Dutch SMEs in the debt paradox

Dutch SMEs

in the

debt paradox

Bachelor Finance & Bachelor Tax Economics

University of Amsterdam

Faculty Economics & Business Dr. J.E. Ligterink

Amsterdam Law School Dr. Mr. G.J.W.M Kampschöer

Abstract

This paper analyzes the relationship of the marginal corporate income tax rate on capital structure policy of the Dutch private small- and medium enterprises (SMEs); focusing on the

firm’s debt-equity choice. The current theories try to explain the level of debt in the firm’s capital structure, resulting in a variety of leverage cost and benefit related determinants. Prior

research has shown these determinants fail to explain the debt levels completely. Hence, many issues remain unresolved, whereas the theories do not satisfy reality: the paradox of debt. Since the majority of preceding research focused on large quoted companies, the data sample used in this study consist of 2616 firm-year observations for 327 SMEs, covering the

period 2003-2010. A panel least square regression has been performed. The results do not support the theory, contradicting most of the hypotheses and prior research on capital

structure.

Frank Boerma Amsterdam, 24-01-2014

Contents 1 Introduction

2 Theoretical Framework 2.1 Trade-off theory

2.1.1 Debt tax shield benefits 2.1.2 Financial distress costs 2.1.3 Target ratio adjustment costs 2.2 Pecking order theory

2.2.1 Information asymmetries 2.2.2 Agency conflicts

2.2.3 Risk shifting, asset substitution & debt overhang 2.2.4 Strategic considerations

2.3 SMEs

2.3.1 SMEs in the Netherlands 2.3.2 Dutch corporate tax

2.3.3 Large public firms vs. private SMEs 2.3.4 SMEs’ capital structure

3 Hypotheses 3.1 Tax variables 3.2 Controlling variables 4 Methodology 4.1 Sample selection 4.2 Dependent variables 4.3 Modelling the leverage

4.4 Testing the model: PLS regression 5 Results 5.1 Tax effects 5.2 Controlling effects 6 Conclusion References Appendix

1 Introduction

For decades, globalization has increasingly led to an interdependence and unionization of economies worldwide. In addition, capital and investment transfers as well as migration and movement of labor between countries, require little effort. Companies relocate more and more easily. Globally, public authorities attempt to create therefore a favorable fiscal climate, in order to attract firms to establish within their region, meanwhile responding to tax-related changes in other countries (Devereux, 2008). These conditions have resulted in transitions in tax regulations in the Netherlands as well.

Firstly, Dutch corporate tax has been reduced by approximately ten percent in the last decade, to a competitive corporate statutory tax rate of 25.5 percent. Furthermore, contrary to several other European Union countries, tax policy in the Netherlands permits firms tax loss carry-forwards as well as tax loss carry-backs. Moreover, due to the recent financial and economic crisis worldwide, tax regulations have temporarily been liberalized. Since 2007 in particular, these liberalizations concern the treatment of corporate tax losses and a temporarily corporate income tax rate reduction to about 25 percent (Staatscourant 3804, 2012).

Corporations can finance new projects in several ways and this capital structure decision making process has been studied for many years. In general, firms can make use of either in-house cash, raise their debt or consider an issuance of new shares (Hovakimian & Opler, 2001). The net present value of new projects depends on the way they are financed, due to a variety of market imperfections. Therefore, total firm value depends on its capital structure. One of these market imperfections involves the tax policy, including current corporate income tax rate and the tax regulations with regards to the treatment of corporate income losses.

The main advantage of raising debt involves the interest payments’ deductibility of the operating income. This part of income, being untaxed, increases the net present value of future projects, raising the value of the whole firm (Myers, 1984). It is widely accepted that firms pursue financial policies that provide these tax benefits. The higher the marginal corporate income tax rate, the higher the potential tax benefits. Graham (2000) has shown that these so-called tax shields are not fully exploited. Most of the current literature however, discussing the firms’ capital structure, examines a sample of large publicly traded firms, usually established in the United States (Donaldson, 1961; Fama & French, 2002; Flannery & Rangan, 2006; Graham, 2002; Harris & Raviv, 1991; Titman & Wessels, 1998).

Medium-Sized Enterprises (SMEs). The dataset sample examined covers the years 2003-2010. As mentioned above, this specific research period encloses several interesting externalities. The main objective of this paper is to examine the effects of the marginal corporate income tax rate on the SME´s debt-equity choice, the leverage of the firm. Both the trade-off theory and the pecking order theory, describing the financial behavior of SMEs with some accuracy, will be of major importance in establishing the hypotheses and the controlling variable behavior. The theories are jointly tested in order to explain the effect of the marginal tax rate, taking into account the SME’s characteristics.

The aim of this paper is to contribute, through the selection of a distinctive dataset sample, to the existing empirical literature on the SMEs’ capital structure decision making. Where prior research often examines cross-country datasamples, this study focusses on a single country. The alter tax regulations however, whether or not temporarily, hopefully contribute to the understanding of capital structure behavior of the SMEs.

The remainder of the paper is structured as follows. Section two starts with a review of the literature, discussing the widely accepted trade-off theory and the pecking order theory in addition to the SMEs’ capital structure issues. Section three addresses the hypotheses, including a presentation of the explanatory and controlling variables. Section four illustrates the methodology, addressing the sample selection, the model specification and testing. Section five will reflect the results. Finally, section six presents the conclusion and discussion of this research paper.

2 Theoretical Framework

2.1 Trade-off theory

The founders of the theory of capital structure, Modigliani and Miller (1958), showed that in an efficient working market, debt policy is irrelevant. The firm’s market value is independent of its capital structure. Later they adjusted their model by adding corporate taxes as a variable (1963). Considering the tax benefits of debt financing, they demonstrated a firm’s optimal capital structure with 99.99 percent debt financing. The costs of financial distress had not been considered here.

Kraus and Litzenberger (1973, p. 918) initiated the theory of the trade-off. They considered: ‘The market value of a levered firm is shown to equal the unlevered market value, plus the corporate tax rate times the market value of the firm’s debt, less the complement of the corporate tax times the present value of bankruptcy costs.’ Currently, this theory is referred to as the static trade-off theory, while leverage has been determined by only two key factors: the tax benefits of debt and the bankruptcy costs of debt. Although this does not represent a realistic business model, it is however an interesting starting point in this debate on capital structure.

According to Myers (1984), a firm that follows the trade-off theory sets a target debt-to-value ratio and then gradually moves towards the target, referred to as the dynamic trade-off theory. This target is determined by considering debt tax shields benefits against costs of bankruptcy; firms are trading off the marginal benefits against the marginal costs. Although this leads to an improvement of the latter model, Flannery & Rangan (2006) stated that: ‘The typical firm closes about one-third of the gap between its actual and its target debt ratios each year’.

Many other market imperfections influence the firm’s trade-off in capital structure decision making. Besides the benefits of taxes and the costs of financial distress, costs related to transactions, agency conflicts, issues of information asymmetry, as well as strategic considerations have to be taken into account (Graham, 2003). The latter three will be discussed in the next section considering the pecking order theory.

In summary; in the majority of current literature is consensus about two important hypotheses. Firstly, companies seek to obtain an optimum capital structure. Secondly, firms balance the advantages and disadvantages of an additional monetary unit of debt (Lopez-Gracia & Sogorb-Mira, 2008). The term trade-off theory is used by a variety of authors, describing a family of related theories. The theory will be discussed by its leading individual

elements. The sections below consider debt tax shield benefits, financial distress costs and target adjustment costs respectively.

2.1.1 Debt tax shield benefits

Interest payments on debt are deductible from the operating income. These expenses reduce the taxable income and can be seen as the main advantage of debt financing. These benefits have been introduced as the debt tax shield, since the interest expenses metaphorically protect the taxable income from being taxed (Myers, 1984). Due to a lack of data, several non-debt tax shields, such as depreciation expenses, will not be considered in this paper. The examined dependent variable equals the income after expenses, and before interest and taxes (ebit). However, this could influence the results, discussed later.

The level of the tax shield benefits depends on several country-specific factors. The three most important elements are the tax rates, including the corporate income tax rate and the rate on capital gains, the scope of tax loss carry-backs and -forwards and the expected cash flows in prospective years. The tax shield benefits are considered to be an upward slope linear function of these three elements (Bartholdy & Mateus, 2011).

Frank & Goyal (2007) put the relationship between corporate tax and leverage in a proper perspective: ‘Corporate income taxes are only about a century old. Debt financing was common long before the introduction of the corporate income taxes. Thus, we know that taxes do not provide a complete justification for the use of debt financing.’ This obviously does not imply that taxes can be safely ignored when analyzing modern corporate use of debt.

Modigliani & Miller (1963) demonstrated that, in the absence of the costs of debt financing, firms should finance their projects entirely through debt. They claimed that, in this theoretical situation, the interest payments totally offset the taxable income; the tax shield has been maximized. However, costs of default, transaction costs and costs related to information asymmetries and strategic considerations disrupt this assumption. The latter two debt-related disadvantages will be discussed in the next section, considering the pecking-order theory. The first three elements will be discussed below.

2.1.2 Financial distress costs

As mentioned above, debt capacity is limited because corporations trade-off the tax savings generated by the deductibility of interest payments against the expected value of the costs incurred in the event of bankruptcy. Financial distress costs are costs associated with the company’s failure of meeting its obigations. These costs include, among others, the inability

to negotiate long-term supply contracts, due to the lack of trust. Moreover, a firm could face the loss of customers, not having confidence in warrenties, repairs or returns. It may need to inefficiently liquidate productive assets in order to generate cash. Briefly, the probability of financial distress is an indirect cost, consisting of interest conflicts between debt- and other stakeholders and costly disruptions in the relationship of the firm with customers, suppliers and employees. Creditors will require compensation for this increase in risk and ask for higher interest rates (Miller, 1977).

The quantity of the bankruptcy costs should raises as a firm moves towards its optimal target debt ratio (Titman & Wessels, 1988). Although a firm situated below its optimal target could benefit from an increase in leverage, it is not as critical as a firm that is situated above its target (Hovakimian & Opler 2001). The costs involved in adjustment to this target debt ratio will be discussed next.

2.1.3 Target ratio adjustment costs

Myers (1984) points out that the trade-off approach implies the rate of real leverage converging to a target or optimal level. When firms are perturbed from this optimum, this view argues that companies respond by rebalancing their leverage back to the optimal level (Leary & Roberts, 2005). Fama and French (2002) note that the firms’ debt ratios adjust slowly toward their targets. The distance or gap between both real and target ratio levels supposedly depends on the firms’ target ratio adjustment costs. For the adjustment to be gradual rather than abrupt, the marginal costs of adjusting should increase when the adjustment is larger (Frank & Goyal, 2007).

Target ratio adjustment costs consist, among other elements, of registration and underwriting fees, as well as legal and accounting fees. SMEs face relatively a high level of these costs, as they face larger financial restrictions in capital markets and more agency problems, this will be examined further in the next section, discussing the pecking order theory.

2.2 Pecking order theory

The pecking order theory is often suggested as the competitive counterpart of the trade-off theory. It demonstrates a hierarchy in funding sources in considering a new project’s financing. The theory has long roots in the descriptive literature. It was initially suggested by Donaldson (1961) and later modified by Myers (1984). Donaldson (1961) suggested:

‘Management strongly favored the internal generation as a source of new funds to the external funds, except for occasional unavoidable bulges in the need for funds.’ Furthermore, it does not consider the firm to have a well-defined target debt ratio, in contrary to the dynamic trade-off theory. However, it is a valuable contribution in the explanation of the short- and long-term debt ratios, since a synergy of the elements of both theories has much more explanatory power.

2.2.1 Information asymmetries

In addition to the work of Donaldson (1961), Myers (1984) provided a theoretical underpinning, arguing for a preference of retained earnings over debt, and favoring debt in turn over equity, due to adverse selection. The issue of adverse selection derives from a broader concept, the issue of the stakeholders’ asymmetry of available information.

Asymmetry of information arises at multiple levels. Internally, information could be out of balance between the principal and the agent. Externally, the same problem is experienced between the management and shareholders or creditors, as well as between shareholders and creditors. The company’s in- and outsiders usually do not have the same available information. The manager’s information about the firms and their future cash flows is likely to be superior to that of outside investors, thus resulting in the problem of adverse selection.

Adverse selection, an important element of the pecking order theory, has been developed by Myers and Majluf (1984) and Myers (1984). In summary; it refers to the issue of a negative outcome of demand and supply due to an inequality of available information for both parties. The key idea is that the owner-manager of the firm knows the true value of the firm’s assets and growth opportunities, whereas the outside investor can only guess these values (Frank & Goyal, 2009). Moreover, a debt issue can reveal information to uninformed investors as well, whereas a lot of debt is often associated with the risk of default.

In case of an equity issue, managers actually are signaling an overpriced value of the stock, since the firm wouldn’t offer stock at an undervalued price. Therefore, an external investor would require compensation for the increase in risk. The adverse selection problem induces a premium in the cost of external capital and one that is increasing as the information sensitivity increases. As a result, external equity can be too expensive and a firm will even give up projects with a positive net present value in order to search for other external funds.

This results in the following expected ranking in the preference of funding sources. Firstly, firms should prefer internal finance. Secondly, if external finance is required, firms

should prefer the safest security first, whereas the risk premium is the lowest. The following ranking, varying from the cheapest source of financing to the most expensive has been observed: internal funds (retained earnings), risk free debt, risky debt, preferred debt/convertible debt, common stock (Myers, 1984)

2.2.2 Agency conflicts

In addition Jensen (1986), demonstrated the consequences of the conflicts of interest within the company. He introduced the term agency costs. Agents are supposed to act in their own interest due to the fact that the agent is not the sole owner, reducing the firm’s total value. Two important agency problems generally arise: the horizon problem, and the conflict of moral hazard. Furthermore conflicts of interest between different external investors reflect issues like risk shifting or asset substitution. These issues will be discussed below. The strategic decisions firms often make in the use of debt will be discussed as well.

2.2.3 Risk shifting & debt overhang

Risk shifting or asset substitution, takes place when managers make disproportional risky investment decisions that maximize the shareholder’s or owner’s value at the expense of the debt holders’ interests (Jensen & Meckling, 1976). Managers have incentives to cause their firms to grow beyond the optimal target debt ratio, while this growth subsequently increases the managers’ power by increasing the resources under their control (Jensen, 1986). It is also associated with an increase in the managers’ compensation level, since changes in compensation are positively related to growth in sales (Murphy, 1985). This increase of the firm’s risk will result in a decrease of the amount of debt suppliers. In case of large dividend pay-outs or asset reductions, used to serve as collateral for creditors, these strategies even harm the debt holders.

Another conflict of interest between creditors and shareholders refers to the problem of debt overhang. Total firm value will increase when new projects with a positive net present value are undertaken. The problem is that former shareholders refrain from taking these strategies, foregoing the projects, and harming the benefits of the debt holders. >> The result is an underinvestment and a decrease of the firm’s profitability and future flexibility. Highly levered firms and dispersed debt holders will experience these debt overhang problems relative often. This issue could be solved when debt holders are willing to put up most of the finance of the project. Due to asymmetric information problems this equilibrium won’t be reached.

2.2.4 Strategic considerations

>> Conflicts of interest between shareholders and managers generally arise when the

organization generates substantial free cash flow, whereas cash flow excesses the required level to fund all projects. Jensen discussed t he problem how to motivate managers to disgorge the cash rather than investing it at a level below the cost of capital or wasting it on organization inefficiencies (Jensen, 1986).

An increase of the leverage could provide a disciplinary role. Managers in highly levered firms have to pay out more cash to their creditors. Therefore they have fewer funds left for wasteful activities, or less margin for error. Debt helps prevent such firms from wasting resources on low-return projects (Jensen, 1986). However, as mentioned above, too much debt can lead to the issue of asset substitution as well (Stulz, 1990; Harris & Raviv, 1991)

2.3 SMEs

This paper focuses on Small- and Medium-Sized Enterprises (SMEs), while SMEs represent a major proportion of the total amount of enterprises in developed nations, they have nonetheless been ignored in a major part of empirical research on capital structure (Sogorb-Mira, 2005). While SMEs, typically non listed companies, are generally expected to have larger variability in their capital structure in cross-country comparisons compared to their larger counterparts (Hall et al., 2004). Moreover, many SME studies use data of firms that would be classified as large by any definition of small business (van der Wijst, 1993).

2.3.1 SMEs in the Netherlands

Since the current sample covers the period of 2003-2010, this study has adopted the European Commission’s new definition of SME: ‘Companies that employ fewer than 250 persons and which have an annual turnover not exceeding 50 million euro, and/or an annual balance sheet total not exceeding 43 million euro’ (Article 2 of the Annex of Recommendation 2003/361/EC).

According to the European Commission (2002), in the enlarged European Union of 25 countries, 99.8 percent of the total number of companies was SME, providing 97 million jobs or 69.7 percent of total European employment. 570 thousand SMEs were established in the Netherlands by the year 2002 or 99.6 percent of all enterprises. Over 10 % of these enterprises

represent the manufacturing sector.

2.3.2 SME’s corporate taxation

Like many other European countries, the Netherlands try to execute competitive corporate income rates. As mentioned above, corporate tax regulations in the Netherlands have been altered in recent years (Stcrt. 2012, 3804). First, the Dutch government has reduced the corporate income tax rate heavily, since 2004 in particular. Table I illustrates this reduction in the tax rates since 2001. Although after-2010 financial statements are not considered in the sample period, these tax rate changes have been considered in the firms’ investment, financial and other corporate structure decisions in preceding years, influencing debt policy in the sample period (Stcrt. 2012, 3804). Second, the treatment of tax losses has been limited. The tax loss carry-back has been reduced to one year, whereas the tax loss carry-forwards have been reduced to a period of nine years. However, the possibility of carry-backs is still a competitive element of the Dutch tax policy (Stcrt. 2012, 3804). Third, due to the financial and economic crisis, tax rules have been liberalized temporarily for the years 2009-2011. During this period, firms can either choose a three-instead of a one-year tax loss carry-back. However, in those cases the tax loss carry-forwards are restricted from a nine- to a six year-period (Staatscourant 3804, 2012).

Bron: Jongbloed Fiscaal Juristen N.V.

2.3.3 Large public firms vs. private SMEs

It’s widely accepted that taxes do matter for the firm´s capital structure (Myers, 1984; Shevlin, 1990; Rajan & Zingales, 1995; Graham, 2003). However, as mentioned most of the research has been done on samples of large publicly traded U.S. firms. This study will focus

Year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Amount (€) > 22,689 29 29 27 25.5 - - - -22,689 > 34.5 34.5 31.5 29.6 - - - -> 25,000 - - - - 20 - - - -25,000 - 60,000 - - - - 23.5 - - - -60,000 > - - - - 25.5 - - - -> 200,000 - - - 20 20 20 20 20 200,000 > - - - 25.5 25.5 25 25 25 > 275,000 - - - 20 - - - - -275,000 > - - - 25.5 - - - - -TABLE I

on small private firms only. Public and private firms differ in several aspects, making this distinction highly relevant.

First, large publicly traded firms require far less effort to access the financial markets, while SMEs are usually financed by one bank. In contrast with public firms, private firms do not have a quotation on the stock market, and are unable to issue publicly traded shares or bonds, limiting their access to the financial markets. In turn they use non-traded debt such as bank loans and leasing- or trading credit. As a consequence, SMEs are usually less leveraged, as they tend to be financially restrained by creditors (Lopez-Gracia & Sogorb-Mira, 2008).

Second, small firms are often not as profitable as large firms, thus reducing the amount of income to be offset with deductible interest payments. At least some small firms face lower marginal tax rates than larger firms, implying smaller benefits from the tax shield (Michaelas et al., 1999). For this reason, Petit and Singer (1985) have argued that tax considerations are of little attention for SMEs. However, small firms have a limited amount of different debt sources, limiting other tax-shields in the way, and increasing the potential tax benefits of the deductible interest payments on the other side. Moreover, bank financing may be more flexible due to the closer relationship between the company and the financial institution, therefore increasing these tax benefits as well (Bartholdy and Mateus, 2011).

Third, the potential internal agency and asymmetric information problems faced by the firm increase with the firm’s size, especially in the case of publicly traded firms. This is because SMEs tend to have less of a separation between ownership and management. Moreover, smaller firms often have other approaches in solving these agency and asymmetric information problems (Bartholdy & Mateus, 2011). In addition, Cassar and Holmes (2003) argue that SMEs face higher levels of asymmetric information problems between in- and outsiders, especially those with a relatively short history due to a lack of historical performance data on which creditors can make investment decisions.

Finally, small firms have fewer options for financial diversification, resulting in a higher probability of financial distress; facing a reduction in the amount of debt they can carry. They face a limited amount of resources to be invested in a variety of assets (Cassar & Holmes, 2003). Moreover SMEs might experience bankruptcy costs as higher as they often are family owned, since these companies likely represent a greater amount of sentimental value to the owners, reducing the leverage of the optimal capital structure.

So far, the two leading capital structure theories have been discussed, as well as the SME situation within the Dutch corporate environment, in comparison to their large public counterparts. A major gap in the empirical research has been noted in the capital structure and their determinants of SMEs. Rajan and Zingales (1995) already stated that the emphasis on large companies has resulted in the lack of knowledge about the young and small firms, as well as Ang (1991), who pointed out that the financial theories were not developed with the small business in mind. As a result, the empirical literature has been inconsistent about SMEs’ capital structure. Watson & Wilson (2002) argue that the explanatory power of the static trade-off theory is typically expected to be low. Van der Wijst and Thurik (1993) concluded that the theoretical determinants appear to be relevant for the SMEs, but they state that the influences encountered are far less straightforward than the hypothesized effects in the theories. However, several authors illustrate that the implications of the capital structure theories can be applied to the small firm context as well (van der Wijst & Thurik, 1993; Hall et al. 2004; Ang, 1991 and Holmes, 1991) emphasize that the pecking order theory can be easily applied in SMEs.

Myers (1984) argues that SMEs won’t trade off the costs and benefits of debt, as the trade-off theory suggests, but that debt ratios will be determined by the firms’ cumulative requirements for external finance, as the pecking order theory suggests. Myers concludes that debt ratios vary across industries because asset risk, asset type and the need for external funds vary across industries as well. This implies that the firms’ industry does not directly determine its capital structure, but that it may be the natural outcome of the composition of the firm’s assets. Moreover, SMEs should not even have the financial diversification possibilities to have the ability for levering up to their optimal capital structure.

In spite of Myers argumentation, Harris and Raviv (1991) suggest that the industry in which the firm operates determines the debt ratio target. Ang (1991) supports this argument, considering the need of inter-firm comparisons as a cost-effective way of improving financial management, due to the lack of the full range of financial management skills within many SMEs.

The next sections will discuss the hypotheses constructed, considering the various elements of the theories discussed above.

3 Hypotheses

The trade-off theory has been discussed, illustrating the costs and benefits to trade off in order to target a firm’s optimal debt ratio, considering the benefits of the debt tax shield, the costs of financial distress and the adjustment costs in achieving this target. In addition, the pecking order theory examined the problem of information asymmetries, considering adverse selection, agency conflicts, risk shifting and considerations of strategy.

An extensive amount and variety of debt ratio determinants emerge from prior theoretical and empirical research (Modigliani, 1961; Donaldson, 1961; Kraus and Litzenberger, 1973; Myers, 1977; Graham, 1996; Graham, 2000). Based on these studies and the review on capital structure policy discussed in the previous sections, the hypotheses will be formulated. The variables discussed below, are expected to influence the capital structure of the SME, the debt-equity ratio. First, in order to illustrate a clear complete review, the three explanatory tax variables will be discussed and their resulting hypotheses, followed by the thirteen controlling variables.

3.1 Hypotheses

As mentioned earlier, interest payments are deductible from the taxable corporate income. This suggests a positive relationship between the amount of corporate tax to be paid and the amount of debt. The three hypotheses concerning the tax-related variables are discussed below.

H1 The marginal tax rate is positively correlated with leverage

The ability to carry losses and credits forward and backward makes it unlikely that examining current-period financial statements will provide an accurate assessment of a company’s marginal tax rate (Graham, 1996a, 2002). In an attempt to deal with this issue, prior research provided several solutions. One could add a dummy variable equal to one if a firm has a net operating loss carry-back or –forward and equal to zero otherwise like Scholes et al. (1990).

Graham (1996a, b), following Shevlin (1990), has developed a method to estimate the firm’s specific marginal tax rate. This more accurate method has been used in this paper. Marginal tax rate is defined as the present value of current and expected future taxes paid on an additional dollar of income earned today. It is desirable to estimate the marginal tax rate by supplementing historical data with a forecasted stream of future taxable income and then calculate the tax bill over the entire horizon in a manner consistent with the current tax rules.

This method of estimating involves three sets of inputs: the current tax rules, in particular the treatment of operating losses, the statutory tax rate and an estimation of the expected future earnings (Graham, 1996a). An extended calculation of this variable is found in the Appendix. Figure I illustrates the average marginal tax rates by year since 2006. A higher marginal tax rate will result, in case of an increased debt level, in a decrease in corporate taxable earnings. Therefore, the first hypothesis states that the marginal tax rate is positively correlated with the firm’s leverage.

H2 Kink is negatively correlated with leverage

Graham (2000) initiated the use of this variable in his study. The proxy should quantify to what extent the firm exploits it tax shield conservatively or aggressively. He defines this variable by observing the ‘kink’ in its benefit function, that is, the point where marginal benefits begin to decline and therefore the function begins to slope downward. The equation of kink is as simple as the ratio of EBIT and the actual interests paid:

𝐾𝑖𝑛𝑘 = 𝐸𝐵𝐼𝑇

𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡

The values of Kink should be interpreted as follows:

𝐾𝑖𝑛𝑘 = 1

If kink equals the value of one, the amount of interest expenses paid equals the earnings before tax. In this situation, the amount of interest offsets the amount of EBIT; the tax shield

is fully exploited. Debt is minimizing taxes here, resulting in a marginal tax rate equaling the statutory tax rate.

𝐾𝑖𝑛𝑘 < 1

A value of kink less than one, a firm operates on the downward sloping part of its tax rate function. In this situation, the interest payments are larger than the earnings before tax, resulting in a marginal tax rate below the statutory tax rate. This implicates reduced tax benefits on the last portion of its interest deductions, representing an aggressive debt policy.

𝐾𝑖𝑛𝑘 > 1

If kink is greater than one, the amount of earnings is positive after interest expenses. The firm could increase its interest expenses, increasing more benefits of these incremental deductions. With the potential tax shield not fully exploited, this situation implies a conservative debt policy.

In summary, an increase of the value of kink is expected to be positively correlated with debt conservatism, whereas a decrease in kink is expected to be negative correlated with debt conservatism. The second tax-related hypothesis therefore states that kink and leverage should have a negative relationship.

H3 Standardized Kink is positively correlated with leverage

As discussed above, Graham (2000) claims that firms with large values of kink use debt conservatively. In addition however, he notes that firms with large values of kink should remain on the flat part of their benefit functions even if they receive a negative shock to earnings. As referred to by Bartholdy & Mateus (2011): ‘If two firms have the same value for the kink variable, but one has more volatile earnings than the other, the firm with more volatile earnings has a less conservative debt policy approach.’ The latter firm is taking more risk and will be more uncertain about future income. Therefore, the degree of debt conservatism is a function of the degree of earnings volatility as well.

To capture this effect, a standardized kink variable has been added, adapted from Graham (2000). The proxy of this standardized kink equals the ratio of the amount of interest payments multiplied by kink, divided by the standard deviation of EBIT:

𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑𝑖𝑧𝑒𝑑 𝐾𝑖𝑛𝑘 = (𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡)(𝐾𝑖𝑛𝑘)

𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝐸𝐵𝐼𝑇

A high level of earnings volatility, i.e. a relative large value of the standard deviation of EBIT, results in a low value of Standardized Kink. A low level of earnings volatility should result in

a more aggressive debt policy. Therefore the third tax-related hypothesis states that Standardized Kink is expected to be positive correlated with leverage.

3.2 Controlling variables

Although the objective is to examine the relationship of corporate taxation on the leverage policy, the explanatory power of the other controlling variables should not be underestimated. Otherwise, the tax variables would be proxies for other variables to a great extent. Moreover, with a set of variables as complete as possible, the coefficients of the tax variables will be of more value.

Asset tangibility

Tangible assets, such as property, plant and equipment are easy to value for outside investors. Since tangible assets are more liquid than intangible assets, such as goodwill, these assets have a strong value as collateral for loans; as they lower the present value of expected distress costs. Moreover, Siglitz and Weis (1981) state that the collateral overcomes the problems of both moral hazard and adverse selection. This reduction in risk for the outsiders decreases the agency costs of debt associated with risk shifting as well. The reduced expected costs of distress and debt-related agency problems predict a positive relation between asset tangibility and leverage.

In contrast, a relatively high percentage of fixed assets in relation to the amount of total assets imply a higher level of operating risk as well. This suggests an increase of the probability of default, reducing the positive effect on leverage. It’s not expected, but this effect could offset the first argument and may switch the parameter of the proxy from positive to negative.

Also, Berger and Udell (1998) have argued that firms having a close relationship with their customers need to provide less collateral. This close relationship, i.e. more informed monitoring, should be a substitute for physical collateral. Default is less expected here. Hence this variable could be of less significance than expected.

Finally, a high level of fixed assets can solve the problem of underinvestment as suggested by Myers (1977). Moreover such a high level could reduce agency costs associated with risk shifting when used as committed collateral. Both Rajan & Zingales (1995) for G7 countries and Hall et al. (2004) for European SMEs found this factor positive and significant for long-term debt.

more willing lenders should be to supply loans, the higher the leverage should be. The proxy of asset tangibility is set at the ratio of fixed assets to total assets.

Asset intangibility

As noted above, discussing the asset tangibility, tangible assets are expected to have a positive effect on leverage. Intangible assets, such as goodwill, patents and trademarks, are expected to have the opposite negative effect. These assets have a poor value on the collateral of loans, as they are often associated with the distress costs.

In addition, since these assets are difficult to value by external creditors and investors, this issue can be seen as a problem of asymmetric information as well, intensifying the negative effect on leverage. Finally, a high percentage of intangible assets may represent future growth opportunities, facing mixed effects on leverage, as will be explained below. Titman & Wessels (1988) found a negative relation between this variable and leverage. The proxy of asset intangibility is set at the ratio of intangible assets to total assets.

Growth

Highly levered firms are likely to pass up profitable investment opportunities, since they face higher costs of financial distress and see themselves confronted with lenders requiring being compensated for the increase in risk (Myers, 1977). Therefore, in these situations firms will use a greater amount of other financial sources. This theory suggests a negative correlation with leverage. In addition, the pecking order theory implies that firms with many investments opportunities should accumulate more debt over time. Hence, growth and leverage are positively related according to this theory.

However, one could argue that a fast growing firm could be viewed as strong and healthy, facing many profitable investment opportunities. In this perspective the probability of bankruptcy should be lower and therefore result in a positive effect on leverage. For this variable Hall, Hutchinson & Michealas (2004) find both positive correlations in some countries and negative correlations for other countries. In summary, the debt level could be either decrease or increase when growth opportunities increase. The proxy of growth is set at the percentage change in total assets:

𝐺𝑟𝑜𝑤𝑡ℎ = 𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑡− 𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑡−1 𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠𝑡

Firm size

There are arguments for both a positive and negative expected effect of size on leverage. On the one hand, one could argue that large firms are more diversified and less loaded with volatile assets; reducing the risk of default. In addition, large firms generally have relatively less fixed costs. As mentioned above, a high level of fixed costs is often associated with default risk. Therefore, with a low level of fixed costs, the probability of fail should be smaller, as well as the costs of financial distress. A positive effect on leverage is to be expected, based on these arguments.

In contrast, relative large firms usually face more asymmetric information and agency problems between the firm’s insiders and the capital markets. These agency costs of debt can be reduced by issuing short-term debt. The size of the firm should be negative correlated to leverage here. Rajan & Zingales (1995) found a significant and positive effect for G7 countries. Hall et al. (2004) find the same effect for several European SMEs.

Therefore, it is expected that the negative effect of the second argument does not offset the first argument. Firm size should have a positive effect on leverage. The proxy of this variable has been defined as the natural logarithm of total assets.

Profitability

Like several other explanatory variables, there are arguments for both a positive expected effect and a negative effect on leverage. First, in times of relative high profitability, there is more room to exploit the benefits of the tax-shields, according to the trade-off theory. Moreover, profitable firms, generating a positive cash flow, will be associated with the probability of default and bankruptcy costs to a lesser extent. Therefore, investors are more willing to lend to profitable firms (Myers, 1984). Finally, as demonstrated by Jensen (1986), in control over large amounts of internal funds, managers may undertake projects with a negative NPV, entrenching themselves into the firm. Hence, it has been suggested that an increased level of debt serves as a bonding mechanism, reducing the possibility of wasting funds. These three arguments suggest a positive relationship between profitability and leverage.

However, Rajan & Zingales (1995) discussed the ineffectiveness of this last argument, arguing the preference of managers to avoid the disciplinary role of debt, suggesting a negative correlation between profitability and debt. Second, the pecking order theory suggests the preference of internal funds to debt in financing new projects, due to information asymmetries. Profitable firms will need relatively fewer external funds and use their internal

financial sources first.

Titman & Wessels (1988) found a negative relation between profitability and leverage, while, more recently; Rajan & Zingales (1995) found mixed evidence for various countries, just as Hall et al. (2004) for European SMEs. Most recently, Bartholdy & Mateus (2011) expect no asymmetric information problems with SMEs, due to continuous monitoring and renegotiation of these primarily bank-financed firms, tending to a positive correlation.

Profitability has been defined by earnings before interest and taxes (EBIT, or operating income) over total assets, just as Titman & Wessels (1998) and Rajan & Zingales (1995). An alternative measure is return on assets (ROA), calculated as earnings after taxes and interest over total assets, which will be included as well.

Return on assets

An alternative unit of measurement for the profitability of the firm is the return on assets. It is expected to capture the same effect as profitability. As it is an available, easy to calculate variable, this inclusion is valuable. The proxy slightly differs, as the proxy is set at the earnings after the deduction of interest and taxes. The relationship is expected to be positive as well.

Firm risk

Risk is a variable to be included, but there is doubt of its explanatory power. Higher firm risk indicates higher volatility of earnings and a higher probability of bankruptcy (Bardley et al., 1984; Titman and Wessels, 1988). However, this variable might correlate with growth opportunities. Like the expected outcome of the variable growth opportunities, this controlling variable’s expected relationship with leverage is unclear. Risk is defined as the standard deviation of return on assets.

Liquidity

Accumulated cash and other very liquid assets can serve as a measure of liquidity. As discussed, according to the pecking order theory, this financial source is favorable in the financing of future projects. Hence, a negative effect on leverage is expected. The proxy of liquidity is set as total current assets divided by total current liabilities.

Bankruptcy

function of the probability of bankruptcy. While bankruptcy is very costly, creditors require to be compensated for the increase in risk. Therefore, a negative correlation is expected. Altman (1968) constructed a formula, in attempt to predict bankruptcy. The initial equation applies to publicly trading firms, SMEs have not been taken into account. Following Bartholdy & Mateus (2011), a modified version of the function is used, as illustrated with the equation below: 𝑍 = 1.2 𝑤𝑜𝑟𝑘𝑖𝑛𝑔 𝑐𝑎𝑝𝑖𝑡𝑎𝑙1 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 + 1.4 𝑟𝑒𝑡𝑎𝑖𝑛𝑒𝑑 𝑒𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 + 3.3 𝑒𝑏𝑖𝑡 𝑡𝑜𝑡𝑎𝑙 𝑠𝑎𝑙𝑒𝑠 +0.999 𝑠𝑎𝑙𝑒𝑠 𝑡𝑜𝑡𝑎𝑙 𝑠𝑎𝑙𝑒𝑠

¹Working capital is defined as current assets minus current liabilities

The initial Z-score zones of discrimination:

𝑍 > 2.99 𝑆𝑎𝑓𝑒 𝑧𝑜𝑛𝑒

1.81 < 𝑍 < 2.99 Grey zone

𝑍 < 1.81 𝐷𝑖𝑠𝑡𝑟𝑒𝑠𝑠 𝑧𝑜𝑛𝑒

The nominal interest rate

The cost of borrowing rises as the nominal interest increases. The demand for debt is expected to decrease when the cost of borrowing rises. The nominal interest rate is set at the Euribor three-month rate.

Difference in the interest

The spread between interest rates on long- and short-term debt could be correlated with leverage. As the dependent variables consist of long-term, short-term and total debt, a distinction in the interest spread is expected to be valuable. The variable is defined as the difference between the risk-free rate, defined as the Euribor three-month interest rate, and the ten-year Euribor bond rate. A decrease in the spread implies an increase in the costs of long-term debt, whereas an increase has a similar effect on short-long-term debt. A positive correlation is expected on long-term debt, a negative relation to short-term debt.

Following Peterson & Rajan (1994), age can be an explanatory variable of the firm’s reputation. With an increase of the firm’s maturity, the amount of available information, e.g. financial history, increases as well. In general, a young firm will find it more difficult to obtain external financing. Therefore, a positive correlation is expected. However, a non-linear relation is suggested, while a firm’s reputation is not build in a few years. Moreover, the most financial transparent firms are the ones accessing bond markets, which are not included in the sample of this study. The natural logarithm of the amount of years since incorporation serves as a proxy for the reputation, or age, of the firm.

Inflation

Since inflation rates and the nominal interest rates are highly correlated, these controlling variables likely capture the same effect. Nevertheless this variable is included.

Table II illustrates the expected parameters, as well as the proxies of the explanatory and the controlling variables of the model, as illustrated in the discussion of the hypotheses.

4 Methodology

The main objective of this paper is to examine the effect of the marginal corporate income tax rate on the SME’s capital structure choice. This section discusses the selection of the sample, an illustration of the dependent variables, the specification of the model and the method of testing respectively.

4.1 Sample selection

The dataset has been collected from the Amadeus database of Bureau van Dijk. This database contains income statements and balance sheet data of nearly 20 million European firms, including over 900,000 companies within the Netherlands. The initial sample contained 6142 privately owned Dutch firms.

Exp. effect on Proxy

Tax variables leverage

MTR + marginal tax rate

KINK - ebit (amount interest to make earnings zero) /

actual interest payments

STAND + (interest * kink )/ standard dev. ebit

Controlling variables

AT + fixed (tangible) assets / total assets

AIT - intangible assets / total assets

GO - Δtotal assets / total assets

SIZE + natural log(total assets)

PROFIT + ebit / total assets

ROA - roa / total assets)

FR - std.dev.(roa) / total assets

LQ - total curr. assets / total curr. liabilities

BP - modified Altman's Z-score

INT - three-month Euribor interest rate

DIFFINT ltd +, std - difference Euribor ten-year & three-month interest rates

AGE + natural log(number of years since corporation date)

INF - 3-month Euribor interest rate

Dependent variables

LTD long-term debt

STD short-term debt

TD total debt

TABLE II

In order to yield a useful homogenous SME sample, several selection criteria have been imposed. Only private manufacturing firms with limited liability registered in the Netherlands have been included. Manufacturing firms are likely to have a relative high level of capital investments, in comparison with sectors as retail trade and wholesales. Public firms have been excluded. The required minimum amount of equity was set at €18,000, due to legal requirements. Firms with negative worth or taxes were excluded.

The research period ranges from 2003-2010 and only firms with data available for at least 4 consecutive years have been included. Observations with missing values have been discarded. In the end, the sample consisted of 327 firms with 2616 firm-year observations.

Descriptive statistics

Table III below presents the descriptive statistics of the dependent, explanatory and controlling variables. On average, short-term debt accounts for 4.5%, long-term debt for 7.1% and total debt 13.8%, whereas these amounts are scaled by the level of total assets. These percentages are reported, split by year in Table IV below. Differences in the cumulation of the short- plus the long-term and the total debt levels can be explained by an unequal amount of observations. The average value of the marginal tax rate is 24.8%. The maximum value is 29.6%, whereas this value represents the statutory tax rate of the year 2006. This year accounts for the first marginal tax forecast, as illustrated in the Appendix.

4.2 Dependent variables

Since the scope of the capital structure of SMEs is generally quite narrow, the focus of this paper is on the ratio of debt and equity: the leverage of the firm. The firm’s short- and long-term debt levels, as well as the firm’s total debt level will be the dependent variables. Short-term debt usually has a maturity of less than one year; long-Short-term debt exceeds this maturity of one year. Short-term, long-term and total debt will be scaled by the firm’s amount of total assets as a proxy for estimation. The average levels of the debt ratios are summarized and arranged by year in Table IV.

Variable Mean Median Maximum Minimum Std. dev.

Short-term debt 0.045 0.000 0.828 0.000 0.106

Long-term debt 0.071 0.001 0.644 0.000 0.110

Total debt 0.138 0.041 0.816 0.000 0.176

Marginal tax rate 0.248 0.255 0.296 0.000 0.056

Kink 83.889 5.877 19779.850 -1247.600 754.261 Stand - - - - -Asset tangibility 0.281 0.246 0.967 0.000 0.197 Asset intangibility 0.024 0.000 0.680 0.000 0.075 Growth opportunities -0.692 0.033 0.947 -1102.900 25.782 Size 4.720 4.653 7.362 2.033 0.656 Profitability 0.097 0.078 3.702 -1.912 0.160 Return on assets 0.067 0.054 3.668 -1.426 0.140 Firm risk 0.066 0.041 2.499 0.000 0.125 Liquidity 1.858 1.508 65.274 -9.588 2.248 Bankruptcy probability 2.743 2.544 22.771 -4.536 1.501 Interest 2.581 2.255 4.640 0.810 1.266 Diff. Interest 1.736 1.830 2.980 0.050 1.005 Age 39.496 30.000 308.000 3.000 32.001 Inflation 1.593 1.610 2.490 1.170 0.442 TABLE III

The proxies of the firms leverage, the short- and long-term debt ratios, do consider some measurement issues. Firstly, some first roll over short-term debt and use it for long-term financing due to lower rates or more flexibility. Secondly, some short-term loans are considered short-term debt, but get renewed each year and technically serve as long-term debt (Bartholdy and Mateus, 2011).

An alternative measure of debt would be an accumulation of the total amount of current- and non-current liabilities. However, this proxy encloses requirements of working capital as well. In order to deal with this issue and make a valuable contribution to the first regression as well, another dependent variable has been adopted. The dependent variable, total loans, consists of the amount of total liabilities minus total debtors.

4.3 Modelling the leverage

According to the trade-off theory, in the absence of the market imperfections, the agency-, transaction-, information asymmetries costs, the firm could easily react to any variation in its debt objective. At a given time 𝑡, the debt level 𝐷_𝑖𝑡 of a company 𝑖 sould equal its debt target

𝐷_𝑖𝑡∗_{, that is 𝐷}

𝑖𝑡 = 𝐷𝑖𝑡∗. However, the target ratio adjustment costs or transaction, as discussed

earlier, implies a partial adjustment to this equation:

𝐷_𝑖𝑡− 𝐷_𝑖𝑡−1 = 𝛾(𝐷_𝑖𝑡∗ _{− 𝐷} 𝑖𝑡−1) (1) 𝐷_𝑖𝑡 𝐷𝑒𝑏𝑡 𝑟𝑎𝑡𝑖𝑜 𝑖 𝑓𝑖𝑟𝑚 𝐷_𝑖𝑡−1 𝐷𝑒𝑏𝑡 𝑟𝑎𝑡𝑖𝑜, 𝑜𝑛𝑒 𝑝𝑒𝑟𝑖𝑜𝑑 𝑙𝑎𝑔𝑔𝑒𝑑 𝑡 𝑡𝑖𝑚𝑒 𝛾 𝑇𝑎𝑟𝑔𝑒𝑡 𝑎𝑑𝑗𝑢𝑠𝑡𝑚𝑒𝑛𝑡 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡

Year Short-term Long-term Total

2003 0.060 0.082 0.165 2004 0.061 0.085 0.173 2005 0.068 0.092 0.184 2006 0.071 0.083 0.182 2007 0.067 0.096 0.191 2008 0.015 0.041 0.060 2009 0.012 0.032 0.037 2010 0.007 0.014 0.014 TABLE IV

𝐷_𝑖𝑡∗ _{𝐷𝑒𝑏𝑡 𝑡𝑎𝑟𝑔𝑒𝑡 𝑟𝑎𝑡𝑖𝑜}

The debt target ratio 𝐷_𝑖𝑡∗ is the dependent variable to be modelled. The model is considered as a lineair function of the explanatory and controlling variables as discussed above:

𝐷𝑖𝑡= 𝛼 + 𝛽𝑇𝐴𝑋𝑇𝐴𝑋 + 𝛽𝑍𝑍𝑖𝑡+ 𝑒𝑖𝑡 (2) 𝛽_𝑇𝐴𝑋 𝐶𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑡𝑎𝑥 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 𝑇𝐴𝑋 𝐷𝑒𝑓𝑖𝑛𝑒𝑠 𝑡ℎ𝑒 3 𝑡𝑎𝑥 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 (𝑀𝑇𝑅𝐸𝐵𝐼𝑇, 𝑆𝑇𝐴𝑁𝐷, 𝐾𝐼𝑁𝐾) 𝛽_𝑍 𝐶𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑖𝑛𝑔 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 𝑍_𝑖𝑡 𝐷𝑒𝑓𝑖𝑛𝑒𝑠 𝑡ℎ𝑒 14 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑖𝑛𝑔 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠 𝑒𝑖𝑡 𝐸𝑟𝑟𝑜𝑟 𝑡𝑒𝑟𝑚

4.4 Testing the model: PLS

Equation (2) is the model to be tested. A Hausman test has been performed and the random effects model has been rejected, hence the fixed effects model is preferred. Autocorrelation and hereoskedasticity are accounted for in the estimation procedure.

5 Results

Table VIa reports the results of equation (2)of the panel data, performing a panel least square estimation. The model is performed once for each explanatory variable, on long-term debt, short-term debt and the total debt level. Due to collinearity problems, the variable STAND has been omitted, resulting in a total of six regressions.

The sample consists of 2616 firm-year observations over the period 2003-2010. The regression estimated: 𝐷𝑖𝑡= 𝑐 + 𝛽𝑇𝐴𝑋𝑇𝐴𝑋 + 𝛽𝑍𝑍𝑖𝑡+ 𝑒𝑖𝑡. 𝐷𝑖𝑡is the debt level of firm ‘i’ in year ‘t’. 𝑐 is the constant term. 𝑇𝐴𝑋 is the taxation proxy and 𝑍𝑖𝑡 is the proxy for the controlling variables. Superscript indicates significance at the 0.01(***), 0.05(**) and 0.10(*) percent levels. T-values are reported in parentheses. A panel least square regression is performed.

Tax variables MTR KINK MTR KINK MTR KINK

Controlling 0.117 * -2.4E-06 -0.145 2.0E-06 -0.004 -2.5E-06

variables [1.42] [-0.57) [-1.04] [0.16] [-0.02] [-0.16] AT -0.115 ** -0.071 * 0.202 * 0.158 ** 0.016 -0.039 [-2.00] [-1.93] [1.97] [2.11] [0.11] [-0.41] AIT 0.319 * 0.190 ** 0.037 0.056 -0.288 0.153 [1.77] [2.13] [0.18] [0.43] [-1.01] [0.76] GO -5.7E-05 -1.2E-04 -0.018 -0.022 -0.026 -0.093 *** [-0.39] [-1.22] [-0.45] [-0.83] [-0.45] [-2.87] SIZE 0.011 0.037 0.382 *** 0.155 *** 0.520 *** 0.328 *** [0.27] [1.49] [5.03] [3.01] [4.87] [5.15] PROFIT 0.044 0.067 -0.226 -0.083 -0.374 -0.324 * [0.48] [1.24] [-1.29] [-0.59] [-1.52] [-1.86] ROA -0.068 -0.071 -0.059 -0.004 -0.181 0.183 [-0.93] [-1.40] [-0.34] [-0.03] [-0.74] [1.08] FR 0.063 0.116 -0.428 * -0.054 0.023 0.155 [0.41] [1.47] [-1.79] [-0.30] [0.07] [0.69] LQ -0.004 -0.001 0.044 *** 0.020 ** -0.012 -0.022 ** [-1.06] [-1.00] [3.43] [2.58] [0.66] [-2.36] BP -0.014 -0.008 0.035 -0.005 0.065 ** -0.009 [-1.47] [-1.27] [1.61] [-0.32] [2.12] [-0.50] INT -0.002 -0.003 -0.007 -0.006 -0.014 ** -0.008 [-0.71] [-1.24] [-1.44] [-1.10] [-2.06] [-1.29 DIFFINT -0.015 *** -0.019 *** -0.003 -0.002 -0.024 *** -0.022 *** [-3.90] [-7.90] [-0.47] [-0.42] [-2.80] [-3.75] AGE -0.012 *** -0.010 *** -0.008 -0.006 * -0.024 ** -0.024 *** [-2.37] [-6.51] [.1.23] [-1.73] [-2.49] [-5.61] INFL - - - -- - - -Constant 0.542 * 0.345 *** -1.563 *** -0.477 * -1.448 ** -0.392 [1.95] [2.72] [-3.67] [-1.90] [-2.42] [-1.26] Adjusted R-squared 0.439 0.583 0.753 0.673 0.769 0.757 Total debt TABLE VIa

Panel least square regression Short-term debt Long-term debt

As mentioned in section 4.2, discussing the dependent variables, total loans is adopted as a measure of a firm’s debt as well. Below, in TABLE VIb, these results are presented.

In terms of explanatory power, the (adjusted) R-squared for the short-, long-term and total debt levels are distributed between 44% and 77%, while for the level of total loans the R-squared is showing a range from around 86% to 91%. Hence, the latter regression has provided a better fit.

5.1 Tax effects

Considering all four debt levels, only at the short-term debt level, the marginal tax rate appears to be weakly significant. KINK is not significant at any debt level. The explanatory variable standardized kink has been omitted of the regression while this variable, due to their

Tax variables MTR KINK Controlling 0.194 3.36E-06 variables (1.58) (0.44) AT -0.232 *** -0.201 *** (-2.72) (-3.02) AIT 0.077 0.422 *** (0.29) (2.65) GO -5.5E-05 -1.2E-04 (-0.26) (-0.65) SIZE 0.017 0.035 (0.27) (0.79) PROFIT -0.204 -0.131 (-1.51) (-1.35) ROA -0.431 *** -0.132 (-3.93) (-1.45) FR 0.156 0.059 (0.69) (0.42) LQ -0.038 *** -0.019 *** (-7.36) (-8.61) BP 0.041 *** 0.028 ** (2.86) (2.56) INT 0.001 -0.003 (0.28) (-0.47) DIFFINT -0.003 -0.005 (-0.45) (-0.86) AGE -0.014 * -0.013 *** (-1.89) (-4.58) INFL - 0.010 - (0.65) Constant 1.093 0.970 (2.65) (4.26) Adjusted R-squared 0.907 0.862 TABLE VIb

Panel least square regression Total loans

similarity calculation, captured the same effect as the variable KINK. Under the assumptions of this model, there is hardly any evidence of the tax effects. The hypothesis H1 has to be rejected for the long-term and the total debt level, contradicting earlier results from Bartholdy & Mateus (2011). See Table X below as well for the outcome of more results of prior research on SME capital structure.

For the short-term debt level we can accept the hypothesis. The positive correlation at a 0.10 percent significance level for the short-term debt level could be explained by the fact that it is easier to adjust the amount of short-term debt. When the earnings for a particular year appear to be high, the firm increases the amount of the short-term debt, taking the advantage of the tax shield. The opposite applies when the earnings are low.

5.2 Controlling effects

While the focus is on the effect of the marginal tax rate on the leverage of the SME, the importance of the implications of the controlling variables is considered as well. In contrast to the explanatory variables, there are controlling variables with a significant result.

Asset tangibility appears negative and strongly significant for total loans and significant for short-term debt. It has the ‘wrong’ sign however. For long-term debt the weakly significant parameters are positive however. In spite of the resulting significance of this variable, the negative signs were not expected. Asset intangibility is only weakly significant and with a positive sign at a short-term debt level. Like asset tangibility, this variable deals with the unexpected sign, whereas it may suffer the same explanation.

Growth opportunities are not significant at any debt level, the same result is observed with profitability. The hypotheses concerning these variables, considering expectations for both a negative and positive effect on the dependent leverage, are not supported in any way.

Due to collinearity with both MTR and KINK, the controlling variable inflation is automatically omitted of the estimations. The calculation of MTR includes the estimated EBIT, discounted for several years with the three-month Euribor rate, or risk-free interest rate. Since this rate is highly correlated to the inflation, this should be the explanation of this estimation issue.

The coefficient of the firm’s risk level is only weakly significant at the long-term debt level and appears with a negative sign. This result could be interpreted as evidence for the trade-off theory, whereas firm risk should contribute to the probability of bankruptcy. However, the variable accounting for this effect, the modified Altman’s Z-score, shows a much higher level of significance. Whereas the firm’s bankruptcy probability is significant for

total debt and strongly significant for total loans, the positive sign is unexpected.

Size is strongly significant for long-term debt and significant for total debt as well. As expected, the sign is positive. Liquidity is strongly significant and positive for long-term debt. This is consistent with the expectations; however the variable is strongly significant and negative for total loans. Interest rate is significant for total debt only and with a negative sign as expected. ROA is strongly significant and with a negative sign, only for total loans however.

Return on assets, the alternative measure for profitability, which appears not to be significant, is according the hypothesis. The interest spread, the difference between the short- and long-term interest rate is significant for total debt. As expected, for short-term debt, it has the negative sign as well as a strong significance coefficient. The firm’s age is strongly significant for short-term and total debt. For total loans it appears weakly significant. For all three debt levels it has the unexpected negative sign however.

Table X below illustrates the significant results of capital structure determinants of SMEs from prior research.

L ope z-G ra ci a & S ogor b-M ir a, 2008 M ic ha el as , C hi tte nde n & P out zi our is , 1999 H al l, H ut chi ns on, M ic ha el as H eym an & D el oof & O oghe , 2007 C hi tte nde n & H al l & H ut chi ns on, 1996 B ar thol dy & M at eus , 2011 st + - + -lt + + + + t + + + + st lt t st + -lt + + t + + + - + st + - + lt + + t - + -st - - + lt - + t - - + st -lt -t -st + lt + -t + st lt t st lt -t -st -lt -t -st lt t st - - -lt - -t - -st lt t st + lt t + DIFFINT AGE INF MTR TABLE X

Prior research on SME's leverage determinants

AIT AT GO SIZE PROFIT ROA FR LQ BP INT

6 Conclusion

In this research paper, the capital structure of Dutch small- and medium enterprises has been examined. The main objective was to examine the effects of the marginal corporate income tax rate on the SME’s debt levels. The trade-off theory and the pecking order theory have jointly been tested.

As the results of this paper demonstrate, no strongly significant tax effects on the capital structure of small- and medium enterprises have been observed. Hence, the hypotheses regarding the tax effects are not clearly supported. Only at the short-term debt level, the marginal tax rate appear to be weakly significant. The importance of short-term debt in capital structure decision making has been noticed by Scholes et al. (1990). They argue that firms, facing uncertainty in their tax situation, may prefer to use short-term debt when their tax rate appears to be high. Bartholdy and Mateus (2011) suggest that in this setting, short-term debt will be the least costly and easiest way to adjust debt levels temporarily to a firm’s optimum, while the potential cost of retiring outstanding debt in the future is avoided. Hence, in addition, a correct measure of debt to capture the tax effects in capital structure should include short-term debt as part of the calculation.

Blouin et al. (2010) documented and described the possible measurement errors used to estimate future taxable income in MTR computations. They showed that these problems have a significant impact on these corporate MTR computations. As described in the Appendix, there is a clear margin for error, estimating future taxable income, in order to calculate the MTR. These issues, among others, could partly explain the insignificance of the marginal tax rate parameters within the current research.

Furthermore, only a minority of the controlling variables are showing any level of significance at the multiple regression results. There are three variables, strongly significant and with the expected corresponding sign: size, for long-term debt and total loans, liquidity, for total loans, difference in interest, for short-term debt, and return on assets, for total loans. This implicates some evidence for the trade-off theory. The results appear to be consistent with the hypothesis and support prior research (Rajan & Zingales, 1995; Hall, Hutchinson & Michaelas, 2004).

Since bankruptcy probability has the wrong sign for all debt levels, implications are difficult to make here. Asset tangibility, -intangibility, age and bankruptcy probability are strongly significant, however in the unexpected directions. These results could indicate that the costs of financial distress play a much greater role in short-term debt considerations.

Growth opportunities and profitability appears to be insignificant at all debt levels. . Actually, this result is not consistent with the hypotheses, derived from the trade-off and the pecking order theory, as this does not support the underinvestment problem of Myers (1977) as well. Furthermore, the result contradicts prior research (Rajan & Zingales, 1995) Liquidity is strongly significant and positive for long-term debt, hence consistent with the expectations and supportive for the pecking order theory.

Two important notes can be made in the discussion of the results. Firstly, the majority of the variables derived from the trade-off theory and the pecking order theory do not significantly determine the capital of small private firms. However, as Bartholdy and Mateus (2011) stated, the insignificant results do not necessarily implicate that these variables do not determinant the debt target ex-ante. The implications of these factors are only just not observable ex-post from the balance sheet. If the limitations of data, regarding small private firms, can be overcome, one might find more significant results.

Secondly, a few determinants of the SME’s capital structure appear indeed to be relevant. However, the results are less straightforward than the hypothetical effects discussed above. In conclusion, it would be incorrect to conclude that agency and information asymmetric information problems are not important in determining the optimal capita structure for SMEs. It may not be possible to test for these issues based on just balance sheet data. Small private firms may deal with these problems in a different way than large public