1
Firm-specific determinants of leverage for firms
on the Dutch market
By
Aarif Fatehmahomed1 S 2350076 MSc Finance
Faculty of Economics and Business University of Groningen Supervisor: Dr. D. Ronchetti
April 2017
Abstract
The advantages and disadvantages of debt are factors that have been widely studied theoretically and empirically. While leverage was deemed to be irrelevant when Modigliani and Miller (1958) developed their first model, it became an important factor in more recent corporate finance models and empirical studies. In this paper, an effort is made to add to contemporary empirical research by studying how former identified firm-specific factors influence leverage among Dutch firms. The findings here suggest that, conditionally on the institutional characteristics of the Dutch market, tangibility and profitability are important determinants of leverage, while growth opportunities, taxes and a public market listing also have some influence on leverage. There is no evidence found for a relationship between size and leverage when the proxy for size is based on net sales.
Keywords: Firm leverage, debt, capital structure determinants
2
Introduction
Several theoretical and empirical studies on the firm capital structure theories have brought forward several firm-specific factors that are expected to have relationships with firm leverage. Most findings tend to point to special relationships between leverage and several firm-specific factors, such as size, tangibility, profitability, growth opportunities, risk, and taxes, but there are still inconsistencies in empirical results, and it is not clear how to isolate empirically the contribution to leverage given by institutional, interindustry, and intra-industry factors. This thesis reports an empirical study of the impact of the mentioned firm-specific factors on the leverage of Dutch firms. The study finds a positive relationship between tangibility and leverage, and a negative relationship between profitability and leverage (Rajan and Zingales, 1995; Gaud, Jani, Hoesli, and Bender, 2005; de Jong, Kabir, and Nguyen, 2008; Brav, 2009). Taxes and a public listing are also found to have influence on leverage, but their impact is small. Growth opportunities seem to be especially important for listed firms. While Titman and Wessels (1988) find no evidence for this, the evidence found here is in line with Rajan and Zingales (1995), Gaud, Jani, Hoesli, and Bender (2005), Harjeet, Tong, and Dogan (2008), de Jong, Kabir, and Nguyen (2008), Brav (2009), and Dasilas and Papasyriopoulos (2015), but in contrast with Sogorb-Mira (2005) and Degryse, de Goeij, and Kappert (2012).
Interest rates, in particular long-term interest rates, display correlations with several economic factors, such as economic growth and inflation, and are expected to influence leverage since they are the basic component of the costs of debt (Michaelas, Chittenden, and Poutziouris, 1999; Graham and Harvey, 2001; Koller, Goedhart, and Wessels, 2010; Öztekin, 2015). According to the OECD2, of which the Netherlands is also a member, the long-term interest rate is a key short-term economic indicator. Therefore, this thesis analyzes firm-specific determinants of leverage during periods characterized by distinct interest rate levels. During the considered period in this thesis there were no major changes in Dutch legislation or other institutional factors that may affect leverage. In addition, some theories and empirical studies suggest that country effects are less important than firm effects in explaining capital structures and in general the legal, institutional, and financial environment remains constant when firms in one country are
3
analyzed (Daskalakis and Psillaki, 2008; Myers, 1984; Harjeet, Tong, and Dogan, 2008). Therefore, changes in legislation or institutional factors are not used as determinants of leverage in this thesis.
The Dutch market is a market-based economy. Countries with market-based economies have generally better developed markets for equity finance than countries with bank-based economies. In market-based systems there is more competition among financiers, which results in more widely available information about firms, easier trading of securities, and larger and more liquid markets for securities. In contrast to bank-based systems, where the management of firms is disciplined by large block shareholders or creditors, in the market-based system management is disciplined by the capital market (Bijlsma and Zwart, 2013). The market-based system in the Netherlands suggests that the market mechanism is not constricted and that this may have a positive influence on the freedom of firms in making capital structure decisions. Furthermore, this also suggests that capital structure theories that are developed on basis of market-based economies, such as the United States and the United Kingdom, are appropriate to analyze capital structure decisions of Dutch firms.
The data used in this paper is extracted from the Orbis database, which is provided by Bureau van dijk3. The final sample, consisting of 211 firms, is selected based on its characteristics and available data and covers the period from 2007 to 2015. Although inter-industry effects may affect leverage levels on itself, the impact of firm characteristics on leverage is mostly in line for the most industries (Degryse, de Goeij, and Kappert, 2012). However, financial firms, insurance companies, and utility companies are excluded from the sample, because they have different regulations (Rajan and Zingales, 1995; Fama and French, 2002; Gaud, Jani, Hoesli, and Bender, 2005) and they may receive government support. In addition, excluding firms in the mentioned industries may also contribute substantially to alleviate any omitted variable problems created by industry effects in financing decisions (Fama and French, 2002). It is also assumed that firm-specific characteristics of the firms in the final sample, such as the amount of tangible assets and profitability, already account for smaller industry differences. Furthermore, micro-enterprises, are also excluded from the sample (Dasilas and Papasyriopoulos, 2015). To measure the effect of the firm-specific determinants firm-level, i.e. cross-sectional,
4
ordinary least squares regressions are performed with leverage as the dependent variable and the firm-specific factors as the explanatory variables (Rajan and Zingales, 1995; de Jong, Kabir, and Nguyen, 2008). The regression is performed separately for three time periods, i.e. the total sample period from 2007 to 2015, the period from 2007 to 2011, which is related to financial crises of the last decade, and the period from 2012 to 2015, which is related to the period after the financial crises. The explanatory variables are averaged to reduce noise and to account for slow adjustments, and they are lagged one period to reduce the problem of endogeneity (Rajan and Zingales, 1995).
The thesis is structured as follows. I first give an overview of some theoretical models and empirical studies of the firm capital structure. Based on this I formulate the hypotheses that are tested in this thesis. I then continue with a description of the data and the methodology that are used for the analysis, the results of the analysis, and finally a conclusion with regard to the findings.
The capital structure of a firm
In this section I summarize results of theoretical and empirical studies of the firm capital structure that motivate this thesis. I first describe some theories that are often used in corporate finance to explain the choices that managers make with regard to capital structure of firms. These are the agency theory, the tradeoff theory, the pecking-order theory, and the asset substitution effect. I then highlight some benefits and drawbacks of using debt and how the theories of capital structure are used to explain how managers choose an optimal level of leverage. Finally, I explain which determinants are expected to influence corporate leverage and why those determinants are expected to influence corporate leverage.
Capital structure theories
5
contracts wherein the separation of ownership and control is established (Jensen and Meckling, 1976). However, if principals, as owners of the firm, and agents, as managers of the firm, are both acting utility maximizing, there is a good reason to believe that the interests of both will not be in alignment (Jensen and Meckling, 1976). When management does not act in the best interest of the firms’ owners, it is called Moral Hazard (Tirole, 2006). There are four main categories of moral hazard. The first is when managers put insufficient effort to get the best outcome for the firm. The second is when managers make extravagant investments that can be detrimental to the owners. The third is when managers use entrenchment strategies. Finally, the fourth is when managers engage in self-dealing to increase their private benefits.
Using debt in the capital structure of a firm, what is also known as leveraging, can have a mitigating role on the occurrence of moral hazard. Debt does not only provide tax advantages and enhanced investment opportunities (Modigliani and Miller, 1963; Jensen and Meckling, 1976), but it can also work as a governing mechanism (Tirole, 2006). Because cash is required to service debt, debt prevents management from consuming cash. At the same time, it incentivizes management to realize positive cash flows, it threatens management of losing control, and it incentivizes management to perform better when they own equity of the firm themselves (Tirole, 2006). For the uninformed reader, this would initially suggest that the use of debt is an easy and effective way to prevent managers of engaging in moral hazard. However, if other mitigating factors are not in place, agency costs of debt will still discourage the use of corporate debt (Jensen and Meckling, 1976).
The loss of wealth due to restrictions in investment decisions, costs of monitoring and bonding, and potential bankruptcy and reorganization are all agency costs of debt (Jensen and Meckling, 1976). The benefits of debt on the one hand and the agency costs of debt on the other, require firms to make deliberate decisions when they consider the usage of debt. Higher tax advantages of debt will lead to higher optimal debt-equity ratios, while higher non-debt tax advantages will lead to lower optimal debt-equity ratios. A firm’s capital structure will be determined by a tradeoff between the tax advantages of debt and the financial costs that occur with too much debt. This is also known as the “static tradeoff theory” (Myers, 1984; Tirole, 2006).
6
and they want to avoid informational asymmetries and stock price reactions to seasoned equity issues, they will rather use debt than equity. To fund projects, firms prefer to use internal capital as primary source and then turn to debt as second, convertibles as third, and equity as fourth. This is known as “pecking-order hypothesis” or the “pecking-order theory” (Tirole, 2006).
While equity holders have limited liability, they can profit unlimitedly from the upside potential of risky projects as long as the returns are higher than the costs of capital. Therefore, even if debt has to be increased and total risk of the firm increases, equity holders prefer risky projects with high returns over less risky projects with lower returns. On the other hand, current debt holders have limited returns and a probability of losing their investment in the firm partially or completely. Therefore, current debt holders do not prefer the firm to increase debt to invest in high risk projects. However, if management’s compensation, rewards, or investments are related to equity performance, they may become compliant to increase risk taking. In corporate finance this tendency to prefer high risk projects or assets above safer investments is often called “asset substitution” (Tirole, 2006). On the other hand, the theory of asymmetric information suggests that that sufficiently risk-averse managers can also use debt as a signal to inform investors that they will invest in high quality projects (Blazenko, 1987). By comparing different theoretical models however, Harris and Raviv (1991) conclude that there is no consistent prediction of how equity holders react to debt issuance.
Benefits of using debt
Modigliani and Millers’ (1958) propositions regarding the irrelevance of the capital structure unleashed a lot of research on the dynamics of capital structure. According to their propositions, assuming perfect capital markets, i.e. commodities that are perfect substitutes have the same price, it does not matter what capital structure a firm uses to fund investment opportunities. Regardless of how an investment opportunity is funded, either by debt or equity, changes in capital structure will not affect the value of the firm. The value of the firm is determined by the discounted value of the firm´s expected returns. In addition, the return on the firms’ stocks rises linearly with the debt to equity ratio of the firm and thus the return on the stock rises as the debt to equity ratio increases. The rest of this section gives an overview of theories and empirical findings that counter this theory of capital structure irrelevance.
7
(1958), they conclude that the tax advantages of riskless debt financing are greater than they originally suggested. The use of debt in the capital structure of a firm provides tax gains to the firm. These tax gains will affect the cash flows of the firm and will ultimately have an influence on the value of the firm. Jensen and Meckling (1976) argue that not only the use of riskless perpetual debt alters the value of the firm, but that even the use of risky debt alters the value of the firm through the tax subsidies on interest payments. Furthermore, even if there are no tax gains as a result of debt, in the absence of internal resources debt enables firms to invest in unique and profitable projects. In addition to tax advantages and enhanced investing possibilities, issuing debt or increasing leverage can also function as a governing mechanism (Jensen, 1986; Tirole, 2006). With the “control hypothesis” Jensen (1986) shows that debt can motivate managers and their organization to improve efficiency. The promise of managers to pay out cash flows in order to service debt is more effective than the promise to pay out dividends. When debt is not serviced, managers of the firm give debt holders the opportunity to take the firm to court. In this way debt reduces the agency costs of free cash flow, because the cash available for discretion spending by managers is reduced. Likewise, the requirement of servicing debt out of free cash flows helps to motivate managers to overcome organizational resistance to cut back on costs. In addition, debt does not only threaten management to lose control if the firm does not become more efficient, but it can also motivate management to become more efficient if they own equity of the firm themselves (Tirole, 2006).
Drawbacks of using debt
8
major components of agency costs are the main reasons why we do not observe a lot of large corporations with very high amounts of debt (Jensen and Meckling,1976). These components of agency costs are categorized as: (1) the incentive effects associated with highly leveraged firms, (2) the monitoring costs that these incentive effects engender, and (3) the bankruptcy costs as a result of high leverage (Jensen and Meckling, 1976, p. 334).
The incentive effects concern the anticipation of bondholders on the maximizing behavior of managers. Assuming that managers also own shares of the firm, shareholders in aggregate will prefer risky projects with high expected payoffs over safer projects with low expected payoffs. Since equity holders are residual claimants, they benefit relatively more than bondholders from the upside potential of investments, while they lose relatively less than bondholders from project failure. Bondholders in turn, may anticipate on this by limiting the amount of credit they provide to the firm. As a result, investment possibilities decrease, which lowers the total value of the firm. This could be avoided if bondholders were assured that managers would not unnecessarily increase the risk of the firm.
The monitoring costs concern the costs that arise as a result of provisions that bondholders impose to mitigate the incentive effects and avoid high risk behavior. Although it is difficult for bondholders to protect themselves completely, they often include constraints on management’s decision making regarding factors such as dividends, future debt issues, and maintenance of working capital in bond issues to obtain some protection. However, these provisions are costly to write and to implement. In addition, they are costly, because they can reduce profitability in cases where management is limited to take optimal actions.
9
Myers (1977) adds the suboptimal investment policy, to the list of agency costs. He shows that in some cases, even in markets with full information and no bankruptcy costs, firms will pass up positive net present value (NPV) projects. When a firm has debt outstanding, shareholders only want to invest in projects that have a return that is higher than the amount that needs to be paid back to creditors. However, passing up all positive NPV projects that have a lower return than equity holders require will reinforce the risk of not being able to service current debt. This under-investing increases risk for creditors and thus decreases the value of debt. In addition, it also decreases the value of equity and of the firm in totality, because the firm is forgoing positive NPV projects or is investing poorly. At the same time, shareholders may still want to issue more debt to invest in high risk projects with high expected returns, since the decrease in equity value will be more than offset by the expected gains at the cost of debt holders. On the other hand, if debt holders anticipate on this behavior, the costs of new debt will increase. This brings the disadvantage of increased risk taking back to equity holders. This effect is similar to the aforementioned “asset substitution effect” (Harris and Raviv, 1991).
Optimal leverage theories
As described above, tax shields (Modigliani and Miller, 1963; Jensen and Meckling, 1976), enhanced investment possibilities (Jensen and Meckling, 1976), and the control hypothesis (Jensen, 1986) encourage the use of debt on the one hand, but limiting agreements (Modigliani and Miller, 1963), agency costs (Jensen and Meckling, 1976), and under-investment (1977) discourage the use of debt on the other. Margaritis and Psillaki (2007) find that leverage in itself is associated with improved efficiency and they suggest that this is evidence of Jensen and Meckling’s (1976) predictions of the agency cost hypothesis. Leverage has a disciplinary effect and will require the firm to attain best practices, to reduce inefficiencies, and to subsequently perform better. D’Mello and Miranda (2009) find that when unlevered firms introduce leverage in the capital structure, overinvestment in cash holdings and real estate decreases and as a result equity value and firm value increases. Given the pros and the cons of leverage, the question that remains is how much leverage is optimal for firms.
10
costs on the other. However, actual levels of debt may differ from the optimal level, because of two effects that indicate the agency costs of managerial discretion on optimal leverage (Morellec, 2004). First, the willingness of shareholders to improve investment policy by using debt depends on the profitability of assets in place acquired by overinvestment, on taxes, on financial distress costs, and on the number of growth options available to the firm. Secondly, the debt policy that maximizes firm value may not eliminate the free cash flow problem, because debt also induces underinvestment and bankruptcy costs. While shareholders would like the manager to select the optimal level of leverage for firm value maximization, the preferences of the manager and the costs of control challenges will induce the manager to deviate from the optimal level of leverage.
Another theory that explains the capital structure decisions of management is the pecking order theory (Myers and Majluf, 1984; Myers, 1984; Tirole, 2006). This theory is not based on an optimal level of leverage, but on the assumption that there is an asymmetry of information between managers and stockholders and that managers act in the interest of existing stockholders. As mentioned before, according to this theory management prefers to use internal sources such as cash and marketable securities, i.e. financial slack, as first resource to fund projects and then turn to debt, possible hybrid securities such as convertible bonds, and equity respectively. The leverage level of a firm will thus be a reflection of the firm’s cumulative demand for external capital (Myers, 1984).
In contrast to both theories above, Baker and Wurgler (2002) and Rajan and Zingales (1995), find evidence for the market-timing theory where high leverage is a cumulative result of raising funds when market values are low and low leverage is a cumulative result of raising funds when market values are high. Furthermore, firms tend to repurchase equity when the market value is low. Baker and Wurgler (2002) suggest that these findings indicate that firms are successful in timing the market with regard to the costs of issuing equity and with regard to investors’ enthusiasm about earnings prospects.
11
evidence of the pecking-order theory. Although Harris and Raviv (1991) could find no consistency in models’ predictions of the existence of a pecking-order, they do find that there is empirical evidence among different researches for the existence of the pecking-order theory. Flannery and Rangan (2006) show that market timing and pecking-order considerations only explain less than 10% of changes in capital structures. More than 50% of observed capital structure changes can be attributed to targeting behavior. Firms converge to their targets with a rate of more than 30% per year. In contrast to Baker and Wurgler (2002), Flannery and Rangan (2006) argue that their results show that this behavior of adjusting capital structures is mainly explained by the tradeoff theory and that the pecking-order theory and market timing theory add very little explanation.
Faulkender, Flannery, Watson Hankins, and Smith (2012) show that firms converge to their target leverage ratios based on the marginal costs of implementing capital changes. Their results show that firms that are over-levered close more of the gap between actual leverage and target leverage than firms that are under-levered. They argue that this indicates that the benefits of decreasing leverage may be larger than the benefits of increasing leverage and that financial constraints are the main considerations for firms in regard to their adjustment speed to target leverage ratios. Firms with almost zero cash flow close the gap between actual and target leverage ratios at a slower rate than firms with greater cash flows and greater deviations from target leverage ratios. The adjustment speed is highest for firms that are over-levered.
Although there is some empirical support for the different theoretical models for a firm capital structure, none of them fully explains the capital structure decisions of firms and their optimal level of leverage. There is even evidence that executives in the U.S. and Canada are not very concerned with issues such as asset substitution, asymmetric information, transaction costs, free cash flows, or personal taxes, but that they rely on practical and informal rules when making capital structure decisions (Graham and Harvey,2001).
Determinants of firm leverage
12
capital structure dynamics in similar economies with developed legal environment and high level of economic development (de Jong, Kabir, and Nguyen, 2007).
Based on the three most known theoretical capital structure models, i.e. the static trade-off theory, the agency theory, and the pecking-order theory, researchers have identified several firm-specific determinants of corporate leverage. Although firm-firm-specific factors, such as tangibility, size, risk, profitability, and growth opportunities, are found to be strong and in line with capital structure theories across a large number of countries, including the Netherlands, in some countries one or more firm-specific factors are not significantly related to leverage and for a very small number of countries the results are not consistent with theories (de Jong, Kabir, and Nguyen, 2007). Firm-specific determinants can be defined as those variables that are particularly relative to firm operations and fundamentals (Daskalakis and Psillaki, 2008).
When the variation of firms’ capital structures across countries (de Jong, Kabir, and Nguyen, 2007) or the adjustment speed towards a target leverage ratio (Rajan and Zingales, 1995; Gaud, Jani, Hoesli, and Bender, 2005) are considered, country-specific or institutional factors like GDP growth, bond market development, and creditor protection also play an important role in explaining leverage levels. However, Daskalakis and Psillaki (2008) find that, based on their empirical study on Greece and France, firm effects are more important than country effects in capital structure decisions. This in line with arguments put forward by Myers (1984). This may relate to the fact that in some countries, such as the Netherlands, all firms have to comply with the same legislation or legislation that only differ due the characteristics of the firm itself, but do not have a major influence on capital structure choices. This is in contrast to some other countries, such as Switzerland and the United States, where different states within the country may have major differences in legislation that do affect capital structure choices of firms.
13
advantages. Interest rates cannot be influenced by firms, but they are expected to have an important influence on firm leverage, because they are the basic component of the cost of capital for firms. During the period under study in this thesis there are major changes in interest rates. Because of the reasons mentioned here, taxes paid and interest rates are also considered to be determinants of leverage in this thesis.
Firm-specific leverage factors
This thesis analyzes how the characteristics of firms in the Netherlands influence their leverage ratios. Therefore, rather than comparing country specific determinants or institutional factors between countries, this study will focus on how former identified firm-specific factors relate to firm leverage in one country (Sogorb-Mira, 2005; Daskalakis and Psillaki, 2008; Titman and Wessels, 1988; Gaud, Jani, Hoesli, and Bender, 2005). The considered firm-specific determinants are size, tangibility, growth, profitability, volatility of profitability, and taxes, and in the remaining part of this section I state the hypotheses that I consider in the thesis, relating them to previous findings in the financial literature on leverage. Institutional, inter-industry, and intra-industry factors are more important when studying capital structure across countries or across industries (Rajan and Zingales, 1995; Degryse, de Goeij, and Kappert, 2011). In addition, some theories and empirical studies suggest that firm effects are more important in explaining capital structures decisions than country effects are (Daskalakis and Psillaki, 2008; Myers, 1984) and that the legal, institutional, and financial environment remains constant when only one country is analyzed (Harjeet, Tong, and Dogan,2008).
Although institutional factors are not expected to change often and quickly, dramatic changes in a country’s financial conditions, like the financial crisis that sprung in 2007 and had its effects well into 2010, can alter the capital structure decisions of firms (Dasilas and Papasyriopoulos, 2015). Since the long-term interest rate is a key short-term economic indicator4 and interest rates are a component of the cost of capital for firms, the effect of changing long-term interest rates will be considered by comparing the leverage ratios and the economic significance of the firm-specific determinants of leverage during the period 2007-2011 with those
14 during the period 2012 -2015.
Profitability
Harris and Raviv (1991) find that most models predict a positive relationship between leverage and profitability, but most of the empirical studies they review show that leverage does not increase with increases in profitability. According to the prediction of the trade-off model, profitability and book leverage are positively related. Margaritis and Psillaki (2007) find that profitability and leverage are positively related, which is line with the theoretical predictions. Also Dasilas and Papasyriopoulos (2015) find a positive correlation between leverage and profitability. However, according to the predictions of the pecking order model profitability and book and market leverage are negatively related (Rajan and Zingales, 1995; Fama and French, 2002; Gaud, Jani, Hoesli, and Bender, 2005; Brav, 2009; Öztekin, 2015). Rajan and Zingales (1995) find that, except for Germany, among all G7 countries profitability is negatively correlated with leverage. Furthermore, they find that in the United States, Japan, Italy, and Canada the negative influence of profitability on leverage is stronger when firm size increases. For large firms in the United Kingdom, however, profitability and leverage are more positively correlated and in Germany and France there is no relationship for large firms. Degryse, de Goeij, and Kappert (2011) find that Dutch SMEs use profits to particularly reduce short-term debt, but at the same time increase long-term debt to invest in long-term assets and fund growth while they are profitable. Titman and Wessels (1988), Sogorb-Mira (2004), Gaud, Jani, Hoesli, and Bender (2005), Daskalakis and Psillaki (2008), and Brav (2009) also find evidence for a negative correlation between profitability and leverage, which is in line with the pecking order theory. There is also evidence that firms use target leverage ratios, both book-valued and market-valued, and increase their leverage to meet their target leverage ratios when profitability increases (Titman and Wessels, 1988; Flannery and Rangan, 2006; Brav, 2009). Based on the majority of the empirical work reviewed here, the first hypothesis of this paper is formulated as follows:
H1: Firm profitability and leverage of Dutch firms are negatively related.
Size
15
firms tend to be more diversified, tend to fail less often, and are expected to be more able to weather economic downturns (Rajan and Zingales, 1995; Margaritis and Psillaki, 2007). On the other side, size can also be a proxy of available public information as larger firms tend to have less informational asymmetry between themselves and investors. This should enable them to issue equity more easily and thus would have a negative effect on leverage. However, data shows that this is not true since in most of the G7 countries the largest firms issue significantly less equity than the smallest firms.
Empirical evidence shows that size and leverage are indeed positively correlated (Rajan and Zingales, 1995; Sogorb-Mira, 2004; Gaud, Jani, Hoesli, and Bender, 2005; Margaritis and Psillaki, 2007; Harjeet, Tong, and Dogan, 2008; Degryse, de Goeij, and Kappert, 2011). Among the G7 countries, only in Germany large firms have substantially less debt than small firms (Rajan and Zingales, 1995). Although Rajan and Zingales (1995) conclude that they do not really understand why size and leverage are positively correlated, others attribute this positive relationship to the facts that leverage is costlier for smaller firms in the sense of transaction costs, agency costs, and information asymmetry costs, that smaller firms have less collateral, and that smaller firms appear to be risky and more sensitive to economic downturns (Titman and Wessels, 1988; Michaelas, Chittenden, and Poutziouris, 1999; Degryse, de Goeij, and Kappert , 2011). These reasons might also explain why small firms use more short-term debt than large firms do (Titman and Wessels, 1988; Degryse, de Goeij, and Kappert, 2011; Faulkender, Flannery, Watson Hankins, and Smith, 2012). Based on these former studies, the second hypothesis to be tested here, is formulated as:
H2: Firm size and leverage of Dutch firms are positively related.
Tangibility of assets
16
(2011) also show that it is easier for SMEs with collateral to borrow from banks. While Titman and Wessels (1988) find no evidence for a relationship between collateral value and leverage and there is even some evidence of a negative relationship between tangibility of assets and leverage (Daskalakis and Psillaki, 2008), most empirical studies reviewed here find a positive relationship between tangible assets and leverage (Rajan and Zingales, 1995; Gaud, Jani, Hoesli, and Bender 2005; Margaritis and Psillaki, 2007; Harjeet, Tong, and Dogan, 2008; Brav, 2009). Therefore, the third hypothesis in this study is formulated as:
H3: The tangibility of assets and leverage of Dutch firms are positively related.
Growth opportunities
Myer’s (1977) model predicts that the amount of debt issued by a firm will be inversely related to the ratio of the value of the firm that is determined by growth opportunities to the total value of the firm. His argument for this is that growth opportunities are a component of the firm’s value, but investments in those opportunities are dependent on future expenditure of the firm. This implies that firms with more growth opportunities and less tangible assets should use less debt. Rajan and Zingales (1995), Gaud, Jani, Hoesli, and Bender (2005), Harjeet, Margaritis and Psillaki (2007), Tong and Dogan (2008), and Brav (2009) find evidence in line with this. Their empirical analyses show that growth opportunities and leverage are negatively correlated. Margaritis and Psillaki (2007) argue that a high number of intangible assets is common for firms with high expected growth opportunities and that this results in low leverage. High leverage increases the probability of adverse selection and firms, especially growth firms, want to avoid this.
17
find a positive relationship between growth options and leverage, the majority of the empirical evidence and the models presented here lead to the formulation of the fourth hypothesis in this study as:
H4: Growth opportunities and leverage of Dutch firms are negatively related.
Volatility of earnings
Both, the pecking order theory and the trade-off theory predict that firms with more volatile earnings and net cash flows are less levered and have lower dividend payouts. This lowers the probability that a firm has to issue new risky securities or has to forgo profitable investments when cash flows are low (Fama and French, 2002). Although Fama and French (2002) find some evidence for these relationships, their results may be biased because they use size as proxy for volatility and obtain multicollinearity problems with other factors. They recognize that size may also be a proxy for age and access to capital markets, which both influence financing decisions. De jong, Kabir, and Nguyen (2008) relate volatility of earnings to business risk since more volatile earnings indicate a higher probability of bankruptcy. Therefore, they expect that there is a negative relationship between business risk or volatility of earnings and leverage. However, their empirical findings are mixed and show that only in 14 of 42 countries in their study there is a significant negative relationship between business risk or volatility of earnings and leverage. Gaud, Jani, Hoesli, and Bender (2005) argue the other way around and suggest that a negative relationship between operating risk and leverage is expected, because firms that have high operating risk can reduce the volatility of net profits by reducing the level of debt. This will cause bankruptcy costs to decrease and the probability of fully benefiting from the tax shield to increase. In addition, firms that have volatile results try to accumulate cash during good years to avoid possible under-investment issues in the future. Although Titman and Wessels (1988) find no support for this negative relationship, the literature review here supports this relationship and therefore the fifth hypothesis formulated here is:
H5: Volatility of earnings and leverage of Dutch firms are negatively related.
Paid taxes
18
structure decisions. Tax shields will encourage firms to use debt in their capital structure on the one hand, but financial distress costs and agency costs on the other hand hamper firms to leverage unlimitedly (Modigliani and Miller, 1963; Jensen and Meckling, 1976; Myers, 1977). Although most empirical studies reviewed here acknowledge that taxes might have influence on corporate leverage, they do not specifically use taxes as a determinant of leverage (Gaud, Jani, Hoesli, and Bender, 2005; Brav, 2009; Titman and Wessels, 1988; Harjeet, Tong, and Dogan, 2008). Rajan and Zingales (1995) do show that taxes cannot easily be dismissed as having an influence on corporate leverage, but they do not measure the exact effect of taxes on leverage. On the other hand, MacKie-Mason (1990) finds substantial effects of taxes on financing decision, but they suggest that the overall capital structure studies are not appropriate to study tax effects, since the effective tax rate is dependent on several other factors. They show that when tax shields are likely to reduce a firm’s tax rate, firms with high tax shields that have already exhausted their tax benefits are less likely to use debt. Degryse, de Goeij, and Kappert (2012) find that taxes have a negative effect on total and long-term debt, but a slightly positive effect on short-debt. They argue that an explanation for this might be that high taxes stem from high profits, which at the same time decreases the need for debt. Faulkender and Smith (2016) find empirical evidence in line with the trade off theory. Multinational firms in countries with higher tax rates do have higher leverage ratios and lower interest coverage ratios. Michaelas, Chittenden, and Poutziouris (1999) find that tax effects do not appear to have an effect on the debt ratios of small firms, but tax considerations may become important for long-term capital structure decisions. While Cheng and Green (2008) do find a significant positive impact of tax rates on the debt ratios of firms in the EU, their finding show that this impact is very small. In contrast to the other studies and theoretical models, Graham and Harvey (2001) find that according to firms’ CFOs the corporate tax advantage of debt is only moderate important in capital structure decisions. The literature here implies that firms will most likely employ tax benefits of debt and increase debt to lower taxes. Based on this expectation the fifth hypothesis in this study is formulated as:
H6: Taxes paid and leverage of Dutch firms are negatively related
Interest rates
19
firms. Graham and Harvey (2001) find that large and small firms have different priorities when they consider factors to adjust for risks. However, both do consider the interest rate risk as an additional risk factor next to firm characteristics. This may be due to the fact that most firms in their sample, which consists of firms in the U.S. and Canada, use present value techniques to evaluate projects, but surprisingly also use company-wide discount rates rather than project specific discount rates. In general, large firms, which probably have large or sophisticated treasury departments, try to time the market in the sense of interest rates. CFOs tend to borrow short-term rather than long-term when they think that short rates are low relative to long rates or when they expect that long-term rates will decline. On the other hand, higher expected inflation makes debt issuances cheaper which should result in higher leverage (Öztekin, 2015). In general, inflation and interest rates are positively correlated (Koller, Goedhart, and Wessels, 2010). Michaelas, Chittenden, and Poutziouris (1999) find that short-term debt and economic conditions are negatively related, while average long-term debt ratios are positively related with economic growth. None of the empirical studies that are reviewed here have used interest rates as a determinant of firm leverage. However, since interest rates do have a relationship with debt, the effect of interest rates on leverage will be considered here by comparing the leverage ratios and the economic significance of the firm-specific determinants between periods with different average long-term interest rates.
Data and model for leverage of Dutch firms
This section describes the data used in the study, the adopted model for leverage and its estimation procedure.
Data
The data used in this paper is extracted from the Orbis database, which is provided by Bureau van Dijk5. This database provides financial data for public and private firms during the last ten years. Since most firms have not presented their financial information of 2016 yet, the analysis here will cover the period starting in 2007 and ending in 2015. Financial firms, insurance companies, and utility companies such as gas, water, electricity, post, and telecommunication
20
companies are excluded from the sample, because of several reasons. First, the leverage of these firms is strongly influenced by investor insurance schemes such as deposit insurance. Second, their debt-like liabilities have different characteristics than the debt issued by other firms. Third, there are other regulations such as minimal capital requirements that apply to them. And fourth, the businesses of these firms may be highly regulated (Rajan and Zingales, 1995; Gaud, Jani, Hoesli, and Bender, 2005). In addition, financial firms and utilities are assumed to be inappropriate to test the predictions of leverage models and the financing decisions of utilities may be criticized, because these decisions may be a byproduct of regulation. Excluding these firms may also contribute to alleviate any omitted variable problems created by industry effects in financing decisions (Fama and French, 2002). Furthermore, firms are selected on the availability of at least data for long-term debt and total assets during the total period of interest. Micro-enterprises, are also excluded from the sample (Dasilas and Papasyriopoulos, 2015). According to recommendation 2003/361 of the European Commission6 (2003), which is also used in Sogorb-Mira (2005) and Dasilas and Papasyriopoulos (2015), micro-enterprises have less than 10 employees and a turnover or balance sheet total equal to or less than € 2 million. Here, firms with total sales less than €2 million are not included. All the selection criteria used here result in a final sample consisting of 211 firms. However, depending on the sort of analysis, i.e. only leverage levels or statistics of a common sample that is used for estimation, the number of observations varies as a result of missing information.
Firm-specific variables
The ratio of total liabilities to total assets is the broadest definition of leverage, but it does not give an indication of default risk on the short term and it may overstate the amount of leverage by including items that are used for transactions rather than for financing. A better definition of financial leverage is provided by the ratio of total debt to total assets (Rajan and Zingales, 1995). Therefore, the dependent variable, the total debt ratio (TDRi), used in study will be defined as such, i.e. total debt divided by the total assets of the firm (Michaelas, Chittenden, and Poutziouris, 1999; Gaud, Jani, Hoesli, and Bender, 2005; Sogorb-Mira, 2005; Degryse, de Goeij, and Kappert, 2012; Dasilas and Papasyriopoulos, 2015). In addition, the long-term debt ratio (LDRi) and the short-term debt ratio (SDRi) are also taken into consideration separately and
21
are defined as long-term debt divided by total assets and short-term debt divided by total assets, respectively (Michaelas, Chittenden, and Poutziouris, 1999; Sogorb-Mira, 2005; Degryse, de Goeij, and Kappert, 2012; Dasilas and Papasyriopoulos, 2015). Table II in the analysis section presents the leverage ratios at three points in time, i.e. 2007, 2011 and 2015.
22 are defined here.
Table I: Descriptive statistics
This table presents the descriptive statistics of the common samples used in the estimation of the model for three different periods. TDR, LDR, and SDR are the annual leverage ratios of each indicated year, defined as the ratio of total debt to total assets, long-term debt to total assets, and short-term debt to total assets, respectively, and represent the dependent variable, Li, in the specification of equation (1). SIZE represents size defined as the logarithm of sales, TANG represents tangibility defined as the ratio of fixed assets to total assets, GROWTH represents growth defined as the ratio of intangible assets to total assets, PROFIT represents profitability defined as the ratio of earnings before interest and taxes (EBIT) to total assets, TAXES represents taxation defined as the ratio of all taxes of an accounting year to profits before tax, and VOL represents the volatility of profitability defined as the standard deviation of profitability. SIZE, TANG, GROWTH, PROFIT, VOL, and TAXES are the explanatory variables in equation (1) and are all, except for VOL, averages of the annual values of each firm during the indicated period. VOL is the standard deviation of the annual values of profitability of each firm during the indicated period.
Panel A: regression of firm-specific determinants during 2007 to 2014 on firm leverage in 2015 Variable TDR LDR SDR SIZE TANG GROWTH PROFIT VOL TAXES Mean 0.22 0.18 0.04 12.60 0.28 0.15 0.05 0.06 0.13 Median 0.17 0.14 0.00 12.25 0.26 0.04 0.04 0.04 0.19 Std.Dev. 0.24 0.19 0.13 2.03 0.22 0.19 0.08 0.06 1.03 Min 0.00 0.00 0.00 8.34 0.00 0.00 -0.42 0.00 -7.70 Max 1.91 1.46 1.42 18.93 0.92 0.90 0.44 0.57 5.00 N 145 145 145 145 145 145 145 145 145
Panel B: regression of firm-specific determinants during 2012 to 2014 on firm leverage in 2015 Variable TDR LDR SDR SIZE TANG GROWTH PROFIT VOL TAXES Mean 0.22 0.18 0.04 12.64 0.28 0.14 0.04 0.04 0.24 Median 0.17 0.14 0.00 12.20 0.25 0.04 0.04 0.02 0.20 Std.Dev. 0.24 0.19 0.13 2.06 0.22 0.19 0.11 0.08 2.17 Min 0.00 0.00 0.00 8.42 0.00 0.00 -0.87 0.00 -20.25 Max 1.91 1.46 1.42 19.23 0.93 0.87 0.42 0.80 11.61 N 148 148 148 148 148 148 148 148 148
23
Interest rate
From the credit rating of a firm, the interest rate payable on its debt funding can be estimated. The difference between risk-free bonds and corporate bonds, the credit spread, is greater for firms with lower credit ratings, because they have a higher probability of default. An estimate for the cost of debt can be obtained by adding the credit spread to the risk-free rate (Koller, Goedhart, and Wessels, 2010). Therefore, the Dutch long-term interest rate will be considered here to detect any influence of interest rates. Figure I displays the trend of the long-term interest rate during the last 11 years. Because a cross sectional, i.e. firm level, analysis is performed in this study to analyze firm-specific determinants, the average interest rate of a period, which is one value for all firms, cannot be used in the regression model. To detect any influence of the interest rate, the difference in leverage ratios in 2007,2011 and 2015 is considered as well as the change in relevance and significance of the firm-specific determinants of leverage during the period from 2007 to 2011 and the period from 2012 to 2015.
Fig.1. Long-term interest rate in the Netherlands
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 In te re st r at e Year
24
Model for leverage
In this study, the effect of each firm-specific determinant, that is deduced from theoretical frameworks such as the trade-off theory, the pecking order theory, and the agency theory is measured by performing firm-level, i.e. cross-sectional, ordinary least squares regressions with leverage as the dependent variable and the several firm-specific factors as the explanatory variables (Rajan and Zingales, 1995; de Jong, Kabir, and Nguyen, 2008). The linear regression model used to estimate the relationship is formulated as:
Li = α +β1 SIZEi + β2 TANGi + β3 GROWTHi + β4 PROFITi
+ β5 VOLi +β6 TAXESi + β7 DLISTING + εi (1)
where Li is the leverage ratio for the i-th firm in the sample, with size equal to 211; SIZEi, TANGi,
GROWTHi, PROFITi, VOLi, TAXi is its logarithm of net sales, tangible assets, intangible assets,
profitability, volatility of profitability, and taxes, respectively; DLISTING is a dummy for a listing on the stock market; and α, β1, β2, β3, β4, β5, β6, andβ7 are the parameters of the linear model
specification.
All three measures of leverage, TDR, LDR, and SDR, are used as the dependent variable Li. The regression is performed separately for three time periods to detect whether the influence of
the firm-specific determinants differ between periods with different interest rates. First, the total sample period is used, where the independent variables cover the period from 2007 to 2014 and the dependent variable is the leverage ratio in 2015. Secondly, a regression is performed where the independent variables cover the period from 2012 to 2014 and the dependent variable is the leverage ratio in 2015. This period relates to the recent period after the financial crisis. Finally, a regression is performed where the independent variables cover the period from 2007 to 2010 and the dependent variable is the leverage ratio in 2011. This period relates to the period of the financial crisis and the first wave of declining interest rates (see figure I).
25
may occur, because public firms have easier access to public equity markets, have less information asymmetry problems, and have less ownership concentration. As a result, private firms have higher leverage ratios, because they have a lower ability to access the public equity market and have to rely more on debt (Brav, 2009).
Empirical analysis
This section discusses the results of analysis. First the leverage ratios at different points in time are considered, followed by the results of the OLS regressions that are performed to estimate the influence of the firm-specific factors on firm leverage.
Leverage ratios
26
Table II: Leverage among Dutch firms
Leverage ratios are calculated based on annual data and the values presented are the medians (means) for all firms in the common sample of the specific year. TDR represents the total debt ratio and is defined as total debt over total assets, LDR represents the long-term debt ratio and is defined as long-term debt over total assets, and SDR represents the short-term debt ratio and is defined as short-term debt over total assets.
Year Number of firms
Leverage ratio
TDR LDR SDR
2007 197 0.24 (0.26) 0.15 (0.18) 0.04 (0.08) 2011 169 0.18 (0.25) 0.15 (0.21) 0.00 (0.03) 2015 150 0.17 (0.22) 0.15 (0.18) 0.00 (0.04)
Linear regression of firm-specific factors on leverage
The results of the estimated model are presented in table III. Overall the explanatory power of the model is considerably higher for the period from 2007 to 2015 and for the period from 2012 to 2015 than for the period from 2007 to 2011. For the first two periods the adjusted R2 range from 0.11 to 0.58, while for the last period the adjusted R2 range from 0.03 to 0.20, where only the estimation with the SDR as dependent variable has an adjusted R2 above 0.05, i.e. 0.20. This suggests that the model that is used here and that uses several firm-specific factors that are found to be important determinants of leverage in other countries as explanatory variables, explains the differences in leverage among Dutch firms weakly to moderately during longer and more economic stable periods, but does a very poor job during economic turbulent periods. All explanatory variables that are used in the model, except for size, have a relationship with leverage. However, the relationships vary from weak to strong as is described further in this section.
27
period from 2012 to 2015 and by 0.21 in the period from 2007 to 2011. The hypothesis that the estimate of the relationship is equal to zero can be rejected at the 1% percent confidence level for the TDR and the LDR in the period from 2007 to 2015 and in the period from 2012 to 2015. The hypothesis that the estimate is equal to zero can be rejected at the 10% confidence level for the TDR and at the 5% confidence level for the LDR in the period from 2007 to 2011. The statistical significance, i.e. the rejection of the hypotheses that the estimates are equal to zero, and the economical relevance, i.e. the magnitudes of the estimates, suggest that the findings here are in line with theoretical predictions (Myers, 1977; Blazenko, 1987; Harris and Raviv, 1990) and former empirical findings (Rajan and Zingales, 1995; Gaud, Jani, Hoesli, and Bender 2005; Margaritis and Psillaki, 2007; de Jong, Kabir, and Nguyen, 2008; Harjeet, Tong, and Dogan, 2008; Brav, 2009). Collateral is important to obtain debt, and in particular to attract long-term debt. Although Degryse, de Goeij, and Kappert (2012) find higher estimation coefficients for the determinant tangibility among Dutch SMEs, 0.35 and 0.55 for the TDR and LDR respectively, the findings here support their suggestions relatively well since the positive effect of collateral on the TDR seem to stem mainly from the LDR.
28
Hoesli, and Bender, 2005; Daskalakis and Psillaki, 2008; de Jong, Kabir, and Nguyen, 2008; Brav, 2009). The estimates here are larger than the average of those in de Jong, Kabir, and Nguyen (2008) (estimates are between -0.07 and -1.079) and Degryse, de Goeij, and Kappert (2012) (the estimate is -0.04). This may be because of the fact that Degryse, de Goeij, and Kappert (2012) study only SMEs and use earnings before interest taxes, and depreciation (EBITD) over total assets as a proxy for profitability and de Jong, Kabir, and Nguyen, 2008 study only listed firms in industrialized and emerging countries and use long-term leverage as the dependent variable. However, since de same proxy for profitability is used here as in de Jong, Kabir, and Nguyen, 2008, the results here may also support the suggestions that profitability is more important for small firms than for large firms (Titman and Wessels, 1988; Michaelas, Chittenden, and Poutziouris, 1999; Degryse, de Goeij, and Kappert, 2011), because the sample used here includes SMEs and large firms. If this is the case, then the mixed sample that is used in this study may also explain why the negative effect of profitability on the TDR seem to stem from both the LDR and the SDR in this study, since it is found that small firms use more short-term debt than large firms (Titman and Wessels, 1988; Degryse, de Goeij, and Kappert, 2011; Faulkender, Flannery, Watson Hankins, and Smith, 2012).
There is some evidence that taxes negatively influence the TDR and the LDR in the period from 2007 to 2015 and the period from 2007 to 2011, but the economic relevance (estimates are between -0.01 and -0.03) is very small. Only the TDR and the LDR have statistical significance estimates and only the estimate for TDR in the period from 2007 to 2015 (-0.03) is significant at the 1% confidence level. The other three significant estimates are only significant at the 10% confidence level. The findings are in line with both, Cheng and Green (2008) who find a small but significant positive impact of tax rates on leverage for firms in the EU, and with Graham and Harvey (2001) who find that CFOs consider the corporate tax advantage of debt only as moderate important. The positive effect of tax rates on leverage in Cheng and Green (2008) implies a negative relationship between taxes paid and leverage and thus is line with findings here.
29
the same period with a value of -0.14 (and a S.E. of 0.06), and for the SDR in the period from 2012 to 2015 with a value of -0.08 (and a S.E. of 0.05). The economic relevance of growth opportunities for the TDR (with an estimate of -0.21) is greater than the economic relevance of growth opportunities for long-term leverage in de Jong, Kabir, and Nguyen (2008) (estimates are between -0.01 and -0.145). While the negative relationship between growth opportunities and the TDR is line with theories and most of the findings in the empirical studies that are reviewed here (Myers, 1977; Rajan and Zingales, 1995; Gaud, Jani, Hoesli, and Bender ,2005; Harjeet, Margaritis and Psillaki, 2007; Tong and Dogan, 2008; Brav, 2009), it is in contrast with some other empirical findings (Sogorb-Mira, 2005; Degryse, de Goeij, and Kappert, 2012). The negative relationship between growth opportunities and SDR is line with Sogorb-Mira (2005) who studies Spanish SMEs.
Although the economic relevance of a listing on the stock market is relatively low, the statistical significance for the positive relationship between listed firms and leverage is relatively strong for the TDR (estimate value is 0.09 with a S.E. of 0.04) and the SDR (estimate value is 0.06 with a S.E. of 0.02) in the period from 2007 to 2015 and the TDR (estimate value is 0.10 with a S.E. of 0.04) and SDR (estimate value is 0.07 with a S.E. of 0.02) in the period from 2012 to 2015 and for the SDR (estimate value is 0.08 with a S.E. of 0.01) in the period from 2007 to 2011. This is in contrast to the argument that public firms have lower leverage ratios and private firms have higher leverage ratios, because public firms can more easily access equity capital and private firms have to rely more on debt (Brav, 2009). The findings here show that listed firms also have slightly higher leverage and that probably this is because they can more easily attract short-term debt given the significance of the SDR in all three periods.
30
2008), it can also be viewed from other perspectives, since the positive relationship seem to apply especially to the SDR. If volatility increases risk of the firm and this makes it more difficult to attract long-term debt, than firms may tend to use more short-term debt if their borrowing capacity allows this. It is also possible that, since long-term interest rates were declining during this period, managers expected the interest rate to decline even further and therefore preferred to borrow short-term rather than long-term. This would be in line with the survey of Graham and Harvey (2001). Another possibility is that firms that have volatile earnings make more use of trade credit. This relates to the argument that short-term debt largely consists of trade credit which is under influence of different determinants than the firm-specific factors (de Jong, Kabir, and Nguyen, 2008).
In almost all of the regressions with SDR as the dependent variable, heteroscedasticity is detected when the white test for heteroscedasticity is performed. Therefore, heteroscedasticity robust standard errors are used for the estimation in these situations. This may be the reason that on average the estimations with SDR as dependent variable have higher R2 and that the estimates of these estimations have higher statistical significance. The estimation for the TDR in the period from 2007 to 2014 is also estimated with heteroscedasticity robust standard errors. The values of the white tests of the regressions are given in table V in the appendix.
31
Table III: Regression results
The regression model is estimated by computing OLS regressions for each leverage ratio, TDR, LDR, and SDR, as the dependent variable in the indicated year. The regressors, except for VOL, are the averages of the annual values of the variables of each firm during the indicated period. Standard errors are given in parentheses. SIZE represents size defined as the logarithm of sales, TANG represents tangibility defined as the ratio of fixed assets to total assets, GROWTH represents growth defined as the ratio of intangible assets to total assets, PROFIT represents profitability defined as the ratio of earnings before interest and taxes (EBIT) to total assets, TAXES represents taxation defined as the ratio of all taxes of an accounting year to profits before tax, and VOL represents the volatility of profitability defined as the standard deviation of profitability. DLISTING is a dummy variable that takes the value of one for listed firms and zero for private firms. The regressions include an intercept for which the values are not given here.
Panel A: regression of firm-specific determinants during 2007 to 2014 on firm leverage in 2015
Variable TDR LDR SDR SIZE 0.01 (0.01) 0.01 (0.01) 0.01 (0.00) TANG 0.25 (0.08) *** 0.30 (0.07) *** 0.00 (0.04) GROWTH -0.21 (0.11) * -0.01 (0.08) -0.14 (0.06) ** PROFIT -0.95 (0.32) *** -0.56 (0.19) *** -0.36 (0.18) ** VOL 1.13 (0.75) 0.08 (0.26) 1.25 (0.59) ** TAXES -0.03 (0.01) *** -0.02 (0.01) * -0.01 (0.01) DLISTING 0.09 (0.04) ** 0.02 (0.03) 0.06 (0.02) *** Adjusted R² 0.27 0.14 0.47 Number of observations 145 204 145
Panel B: regression of firm-specific determinants during 2012 to 2014 on firm leverage in 2015
Variable TDR LDR SDR SIZE 0.01 (0.01) 0.00 (0.01) 0.00 (0.00) TANG 0.28 (0.09) *** 0.28 (0.06) *** 0.03 (0.03) GROWTH -0.13 (0.10) 0.02 (0.08) -0.08 (0.05) * PROFIT -0.76 (0.20) *** -0.36 (0.16) ** -0.38 (0.13) *** VOL 0.82 (0.28) *** 0.07 (0.22) 0.85 (0.30) *** TAXES 0.00 (0.01) -0.01 (0.01) 0.00 (0.00) DLISTING 0.10 (0.04) *** 0.01 (0.03) 0.07 (0.02) *** Adjusted R² 0.33 0.11 0.58 Number of observations 148 210 148
Panel C: regression of firm-specific determinants during 2007 to 2010 on firm leverage in 2011
Variable TDR LDR SDR SIZE -0.01 (0.01) 0.00 (0.01) -0.01 (0.00) * TANG 0.21 (0.12) * 0.21 (0.09) ** 0.02 (0.02) GROWTH 0.05 (0.13) 0.08 (0.11) -0.01 (0.03) PROFIT -0.63 (0.32) ** -0.55 (0.24) ** 0.00 (0.06) VOL -0.50 (0.51) -0.31 (0.42) -0.06 (0.11) TAXES -0.03 (0.02) * -0.03 (0.02) * 0.00 (0.00) DLISTING 0.05 (0.05) -0.02 (0.04) 0.08 (0.01) *** Adjusted R² 0.03 0.05 0.2 Number of observations 165 205 165
32
Robustness of the results to different measures of leverage
The effect of firm-specific variables is also tested separately on a market-based measure of leverage. The total sample here includes only 73 public listed firms. This number decreases when the firms are analyzed due to a lack of information for all variables. To calculate the market leverage, the (book) value of total assets is replaced with the market value of total assets. To obtain market values of firms’ total assets, the book value of equity as a component of total assets is replaced with the market value of equity. Market capitalization is used as a measure of the market value of equity. The same methodology and linear regression model as before are used, with the exception of the dummy variable for listed firms. The model is formulated here as:
MLi = α +β1 SIZEi + β2 TANGi + β3 GROWTHi
+ β4 PROFITi + β5 VOLi +β6 TAXESi + εi (2)
The three measures of leverage, that are used as dependent variable MLi, are now TDRM, LDRM
and SDRM, which are the ratio of total debt to the market value of total assets, the ratio of long-term debt to the market value of total assets, and the ratio of short-long-term debt to the market value of total assets, respectively. The results of the regressions are presented in table IV in the appendix.
The explanatory power of the model for the different time periods that are used suggests the same as before. While the model has a low explanation power, i.e. low adjusted R2, on average the adjusted R2 for the estimations in the period from 2007 to 2015 and from 2012 to 2015 are higher than those for the estimations in the periods from 2007 to 2011. For the first two periods the adjusted R2 range from 0.06 to 0.27, with only one value of 0.06 and the rest with a value between 0.18 to 0.27. In the third period used, the adjusted R2 range from 0.02 to 0.12. The suggestion is that also in the case of using market leverage and listed firms the model with the firm-specific factors that is used here explains leverage among Dutch listed firms relatively better in longer periods and periods with more economic stability. The listed firms used here, are Dutch firms, but are not necessarily listed on the Dutch stock market. Some of them are listed on other markets such as the NYSE or stock markets in other European countries.
