Investigating the Relevance of the Pecking Order Theory and the Agency Costs of Free Cash Flow Theory in the Eyes of Investors

Transcription

1 Investigating the Relevance of the Pecking Order Theory and the Agency Costs of Free Cash Flow Theory in the Eyes of Investors A longitudinal study on firms traded at NYSE Euronext Amsterdam Master Thesis Delft University of Technology Faculty of Technology Policy and Management August 2011 Author Wouter Goossens Management of Technology Graduation Committee Prof.dr. Alfred Kleinknecht - Chairman Dr. Zenlin Kwee - First supervisor Dr. Laurens Rook - Second supervisor Department of Innovation Systems

2 Table of contents Acknowledgement... 3 Summary Introduction Practical research background Theoretical research background Research aim, questions and approach Scientific and managerial relevance of this research Outline of the thesis Literature review Conceptual design and hypotheses development Conceptual model Definition and clarification of the variables Operationalization of the variables as proposed in the literature Operationalization of the variables as performed in this research Hypotheses development Methodology Empirical research design Empirical models Data collection and data sources Case selection approach Detecting and removing outliers Dealing with influential cases Descriptive statistics Sample description Aggregated descriptive statistics of the predictors Determining the market value of cash holdings Results Multivariate tests Sensitivity tests Limitations Limitations concerning the conceptual design Limitations concerning the methodology Limitations concerning the regression results Conclusion and recommendations References Appendices

3 Acknowledgement This thesis is the result of six months of successful collaboration between my graduation committee and me. I would like to thank my supervisors Prof. dr. Alfred Kleinknecht, Dr. Zenlin Kwee and Dr. Laurens Rook for guiding my through this project. A special thanks goes out to my first supervisor, Dr. Zenlin Kwee. I m very grateful to her for giving me the opportunity to work on this fascinating research topic. Moreover, I m very thankful to her for all the powerful feedback I got at all our meetings. Further, I would like to thank my parents, Erik and Elly, my sister Carlijn and her love Philippe for their unconditional love and support. I am grateful to be part of such a lovely family. Finally, I would like to thank my love Regina. Her love, support and happiness are unconditional and I feel very blissful being together with her. Wouter Goossens Delft, June

4 Summary Nowadays companies hold increased amounts of cash (Dittmar 2008; Schauten, van Dijk and Van der Waal 2008). Literature gives two contradictory theories that may influence the valuation of these cash holdings by investors in the market. The pecking order theory predicts that managers have more information about the state of the firm than outside investors and therefore outside investors require a premium on the money they invest in the firm (Myers and Majluf 1984). This premium is referred to as information costs because it reflects the additional amount of money firms are charged for by investors to compensate for the information gap that exists between investors and the managers of the firm. Contrary to the pecking order theory, the agency cost of free cash flow theory predicts that managers do not always have the intention to maximize shareholder value; they rather use the company s cash to maximize their own utility (Jensen 1986). The costs that arise from agency problems are referred to as agency costs. Although the pecking order theory itself does not predict how the existence of information asymmetry may affect the market value of cash holdings, it can be argued that because firms have to pay a premium in case of external financing, investors will appraise the capital that is internally available to the firm. Therefore, information costs are likely to have a positive influence on the market value of cash holdings. On the other hand, based on the agency costs of free cash flow theory, it can be argued that when investors believe that a firm experiences agency problems they are likely to have a negative perception on the market value of cash holdings. This study aims to investigate the relevance of these contradicting theories in the eyes of the investor by analyzing the impact of information costs and agency costs on the market value of firms cash holdings. In an attempt to answer this issue, a longitudinal quantitative study is performed on a sample of firms traded at NYSE Amsterdam Euronext. We performed an extensive literature review to give a clear and logical presentation of the relevant research work done thus far. Based on literature findings we developed a conceptual model. The conceptual model showed the predicted relationships among variables. The operationalization of the variables was established after careful evaluation of suggestions given in the existing literature. Based on the predicted relationships in the conceptual model we developed seven hypotheses which were statistically tested by the performance of various cross sectional multiple hierarchical linear regression analysis. The central research question guiding this research is: By performing an empirical analysis on the market value of cash holdings of firms, do investors believe the pecking order theory or the agency costs of free cash flow theory to be the most relevant to firms? Driven by the challenge to conceptualize and operationalize the key principles the following sub questions were derived: (1) What is the market value of a company s cash holdings? (2) Which firm-specific characteristics influence the market value of cash holdings and how can we control for this? (3) Do industry characteristics affect the relevance of the pecking order theory and the agency costs of free cash flow theory and how can we control for this? (4) Do macroeconomic effects affect the market value of cash holdings and how can we control for this? (5) To what extent do information costs and agency costs influence the market value of cash holdings? Chapter 2 provides the outcome of an extensive literature study. First an overview of prior research on the pecking order theory is provided. Second, we provide an overview of prior research on the agency costs of free cash flow theory. Third, an overview of prior research on factors that influence the market value of cash holdings is given. Fourth, we provide an overview on existing empirical models that can 4

5 be used to calculate the market value of cash holdings. Finally, in the concluding section we summarize the most important literature findings and argue how our research contributes to the existing literature. In Chapter 3 we develop a conceptual model. As follows from the main research question, the primary variable of interest is market value of cash holdings and therefore, in our conceptual model, this variable is adopted as dependent variable. The literature study showed that the variables: financial constraints, growth opportunities, book value of cash holdings, corporate governance, information costs and agency costs may have an influence on the market value of cash holdings. Since this research aims to analyze the effect of information costs and agency costs on the market value of cash holdings the variables information costs and agency costs are adopted as independent variables. The variables financial constraints, growth opportunities, total book value of cash holdings, and corporate governance are adopted as covariate variables. The operationalization of the main variables is established after careful evaluation of all the suggestions given in the existing literature. The covariate variable corporate governance could not be operationalized because we were not able to get the required data on its indicators. All variables were corrected for firm size because we believe that the relationship between the dependent variable and each of the predictors may be affected by the size of the company. Moreover, as argued by Fama and French, correcting for firm size avoids that the regression results will be dominated by the larger firms. We used net assets ( ) as a proxy for firm size. At the end of this chapter, based on the predicted relationships in the conceptual model, seven hypotheses are developed. Chapter 4 discusses and outlines the research methodology. To access articles and publications from top journals in the fields of finance, economics and business management we used academic search engines such as Wiley Interscience, Google Scholar and ScienceDirect. To gather the required financial data we used the financial databases Thomson One Banker, Worldscope and I/B/E/S. The financial numbers extracted from these financial databases were converted into real values in 2009 using the Consumer Price Index (CPI). Subsequently, in this chapter we present the three empirical models that we used to test our hypotheses. We decided to test our hypothesis on three different empirical models to ensure the robustness of our research findings. Additionally, we point out how we selected testable cases. Our testable sample contained 625 observations. These observations represent firms across different industries over the period. The distribution of firms across industries in the testable sample was almost similar to the distribution in the original data set. Hence, we trust that the testable sample is a homogeneous representation of the original data set. At the end of this chapter we discuss how we dealt with outliers and influential cases. As discussed in the book of Field (2009) outliers and influential cases are extreme cases that have to be removed from the sample because they bias the regression output. To detect and remove outliers and influential cases we followed the approach described in the book of Field. After we removed all outliers and influential cases about 80% of the testable sample was left. Chapter 5 provides the empirical setting of this research. It starts with the presentation of some aggregated descriptive statistics on characteristics of the firms included in our sample. Subsequently, we present the aggregated descriptive statistics of all predictors included in the regression analysis. Finally, in this chapter we present the results of a regression analysis that was performed to determine the market value of cash holdings. We found that on average investors place a value of euro on every single euro of cash that a company has on its balance sheet. In other words, the average market value of a euro of cash holdings is euro. 5

6 Chapter 6 provides the results of the empirical tests that are performed. The chapter first presents the results of the multivariate tests which relate to the first four hypotheses. Tests with three empirical models consistently showed that information costs have a significant positive influence on the market value of cash holdings. The empirical models gave mixed results with respect to the influence of agency costs on the market value of cash holdings. Overall, we found that information costs have a larger impact on the market value of cash holdings than agency costs. After presenting the results of the multivariate tests the chapter continues with providing the results of the sensitivity tests. These tests relate to hypotheses 5 6 and 7. We found that industry characteristics and macroeconomic circumstances do not affect the relationship between the market value of cash holdings and information costs or agency costs. Finally, a number of experiments were performed to assess the influence of information costs as a moderating variable. We found that besides the direct influence of information costs on the market value of cash holdings this variable also acts as a moderating variable in the relationship between agency costs and the market value of cash holdings. Chapter 7 focuses on the limitations of this research. First, we present the limitations of our conceptual model are discussed. Second, we present the limitations of the research methodology that was used in this research. Finally, we discuss the limitations of our empirical results. Chapter 8 summarizes the key findings of this research and concludes on our research contributions. Moreover, in this chapter we present a number of suggestions for future research. Regression analysis performed with three different empirical models consistently showed that information costs make a significant positive influence on the market value of cash holdings. Tests on the influence of agency costs on the market value of cash holdings gave mixed results. For those tests at which agency costs made a significant contribution to the market value of cash holdings its impact appeared to be lower than the impact made by information costs. Based on the fact that we found strong evidence for the influence of information costs while the evidence for the influence of agency costs is rather weak and its impact appeared to be lower than that of information costs we believe that information costs is more relevant as a predictor of the market value of cash holdings. Since information costs relate to the pecking order theory we conclude that investors believe the pecking order theory to be more relevant to firms than the agency costs of free cash flow theory. This conclusion is valid for non-financial firms under the assumption that a firm s actual level of cash holdings has no influence on the impact of agency problems and information asymmetry. Additional empirical tests indicated that the classification of firms into growth industries and mature industries and the addition of dummy variables to represent industry sector and macroeconomic circumstances all have no influence on the conclusion that information costs is more relevant as a predictor of the market value of cash holdings than agency costs. However, tests with the inclusion of information costs as a moderating variable showed that besides the direct influence of information costs on the market value of cash holdings it also acts as a moderating variable in the relationship between agency costs and the market value of cash holdings. We found that in case information costs are present, the interaction term of agency costs and information costs makes a significant negative contribution to the market value of cash holdings. Although the statistical analysis performed thus far tells us in what direction the interaction term relates to the dependent variable the exact underlying structural mathematical equation linking the three variables is still unknown. We recommend future researchers to take a closer look at this. Moreover, we suggest that it would be interesting to repeat this analysis on a larger sample and over a longer time period and to experiment with different mathematical configurations of the predicted relationships. 6

7 1. Introduction 1.1 Practical research background Companies show a tendency of holding increased amounts of cash The global economy shows an interesting practical trend that caught our attention. First of all, for years, public companies show a tendency of holding increased amounts of cash. The average cash holdings of U.S. public firms, as a percentage of total assets, have increased from 10% to 23% from 1960 to 2008 (Dittmar 2008). For European public firms the average increased from 13% to 17% from 2000 to 2008 (Schauten, van Dijk and Van der Waal 2008). This trend raises the question why firms are holding substantial amounts of cash. Bates, Kahle, and Stulz studied this phenomena and found that companies hold more cash because their operating cash flows have become riskier, their inventories and accounts receivable have fallen, and their R&D expenditures have increased (Bates, Kahle and Stulz 2009). Market imperfections and the market value of cash holdings From what is described above, it can be concluded that nowadays companies hold increased amounts of cash. Since cash holdings have a substantial influence on the overall market value of a firm (Fama and French 1998; Pinkowitz, Stulz and Williamson 2006; Pinkowitz and Williamson 2003), it is important to develop a clear understanding of the factors that influence the market value of companies cash holdings. In a perfect market, every euro of cash should be valued as one euro by the market. In real world, however, phenomena such as taxes, information asymmetry and moral hazard make markets imperfect and this may cause the market value of cash holdings to differ from its book value (Nyborg 2010). 1.2 Theoretical research background The pecking order theory Literature gives two contradictory theories that may influence the valuation of cash holdings by the market. Myers and Majluf (1984) developed the so called pecking order theory. This theory predicts that managers have more information about the state of the firm than outside investors and therefore outside investors require a premium on the money they invest in the firm. Hence, the premium charged by equity investors will be higher than the premium charged by debt investors because the investment done by debtors is less sensitive to fluctuations in the value of the firm (debtors receive fixed amounts of interest payments at pre-specified moments in time). The size of the premium that investors charge to firms depends on the level of information asymmetry between managers of the firm and investors. The less investors know about the exact state of the firm, the higher the level of information asymmetry and therefore the higher the premium. The premium charged by investors thus reflects costs that arise from the existence of information asymmetry and therefore these costs are called information costs. The pecking order theory does not directly predict how the existence of information costs influences the market value of cash holdings. However, from this theory, it can be derived that the existence of information cost makes it expensive for a firm to raise capital externally causing investors to appraise the capital that is internally available to the firm. 7

8 The agency costs of free cash flow theory Contrary to the pecking order theory, Jensen (1986) developed the agency costs of free cash flow theory. This theory predicts that managers do not always have the intention to maximize shareholder value; they rather use the company s cash to maximize their own utility. Agency costs exist in three different varieties. First, managers have incentives to enjoy the full value of anything they get from the firm, such as private jets or country club memberships, since they bear only a fraction of these benefits (on the job consumption). Second, managers might strive for empire building since a larger firm gives managers greater prestige, power and compensation (empire building). Third, managers have incentives to govern the company according to their own preferences and make themselves indispensable by taking on projects that others cannot manage as well (entrenchment). In sum the agency costs of free cash flow theory states that the managers waste a company s cash holdings by making decisions that maximize their own utility instead of the utility of shareholders. From this theory it can be derived that when investors believe that a firm experiences agency problems they do not like it when that firm holds substantial amounts of cash. Industry characteristics and the relevance of both theories Now that both theories are clearly defined, it is important to realize that industry characteristics may have an influence on the relevance of both theories. Managers of firms in growth industries are likely to have better information regarding R&D efforts than investors and may not want to divulge sensitive information (Nyborg 2010). This makes it reasonable to believe that the costs from information asymmetry (i.e. information costs) are higher for firms in growth industries and therefore, the pecking order theory seems to be particularly relevant to firms in growth industries. On the other hand, agency costs are considered to be more important for firms in mature industries. Firms in mature industries tend to generate cash in excess over their investment needs. As a result in mature industries, there is typically more free cash flow available for management to play around with (Nyborg 2010). This makes it reasonable to believe that agency costs are higher for firms in mature industries and therefore, the agency costs of free cash flow theory seems to be particularly relevant to firms in mature industries. Macroeconomic effects and the market value of cash holdings Macroeconomic effects may affect the market value of cash holdings because the global balance between demand and supply of cash is likely to influence the ease with which firms can raise capital in the market. The overall status of the global or a national economy is often measured based on national income where an extreme decline of national income indicates economic downturn and an extreme growth of national income indicates economic growth (Kleinknecht 2009). Literature gives two important macroeconomic theories that aim to explain cyclical fluctuations in national income. Juglar developed a theory that explains fluctuations in national income by a model that considers three phases (respectively: prosperity, crises and liquidation) over a period of 7 to 10 years (Kleinknecht 2009). Alternatively, Schumpeter and Kondratieff develop a theory that explains fluctuations in national income based on technologically trajectories lasting 45 to 60 years (Kleinknecht 2009). 8

9 In times of economic downturn, the market is worried about the future which disturbs the flow of cash between firms, banks and other type of investors. In these times, capital becomes more difficult to obtain which makes it relatively expensive for firms to raise money in the market (Kleinknecht 2009). Hence, investors are likely to place a premium on firms cash holdings. On the other hand, in times of economic boom, there typically are excessive flows of capital which makes it relatively cheap for firms to raise money externally (Kleinknecht 2009). Therefore, in times of economic growth, investors are likely not to place any additional value on companies cash holdings. 1.3 Research aim, questions and approach Research aim This study aims to investigate whether investors believe the pecking order theory or the agency costs of free cash flow theory to be the most relevant to firms. Therefore, the central research question guiding this research will be: By performing an empirical analysis on the market value of cash holdings of firms, do investors believe the pecking order theory or the agency costs of free cash flow theory to be the most relevant to firms? Research questions In order to be able to answer the central research question this research is conducted in an attempt to answer the following research questions: RQ1: RQ2: RQ3: RQ4: RQ5: What is the market value of a company s cash holdings? Which firm-specific characteristics influence the market value of cash holdings and how can we control for this? Do industry characteristics affect the relevance of the pecking order theory and the agency costs of free cash flow theory and how can we control for this? Do macroeconomic effects affect the market value of cash holdings and how can we control for this? To what extent do information costs and agency costs influence the market value of cash holdings? Table gives an overview about how each of the research questions is addressed. The table also shows the type of research that is conducted for studying the research question as well as the short answer to the research question. 9

10 Table 1.3.1: Overview of the research questions addressed in this research Research question Nature Chapter Short answer RQ1: Determine market value of cash holdings RQ 2: Influence of firm-specific characteristics on the market value of cash holdings RQ3: Influence of industry characteristics on the relevance of both theories RQ4: Influence of macroeconomic effects on the market value of cash holdings RQ5: Influence of information costs and agency costs on market value of cash holdings Descriptive 5 On average, a euro of cash written in the firm s books is worth euro in the market Exploratory 2,3 Financial constraints Growth opportunities Corporate governance Financial policy Firm size Agency costs Information costs Hypothesis testing Hypothesis testing Hypothesis testing 6 Industry characteristics turned out to have no significant influence on the relevance of both theories 6 Macroeconomic effects turned out to have no significant influence on the relevance of both theories 6 Information costs turned out to have a significant positive impact on the market value of cash holdings. We did not find strong evidence for the influence of agency costs on the market value of cash holdings Research approach This study aims to investigate the relevance of the pecking order theory and the agency costs of free cash flow theory in the eyes of the investor by analyzing the impact of information costs and agency costs on the market value of firms cash holdings. In an attempt to answer this issue a longitudinal quantitative study was performed on a sample of firms traded at NYSE Amsterdam Euronext. We performed an extensive literature review to give a clear and logical presentation of the relevant research work done thus far. Based on literature findings we developed a conceptual model. The conceptual model showed the predicted relationships among variables. The operationalization of the variables was established after careful evaluation of suggestions given in the existing literature. Based on the predicted relationships in the conceptual model we developed seven hypotheses which were statistically tested by the performance of various cross sectional multiple hierarchical linear regression analysis. 1.4 Scientific and managerial relevance of this research Scientific relevance This research is relevant from a scientific point of view for three reasons. First of all, various researchers have previously analyzed the impact of information costs and agency costs on the market value of cash holdings separately or indirectly (Dittmar and Mahrt-Smith 2007; Drobetz, Grüninger and Hirschvogl 2010; Faulkender and Wang 2006; Pinkowitz, Stulz and Williamson 2006). However, to my notice, none of these studies has directly measured the influence of both in one empirical design. An empirical analysis that includes direct measurement of both types of costs allows comparing the impact of both. 10

11 Secondly, since the conceptual model used in this study is developed based on findings from various previous studies the model will incorporate key variables for which it is empirically proven to affect the market value of cash holdings. Finally, this research extends the existing literature on cash holdings because it uses a different set of data. While previous research mainly used U.S. public firms, this research focuses specifically on Dutch public firms. Although, the studies of Pinkowitz et al. (2006), Schauten et al. (2008) and Drobetz et al. (2010) use data on Dutch companies, they report research outcomes on a country level while this research reports results on a firm level. Managerial relevance This research is relevant from a managerial point of view for two reasons. First of all, it sheds a light on whether cash holdings are valued above or below book value and this provides management with insights on whether cash holdings do add real value to a firm. Secondly, this research provides an overview on factors that influence the market value of cash and estimates the extent by which these factors impact the market value of cash holdings. These insights might be beneficial to a company s management when deciding on the company s financial policy. Table summarizes the managerial and scientific relevance of this research and links these contributions to chapters out of this thesis. Table 1.4.1: Overview of factors that explain the relevance of this thesis Type of relevance Description of relevance Chapter Scientific relevance Novel approach where agency costs and information costs are incorporated into one conceptual model 3 Conceptualization and operationalization based and built upon existing literature 3 Unique data set with a specific focus on Dutch public firms with research outcomes reported on a firm level instead of a country level 5,6 Managerial contribution Insight into whether cash holdings do add value to a firm Insight into factors influencing the market value of cash holdings 1.5 Outline of the thesis The outline of this thesis is depicted in figure Part I gives an introduction to our research topic. In Chapter 1, we present the practical and theoretical research background related to our main research question. Subsequently, we present our main research question and derive a number of sub questions that helped to answer the central question. We also give a brief summary of the research approach of this project. Part II starts with Chapter 2 in which we summarize the key findings of our literature study. Based on these findings, in Chapter 3 we develop a conceptual framework and present the operationalization of the key concepts. Subsequently, we present the hypotheses derived from the predicted relationships in our conceptual model. Part III starts with Chapter 4 in which we outline and discuss the research methodology. This chapter elaborates on the research design and presents the three empirical models that we used. Subsequently,

12 we explain how we collected data and we present our data sources. This chapter continues with an explanation of the selection process that we applied to obtain a testable sample. Chapter 4 ends with an explanation of how we dealt with outliers and influential cases. Next, in Chapter 5 we present several descriptive statistics. We start this chapter by presenting some aggregated descriptive statistics on characteristics of the firms included in our sample. Subsequently, we present some descriptive statistics of the predictors that are incorporated into the empirical models. Finally, in this chapter we show what cash holdings are worth to investors. Part IV discusses the empirical results. In Chapter 6 we present the key findings from our empirical analysis. We conclude our study in Part V. This section starts with Chapter 7 in which we discuss the limitations of this research. Finally, in Chapter 8 we present the key results of our research and reflect on our central research question. Moreover, we present a number of suggestions for future research. Figure 1.5.1: Thesis outline 12

13 2. Literature review Empirical evidence for the pecking order theory One strand of literature provides evidence for the pecking order theory. Shyam-Sunder and Myers (1999) study the pecking order theory and conclude that the pecking order theory gives a good description of firms financing behavior. They find that firms that do not have cash at hand prefer debt financing over equity financing. However, in their paper they also mention that it can be questioned if their conclusion would hold in case the sample would consists entirely of growth firms. Faulkender and Wang (2006) provide limited evidence for the pecking order theory. They find that investors of all equity firms place a value of about $1.43 on companies cash holdings. They document that the reason for investors to assign a premium to cash holdings is because it avoids a company from paying information costs when raising capital in the market. However, they also find that the market value of cash holdings declines with larger cash holdings, higher leverage, better access to capital markets, and greater cash distributions via dividends rather than via share repurchases. Bharath, Pasquariello, and Wu (2009) test whether information asymmetry is an important determinant of capital structure decisions, as suggested by the pecking order theory. They find that firms that face high levels of information asymmetry issue forty to seventy percent more debt than firms that face low levels of information asymmetry. Moreover they find that firms prefer to issue equity when the level of information asymmetry is low. These findings provide evidence for the existence of an order of preferences to raise capital as predicted by the pecking order theory. Lemmon and Zender (2006) develop and test a modified version of the pecking order theory, which takes into account a firm s debt capacity. Lemmon and Zender argue that firms financing choice may depend on its debt capacity because they believe that some firms may save some of their debt capacity to fulfill future financing needs. They find that internal funds are the first choice of financing for all firms. Moreover they find that firms unconstrained by concerns over debt capacity primarily use debt to fill their financing deficit, while those with limited debt capacity rely heavily on external equity financing. Based on these findings they conclude that a modified version of the pecking order theory that takes into account a firm s debt capacity is a good descriptor of firms financing behavior. In contrast to all who found support for the pecking order theory, Frank and Goyal (2003), Fama and French (2005) and Leary and Roberts (2010) document that the pecking order theory is not able to explain firms financing behavior. Their findings show that the pecking order theory does not survive when a broader sample of firms or a longer time series is used. Table summarizes the prior research on the pecking order theory. 13

14 Table 2.2.1: Prior research on the pecking order theory Author Focus Sample Key finding Citation index Bharath, Pasquariello, and Wu (2009) Fama and French (2005) Faulkender and Wang (2006) Frank and Goyal (2003) Leary and Roberts (2010) Lemmon and Zender (2006) Myers and Majluf (1984) Shyam-Sunder and Myers (1999) The influence of information asymmetry on financing decisions Empirical relevance of the pecking order theory Corporate financial policy and the market value of cash holdings Empirical relevance of the pecking order theory Empirical relevance of the pecking order theory using various empirical models Empirical relevance of a modified version of the pecking order theory Determinants of firms capital structure decisions Empirical relevance of the pecking order theory U.S. public firms over the period 1973 to 2002 Firms listed on the NYSE, AMEX, and Nasdaq over the period 1973 to 2002 U.S. public firms over the period 1971 to 2001 U.S. public firms over the period 1971 to 1998 U.S. public firms over the period 1980 to 2005 U.S. public firms over the period 1971 to 2001 No empirical analysis U.S. public firms over the period 1971 to 1989 Firms prefer to issue debt when the level of information asymmetry is high and they prefer to issue equity when the level of information asymmetry is low More than half of the sample firms violate the pecking order theory Investors assign a premium to cash holdings because it avoids a company from paying information costs when raising capital in the market Large firms generally follow the pecking order theory while small firms especially those analyzed in later years do violate the pecking order theory The empirical relevance of the pecking order depends crucially on one s interpretation of the hypothesis and the empirical model used A modified version of the pecking order theory, which takes into a firm s debt capacity, is a good descriptor of firms financing behavior Firms prefer to raise capital in the following order: internal, debt, equity The pecking order theory is an effective first-order descriptor of corporate financing behavior Empirical evidence for the agency costs of free cash flow theory Another strand of literature provides evidence for the agency costs of free cash flow theory. D Mello and Miranda (2010) investigate the role of long-term debt in influencing overinvestments by analyzing the pattern of abnormal capital expenditures around new debt offerings by unlevered firms. They find that issuing debt leads to a reduction in abnormal capital expenditures. Moreover, they find that the reduction in overinvestments has a positive impact on equity value. D Mello and Miranda argue that the decline in overinvestments can be explained by their belief that debt service obligations reduce the number of discretionary funds under managerial control. Nohel and Tarhan (1998) study the impact of share repurchases using excess amounts of cash on operating performance. They find that operating performance improves after share repurchases and argue that this improvement is due to more efficient employment of assets. Nohel and Tarhan believe that more efficient employment of assets signals a decline in agency costs of free cash flow problems

15 Pinkowitz, Stultz and Williamson (2006) examine the impact of various corporate governance regimes across different countries on the value of cash holdings and dividends. Consistent with the presence of agency problems, they find that investors in countries with poor governance scores place a lower value on a dollar of corporate cash holdings than investors in countries with above average governance scores. Similarly, Dittmar and Mahrt-Smith (2007) find that investors place a lower value on the cash holdings of poorly governed firms. They argue that these findings convey investors expectation of firm value to be lost from the waste of cash reserves. Drobetz, Grüninger and Hirschvogl (2010) study the influence of information asymmetry on the market value of cash holdings. They find that the market value of cash decreases significantly when a firm faces higher levels of information asymmetry. Based on this finding Drobetz et al. conclude that agency costs have a greater impact on the market value of cash holdings than information costs. Table summarizes the prior research on the agency costs of free cash flow theory. 15

16 Table 2.2.2: Prior research on the agency costs of free cash flow theory Author Focus Sample Key finding Citation index D Mello and Miranda (2010) Dittmar and Mahrt-Smith (2007) Drobetz, Grüninger and Hirschvogl (2010) Jensen (1986) Nohel and Tarhan (1998) Pinkowitz, Stultz and Williamson (2006) The influence of debt issuing on overinvestments and agency costs The influence of corporate governance on the market value of cash holdings The influence of information asymmetry on the market value of cash holdings The influence of agency costs on financing decisions The influence of share repurchases on operating performance and agency costs The influence of corporate governance on the market value of cash holdings U.S. public firms that have started to issue debt between 1968 and 2001 and that have been unlevered at least for three years immediately preceding the offering U.S. public firms over the period 1990 to 2003 Public firms over the period 1995 to 2005 No empirical analysis All tender offer stock repurchases announced between 1978 and 1991 listed in the appendix of Comment and Jarrell (1991), and the Wall Street Journal Index. Public firms over the period 1983 to 1998 Issuing debt leads to a reduction in abnormal capital expenditures which can be explained by debt service obligations that reduce discretionary funds under managerial control Investors place a lower value on the cash holdings of firms that are poorly governed can be explained as a convey of investors expectation of firm value to be lost from the waste of cash reserves The market value of cash significantly reduces when a firm faces a higher level of information asymmetry, therefore agency costs have higher impact on the market value of cash holdings than information costs Conflicts of interest between shareholders and managers are especially severe when a company holds substantial amounts of free cash flow Operating performance improves after share repurchases. This improved performance does not result from higher growth opportunities but from the more efficient employment of assets, supporting the agency costs of free cash flow theory Investors in countries with poor governance scores place a lower value on a dollar of corporate cash holdings than investors in countries with above average governance scores Factors that influence the level of information asymmetry There also exists a strand of literature that focuses on factors that influence the level of information asymmetry. Diamond and Verrecchia (1991) develop a theoretical model to predict how the level of information asymmetry influences firms cost of capital. Their model predicts that the level of information asymmetry of large firms is lower than that of small firms. Diamond and Verrecchia argue that the reason for large firms to experience less information asymmetry is because these firms are typically more mature and thus have a longer track record. Moreover, they argue that the level of information asymmetry of large firms is lower because large firms typically get more public attention. Empirical evidence for the idea that the level of information asymmetry depends on firm size is

17 delivered by Brennan and Hughes (1991). Brennan and Hughes investigate whether the flow of information about firms is an increasing function of firm size. They find that there is an increased level of information out in the market for larger firms. Aboody and Lev (2000) study whether the level of information asymmetry is related to R&D expenditures. They find that R&D expenditure is positively related to the level of information asymmetry. Aboody and Lev argue that R&D projects are projects that are unique to the particular firm whereas most other projects often share common characteristics across firms or industry. The uniqueness of R&D projects as well as the closed environment in which R&D typically occurs makes it difficult for investors to derive information about the value of a firm s R&D activities. Aboody and Lev also find that the influence of R&D on information asymmetry is especially relevant for firms with great growth opportunities. They argue that this type of firm typically has lots of R&D activities going on and therefore the level of information asymmetry is likely to be higher. Brennan and Subrahmanyam (1995) examine whether the level of information asymmetry is related to the number of financial analysts following the company. They find that a larger number of financial analysts (increased financial analyst coverage) following a firm reduces information asymmetry and consequently the level of information costs. Brennan and Subrahmanyam argue that an increase in financial analyst coverage intensifies the competition among financial analysts which in turn contributes to the quantity and quality of publicly available information about firms. Table summarizes the prior research on factors that influence the level of information asymmetry. Table 2.2.3: Prior research on factors that influence the level of information asymmetry Author Focus Sample Key finding Citation index Aboody and Lev (2000) Brennan and Hughes (1991) Brennan and Subrahmanyam (1995) Diamond and Verrecchia (1991) The relation between R&D, information asymmetry and insider gains The influence of stock price and stock splits on the availability of public information The relation between the number of financial analysts following a firm and the amount of information costs The relation between information disclosure, liquidity and cost of capital All sale transactions made by insiders and reported to the SEC over the period 1985 to 1997 Firms listed on the NYSE, AMEX or NASDAQ index over the period 1976 to 1986 Firms listed on the NYSE index in 1988 No empirical analysis R&D expenditure is positively related to the level of information asymmetry The number of analysts following a firm is inversely related to its share price Increased financial analyst coverage reduces information asymmetry and consequently information costs Reducing information asymmetry reduces a firm s cost of capital and this relationship is especially relevant for large firms Factors that influence the level of agency costs Another strand of literature focuses on factors that influence the level of agency costs. Singh and Davidson (2003) assess whether agency costs vary with the extent of managerial ownership, the number of block holders, board size and the number of outside board commissioners on a sample of large public firms. Singh and Davidson find that higher managerial ownership has a positive influence

18 on one of the two indicators of agency costs (i.e. SGA expense and asset utilization are used as indicators but only for the indicator SGA expense a significant relationship was obtained) Therefore, they report that their analysis only provides limited evidence for the influence of managerial ownership on agency costs. Moreover, they find that a larger board size is associated with higher agency costs. Singh and Davidson find that the number of outside block holders and the number of outside board commissioners do not influence the level of agency costs. The finding of Sing and Davidson that block holders are unrelated to agency costs are somewhat contradictory to findings by other researchers. Holderness and Sheeham (1985), Barclay and Holderness (1991) and Shome and Singh (1995) provide empirical evidence for the prediction that block purchases lead to an increase in stock price. Barclay and Holderness find that purchases of blocks of shares are associated with an average abnormal stock price of 16.5%. Allen and Philips (2000) report that block purchases lead to an increase in operating and financial performance. Even though Allen and Philips, Sing and Davidson, Holderness and Sheeham, Barclay and Holderness and Shome and Singh do not directly analyze the impact of the number of block holders on agency costs their research outcomes seem to indicate that a larger number of block holders decreases agency costs. Table summarizes the prior research on factors that influence the amount of agency costs. Table 2.2.4: Prior research on factors that influence the level of agency costs Author Focus Sample Key finding Citation index Allen and Philips (2000) Barclay and Holderness (1991) Holderness and Sheeham (1985) Singh and Davidson (2003) Shome and Singh (1995) The influence of block ownership and firm performance Control of corporations by active block holders Influence of block holdings by six controversial investors on stock prices The relation between agency costs and extend of managerial ownership, the number of block holders, board size and the number of outside board commissioners The influence of block holdings on share prices Large public firms over the period Firms listed on the NYSE index over the period 1978 to 1982 Stock price changes associated with the first public announcements of stockholding by any of the six investors between 1977 and 1982 Firms listed on the NYSE, AMEX or NASDAQ index over the period 1976 to 1986 having annual sales revenue of $100m or more U.S. public firms in 1994 Block purchases lead to an increase in operating and financial performance Purchases of blocks of shares are associated with abnormal stock prices Average stockholders of target firms earned statistically significant positive abnormal returns when it was first announced that one of these six controversial investors had purchased blocks of the stock Managerial ownership is positively related to asset utilization but does not serve as a significant deterrent to excessive discretionary expenses. Board size is positively associated with agency costs. The number of outside block holders and the number of outside board commissioners have no impact on agency costs The stock market reacts positively on announcement of block purchases Empirical studies on the influence of financial constraints on the market value of cash holdings Besides studies on the pecking order theory and the agency costs of free cash flow theory, there also exists a vast body of literature on other factors that affect the market value of cash holdings. One

19 strand of literature studies the influence of firms financial constraints on the market value of cash holdings. Faulkender and Wang (2006) and Almeida, Campello and Weisbach (2004) both find that cash holdings are more valuable for constrained firms than for unconstrained ones. Faulkender and Wang find that shareholders of an average firm that is classified as financially constrained place a value of $1.04 on a dollar of cash, while shareholders of the average financially unconstrained firm only place a value of $0.77 on a dollar of cash. Contrary to the findings of Faulkender and Wang and Almeida, Campello and Weisbach, Pinkowitz and Williamson (2003) find that the market value of a dollar of cash holdings of unconstrained firms is about $1.17 while the market value of a dollar of cash holdings of constrained firms is about $0.81. Denis and Sibilkov (2010) study why cash holdings are more valuable for constrained firms. They find that investors place a premium on the value of cash holdings of constrained firms mainly because these cash holdings allow constrained firms to undertake value increasing projects that might otherwise be bypassed. Table summarizes the prior research on the influence of financial constraints on the market value of cash holding. Table 2.2.5: Prior research on the influence of financial constraints on the market value of cash holdings Author Focus Sample Key finding Citation index Almeida, Campello and Weisbach (2004) Denis and Sibilkov (2010) Faulkender and Wang (2006) Pinkowitz and Williamson (2003) The influence of financial constraints on the market value of cash holdings The influence of financial constraints on the market value of cash holdings The influence of corporate financial policy on the market value of cash holdings Determinants of the market value of cash holdings U.S. public manufacturing firms over the period 1971 to 2000 U.S. public firms over the period 1985 to Firms are required to have at least $25 million in total book assets in 1994 dollars. U.S. public firms over the period 1971 to 2001 U.S. public firms over the period 1950 to 1999 Cash holdings are more valuable for constrained firms Investors place a premium on the value of cash holdings of constrained firms because these cash holdings allow constrained firms to undertake value-increasing projects that might otherwise be bypassed Cash is more valuable for firms with large financial constraints Cash is more valuable for firms with low financial constraints Empirical studies on the influence of growth opportunities on the market value of cash holdings Another strand of literature studies the influence of firms growth opportunities on the market value of cash holdings. Myers and Majluf (1984) already argued that cash holdings are more valuable to firms with good investment opportunities, but their paper only provides theory and does not give empirical evidence. Opler, Pinkowitz, Stulz and Williamson (1997) find that firms with high growth opportunities and riskier cash flows held relatively high ratios of cash but they do not investigate whether there exist a relationship between growth opportunities and the market value of cash holdings

20 Empirical evidence for the existence of a relationship between growth opportunities and the market value of cash holdings is provided by Faulkender and Wang (2006) and Pinkowitz and Williamson (2003) who both find that the market value of cash is valued at a premium for companies with great growth opportunities. Pinkowitz and Williamson determine that a dollar of cash hold by firms with high growth opportunities has a market value between $1.11 and $1.84. Pinkowitz and Williamson argue that cash holdings may be valued at a premium by investors of firms with great growth opportunities because they make it possible for firms to fulfill their investment opportunities and it avoids that firms have to access the capital market where information costs have to be paid. Table summarizes the prior research on the influence of growth opportunities on the market value of cash holding. Table 2.2.6: Prior research on the influence of growth opportunities on the market value of cash holdings Author Focus Sample Key finding Citation index Opler, Pinkowitz, Stulz and Williamson (1997) Faulkender and Wang (2006) Pinkowitz and Williamson (2003) Myers and Majluf (1984) Determinants and implications of corporate cash holdings The influence of corporate financial policy on the market value of cash holdings Determinants of the market value of cash holdings Empirical relevance of the pecking order theory U.S. public manufacturing firms over the period 1971 to 2004 U.S. public firms over the period 1971 to 2001 U.S. public firms over the period 1950 to 1999 U.S. public firms over the period 1971 to 1989 Firms with strong growth opportunities and riskier cash flows hold relatively large amounts of cash. Firms that have the greatest access to the capital markets, such as large firms and those with high credit ratings, tend to hold small amounts of cash. Cash is more valuable for firms with greater growth opportunities Firms with greater growth opportunities or more volatile investment programs have their cash holdings valued at a premium Cash holdings are more valuable to firms that get the opportunity to invest in positive NPV projects Empirical studies on the influence of corporate governance on the market value of cash holdings Another strand of literature studies the influence of corporate governance on the market value of cash holdings. Schauten, van Dijk, and van der Waal (2008) study the influence of several corporate governance mechanisms on the market value of excess cash. They find that excess cash held by firms with less anti-takeover provisions is valued higher than excess cash held by more protected firms. Moreover they find that the governance mechanisms shareholder rights, disclosure and board functioning do not have a significant influence on the market value of excess cash. Dittmar and Mahrt-Smith (2007) investigate how corporate governance impacts firm value by comparing the market value of cash holdings of poorly and well-governed firms. They find that, on average, a dollar of cash has a market value of $1.09. However, according to their analysis, this value can decrease to $0.42 in case a firm is poorly governed. Alternatively, a dollar of cash of a wellgoverned firm has a market value as much as $1.62. Dittmar and Mahrt-Smith argue that these findings convey investors expectation of firm value to be lost from the waste of cash reserves

21 Similarly, Pinkowitz et al. (2006) investigate whether the market value of cash holdings is lower for minority shareholders in countries where investor protection is low. They find that a dollar of cash is worth $0.91 in countries with above average investor protection while it is worth only $0.33 in countries where investor protection is low. Pinkowitz et al. argue that the value of cash holdings is lower in countries where investor protection is low because firms in these countries are likely to suffer more from agency problems. Table summarizes the prior research on the influence of corporate governance on the market value of cash holding. Table 2.2.7: Prior research on the influence of corporate governance on the market value of cash holdings Author Focus Sample Key finding Citation index Dittmar and Mahrt-Smith (2007) Pinkowitz, Stultz and Williamson (2006) Schauten, van Dijk, and van der Waal (2008) The influence of corporate governance on the market value of cash holdings The influence of corporate governance on the market value of cash holdings The influence of corporate governance on the market value of excess cash holding U.S. public firms over the period 1990 to 2003 Public firms over the period 1983 to 1998 Firms listed on the FTSEurofirst 300 index over the period 2000 to 2004 Investors place a lower value on the cash holdings of firms that are poorly governed and which can be explained as a convey of investors expectation of firm value to be lost from the waste of cash reserves Investors in countries with poor governance scores place a lower value on a dollar of corporate cash holdings than investors in countries with above average governance scores Excess cash held by firms with less anti-takeover provisions is valued higher than excess cash held by more protected firms The governance mechanisms shareholder rights, disclosure and board functioning do not have a significant influence on the market value of excess cash unknown Empirical studies on the influence of corporate financial policy on the market value of cash holdings Faulkender and Wang (2006) study how firms financial policy might affect the market value of cash holdings. They argue that the value of cash to shareholders varies considerably depending upon whether cash is more likely to be used to: (1) distribute earnings to shareholders via dividend payments or share repurchases, (2) service debt or other liabilities or (3) finance positive NPV projects decreasing the amount of money that needs to be raised in the capital market. For firms whose cash reserves appear to greatly exceed their future needs, a dollar of cash holdings is more likely to be distributed to equity holders through dividends or stock repurchases. However, because of personal taxes on dividends and capital gains ( ) only a fraction (1- ) ends up in the hands of shareholders. For firms that use cash holdings to pay down debt, cash will increase debt value and not equity value. Thus, equity holders are likely to place a lower value on cash for high leveraged firms relative to the value of cash holdings for a firm with little debt. For firms that use cash holdings to finance growth opportunities investors are likely to place a premium on a firm s cash holdings because the availability of cash avoids that firms have to access the capital market where information costs have to be paid. 21

22 Faulkender and Wang (2006) link firm characteristics to each of the three financial policy regimes and develop hypotheses which they empirically test. They find that the market value of a dollar of cash holdings of the average firm is $0.94. This finding suggests that less than the full value of a dollar of cash holdings is incorporated in the stock price which is consistent with the idea that shareholders value cash at its after tax value. Furthermore, they find that the market value of cash holdings decreases with the amount of leverage which supports their belief that shareholders of firms with high debt obligations place a discount on the company s cash holdings. Finally, they determine that the market value of cash increases with a firm s financial constraints and growth opportunities. These findings confirm the believe that shareholders place a premium on a company s cash holdings once it can be used to finance growth and external financing can be avoid. Based on these findings their overall conclusion is that a firm s financial policy influences the market value of cash holdings. Table summarizes the prior research on the influence of corporate financial policy on the market value of cash holding. Table 2.2.8: Prior research on the influence of corporate financial policy on the market value of cash holdings Author Focus Sample Key finding Citation index Faulkender and Wang (2006) The influence of corporate financial policy on the market value of cash holdings U.S. public firms over the period 1971 to 2001 The marginal value of cash declines with larger cash holdings, higher leverage, better access to capital markets, and as firms choose greater cash distribution via dividends rather than repurchases 202 Empirical studies on the influence of target levels for cash holdings on the market value of cash holdings In their paper, Pinkowitz and Williamson (2003) discuss whether firms have target levels for their cash holdings and how the existence of target levels would affect the market value of cash holdings. Pinkowitz and Williamson do not take a position on whether firms have an optimal level of cash holdings but they explain the influence that target levels for cash holdings may have on the market value of cash holdings. They discuss that it could be that firms specifically target the level of their cash holdings and attempt to increase or decrease their cash balances in case they stray from their target. Moreover, they discuss that, alternatively, it could be that an optimal cash level does not exist and firms simply accumulate cash when internal cash flows exceed investments and draw down cash when the reverse is true. Pinkowitz and Williamson argue that if there is no optimal level of cash holdings, then calculations on what cash is worth to the market should accurately reflect the real market value of cash holdings. However, they also explain that if an optimal amount of cash holdings would exist, estimates of the market value of cash holdings are conditional on where the firm is relative to its optimum. If a firm is at its target changes in cash holdings do not contribute much to the market value of a firm. Alternatively, if a firm is far away from its target changes in cash holdings are likely to contribute more to the market value of the firm because it controls for movements towards the target level. In order to analyze the influence of the existence of target levels for cash holdings, besides the variable current cash holdings, in their model Pinkowitz and Williamson (2003) also incorporate variables that account for two year past and future changes in cash holdings. They find that both past and future changes only contribute very little to the market value of a firm. Table summarizes the prior research on the influence of target levels for cash holdings on the market value of cash holding. 22

23 Table 2.2.9: Prior research on the influence of target levels for cash holdings on the market value of cash holdings Author Focus Sample Key finding Citation index Pinkowitz and Williamson (2003) Determinants of the market value of cash holdings U.S. public firms over the period 1950 to 1999 Past and future changes of cash holding levels only contribute just a little to the market value of a firm 60 Existing methodologies to measure the market value of cash holdings Fama and French (1998) study how firm value is related to dividend payments and debt. More specifically, they examine how taxes on dividends and debt influence the market value of a firm. They develop a regression model that allows for estimation of the market value of a firm s equity as a function of a number of firm-specific characteristics. Pinkowitz and Williamson (2003) extend this regression model by splitting the independent variable assets into its cash and non-cash components. This modification allows to specifically analyze how cash holdings contribute to the market value of a firm. The model of Pinkowitz and Williamson has an adjusted R-square value of which means that the model explains about 47% of the market value of a firm. Faulkender and Wang (2006) develop an alternative model to determine the market value of cash holdings. In contrast to the methodology of Fama and French, Faulkender and Wang use a methodology with excess equity return rather than overall equity value as indicator of the market value of a firm s equity. Their model has an adjusted R-square value of which means that the model explains about 21% of the market value of a firm. Unfortunately, neither Pinkowitz and Williamson and Pinkowitz nor Faulkender and Wang do report any information on outliers. Both studies only mention that in order to reduce the effect of outliers they trim their sample at 1% by dropping 0.5% in each tail of each variable. Table summarizes the prior research on methodologies to measure the market value of cash. 23

24 Table : Prior research on methodologies to measure the market value of cash holdings Author Focus Sample Key finding Citation index Fama and French (1998) Faulkender and Wang (2006) Pinkowitz and Williamson (2003) The relation between firm value, dividend payments, debt levels and taxes The influence of corporate financial policy on the market value of cash holdings Determinants of the market value of cash holdings U.S. public firms over the period 1965 to 1992 U.S. public firms over the period 1971 to 2001 U.S. public firms over the period 1950 to 1999 Using a linear regression model an estimation of the market value of a firm s equity can be derived from a firm s current and future values of dividends, interest, earnings, investment and R&D expenditures Using a linear regression model an estimation of the market value of a firm s equity can be derived from a firm s cash holdings, earnings, R&D expenditures, dividend payment, leverage and net financing An Extension of the valuation model of Fama and French (1998) allows to determine the market value of cash holdings as function of a firm s overall value market value Concluding remarks This research intends to build further upon previous findings as discussed in this chapter. Various previous researchers have empirically studied the relevance of the pecking order theory (Bharath, Pasquariello and Wu 2009; Fama and French 2005; Faulkender and Wang 2006; Frank and Goyal 2003; Leary and Roberts 2010; Lemmon and Zender 2006; Myers and Majluf 1984; Shyam-Sunder and Myers 1999) and the agency costs of free cash flow theory (D'Mello and Miranda 2010; Dittmar and Mahrt-Smith 2007; Drobetz, Grüninger and Hirschvogl 2010; Jensen 1986; Nohel and Tarhan 1998; Pinkowitz, Stulz and Williamson 2006). This research contributes to the debate about the relevance of the pecking order theory and the agency costs of free cash flow theory by developing and testing hypotheses in line with this discussion. Moreover, this research contributes through the operationalization of the key concepts of both theories based on and built upon the existing literature. Additionally, this research contributes to the debate about the market value of cash holdings. Previous research has primarily focused on the influence of single factors on the market value of cash holdings (Almeida, Campello and Weisbach 2004; Denis and Sibilkov 2010; Dittmar and Mahrt-Smith 2007; Drobetz, Grüninger and Hirschvogl 2010; Faulkender and Wang 2006; Pinkowitz, Stulz and Williamson 2006; Pinkowitz and Williamson 2003; Schauten, van Dijk and Van der Waal 2008). This research contributes to these earlier studies by capturing these key factors into one empirical model. By doing this, it meets the recommendation of Opler et al. (1997) who suggested further research to focus on the market value of cash holdings. Moreover, as suggested by Faulkender and Wang (2006) this research uses a different set of data to analyze the market value of cash holdings. By using a different data set this research also provides additional tests on the validity of the valuation model of Fama and French (1998) as was given as a recommendation in their paper. Similar to earlier studies on the market value of cash holdings, this research has a strong explanatory power because of its longitudinal character

25 Finally, while previous research has primarily focused on the direct influence of macroeconomic and industry specific factors on the market value of cash holdings (Almeida, Campello and Weisbach 2004; Denis and Sibilkov 2010; Dittmar and Mahrt-Smith 2007; Drobetz, Grüninger and Hirschvogl 2010; Faulkender and Wang 2006; Pinkowitz, Stulz and Williamson 2006; Pinkowitz and Williamson 2003; Schauten, van Dijk and Van der Waal 2008), this research also assesses whether macroeconomic and industry specific factors influence the relationship between agency costs and information costs and the market value of cash holdings. 25

26 3. Conceptual design and hypotheses development 3.1 Conceptual model The literature review, as presented in the previous chapter, helps to identify and highlight relevant findings that serve as a foundation on which the conceptual model can be build. As follows from the main research question, the primary variable of interest is market value of cash holdings and therefore, in our conceptual model, this variable is adopted as dependent variable. From the literature review it has been investigated that the variables: financial constraints, growth opportunities, book value of cash holdings, corporate governance, information costs and agency costs influence the market value of cash holdings in either a positive or negative way. Since this research aims to analyze the effect of information costs and agency costs on the market value of cash holdings the variables information costs and agency costs are adopted as independent variables. The variables financial constraints, growth opportunities, total book value of cash holdings, and corporate governance are adopted as covariate variables (also called: control variables). The predicted relationship between the independent variables and the dependent variable is conceptualized in figure For completeness, the figure also shows the factors that have an influence on each of the independent variables. However, it is important to note that an analysis on these relationships is outside the scope of this research. This research focuses specifically on the relationships presented in the dotted line box. Figure Conceptual model 26

27 3.2 Definition and clarification of the variables Before proceeding with the operationalization of the key variables it is necessary to clarify and define the main concepts. Definition and clarification of the dependent variable Market value of cash holdings The overall value of a company s cash holdings in the eyes of the investor. In perfect markets the market value of cash equals its book value which means that cash with a book value of one euro has a value of one euro in the market. In real world, however, factors such as taxes, moral hazard and information asymmetry make markets imperfect which cause the market value of cash to be different from its book value. Definition and clarification of the independent variables Information costs Informational costs associated with external financing due to the existence of information asymmetry between managers and investors. Cash holdings may be valuable to firms with high information costs because it avoids firms from raising capital externally. Therefore, information costs are expected to have a positive influence on the market value of cash holdings. Agency costs Managerial costs associated with managers who make decisions that maximize their own utility at the expense of the utility of shareholders. Cash holdings may not be valuable to firms with high agency costs because in case managers do not strive to maximize shareholder value cash holdings may be wasted on non-value adding projects. Therefore, agency costs are expected to have a negative influence on the market value of cash holdings. Definition and clarification of the covariate variables Financial constraints The difficulty for firms to raise capital in the market. Cash holdings may be valuable to firms with larger financial constraints because they allow the firm to invest when other sources of funds are costly, limited, or unavailable. Therefore, financial constraints are expected to have a positive influence on the market value of cash holdings. 27

28 Growth opportunities The chances a firm gets to undertake value adding projects. Cash holdings may be valuable to firms with great growth opportunities because they make it possible for firms to undertake positive NPV projects that add additional value to the firm. Therefore, growth opportunities are expected to have a positive influence on the market value of cash holdings. Book value of cash holdings The amount of money a firm holds in cash or cash equivalents with an insignificant risk of change in value. The book value of cash holdings includes cash and cash equivalents. Cash is the amount of money that a company holds as a deposit in its saving account. Cash equivalents are financial securities that can be converted into cash immediately and have an insignificant risk of change in value, for instance: short term government bonds, money market holdings or commercial papers. Common stocks cannot be considered as cash equivalents because this type of investment includes a certain risk of change in value. The market value of cash holdings is higher for firms with a higher book value of cash. Corporate governance The ability of shareholders to monitor and influence actions taken by a company s management. Cash holdings may not be valuable to firms that are poorly governed because the absence of good corporate governance mechanisms makes it difficult for shareholders to monitor and influence actions taken by a company s management. Contrary, when shareholders are able to monitor and influence actions taken by management they can steer and control these actions and this may prevent inefficient management behavior. Therefore, corporate governance is expected to have a positive influence on the market value of cash holdings. The variables agency costs and corporate governance are expected to be interrelated. Firms that are well governed are expected to suffer less from agency costs because the availability of good governance mechanisms will allow shareholders to carefully monitor management behavior and to take action in case managers do not seem to act in shareholders interest. 3.3 Operationalization of the variables as proposed in the literature Operationalization of the dependent variable Operationalization refers to the operations by which we measure a concept. For the operationalization of the dependent variable we use the model developed by Fama and French (1998) and Pinkowitz and Williamson (2003). This model allows us to derive the market value of cash holdings from a model that regresses the overall market value of a firm s equity as function of a number of firm-specific characteristics. The model also incorporates one year past and future values because Fama and French and Pinkowitz and Williamson believe that investors base the value of a firm s equity not only on the firm s current status but also on past returns and expectations about future earnings. To correct for firm size all variables are divided by net assets. 28

29 The regression model is: Where: Market value of equity (=market capitalization value) Earnings before interest and extraordinary items but after tax and depreciation Book value of net assets (= total assets cash) R&D expenditures Interest expense Dividends paid Book value of cash holdings And: * To simplify the notation the firm subscript that should appear on all variables in the regression and the year subscript that should appear on all regression coefficients are omitted in the equation. The regression coefficient of interest in this model is. This coefficient on cash holdings indicates the marginal contribution of cash holdings to the market value of a firm s equity. Because cash is carried on the books euro for euro, this coefficient represents the value investors place on every euro of cash written in the books. Once the marginal value of cash holdings is known the absolute market value of cash can simply be determined by multiplying this value with the total book value of cash holdings. Operationalization of the independent variables For the operationalization of the independent variables and the covariate variables we follow Goertz (2006) who developed a hierarchical order to operationalize variables. Each variable is specified at three levels: the concept level, the item level and the indicator level. The concept level and the item level correspond to the definition of the variable as given in section 3.1. The indicator level is the level of operationalization. Information costs Literature proposes a variety of ways to measure information costs. However, there is no general agreement on which measure is the best proxy for information costs. We derive a proxy for information costs from two items that relate to the level of information asymmetry between the company s management and investors, respectively: financial analysts forecast inaccuracy and disagreement among financial analysts (also called: analyst dispersion). To measure these two items we follow DaDalt, Gay and Nam (2002). Financial analysts forecast inaccuracy is measured as the absolute value of the average earnings forecast error, normalized by the firm s stock price. Financial analyst dispersion is measured as the standard deviation of a firm s earnings per share estimates, normalized by the firm s stock price. The higher the financial analysts forecast inaccuracy and the 29

30 financial analyst dispersion the higher the level of information asymmetry and therefore the higher the information costs associated with external financing. The operationalization of this variable is depicted in Figure Figure 3.3.1: Operationalization of information costs Agency costs To establish indicators of agency costs we follow Singh and Davidson (2003) who studied the relation between agency costs, ownership structure and governance mechanism on a sample of large public U.S. firms. Singh and Davidson defined two items of agency costs. The first one is asset utilization measured by the ratio of annual sales to total assets. This indicator measures management s ability to employ assets efficiently. Low asset utilization indicates that management deploys its assets inefficiently and, according to Singh and Davidson, this signals agency costs. As a second item they defined a firms Selling, General and Administrative expense structure (SGA expense structure) measured by the ratio of SGA expense to annual sales. SGA expenses include salaries, advertisements, rents, utilities, lease payments and supplies and these directly reflect expenses on office buildings, furnishings and automobiles and therefore they reflect managerial discretion in spending company resources (Singh and Davidson III 2003). High excessive pay and perquisite consumption indicates that managers spend excessive amounts of cash on non-value adding projects cash and, according to Singh and Davidson; this is another signal of agency costs. The operationalization of this variable is depicted in Figure Figure 3.3.2: Operationalization of agency costs 30

31 Operationalization of the covariate variables Financial constraints The literature proposes a variety of ways to identify the level of financial constraints faced by firms. However, there is no general agreement on which measure is the best proxy for financial constraints. We follow Almeida et al. (2004), Denis and Sibilkov (2010), Pinkowitz and Williamson (2003) and Faulkender and Wang (2006) to establish indictors of financial constraints. The first item is dividend payout ratio. Firms with high payout ratios have less cash left to finance their debt obligations and therefore they face larger financial constraints (Almeida et al. 2004). Payout ratio will be measured by the sum of a firm s total dividend payments and share repurchases. The second item is firm size. Larger firms are thought to be better known and have better access to capital markets than smaller firms. The larger the firm the fewer it will face constraints when raising capital (Faulkender and Wang 2006). Firm size is measured by a firm s total assets. The third item is related to a firm s ability to access public debt markets. Firms with a good bond rating have good access to public debt markets and are able to raise funds from a source of capital that those with a bad bond rating or with no bond rating are not be able to get (Faulkender and Wang 2006). Therefore, the higher a firms bond rating the lower its level of financial constraints. The operationalization of the variable financial constraints is depicted in Figure Figure 3.3.3: Operationalization of financial constraints Growth opportunities As an indicator of growth opportunities we use market to book ratio. Together with Tobin s q market to book ratio is one of the most widely used indicators to measure growth opportunities (Pinkowitz and Williamson 2003). A high market to book ratio indicates that a firm has explored great growth opportunities and is expected to get many more opportunities to undertake positive NPV projects in the future. Therefore, the higher a firms market to book ratio the greater its growth opportunities. The operationalization of this variable is depicted in Figure Figure 3.3.4: Operationalization of growth opportunities 31

32 Book value of cash holdings As an indicator for the book value of cash holdings we use the value of cash holdings as written in the company s financial statements. The operationalization of this variable is depicted in Figure Figure 3.3.5: Operationalization of book value of cash holdings Corporate governance To establish indicators of corporate governance we use the findings by Schauten et al. (2008) who studied the influence of corporate governance on the market value of excess cash holdings on a sample of large European firms. Schauten et al. use data on shareholder rights, takeover defenses, disclosure and board functioning to measure corporate governance and find that out of these four indicators only takeover defenses has a significant influence on the market value of excess cash holdings. Takeover defense is measured by the number of anti-takeover provisions. Anti-takeover provisions are management rights to prevent a firm from being acquired by a successful bidder. Firms with high antitakeover provisions are seen as poorly governed firms. The operationalization of the variable corporate governance is depicted in Figure Figure 3.3.6: Operationalization of corporate governance Table summarizes the operationalization of the key variables as proposed by the literature. 32

33 Table 3.3.7: Overview of the operationalization of the key variables as proposed by the literature Variable Item Indicator Variable type Unity Scale Literature reference Data source Financial constraints Growth opportunities Book value of cash holdings Payout ratio Firm size Ability to access debt markets Opportunity to undertake positive NPV projects Internal value of cash holdings Total dividends and share repurchases Total assets Long term bond rating Covariate Thousands of euros Thousands of euros Alphanumeric Ratio Ratio - Almeida et al. (2004); Denis and Sibilkov (2010); Pinkowitz and Williamson (2003); Faulkender and Wang (2006) Market to book ratio Covariate Dimensionless Ratio Pinkowitz and Williamson (2003) Value of cash as written in FS Covariate Thousands of euros Ratio Pinkowitz and Williamson (2003); Pinkowitz et al. (2006); Faulkender and Wang (2006); Drobetz et al. (2010) Corporate governance Takeover defenses Anti-takeover provisions Covariate Dimensionless Ratio Schauten et al. (2008) Deminor Information costs Financial analyst forecast inaccuracy Financial analyst dispersion Absolute value of the average earnings forecast error, normalized by the firm s stock price Standard deviation of a firm s EPS estimate, normalized by the firm s stock price Independent Dimensionless Dimensionless Ratio Ratio DaDalt et al. (2002) I/B/E/S Agency costs Market value of cash holdings Asset utilization SGA expense structure Annuals sales to total asset SGA expense to annual sales Independent Dimensionless Dimensionless Ratio Ratio Singh and Davidson (2003) - - Dependent Thousands of euros Ratio Fama and French (1998) Pinkowitz and Williamson (2003) Thomson/Worldscope Thomson/Worldscope Thomson/Worldscope Thomson/Worldscope Thomson/Worldscope 33

34 3.4 Operationalization of the variables as performed in this research In the previous section we presented the operationalization of the variables as proposed by the literature. However, for reasons outlined below the operationalization of some of the variables differs slightly from the way it was proposed in the literature. First of all, the financial databases that we used did not contain data on takeover defenses; the indicator of corporate governance. Collecting data on this indicator manually would take a tremendous amount of time. Hence, we decided to leave the covariate variable corporate governance out of our empirical design. Secondly, it became clear that most of the firms traded at NYSE Euronext Amsterdam Stock Exchange did not hold corporate bonds. As a result, data on bond ratings could only be gathered for a few companies. Hence, in order to avoid that we would end up with a relatively small set of observations it was decided not to incorporate long term bond rating as an indicator of financial constraints in our empirical design. Thirdly, all variables are corrected for firm size because we believe that the relationship between the dependent variable and each of the predictors may be affected by the size of the company. Moreover, as argued by Fama and French, correcting for firm size avoids that the regression results will be dominated by the larger firms. We used net assets ( ) as a proxy for firm size. Finally, an important remark has to be made with respect to the use of the regression model of Fama and French (1998) and Pinkowitz and Williamson (2003). As discussed earlier, their model takes into account a time factor (i.e. one year past and future changes). However, taking into account this time factor strongly limits the number of observations that can be included in the sample because including these time factors would mean that we can only select those companies that have been listed for at least three years in a row. Thus, in order to avoid that we would end up with a relatively small set of observations it was decided not to incorporate the time factors into the model that we use to determine the market value of cash holdings. Table gives an overview of the operationalization of the key variables as performed in this research. 34

35 Table 3.4.1: Overview of the operationalization of the key variables as performed in this research Variable Item Indicator Variable type Unity Scale Literature reference Data source Financial constraints Growth opportunities Book value of cash holdings Information costs Agency costs Payout ratio Firm size Opportunity to undertake positive NPV projects Internal value of cash holdings Financial analyst forecast inaccuracy Financial analyst dispersion Asset utilization SGA expense structure Total dividends and share repurchases Total assets Covariate Thousands of euros Thousands of euros Ratio Ratio Almeida et al. (2004); Denis and Sibilkov (2010); Pinkowitz and Williamson (2003); Faulkender and Wang (2006) Market to book ratio Covariate Dimensionless Ratio Pinkowitz and Williamson (2003) Value of cash as written in FS Covariate Thousands of euros Ratio Pinkowitz and Williamson (2003); Pinkowitz et al. (2006); Faulkender and Wang (2006); Drobetz et al. (2010) Absolute value of the average Independent Dimensionless Ratio DaDalt et al. (2002) earnings forecast error, normalized by the firm s stock price Standard deviation of a firm s EPS estimate, normalized by Dimensionless Ratio the firm s stock price Annuals sales to total asset SGA expense to annual sales Independent Dimensionless Dimensionless Ratio Ratio Singh and Davidson (2003) Thomson/Worldscope Thomson/Worldscope Thomson/Worldscope I/B/E/S Thomson/Worldscope Market value of cash holdings - - Dependent Thousands of euros Ratio Fama and French (1998) Pinkowitz and Williamson (2003) Thomson/Worldscope Note: All variables presented in table were corrected for firm size. We used net assets ( ) as a proxy for firm size. 35

36 3.5 Hypotheses development The conceptual model depicts the expected relationships among variables derived from theory. From the conceptual model testable hypotheses are developed. The pecking order theory predicts that firms prefer to raise capital in the following order: internal, debt and equity (Myers and Majluf 1984). When companies have to raise capital in the external market they pay a premium on each euro of cash because investors want to get compensated for not knowing the exact state of the firm. The premium that has to be paid is referred to as information costs. The amount of information costs is assumed to be directly related to the level of information asymmetry. The higher the information costs the more expensive it is for companies to raise money externally and therefore we predict that information costs positively influence the market value of capital that is internally available to the firm. This reasoning results in our first hypothesis: Hypothesis 1: Information costs positively influence the market value of cash holdings. The agency costs of free cash flow theory predicts that cash will be wasted by the company s management because managers act in their own interest rather than that the aim to maximize the wealth of shareholders (Jensen 1986). Agency costs are exploited through on the job consumption, empire building or entrenchment. The larger these agency problems the more money gets wasted by company s management and therefore we expect that agency costs negatively influence the market value of cash. This reasoning results in our second hypothesis: Hypothesis 2: Agency costs negatively influence the market value of cash holdings. The first two hypotheses contain opposing expectations concerning the market value of cash holdings as a result of two contradictory applicable theories. If it can be proven that there exists a relationship between agency costs (related to the agency costs of free cash flow theory) and the market value of cash holdings, we may still find that there also exists a relationship between information costs (related to the pecking order theory) and the market value of cash holdings. In case we do not detect any relationship we cannot rule out that both relationships are at work but cancel each other out. Besides investigating the existence of a relationship between agency costs and information costs and the market value of cash holdings we also aim to determine which of the two has the largest impact on the market value of cash holdings, or in other words; which of the two is most relevant in the eyes of the investor. This reasoning results in our third and fourth hypotheses. Our third hypothesis is: Hypothesis 3: Agency costs have a larger impact on the market value of cash holdings than information costs. If it cannot be proven that agency costs have a larger impact on the market value of cash holdings than information costs it does not necessarily mean that the reverse is true. In order to test whether information costs has a larger impact on the market value of cash holdings than agency costs we developed a fourth hypothesis: Hypothesis 4: Information costs have a larger impact on the market value of cash holdings than agency costs. 36

37 As discussed in section 1.2 the relevance of the pecking order theory and the agency costs of free cash flow theory may be influenced by industry characteristics. Information costs are likely to be more important for firms in growth industries while agency costs are likely to be particularly relevant to firms in mature industries (Nyborg 2010). This reasoning results in our fifth and sixth hypotheses. Our fifth hypothesis sounds: Hypothesis 5: Information costs are more important than agency costs for firms in growth industries. Our sixth hypothesis is: Hypothesis 6: Agency costs are more important than information costs for firms in mature industries. Section 1.2 also elaborates on the potential influence of macroeconomic effects on the market value of cash holdings. Since this research focuses on one specific country, we take macroeconomic effects into account on a national level rather than on a global level. In times of economic downturn it becomes more difficult to obtain cash which makes it relatively expensive for firms to raise money in the market (Kleinknecht 2009). Hence, cash holdings are likely to be valued at a premium by investors. On the other hand, in times of national economic boom the market rather has positive expectations about the future and cash is easily transferred among firms, banks and other type of investors (Kleinknecht 2009). Thus, in times of economic growth it is much easier for firms to raise money in the market and therefore it is likely that investors do not place any additional value on companies cash holdings. The time period that we cover in this research does not allow us to test how the market value of cash holdings may differ when comparing periods of economic downturn with periods of economic growth. Therefore, we will just generally analyze whether macroeconomic status influences the market value of cash holdings. This reasoning results in our seventh hypothesis: Hypothesis 7: Macroeconomic status influences the market value of cash holdings. 37

38 4. Methodology 4.1 Empirical research design Research approach The main purpose of this research is to investigate quantitatively whether investors believe the pecking order theory or the agency costs of free cash flow theory to be the most relevant to firms. A problem statement was developed after we investigated a number of interesting global economic trends and the related theory. An extensive literature review was performed to give a clear and logical presentation of the relevant research work done thus far. Based on literature findings we developed a conceptual model. The conceptual model shows the predicted relationships among variables. The operationalization of the main variables was established after careful evaluation of suggestions given in the existing literature. Based on the predicted relationships in the conceptual model we developed seven hypotheses which will be statistically tested. The outcome of these statistical tests will be used to establish a satisfying empirical design. Figure gives a stepwise overview of the research approach that we applied. Figure 4.1.1: Stepwise overview of the research approach Research strategy The research strategy that we used is year by year cross-sectional hierarchical multiple linear regression. Hierarchical multiple linear regression fits well with a hypothesis testing type of research design because it allows to test the hypothesized relationships among variables (Sekaran 2003). In hierarchical regression the covariate variables are entered into the model first. After the covariate variables are added the independent variables are added in a stepwise manner. The advantage of crosssectional regression compared to time-series regression is that it does not require that firms survive for long periods, so they can cover many more firms than time series regression (Sekaran 2003). Reliability Reliability refers to the extent to which the research and its results are without bias and consistent across time and across researchers (Sekaran 2003). In other words it refers to the robustness of our results. To ensure the robustness of our research we tested the hypotheses using three different empirical models. The various models that are used are explained in more detail in the next section. The use of different empirical models allowed us to derive valid conclusions from the research outcome. To assess the reliability of each of the empirical models we interpreted their R-square value, the adjusted R-square value and by calculating the difference between these two values. R-square is a measure of how much of the variability in the outcome is accounted for by the predictors (Field 2009). The adjusted R-square tells us something about how well the model generalizes (Field 2009). The differences between the value of R square and the adjusted R square indicates to what extend the model s explanatory power would shrink if the model was derived from the population instead of the 38

39 sample (Field 2009). A reliable regression model has a low difference of R-square value and adjusted R-square indicating that its outcome is well defined and generalizable. To assess the reliability of the individual predictors that are adopted in each of the empirical models we looked at the confidence intervals of the regression coefficients of the predictors. These confidence intervals tell us how accurately the regression coefficients represents reality (Field 2009). A reliable predictor has a regression coefficient with small confidence intervals indicating that the obtained coefficient is an accurate estimate of its real value. Validity Validity is concerned with whether findings are really about what they appear to be about. In other words, it tests how well an instrument that is developed measures the particular concept it is intended to measure (Sekaran 2003). There are three types of validity: internal validity, external validity and concept validity (Sekaran 2003). Internal validity addresses the question, To what extent does the research design permit us to say that the a change in the dependent variable is caused by the independent variables? (Sekaran 2003). This research has a weak internal validity of zero since the regression analysis that is performed only indicates to what extend the dependent variable is correlated with the independent variables, it is not identified whether a change in the dependent variable is caused by the independent variables. External validity refers to the extent of generalizability of the results (Sekaran 2003). The external validity of this research is very high, for three reasons. First of all, the research setting is a real-world setting. In contrast to findings obtained using other research strategies, such as experiments, the findings obtained through regression analysis are sure to represent reality. Secondly, this research covers a ten year time span. An analysis over a longer period of time contributes considerably to the external validity of the research. Finally, although the sample studied in this research mainly includes Dutch public firms, the relationships between the various concepts as depicted in the conceptual model might also be applicable to non-dutch companies. The final form of validity is concept validity. Concept validity refers to the extent to which the concept reflects the empirical fact in a research setting (Sekaran 2003). Concepts need to be operationalized in such a way that they are measured correctly and adequately. Conceptualization and operationalization of the variables has already been established following an iterative process based on findings in previous studies. Hence, concept validity is warranted through close collaboration and discussion with academic experts in the field of finance and economics. 4.2 Empirical models To ensure the robustness of the research findings we tested our hypotheses on three different empirical models. The first two models are developed by us based on findings in the literature. The third model is an extended version of the valuation model developed by Fama and French (1998) and Pinkowitz and Williamson (2003). Model 1 The first regression model incorporates the indicators of the variables as they were derived from literature. Using the indicators instead of the overarching variables avoids that the data loses its absolute value. The baseline model is: 39

40 Model 2 The second regression model incorporates the variables constructed from the indicators. Since the data on these variables is constructed through the combination of multiple indicators it has lost its absolute value and rather represents a relative value. The baseline model is: Since the indicators were measured with different measurement ranges we needed to add a scaling factor to the indicator with the smallest measurement range in order to ensure that the indicators equally impact the overarching variable. We obtained the following formulas: Model 3 The third regression model is an extended version of the regression model developed by Fama and French (1998) and Pinkowitz and Williamson (2003). In the regression model of Fama and French and Pinkowitz and Williamson the regression coefficient on cash holdings indicates the value investors place on every euro of cash written in the books (i.e. the market value of cash holdings). As specified in our hypothesis we are interested in the impact of agency costs and information costs on the market value of cash holdings. In order to measure these effects two additional interactions terms are added to the original regression model. One of the interaction terms is information costs multiplied by cash, the other interaction term is agency costs multiplied by cash. The regression coefficients that belong to these interaction terms are the ones of main interest. The coefficient on the interaction term of information costs and cash measures the influence of information costs on the market value of a single dollar of cash holdings. The coefficient on the interaction term of agency costs and cash measures the influence of agency costs on the market value of a single dollar of cash holdings. To control for a direct influence of information costs and agency costs on firm value both variables are also added as individual predictors. The baseline model is: Concluding remarks All three models allow to investigate the influence of information costs and agency costs on the market value of cash holdings. Hence, all three models will be used to test the first four hypotheses. 40

41 The test results related to hypotheses 1, 2, 3 and 4 are reported in section 7.1. However, because the third model does not include the market value of cash holdings as dependent variable this model cannot be used to test hypotheses 5, 6 and 7. The test results related to hypotheses 5, 6 and 7 are reported in section 7.2. Table gives an overview of the models that are used to test a particular hypothesis. Table 4.2.1: Overview of the models that are used to test a particular hypothesis Hypothesis Model 1 Model 2 Model 3 H1: Information costs positively influence the market value of cash holdings H2: Agency costs negatively influence the market value of cash holdings H3: Agency costs have a larger impact on the market value of cash holdings than information costs H4: Information costs have a larger impact on the market value of cash holdings than agency costs H5: Information costs are more important than agency costs for firms in growth industries - H6: Agency costs are more important than information costs for firms in mature industries - H7: Macroeconomic status influences the market value of cash holdings Data collection and data sources Collection of relevant literature We used academic search engines such as Wiley Interscience, Google Scholar and ScienceDirect to access articles and publications from top journals in the fields of finance, economics and business management. A major advantage of using academic search engines is that they give access to a wide range of publications. Besides, it is a relatively fast way to find useful articles of high quality. When looking for articles, special attention was paid to the year of publication and the journal impact factor. We used the so called Snowball method to collect articles. This method aims to find useful articles and scan them on other useful citations and references. After we completed this ad hoc literature scan the most relevant articles were selected and read in more detail. A complete overview of the articles that we read as part of our literature study can be found in appendix A. Collection of financial data The financial databases Thomson One Banker, Worldscope and I/B/E/S were used to gather data on the key variables. The focus of this research is companies traded at NYSE Euronext Amsterdam Stock Exchange for at least one year during the period from 2000 to 2009 so we only collected data on those companies that fulfilled these requirements. The data extracted from the financial data bases are all nominal values. Therefore, the data includes annual inflation rates. Since the annual inflation rates may differ annually for the time span of this research all nominal values were converted into real values in 2009 using the Consumer Price Index (CPI). 41

42 4.4 Case selection approach Original data set The original data set contains 3800 observations. These observations represent securities of companies (i.e. corporate common stocks or common stock of investments funds) listed on Euronext for at least one year during the period from 2000 to Table gives an overview of the number of observations per industry with companies classified based on Standard Industrial Classification Codes (SIC Codes). Table 4.4.1: Number of observations per industry with companies classified based on SIC codes SIC Code Industry Number of observations Fraction 01xx - 09xx Agriculture, Forestry and Fishing xx - 14xx Mining xx - 17xx Construction xx - 39xx Manufacturing xx - 49xx Transportation, Communication, Electric and Gas xx - 51xx Wholesale trade xx - 59xx Retail trade xx - 69xx Financial service xx - 89xx Service xx - 99xx Public administration unknown unknown Total Case selection process The process of case selection is depicted in figure The figure shows all steps that have been taken to come to a homogeneous representation of the original sample. Moreover, the figure indicates the number of observations that were removed at each of the individual selection steps. In general, cases are selected based on the availability of data on the variables incorporated in the empirical models. In case relevant data was not provided by the financial data bases we tried to look it up online in firms annual reports. However, because of time constraints, in case the data could not be extracted from the financial data bases or from online annual reports we did not to look further for it and the observation was removed from our sample. Similar to Fama and French (1998) Pinkowitz and Williamson (2003) and Faulkender and Wang (2006) we excluded financial firms (SIC code 6000 to 6999) from the data set because these firms use their cash for different purposes than non-financial firms. As a second step we looked at the availability of data on stock prices and excluded those observations for which this data was not available. As a third step we looked at the data on SGA expense which was required to construct one of the two indicators of agency costs. In case an observation did not contain data on SGA expense we searched for it ourselves by looking into the company s annual reports. If the number was found it was added to the data set, if not; the particular observation was excluded from the data set. As a fourth step we looked at the data on EPS which was required to construct one of the two indicators of information costs. Since this data is not publicly published it was not possible to gather this data ourselves. Therefore, in case an observation did not contain data on actual EPS or EPS estimates it was excluded from the data set. 42

43 As a fifth step we looked at missing data on the standard deviation of EPS estimates which was required to construct the other indicator of information costs. It was decided to give those observations for which data on the standard deviation of EPS estimates was missing the firm s average value of this measure. To illustrate: in case the standard deviation of EPS estimates for Heineken NV was unknown in 2000 while values over the period 2001 to 2008 were known we would have taken the average value of the standard deviation of EPS estimates over the period 2001 to 2008 as an estimate of the value in the year As a sixth step we looked at the data on R&D expense. It turned out that a large number of observations did not contain information on R&D expense. This may be due to the fact that many companies included in the data set were not much involved in the development of technology and therefore they did not report on R&D as an individual expense. It was decided to give those observations that did not report on R&D expenditures a value of zero for R&D expense. As a final step we looked at the availability of data on the remaining variables. In case data was missing we tried to look this up ourselves. If the number was found it was added to the data set, if not; the particular observation was excluded from the data set. Figure 4.4.1: Schematic overview of the case selection process 43

44 Testable sample The sample contains a total number of 625 observations. These observations represent 129 different companies. Table gives an overview of the number of observations per industry with companies classified based on Standard Industrial Classification Codes (SIC Codes). Table 4.4.2: Number of observations per industry with companies classified based on SIC codes SIC Code Industry Number of observations Fraction 01xx - 09xx Agriculture, Forestry and Fishing xx - 14xx Mining xx - 17xx Construction xx - 39xx Manufacturing xx - 49xx Transportation, Communication, Electric and Gas xx - 51xx Wholesale trade xx - 59xx Retail trade xx - 69xx Financial service xx - 89xx Service xx - 99xx Public administration 0 0 Total When comparing table with table and looking at the column labeled Fraction it can be seen that, except for the financial service industry, the distribution of firms across industries in the sample is more or less equal to the distribution in the original data set. Hence, we trust that the sample is a homogeneous representation of the original data set. 4.5 Detecting and removing outliers The data on each of the variables incorporated in the empirical models includes a number of extreme cases that have values very different to the rest. These outliers bias the mean and inflate the standard deviation and therefore they should be excluded from the sample (Field 2009). To detect the outlying cases we looked at the z-scores of data on each of the indicators. To convert data into z-scores we took the actual value of a single case and subtracted it by the mean of all cases divided by the standard deviation of all cases. We ignored the positive and negative sign of the z-score so we determined the z-score as an absolute number. After we obtained the absolute value of the z- scores we counted how many cases had a z-score within certain important limits. When the data was normally distributed about 5% should have an absolute value greater than 1.96, and 1% should have an absolute value greater than 2.58 and none of the cases should have a z-value greater than 3.29 (Field 2009). All cases with a z-value greater than 1.5 are considered as outliers and are removed from the sample set. For the empirical models 1 and 2 include we used the same sample set since both models include variables that are derived from similar underlying indicators. Empirical model 3, however, includes variables that are different from the ones included in the models 1 and 2. Hence, for model 3 outliers will be detected and removed by analyzing different variables than the ones we looked at for the models 1 and 2. Table gives a summary of the z-scores and their frequency for the variables incorporated in the empirical models 1 and 2. 44

45 Table 4.5.1: Summary of z-scores and their frequency for variables of the empirical models 1 and 2 Inaccuracy Dispersion Asset utilization SGA expense frequency percent frequency percent frequency percent frequency percent absolute z-score < absolute z-score < absolute z-score > absolute z-score > absolute z-score > missing cases Total Payout ratio Firm size Growth opport. BV cash frequency percent frequency percent frequency percent frequency percent We detected a total number of 115 outliers. After we removed all outliers 510 cases were left which is 81.6% of the initial sample. We used the same approach to detect and remove outliers on the variables incorporated in empirical model 3. After analyzing data on the variables included in Model 3 we found a total number of 127 outliers. After removing these outliers 498 cases were left which is 80.0% of the initial sample. 4.6 Dealing with influential cases Besides looking at outliers it is also necessary to detect and remove influential cases. Influential cases are observations that exert undue influence over the parameters of the model (Field 2009). Influential cases might have undue influence on a predictor s regression coefficient and therefore they have to be removed from the sample set (Field 2009). Influential cases are detected following the approach of Belsley, Kuh and Welsch (1980). Belsley et al. recommended the following: If: With: 1 Covariance ratio of observation i Number of predictors Sample size Then: The case is considered as an influential case and deleting the i th observation will improve the precision of some of the model s parameters. So in each regression analysis we analyzed the covariance ratios of all observations and excluded cases following the recommendations of Belsey et al.. For each of the tests the number of influential cases that we removed was less than

46 5. Descriptive statistics 5.1 Sample description Our sample includes data on securities that represent companies traded at NYSE Euronext Amsterdam Stock Exchange for at least one year. Table presents some aggregated statistics on characteristics of the firms included in our sample. Numbers are given in thousands of euros expect for the numbers related to ratios. Table 5.1.1: Aggregated statistics on characteristics of the firms included in our sample N Minimum Maximum Mean St. dev. Variance Total assets E8 Market capitalization E8 EBIT Cash holdings Leverage ratio* R&D intensity** *Leverage is measured as the ratio of total debt to total equity **R&D intensity is measured as the ratio of R&D expenses to sales revenue (OECD 2005) Table shows that the amount of assets varies among the firms included in this sample. This indicates that the sample contains both small and large firms. The row labeled market capitalization shows that the sample includes companies with high and low market values. The average company has a market value of about 4.9 million euro. Moreover, when looking at the statistics on EBIT it can be observed that the sample includes both companies that are highly profitable and companies that were not profitable at all. The company with the highest profit had an annual net income of about 40 million euro while the company with the highest negative profit made an annual loss of about 2.1 million euro. Furthermore, the table shows that there exists high variance in leverage ratio. This indicates that the companies included in this sample differ in the way they finance their business activities. Companies with a high leverage ratio seem to prefer debt financing instead of equity financing while those with a low leverage ratios prefer to finance their business activities with equity rather than with debt. The row labeled "cash holdings" shows that there exists a substantial difference in the amount of cash held among companies included in our sample. The average company holds about 20 million euro cash. Finally, it is interesting to look at the statistics on R&D intensity. The average of R&D intensity is relatively low but this is because we assumed those companies that did not report on R&D expenditures to have no R&D expense. The table shows that there exists high variance in R&D intensity which indicates that the sample includes firms that rely heavily on R&D efforts and firms for which R&D is less important. 46

47 5.2 Aggregated descriptive statistics of the predictors Table presents some aggregated descriptive statistics of all predictors included in the regression analysis with the empirical models 1 and 2. Numbers are given in thousands of euros expect for the numbers on ratios. Table 5.2.1: Aggregated descriptive statistics of the predictors included in the regression analysis with model 1 and 2 N Minimum Maximum Mean St. dev. Variance Forecast inaccuracy Forecast dispersion Asset utilization SGA expense structure Payout ratio Firm size E8 Growth Table shows some aggregated descriptive statistics of all predictors included in the regression analysis with empirical model 3. Numbers are given in thousands of euros expect for the numbers on ratio s. Table 5.2.2: Aggregated descriptive statistics of the predictors included in the regression analysis with model 3 N Minimum Maximum Mean St. dev. Variance Cash EBIT R&D expense Interest expense Dividends paid Information costs Agency costs Information costs*cash Agency costs*cash Determining the market value of cash holdings The market value of cash holdings was determined from the outcome of a regression analysis with the regression model of Fama and French (1998) and Pinkowitz and Williamson (2003). In this section we do not entirely discuss the framework that we used for the performance of the regression analysis. This framework is outlined in appendix B. Furthermore, in this section we neither present all graphs and figures of the regression output. Instead, we only show the figures that are required to conclude on the market value of cash holdings. All other graphs and figures can be found in appendix C. We detected and removed a total number of 18 influential cases. To check the existence of collinearity between predictors we looked at the collinearity diagnostics table of the regression output. There appeared to exist collinearity between the variables interest 47

48 expense and earnings before interest and tax. However, after we excluded the influential cases the collinearity disappeared. We obtained an R-square value of which means that the regression model is able to explain 58.2 percent of the variation of the market value of equity. The model has a significant F-ratio of which indicates that the model is significantly better at predicting the outcome (i.e. the market value of equity) than using its mean as a best guess. For all variables in the model the VIF values are well below 10 and the tolerance statistics are well above 0.2. Therefore, we can safely conclude that there exist no multicollinearity problems in our data. Figure gives an overview of the regression coefficients. Unstandardized Coefficients Standardized Coefficients Coefficients a 99.0% Confidence Interval for B Correlations Collinearity Statistics Std. Lower Upper Zeroorder Model B Error Beta t Sig. Bound Bound Partial Part Tolerance VIF 1 (Constant) cash EBIT RD interest Div a. Dependent Variable: marketcap1 Figure 5.3.1: Regression coefficients Cash has a b-value of which indicates that as cash increases by one unit the market value of equity increases by units. Because both variables were measured in euro we can say that when all other variables are held constant, every single euro of cash held by a company increases the market value of equity by euro. Thus, investors place a value of euro on every euro held by the company. The t-test for the variable cash gives a significant value of which allows to reject the null hypothesis that says that the variable cash has a b-value of zero. Therefore, we may assume that cash contributes significantly to the model s ability to predict the outcome (i.e. the market value of equity). The confidence intervals of the b-value of cash are constructed such that 99.0% of the cases from the sample contain the true value of b. A good regression model has small confidence intervals indicating that the value of b in the sample is close to the value of b in the population (Field 2009). In this model, the variable cash has relatively tight confidence intervals which indicate that the b-value of cash is likely to be representative for the true population. Finally, we checked the assumptions of the model. A regression model includes a number of assumptions so when a performing a regression analysis it has to be checked whether these assumptions indeed have been met. To check whether the residuals are independent we looked at the Durbin-Watson coefficient. Field (2009) suggests that this coefficient should have a value close to 2 in order to meet the assumption that the residuals are independent. In this analysis we obtained a Durbin-Watson coefficient of which is relatively close to 2 and therefore the assumption has almost certainly been met. 48

49 We plotted a graph of the standardized residuals (ZRESID) against the standardized predicted outcomes (ZPRED) to check the assumptions of linearity and homoscedasticity. The obtained pattern does not indicate non-linearity or heteroscedasticity and therefore the assumptions of linearity and homoscedasticity have almost certainly been met. To test the normality of residuals we looked at the histogram of the standardized residuals. The distribution of the standardized residuals roughly has a bell shaped curve (although there is a small error in the curve around zero) and this proves that the residuals are normally distributed. None of the partial plots shows any obvious outliers which indicates the absence of influential cases and proves the existence of homoscedasticity for each of the independent variables. In summary, because we trust we have met all the assumptions the model is a valid model to determine the market value of cash. We can conclude that investors place a value of euro on every single euro of cash that a company has on its balance sheet. 49

50 6. Results 6.1 Multivariate tests This section presents the results of tests related to hypotheses 1, 2, 3 and 4. Model 1 Table shows the main regression output for model 1. The table reports the b-value (B) and its standardized error (SE B), the standardized regression coefficient ( ), the outcome of the t-test (t-test) and its significance (Sig.). Moreover, for each subsequent hierarchical step of the regression the R- square (R 2 ) and the change in R-square (ΔR 2 ) are given. Table 6.1.1: Main regression output of model 1 B SE B t-test Sig. Step 1 Constant Payout ratio Growth opportunities Step Constant Payout ratio Growth opportunities Forecast inaccuracy Forecast dispersion Step 3 Constant Payout ratio Growth opportunities Forecast inaccuracy Forecast dispersion SGA expense structure Asset utilization R 2 step 1: ΔR 2 step 2: (0.212) ΔR 2 step 3: (0.229) We detected and removed a total number of 9 influential cases. It turned out that the predictors forecast inaccuracy and forecast dispersion were strongly correlated. To solve this problem we removed forecast inaccuracy from the regression model because we found that this variable does not make a significant contribution to the model. All predictors had a VIF value well below 10 and their tolerance statistic was well above 0.2. Therefore, we could safely conclude that there existed no more collinearity among the predictors. We obtained a significant R-square value of which means that the model is able to explain 22.9 percent of the variation in the outcome variable. The covariate variables are able to explain 20.8 percent of the variation in the outcome variable. The positive change of the adjusted R-square ratio after every subsequently regression step indicates that both of the independent variables contribute to the explanatory power of the regression model. The table shows that the covariate variables (i.e. growth opportunities and payout ratio) both make a significant contribution as predictors of the outcome variable. 50

51 Forecast dispersion is one of the two indicators of information costs (see figure 3.3.1). The table shows that the independent variable forecast dispersion is positively related to the market value of cash holdings. This result is in line with what we expected from theory (Myers and Majluf 1984). SGA expense structure is one of the two indicators of agency costs (see figure 3.3.2). The table shows that this indicator does not make a significant contribution to the model. This finding is not as we expected. The other indicator of agency costs is asset utilization. It turned out that asset utilization is negatively related to the market value of cash holdings while we expected this relationship to be positive. While theory suggests that an increase in asset utilization signals a decrease in agency costs and agency costs negatively affect the market value of cash holdings (Jensen 1986) this empirical analysis provides evidence for the existence of an opposite relationship. We checked whether the regression model met all its assumptions and found that they were all met. Model 2 Model 2 includes variables constructed from the indicators that we regressed in the first model. Table shows the main output of this regression analysis. Table 6.1.2: Main regression output of model 2 B SE B t-test Sig. Step 1 Constant Financial constraints Growth opportunities Step 2 Constant Financial constraints Growth opportunities Information costs Step 3 Constant Financial constraints Growth opportunities Information costs Agency costs R 2 step 1: ΔR 2 step 2: (0.180) ΔR 2 step 3: (0.183) We detected and removed a total number of 13 influential cases. There existed no multicollinearity problems. All predictors had a VIF value well below 10 and their tolerance statistic was well above 0.2. We obtained a significant R-square value of The positive change of the adjusted R-square ratio after every subsequently regression step indicates that both of the independent variables contribute to the explanatory power of the regression model. The table shows that both of the covariate variables make a significant contribution as predictors of the outcome variable. 51

52 The table shows that the independent variable information costs make a significant positive contribution as predictor of the market value of cash holdings. This finding is in line with what we expected from theory (Myers and Majluf 1984). This regression analysis indicates that the independent variable agency costs are not a significant predictor of the market value of cash holdings. This findings is in contrast with what we expected since based on theory we expected agency costs to have a negative influence on the market value of cash holdings (Jensen 1986). We checked whether the regression model met all the assumptions we made upfront and found that all the assumptions were met. Model 3 Table shows the main regression output for model 3. Table 6.1.3: Regression output of model 3 B SE B t-test Sig. Step 1 Constant Cash EBIT R&D expense Interest expense Dividends paid Information costs Agency costs Step 2 Constant Cash EBIT R&D expense Interest expense Dividends paid Information costs Agency costs Information costs*cash Step 3 Constant Cash EBIT R&D expense Interest expense Dividends paid Information costs Agency costs Information costs*cash Agency costs*cash R 2 step 1: ΔR 2 step 2: (0.684) ΔR 2 step 3: (0.688) We detected and removed a total number of 8 influential cases. There existed no multicollinearity problems. All predictors had a VIF value well below 10 and their tolerance statistic was well above

53 The model had a significant R-square value of The positive change of the adjusted R-square ratio after every subsequently regression step indicates that both of the independent variables contribute to the explanatory power of the regression model. The table shows that the interaction term of information costs and cash makes a significant positive contribution as a predictor of the market value of a firm which indicates that information costs has a positive influence on the market value of cash holdings. This finding is in line with what we expected from theory (Myers and Majluf 1984). The table shows that the interaction term of agency costs and cash holdings makes a significant positive contribution as a predictor of the market value of a firm which means that agency costs has a positive influence on the market value of cash holdings. However, from theory we expected the relationship between agency costs and the market value of cash holdings to be negative (Jensen 1986). We checked whether the regression model met all its assumptions and found that they were all met. Summary of the results Tests with three empirical models consistently showed that information costs have a significant positive influence on the market value of cash holdings. This provides strong support for our first hypothesis. The empirical models gave mixed results with respect to the influence of agency costs on the market value of cash holdings. While, from theory, we expected agency costs to have a negative influence on the market value of cash holdings model 1 and 3 give some evidence for the existence of an opposite relationship between these variables. For model 2 we find that agency costs does not make a significant contribution as a predictor of the market value of cash holdings. Overall, we can conclude that we did not find support for our second hypothesis. Comparing the regression coefficients obtained for the variable(s) that indicate information costs with the ones obtained for the variable(s) that indicate agency costs allows us to analyze which of the two types of costs has the largest impact on the market value of cash holdings. Since both variables have the same unity we may directly compare their b-values. Tests with all three empirical models showed that the regression coefficient of information costs is larger than the one of agency costs. For model 2 we even find that agency costs does not make a significant contribution as a predictor. Based on these findings we can safely conclude that information costs have a larger impact on the market value of cash holdings than agency costs. Therefore, we find support for hypothesis four but not for hypothesis three. Table summarizes the main regression output related to the first four hypotheses. Table 6.1.4: Summary of test results related to hypotheses 1, 2, 3 and 4 Hypothesis Model 1 Model 2 Model 3 H1: Information costs positively influence the market value of cash holdings support support support H2: Agency costs negatively influence the market value of cash holdings no support no support no support H3: Agency costs have a larger impact on the market value of cash holdings than information costs no support no support no support H4: Information costs have a larger impact on the market value of cash holdings than agency costs support support support 53

54 6.2 Sensitivity tests This section present the results of tests related to hypotheses 5, 6 and 7. For the sensitivity tests we will not use model 3 because this model does not allow to identify the direct influence of factors on the market value of cash holdings since this model does not incorporate the market value of cash holdings as dependent variable. Therefore, for the remaining analysis we will only use the models 1 and 2. Industry effects I Theory suggests that the impact of agency costs and information costs on the market value of cash holdings may differ across industries. It is predicted that agency costs may be more relevant in mature industries (i.e. industries with steady cash flows and relatively few growth opportunities) while information costs may be more relevant in growth industries (i.e. industries with many investment opportunities and where companies typically invest substantial amounts of money in R&D) (Nyborg 2010). Fama and French (1998) argue that relative price to earnings ratio is one of the measures that can be used to indicate whether a firm is active in a growth industry or in a mature industry so we looked at firms price to earnings ratio relative to the sample average to identify whether a firm was active in a growth industry or a mature industry. The 33.3% of firms with the highest relative price to earnings ratio were considered as firms active in growth industries. The 33.3% of firms with the lowest relative price to earnings ratio were considered as firms active in mature industries. Again, model 1 suffered from multicollinearity problems which restricted us to use this model. For model 2 we did not detect any multicollinearity issues. The results obtained with model 2 did not provide evidence for the predictions that information costs are more relevant to firms in growth industries and that agency costs are more relevant to firms in mature industries. A complete overview of the regression results for model 2 can be found in appendix D.1. Industry effects II Additionally, we also tested whether industry sector has an influence on the market value of cash holdings. To check for this we added a dummy variable to the original empirical models which represented the various industry sectors firms were active in. Firms were assigned to a specific industry sector based on their SIC code. Our sample contained seven different industry sectors. Based on the number of observation in each of the industry sectors we added a scaling factor to each of the category labels so that each specific industry sector made an equal contribution to the dummy variable. For both models we found that the dummy variable did not make a significant contribution as a predictor. Therefore, we can conclude that empirical model 1 and 2 are both not sensitive to the industry sectors firms are active in. An overview of the regression results for model 1 and 2 can be found in appendix D.2. Macroeconomic effects Theory suggests that the relation between the market value of cash holdings and its predictors may also be influenced by macroeconomic circumstances (Kleinknecht 2009). To investigate whether the empirical models are sensitive to macroeconomic influences we added a dummy variable to each of the original empirical models which represented the various macroeconomic circumstances during the time span of our research period. We used annual changes in GDP as an indicator of macroeconomic status. Based on the number of observation in each year we added a scaling factor to each of the category labels so that each year made an equal contribution to the dummy variable. 54

55 For both models we found that the dummy variable did not make a significant contribution as a predictor. Therefore, we can conclude that that empirical model 1 and 2 are both not sensitive to macroeconomic circumstances. An overview of the regression results for model 1 and 2 can be found in appendix D.3. Additional experiment with an alternative configuration of the relationships among variables In tests with the original empirical models we did not find the expected relationship between agency costs and the market value of cash holdings. One reason for this might be that we constructed the relationship around the variable agency costs in the wrong way. Based on this thought we came to the idea that information costs and agency costs may be interrelated. Since we only found weak evidence for the existence of a direct relationship between agency costs and the market value of cash holdings (while the evidence for the existence of a direct relationship between information costs and the market value of cash holdings was very strong) we believed that it could be that information costs acts as a moderating variable in the relationship between the variables agency costs and market value of cash holdings. Figure depicts this alternative configuration. In this configuration agency costs and the market value of cash holdings are unrelated to each other unless information costs are present. Figure 6.2.1: Alternative relationship among the variables agency costs, information costs and market value of cash holdings We tested the relevance of this relationship on empirical model 2 and found that there existed strong collinearity between the moderating variable and the independent variables agency costs and information costs. This multicollinearity problem caused the regression results to be biased and unreliable. However, to be able to investigate the relevance of the moderating variable we performed an alternative regression in which we excluded the independent variables agency costs and information costs. The exclusion of these variables solved the multicollinearity issues. We found that the moderating variable makes a significant negative contribution as predictor of the market value of cash holdings. This means that, in case information costs are present, the interaction term of information costs and agency costs is significantly negative related to the market value of cash holdings. A complete overview of these regression results can be found in appendix D.4. Summary of the sensitivity tests Tests on sample splits of growth industries and mature industries did not provide evidence for the prediction that information costs are more relevant to firms in growth industries and that agency costs are more relevant to firms in mature industries. Therefore, we do not find support for hypotheses 5 and 6. We found that empirical models 1 and 2 are not sensitive to the industry sectors firms are active in. 55

56 From tests on the impact of macroeconomic circumstances on the market value of cash holdings we concluded that empirical models 1 and 2 are not sensitive to macroeconomic circumstances. Therefore, we do not find support for hypothesis 7. Table summarizes the main regression output related to hypotheses 5, 6 and 7. Table 6.2.1: Summary of test results related to hypotheses 5, 6 and 7 Hypothesis Model 1 Model 2 Model 3 H5: Information costs are more important than agency costs for firms in growth industries - no support - H6: Agency costs are more important than information costs for firms in mature industries no support no support - H7: Macroeconomic status influences the market value of cash holdings no support no support - Tests with the inclusion of information costs as a moderating variable in the relation between agency costs and the market value of cash holdings showed that the relationship between agency costs and the market value of cash holdings is conditional upon the level of information asymmetry. We found that in case information costs are present, the interaction term of information costs and agency costs makes a significant negative contribution to the market value of cash holdings. 56

57 7. Limitations 7.1 Limitations concerning the conceptual design Quality of the regression model of Fama and French and Pinkowitz and Williamson We assume that the model of Fama and French (1998) and Pinkowitz and Williamson (2003) is the most appropriate model to establish an estimation on the market value of cash holdings for two reasons. First of all, the regression model of Fama and French and Pinkowitz and Williamson obtained an adjusted R-square value of which indicates that their model gives a relatively good fit on the data. Secondly, various other researchers have used this model for valuation purposes (Denis and Sibilkov 2010; Drobetz, Grüninger and Hirschvogl 2010; Pinkowitz, Stulz and Williamson 2006; Schauten, van Dijk and Van der Waal 2008). However, this assumption may be somewhat critical since both studies do not report information on outliers while information on outliers has to be analyzed before one can argue about the quality of the regression model. The influence of corporate financial policy on the market value of cash holdings Faulkender and Wang (2006) investigated that a firm s financial policy does influence the market value of cash holdings. Faulkender and Wang make estimates on a firm s financial policy by analyzing firm-specific characteristics such as financial leverage, growth opportunities and dividend payout ratio. However, this variable will be omitted from our conceptual model because we expect these factors to be highly correlated with the independent variables of our conceptual model, causing multicollinearity problems. Nevertheless, following the efficient market hypothesis developed by Fama (1970), we do believe that investors take into account corporate financial policy as one of the factors on which they base their investment decision since they receive information on this at official shareholder meetings or from publications in financial newspapers, magazines or journals. The influence of tax implications on the market value of cash holdings The influence of tax implications will also be omitted from our conceptual model. The impact of tax on the market value of cash holdings is closely related to a firm s financial policy. If a firm decides to distribute cash as earnings to shareholders the tax will be taxed on a corporate and personal level. If, however, the firm decides to use cash to finance new projects cash will only be taxed at the corporate level and shareholders will not directly be taxed. The influence of tax on the market value of cash holdings, thus, depends heavily on where a company uses its cash for. In addition tax rates differ across countries and therefore investors will value the impact of tax differently. Since we have limited information on a firm s financial policy and tax rates, both on corporate and personal level, differ across countries it is not possible to analyze the influence of tax implications on the market value of cash holdings. Therefore, the influence of tax implications will be omitted from our conceptual model. The operationalization of the variable agency costs Following Singh and Davidson (2003) we used SGA expense structure and asset utilization as indicators of agency costs. However, we believe that the use of SGA expense structure as an indicator of agency costs may be somewhat critical. Firms which business activities rely heavily on marketing are likely to make large advertising expenses which give them a relatively high SGA expenses structure. For these types of firms it is questionable whether high SGA expense structure really signals the existence of agency costs. 57

58 The operationalization of the variable information costs Following DaDalt, Gay and Nam (2002) we used financial analyst forecast inaccuracy and financial analyst forecast dispersion as indicators of information costs. To compute both measures we looked at data on earnings per share at the latest month of every year. So in fact for each observation we took data on earnings per share only for one specific month. It can be questioned if this monthly measure is an accurate representation of the (average) value for the whole year. 7.2 Limitations concerning the methodology As discussed to statistically analyze the financial data we will perform a cross sectional hierarchical multiple linear regression test. This quantitative testing technique brings along some critic which is important to discuss. Theory deficit According to the philosopher David Hume in order to analyze whether one event regularly succeeds another event we have to involve theory about the phenomena, derive hypotheses and do many observations to verify the hypotheses (Schimmelfennig 2010). Hume believes that causality is an entirely theoretical concept that cannot be observed. The researcher can only aspire to observe the regular succession of events. In this research our hypotheses, derived from the conceptual model, predict the regular succession of a number of events. The statistical tests allow one to study the existence of a relationship between these two events. However, once the hypotheses are statistically verified we only know that variable Y regularly succeeds variable X. Why this correlation exists and how this correlation is brought about remains hidden. Therefore, this research might be accused having a theory deficit. Cum ergo propter hoc fallacy Additionally, analyzing causality as a regular succession may also give rise to the so called Cum ergo propter hoc fallacy (Schimmelfennig 2010). This Latin expression means that the fact that variable Y is influenced by variable X does not necessarily imply that Y is caused by X. To illustrate: if we observe that the existence of information costs negatively influences the market value of cash it does not necessarily mean that a low valuation of cash by the market is caused by information costs. Spurious relationship Finally, it is important to realize that there may exist a spurious relationship between some of the variables. A spurious relationship exists when the correlation between variable X and Y does not exist independently of a third variable, let s say Z. To illustrate: it may be that the relation between information costs and the market value of cash holdings does not exist independently of the number of growth opportunities. To check for the existence of spurious relationships, following an iterative process, we performed a number of additional statistical analyses where an intervening variable was added to the conceptual design. 7.3 Limitations concerning the regression results Mathematical configuration of the predicted relationships Our literature study gave us a comprehensive view on factors that may have an influence on the market value of cash holdings. However, although the literature gave the direction of the relationships between these predictors and the market value of cash holdings we did not found any sources that had experimented with the mathematical configuration of the relationships. Therefore, in this research we 58

59 assumed that the relationship between the market value of cash holdings and each of the predictors is linear. However, it might be interesting to test relationships with alternative mathematical configurations, such as: exponential-, quadratic- or logarithmic functions. Explanatory power of the empirical models The explanatory power of the empirical models 1 and 2 is rather low since we obtained R-square values in the range of (for model 2) and (for model 1). A reason for this might be that the empirical models do not include all influential variables. As discussed before, we were not able to incorporate the covariate variable corporate governance into our empirical design because it was not possible to gather data on the indicator anti-takeover provisions. Besides, even though we carefully scanned the literature on factors that have an influence on the market value of cash holdings, some additional influential variables may be missing. Interpretation of the moderating variable We found that information costs acts as a moderating variable in the relationship between agency costs and the market value of cash holdings. Based on the regression coefficient that we obtained for the moderating variable we concluded that in case information costs are present, the interaction term of agency costs and information costs makes a negative significant contribution to the market value of cash holdings. However, evidence for the exact nature of the relationship between these three variables is limited since the statistical analysis performed thus far only tells us in what direction the interaction term (i.e. information costs * agency costs) relates to the dependent variable (i.e. market value of cash holdings.) but the exact underlying structural mathematical equation linking the three variables is still unknown. 59

60 8. Conclusion and recommendations This study aimed to investigate the relevance of the pecking order theory and the agency costs of free cash flow theory in the eyes of the investor by analyzing the impact of information costs and agency costs on the market value of firms cash holdings. In an attempt to answer this issue a longitudinal quantitative study was performed on a sample of firms traded at NYSE Amsterdam Euronext. We performed an extensive literature review to give a clear and logical presentation of the relevant research work done thus far. Based on literature findings we developed a conceptual model. The conceptual model showed the predicted relationships among variables. The operationalization of the variables was established after careful evaluation of suggestions given in the existing literature. Based on the predicted relationships in the conceptual model we developed seven hypotheses which were statistically tested by the performance of various cross sectional multiple hierarchical linear regression analysis. Tests of seven hypotheses and various empirical designs have provided empirical evidence on the question whether investors believe the pecking order theory or the agency costs of free cash flow theory to be the most relevant to firms. Regression analysis performed with three different empirical models consistently showed that information costs make a significant positive influence on the market value of cash holdings. Tests on the influence of agency costs on the market value of cash holdings gave mixed results. For those tests at which agency costs made a significant contribution to the market value of cash holdings its impact appeared to be lower compared to the impact made by information costs. Based on the fact that we found strong evidence for the influence of information costs while the evidence for the influence of agency costs is rather weak and its impact appeared to be lower than that of information costs we believe that information costs is more relevant as a predictor of the market value of cash holdings. Since information costs relate to the pecking order theory we conclude that investors believe the pecking order theory to be more relevant to firms than the agency costs of free cash flow theory. Additional empirical tests indicated that, the classification of firms into growth industries and mature industries and the addition of dummy variables to represent industry sector and macroeconomic circumstances all have no influence on the conclusion that information costs are more relevant as a predictor of the market value of cash holdings. However, tests with the inclusion of a moderating variable showed that besides the direct influence of information costs on the market value of cash holdings it also acts as a moderating variable in the relationship between agency costs and the market value of cash holdings. We found that in case information costs are present, the interaction term of information costs and agency costs makes a significant negative contribution to the market value of cash holdings. Although the statistical analysis performed thus far tells us in what direction the interaction term relates to the dependent variable the exact underlying structural mathematical equation linking the three variables is still unknown. We recommend future researchers to take a closer look at this. In order to avoid misinterpretation of our claim it is important to stress the circumstances in which our claim holds. First of all an important remark has to be made with respect to the relevance of our conclusion across industry sectors. Our conclusion is applicable to firms in those industry sectors that were covered in our analysis. These are: Agriculture, Forestry and Fishing; Mining; Construction; Manufacturing; Transportation, Communication, Electric and Gas; Wholesale trade; Retail trade and Service. Firms active in the financial service industry were excluded from our sample so therefore our 60

61 conclusion does not hold for financial firms. Secondly, it is important to mention that we did not investigate whether the validity of our conclusion depends on the actual level of cash hold by firms. Some believe that there exists an optimal level of cash holdings for firms and agency problems and information asymmetry are particularly relevant for those firms that hold cash in excess of their optimal level (Dittmar and Mahrt-Smith 2007; Drobetz, Grüninger and Hirschvogl 2010). However, since we did not want to take a position on whether there exist optimal levels of cash we did not investigate if the validity of our conclusion depends on the actual level of cash hold by firms. Nevertheless, as a suggestion for further research we believe that it may be interesting to study whether the relevance of both theories is affected by the actual level of cash hold by firms. Finally, it is important to mention that our conclusion is in contrast with what was found in an earlier study by Drobetz et al. (2010). In their paper Drobetz et al. conclude that the agency costs of free cash flow theory is more relevant than the pecking order theory. However, Drobetz et al. measure the relevance of both theories based on the level of information asymmetry that firms experience. Hence, they insist that the agency costs of free cash flow theory and the pecking order theory are both related to adverse selection. In contrast to their view, we believe that only the pecking order theory is related to adverse selection while the agency costs of free cash flow theory does not directly tell us something about the costs that arise as a result of hidden information (i.e. adverse selection); it rather tells us something about costs that arise as a result of hidden actions (i.e. moral hazard). From a scientific point of view we believe that, through direct measurement and comparison of the costs related to both theories, this study provides new evidence in the debate of the relevance of the pecking order theory and the agency costs of free cash flow theory. Moreover, through the performance of statistical tests on different empirical models that incorporates all key variables, this study contributes to debate about factors that influence the market value of cash holdings. From a managerial point of view we believe that this study provides evidence for managers of the firm that cash holdings really add value to their firm. Our results show that a euro of cash is worth euro in the market. When comparing our results with studies which also estimated the market value of cash holdings; it becomes clear that our estimation for the market value of cash holdings is not in range with results obtained in previous studies. Pinkowitz and Williamson (2003), for instance, found that the average market value of a dollar of cash holdings is $0.97 and Dittmar and Mahrt-Smith (2007) found that, on average, a dollar of cash is worth $1.09 to the market. As a possible reason for this substantial difference it might be that the distribution of firms across industries in the samples analyzed by Pinkowitz and Williamson Dittmar and Mahrt-Smith differs from ours. Alternatively, it could be that investors in the U.S. market have a different view on the value of cash than investors in the European market. Since our result differs significantly from what was found by others; managers should not blindly use these absolute numbers in establishing their firm s cash flow policy. Additionally, this study provides managers an answer to the question which factors affect the market value of cash holdings and the relative size of their impact. We found that financial constraints, growth opportunities and information costs all make a significant contribution as predictors of the market value of cash holdings. Tests with empirical model 1 showed that information costs have the largest impact on the market value of cash holdings, followed by growth opportunities and financial constraints. Tests with empirical model 2 showed that growth opportunities costs have the largest impact on the market value of cash holdings, followed by information costs and financial constraints. As a suggestion for future research we recommend to repeat this analysis on a larger sample and over a longer time period. Comparative studies such as the research of Dittmar and Mahrt-Smith (2007), 61

62 Drobetz et al. (2010), Pinkowitz and Williamson (2003) and Pinkowitz et al. (2006) all use samples that consist of more than 5000 observations and most studies base their analysis on a time period much longer than the 10 years that is taken under consideration in our study. Another suggestion for future research is to experiment with different mathematical configurations of the predicted relationships. As discussed earlier, in this research we assume that the relationship between variables is linear while it could be that some variables hold a relationship in a different mathematical configuration, for instance: quadratic, exponential or logarithmic. 62

63 References Aboody, D., and B. Lev, Information asymmetry, R&D, and insider gains, The Journal of Finance 55, Allen, J. W., and G. M. Phillips, Corporate equity ownership, strategic alliances, and product market relationships, The Journal of Finance 55, Almeida, H., M. Campello, and M. Weisbach, The cash flow sensitivity of cash, The Journal of Finance 59, Barclay, M. J., and C. G. Holderness, Control of corporations by active block investors, Journal of Applied Corporate Finance 4, Bates, T., K. Kahle, and R. Stulz, Why do US firms hold so much more cash than they used to?, The Journal of Finance 64, Belsley, D. A., E. Kuh, and R. E. Welsch, Regression diagnostics(wiley Online Library). Bharath, S. T., P. Pasquariello, and G. Wu, Does asymmetric information drive capital structure decisions?, Review of Financial Studies 22, Brennan, M. J., and P. J. Hughes, Stock prices and the supply of information, Journal of Finance, Brennan, M. J., and A. Subrahmanyam, Investment analysis and price formation in securities markets, Journal of Financial Economics 38, D'Mello, R., and M. Miranda, Long-term debt and overinvestment agency problem, Journal of Banking & Finance 34, DaDalt, P., G. Gay, and J. Nam, Asymmetric information and corporate derivatives use, Journal of Futures Markets 22, Denis, D. J., and V. Sibilkov, Financial constraints, investment, and the value of cash holdings, Review of Financial Studies 23, 247. Diamond, D. W., and R. E. Verrecchia, Disclosure, liquidity, and the cost of capital, The Journal of Finance 46, Dittmar, A., Corporate cash policy and how to manage it with stock repurchases, Journal of Applied Corporate Finance 20, Dittmar, A., and J. Mahrt-Smith, Corporate governance and the value of cash holdings, Journal of Financial Economics 83, Drobetz, W., M. C. Grüninger, and S. Hirschvogl, Information asymmetry and the value of cash, Journal of Banking & Finance 34,

64 Fama, E., and K. French, Taxes, financing decisions, and firm value, The Journal of Finance 53, Fama, E. F., Efficient capital markets: A review of theory and empirical work, The Journal of Finance 25, Fama, E. F., and K. R. French, Value versus growth: The international evidence, The Journal of Finance 53, Fama, E. F., and K. R. French, Financing decisions: who issues stock?, Journal of Financial Economics 76, Faulkender, M., and R. Wang, Corporate financial policy and the value of cash, The Journal of Finance 61, Field, A. P., Discovering statistics using SPSS(SAGE publications Ltd). Frank, M., and V. Goyal, Testing the pecking order theory of capital structure, Journal of Financial Economics 67, Goertz, G., Social Science Concepts: a User's Guide. Holderness, C. G., and D. P. Sheehan, Raiders or saviors? The evidence on six controversial investors* 1, Journal of Financial Economics 14, Jensen, M., Agency costs of free cash flow, corporate finance, and takeovers, The American Economic Review 76, Kleinknecht, A., Macro-Economics part I, Course material: The Economic Foundations of the Firm (MOT1420) - Delft University of Technology. Kleinknecht, A., Macro-Economics part II, Course material: The Economic Foundations of the Firm (MOT1420) - Delft University of Technology. Leary, M. T., and M. R. Roberts, The pecking order, debt capacity, and information asymmetry, Journal of Financial Economics 95, Lemmon, M. L., and J. F. Zender, Debt capacity and tests of capital structure theories, Journal of Financial and Quantitative Analysis, Myers, S. C., and N. S. Majluf, Corporate financing and investment decisions when firms have information that investors do not have* 1, Journal of Financial Economics 13, Nohel, T., and V. Tarhan, Share repurchases and firm performance: new evidence on the agency costs of free cash flow, Journal of Financial Economics 49, Nyborg, K. J., Capital structure: Information asymmetry and agency costs, Course material: Advanced Corporate Finance I - University of Zurich. 64

65 OECD, Guidelines for collecting and interpreting data on innovation, OECD Publishing 3rd. Opler, T., L. Pinkowitz, R. Stulz, and R. Williamson, The determinants and implications of corporate cash holdings, Working paper. Pinkowitz, L., R. Stulz, and R. Williamson, Does the Contribution of Corporate Cash Holdings and Dividends to Firm Value Depend on Governance? A Cross country Analysis, The Journal of Finance 61, Pinkowitz, L., and R. Williamson, What is a dollar worth? The market value of cash holdings, Working paper. Schauten, M., D. van Dijk, and J. Van der Waal, Corporate governance and the value of excess cash holdings of large European firms, Working paper Erasmus University Rotterdam. Schimmelfennig, F., Module 2: Causality and Causual mechanisms, Course material: Research Methods I - ETH Zurich. Sekaran, U., Research methods for business: A skill building approach, JOHN WILEY & SONS, INC., Shome, D. K., and S. Singh, Firm value and external blockholdings, Financial Management 24, Shyam-Sunder, L., and S. C. Myers, Testing static tradeoff against pecking order models of capital structure, Journal of Financial Economics 51, Singh, M., and W. N. Davidson III, Agency costs, ownership structure and corporate governance mechanisms, Journal of Banking & Finance 27,

66 Appendices 66

67 Appendix A Literature review: Summaries of the most relevant articles Index 1. Aboody, D., and B. Lev, Information asymmetry, R&D, and insider gains, The Journal of Finance 55, Aghion, P., S. Bond, A. Klemm, and I. Marinescu, Technology and financial structure: are innovative firms different?, Journal of the European Economic Association 2, Almeida, H., M. Campello, and M. Weisbach, The cash flow sensitivity of cash, The Journal of Finance 59, Ang, J. S., R. A. Cole, and J. W. Lin, Agency costs and ownership structure, The Journal of Finance 55, Bates, T., K. Kahle, and R. Stulz, Why do US firms hold so much more cash than they used to?, The Journal of Finance 64, Bharath, S. T., P. Pasquariello, and G. Wu, Does asymmetric information drive capital structure decisions?, Review of Financial Studies 22, Brown, J. R., and B. C. Petersen, Cash holdings and R&D smoothing, Journal of Corporate Finance. 8. Chan, L. K. C., J. Lakonishok, and T. Sougiannis, The stock market valuation of research and development expenditures, The Journal of Finance 56, D'Mello, R., and M. Miranda, Long-term debt and overinvestment agency problem, Journal of Banking & Finance 34, Denis, D. J., and V. Sibilkov, Financial constraints, investment, and the value of cash holdings, Review of Financial Studies 23, Dittmar, A., and J. Mahrt-Smith, Corporate governance and the value of cash holdings, Journal of Financial Economics 83, Drobetz, W., M. C. Grüninger, and S. Hirschvogl, Information asymmetry and the value of cash, Journal of Banking & Finance 34, Fama, E., and K. French, Taxes, financing decisions, and firm value, The Journal of Finance 53, Fama, E. F., and K. R. French, Financing decisions: who issues stock?, Journal of Financial Economics 76, Faulkender, M., and R. Wang, Corporate financial policy and the value of cash, The Journal of Finance 61, Frank, M., and V. Goyal, Testing the pecking order theory of capital structure, Journal of Financial Economics 67, Hall, B. H., and J. Lerner, The financing of R&D and innovation, NBER working paper. 18. Jensen, M. C., and W. H. Meckling, Theory of the firm: Managerial behavior, agency costs and ownership structure, Journal of Financial Economics 3, Jensen, M., Agency costs of free cash flow, corporate finance, and takeovers, The American Economic Review 76, Leary, M. T., and M. R. Roberts, The pecking order, debt capacity, and information asymmetry, Journal of Financial Economics 95, Lemmon, M. L., and J. F. Zender, Debt capacity and tests of capital structure theories, Journal of Financial and Quantitative Analysis,

68 22. Myers, S. C., and N. S. Majluf, Corporate financing and investment decisions when firms have information that investors do not have* 1, Journal of Financial Economics 13, Nohel, T., and V. Tarhan, Share repurchases and firm performance: new evidence on the agency costs of free cash flow, Journal of Financial Economics 49, O'Brien, J. P., and T. B. Folta, A transaction cost perspective on why, how, and when cash impacts firm performance, Managerial and Decision Economics 30, Opler, T., L. Pinkowitz, R. Stulz, and R. Williamson, The determinants and implications of corporate cash holdings, Working paper. 26. Pinkowitz, L., R. Stulz, and R. Williamson, Does the Contribution of Corporate Cash Holdings and Dividends to Firm Value Depend on Governance? A Cross country Analysis, The Journal of Finance 61, Pinkowitz, L., and R. Williamson, What is a dollar worth? The market value of cash holdings, Working paper. 28. Singh, M., and W. N. Davidson III, Agency costs, ownership structure and corporate governance mechanisms, Journal of Banking & Finance 27, Schauten, M., D. van Dijk, and J. Van der Waal, Corporate governance and the value of excess cash holdings of large European firms, Working paper Erasmus University Rotterdam. 30. Shyam-Sunder, L., and S. C. Myers, Testing static tradeoff against pecking order models of capital structure, Journal of Financial Economics 51,

69 Literature review strategy Literature collection A literature review was conducted to gain familiarity with the research topic. Moreover, the literature review helped to gather all information needed to develop the conceptual model. Literature is a secondary source of information, as it already exists and does not have to be collected by the researcher (Sekaran 2003). Academic search engines such as Wiley Interscience, Google Scholar and ScienceDirect were used to access articles and publications from top journals in the fields of finance, economics and business management. A major advantage of academic search engines is that they give access to a wide range of publications. Besides, it is a relatively fast way to find useful articles of high quality. When looking for articles, attention was paid to the year of publication and the journal impact factor (also called: citation index). Literature analysis Initially, literature was analyzed following the snowball method, where literature references are scanned for useful publications. Once we finished the scanning process the most relevant articles were selected and analyzed in more detail. For each of these articles general information was listed in a table. In order to obtain deep understanding on what the articles were about and what information was useful for our own research for each article we tried to find the answer to the following six quetsions: 1. What has been studied/examined/researched or investigated? 2. Which variables are important and how are these defined? 3. What research method is been used? 4. What are the outcomes/findings of the research (details / numbers)? 5. What has been concluded based on these outcomes/findings? 6. What are the limitations of the particular research and are there any recommendations given for further research? 69

70 Results 1. Information asymmetry, R&D, and insider gains Citations (impact factor) 471 Author Aboody, D., and B. Lev Year of publication 2000 Journal The Journal of Finance Key words Innovative firms, R&D intensity, Information asymmetry Research question How do R&D expenditures and insider gains influence the level of information asymmetry? Data All sale transactions made by insiders and reported to the SEC over the period 1985 to 1997 Key findings R&D is a major contributor to information asymmetry. Recommendations - Analysis Aboody and Lev study how R&D expenditures and insider gains influence the level of information asymmetry. Aboody and Lev expect R&D intensive firms to face higher levels of information asymmetry. They argue that many R&D projects, such as radically new drugs under development or software programs, are unique to the developing firm, whereas most capital investments, such as commercial property or airplanes, share common characteristics across firms within an industry. Consequently, investors can derive little or no information about the productivity and value of a firm's R&D from observing the R&D performance of other firms. Moreover, they believe that while most physical and financial assets are traded in organized markets, where prices convey information about asset productivity and values, there are no organized markets for R&D and hence no asset prices from which to derive information. The level of information asymmetry is indicated based on insider gains and by analyzing the public disclosure of insider trades. To analyze the relation between R&D, insider gains and information asymmetry Aboody and Lev construct four monthly portfolios: (1) Firms engaged in R&D whose insiders were "net purchasers" of shares in a given month during the period from 1985 to 1997 (a net purchaser firm-month is one where the number of shares purchased by the firm's officers during the month exceeded the number of shares sold); (2) firms engaged in R&D whose insiders were "net sellers" of shares (i.e., number of shares sold by insiders during the month exceeded purchases); (3 firms without R&D whose insiders were net purchasers; and (4) firms without R&D whose insiders were net sellers of shares. For each of the four portfolios Aboody and Lev calculate stock returns and the total volume of shares traded. Aboody and Lev find that insider gains in firms conducting R&D are substantially larger than insider gains in firms with no R&D activities. These differences in insider gains are both statistically and economically significant, and hold after controlling for various known risk factors. Moreover they find that investors' reaction to the public disclosure of insiders' trade is stronger for R&D firms than for No-R&D firms, corroborating their hypothesis that R&D activities enhance information asymmetry, and that this asymmetry is not eliminated by insiders' trade and investors' information search. 70

71 Based on these findings Aboody and Lev conclude that R&D is a major contributor to information asymmetry. Aboody and Lev do not provide any recommendations for further research. 2. Technology and financial structure: are innovative firms different? Citations (impact factor) 66 Author Aghion, P., S. Bond, A. Klemm, and I. Marinescu Year of publication 2004 Journal Journal of the European Economic Association Key words Innovative firms, R&D intensity, Capital structure Research question Does the capital structure of R&D intensive companies differ from that of non R&D intensive companies? Data UK public companies listed on the London Stock Exchange over the period Key findings The capital structure of firms with R&D activities differs from that of firms with no R&D activities. Firms with R&D activities rely more on equity financing. However, firms with high R&D intensity issue more debt than those with low R&D intensity. Recommendations Use an alternative indicator of R&D intensity. Use a different data set. Analysis Aghion, Bond, Klemm, Marinescu investigate whether financing choices differ systematically with R&D intensity. More specifically, their aim is to explore whether innovative firms make different financing choices than less innovative firms. Aghion et al. follow the OECD Frascati Manual for classification of R&D intensity. They measure R&D intensity as R&D expenditure relative to sales revenues. The extent of debt financing is measured as total debt financing relative to total assets. Equity financing is measured as the total amount of equity outstanding relative to total assets. Aghion et al. perform a regression analysis to analyze the relation between financing choices and R&D intensity. The dependent variables are: debt issued and equity issued. The independent variable of interest is: R&D intensity. The control variables are: number of employees, sales growth and profitability. Aghion et al. find that on average firms with R&D activities use 20% less debt financing than firms with no R&D activities. Furthermore, they find that firms with high R&D intensity use more debt financing than firms with low R&D intensity. Based on their findings Aghion et al. conclude that the capital structure of firms with R&D activities differs from that of firms with no R&D activities. Firms with R&D activities rely more on equity financing. However, firms with high R&D intensity issue more debt than those with low R&D intensity. 71

72 As an idea for further research Aghion et al. recommend to use different indicators to indicate a firm s R&D intensity. As a limitation of their research Aghion et al. have only considered large and mature public companies form the UK. Additional research on a sample of smaller firms outside the UK might lead to additional insights on the relation between firms innovativeness and their financing choices. 3. The cash flow sensitivity of cash Citations (impact factor) 537 Author Almeida, H., M. Campello, and M. Weisbach Year of publication 2004 Journal The Journal of Finance Key words Cash holdings, Financial constraints Research question Are cash holdings more valuable for constrained firms than for unconstrained firms? Data U.S. manufacturing firms over the period Key findings Cash holdings are more valuable for constrained firms than for unconstrained firms. Recommendations Use cash flow sensitivity as a variable in research related to agency costs, financial constraints and financing policy. Analysis Almeida, Campello and Weisbach investigate to which extent firms save cash out of cash inflows (referred to as cash flow sensitivity). Furthermore, they study whether the cash flow sensitivity of cash provides an empirical useful measure of financing constraints. Almeida et al. believe that firms that are unconstrained will not display a systematic propensity to save cash, while constrained firms are expected to have positive cash flow sensitivity. Almeida et al. define cash holdings as the total amount of cash holdings and marketable securities relative to total assets. They define cash flow as earnings before extraordinary items and depreciation relative to total assets. A firm s level of financial constraints is measured based on: payout policy, asset size, bond ratings and commercial paper ratings. Almeida et al. perform a regression analysis to analyze cash flow sensitivity for subsamples of constrained and unconstrained firms. They find that the cash flow sensitivity of cash is close to and not statistically different from zero for the unconstrained firms, but positive and significantly different from zero for the constrained firms. Based on these finding they conclude that cash holdings are more valuable for constrained firms than for unconstrained firms. Almeida et al. argue their research has proven that cash flow sensitivity of cash is a useful variable in empirical research. Therefore, they recommend using cash flow sensitivity as a variable in research related to agency costs, financial constraints and financing policy. 72

73 4. Agency Costs and Ownership Structure Citations (impact factor) 551 Author Ang, J. S., R. A. Cole, and J. W. Lin Year of publication 2000 Journal The Journal of Finance Key words Agency costs, Agency costs of free cash flow theory, Corporate governance Research question How to measure agency costs and is the agency costs of free cash flow theory as developed by Jensen and Meckling applicable in practice? Data Small U.S. business that participated in the National Survey of Small Business Finances in 1997 Key findings Agency costs are found to be: (1) significantly higher when an outsider rather than an insider manages the firm, (2) inversely related to the manager s ownership share, (3) increasing with the number of non-manager shareholders, (4) to a lesser extent, lower with greater monitoring by banks. Recommendations - Analysis Ang, Cole and Lin establish two measures of agency costs and investigate whether the agency costs of free cash flow theory as developed by Jensen and Meckling (1976) is applicable to their data sample. To measure agency costs of the firm, Ang et al. use two alternative efficiency ratios: (1) the expense ratio measured as operating expense relative to annual sales and (2) the asset utilization ratio measured as annual sales relative to total assets. The first ratio is a measure of how effectively the firm s management controls operating costs, including excessive perquisite consumption, and other direct agency costs. The second ratio is a measure of how effectively the firm s management deploys its assets. In contrast to the expense ratio, agency costs are inversely related to the sales-to-asset ratio. A firm whose sales-to-asset ratio is lower than the base case firm experiences positive agency cost. These costs arise because the manager i) makes poor investment decisions, ii) exerts insufficient effort, resulting in lower revenues; iii) consumes executive perquisites, so that the firm purchases unproductive assets, such as excessively fancy office space, office furnishing, automobiles, and resort properties. Both ratios are compared to a base case. The base case is a firm that is owned and managed by one and the same person. Ang et al. control for differences across industries, the effects of economies of scale, and differences in capital structure. Ang et al. perform a regression analysis to examine how agency costs vary with the separation of ownership and control, i.e., whether the firm s manager is a shareholder or an outsider with no ownership stake. Ang et al. find that both the median and average ratios of operating expenses to annual sales are considerably higher for firms managed by outsiders than for firms owned by shareholders. Moreover, they find that the sales-to-asset ratios are higher in all categories of shareholder-managed firms versus outsider-managed firms. 73

74 Based on these findings Ang et al. conclude that both the ratio of operating expenses to annual sales and the ratio of annual sales to total assets are adequate proxies for small corporations agency costs. Moreover, the results of Ang et al. provide direct confirmation of the predictions made by Jensen and Meckling (1976). Agency costs increase with a reduction in managerial ownership, as predicted by Jensen and Meckling. These results hold true after controlling for differences across industries, the effects of economies of scale, and differences in capital structure. They also find some evidence that delegated monitoring of small firms by banks reduces agency costs. Ang et al. do not provide any recommendations for further research. 5. Why do US firms hold so much more cash than they used to? Citations (impact factor) 168 Author Bates, T., K. Kahle, and R. Stulz Year of publication 2009 Journal The Journal of Finance Key words Cash holdings, Excess cash Research question Why do U.S. firms hold so much more cash than they used to? Data Industrial U.S. firms over the period Key findings Companies hold more cash because: cash flows have become riskier, firms hold fewer inventories and receivables and firms are increasingly R&D intensive. Furthermore, increased R&D expenditures require firms to hold a greater cash buffer against future shocks. Recommendations Be aware of the growing importance of cash and they recommend to take this fact into account when evaluating the financial condition and assessing the capital structure decisions of firms. Analysis Bates, Kahle, and Stulz study why the amounts of cash hold by companies have increased over time. Bates et al. calculate that the average cash to asset ratio has more than doubled over their sample period, from 10.5% in 1980 to 23.2% in Bates et al. define cash flow as EBITDA (Earnings before Interest Taxes, Depreciation and Amortization) minus interest minus taxes minus dividends. They define cash ratio as the total amount of cash and marketable securities relative to total assets. Bates et al. perform a regression analysis on different subsamples, using a regression model that was developed by Opler, Pinkowitz, Stulz, and Williamson (1999). First they split their sample into small and large size firms to investigate whether the amount of cash holdings depends on firm size. Secondly, they split their sample into new listed and old listed firms to determine whether the increase in cash holdings is closely related to the disappearing dividends and new listings phenomena as documented by Fama and French (2001, 2004). Thirdly, they split their sample based on firm s idiosyncratic risk to investigate whether the increase in cash holdings can be related to idiosyncratic risks. Finally, they analyze how firm specific characteristics could have caused firms to hold more cash. 74

75 Bates et al. find that the increase in cash holdings is concentrated among firms that do not pay dividends, firms in more recent IPO listing cohorts, and firms in industries that experience the greatest increase in idiosyncratic volatility. Furthermore, they find that the main reasons for the increase in cash ratio are that: inventories have fallen, cash flow risk for firms has increased, capital expenditures have fallen, and R&D expenditures have increased. Bates et al. argue that R&D activities typically involve large risks and uncertainties and therefore increased R&D expenditures require firms to hold a greater cash buffer to protect them against future shocks. Based on their findings Bates et al. conclude that cash holdings have increased because: cash flows have become riskier, firms hold fewer inventories and receivables and firms are increasingly R&D intensive. Bates et al. argue that researchers should be aware of the growing importance of cash and they recommend to take this fact into account when evaluating the financial condition and assessing the capital structure decisions of firms. 6. Does asymmetric information drive capital structure decisions? Citations (impact factor) 65 Author Bharath, S. T., P. Pasquariello, and G. Wu Year of publication 2009 Journal Review of Financial Studies Key words Information asymmetry, Pecking order theory, Capital structure Research question Is information asymmetry an important determinant of capital structure decisions, as suggested by the pecking order theory? Data U.S. firms over the period Key findings Firms prefer to access capital markets and issue equity when the level of information asymmetry is low. Recommendations - Analysis Bharath, Pasquariello, Wu test whether information asymmetry is an important determinant of capital structure decisions, as suggested by the pecking order theory. Bharath et al. develop a novel information asymmetry index to measure the level of information asymmetry. This information asymmetry index is based on the common cross-sectional variation of either the level of or the annual change in four direct measures of information asymmetry (the adverse selection component of both quoted and effective bid-ask spreads, the probability of informed trading, and the relation between daily volume and first-order return autocorrelation) and three broader measures of market liquidity (the liquidity ratio, the illiquidity ratio, and the reversal coefficient). Barath et al. follow the model developed by Shyam-Sunder and Myers (1999) to test the pecking order theory. They partition their sample into low information asymmetry and high information asymmetry subsamples. 75

76 Barath et al. find that firms with a low information asymmetry index issue 30 to 60% less debt than firms with a high information asymmetry index. Based on this finding they conclude that firms prefer to access capital markets and issue equity when the level of information asymmetry is low. Bates et al do not give any recommendations for further research nor do they discuss the limitations of their research. 7. Cash holdings and R&D smoothing Citations (impact factor) 10 Author Brown, J. R., and B. C. Petersen Year of publication 2010 Journal Journal of Corporate Finance Key words Capital structure, R&D smoothing, Financial constraints Research question Do firms use their cash reserves to smooth their R&D expenditures? Data Publicly traded manufacturing firms in U.S. manufacturing over the time period Key findings Young firms and firms with high level of financial constraints use their cash reserves to smooth R&D activities. Recommendations Repeat this research using a different data set. Analysis Brown and Petersen examine whether firms use cash reserves to smooth their R&D expenditures. Brown and Petersen measure change in cash holdings, with cash holdings being the total amount of cash, to indicate whether cash holdings have been used to smooth R&D expenditures. R&D expenditure is measured based on the total amount of R&D expenditures in a particular year. To establish a measure of financial constraints Brown and Petersen follow Almeida, Campello and Weisbach (2004) who measure financial constraints based on payout ratio, firm size, and the absence/presence of a bond rating. Brown and Petersen perform a regression analysis to investigate whether firms use cash reserves to smooth R&D expenditures. They divide their sample into positive R&D and no R&D subsamples based on whether the firm reports positive R&D expenditures in a given sample period. Brown and Petersen argue the positive R&D sample is useful for the primary plots, summary statistics and regressions. The no R&D sample is not useful for directly testing the importance of R&D smoothing with cash holdings, but it is valuable for understanding how the level and variability of cash holdings differs across firms. Next they divide firms into young and mature categories based on the number of years since the firm first appears in COMPUSTAT. To account for periods of high and low R&D investment they also divide the sample into three periods: , and because they expect that cash holdings will be most important for R&D smoothing. Additionally, they sort firms into various groups based on the likelihood they face binding financing constraints. 76

77 Brown and Petersen find that approximately 75% of the younger firms used cash holdings to dampen the volatility in R&D. Overall, the results for , , and the narrow window support an interpretation that firms facing financing constraints actively manage their cash holdings to smooth R&D expenditures while firms with low financial constraints appear to smooth R&D expenditures without the use of cash holdings. Based on these findings Brown and Petersen conclude that young firms and firms with high level of financial constraints use their cash reserves to smooth R&D activities. Brown and Petersen believe that they are the first who analyze whether firms use cash reserves to smooth their R&D expenditures. Since their research focuses only on a sample of public US firms they recommend to repeat this research using a different sample set. 8. The stock market valuation of research and development expenditures Citations (impact factor) 550 Author Chan, L. K. C., J. Lakonishok, and T. Sougiannis Year of publication 2001 Journal The Journal of Finance Key words Capital structure, R&D valuation Research question Do stock prices fully reflect the value of firms' intangible assets, especially research and development efforts? Data All U.S. common stocks listed on the New York and American Stock Exchanges and on Nasdaq over the period Key findings Returns to investors of firms that do invest in R&D are not higher than returns to investor of firms that do not invest in R&D and therefore stocks fully reflect the value of intangible assets. However, firms with high R&D intensity seem to generate excess returns. Recommendations Analyze why it could be that firms with high R&D intensity generate excess returns. Analysis Chan, Lakonishok, and Sougiannis examine whether stock prices fully reflect the value of firms' intangible assets, especially research and development efforts (R&D efforts). Chan et al. believe that the market value of a firm should ultimately reflect the value of all its assets. When most of the assets are physical, such as plant and equipment, the link between asset values and stock prices is relatively apparent. In modern economies, however, a large part of a firm's value may reflect its intangible assets, such as brand names. Under generally accepted U.S. accounting principles, many types of intangible assets are not reported in firms' financial statements. When a firm has large amounts of such intangibles, the lack of accounting information generally complicates the task of equity valuation. Chan et al. express R&D spending as the total annual R&D expenditures relative to sales, net income, total dividends or book value of equity. Each firm's R&D capital is estimated from its past history of R&D expenditures as follows:

78 Effectively Chan et al. assume that the productivity of each dollar of spending declines linearly by 20% per year. Chan et al. use two measures to indicate the level of R&D intensity, respectively: R&D expenditures relative to sales and R&D expenditure relative to the market value of equity. To measure excess return each stock is matched with a control portfolio of stocks based on size and book-to market and then past three-year return. The difference is calculated between the stock's annual buyand-hold return and the return on the control portfolio. Chan et al. use descriptive statistics to analyze whether firms that do invest in R&D generate higher returns than firms that do not invest in R&D. Additionally; they use descriptive statistics to analyze whether high R&D intensity leads to excess returns. Chan et al. argue that it is possible that any excess returns earned by R&D-intensive stocks reflect risk differentials or firm specific characteristics. Therefore, by incorporating Fama and French (1993, 1996) multifactor model they adjust their analysis for risk sensitivities. To account for firm specific characteristics they incorporate Fama and MacBeth (1973) cross-sectional regressions. Chan et al. do not find a direct link between R&D spending and future stock returns. They find that stocks doing R&D have an average return of 19.65% per year, and stocks doing no R&D have an average return of % per year. For firms engaged in R&D Chan et al. find that firms with high R&D intensity, where R&D intensity is measured as R&D expenditures relative to the market value of equity, obtain an average excess return of 6.12 % per year. However, when using R&D expenditure relative to sales as a measure of R&D intensity, R&D intensive firms do not obtain any excess return. After the performance of a risk-adjusted analysis they find that the returns on the portfolio with high R&D intensity relative to market are not entirely due to increases in risk. After controlling for firm specific characteristics Chan et al. find that large returns on the portfolio with high R&D intensity relative to market are independent of firm size and growth opportunities. Based on the finding that returns to investors of firms that do invest in R&D are not higher than returns to investor of firms that do not invest in R&D Chan et al. conclude that stocks fully reflect the value of intangible assets such as R&D. However, they find some evidence that firms with high R&D intensity generate excess returns. Chan et al. do not explain why it could be that firms with high R&D intensity generate excess returns. They recommend further research to seek for an explanation. 78

79 9. Long-term debt and overinvestment agency problem Citations (impact factor) 1 Author D'Mello, R., and M. Miranda Year of publication 2010 Journal Journal of Banking & Finance Key words Agency costs of free cash flow theory, Capital structure Research question Does debt issuing affect overinvestments and agency costs? Data Firms that have started to issue debt between 1968 and 2001 and that have been unlevered at least for three years immediately preceding the offering. Key findings Issuing debt leads to a reduction in abnormal capital expenditures. This decline in overinvestments is explained by debt service obligations that reduce discretionary funds under managerial control. Reduction in overinvestments also has a positive impact on equity value. Recommendations - Analysis D'Mello and Miranda investigate the role of long-term debt in influencing overinvestments by analyzing the pattern of abnormal investments around a new debt offering by unlevered firms. They believe that before being levered when the disciplining role of debt is missing, firms retain excessive amounts of cash. A firm is assumed to issue debt if the change in book value of total long-term debt exceeds five percent of the pre-issue book value of assets. Their test requires firms to be unlevered for an extended period before the offering and therefore they restrict their sample to firms that issue debt after having no long-term debt for at least three consecutive years. Overinvestment is defined as having cash and capital expenditure ratio s higher than the industry average. D'Mello and Miranda determine whether firms systematically overinvest by comparing the cash and capital expenditure ratios (liquidity ratio s) of sample firms in a given year to those of median industry firms in that year. D'Mello and Miranda find that the liquidity ratio of the median unlevered sample firm exceeds its benchmark by 20% in the interval before the debt offering. Once firms start to issue debt overinvestment declines and within three years of the offering the sample firms liquidity ratios are similar to their industry benchmarks. D'Mello and Miranda argue that the decline in overinvestments, after debt issuing, is significantly related to debt service obligations. Furthermore, they find that a reduction in cash overinvestments leads to higher equity market values. Based on these findings D'Mello and Miranda conclude that issuing debt leads to a reduction in abnormal capital expenditures. This decline in overinvestments is explained by debt service obligations that reduce discretionary funds under managerial control. Reduction in overinvestments also has a positive impact on equity value. D'Mello and Miranda do not give any recommendations for further research. 79

80 10. Financial constraints, investment, and the value of cash holdings Citations (impact factor) 40 Author Denis, D. J., and V. Sibilkov Year of publication 2010 Journal Review of Financial Studie Key words Financial constraints, MV of cash holdings, Growth opportunities Research question Are cash holdings more valuable to firms with low financial constraints than firms with high financial constraints? And, why may this be the case? Data U.S. public companies with financial data available on COMPUSTAT over the period 1985 to Firms are required to have at least $25 million in total book assets in 1994 dollars. Companies in the financial and utility industries are excluded. Key findings Investors place a premium on the value of cash holdings of constrained firms because these cash holdings allow constrained firms to undertake value-increasing projects that might otherwise be bypassed. Recommendations - Analysis Several studies have reported that cash holdings are more valuable for firms with large financial constraints. Denis and Sibilkov examine why this may be the case? A firm s level of financial constraints is measured based on dividend payout ratio, firm size, debt rating and paper rating and KZ-index. Following Lang, Ofek, and Stulz (1996), they measure investment expenditures as net investment, defined as capital expenditures net of depreciation. Denis et al. follow the regression model of Fama and French (1998) for their analysis but like Pinkowitz, Stulz, and Williamson (2006), they modify the Fama-French specification by splitting the total change in assets into its cash and non-cash components. Similar to earlier studies they first find that constrained firms hold substantially more cash than unconstrained firms. These findings are consistent with two interpretations. Greater cash holdings might be more valuable to constrained firms because they allow the firm to invest when other sources of funds are costly, limited, or unavailable. In other words, greater cash holdings allow firms that experience external financial constraints to avoid underinvestment and reduced growth. Alternatively, however, it might be the case that the higher value associated with greater cash holdings is a reflection of the market rewarding the firm for holding cash rather than overinvesting that cash in unprofitable projects. To distinguish between these alternative interpretations, Denis and Sibilkov examine whether greater cash holdings are associated with greater investment and whether this association is stronger for constrained than for unconstrained firms. Denis and Sibilkov find that firms with higher cash holdings invest more regardless of whether they are constrained or unconstrained. Although these findings are consistent with the idea that cash holdings allow constrained firms to increase investment, the results do not explain the higher value of cash for constrained firms, since the effect of cash on 80

81 investment is insignificantly different between constrained and unconstrained firms. Therefore, they also analyze whether the profitability of investment is different for constrained and unconstrained firms. They find that the marginal value of investment is greater for financially constrained than for unconstrained firms. The significantly greater impact of investment on value for constrained firms implies that, although greater cash holdings are associated with greater investment for both constrained and unconstrained firms, the financially constrained firms appear to have more valuable marginal investment opportunities. Based on these findings Denis and Sibilkov conclude that investors place a premium on the value of cash holdings of constrained firms because these cash holdings allow constrained firms to undertake value-increasing projects that might otherwise be bypassed. Moreover, they find that the marginal value from these investments if higher for constrained firms. Denis and Sibilkov do not give any recommendations for further research. 11. Corporate governance and the value of cash holdings Citations (impact factor) 269 Author Dittmar, A., and J. Mahrt-Smith Year of publication 2007 Journal Journal of Financial Economics Key words Agency costs of free cash flow theory, Corporate governance, MV of cash holdings, Moral hazard Research question Do investors place a lower value on the cash holdings of firms that are poorly governed? Data US publicly traded firms from 1990 to 2003 for which the required data items are available on COMPUSTAT. Key findings Investors place a lower value on the cash holdings of firms that are poorly governed and this conveys investors expectation of firm value to be lost from the waste of cash reserves. Recommendations - Analysis Dittmar, and Mahrt-Smith investigate how corporate governance impacts firm value by comparing the value and use of cash holdings in poorly and well-governed firms. Corporate governance is measured based on investor oversight by large institutional shareholders and managerial entrenchment resulting from antitakeover provisions. The degree of managerial entrenchment due to takeover protection is indicated by two measures. The first measure is the Gompers, Ishii, and Metrick (2003) corporate governance index, which measures the number of antitakeover provisions in a firm s charter and in the legal code of the state in which the firm is incorporated. The data for the index is assembled and reported about every two years by the Investor Responsibility Research Center (IRRC) and the index varies between zero and 24. As a second measure, they replace the Gompers, Ishii, and Metrick index with the index developed in Bebchuk, Cohen, and Ferrell (2005). This index is based on the same raw data as the Gompers, Ishii, and Metrick (2003) corporate governance index, but uses only six provisions that Bebchuk, Cohen, and Ferrell show have the greatest impact on firm. In their analysis, Dittmar, and Mahrt-Smith distinguish 81

82 between cash holdings and excess cash where cash holdings is the total amount of cash internally available to the firm and excess cash are cash reserves held in excess of those needed for operations and investments. Dittmar, and Mahrt-Smith follow the regression model of Fama and French (1998) and Faulkender and Wang (2005) to determine how a change in cash holdings leads to a change in the market valuation of a firm. They partition their sample into poor governed and well governed subsamples. For their analysis on operating performance they split up the sample into firms with large excess cash and poor governance and firms with low excess cash and good governance. Dittmar, and Mahrt-Smith find that a dollar of cash is worth approximately $1.09 on average. However, according to their analysis, this value can decrease to $0.42 in case a firm is poorly governed. Alternatively, a dollar in a well-governed firm is worth as much as $1.62. They also find that the value of excess cash of a well monitored firm is 40% greater than that of a lesser monitored firm. Additionally, they find that firms with both high excess cash and poor governance subsequently experience particularly low operating performance: the accounting returns of a sample of firms that draw down their large excess cash reserves is significantly diminished if the firms have poor corporate governance. This negative impact of excess cash on operating performance is cancelled out if the firm is well governed. Based on these findings Dittmar, and Mahrt-Smith conclude that investors place a lower value on the cash holdings of firms that are poorly governed. They argue that this conveys investors expectation of firm value to be lost from the waste of cash reserves. Dittmar and Mahrt-Smith argue that their paper does not attempt to comprehensively examine all reasons for firms to hold cash; they rather focus on the cost of holding excess cash for poorly governed firms. They do not provide any recommendations for further research. 12. Information asymmetry and the value of cash Citations (impact factor) 3 Author Drobetz, W., M. C. Grüninger, and S. Hirschvogl Year of publication 2010 Journal Journal of Banking & Finance Key words MV of cash holdings, Information asymmetry, Pecking order theory, Agency costs of free cash flow theory Research question Does there exists a relationship between the value of cash holdings and the extend of information asymmetry? Data All firms from the different countries for which I/B/E/S provides analysts forecasts and for which we can retrieve company data from Worldscope from the period from 1995 to Key findings The market value of cash significantly reduces when a firm faces a higher level of information asymmetry; therefore agency costs have a higher impact on the market value of cash holdings than information costs. Recommendations - 82

83 Analysis Drobetz, Grüninger, and Hirschvogl investigate whether there exists a relationship between the value of cash holdings and the extend of information asymmetry. Their aim is to determine the relevance of the pecking order theory and agency costs of free cash flow theory. Drobetz et al. give the following definitions of both theories in relation to information asymmetry: according to the pecking order theory, adverse selection problems make external financing costly and imply a higher market value of a marginal dollar of cash in states with higher information asymmetry. Drobetz et al. argue that the free cash flow theory, in contrast, predicts that excessive cash holdings bundled with higher information asymmetry generate moral hazard problems and lead to a lower market value of a marginal dollar of cash. Drobetz et al. measure the extent of information asymmetry based on dispersion of analysts forecasts an indicator a developed by Krishnaswami and Subramaniam (1999). This indicator measures the standard deviation of earnings per share forecasts across analysts that cover a firm. Greater disagreement among analysts indicates a higher level of information asymmetry. In their analysis Drobetz et al. control for financial constraints and corporate governance. To indicate a firm s level of financial constraints they use two measures at the country-level (stock market capitalization to GDP and private bond market capitalization to GDP) and one variable at the firm-level (firm size). To indicate the level of corporate governance they use four measures at the country-level: anti-director rights index, rule of law index, corruption index, and legal system classification and one variable at the firm-level: percentage of closely held shares. Drobetz et al. use the regression model developed by Fama and French (1998) to perform their analysis. They extend this model by incorporating the variable information asymmetry. Drobetz et al. divide their sample into several subgroups to test whether financial constraints and corporate governance structures have an impact on how information asymmetry influences the value of cash. Drobetz et al. find that the market value of cash significantly reduces when a firm faces a higher level of information asymmetry. Secondly, the find that given high information asymmetry, the market value of cash is higher if investor protection is better and the quality of corporate governance is higher. The findings regarding the influence of financial constraints are not quit clear since the indicators used generate contradictory results. While the firms size indicator gives a higher value of cash holdings for small firms (large financial constraints) the stock market capitalization indicator gives a lower value of cash holdings for a low stock market capitalization value (which indicates the opposite (large financial constraints). Drobetz et al., thus, find that the market value of cash significantly reduces when a firm faces a higher level of information asymmetry. Based on this Drobetz et al. conclude that agency costs have a higher impact on the market value of cash holdings than information costs. Even though Drobetz et al. find that the agency costs of free cash flow outweigh the benefits of cash as an internal source of finance they do not find evidence for the Pecking order theory being inapplicable to firms. Drobetz et al. do not provide any recommendations for further research. 83

84 13. Taxes, financing decisions, and firm value Citations (impact factor) 285 Author Fama, E. F., and K. R. French Year of publication 1998 Journal The Journal of Finance Key words Firm value, Firm characteristics Research question How is firm value related to dividend payments and debt? Data Firms with data available on all variables in COMPUSTAT for the period 1965 to Key findings Firm value is positively related to dividends and negatively related to debt. Recommendations Apply the valuation model also on different set of sample data to investigate whether it still gives a good estimation of a firm s market value. Analysis Fama and French (1998) study how firm value is related to dividend payments and debt. More specifically, they aim to investigate how taxes on dividends and debt influence firm value. They develop a regression model that allows to estimate firm value as function of a number of firm specific characteristics. The dependent variable in the regression model of Fama and French is the spread of value over cost, where is the total market value of a firm and is the book value of its assets. Fama and French also attempt to explain the two year change in the spread,. The independent variables include past, current and future values of dividends, interest, earnings, investment and R&D expenditures. Fama and French incorporate past and future values because they believe that investors base the value of a firm not only on the firm s current status but also on past returns and expectations about future earnings. Similar to Fama and MacBeth (1973), Fama and French base their statistical inferences (standard errors) on the average slopes from the regression estimated separately for each year of their sample period. Fama and French argue that the main advantage of this approach is that the year by year variation in the slopes includes the effects of estimation error due to the cross-correlation of the residuals for individual firms. Moreover, they argue that these standard errors are derived from larger samples (an average of about 2400 firms per annual regression) which increase the accuracy of the slopes and reduce their year by year variance. Fama and French mention that using as dependent variable might create problems. They argue that the results are likely to be dominated by the largest firms and heteroskedasticity (when the random variables have different variances) is likely to cloud the inferences. Moreover, they argue that the two year spread probably faces similar problems. In order to avoid these errors Fama and French scale both the dependent and independent variables by total book assets. 84

85 The regression model is: is the current value of the particular variable. is the compact notation for the two year past change,. is the compact notation for the two year future change. Where: = Market value of a firm = Earnings before interest and extraordinary items but after tax and depreciation. = Book value of assets = R&D expenditures = Interest expense = Total dividends paid * To simplify the notation the firm subscript that should appear on all variables in the regression and the year subscript that should appear on all regression coefficients are omitted in the equation. Fama and French find that firm value is positively related to dividends and negatively related to debt. Fama and French recommend to apply their valuation model also on different set of sample data to investigate whether it still gives a good estimation of a firm s market value. 14. Financing decisions: who issues stock? Citations (impact factor) 227 Author Fama, E. F., and K. R. French Year of publication 2005 Journal Journal of Financial Economics Key words Pecking order theory, Information asymmetry Research question Do firms follow the pecking order theory? Data NYSE, AMEX, and Nasdaq firms over the period is 1973 to 2002 for which the required data items are available on COMPUSTAT. Financial firms and utilities are excluded Key findings Firms violate the pecking order theory. Recommendations Specify the conditions under which the pecking order theory holds. 85

86 Analysis According to the pecking order theory investments are financed first with retained earnings, then debt, and with equity only as a last resort. Fama and French empirical examine whether this theory holds. Fama and French describe seven ways for a firm to issue equity: (1) mergers via an exchange of stock, (2) employee stock options, grants, and other employee benefit plans, (3) subscription rights issued to stockholders, (4) warrants attached to other securities, (5) convertible bonds, (6) dividend reinvestment and other direct purchase plans, and (7) private placements. To analyze the relevance of the pecking order theory Fama and French follow the pecking order model developed by Shyam-Sunder and Myers (1999). To analyze annual equity issues they develop a model that consists of two measures. The first is the change in the book value of stockholders equity in excess of the change in COMPUSTAT s adjusted balance sheet retained earnings (referred to as dsb). The second is the change in the number of shares outstanding over the fiscal year times the average of the stock prices at the beginning and end of the fiscal year (referred to as dsm). This model incorporates all seven ways a firm could issue equity. Fama and French find that equity issues are common place. During 1973 to 1982, 54% of the sample firms make net equity issues each year, issuing to 62% for 1983 to 1992 and 72% for 1993 to The fractions of firms making gross equity issues are much higher, 67%, 74%, and 86%. Moreover they find that during 1983 to 1992 and 1993 to 2002, the net equity issues of net issuers among small firms are on average larger than their net new issues of debt. The net equity issues of big firms that are net issuers are about one-third the size of their net debt issues during 1983 to 1992, and they are on the order of net debt issues during 1993 to Based on their findings Fama and French conclude that during more than half of their sample firms violate the pecking order theory. They argue that one story for their result is that the pecking order breaks down at least in part because there are ways to issue equity with low transactions costs and modest asymmetric information problems. Three of the alternatives to issue equity (issues to employees, rights issues, and direct purchase plans) seem to have both low transactions costs and minor asymmetric information problems, and a fourth (mergers financed with stock) may fall into this category. Fama and French clarify that at least during the last ten years of the sample period, mergers are important in explaining the magnitude of equity issues. Furthermore, they believe that issues to employees are probably important in explaining both the magnitude and frequency of issues throughout the sample period. Fama and French argue that the Pecking order theory has large problems when applied on a general set of data. They believe that further research should not focus on trying to empirically prove the generalizability of the theory but researchers should rather try to specify the conditions under which the theory holds. 86

87 15. Corporate financial policy and the value of cash Citations (impact factor) 202 Author Faulkender, M., and R. Wang Year of publication 2006 Journal The Journal of Finance Key words Corporate financial policy, Financial constraints, MV of cash holdings Research question Does a firm s financial policy affect the market value of cash holdings? Data U.S. public companies with financial data available on COMPUSTAT over the period 1971 to Companies in the financial and utility industries are excluded. Key findings The marginal value of cash declines with larger cash holdings, higher leverage, better access to capital markets, and as firms choose greater cash distribution via dividends rather than repurchases. Recommendations Repeat this analysis using a different data set. Analysis Faulkender and Wang study how a firm s financial policy might affect the market value of cash holdings. They argue that the value of cash to shareholders varies considerably depending upon whether cash is more likely to be used to: (1) distribute earnings to shareholders via dividend payments or share repurchases, (2) service debt or other liabilities or (3) finance positive NPV projects decreasing the amount of money that needs to be raised in the capital market. For firms whose cash reserves appear to greatly exceed their needs in the foreseeable future, an additional dollar of cash reserves is more likely to be distributed to equity holders through dividends and/or stock repurchases. However, because of personal taxes on dividends and capital gains only the fraction (1- ) ends up in the hands of shareholders. Alternatively, if firms use their cash to pay down debt or other liabilities, cash will increase debt value and not equity value. Thus, the equity market (shareholders) will place a lower value on cash for high leveraged firms relative to the value of cash for a firm with little debt. In contrast, for those firms that have positive NPV investment opportunities investors will place a premium on a firm s cash holdings because the availability of cash avoids that firms have to access the capital market where a transaction costs have to be paid. Faulkender and Wang indicate a firm s level of financial constraints based on payout ratio, firm size, long term bond rating and commercial paper rating. In contrast to the methodology of Fama and French, Faulkender and Wang use a methodology that is based on excess equity return rather than the difference between market and book value. Following Grinblatt and Moskowitz (2004) and Daniel and Titman (1997), their dependent variable is a stock s excess return over the fiscal year, which is defined to be stock s excess return stock i s return during fiscal year t less the return of stock i s benchmark portfolio during fiscal year t. Benchmark portfolios are designed to offset the expected return component of stock i due to its size and market-to book ratio at the beginning of the fiscal year. In their model Faulkender and Wang regress excess return on changes in firm characteristics (cash holdings, earnings, R&D expenditures, dividend payment, leverage and net financing.), focusing on the estimated coefficient that corresponds to the variable 87

88 measuring the ratio of the unexpected change in cash to the firm s equity value. Since both the dependent and independent variables are standardized by the previous year market value of equity,, the coefficient measures the dollar change in shareholder value resulting from a one dollar change in the amount of cash held by the firm. Faulkender and Wang argue that their model is an improvement on the model developed by Fama and French (1998). The regression model is: Where: = excess return = Market value of cash holdings = Earnings before interest and extraordinary items but after tax and depreciation. = net assets = total assets cash = Book value of assets = R&D expenditures = Interest expense = Total dividends paid = The previous year market value of cash holdings = Net financing= total equity issued total share repurchases + total debt issued - debt redemption = Leverage captures the effect of changes in the market value of cash for different levels of cash holdings and captures the effect of leverage on the marginal value of cash holdings. * To simplify the notation the firm subscript that should appear on all variables in the regression and the year subscript that should appear on all regression coefficients are omitted in the equation. The regression coefficient can be interpreted as the change in firm value associated with a one dollar change in cash. To reduce the effect of outliers Faulkender and Wang trim their sample at 1% by dropping 0.5% in each tail of each variable. The model of Faulkender and Wang has an adjusted R- square value of 0.20 where the adjusted R-square value is the average of the adjusted R-square value of 26 single cross sectional regressions. Faulkender and Wang find that the average marginal value of cash across all firms is $0.94. This suggests that less than the full value of an additional dollar of cash is incorporated in the stock price, consistent with the idea that shareholders value cash at it s after tax value. Furthermore, they find that for a firm with no leverage and cash holdings equal to 5% of their equity market capitalization, the value of an additional dollar of cash is $1.43 relative to $1.36 for an otherwise equivalent firm with cash holdings equivalent to 15% of the value of their equity. Thus, as firms cash levels and leverage increase, the marginal value of cash decreases significantly. Faulkender and Wang also find that for those firms that distribute cash, the marginal value of cash is $0.13 higher if they do so by stock 88

89 repurchase rather than by dividend payments. They argue that this number is consistent with a dividend tax rate that is 13% higher than the capital gains tax rate on repurchases for the marginal shareholder. Finally, the find that the average marginal value of cash for those firms that are likely to have more difficulty accessing capital is significantly higher than for those firms with less financial constraints. Shareholders of a mean firm that is classified as financially constrained, with payout ratio as indicator for financial constraints, place a value of $1.04 on an extra dollar of cash, while shareholders of the mean financially unconstrained firm only place a value of $0.77 on an extra dollar of cash. The difference in the marginal value of cash between constrained firms and unconstrained firms is especially large among those firms that appear to have valuable investment opportunities but low levels of internal funds. Based on these findings Faulkender and Wang conclude that the marginal value of cash declines with larger cash holdings, higher leverage, better access to capital markets, and as firms choose greater cash distribution via dividends rather than repurchases. In their research Faulkender and Wang focus on public U.S. firms only. In order to investigate whether their valuation model is also applicable on other samples they recommend further research to repeat their analysis using a different set of sample data. 16. Testing the pecking order theory of capital structure Citations (impact factor) 621 Author Frank, M., and V. Goyal Year of publication 2003 Journal Journal of Financial Economics Key words Pecking order theory, Information asymmetry, Firm size, Time Research question Do firms follow the pecking order theory? Data U.S. public companies with financial data available on COMPUSTAT over the period 1971 to Companies in the financial and utility industries are excluded. Key findings Large firms generally follow the pecking order theory while small firms especially does analyzed in later years do violate the pecking order theory. Recommendations - Analysis Frank and Goyal investigate if companies do follow the pecking order theory and whether this depends on firm size and time. Firm size is measured as total assets which is the sum of book debt plus market equity. For their empirical analysis Frank and Goyal follow the pecking order model developed by Shyam- Sunder and Myers (1999). They partition their sample into large firm and small firm subsamples and perform a year by year cross sectional analysis. Frank and Goyal find that the greatest support for the pecking order is found among large firms in earlier years. They investigate that over time, support for the pecking order declines for two reasons. 89

90 More small firms are publicly traded during the 1980s and 1990s than during the 1970s. Since small firms do not follow the pecking order, the overall average moves further from the pecking order. Based on their findings Frank and Goyal argue, however, that the time period effect is not entirely due to more small firms in the 1990s. Even when attention is restricted to the largest quartile of firms, support for the pecking order theory declines over time. Based on their findings Frank and Goyal conclude that large firms generally follow the pecking order theory while small firms especially those analyzed in later years do violate the pecking order theory. Frank and Goyal do not provide any recommendations for further research. 17. The financing of R&D and innovation Citations (impact factor) 37 Author Hall, B. H., and J. Lerner Year of publication 2009 Journal NBER Working paper Key words R&D, Capital structure Research question How do firms R&D efforts affect capital structure? Data No empirical analysis Key findings R&D intensive firms systematically face a higher cost of capital. Recommendations - Analysis Hall and Lerner study how R&D efforts affect capital structure. Hall and Lerner conclude, based on findings in previous literature, that R&D intensive firms systematically face a higher cost of capital. As a reason for a higher cost of capital for R&D intensive firms Hall and Lerner argue that banks and other debt holders prefer to use physical assets to secure loans and that this makes debt holders reluctant to lend money to firm s when the project involves substantial R&D investment rather than investment in plant and equipment. Additionally, they also argue that servicing debt usually requires a stable source of cash flow, which makes it more difficult to find the funds for an R&D investment program that must be sustained at a certain level in order to be productive. 90

91 18. Theory of the firm: Managerial behavior, agency costs and ownership structure Citations (impact factor) Author Jensen, M. C., and W. H. Meckling Year of publication 1976 Journal Journal of financial economics Key words Agency costs of free cash flow theory, Agency costs Research question What are agency costs? Data No empirical analysis. Key findings Agency costs are costs from: (1) the monitoring expenditures by the principal, (2) the bonding expenditures by the agent and (3) the residual loss. Recommendations - Analysis Jensen and Meckling define the concept of agency costs. Jensen and Meckling define an agency relationship as a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent. If both parties to the relationship aim to maximize their utility, there is good reason to believe that the agent will not always act in the best interests of the principal. The principal can limit divergences from his interest by establishing appropriate incentives for the agent and by incurring monitoring costs designed to limit the aberrant activities of the agent. In addition in some situations it will pay the agent to expend resources (bonding costs) to guarantee that he will not take certain actions which would harm the principal or to ensure that the principal will be compensated if he does take such actions. Jensen and Meckling argue, however that it is generally impossible for the principal or the agent at zero cost to ensure that the agent will make optimal decisions from the principal s viewpoint. In most agency relationships the principal and the agent will incur positive monitoring and bonding costs (non-pecuniary as well as pecuniary), and in addition there will be some divergence between the agent s decisions and those decisions which would maximize the welfare of the principal. The dollar equivalent of the reduction in welfare experienced by the principal as a result of this divergence is also a cost of the agency relationship, and Jensen and Meckling refer to this latter cost as the residual loss. Jensen and Meckling define agency costs as the sum of: (1) the monitoring expenditures by the principal, (2) the bonding expenditures by the agent and (3) the residual loss. 91

92 19. Agency costs of free cash flow, corporate finance, and takeovers Citations (impact factor) 9549 Author Jensen, M. Year of publication 1986 Journal The American Economic Review Key words Agency costs of free cash flow theory, Agency costs Research question How do agency costs affect cash flows, corporate finance and takeovers? Data No empirical analysis. Key findings Free cash flow is cash flow in excess of that required to fund all projects that have positive net present values when discounted at the relevant cost of capital. Conflicts of interest between shareholders and managers over payout policies are especially severe when the organization generates substantial free cash flow. Recommendations Test the empirically relevance of the agency costs of free cash flow theory. Analysis Jensen develops the agency costs of free cash flow theory. According to Jensen free cash flow is cash flow in excess of that required to fund all projects that have positive net present values when discounted at the relevant cost of capital. Jensen discusses that conflicts of interest between shareholders and managers over payout policies are especially severe when the organization generates substantial free cash flow. Moreover he argues that the problem is how to motivate managers to disgorge the cash rather than investing it at below the cost of capital or wasting it on organization inefficiencies. The theory developed by Jensen explains that debt reduces the agency costs of free cash flow by reducing the cash flow available for spending at the discretion of managers. Furthermore, issuing large amounts of debt to buy back stock also sets up the required organizational incentives to motivate managers and to help them overcome normal organizational resistance to retrenchment which the payout of free cash flow often requires. The threat caused by failure to make debt service payments serves as an effective motivating force to make such organizations more efficient. Jensen recommends that the agency costs of free cash flow should be empirically tested in order to assess its practical relevance. 92

93 20. The pecking order, debt capacity, and information asymmetry Citations (impact factor) 66 Author Leary, M. T., and M. R. Roberts Year of publication 2010 Journal Journal of Financial Economics Key words Pecking order theory, Information asymmetry Research question Which theory does best explain a firm s financing decisions? Data U.S. public companies with financial data available on COMPUSTAT over the period 1980 to Companies in the financial and utility industries are excluded. Key findings The empirical relevance of the pecking order depends crucially on one s interpretation of the hypothesis and the empirical model used. Recommendations - Analysis Leary and Roberts analyze the empirical relevance of the pecking order theory and analyze and discuss various interpretations of the pecking order hypothesis. Equity issuance is defined as the issuance of common and preferred stock, net of repurchases, during period t, divided by total assets. Debt issuance is defined as a change in total debt (long-term plus short-term) from year t -1 to t divided by total assets in year t -1 in excess of 5%. Leary and Roberts analyze the robustness of the pecking order theory by using various empirical models. The baseline empirical model, developed by Shyam-Sunder and Myers (1999), holds a relatively strict interpretation of the pecking order theory. It requires firms to maintain constant cash reservoirs and debt capacities while adhering to the pecking order s financing hierarchy. Besides an analysis with the baseline model Leary and Roberts also perform an analysis with an extended version of the baseline model. This model holds a more liberal interpretation of the pecking order theory because it incorporates the assumption that firms may issue debt in a given year up to the point where their leverage ratio is equal to that of an average investment grade rated firm in the same industry and during the same year. Finally, Leary and Roberts also perform an analysis that is even more liberal because it allows firms debt capacities to vary with factors typically attributable to alternative theories than the pecking order theory. From the analysis performed with the baseline model Leary and Roberts find that 77% of their sample firms follow the pecking order in choosing between internal and external finance, but only 17% follow the pecking order in choosing between debt and equity. From the analysis performed with the more liberal model Leary and Roberts find that fewer than 20% of firms adhere to the pecking order s prediction for debt and equity issuances. Only when the using the most liberal model Leary and Roberts find that the pecking order accurately classifies the debt equity decisions of 48% of the sample firms. Based on their findings Leary and Roberts conclude that the empirical relevance of the pecking order depends crucially on one s interpretation of the hypothesis and the empirical model used. 93

94 Leary and Roberts do not provide any recommendations for further research. 21. Debt capacity and tests of capital structure theories Citations (impact factor) 171 Author Lemmon, M. L., and J. F. Zender Year of publication 2006 Journal Journal of Financial and Quantitative Analysis Key words Pecking order theory, Financial distress Research question Does a modified version of the pecking order theory better explain a firm s financing decision? Data U.S. public companies with financial data available on COMPUSTAT over the period 1971 to Key findings A modified version of the pecking order theory, which takes into account financial distress costs, is a good descriptor of firms financing behavior. Recommendations Extend the current analysis by incorporating the possible relation between growth and profitability and a firm s financing decision. Analysis Lemmon and Zender examine the impact of incorporating a measure of debt capacity in empirical models of the pecking order theory. Debt capacity was originally defined by Myers (1977) as the point at which an increase in the use of debt reduces the total market value of the firm s debt. Lemmon and Zender use a predictive model of whether a firm has rated debt outstanding as the primary indication of the extent of a given firm s debt capacity. Financing deficit is defined as the growth in assets less growth in current liabilities less growth in retained earnings. Lemmon and Zender develop a modified version of the pecking order theory model of Shyam- Sunder and Myers (1999). The model of Lemmon et al. differs from the Shyam- Sunder and Myers model in the sense that it separately examines firms that are expected to be constrained by concerns over debt capacity and those that are not. Moreover, the model of Lemmon and Zender also includes an additional independent variable; the square of the financing deficit. This leads to the following regression equation: Where is the net debt issued by firm i in period t, is the corresponding financing deficit and is an error term, is the y-as intercept, is the slope coefficient on the financing deficit the sign indicates the nature of the financing hierarchy. Lemmon and Zender predict that for firms that follow the pecking order and are unconstrained by concerns over debt capacity, the equation should perform very well (a coefficient estimate near 1 and a high R-squared). For pecking order firms that are constrained by concerns over debt capacity the equation should result in an estimate of the coefficient that is negative and significant. Lemmon and Zender find that for the median firm the slope coefficient on the financing deficit is 0.53, indicating that small deficits are financed by about half debt and half equity. The coefficient 94

95 estimate on the squared deficit is -0.24, indicating that firms relies much more heavily on equity financing when deficits are large. The results are consistent with the predictions of the pecking order theory in the presence of concerns about debt capacity. Therefore, Lemmon and Zender conclude that a modified version of the pecking order theory, which takes into account financial distress costs, is a good descriptor of firms financing behavior. Lemmon and Zender assume that asset growth and profitability are exogenous to the financing decision in this analysis. Theoretically, with perfect markets, this would be correct. However, in real world markets are imperfect. Therefore, Lemmon and Zender argue that further research should extend the current analysis by incorporating the possible relation between growth and profitability and a firm s financing decision. 22. Corporate financing and investment decisions when firms have informationthat investors do not have Citations (impact factor) 6847 Author Myers, S., and N. Majluf Year of publication 1984 Journal Journal of Financial Economics Key words Pecking order theory, Information asymmetry Research question Do firms have any preferences regarding the funds they use to finance their business activities? Data No empirical analysis. Key findings Firms prefer to raise capital in the following order: internal, debt, equity. Recommendations Test the empirically relevance of the pecking order theory. Analysis Myers and Majluf develop the pecking order theory. The pecking order theory is based on the principle of asymmetric information and they believe that a firm s management acts in the interests of passive, old stockholders. The theoretical model that is derived by Myers and Majluf explains several aspects of corporate behavior, including the tendency to rely on internal sources of funds and to prefer debt to equity if external financing is required. Based on this theoretical model of Myers and Majluf generate the following predictions: (1) it is generally better to issue safe securities than risky ones. Firms should go to bond markets for external capital, but raise equity by retention if possible. That is, external financing using debt is better than financing by equity. (2) Firms whose investment opportunities outstrip operating cash flows, and which have used up their ability to issue low-risk debt, may forego good investments rather than issue risky securities to finance them. This is done in the existing stockholders interest. (3) Firms can build up financial slack (The sum of cash on hand and marketable securities) by restricting dividends when investment requirements are modest. The cash saved is held as marketable securities or reserve borrowing power. (4) The firm should not pay a dividend if it has to recoup the cash by selling stock or some other risky security. (5) When managers have superior information, and stock is issued to 95

96 finance investment, stock price will fall, other things equal. This action is nevertheless in the (existing) stockholders interest. If the firm issues safe (default-risk-free) debt to finance investment, stock price will not fall. Myers and Majluf recommend that the pecking order theory should be empirically tested in order to assess its practical relevance. 23. Share repurchases and firm performance: new evidence on the agency costs of free cash flow Citations (impact factor) 244 Author Nohel, T., and V. Tarhan Year of publication 1998 Journal Journal of Financial Economics Key words Agency costs of free cash flow theory, Signaling, Moral hazard Research question Do share repurchases influence operating performance? And, why could this be the case? Data All tender offer stock repurchases announced between 1978 and 1991 listed in the appendix of Comment and Jarrell (1991), and the Wall Street Journal Index. Key findings Operating performance improves after share repurchases. This improved performance does not result from higher growth opportunities but from the more efficient employment of assets, supporting the agency costs of free cash flow theory. Recommendations - Analysis Nohel and Tarhan study post-announcement industry-adjusted performance of repurchasing firms to differentiate between the information signaling theory and the agency costs of free cash flow theory. According to the information signaling theory a firm s intention in announcing a repurchase is to signal to outsiders that the performance prospects are improving. Alternatively, according to the free cash flow theory a firm s intention in announcing a repurchase is to distribute free cash flow to shareholders which decreases the risk of potential agency costs from bad management behavior of the firm. Nohel and Tarhan argue that if a firm s intention in announcing a repurchase is to signal to outsiders that the performance prospects are improving, then we should see an improvement in operating performance following the repurchase, relative to what was expected. Alternatively, if the intention of management is to distribute free cash flow, which decreases the risk of potential agency costs, investors should react to the repurchase announcement but the repurchasing firms may or may not exhibit improved performance. Performance is measured as the ratio of cash flow to the market value of the assets that generate the cash flow. Nohel and Tarhan examine the post-announcement industry-adjusted performance of repurchasing firms. They perform a year by year cross sectional regression analysis to determine post-repurchase performance on pre-repurchase performance. Nohel and Tarhan sort their sample into subsamples according to a firm s growth opportunity, using Tobin s q as an indicator of growth opportunity 96

97 because they believe that high growth firms and low growth firms distribute cash for different reasons. They argue that high-growth firms may be using the repurchase to signal improving investment opportunities, while low-growth firms may be distributing excess cash. Nohel and Tarhan find that there is a significant improvement in the performance of repurchasing firms, relative to a set of control firms. The operating performance of the median low-growth firm is 23.30% larger relative to a set of control firms. The operating performance of the median high-growth firm is only 1.94% higher relative to a set of control firms. Based on these findings Nohel and Tarhan conclude that operating performance improves after share repurchases. They argue that this improved performance does not result from higher growth opportunities but from the more efficient employment of assets, supporting the agency costs of free cash flow theory. Nohel and Tarhan do not provide any recommendations for further research. 24. A transaction cost perspective on why, how, and when cash impacts firm performance Citations (impact factor) 2 Author O'Brien, J. P., and T. B. Folta Year of publication 2009 Journal Managerial and Decision Economics Key words Firm performance, Corporate Governance, Firm characteristics Research question How do cash holdings, R&D intensity, monitoring quality and environmental uncertainty affect firm performance? Data Firms listed in COMPUSTAT between the years 1991 and 2003 that had sales and assets of at least one million dollars and also had information available on the market value of their equity. Financial and utility industries are excluded. Key findings Firm performance increases with increased cash holdings and increased R&D intensity while monitoring and environmental uncertainty have a negative impact on performance. There exists a positive and significant three way interaction between cash, R&D intensity and uncertainty. There also exists a positive and significant three way interaction between cash, monitoring and uncertainty. Recommendations Repeat this research using a different data set. Analysis O'Brien and Folta analyse how cash holdings, R&D intensity, monitoring quality and environmental uncertainty affect firm performance. Besides their empirical study O'Brien and Folta provide a good description of the difference between the transaction costs theory and agency costs theory. They explain that the transaction costs theory and the agency costs theory are highly complementary theoretical perspectives (Williamson, 1988). As both are primarily concerned with the proper application of managerial discretion, both consider incentives and contract structures, and both view the board of directors as a control instrument that monitors managers. Where the two differ, however, is in the basic unit of analysis. For the agency costs theory, the basic unit is the individual agent. For 97

98 the transaction costs theory, in contrast, the basic unit of analysis is the transaction. In terms of corporate governance, the focal transaction is the money invested in the firm. O'Brien and Folta measure firm performance based on Tobin s q. They argue that this measure is the most appropriate because it incorporates expectations regarding future firm performance and future growth opportunities. O'Brien and Folta define cash as the magnitude of cash holdings measured based on total cash and short-term investments divided by total assets. R&D was defined as the intensity of investment and measured based on total R&D expenditures divided by total assets. O'Brien and Folta constructed an indicator to measure the level of monitoring by giving firms a score from zero to three based on the number of characteristics linked to good monitoring that exhibited. These characteristics included: there is are one or more outside block holders who own at least 5% of the firm; more than 60% of the board of directors is comprised of outsiders; and the roles of CEO and Chairman of the Board are separated. O'Brien and Folta indicate the level of environmental uncertainty based on the average stock volatility of the firm s industry. O'Brien and Folta also incorporate an indicator that controls for fixed firm effects in their analysis. O'Brien and Folta perform a year by year cross sectional regression to analyze their predictions. They split up their sample into subsamples of high and low cash holdings, high and low R&D intensity, high and low uncertainty and good and poor monitoring. O'Brien and Folta find that increased cash holdings lead to higher firm performance. Secondly, they find that monitoring has negative impact on firm performance. Thirdly, they find that increased R&D efforts lead to higher firm performance. Fourthly, they find that uncertainty has a negative impact on firm performance. Fifthly, they find a positive and significant three way interaction between cash, R&D intensity and uncertainty. Finally, they also find a positive and significant three way interaction between cash, monitoring and uncertainty Based on these findings O'Brien and Folta conclude that firm performance increases with increased cash holdings and increased R&D intensity while monitoring and environmental uncertainty have a negative impact on performance. As a recommendation O Brien and Folta argue this research should be repeated using a different indicator for firm performance. 98

99 25. The determinants and implications of corporate cash holdings Citations (impact factor) 718 Author Opler, T., L. Pinkowitz, R. Stulz, and R. Williamson Year of publication 1997 Journal Working paper Key words MV of cash holdings, Growth opportunities, Firm size, Riskiness, Financial constraints Research question What are the determinants and implications of corporate cash holdings? Data U.S. public companies with financial data available on COMPUSTAT over the period 1971 to Companies in the financial and utility industries are excluded. Key findings Firms with strong growth opportunities and riskier cash flows hold relatively large amounts of cash. Firms that have the greatest access to the capital markets, such as large firms and those with high credit ratings, tend to hold small amounts of cash. Recommendations Test this model on a different data set. Perform research on the market value of cash holdings. Analysis Opler, Pinkowitz, Stulz and Williamson examine the determinants and implications of cash holdings and marketable securities. Opler et al. measure cash holdings as the ratio of cash and marketable securities to total assets minus cash and marketable securities. Opler et al. perform year by year cross sectional regression tests, following the Fama and MacBeth (1973) regression method, to predict levels of cash holdings. The Fama and MacBeth regression method effectively treats each year as an independent cross-section and eliminates the problem of serial correlation in the residuals of a time series cross-sectional regression. Table 1 gives an overview of the variables used in their analysis. The dependent variable is cash holdings. 99

100 Table 1: Description of variables used in the analysis of Opler et al. Opler et al. find that cash holdings decrease significantly with size, net working capital, leverage, whether a firm pays dividends, and whether it is regulated. Cash holdings increase significantly with the cash flow-to-assets ratio, the capital expenditures-to-assets ratio, industry volatility, and the R&Dto-sales ratio. Based on these findings Opler et al. conclude that firms with strong growth opportunities and riskier cash flows hold relatively large amounts of cash. Firms that have the greatest access to the capital markets, such as large firms and those with high credit ratings, tend to hold small amounts of cash. Opler et al. recommend that further research could focus on the market value of cash holdings. Moreover, they also recommend that their model should be applied on a different sample of data to analyze whether it still derives valid predictions. 100

101 26. Does the Contribution of Corporate Cash Holdings and Dividends to Firm Value Depend on Governance? A Cross Country Analysis Citations (impact factor) 153 Author L. Pinkowitz, R. Stulz, and R. Williamson Year of publication 2006 Journal The Journal of Finance Key words MV of cash holdings, Corporate governance, Agency costs of free cash flow theory Research question Does the contribution of corporate cash holdings and dividends to firm value depend on governance? Data Public companies with financial data available on Worldscope over the period 1983 to Key findings Investors in countries with poor governance scores place a lower value on a dollar of corporate cash holdings than investors in countries with above median governance scores. Recommendations - Analysis Pinkowitz, Stulz and Williamson investigate whether the market value of cash holdings is lower for minority shareholders in countries where investor protection is low. Pinkowitz et al. indicate the level of investor protection based on legal rights granted to the investor and the extent to which these rights are respected and enforced. They measure the level of investor protection based on the average of various indices of investor protection. They use the regression model that was developed by Fama and French (1998) and Pinkowitz and Williamson (2003) to determine the market value of cash. Pinkowitz et al. find that that in countries with high investor protection, a dollar of cash holdings is worth roughly a dollar to minority investors. In contrast, in countries with poor investor protection, a dollar of cash holdings is worth much less. They estimate that a dollar of liquid assets is worth $0.91 in countries with above-median investor protection while it is worth only $0.33 in countries where investor protection is low. Based on these findings Pinkowitz et al. conclude that investors in countries with poor governance scores place a lower value on a dollar of corporate cash holdings than investors in countries with above median governance scores. They believe that the value of cash holdings is lower in countries where investor protection is low because of increased agency costs. Pinkowitz et al. do not provide any recommendations for further research. 101

102 27. What is a dollar worth? The market value of cash holdings Citations (impact factor) 60 Author L. Pinkowitz, R. Williamson Year of publication 2003 Journal Working paper Key words MV of cash holdings, Financial constraints, Growth opportunities, Target level for cash holdings, Financial distress Research question What is the market value of a dollar of cash held by a company? What factors influence the market value of cash holdings? Data U.S. Public companies with financial data available on COMPUSTAT over the period 1950 to Companies in the financial and utility industries are excluded. Key findings There are cross-sectional differences in the valuation associated with cash holdings. Firms with greater growth opportunities or more volatile investment programs have their cash holdings valued at a premium. Additionally firms which are likely to face financial distress have a significant discount placed on the value of their cash holdings. Also, cash holdings in unconstrained firms are more valuable than cash in a constrained firm. Recommendations Study the impact of financial constraint on a firm s cash policies and the impact of agency costs on the value of cash holdings. Analysis Pinkowitz and Williamson study the market value of cash holdings. They analyze whether the market value of cash holdings depends on a firm s growth opportunities, the volatility of its investment opportunities, the level of financial distress and a firm s financial constraints. Pinkowitz and Williamson define cash as cash plus marketable securities. They define a firm s growth opportunity as the firm s ability to undertake positive NPV projects. They predict that firms with greater growth opportunities should have higher values placed on their cash. Growth opportunity is measured as based on sales growth, dividends, R&D expenditures, and capital expenditures and dividend payments. Furthermore, Pinkowitz and Williamson predict that firms with greater predictability in their investment program should have lower values placed on their cash. The volatility of investment opportunities of a firm is measured based on the standard deviation of either their capital expenditures or R&D expenses. Additionally Pinkowitz and Williamson argue that when the risk of financial distress is present, conflicts can arise between stockholders and bondholders, they predict that firms with greater probability of financial distress conflicts have lower values placed on their cash. As proxies for the likelihood of financial distress, they use Altman s (1968) Z-score, the interest coverage ratio, and firm leverage. Finally they also predict that holding growth options constant, firms with large financial constraints should have higher values placed on their cash. They utilize several proxies to indicate a firm s level of financial constraints. First, firms which are small or do not pay dividends may have a high cost of external finance and limited access to the capital markets. Additionally, they expect that firms with low interest coverage ratios or high amounts of leverage will have less access. They also use measures developed by Whited and Wu (2002) and Almeida, Campello and Weisbach (2003) to determine the level of financial constraints. 102

103 They extend the regression model that was developed by Fama and French (1998) to determine the market value of cash. Pinkowitz and Williamson estimate the relation between market value and cash holdings in Fama and French s model simply by split the change in assets into its cash and non-cash components. They partition their sample into subsamples based on growth opportunities, investment volatility, financial distress and level of financial constraints. The regression model is: Where: = Market value of equity=total shares outstanding*share price = net assets = total assets liquid assets cash holdings and marketable securities. * To simplify the notation the firm subscript that should appear on all variables in the regression and the year subscript that should appear on all regression coefficients are omitted in the equation. In their analysis Pinkowitz and Williamson do not take a position on whether firms have an optimal level of cash holdings. However, they modify their initial regression model to examine whether the existence of a target level for cash holdings influences the market value of cash. They incorporate two more variable that account for two year past and future changes in cash holdings. They argue that if target levels exists and a firm is either at its target or there is no target changes in cash holdings do not contribute much to the overall firm value which represented by small regression coefficients. However, if a target level exist changes in cash holdings are likely to contribute more to the overall value of the firm because it controls for movements towards or away from the target level. The modified regression model now becomes: Where: 103

104 * To simplify the notation the firm subscript that should appear on all variables in the regression and the year subscript that should appear on all regression coefficients are omitted in the equation. To reduce the effect of outliers Pinkowitz and Williamson trim their sample at 1% by dropping 0.5% in each tail of each variable. Following Fama and French (1998) they trim based on the full sample so that while they trim 1% of the observations for each of the 18 independent variables the loose only 8.6% of the total observations. The model of Pinkowitz and Williamson has an adjusted R-square value of where the adjusted R-square value is the average of the adjusted R-square value of 46 single cross sectional regressions. Pinkowitz and Williamson find that for the median firm of their sample; shareholders place a value of $0.97 on a marginal dollar of cash. They also find that the value of cash holdings of firms with high growth opportunities ranges from $1.11 to $1.84. Secondly, they find that firms with low volatility in their investment program have lower valuations associated with their cash holdings. They find that the value of cash holdings of firms with low volatility in their investment program ranges from $0.65 to $0.87 while the value of cash holdings of firms with high volatility in their investment program ranges from $1.58 to $1.17. Despite this result they argue that it is likely that their measures of volatility are not great proxies for measuring whether investment opportunities are able to be delayed and therefore the evidence for this statement is not so strong. Next they test whether firms with greater likelihood of financial distress have lower valuations of their cash holdings due to agency costs between stockholders and bondholders. Overall, their test results are consistent with the idea that firms with greater likelihood of financial distress have lower valuations of their cash holdings. Finally they test whether firms with high financial constraints higher values placed on their cash holdings. Contrary to their expectation, the results suggest that cash is more valuable for firms which are less financially constrained. They find that the value of cash holdings of firms with large financial constraints is about $ 0.81 while the value of cash holdings of firms with little financial constraints is about $1.17. As an explanation for this outcome Pinkowitz and Williamson argue that their empirical measures of constraint may be flawed. As an alternative explanation they argue that that it may also be the case that shareholders view a firm s access to the capital markets as a determining factor regarding whether the firm will be able to take advantage of investment opportunities. If the investment decision is linked to the financing decision, then firms with low financial constraints will probably have greater value associated with cash holdings. Based on these findings Pinkowitz and Williamson conclude that there are cross-sectional differences in the valuation associated with cash holdings. Firms with greater growth opportunities or more volatile investment programs have their cash holdings valued at a premium. Additionally firms which are likely to face financial distress have a significant discount placed on the value of their cash holdings. Also, cash holdings of firms with low financial constraints are more valuable than cash holdings of firms with large financial constraints. Pinkowitz and Williamson recommend future research to focus on examining the impact that financial constraint has on a firm s cash policies and the impact of agency costs on the value of cash holdings. They argue that theoretical models of the value shareholders place on the liquid assets of constrained and unconstrained firms would be very useful to increasing our understanding in this area. 104

105 28. Agency costs, ownership structure and corporate governance mechanisms Citations (impact factor) 155 Author Singh, M., and W. N. Davidson Year of publication 2003 Journal Journal of Banking and Finance Key words Agency cost of free cash flow theory, agency costs, corporate governance Research question Is the agency costs of free cash flow theory applicable to large firms? Data NYSE, AMEX, and NASDAQ listed large US corporations having annual sales revenue of $100m or more over the period 1992 to 1994 Key findings Higher inside ownership aligns managerial and shareholders interests and lowers the agency costs in large corporations when agency costs are defined in terms of asset utilization. However, the relation is generally insignificant when agency costs are defined as discretionary expenses. Recommendations - Analysis Sing and Davidson extend the work of Ang, Cole and Lin (1999) by analyzing the practical applicability of the agency costs of free cash flow theory to large firms. Sing and Davidson argue that the role of ownership structure on agency costs of large public firms might differ from that of small firms since large public firms are continuously monitored. Because Sing and Davidson focus on large publicly traded corporations, they do not have a zeroagency-cost base case where a firm is fully owner managed. They, therefore, relate absolute levels of asset utilization efficiencies and operating expenses to firm ownership while controlling for governance characteristics. Furthermore, they use a slightly different definition of operating expenses. As their aim is to capture agency induced managerial expense as a measure of agency cost, they focus on a firm s selling, general, and administrative (SG&A) expenses instead of total operating expenses used by Ang et al. (1999). SG&A expense, representing the costs related to the management function and to the sale of products, includes managerial salaries, rents, insurance, utilities, supplies, and advertising costs. Sing and Davidson argue that higher levels of SG&A expenses are a close approximation of managerial pay and perquisite consumption in terms of higher salaries, large office complexes, and other organizational support facilities. These costs, to a large extent, reflect managerial discretionary expenses and may be a closer proxy for agency costs. Sing and Davidson control for the number of board members, firm size, leverage and type of industry Sing and Davidson find that firms with high inside ownership are more efficient in their asset utilization and have lower managerial discretionary expenditures relative to firms with below median inside ownership. Firms with above average inside ownership have asset turnovers of 1.58 and those with below inside ownership have asset turnover of Asset turnovers remain statistically significant between the two inside ownership categories, but the difference is insignificant in

106 Moreover they find that inside ownership is unrelated to SG&A expenses in However, in 1994, the mean SG&A ratio is 0.5 for firms with above median inside ownership and is 0.27 for below median inside ownership. Overall these findings provide some evidence that inside ownership helps to align the interests of shareholders and managers. Sing and Davidson also compare the asset turnover ratios and SG&A expense ratios for firms with above median to below median outside block ownership. For the entire sample Sing and Davidson find that the asset turnover is nominally larger for firms with above average block ownership. This difference is consistent with their expectation but is insignificant. The SG&A expense ratio is higher for firms with above median outside block ownership. This result is inconsistent with their expectations but insignificant. Based on these findings Sing and Davidson conclude that higher inside ownership aligns managerial and shareholders interests and lowers the agency costs in large corporations when agency costs are defined in terms of asset utilization. However, the relation is generally insignificant when agency costs are defined as discretionary expenses. Sing and Davidson argue that one possible reason for the SG&A expense ratio not being significantly influenced by ownership and governance variables may be that it is not as visibly related to cash flows generated by firms as is sales revenue. Furthermore, Sing and Davidson conclude that the proportion of equity held by outside block owners does not relate to agency costs as measured by asset utilization and discretionary expense ratios. Sing and Davidson argue that his insignificance relationship measures may be because these agency variables may not completely capture the performance metrics that are evaluated by the outside block holders when evaluating firm performance. Sing and Davidson do not provide any recommendations for further research. 106

107 29. Corporate governance and the value of excess cash holdings of large European firms Citations (impact factor) unknown Author Schauten, M., D. van Dijk, and J. Van der Waal Year of publication 2008 Journal Working paper Key words Corporate governance, MV of cash, Excess cash Research question Does there exist a relation between the quality of corporate governance and the value of excess cash? Data Firms included in the FTSEurofirst 300 Index for the years Key findings Excess cash held by firms with less anti-takeover provisions (low management rights) is valued higher than excess cash held by more protected firms (high management rights). The governance mechanisms shareholder rights, disclosure and board functioning do not have a significant effect on the market value of excess cash. Recommendations - Analysis Schauten, van Dijk, and der Waal study the relation between the quality of corporate governance and the value of excess cash. Excess cash is defined as the difference between a firm s predicted level of cash holdings and its actual level of cash holdings. Schauten et al. use Deminor ratings to indicate the quality of corporate governance. These Deminor ratings are based on 300 different governance indicators. Schauten et al. classify the governance indicators into four categories: Rights and duties of shareholders (referred to as Shareholder rights); range of takeover defenses (referred to as Takeover defenses); disclosure on financial matters and corporate governance (referred to as Disclosure); and Board structure and functioning (referred to as Board). Schauten et al. first use the model developed by Opler et al. (1999) to determine the level of normal or optimal cash holdings and define excess cash as the difference between the actual cash holdings and the predicted normal cash holdings. Next they perform a regression analysis using the regression models developed by Fama and French (1998) to determine the value of excess cash. Additionally, Schauten et al. also perform a regression analysis using the model of Faulkender and Wang (2006). Schauten et al. find that firms with more anti-takeover provisions hold relatively less (excess) cash and spend their excess cash more quickly. Schauten et al. argue that this indicates that well governed firms operate under the fear of the capital market for misallocation of their excess cash holdings. Furthermore, they find that the value of 1 of excess cash is 0,89 for firms with more anti-takeover provisions (high management rights) and 1,45 for firms with less anti-takeover provisions (low management rights). Schauten et al. argue that the value of excess cash of firms with high management rights is relatively low, because the capital market cannot correct nor prevent the misuse of these cash holdings. Cash holdings of these firms are accordingly valued below face value. On the other hand, firms with low management rights run the risk of being taken over if they destroy value (now or probably in the future) by investing in negative NPV projects or by operating extremely 107

108 inefficient. Because of this threat of control over the amount of excess cash, the probability that it will be allocated wrongly is smaller and hence excess cash is valued higher. Additionally, Schauten et al. also find that the other corporate governance measures do not have a significant effect on the valuation of excess cash. Based on these findings Schauten et al. conclude that excess cash held by firms with less anti-takeover provisions (low management rights) is valued higher than excess cash held by more protected firms (high management rights). They also conclude that the governance mechanisms shareholder rights, disclosure and board functioning do not have a significant effect on the market value of excess cash. Schauten et al. do not provide any recommendations for further research. 30. Testing static trade-off against pecking order models of capital structure Citations (impact factor) 895 Author Shyam-Sunder, L., and S. C. Myers Year of publication 1999 Journal Journal of Financial Economics Key words Pecking order theory, MV of cash holdings, Asymmetric information Research question Do firms follow the pecking order theory? Data U.S. public companies with financial data available on COMPUSTAT over the period 1971 to Key findings The pecking order theory is an effective first-order descriptor of corporate financing behavior. Recommendations Analyze the empirical relevance of the pecking order theory using a different data set. Identify the exact circumstances under which the theory holds. Analysis The study of Shyam-Sunder and Myers focuses on the development of a regression test of the pecking order. The financing deficit is constructed from an aggregation of dividends, investment, change in working capital and internal cash flows. If the pecking order theory is correct, then the construction of the financing deficit variable is a justified aggregation. Under the pecking order, each component of financing deficit should have the predicted dollar-for-dollar impact on corporate debt. The main conclusion from the empirical tests performed by Shyam-Sunder and Myers is that the pecking order theory is an effective first-order descriptor of corporate financing behavior. Shyam-Sunder and Myers recommend that further research should focus on analyzing the relevance of the pecking order theory using a different set of data. Moreover, they suggest that further research should aim to further identify the exact circumstances under which the theory holds. 108

109 Appendix B Framework for the performance of a regression analysis Figure B.1 shows the framework that we used for the performance of a regression analysis. Figure B.1: Framework for the performance of a regression analysis. Table B.1 gives a more detailed explanation of all the actions that are included in the framework. 109

110 Table B.1: Explanation of all actions included in the framework Action Explanation Related statistic(-s) 1.1 Data on all variables must be of quantitative nature The independent variables should have some variation in value Standard deviation, Variance 3.1 These extreme cases bias the mean and inflate the standard deviation and Standardized residual therefore they should be excluded from the sample 3.2 These extreme cases (i.e. influential cases) might have undue influence Partial plots on a predictor s regression coefficient and therefore they have to be removed from the sample set 3.3 There should be no perfect linear relationship between two or more of the VIF, Tolerance independent variables 4.1 Model statistics tell us something about the overall fit of the regression R-square, Adjusted R-square model 4.2 ANOVA stands for analysis of variance and tells us whether the model is significantly better at predicting the outcome than using the means as a best guess F-ratio 4.3 Statistics on the regression coefficients tell us something about the relationship between the dependent variable and the independent variables. Moreover these statistics tell us how accurate the predicted relationships are. b-value, t-test, Confidence intervals 5.1 For any two observations the residual terms should be uncorrelated Durbin-Watson coefficient 5.2 The variance of the residual term of the independent variables should be constant Plot of standardized residual against standardized predicted outcomes 5.3 The residuals in the model should be random, normally distributed variables with a mean of zero Histogram of standardized residuals 5.4 The mean values of the outcome variable for each increment of the predictors should lie on a straight line Plot of standardized residual against standardized predicted outcomes 110

111 Appendix C Regression analysis: The market value of cash holdings The following figures contain the output of the regression analysis with the regression model of Fama and French (1998) and Pinkowitz and Williamson (2003). Figure C.1 depicts the model summary. The model summary table contains indicators which are used assess the overall fit of the regression model (Field 2009). Model Summary b Std. Error Change Statistics R Adjusted of the R Square F Sig. F Durbin- Model R Square R Square Estimate Change Change df1 df2 Change Watson a a. Predictors: (Constant), Div1, cash1, interest1, EBIT1, RD1 b. Dependent Variable: marketcap1 Figure C.1: Model summary Figure C.2 depicts the ANOVA table. The ANOVA table contains an analysis of variance (ANOVA) that tests whether the model is significantly better at predicting the outcome than using the means as a best guess (Field 2009). ANOVA b Model Sum of Squares df Mean Square F Sig. 1 Regression a Residual Total a. Predictors: (Constant), Div1, cash1, interest1, EBIT1, RD1 b. Dependent Variable: marketcap1 Figure C.2: ANOVA table Figure C.3 shows depicts a table that shows the regression coefficients. Unstandardized Coefficients Standardized Coefficients Coefficients a 99,0% Confidence Interval for B Correlations Collinearity Statistics Std. Lower Upper Zeroorder Model B Error Beta t Sig. Bound Bound Partial Part Tolerance VIF 1 (Constant) cash EBIT RD interest Div a. Dependent Variable: marketcap1 Figure C.3: Regression coefficients 111

112 Figure C.4 gives an overview of the correlation coefficients between the individual variables. Coefficient Correlations a Model Div1 cash1 interest1 EBIT1 RD1 1 Correlations Div cash interest EBIT RD Covariances Div cash interest EBIT RD a. Dependent Variable: marketcap1 Figure C.4: Correlation coefficients Figure C.5 depicts a graph of the standardized residuals (ZRESID) against the standardized predicted outcomes (ZPRED). This plot is used to assess the assumptions of linearity and homoscedasticity. To conclude on this, the pattern in figure C.5 was compared with patterns that meet or violate the assumption of linearity and homoscedasticity as presented in figure C.6. Figure C.5: Plot of the standardized residuals (ZRESID) against the standardized predicted outcomes (ZPRED) 112

113 Figure C.6: Plots of ZRESID against ZPRED with patterns that meet or violate the of linearity and homoscedasticity Figure C.7 depicts the histogram of the standardized residuals which is used to check if the residuals were normally distributed. Figure C.7: Histogram of the standardized residuals 113

114 Figure C.8 till C.12 show all partial plots. Figure C.8: Scatterplot of cash Figure C.9: Scatterplot of EBIT Figure C.10: Scatterplot of interest expense Figure C.11: Scatterplot of dividends paid Figure C.12: Scatterplot of R&D expense 114