Disclosure of additional information in audit reports and it’s relation with investor decisions

(1)

Master thesis investor ownership Pagina 1

Disclosure of additional information in audit reports

and it’s relation with investor decisions

Amsterdam Business School

Faculty of Economics and Business

Student: Anies Arends

Student number: 10890173

Thesis supervisor: Dr. A. Sikalidis

Education program: Msc. Accountancy & Control

(2)

Master thesis investor ownership Pagina 2 Statement of Originality

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

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

(3)

Abstract

In this study, a database research is conducted to examine if there is a significant effect between the overall materiality threshold and the decision making of investors in their ownership of shares. I focus on two investor groups: institutional investors (long-term) and transient investors (short-term). My expectations are that the overall materiality threshold has a significant positive relationship with the institutional investor decisions and that the overall materiality threshold has a stronger effect on institutional investor decisions compared with transient investor decisions. I expect this, because in general institutional investors are more skeptic and have stronger analytical abilities compared to the transient investors. In my research I found that the overall materiality is not positively related with institutional investor decisions. These findings might indicate that an auditor requires more payment and has to put more effort in an audit if the expectations of material misstatements are high, which in turn leads to less confidence for the investors. However, my expectations were correct that the effect in this relationship is stronger for the institutional investor decisions.

Key words: Overall materiality disclosure, Materiality, Institutional investor decisions, transient investor

(4)

1 Introduction

The auditor has to apply materiality for the planning and performance of the audit, this means that possible uncorrected misstatements need to be indentified and consider the possibility that these misstatements will occur. Additionally, the auditor has to apply materiality for his own opinion in the auditor‘s report. The process in determining the materiality consists of three phases: First, the materiality level for the financial statement as a whole needs to be established, this is called the overall or planning materiality. Second, the auditor has to establish a lower materiality level compared to the materiality for the whole financial statement that needs to be used as a basis in audit tests, this is referred as tolerable misstatement or performance materiality. Finally, an evaluation takes place of the audit results (Messier, Glover & Prawitt 2014, pp.84–89).

The Financial Reporting Council (FRC, 2013) is responsible for UK interests on international level and has required a disclosure of the overall materiality in an auditor‘s report since 2013. This requirement is applied on all companies in the UK with a premium listing of equity shares. However, there is no understanding to the extent of how this additional disclosure of information will benefit the users of the financial statement (Eilifsen & Messier, 2015, p.20). This might be interesting to apply within the US, because this country has one of the largest economies in the word with a large quantity of potential investors.

This paper is focused on the first phase of the materiality process, because there are not that many studies on materiality. The aim of this paper is to find out if additional disclosure in the auditor‘s report (overall materiality) could have an effect on the decision of the users, specifically the investors. This is not studied yet. Therefore it could be interesting to test if the additional disclosure of overall materiality in audit reports has an effect on the decision making of the investors and see if this possible effect differs between short and long-term investors. During this study the notion ‗overall materiality‘ will be used for the materiality being established for the whole financial statement.

(6)

2 Background

2.1 Materiality determination

There is much prior research conducted about the notion of materiality in an audit context. Many decades ago the whole idea of materiality was a big black box, since materiality is determined by what is called professional judgment. These studies have determined how various factors can affect the professional judgment (Moriarity & Barron, 1979, p.114). An auditor that detects certain breaches in a financial statement will inform the manager about this issue. Both agents have to compromise if the manager will correct these issues (Keune & Johnstone, 2012, p.1641).

There are three studies that analyzed some possible important financial factors that might have an impact on materiality decisions by auditors. All three studies found that net income is the most important factor for user decisions and used as the basis to judge whether an item is material or not (Boatsman & Robertson, 1974), (Hofstedt & Hughes, 1977) and (Sweeny, 1979). Additionally, a research conducted by Tuttle, Coller & Plumlee (2002, p.24) has analyzed the relationship between the level of misstatements and the mean market price and found that the mean market price is positively associated with the level of misstatements. This basically means that misstatements in practice can have external consequences. Friedberg et al. (1989) document that income is most frequently used in audit manuals. They distinguish income in ―income from continuing operations‖ and ―income before taxes‖, the authors also mentioned that the total assets are frequently used in practice to provide a basis for materiality.

Considering these important factors it is vital to know when an item is considered as a misstatement issue. However, there are no absolute values for overall materiality. Most audit firms can use benchmarks to decide the range of their overall materiality. For example there are some firms that use a 5 to 6 percent threshold on the total amount of income before taxes, but other firms will use a range of 5 to 10 percent (Eilifsen & Messier, 2015, p.12). The overall materiality is then used as a basis in determining the tolerable misstatement (performance materiality). The percentage used on the total amount of the overall materiality is frequently around 50 to 75 percent in practice. Some even use 90 percent (Eilifsen & Messier, 2015, p.13). It is perfectly clear to understand why the degree of materiality has no

(7)

Master thesis investor ownership Pagina 7 absolute value considering the above and that is why it must be decided by professional judgment.

2.2 materiality threshold

For most auditors the main question is what the materiality threshold should be. The materiality benchmarks only give an idea in what range the percentage should be based on a certain item. Some firms use a certain range of 5-10 percent on the total amount of income before taxes. Should the auditor use 5 percent (the lower fence of the benchmark) or 10 percent (the higher fence of the benchmark)? A recent study conducted by Budescu, Peecher and Solomon (2012) should answer this question. They study the relationship between the extent and nature of audit evidence, materiality thresholds and misstatement type on achieved audit risk. They tested the assumption when an auditor uses a high materiality tresshold, the audit risk will be lower, so the auditor does not need to gather as much evidence compared to a low materiality threshold.

In their results the authors found that for all cases when biased evidence exceeds the materiality threshold, the audit risk will increase. This results in reducing the degree to which noise in the evidence distorts the auditor‘s normative expectations. Thus, standards that imply that the increasing extent of evidence decreases audit risk only holds when the bias of evidence is lower than the materiality threshold. To summarize, low materiality thresholds by the auditor represents more required audit evidence and reduces audit risk when the bias of evidence is lower than the threshold. Vice versa, a high materiality threshold indicates less required evidence and acceptance of a higher audit risk. Thus, it is expected that auditors use a low overall materiality threshold when they are unfamiliar with a firm or when they expect a high possibility of error or/and fraud which lead to possible material misstatements.

2.3 The materiality reporting process

By providing an understanding on how materiality is determined and what factors are considered it might be interesting to know the process after a detection of misstatement. A study by Keune & Johnstone (2012 p.1644) analyzes how materiality judgments of detected misstatements are influenced by incentives of auditors and managers and audit committee characteristics. They found that managers are more likely to neglect a qualitative material misstatement when analysts are on their heels when the audit fees are low. In other words,

(8)

Master thesis investor ownership Pagina 8 pressure from analysts will create incentives for managers to use aggressive reporting (Keune & Johnstone, 2012, p.1670). They also found that when audit fees are high, the auditor is less likely to neglect a qualitative and quantitative material statement (Keune & Johnstone, 2012, p.1673).So incentives and audit committee characteristics influence the auditor‘s decision for the needed correction of misstatements.

2.4. The audit report

There used to be a lack of external audit information, the external stakeholders did not receive sufficient information about the conducted internal audits. Previous authors have therefore considered a need of internal audit reports (IAR) (Ege 2015; Archambeault et al., 2008). The authors found that providing an IAR to the external stakeholders might lead to better forms of assurance. The disclosure provided in the audit report could contain information about responsibilities, activities and composition of the internal audit function.

There was a prior study by Holt & DeZoort (2009) about the user side of an IAR setting and they evaluate to what extent a descriptive IAR affects investor decision making. Their findings indicate that investors provided with an IAR have a much greater confidence in the reliability of the financial reporting process in an organization. Additionally, the authors document that the effect of the IAR on investor confidence in financial reporting is strongest for companies were the risks of possible frauds is high. Another finding indicates that an IAR has an equal level of usefulness compared to other disclosure materials like the audit committee report or management‘s report on internal controls. This study will contribute to the prior study conducted by Holt & DeZoort (2009) by adding the requirement of the FRC to disclose the overall materiality and assess whether this additional disclosure in the audit report has an influence on the investor decision making.

2.5 The research question

In summary there are audit firms that all use different overall materiality levels when they face a similar situation which basically means that determination of materiality is based on professional judgment by considering factors like net income, total assets, equity etc. The decision from an auditor to decide if misstatement needs correction can be affected by incentives for both the auditor and the manager and the characteristics of the audit committee characteristics. This paper will focus on what happens after the process in determining the

(9)

Master thesis investor ownership Pagina 9 overall materiality, questioning if disclosure of the overall materiality in the audit report for US listed companies has an influence on investor decision making. This study should provide an understanding if the investors can actually use this additional disclosed information.

To narrow this research down, this study will only focus on the investors. For example, ―Does the investor gain more confidence with this additional disclosure‖? And is any of this information actually useful to influence investor decision making? We will come to the following research question for this study: Does the disclosure of the overall materiality

in the audit reports have a significant influence on investor decision making?

3 Hypothesis development

To answer the research question, this paper contains two hypotheses that will be tested with a statistical regression model. The first hypothesis is focused on finding a possible influence of overall materiality disclosure on the decision making of institutional investors. The second hypothesis aims to find if this possible influence on the decision making between long-term investors and short-term investors varies.

3.1 Confidence

A research conducted by Mock et al (2013, pp.323-351) resulted in the fact that investors have a desire for more information in the audit report. They used surveys to gather this information. This research could relate to the decision of the FRC about the requirement of additional information in the audit report (overall materiality disclosure), because in the same year this decision came to the surface.

Considering these findings, there is a study by Christensen, Glover & Wolfe (2014, p.86) that a critical audit matter (CAM) can have consequences for a firm. They document that a CAM about some uncertain fair value estimates will most likely influence investors decisions and results in halting investments. The authors compare the CAM with management‘s fair value footnote disclosures to find out if the CAM disclosures in the audit report have more influence on the investment decisions of investors than these footnote disclosures. The authors found positive evidence on their expectations and concluded that CAM disclosures are easier to analyze and draw more attention than footnote disclosures.

(10)

Master thesis investor ownership Pagina 10 Previous research by Goh, Krishnan & Li (2013, p.970) studies the relationship between internal control and going concern audit opinions. The authors state that according to the Sarbanes-Oxley Act of 2002 (SOX), independent auditors are required to provide an opinion on their clients‘ internal control over the financial statements. The Authors of this paper explore the association of two different audit opinions: the going concern audit opinion (GCO and the adverse internal control material weakness opinion (MWO). The GCO reflect the auditors‘ view that the client is able to continue her operations as a going concern for a period of 12 months beyond the year of the financial statement. The MWO reflects the auditors view on some possible material weaknesses in the internal controls and thus the likelihood of material misstatements not being prevented or detected increases.

Jiang & Son (2015, p.318) conducted a research on whether audit fees reflect risk premiums in the presence of control risk after controlling for audit effort through audit delay. The results indicate that the auditors are willing to adjust their risk premiums in the audit fees and their audit effort as well in a response to an altered control risk. The authors show in their analyses that the extent of risk premium adjustments varies depending on how serious the underlying internal control problems are. The results of the study of Jiang & Son suggest that the audit fee is positively associated with audit effort. Additionally, the audit fee is positively associated with the control risk.

It is reasonable to assume that a low threshold of overall materiality disclosed in the audit report is positively related to a non-going concern opinion by the auditor, because with a lower materiality threshold the possibility to detect material misstatements is higher. Hence, chances are higher for items in the financial statement with a material amount that exceeds the low materiality threshold. Moreover, it is expected that the audit fee is positively related with a low threshold of overall materiality, because the auditor needs more evidence when materiality threshold is low and this in turn also increases the audit effort. Finally, one might expect that investors have more faith in disclosed information, when the auditor has put a lot of effort in the audit

It is likely that the additional information (e.g. the audit opinion, responsibilities, fees etc.) provided by the auditor is useful to an investor for a basis of his/her decision making. It is to be expected that the disclosure of the overall materiality in the audit report will lead to more investments if the information of the firm is faithful and accurate and a lot of audit effort is put in the audit. This leads to the following hypothesis:

(11)

Master thesis investor ownership Pagina 11 H1: It is more likely that institutional investors will invest more in a firm when the reported

overall materiality in the audit report reflects high audit effort.

Support for this hypothesis is only expected when there is a low amount of material misstatements present, it is expected that high audit effort has a higher chance in finding material misstatement, because the materiality level is stricter compared with low audit effort. In cases of high material misstatements, investments will drop and then it is expected that H1 will be rejected.

3.2 Investor behavior

Prior research by Mehta & Sharma (2015, p.26) has explained how investor behavior emerges in India. However, these results are also compatible on a broad international scale. They draw their results on the economic theory of investment behavior. The decision that an individual (investor) makes is termed ‗macroeconomic aggregate function‘. The various macroeconomic aspects of the decision making are drawn from the utility theory. The most popular proverbs of the utility theory are: a) Investors are risk averters and rational, b) Investors want to maximize their wealth and are able to deal with complex choices. Portfolios that maximize the expected returns are the main priority of choice for an investor by lowering their risk and this is called ‗the expected utility of investment‘.

Some prior studies have distinguished investor groups to found a variation in their tested relationship. For example, Mehta & Sharma (2015, p.34) found that middle age group investors are less risk averse than young investors. Furthermore the authors concluded that a majority of investors based on age and income hold their position in equity stocks until the stock price reaches a certain level. When the stock price reaches a desired level, the investors want to change their position to a term position. The majority of investors prefer short-term position and want to sell their stocks as soon as possible.

Prior research by Koh (2003, p.105) tested the relationship between institutional ownership and Australian firms‘ aggressive earnings management strategies. The author found that a small ownership is positively related with the firms‘ aggressive earnings management, while a large ownership is negatively associated with aggressive earnings management. This finding suggests that the level of ownership influences a firms‘ earnings management. The investors with a small ownership are consistent with short-term orientated investors and investors with a large ownership are consistent with long-term orientated

(12)

Master thesis investor ownership Pagina 12 investors. Koh (2003, p.107) documented that the short-term investors are often referred to as myopic or transient investors, these investors often focus on the current earnings instead of long-term earnings incorporated in stock prices. Additionally, Koh (2003, p.109) documented that the long-term investors are referred to as the institutional investors, these investors have the ability to monitor firms and have the intention holding their shares over a long time, without a specific focus on earnings only. Koh (2003, p.124) argues that this information clearly suggests that transient investors are easily attracted by aggressive earnings management, while the institutional investors are not tricked so easy and are thus more skeptic about the firm.

It is clear that the possible risks and practices of a firm (e.g. earnings management) are very important for investor decisions. It is reasonable to assume that the outcome of H1 is more significant for the institutional investors than the transient investors. This expectation is based on the fact that institutional investors have a broader focus on the firm than transient investor who are mainly focused on current earnings. The overall materiality threshold is consistent with the auditors‘ knowledge and trust in a firm. Hence, a low materiality threshold in general is related with higher expectations of error, higher possibilities of fraud and a lower knowledge of the firm by the auditor.

H2: The disclosure of overall materiality reflecting high audit effort has a stronger effect on institutional investor investments compared with the transient investor investments.

4. Data sources

This research will be conducted by using publically available data from Compustat, CRSP, AuditAnalytics and Orbis databases. To answer the research question this study will provide panel data about investment decisions from investors in the years that overall materiality disclosure in not required and the years that overall materiality disclosure is required. The range of the panel data is 5 years before the required disclosure in the audit report (2008-2012) and from the moment of required disclosure until the end of 2014 (2013-2014). The software of SPSS will be used to generate and maintain the collected data. This study uses data from the U.S. because of the large availability. Depending on the available results, the definite data range in the sample will be decided later in this paper. Data on consolidated companies within the NYSE, NYSE ARCA and NYSE MKT stock markets are gathered and used to test both hypotheses.

(13)

Master thesis investor ownership Pagina 13 4.1 Regression models

In prior literature standard regression models are used to measure a significant relationship between the independent variable and the dependent variable. Other variables which can influence the dependent variable are also included in a regression model. In this study there are two regression models: The first model is based on institutional investor decisions and the second model is based on transient investor decisions.

First model

INVDECI= bₒ + b1DISCOMAR + b2ANION + b3STOCKPR + b4AUQUA + b5DIVY +

b6PROFIT + b7 FVAL + b8TURN + Ԑ1

Second model

INVDECT= bₒ + b1DISCOMAR + b2ANION + b3STOCKPR + b4AUQUA + b5DIVY +

4.2 Variables

In this study the disclosure of the overall materiality in the audit report will be used as the independent variable and the investor decisions are used as the dependent variable. The decision making of the investor (INVDECI) is mainly focused on the percentage of shares held by institutional investors (banks, hedge funds and venture capitalists). This variable is computed as the number of shares held divided by the total number of equity shares. The results from the first model will be tested for a deviation between institutional investors and transient investors (industrial companies, individuals or families, unnamed private shareholders and other unnamed shareholders). Thus the dependent variable for the second model (INVDECT) is used. This variable is computed exactly like the INVDECI variable. The disclosed overall materiality in the audit report (DISCOMAR) is used as the independent variable in this study and is defined as the total amount of annual audit fees paid by parental companies scaled by total assets. The goal of this study is to find out if investors will continue their investments based on the overall materiality disclosure in the audit report. There are a great number of other factors that influence the investor decisions, these factors will be chosen as the control variables.

(14)

Master thesis investor ownership Pagina 14 4.3 Control factors

There are numerous control factors that can influence investor decision making. A prior study by Goh, Krishnan & Li (2013) tested the relationship between internal controls and the auditors‘ opinion. This research indicates that the audit opinion is based on how strong the internal controls are within the organization. A non-going concern opinion will be given when there are weaknesses in the internal controls discovered by the auditor, these weaknesses can lead to possible material misstatements. The auditor opinion (ANION) will be used as the first control variable, because an investor will not invest in a company when there are a lot of material misstatements.

Another possibility to influence investor decisions is documented by Mehta & Sharma (2015, p.34). The authors conclude that stock price level has a significant effect on the investor decision. In their study they conclude when stock price reaches a certain level, the investors would like to change their position to a short term basis if they are holding a long term position. Additionally, they conclude that young investors are more risk averse than middle age investors, so a rise in stock price has a stronger effect on the decision making on young investors compared to middle age investors. This suggests that investor decisions can be influenced by stock price level, especially for the young investors. For the second control variable in this study the stock price level (STOCKPR) will be used in the equation.

The third factor that can influence investor decisions is the audit quality of the audit in a firm. Koh (2003, p.114) documents that a study by Becker et al. (1998) indicates, compared with big 6 auditors, a stronger positive relationship between non big 6 (Big 4 in the present) auditors and discretionary accruals. The author used a dummy variable to control for the effect of auditor quality on discretionary accruals based on these findings by Becker et al. In this study the same control variable is used representing audit quality (AUQUA). It is expected that investors have more confidence to invest in a firm when the audit is conducted by a big 4 firm. In this study data is used from consolidated publicly listed companies and it is expected that most of these companies are audited by big 4 firms.

In a recent study by Hawas & Tse (2016, p.100) the relationship between the corporate governance structure and the decisions of major investors was tested from 2005 to 2009. The authors developed a corporate governance index based on requirements of the combined code within the UK. Their study indicates that there is a significant positive relationship between corporate governance and the total major shareholdings. Additionally, they provide evidence that this relationship was insignificant before the financial crisis and significant after the

(15)

Master thesis investor ownership Pagina 15 financial crisis. To control for potential omitted-variable bias, they included some variables from recent studies. Some of these control variables will be used in this paper to control for a potential omitted-variable bias. These variables are: dividend yield, profitability, firm value and turnover. The authors have also included stock price as a control variable, this variable is also included in this paper. The explanation for including this variable in this paper is already justified according recent studies by Mehta & Sharma (2015) and Holt & DeZoort (2009). For the investor decisions they use the total major shareholdings as the dependent variable (TOTAL_MAJ). This dependent variable is categorized for different industries. In this paper the dependent variable (INVDEC) represents the total percentage of ownership in a firm by institutional investors

Dividend yield is included as a control variable because institutional investors have a preference to invest in stock with a low-dividend yield, individual investors prefer to invest in stock with a high-dividend yield. Hence, dividend yield does have an influence on investor decisions. Hawas & Tse (2016, p.109) included profitability and firm value in their study because investors prefer to invest in companies with a high return on assets and a high Tobin‘s Q. The authors have included turnover to control for stock liquidity preferences, because recent studies have pointed out that fund managers tilt their holdings more towards liquid stocks. In this paper the control variables: dividend yield (DIVY), profitability (PROFIT), firm value (FVAL) and turnover (TURN) are included, because the main focus lies in whether the investor decides to invest or not to invest in a company according to the disclosed overall materiality in the audit report. However, investor decisions can also consist of certain stock preferences. Moreover, it is clear that investors will not invest in stocks with a low profitability. Additionally, other factors like the value of the firm and turnover can have an influence on investor decision making. Therefore, these control variables are necessary to control for a biased result.

First, to measure for dividend yield Hawas & Tse scaled dividends per share by the market price at the end times 100. Second, the authors used the ROA to measure for profitability and measured this by scaling net income by the total assets. Finally, the authors measured Tobin‘s Q is measured by the market value of equity plus total debts scaled by total assets and turnover is measured by scaling the shares traded over the year by the number of outstanding share.

(16)

Master thesis investor ownership Pagina 16 It is expected that stock price has a significant influence on investor decisions, because the amount of their returns is based on the stock price. A higher stock price leads to better returns. As for the other control variables there is no expectation about the significant level of influence on the investor decisions. The statistical regression model in this paper measures if there is a significant relation between the disclosed overall materiality in the audit report and the total length of investments by institutional investors, this aims to answer the first hypothesis (H1).

The second regression model is aimed to test the second hypothesis (H2). It is expected that the DISCOMAR variable has a stronger significant relationship with institutional investor decisions compared with the transient investor decisions. The second model includes the same control variables than the first model. The only difference lies in the dependent variable, in this case the focus lies on the short-term investors (INVDECT).

4.3 Statistical model

The hypotheses in this paper will be tested with the help of a statistical regression model. This model is chosen because in my opinion it is the best model to test if there is a significant relationship between the disclosed overall materiality in the audit report and the decision of the investor. Moreover, it is possible to include many control variables to control for biased test results. With all the necessary variables to get an answer for this research, the proxy measures need to be decided. All data for the proxy measures in this paper is collected from public available databases.

Goh, Krishnan & Li (2013) examined the relationship between Australian firms‘ internal controls and the auditor opinion. In their study they use as a proxy for the auditor opinion by using the going concern opinion and an adverse opinion to examine this relationship. As a proxy for the ANION variable in this paper, the going concern opinion is used as a proxy measure to control for a biased result between the disclosed overall materiality and investor decision making. This is necessary because it is most likely that investors will not invest in firms with a non-going concern opinion.

Hawas & Tse (2016, p.109) used as a proxy for dividend yield the dividends per share scaled by the market price at the end of the year times 100. In this paper the same measurement will be used, because this percentage reflects how much dividend is earned per share at the end of the year as a percentage of the market price. The higher this percentage,

(17)

Master thesis investor ownership Pagina 17 the more advantageous this is for the investors. Hawas & Tse (2016, p.109) used the return on assets (ROA) and Tobin‘s Q as a proxy to measure profitability and firm value, respectively. The same measurements are applied in this paper, because the ROA represents how well the firm operates, this is calculated by net income as a percentage of the total assets. Tobin‘s Q is measured by the market value of equity plus total debts scaled by total assets. Tobin‘s Q is a ratio that represents the value of the firm as a percentage to the total assets, this is an appropriate measure to capture firm value and this measure is also used in this paper to proxy for firm value.

The results of the relationship between the disclosed overall materiality and investor decision making in this study will be tested for a deviation between transient and institutional investors consistent with the study of Koh (2013). The author used as a proxy for institutional ownership the total number of shares held by institutional investors scaled by the total number of outstanding shares. The same method will be used to measure the INVDECI variable. To measure the dependent variable (INVDECT) in this paper the same kind of proxy is used consistent with how institutional ownership is measured by the author. INVDECI is proxied by calculating the total number of shares held by institutional investors divided by the total number of shares outstanding. These annual percentages of shares will be compared with each other to see if the number of investments over the years have increased or decreased. The author has obtained the financial information from the Compustat GLOBAL Vantage database supplemented by the Connect 4 database. In this paper the data will be gathered from the Orbis database.

Hawas & Tse (2016, p.109) proxied for turnover by dividing the number of shares traded over the year by the number of outstanding shares. This is a ratio that is commonly applied to calculate turnover in practice. The turnover ratio scaled by total assets is used in this paper to proxy for the turnover variable. A recent study by Mehta & Sharma (2015) provided evidence that the level of stock price has an influence on investor decisions. Moreover, the results of these decisions deviate between young investors and middle age investors. The majority of the investors are the middle age investors that age around 36-45 and 46-55. The minority investors are of young age, these young investors age below 25 and between 26-35. Hawas & Tse (2016, p.109) included the stock price in their study to control for a biased result. They proxied for this variable using the annual stock price. The annual book value per share is used in this paper to proxy for stock price, because investors only sell their shares when stock price reaches a certain level.

(18)

Master thesis investor ownership Pagina 18 Koh (2013) measures audit quality by making a dummy variable and as a proxy is a distinction made between big 6 and non big 6 auditors. In this study the same measurement will be used to proxy for audit quality. If the company is audited by a big 4 firm the dummy is equal to 1. Should the company be audited by a non big 4 firm the dummy is equal to 0.

As a proxy for the independent variable, the audit fees paid by the parental companies scaled by the total assets are used. The audit fees are assumed to be negatively associated with the overall materiality threshold. It is reasonable to assume that a low threshold of overall materiality is generally used when the auditor expects a high amount of errors and/or frauds and is uncertain regarding the risks within a company. This requires more effort in the audit by the auditor and thus the auditor is willing to adjust the risk premium which adds up in the audit fee. Only total audit fees paid by parental firms above € 200.000 are considered to be high.

Two models are used in order to answer the research question and test the underlying hypotheses in this paper. The first hypothesis aims to find a positive significant relationship between audit effort and the decision making of institutional investors. When audit effort is high, the auditor is able to charge more audit fees. A long term investor is expected to have more confidence when the auditor puts high effort in the audit of a firm. Ownership data from banks, hedge funds and venture capitalists are used as a proxy for INVDECI. The first model to test H1 is the following:

INVDECI= bₒ + b1DISCOMAR + b2ANION + b3STOCKPR + b4AUQUA + b5DIVY +

The second hypothesis is aimed to find a comparison between institutional investor decision making and transient investor decision making. It is expected that high audit effort has more significant influence on institutional investor decision making compared with transient investor decision making, because it is expected that institutional investors have a broader view on a firm while the transient investors are mainly focused on current earnings. If the outcome of this model is lower than the second model, H2 will be supported. If the outcome is higher, h2 should be rejected. Ownership data from industrial companies, individuals or families, unnamed private shareholders and other unnamed shareholders are used as a proxy for INVDECT. The following model is used to test H2:

(19)

Master thesis investor ownership Pagina 19 INVDECT= bₒ + b1DISCOMAR + b2ANION + b3STOCKPR + b4AUQUA + b5DIVY +

4.4 Data management

A total of 38796 records were found in the CRSP/Compustat merged database. Only the records with a data in the years 2012, 2013 and 2014 are used in the model, because data before 2012 is irrelevant for the research question and the hypotheses. Another reason to exclude data before the year 2012 is because it is difficult to get all the data of one single company within this data range. Thus after eliminating this data, 16308 records remained. Companies that have zero dividend per share in a year, will automatically have a dividend yield of zero. All records with an auditor opinion code representing other than a going concern opinion are eliminated, a total of 10413 records remained.

Finally all of the records from companies were it was not able to gather all the necessary data from the other databases and did not contain a data range between 2012 and 2014 were all eliminated. From these records, a complete database set is created of 165 records. Due to name matching difficulties for one company, three records were eliminated. Due to insufficient data on three companies nine records were eliminated. A total of 153 records remain. For each year a sample of 51 companies is used in both models.

In the AuditAnalytics database a total of 84981 records are found. A large amount of records did not contain the name of the parent company, these records were filtered out and as a residual 1565 records are remaining. With this dataset the audit fees are added in the complete database set. In the Orbis database there are 2884 companies found with stock and ownership data, for both models. With this dataset the information about stocks are added in the complete database set.

5 Results

5.1 Analyses of the first regression model

The following outputs from SPSS are aimed to answer the first hypothesis. The first table gives an overview of all the variables used in the regression model as well as the total number of observations (N = 153). The second table indicates how much of the outcome variable can be explained by the regression model. The third table indicates which variable correlates best

(20)

Master thesis investor ownership Pagina 20 with the outcome variable and to test if there are any multicollinearity issues. The results show that the DISCOMAR variable correlates best with the outcome variable, there are also no multicollinearity issues regarding this regression model. The fourth table gives an indication how strong this model is in predicting the outcome. Results show that this model is reliable (F > 1) (P < .001). The fifth table represents the coefficient and sig. values for each variable. Results indicate that the independent variable (audit fees) has no significant relationship with the dependent variable (% ownership institutional investors), thus the first hypothesis is rejected based on these results. In fact, Turnover has the best explanatory power on the outcome of this study.

Table 1: Statistics

INVDECI DISCOMAR ANION STOCKPR AUQUA DIVY PROFIT FVAL TURN

N 153 153 153 153 153 153 153 153 153 Mean 25,74 133,65 1,00 31,58 0,92 3,46% 2,74% 0,51 145,96 Median 25,84 88,53 1,00 26,76 1,00 3,12% 2,62% 0,45 42,87 Std. Deviation 8,42 204,43 0,00 31,61 0,27 3,56% 3,31% 0,31 261,03 Minimum ,10 ,74 1 -,69 0 0,00% -7,17% ,03 ,68 Maximum 49,30 1364,59 1 199,87 1 32,81% 18,18% 2,01 1583,26

Table 1 shows the mean, standard deviation, median, minimum and maximum of all the variables used in the first model that aims to test my first hypothesis. The median value represents the exact middle of off all the total summed values from one variable, while the mean represents the average value. Standard deviation (SD) represents the variation of these values from the mean. A low standard deviation indicates that the data values are close to the mean value. The minimum values represent the lowest values of a variable, while the maximum values represent the highest values of a variable. For the first hypothesis the relationship between institutional investor decisions and disclosure of the overall materiality will be tested. The percentage of ownership by institutional investors is the proxy for the decision making, while the audit fees scaled by the total assets is the proxy for overall materiality disclosure. Recently in this paper it is explained that audit fees are related with audit effort and the more effort is done by the auditor, the more confidence is expected from the institutional investor. We can reasonably assume that high audit effort leads to more audit tests and this will influence the overall materiality that is used by the auditor. The mean value of the dependent variable (x ; 25,74) indicates in this sample that on average institutional

(21)

Master thesis investor ownership Pagina 21 investors (banks, hedge funds and venture capitalists) own almost 26% of the shares in a firm, which is more than ¼ of all total shares. The audit fee values represent millions in value scaled by the total assets, these fees are the total fees as a percentage of the total assets paid by a parental company within a whole year. The mean high, because the sample contains only large consolidated companies. Hence, it can be expected that audit expenses are high. The auditor opinion- internal control has a mean value of 1,00. The value 1 indicates a going concern opinion, other values indicate other opinions. Only companies with a going concern opinion are used in this study. It is no surprise that the standard deviation is nil. Turnover represents a mean of almost 146, which is also high. This is because of all the companies in this sample almost all the shares were traded annually and barely had any outstanding shares. The sample consists of 153 observations (N: 153).

Table 2: Model Summaryb

Model R R Square Adjusted R Square Std. Error of the Estimate 1 ,548a ,300 ,266 7,21400

a. Predictors: (Constant), TURN, POFIT, STOCKPR, AUQUA, DISCOMAR, DIVY, FVAL

b. Dependent Variable: INVDECI

Table 2 represents the summary of the first model that is used to test the first hypothesis. The R symbol indicates the value of the multiple correlation coefficient between the predictors and the outcome. The R Square indicates that 30% of the variation of the dependent variable is explained by the model. Hence, 30% of the outcome is influenced by the variables used, while the rest is not explained.

(22)

Master thesis investor ownership Pagina 22 The values in table 3 represent the Pearson‘s correlation coefficients between every pair of variables. This table is used to assess if there is multicollinearity in the data. According to prior literature there is multicollinearity if the Pearson correlation is higher than 0.90 ( r > .9). According to these results DISCOMAR correlates best with the outcome (r = -.413) . Between the rest of the variables there is no indication of any multicollinearity issues. Hence, there are no predictors that correlate too high with each other. It is likely that DISCOMAR is the variable in this regression model that will best predict the variance in the percentage of ownership in shares held by institutional investors. Table 3: Correlations

INVDECI DISCOMAR ANION STOCKPR AUQUA DIVY PROFIT FVAL TURN Pearson Correlation INVDECI 1,000 -,413 -,029 ,190 -,225 -,032 -,249 ,037 DISCOMAR -,413 1,000 -,215 -,156 ,023 -,106 ,475 ,381 ANION 1,000 STOCKPR -,029 -,215 1,000 ,124 -,180 ,093 -,310 -,184 AUQUA ,190 -,156 ,124 1,000 -,459 -,048 -,195 -,302 DIVY -,225 ,023 -,180 -,459 1,000 ,248 ,097 -,064 PROFIT -,032 -,106 ,093 -,048 ,248 1,000 ,379 ,052 FVAL -,249 ,475 -,310 -,195 ,097 ,379 1,000 ,402 TURN ,037 ,381 -,184 -,302 -,064 ,052 ,402 1,000

Sig. (1-tailed) INVDECI ,000 ,000 ,362 ,009 ,003 ,347 ,001 ,327

DISCOMAR ,000 ,000 ,004 ,027 ,389 ,095 ,000 ,000 ANION ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,000 STOCKPR ,362 ,004 ,000 ,064 ,013 ,127 ,000 ,012 AUQUA ,009 ,027 ,000 ,064 ,000 ,278 ,008 ,000 DIVY ,003 ,389 ,000 ,013 ,000 ,001 ,115 ,215 PROFIT ,347 ,095 ,000 ,127 ,278 ,001 ,000 ,260 FVAL ,001 ,000 ,000 ,000 ,008 ,115 ,000 ,000 TURN ,327 ,000 ,000 ,012 ,000 ,215 ,260 ,000

(23)

Master thesis investor ownership Pagina 23 Table 4: ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 3235,034 7 462,148 8,880 ,000b Residual 7546,055 145 52,042 Total 10781,089 152

a. Dependent Variable: INVDECI

b. Predictors: (Constant), TURN, PROFIT, STOCKPR, AUQUA, DISCOMAR, DIVY, FVAL

Table 4 is an ANOVA that explains that the regression model that is used to test the first hypothesis is significantly better at predicting the outcome rather than using the mean as the best guess. The F value is greater than 8 and should be greater than 1 according to prior literature. Hence, the regression model is highly significant (F = 8,880, p < .001). The degrees of freedom (df) are equal to the number of variables used in the regression model.

Table 5: Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig. 95,0% Confidence Interval for B B Std. Error Beta Lower Bound Upper Bound 1 (Constant) 28,786 3,427 8,399 ,000** 22,012 35,559 DISCOMAR -,018 ,004 -,444 -5,127 ,000** -,025 -,011 STOCKPR -,047 ,021 -,178 -2,294 ,023** -,088 -,007 AUQUA 3,484 2,664 ,112 1,308 ,193 -1,781 8,749 DIVY -,404 ,205 -,170 -1,971 ,051 -,808 ,001 PROFIT ,095 ,221 ,037 ,432 ,666 -,341 ,532 FVAL -4,708 2,659 -,175 -1,771 ,079 -9,964 ,547 TURN ,009 ,003 ,264 3,166 ,002** ,003 ,014

a. Dependent Variable: INVDECI

*P =10% **P = 5% (one tailed values of current sig values)

Table 5 represents the coefficient values between the outcome variable (INVDECI) and a predictor variable. According to the results in this table, the audit fees are significantly related with the decision making of institutional investors (P .000), because P < .001. The results

(24)

Master thesis investor ownership Pagina 24 show a strong negative relationship between the institutional investor decision and the audit fees. The negative standardized coefficient (β = -.444) indicates that if the total amount of audit fees paid increases with one standard deviation (204,43), the percentage of ownership by institutional investors decreases by -.444 standard deviations. Hence the percentage of ownership will decrease by 3,738 (-.444*8,42). This is a large insignificant decrease and indicates that audit fees paid are actually negatively associated with institutional investor decisions. Hence, investments will go down if there is more audit effort put in the firm. Based on these results the first hypothesis is rejected.

The annual book value per share (STOCKPR) has a significant p-value (P = .023). The results indicate that the annual book value per share is negatively associated with the institutional investor decisions which were expected. A rise in the stock price leads to a lower percentage of ownership, because the investors will sell their shares when the stock price rises. The standardized coefficient (β = -.178) indicates that if the annual book value per share increases with one standard deviation, the percentage of ownership by the institutional investor decreases by 1,498 (-.178*8,42). The dummy control variable (AUQUA) shows an insignificant relationship with the institutional investor decisions (P = .193). This result indicates that the investor pays no mind if the firm is audited by a big4 firm or a non big4 firm. However, the relationship is positive, which means that when a firm is audited by a big firm there is a slight insignificant increase (β = .112) of 0,943 in the total percentage of ownership by the institutional investors

DIVY, FVAL and PROFIT are insignificantly related with institutional investor decisions. These variables have a p-value higher than 5% (P > .05). Dividend yield (P = .051) is negatively associated with the outcome variable, which means that a higher dividend yield will lead to less ownership of shares (β = -.170) by institutional investors. Should DIVY increase with 1 standard deviation (SD = 3,56 %) than the total percentage of ownership held by institutional investors will decrease by 1,431%. FVAL (P < .079) is negatively associated with institutional investor decisions. A higher value of the firm will lead to a lower percentage of ownership (β = -.175) by institutional investors. When FVAL increases with 1 standard deviation (SD = 0,31) the percentage of ownership by held by institutional investors decreases by 1,473%. PROFIT (P = .666) is positively associated with institutional investor decisions. The results show a positive relationship (β = .037) and indicates that in any case PROFIT increases with 1 standard deviation (SD = 3,31%), the ownership from institutional investors increases by 0,311%.

(25)

Master thesis investor ownership Pagina 25 Finally, TURN (P = .002, P < .05) shows a positive significant relationship with investor decisions (β = .264), this could be expected because when a firm trades a high amount of shares this indicates that the firm has many investors. This gives a potential investor more confidence to invest in that firm, because institutional investors will not invest in a firm if they don‘t have enough confidence in a firm. If TURN increases with 1 standard deviation (SD = 261,03) the total percentage of shares held by the institutional investors increases by 2,222%.

5.2 Analyses of the second regression model

The following outputs from SPSS are aimed to answer the second hypothesis. The sixth table gives an overview of all the variables used in the regression model as well as the total number of observations (N = 153). The seventh table indicates how much of the outcome variable can be explained by the regression model. The eighth table indicates which variable correlates best with the outcome variable and to test if there are any multicollinearity issues. The results show that the AUQUA variable correlates best with the outcome variable, there are also no multicollinearity issues regarding this regression model. The ninth table gives an indication how strong this model is in predicting the outcome. Results show that this model is reliable (F > 1) (P < .001). The tenth table represents the coefficient and sig. values for each variable. Results indicate that the independent variable (DISCOMAR) has a significant relationship with the dependent variable (INVDECT). Results indicate a higher decreasing effect on institutional investor decisions compared with the lower increasing effect on the transient investors decision from the first model. Thus the second hypothesis is supported based on these results.

Table 6: Statistics

INVDECT DISCOMAR ANION STOCKPR AUQUA DIVY PROFIT FVAL TURN

N 153 153 153 153 153 153 153 153 153 Mean 5,97 133,65 1,00 31,58 0,92 3,46% 2,74% 0,51 145,96 Median 2,80 88,53 1,00 26,76 1,00 3,12% 2,62% 0,45 42,87 Std. Deviation 12,15 204,43 0,00 31,61 0,27 3,56% 3,31% 0,31 261,03 Minimum 0,00 ,74 1 -,69 0 0,00% -7,17% ,03 ,68 Maximum 89,37 1364,59 1 199,87 1 32,81% 18,18% 2,01 1583,26

(26)

Master thesis investor ownership Pagina 26 Table 1 shows the mean, standard deviation, median, minimum and maximum of all the variables used in the second model that aims to test my second hypothesis. The median value represents the exact middle of off all the total summed values from one variable, while the mean represents the average value. Standard deviation (SD) represents the variation of these values from the mean. A low standard deviation indicates that the data values are close to the mean value. For the second hypothesis the relationship between transient investor decisions and disclosure of the overall materiality will be tested. The percentage of ownership by transient investors is the proxy for the decision making, like the first regression model the audit fees paid by parental companies scaled by total assets is used as the proxy for overall materiality disclosure. Recently in this paper it is explained that it is expected that the outcome of the second regression model is less affected compared to the outcome in the first regression model with the same independent variable and the same control variables. In other words, there should be less variance on the percentage of ownership held by transient investors, because these investors are more focused on the short-term and are thus less skeptic than institutional investors. The mean value of the dependent variable ( x ; 5,97) indicates in this sample that on average transient investors (industrial companies, unnamed individuals, unnamed shareholders and unnamed private shareholders) own almost 6% of the shares in a firm, which is less than 1/10 of all total shares.

The exact same mean, median, minimum, maximum and SD results are given for the control variables and the independent variable as in the first regression model, because the same data sample is used and only the dependent variable has changed. The audit fee values represent millions in value, these fees are the total fees paid by a parental company within a whole year scaled by total assets. The mean value is high, because the sample contains only large consolidated companies. Hence, it can be expected that audit expenses are high. The auditor opinion- internal control has a mean value of 1,00. The value 1 indicates a going concern opinion, other values indicate other opinions. Only companies with a going concern opinion are used in this study. It is no surprise that the standard deviation is nil. Turnover represents a mean of almost 146 million. This is because of all the companies in this sample almost all the shares were traded annually and barely had any outstanding shares. Consistent with the first regression model, The same sample of 153 observations (N: 153) is used.

(27)

Master thesis investor ownership Pagina 27 Table 7: Model Summaryb

Model R R Square Adjusted R Square Std. Error of the Estimate 1 ,729a ,531 ,508 8,52021

a. Predictors: (Constant), TURN, PROFIT, STOCKPR, AUQUA, DISCOMAR, DIVY, FVAL

b. Dependent Variable: INVDECT

Table 7 represents the summary of the second model that is used to test the second hypothesis. The R symbol indicates the value of the multiple correlation coefficient between the predictors and the outcome. The R Square indicates that more than 50% of the variance of the dependent variable is explained by the model. Hence, more than 50% of the outcome is influenced by the variables used, while less than half of the outcome cannot be explained by the model.

(28)

Master thesis investor ownership Pagina 28 The values in table 8 represent the Pearson‘s correlation coefficients between every pair of variables. This table is used to assess if there is multicollinearity in the data. According to prior literature there is multicollinearity if the Pearson correlation is higher than 0.90 ( r > .9). According to these results AUQUA correlates best with the outcome (r = -.618). Between the rest of the variables there is no indication of any multicollinearity issues. Hence, there are no predictors that correlate too high with each other. It is likely that AUQUA is the variable in this regression model that will best predict the variance in the percentage of ownership in shares held by the transient investors. Table 8: Correlations

INVDECT DISCOMAR ANION STOCKPR AUQUA DIVY PROFIT FVAL TURN

Pearson Correlation INVDECT 1,000 ,314 -,094 -,618 ,443 ,228 ,291 ,382 DISCOMAR ,314 1,000 -,215 -,156 ,023 -,106 ,475 ,381 ANION 1,000 STOCKPR -,094 -,215 1,000 ,124 -,180 ,093 -,310 -,184 AUQUA -,618 -,156 ,124 1,000 -,459 -,048 -,195 -,302 DIVY ,443 ,023 -,180 -,459 1,000 ,248 ,097 -,064 PROFIT ,228 -,106 ,093 -,048 ,248 1,000 ,379 ,052 FVAL ,291 ,475 -,310 -,195 ,097 ,379 1,000 ,402 TURN ,382 ,381 -,184 -,302 -,064 ,052 ,402 1,000 Sig. (1-tailed) INVDECT ,000 ,000 ,124 ,000 ,000 ,002 ,000 ,000 DISCOMAR ,000 ,000 ,004 ,027 ,389 ,095 ,000 ,000 ANION ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,000 STOCKPR ,124 ,004 ,000 ,064 ,013 ,127 ,000 ,012 AUQUA ,000 ,027 ,000 ,064 ,000 ,278 ,008 ,000 DIVY ,000 ,389 ,000 ,013 ,000 ,001 ,115 ,215 PROFIT ,002 ,095 ,000 ,127 ,278 ,001 ,000 ,260 FVAL ,000 ,000 ,000 ,000 ,008 ,115 ,000 ,000 TURN ,000 ,000 ,000 ,012 ,000 ,215 ,260 ,000

(29)

Master thesis investor ownership Pagina 29 Table 9: ANOVAa Model Sum of Squares df Mean Square F Sig. 2 Regression 11923,058 7 1703,294 23,463 ,000b Residual 10526,125 145 72,594 Total 22449,183 152

a. Dependent Variable: INVDECT

b. Predictors: (Constant), TURN, PROFIT, STOCKPR, AUQUA, DISCOMAR, DIVY, FVAL

Table 9 is an ANOVA that explains that the regression model that is used to test the second hypothesis is significantly better at predicting the outcome rather than using the mean as the best guess. In fact the second regression model is even stronger than the first regression model The F value is almost 3 times as high compared with the first model and prior literature states that F should be greater than 1. Hence, the regression model is highly significant (F = 23,463, p < .001). The degrees of freedom (df) are equal to the number of variables used in the regression model. Table 10: Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig. 95,0% Confidence Interval for B B Std. Error Beta Lower Bound Upper Bound 2 (Constant) 16,472 4,048 4,070 ,000** 8,472 24,473 DISCOMAR ,013 ,004 ,219 3,089 ,002** ,005 ,021 STOCKPR ,020 ,024 ,051 ,806 ,422 -,029 ,068 AUQUA -19,192 3,146 -,426 -6,100 ,000** -25,410 -12,973 DIVY ,770 ,242 ,225 3,185 ,002** ,292 1,249 PROFIT ,658 ,261 ,179 2,522 ,013** ,142 1,174 FVAL -2,044 3,140 -,053 -,651 ,516 -8,250 4,163 TURN ,010 ,003 ,206 3,014 ,003** ,003 ,016

a. Dependent Variable: % ownership transient investors *P =10% **P = 5% (one tailed values of current sig values)

(30)

Master thesis investor ownership Pagina 30 Table 10 represents the coefficient values between the outcome variable (% ownership transient investors) and a predictor variable. The second hypothesis is aimed to find a weaker relationship compared to the outcome for the first hypothesis. According to the results in Table 10, the audit fees paid are significantly related with the decision making of transient investors (P = .002, first model P < .001), because P < .05. Compared with the institutional investors, the results show a positive relationship between the investor decision and audit fees paid. The positive standardized coefficient (β = .219, first model β = -.444) indicates that if the total amount of audit fees paid increases with one standard deviation (204,43), the percentage of ownership by transient investors increases by .219 standard deviations. Hence the percentage of ownership will increase by 2,660 (.219*12,15). This is a strong significant increase, this indicates that DISCOMAR is positively associated with transient investor decisions and compared with the first regression model, DISCOMAR is negatively associated with institutional investor decisions. In the first regression model, the stock possession of the institutional investors decreases by 3,738 when the audit fees paid rise with 1 SD. Based on these results audit fees have a stronger effect on institutional investor ownership compared with transient investor ownership (first model P < .001, second model P = .002). There is support for the second hypothesis.

The annual book value per share (STOCKPR) has an insignificant p-value (P = .422, first model P = .023). The results indicate that the annual book value per share is positively associated with the transient investor decisions which were not expected. A rise in the stock price should lead to a lower percentage of ownership, a possible explanation is that transient investors invest less money in shares compared with institutional investors. Because institutional investors risk more money, these investors are more tempted to sell their shares to gain high profit. Transient investors are probably more willing to wait a bit longer until stock prices reach a higher value in order to gain a decent profit. The standardized coefficient (β = .051 first model β = -.178) indicates that if the annual book value per share increases with one standard deviation, the percentage of ownership by the transient investor increases by 0,61 (.051*12,15). The dummy control variable (AUQUA) shows a significant relationship with the transient investor decisions (P < .001, first model P = .193). These results indicate that a transient investor gives a high priority if the firm is audited by a big4 firm or a non big4 firm. Compared with the first model the results are now negative, which means that when a firm is audited by a big firm there is a strong significant decrease (β = -.426, first model β = .112) of 5,175 in the total percentage of ownership by the transient

(31)

Master thesis investor ownership Pagina 31 investors. These results were unexpected, a possible reason could be the fact that firms being audited by big4 firms have more complicated systems compared to firms that are audited by non-big4 firms. The transient investors will have more confidence about current earnings being audited by big4 firms. Hence, it might be different for small individual investors to analyze the results and mainly focus on the current earnings to ensure that they get their money back with a decent profit, while the institutional investors are likely to have more experience and have stronger analytical skills.

DIVY, PROFIT and TURN also have a high significant relationship between transient investor decisions. These variables have a p-value lower than 5% (P < .05). DIVY (P = .002, first model P = .051) is positively associated with the outcome variable, which means that a higher dividend yield will lead to a higher ownership of shares (β = .225, first model β = -.170) by transient investors. Should DIVY increase with 1 standard deviation (SD = 3,56%), then the total percentage of ownership held by transient investors will increase by 2,733%. PROFIT is positively associated with transient investor decisions (P = .013, first model P = .666). The results show a positive relationship (β = .179, first model β = .037) and indicate that in any case PROFIT increases with 1 standard deviation (SD = 3,31%), the ownership from institutional investors increase by 2,174%. The transient investors are more focused on profitability than institutional investors, which was explained earlier. Hence, these significant positive results are no surprise. TURN (P = .003, first model P = .002) is positively associated with transient investor decisions (β = .206, first model β = .264), this could be expected because when a firm trades a high amount of shares this indicates that the firm has many investors. This gives a potential investor more confidence to invest in that firm, because transient as well as institutional investors will not invest in a firm if they don‘t have enough confidence in a firm. If TURN increases with 1 standard deviation (SD = 261,03) the total percentage of shares held by the transient investors increases by 2,502%.

Finally, FVAL shows a negative insignificant relationship between firm value and transient investor decisions (P = .516, first model P < .079). A higher value of the firm will lead to a lower percentage of ownership (β = -.053, first model β = -.175) by transient investors. When FVAL increases with 1 standard deviation (SD = 0,31) the percentage of ownership by held by transient investors decreases by 0,643%.

Disclosure of additional information in audit reports and it’s relation with investor decisions