Ownership structure and its impact on corporate risk taking

Transcription

1 Ownership structure and its impact on corporate risk taking Emil Larma s Department of Finance and Statistics Hanken School of Economics Helsinki 2016

2 HANKEN SCHOOL OF ECONOMICS Department of: Finance and Statistics Type of work: Thesis Author and Student number: Emil Larma s Date: Title of thesis: Ownership structure and its impact on corporate risk taking Abstract: The purpose of the study is to test whether the ownership structure in Finnish firms can explain their risk taking. The study analyses both the ownership concentration and the type of the largest investor and their impact on corporate risk taking. I use two different samples, one consisting of the 500 largest firms in Finland for years and another sample with Finnish listed firms for years The descriptive statistics show that the largest owner in listed Finnish companies own on average around 31% of the voting rights, whereas a majority owner (>50% of voting rights) exists, on average, in 25% of the listed companies. Further, both samples show that the largest owner in Finnish companies is most commonly a family. The regressions indicate that the ownership concentration, meaning the level of ownership of the largest owner, does not affect risk taking, instead it is the owner type, which has a correlation with firm risk taking. The results imply that municipalities, venture capital firms, foreign investors, families, foundations and corporations have a positive impact on risk taking, meaning that they are present as investors in companies with higher risk. On the other hand, companies with dual class shares tend to have considerably lower risk taking compared to other firms. Keywords: risk taking, ownership structure, ownership concentration, ownership types, panel data, dual class shares, family ownership, foreign investors, publicly listed companies, non-listed companies

3 CONTENTS 1 INTRODUCTION Aim of the study Limits to the study Contribution Structure of the study AGENCY THEORY Agency problems Compensation structures Free-rider problem Moral hazard theory Moral hazard theory and ownership concentration THEORY ON RISK Risk and return Risk and utility functions Firm specific risk factors Capital structure Investments Measuring risk OWNERSHIP STRUCTURE AND INVESTOR TYPES Ownership structure Dual class shares Pyramid structures Cross-ownership Investor types Insiders Foreign ownership Family ownership Institutional investors The state as investor Private equity investors Summary on investor types...24

4 5 PREVIOUS STUDIES Wright, Ferris, Sarin & Awasthi (1996) Data Method Results Coles, Daniel & Naveen (2006) Data Method Results Anderson, Mansi & Reeb (2003) Data Method Results Summary of previous studies DATA Talouselämä Risk variable Ownership variable Control variables Size Growth Profitability Industry dummy Summary on variable definitions and sources Descriptive data Data on listed companies Risk variables Ln Volatility Ln Gearing Ln Equity ratio Standard deviation on ROA Ownership variables Ownership concentration Investor type Control variables...42

5 CEO ownership Majority ownership Dual class share Foreign investor Size Growth Profitability Summary on variable sources and definitions Descriptive data Data discussion Survivorship bias Endogeneity Extreme values METHODOLOGY Models Talouselämä models Listed sample models Estimation methods Hausman-test Fixed-effects model Random-effects model Robust standard errors Hypothesis and expected signs REULTS Talouselämä Listed sample Results for the basic sample Results on regressions with lagged ownership Results on regressions without the main ownership variables Comparing results with hypothesis and previous research Model diagnostics Normality Multicollinearity Autocorrelation... 73

6 8.4.4 Heteroscedasticity CONCLUSIONS Discussion Critical discussion on the study Suggestions for further research SVENSK SAMMANFATTNING REFERENCES APPENDICES Appendix 1 Companies in the listed sample Appendix 2 Companies excluded from the listed sample Appendix 3 Multicollinearity table listed sample Appendix 4 Multicollinearity table Talouselämä sample TABLES Table 1 The relationship between families and minority owners as described by Holan & Sanz Table 2 List of ownership dummies used in the Talouselämä 500 data set Table 3 Variable sources for Talouselämä 500 data Table 4 Definitions on variables used in Talouselämä 500 data Table 5 Descriptive statistics on data used in the Talouselämä 500 sample Table 6 Investor types in the listed sample...42 Table 7 Summary on primary and secondary sources for variables Table 8 Summary on variable descriptions Table 9 Descriptive data on listed firms... 46

7 Table 10 Hypothesis for the regressions, where hypothesis 1 applies for both data sets, whereas hypothesis 2-6 only apply for the data on listed companies...58 Table 11 Expected signs for Talouselämä 500 variables...58 Table 12 Expected signs for variables in listed sample data Table 13 Talouselämä 500 results Table 14 Results for regressions where the main variable for ownership has been removed Table 15 The results impact on the hypothesis FIGURES Figure 1 Distribution of ownership dummies Figure 2 Data distribution based on industry type Figure 3 Distribution of the largest owner by type Figure 4 Distribution of the second largest owner by type Figure 5 Distribution of the third largest owner by type Figure 6 Distribution according to industry type... 50

8 1 1 INTRODUCTION A modern company is a complex organization with many internal and external stakeholders who have to interact with each other to make the company work efficiently. Research in corporate governance tries to find solutions on how stakeholders should interact with each other as to minimize frictions between parties. The decisions made within the firm will not always be aligned with those of the owners, assuming that individuals will secure their own utility before that of the firm. Conflicts can arise between executives and financiers of the firm, managers and their subordinates or between owners and executives. In this thesis, I will concentrate on the conflicts that can arise between owners and managers decision-making, with emphasis on risk taking. Recent publications within finance, business and corporate governance have studied whether ownership structure has an impact on various firm characteristics, in particular, studies on ownership concentration, meaning the largest shareholders, and their possibility to impact firm decisions. Several studies have proven a relationship between ownership concentration and firm variables such as firm performance (Morck et al. 1988), firm value (Slovin and Sushka 1993), competitiveness (Gadhoum 1999), managerial ownership (Denis et al. 1997), CEO pay-performance sensitivity (Coles et al. 2006) and legal protection of investors (John et al. 2008). There are however, only a few studies that have touched upon how ownership concentration affects corporate risk taking. The agency theory considers that the ownership structure affects the ability of owners to influence corporate risk taking (Jensen and Meckling 1976). Consequently, large shareholders have powerful incentives to collect information and monitor managers in order to maximize their profits (Shleifer and Vishny 1986). As ownership increases, ceteris paribus, owners have higher incentives to take on risky projects, which could substantially raise firm profits. However, concentrating much wealth into one firm may force large shareholders to take a more risk averse position than if they had diversified portfolios (John et al. 2008). By taking a more risk averse position on investments, investors can secure their wealth by approving long term projects with lower profit margins. The net effect of ownership structure on risk taking will therefore become more difficult to interpret, and will depend on the largest shareholders trade-off between risk taking and securing future wealth.

9 2 Furthermore, this topic is interesting as corporate risk taking is essential for long-term economic growth. Therefore, it is of special value to understand the determinants of corporate risk-taking as it helps to identify which policy changes can improve it and thereby the economic welfare. This topic is also interesting, as a similar study has not been performed on the Finnish market, where family ownership is relatively common also in listed firms. Further, the thesis is done in collaboration with Boardman Oy, which is a company focusing on Finnish ownership and it is in their interest to find various relationships on how ownership structure of a firm can affect firm decision making. I have therefore decided to have a thorough look at the ownership structure of Finnish firms and studied if a relationship can be found between firm risk taking and ownership structure. 1.1 Aim of the study The purpose of the study is to test whether the ownership structure of Finnish firms can explain their risk taking. 1.2 Limits to the study The geographical limit for data is constricted to only include Finnish firms. This has some positive and negative effects, firstly it enables me to analyse the whole spectrum of Finnish listed firms, which consists of a large percentage of investments that the average Finnish individual invests in. It has been shown in previous research that the majority of investments are done in an individual s home country. Thus, the data should consist of a diversified set of investors and investor types, which is essential for my study. I have however, not only included listed companies, instead the data also includes a data set on unlisted firms. The study is limited to a period of 7 years. The main reason for including 7 years is that the period should be long enough to gain an insight in how ownership and risk has changed through time and how they affect each other. On the other hand, some of the data, especially considering ownership, were gathered manually, which makes it very time consuming to include more years. I have excluded all financial firms from the sample, firstly, as their capital structure is considerably different compared with non-financial firms and secondly, as there is a

10 3 wide range of previous research that has studied the relationship between firm risk taking and ownership structure for financial firms. Therefore, financial firms will not be included in the same model. 1.3 Contribution The relationship between insiders and risk taking has been studied previously, however, other ownership forms haven t received as much attention. Therefore this study takes a much closer look at ownership and how ownership can be defined compared with previous research. Insider ownership is regardless a very important form of ownership as the CEO has a big impact on risk taking. This thesis is somewhat unique as it has detailed information on Finnish CEO ownership. This study also contributes to previous research as it takes into account privately held firms, which is not very common in previous research. The main reason for normally excluding privately held firms is data availability, which is commonly very poor. This is therefore a relatively unique set of data as it includes financial numbers and ownership structures of privately held firms. Further, this paper aims at finding the ultimate owners of companies. I have used the ownership structure given from databases, and after that manually gone through the data and looked up the ultimate owners of holding companies as well as pooled together voting rights from same family members, as one can assume that they vote similarly on the annual general meeting. By pooling together these voting rights, I have discovered that the majority of firms actually have families as one of the three largest investors. Gathering and sorting the data has been very time consuming and hence, I believe that this is the part of the study with the largest contribution as the data is rather unique. As a last point, I will also contribute to previous research as I will use lagged ownership variables to see if this can explain firm risk taking. This has been performed in only a few studies, which usually have found a relationship here as well, as one can assume that leverage is decided for the long term.

11 4 1.4 Structure of the study The structure of the study is as follows; the first three chapters describe the theory relating to the paper. The first chapter describes the agency theory, followed by a section with the definition of risk and finally the theory parts ends with a discussion on the various ownership structures and the most common investor types. Chapter 5 discusses three articles that on the topic, or which have elements that are important for this study. Chapter 6 contains data description on the two data sets that are used in the study. Chapter 7 will describe the methodology used as well as the various regression models. In chapter 8 I will present the results and key findings in the study, whereas chapter 9 will make ending conclusions as well as a critical discussion on the study.

12 5 2 AGENCY THEORY Studies within business and economics have for a long time documented the issues of various agency conflicts within modern companies. These conflicts arise from the classical agency relationship, which is especially visible in publicly held firms. The underlying problem in the agency relationship is the separation between ownership and control, meaning that owners do not have access to the managers decision-making (Berle & Means 1932). In this relationship both parties will strive to maximize their own utility, assuming that both parties behave rationally, which will lead to a situation where the agent (manager) will not act in the principals (owners) best interest (Jensen & Meckling 1976). The principal will not be able to align the agents interests with his own without introducing suitable incentives for the agent. Hence, it is generally impossible for either the principal or agent to secure that the agent will make decisions, which are aligned with the principals, without bearing any costs. Most agency relationships will include decisions made by the agent, which are not going to maximize the principals wealth and the loss in the principals wealth between a good and bad decision made by the agent is defined as agency costs. (Jensen & Meckling 1976) The relationship can be described by a simple situation. Assuming that a company only has one owner, who also is the executive, he will always make decision that maximizes his utility. The decisions can include monetary as well as non-monetary decisions, however, the company s actions will always be aligned with the owners interests. When the executive (also only owner) decides to sell a part of his shares to another individual, a new agency relationship will occur, as the original and external owners interests aren t aligned. The original owner and executive will now be forced to share the cash flow gains originating from the company with the new owner, however it also means that all costs occurring from the firm will be shared. Hence, the original owner will gain less from dividend distributions as his ownership has been diluted. Instead, he will maximize his own utility by using corporate expenditure on personal gains, as he now bears these costs together with the external owner. The minority owner, on the other hand, will recognize that this problem might arise and he will therefore be prepared to use resources to monitor the agents behaviour. This relationship can basically be applied to corporations with hundreds of owners as there always will be conflicts of

13 6 interests between both majority- and minority owners as well as between owners and the managers. Another type of agency cost arises due to difference in the willingness to take on risky projects. Investors will commonly think about the systematic risk in the industry in question or the economy in general, as they usually have a diversified portfolio and can therefore minimize the company specific risk by diversification. Executives, on the other hand, usually have a large portion of their wealth tied to the company, both in form of basic salary, as well as potential compensation programs tied to the firm s shares. Therefore, the executives might lose a substantial part of their wealth if they would obtain very risky investments, which eventually could lead to great personal losses if projects are unsuccessful. Executives will be more risk averse, compared to the owners, as they will be able to keep the company risk levels at levels they are comfortable with. This will lead to a situation where executives will usually prefer projects with lower risk and longer investment horizons, whereas the owners often have a rather opposite view on risk appetite. (Easterbrook 1984) 2.1 Agency problems The term agency problem has a wide definition and exists in many types of situations in companies, as well as on various levels of decision making within the organization. The most common forms of agency problems are related to the conflicts of interest between owners and executives. Hereafter I will discuss the most relevant agency costs Compensation structures Owners of listed companies create different compensation instruments in order to align executives interest with theirs. The most common are bonuses, stocks and options. Bonuses work as short term incentive instruments. They encourage the executives to increase firm performance for the coming year, as the bonus usually is tied to the company s end of year profit. Bonuses can however also be value destroying, as managers might maximize firm profits for the coming year, ignoring long-term profits (Cai et al. 2010). Options on the other hand, are long-term compensation instruments, encouraging managers to improve long-term firm performance. A manager taking part in an option program has the opportunity to buy shares in the company at a predetermined price at a future date. Assuming that the manager will maximize his own utility, taking part of the option program will align his interests together with those of

14 7 investors. Another long-term compensation instrument is company stocks, which also encourage maximizing long-term firm performance. Stocks increase manager s wealth, just like options, when firm performance increases, however, it also has a negative effect on the executive s wealth if the manager makes bad decisions in the firm. Hence, stocks are seen as the instrument, which most effectively aligns shareholder and manger interests. Compensating executives with bonuses, options and stock, is however not free of charge, hence the owners will need to bear the costs of compensation (agency cost) in order to align executives interest with theirs. Bonuses can be seen as a relatively cheap incentive instrument, whereas stock can be rather expensive. If the company in question is valued at say 1 billion euro, paying a million in bonus, will not be a huge cost for shareholders, whereas making the executive a major shareholder can become much costlier due to the dilution of old shareholders Free-rider problem The free-rider problem arises as s conflict of interest between majority- and minority owners. Assume a situation where the ownership structure is very dispersed, meaning that there is no controlling shareholder/shareholders, and where managers don t behave according to shareholders best interest. None of the individual shareholders will have enough resources to manage the executives, and we will have a classical agency conflict, which can partly be solved with a proper compensation structure. On the other hand, if we have a situation with one or several large owners, they will be willing to protect their wealth by monitoring executives, as their stake in the company is of considerable size. However, this means that investors with less ownership will not have any interest in monitoring the company, as they know that another investor has a considerably larger stake in the company. This is where small investors will freeride, and not share the monitoring costs associated with monitoring executives and the company together with other investors. Although the large shareholders will be forced to bear all the monitoring costs, they will commonly have access to management information that is not available for other investors and therefore they will have a better view of the company and its future. (Grossman & Hart 1980) In other words, a large shareholder might still get more out of the situation, in terms of semi disclosed information, compared with the monitoring costs they have to bear.

15 Moral hazard theory A moral hazard situation is generally defined as a situation where one person has the power to take a decision regarding risk whereas another person bears the costs of those risks. Economist Paul Krugman described moral hazard as "any situation in which one person makes the decision about how much risk to take, while someone else bears the cost if things go badly" (Krugman 2009). The moral hazard theory is therefore closely linked with the agency theory, or can be described as part of it, as it concentrates on the question of managerial risk taking. Executives usually have an urge to quickly grow company size as this will give them certain benefits, such as higher bonuses and corporate benefits. The effect is called empire building, where the manager strives to grow firm turnover as quickly as possible, whereas profitability might deteriorate (Jensen 1986). Growth can be achieved by taking on risky projects with high possible outfalls. However, this type of strategy does not satisfy certain large stockholders, as the risk taking is too high for their preference. At this stage, the shareholders could encourage executives to lower their risk taking by introducing new compensation structures. Options, will naturally not be the preferred instrument as it usually significantly increases executive risk taking (DeFusco et al. 1990, Sanders 2001) whereas stocks have proven to decrease executive risk taking (assuming that risk taking has been too high in the past) and should be preferred (Coles et al. 2001). The other side of the moral hazard theory is that we can experience executives who only invest in low risk projects. These types of managers are especially dominant in companies with high industry leverage, such as banks (Demsetz et al. 1997). The higher level of leverage means that mangers feel that they are in a riskier situation and will therefore be risk averse in order to secure their wealth (salary). Many studies (e.g. Demsetz et al. 1997, Low 2009) show that agency conflicts arise as managers are not willing to bear as much risk as the owners would like to undertake. Here again, the situation can be corrected for by introducing appropriate compensation instruments. Stocks and options will hopefully increase managers risk-taking as they will notice the upside potential that riskier projects entails to.

16 Moral hazard theory and ownership concentration As noticed from the moral hazard theory, large shareholders have an incentive to align firm risk taking with their own. This in turn speaks for a relation between large blockholder ownership and firm risk taking, which is the purpose of this study. A large concentration of large owners, as well as one large shareholder should therefore be able to explain the risk taking in a firm. If we take this one step further, it also means that if one of the largest owners also happens to be an insider in the company, one could assume that a relationship should also be found between ownership concentration and firm risk taking.

17 10 3 THEORY ON RISK What is risk and how can you define it? One way to look at it is volatility of an uncertain outcome such as the value of an asset. There are three basic types of risks that a company is exposed to; business, strategic and financial risk. Business risk consists of all the risk that the managers deliberately take on the firm, in order to grow firm value, strategic risk stem from the general movement in the economy, whereas financial risks are related to the risks on the financial market (Jorion 1997). This chapter will start with a definition of risk and return as well a short introduction to utility functions, followed by discussion on firm specific risk factors. The chapter ends with a short description on how risk can be measured statistically. 3.1 Risk and return Risk and return are one of the most important terms when it comes to investing. A general rule within investing is that less risk commonly yields lower returns, whereas higher risk can yield higher returns, but also great losses. In other words, an investor cannot yield high returns without taking on more risk. Risk does not only imply the potential yield (low/high) on the investment but also on the potential loss of the principal amount invested. The definition of risk and return refers to the relationship between risk and expected return, not between risk and realized returns. Generally, people perceive expected return as an outcome that will most likely happen, whereas it actually refers to an average value. Therefore, the outcome might be significantly higher or lower than the expected return on a specific investment. (Hull, 2010) An individual investor can relatively easily perceive what risk is and how risky investments are for them. A company on the other hand, has a more complex relationship between risk and return. Theory says that a company should always choose projects that have a positive net present value, meaning that the investment yields more than the expected return set by the firm. The company needs to put more attention to the individual projects risks, whereas investors are more diversified towards the individual project as they usually have a diversified investment portfolio. We can hence conclude that there is a positive correlation between risk and return and the only way to increase expected returns is to bear more risk.

18 Risk and utility functions We can generally divide investors in three groups according to their attitude towards risk; risk-averse, risk-neutral and risk seeking. Investors are assumed to behave rationally, which means that they will be risk averse. This means that given a situation with many projects with the same expected return, the investor will always choose the project with least risk. Humans have varying attitudes towards risk, although the average person is considered risk averse. Hence, people have different preferences when it comes to risk, some are comfortable with low returns and low yields, whereas others enjoy bearing risk with potential high returns. According to this logic, we can assume that all individuals have a utility function that varies according to personal risk preferences (Pratt 1964). Individuals will try to maximize their own utility functions, however, the individual utility functions will vary depending on the individuals risk taking preferences. The individual utility functions are therefore the underlying factor that differentiates one investor from another when it comes to risk taking. This thesis aims at looking whether a group of investors with same characteristics (e.g. family, state or investment advisors) have a similar appetite for risk, or in other words similar utility functions. 3.3 Firm specific risk factors Next, I will describe the background theory on how firms choose their capital structure followed by a discussion around investments and how they are related to firm risk taking Capital structure The basic theory on capital structure was laid out by Modigliani & Miller in 1958 and is still used today. The proposition says that firm value is independent from the capital structure, hence, the name irrelevance theory. They argue that the market value of the firm stems from its earnings and from the risk of its underlying assets and that the value is hence independent from the way the assets are financed. The following statements are assumed in a perfect market; no taxes, no transaction costs, no bankruptcy costs, all market players have the same lending costs and that the firm value is not predictable based on financial decisions.

19 12 The M&M proposition has some drawbacks when applying it to reality, as perfect markets are not as perfect as in theory. The foremost variable that differs between theory and current markets is the existence of bankruptcy costs, as they have a significant effect on modern corporations. Kraus & Litzenberg (1973) took into account the effect of bankruptcy costs and developed their trade-off theory. According to their model, optimal capital structure in a firm is chosen based on a trade-off between benefits received from the tax shield and the disadvantages from bankruptcy costs. Hence, a modern company has three alternatives on how to finance their need of capital; 1) retained earnings, 2) debt and 3) share emission. There are a few theories that try to explain how a company makes decisions between financing alternatives and the most famous once, the pecking-order theory, signalling theory and market-timing theory will be discussed next. The pecking-order theory assumes that there is a certain ranking that specifies in which order a company should seek for capital. The theory is based on the information asymmetries that exist between parties. According to the theory a firm prefers to finance itself by retained earnings, followed by external capital, such as bank loans or bonds. As a last resource of financing, a firm goes to the market for capital by issuing new shares. (Myers & Majluf 1984) By using this order of financing, the company will minimize the need to disclose firm specific information to outsiders. According to the signalling theory the firm signals the market about their financial and future state, through their choice of financing. If a firm chooses to use debt as financing instrument, it signals that the company is in good financial state as it shows investors that the firm has enough capital to pay back its debt and does not have to take into account potential bankruptcy costs. (Ross 1977) The market timing theory assumes that there is no difference between financing the company with equity or debt. Instead, the company choses between financing instruments depending on the company s market value. Hence, if mangers consider that the firm is overvalued, they will issue new shares, instead of using any of the other instruments. (Baker & Wurgler 2002) The choice of capital structure in a firm is based on many factors, such as those described above. The capital structure is not chosen by random, instead it is a conscious choice made by the management of the firm. Accordingly, the capital

20 13 structure is affected by the company s financial health, as well as by the state of the general market. Leverage has a significant effect on a company s risk taking, hence, managers can alter firm leverage and choose the preferred level of risk taking that the company bears Investments An investment can be defined as a transaction where an investor pays a principal at the current date in order to possibly receive remuneration in the future. An investment is considered irreversible, meaning that after the initial payment, one cannot undo the investment. Further, it is commonly not possible to predict the future return, it can either be considerably better than expected or it can yield nothing, thus loosing also the principal. A company s investments can be defined as how the managers choose to allocate capital between various projects. The managers also need to decide on the type of investment that capital will be allocated to. A company s investments can be divided into two main categories; Capital Expenditure (CapEx) and Research and Development (R&D). Capital expenditure is considered as a newly purchased capital asset or an investment that improves the usefulness of an existing asset. Purchase of new assets can include an investment into new headquarters or acquisition of new machinery for factories. Capex also includes larger improvements in existing assets, such as machinery, which considerably lengthens the usefulness of the machine. Although capex is commonly associated only with investments in fixed assets, investments in immaterial assets, such as acquisition of patents, can be included in capex. Capital expenditures vary from normal costs in the company, as they tend to be of considerable size, and therefore they are accounted for in a different fashion compared to other costs. Therefore, capital expenditures are usually depreciated over time or amortized (if we speak about immaterial assets). Capital expenditure varies significantly between industries, as the need for material assets differs considerably between industries. Industries, such as airlines, have significant assets, and will require considerable capital expenditures annually, whereas other industries, such as service businesses, have a light need of material assets, and will therefore have low annual capital expenditures. The risk that is associated with

21 14 capex, is that one can never know how profitable an investment will turn out in the future, therefore large capital expenditures can usually be associated with more risk. Research and development is the second type of investment, which is used for innovations in new products and the development of existing products. R&D investments are considered significantly riskier than capital expenditures, partly as the future outcome is more unclear and partly as there is no underlying asset that could be divested in order to get some of the original principal back. R&D investments are commonly characterized as a longer investment compared with Capex, which also increases risk. As an example; a technology developed today, might be out of date when it is launched after several years of development. Hence, investments in R&D are usually considerably smaller, compared to capex, as the risk is perceived as much higher. A manager can therefore adjust the risk taking in the firm depending on what type of investments that are made. If the manager is risk-averse, he will prefer capital expenditures that only just keep the machines and other fixed assets in working condition. Alternatively, the manager can make larger capital expenditures or/and increase R&D expenditure to increase the risk in the company, but at the same time increase the expected returns on these investments. 3.4 Measuring risk As earlier mentioned in the chapter, risk is commonly measured as volatility. This is not only restricted to firm value, instead it is used for various variables, such as price volatility on commodities. Volatility (σ) is defined as the standard deviation on a variables return for a time period, where the return is continuously compounded (Hull 2010). Volatility is given through time and is defined as the standard deviation of: Where S t is the price at time T and S 0 is the price at the beginning of the period. The function equals the total cumulated continuously compounded return for time T, not the return per time unit. If volatility is expressed per day or year, then, T also needs to be accounted in days or years. The variance is another variable for measuring risk and is defined as the volatility squared.

22 15 4 OWNERSHIP STRUCTURE AND INVESTOR TYPES This part of the theory will discuss the various controlling investor types, such as families, insiders, institutions and the state. I will analyse their investor rationale as well as their impact on company risk taking. Before moving on to the investor types themselves, I will first describe the various defensive actions that owners can use to secure their influence in companies. 4.1 Ownership structure As a company gets new owners, e.g. through an initial public offering (IPO), the founders ownership will be diluted. There are a couple of actions that the founders can do in order to keep high voting rights in the company, and therefore control. The most common measures include pyramid structures, dual share classes and cross-holdings. The purpose of these actions is to separate voting rights from cash flow rights, by transferring significant voting rights but only a fraction of the cash flow rights. Regardless of what type of structure is used, the ownership can be described as a controlling-minority structure (CMS). The separation of voting and cash flow rights will also create some conflicts of interest between parties. As an investor has more voting rights than equity rights, he can influence the company in ways that maximizes his own utility, e.g. the investor could raise or lower risk taking according to his own preferences. By doing this the investor is expropriating minority shareholders in the firm. There is, however, little that can be done to prevent the controlling-minority structures described below to expropriate other shareholders Dual class shares Perhaps the easiest form of CMS is the creation of a dual class share structure, where a company has two or more stock types. Each stock class will have a different amount of voting and equity rights, where one of the stock classes will have e.g. 10-times the voting rights compared with the other stock class, whereas the equity rights are equal. Hereby, a founder can keep a significant part of the voting rights to himself while the firm raises new capital by a share issue of the other stock class. This structure gives the controlling shareholder a strong foothold in the company, making it difficult for other shareholders to have their voices heard in the company. The controlling shareholder

23 16 can therefore use his voting rights and effectively influence manager s decisions. This form of CMS is the only one that does not require the creation of multiple firms in order to secure large voting rights. (Bebchuk et al. 2000) Although the structure is simple, dual class stock is not the most common CMS structure. One reason might be that it is relatively difficult for a company to introduce dual classes at a later date. It is obviously easier for large founding owners to introduce dual stocks before an IPO, as they have a large influence in the whole IPO process. However, if we assume a situation where an investor has increased ownership gradually to say, 30%, it will be difficult for this investor to introduce dual class stocks to the firm. It will be easier for the investor to use other CMS structures that are discussed later. Another reason why dual class stocks are not that common is due to the restrictions that corporate law puts on the allocation between voting and equity rights. Laws usually restrict the ratio of voting rights per equity right per share, which makes this type of CMS relatively inefficient (Bebchuk et al. 2000). Legal jurisdictions cannot wholly explain the relatively low popularity of dual class shares. La Porta et al. (1999) show in their study that even in countries with dual share classes, CMS companies usually do not reduce the control rights to the legal minimum. Dual class stocks are particularly popular in Sweden and South Africa. In Sweden, the best example is the Wallenberg companies. The Wallenberg Group owns a significant share of large, international companies such as ABB, Electrolux and SEB through its Family Trust and principal holding company Investor. The family holds some 40% of the listed shares on the Stockholm Stock Exchange. The family s voting rights in holdings through Investor consists of 50%, whereas the equity rights consist of 23% (Wallenberg.com). Dual class stocks are used by some 9% of companies on the Helsinki Stock Exchange (Euroclear). Probably the best example of this CMS structure in Finland is the Herlin family, which holds more than 50% of voting rights in KONE, through its ownership of class-a shares, which were originated prior to the listing of KONE Pyramid structures An investor can create a CMS structure with a single stock class, by building a pyramid structure of multiple corporations. In a pyramid structure of two companies, the investor has a controlling minority ownership in an investment company, which in turn

24 17 is the controlling minority shareholder in an operating company. The same applies for a structure with two or more holding companies. It is much less costly for the investor to become a controlling shareholder in the operating company through the ownership in the holding company, compared to a direct holding in the operating company. This structure also implies that the investor has higher voting rights compared with equity rights. (Bebchuk et al. 2000). La Porta et al. (1999) find in their study that pyramid structures are the most common mechanisms for concentrating control in CMS structures. There are probably two main reason for the use of pyramid structures; firstly, as mentioned earlier it is far less costly to buy a controlling minority ownership through a pyramid structure compared to a direct holding; secondly, a pyramid structure can make the investor more anonymous for the general public. A pyramid structure consisting of one holding company is also popular in Finland due to taxation reasons as an investor can raise part of the income from the holding company as tax-free. The use of pyramid structures is common both in Asian countries (Claessens et al. 1999) and some European countries (Bianchi et al and Holmen & Hogfeldt 1999). Hong Kong is especially known for families with large pyramid structures Cross-ownership Cross ownership refers to a situation where companies, doing business together, own part of the other company. In other words, companies with business relationships are linked horizontally through cross-holdings that reinforce the decision-making power in both companies. Cross-ownership differs from pyramid structures as the voting rights for controlling the companies remain across the range of companies, instead of in the hands of a single shareholder or company. (Bebchuk et al. 2000). Cross-ownership is mainly found in Asia, and less common in the rest of the world. 4.2 Investor types In this chapter, I have already discussed the various ways in which shareholders can create ownership structures, where they keep significant voting rights in the company, while equity rights remain relatively low. In this part of the chapter, I will have a look at how one can assume that different investor types use this situation to their advantage.

25 Insiders An insider can be described as a person who is either a CEO, member of the executive board or the board of directors and holds an ownership stake in the company. By having both control rights, through the position as executive, and equity rights, through shareholdings, a mangers interest should be more in line with those of shareholders. As discussed previously, the moral hazard theory gives a good theoretical reference on the theory on risk taking behaviour of insiders. The existing literature does argue for a positive relationship between firm risk taking and equity ownership, where an increase in insider ownership aligns manager s interest with those of shareholders (e.g. Chen & Steiner 1999, Shleifer & Vishy 1986 and Gadhoum & Ayadi 2003). On the other hand, Wright et al. (1996) show in their study that insider ownership does increase risk taking, however, the relationship may become negative at high levels of insider ownership. In other words, low level of insider ownership increases firm risk taking, whereas high levels of insider ownership is associated with lower risk taking. This would imply that insiders with large holdings perceive that a too large part of their wealth is tied to the company and their willingness to invest in riskier projects decreases Foreign ownership The basic definition of a foreign investor is a person who invests in assets that exists outside of the person s country of residence. Foreign investors are commonly not looking for a controlling ownership as they are often looking for geographical diversification. However, at some instances foreign ownership can lobby for foreign persons to the corporate board. This is perceived to have a positive effect on the company as an outsider might bring new thinking and ideas into the corporation. Otherwise, foreign owners usually have little effect on the companies themselves. The implications are, though, very different if a Finnish company/subsidiary is acquired by a foreign company. Finnish takeover targets usually go through a considerable change in company structure, as the company s functions are streamlined according with its acquirer. Further, it has been proven that international acquisitions of Finnish firms have a positive effect on target companies (Ylä-Anttila et al. 2004). Finnish law had strict regulations in the past regarding the level of foreign ownership in Finnish companies, which resulted in low foreign ownership (3-9%). However, these

26 19 regulations changed in 1992, when the market was opened freely for foreign ownership and as a consequence foreign ownership increased from 9% in 1992 to 65% in 1999 (Ylä-Anttila et al. 2004). Some studies have also argued that the Helsinki stock exchange is among the most international, due to high percentage of foreign investors. This effect is largely due to the Nokia effect, as Nokia consisted at its height of some 70% of the market capitalization on the exchange, whereof 90% were foreign investors. Therefore, we had a considerable bias that was not taken into account. The relationship between foreign investors and risk taking is quite unclear. Some foreign investors seem to have interest in affecting the people chosen to the board, whereas most investors seem to be inactive. Ylä-Anttila et al made a study on foreign investors on the Finnish market and they found a positive effect between foreign investors and firm performance. I however, think that the data might have a bias, as it only includes the years from , right after the regulation change. Foreign investors could therefore choose to invest only in the most promising companies, such as Nokia, which probably skewed the data Family ownership Family ownership is very prominent in many Finnish firms, both listed and privately held. Examples include listed firms such as KONE, Lemminkäinen and Ahlstrom, and privately held firms such as Fazer, Paulig and Etola. Family ownership can be seen as a special case of insider ownership, as the family commonly has a representative either as the CEO or in the Board of Directors. Looking at the situation from a shareholder protection point of view, Burkat et al reported a correlation between minority shareholder protection and ownership concentration. The study shows that agency costs increase as the juridical protection decreases, which in turns gives a controlling family ownership even higher ownership concentration and control. On the other hand, it has been shown that families usually act as a substitute of sorts for juridical protection, as the family wants to protect its wealth tied to the firm (Holan & Sanz 2006). The relationship between family dynamics and juridical protection can be illustrated according to table 1.

27 20 Strong juridical protection of minority shareholders Weak juridical protection of minority shareholders Good family dynamics The minority owners interests are taken into account, easy to keep existing and find new minority owners The family works as a substitute for the juridical protection, but can be weakened by bad family relations Dysfunctional family Current minority owners follow the company, difficult to find new owners Expropriation and misunderstandings with minority shareholders Table 1 The relationship between families and minority owners as described by Holan & Sanz 2006 The table speaks for itself, good juridical protection and functioning family dynamics, is represented by minimum agency costs between owners and managers. In the opposite corner, where both the juridical protection and family relationships are weak, the relationship is characterized by a situation with increasing risk for major agency costs between parties. It is somewhat unclear whether family ownership can create value in a company, in other words, do they create value by taking on risk or do they destroy value by being too conservative as owners. Families have large incentives to monitor the management, when they are controlling shareholders, which should decrease the classical agency costs. Therefore, the agency theory predicts a positive correlation between family ownership and value creation in the company. On the contrary, they have a large risk in the company and might stay as conservative and risk averse investors. The current literature is somewhat contradictory as some studies find that companies with founders within the management are more profitable, whereas other studies show that the company value decreases as family members are nominated to the company management. Palia & Ravid (2002) find that companies lead by the founders tend to be more profitable compared to other firms. Adams et al. (2009) also find a positive correlation between founders as CEOs and firm performance, and show that the founders usually step down from the position as CEO only when firm performance is high. A further

28 21 study by Anderson & Reeb (2003) prove that firm performance is higher for companies with families as controlling shareholders compared with other form of controlling shareholder types. Additionally, moral hazard problems appear to decrease in firms with families as controlling shareholders. On the other hand, Smith & Amokau-Adu (1999) show that firm value decreases when a family member is nominated as executive, whereas there is no reaction to firm value as outsiders are nominated as executives. The reason for the negative reaction appeared to be the newly appointed executive s low age, which reflects to low experience. Bennedsen et al. (2006) shows that the short-term (two days) stock return decreases by 1% when a family member is nominated, whereas it increases by 2%, when an external person is nominated. Further, Anderson et al. (2003) showed that families tend to avoid risk, as they want to secure the value of the firm for the next generation. To summarize, the relationship between family ownership and risk and return is rather unclear based on previous research. Given that shareholder protection is stable and fixed during the time period of the sample in Finland, we can assume that the amount of agency conflicts in a company is given by the dynamics within the family, as can be seen from table 1, instead of fluctuating and unclear shareholder protection Institutional investors The group of institutional investors consists mainly of various funds, such as pension, mutual and insurance funds. These investors are a significant group of investors, as they possess a large amount of the capital under management. This means that they have a large amount of capital that needs to be invested. Institutional investors usually have regulations that limit the managers acting and require the manager to have sufficient information about companies that they invest in. This implies that they usually have a close relationship with companies that they invest in, in order to be updated on company issues, which in turn suggests that they possess inside information that can be used as a valuation tool. Getting hold of information from the company obviously takes time and is somewhat costly for the institutional investor, however, it is usually very fruitful as the potential upside from a close relationship with managers can be very high. (Schnatterly et al. 2008) Institutional investors commonly own large blocks of shares, which can be both a positive or negative aspect considering how they can affect manager s behaviour.

29 22 Edmans (2009) argue that large block holders with long investment horizons, such as institutional investors, have a strong incentive to manage the firm s fundamental value. The investors can then trade on inside information, causing the share price to reflect fundamental values rather than current earnings. This does not satisfy managers, as share prices will seem to be off, and will therefore encourage mangers to invest in longterm growth projects, instead of short-term profits. Institutional investors are thus mostly governed by their internal regulations, which largely affect their ability to impact company management. On the other hand, institutional investors usually have a close relationship with companies they invest in and therefore have a lot of information regarding e.g. risk taking and that they therefore can choose to invest only in companies which risk profile suit them. Therefore, it is difficult to assess how an institutional investor will affect company risk taking The state as investor The state is a major player in some Finnish listed companies. The state has a majority ownership in three listed companies (Fortum, Neste and Finnair), as well as a minority ownership in some 10 companies. There are also numerous large privately held companies where the state has significant ownership, and can hence influence decision making. In state-owned non-listed companies, the state tends to drive for decisions with political value, instead of maximizing the value of the company itself. Listed companies with high state ownership can t be controlled in the same way as private companies, as companies on the stock exchange have stricter regulations. Further, the ownership policy must be consistent and equal throughout all companies, as investors will get rid of their shares, if large changes are made by influence of the state, for projects that maximize the value for the state instead of the value of the company. Pedersen & Thomsen (2003) show that the state puts considerable thought on its own political goals, and hereby drives matters such as pricing of products and employment, in ways that maximize their political agenda, instead of firm value. These decisions decrease the efficiency of the company and will often be reflected in negative firm performance and value.

30 23 Foreign investors, as well as some domestic investors, usually see state ownership as a negative factor, which is usually associated with risk. Therefore, these companies will be perceived as riskier and investors will require a higher rate of return on these shares. La Porta et al. (2002) studied how state ownership affects banks and their performance. They found that the state drives their political agenda through control in banks, which in turn leads to inefficiencies and slow economic and financial development in both wealthy and poor countries. As a conclusion, one can say that large state ownership tends to have a slightly negative impact on at least firm performance, but there has been no clear research on how risk is affected. However, given the lower firm performance, it could be argued based on the theory between risk and return, that firm risk taking is also lower when the state has larger shareholdings Private equity investors Generally, Private Equity (PE) investors invest in both listed and unlisted firms. In Finland though, most investments are done in unlisted companies, due to the small number of firms on the Helsinki stock exchange, and hence the possibilities to find new opportunities from listed firms are slim. Private equity firms make acquisitions, commonly called buyouts, in order to gain a majority stake in the target company. The investment period tends to be between 3 and 10 years, where the old/new management is usually required to buy a small amount of shares that incentivize the management to perform well. The investments are made through limited partnerships, where the investors have very limited liabilities, whereas the PE-firm takes on the administrative tasks of the investments. The PE-firm usually only has a minor equity stake in the limited partnership, whereas the limited partners own the majority equity stake. The voting rights, are however, distributed fully to the PE-firm, leaving the limited partners without any decisions making power. The companies owned by PE firms tend to take on a substantial amount of debt, as this increases the return that the PE firm and its limited partners get on the investment. The Finnish private equity market is rather underdeveloped as only 0,4% of GDP was invested in risk capital investments during , whereas the same number was 0,8% in Sweden, where the private equity market is much more developed.

31 24 There are a rather limited number of studies that show what private equity backed companies risk and return characteristics look like. This is due to the limited availability on data as PE firms disclose very seldom any data on their investments. Ljungqvist and Richardson (2003), argue that although private equity backed firms might have higher risk, they still have a better risk-adjusted return as the general return on PE-investments are high and diversified. Emery (2003) showed that private equity firms have a low correlation with stock exchanges, making it an attractive investment class. However, the general thought is that private equity backed firms tend to take on more risk, as they look for considerable growth Summary on investor types The issues between minority and majority owners will always exist in companies, where the majority owner will control the situation, irrelevant form the existence of controlling-minority structures. Depending on who the controlling owner is, they will have different preferences regarding how the firm should be lead and what level of risk that should be chosen. The relationship between insider ownership and risk can be described by a curve where zero ownership may reflect in low risk taking, a modest ownership (compared with the insider s total wealth) increases risk taking, whereas too high ownership can again lower risk taking. Families are usually perceived to hold a low or medium level of risk, and can at times work as a substitute for juridical protection for minority shareholders. Institutional investors on the other hand, are characterized as investors with significant capital holdings and block holders. They also tend to have more detailed information on the firms as they commonly have close relationships with managers. The state is perceived as an investor that invest in strategically important companies, however the state tends to drive a political agenda, instead of maximizing firm value for shareholders. This tends to lead to less efficiency in firms with the state as a controlling owner. Lastly, Private Equity firms are generally assumed to have higher risk taking profiles in their portfolio companies compared with other firms.

32 25 5 PREVIOUS STUDIES In this chapter, I will describe three research papers that are relevant to this thesis. None of the papers can be directly comparable to my study, but they have elements that are closely linked with this paper. The first paper has the closest similarity to my study as it takes into account both the ownership structure and various owner types and how they affect corporate risk taking. The second article studies the relationship between managerial wealth and its impact on risk taking. The various measures of risk used in this study are of high relevance for my study. Finally, the third paper describes founding family ownership and its relation to risk, calculated through publicly traded debt. This paper is of relevance as many Finnish firms have considerable founding family ownership. 5.1 Wright, Ferris, Sarin & Awasthi (1996) Wright et al. write in their article Impact of corporate insider, blockholder and institutional equity ownership on firm risk taking about ownership concentration and its impact on risk taking. The study puts considerable weight on the relationship between insiders and risk taking. The authors find that low to medium managerial ownership increases firm risk taking, whereas high managerial ownership decreases firm risk taking, as managers portfolios become less diversified. The study also looks at the relationship between blockholders and institutional ownership on firm risk taking. This paper is of special interest as it puts considerable weight on insider ownership and its effect on risk taking at different levels Data The study uses data on listed companies from 1986 and 1992, to test whether their hypothesis is stable through time. The data on a company had to meet three restrictions in order to be taken into account in the study; 1) insider equity ownership was available, 2) financial analysts forecasts on earnings per share (EPS) had to be available and 3) financial data had to be available in COMPUSTAT so that Tobin s Q could be calculated. A last requirement stated that companies needed to have fiscal years from January-December, so that statistical estimation would not have a bias. The final data that filled all requirements included 358 firms from 1986 and 514 firms for The data was later extended to include financial data from 1988 and 1994, as

33 26 ownership structure in a given year may affect corporate risk taking in following years. Hence, ownership data from 1986 was used to evaluate firm risk raking in 1988 and the influence of firm ownership in 1992 on risk taking in Method The method included a moderated cross-sectional regression analysis, in which risk taking, defined as the standard deviation of the spread between financial analysts forecasts on EPS, was regressed against separate variables of insider ownership. The study used Tobins Q as a moderator variable to test whether or not a company had growth potential, where after the data was divided in two groups. The authors believe that a nonlinear relationship exists between insider ownership, therefore ownership was divided into groups where managers had low ownership (0-7,5%) and high ownership (7.5%<). Control variables were used to control for size (assets) and industry effects (two-digit SIC code dummies) Results The results prove that there is a nonlinear relationship between insider equity ownership and corporate risk taking. Low insider equity ownership correlates positively with corporate risk taking, whereas high insider ownership tends to reduce firm risk taking. The relationship between institutional equity ownership and risk taking is positive, suggesting that institutional investors do use their position as controlling owners on firm risk taking. Further, the results show that blockholders have no statistically significant effect on risk taking, meaning that a blockholder, on average, does not seem to influence corporate risk taking, although blockholders would have the possibility to do so. 5.2 Coles, Daniel & Naveen (2006) Coles et al. article Managerial incentives and risk-taking discusses the relationship between managerial compensation and investment policy, debt policy and firm risk. The authors use managerial ownership to test whether it incentivizes the manager to take on more risk in the company. The primary variable used for managerial risk is the sensitivity of CEO wealth to stock return volatility (vega). The study finds a positive correlation between CEO compensation and firm risk, measured with a number of variables. Further, riskier policy decisions commonly lead to compensation structures

34 27 with higher vega. This paper is of specific interest as it includes relevant measures of risk Data The study uses data from 1992 to 2002 for companies on the following indices; S&P 500, S&P 400 Midcap and S&P Smallcap 600 (finance and utilities firms are eliminated). The compensation data is taken from Standars & Poor s Execucomp database and contains the following variables both for the CEO and for the top five managers; salary, bonus and total compensation. The total data sample consists of 10,687 observations after data requirements have been met. The authors define delta as the change in dollar value of the executive s wealth for a one percentage point change in stock price. Vega is defined as the change in dollar value of the executive s wealth for a 0,01 change in the annualized standard deviation of stock return. The following variables are used as variables measuring investments and financial policy; (1) R&D, defined as research and development expenditures related to assets, (2) CAPEX, defined as capital expenditures (capital expenditures less sales of property, plant and equipment) related to assets, (3) Segments defined as the number of different businesses in which the firm operates, (4) Herfindahl Index captures income concentration between segments and is defined as the sum of the square of segment sales divided by the square of firm sales and (5) Book leverage defined as total book debt scaled by book value of assets. The authors assume that the variables presented above should be captured in stock return volatility (Firm Risk), defined as the logarithm of the variance of daily stock returns. Following variables where used as control variables; Logarithm of sales, Market-to-Book, Surplus Cash, Sales Growth, Stock Return, ROA, Dividend Cut, CEO Turnover, Net PPE, Z-Score (proxy for the probability of bankruptcy), CEO Tenure and CEO Cash Compensation Method The authors use panel data with the fixed-effects model. The study regress each of the five risk variables presented above, against vega, delta and the other control variables. Further, the regressions are made separately for both CEOs and managers. All regressions are also run through robustness checks, using 3SLS regressions. This leads to a wide range of regressions, making the results even more thorough and robust.

35 Results The authors find a strong causal relationship between managerial compensation, and investment and debt policy. The results also show that higher vega implies riskier policy choices, which can be noticed from more investments in R&D and less investments in capital expenditures, higher focus on chosen business lines and higher firm leverage. These results are in line with hypotheses that higher sensitivity on share price volatility in the managerial compensation structure incentivizes managers to invest heavier on risky investments and takes on more debt. The authors also find that stock price volatility has a statistically positive relationship with R&D as well as company focus and leverage, and a negative correlation with capital expenditures. 5.3 Anderson, Mansi & Reeb (2003) Anderson et al. article Founding family ownership and the agency cost of debt discusses the relationship between family ownership and firm risk. The study looks specifically on how debt financing costs differ between companies where founding families have large ownership stakes and other companies. The authors expect that companies with founder family ownership will have lower costs of debt as they differ from other companies in three distinct ways; 1) families have commonly undiversified portfolios 2) family firms want to ensure a successful transfer of the firm to the next generation as a large part of the families wealth is tied to the company and 3) the family has a great pride in the company and will therefore ensure firm survival instead of concentrating on value maximization. Due to these differences, the authors argue that family firms will be able to create incentive instruments that better align interest in the firm, thus lowering agency costs. This article is relevant for my study as it shows the importance of family ownership, which is considered substantial in Finland Data The study contains data on companies that could be found both in the Lehman Brothers Bond Database (LBBD) and the S&P 500 Industrial Index between years 1993 and The LBBD provides monthly observations on outstanding publicly traded debt. The ownership data is collected by hand from annual reports, and the authors aimed at finding all the founding families ownership, although it might be widely spread. Any missing data was filled in from the Compustat Industrial Files. The final data sample consists of 1,052 firm-year observations on 252 firms.

36 29 The authors argue that there is no generally accepted variable to measure ownership and therefore they decided to use a binary variable for firms with family ownership. Other variables such as size of family ownership compared with blockholders, and family ownership as a fraction of outstanding shares, were used to increase robustness. The cost of debt on the other hand is measured by the yield spread. This is calculated as the difference between the weighted-average yield to maturity on the firm s outstanding traded debt and yield to maturity on a Treasury security with corresponding duration. The control variables consisted of Performance (Cash flow/assets), Risk (the standard deviation of the firm s cash flows scaled by long-term debt for the previous 5 years), Leverage (Debt/Assets), Size (logarithm of assets), Duration as well as a conversion number for Credit Ratings Method The method used in the study is relatively straight forward; the authors use crosssectional regressions to test the relationship between the various family ownership variables and the yield spread on bonds. The authors also test for the effect of outliers on the result by eliminating observations based on the R-Student and DFFITS scores. The authors find that there is no substantial difference in the results after removing outliers Results The authors find a statistically negative correlation between founding family ownership and bond yields, confirming that founding family ownership decreases the cost of debt. The study finds that firms with founding family ownership receive financing on the market at a 0,32% lower cost compared with other firms. The firm receives the largest gains when family ownership is beneath 12% of the firm s shares. However, at ownership levels above 12%, the cost of debt increases but it is still lower than for other companies. Finally, the study finds that debt costs rise as a founding family member is appointed as CEO, but costs still stay lower compared with other firms.

37 Summary of previous studies Authors Purpose Period Wright, Ferris, Sarin & Aswasthi (1996) Studies the relationship between ownership structure and firm risk taking 1986, 1992 Coles, Naveen & Nav een (2006) Studies the relationship between CEO compensation, investment and risk taking decisions on firm risk taking Anderson, Mansi & Reeb (2003) Studies the relationship between founding family ownership and the cost of firm debt Market USA S&P 500 S&P Midcap 400 S&P Smallcap 600 S&P 500 Industrial Index Dependent variable Volatility on analytics estimates on EPS Firm risk, Book Leverage R&D, Capex Log (segments) Herfindahl index Y TM spread between firm bonds and comperable Treassury bonds Results Finds a non-linear relationship between insider ownership and risk taking. Low ownership increases risk, whereas too high ownership decreases risk taking Positive relationship between CEO compensation, R&D (immaterial investments) and leverage, but a negative relationship with Capex (material investments) Firms with considerable founding family ownership receive cheaper debt financing on the market

38 31 6 DATA This chapter will present the data used in this study. The chapter will be divided into two separate main chapters as I use two sets of data in the study. The first chapter will introduce the Talouselämä 500 data, whereas the second part will present the data on listed companies. Both chapters include descriptive data. 6.1 Talouselämä 500 This set of data is based on the ETLA 500 list of the 500 largest companies in Finland and includes both listed and privately held companies. The list is published annually. I have used Talouselämäs modified version of the list, which includes ownership dummies for companies. The data was received from Talouselämä, however it required quite a bit of modifications, which will be discussed below. The data includes six years and spans from 2009 to 2014, as Talouselämä started using ownership dummies from year 2009 onwards. I have excluded all financial companies, as their capital structure is very different compared with other firms, an also excluded companies with no values for the risk variables Risk variable Due to the restrictions in data availability, I will use two measures of risk; Gearing and Equity ratio. These two variables measure the capital structure of the company. A company is said to be riskier when the equity proportion is lower, meaning that debt as a proportion of total assets increases. Gearing is defined as follows: where net debt consists of long term interest-bearing debt less cash and cash equivalents (ETLA). Thus, risk increases as gearing increases. Equity ratio is defined as follows: where total equity includes the equity value, provisions, minority interest, and concern reserve. Total Equity and Liabilities includes the amount stated on the balance sheet

39 32 adding the pension-fund liability deficit (if not already included in the balance sheet) and removing advance payments (ETLA). This variable works in the opposite way compared with gearing as risk increases when the equity ratio decreases Ownership variable Talouselämä started adding certain ownership dummies for companies in 2009 (family and foreign) and extended the amount of dummies in 2011 (municipality, shares publicly traded, cooperative, venture capital and state). The data did not include dummies for all companies for all years, and hence I had to manually go through most of the companies to confirm that the dummies where correct. The ownership data was gathered from firm annual reports, webpages, Kauppalehti and Solidium. The purpose of the ownership dummy is to give an indication on the characteristics of the largest owner. I want to test whether a certain group of investors, whether it be family, municipality or venture capital, tend to make similar decisions in the company when it comes to risk taking. As many firms also can have two dominant ownership characteristics, it is possible for firms to have multiple ownership dummies. For example, Ahstrom is a listed company, which is controlled by a family, therefore Ahstrom has an indicator variable for family ownership and one to note that its shares are listed. Talouselämä did not have sufficient dummies to correctly categorize all companies, therefore I added the following dummies; 1) Employees for companies that are owned by its employees such as KPMG and PwC and 2) Subsidiary for companies that are owned 50/50 by two listed companies such as Steveco. Below is a list of all the ownership dummies used for this data set. Ownership Dummies Cooperative/association Employees Family ownership Foreign ownership Municipality Shares publicly traded State Subsidiary Venture capital ownership Table 2 List of ownership dummies used in the Talouselämä 500 data set

40 Control variables Size A variable to control for size is needed as companies with different size characteristics can be assumed to have different risk preferences. A small and growing company might be more prone to take on debt, as they don t have enough retained earnings to finance their operations. Larger firms, on the other hand, commonly strive for less growth and want to attain a more conservative capital structure, as well as have more retained earnings. I will use the logarithm of sales as a proxy for size for two reasons, firstly, previous research (Coles et al 2006) have also used this variable and secondly, the data given from Talouselämä does not have any other variables that could be used as a proxy for size. As earlier said, smaller companies often want to take on more risk as they strive for greater growth compared with larger companies, therefore I assume a negative relationship between size and risk Growth A control variable for growth is useful as companies that strive for growth commonly take on more risk, in order to facilitate high growth. As classical risk theory suggests, higher returns and growth can usually only be implemented by taking on significant risk. I will use the logarithm on sales growth as a proxy for firm growth. This is also in line with previous research (John et al. 2008). I assume that growth has a positive correlation with risk, meaning that firms with more growth also take on more risk Profitability The relationship between profitability and risk is somewhat unclear, as discussed in the theory chapter (3.1). The classical pecking order theory suggests that a company with higher profitability, which in turn increases retained earnings, will have less debt and therefore less risk. The trade-off theory, on the other hand, suggests that a company should maximize its tax shield by taking on more debt, which would be reflected in higher risk. It is therefore difficult to predict which sign the variable should have. I have used return on investments (ROI) as a proxy for profitability, as it was the most suitable variable for this based on the data received from Talouselämä.

41 Industry dummy Companies in different industries have different characteristics that might make a bias in the regressions if they are not taken into account. One example is the financial leverage of a firm, which can vary vastly between industries. I have therefore included industry dummies based on Kauppalehti s industry classifications Summary on variable definitions and sources Below is a table that describes both the primary and secondary sources for information on the Talouselämä 500 data. The ownership variable required the most attention in the data gathering. Variable Primary source Secondary source Ownership Talouseläm ä Annual reports, Kauppalehti, Solidium Gearing Talouseläm ä - Equity ratio Talouseläm ä - Net sales Talouseläm ä - Sales growth Talouseläm ä - ROI Talouseläm ä - Industry dummy Talouseläm ä Kauppalehti Table 3 Variable sources for Talouselämä 500 data

42 35 The variables used in the Talouselämä 500 data are defined as follows. Variable Ln Equity ratio Ln Gearing Municipality Publicly traded shares Cooperative Venture capital Family Foreign State Employee Subsidiary Ln Sales Ln Sales Growth ROI Definition Natural logarithm of: equity / total assets Natural logarithm of: (interest bearing debt - cash and equiv alents) /equity Receiv es v alue of 1 if a municipality is a major owner Receiv es v alue of 1 if the shares are publicly traded Receiv es v alue of 1 if a cooperativ e is a major owner Receiv es v alue of 1 if a v enture capital firm is a major owner Receiv es v alue of 1 if a family is a major owner Receiv es v alue of 1 if a foreign inv estor is a major owner Receiv es v alue of 1 if the state is a major owner Receiv es v alue of 1 if it is owned by its employ ees Receiv es v alue of 1 if it is a subsidiary owned 50/50 by two firms Natural logarithm on firm sales Natural logarithm on firms sales growth (Total profit + Interest expense) / Inv ested capital Table 4 Definitions on variables used in Talouselämä 500 data Descriptive data The original data included firm-year observations. The one extra company stems from an additional company that was added to the year of 2010 by Talouselämä in year All financial firms were removed from the data (170), as these firms have a very different capital structure compared with other firms, which would distort the results. I also removed observations that did not have any values for gearing or equity ratio (1 005). It is mostly non-listed companies whose equity and gearing values are missing and it is unfortunate that this removes a third of all the observations from the sample. The final set of data consists of firm year observations. The following variables were winsorized at the 1% percentile at both the higher and lower end of the scale: Gearing, Equity ratio and ROI. This should improve the statistical significance of the results as extreme values have now been removed from the sample. The gearing ratio had values below -1 and hence I have used the formula ln(2+gearing), in order to use all off the values and not having to omit any variables in the regressions. The table below shows the descriptive statistics on the data set that is used in the regressions.

43 36 Variable Min Max Mean Median Std. Dev. Gearing -1,3 6 6,53 0,59 0,29 1,23 Equity ratio 0,06 0,89 0,46 0,44 0,20 Municipalty 0 1 0,06 0 0,24 Sharespublic 0 1 0,22 0 0,41 Cooperative 0 1 0, ,3 2 Venturecapital 0 1 0,04 0 0,21 Family Ownership 0 1 0, ,46 Foreign Ownership 0 1 0,29 0 0,45 State 0 1 0,09 0 0,29 Employee 0 1 0,01 0 0,09 Sales (EURmm) 7 7, ,00 857, , ,00 Sales Growth -41 % 7 8 % 5 % 4 % 1 8 % ROI -27 % 1 05 % 1 5 % 1 0 % 1 7 % Table 5 Descriptive statistics on data used in the Talouselämä 500 sample The table above gives arise to some interesting facts about the data. The descriptive statistics show that 8 out of the 13 variables consist of dummies that are given to companies according to their ownership structure. Family ownership as well as foreign ownership are the largest ownership types in the sample. At the other end, employee, venture capital and municipality owned companies are least represented in the sample. Gearing has a relatively high spread between the minimum and maximum values, although the variable has been winsorized. This implies that there are very large differences in capital structures between companies. The same logic also applies for the equity ratio where the lowest value is close to zero, whereas the highest value is close to 100%. The sales figures show that the 500 largest companies in Finland in general are relatively small as the median revenue is only EUR 227mm and lowest value is only EUR 78mm. The large spread can also be seen from the rather large standard deviation. There is also a rather large gap in sales growth, as the mean is 5%, whereas standard deviation is 18%. Further, return on investments (ROI) also has high extreme values, although winsorized, as the largest ROI is 105%. Gearing, Equity ratio, Sales, Sales growth and ROI variables are taken as the logarithm for the regressions, which should make the data more normally distributed.

44 37 5 % 4 %1 % Family Ownership 8 % 27 % Foreign Ownership Sharespublic 10 % Cooperative 19 % 26 % State Municipalty Venturecapital Employee Figure 1 Distribution of ownership dummies The data consists of a total of dummies, meaning that each company has on average 1.15 dummies. The graph above clarifies the distribution of ownership dummies that were assigned to the companies in the sample. The largest owner type is families (private individuals. The second largest group consists of companies or subsidiaries of large international corporations. The graph also shows that only 19% of the largest companies in Finland are listed on the stock exchange, which is somewhat surprising. The listed companies get a lot of attention in the media, most likely because their information is public, however more attention could be given to privately held companies as they represent over 80% of the largest companies in Finland. Cooperatives are also an important form of owners in Finland, as some 10% of companies in the sample have this ownership structure. The rest of the companies are owned by the state (8%), municipalities (5%), venture capital (4%) and employees (1%). 12 % 18 % 8 % 4 %1 % 1 % 1 % Industrial Products & Services Industry 19 % 36 % Consumer Service Consumer Goods Energy Technology Telecommunication Oil & Gas Healthcare Figure 2 Data distribution based on industry type

45 38 The figure above gives some light on the various industries that the companies are present in. The largest segment consists of Industrial products & services followed by general Industry and Consumer service. The smallest segments consist of Oil & gas and Healthcare firms. 6.2 Data on listed companies This data set is based on listed Finnish firms. Listed companies have considerably better reporting regarding ownership structure compared with privately held companies, which make these companies easier to analyse. The data is restricted to eight years and consists of years The restriction is made, as ownership data before 2008 is more difficult to access from databases due to regulatory reasons. Before 2008 Finnish companies did not have the same obligations to report ownership structures and hence the data starts from year Some Finnish investors have foreign investment companies that invest in Finnish firms. I have taken this into account when going through the material and changed the nationality of foreign investment firms held by Finnish investors. Thus, companies with a foreign ownership indicator variable, does not include these Finnish holding companies. The sample only consists of companies that have been listed during the whole time period, as this makes data handling considerably easier. Financial firms have been excluded from the sample. Appendix 1 shows the entire list of 92 companies that were used in the regressions, whereas the table in appendix 2 shows companies that have been excluded as well as the reason for their exclusion from the data set Risk variables As earlier mentioned, there is no right way to measure risk and it is difficult to define an exact definition for risk. I have therefore used multiple variables for risk as I hope that this will yield more robust results Ln Volatility This variable measures the volatility of the stock price of a company and is defined as the logarithm of volatility on weekly returns. The volatility is calculated from prices

46 39 spanning for one year. Previous studies (Anderson & Reeb 2008) have used this measure to describe risk however, previous research has commonly used daily data to calculate the returns. This is not wise on Finnish data, as the Helsinki Stock Exchange includes many small cap companies that have very thin trading. Therefore, I regard weekly returns as a more appropriate interval for calculating returns. Higher volatility on stock returns implies that there is more uncertainty regarding the firm s future success. This can in other words be regarded as a definition of risk. One can expect that a company with stable share price development has less uncertainty regarding its future development. When talking about the share price, one has to bear in mind that it does not only take into account the company s performance and prediction on its future outcome. Instead, it also consists of the general state of the economy and its movements. As the data at hand includes years , it is obvious that general movements in the economy has had a large effect on stock price volatility and should be kept in mind when analysing results from this variable. The logarithm of weekly returns would probably be better suited for time periods with more stable movements in the economy, in order to capture the firm specific effects. Therefore, I do not expect significant results for this variable Ln Gearing The gearing variable was already discussed in the previous chapter and will be included for the same reasons Ln Equity ratio The gearing variable was already discussed in the previous chapter and will be included for the same reasons Standard deviation on ROA The standard deviation on ROA measures the volatility on returns that a share has in relationship to its average assets. It therefore describes how effectively a company can produce returns in relationship to its assets. This measure of risk has been used by previous research (John et al. 2008) and can also be seen as a measure of profitability.

47 40 This makes ROA especially good for comparing firms within the same industry. I have included industry dummies to take into account the difference of capital structures of firms and the difference in capital structure between industries should be taken into account in the results. The standard deviation on ROA can also be seen as a measure of how stable the returns are. This can be analysed both as a risk measure, but I see it also as an indicator on how stable returns the firm can achieve. It however, does not give an indication on the level of returns that the company is producing. Return on assets can be calculated for various periods of time, such as quarterly, halfyear or yearly basis. I have decided to use it yearly data, due to data restriction for Finnish firms. The variable is counted as the logarithm on standard deviation on ROA for three years. The standard deviation has been calculated for three years; for the year preceding the observation (t-1), the year in question (t) and the year following the observation (t+1). This is done for all years, except for 2014, which only includes two years, as financial data was not available for firms for year 2015 at the time of data collection Ownership variables Gathering the ownership material has been done thoroughly for all companies. The raw data on ownership structure has been taken from Thomson One Banker, which then has been gone through manually. The 20 largest owners for each company were analysed closer, in order to find the ultimate owners of these companies. Ownership was gathered manually for all firms with dual class shares, as this information was not available from Thomson One Banker. I added together all the voting rights of members of the same family, as this gives a better view of the family s voting rights in total (one can assume that members of the same family vote together), once the ultimate owners were identified. Families became the largest owners in many cases once all the shares were calculated together. Especially families such as Ehrnrooth, Herlin and Erkko commonly have shareholdings through multiple holding companies, which eventually made them the largest owners, although the individual family members did not have a controlling share ownership. The consolidated ownership data included the three largest owners, which enables me to have a look at the characteristics of the largest owners as well as the sum of the three

48 41 largest owners. I have used multiple variables to describe ownership and they will be discussed below Ownership concentration The ownership concentration in a firm is a proxy for how much control an investor has in a specific company. When ownership concentration is higher, it means that the investor has more voting power in the company and thus has more possibilities to impact corporate risk taking. I have used five different proxies for ownership control; (1) the voting percentage of the largest owner (2) the voting rights of the three largest owners (3) an indicator that receives the number one if the largest owner has voting rights in excess of 10%, (4) an indicator that receives the number one if the largest owner has voting rights in excess of 15% and (5) an indicator that receives the number one if the largest owner has voting rights in excess of 20%. The use of dummies to represent different thresholds of ownership is used in most previous research as it gives a good representation on the different levels of ownership and how they affect risk taking. Further, the five variables described above are mutually exclusive, meaning that they will not be used in the same regressions, as they aim at describing the same phenomena, and have high collinearity Investor type Another way of representing the ownership structure is by using dummies for different ownership types. The owners have been divided into groups according to investor types. The variable has been determined by the information received from Thomson One banker however, it has been slightly modified, in order to decrease the number of ownership types, as some investor types had limited number of observations. Table 7 shows on the left hand side the original investor types that where received from Thomson One Banker, whereas the right hand side shows the investor type groups used in the regressions.

49 42 Table 6 Investor types in the listed sample Control variables Next I will present all the control variables that I used in the data set for listed firms as well as how they have been collected CEO ownership CEO ownership is a logical variable to describe ownership as a CEO has considerable authority to impact the risk taking of a firm. When the CEO buys shares in the company, he automatically becomes an insider, who hopefully will think about risk taking from a shareholders perspective. Thus, this could alter the risk taking of the firm. CEO ownership has been collected manually from company annual reports, which has been very time consuming. Insider ownership is also commonly used in previous research. The variable is presented both in total market value and as a percentage of total outstanding shares. The variable can have a significant impact on risk taking as high managerial ownership might reduce the mangers risk taking as more of his total wealth

50 43 is tied to the firm. On the other hand, smaller ownership values might increase risk taking. Due to this trade-off it is difficult to predict the sign for this variable Majority ownership Majority ownership is an indicator variable that takes the value of one when the largest owner controls more than 50% of the shares in a company. The estimated sign for majority ownership is difficult to assess, but assuming that the investor owning more than 50% of a company might not have a diversified portfolio, I therefore predict a negative relationship between majority ownership and risk taking Dual class share Dual class share is an indicator variable that takes the value of one when the company has two series of shares. This variable is included as the voting rights of the largest owners seem to be considerably larger compared with companies with only one series of shares. Regarding the estimated sign, I predict a negative relationship between dual class shares and risk taking, based on the same argument as for majority ownership; one can assume that the person also has a large portion of his wealth tied to the company and therefore also an undiversified portfolio Foreign investor Foreign investor is an indicator variable that takes the value of one when the company has a foreign investor as the largest investor, or among the three largest investors when using the ownership concentration variable describing the three largest owners. Foreign investors usually have a diversified portfolio, otherwise they would not invest abroad, and therefore I predict that the relationship between foreign investors and risk taking is positive Size As discussed in the data chapter on the Talouselämä data, the regression needs a proxy variable for size, as companies of different sizes have different risk preferences. I have used the logarithm on annual sales as the proxy for a firm s size, which is in line with some previous research. As described earlier, I predict a negative relationship between the size of a company and its risk taking.

51 Growth This variable was also discussed previously for the Talouselämä data. Companies with high growth can commonly be characterized by higher risk and therefore this control variable is relevant for this study. I use annual sales growth as a proxy variable for growth. As described earlier, I predict a positive relationship between risk and growth Profitability Profitability was as well discussed for the Talouselämä. A firm does not commonly receive high profitability figures, without taking on considerable risk. This is particularly the case for smaller, less developed companies. I have used return on average assets as a proxy variable for risk. As described earlier, it is challenging to estimate the relationship between risk and profitability Summary on variable sources and definitions This chapter illustrates the sources for data on variables as well as the exact definitions of these variables. The first table is simplified and only includes the main category for all variables, instead of all the individual variables. Most of the data was extracted from data bases, however, the ownership data needed a large amount of manual work. Variable Primary source Secondary source Risk variables Factset - All ownership variables Thom son One Annual reports Control variables Factset - Industry dummies Factset US Departm ent of Labour Table 7 Summary on primary and secondary sources for variables The table below lists all the variables used in the data set as well as the definition for each variable. All variables were not used in the same regression and this will be explained in further detail in the methodology chapter.

52 45 Table 8 Summary on variable descriptions