0 Amsterdam Business School
Market reaction to going concern disclosures: The influence of
institutional ownership
Name: Ilse Carina Eland
Student number: 6278426/10000389
Date: 20/06/2015
Word count: 13.104
Thesis supervisor: Dr. Jeroen van Raak
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
1 Statement of originality
This document is written by student Ilse Carina Eland, 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 Economic and Business is responsible solely for the supervision of completion of the work, not for the contents.
2 Abstract
This paper examines the effect of institutional ownership on the reaction of the market after and before a going concern disclosure in the United States. Prior literature shows mixed results on the reaction of the market after a going concern and therefore more detail is given about institutional holdings. This paper makes a subdivision based on small, medium and large institutional holdings. Results show that there is no evidence for anticipating behaviour of institutional investors. However institutional investors do react less negative after a going concern opinion than non-institutional investors. Results suggest that the subtypes of institutional ownership react in a different way to going concern opinion disclosures.
3 Table of Contents
1. Introduction ... 4
2. Literature review ... 6
2.1 Definitions of institutional ownership ... 6
2.2 Going concern and current issues ... 6
2.3 AICPA, IAASB, PCAOB and Going Concern ... 8
2.4 Results prior research about market reaction ... 9
2.4.1 United Kingdom ... 9 2.4.2 Australia ... 10 2.4.3 United States ... 11 2.5 Institutional investors ... 13 2.6 Hypotheses ... 14 3. Research methodology ... 15 3.1 Data description ... 16 3.2 Empirical model ... 18 3.2.1 Dependent variable ... 20
3.2.2 Independent and control variables ... 20
4. Results ... 23
4.1 Descriptive statistics ... 23
4.1.1 Descriptive statistics total sample ... 23
4.1.2 Descriptive statistics divided sample ... 25
4.2 Pearson correlation test ... 27
4.3 regression results ... 29
5 Additional analysis ... 35
6 Conclusion ... 37
4 1. Introduction
The financial crisis of 2007 influenced the audit profession. After some major banks filed for bankruptcy after receiving a clean opinion, criticism by society on the audit profession rose. Questions as “where were the auditors” were asked since the audit report lacked of a warning of bankruptcy (Humphrey et al., 2011). When an auditor questions the ability of a firm to survive in the upcoming 12 months, a going concern opinion should be disclosed (Louwers et al., 1999). According to Louwers et al. (1999) the decision to issue a going concern opinion is one of the most difficult decisions in the audit profession. Although it is not an auditor’s main responsibility to predict a firms’ bankruptcy, investors expect from auditors to give a timely warning when firms have a high risk of being unable to survive in the following 12 months (Chen and Church, 1996).
Results of research about market reaction after disclosure of a going concern are mixed. Chen and Church (1996), Carlson et al. (1998), Citron et al. (2008) and Taffler et al. (2004) found a negative market reaction after disclosure of a going concern. However, Blay and Geiger (2001) and Herbohn et al. (2007) did not find significant evidence of a strong market reaction after the going concern disclosure. They did find evidence of anticipating behaviour of the market in the months prior to the going concern disclosure.
Prior literature suggests that in countries where the market is transparent, investors will already anticipate before a going concern is disclosed (Herbohn et al., 2007; Blay and Geiger, 2001). Also research other than going concern disclosures, show anticipating behaviour of investors. Ball and Brown (1968) conducted a research about accounting numbers and their information content for investors. They concluded that earnings and returns are related and that a lot of the abnormal returns were incorporated in prices before an earnings release. Ball and Shivakumar (2008) conducted a research about the amount of new information in earnings. Their results suggest that there is some new information in earnings for investors, but only to a limited extent. According to Bharath et al. (2008) private investors like institutional holdings have more information than public investors like individuals. This is the case because some firms are not willing to share information with the public but only with a limited amount of investors. To give more clarity about the market reaction after a going concern discosure, and because of mixed results in prior research, this paper looks at the specific role of institutional ownership regarding going concern disclosures.
Based on the research from Bharath et al. (2008) expectation is that institutional investors will have more information than non-institutional investors. However there is a lack in the literature about the types of investors in combination with going concern disclosures.
5
Therefore this paper conducts a research about the type of investors and their reaction to a going concern disclosure. The research question is: Do institutional investors react less after the disclosure of going concern issues than non-institutional investors?
This research contributes to the existing research in three ways. First this research has an empirical contribution since it adds extra information about the different types of investors in the somewhat limiting existing literature. A difference is made between the institutional investors and the non-institutional investors in order to get a more detailed view about the different types of investor reactions. Menon and Williams (2010) did conduct a research about the relationship between institutional holdings and market reaction. However they only looked at what happened after disclosure of a going concern and did not control for the time before disclosure, what this research does. Second, results in prior literature are mixed and differ per country and in countries self. Where some papers show anticipating behaviour of the investor group as a whole, others claim that there is a strong market reaction on the disclosure of a going concern. This research will focus on the different types of investors who can explain the difference among the reactions in different countries and prior literature. There has been some research conducted about the reaction of institutional investors on earnings announcement, however research on market reaction of institutional investors on going concern disclosures is very limited (Menon and Williams, 2010). This paper will contribute to that limited amount of research. Third, this paper makes a distinction between low, medium and large institutional holdings based on the research of Ali et al. (2008). This distinction is made since Ali et al. (2008) found different results per subgroup. The research of Menon and Williams (2010) focused on data prior the financial crisis. This paper looks also at the impact of the financial crisis by including data from the years 2007, 2008 and 2009.
Results show that there is no evidence for anticipating behaviour of institutional investors. However institutional investors do react less negative after a going concern opinion than non-institutional investors. Results suggest that the subtypes of institutional ownership react in a different way to going concern opinion disclosures.
The remainder of this paper is organized as follows. Sections 2 provides background literature and the hypotheses development. In section 3 the research method with the empirical model is discussed. Section 4 and 5 show the results of this research and in section 6 a summary and conclusion is given.
6 2. Literature review
In the literature review more background information is given. First different definitions of institutional investors and going concern disclosures are given. Secondly current issues about going concern opinions are discussed. Section 2.3 covers the different international regulations and the laws in the United States. In section 2.4 results of prior literature about the market reaction after a going concern opinion are given. Section 2.5 discusses the research conducted about institutional investors and earnings announcements. Finally the hypotheses are developed.
2.1 Definitions of institutional ownership
Although institutional holdings are an important notification in many research papers, there doesn’t seem to exist one clear definition of institutional ownership. Besides the websites of Investopedia or Wikipedia no clear definition is given of institutional investors. Only the website of Nasdaq (2015) gives an explanation of the 13F filings. According to this website information about institutional holdings is filed by major institutions on form 13F with the Securities and Exchange Commission. The website provides more information about the major institutions by adding:
“ Major institutions are defined as firms or individuals that exercise investment discretion, over the assets of others, in excess of $100 Million.” (Nasdaq, 2015)
Examples of these major institutions are insurance companies, pension funds, financial holding companies, banks and mutual/portfolio fund managers. Utama and Cready (1997) made also use of the 13F requirements to define an institution. They explained institutions are entities other than natural persons with investment discretion over at least $100 million of equity securities. According to Grinstein and Michaely (2005) institutional owners can influence the corporate financial policies in a company. They mention that institutional investors are one of the major investor groups in the United states and that institutions held more than 50% of US industrial firms’ equity in 1996.
So according to the website of Nasdaq the main differences between private and public investors is that public investors are individuals while private investors are companies and have a higher stake.
2.2 Going concern and current issues
This section of the literature review gives more information about the going concern opinion. What are the expectations of society with regard to the accountant about the issuance of a
7
going concern opinion and what are the risks of disclosure of one.
Financial statements are provided and prepared with the idea that the firm will continue in existence (Kauser et al., 2006). When an auditor doubts the survival of a firm in the upcoming 12 months, he or she should disclose a going concern opinion (Louwers et al., 1999). The going concern opinion should provide information to the financial statement readers, however there is a debate if these going concern opinions are useful (Menon and Williams, 2010). Some claim that financial statement users can make a better estimation of the survival of a firm than auditors. There also exists a viewpoint that auditors do not have the expertise to judge the going concern status of a firm and should focus on the assurance audits (Jones, 1996). However according to others, auditors have access to information unavailable for financial statement users and they have more knowledge to assess the going concern of a firm (Menon and Williams, 2010). In this way the work of an auditor and the opinion about the existence of a firm will influence the decisions of (potential) investors. According to Kauser et al. (2006) the auditors’ going concern is especially important since it is a statement from an independent party and gives a direct public-domain signal of a firm’s distress. According to Menon and Williams (2010) a way of judging the going concern and its usefulness will be to investigate the reaction of investors after disclosure of a going concern by measuring stock price.
Auditors are not responsible for predicting bankruptcy and a going concern opinion does not lead automatically to bankruptcy. However investors do expect a warning from auditors with expected financial failure (Chen and Church, 1996). Regulators, politicians, clients and media often criticize auditors for not giving an early warning and are calling of an audit failure if a company without a going concern opinion will suffer from bankruptcy (Carey et al., 2008). According to Carey et al. (2008) , auditors are more careful in providing a going concern opinion since they fear client loss and therefore less audit fees.
Some auditors are more reluctant to disclose a going concern opinion since they fear a self-fulfilling prophecy effect. Self-fulfilling prophecy means the going concern disclosure itself will lead to bankruptcy where otherwise when a firm would have obtained a clean opinion, it would have survived (Louwers et al., 1999). Prior literature shows mixed results on the existence of a self-fulfilling prophecy effect. Findings of the research of Louwers et al. (1999) suggest that auditors do not have to fear for a self-fulfilling prophecy effect. Chen and Church (1996) claim the opposite of the self-fulfilling prophecy effect by mentioning that the going concern has information value which can be used by the company to get financially healthy again.
8
Carey et al. (2012) describe the difference between a type 1 and a type 2 error. A type 1 error occurs when a company survives after the disclosure of a going concern opinion from the auditor. According to Carey et al. (2012) this should not be stated as an error since sometimes companies survive because of large capital injections and would have otherwise filed for bankruptcy. The auditor cannot predict a capital injection. A type 2 error occurs when a firm without a going concern opinion will fail. According to Carey et al. (2012) this is not always the fault of the auditor since some events cannot be predicted by the auditor.
2.3 AICPA, IAASB, PCAOB and Going Concern
In this section the different standards about going concern opinions are explained. Since this research will be conducted with U.S. data, a more detailed look is given about the regulations in the U.S.. Also some international standards are discussed.
The International Federation of Accountants (IFAC) makes international standards for auditors all over the world and is also concerned about ethics, quality and professionalism of auditors. One department of the IFAC is the International Auditing and Assurance Standards Board (IAASB) which develops accounting standards called International Standards on Accounting Standards (ISA). ISA 570 describes the responsibilities of the auditor with a going concern opinion. According to ISA 570 auditors are responsible for finding evidence about the going concern estimates of management. Even when management has not disclosed anything about the going concern, the auditor needs to investigate the survival probability of a firm (ISA 570). However this responsibility is limited. According to ISA 200 the auditor cannot foresee unexpected future events or cannot control for or influence them. An absence of a going concern opinion is not a guarantee for an entity’s survival (ISA 200).
In the United States the American Institute for Certified Public Accountants (AICPA) is concerned with the development of accounting standards (AICPA, 1988). These accounting standards are called Statement on Auditing Standards (SASs) and are developed by the Auditing Standards Board (ASB) of AICPA. In the US, SAS 34 was in 1988 replaced by SAS 59. Where SAS 34 was more passive about the auditor’s search for going concern issues and timely information, SAS 59 required more active auditor participation (Holder-web and Wilkens, 2000). According to SAS 59 the auditor needs to actively search to assure the going concern of a firm (AICPA, 1988). SAS 59 explains that if a firm will not survive in the upcoming 12 months without a going concern opinion, the auditor cannot be held responsible for that. However, according to Vanstraelen (2003) auditors in the United States are more often accused and in need to defend themselves in court than auditors in Europe.
9
As a reaction to different accounting scandals in the U.S., the Sarbanes- Oxley act was in 2002 introduced which established the PCAOB (Humphrey et al., 2011). The PCAOB is an independent non-profit organization which has as objective to oversee the audits of public companies in order to protect the interests of investors and public. Section AU 341 (2015) deals with the going concern. This section explains that an auditor should look at possible risks, not only in the financial statements but also in the macro-economic circumstances. If the auditor has substantial doubt about the continuity of the firm, an additional explanatory paragraph should be included with the going concern status. Foster and Ward (2012) investigated if these post-SOX changes in the going concern procedures are more useful. They concluded that going concern modifications after SOX added more incremental information to bankruptcy models than before SOX. The auditors anticipated better for future bankruptcy of companies after SOX.
This section shows the duties and responsibilities of an auditor with regard to the disclosure of a going concern opinion. These duties are prescribed by international standards setters and national law which can differ per country. For this paper only data is used from the United States.
2.4 Results prior research about market reaction
In this section some results of prior research about the market reaction on a going concern disclosure are discussed. These results are grouped per country because laws and regulations per country differ which can influence the results as described in section 2.2. These three countries discussed contain the most research conducted about the market reaction after a going concern opinion. First research in the United Kingdom is discussed, followed by Australian research and papers in the United States. This research is conducted with data from the United States. However to get a better understanding about the influences of the market , research results from the United Kingdom and Australia are included in this section.
2.4.1 United Kingdom
One of the first research conducted is by Firth (1978) about the impact of qualified audit reports on decision making of investors. According to Firth (1978) market reactions differ per type of qualified audit report but in general there is a negative market reaction on these reports. So the market will react more negative on a qualified audit report than it will on an unqualified audit report. Citron and Taffler (1992) added on the research of Firth by investigating the impact from a going concern opinion itself. According to them there is a positive relationship between company failure and the issuance of a going concern opinion.
10
However auditors have the tendency only to issue a going concern opinion if the likelihood of failure is very high in order to avoid risk of losing their clients.
Taffler et al. (2004) looked at the change in the stock price after the going concern disclosure on the long term. Results show that the market shows a negative reaction after a going concern disclosure. This post announcement drift can be explained by two reasons. One of these reasons could be that the market is in denial of the underperformance of a firm (Taffler et al.,2004). However according to Taffler et al. (2004) the reason for this post announcement drift is impacted by high transaction costs. Opportunities for arbitrage are restricted by these costs by government. There is a highly governmental regulated market in the United Kingdom, so even if the shareholders were aware of negative performance of a firm, they were not able to sell their shares due to high transaction costs. This research shows also that institutional holdings remained unchanged after disclosure of a going concern even in case of bankruptcy. Institutional investors and directors do not reduce their holdings after disclosure of a going concern opinion. According to Taffler et al. (2004) there are information processing errors explaining the under reaction of institutional investors.
Citron et al. (2008) examined the price sensitivity of going concern disclosures in the London market. They also found a significant market price reaction on the going concern disclosure. They did not find evidence that the reaction when disclosed earlier is higher than later disclosed information. Paradoxically, they did find that managers delay disclosure to the last possible date when a greater degree of financial distress is applicable. This is according to Citron et al. (2008) due to the inability of managers to give a timely information disclosure or because of the absence of regulatory penalties when information could have been disclosed earlier.
Most research conducted in the United Kingdom show a negative reaction of the market after disclosure of a going concern opinion. However different reasons are given by prior literature. Where one claims the transaction costs as main reason (Taffler et al., 2004), others point out the inability of managers and penalties (Citron et al., 2008). Also the research of Ogneva and Subramanyam (2007) who conducted the same research of Taffler et al. (2004) in the United States and Australia shows conflicting results.
2.4.2 Australia
Herbohn et al. (2007) investigate the market reaction prior, at the time of disclosure and after the disclosure of a going concern opinion. They did not find significant evidence of a market reaction short and in 12 months after the going concern disclosure. However they did find
11
evidence of anticipating behavior of the market in the 12 months prior to the going concern disclosure. According to Herbohn et al. (2007) this suggest that the market is already aware of the difficulties a firm is struggling with. This research was conducted in Australia and Herbohn et al. (2007) claim that the market in Australia is very transparent which explains the anticipating behaviour of the market. Carey et al. (2008) examined the issuance of a first time going concern disclosure and its potential costs for the auditor. According to them costs will rise for an auditor after disclosure of a going concern as a result of lost fees due to auditor switch of the distressed firm. Compared to Herbohn et al. (2007) this finding can give incentive for an auditor to not disclose a going concern opinion.
2.4.3 United States
This paper investigates the role of institutional ownership based on reaction after a going concern disclosure. Only data from the United States is used and therefore more information about prior literature in the United States is given in this section.
One of the first papers discussed is the one of Mutchler (1985) which does not look at the reaction of the market but examined the knowledge of the market about information about a firm. She conducted a research in order to discover to what extent investors can predict a going concern problem based on public available information. Results suggest that with a majority of firms from the United States the going concern disclosure did not add additional information. In some specific cases the going concern disclosure had a marginal information content. However, Mutchler (1985) did not mention an example of such a specific case. She also claims that even with a 100% prediction rate it is still unclear if the public available information in combination with the information in a going concern opinion disclosed by auditors were seen as important by financial statement users.
Some studies show that the market got some extra information after disclosure of a going concern. Carlson et al. (1998) and Jones (1996) conducted a research about the reaction of the market after disclosure of a going concern opinion on the short term. Both studies concluded that the information in the going concern opinion is useful for the market. By comparing the sample of distressed firms with firms that got an unqualified opinion, results of the study of Jones (1996) show that the market reaction after a going concern opinion which was not expected is even more negative. This result is supported by Chen and Church (1996). Carlson et al. (1998) controlled in their research for concurrent financial information disclosures around the days of the disclosure of the financial reports. Chen and Church (1996) also controlled in their sample for variables which could otherwise influence the
12
market reaction. These variables are: probability of bankruptcy, the market's reaction to the media's disclosure of financial difficulties prior to bankruptcy and changes in stock price prior to the issuance of the auditor's report. By controlling for these variables the market has no information prior the going concern disclosure and it was predicted in the hypotheses that the going concern disclosure would have a lot of impact on the market. Chen and Church (1996) found evidence that some companies manage to survive thanks to the going concern disclosure. An important note to make is that this research only looked at bankruptcy and not at other forms of financial distress.
Other studies didn’t find a significant market reaction after disclosure of a going concern (Blay and Geiger, 2001; Kauser et al., 2009). Blay and Geiger (2001) found no significant market reaction and conclude that a going concern opinion only provides incremental information to the market. This research controls also for the fact that management will sometimes choose for bankruptcy for strategic reasons. Kauser et al. (2009) conducted a research about the market response to a first-time going concern opinion and its withdrawal. According to them the market in the United States does not respond on a timely basis to a going concern opinion and this will lead to a market under reaction of -14% in the 12 months after disclosure. However withdrawals of going concern disclosures are processed in a timely basis. According to Kauser et al. (2009) an explanation for the under reaction has to do with high transaction costs which limit arbitrage opportunities. The same reason Taffler et al. (2004) gave for their study in the United Kingdom. So this prevents that stock prices are influenced by this information on a timely basis. This research is in conflict with the research of Ogneva and Subramanyam (2007) who deny the existence of a market anomaly.
Whisenant et al. (2003) did not investigate the relation between market reaction and going concern opinions but investigated the market reaction to disclosure of reportable events. They identify reportable events as internal control weaknesses related to reliability of management or financial statement reliability. According to them auditor changes in combination with disclosure of reportable events have information content for investors. Also they suggest that there is especially for financial statement reliability issues a negative stock price relation.
According to Menon and Williams (2010) going concern opinion reports provide useful information. According to them recent research did not find a negative market reaction on going concern disclosure (Blay and Geiger, 2001; Herbohn et al., 2007). Menon and Williams (2010) conducted their research in order to find the reason why these going concern reports provide useful information. Going concerns which cite financing problems get a more
13
negative market reaction than other going concerns since investors were not aware of these problems. They also claim that institutional investors make use of the information in going concern disclosures. This would be in contrast to the research of Bharath et al. (2008).
Prior literature in the United States about the market reaction on going concern disclosures provides mixed results. Where Blay and Geiger (2001) claim that the going concern opinion will only give some additional information, Menon and Williams (2010) found evidence that both investors and institutional holdings find going concern disclosures useful and react on those in the stock price.
2.5 Institutional investors
In this section prior literature about institutional investor behaviour is discussed. Since there has not been research conducted about the role of institutional ownership and going concern disclosures, besides the research of Menon and Williams (2010), this section will have a look at prior literature about institutional ownership and their reaction on earnings announcement days.
According to Menon and Williams (2010) there are two main reasons why institutional investors react in a different way than individual investors on accounting information. First of all, they argue that institutional investors have more access to information than individual investors. This reason is supported by Bharath et al. (2008) where they explain that firms are not willing to share information with the public but with only a limited amount of investors. As second reason Menon and Williams (2010) argue that there are differences in the amount of time and effort both institutional and individual investors can put in their investment decisions. According to Menon and Williams (2010) it is the second reason why institutional investors react more negative on a disclosure of a going concern opinion. However according to Kauser et al. (2006) institutional investors act on a timely basis and reduce their holdings in stocks of a firm with a going concern opinion over a two-year period surrounding the announcement.
Utama and Cready (1997) conducted a research about the relationship between ownership structure and trading volume at earnings announcements dates. They found information asymmetry between institutional investors and non-institutional investors. According to them this implies that individual investors are less well-informed than institutional investors and find earnings announcement more useful because this is new information to them. Research of Ali et al. (2008) continued the research of Utama and
14
Cready (1997). They divided the institutional investors in institutional holding low, medium and large stake. They argued that institutional holdings with low stakes have little incentive to discover private information since their stake is too small. On the other hand institutional holdings with large stakes are according to Ali et al. (2008) more dedicated investors or restricted to trade on announcement days due to corporate insider trading policies. Only the medium stake investors are more motivated to acquire more private information about the firm and have incentive to trade around earnings announcements because they have private information. So Ali et al. (2008) divided the institutional investor group in subgroups and found different reactions at earnings announcements.
2.6 Hypotheses
According to Menon and Williams (2010) institutional investors will react to disclosure of a going concern opinion. However they only used data after disclosure of a going concern opinion and did not control for anticipating behaviour. Other research about behaviour of institutional investors and earnings announcements show more anticipating behaviour of institutional investors since they have more inside knowledge (Kauser et al., 2006; Utama and Cready, 1997 and Ali et al.,2008). Therefore I predict that institutional investors will show more anticipating behaviour prior to the disclosure of a going concern opinion since they have more access to private information (Bharath et al.,2008). Therefore this research is based on the following hypotheses:
H1: Institutional investors will react less negative on a going concern disclosure than non-institutional investors
The expectation is based on prior literature that claims an information advantage and insider knowledge of institutional investors (Utama and Cready, 1997). Because institutional investors tend to be more aware of financial difficulties in advance, I expect that they will react on this information in a more timely manner.
H2: Institutional investors will show more anticipating behavior prior to the going concern disclosure than non-institutional investors
However according to the second argument of Menon and Williams (2010), institutional investors are better able to react on the disclosure of a going concern. The reason they gave is
15
that institutional investors do not have the resources and time to analyze a firm and will therefore depend more on the auditors’ opinion. Based on this research the following hypothesis is written.
H3: When the degree of institutional ownership is higher, the information after a going concern opinion will be processed faster than when a company has more individual investors.
Based on the research of Ali et al. (2008) a subdivision is made between the different types of investors as described in section 2.5. According to Ali et al. (2008), medium stake institutional investors are the ones who acquire private information and are putting an effort to get more knowledge about the company. Therefore the last hypothesis is about the role of the subtypes of institutional holdings.
H4: Medium institutional investors will react in a different way to going concern disclosures than small and large institutional holdings.
3. Research methodology
In this section, the collection of data is discussed. In the second part the model and the different variables are explained. The goal of this paper is to find information about the reactions to going concern disclosures of institutional investors relating the reaction of the total market. For more clarity figure 1 describes the goal of the paper in a model.
Figure 1 Model of the research explained
Company health Market reaction Control Variables Going concern opinion Abnorm al returns Size Big 4 Financial distress Percentage of instituional holdings
16 3.1 Data description
In this research the data used for the model is panel data. Panel data is a combination of cross-sectional and time- series. Cross-sectional means that the data consists of multiple firms. Time-series looks at multiple years per firm (Veenman, 2013). Both cross-sectional and time-series are used in this research.
As explained in the literature review, this research is conducted with data from the United States. Reason for this, as discussed in section two, is that mainly in the United States the results from prior literature about market reaction of a going concern disclosure are mixed and it is therefore the most interesting country to conduct the research. The time period is between fiscal year 2006 and end 2013, since SOX became effective in 2002 (Humphrey et al., 2011) . The reason for this time slot is discussed in the literature review, after SOX there were some regulation changes regarding going concern disclosures. By using only data in the after- SOX period, results are not affected by the regulation change.
This research is conducted with the use of abnormal returns of U.S. firms in the period 2006 till end 2013. Data is abstracted from Compustat, Audit Analytics, CRSP, Eventus and the Thomson Reuters database. Information about the abnormal returns comes from the Eventus database1. Information about the going concern opinions and other financial statement items is extracted from Audit Analytics. Information about the stock price, ebit and z-score can be found in Compustat. Finally Thomson Reuters provides data about institutional ownership.
Based on the research of Menon and Williams (2010), firms that already have received a going concern before are excluded from the dataset. According to Menon and Williams (2010) the surprise effect can be measured in a better way when using first time going concern disclosures. Blay and Geiger (2001) limited their sample to only first-time going-concern report recipients because they argue that going concern firms are more likely to exhibit share price adjustments to the new information form the auditor. So in this sample only first time going concern opinions are included. A firm can receive an unmodified opinion the year after a going concern opinion was issued. However if a firm obtains in the subsequent years another going concern opinion, it is excluded from this research.
To test the hypotheses a sample of 176.727 firm-year observations from Audit Analytics are used. These are observations from both firms with and without going concern
1 In order to retrieve information from the Eventus database, a special company code is needed named permno. This code is not provided by the data retrieved from Audit Analytics and Compustat. Therefore I used the adjusted cusipnumbers from Compustat to obtain the permno in CRSP database. After that the company codes could be uploaded in Eventus.
17
disclosures and for multiple fiscal years, 2006 till 2013. Of this number 1.560 duplicates are dropped. These duplicates consisted of the same cik number and year. Reason that these duplicates exists is that Audit Analytics contains data on multiple auditors who are auditing one firm. 144.221 observations without a going concern disclosure are deleted. This research only includes first time going concern disclosures and the total amount of firms receiving a going concern for the first time is 1.316 in the sample. Due to the merge of the data from Audit Analytics and the data retrieved from Compustat only 902 firms are matched. The merge is based on the CIK code of Compustat which should equal the company f_key of audit analytics, however some of the codes did not match and were deleted.
Not all firms from this combined dataset could be merged with the information about institutional holdings as defined in the Thomson Reuters database. Since the data retrieved from Thomson Reuters is not specified per firm but per manager code, which states for the institutions, the total amount of observations becomes 2.355 after merging. In total 855 firms are merged . In order to merge the data from this combined dataset with the abnormal returns retrieved from Eventus a permnumber is needed from the CRSP database. This number can be retrieved with the Cusip number in the combined database. CRSP demands for an 8 digit Cusip numer, in Stata the 9 digit Cusips are adjusted. However some of the Cusipnumbers in the combined database are less than 8 digits and are not recognized in the CRSP database. Also Eventus did not recognize all of the inputted permnumbers and therefore a lot of observations are lost. The total amount of observations after merging with the data from Eventus is 1.859 consisting of 411 companies both with and without institutional holdings. There are more observations per company since the institutional holdings are separate observations. Lastly companies without information about total shares outstanding are dropped resulting in a total of 346 companies. Table 1 provides an explanation of the sample selection procedure and gives a distribution of the sample by fiscal year.
The total amount of the sample consists of 346 companies since a lot of observations had to be dropped. Of this total amount, 40 companies included information about institutional holdings. This sample would seem small, however prior literature has also a small amount of observations (Chen and Church,1996; Jones, 1996; Taffler et al., 2004; Blay and Geiger, 2001).
18 Table 1 Sample selection and distribution by fiscal year
A: Sample selection procedure
Original Sample Audit Analytics 176.727 observations
Duplicates deleted -1.560 observations
First time Going Concern firms in Audit Analytics 1.316 companies
Merge Audit Analytics and Compustat 902 companies
Merge with Thomson Reuters 855 companies
Merge with Eventus 411 companies
Final sample 346 companies
B: Sample distribution by fiscal year
fyear N 2006 30 2007 54 2008 100 2009 57 2010 30 2011 27 2012 38 2013 10 Total 346
Table 1B shows the sample distribution by fiscal year. For this research data from the United States is used. 2008 has more observations that the other fiscal years since it was the time of the financial crisis and more going concern disclosures were issued.
3.2 Empirical model
In this research a regression test is performed in order to discover other variables which can influence the reaction of institutional owners prior to the going concern. These variables are based on the research of Chen and Church (1996), Blay and Geiger (2001), Ali et al.(2008), Berger et al (1996), Altman (1968), Utama and Cready (1997) and Menon and Williams (2010).
19
The final model looks as follows:
CAR = β0 + β1firmsize + β2ebit + β3Δincome + β4cfops + β5z-score + β6exitvalue
+β7lateFiler + β8big4 + β9prc + β10instown + β11change + β12crisis +
β13GCxinstS+ β14GCxinstM + β15GCxinstL + ԑ.
Where
Dependent variable
CAR = Cumulative abnormal return
Independent & control variables
firmsize = Firm size, measured as the natural log of total assets ebit = Earnings before interest and tax
Δincome = Change in income from precious year cfops = Cash flow of operations per share
z-score = Z-score based on the research of Altman (1968) exitvalue = Exit value of assets
latefiler = Latefiler of company disclosure big4 = 1 if audited by big 4, 0 otherwise
prc = Share price
instown = % of institutional holding from total shares outstanding change = Net change in shares of institutions since prior report crisis = 1 if year= 2007/2008/2009, 0 otherwise
GCxInstS = Institutional holdings< 1%
GCxInstM = Institutional holdings between 1% and 5% GCxInstL = Institutional holdings >5%
20 3.2.1 Dependent variable
The research of Jones (1996) makes use of the stock price as measurement for the market reaction since it captures the reaction of investors (Jones, 1996). He argues that the stock price is an estimation of future probabilities. This paper is an event study since it looks at the influence of an event to the stock prices. Based on the research of Menon and Williams (2010) the dependent variable is CAR. CAR measures the cumulative abnormal returns surrounding the disclosure of a going concern. The cumulative abnormal returns are the actual returns minus the expected returns (Menon and Williams, 2010). Data about the CAR is retrieved from the Eventus database in WRDS. Independent variables are used in order to measure conditions that can explain the market’s reaction to CAR disclosures. In this research, six time windows are used.
CAR1 measures the abnormal returns in the month till 5 days prior disclosure. For this CAR a time period from -30 till -5 days are used in order to see if there is anticipating behaviour. The second CAR measures the abnormal returns in the 10 days prior disclosure. CAR3, CAR4 and CAR 5 measures the market reaction in the days surrounding disclosure and CAR 6 measures a period from 0 till +30 days.
Variable Event window
CAR1 -30,-5 CAR2 -10,0 CAR3 -3,+3 CAR4 -1,+1 CAR5 0, +3 CAR6 0, +30
3.2.2 Independent and control variables
In order to provide explanation for the dependent variable, independent and control variables are used. These variables are explained in this section and are based on prior literature.
The log of the total assets of a firm is a proxy for firmsize. Blay and Geiger (2001) claim that information releases of smaller companies contain more information than for larger companies with more media attention. According to Menon and Williams (2010) larger firms have richer information environments, and less surprise is expected with their going concern disclosures. Therefore firmsize controls for effect which are not fully captured by the adjusted returns.
21
Ebit is the earnings before interest and tax, scaled by total assets, based on the research of Menon and Williams (2010). Information about the ebit is retrieved from the Compustat database.
Cfops are the cash flows of operations per share measured in the year of going concern disclosure. This variable is calculated by dividing the total shares on the Cash flow of operations.
Δincome is used to measure the change in income for the previous year and the year of the going concern disclosure scaled by total assets. This variable is also based in the research of Menon and Williams (2010).
Z-score is a measure for calculating bankruptcy and financial distress based on the research of Altman (1968). The Altman z-score is constructed by using balance sheet and income statement variables available in Audit-Analytics and Compustat database.
Z-score = A x 3.3 + B x 0.99 + C x 0.6 + D x 1.2 + E x 1.4 Where,
A=EBIT/Total Assets;
B=Net Sales /Total Assets;
C=Market Value of Equity / Total Liabilities
D=Working Capital/Total Assets
E=Retained Earnings /Total Assets Common interpretation of Z Score:
> 3.0 - safe based on these financial figures only.
2.7 to 2.99 - On Alert.
1.8 to 2.7 - Good chances of going bankrupt within 2 years.
< 1.80 - Probability of Financial distress is very high
Exitvalue of assets is calculated using the formula of Berger et al (1996). According to the expectation of Menon and Williams (2010), investors will place more weight on the exit value in highly distressed companies by revising the bankruptcy probability. Exitvalue is scaled by total assets.
22
Exitvalue = 1,0 x cash + 1,0 x marketable securities + 0,72 x receivables + 0,55 x inventory + 0,54 x fixed assets - 1,0 x Payables – 1,0 x total debt
Latefiler is a control variable and measures the timeliness of disclosures since late filing companies are expected to be poor performers (Citron et al., 2008).
The variable big 4 is included since investors believer that big 4 auditors have a higher credibility than smaller auditors (Menon and Williams, 2010).
Prc measures the differences in stock price prior to the going concern disclosure (Chen and Church, 1996).
Instown is measured as a percentage of institutional ownership in firms. So the variable is the proportion of a stock held by institutional investors during the disclosure of a going concern. This calculation is based on the research of Utama and Cready (1997) who calculated institutional ownership as the percentage of a firm’s outstanding common shares held by institutional investors that is reported in the Thomson Reuters database. The Thomson Reuters database, formerly known as CDA/Specturm s34, contains information about institutional holdings with $100 million or more assets.
Change is the net changes in shares of institutions since prior report. Expectation is that these changes will occur before the disclosure of a going concern instead of the time after since institutional investors will anticipate on the event.
Since this research is conducted with data from 2006 till 2013 the years of the financial crisis are included. Therefore an extra variable is made to measure the impact of the financial crisis on the abnormal returns. This dummy variable is calculated by giving the years 2007, 2008, 2009 an one and the rest of the years a zero.
The last three variables are based on the research of Ali et al. (2008) . They divided the group of institutional owners in small, medium and large holdings. Ali et al. (2008) measured institutional investors’ reactions on trading announcements in general, however this research looks at the reaction before and after disclosure of a going concern. The variable GCxinstL= is the percentage of outstanding shares that are held by institutions with high stakes in a company. High stakes are measured as the individual institution with more than 5% . Every institution with more than 5% stake in a company will be part of the GCxinstL. GCxinstM is the percentage of shares held by institutional investors with a stake between 1% and 5% in a firm. GCxinstS is the percentage of shares held by institutional investors with a stake less than 1% but more than 0%. So per company a subdivision is made between small, medium and large institutional holdings.
23 4. Results
In this section the results of the performed regressions are discussed. First descriptive statistics are discussed. Before the regression a correlation test is conducted.
4.1 Descriptive statistics
4.1.1 Descriptive statistics total sample
Table 2 provides descriptive statistics about the total sample used in this research. These variables are further discussed in section 3.2. The variables latefiler, big4 and crisis are dummy variables ranked with a zero or one. Table 2 shows that 76% of the sample of going concern firms filed too late. This observation is supported by the research of Citron et al. (2008) who concluded that managers delay disclosure to the last possible date when a greater degree of financial distress is applicable. 58% of the firms are audited by a big 4 firm and 61% of the first going concerns disclosed happened during the financial crisis.
The firms in this sample are in heavy financial problems. The average z-score is -6,80 which shows a very high probability of distress. According to Altman (1968) a Z-score of 1.8 to 2.7 already indicates a good chances of going bankrupt within 2 years. Firms with a score lower than 1,8 have a very high probability of financial distress. In this sample the score is even a negative amount. Prior literature also had a sample with a very negative z-score (Kauser et al. ,2009; Taffler et al., 2004) and very distressed firms. The change in income is negative which means that the income decreased since the year prior to the first time going concern of the company. The average earnings before interest and tax is also negative for the sample which is consistent with the sample of Menon and Williams (2010). However in this sample the exitvalue is slightly positive where in the research of Menon and Williams (2010) the exitvalue is negative.
The variable change shows a positive mean. So the in shares of institutions since prior report has increased. Possible explanation is that the institutions did not want to get rid of their stock since prior report. The amount of institutional ownership is 12% which is low in comparison with the sample of Ali et al. (2008). Grinstein and Michaely (2005) claimed that in the United States institutions held more than 50% of US industrial firms’ equity. This sample does not represent that. However this sample also includes companies without institutional holdings so therefore the low percentage can be explained. Kauser et al. (2006) showed that in a year prior to a going concern report, institutional holdings already sell their stock and that therefore the percentage of institutional holdings is lower in the days prior to disclosure.
24 Table 2 Descriptive statistics total sample
Full sample (n=346)
Variable Mean Standard deviation Min Max
firmsize 4.248 1.84 1.522 10.716 ebit -.501 .585 -2.073 .162 d_Income -33.175 103.257 -378.507 419.537 cfops -.278 1.058 -2.837 3.008 zscore -6.806 8.714 -29.524 2.323 exitvalue .242 .565 -2.817 1.786 latefiler .757 - 0 1 big4 .580 - 0 1 prc 1.744 2.472 -1.19 20.78 instown .124 .708 0 9.069 change 4622.457 84076.07 -563758 1250000 crisis .609 - 0 1 GCxinsS .393 1.434 0 12.143 GCxinsM .857 2.921 0 24.136 GCxinsL .957 4.343 0 40.468 Variable definitions:
firmsize = Firm size, measures as the natural log of total assets
ebit = Earnings before interest and tax
Δincome = Change in income from precious year
cfops = Cash flow of operations per share
z-score = Z-score based on the research of Altman (1968) exitvalue = Exit value of assets
latefiler = Latefiler of company disclosure
big4 = 1 if audited by big 4, 0 otherwise
prc = Share price
instown = % of institutional holding from total shares outstanding change = Net change in shares of institutions since prior report
crisis = 1 if year= 2007/2008/2009, 0 otherwise
GCxInstS = Institutional holdings< 1%
GCxInstM = Institutional holdings between 1% and 5% GCxInstL = Institutional holdings >5%
25 4.1.2 Descriptive statistics divided sample
This subsection of the results chapter discusses the difference between companies with institutional ownership and companies without institutional holdings. Companies with institutional holdings have a value of one and companies without institutional ownership are valued as zero.
A t-test is conducted about the different variables for firms with and firms without institutional holdings to measure the difference. Table 3 provides information about the comparison between institutional investors and non-institutional investors.
Table 3 Comparison institutional investors and non-institutional investors
Institutional holdings n=40 non-institutional holdings n=306
Variables Event window Mean mean Pr([T]>[t])
CAR1 -30,-5 0.216 -0.023 0.0059*** CAR2 -10,0 0.061 -0.024 0.0639* CAR3 -3,+3 -0.056 -0.026 0.5332 CAR4 -1,+1 -0.053 -0.179 0.4028 CAR5 0,+3 -0.051 -0.036 0.6566 CAR6 0,+30 0.077 -0.005 0.4225 firmsize 4.601 4.202 0.1995 ebit -0.413 -0.531 0.3129 d_Income -36.695 -32.695 0.8114 cfops -0.186 -0.290 0.5623 z-score -5.416 -6.988 0.2842 exitvalue 0.261 0.240 0.8226 latefiler 0.7 0.764 0.3709 big4 0.6 0.578 0.7956 prc 1.958 1.716 0.5616 change 39984.25 0 0.0045*** crisis 0.775 0.588 0.0227** Ter/der** 0.475 0.313 0.0418**
*,**, and,***indicates statistical significance at the 10%. 5%. And 1% levels, respectively Ter/der measures whether the company is terminated or deregistered.
26
In the month prior to the disclosure of a going concern, companies with institutional holdings still have positive abnormal returns. However firms without institutional ownership already show negative abnormal returns. This significant result may suggest that institutional owners do not anticipate in the month prior to the disclosure of a going concern and other investors do. However it can also be the case that institutions anticipate earlier than a month (Kauser et al., 2006) or that institutions are dedicated investors (Ali et al., 2008). This result is in contrast with the research of Kauser et al. (2006) who claimed that institutional owners react on a timely basis and will reduce their holdings. Hypothesis 2, Institutional investors will show more anticipating behavior prior to the going concern disclosure than non-institutional investors, is therefore not supported. The abnormal returns for firms with institutions are negative in the days surrounding the disclosure but not prior.
The abnormal returns for companies with institutional holdings are not significantly different than those of companies without institutional holding surrounding the days of disclosure of a going concern opinion. Hypothesis 1, Institutional investors will react less negative on a going concern disclosure than non-institutional investors, cannot be supported or rejected and no conclusions can be drawn from this observation.
In the month after the disclosure of the going concern, CAR6, there is not a significant difference between companies with and without institutional ownership. Therefore hypothesis 3, When the degree of institutional ownership is higher, the information after a going concern will be processed faster than when a company has more individual investors, cannot be supported or rejected.
For most of the independent and control variables, there is not a significant difference between the sample with only institutional holdings and the sample without institutional holdings. The degree of companies suffering from the crisis is significant higher in the sample of companies with institutional ownership. This can be seen in table 3 where there is a significant difference between the crisis for firms with and without institutional holdings. That there is a significant difference in the change of shares of institutions is expected since there is no change in shares of institutions in companies without institutional owners. Also significantly more companies are terminated or deregistered if they had an amount of institutional holdings.
In summary as described in table 3 there is no evidence found that companies with institutional holdings have less negative abnormal returns in the days after going concern disclosure. Anticipating behavior is not found in the results , it even seems that individual investors anticipate more on a going concern than institutional investors. However it can also
27
be the case that institutional investors are already anticipating earlier than 30 days prior disclosure of a going concern opinion (Kauser et al., 2006). Also no conclusions can be drawn about the difference in processing of information between institutional investors and non-institutional investors since results are not significant.
4.2 Pearson correlation test
A Pearson’s correlation test is run to assess the relationship between firmsize, ebit, change in income, cfops, z-score, exitvalue, latefiler, big4, prc, change, instown, crisis,GCxinsS, GCxinsM and GCxinsL.
Table 4 shows the Pearson correlation. The amounts with a star behind are statistically significant at the 0,05 level. The level of statistical significance is shown below the correlation number. In this model there are multiple variables used. Therefore there does exist correlation between some of the variables as described in table 4. Due to a high correlation, some of the variables below are discussed and tested further. Cohen (1988) is often cited for determining the acceptable correlation amount. He made several correlation boundaries. If an absolute number is between 0,1 and 3, there exist only a small correlation. Between 0,3 and 0,5 is a medium correlation is detected and a correlation above the 0,5 is seen as strong. Based on this research variables with a correlation higher than 0,5 are discussed below.
Firmsize shows a high correlation with ebit, and z-score. Also earnings before interest and tax shows a high positive significant correlation with z-score. Therefore after every regression a vif command is used to test for multi-collinearity. Vif stands for variance inflation factor. If a vif is higher than 10, more investigation is needed. This was not the case in the regressions run for this research2.
The correlations between the subtypes of institutional ownership are positive and significant. Also these correlations are above the 0,5 boundary. However Ali et al. (2008) also had high correlated subtypes of institutional ownership. Also the correlation between Institutional ownership and the subtypes of institutional owners are positive like the research of Ali et al. (2008). Main difference is that Ali et al. (2008) did not have a correlation higher than 0,5 between InstL and the other 2 subtypes
2 After every regression the vif command is used in order to assess the multicollinearity. In none of the cases the vif was higher than 10 and therefore no further investigation was needed.
28 Table 4 Pearson correlation table
CAR=β0 +β1firmsize+ β2ebit+ β3Δincome+ β4cfops+ β5z-score+ β6exitvalue+ β7lateFiler+ β8big4+ β9prc+ β10instown+ β11change+ β12crisis+ β13GCxinstS+ β14GCxinstM+ β15GCxinstL+ԑ.
Correlation Firmsize EBIT change
Income CFOPS Z-score
Exit
Value Latefiler Big 4 Chprice InstOwn Change Crisis S M L
Firmsize 1 EBIT 0.6492* 1 0.0000 d_Income -0.8459* -0.4316* 1 0.0000 0.0000 CFOPS 0.7610* 0.5240* -0.8156* 1 0.0000 0.0000 0.0000 Z-score 0.6061* 0.8716* -0.4396* 0.5114* 1 0.0000 0.0000 0.0000 0.0000 ExitValue -0.2020* -0.1664* 0.1050* -0.1617* -0.0720* 1 0.0000 0.0000 0.0001 0.0000 0.0061 latefiler -0.3734* -0.0184 0.4078* -0.3401* 0.0037 -0.0408 1 0.0000 0.4739 0.0000 0.0000 0.8854 0.1175 Big 4 0.2869* 0.0423 -0.3429* 0.2867* -0.0346 -0.2034* -0.4361* 1 0.0000 0.0989 0.0000 0.0000 0.1810 0.0000 0.0000 Chprice 0.7277* 0.4070* -0.6474* 0.6001* 0.3685* -0.0797* -0.5789* 0.3200* 1 0.0000 0.0000 0.0000 0.0000 0.0000 0.0023 0.0000 0.0000 InstOwn -0.0822* -0.0373 0.0960* -0.0807* -0.0112 0.0267 0.0396 -0.0230 -0.0713* 1 0.0013 0.1459 0.0002 0.0016 0.6646 0.3061 0.1227 0.3697 0.0057 Change 0.0022 0.0098 -0.0067 0.0076 0.0073 -0.0002 -0.0273 0.0031 0.0320 0.0346 1 0.9327 0.7014 0.7948 0.7660 0.7766 0.9930 0.2872 0.9035 0.2155 0.1774 Crisis 0.2166* -0.0036 -0.1987* 0.1479* 0.0117 -0.0836* -0.1408* 0.1104* 0.0018 0.0499 -0.0269 1 0.0000 0.8882 0.0000 0.0000 0.6521 0.0013 0.0000 0.0000 0.9430 0.0515 0.2942
29 4.3 Regression results
This subsection provides the results of the research conducted. These results are compared with results of prior literature and discussed.
Table 5 gives an overview of the abnormal returns averages in the different time windows.
Table 5 Descriptive abnormal returns
Variable Event window
Mean Std. Dev. Min Max
CARwindow1 -30,-5 .0038 .520 -1.535 5.872 CARwindow2 -10,0 -.0146 .271 -1.154 1.470 CARwindow3 -3,+3 -.0302 .279 -.751 2.838 CARwindow4 -1,+1 -.0220 .251 -.715 3.395 CARwindow5 0, +3 -.0378 .208 -.753 1.338 CARwindow6 0, +30 .0044 .604 -1.561 7.522
This table shows that there isn’t evidence of anticipating behavior of the market in the longer term since the mean is still positive in the month before the disclosure. This is consistent with the research of Kauser et al. (2009) who also measured a market under reaction to first- time going concern opinion firms in the United States. However the days before and after the disclosure of a going concern the market reacts negatively as can be seen in the negative mean of CAR2, CAR3, CAR4 and CAR5. Jones(1996) also measured negative abnormal returns in the days surrounding the disclosure of a going concern opinion.
Table 6 shows the results of the regression based on the different event windows. This regression only looks at the institutional holdings as a whole in the different event windows. Only the significant results are discussed and compared with prior literature. In the month prior the disclosure of a going concern opinion only the share price shows a significant positive effect on the abnormal returns. The change in shares of institutions shows a negative significant effect in the month after the disclosure of a going concern opinion meaning that the abnormal returns are less positive if institutions are selling their shares after the disclosure. Exit value has a significant negative effect in the same time window as change. This result is inconsistent with the research of Menon and Williams (2010).
30 Table 6 OLS Regression on the market reaction to going concern disclosures, instown
CAR = β0 + β1firmsize + β2ebit + β3Δincome + β4cfops + β5z-score + β6exitvalue
+β7lateFiler + β8big4 + β9 prc + β10instown + β11change+ β12crisis + ԑ.
Variables CAR1 CAR2 CAR3 CAR4 CAR5 CAR6
Event window
-30,-5 -10,0 -3,+3 -1,+1 0, +3 0, +30
Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
firmsize -.012 -.009 .002 .002 .009 .015 ebit .011 -.002 -.004 -.016 -.004 -.059 d_Income .000 .000 -.000 .000 .000 -.000 cfops .009 -.005 .004 -.000 .013* (0.093) -.000 z-score -.002 .001 .000 .000 -.000 .003 exitvalue .015 -.009 -.011 -.003 -.000 -.064* (0.096) latefiler .003 -.014 -.051** (0.039) -.041** (0.012) -.030 -.074 big4 .002 -.008 -.002 -.009 -.001 -.022 prc .012* (0.079) .008** (0.041) .0008 -.001 -.001 -.003 instown .024 .032** (0.032) .028* (0.051) -.018* (0.064) .020* (0.081) .060* (0.058) change .000 -1.60e-07 -6.08e-08 -3.30e-09 -2.92e-08 -5.48e-07
**(0.035) crisis -.023 .004 -.014 -.021 -.031** (0.049) .0157 intercept .000 .014 -.018 .006 -.052 -.041 R2 0.0421 0.0321 0.0251 0.0199 0.0365 0.0711 F-value 0.69 1.09 0.9 1.21 1.4 1.5
31
In the 10 days prior to the going concern opinion, and in the time windows surrounding and after disclosure, the variable institutional holdings show a positive significant effect except for time window 3. So the abnormal returns as described in table 5 are less negative or more positive if the degree of institutional holdings is higher. This results suggest that institutional investors are not anticipating for a going concern opinion and does not support hypothesis 2. Possible explanation can be that institutional investors do not have more information prior to disclosure. However Kauser et al. (2006) gave as explanation that institutional investors are already reducing their holdings in the 12 months prior a going concern disclosure. This time span is not included in this research. Table 6 also shows that institutional holdings have a significant positive effect on the cumulative abnormal returns in the days surrounding the disclosure meaning that the abnormal returns are less negative with a higher degree of institutional ownership. Therefore hypothesis H1 can be supported since institutional investors react less negative after disclosure of a going concern opinion. This result is consistent with the research of Taffler et al. (2004) showing that institutional investors do not reduce their holdings after a going concern opinion. However Menon and Williams (2010) measured a significant negative coefficient for institutional holdings. They conducted their research with a time span of day 0 till +2 which can explain the difference for some amount.
Latefiler has a significant negative effect on the abnormal returns in the days surrounding disclosure of a going concern opinion. So if a company files too late, the abnormal returns will be even more negative. This result is consistent with the research of Jones (1996). Exitvalue, crisis and change in shares of institutional holdings show only a significant effect in the time window after the disclosure of a going concern opinion. Cfops show a small significant positive effect, meaning that the abnormal returns are less negative if the cfops are higher. This is expected because with a higher cash flow of operations per share the abnormal returns will become less negative. The control variable crisis shows a negative significant impact. This result suggest that during the crisis the abnormal returns were even more negative after a going concern disclosure.
The share price has a positive significant effect on the abnormal returns. This results shows that a higher share price will lower the negative abnormal returns or have an increasing effect on positive abnormal returns prior to the disclosure of a going concern opinion.
