Differences in level of voluntary disclosure provided by family and non-family firms

Transcription

1 Differences in level of voluntary disclosure provided by family and non-family firms Sjef Tijssen S Tilburg University

2 Differences in level of voluntary disclosure provided by family and non-family firms Master Thesis Department Accountancy Faculty of Economics and Business Studies Tilburg University Sjef Tijssen S KPMG Eindhoven Date of Completion: Supervisors: Dr. B.R.C.J. van den Brand Tilburg University Drs. M.J. Hendriks KPMG Eindhoven 2

3 Abstract This study investigates whether there are significant differences between the level of voluntary disclosure provided by family and non-family firms. To be able to conduct this study, historical data from 1930 is used. This overcomes the problem related to the difficulty associated with separating voluntary and mandatory levels of disclosure between the countries included in this sample, being Belgium, the Netherlands and the United States. After eliminating all non-usable data, a sample of 203 companies is obtained, of which 75 are family firms. In order to estimate the level of voluntary disclosure a predetermined voluntary disclosure index and a simple count of the number of pages and items included in the annual reports is used. After comparing the samples of family and non-family firms and conducting the corresponding T-tests the conclusion is reached that family firms on average have lower levels of voluntary disclosure. When controlling for other variables which might be of influence on the level of voluntary disclosure, a positive relation with the involvement of an auditor is found. Furthermore, the type of legal system of a country seems to have a significant effect on the level of voluntary disclosure. The results found in this study contribute to the discussion on whether family firms provide more or less voluntary disclosure than their non-family counterparts. 3

4 Table of Contents Chapter 1 Introduction Research Topic Research Method Main Findings Academic and societal relevance Outline... 8 Chapter 2 Hypothesis Development Family Firms: Economic Relevance Definition: General Findings Agency Theory: General Agency Theory Agency Theory in Family Firms Voluntary Disclosure General Voluntary disclosure in family firms Hypothesis Chapter 3 Historical Relevance and Background Historical Accounting Research Unregulated Setting Historical Background Netherlands Belgium United States Chapter 4 Methodology Sample Selection Year Countries Companies

5 4.2 Variables Dependent Variable Independent Variable Variables with potential influence on level of voluntary disclosure Research design Chapter 5 Findings Findings from Annual reports Belgium The Netherlands The United States Table Summary T-test analysis Findings from descriptive and possible influential variables Belgium The Netherlands The United States General Chapter 6 Results Regression analysis Belgium The Netherlands The United States Institutional Variables General Results Chapter 7 Conclusion Summary Main Conclusions Limitations Future Research References Appendix 1 List of Firms Appendix 2 Voluntary Disclosure Index

6 Chapter 1 Introduction 1.1 Research Topic Met het familiebedrijf de crisis door! This is the title of a recent study which shows that the crisis has had a positive effect on 40% of Dutch family firms (Flören and Jansen, 2011). This study is the largest qualitative study ever conducted in relation to family firms in the Netherlands and therefore it underlines the fact that studies regarding family firms are of interest nowadays. One possible explanation for this phenomenon is that an increasing number of researchers has pointed out the importance of family firms in the world economy. Villalonga and Amit (2006) for instance found that family firms are at least as common among public corporations around the world as are widely held and other non-family firms. Moreover, among privately held firms, family ownership is nearly universal (Burkart et al, 2003). Therefore, it is rather surprising that besides this prevalent notion of the economic importance of family firms, there is relatively little structured information available concerning this group of firms. The studies that do cope with these differences between family and non-family firms show that these are mainly caused by differences in agency related factors (Ali et al., 2007; Chen et al., 2008). A result of the existence of these agency related problems is the demand for quality financial reporting. As a consequence, besides the information that is nowadays required to disclose by law, firms voluntarily disclose more information (Meek et al., 1995). Concerning family firms, there are both studies arguing that these agency problems will lead these firms to voluntarily disclose more than there non-family counterparts (Ali et al., 2007; Wang, 2006) as well as studies arguing the opposite (Chen et al., 2008; Chau and Gray, 2002; van den Brand, 2007). In order to find an answer to this unsolved discussion this study s research question will be: Is there a significant difference in the level of voluntary disclosure provided by family and non-family firms? 1.2 Research Method To be able to provide an answer to this question a thorough literature review will be held. This literature review will give information on why there might exist differences with respect to the level of voluntary disclosure of these two groups. Furthermore, it will enable us to identify variables that might have an influence on the level of voluntary disclosure and which therefore need to be controlled for in the empirical investigation which constitutes the second part of this study. An important difference that has to be made in order to conduct this study is the one between voluntary and mandatory levels of disclosure. Since this study is based in an international setting, including the United States, Belgium and the Netherlands, one problem will be the significant 6

7 differences which exist across countries related to the level of disclosure requirements (Gray et al., 1995). This study will overcome this issue by studying historical data from In that time there were little or no mandatory requirements concerning levels of disclosure in neither of the sample countries. These countries are included for different reasons. Firstly the availability of data is an important factor and it is one of the more important reasons to included the Netherlands in the study because a large sample can be obtained from this country. The United States and Belgium both provide a smaller sample, but as is shown in an article by Rajan and Zingales (2003), these countries were among the most financially developed countries in the world in the beginning of the 20 th century. In addition, these two European countries on the one hand and the United states on the other, both have a different legal system. In their article, Archambault and Archambault (2003) describe the classification of countries in common and code law. These different legal systems are expected to have an effect on the level of disclosure of companies and therefore these countries will provide a good sample on which to conduct this study. Banks and other financial institutions are not included in the sample because of their different reporting which make them hard to compare with other companies (Barton and Waymire, 2004). Furthermore, governmental organizations will not be included because of the possible influence on a company s policies. Additionally, railroads will be excluded because of their highly idiosyncratic reporting practices that makes it difficult to develop a measure of financial reporting quality that is comparable with other industries (Barton and Waymire, 2004). As a result, 203 firms remain, among which are 75 family firms. This sample size is a bit smaller, but still in line with other studies like, Anderson and Reeb (2003A) 403 firms, 142 family, Wang (2006) 542 firms, 207 family, and Gallo et al. (2004) 305 firms, 104 family. The annual reports of the remaining firms in the sample are used to determine the level of voluntary disclosure of the respective firms. This level of voluntary disclosure is measured by the use of a voluntary disclosure index (Barton and Waymire, 2004; van den Brand, 2007) and by a simple count of the number of pages and items included in the annual reports. The empirical investigation will consist of an analysis of the collected data in order to provide an answer to the research question. This will be done by using a comparison of means with corresponding T-test analyses and by means of linear regressions. 1.3 Main Findings Studying the existing literature and combining the different conclusion led to the expectation that there are indeed significant differences in the level of voluntary disclosure provided by family and non-family firms. Furthermore, the literature review has shown evidence supporting certain relations to exist between several variables and the level of voluntary disclosure. By investigating both the level of voluntary disclosure as measured by the disclosure index, as well as estimated by the number of pages and items included in the annual reports it is found that on average family firms tend to provide less 7

8 voluntary disclosure. Moreover, there is evidence found to support the argument that a country s legal system has an influence on the level of voluntary disclosure, for family firms as well as for non-family firms. Furthermore, it is shown, at least for the U.S. and the Netherlands, that auditor involvement is one of the most important variables influencing the level of voluntary disclosure. The reason why this finding is not supported by the Belgian sample is that in 1930 there did not exist any audit control in Belgium. Therefore, audit involvement was not incorporated in analyses of the Belgian sample. 1.4 Academic and societal relevance This study investigates whether there are significant differences in the level of voluntary disclosure provided by family and non-family firms. As explained before, family firms are among the most common types of firms in the world (Villalonga and Amit, 2006; Wang, 2005), nevertheless there is still very little structured information available regarding this type of firm and the effect it has on disclosure practices. Most studies on disclosure practices are namely solely based on factors contributing to explain this phenomenon. Therefore, this study contributes to the existing literature by providing further evidence related to family firm ownership as an independent variable explaining voluntary disclosure. Furthermore, accounting history is very much a nation-centered research area (Richardson and MacDonald, 2002). Therefore, by incorporating multiple countries, especially with different legal systems, this study will be able to contribute to the literature regarding international differences in relation to family firms. 1.5 Outline The outline of this study is as follows. In chapter 2 there will be given a thorough review of the most recent literature regarding voluntary disclosure practices of family firms and the related agency and voluntary disclosure theory. In chapter 3 there will consequently be held a discussion about the relevance of historical accounting followed by a short summary of the most important developments with regard to auditing around the world as well as related to the countries included in this study. Chapter 4 will analyze the existing literature ones again in order to determine which other variables could be of influence on the research model, which will also be described in that chapter. In chapter 5 the means of the family and non-family firms with regard to certain predetermined variables will be compared and tests will be conducted to find out whether there exist certain differences between these two groups. Chapter 6 will subsequently show the linear regressions in order to see which variables are of influence on determining the level of voluntary disclosure. In chapter 7 there will be given a short summary and the main conclusions and limitations of this study will be addressed followed by some recommendations for future research. 8

9 Chapter 2 Hypothesis Development In this chapter existing literature on the subject of this study will be discussed. Some general issues related to family firms are addressed first. Thereafter, agency theory and voluntary disclosure theory will be discussed, leading to a hypothesis. 2.1 Family Firms: Economic Relevance As mentioned in the introduction there is an increasing number of researchers that has underlined the importance of family firms in the world economy. Besides the evidence provided by Villalonga and Amit (2006) it is for instance shown that in the US, family firms are present in one-third of the S&P 500 and own 18% of outstanding equity within this group of firms (Anderson and Reeb, 2003A). Furthermore, La Porta et al. (1999) show that 30% of large companies in the world is family controlled. With respect to smaller companies La Porta et al. (1999) find that even a higher percentage is family controlled. This is in line with findings from Chu (2009), who shows that in Taiwan, family firms represent half of public small and medium sized enterprises and account for more than 11% of outstanding equity (Chu, 2009). When looking at even smaller, private, firms, it is argued that family ownership is nearly universal (Burkart et al, 2003). The fact that, besides this evidence on the importance of family firms in public as well as private markets, there still exists little comparable data concerning these firms is rather surprising. One of the possible explanation could be the large number of different definitions used defining family firms (Miller et al, 2007) Definition: To conduct a study comparing voluntary disclosure practices of family and non-family firms of course it is needed to firstly define family firms. However, as stated before, this has been a topic of debate. In order to investigate these different definitions, Dyer Jr. (2006) compares nine, rather recent studies, which, as he finds, all use different definitions to define a family firm. Moreover, Miller et al. (2007) show that over time at least twenty different studies coping with this subject have used a different definition of family firms. Villalonga and Amit (2006) do for instance use a definition which is similar to the one used by Anderson and Reeb (2003A), nevertheless, they do incorporate an additional criteria which does alter the sample of these respective studies. An even more distinct definition is used in a study by Suehiro and Wailerdsak (2004) who investigate family firms in Thailand. They agree with Dyer Jr. (2006) who states that classifying all family firms in the same category might lead to misleading conclusions. Therefore they use four different categories in which they place family firms (Suehiro and Wailerdsak, 2004). A similar, though again, slightly different categorization is used by Mishra et al. (2001) who uses four alternative variables to measure family control. The above examples clearly show that indeed there are as many definitions used as there are studies conducted. The first issue with these definition differences is the fact that some studies are likely 9

10 to have included firms in their family firm sample which are not included in other studies. As a consequence this might have led to the comparison of apples and oranges and the mixed results may therefore not really be explained by the variables as stated in those respective studies (Dyer Jr., 2006). A second problem with these definition differences is that the incorporation of different kinds of businesses might have led to different results relating to, for instance, firm performance. As Miller et al. (2007) show in their paper, when not including single-founder businesses in the family firm category, there is no evidence supporting superior market valuations. Moreover, the paper states that including singe-founder firms as family firms, as has been common among previous studies, is surprising as such, given the absence of any other family involvement. Family firms with multiple family members serving as owners or managers can for instance make more use of governance conditions than single-founders without other family members in the firm (Miller et al., 2007). Together, both these papers show that the differences between family and non-family firms vary greatly with the definition of family firms used (Miller et al., 2007; Dyer Jr., 2006). In an attempt to overcome these issues, in 2009, the expert group on family business has formulated a fixed definition of what constitutes a family business. This should make it easier to compare findings in different studies dealing with the differences in family versus non-family firms (Flören et al, 2011). However, since this study will be using data from previously conducted studies which used the definition of Anderson & Reeb (2003A) this latter mentioned definition will also be used in this study. According to this definition a family firm is a firm from which the founder or a member of his or her family owns shares in the firm and/or a member of the family is present on the board General Findings By using this definition, Anderson & Reeb (2003A) come to the conclusion that family firms outperform their non-family counterparts. Nevertheless, because several studies have come to rather different conclusion on this subject, the debate whether or not these firms do indeed outperform their non-family competitors is still ongoing (Pindado et al., 2011; Lee, 2006). There are a number of papers which find that family firms in general outperform (Villalonga and Amit, 2006; Maury, 2006; Martinez et al., 2001; Anderson and Reeb, 2003A), while others come to the opposite conclusion (Miller et al., 2007; Faccio et al., 2001; Barth et al., 2005). These contradicting findings can be explained by the fact that there are advantages as well as disadvantages related to family control (Anderson and Reeb, 2003A). An often mentioned advantage is the fact that families are long-term investors committed to the success of the firm they invest in (Pindado et al, 2011; Isakov and Weisskopf, 2009). One of the disadvantages on the other hand, as described in previous literature, is the different priorities families might have from outside shareholders which may hinder value creation and growth of these companies (Barontini and Caprio, 2006; Gompers et al., 2004). In general, the advantages and disadvantages are highlighted by the magnitude of agency problems and the differences in that respect between family and non-family firms. 10

11 2.2 Agency Theory: General Agency Theory As previously mentioned, advantages, disadvantages and other differences between family and non-family firms are often the result of differences in agency related issues. These issues are related to the agency theory which is based on the separation of ownership and control (Jensen and Meckling, 1976). A good definition to clarify this theory is given by Jensen and Meckling (1976) who define this agency relationship as: a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent. At first sight this might not have to be an issue when both the principal and the agent act in the best interest of the firm, hence the interest of the principal. Nevertheless, in case both parties (agent and principal) involved are trying to maximize their own utility, this could lead to the agent not acting in the best interest of the principal (Jensen and Meckling, 1976). These possible different interests can be reduced by implementing incentives in order to align the interest of the agent with the principal and/or by monitoring the actions of the agent to limit the possibilities of shirking (Demsetz and Lehn, 1985). However, to realize these possible solutions and make sure the agent will make the most optimal decisions for the principal, certain costs will have to be made. Jensen and Meckling (1976) separate 3 kinds of costs associated with this principal agent problem. The first are monitoring expenditures which, as explained before, are incurred to check whether the agent does indeed act in the best interest of the principal. The second type of expenditures they connect to agency costs are the so called bonding costs. These are expenses to guarantee that the agent will not act in disfavor of the principal and in case he does, to ensure compensation for those actions. The final type of agency cost is called residual loss and is occurred because of the ever remaining difference in decisions taken by the agent and the decisions that would be optimal for the principal. A good example of this agency theory can be found by studying the relationship between managers and shareholders of a firm. These principles (shareholders) and agents (managers) are often confronted with the two types of agency problems namely, adverse selection and moral hazard. Adverse selection is based on the case where the agent, beforehand, has more information available then the principal. A good and famous example of this adverse selection is given by (Akerlof, 1970) which describes the lemons problem in the market for second hand cars. In that case the salesman (agent) has more information than the client (principal) about the car and is able to act upon this difference in knowledge in its own interest. In the case of the manager vs. shareholder conflict this adverse selection can be used by a manager, who is more directly informed about the firm s performance, to make arrangements with the shareholders in its own best interest, for instance, resulting in a higher bonus. The other agency problem, moral hazard, has to do with the possibility of the agent to alter the results of an already agreed upon contract or other agreement. This could for instance arise because of a manager s tendency to obtain assets out of the firm for his own use. Moreover, in case there do not 11

12 exist any incentives for the agent, as described before, this moral hazard could lead to the agent not acting upon opportunities for the firm because of a lack of interest (Jensen and Meckling, 1976). For the purpose of this study it would be interesting to see how this agency relationship effects family firms Agency Theory in Family Firms When studying differences in agency problems between family and non-family firms it shows that these are shaped by differences in separation of ownership and management (type 1 agency problem) and the existence of controlling and non-controlling shareholders (type 2 agency problem) (Ali et al., 2007). The separation of ownership and management (type 1 agency problem) may lead managers to act in their own interest which might not be in the best interest of the shareholders. In family firms this is likely to be a less severe issue because of the tighter alignment of ownership and control and the owner s ability to directly monitor management (Klein et al., 2005; Demsetz and Lehn, 1985). Furthermore, within family firms the owner and manager are likely to attach value to their relationship that goes beyond the economic contract (Gomez-Mejia et al., 2001). On the other hand, family firms are more likely to have larger agency problems arising between controlling and noncontrolling shareholders (type 2 agency problem). This can for instance occur when the controlling shareholder, the family, extracts money from the firm for their own benefit at the expense of other, non-controlling shareholders (Ali et al., 2007). According to Faccio et al., (2001) and Anderson and Reeb (2004), this exploitation of minority shareholders is likely to occur when there is no, necessary, oversight. Mainly because the family only bears a fraction of the total costs involved. This extraction of wealth can be done in different ways, for instance by excessive compensation of family members, risk avoidance or the use of a special dividends (Anderson and Reeb, 2003B). Whether this creative selfdestruction is more or less serious than the agency problems in other firms remains an unanswered question (Morck and Yeung, 2003). Nevertheless, compelling research has shown that on average the advantage of the less severe type 1 agency problem more than offsets the disadvantage related to the more severe type 2 agency problem within family firms (Barontine and Caprio, 2006; Ali et al., 2007; Wang, 2006). Dyer Jr. (2006) however argues that the owner-manager issue, contributing to less severe type 1 agency problems, does not have to be a unique aspect of a family firm. According to Dyer Jr. (2006), non-family firm owners can also manage their enterprise and thus receive the same benefits as family firm owner-managers. Implying that the owner-management effect on firm performance is not necessarily unique to a family firm (Dyer Jr., 2006). This idea is supported by Hutton (2007), who argues that what really matters in this case is the personal identities of the owner and the manager. Moreover, although recognized before as being an advantage to have less severe type 1 agency problems, thus more aligned owners and managers, Dyer Jr. (2006) argues that when family members are not equally contributing to the firm s needs, this might lead family members to fight for their own interests. In other words, a family manager does not have to have identical concerns as other family shareholders and can 12

13 therefore also want to pursue its own agenda (Gomez-Mejia et al., 2001). Using this perspective, type 1 agency costs may rise because of a conflict of interest accompanied by family involvement. As a result, this might even increase type 1 agency costs of family firms over non-family firms. Besides, even if these costs do not exceed the ones from non-family firms, this might mitigate the advantage gained on the less severe type 1 agency problems and with that, offset the, on average, favorable position towards agency problems of family firms. Therefore, besides the primary question whether or not this less severe type 1 agency problem can be contributed to family firms, it is the question whether or not this will always be beneficial for the firm (Dyer Jr., 2006)? To provide an answer to these questions we can look at Villalonga and Amit (2006) who incorporate the governance structure of family firms with both these types of agency problems in order to study the influence on firm performance. Combining both these variables they come to the conclusion that family firms outperform only when the founder serves as a CEO. This is because of the fact that in that case the type 2 agency problem is less costly than the type 1 agency conflict in nonfamily firms. However, they also come to the conclusion that a firm with a descendant acting as CEO might perform better because the owner-manager conflict (type 1 agency problem) is less costly than the conflict between shareholders (type 2 agency problem) (Villalonga and Amit, 2006). This view on the different results found with respect to the effect of agency problems on family firms is supported by Miller and Miller (2006) and Mishra et al. (2001). The paper by Miller and Miller (2006) contributes the possible positive effect on performance of founder and descendant to the longer tenure and highly trustable environment which mitigates the owner-manager conflict. Mishra et al. (2001), on the other hand, address the possible pitfalls associated with family members benefiting and overcompensating themselves at the cost of other shareholders. Braun and Latham (2009) tend to agree with Miller and Miller (2006) and argue that what is good for the family, might in the long term also be good for other stakeholders in the firm. Furthermore, they counter argue the type 2 agency problem by stating that a family will not be encouraged to exploit firm resources for its own private benefit. This is consistent with the alignment effect, discussed later, that family firms create their own incentives for the family to maximize overall shareholders wealth (Wang, 2006). Anderson and Reeb (2003A; 2003B) support this view by stating that since, for families, the welfare of the firm is so closely related to their own wealth, they will have strong incentives to monitor their managers and mitigate moral hazard. When studying the evidence found it can be concluded that although in a different mix and with different magnitudes, family firms as well as non-family firms have to cope with agency problems. Nevertheless, there are certain factors that can mitigate the differences between family and non-family firms and their agency problems. These factors will have an effect on the corporate disclosure practices of these firms (Ali et al., 2007). For instance, in order to mitigate the conflicts of principal and agent and align the different interests, accounting earnings are used (Wang, 2006). Furthermore, the separation of ownership and control demands for regular information in order to protect creditors and shareholders 13

14 in case of liquidation (van den Brand, 2005). This in turn has led to a demand for quality financial reporting, also to be able to monitor contracts with managers (Wang, 2006; Watts and Zimmerman, 1983, Gray et al., 1995). 2.3 Voluntary Disclosure General In a response to the before mentioned demand for quality financial reporting, firms supply information from which the financial statements are an important part. Besides the information that is nowadays required to disclose by law, firms voluntarily disclose more information (Meek et al., 1995). This voluntary disclosure, which exists out of disclosures in excess of requirements, contains information which is relevant for users of annual reports and from which its disclosure is up to the choice of management (Meek et al., 1995). A possible explanation why firms voluntarily disclose accounting information is the existence of asymmetric information, as described by the principal-agent problem (Gray et al., 1995). Because of the uncertainty between on the one side, parties like employees, customers, investors and regulatory agencies and on the other side, management, there is a cost for the firm to be perceived as a, previously mentioned, lemon (Meek et al., 1995; Gray et al., 1995). To overcome this issue firms will voluntarily disclose more information to get rid of the uncertainty and obtain the most optimal terms when raising capital (Firth, 1980). According to Gray et al. (1995) the reduction of this information risk will lead investors to accept a lower rate of return which subsequently will lead to a reduction of the cost of capital. This consequence of voluntary disclosure is also found by Chen et al. (2008) and Meek et al. (1995). Chen et al. (2008) find that voluntary disclosure reduces transaction costs between related parties and therefore reduces the cost of capital. Meek et al. (1995), also support the idea of Gray et al. (1995) by concluding that voluntary disclosure enables firms to raise capital on the best available terms which often implies getting the lowest cost. Nevertheless, besides this literature that at first sight seems to be anonymous in the effect of voluntary disclosure on the cost of capital, Botosan (1997) does not find any evidence between his measure of disclosure levels and the cost of capital. However, when further studying his findings he has to conclude that voluntary disclosure does have a negative effect on the cost of capital in case the firm has low analyst following. All in all the literature shows a reduction in the cost of capital as a result of an increase in voluntary disclosure (Chen et al., 2008). As a result, companies that face a lot of capital market pressure for the disclosure of information, for instance companies raising capital internationally, are likely to disclose more voluntary information in their annual reports in order to reduce the cost of capital (Gray et al., 1995). Up till now it seems that the higher the level of voluntary disclosure provided by firms the better it is in the end for all shareholders because of the lower cost of capital that can be obtained. This is indeed what Chen et al. (2008) indicate by stating that the general assumption is that all shareholders prefer more voluntary disclosure to less. Nevertheless, others argue that it is not quite that simple and 14

15 argue that executives have to find a balance between the lower cost of capital on the one hand and the cost of collecting and processing the information, litigation costs and political costs on the other (Gray et al., 1995; Meek et al., 1995). Furthermore, there are potential costs related to competitors being able to obtain information from other firms. This latter mentioned proprietary cost could induce firms to disclose less information and consequently accept a higher capital cost rate. This would especially be the case for firms operating in different segments that otherwise could hide the performance in each separate business section (Healy and Palepu, 2001). Taking these other costs into consideration, companies will be expected to voluntarily disclose information when the benefits of doing so exceed the direct and indirect costs involved (Meek et al., 1995). Some of the factors that have an influence on these possible benefits and costs are summarized in Healy and Palepu s (2001) article. They differentiate between; capital market transactions, corporate control contests, stock compensation, litigation, proprietary costs, and management talent signaling (Healy and Palepu, 2001). Besides the differences in costs and benefits associated with these factors, it becomes clear when studying previous literature, that the explanation for voluntary disclosure practice depends on the type of information as well as on the type of shareholders involved (Chen et al., 2008, Meek et al., 1995). Some provide voluntary disclosure in order to inform about the stewardship of managers with respect to the company s resources while others provide it to help investors determine the timing and uncertainty of cash flows (Gray et al., 1995). These differences in the use of voluntary disclosure become particularly clear when investigating different concentrations of ownership (Chen et al., 2008). Nevertheless, studies coping with this subject seem to reach different conclusions. On the one hand there are studies arguing that more concentrated ownership is related to less voluntary disclosure while on the other hand others support the idea of concentrated ownership leading firms to provide more information (Chen et al., 2008). In line with this discussion there are some recent studies investigating this relationship by looking at family and non-family firms and their respective voluntary disclosure practices (Wang, 2006; Chen et al., 2008; Ali et al., 2007; Chau and Gray, 2002; Van den Brand, 2007) Voluntary disclosure in family firms Several theories predict that family influence in a firm can have an effect on the demand and supply of voluntary disclosure in two different ways (Wang, 2006). The first being the entrenchment effect and the second the alignment effect, which was already shortly mentioned before. The entrenchment effect and the alignment effect have some opposite characteristics. The entrenchment effect has some similarities with the agency type 2 problems as it is based on the prediction that family members in a family firm extract wealth from other shareholders or other investors. In order to, for example, hide the negative effects associated with family members serving certain positions in management and/or interfamily monetary transactions, little or no voluntary disclosure will be provided (Ali, 2007; Wang, 2006). The alignment effect on the other hand counters this perception and is based 15

16 on the prediction that families align their interests with the ones from other shareholders leading to higher levels of voluntary disclosure (Wang, 2006). Which of these effects has the strongest influence on the voluntary disclosure practice of family firms might not seem clear at first sight. The entrenchment effect is likely to result in a high demand for more voluntary disclosure by involved parties because of their believe that family firms have something to hide or in other ways have an inferior governance practice in place. The alignment effect on the other hand is likely to have the opposite influence on the demand of voluntary disclosure since involved parties have the believe that the family has the same interests as them (Wang, 2006). When studying the entrenchment effect more in depth one might conclude that family firms should have little managerial entrenchment since the power is shared between people of the same family who are able to provide a natural check on one another (Wang, 2006). Nevertheless, it could also have the opposite effect and lead to greater opportunities for entrenchment which may result in higher agency costs between family and other, non-controlling, shareholders (Hutton, 2007). This latter view is supported by Gomez-Mejia et al. (2001) who also suggest that agency costs are likely to increase as a result of higher levels of managerial entrenchment. However, this result is not universally supported because other studies find no evidence of entrenchment of founding families in family firms. Nevertheless, these studies are based solely on the U.S. and might therefore be related to better shareholder protection which might make it difficult to generalize these findings (Chen et al, 2008). Moreover, it is not even clear whether entrenchment has an impact on voluntary disclosure decisions at all. According to Chen et al. (2008) it is unlikely that voluntary disclosure can reveal or hide evidence of entrenchment. Besides these mixed findings concerning the entrenchment effect, the alignment effect also has some arguments against and in favor. A possible negative alignment effect for family firms could be, as mentioned before, the possibility that involved parties have the believe that the family has the same interests and therefore do not demand any form of high quality earnings or voluntary disclosure (Wang, 2006). However, since a very important incentive for families is to pass on the firm to future generations, they are more likely to focus on the long-term and supply high levels of voluntary disclosure instead of engaging in earnings management (Anderson and Reeb, 2004; Wang, 2006). By supplying these levels of voluntary disclosure they avoid possible damage to the reputation, overall wealth and performance of their firm. Moreover, because they see the firm s health as an extension of their personal health this is likely to be a strong motivation (Anderson and Reeb, 2003B). Taking both these effects into consideration, Wang (2006) finds higher earnings quality, implying more voluntary disclosure, associated with family ownership. The conclusion that family firms have higher earnings quality is supported by Ali et al. (2007). On the one hand they find that family firms make less voluntary disclosures when it comes to corporate governance. According to their study this is caused by the entrenchment effect related to getting members of the family on boards without other shareholders interfering. However, on the other hand 16

17 they conclude that overall earnings quality is higher for family firms since they do make better and more disclosure about their financial performance. Contrary to the findings of Ali et al. (2007) and Wang (2006), there are some studies reporting lower voluntary disclosure by family firms (Chen et al., 2008; Chau and Gray, 2002; van den Brand, 2007). Chau and Gray (2002) do not specifically focus on family firms, but look in a broader sense to general ownership structures and the relation with disclosure practices. Nonetheless, they do conclude that outside ownership is positively related to voluntary disclosure, which implies that for insider or family firms the level of voluntary disclosure is likely to be less (Chau and Gray, 2002). One of their explanations for this finding is in line with the alignment effect. They namely argue that there is little motivation for family firms to voluntarily disclose information since the demand for that information is less than that for their non-family counterparts. In a paper by van de Brand (2007) this finding is confirmed and his results also show less information disclosure by family firms in comparison with nonfamily firms (van den Brand, 2007). More in depth research shows that family firms, compared to non-family firms, provide fewer earnings forecasts but more earnings warnings (Chen et al., 2008). It becomes clear when studying this previously conducted research that family firms do have different preferences than their non-family counterparts when it comes to voluntary disclosure. These different preferences are explained by differences in the tenure of the investment horizon, agency problems between owners and managers, possibilities of monitoring management and different concerns related to reputation and litigation costs (Chen et al., 2008). A family firm s longer investment horizon in combination with lower information asymmetry between owner and manager and the better possibilities of monitoring management leads to the conclusion that family firms prefer less voluntary disclosure. However, there is also an argument to be made in favor of family firms preferring more, rather than less, voluntary disclosure. This is based on the fact that usually family firms own a large part of the equity of the firm and are therefore closely linked to both the benefits of voluntary disclosure as to the costs associated with nondisclosure (Chen et al., 2008). One of the important benefits of voluntary disclosure is, as elaborately explained before, the reduction in the cost of capital (Meek et al., 1995; Gray et al., 1995). Important costs related to nondisclosure, in particular related to family firms, are the direct costs associated with litigation and the consequential damage that it may cause to the family s and firm s reputation. Voluntarily disclosing information, especially in relation to bad news, may preempt litigation and is therefore a good motivation for family firms to prefer more disclosure (Chen et al., 2008). Nevertheless, this argument related to bad news disclosure leading to more voluntary disclosure by family firms may easily be counterbalanced by greater concerns about the careers of managers in non-family firms. These managers face a greater possibility of losing their job when it is discovered that they are withholding bad news (Chen et al., 2008). 17

18 These contradicting results show that it is not as straightforward as it might seem to conclude whether or not family firms provide more or less voluntary disclosure. Chen et al. (2008) suggest that family firms are more likely to face costs rather than benefits from disclosure of information and therefore prefer less disclosure. This view is supported by their results that relative to non-family firms, family firms, whether it is good or bad news, tend to provide less voluntary disclosure. This seems at odds with the before mentioned findings by Ali et al. (2007) and Wang (2006). 2.4 Hypothesis It remains an empirical question whether family firms provide more or less voluntary disclosure. However, one thing that is certain is the fact that family firms have different preferences and incentives when it comes to providing voluntary disclosure compared to non-family firms. With these findings in mind the following hypothesis is formed to help answer the main research question of this study. However, since it is thus still unclear whether family firms provide more or less voluntary disclosure the hypothesis can only be non-directional. As is the case in the paper of Chen et al. (2008). Therefore, the hypothesis is: H1: The level of voluntary disclosure significantly differs between family and non-family firms. However, as mentioned in the introduction, in order to test this hypothesis, and thus answer the main research question, a difference would have to be made between voluntary and mandatory disclosure levels. Since this study is based in an international setting one problem will be the significant differences which exist across countries related to disclosure requirements (Gray et al., 1995). However, this study will overcome this issue by studying historical data from 1930 when there were little or no mandatory disclosure requirements. 18

19 Chapter 3 Historical Relevance and Background This chapter will give an outline of the historical relevance and background of accounting. The importance of historical accounting research will be discussed followed by a brief description of relevant accounting history in the Netherlands, Belgium and the United States. 3.1 Historical Accounting Research As stated in the previous chapter the difficulty in separating different levels of disclosure requirements can be overcome by focusing on a period in time in which none or little such requirements existed. Chow (1982) also investigates characteristics of voluntary practices and even argues that it is, necessary, to select a time period without external audit requirements. However, there are some issues to take into account when using historical data in research (Bricker and Chandar, 2003). According to Bricker and Chandar the use of historical data comes with both potential and difficulties because it is more than simply taking the data of an earlier period. Besides the use of this data it namely also requires an understanding of the environment of that period. Furthermore, it is important to be aware of possible biases related to how we think nowadays which might be completely different than in earlier periods (Bricker and Chandar, 2003). Problems related to these issues are common in all historical investigations, nevertheless they will be more prevalent in studies over a longer time period in which the environmental conditions might change and for which solely relying on the date would be inadequate (Bricker and Chandar, 2003). This latter finding might seem at odds with the argument from Rajan and Zingales (2003) who find that the state of a financial sector s development does not change significantly over time. This argument is supported by their findings that most countries were better developed financially in 1913 than in 1980 (Rajan and Zingales, 2003). However, although maybe the state of the financial sector does not change to much, this does not mean that other factors may not be of influence on these findings. Therefore, in order to successfully conduct historical research one needs to get an understanding of the context of the period. Bricker and Chandar (2003) mention two steps that are necessary to use when conducting such a study. The first is to extensively read trough the existing literature both about the period at hand and originating from that period. The second step is to get an idea about the companies to include in the sample and in particular to understand the company s ownership and control relationships. Furthermore, with regard to sample selection in historical research it is important to be aware of the fact that it is likely that more data will have been preserved from successful companies (Watts and Zimmerman, 1983). Besides these remarks related to historical research some other papers argue that it indeed can be very useful for today s purposes (Bricker and Chandar, 2003; Basu, 2003; Napier, 1989; Richardson and MacDonald, 2002; Watts and Zimmerman, 1983). Even Bricker and Chandar conclude in the end of their 2003 article that, when carefully considering the context of the period used in the study and taken into account possible biases related to current ideas, historical studies do contribute in finding answers to current accounting issues. Watts and Zimmerman (1983) also state that besides the possible ex-post 19

20 selection bias related to data coming more from successful firms, historical research does provide interesting insights. This statement is supported by Basu (2003) who sees accounting history research as a potential for more powerful tests of the role of accounting. However, at the same time, in line with Bricker and Chandar (2003) he points at the importance of understanding both the economic and institutional environments and the use of appropriate data (Basu, 2003). Nevertheless, since for instance the role of government was smaller in the past it is indeed easier to conduct research related to voluntary disclosure practices. This can give us a better insight and avoid us to repeat past mistakes or better understand certain connections because firms in the past had incentives to experiment with accounting methods in order to help them survive and prosper (Basu, 2003). Other advocates of conducting historical accounting research are Flesher et al. (2005), Walton (1995) and Foreman-Peck, Flesher et al. (2005) particularly focus on the materials that can be found in historical periods which can give a perspective for today s accounting researchers. On the other hand they warn for the possible bias in viewing past events in the limited perspective of the twentieth century as did Bricker and Chandar (2003). They support this by stating that even when incorporating the stock market crash of 1929 there is still a lot of literature missed from preceding periods (Flesher et al., 2005). Nevertheless, as mentioned before they do also advocate the relevance of historical audit research in gaining an appreciation for today s events and processes. This is supported by Walton (1995) who states that it is necessary to examine the historical development in order to be able to understand the present of any particular country. The fact that a particular country is mentioned in his argument has to do with the fact that accounting history is very much a nation-centered research area (Richardson and MacDonald, 2002). However, Foreman-Peck (1995) does not focus just on one particular country but he argues that also to understand European accounting in general, one needs to appreciate the economic history. This finding is in line with arguments in favor of the importance of historical accounting research in an international setting (Richardson and MacDonald, 2002). According to Richardson and MacDonald (2002) the pace of change and the need for understanding, which can be fulfilled by historical accounting research, is nowhere greater than in an international setting. Another important area in which historical accounting studies can contribute, besides thus an international setting, is in agency theory (Bricker and Chandar, 2003). This is because agency theory has several features which make it particularly suitable to use in historical accounting studies (Bricker and Chandar, 2003). These two areas of historical accounting are combined in an article by Overfelt et al. (2010). They however leave a more in depth study of the disclosure of information in an unregulated setting to future research. Nevertheless, they do combine the importance of studying voluntary disclosure differences, as discussed in chapter 2, with the importance of doing historical accounting research as discussed here. This study will also be combining these two insights. 20