349 1.120 0.304 1.514 0.1299 Ln Size 7702 7.203 1.829 1349 5.246 1.338 -37.560 0.0000 Profitable 7702 0.841 0.365 1349 0.699 0.458 -12.583 0.0000 ROA 7702 0.025 0.143 1349 0.015 0.199 -9.173 0.0000 Pos Disc Acc 7702 0.337 0.472 1349 0.340 0.474 0.244 0.8066 Pos Disc Prod 7702 0.348 0.476 1349 0.416 0.493 4.827 0.0000 Pos Disc Exp 7702 0.392 0.488 1349 0.437 0.496 3.114 0.0018 Pos Disc CFO 7702 0.645 0.478 1349 0.551 0.497 -6.581 0.0000

(28)

28

Table 3, Spearman Correlation

1 2 3 4 5 6 7 8 9 10

1 Non-GAAP Eps 1.0000

2 Big 4 0.2219*** 1.0000

3 Pos Excl Use 0.0298*** 0.0188* 1.0000

4 Pos Other Excl Use -0.0159 -0.0098 -0.8940*** 1.0000

5 Pos Special Items Use 0.0198* 0.0054 -0.7846*** -0.8321*** 1.0000

6 MBE 0.2953*** 0.0775*** -0.1095*** 0.1174*** -0.0864*** 1.0000

7 Sales Growth -0.0105 0.0012 -0.3170*** 0.3307*** 0.3099*** 0.2306*** 1.0000

8 Book-To-Market -0.0872*** -0.0915*** 0.2193*** -0.2411*** -0.2311*** -0.1810*** -0.2656*** 1.0000

9 Ln Size 0.6369*** 0.3833*** -0.1077*** -0.1643*** 0.1616*** 0.2456*** 0.1309*** -0.3682*** 1.0000

10 Profitable 0.6588*** 0.1308*** 0.0588*** 0.0317*** -0.0482 0.2344*** 0.0362*** -0.0228** 0.3977*** 1.0000

(29)

4.3 Non-GAAP earnings are higher than GAAP earnings

To answer H1, table 4 presents a t-test to compare the means of the non-GAAP earnings with the GAAP earnings for the full sample. Non-GAAP earnings have a mean which is 0.0713 higher than GAAP earnings. This difference is significant at the 1% level, with a t-value of 2.7627.

Table 4, H1

Variable N Mean Std. Dev.

Non-GAAP EPS 9051 1.0716 0.1557

GAAP EPS 9051 1.0003 0.02060

T-Value = 2.7627

P-Value= 0.0057

This table presents T-test to compare the means between non-GAAP EPS and GAAP EPS.

This result is consistent with prior literature about non-GAAP earnings. Where the literature which examines non-GAAP earnings in the pre-Regulation G period, Bradshaw and Sloan, 2002; Brown and Sivakumar, 2003; Doyle et al, 2003; van Raak, 2005, find evidence that non-GAAP earnings are higher than GAAP earnings. And in the post-Regulation G, which caused the difference between non-GAAP earnings and GAAP earnings to decrease, period Heflin and Hsu (2008) and Kolev et al. (2008) provide evidence that non-GAAP earnings are higher than GAAP earnings.

The results presented in table 4 are consistent with H1 and confirm that in the full sample used in this paper non-GAAP earnings are higher than GAAP earnings. Therefore, H1 is confirmed.

On the one hand according to Chen et al. (2012) non-GAAP earnings that are reported higher than GAAP earnings are an indicator of opportunism in the non-GAAP earnings. On the other hand Regulation G is implemented (SEC, 2003), because of Regulation G the quality of non-GAAP earnings increased (Heflin and Hsu, 2008; Kolev et al., 2008). Therefore, it can be argued that when non-GAAP earnings are reported higher than GAAP earnings it does not have to imply that the there is opportunism in the non-GAAP earnings. The non-GAAP earnings can provide more useful information for investors.

4.4 Difference GAAP earnings and GAAP earnings between Big 4 and non-Big 4 audited firms.

To answer H2, table 5 Panel A presents a t-test to compare the mean between the non-GAAP earnings and GAAP earnings for firms audited by a Big 4 auditor. The results show that non-GAAP EPS are higher than GAAP EPS, significant at the 5% level.

(30)

Table 5, Panel B presents a t-test to compare the mean between the non-GAAP earnings and GAAP earnings for firms audited by a non-Big 4 auditor. The results show that non-GAAP EPS are higher than the GAAP EPS, significant at the 5% level.

Table 5, Panel C presents a t-test to compare the means of the difference, non-GAAP earnings minus GAAP earnings, which is the variable Exclusions, between firms audited by a Big 4 auditor and firms audited by a non-Big 4 auditor. The means of exclusions is higher for firms audited by a non-Big 4 auditor than for firms audited by a Big 4 auditor. However, this results is not significant.

Table 5, H2

Panel A: Firms audited by Big 4 auditor Variable N Mean Std. Dev.

Non-GAAP EPS 7702 1.1840 1.5265

GAAP EPS 7702 1.1193 2.0320

T-Value = 2.2325 P-Value = 0.0256

Panel B: Firms audited by non-Big 4 auditor Variable N Mean Std. Dev.

Non-GAAP EPS 1349 0.4299 0.9693

GAAP EPS 1349 0.3204 1.2866

T-Value = 2.4979 P-Value = 0.0126

Panel C: Difference between Exclusions Variable N Mean Std. Dev.

Big 4 7702 0.0653 1.4266

Non-Big 4 1349 0.09818 0.9276

T-Value = 0.8100 P-Value = 0.4180

Panel A presents a T-test to compare the means of non-GAAP EPS and GAAP EPS for firms audited by a Big 4 auditor. Panel B presents a T-test to compare the means of non-GAAP EPS and GAAP EPS for firms audited by a non-Big 4 auditor. Panel C presents a T-test to compare the difference between non-GAAP and GAAP EPS, which is Exclusions for firms audited by a Big 4 auditor and firms audited by a non-Big 4 auditor.

These results imply that for both groups, firms audited by a Big 4 auditor and firms audited by a non-Big 4 auditor, non-GAAP earnings are higher than GAAP earnings. But there is no evidence that the exclusions to create non-GAAP earnings are significantly higher for firms audited by non-Big 4 auditor.

According to Chen et al. (2012), the first evidence of opportunism in the presentation of non-GAAP earnings is that non-GAAP earnings are higher than GAAP earnings.

Therefore, the results presented in Table 5 provide evidence that both groups use non-GAAP earnings opportunistically. The result of table 5, panel c reject H2, the difference between GAAP earnings non-GAAP earnings are lower for firms audited by Big 4 auditors. However,

(31)

this results can be caused by the effect of Regulation G. Heflin and Hsu (2008) and Kolev et al. (2008), present evidence that after the implementation of Regulation G non-GAAP earnings are presenting more useful information for investors. Therefore, the exclusions are not necessarily used for opportunistic reasons and can present more useful information for investors.

4.5 Meeting and beating analyst forecasts with non-GAAP earnings.

To answer H3, table 6 presents logistic regressions to examine the relation between the presence of a Big 4 auditor and the use of exclusions to meet or beat analyst forecasts. In column 1 of table 6 the relation between the presence of a Big 4 auditor and the use of positive exclusions to meet or beat analyst forecasts is presented. Untabulated results show that the odds ratio for Big 4 is 1.29, which implies that when a firm is audited by a Big 4 auditor is has 29% more chance to meet or beat analyst forecast. Furthermore, the untabulated results show that the odds ratio for Pos Excl Use is 1.10, which implies that when a firm uses positive exclusions it has 10% more chance to meet or beat analyst forecasts. However, this result is not significant. The results presented in table 6 and 7 are robust and are controlled for firm and year effects.

The variable of interest to answer H3 is the interaction term Big 4 * Pos Excl Use. The coefficient of this variable is negative and significant at the 1% level. This result implies that when a firm is audited by a Big 4 auditor the use of positive exclusions to meet or beat analyst forecasts becomes smaller. Therefore, the chance that a firm audited by a non-Big 4 auditor uses positive exclusions to meet or beat analyst forecasts is bigger. This result supports H3, firms audited by a non-Big 4 auditor use exclusions to increase the non-GAAP earnings more often to meet or beat analyst forecast than firms audited by a Big 4 auditor.

This result is consistent with the expectation. Using the discretion in the exclusions of non-GAAP earnings to make the non-GAAP earnings higher signals opportunism (Doyle et al, 2013). Therefore, the results shown in column 1 from table 6 imply that firms audited by non-Big 4 auditor use more opportunism in the non-GAAP earnings.

In Column 2 of table 6 exclusions are separated into other exclusions and special items. The variable of interest is the interaction term Big 4 * Pos Other Excl Use, which is negative and significant. This implies that when a firm is audited by a Big 4 auditor it uses less other exclusions in their exclusions, and therefore, which untabulated results confirm, use more special items in their exclusions. Since other exclusions are items that are not flagged as nonrecurring and special items are, this results provide evidence that firms audited

(32)

by non-Big 4 auditors exclude more recurring items in their non-GAAP earnings than firms audited by a Big 4 auditor.

The results in Column 2 of table 6 are consistent with the expectation. Firms audited by non-Big 4 auditors exclude more recurring items in their non-GAAP earnings. This indicates more opportunism since recurring items should be included in the non-GAAP earnings, because they provide useful information for investors. Furthermore the results provide evidence that firms audited by Big 4 auditors exclude more nonrecurring items, which indicates that they construct their non-GAAP earnings to provide more information for investors. Therefore, the results in column 2 from table 6 provide evidence that the GAAP earnings of firms audited by Big 4 auditors use more opportunism in their non-GAAP earnings.

The results in table 6 suggest more opportunism in the non-GAAP earnings of firms audited by non-Big 4 firms. This is consistent with the expectation of H3. And also consistent with the expectations of prior literature. Doyle et al. (2013) states that using non-GAAP earnings was another way of firms to meet or beat earnings thresholds, instead of using earnings management. Becker et al. (1998) provides evidence that when a firm is audited by a Big 6 2audit firm there is less earnings management. Which is consistent with the results provided in table 6 for H3.

Next to that, the results from table 6 column B can be put into context with the results found by Frankel et al. (2011). They provide evidence that when the board is more

independent the exclusions made in non-GAAP earnings are less opportunistic. The results of Frankel et al. in context with the results I provide in this section imply that when the control on non-GAAP earnings is lower the opportunism in the non-GAAP earnings is higher. This suggest that disclosing non-GAAP earnings with weak controls leads to managers reporting them opportunistically.

Furthermore, the results in this section can be put into context with the results for H1 and H2. Non-GAAP earnings are higher for both groups, however the results in table 6 show that the exclusions in non-GAAP earnings for firms audited by non-Big 4 auditors contain more recurring items than the exclusions from firms audited by Big 4 auditors. Therefore, a positive difference between non-GAAP and GAAP earnings can be an indicator of

opportunistically reported non-GAAP earnings for firms audited by non-Big 4 auditors.

2In the research period of Becker et al. (1998) there were the Big 6 audit firms, which are consistent with the

Referenties

GERELATEERDE DOCUMENTEN

For the Re th = 3500 case, similar trends in the throat area are seen while the jet breakdown location from HR simulations matches more closely to the experiments. This location

Op basis van de resultaten van het innova- tieproject kunnen de volgende acties bijdra- gen aan verbetering van de kwaliteit van biologische producten: voortzetting en ver- betering

The existing challenges in the phenotyping of hPSC-CM function as described above will be addressed in this thesis. Overcoming these challenges by developing a reliable and

Het Advocacy Coalition Framework beschrijft voornamelijk perioden en factoren van stabiliteit in en (John, 2013; Sabatier & Weible, 2007, p. 198), maar verschaft

The performance on the perception task on the unaware cue-present trials in comparison with the delayed cue-target discrimination task could have been higher, because of the guess

In addition, there is also comparison on amount of waste collected in kilograms, the distance driven by the trucks, the mean filling level of the sites that are visited, the

These stemborers, along with other lepidopteran maize pests such as the invasive Spodoptera frugiperda (J.E. Smith) (Noctuidae), are often found in mixed populations in maize

Experimental and simulation data of the compressive strain with the pressure chamber diameter and the applied pressures as variables; combinatorial screening of 3D cellular behaviors