Determinants of capital structure : an empirical analysis of Dutch listed firms

(1)

UNIVERSITY OF AMSTERDAM

Determinants of Capital

Structure

An Empirical analysis of Dutch listed firms

Reinout Mensing

1/2/2014

(2)

1. Introduction

Engaging in business activities is essential for firms to generate revenue and to satisfy customers. The sources of funding that firms utilize to finance these business activities constitute their capital structure. A firm’s choice of capital structure is a valuable tool in the pursuit of value maximization. Having the potential of increasing profitability, capital structure has become an extensively researched field of finance (Bradley, Jarrel, & Kim, 1983). Four theories on capital structure prominently prevail in the existing literature.

The capital structure irrelevance theory, proposed by Modigliani and Miller in 1958, has made the discussion concerning capital structure decisions prosper in today’s global economy. The trade-off theory, pecking order theory and principal-agent theory followed up on the capital structure irrelevance theory, all approaching the capital structure puzzle from a different angle (Myers, 1984). The trade-off theory explains the capital structure by balancing possible tax benefits from debt financing with financial distress cost of leverage (Kraus & Litzenberger, 1973). The pecking order theory describes an order of preference regarding sources of funding (Myers, 2001) and the principal-agent theory is based on agency costs. Uniting the different theories on capital structure is the fundamental assumption that firms which to maximize their utility. Firms balance the possible advantages and disadvantages in an effort to find their optimal capital structure.

The Dutch economy is a small, open and internationally orientated economy. Due to the stable economic environment and institutional settings, countless multinationals are based in the Netherlands (van Dijk & Weyzig, 2007). Because of the European Union, the Dutch institutional settings are comparable to other European countries such as Germany (Chen, Lensink, & Sterken, 1998). However, corporate governance structures, possible tax advantages and other country specific characteristics can deviate. Furthermore, the underdevelopment of the Dutch corporate bond market is of substantial influence on the financing decisions for Dutch firms (De Haan & Hinloopen, 2003) .

With ever changing economic environments, research in this field rapidly becomes obsolete. The last significant large-scale study on Dutch capital structure decisions was

conducted by De Haan & Hinloopen and dates back to 1994. Turmoil on the financial markets over the last decade may have changed firm’s financing behavior. The bankruptcy of

powerful companies such as Lehman Brothers might have caused firms to develop a more conservative perception of liquidity, solvency and consequently capital structure (Bordo,

(4)

2008). The changing economic setting potentially alters the determinants of capital structure and thus an empirical analysis of the recent years is required as a supplement to existing research.

A firm’s capital structure decisions are influenced by countless firm determinants. Among these factors are managerial behavior, asset structure, profitability, corporate governance structure, institutional settings and the tendency of the market (Stonehill, et al., 1975). This thesis researches the extent to which firm characteristics are determinant for capital structure decisions in the Netherlands. In order to identify potentially significant firm characteristics, this thesis will first review the existing theories on capital structure. Once the possible determinants are identified, a statistical analysis of firms listed on the Euronext Amsterdam, examines the degree to which they explain capital structure decisions for Dutch firms.

The empirical model used to explain capital structure decisions in the Netherlands is of statistical significance. The most determinant firm characteristics are profitability, liquidity and the non-debt tax shield. This thesis provides evidence for the existence of a financial pecking order in the Netherlands but does not confirm the hypotheses formed by the principal-agent theory and the trade-off theory.

(5)

2. Related Literature

2.1 Capital structure

The capital structure is defined by the relative proportion of equity and debt that a firm uses to finance its activities. Equity financing is a source of funding in which capital is

obtained through the issuance of shares. The firm sells a stake of ownership in order to raise capital. Debt financing is a source of funding in which a firm borrows money from an investor, to whom, the initial amount borrowed plus interest will be reimbursed at maturity (Berk & DeMarzo, 2011). A common indicator for the proportion of sources of funding is the leverage ratio. Capital structure theory attempts to explain the mix of securities and financial resources used by corporations to finance their activities (Myers, Capital Structure, 2001). 2.2 Modigliani-Miller theory

Since the publication of their influential paper, Modigliani and Miller are considered the founders of modern capital structure theory. Assuming perfect and frictionless capital markets, Modigliani and Miller proved that a firm’s value is unaffected by the choice of capital structure (Myers, Capital Structure, 2001). For this reason, the Modigliani-Miller theorem is also known as the Capital structure irrelevance theorem. The first proposition of the Modigliani-Miller theorem states that “the market value of any firm is independent of its capital structure and is given by capitalizing its expected return at the rate Pk appropriate to its

class” (Modigliani & Miller, 1958).

In order for this proposition to be valid, several conditions must hold. The first

condition is that investors and firms can trade the same set of securities at competitive market prices equal to the present value of their future cash flows. The second condition presumes that there are no taxes, transaction costs, or issuance costs associated with security trading. The third condition states that a firm’s financing decisions do not change the cash flows generated by its investments, nor do they reveal new information about them (Berk &

DeMarzo, 2011). When these three conditions hold the total market value of a firm is equal to the market value of its assets, whether the firm is unlevered or levered. Even though these assumptions about perfect capital markets are not met in the real economy, Modigliani and Miller laid the foundations for modern capital structure theory (Myers, Capital Structure, 2001).

(6)

2.3 Trade-off theory

One of the imperfections of the real capital market is corporate tax. Firms are obliged to pay taxes on the income they have earned from their investments. As corporate tax payments violate the perfect capital market assumption of Modigliani and Miller, capital structure is no longer irrelevant. The interest expense resulting from debt finance is tax deductible. Therefore the amounts of debt firms decide to take on when financing their activities can be used as a tool to minimize tax payments. The amount by which taxable income is reduced is called the

tax shield (Berk & DeMarzo, 2011). The

trade-off theory maximizes the tax advantage form interest payments, while taking into account, the costs of financial distress (Kraus & Litzenberger, 1973). Financial distress costs are costs incurred when firms finance their using debt instead of equity funding. A higher leverage ratio is associated with higher financial distress costs due to the increasing

probability of bankruptcy. Aside from the direct legal and administrative costs of bankruptcy, financial distress can have severe consequences for a company. Loss of customers, loss of suppliers, fire sale of assets and loss of receivables are listed by Berk & DeMarzo (2011) as possible consequences of financial distress.

Firms wish to exploit the possible tax advantage from debt financing by maximizing the tax shield. Interest expense however, is not the only tax-deductible expense. The taxable income is also determined by depreciation and amortization expense. When depreciation and amortization expenses are deducted from the taxable income, the possible tax advantage that can be gained through the deduction of interest expense is reduced (Titman & Wessels, 1988).

The trade-off theory diverges into the static and dynamic trade-off model. The static trade-off model balances the tax benefits and financial distress costs for separate accounting periods while the dynamic trade-off model implies that companies should target their optimal leverage ratio that maximizes tax benefits and minimizes financial distress costs and adjust their financing behavior to adhere to the target leverage ratio. This Optimal leverage ratio is achieved when the marginal tax benefits equal the marginal costs of debt (Myers, 1984).

(7)

2.3.1 The trade-off theory and the Dutch capital market

The Netherlands is a small open and internationally orientated economy. The Dutch capital market is subject to corporate legislation specifically designed for the Netherlands. Firms based in the Netherlands adjust their financing behavior to the Dutch institutional and legislative setting in an effort to minimize their costs. The Netherlands is often considered a tax haven for international corporations (van Dijk & Weyzig, 2007). Other than economical and financial stability, the Netherlands offers a favorable tax framework. Multinationals from all over the world are based in the Netherlands to minimize their tax payments. The corporate tax rate in the Netherlands is relatively low in international perspective (Chen, Lensink, & Sterken, 1998). In addition, interest expenses on debt are 100% tax deductible in the Netherlands.

The regulation on provision for bad debt and pension liability, found on the consolidated balance sheet, is an example of the unique corporate tax environment of the Netherlands. Consistent with the Dutch legislation, the provision for bad debt and pension liability can either be subtracted from the account receivables or be 100% deducted from the income. Making up a significant portion of the balance sheet, the provision for bad debt and pension liability substantially reduces the taxable income (Chen, Lensink, & Sterken, 1998). Through earnings and capital structure management, firms use this regulation to their

advantage. The extensive utilization of the law on provision for bad debt and pension liability suppresses the use of debt (McNichols & Wilson, 1988).

According to Chen and Jiang (2001), bankruptcy costs are relatively high in the Netherlands. This is due to the relatively advantageous position of the creditor in Dutch bankruptcy law. The Dutch bankruptcy law system is a liquidation-based system in which payments to creditors are of relatively high priority compared to the reorganization of the bankrupt firm. However, the institutional settings are of minor importance when explaining the probability of bankruptcy. A large scale study conducted by Couwenberg and de Jong (2008) showed that firm characteristics are the major determinants for the probability of bankruptcy. Firm size, asset structure and capital structure are found to have a strong effect on the probability of bankruptcy. A higher probability of bankruptcy increases bankruptcy costs and affects the trade-off decision between tax benefits and financial distress costs.

.

(8)

2.4 Pecking order theory

The financial pecking order theory explains the order of preference of financing sources. Meyers (2001) assumed capital markets with asymmetric information. The problem of asymmetric information arises when investors do not know the true value of the existing assets or investment opportunity as opposed to firm managers who have more information available (Myers, Capital Structure, 2001). The information asymmetry is one of the explanatory factors in the financial pecking order theory.

According to Stewart Meyers, firms prefer internal finance rather than external finance. However, if external finance is required, firms first issue debt and use equity finance as a last resort (Myers, The capital Structure Puzzle, 1984). Sources of internal funding are retained earnings or the liquidation of excess assets while external finance consist of bank loans, corporate bonds and share capital. The reason that internal finance is preferred over external finance is that external finance has higher costs. Aside from the administrative and underwriting costs, external financing is subject to asymmetric information cost.

Manager’s information about the financial position of the firm is likely to be superior to that of outside investors (Berk & DeMarzo, 2011). Managers give signals about the financial position of the firm through their actions. When a firm is in need of external

financial resources it can choose to issue debt or equity. Issuing new shares signals ambiguous information about the firm’s financial position to outside investors. Investors might interpret the issuance of new shares as a sign that the firm has lucrative investment opportunities. However, investors could also interpret the issuance of new shares as a sign that the firm is looking to sell its overpriced equity. This information asymmetry between managers and investors causes the equity of healthy firms with good investment opportunities to be undervalued (Myers, Capital Structure, 2001).

The ambiguous signal causes share prices to drop. Assuming that managers act in the interest of existing shareholders, the manager will refuse to sell undervalued shares. As a consequence of information asymmetry firms would rather issue debt. Because debt has a prior claim on assets, it minimizes the risk of asymmetric information for investors (Myers, Capital Structure, 2001). The costs arising from equity finance due to the information asymmetry cause debt to be of higher preference in the financial pecking order.

(9)

2.4.1 The pecking order theory and the Dutch capital market

Several empirical studies have confirmed the presence of a financial pecking order in the Netherlands. In accordance with pecking order theory, the Dutch economy experienced a major increase in liquidity holdings in the period from 1981 to 1990. Companies buffered their profits, doubling the ratio of broad money to national income. This development indicates the need for financial stability and increasing preference for internal financing sources (De Haan, Koedijk, & Vrijer, 1994). The development of increased money holdings is in line with the financial pecking order theorem. A study by De Haan (1994) showed that 54% of all Dutch firms prefer internal capital.

The strong preference of internal finance over external finance by Dutch firms was confirmed by a study of the Dutch national bank. A binomial logit regression analysis for the choice of internal or external finance on a sample of Dutch firms from 1984 to 1997 proved significant preference for internal finance. The use of external finance depended heavily on the availability of internal cash flow and the liquidity of assets (De Haan & Hinloopen, 1999). If external financing sources were necessary, debt financing was preferred over equity

financing. A reason for this might be that share issues are relatively costly and difficult to place in the Netherlands (Stonehill, et al., 1975).

According to the empirical study of the Dutch financing decisions by De Haan and Hinloopen (2003), Dutch firms have a unique financing pecking order: internal finance, bank loans, share issues and bond issues. As is the case for most countries, the most preferred source of finance, is internal finance. The main source of external finance in the Netherlands is bank credit; therefore the Dutch economy can be described as bank-based. About 97% of direct debt financing to the private sector in the Netherlands comes from banks or other financial institutions (Chen & Jiang, 2001) . Corporate bonds are a minor source of funding due to the relatively underdeveloped Dutch corporate bond market (De Haan & Hinloopen, 2003). The fact that share issues are higher in the hierarchy of sources of funding is in conflict with the traditional pecking order theory.

(10)

2.5 Principal-agent theory

Stock market listed firms aim to take the interests of their different stakeholder into account. The stakeholders in the agency theory are corporate management, shareholders and debt holders. Conflicts of interest are mainly caused by the separation of ownership, management and control. Each stakeholder has different incentives and is most concerned with his own value maximization. The classical conflict of interests between corporate management and shareholders is often referred to as the principal-agent problem (Jensen & Meckling, 1976). In this case, corporate management is the agent, while the shareholders are the principal. The agency costs resulting from these conflicts have their impact on capital structure.

The divergence of ownership and control in a professionally managed firm might result in managers failing to maximize firm value to the discontent of shareholders (Berger & Bonaccorsi di Patti, 2006). Jensen & Meckling (1976) assume that managers aim to maximize their own utility. The conflict of interests between managers and shareholders can take several forms. With high levels of free cash flow, managers might invest in building their own empire rather than investing in projects that maximize shareholder value. In case of the

overinvestment problem, managers might invest in projects with negative net present value in an effort to expand their power and control (Jensen, 1986).

As an incentive to undertake value enhancing activities, managers are partly compensated based on their performance (Fama, 1980). However, this might motivate managers to invest in short-term projects with early results that enhance their reputation and payments, instead of investing in more profitable long-term projects. Another problem arising from conflicting interests of managers and share holders is the investment in projects, taking on unnecessary risks of destroying shareholder value, in an effort to enhance management’s bonus payments (Niu, 2008).

According to Jensen (1986), increasing the leverage ratio can reduce agency costs. Since debt forces managers to pay out cash, levering up reduced the amount of free cash flow available to managers. If managers spend the free cash flow on wasteful expenditures, the probability of default on debt payments increases, weakening the manager’s position within the firm. This prevents managers to invest in projects with negative net present value and provides stimulation to utilize assets efficiently (Kochhar, 1996). The controlling role of debt in an organization with separation of management and ownership helps to align the interest of the different stakeholders.

(11)

Another form of agency costs arises from the conflict of interest between shareholders and debt holders. The phenomenon of this conflict is the transfer of wealth from bondholders to shareholders due to decisions made by shareholders or their representatives (Niu, 2008). Because shareholders and their representatives can influence company decisions, they have the power to increase their wealth at the expense of bondholders. First, shareholders can reduce planned investments, and payout dividends using the withheld investment financing capital. Furthermore, shareholder can increase dividend payments by issuing new senior debt with higher priority (Kalay, 1982). Issuing new debt increases the risk profile for bondholders and transfers their wealth to shareholders.

Secondly, debt contracts give incentive to shareholders to invest in risky assets. The substitution of existing assets for riskier assets is known as the asset substitution problem. The excess value created when these investments are successful will be owned by shareholders, while the consequences of failure are mainly born by debt holders due to shareholder’s limited liability (Niu, 2008). Investing in risky projects with negative net present value in which the value decrease consists of a decrease of the value of debt and an increase in the value of equity is described as the overinvestment problem or the moral hazard problem. Thirdly, shareholders can refrain from investing in low risk assets that generate steady cash flows. Not investing in safe projects with positive net present value in which the value increase consists of an increase in the value of debt and a decrease in the value of equity is commonly known as the underinvestment problem (Parrino & Weisbach, 1999).

Bondholders will take the possibility of wealth transfers and increased risk profiles into account when signing debt contact. The opportunistic behavior by shareholders and their representatives cannot be pursued with impunity. This effectively means that bondholders will demand a higher risk premium, thereby increasing the cost of debt financing. The solution to this agency problem would less debt financing, thereby decreasing the leverage ratio and influencing the firm’s capital structure (Niu, 2008).

However, the magnitude of the impact of these agency costs on capital structure is widely disputed. In their research, renowned economists such as Fama, Miller and Myers acknowledge the existence of agency costs but describe the impact on capital structure as relatively unimportant. Research conducted by Gertner and Scharfstein proved that agency costs are mainly of significant influence on companies in financial distress (Gertner & Scharfstein, 1991). Furthermore, established firms with good track records of meeting debt obligations are substantially less subject to agency costs than young firms with little reputation (Chen, Lensink, & Sterken, 1998).

(12)

2.5.1 The principal-agent theory and the Dutch capital market.

The agency theory depends heavily on the corporate governance structure. The role of debt and equity holders therefore determines the relevance of the agency theory in the Netherlands. When examining the conflicts of interest between the different stakeholders, the degree to which shareholders can influence and control management is decisive in determining the impact of agency costs. Especially in comparison with Germany, Dutch shareholdings of companies are widely dispersed (Chen, Lensink, & Sterken, 1998). This effectively means that Dutch firms have many small shareholders instead of several large shareholders. As large shareholders have the ability to exert more power, this implies that the agency theory in the Netherlands is of limited impact on capital structure decisions.

The ownership structure in the Netherlands is rather dispersed: individual 19%, private companies 19%, foreign investors 37%, institutional investors 22% and banks 2% (Corhay & Rad, 2000). The minor role of shareholders in the Netherlands is confirmed by the limited role pension fund have in corporate governance. Ranked as institutional investor, pension funds own about 8% of all corporate shares in the Netherlands. Hence, pension funds could be of significant influence on corporate policy. However, pension funds are mainly interested in the safeness of their investments rather than being of influence on corporate management (Chen, Lensink, & Sterken, 1998).

The agency costs of managers maximizing their utility at the expense of shareholder value are negligible in the Netherlands. This can be attributed to the Dutch corporate

governance structure of an executive and supervisory board. The supervisory board consists of outside experts like bankers, retired politicians and corporate executives of other firms. Compared to similar economies, the supervisory board in the Netherlands has a more

extensive role. Namely, the supervisory board is legally obliged to guard the firm as a whole, not only the interests of a particular stakeholder (Corhay & Rad, 2000). The extended duty of the supervisory board restricts the opportunistic behavior of managers and thereby suppresses possible agency costs.

The country specific characteristics of the Netherlands make that agency costs are of minor importance in explaining capital structure decisions. Moreover, Chen, Lensink and Sterken (1998) argue that agency costs hardly exist and plead for the irrelevance of the agency theory for Dutch firms.

(13)

3. Hypotheses

The different theories on capital structure provide insights in the determinants of capital structure decisions. Over the years, economists have conducted empirical studies to test the degree to which these theories apply to specific stock markets, countries and industries. Hypotheses about capital structure determinants can be formulated when considering the existing theoretical and empirical research on capital structure and taking into account the Dutch institutional setting and the country specific characteristics. The hypotheses will describe the expected relation of firm characteristics with the capital structure. The proxy for capital structure is the leverage ratio.

Asset tangibility

Tangibility of assets is defined as the degree to which assets have a physical nature. Asset tangibility is considered to be a capital structure determinant due to the influence it has on a firm’s risk profile. Tangible assets can serve as collateral for debt financing, and therefore provide security to potential investors. As the riskiness of providing capital decreases for investors, so will the interest rate (Titman & Wessels, 1988). Furthermore, tangible assets are expected to retain more value in liquidation. The reduced costs of capital enable firms to operate with higher levels of debt, without experiencing increased financial distress costs.

The trade-off theory predicts that firms with high asset tangibility will generally have a higher leverage ratio (Rajan & Zingales, 1995). The agency theory is based on the presence of information asymmetry. When tangible assets function as collateral in debt contracts, information asymmetry is minimized. The result is decreased agency costs. As tangible assets provide security to investors and minimize information asymmetry, tangibility is expected be positively related to leverage ratio.

Non-debt tax shield

The trade-off theory suggests that companies attempt to maximize their tax benefits created by the tax shield. Interest expense is not the only tax-deductible expense that is tax deductible. As depreciation and amortization costs are also tax deductible, the size of the non-debt tax shield is expected to have a negative relation to leverage ratio (Titman & Wessels, 1988).

Growth opportunity

(14)

According to Berk & DeMarzo, managers communicate information about the firm’s financial position to investors through signals. Because of information asymmetry between investors and managers, these signals are interpreted ambiguously. Resulting from this, firms with growth opportunities are likely to have undervalued equity. Because managers are averse to selling their undervalued equity, firms with growth opportunities are more likely to finance their activities with debt, increasing the leverage ratio (Berk & DeMarzo, 2011).

In contradiction, the agency theory suggests that firms with growth opportunities, which have more flexibility in their choice of future investments, are more likely make suboptimal investment decisions in an effort to transfer value from debt to equity holders (Myers, Capital Structure, 2001). The relation between growth opportunity and leverage should thus be negative. Moreover, growth opportunities are capital assets that are not available to serve as collateral and do not generate taxable income. Resulting from this are increased agency costs and confirmation for the negative relation between growth opportunity and leverage ratio (Titman & Wessels, 1988).

Firm size

According to the off theory, firm size is positively related to leverage ratio. The trade-off between financial distress costs and tax benefits is dependent on firm size because of the influence of firm size on stability. Large firms are assumed to have relatively stable cash flows and well diversified portfolios. Due to well established sources of income and stability of large firms, default risk on debt contracts decreases as firm size increases.

Small firms on the other hand are expected to have more volatile and unstable earnings, which makes the firm’s ability to meet its debt obligations unreliable (Titman & Wessels, 1988). As a compensation for the increased default risk, investors will demand an increased risk premium on their interest rates. Higher interest rates cause the costs of debt capital to rise and thus, the costs of debt financing are generally higher for small firms. As a result, the relation between firm size and leverage ratio is expected to be positive (Rajan & Zingales, 1995).

(15)

Profitability

According to the pecking order theory, firms prefer internal finance over external finance. When external finance is required, debt is of higher preference than equity (Myers, Capital Structure, 2001). Profitable firms have the opportunity to retain earnings and can therefore finance a larger proportion of their activities using internal finance. Less profitable firms have less internal capital available and are thus expected to attract more external capital. As a result, profitability is hypothesized have a negative relation to leverage ratio (Titman & Wessels, 1988).

Liquidity

According to the pecking order theory firms prefer internal finance over external finance. This implies that firms with sufficient cash and cash equivalents available will be less inclined to use external financing sources. As a result, highly liquid firms are expected to have a

relatively low leverage ratio. Consequently, the relation between liquidity and leverage ratio is expected to be negative (Berk & DeMarzo, 2011).

Tax rate

The corporate tax rate is determinant for the potential value of the tax shield. A high corporate tax rate equals a large potential tax shield. As the corporate tax rate increases, the leverage ratio is expected to increase as well. Therefore, the hypothesized relationship between tax rate and leverage ratio is positive (Kraus & Litzenberger, 1973). According to KPMG global, the corporate tax rate in the Netherlands was about 25% in the period 2002-2012.

Table 1: Summary of hypotheses

Firm Characteristic Hypothesized relation to leverage ratio

Asset tangibility Positive

Non-debt tax shield Negative

Growth opportunity Negative/Positive

Firm size Positive

Profitability Negative

Liquidity Negative

Tax rate Positive

(16)

4. Methodology

4.1. Data

The source of the dataset, used for the empirical analysis of capital structure determinants for Dutch listed firms, is the online financial database; Wharton Research Database Services. The firms included in the sample are Dutch firms listed on the Euronext Amsterdam stock

exchange. The sample constitutes 25 of the largest Dutch listed firms, measured over the period 2002-2012.

4.2 Variables

The related literature provides insight in the firm characteristics that are presumed to have a relation to capital structure decisions. The dependent variable leverage ratio is described by two different proxies, the debt-to-equity ratio and the debt-to-asset ratio. Most proxies used to calculate the explanatory variables can de naturally derived from their definitions. However, as growth opportunity and size can take on several definitions, two different sets of variables that are commonly used in literature are utilized to describe their relation to the dependent variable. The proxies used for the variables that represent the different firm characteristics are as following: Leverage ratio: 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 = 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 𝐿𝐿𝐿𝐿𝑟𝑟𝑟𝑟𝑟𝑟 = 𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝑇𝑇𝑟𝑟𝑙𝑙𝐿𝐿 𝑟𝑟𝐿𝐿𝐿𝐿𝑡𝑡 𝑑𝑑𝐿𝐿𝑑𝑑𝑟𝑟_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎} 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 = 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 𝐿𝐿𝐿𝐿𝑟𝑟𝑟𝑟𝑟𝑟 =_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝑐𝑐𝑟𝑟𝑡𝑡𝑡𝑡𝑟𝑟𝑙𝑙 𝐿𝐿𝑒𝑒𝑒𝑒𝑟𝑟𝑟𝑟𝑒𝑒}𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝑇𝑇𝑟𝑟𝑙𝑙𝐿𝐿 𝑟𝑟𝐿𝐿𝐿𝐿𝑡𝑡 𝑑𝑑𝐿𝐿𝑑𝑑𝑟𝑟 Asset tangibility: 𝑇𝑇𝐿𝐿𝑇𝑇 = 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟 𝑟𝑟𝐿𝐿𝑙𝑙𝐿𝐿𝑟𝑟𝑑𝑑𝑟𝑟𝑇𝑇𝑟𝑟𝑟𝑟𝑒𝑒 = 𝑃𝑃𝐿𝐿𝑟𝑟𝑃𝑃𝐿𝐿𝐿𝐿𝑟𝑟𝑒𝑒, 𝑃𝑃𝑇𝑇𝐿𝐿𝑙𝑙𝑟𝑟 & 𝐿𝐿𝑒𝑒𝑒𝑒𝑟𝑟𝑃𝑃𝑡𝑡𝐿𝐿𝑙𝑙𝑟𝑟_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎} 16

(17)

Non-debt tax shield: 𝑇𝑇𝐿𝐿𝑇𝑇𝑁𝑁 = 𝑇𝑇𝑟𝑟𝑙𝑙 − 𝑑𝑑𝐿𝐿𝑑𝑑𝑟𝑟 𝑟𝑟𝐿𝐿𝑡𝑡 𝑎𝑎ℎ𝑟𝑟𝐿𝐿𝑇𝑇𝑑𝑑 = 𝐿𝐿𝐿𝐿𝑃𝑃𝐿𝐿𝐿𝐿𝑐𝑐𝑟𝑟𝐿𝐿𝑟𝑟𝑟𝑟𝑟𝑟𝑙𝑙 & 𝐿𝐿𝑡𝑡𝑟𝑟𝐿𝐿𝑟𝑟𝑟𝑟𝑎𝑎𝐿𝐿𝑟𝑟𝑟𝑟𝑟𝑟𝑙𝑙_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎} Growth opportunity: 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐿𝐿𝑃𝑃𝐿𝐿𝐺𝐺 = 𝐺𝐺𝐿𝐿𝑟𝑟𝐺𝐺𝑟𝑟ℎ 𝑟𝑟𝑃𝑃𝑃𝑃𝑟𝑟𝐿𝐿𝑟𝑟𝑒𝑒𝑙𝑙𝑟𝑟𝑟𝑟𝑒𝑒 =𝐺𝐺𝐿𝐿𝑃𝑃𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑡𝑡𝑃𝑃𝐿𝐿𝑙𝑙𝑑𝑑𝑟𝑟𝑟𝑟𝑒𝑒𝐿𝐿𝐿𝐿𝑎𝑎_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎} 𝐺𝐺𝐺𝐺𝐺𝐺𝑃𝑃𝐿𝐿 = 𝐺𝐺𝐿𝐿𝑟𝑟𝐺𝐺𝑟𝑟ℎ 𝑟𝑟𝑃𝑃𝑃𝑃𝑟𝑟𝐿𝐿𝑟𝑟𝑒𝑒𝑙𝑙𝑟𝑟𝑟𝑟𝑒𝑒 = (𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝑐𝑐𝑟𝑟𝑡𝑡𝑡𝑡𝑟𝑟𝑙𝑙 𝐿𝐿𝑒𝑒𝑒𝑒𝑟𝑟𝑟𝑟𝑒𝑒 𝑁𝑁ℎ𝐿𝐿𝐿𝐿𝐿𝐿𝑎𝑎 𝑟𝑟𝑒𝑒𝑟𝑟𝑎𝑎𝑟𝑟𝐿𝐿𝑙𝑙𝑑𝑑𝑟𝑟𝑙𝑙𝐿𝐿)_{𝐿𝐿𝐿𝐿𝐿𝐿𝑙𝑙𝑟𝑟𝑙𝑙𝐿𝐿𝑎𝑎 𝑃𝑃𝐿𝐿𝐿𝐿 𝑎𝑎ℎ𝐿𝐿𝐿𝐿𝐿𝐿}⁄ Firm size: 𝑁𝑁𝑆𝑆𝑆𝑆𝐿𝐿𝐿𝐿𝑆𝑆𝑃𝑃 = 𝐹𝐹𝑟𝑟𝐿𝐿𝑡𝑡 𝑎𝑎𝑟𝑟𝑎𝑎𝐿𝐿 = ln (𝐿𝐿𝑡𝑡𝑃𝑃𝑇𝑇𝑟𝑟𝑒𝑒𝐿𝐿𝐿𝐿𝑎𝑎) 𝑁𝑁𝑆𝑆𝑆𝑆𝐿𝐿𝑁𝑁𝐿𝐿𝐿𝐿𝐿𝐿 = 𝐹𝐹𝑟𝑟𝐿𝐿𝑡𝑡 𝑎𝑎𝑟𝑟𝑎𝑎𝐿𝐿 = ln (𝑙𝑙𝐿𝐿𝑟𝑟 𝑎𝑎𝐿𝐿𝑇𝑇𝐿𝐿𝑎𝑎) Profitability: 𝑃𝑃𝐺𝐺𝐺𝐺𝐹𝐹 = 𝑃𝑃𝐿𝐿𝑟𝑟𝑝𝑝𝑟𝑟𝑟𝑟𝐿𝐿𝑑𝑑𝑟𝑟𝑇𝑇𝑟𝑟𝑟𝑟𝑒𝑒 =_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎}𝐺𝐺𝐿𝐿𝐿𝐿𝐿𝐿𝑙𝑙𝑒𝑒𝐿𝐿 Liquidity: 𝐿𝐿𝑆𝑆𝐿𝐿 = 𝐿𝐿𝑟𝑟𝑒𝑒𝑒𝑒𝑟𝑟𝑑𝑑𝑟𝑟𝑟𝑟𝑒𝑒 = 𝐺𝐺𝐿𝐿𝑎𝑎ℎ & 𝑐𝑐𝐿𝐿𝑎𝑎ℎ 𝐿𝐿𝑒𝑒𝑒𝑒𝑟𝑟𝐿𝐿𝐿𝐿𝑇𝑇𝐿𝐿𝑙𝑙𝑟𝑟𝑎𝑎_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎} Tax rate: 𝑇𝑇𝐿𝐿𝐺𝐺 = 𝑇𝑇𝐿𝐿𝑡𝑡 𝐿𝐿𝐿𝐿𝑟𝑟𝐿𝐿 = 𝑐𝑐𝑟𝑟𝐿𝐿𝑃𝑃𝑟𝑟𝐿𝐿𝐿𝐿𝑟𝑟𝐿𝐿 𝑟𝑟𝐿𝐿𝑡𝑡 𝐿𝐿𝐿𝐿𝑟𝑟𝐿𝐿 17

(18)

4.3 Regression model

A multiple regression analysis is utilized to estimate the relation between firm characteristics and leverage ratio. The leverage ratio is the dependent variable, whereas the capital structure determinants are explanatory variables. The relevant data were retrieved from the Wharton Research Data Services Database and imported to Windows Excel. Subsequently, the data were converted into the variables as described in chapter 4.2. The formatted Excel spreadsheet was imported to the statistical analysis program STATA, which was used to produce the statistical output.

A multiple regression analysis explains the degree to which the dependent variable is influenced by the explanatory variables. For each explanatory variable, a statistical relevance test is performed. The statistical test produces coefficients of regression, standard errors, t-statistics, p-values and confidence intervals for each of the explanatory variables. These measures describe the degree to which the movements in the dependent variable are explained by a particular explanatory variable. When the explanatory variable is of statistical

significance, tested at 5% significance level, the absolute value of the t-statistic is larger than 1.96 and the p-value is smaller than 0.05. All the statistical tests throughout the entire thesis, test at a significance level of 5%. Both regression analyses employ the option of robust standard errors, which corrects for heteroskedasticity.

In a multiple regression analysis, with more than one explanatory variable an overall test of the model is conducted. The F-test measures whether the model, formed by the explanatory variables, has a relation to the dependent variable. The null-hypothesis for the F-test is that all the regression coefficients are equal to zero and the alternative hypothesis is that one or more of the regression coefficients significantly deviates from zero:

H0: β1 = β2 = … = βn = 0

H1: β1 and/or β2 and/or … βn ≠ 0

In the regression analyses of the sample, the F-statistic is distributed F (8, 218). The critical, value at 5% significance, for the F-statistic with degrees of freedom F (8,218) is 1.98. If the value of the F-statistic exceeds 1.98, the statistical model has proven to be of statistical significance in explaining the dependent variable. Concluding, the null-hypothesis is rejected if F > 1.98.

(19)

The F-test is a measure of the relevance of the overall model. The significance of each individual explanatory variable can be tested using the statistic. STATA produces the t-statistic for each explanatory variable. The critical value, at 5% significance, for the absolute value of the t-statistic is 1.96. This means that when the absolute value of t > 1.96, the individual explanatory variable is statistically significant.

The R-squared is measure designed to indicate the degree to which the dependent variable is explained by the regression model. The R-squared value is the percentage of the variance on the dependent variable that is explained by the model. The explanatory power of the set of firm characteristics increases as the R-squared increases. The data used for the multiple regression analyses are summarized in table 2.

Table 2: Summary of data

Variable Mean Std. Dev. Min Max

LEVDA 0.1940461 0.1226495 0 0.55189672 LEVDE 0.7487237 0.7635508 0 5.132365 TAN 0.5230921 0.3406119 0 1.314896 NDTS 0.433855 0.0205586 0.0063762 0.1128987 GROCAPEX 0.494515 0.043252 0 0.3023481 GROPE 4.830569 14.94291 -116.1338 71.73108 SIZEEMP 3.065368 1.377414 -1.629641 5.547709 SIZESALE 8.568894 1.447612 5.259727 13.06085 PROF 1.141509 0.5428891 0.2616497 2.533875 LIQ 0.1067229 0.0908661 0.0060729 2.522875 TAX 0.25 0.25 0.25 0.25 19

(20)

5. Results

5.1 Correlations

The correlation table describes the correlation of all the variables used to research the statistical relevance of the different firm characteristics. The correlation describes the extent to which the movements of variables can be linked. When the value of the correlation coefficient is positive, the movement of the researched variables is in the same direction whereas the movement of the researched variables is in the opposite direction when the sign of the correlation coefficient is negative. The absolute value of the coefficient of correlation is between 0 and 1. A correlation coefficient close to 1 indicates a strong correlation between the two variables and a correlation coefficient close to 0 indicates a weak correlation. The correlation between the different variables is summarized in table 3.

It is remarkable that the correlation between LEVDA and most explanatory variables is stronger than their correlation with LEVDE. This implies the debt-to-asset ratio is more correlated with the firm characteristics than the debt-to-equity ratio is. The correlation between LEVDE and LEVDA is 0.8273. As LEVDE and LEVDA are both measures for the leverage ratio, the strong correlation between these variables is not surprising. The problem of multicollinearity can occur when the explanatory variables are extremely correlated. As this is not the case, this regression model will not suffer from the multicollinearity problem.

Table 3: Summary of correlations

Variable LEVDA LEVDE TAN NDTS GROCAPEX GROPE

LEVDA 1.000 LEVDE 0.8273 1.000 TAN 0.2345 0.0476 1.000 NDTS 0.4603 0.3664 0.6130 1.000 GROCAPEX 0.2422 0.0691 0.6911 0.4584 1.000 GROPE -0.0753 -0.0893 -0.0070 -0.1012 -0.0397 1.000 SIZEEMP -0.0349 0.0797 -0.1192 -0.0190 -0.2816 -0.1189 SIZESALE -0.0765 0.0473 0.0352 0.0284 -0.1298 -0.1163 PROF -0.3357 -0.1764 -0.2712 -0.2537 -0.2808 -0.1170 LIQ -0.2794 -0.1923 -0.2211 -0.2353 -0.207 -0.2070

Variable SIZEEMP SIZESALE PROF LIQ SIZEEMP 1.000

SIZESALE 0.8162 1.000

PROF 0.2459 0.1278 1.000

LIQ -0.2094 -0.080 -0.0060 1.000

(21)

5.2 Regression analyses

The regression analyses provide insights in the statistical significance of the explanatory variables when explaining the movement and variance of the dependent variable. Two

different regression analyses were performed using the same set of explanatory variables. The dependent variable was different in both regressions. Table 4 and table 5 summarize the regression output from STATA with LEVDA as dependent variable. Table 6 and table 7 summarize the STATA output with LEVDE as dependent variable.

5.2.1 Dependent variable: Total long-term debt / Total assets

LEVDA = β0 + β1TAN + β2NDTS + β3GROCAPEX + β4GROPE + β5SIZEEMP + β6SIZESALE +

β7PROF + β8LIQ

LEVDA is defined as the value of total long-term debt divided by the total value of assets. This is a measure that represents the debt ratio and explains the degree to which firms finance their activities with debt financing. Table 4 shows the F-value and R-squared value that result from the regression analysis with LEVDA as dependent variable. As described in the

methodology the critical value of the F-statistic is 1.98. The F-statistic, as displayed in table 4, is 16.05. The F-statistic exceeds the critical value and thus the null-hypothesis, that assumes that the model is insignificant, can be rejected.

The R-squared value is the percentage of the variance of the debt-to-asset ratio that is explained by the model. Thus, 32.87% of the variance in LEVDA is explained by the set of firm characteristics.

Table 4:

Table 5 shows the regression coefficients, standard errors of regression and the t-values that result from the multiple regression analysis. Variables are considered to be of statistical significance when the absolute value of the t-statistic is larger than 1.96. The underlined variables, NDTS, PROF and LIQ are statistically significant firm characteristics. The NDTS has a positive coefficient of regression which means that NDTS and LEVDA

Dependent variable LEVDA

F-value 16.05

R-squared 0.3287

(22)

move in the same direction. When the NDTS increases, so does the leverage ratio. As the non-debt tax shield decreases the possible tax benefits from non-debt financing, the hypothesized relation between NDTS and LEVDA was negative. This hypothesis does not hold in the sample used for the regression analysis. The regression analysis even shows a significant positive relationship.

The strong negative relation between PROF and LEVDA means that as profitability increases, firms use less debt financing to finance their business activities. This relation is in line the hypothesis that was derived from the existing literature on capital structure decisions. The last statistically significant variable is LIQ. As liquidity increases, firms tend to have a lower leverage ratio. The assumption that firms with more cash and cash equivalent on their balance sheet have lower leverage ratios is thereby confirmed.

Table 5:

LEVDA Coefficient Standard error,

Robust t-value TAN -0.0523591 0.0345578 -1.52 NDTS 2.248174 0.4540524 5.17 GROCAPEX 0.1260046 0.2627404 0.48 GROPE -0.000612 0.0006359 -0.96 SIZEEMP 0.0105827 0.0092826 1.14 SIZESALE -0.0142099 0.008456 -1.68 PROF -0.599722 0.0137741 -4.35 LIQ -0.2770792 0.0742252 -3.73 CONSTANT 0.3128592 0.0612767 5.11

5.2.2 Dependent variable: Total long-term debt / Total common equity

LEVDE = β0 + β1TAN + β2NDTS + β3GROCAPEX + β4GROPE + β5SIZEEMP + β6SIZESALE +

β7PROF + β8LIQ

LEVDE is defined as the total value of long-term debt divided by the total value of assets. This ratio is commonly known as the debt-to-equity ratio and explains the relative proportion of debt and equity that a firm has on its balance sheet. As prevailed from the table 3, the correlation between the firm characteristics and LEVDE is not as strong as their correlation with LEVDA. This implies that the movements in LEVDA are better explained by the model than the movements in LEVDE. This assumption is confirmed by the R-squared ratio of 0.2264. The R-squared ratio is relatively low when compared to the R-squared value of 0.3287 produced when LEVDA is the dependent variable.

(23)

The value of the F-statistic for this regression, as shown in table 6, is 7.54. As the critical value for the F-statistic in 1.98, the null-hypothesis, which assumes that the model is not of statistical significance, can be rejected. The significance of the individual explanatory variables is described by the t-statistic.

Table 6:

The correlation between LEVDE and LEVDA is 0.8273 and therefore it is expected that the firm characteristics have approximately the same influence on LEVDE as they have on LEVDA. The significant variables in this regression are the variables of which the absolute value of the t-statistic exceeds the rejection region of 1.96. Thus, the underlined variables; TAN, NDTS, PROF and LIQ are statistically significant. Except for the variable TAN, these variables correspond to the regression with debt-to-asset ratio a dependent variable.

In contrast with the related literature, the coefficient NDTS suggests a positive relation between the non-debt tax shield and leverage. TAN, PROF and LIQ are negatively related to the debt-to-equity ratio. The negative relation between TAN and LEVDE suggests that firms with more tangible assets tend to have a lower debt-to-equity ratio. The relationships

described by PROF and LIQ are similar for both regression analyses.

Table 7:

LEVDE Coefficient Standard error,

Robust t-value TAN -0.7347275 0.2081383 -3.53 NDTS 17.89283 4.157363 3.96 GROCAPEX 0.1982113 1.59223 0.12 GROPE -0.0030796 0.003814 -0.81 SIZEEMP 0.0285209 0.0688582 0.41 SIZESALE 0.0048758 0.623058 0.08 PROF -0.2263941 0.1012102 -2.24 LIQ -1.28405 0.3850919 -3.33 CONTANT 0.662924 0.4820684 1.38

Dependent variable LEVDE

F-value 7.54

R-squared 0.2264

(24)

5.2.3 Robustness check

Throughout both regression analyses the option for robust standard errors was employed. This option corrects for possible heteroskedasticity, inconsistency of the variance in the error terms. The regression analyses were performed using proxy variables as measures for firm characteristics. The proxies used prevailed naturally from the related literature. However, some firm characteristics can be defined in several ways. To test the fragility of the regression model, minor changes in the proxies for the variables were made and the new regression models were tested. Some different proxies used to test the robustness of the model were:

𝑃𝑃𝐺𝐺𝐺𝐺𝐹𝐹 = 𝑃𝑃𝐿𝐿𝑟𝑟𝑝𝑝𝑟𝑟𝑟𝑟𝐿𝐿𝑑𝑑𝑟𝑟𝑇𝑇𝑟𝑟𝑟𝑟𝑒𝑒 = 𝐿𝐿𝐸𝐸𝑆𝑆𝑇𝑇𝐿𝐿𝐿𝐿 𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎

𝑇𝑇𝐿𝐿𝑇𝑇 = 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟 𝑟𝑟𝐿𝐿𝑙𝑙𝐿𝐿𝑟𝑟𝑑𝑑𝑟𝑟𝑇𝑇𝑟𝑟𝑟𝑟𝑒𝑒 = (𝑃𝑃𝐿𝐿𝑟𝑟𝑃𝑃𝐿𝐿𝐿𝐿𝑟𝑟𝑒𝑒, 𝑃𝑃𝑇𝑇𝐿𝐿𝑙𝑙𝑟𝑟 & 𝐿𝐿𝑒𝑒𝑒𝑒𝑟𝑟𝑃𝑃𝑡𝑡𝐿𝐿𝑙𝑙𝑟𝑟 + 𝑟𝑟𝑙𝑙𝐿𝐿𝐿𝐿𝑙𝑙𝑟𝑟𝑟𝑟𝐿𝐿𝑒𝑒)_{𝑇𝑇𝑟𝑟𝑟𝑟𝐿𝐿𝑇𝑇 𝐿𝐿𝑎𝑎𝑎𝑎𝐿𝐿𝑟𝑟𝑎𝑎}

Using different proxies for the firm characteristics caused only minor changes in the regression results. The statistically significant variables that prevailed from the regression analysis were the same as in the original regression models.

The statistical analysis program STATA provides the option to perform a robust regression. This robust regression makes the regression model less sensitive to outliers in the data. The robustness option assigns weights to more consistent and better behaved variables. This regression analysis is also called the iteratively reweighted least squares. Comparing the regression result from the standard and the robust regression analyses provides an indication for the robustness and trustworthiness of the model. Table 8 and table 9 compare the t-statistics for both leverage ratios between the standard and the robust regression.

Table 8 compares the robust and standard regression analyses performed on the dependent variable LEVDA. The t-statistic show only minor differences, which indicates that the models correspond well. In the robust regression model, SIZESALE has become a

significant variable. Table 9 compares the robust and standard regression analyses performed on the dependent variable LEVDE. The deviations in the t-statistic are considerably larger compared to the regression on LEVDA. This indicates that the model is more fragile and less robust. This confirms the earlier discovery, that the explanatory power of the model is

stronger for LEVDA than LEVDE, based on the R-squared value. Due to the large deviations of the t-statistics, GROPE and SALE became significant variables.

(25)

Table 8: comparison standard/robust dependent variable: LEVDA Dependent variable: LEVDA

Variable Standard t-statistic Robust t-statistic Difference TAN -1.52 -1.08 0.44 NDTS 5.17 5.51 0.34 GROCAPEX 0.48 0.30 0.18 GROPE -0.96 -1.49 0.53 SIZEEMP 1.14 0.95 0.19 SIZESALE -1.68 -2.08 0.4 PROF -4.35 -4.57 0.22 LIQ -3.73 -2.99 0.74 F-value 16.05 13.72 2.33

Table 9: comparison standard/robust dependent variable: LEVDE Dependent variable: LEVDE

Variable Standard t-statistic Robust t-statistic Difference TAN -3.53 -1.73 1.8 NDTS 3.96 2.62 1.34 GROCAPEX 0.12 0.74 0.62 GROPE -0.81 -3.17 2.36 SIZEEMP 0.41 1.55 1.14 SIZESALE 0.08 -2.07 2.15 PROF -2.24 -2.96 0.72 LIQ -3.33 -3.01 0.32 F-value 7.54 6.34 1.2 25

(26)

The theoretical hypotheses of the relation between the significant firm characteristics and leverage ratio are summarized in table 10. The relation between the nod-debt tax shield and both leverage ratios is positive, while the hypothesized relation was negative. The

hypothesized relation between profitability and liquidity was negative. This relation is confirmed by the regression result. Tangibility is only significant for the debt-to-equity ratio and describes a negative relation, even though the hypothesized relation between the two variables was positive.

Table 10: Summary of hypotheses significant firm characteristics

Firm characteristic Hypothesized relation to leverage ratio

Relation to LEVDA Relation to LEVDE

Non-debt tax shield - + +

Profitability - - -

Liquidity - - -

Tangibility + -

(27)

6. Discussion & conclusion

Theories on capital structure decisions indicate firm characteristics as possible determinants for capital structure decisions. The theoretical effects of these firm characteristics on leverage ratio however, do not always correspond to the empirical results. Earlier research on capital structure determinants in the Netherlands provided evidence for the existence of a clear financial pecking order (De Haan & Hinloopen, 2003). Firm characteristics that are indicators for the financial pecking order for funding business activities are liquidity and profitability. Dutch firms have a strong preference for internal financing sources (De Haan & Hinloopen, 1999). However, the extent to which internal finance is used depends heavily on its

availability. Profitable firms have the opportunity to retain earnings and build up internal capital. Consequently, profitable firms will have a tendency to have lower leverage ratios.

Firms with highly liquid assets can liquidate their assets whenever they want to invest and thus have less need for external finance. Hence, highly liquid assets are associated with a low leverage ratio (Niu, 2008). According to the pecking order theory, liquidity and

profitability are thus negatively related to leverage ratio. In both regression analyses, these firm characteristics showed significant negative relation to leverage ratio. The empirical research thus confirms the existence of the financial pecking order theory.

The trade-off theory is based on the trade-off between tax benefits and financial distress costs of a high leverage ratio (Myers, 2001). The suggested relation between the non-debt tax shield and leverage ratio is negative because the no-non-debt tax shield reduces the possible tax advantage of debt. The regression analyses cannot confirm this negative relation. On the contrary, a positive relation is found between non-debt tax shield and leverage ratio. This is in contradiction with the existing literature and thus provides evidence that the trade-off theory is not decisive for capital structure decisions in the Netherlands.

The variables associated with the principal-agent theory, size and growth opportunity, were of insignificant influence on leverage ratio. This implies the absence of the principal-agent theory in the Netherlands. This might be attributed to the extended controlling and supervising role of the supervisory board in the Netherlands (Corhay & Rad, 2000). The absence of the principal-agent theory is consistent with the research of Chen, Lensink and Sterken (1998).

(28)

The regression model that was used to examine the significance of different firm

characteristics could be subject to proxy selection bias. As some of the firm characteristics can be specified using different proxy variables, the outcome of the regression is partly dependent on the proxy selection. As described in the results, multiple regressions were performed using slightly different proxy variables to test the dependence of the regression on proxy selection. The regression analyses indicated the same firm characteristics to be

significant. Another problem with proxy selection is that the proxy might not perfectly represent the firm characteristic. Therefore a measurement error can occur in the variable and consequently causes a small bias.

Furthermore, robustness checks provided insight in the frangibility of the regression model. The robust regressions showed that the model, estimating LEVDA as dependent variable, was more robust. The robust t-statistics hardly deviate from the earlier regression results. The robust regression with LEVDE as dependent variable caused more problems. Some t-statistics deviated significantly. The smaller degree of explanatory power of the regression model for the dependent variable LEVDE was already predicted by the lower R-squared value. The strong deviations of the t-statistics make that the model for LEVDE is less reliable than the model for LEVDA.

Future research could minimize the problems that occur in the regression analysis by further investigating to correspondence on the proxy variables used to approximate the firm characteristics. Increasing the number of firms used in the sample could lead to more reliable regression results as well.

(29)

7. Bibliography

Berger, N. A., & Bonaccorsi di Patti, E. (2006). Capital Structure and firm performance: A new approach to testing agency theory and an application to the banking industry. Journal of Banking & Finance , 1065-1102.

Berk, J., & DeMarzo, P. (2011). Corporate Finance. Pearson.

Bordo, M. D. (2008). An historical perspective on the crisis of 2007-2008. Camebridge: National Bureau of Economic Research.

Bradley, M., Jarrel, G., & Kim, A. (1983). On the existence of an optimal capital structure. The Journal of Finance , 857-878.

Chen, L. H., & Jiang, G. J. (2001). The Determinants of Dutch Capital Structure Choice. SOM-theme E .

Chen, L. H., Lensink, R., & Sterken, E. (1998). The Determinants of Capital Structure: Evidence from Dutch panel Data.

Corhay, A., & Rad, A. T. (2000). International acquisitions and shareholders wealth evidence from the Netherlands. International Review of Financial Analysis , 163-174.

De Haan, L., & Hinloopen, J. (1999). Debt or Equity? An empirical study of securities issues by Dutch companies.

De Haan, L., & Hinloopen, J. (2003). Preference hierarchies for the internal finance, bank loans, bond and share issues: evidence for Dutch firms. Journal of Empirical Finance , 661-681.

De Haan, L., Koedijk, K. G., & Vrijer, E. J. (1994). Buffer stock money and pecking order financing: Results from an interview study among Dutch firms. De Economist , 287-305.

Fama, E. F. (1980). Agency Problems and the Theory of the Firm. Chicago Journals , 228-307. Gertner, R., & Scharfstein, D. (1991). A theory of the Workouts and the Effects of Reorganization Law. Journal of Finance , 1189-1222.

Jensen, M. C. (1986). Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers. The American Economic Review , 323-329.

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behaviour, agency costs and ownership structure. Journal of Financial Economics , 305-360.

Kalay, A. (1982). Stockholder-bondholder conflict and dividend constraints. Journal of Financial Economics , 211-233.

Kochhar, R. (1996). Explaining Firm Capital Structure: The Role of Agency Theory vs. Transaction Cost Economics. Strategic Management Journal , 713-728.

Kraus, A., & Litzenberger, R. H. (1973). A State-Preference Model of Optimal Financial leverage. The Journal of Finance , 911-922.

(30)

McNichols, M., & Wilson, P. G. (1988). Evincende of Earning Management for the Provision for Bad Debts. Journal of Accounting Research , 1-31.

Modigliani, F., & Miller, M. (1958). The Cost of Capital, Corporation Finance and the Theory of Investment. The American Economic Review , 261-297.

Myers, S. C. (2001). Capital Structure. Journal of Economic Perspectives , 82-102. Myers, S. C. (1984). The capital Structure Puzzle. The Journal of Finance , 574-592. Niu, X. (2008). Theoretical and Practical Review of Capital Structure and its Determinants. International Journal of Business and Management .

Parrino, R., & Weisbach, M. S. (1999). Measuring investment distortions arising from stockholder-bondholder conflicts. Journal of Financial Economics , 3-42.

Rajan, R. G., & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. The Journal of Finance , 1421-1460.

Shyam-Sunder, L., & Myers, S. C. (1999). Testing static tradeoff against pecking order models of capital structure. Journal of Financial Economics , 219-244.

Stonehill, A., Beekhuisen, T., Wright, R., Remmers, L., Toy, N., Pares, A., et al. (1975). Financial goals and debt ratio determinants: A survey of practice in five countries. Financial Management , 27-41.

Titman, S., & Wessels, R. (1988). The determinants of capital structure choce. The Journal of Finance , 1-19.

van Dijk, M., & Weyzig, F. (2007). Tax haven and development partner. SOMO.

Determinants of capital structure : an empirical analysis of Dutch listed firms