The influence of modified audit opinion on the
stock market and cost of debt: U.S. evidence
Student name: Carmen Necula Student number: 10622756
Supervisor: Dr. Alexandros Sikalidis
MSc Accountancy and Control, Accountancy Track Faculty of Economics and Business
Academic year: 2013-2014 June 22, 2014
2 Abstract
This study examines the impact of the announcement of a modified audit opinion on the market, more specifically on the stock returns and on the cost of debt, and the way it is perceived by the investors and creditors. The sample contains publicly held companies from U.S. with collected data from 2012, that had a continuous auditing, excluding those from the financial and real estate sector. The main focus was on the cumulative abnormal return and the interest rate for the analysis of stock market and respectively the cost of debt. Contrary to the majority of previous studies, the results suggest that there is no significant influence of a modified audit opinion over the stock returns and the cost of debt, since the investors and creditors do not perceive this information as value relevant for their decisions. However, the results may be biased by other concurrent information, by the methodology approach or the sample might not be representative for this research.
Key words: modified audit opinion, stock market, cost of debt, abnormal return, interest rate,
3
Contents
1. Introduction ... 4 1.1Background ... 4 1.2 Research question ... 6 1.3 Motivation ... 72. Literature review and hypotheses development ... 8
2.1 Theoretical background ... 9 2.1.1 Auditor opinion ... 9 2.1.2 Stock market ... 10 2.1.3. Cost of debt ... 11 2.2 Literature review ... 12 2.3 Hypotheses development ... 16 2.3.1 Hypothesis I ... 16 2.3.2 Hypothesis II ... 17 3. Research methodology ... 18 3.1 Sample selection ... 18 3.2 Variables measurement ... 20 3.2.1. Dependent variables ... 20 3.2.2 Independent variables ... 21 3.2.3 Control variables ... 21 3.3 Regression models ... 22 4. Results... 24
4.. The stock market reaction over the announcement of a modified audit opinion ... 24
4.2 The effect of modified audit opinion on the cost of debt ... 30
5. Limitations ... 35
6. Summary and conclusion ... 36
References ... 38
4 1. Introduction
1.1 Background
Nowadays, many companies are disclosing the audit report either voluntarily for improving their relationship with their stakeholders, or because they are required to do that, in order to give a fair assurance over the financial statements. The audit report is not always in the auditee‟s favor, because the auditor may find some inconsistencies that can affect the real image of the company. Depending on their significance, these are included in the explanatory notes to an unqualified audit opinion. No matter the type of the auditor‟s opinion is, this brings new information into the stock market attention. This can lead to some changes in stock prices so the value relevance of the auditor‟s opinion could be an important aspect that deserves to be analysed. Another aspect about this, is the information decision usefulness, more precisely, the way in which the auditor‟s opinion influences the investors‟ decision making process, about buying, selling or holding stocks within a specific company. Another category of users that could make use of the audit report is represented by the creditors, who have to be aware of the financial position of a company when deciding to give them credits.
The relationship between stock market and auditor‟s opinions has been studied by other authors too, using different samples and analyzing specific aspects of this topic. One of them is the going concern opinion, as part of a modified audit opinion.
For instance, Ogneva and Subramanyam (2007) examines the returns following disclosure of going concern opinions using a sample of U.S. and Australian firms, but they didn‟t find any pricing anomaly related to going concern opinion in either of these countries. This can be interpreted as the lack of under-reactions of the stock market following the disclosure of a going concern opinion.
Different results related to this topic, are emphasized in Taffler et al. (2004), which found a highly significant adverse price reaction over 1 year period, after a going concern modified audit report, between -24% and -31%, based on expected return benchmark used. They did find significant market under-reaction associated with a going concern opinion in U.K., but this conclusion is consistent with irrational investor behavior.
The information content is analyzed in Frederick Jones (1996), where the U.S. sample is used to prove that the independent auditor‟s going concern evaluation provides information to investors and supports the requirement of disclosing those uncertainties within the audit report.
5
Another interesting study about the information content of auditor‟s opinion is Dodd et al. (1984) because they exclude the adverse opinion from their sample, containing New York Stock Exchange firms. In their analysis, they took into consideration only qualified opinions and disclaimer of opinion and the results suggest that audit opinions have little impact on the value of common stock, but this doesn‟t mean they are less important.
One of the few studies that focused on the difference between the unqualified and the qualified opinions is Chen et al. (2000). The conclusion stated in this paper is that there is a negative association between MAOs and cumulative abnormal returns. Furthermore, they found no difference between market reaction to qualified opinions and market reaction to unqualified opinions with explanatory notes.
The cost of capital with its both cost of debt and cost of equity, has been in the main attention of many researchers from many years, as well.
One recent paper where the debt contracting has been the main focus is Chen et al (2012). The authors have examined the effect of qualified audit opinions on private debt contracts, proving that there is an association: “a qualified audit opinion is associated with an average increase of 18 basis points in the interest rate of loan facilities issued in the year following a QAO.”1 This have consequences in the future too, as they observed the interest rate effects of a QAO persist for three years, even after a clean audit report.
There are also studies that have made an analysis on the audit status of companies and the influence of it on the cost of debt. A relevant study in this concern is Kim et al (2007), based on privately held Korean companies, that are not required to get an external audit, but some of them are voluntarily doing it. They found that those firms with an external audit, pay a significantly lower interest rate on their debt as opposed to firms without an external audit. This is also supported by the fact that a change in a company‟s audit status from no audit to external audit (either voluntarily or mandatory) leads to significant savings in the cost of borrowing.
The relationship between accounting, disclosure quality and debt contracting is another main topic for a large number of papers. Bharath et al (2004) has examined the commercial bank loan contracts of publicly traded US firms and it has emphasized that “poor accounting quality
1 Chen, Peter F.; Tenho, Shaohua; Ma, Zhiming; Stice, Derrald E, 2012. Qualified audit opinions and debt contracting. The 3rd International
6
reflects limited information about borrowers‟future operating cash flows”2. This is translated by higher risks, which encourage the bank to increase their loan transaction cost in order to compensate for the information risk. Since the accounting quality is closely related to a quality audit report, I expect that a poor audit quality have similar effects on the interest rate.
These expectations are met in Karjalainen (2011), which examined a large sample of private Finnish firms in order to emphasize the value relevance of the audit quality in debt pricing. Their findings indicate that Big 4 audits correspond with a decreased cost of debt, and that for firms with modified audit reports this is higher than for firms with a clean audit report.
Regarding the level of disclosure, Jere et al (2005) and Sengupta (1998) both have proved there is an inverse relationship between interest cost of issuing debt and the quality of corporate disclosure. Considering this, I assume that a high quality audit report will provide significant information for the creditors, and their decision making process will be more accurate and the level of riskiness decreases significantly.
Considering there is no much evidence of the difference between the effects of both clean and modified audit opinions on the stock market reaction in the US market, I think my research will bring a significant contribution into the literature and also a societal one, offering a better overview over the auditor‟s opinion. As respects the cost of debt, there is a significant contribution as well, since this is the first study to analyse the modified audit opinion influence on cost of debt for U.S firms.
1.2 Research question
In my study, I want to focus on the effects of the auditor‟s opinion on the stock market and on the cost of debt, considering the unqualified or clean opinion and the modified audit opinion. Some other purposes related to this research question are investigating the value relevance of the auditor‟s opinion and the usefulness of information disclosed within the audit report, for the investors when, making their decisions.
Based on the above discussion this study examines the following research questions: - How does the auditor‟s opinion over the financial statements influence the stock market? - What is the effect of the auditor‟s opinion on the cost of debt?
2
7 1.3 Motivation
The main aim of this research is to bring into investors‟ and creditors‟ attention important information, given the changes that may happen on the stock market when different types of audit opinions are disclosed. The reason for focusing on these two categories of users, is that I consider them the most important for a company in order to keep their business profitable. Investors can significantly increase the profitability and creditors can help the companies make investments that have future important benefits. Furthermore, considering that debt financing is the predominant form of external financing for publicly traded firms in U.S. (Sengupta 1998), I assume my sample will be relevant for analysing the influence of a modified audit opinion.
In this study, I will take a different approach from prior literature, which has primarily focused on the stock market reaction by analyzing either the unqualified and qualified opinions or the going concern opinions of auditors. There are few studies which actually highlight the differences between types of opinions and the way they are perceived by investors and creditors, which encouraged me to focus on two types of opinions: the unqualified opinion and the modified audit opinion. The reason I chose these two types of opinions, is that the modified audit opinion can be viewed as a bad news and the unqualified opinion as a good news so they can be considered being opposite. The interesting part here is to analyze if the stock market reaction on auditor‟s opinion differs after disclosing one type of opinion or another and in which way.
Regarding the approach for analyzing the effect on the cost of debt, I did not find many studies regarding its relationship with the modified audit opinion, so I consider it would be interesting to show whether the effect of disclosing such an opinion is similar with those from Karjalainen (2011), already mentioned in the Background. Furthermore, I will also consider the BIG4 variable in my model in order to emphasize the influence of auditor‟s rank on the cost of debt, and the value relevance of audits to creditors.
This research will contribute to prior literature in a significant way, because it will bring into the reader‟s attention the differences in the market reaction following those types of opinions and the changes regarding the debt contracting. Also, through answering the research questions, the value relevance of the auditor‟s opinion and the usefulness of the information provided will be clearly emphasized.
Taking into consideration the previous papers, which are not aligned with the same conclusion regarding the effect of the audit opinion over the stock market, I think extending the
8
analysis to another more recent period - 2012 - and using as sample firms listed on U.S. market, will bring another contribution, from a scientific point of view. The paper will analyze recent data from U.S. which, although it has been analyzed before, there have been found contradictory results of the going concern opinion.
The stock market reaction to an auditor‟s opinion is important for both the company and for its stakeholders. This study could bring into stakeholders‟ attention that there are some other factors that can influence the reaction of stock market, besides auditor‟s opinion, so they should analyze other determinants too, when making their decisions.
The cost of debt is another important aspect for every company since loans are one of the way to get external financing. The extent to which a modified audit opinion could influence the cost of debt, would increase the accuracy in decision making of the lenders when evaluating debt covenants and the financial statements of a specific firm.
The rest of the paper is organized as it follows: section 2 reviews prior literature regarding the audit opinion and its influence on the market, provide some theoretical background for the most important concepts, and present the hypotheses of this research; section 3 develops the methodology used for this study and the empirical models; section 4 describes the reports and analyzes the results of the regression models; section 5 contains the limitations of the analysis; and finally, section 6 presents the summary and the conclusion of this study.
2. Literature review and hypotheses development
The structure of this chapter is as it follows:
I will first provide some theoretical concepts that are most important to know for understanding the relationships for the analysis;
Then I will present some key papers that have studied important aspects related to the research questions and I will rely on their results to find an easier way to my own findings;
9 2.1 Theoretical background
In my research, I will focus on two types of auditor‟s opinion: the unqualified opinion, which is the best opinion a company can receive, and the modified audit opinion, which can be unfavorable for the company, depending on the importance of the explanatory notes included in the report. The purpose of this comparison is to better highlight the differences between the two types of auditor‟s opinions and between their effects on the stock market returns and cost of debt.
2.1.1 Auditor opinion
The auditor‟s report is known as a primary means of communication between the auditor and the stakeholders of the audited entity, according to IAASB. The users of audited financial statements rely on the audit report and they consider it meaningful for their decision making process. But, since the global financial crisis began, the users are increasingly asking for more inside information to be included within the audit report, for a better transparency.
One of the most important part of the audit report is the auditor‟s opinion because it represents the assurance over the financial statements. The audit opinions can be classified as it follows: unqualified opinion, modified unqualified opinion, qualified opinion, adverse opinion and disclaimer opinion.
The unqualified opinion, also known as a clean opinion, states that the financial statements are representing in a fair and accurate way the view of the business, according to the accounting principles.
The modified unqualified opinion, also known as the standard auditor‟s report with explanatory notes is used when the auditor wants to bring some specific information into users‟ attention which may affect their decision making process. A very known opinion in this case is the going concern report, which states that the auditor has great doubts about the ability of the company to remain in business.
A qualified opinion contains some misstatements detected by the auditor, considered to be material but not pervasive.
For an adverse opinion, the auditor has found enough evidence to prove that the misstatements are material and pervasive.
10
The disclaimer opinion states that the auditor didn‟t find sufficient evidence over the financial statements and is unable to complete the audit mission.
I chose to focus on two types, more precisely, on the unqualified opinion and the modified unqualified opinion, because first of all, they are the most common among companies and secondly, because I consider they bring different information that could matter in a way or another for the constituencies. This means that it‟s expected to get different results with respect to the effects of their disclosure on the stock returns and cost of debt.
2.1.2 Stock market
In order to better understand the relationship I want to analyze, I will provide an explanation for stock market concept.
The stock market, also known as the equity market, is the market in which shares of publicly held companies are issued and traded through exchanges, giving investors a part of ownership and allowing them to participate in the achievements of the companies they own shares, but they are also exposed to the risk of losing money if the company suffers losses. By stock market reaction we can understand the changes that are happening in the market, through the movement in the price, volume or returns of stocks.
There have been studies, as mentioned in the Background, that proved the stock market underreacts following the going concern audit opinion disclosure. For instance, Taffler et al. (2007) is one of the papers that brings into discussion this anomalous market behavior, consisting in firms that are underperforming various return benchmarks. This is caused by “denying the bad news conveyed by a going concern audit opinion and trading at prices inconsistent with underlying value”3, which is explained by the existence of irrational or naïve investors. This fact is inconsistent with the market efficiency concept, which states that prices fully reflect all the publicly available information. The measurement procedure for this under-reaction of stock market is represented by the difference between returns when the investors take into consideration the going concern audit opinion and the returns when investors are ignoring this disclosure.
3Taffler, R.J., Lu, J., Kausar, A., 2004. In denial? Stock market underreaction to going-concern audit report disclosures. Journal of Accounting
11
The question that comes next is whether there are possible over-reactions of the stock market and the answer could be affirmative as well. One representative study on this topic is Schaub (2006), which explains this concept as the phenomenon through which investors initially compound the true impact of new information on the firm, in this case the disclosure of going concern opinion, determining stock prices go too low following bad news or too high, based on good news. This concept of overreaction is also against the stock market efficiency, as long as the investment is not profitable. The results show that greater initial price reactions are followed by larger stock price adjustments, which is consistent with the over-reaction hypothesis.
The aspect I want to emphasize here, is that the stock market efficiency is only a theory which can hardly be applied in an imperfect economic environment and that stakeholders should expect anomalies to happen.
2.1.3. Cost of debt
Capital is a fundamental element of a company for its business survival and for a high turnover. There are two major ways to raise money for a company: debt financing and equity financing, but the predominant one among companies is the debt financing.
Jensen and Meckling (1976) provides an important theoretical background over the cost of debt and its particular aspects, like the incentive effects associated with highly leveraged firms, the monitoring costs and the bankruptcy and reorganization costs.
For the incentives, they explain that creditors would not borrow a company for an amount much higher than the entrepreneur‟s investment, because the latter tends to engage in activities with a high profitability but a low probability of success, since he will benefit of most of the gains if the things turn out well, and if not, the creditors will bear the most of the costs.
In order to limit this managerial behaviour, the creditors could ask for covenants with provisions that impose constraints on management‟s decisions, including limitations of the riskiness of the investment project. All the costs associated with these type of covenants are the monitoring costs.
In case of bankruptcy or litigation the cost of debt increases, because there is much more risk involved for the lenders and they will pay for fixed claims for a price inversely related to the probability of the incurrence of these costs to probability of bankruptcy.
12
According to Sengupta (1998) there are two alternative measures of cost of debt: - the yield to maturity on new debt issues, and
- the total interest cost of new debt issues
In my study, I will consider the cost of debt as the interest rate, determined as the interest expense scaled by total liabilities.
2.2 Literature review
The subject of this paper is a difficult and controversial one, because there were different findings that came out of related studies regarding the effect of audit opinion over the stock market, as I have already mentioned in the Background. Regarding the connection with the cost of debt, this is quite a new topic since there were few studies that focused on it, as far as I documented.
One important study that outlines the stock market reaction to a going concern opinion is Herbohn et al. (2007), which focuses on the information content of the audit report, but also investigates the stock market anomaly following the going concern opinion, using Australian firms. The authors state that a going concern opinion of an audit report “is a significantly unambiguous „badnews‟ event for stock market participants and an adverse stock market reaction is expected.”4. But this expectation wasn‟t proved through their study, because they did find that the stock market has a significant negative reaction in the pre-event period rather than the post-event period. They explain this by Australian markets being well informed, so the announcement of a going concern opinion is a culmination of multiple prior events, and disclosing it doesn‟t actually give additional information to the market. I consider this study a interesting one, as long as the results are not consistent with the ones expected at the beginning.
The going concern opinion has also been analyzed from another point of view, specifically the ability to predict the bankruptcy, as stated in Chen and Church (1996). This can show a significant level of information usefulness for investors, who didn‟t know about the real situation of the company where they own shares. The authors found that, as opposed to receiving unqualified opinions, firms that receive going concern opinions “experience less negative excess returns in the period surrounding bankruptcy filings”5
. They focused only on one financial
4Herbohn, K., V. Ragunathan& R. Garsden (2007). The horse has bolted: revisiting the market reaction to going concern modifications of audit
reports, Accounting and Finance, 47:3, 473–493
5 Chen, K.C., & Church B.K. (1996). Going concern opinions and the market’s reaction to bankruptcy filings. The Accounting Review, Vol. 71
13
difficulty occurring after disclosing a going concern opinion, in order to prove the value information, but there are others too, like quarterly losses, suspension of dividends, default on debt etc.
Another view over information usefulness of the going concern opinions is given by Dennis O‟Reilly (2010), which shows the importance of disclosure for valuing company‟s stocks by investors. What makes this study more interesting is the fact of being the first one that uses an experimental approach for reaching the findings. More precisely, U.S. financial analysts were required to participate by email at two stock price estimates of a financial distressed company, one before and one after including the auditor‟s opinion. The author found significant stock price reductions, after receiving a going concern opinion, and the usefulness of the auditor‟s opinion is proved to be “greater when it provides a signal that differs from that which the market is expecting”6
. It will be interesting to see whether a qualitative research like this one is consistent with my conclusion, also based on the US sample.
Chen et al. (2000) focused on the differences between the qualified and unqualified opinions, by analyzing the valuation effect of modified audit opinions (MAOs), in the Chinese stock market. The environment here has some specific characteristics, which differ from the ones in U.S.. Some of them are related to the Chinese investors, who are not experienced in using financial information, others to Chinese auditors who usually disclose an unqualified opinion with explanatory notes instead of a qualified opinion, and other characteristics are related to the regulators, who are more strict and more offended in case of GAAP violations. By analyzing different forms and contents of modified audit opinions, the authors want to emphasize the role of the auditor in the market. They found that in an emerging market like the Chinese one, MAOs are negatively associated with the returns but there is no difference observed between the qualified opinion and the unqualified opinion with explanatory notes. Furthermore, investors don‟t react negatively until the second year after disclosure.
Allen et al. (2011) focused on the auditor‟s going concern opinion as a communication of risk investors are exposed to. This leads to a substantial shift in the structure of the market valuation for firms financially distressed and more importantly, “the results hold even after
6 Dennis M. O'Reilly, (2010) Do investors perceive the going-concern opinion as useful for pricing stocks?, Managerial Auditing Journal, Vol. 25
14
controlling for several other measures of financial distress”7. The analysis is conducted with firms that are receiving for the first time the going concern opinion, which as opposed to the other firms, have a greater valuation weight of the book value of equity. The authors succeed to prove the value relevance of this type of opinion, as long as it brings up to investors the potential risk of abandonment or adaption of firm assets.
As I noticed, there are many studies developed on one specific modifed report meaning the going concern opinion, but less on the modified audit opinion as a whole, under all its circumstances, as stated in Syou-Ching et al. (2009):
- going concern consideration
- change in the application of accounting principles for financial reporting as required under law - change in the application of accounting principles for financial reporting on a voluntary basis - uncertainty of events that may have significant impact on the future financial condition of the firm
- the adoption of another auditor‟s audit report for the current year audit.
The reason for which the going concern opinion has got more attention from researchers could be that the information provided is much more relvant for investors as opposed to the one given by other issues, which may not influence in the same way the investment decisions. This assumption is proved by Syou-Ching et al. (2009) which mentioned that “of the five circumstances mentioned above, the one based on the going concern consideration had the highest significant impact on stock prices”8
. In their research, the authors have adopted Ohlson‟s model for investigating the information content of MAO, in order to control for concurrent information that could bring significant biases to the analysis, since they consider the event approach a “validity trap”.
As Firth (1978) states, investors react differently to the various types of audit qualification, meaning the type of audit opinion gives different information, and that the price reactions occurred in the expected direction, immediately after releasing the audit reports. This is an important study, that appears as reference for many recent articles, and that‟s one more reason
7
Allen D. Blay, Marshall A. Geiger, David S. North (2011) The Auditor's Going-Concern Opinion as a Communication of Risk. A Journal of Practice & Theory: May 2011, Vol. 30, No. 2, pp. 77-102.
8
Syou-Ching Lai, Cecilia Lin, Hung-Chih Li, Frederick H. Wu (2009). The Information Contents of Modified Unqualified Audit Opinions under
the Control of Concurrent Information: The Case of Taiwan, Journal of Accounting and Corporate Governance, Volume 6 Number 1, June 2009
15
to mention it, for proving the differences between the two types of audit opinions I chose to analyze.
One of the main articles I have relied on in my research is Chen et al (2012), that came with significant findings regarding the audit opinion influence on the debt costing. They have focused only on the qualified audit opinion, which is associated with decreased use of performance pricing provisions in debt contracts. They also bring into discussion the importance of the quality accounting information, with an inverse relationship: the higher the quality is, the lower the cost of monitoring the borrower, supported by lenders, which leads to a lower interest rate demanded by them. Since the qualified audit opinion is giving outside investors a signal over the financial statement quality as being lower than a clean opinion, this will determine lenders to rely less on the financial statement numbers, and make use of other monitoring mechanisms, that imply higher costs.
Another study I have considered in my research is Karjalainen (2011), which has emphasized that both audit quality and the auditor‟s opinion is perceived as value relevant sources of information in debt pricing. The author has also identified a mixed evidence in the prior literature regarding the extent to which the auditor opinion can affect stock prices and cost of debt. This can be explained by the methodological problems as “identifying the first day of trade on the audit report information”9
and also the “difficulty of isolating the market reaction to the information content of the auditor‟s opinion from the reaction to other concurrent information”10
.
The relationship between the auditor choice and debt pricing, was analysed in Pittman and Fortin (2004), and later in 2007, the authors extended their research from publicly to private traded companies. In the first paper, they have defined the cost of debt as the interest rate on the debt of the firm, calculated as the interest expense for the year divided by the average short and long term debt during the year. But in the second one, they used as a dependent variable the yield spread as “the difference in basis points between the at-issue yield to maturity on the corporate bond and that of a U.S. treasury bond issued on the same date with comparable maturity”11.
9 Karjalainen, J. (2011), Audit Quality and Cost of Debt Capital for Private Firms: Evidence from Finland. International Journal of Auditing, 15:
88–108
10 Manuel Cano-Rodríguez (2010). Big Auditors, Private Firms and Accounting Conservatism: Spanish Evidence, European Accounting Review,
19:1, 131-159
11 Fortin, S. and Pittman, J. A. (2007), The Role of Auditor Choice in Debt Pricing in Private Firms. Contemporary Accounting Research, 24:
16
As opposed to their first research, they were unable to support the value relevance of BIG4 audits in pricing bond issues. They concluded that the auditor characteristics are irrelevant to the debt-contracting process for the private firms: “neither the presence of BIG4 auditor nor auditor tenure explains debt pricing12”.
One recent article that got my attention is Armitage and Marston (2014) which focuses on the views of finance directors about the link between disclosure and the cost of capital, using a qualitative methodology, through conducting 16 semi-structured confidential interviews. The main question was whether greater disclosure to a rating agency or bank would affect the cost of a bond issue or a bank loan. In most of the cases the answer was affirmative: nine of the interviewees said that providing more information would increase the availability of debt or reduce the cost of debt, one said it would have no effect and the rest of six were not sure. Considering these views, I assume that providing an external audit, would have the same effect, since this could be one way of increasing the corporate disclosure, especially when the auditor is from Big4.
2.3 Hypotheses development
In order to provide a clear answer for each of the two research questions of my study, I will develop two separate hypotheses that will be explained later in this section. The link between them is the modified audit opinion, which remains the aim of the research along the whole study, since I want to emphasize the differences from a clean opinion.
2.3.1 Hypothesis I
The first effect of a modified audit opinion I want to measure is the one on the stock market, by analysing the daily returns based on stock prices.
As I noticed in previous studies, Chen and Church (1996) and Allen et al. (2011), the going concern opinion is seen as a bad news because it announces financial difficulties of the company, but the question is whether the unqualified opinion is considered by investors a good news, which could encourage them to buy the specific stocks.
12 Fortin, S. and Pittman, J. A. (2007), The Role of Auditor Choice in Debt Pricing in Private Firms. Contemporary Accounting Research, 24:
17
The Chen et al. (2000) paper has a similar purpose, more precisely, analyzing if this type of information provided through disclosing a modified audit opinion can be evaluated by investors in a proper and timely way. According to Chen et al. (2000) there are differences between the qualified opinion and unqualified opinion but these may not be observed by the investors. The authors refer to one article from news to emphasize the “difference between the two types of opinions as lawful but unreasonable (unqualified with explanations) versus unlawful (qualified opinions)”13
. One problem here would be that some of the auditors are choosing an unqualified opinion with explanatory notes – referred to as the modified audit opinion – instead of a qualified opinion in order to keep their clients and in the same time conforming with auditing standards. This assertions is made for the Chinese environment, which has an emerging market.
Based on this, the first hypothesis is as it follows:
H1: Ceteris paribus, the announcement of modified audit opinions is negatively
associated with market returns.
2.3.2 Hypothesis II
Another important effect of the auditor‟s opinion over the market that worths to be analysed is the change in the cost of debt. This relationship between auditing and cost of debt has not been studied by too many researchers, as most of the papers are focusing on the connection with the accounting quality given by the way of disclosing the information to the market, as already mentioned in the Literature review.
The qualified audit opinion has been proved to have a negative reaction among creditors, which implicitly lead to a higher cost of debt, according to Chen et al (2012). Pittman and Fortin (2004) examines the impact of auditor choice on debt pricing and concludes that the services provided by Big 4 Auditors leads to enhancing the credibility of financial statements, which reduces the debt monitoring costs.
Another factor besides the auditor choice is the auditor opinion, as the main subject of this study. So I found it interesting to research on this relationship. The interest rate is the variable that need to be analysed in order to figure out the direction towards it moves, when the auditor‟s
13 CHEN, C. J. P., SU, X. and ZHAO, R. (2000), An Emerging Market's Reaction to Initial Modified Audit Opinions: Evidence from the Shanghai
18
opinion is unqualified or modified. Considering that both auditor‟s choice and auditor‟s opinion represent a direct influence over the audit quality, I expect that when the auditor opinion is modified - there are some issues in providing a fair image of the financial statements that need to be explained in the notes - the interest rate used by the company to pay its current debts, increases.
That being said, the second hypothesis is stated as it follows:
H2: Ceteris paribus, there is a negative relationship between the modified audit opinions
and cost of debt.
These two hypotheses show, in my opinion, the most important aspects of the relationship between auditor‟s opinion and the market, and also the differences between the two types of audit opinions I want to analyze. By supporting the hypotheses, a significant contribution will be brought to the prior literature, considering that there is little evidence of a comparison between opinions.
3. Research methodology
For this study, I will develop a quantitative methodology using statistical models to understand the assumptions stated in the hypotheses. The sample I used for examining the variables is extracted from firms that were continuously audited for the 2011 and 2012.
In order to give have a clear image about the methodology, I will separately present it for the two hypotheses, when there is the case.
Later in this chapter, I will present the steps for the selecting the sample, I will give an overview over the variables used and their measurement and finally the empirical models will be explained.
3.1 Sample selection
In order to keep the consistency in my study, I will use the same sample for proving both hypotheses, with the focus on the audit opinion, which is present in both empirical models as an independent variable.
19
The sample is based on the changes happened in 2012, for US companies. Since there are few papers that analyse the recent period, I consider this study will bring a big contribution to the literature. However, I took into consideration the previous year as well, in order to select only the companies that were continuously audited and to identify the companies that first disclosed a modified audit opinion. Moreover the changes in earnings are also computed based on the difference between those two years. More about the measurement of each variable is provided later in this section.
Regarding the statistical procedure, I got data from Audit Analytics and I selected only the observations that met some of the mandatory conditions, like having a valid date for the audit opinion, which is represented by the “Signature date of opinion” variable. Other important variables that I collected from this database, refer to the identifiers for the auditors and companies selected. Since I first started collecting data from this databases, taking into considerations the restrictions above mentioned, I got for the initial sample 14,659 firm year observations for 2012 and 2011, out of which only 6192 refer to the focus year of this study, 2012.
I used Compustat database for collecting financial and accounting data like assets, liabilities, net income, equity and cash flow from operating activities. Besides these, the codes for the auditor and auditor opinion are also available in Compustat, which give a more detailed information regarding the type of opinion, since there are all included.
In order to assure the consistency in the selection of companies, I have used only the company identifiers of the firms obtained from Audit Analytics, and after this, the sample was reduced to 3,921. But in order to assure the validity of the research, I have applied some other restrictions, regarding the following:
- the stock ownership code 0 for publicly traded companies
- the SIC codes excludes the 6 category for the financial and real estate sector
- the GIC industry code was also restricted, by eliminating the financial area (the category 40) - the unaudited companies were excluded, since the focus of this study is given by the auditor opinion
- the status of the company has to be active, since I am looking at the impact over the stock market
After applying the restrictions afore-mentioned and after excluding those observations with missing data regarding the financial variables like interest expense, cash flow from
20
operating activities, liabilities, common shares outstanding and others, the sample was reduced to 663 firm observations for both years, but only 332 for 2012.
Next step in the sample selection, was to collect data from CRSP regarding the daily stock data for each of the companies identified before. After removing the missing values among observations and after merging all data from the 3 databases, there were left 287 firm observations of 2012. Because of the limited number of observations, the final sample, has only the codes 1 and 4, for the auditor opinion variable, where 1 is for a clean opinion and 4 is for a modified audit opinion. Considering this, the only difference that can be analysed here is the one between these two types of opinion.
As stated in prior studies like Jones (1996) and Herbohn et al. (2007), it is important to consider only the first time going concern audit reports, because this is the best way to analyze the information content of the disclosure. For successive going concern audit reports, the usefulness of the announcement is reduced, because the market is already aware of the real situation of the firm from previous years. This will make necessary the examination of previous audit report for each firm. Because is quite difficult to follow the historical data for each company, I will implement this restriction only by considering the previous year in order to get a more accurate data. Since the going concern is one cause for a modified audit opinion, I will consider the same approach as in the papers mentioned above by determining the companies that first have a modified audit opinion, considering the previous year as well.
3.2 Variables measurement
In this section I will explain the measurement of each variable (more in Appendix no. 1), dependent and independent, as well as the rationale behind the control variables included, and for this I will separately consider the two hypotheses.
3.2.1. Dependent variables
In order to reflect the real impact of disclosing the auditor opinion for the investors, so basically the influence on the stock prices, the most used proxy is the cumulative abnormal returns, which is the dependent variable for the first hypothesis.
For calculating this, the abnormal returns around the announcement date of the audit opinion, will be determined. To reflect the immediate effect of disclosing the audit opinion, I
21
considered the abnormal returns one day before and one day after the event, given by the date of disclosing the audit opinion. The abnormal return is determined after running a regression where the return of the firm i on day t is the dependent variable, explained by the market return on the day t. The period taken into consideration is from day (-120) to day (-30) as in Chen et al (2000). The reason for choosing this period is to control for the effect of other external factors that could increase the noise in the model.
Another aim of this study is to show the impact of auditor opinion over the cost of debt, as stated in the second hypothesis, and for this I will use another dependent variable, the interest rate, measured as the interest expense divided by the average short and long term debt (total liabilites).
3.2.2 Independent variables
For the first hypotheses, one of the independent variables is given by the ΔEPS which reflects the change in earnings per share (EPS2012 – EPS2011) scaled by the beginning price at day (−1).
Another independent variable is given by the interaction term OP*ΔEPS, where the OP takes value of 1 if there is a modified audit opinion and 0 otherwise.
For the second hypothesis, I included as independent variables the leverage (LEV) as the book value of total short- and long-term debt deflated by firm market value, the cash flow (CF) from operations scaled by total assets, and the assets structure (ASSET_STR) measured as the total property, plant and equipment scaled by total assets.
3.2.3 Control variables
In order to mitigate the effects of other factors on the dependent variables, I will consider some control variables.
For the first and second hypothesis related to the influence of auditor opinion over the stock market, I will use the following control variables:
Firstly, I used a dummy variable for the audit opinion (OP) which takes the value 1 for a modified audit opinion and zero otherwise. The focus of this study is on the modified audit opinion so this is the reason why I will control the regression by this variable.
Secondly, I added the dummy variable DIV, which equals 1 for the modified audit opinions followed by a decrease in dividends between 2012 and 2011.
22
Finally, I included a dummy control variable REPEAT which takes the value 1 for the firms that have a modified audit opinion in the previous year as well, and 0 otherwise. This helps to better reflect the impact of a new information for the investors and over the stock prices.
For the second hypothesis, regarding the relationship between audit opinions and cost of debt I took into consideration some other control variables.
Firstly, the same dummy variable OP is for expressing the modified audit opinion, when it takes the value of 1, and 0 otherwise.
Secondly, I included a dummy variable to control for firms that have an auditor from one of the Big 4 audit firms (BIG4), which takes the value of 1 if this is the case, and 0 otherwise. It has been proved that services provided by Big4 auditors give a plus of credibility among creditors.
Thirdly, the negative book equity dummy indicates if the book value of equity is negative, which is an important indicator for the creditors when considering to lend or no a specific company.
3.3 Regression models
For the first hypothesis I used an event-study methodology in order to investigate stock market reaction to the auditor‟s opinion, based on U.S. listed firms.
The descriptive data will be divided in two periods: the pre-event period and the post-event period, where the post-event in this research is the disclosure of the auditor‟s opinion. In order to better emphasize the effect of auditor opinion I think it is better to analyse the immediate changes in the stock prices because in this way, the chances of other events‟ influences decrease significantly.
In order to estimate the daily abnormal returns, I will use the model developed in Chen et al. (2000) paper which is using the following formula:
𝑅𝑖𝑗𝑡 = 𝛼𝑖𝑗 + 𝛽𝑖𝑗𝑅𝑚𝑗𝑡 + 𝜀𝑖𝑗𝑡 (1) where
Rijt= the return of firm i in year j on day t, and Rmjt= the market return in year j on day t.
23
The stock returns (Ri) need to be calculated from changes in stock price (already adjusted for dividends), using the following formula:
𝑅𝐸𝑇𝑡 = (𝑃𝑅𝐼𝐶𝐸𝑡 − 𝑃𝑅𝐼𝐶𝐸𝑡−1)/𝑃𝑅𝐼𝐶𝐸𝑡−1
The firm-year specific parameters of the model (𝛼 and 𝛽) are estimated over a 120-day period, ending 30 days before the announcement date. The 3-day (from day −1 to day 1) cumulative abnormal return (CAR) is calculated as follows:
𝐶𝐴𝑅𝑖𝑗 = 1𝑡=−1𝑅𝑖𝑗𝑡 − 𝑎𝑖𝑗 + 𝑏𝑖𝑗𝑅𝑚𝑗𝑡 (2)
where
CARij= the 3-day cumulative abnormal return for firm i in year j
aij and bij= firm-year specific parameters estimated by using equation 1
Based on the model implemented by Chen et al. (2000) for the first hypothesis, I will develop the new model adapted to the sample of firms and considering only one year of analysis, 2012.
Therefore, the adjusted model looks like it follows:
𝐶𝐴𝑅𝑖𝑗 = 𝛼0+ +𝛼1𝑂𝑃𝑖𝑗 + 𝛼2∆𝐸𝑃𝑆𝑖𝑗 + 𝛼3𝑂𝑃𝑖𝑗 ∗ ∆𝐸𝑃𝑆𝑖𝑗 + +𝛼4𝐷𝐼𝑉𝑖𝑗 + 𝛼5𝑅𝐸𝑃𝐸𝐴𝑇𝑖𝑗 + 𝜀𝑖𝑗 (3) where,
CAR = 3-day cumulative abnormal return from day −1 to 1,
OP = a dummy variable with a value of 1 for MAOs and 0 otherwise, ΔEPS = change in earnings per share scaled by beginning price at day −1, DIV = 1 for MAOs with dividend decreases and 0 otherwise,
REPEAT = 1 for non-initial MAOs and 0 otherwise
For the second hypothesis, which implies the effect of a modified audit opinion on cost of debt, I will consider the model from Pittman and Fortin (2004), in order to adapt it for my case.
The following regression model will be used in order to analyse the relation between auditor‟s opinion and the cost of debt:
𝐼𝑁𝑇_𝑅𝐴𝑇𝐸𝑖𝑡 = 𝛽 + 𝛽1𝑂𝑃𝑖𝑡 + 𝛽2𝐵𝐼𝐺4𝑖𝑡 + 𝛽3𝐿𝐸𝑉𝑖𝑡 + 𝛽4𝐶𝐹𝑖𝑡 + 𝛽5𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽6𝐴𝑆𝑆𝐸𝑇_𝑆𝑇𝑅𝑖𝑡 +
𝛽7𝑁𝐸𝐺_𝐸𝑄𝑈𝐼𝑇𝑌𝑖𝑡 + 𝜀𝑖𝑡 where,
24
𝐼𝑁𝑇_𝑅𝐴𝑇𝐸𝑖𝑡 - Interest rate is interest expense divided by the average of total short and long term
debt during the year,
OP = dummy variable with a value of 1 for MAOs and 0 otherwise,
𝐵𝐼𝐺4𝑖𝑡 - it has a value of one when the firm retains a Big 4 auditor; zero otherwise,
𝐿𝐸𝑉𝑖𝑡 - the book value of total short- and long-term debt deflated by firm market value (sum of
the market value of equity and the book value of total debt), 𝐶𝐹𝑖𝑡 – is cash flow from operations scaled by total assets,
𝑆𝐼𝑍𝐸𝑖𝑡 - firm size is the natural logarithm of one plus total assets,
𝐴𝑆𝑆𝐸𝑇_𝑆𝑇𝑅𝑖𝑡 - asset structure is total property, plant and equipment scaled by total assets,
𝑁𝐸𝐺. 𝐸𝑄𝑈𝐼𝑇𝑌𝑖𝑡 - the negative book equity dummy indicates if the book value of equity is
negative.
4. Results
The focus of this study is on the modified audit opinion for U.S. companies in 2012. Considering this, I expect that all the variables used differ between companies when they disclose a clean opinion or a modified audit opinion. In order to show this difference, I will provide later in this section some relevant tables. Each of the two hypotheses try to explain a different influence of the auditor opinion, one over the stock market and the other over the cost of debt. In order to show a better overview, this section is presenting separately the results of each hypothesis.
4.1. The stock market reaction over the announcement of a modified audit opinion
As mentioned in the methodology part, I used for the first hypothesis an event research, which consists on the analysis before and after the event, which in this case is the announcement date of the auditor opinion. For this, it is necessarily to determine the abnormal returns for each company, depending on its unique date of announcement, and right after that, the cumulative abnormal returns over the 3 days window (-1,+1) are computed. This is the dependent variable, which will remain in the discussion for the analysis.
In order to determine the way in which the cumulative abnormal returns are caused by the type of audit opinion – clean or modified - it is important to take into consideration some other
25
variables which may or may not depend themselves on the type of opinion. Table 1 presents the descriptive statistics of these variables, for each case: when there is a clean opinion, OP equals 0 and for a modified audit opinion the same variable takes the value of 1. As it can be seen below, I had included some other important variables, that are not part of the model, like TA (total assets), NI (net income), ROE (return on equity), AR (abnormal returns for the day 0) and also the two numeric and continuous variables of the model, CAR (cumulative abnormal returns) and ∆EPS (the difference between earnings per share for two consecutive years, 2011 and 2012 scaled by the beginning price of day -1). The reason for including in the table these variables, is because they represent important financial indicators for the company, which are of a big interest for the investors. This is why I expect to notice a difference, but this doesn‟t mean that this is totally due to the difference in audit opinions. Anyway, it is interesting to have an overview of the main statistics for each of these variables.
The variables are corresponding with the 2012 year, except the last one, which states the difference among years in the earnings per share. Considering this, we can see a significant difference between the TA of the two groups, where the MAO firms are bigger in size, as evidenced by the mean and median values. They are also more profitable, as the mean and median of ROE are lower for the clean opinion group, which is contradictory with the Chen et.al (2010). However, the MAO group exhibits a more negative mean and median change in earnings per share than the clean group, fact that was also noticed for the chinese companies in the paper afore-mentioned. Moreover, the abnormal returns show higher mean and median values for the MAO group.
Having said that, apparently the firms that received a modified audit opinion are bigger and more profitable, but this success over the other group of firms, comes with its shortcummings like more risks, and more debts, which can cause the auditor to modify the audit opinion, with some explanatory notes of a significant importance for the investors.
26
Table 1
Descriptive statistics for main variables
OP Variable Mean sd p1 p25 Median p75 p99
0 TA 10,243.99 31,223.97 12.861 715.655 2,144.358 6,733.45 132,244 NI 630.9751 2,200.18 -985 23.706 109.768 449.7605 10,925 ROE 0.1345337 0.3711082 -0.7426036 0.0733614 0.1363148 0.1953806 1.17987 AR 0.0020013 0.0267145 -0.0738255 -0.0084635 0.0001363 0.0085047 0.1090047 CAR -2.842219 2.774461 -7.505228 -3.940635 -3.007319 -1.884142 1.151975 ∆EPS 0.0035035 0.1626191 -0.6953125 -0.0136327 0.0041084 0.0167856 0.312 1 TA 22,848.22 41,546.75 6,955 1,026.317 5,356.978 20,405.3 144,810.5 NI 1,808.556 4,060.8 -441 7.371 260.213 1,226 15,417 ROE 0.2106174 0.7505479 -1.46713 .1047595 0.1443036 0.1810418 3.595.745 AR 0.0083655 0.028134 -0.0498975 -0.004116 0.003105 0.0174242 0.0996667 CAR -3.021553 1.763847 -6.057692 -0.418.663 -3.135211 -1.866331 0.380801 ∆EPS -0.0209531 0.1083877 -0.2700403 -0.0318934 -0.0046982 0.0097703 0.3
Next, I will focus only on the variables that are part of the regression model. As I already mentioned in the Research methodology, most of them are dummy variables, which can take either the value of 1 or 0, depending on the situation applicable to each company. Table 2 shows the descriptive statistics of each of these variables, in order to get an idea about the structure of the sample. As you can see below, there are few companies that had a modified audit opinion, out of which 12 were followed by a decrease in dividends and 17 were having a modified audit opinion in the previous year as well.
27
Table 2
Descriptive statistics: dummy variables
Variables Freq. Percent Mean Std. Dev.
OP 0.0940767 0.2924453 0 260 90.59 1 27 9.41 Total 287 100.00 DIV 0.0418118 0.2005086 0 275 95.82 1 12 4.18 Total 287 100.00 REPEAT 0.0522648 0.2229493 0 272 94.77 1 15 5.23 Total 287 100.00
Table 3 presents the descriptive statistics for the other 3 variables of the model.
It can be seen that the cumulative abnormal return has a negative mean value, which prevent the investors about the financial distress and can influence their decision making process. These kind of signs can be caused by many external or internal factors, but in this study I want to focus only on the effect the audit opinion has on the returns, and implicitly on investors reaction.
Table 3
Descriptive statistics: continuous variables
Variable Obs Mean Std. Dev. Min Max
CAR 287 -2.85909 2.693792 -7.82156 33.1095 ∆EPS 287 0.0012027 0.1583274 -1.210686 1.791667 OP*∆EPS 287 -0.0019712 0.0332496 -0.2700403 0.3
28
The correlations between variables are also important in proving the validity of the implemented model. The results of this analysis are showed in Table 4, with Pearson correlations presented below the diagonal and Spearman correlations presented above the diagonal. I have chosen to run for two types of correlations in STATA in order to enhance the certainty over these relationships. The values reported are not equal, but most of them are close enough and the direction of influence is the same (positively or negatively).
Table 4
Pearson (below) & Spearman (above) correlations
Variable CAR OP ∆EPS OP*∆EPS div_MAO Repeat_MAO
CAR 1.0000 -0.0210 0.0151 0.0442 0.0553 -0.0006 OP -0.0195 1.0000 -0.0998 -0.1109 0.3501* 0.7287* ∆EPS 0.0954 -0.0452 1.0000 0.2988* -0.0567 -0.0529 OP*∆EPS 0.0238 -0.1843* 0.2112* 1.0000 -0.1691* 0.0526 DIV 0.0204 0.3501* -0.0487 -0.2553* 1.0000 0.1856* REPEAT -0.0093 0.7287* -0.0279 -0.1104 0.1856* 1.0000 * significant at the 0.01 level
As we can see in the table above, CAR is positively correlated with ∆EPS but negatively correlated with the dummy variable for MAOs (OP). The CAR is positively correlated with the change in earnings and negatively correlated with OP. Another negative correlation is between ∆EPS and DIV, since a modified audit opinion caused by a decrease in dividends has a negative impact on the earnings as well. However these correlations are not significant. For a significance level of 0.01 we can see positive correlations between OP and the other 2 dummy variables DIV and REPEAT, which is not surprising, but the OP are negatively correlated with the interaction term OP*∆EPS, which can be explained by a negative influence of the changes in earnings over the audit opinion.
In order to reduce any doubts about the multicolinerity of the independent variables, I used Variance Inflation Factor (vif in STATA) which has the mean value of 1.59, which is less than 10, and this means the variables are good and there is no need to drop any to be able to continue with the model. (the output is provided in the Appendix no. 2)
29
For the variables checking, I have performed another test in STATA, the one for specification error which basically checks whether more variables are needed in order to run the regression. In order to prove that there is no specification error I have ran the test (linktest in STATA) and I got a non-significant p-value of _hatsq, which means the model is correctly specified (the output is provided in the Appendix no. 3).
Another way of showing the difference between groups with clean opinions and groups with modified audit opinions is by performing a t-test (Table 5) of cumulative abnormal returns for each of these groups. This test is based on the idea of proving the null hypothesis that both groups have the same mean change in CAR. The alternative hypothesis is that this difference is not equal to zero. For this one, the p-value is greater than 0.1 which means the alternative hypothesis is not supported and implictly the null hypothesis cannot be rejected. Considering this, we might still wonder whether the sample is representative for the first hypothesis.
Table 5
T-test CAR by OP
Group Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] 0 260 -2.842219 0.1720647 2.774461 -3.181043 -2.503395 1 27 -3.021553 0.3394524 1.763847 -3.719308 -2.323799 combined 287 -2.85909 0.1590095 2.693792 -3.172068 -2.546113
diff 0.1793337 0.545525 -0.8944355 1.253103
The main results of the model for the first hypothesis are presented in Table 6, which shows the estimated coefficients for each independent variable, mentioning the significant level where is the case. We can see that the OP is negatively associated with the CAR, but not for a significant level, since the p-value is quite high (0.684). This means that the hypothesis regarding the negative influence of the auditor opinion, more specifically of a modified audit opinion, over the returns is rejected by this model for this particular sample. However the Chen et al. (2000) has proved that investors react negatively to the announcements of modified audit opinions, showing in the regression table a consistently negative coefficient for OP at a significant level for the last two year of their sample (1996 and 1997). The interaction between OP and ∆EPS is positive but not significant (p>0.1) and the other control variables are not significant either in this model, but this is also mentioned in Chen et al (2000).
30
Table 6
Regression results
CAR Coef. Std. Err. t P>t [95% Conf. Interval]
OP -0.3444228 0.845156 -0.41 0.684 -2.008063 1.319218 ∆EPS 1.603401 1.032965 1.55 0.122 -0.4299314 3.636733 OP*∆EPS 0.6602399 5.107573 0.13 0.897 -9.393722 10.7142 Div_MAO 0.5026697 0.8746699 0.57 0.566 -1.219067 2.224407 Repeat_MAO 0.1754674 1.053281 0.17 0.868 -1.897854 2.248789 _cons -2.857504 0.1685286 -16.96 0.000 -3.189242 -2.525765
One explanation for this concluding remark that the hypothesis is not supported could be the lack of sufficient data and the different economic period and region. As it is mentioned in Chen et al (2010), the chinese investors were not that well informed and for them “the auditor‟s opinions are the only source of financial information for their investment decisions”. But for U.S. companies in 2012, this is not applicable anymore, and investors are given a lot more options and reports to found out the information they need, without waiting for an audit report to confirm their expectations.
4.2. The effect of modified audit opinion on the cost of debt
For the second hypothesis, the attention is directed towards the cost of debt, represented by the dependent variable interest rate. As I already explained in the methodology, the model for testing this includes 3 dummy variables: the modified opinion (OP), BIG4 which equals 0 or 1 depending on the auditor‟s rank, and the negative equity (NEG_EQUITY) that equals 1 for a true statement and 0 otherwise. The other 4 variables are continuous, and they basically represent important indicators of a firm‟s wealth which is often taken into consideration by creditors in their decision making process.
In order to be consistent in my study, the same sample period and the same firms were used for the second hypothesis. There is also one independent variable that is used for this second model as well, represented by the auditor opinion, which is actually the main focus of this study. For showing the importance of the modified opinion, I showed in Table 7 the overview on the variables used, depending on the group which a company belongs to. Overall the differences in
31
variables for the two groups are not that big. If we take a look over the interest rate, the mean and median values are greater for those firms with a clean opinion but the leverage and the size are bigger for firms with a modified audit opinion. Furthermore, the mean value of negative equite is significantly greater for the MAO group, which could be an explanation for the need of auditors to provide some explanatory notes.
Table 7
Descriptive statistics
For a better overview of the variables, I sorted them in two categories as continuous and dummy variables, providing separate descriptive statistics to each category as it follows in Table 8 and Table 9.
Table 8 reflects the descriptive statistics for the variables that are of a significant importance for creditors. These basically represent the firm characteristics and they are supposed to control for variation in debt pricing.
OP Variable Mean sd p1 p25 Median p75 p99
0 Interest_rate 0.0207625 0.0154843 0.0002008 0.009074 0.0198404 0.0283679 0.0930809 BIG4 0.85 0.3577601 0 1 1 1 1 Leverage 0.532182 0.1901859 0.1475694 0.3918369 0.5365013 0.6547937 1.12824 Cash-flow 0.1068334 0.0756232 -0.0411275 0.0741339 0.1033383 0.1395552 0.3053475 Size 7.659011 1.850382 2.629079 6.574578 7.671056 8.814535 11.79241 Asset_str 0.283388 0.2168861 0.0167108 0.1222447 0.2190149 0.3894942 0.8648329 Neg_equity 0.0192308 0.1376 0 0 0 0 1 1 Interest_rate 0.0196538 0.018974 -0.0106401 0.0050063 0.0158887 0.0262194 0.0754443 BIG4 0.8888889 0.3202563 0 1 1 1 1 Leverage 0.5552865 0.171736 0.2178218 0.4641084 0.5412867 0.6142384 1.0197 Cash-flow 0.0799327 0.1543938 -0.6306254 0.0589993 0.1016327 0.1607889 0.2048676 Size 8.101633 2.631362 2.073801 6.934706 8.586342 9.923599 11.88319 Asset_str 0.2648026 0.2374075 0.0008627 0.0951337 0.1891273 0.3457143 0.8996205 Neg_equity 0.037037 0.1924501 0 0 0 0 1
32
Table 8
Descriptive statistics: continuous variables
Variable Obs Mean Std. Dev. Min Max
Interest_rate 287 0.0206582 0.0158102 -0.0106401 0.0952311 Leverage 287 0.5343556 0.1883689 0.1248841 1.347441 Cash-flow 287 0.1043027 0.0860692 -0.6306254 0.3227785 Size 287 7.700651 1.935689 2.073801 12.83958 Asset_structure 287 0.2816395 0.2185227 0.0008627 0.9200945
Table 9 presents the descriptive statistics for the dummy variables, where it can be observed the structure of the sample. Most of companies chosen are audited by BIG4 which gives an increased credibility of the audit report and investors are more willing to rely on it. For the negative equity variable there are only 6 firms that met this criterion.
Table 9
Descriptive statistics: dummy variables
Variables Freq. Percent Mean Std. dev
OP 0.0940767 0.2924453 0 260 90.59 1 27 9.41 Total 287 100.00 BIG4 0.8536585 0.3540656 0 42 14.63 1 245 85.37 Total 287 100.00 NEG_EQUITY 0.0209059 0.1433193 0 281 97.9 1 6 2.1 Total 287 100.00
For a better prediction of the influences of each variable the correlations of Pearson and Spearman are provided in Table 10. In order to see whether the correlations between variables are applicable, the significance for the 5% level is provided.
