The effect of the report on the effectiveness of the internal control over financial reporting on the stock price

MSc Accountancy & control, variant Accountancy

Faculty of Economic and Business, University of Amsterdam

The effect of the report on the

effectiveness of the internal control

over financial reporting on the stock

price

Name: Dion Schouten

Student number: 10217703

Final version

Date: 21‐06‐2015

Number of words: 14.266

Supervisor: dr. A. Sikalidis

2

Statement of originality

This document is written by Dion Schouten 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 other 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.

3

Abstract

In this study, I examine the stock price reaction to the announcement of the report on the effectiveness of the internal control over financial reporting. I expect that companies which report a material weakness in the internal control have a negative stock price reaction and companies with effective internal control have a positive stock price reaction. Next to these main hypotheses, I also examine whether the severity of the stock price reaction is different for companies which disclose a material weakness when looking at the borrowing capacity and the quality of the corporate governance. I predict that companies with a high borrowing capacity and a good quality of corporate governance have a less negative (or more positive) stock price reaction. These expectations are studied by conducting a database research. I find evidence for the expectation that companies which report a material weakness in the internal control have a negative stock price reaction. I also find evidence to support the hypothesis that companies with effective internal control experience a positive stock price reaction. Regarding the severity of the stock price reaction, my results show that companies with a high borrowing capacity have a less negative (or more positive) stock price reaction. However, I do not find support for the expectation that companies with good corporate governance have a less negative (or more positive) stock price reaction. These results provide evidence that investors see reporting on the effectiveness of the internal control as value relevant in decision making since they change their expectations after receiving the information.

4

Table of content

1 Introduction 5

2 Literature review and hypotheses development 7

2.1 SOX 7

2.1.1 SOX 302 and SOX 404 7

2.1.2 Kind of internal control weaknesses 8

2.2 Prior literature 9

2.2.1 Internal control 9

2.2.2 The efficient market hypothesis 11

2.2.3 Link between new information and the stock price 12

2.2.4 Link between material weakness disclosure and the stock price 13

2.3 Hypotheses 14

3 Data and method 16

3.1 Sample selection 17 3.2 Research design 18 3.2.1 Empirical models 18 3.2.2 Empirical measures 19 4 Results 22 4.1 Descriptive statistics 22

4.2 The report on the effectiveness of the internal control and the stock price reaction 23 4.3 Effect of disclosing a material weakness in the internal control on the stock price

of a company 24

4.4 Effect of the borrowing capacity on the severity of the stock price reaction for firms which disclose a material weakness in the internal control 26 4.5 Effect of the corporate governance on the severity of the stock price reaction for firms which disclose a material weakness in the internal control 28

5 Conclusion 30

5

1. Introduction

Since the financial crisis, there were a lot of accounting scandals in the news such as Enron, Worldcom and Imtech. Many critics argued that the internal controls were partly to blame because the internal controls were not able to detect and prevent this accounting scandals from happening. In the United States, there was a need to react to these scandals and the Sarbanes‐Oxley Act (SOX) was introduced. SOX is introduced in 2002 and was announced as a reaction to the different accounting scandals but actually the Sarbanes‐Oxley act was already partially made before the scandals became known. Especially interesting for this thesis is section 404 from SOX, which is about the effectiveness of the internal control over financial reporting. If there is a weakness in the internal control, it is mandatory to report this since the introduction of SOX. In SOX section 404, public companies and the external auditor need to report on the effectiveness of the companies’ internal control over financial reporting (PCAOB, 2004). This is a yearly report and is published in the annual report of a company. Another part of the legislation which deals with reporting on the effectiveness of the internal control over financial reporting is SOX section 302. For SOX section 302, the chief executive officer (CEO) and chief financial officer (CFO) need to evaluate in every quarter the effectiveness of the internal control over financial reporting. If there are any deficiencies found, this needs to be reported to the auditor and the audit committee of the board of directors.

With the introduction of SOX 302 and 404, the PCAOB intended to increase the investor confidence in the financial statements and the used processes and controls which are necessary for high quality financial statements (Li, Sun and Ettredge, 2010). If a company has effective internal control, this decreases the chance of having a misstatement in the financial statements and therefore the reliability of the financial statements increases. If SOX 302 and 404 provide useful information to investors to determine the quality of the accounting information, the expectation is that investors change their expectations about the future cash flows which they can generate. This would result in a reaction of the stock price of a company.

Prior research regarding the stock price reaction to the report on the effectiveness of the internal control over financial reporting finds different results. This topic is not very often studied but there are some papers which have studied the impact the report on the effectiveness of the internal control has on the stock price. One of these articles is the article of Rezee et al. (2012). In this research, companies are divided in three groups: companies with effective internal control, companies which report a material weakness and delinquent companies. The results show that companies with a material weakness and delinquent companies have a negative abnormal return, which means a decrease in the stock price. Companies without a material weakness have a positive abnormal return, an increase in the stock price. Another article which has studied the link between the internal control effectiveness and the stock price is that by Beneish et al. (2008). In this article,

6 they have studied whether there is an effect of reporting a material weakness in the internal control under SOX section 302 and 404. The results indicate that there is a 1.8 percent decrease in the abnormal return for firms which report a material weakness under SOX section 302. The result regarding reporting a material weakness under section 404 is different. There is no stock price reaction to reporting a material weakness in the internal control under section 404. The findings of this article are therefore somewhat contradictory.

Although there is some prior literature about the impact of reporting a material weakness in the internal control on the stock price of a firm, all the articles written about this topic use data from 2004‐2006. This is very useful for researching the short term impact of the SOX act but provides no evidence for the longer term effect of SOX. In the longer term, there are two possible scenarios. The first scenario is that investors see SOX section 404 as not really useful anymore, which means that there was a overreaction when the SOX act was introduced. The second scenario is that investors find the report on the effectiveness of the internal control still relevant. With the second scenario, there is still an impact of reporting a material weakness on the stock price of a company.

In this thesis, I will study whether the report on the effectiveness of the internal control is still seen as valuable information for investors. The expectation in this thesis is that companies which report a material weakness in the internal control have a negative stock price reaction. For companies which report effective internal control, the expectation is that the stock price reaction is positive. For the sample of companies which report a material weakness in the internal control, there is also studied whether the severity of the stock price reaction is different depending on the borrowing capacity and the quality of the corporate governance. The hypothesis for the borrowing capacity is that companies with a high borrowing capacity have a less severe negative stock price reaction. Regarding the quality of the corporate governance, the expectation is that companies with a high corporate governance quality have a less severe negative stock price reaction.

To answer the research question, there is data used from the year 2013. The sample for this thesis consists of companies for which it is obligated to report on the effectiveness of the internal control and who are listed at a stock market in the United States. The cumulative abnormal return on a three day event window is used to measure the stock price reaction. All the companies also need to have the required information for the control variables. I find evidence for the relation between the disclosure of a material weakness and the abnormal return. The abnormal return for companies which disclose a material weakness is significant and negative. I also find support for the expectation that companies which report effective internal control have a positive abnormal return. Regarding the severity of the abnormal return for companies with a high borrowing capacity, there is also evidence found in this thesis. However, I do not find support for the expectation that the severity of

7 the stock price reaction for companies with a high quality corporate governance is less negative (or more positive).

This study contributes in three ways to the existing literature. Firstly, this research provides empirical evidence regarding the longer term effect of SOX section 404 by studying the impact of the report on the effectiveness of the internal control over financial reporting on the stock price with more recent data. Secondly, this thesis provides initial evidence about whether there is an impact of the borrowing capacity on the severity of the stock price reaction of companies which report a material weakness in the internal control. Thirdly, this thesis also provides initial evidence on whether there is an impact of the corporate governance quality on the severity of the stock price reaction of companies which report a material weakness in the internal control.

The remainder of this thesis is organized as follows. Section 2 reviews the prior literature and develops the hypotheses. Section 3 describes the sample selection procedure and the research method. In section 4, the results are presented. Finally, section 5 concludes.

2. Literature review and hypotheses development

In this chapter, the literature review is outlined. Firstly, it is described what SOX 302 and SOX 404 are and what the responsibilities of the management and the auditor are. Secondly, the different weaknesses that exist are outlined and the differences between the weaknesses are described. Thirdly, prior literature about the internal control over financial reporting is described. Fourthly, the efficient market hypothesis is explained and why there is the expectation that the stock price will react if there is a material weakness reported in the internal control. Finally, the hypotheses for the thesis are given.

2.1 SOX

2.1.1 SOX 302 and SOX 404

Two sections in the SOX act are especially important for the internal control over financial reporting, namely the sections 302 and 404. In SOX section 302, the chief executive officer (CEO) and the chief financial officer (CFO) need to evaluate the effectiveness of the internal controls every quarter (PCAOB, 2004). Out of this evaluation, management need to make a report about the opinion of the CEO and CFO regarding the effectiveness of the internal controls. The evaluation of the effectiveness of the internal controls must be done within 90 days before the report is filed. Regarding the internal controls, management is responsible for disclosing all significant deficiencies and material weaknesses found in the internal control and must report this to the auditor and the audit committee of the board of directors. Also any fraud from people who have a significant role in the internal control need to be reported and if there were significant changes in the internal control this

8 needs to be stated in the report (PCAOB, 2004). An important element of significant changes in the internal control is the disclosure of remediating a significant deficiency or a material weakness. The auditor has not an active role in the evaluation of the effectiveness of the internal control under SOX 302. His only responsibility is to be informed about any significant deficiencies or material weaknesses that the management has found.

In SOX section 404, which became effective November 2004 for accelerated filers, not only the CEO and CFO are responsible for evaluating the effectiveness of the internal controls over financial reporting but the auditor also needs to evaluate the effectiveness of the internal controls. This section consist of two different subsections, section a and section b. In SOX 404 a, management needs to evaluate the effectiveness of the internal control over financial reporting, which is done almost in the same way as described for SOX 302. In the report, which the management makes after the evaluation, management needs to state their responsibilities for establishing and maintaining internal control over financial reporting (PCAOB, 2004). The opinion of the management regarding the effectiveness of the internal control is also outlined in this report. In the opinion section of this report, there is stated if the internal controls were effective or that there were deficiencies or weaknesses discovered. The difference between SOX 302 and 404 is section b of SOX 404. In SOX 404 section b, the auditor has several responsibilities. Before expressing an opinion on the internal control, the auditor must undertake tests to evaluate some elements. Firstly, the auditor needs to read the opinion of the management stated in section 404 a. Secondly, the auditor needs to gather sufficient competent evidence about the process used and the conclusion reached by the management to evaluate management’s assessment of the effectiveness of the internal control. Thirdly, the auditor also needs to evaluate the effectiveness of both the design and operation of the internal control itself. Finally, the auditor needs to state his opinion about the effectiveness of the internal controls in a report. In this opinion section, the auditor states whether he agrees with the opinion of management and whether he has found any deficiencies or weaknesses regarding the internal control over financial reporting. Just as with an opinion paragraph about the financial statements, an auditor can give an unqualified, adverse or disclaimer of opinion. An unqualified opinion is given when there is no material weakness discovered, an adverse opinion when there is at least one material weakness in the internal control and a disclaimer of opinion is given when the auditor cannot give an opinion due to scope limitations.

2.1.2 Kind of internal control weaknesses

In the internal control over financial reporting, the internal control weaknesses are divided in three categories. The weaknesses are placed in one of the three categories based on the chance and the severity of the potential financial statement misstatements. In increasing order of severity, the

9 weaknesses that can exist are a control deficiency, a significant deficiency and a material weakness (PCAOB, 2004). The first category is a control deficiency. A control deficiency is the least severe type of a weakness in the internal control. Both the chance of a financial statement misstatement and the magnitude of the potential misstatement are the least severe. If there is a misstatement in the financial statements, the decision of a financial statement user would not change if they had known the impact of the control deficiency at the moment they made their decision. A significant deficiency is more severe than a control deficiency but less severe than a material weakness. With a significant deficiency, there is a reasonable possibility that there is a misstatement in the financial statements. The existence of a significant deficiency has a more than inconsequential but less than material impact on the financial statements (Hammersley et al., 2008). The decision that an user of the financial statements would make if the misstatement was corrected could be changed but could also not change, it depends on the magnitude of the misstatement. The differences between a control deficiency and a significant deficiency are the probability and the magnitude of the financial statement misstatements. Both the probability that there are misstatements in the financial statements and the magnitude of these misstatements are higher if a company reports a significant deficiency. The most severe type of a weakness that can exist in the internal control over financial reporting is a material weakness. With a material weakness in the internal control, there is a reasonable possibility that there is a material misstatement in the financial statements. The probability of having a misstatement in the financial statements is therefore not different between a significant deficiency and a material weakness. The difference between these two types is the magnitude of the possible misstatement. With a material weakness, the magnitude of the misstatement is material. For a significant deficiency, as described earlier, the magnitude of the misstatement is more than inconsequential but less than material. Therefore, the impact that a potential misstatement has on the financial statements is higher. If the financial statements were corrected for the material misstatement, an user of the financial statements would have made a different decision.

Since the material weakness is the most severe, this kind of weakness in the internal control should be of the greatest concern to investors. Therefore, in this master thesis only the effect of reporting a material weakness in the internal control over financial reporting is researched.

2.2 Prior literature 2.2.1 Internal control

Since the passage of the Sarbanes‐Oxley act there is already done a lot of research to the effect that this act has on all kind of factors. Most of these articles are researching the effect of the introduction

10 of SOX section 302 or 404. Different effects have been researched so far and a couple of articles, which are important background information for this thesis, are outlined in this paragraph.

There are a couple of articles which have investigated whether there are determinants that make it more or less likely to report a deficiency in the internal control. Ge and McVay (2005) conclude that reporting a material weakness is positively related with business complexity and negatively related with firm size and firm profitability. The findings of Rice and Weber (2012) indicate that the probability of reporting a material weakness is negatively related with the need of external capital, firm size, non‐audit fees and the presence of a large audit firm. Reporting a material weakness is positively related with financial distress, auditor effort, previous reported weaknesses and recent auditor or management changes. A last article in this field is the article of Ashbaugh‐ Skaife et al. (2007). This article researched determinants prior to the mandatory reporting of SOX 404, they only researched SOX 302 reports. Their findings indicate that companies which report a material weakness in the internal control have more complex operations, recent organizational changes, more auditor resignations and fewer resources available for internal control.

Another popular field of research is the effect that reporting a material weakness has on accrual quality. Studying the impact on accrual quality is one of the possibilities of assessing the reliability of the financial statements. The research of Ashbaugh‐Skaife et al. (2008) concludes that the quality of accruals for firms disclosing internal control deficiencies is lower relative to firms without internal control deficiencies. The quality of accruals in the article is measured by accrual noise and the absolute abnormal accruals. Another finding of this article is that firms with internal control deficiencies have significantly larger abnormal accruals, both positive and negative. This suggests that there is more noise for such firms and this leads to unintentional errors and therefore the financial statements are less reliable. In the article of Doyle et al. (2007) the accrual quality is also researched. The finding of this research is that firms disclosing a material weakness in internal control are associated with poorly estimated accruals, the estimated accruals are often not realized as cash flows. However, this finding is driven by company‐level weaknesses, which are more severe than account‐specific weaknesses. They find similar results if they use other measures of accruals quality such as discretionary accruals or average accruals quality. In the article of Bedard et al. (2012) they have studied the effect on accrual quality of remediating a material weakness in the internal control. They conclude that the remediation of a material weakness has a positive effect on abnormal accruals, the abnormal accruals for these companies are lower. Another conclusion of this article is that companies which report a material weakness in two subsequent years are significantly associated with an increase in abnormal accruals. This effect is found for all kind of material weaknesses and not only for the company‐level weakness but also for account‐specific weaknesses. The research of Epps and Guthrie (2010) has studied the differences in discretionary accruals

11 between firms with a material weakness in the internal control and a control group without a deficiency or material weakness. The findings of this article indicate that the existence of a material weakness allows for more earnings manipulation. This is present in income‐increasing and income‐ decreasing discretionary accruals settings. This means that a firm with a material weakness in the internal control has more opportunities to overstate or understate the earnings and therefore, the earnings of these firms are less reliable for financial statement users.

2.2.2 The efficient market hypothesis

One of the key theories in the financial world is the efficient market hypothesis. This theory states that security prices at any time fully reflect all available information (Fama, 1970). This means that security prices are informative for different users of the security prices. For example, an investor can make an optimal comparison between different securities and therefore can make an optimal choose about which securities to buy and which to sell. If the prices in a market fully reflect all the available information, this market is called efficient.

However, there are a few conditions which are assumed in the efficient market hypothesis in its extreme form. If the security prices fully reflect all available information, there are no transaction costs, all available information must be costless available to all the investors and these investors must agree about the impact that this information has on the security prices (Fama, 1970). If the theory can only be used if these assumptions hold, the theory would be worthless since these assumptions are not valid in the real world. There is evidence that the theory holds in specific situations. There is no evidence found against the weak form and the semi‐strong form and only limited evidence found against the strong form (Fama, 1970). Especially the semi‐strong form is important for this thesis since there is evidence found that the efficient market hypothesis holds if there is publicly available information announced. Examples of publicly available information are earnings announcements or reporting on the effectiveness of the internal control over financial reporting. So the security prices fully reflect this kind of information and even if the extreme assumptions do not hold, the efficient market hypothesis can still be used and therefore the theory is used now for almost 50 years.

There is a key implication for this thesis which is an outflow of the efficient market hypothesis. The assumption from this theory is that the stock price is the aggregate of the investors’ expectations on the sum of the future dividends. If there is new publicly available information, investors would change their expectation of the future dividends and therefore there is the expectation that the stock price changes. So if the report on the effectiveness of the internal control over financial reporting is new and informative in the eyes of investors, there is a stock price change.

12 2.2.3 Link between new information and the stock price

As mentioned in the previous section, the expectation is that there is a reaction measurable of the stock price if there is new information announced. In this section, there are some key articles described which have researched the link between new information and the stock price.

The most obvious link between new information and a stock price reaction is measured by studying the announcement of earnings. The first article which provided evidence for this link was the article from Ball and Brown (1968). This study changed the academic profession and the way how research is done. In this research, the stock price reaction to accounting numbers is studied. The finding of this study is that there is a stock price reaction to the announcement of accounting numbers. There is pre‐announcement drift, which means that investors anticipate good and bad news before the announcement date. Although there is pre‐announcement drift measured, the key conclusion of this article is that there is a stock price reaction just after the announcement of the earnings. This implies that there is new information announced in the accounting numbers and that this new information is valuable for investors. Another article about the link between new information and the stock price, is the article from Ball and Shivakumar (2008). In this article there is special attention for the new information, which information is not expected already by investors, of the announcement of earnings. The conclusion is that the earnings announcements provide 6 to 9 percent of the total yearly information incorporated in the share price. Although this percentage is lower than found in earlier research, there is still evidence found for the link between the announcement of new information and the stock price reaction.

The earnings announcements are off course not the only source of information that is publicly announced during a year. There are also other firm‐specific information announcements during a year. In the article of Kalev et al. (2004) 12 different categories of information announcements are distinguished. Examples of information categories are the announcement of a takeover or an asset acquisition. In this research, they have also studied each information category on their own. The findings of this research indicate that for only one of the 12 categories there is no stock price reaction measured. Therefore, almost all categories of information give new information to investors and not only earnings announcements are valuable for investors. There are also stock price reactions to the announcement of information known from the news. For example, after the announcement of a strike from pilots of KLM, the stock price reacted and dropped with a significant amount. Another very recent example is the stock price of Ordina. There was a fraud investigation at this firm. On the moment the fraud investigation became known, the stock price dropped significantly. After the internal fraud investigation was completed, the conclusion was that there were no structural problems. After the announcement of this news, the stock price increased.

13 However, the stock price dropped already significantly after the announcement of the investigation and the increase did not offset this decrease.

2.2.4 Link between material weakness disclosure and the stock price

After providing evidence for the link between new information and the stock price there are in this section articles outlined which have researched the link between the disclosure of a material weakness in the internal control and the stock price.

The article of Rezee et al. (2012) is one of the articles which has studied the link between the stock price and the disclosure of a material weakness in the internal control. The stock price reaction to the SOX 404 report is studied in this article, SOX 302 is not included although they control for this announcement. The results of this research find evidence that companies which report a material weakness in the internal control experience a significant decline in the stock price, the abnormal return decreases. Companies which report effective internal control experience an increase in the stock price, measured by an increase in the abnormal return. De Franco et al. (2005) have also done research on the stock price reaction to the disclosure of a material weakness in the internal control. In this research are not only material weaknesses investigated but also significant deficiencies and the impact that these deficiencies have are studied together. The mean stock price reaction is negative, the mean return is – 1.36 percent. The size‐adjusted returns are even more negative, a mean return of ‐1. 77 percent. This provides evidence on the link between material weakness disclosures and the stock price.

Beneish et al. (2008) have studied the effect on the stock price of reporting a material weakness under both sections regarding the internal control over financial reporting, SOX 302 and SOX 404. To research the link between material weakness disclosures and the stock price, they compare the group which report a material weakness with a group with similar characteristics but without any problem in the internal control. The results are somewhat contradictory. The authors find a significant negative stock price reaction to the announcement of a material weakness in the internal control under SOX 302, the mean abnormal return is ‐1.8 percent. However, there is no stock price reaction to the announcement of a material weakness in the internal control under SOX 404. Although there are several possible explanations given, there is no clear explanation why they do not find a reaction to the disclosure of a material weakness under SOX 404. Hammersley et al. (2008) also studied deficiencies in the internal control, not deficiencies reported under SOX 404 but deficiencies reported under SOX 302. In this research, there are separate groups: a control deficiency group, a significant deficiency group and a material weakness group. Especially important for this thesis is the stock price reaction to the material weakness disclosure. The mean size‐adjusted return for companies reporting a material weakness in internal control is ‐0.95 percent. Compared with the

14 returns of companies reporting a control deficiency or a significant deficiency, the material weakness return is significantly more negative. The stock price reaction to the disclosure of a material weakness is therefore more severe than for the other deficiencies.

2.3 Hypotheses

As outlined in the previous paragraphs, there is evidence that the stock price reacts on the announcement of the report on the effectiveness of the internal control. There are several potential theoretical problems of reporting a material weakness. The reliability of the financial statements is reduced, the cost of remediating the material weakness is high, the cost of capital increases and the audit cost also increase (Rezee et al., 2012). From these potential problems, the reduction of the reliability of the financial statements is the most logical explanation for the stock price reaction. The reasoning behind this is that companies which report a material weakness in the internal control over financial reporting have a higher chance of having a misstatement in the financial statements and the impact of this potential material misstatement has a major influence if there needs to be a correction (PCAOB, 2004). The uncertainty for companies which report a material weakness in the internal control therefore increases, since investors can rely less on the financial statements of the company. This has an influence on the risk associated with the shares of a company which reports a material weakness in the internal control, the risk increases. This increase in risk has an impact on the discount rate which increases and the increase in the discount rate ultimately reduces the stock price of a company. The hypothesis for companies which report a material weakness in the internal control over financial reporting is therefore:

H1: The disclosure of a material weakness in the internal control over financial reporting has a negative effect on the stock price.

Firms which report effective internal control over financial reporting are screening to the market that the reliability and quality of their financial statements is good. This gives investors the confidence that the financial statements demonstrate a true and fair view of the financial condition of the company and the uncertainty of using these statements therefore decreases (Hammersley et all., 2008). For firms which report effective internal control, there are two competing explanations (Rezee et al., 2012). The first explanation is that if the market expects that every company has effective internal controls, there is no stock price reaction measurable to the announcement of having effective internal controls since there is only a reaction measurable if a company has ineffective internal controls. The second explanation is that the market does not expect that every company has effective internal controls and therefore there is a positive market reaction if a company reports effective internal control. In this case, the stock price of effective internal control companies would increase since investors value the information that firms have effective internal controls. In this

15 thesis, the expectation is that the market rewards a company which has effective internal control. This means that there is a positive stock price reaction for firms which report effective internal controls. The hypothesis for companies with effective internal control over financial reporting is:

H2: Companies which report effective internal control over financial reporting have a positive stock price reaction.

However, it is likely that the severity of the stock price reaction is different for companies and two possible explanations for these differences are the borrowing capacity and the quality of the corporate governance. In this research, there is studied whether there is a difference in the stock price reaction between companies with a high borrowing capacity ratio and companies with a low borrowing capacity. For the quality of the corporate governance there is a same comparison made between companies with good corporate governance and companies with bad corporate governance. As explained above, the borrowing capacity is one of the factors which influences the severity of the stock price reaction. The borrowing capacity ratio is a ratio which provides information about the ability of a company to repay its existing debt within a short period. This ratio has also an influence on the company’s ability to attract new debt. The reasoning behind the impact of the borrowing capacity is that when companies report a material weakness in the internal control, they are seen by investors and other parties as higher risk companies. These companies are seen as higher risk companies because companies which have a low borrowing capacity are less able to repay their debt and the risk for investors and debt providers is therefore increased, it is less certain that they make a profit by investing in this company or get their money back (Cotter and Zimmer, 1995). Since reporting a material weakness in the internal control over financial reporting is also a sign that a company is of higher risk, investors want a higher compensation to compensate them for this risk. The risk for companies with a low borrowing capacity is higher in comparison with companies with high borrowing capacity and the market reaction would therefore be more severe for companies with low borrowing capacity. Another explanation is that if companies are less able to repay their existing debt, debt providers are not willing to lend new money to these companies. This will increase the interest rate that the company needs to pay if it wants to attract new capital and this will destroy firm value (Watts and Zimmerman, 1990). The consequence of this is that firms can not attract new debt capital or publish new equity capital. So for companies with a low borrowing capacity ratio, the impact of disclosing a material weakness in internal control is higher than for companies with a high borrowing capacity ratio since they are, after disclosing a material weakness, less able to access the capital market. There is evidence that the market reaction is different for companies with different borrowing capacities. Nguyen (2013) concludes in his research that a market reaction is stronger when a company has a low interest coverage ratio. The hypothesis for the borrowing capacity is therefore as follows:

16 H3: Firms with a low borrowing capacity are expected to have a more severe negative stock price reaction to the disclosure of a material weakness in the internal control over financial reporting.

The corporate governance quality is also a factor which influences the severity of the stock price reaction. If the corporate governance of a firm is of good quality, the expectation is that the stock price reaction is less severe. There are several reasons behind this expectation. The first reasoning is that good corporate governance reduces agency problems. Investors receive more information about the structure and performance of the company and this reduces the risk of making a bad investment decision (Berthelot et al., 2010). A second reason is that if the corporate governance is of good quality, the investors have confidence that the firm is managed effectively. Although the disclosure of a material weakness in the internal control is still bad news, investors have the confidence that the disclosure of a material weakness is an one time problem or is even a problem which is outside the control of management. The corporate governance quality is a relatively recent factor which becomes all more important. There is already evidence for the impact the quality of corporate governance has on the stock price. Ergin (2012) provides evidence that investors find information about corporate governance valuable, he finds evidence that there is a stock price reaction measurable to the disclosure of corporate governance ratings. There is a positive reaction if the quality of the corporate governance is seen as good and a negative stock price reaction if the corporate governance is of low quality. In a research of Goh (2009) there is researched whether the board of directors structure is associated with the remediation of material weaknesses in the internal control. The results of this paper show that companies with a stronger board of directors, measured by the number of independent board members, size and meeting frequency, are better able to remediate material weaknesses in a timely manner. The hypothesis for the corporate governance quality is:

H4: Companies with a good corporate governance quality, estimated by the percentage of non‐independent board members, have a less severe negative stock price reaction when disclosing a material weakness in the internal control over financial reporting.

3. Data and Method

In this chapter, the data which will be used and the method of resource will be outlined. In the first part of this chapter, the sample period and the sample selection are described. In the second part, the empirical models for this thesis and the empirical measures will be described. The databases which are used are also given in this part.

17 3.1 Sample selection

In this thesis, there is an empirical archival research method used. In other words, this research method is called a database research. In this research, the sample constructed has only US listed companies since these companies are obliged to report on the effectiveness of their internal control over financial reporting under SOX 404. Although, there are also other countries in which it is needed to report on the effectiveness of the internal control, these countries do not use SOX explicitly and are therefore not useful for this research. By using US listed companies, it is also easier to compare the results of this thesis with prior literature since all the previous articles written about this topic also use US listed companies which report on the effectiveness of their internal control over financial reporting under SOX. In the sample, companies which have a listing at different stock exchanges in the US are used. Data of financial service companies (SIC codes 6000‐6900) are excluded from my analysis. These companies are excluded since their stock price reacts differently to the announcement of a material weakness in the internal control in comparison with other business segments. This is explained in the Rezee et al. (2012) paper with the explanation that financial institutions with assets of an amount greater than 150 million have been subject to internal control reporting under the Federal Deposit Insurance Corporation Act of 1991. Utility firms can be allowed more revenues by the regulator to cover the cost of compliance with SOX 404. It is also usual to exclude these firms for this kind of research (see Rezee et al., 2012 and Hammersley et al., 2008).The sample period which will be used in this thesis is the period 2013. This time period is especially chosen since the longer term effect of SOX is studied and therefore the most recent data is the best measure for the longer term effect. With this time period, the longer term effect of reporting on the effectiveness of the internal control over financial reporting can be studied.

The initial sample of firms where the auditor concludes that there is at least one material weakness in the internal control over financial reporting in 2013 is 195 companies. After the exclusion of the financial service companies, the sample has 169 observations. Due to data unavailability for some variables, there are 50 observations dropped. There is also researched in the Edgar database whether companies did not file another document in the event window of three days. If a company filed another document, this company is excluded from the sample. This resulted in the exclusion of 11 observations from the sample. The companies which have all the necessary data and have no other filings in the event window are 108 companies. This sample is used to test the first and the third hypotheses. For the fourth hypothesis, there was an additional variable needed, a corporate governance proxy. From the 108 companies, 32 companies had the needed data and therefore the sample for the fourth hypothesis is only 32 observations. For the second hypothesis, there is information needed for a sample of companies which report that the internal control over financial reporting is effective. Since the sample of firms which disclose a material weakness is 108

18 observations, there are randomly picked 108 companies for which the auditor concludes that the internal control is effective. The sample selection procedure is outlined in table 1.

Table 1: Sample selection procedure

Initial sample of firms which disclose a material weakness in the internal control 195

Excluded: Financial service companies 26

Companies which did not have the necessary information in the Compustat database 43

Companies without data in the CRSP database 7

Companies with other filings in the event window 11

Total excluded observations of the initial sample 87

Total observations useable for the research 108

3.2 Research design 3.2.1 Empirical models

I use different tests to assess my hypotheses in this thesis. Firstly, the stock price reaction is measured to the conclusion of the report on the effectiveness of the internal control over financial reporting. There is a t‐test done to see if the report on the effectiveness of the internal control is significantly impacting the stock price. In this test, I test if the abnormal return of a company which discloses a material weakness is significantly different from zero. For the sample of firms which report effective internal control, I also test if the abnormal return is significantly different from zero. This test is done for the sample of companies which report a material weakness and for the sample which report effective internal control. Secondly, there are different control measures added to the first test which could also have an impact on the stock price reaction. For the second test, there is the following empirical model used:

SPR= β₁ MW + β₂ Size + β₃ Profit + β₄ Leverage + β₅ Big4

In this model, SPR is the cumulative abnormal return of a company’s stock price over a three day event window, MW is a dummy variable which equals 1 if a firm reports a material weakness in the internal control over financial reporting and 0 otherwise, size is a control variable which measures the size of the assets of the company, profit is a control variable which measures the profitability of a company in a certain year, leverage is a control variable to measure the debt to asset ratio and Big4 is a variable which indicates whether a big4 auditor is hired or not. How these variables will be measured is outlined in the next section. The above model is used to test the first two hypotheses.

19 There are new variables added to the empirical model which are used to test the last two hypotheses. To test the third and fourth hypotheses, only the companies which report a material weakness in the internal control over financial reporting are used. To test the third hypothesis, there is a new variable added to the first empirical model. The variable which is added to the model is a dummy variable for the borrowing capacity of a firm. The empirical model to test the third hypothesis is:

SPR= β₁ BCR1 + β₂ Size + β₃ Profit + β₄ Leverage + β₅ Big4

In this model, all the variables are the same as explained for the first empirical model. The additional added variable BCR1 is a measure for the borrowing capacity of a firm. BCR1 is a dummy variable which equals 1 if the borrowing capacity ratio of a firm is higher than the median of the sample and equal to 0 if the borrowing capacity is lower than the median. There is no interaction with the material weakness dummy variable needed since this is always equal to 1 because the test is only done for companies which disclose a material weakness in the internal control.

To test the fourth hypothesis, there is a different variable added to the first empirical model, namely a corporate governance variable. The model which will be used to test the fourth hypothesis is:

SPR= β₁ CG1 + β₂ Size + β₃ Profit + β₄ Leverage + β₅ Big4

Just like for the second model, the used variables in the third model are the same as in the first empirical model. The new variable in this model is the CG1 variable. The CG1 variable is a dummy variable which equals 1 if the corporate governance proxy of a company is higher than the median of the sample and 0 if the corporate governance proxy is lower than the median. Just like with the borrowing capacity ratio, there is no interaction with the material weakness dummy since the test is only done for companies which disclose a material weakness. How the borrowing capacity and the corporate governance are measured is explained in the next paragraph.

3.2.2 Empirical measures

The variable of interest in this research is the report on the effectiveness of the internal control over financial reporting. The database which is used to collect the necessary information about the effectiveness of the internal control is the Audit Analytics database. In this database, it is easy to collect data about firms which disclose a material weakness and firms whose internal control is effective. The Edgar database is used to control if the auditor really concludes that there is a material weakness or that the internal control is effective. There are no problems found in the Edgar database. For the first empirical model, the sample of material weakness disclosure firms and the sample of firms with effective internal controls are used. For the other models, only the material weakness disclosure sample is used since there is researched whether there is a difference in the severity of

20 the stock price reaction between companies which disclose a material weakness in the internal control. The data about the report on the effectiveness of the internal control are used to test the first two hypotheses.

Another important variable is the dependent variable, namely the stock price. To measure whether there is a stock price reaction, daily return data are collected from the CRSP database. In this database, the daily return of many listed companies can be found. This database is used since this database is suitable for this research and is often used as a database for event studies (Corrado, 2011). To research whether there is a reaction measurable to the SOX 404 publication, the abnormal return for companies is measured around the event date. This means that the stock price reaction is measured the day before the event date, on the event date itself and the day after the event date. The event date in this research is the day the report on the effectiveness of the internal control over financial reporting is published, which is the file date of the document. To be more specific about the abnormal return measure in this thesis, this is the actual return over the event window minus the expected return corrected with a market index (MacKinlay, 1997). To calculate the cumulative abnormal return, the Eventus database is used. In this database, the market index which is used is a portfolio of companies which are listed on the same stock exchange and with more or less the same market capitalization. This gives a good indication whether firms have outperformed the market or underperformed compared to the market. The three day event window is chosen since this is widely used in prior event studies. This is outlined in a research of MacKinlay (1997), which founds that the mostly used event window is a three day event window. This variable is of interest for every hypotheses since the whole research is about the impact of different variables on the (severity of the) stock price of a company.

To test the third hypothesis, there are additional variables added to study whether the severity of the stock price reaction is different for firms with a high borrowing capacity in comparison with low borrowing capacity firms. There are different possibilities to proxy for the borrowing capacity of a firm. In this thesis, the proxy for the borrowing capacity is the interest coverage ratio. The interest coverage ratio is used to determine how easy a company is able to pay the interest on its existing debt. There are multiple ways to compute the interest coverage ratio such as using earnings before interest and taxes (EBIT), earnings before interest, taxes, depreciation and amortization (EBITDA) or the cash flows (Dothan, 2006). The used inputs to calculate the interest coverage ratio are the earnings before interest and taxes and the interest expense. These inputs are used since these are often used in prior research to compute the interest coverage ratio (for example Dothan, 2006). As explained in the previous paragraph, there is a dummy variable created from the interest coverage ratio. BCR1 equals 1 if a firm has a higher interest coverage ratio than the median

21 and 0 if a company has a lower interest coverage ratio than the median. The data to calculate the interest coverage ratio are from the Compustat database.

For the fourth hypothesis there is a new variable introduced. The new variable of interest for the fourth hypothesis is a corporate governance variable. In the existing literature about corporate governance there are many different variables used. Examples of variables used to measure the quality of the corporate governance are the number of independent board members, the number of meetings or using a one tier or two tier structure. To proxy for corporate governance in this research, the percentage of non‐independent board members inside the board is used. As the research of Goh (2009) concludes, the number of independent board members is associated with the timeliness of the remediation of the material weakness in the internal control. Therefore, these seems a good proxy for the quality of the corporate governance. After the calculation of the percentage of non‐ independent board members, there is a dummy variable created namely CG1. CG1 is equal to 1 if the percentage of non‐independent board members is higher than the median of the sample and 0 if this percentage is lower than the median. The data for this variable are from the ISS database. This was formerly known as the RiskMetrics database.

Next to the main variables of interest, there are also different control variables used in the empirical models. The first control variable is a size variable. There are different ways to measure the size of a company. In this thesis, the size of a company is measured by the natural log of the market value of the common equity. This proxy is used in several articles, for example Rice and Weber (2012), and therefore this proxy seems relevant for this research as well. The necessary items for computing the natural log of the market value of common equity can be found in the Compustat database. A second control variable is a measure for the profitability of a firm. I use the return on assets to proxy for the profitability of a firm. Income is measured by income before extraordinary items and the assets are measured by the total assets of a company. Income before extraordinary items is used since this is a better indication of the long term profitability of a firm. Extraordinary items are often non‐recurring items and unusual and are not very useful in determining the long term profitability of a firm (Cameron and Stephens, 1991). The data which is required to calculate the measure for the return on assets are collected by using the Compustat database. Another control variable which will be used in this research is a leverage variable. The measure for leverage in this thesis is the debt‐to‐ asset ratio. Although there are several critical points given in articles about this measure, for example Welch (2011), it is debatable whether the proposed other measures are better able in measuring the leverage ratio. Another reason why this measure for leverage is used is since this measure is widely used in prior literature and is still seen by a majority of researchers as a good proxy for leverage (for example Guney et al., 2012). The last control variable which will be used in this thesis is a Big4 dummy. The big4 dummy is a proxy for the quality of the external auditor and therefore relevant for

22 this thesis. Prior research shows that there is a difference between the market reaction for companies which hire a big4 auditor and companies which do not hire a big 4 auditor (for example Hammersley et al., 2008). This variable equals 1 if one of the four biggest auditing firms is hired and 0 if another auditing firm is hired. To know whether a company hires a big four auditor or not, the Auditanalytics database is used.

4. Results

In this chapter the results of the thesis are outlined. Firstly, some descriptive statistics is given and the interpretation of these statistics is provided. Secondly, a t‐test is done to study whether the stock price reacts on the report of the effectiveness of the internal control over financial reporting. Thirdly, a regression for the first empirical model is performed and the results are explained. Fourthly, the test for the second empirical model is done and the outcomes are explained. Lastly, the regression for the third empirical model is run and the results are outlined.

4.1 Descriptive statistics

Table 2: Descriptive statistics

Variable Mean Median Standard

deviation Minimum Maximum Abnormal return ‐0.0035032 ‐0.0012 0.0409638 ‐0.1722 0.1894 MW 0.5 0.5 0.5011614 0 1 Ln assets 7.313632 7.098137 1.923877 2.470216 12.75653 ROA ‐0.0025602 0.0327797 0.1819907 ‐1.718732 0.26586 DE 1.405109 1.050172 5.871931 ‐63.16753 38.0293 Big 4 0.8148148 1 0.38935 0 1 Borrowing capacity 0.5 0.5 0.502331 0 1 Corporate governance 0.5 0.5 0.5080005 0 1

In the above table, the mean, median, standard deviation, minimum and maximum for all the variables of interest for this thesis are outlined for the total sample of 216 observations. The mean abnormal return of the total sample is slightly negative, which means that there is a negative

23 abnormal return of 0.35 percent for the whole sample. What is also interesting is that the ROA, which measures the profitability of a firm, has a negative mean. However, the median is positive which means that the most firms are making a profit and that the mean is possibly influenced by some extreme observations. The DE variable, which is a proxy for leverage, shows also some interesting facts. Although there seems to be no problem with the mean and the median, the maximum and minimum values show that there are also for this variable some extreme observations. The mean of the big 4 variable is 0.8148148, which means that approximately 81 percent of the companies in the sample is audited by a big 4 company. There is also descriptive statistics available for the material weakness sample and the effective sample on their own but this is not tabulated. I will only highlight the main differences between the two samples. The mean abnormal return for the material weakness sample is ‐.0125093 whereas the mean abnormal return for the effective sample is .0055028. This means that companies which disclose a material weakness have a negative abnormal return in the event window of approximately 1.25 percent. The mean abnormal return of companies which report effective internal controls is positive, approximately 0.55 percent. Companies which disclose a material weakness are also smaller in size, have a negative ROA and have a lower DE ratio. Companies which have effective internal controls are bigger in size, have a positive ROA and have a higher DE ratio. Companies which have effective controls are also using more often a big 4 auditor in comparison with companies which disclose a material weakness.

4.2 The report on the effectiveness of the internal control and the stock price reaction

To provide evidence regarding the abnormal return of a company which discloses a material weakness in the internal control, a t‐test is performed in which there is tested whether the abnormal return in the event window is significantly different from zero. The expectation is that companies which disclose a material weakness in the internal control have a significant negative abnormal return. For the effective internal control sample the expectation is that the abnormal return is significantly different from zero and that these companies have a positive abnormal return. The results of these tests are outlined in table 3.

Table 3: T-test for the abnormal return

Material weakness sample Effective control sample

Abnormal return ‐0.0125093 0.0055028

T value ‐2.5688 2.2484

Significance level 0.0058*** 0.0133**

Number of observations 108 108

24 The results in the above table indicate that there is a mean abnormal return for the 108 companies with a material weakness in the internal control of approximately – 1.25 percent. The significance level shows that this abnormal return is significantly different from zero at a 0.0058 level. The significance level is a one‐tailed significance level because the test is to research or the abnormal return is negative, which means lower than zero. This provides evidence that the abnormal return is significantly different from zero and that companies which disclose a material weakness in the internal control over financial reporting have a negative stock price reaction. Therefore, this result suggests that companies which report a material weakness in the internal control have a negative stock price reaction suggesting that the first hypothesis should be accepted. For the effective control sample, the table shows that there is a positive abnormal return of approximately 0.55 percent. The significance level of the effective control sample is 0.0133. This is also a one‐tailed significance level because the test is to study or the abnormal return is positive, so greater than zero. The abnormal return of the effective control sample is therefore significantly different from zero at the 0.0133 level. The result for the effective control sample means that there is a significant positive stock market reaction to reporting that the internal control over financial reporting is effective. This provides evidence for the second hypothesis that companies which report effective internal control over financial reporting have a positive stock price reaction. Next to comparing the means with zero, I also tested if the mean of the material weakness sample is significantly different from the mean of the effective internal control sample (results not tabulated). The results show that the means are significantly different from each other at a 1% level, which is highly significant. This provides even more evidence for the first two hypotheses.

To summarize, the above results suggest that the stock price reaction is negative for firms which report a material weakness in the internal control and positive for firms which report effective internal controls. This is what was expected and the results find support for the first two hypotheses.

4.3 Effect of disclosing a material weakness in the internal control on the stock price of a company To test whether the disclosure of a material weakness is significantly related to the abnormal return of a company, there is an OLS regression of the abnormal return on specific variables performed. The dependent variable is the abnormal return in this regression and the independent variables for the first regression are the MW, lnassets, ROA, DE and big4. The meaning of all these variables are outlined in the empirical measure paragraph and will not be explained another time here. For the first regression, the material weakness sample and the effective control sample are both used. Since the lnassets, ROA and DE show some extreme observations, I excluded the lowest 1 percent and the highest 1 percent for these variables. This resulted in dropping 10 observations. The results of the OLS regression for the first empirical model are given in table 4.

25

Table 4: OLS regression of the first empirical model

Variable Coefficient T-value Significance level

MW ‐1.477 ‐2.18 0.031** Big 4 2.303 2.90 0.004*** Ln assets ‐0.021 ‐1.50 0.135 ROA 0.020 1.66 0.098* DE 0.015 1.24 0.216 Constant ‐2.237 ‐1.89 0.061*

Total number of observations is 206 observations for all variables. The coefficients estimate the change in the abnormal return in % as the result of a unit change in a variable. The significance level is based on a two‐tailed test. The */**/*** are significance levels at the 10/5/1% significance level respectively. The empirical model which is tested is as follows: SPR= β₁ MW + β₂ Size + β₃ Profit + β₄ Leverage + β₅ Big4

The R‐squared of this model, which is not tabulated, is 0.1038. The coefficients in the table estimate the change in the abnormal return as a result of a unit change in a variable. The most important coefficient of this regression is the MW variable. This variable equals 1 if a company reports a material weakness and 0 if a company has effective internal control. The coefficient for the MW variable is ‐1.477 and is significant at the 5% level. The meaning of the coefficient is that when a firm reports a material weakness in the internal control, the abnormal return experiences a ‐1.48% reduction. This means that the abnormal return of a company whit a material weakness is lower in comparison with firms without a material weakness, if all the other variables are held constant. This provides evidence for the first two hypothesis that the abnormal return is negative for firms which disclose a material weakness and positive for firms which report effective internal controls.

Some control variables are also significant. Significant control variables in the regression are the big4, ROA and the constant. The big4 coefficient is 2.303. This means that if a company hires a big4 auditor, they experience a 2.3% more positive (or less negative) abnormal return in comparison with companies which do not hire a big4 auditor. The big4 variable is significant at the 1% level. The ROA variable is also significant in this test. The coefficient of ROA is 0.020. The implication of this variable is that if a company is more profitable, since ROA measures the profitability of a firm, the abnormal return increases with approximately 0.02%. The ROA variable is significant at the 10% level. There are however also some control variables which are not significantly related to the abnormal return. The control variables which are not significantly related to the abnormal return are the lnassets and the DE variables. The lnassets variable is a proxy for size whereas the DE variable is a proxy for leverage. Both variables are not significant at any meaningful level. Interesting is that the lnassets variable reports a negative coefficient. This means that bigger companies have a less positive (or more negative) abnormal return. However, this is not the first research in which the size variable

26 is negative. Rezee et al. (2012) also find that the size variable is negatively related to the abnormal return. They are reasoning that the quality of the information is increasing with the size of a company. Smaller companies are associated with lower information quality and have less resources to spend on maintaining effective internal controls. Therefore, investors expect that smaller companies are more often reporting a material weakness and if they do report a material weakness it is not a big surprise and the stock price reaction is less negative. If a bigger company reports a material weakness, this is a surprise to the investor and the negative stock price reaction is more severe. This is a possible explanation why the lnassets variable shows a negative coefficient.

To summarize the results that are presented in table 4, there is support founded for the first two hypotheses. The MW variable, which equals 1 if a company reports a material weakness and 0 if a company has effective internal control, is negative and significantly related to the abnormal return. This provides evidence that companies which report a material weakness in the internal control have a more negative (or less positive) stock price reaction in comparison with companies which report effective controls. Therefore, it provides evidence that companies which do report effective internal control have a more positive (or less negative) stock price reaction. To conclude, the results in table 4 are supporting the first two hypotheses.

4.4 Effect of the borrowing capacity on the severity of the stock price reaction for firms which

disclose a material weakness in the internal control

For the third hypothesis, only the material weakness disclosure sample is used. This is done since I am curious whether the severity of the stock price reaction is different for firms with a high borrowing capacity in comparison with firms with a low borrowing capacity. To research or this is the case, a new OLS regression is run. In this regression the variable of interest is the int.cov. variable, a dummy variable which equals 1 if a company has a higher interest coverage ratio than the median and 0 if a company has a lower interest coverage ratio than the median. Just like for the regression in the previous paragraph, the lowest and highest 1% of the variables ROA, DE and lnassets are removed from the sample since these are extreme observations. The results are presented in table 5.