Abstract
This paper takes into consideration the major theories about capital structures and evaluates these ideas to determine which of them relate to corporate financing choices. What unique is in this paper is that it examines the influence the financial crisis had on the capital structure several years after it occurred in 2008. The main results and conclusions are that the tax rate has a significantly negative relationship with the leverage level, however, this relationship was expected to be positive because of the tax benefits. This research finds that the size of firms is significantly positive related to the leverage ratio. Furthermore, this research finds that profit is significantly negative related to the leverage ratio. These results are matching with the theories. However, the relationship between the age and the leverage ratio does not yield a significant outcome, most likely owed to the large time frame of the data sample. The financial crisis had, as was expected, a significant positive relationship with the leverage ratio. In this research, the relation of debt maturity with leverage ratio remains unexplained.
The effect of firm and country specific factors on the
capital structure
1. Introduction
Corporate financing choices can be determined by different factors that are related to firm and country specific characteristics. Since the paper of Modigliani and Miller (1958), which will be taken a look at in the literature review, there has been a lot of research about corporate financing choices. Studies such as Rajan and Zingales (1995), Wald (1999), and De Jong, Kabir and Nguyen (2008) examined the corporate financing choices of firms. However, these papers are inconclusive concerning which firm and country specific factors influence the financing choices of companies. Furthermore, recent papers of Graham, Leary and Roberts (2015) and Turk Ariss (2016) suggest that further research about corporate financing choices is still needed.
Besides the inconclusiveness that exists about the factors that influence the corporate financing choices in the papers mentioned above, the global financial crisis of 2008 could have an important influence on the corporate financing choice of firms. Reinhart and Rogoff (2009) state that the global financial crisis of 2008 was the biggest financial crisis since the Great Depression in the 1930s. Many financial institutions collapsed, while others survived solely because they were nationalized or received massive state support. The global financial crisis of 2008 affected major financial centers across the entire world. The paper of Demirgüc-Kunt, Martinez Peria and Tressel (2015) states that the global financial crisis of 2008 offers an important opportunity to study the effects that macroeconomic and financial instability might have. This research is adding to previous research, because it investigates the corporate financing choices of firms several years after the financial crisis. There has been little research about the influence of the financial crisis on the corporate financing choices and about what happened the years following the crisis. However, the paper of Demirgüc-Kunt et al. (2015) examined the effect of the financial crisis on the corporate financing choices through the year 2011 and the paper of Mouton and Smith (2016) investigated this effect through the year 2013. In this research, the effect of the financial crisis on the corporate financing choices through the year 2015 is investigated.
One of the corporate financing choices that companies have, is that they can be financed with both internal and external financing. Internal financing is raised by retaining the earnings that are generated. External financing is generally sourced from capital markets and consists of debt and/or equity. The difference between debt and equity is that debt must be paid back to the debtholders and debt is senior to equity, which means that debt has to be paid back in full before the equity claims can be paid out.
The advantage of equity financing is that the firm does not have to pay interest, no negative cash drain, and there is no restriction on its assets. In the case of debt, a creditor might require a pledge of assets for collateral which can restrict the firms' operations. The disadvantage of equity financing is that it reduces the share of profit to existing stockholders.
The capital structure of a company defines how the external financing of a firm in the capital market is built up. When a firm is only financed with equity it is called an ‘unlevered firm’, after adding debt to the company it is called a ‘levered firm’. The amount of debt that the company holds in comparison to the total amount of debt and equity the firm holds specifies the leverage ratio, which is debt/total assets. In this research, we want to look at the influences of different factors on the capital structure of firms. Specifically, to look at what the effect is of differences in country and firm specific factors on the capital structure of companies. Where the leverage ratio will be used as measurement for capital structure.
Besides the global financial crisis of 2008, this paper also investigates the effect of the debt maturity. Debt can exist out of short-term debt that has to be paid back in one year and long-term debt that has to be paid back over a longer period than one year. Therefore, short-term debt carries more risk than long-short-term debt. Debt maturity refers to the point in time the principal is due to be paid. In this research, the leverage ratio looks at the total debt used divided by the market value of total assets. As mentioned, short-term debt is more risky than long-term debt. Therefore, the tests are also done for short-term debt divided by the market value of total assets, the short-term leverage ratio, and the long-term debt divided by the market value of total assets, the long-term leverage ratio. The results of the short-term leverage ratio are compared to the results of the long-term leverage ratio, to find the influence that debt maturity has.
Furthermore, the level of development of a country may have an impact on the capital structure of companies. Countries with a low level of development may have less access to financing for their new investments. This may lead to lower leverage levels for a lower level of development in a country.
The main question of this research will therefore be:
“Which firm and country specific factors can explain the capital structure choice of companies?”
According to the theory of Modigliani and Miller (1958) which is explained further in the literature review it should, in perfect markets, not matter if the firm is financed with debt and/or equity, because in perfect markets the firm can always get new financing. However, when there are market imperfections, such as taxes, it is important to look at the choice for the capital structure. In the real world, there are market imperfects and firms can gain from using more debt, because debt has tax benefits.
Besides the Modigliani and Miller theory, there are two theories that focus on the capital structure choice: The Static off Theory and the Pecking Order Theory. The Static Trade-off Theory suggest that companies should hold the optimal level of debt-to-equity that balances the cost and benefits of debt financing. There are tax benefits on debt, but debt is riskier and could bring costs of financial distress with it. Furthermore, the Pecking Order Theory suggests that firms should first finance with retained earnings before they should get external financing. When the firm gets external financing, the firm should first issue debt before equity is issued. This is should be done to prevent the firm from asymmetric information problems. In this research those three theories, are used to find firm specific factors that can explain the choice of companies for their leverage levels. Besides these theories, other firm and country specific factors will be taken into consideration.
The data used in this research is gathered from the Compustat Global database and the Central Intelligence Agency. The data consists of panel data of 579 firms over a period of 12 years from 2004 till 2015. The total sample consists of 6925 observations.
The main outcomes are that taxes are negatively related to the leverage ratio, which is contrary to what was expected according to the Modigliani and Miller theory. The size of the firm is positively related with the leverage ratio, which is what was expected. The relation of the age of the firm is not significant which can be explained by the large time period the sample has, therefore bankrupt firms and new firms are deleted from the sample. Leaving only older firms in the sample. Profit is negatively related to the leverage ratio, which is also what was expected. Taking the significant factors in consideration, it can be concluded that both the pecking order theory and static trade-off theory can explain the leverage level of firms. The financial crisis has a significant positive relationship with the leverage ratio and the rate of development also has a significant positive relationship with the leverage ratio. The relation that debt maturity has with the leverage remains unexplained in this research.
2. Literature review
In this section, the different theories about the corporate financing choices of companies are taken into consideration. Based on these theories an econometric model is developed to test the hypotheses. As stated in the introduction there are several theoretical perspectives about corporate financing choices. However there still exists inconclusiveness about which firm and country specific factors influence the financing choices of companies. Therefore, more research is needed.
Besides the factors, which can have an influence on the capital structure of the companies and will be taken into consideration, the debt maturity may also have an influence on the capital structure. The firms are namely not only financed with plain debt and equity, there are two types of debt in the dataset that is used; short term debt and long-term debt. To find the relation of the corporate financing choices of firms it is important to classify the debt. The maturity of debt can affect the leverage ratio too, because a firm wants to match the short-term debt to short-short-term investments and long-short-term debt to long-short-term investments to decrease liquidity risks. The liquidity risk arises if the company cannot meet its short-term debt obligations on time. Short-term debt is thus more risky than long-term debt. Liquidity risk increases the costs of financial distress and therefore the higher liquidity risk should lower the short-term leverage ratio.
The research of Titman and Wessels (1988) states that the cost of issuing debt and equity is related to firm size. Small firms, they state, pay more than large firms when issuing new equity, but also when issuing new long-term debt. This would lead to higher short-term debt ratios and also higher leverage ratios in smaller firms. Their research confirms that smaller firms use significantly more short-term debt than larger firms. This conclusion is contrary to the Static Trade-off Theory that suggests that firms decide on their leverage level by balancing the cost and benefits of debt and where larger firms have lower costs of financial distress and can maintain higher leverage levels.
The debt maturity will be taken into consideration by doing the test three times. First for the total leverage level, then for the short-term leverage level and finally for the long-term leverage level. The short-term leverage level can then be compared to the long-term leverage level. This way of measuring the debt maturity is recommended by Serrasqueiro and Caetano (2015). Therefore, the all developed hypotheses are split into three hypotheses: 1) for the leverage ratio, 2) for the short-term leverage ratio and 3) for the long-term leverage ratio. This testing for debt maturity will be a robustness test.
One of the most important theories in the history about corporate financing choices is the theory of Modigliani and Miller (1958). They state that in a perfect capital market there are no transaction costs, no taxes, there is no asymmetric information and there are no costs occurred by financial distress. Therefore, it is irrelevant whether the firm gets financed with debt and/or equity. The main goal of a firm is profit maximization and market value maximization. Whereas the value of a firm can be calculated by discounting the free cash flows the firm generates at the relevant cost of capital, in the perfect market the cost of capital is the same whether it is financed with debt and/or with equity. The firm can, in a perfect market, always get financing for positive net present value projects and therefore the value of the firm and the firm’s stock price is not affected by its capital structure. However, in the real world there are market imperfections such as taxes and when there are taxes, higher debt levels are favourable. This is because of the treatment debt gets by tax officers. Debt financing is cheaper because of the tax break on debt, the fact is that the interest charges are tax deductible. The value of the firm can still be calculated by discounting free cash flows at the relevant cost of capital, but now the cost of capital is adjusted with taxes and are no longer constant when leverage levels change. The value of the firm and its stock price will change when the leverage ratio changes. Taking the explanations and results of the theory of Modigliani and Miller into consideration, when there are higher taxes a company can gain more from higher leverage ratios. This will not only be tested for the influence of the factor taxes on the leverage ratio, but this factor will also be tested for the debt maturity, thus for the short-term leverage ratio and the long-term leverage ratio. Therefore, the first hypotheses will be:
H1: Tax rates are positively related to the leverage ratio
H1a: Tax rates are positively related to the short-term leverage ratio H1b: Tax rates are positively related to the long-term leverage ratio
However, tax rates may not be the only factor that influences the choice of firms for their capital structure. Kraus and Litzenberger (1973) state that in the real world, there are also costs of financial distress. When debt obligations cannot be paid back the firm faces bankruptcy. Thus, debt is riskier than equity. Companies should therefore not only look at the tax benefits of debt and use as much debt as possible, but firms should also look at the costs of debt, for example the costs of financial distress. Firms can gain the most when they choose the optimal balance between the cost and benefits of debt and choose in that way the optimal capital structure. This theory is called the Static Trade-off Theory.
Firms that base their capital structure on the Static Trade-off Theory are therefore expected to use a target debt to equity level. Thus, the paper of Kraus and Litzenberger, as stated above, suggest that the company should hold a trade-off between the costs and benefits of debt. They concluded that the leverage level is dependent on the benefits that taxes have, but is also dependent on the costs of debt, for example the costs of financial distress, which is mentioned in this research as bankruptcy costs. Bankruptcy costs can have a large impact on the leverage level that is optimal for a company. Gruber and Warner (1977) recognize this and examines the bankruptcy costs. They state that there are direct and indirect bankruptcy costs. Direct bankruptcy costs are the costs of lawyers, accountants and other professional fees and the time the management spent administrating the bankruptcy. The direct bankruptcy costs can be measured. The indirect bankruptcy costs are, for example, lost sales and profit and the inability of the firm to get new financing, these bankruptcy costs cannot directly be measured. In their study, Gruber and Warner look at the direct costs of bankruptcy and the risk and return characteristics of defaulted debt claims. They conclude that the ratio of direct bankruptcy costs is lower with higher market values, because direct bankruptcy costs are fixed and thus a relatively lower percentage of the firm value when the market value of the firm is larger. If firms with higher market value have a lower percentage bankruptcy cost, this would mean that they have less risk of going bankrupt when they use higher leverage ratios and take more profit from the tax benefits. From the perspective of the Static Trade-off Theory, this will lead to a leverage level that will be higher for larger firms.
The question can be raised, if it is really the case that companies base their capital structure on the costs and benefits of holding debt. In his research, Miller (1977) states that most companies have less leverage than the optimal leverage level the Static Trade-off Theory would suggests. This paper suggests that this is because taxes are substantial and ongoing and bankruptcies are rare.
Furthermore, Hovakimian, Kayhan and Titman (2011) also investigated the Static Trade-off Theory and determined whether or not the default probabilities are consistent with what the Static Trade-off Theory suggests. Contrary to the predictions of the Static Trade-off Theory, they find that smaller firms choose higher leverage levels than expected. They state that this is the case because smaller firms have less access to the capital markets and are more sensitive to negative profitability and equity value shocks, which makes small firms more sensitive to bankruptcy risk. This already indicates that the Static Trade-off Theory is probably not the only theory that can give a good indication about how firms choose their capital structure.
Besides the static trade-off theory, as previously mentioned, there is also a so-called Pecking Order Theory. The Pecking Order Theory starts with what happens when there is information asymmetry. When there is information asymmetry and the manager knows more about the firms’ value, prospects and risks than the investors know, it will be better for managers to issue debt than equity. The reason for this is that issuing debt signals the investors that the new investment is profitable. Furthermore, issuing debt instead of equity for new investments signals that the current stock price is undervalued. If the stock price would be overvalued issuing equity would be chosen above issuing debt, but that would signal a lack of confidence in the board. Issuing new equity would therefore lead to a drop in the share price. The cost of financing increases with asymmetric information. To solve this asymmetric information problem, firms finance their new projects according to the Pecking Order Theory. With the Pecking Order Theory, the firms prefer internal financing above external financing. When internal financing is not possible, outside financing is used where debt is preferred above equity according to the paper of Myers and Majluf (1984).
The research of Shyam-Sunder and Myers (1999) examined the explanatory power of both theories, the Static Trade-off Theory and the Pecking Order Theory, and concludes that the Pecking Order Theory has a higher explanatory power than in the Static Trade-off Theory. Even when a firm has a well-defined optimal target debt level, the managers are not interested in holding onto these target debt levels.
Contrary to the research of Shyam-Sunder and Myers, Sogorb-Mira and Lopez-Gracia (2003) find that the Static Trade-off Theory has a higher explanatory power than in the Pecking Order Theory. However, they state that both theories can help to explain the choice of firms for their capital structure. The results of this research clearly indicate that there is an optimal target debt level. One of the reasons for this difference between the two papers stated above could be that the paper of Shyam-Sunder and Myers uses a sample of large public firms, where the paper of Sogorb-Mira and Lopez-Gracia uses a sample of small and medium firms. The size of the firm is therefore an important consideration. The second hypothesis will therefore be:
H2: The size of firms is positively related to the leverage ratio
In contrast to the findings of the papers mentioned above, the research of Cotei and Farhat (2009) finds that the Pecking Order Theory and the Static Trade-off Theory are not mutually exclusive. They state that firms may hold a target debt ratio range, but that within this range the Pecking Order Theory may describe the decisions made about the capital structure. They also conclude that the firm may switch over time between the Static Trade-off Theory and the Pecking Order Theory. Thus, besides the size of the firm, the age of the firm might also be an important firm specific factor in the decision that the firm makes about its capital structure.
The paper of Hovakimian, Opler and Titman (2001) also finds that the Static Trade-off Theory and the Pecking Order Theory are not mutually exclusive. They also find that the firms switch from the Pecking Order Theory in the short run to the Static Trade-off Theory in the long run. Where the target trade-off may also change over time, when the firm’s profitability and stock price change. This paper finds that the leverage levels decrease when the firm becomes more mature and larger. This also explains the difference between the paper of Shyam-Sunder and Myers that tested the two theories in large firms and finds a higher explanatory power for the Pecking Order Theory and the paper from Sogorb-Mira and Lopez-Gracia that tested the two theories in small and medium firms and finds a higher explanatory power for the Static Trade-off Theory.
The paper Pfaffermayr, Stöckl and Winner (2013) also finds that the firm’s age has a negative relation to the leverage level. Their argument for this finding is that mature firms tend to have more internal funds available from retained earnings and these firms will subsequently reduce therefore their reliance on debt. Their research concludes that older firms rely less on debt than younger firms. Firm age could therefore have an influence on the leverage ratio. The third hypothesis will therefore be:
H3: The age of the firms is negatively related to the leverage ratio
This factor will, as is done in the first hypotheses, also be tested by three hypotheses, to find the effect of debt maturity.
Vermaelen (1981) states that stock repurchases express positive information, and therefore insider buying is a good signal, where selling is a bad signal. When there is asymmetric information, the manager has more information than the outside investor. There are good and bad managers, Good managers exert productive effort and have therefore higher net present value projects than bad managers who exert destructive or little effort. The outside investor does not know whether the manager is a good or bad manager. It is important for the good managers to let the outside investors know they are the good managers, because only good managers have positive net present value projects for sure and can get financing. When there exists asymmetric information in firms, the good managers can signal that they are the good managers by buying their own shares back. Bad managers will not buy their own shares back, because their shares do not have a positive net present value. Thus, buying back shares means that there will be higher net present values of projects and therefore also higher profits. This would mean, that higher earnings lead to more equity and therefore a lower leverage ratio.
Furthermore, looking back at the Pecking Order Theory stated above, the past profitability of a firm yield a higher amount of earnings which can be retained. Titman and Wessels (1988) state that this would lead to less dependence on debt and therefore a lower leverage ratio. These two theories thus have the same conclusion, higher earnings lead to a lower leverage ratio. The fourth hypothesis will therefore be:
H4: Firms profit is negatively related to the leverage ratio
This factor will, as is done in the first hypotheses, also be tested by three hypotheses, to find the effect of debt maturity.
What was the impact of the financial crisis on the capital structure of firms? The global financial crisis of 2008 had a tremendous effect on the financial markets around the world. The paper of Demirgüc-Kunt et al. (2015) concludes that even the developing countries were affected. They find, in the years following the financial crisis, the leverage ratio went down. This is the case even in countries that did not experience the crisis. Around the time of the financial crisis of 2008, the chance that a firm went bankrupt was much higher. This was the case, because there was a liquidity problem and the bankruptcy of one firm could have a domino effect causing liquidity problems at other firms that the bankrupt firm was working with. This led to lower reliance on risky debt and therefore lower leverage ratios after the financial crisis. However, the paper of Mouton and Smith (2016) also examined the effect that the financial crisis had on the capital structure and did not find a significant result for the effect of the financial crisis. The effect of the financial crisis will be tested by the fifth hypothesis:
This factor will, as is done in the first hypotheses, also be tested by three hypotheses, to find the effect of debt maturity.
Besides the Static Trade-off Theory and the Pecking Order Theory, the level of development of a country may also have an impact on the capital structure of the companies. The countries with a low level of development may have less access to financing for their new investments. Besides little access to finance, those countries may also be suffering from an unstable economy and other corporate restrictions. Several papers investigated the influence of the capital structure in developed and developing countries. For example, the paper of Booth, Aivazian, Demirguc-Kunt and Maksimovic (2001), which investigated the capital structure in developing countries, which they compared to the capital structure of developed countries and concluded that the decisions are affected by the same factors as in developed countries. However, they state that there are differences across countries, such as GDP growth and capital market development. These differences can be explained by country specific factors.
The paper of De Jong et al. (2008) also find that the same factors that influence the capital structure in developed countries also influence the capital structure in developing countries. However, this paper looked further and found that country specific factors suggest higher leverage ratios. They used different factors to separate the developed countries from the developing countries and concluded that creditor right protection, bond market
development and GDP growth rate had a significant impact on the capital structure choice of firms. They further found evidence that legal enforcement, credit/shareholder right protection and macro-economic measures such as capital formation and GDP growth rate had influence on the leverage level of companies. Their conclusions are that developed countries have higher leverage levels and that the factors they tested of the firms in these countries have a higher effect on the capital structure choice. Therefore, the sixth hypothesis will be:
H6: The level of development of a country is positively related to the leverage ratio
This factor will, as is done in the first hypotheses, also be tested by three hypotheses, to find the effect of debt maturity.
3. Methodology
This section defines the methodology of this study. The development of the econometric model is described, the models used in this research to find the influences on the capital structure are explained and, finally, these models are tested for the Gaus-markov assumptions.
3.1 Development of econometric model
There is an econometric model developed to test which factors have an influence on the capital structure of firms. In this econometric model, one dependent variable should be chosen. For example, the leverage level, that explains the influence that the different factors, the explanatory variables, have on the dependent variable. Hence, when taking the debt maturity into consideration the tests should be done three times: 1) for the leverage ratio, 2) for the short-term leverage ratio and 3) for the long-short-term leverage ratio. The econometric model used in this research is stated below:
L = " + T#+ S# + A# + P# + C# + C*T# + D# + ε
Where L is the leverage ratio, T is the tax rate, S is the size of the company, A is the age of the company, P is the profit, C is the crisis dummy variable, C*T measures the moderating effect of the crisis and D is the level of development. In the regression stated above " is the intercept and the # measures the impact of the firm specific factors on de leverage level L. There is also an error term ε, this is needed because the model probably leaves out some determinants of the leverage level. But the error term is also needed because there may be errors in the measurement of leverage or there could be outside influences that cannot be modelled.
The component L is the dependent variable and denotes the leverage level. The leverage level is computed as the total debt divided by the market value of total assets, which is a widely-used measurement for leverage level, and also widely-used in the papers of De Jong et al. (2008), and Fan, Titman and Twite (2012). Thus, the leverage level is the total debt divided by the market value of total assets. The market value of total assets is calculated as book value of total assets, which is the total assets minus the intangibles, minus book value of equity, which is the difference between the total assets and total liabilities, plus market value of equity. The components used for this calculation are sourced from Compustat Global database.
$ = '(')* +,-'
The first explanatory factor that is used in the regression is taxes. According to the first hypothesis, it is expected to be the case that there is a positive relationship between taxes and leverage ratios, because of the interest charges on debt that are tax deductible. The tax rate is denoted in the regression as T. The tax rate used from the Compustat Global database is the total of the income taxes divided by the revenues.
The second explanatory factor that is tested for in the regression is the firm size. The size of the firm is expected to have a positive relation with the leverage level. The size of the firm is denoted as S in the regression. In this research the firm’s size is measured in terms of the amount of sales. In Compustat the sales are used to determine the size of the firm.
The third factor that will be tested in the regression is firm age. This is expected to be negatively related to the leverage level of the firm. Firm age is denoted as A in the regression. The fourth factor that will be tested in the regression is the firm profit. This is expected to be negatively correlated to the leverage level and is denoted as P in the regression. The profit used in this research is the gross profit which is the revenues minus the cost of goods sold, which are sourced from Compustat.
The fifth factor that will be tested in the regression is the financial crisis, denoted as C in the regression. The financial crisis will be a dummy variable for the years 2008, 2009, 2010 and 2011. Besides the influence the financial crisis has on the intercept, which is tested by the dummy, the influence on the slope is also tested by the sixth factor. This factor will measure the moderating effect of the financial crisis.
The seventh factor that will be tested is the level of development, denoted as D. To determine the level of development of countries, the GDP per capita of the different countries will be tested. However, this factor is not available in the Compustat database and will be sourced from Central Intelligence Agency.
To find what the effect is of debt maturity, the ordinary-least-squares tests are taken three times, with three different dependent variables: 1) the leverage level, 2) the short-term leverage level and 3) the long-term leverage level. These results will be compared to each other to find if debt maturity has an effect on the leverage level.
The developed econometric model will be tested by developing three different models: - Model 1: The first model tests for the theory of Modigliani and Miller (1958), which
states that the tax rate has a positive relation with the leverage ratio.
- Model 2: The second model will test for all the factors of the firms that can have an influence with the leverage ratio.
- Model 3: The third model will test for all the factors, just as in model 2. In addition, Model 3 will test for the influence of the financial crisis and for the influence presented by the level of development.
3.2 Models used for testing
The models can be tested using a pooled ordinary least squares model. However, this way of measuring has some limitations, because pooled data assumes that the average values remain constant over time and across the cross-sectional data. When using the pooled ordinary least squares regression, we thus assumed that the data used in this research remains constant over time and across cross-sectional data. This might be an inappropriate assumption. In this research, it is already mentioned that the data changes over time because of the financial crisis. Besides the change over time, the data may also change between firms. The fixed effect model can help with these limitations. The fixed effects model, where the time is fixed, namely allows the intercept in the regression model to differ cross-sectionally but not over time, while all of the slope estimates are fixed both cross-sectionally and over time. With the Redundant fixed effects model, it can be tested, if this way of testing eliminates the limitations of the pooled ordinary least squares regression.
However, using the Hausman test it can be determined if the random effects model is preferred over the fixed effects model. The random effects model is a more complicated case of the fixed effects model. The random effects model assumes that the data used are sourced from a hierarchy of different populations, where these differences relate to that hierarchy. Other than the fixed effects, the random effects model allows for individual effects.
For model 1, the Redundant fixed effects model shows that the probability is 0.0000 when the cross-section data is fixed, which indicates that the fixed effects model is preferred over the pooled model. When the cross-section data and the period data is fixed, the
For model 2, the redundant fixed effects model shows a probability of 0.0000 for fixing both, the cross-section data and the period data. Therefore, the fixed effects model is preferred over the pooled ordinary least squares model. Looking at the random effects model and doing the Hausman test, the probability is 0.0135, indicating that we reject this test at a 5 percent significance level and prefer the fixed effects model above the random effects model. Doing the Hausman test also for the relationship between the tax rate and short-term leverage ratio and the tax rate, the probability is 0.0039. This indicates that the null hypothesis is rejected at a 1 percent significance level and the fixed effects model should be used. Doing the Hausman test for the relation between the tax rate and the long-term leverage ratio, the probability is 0.5915, indicating that the null hypothesis should not be rejected and the random effects model should be used. However, when using the random effects model, we cannot compare these results with the results of the fixed effects model. Besides that, none of the factors are significant and therefore these results are not taken into account.
For model 3, changing the model from a pooled ordinary least squares model to a fixed effects model gives an error in Eviews, therefore it is not possible to use the fixed effects model for model 3. However, when changing the model to a random effects model and doing the Hausman test for model 3 the probability is 0.0421. This indicates that we reject this test with a 5 percent significance level and the pooled ordinary least squares regression is preferred. The probability of the Hausman test when looking at the short-term leverage ratio is 0.0163 and should therefore also be rejected at a 5 percent significance level. However, when doing the Hausman test for the long-term leverage ratio in model 3 gives a probability of 0.7880, indicating that the null hypothesis should not be rejected and the random effects model is preferred. In this research, we want to compare the leverage ratio with the short-term and long-term leverage ratio and using the same model is therefore necessary. For model 3, the pooled ordinary least squares model will therefore be used.
3.3 Best linear unbiased estimator
The models used for testing the hypotheses are tested for the Gauss-markov assumptions. When the Gauss-Markov assumptions are not violated, the models used give the best linear unbiased estimators (BLUE). One of the Gauss-markov assumptions is that a constant should be added to the model. Therefore, a constant is added to all the models, which implicates that the errors have a mean of zero.
Furthermore, another Gauss-makov assumption is that the models have homoskedasticity. In the absence of homoskedasticity there appears to be heteroskedasticity. When heteroskedasticity exists, the errors do not have a constant variance. The White test can be used to see if there exists heteroskedasticity, when the null hypothesis is rejected there exists heteroskedasticity. For model 1, the probability of the White test is 0.9886, which means that the null hypothesis is not rejected and heteroskedasticity should not be a problem. For model 2, the probability of the White test is 0.564837 and for model 3, the probability of is 0.983055. Therefore, heteroskedasticity should also not be a problem for model 2 and model 3. When there would have been heteroskedasticity there should be accounted for by using White’s heteroskedasticity consistent standard error estimates.
Besides the test for heteroskedasticity the autocorrelation is tested. When there exists autocorrelation, there are patterns in the residuals. To find if autocorrelation is a problem, the Durbin-Watson test is used. The Durbin-Watson test statistic value is gathered from the Eviews output. The relevant critical value can be gathered from the table of the Durbin-Watson test. When there is no autocorrelation the test statistic value should be between 4 minus the lower critical value and 4 minus the upper critical value. For model 1 the test statistic value is 0.798010, using the table of the Durbin-Watson test, the test statistic value should be between 2.48 and 2.44. The test statistic value is below the lower critical value and therefore the null hypothesis of no autocorrelation is rejected and the residuals from the model appear to be positive autocorrelated. For model 2 the test statistic value is 0.869333. However, the test statistic value should be between 2.54 and 2.37. Therefore, the test statistic value is also below the critical value and there thus appears to be positive autocorrelation. For model 3 the test statistic value is 0,164949 and the test statistic value should be between 2.56 and 2.35. This test statistic value is also below the critical value and there thus appears to be positive autocorrelation.
3.4 Multicollinearity
In this research there is also tested for multicollinearity. This is done by making a correlation matrix for the explanatory factors, see table 1. When the correlation is between 0.5 and 1 or between -0.5 and -1, there is a strong correlation. The correlation matrix shows only a high correlation between size and profit, this could cause multicollinearity. One of this factor can be deleted, however this is not necessary.
Table 1: Correlation matrix
Tax rate Size Age Profit Crisis Develop
4. Data
The data of the companies that is used in this research is sourced from COMPUSTAT. Financial firms with SIC codes between 6000 and 6999 are excluded from the sample. Financial firms are deleted because they do not have the same capital structure as non-financial companies. A panel data set is used for testing the hypotheses. Panel data is a combination of cross-sectional data and time-series data. Cross sectional data is needed to test for the firm-specific factors and time series data is needed to test for the crisis effect. The sample covers the period from 2004 till 2015. In this way the impact of the financial crisis can be measured: 1) four years before the crisis, 2) four years during the crisis and 3) four years after the financial crisis. This research is adding to previous research, because it investigates the corporate financing choices of firms several years after the financial crisis. There has been little research about the influence of the financial crisis on the corporate financing choices and about what happened the years after the financial crisis
Firms with missing data about the leverage level are deleted from the sample, because this is the dependent variable and without this data the statistical tests cannot be done. Furthermore, firms that did not have all the twelve years of data have been deleted, because the effect of the factors will be measured over the years and firms with only a couple of years of data will give a wrong impression of the impact that the factors will have during the twelve years. By deleting these firms, the sample might have become biased, because firms that went bankrupt in the crisis are deleted and firms that were newly listed after the financial crisis are also deleted. As a result, there may be only strong and old firms left in the final sample.
When looking at the factor firm age there were firms deleted, because some firms were formed of mergers. Leaving merged firms in the sample would cause a distorted view of how old and/or advanced the firms really are. When firms demerged in the past, the founding date of the firm before the breakup is used to calculate the age of the firm because of how advanced the firms really are. Furthermore, firms with outliers of extremely high leverage ratios or negative leverage ratios are deleted, because these outliers are assumed to be measurement faults. Negative leverage ratio’s do not exist and extremely high leverage ratio affect the mean in a way that does not reflect reality. The final sample has data of 39 countries and consists of 580 firms over a twelve years’ period, thus 6,960 observations in total. The descriptive statistics of the final sample and some statistics about the dependent variable are given for each country in table 2.
Table 2: descriptive statistics sample and dependent variable
Country Number of firms Firm years Minimum leverage ratio Maximum leverage ratio mean leverage ratio Median leverage ratio Argentina 7 84 0,0816 1,1850 0,7928 0,7985 Australia 22 264 0,0430 2,1780 0,5912 0,5316 Belgium 6 72 0,0853 1,5577 0,6486 0,6785 Bermuda 10 120 0,0998 1,7131 0,7157 0,6311 Brazil 3 36 0,0415 1,0399 0,4881 0,5097 Switzerland 3 36 0,2276 1,3990 0,5365 0,3995 Chile 12 144 0,0104 1,9374 0,5733 0,5499 China 5 60 0,3717 1,2253 0,8312 0,9046 Germany 30 360 0,1159 1,5563 0,6109 0,6122 Denmark 2 24 0,2369 0,7571 0,4749 0,4858 Spain 6 72 0,1577 1,4586 0,6259 0,6333 Finland 2 24 0,0213 0,8006 0,4889 0,6334 France 13 156 0,2745 1,4370 0,7474 0,6949 United Kingdom 106 1272 0,0172 2,4615 0,7028 0,6857 Hong Kong 8 96 0,3435 1,0224 0,7243 0,7472 Indonesia 4 48 0,2384 1,4848 0,6003 0,4967 India 3 36 0,1990 1,1846 0,5817 0,4518 Ireland 8 96 0,2698 1,3667 0,6280 0,5819 Israel 24 288 0,0584 2,9323 0,6919 0,5945 Italy 12 144 0,0447 1,8388 0,6460 0,6023 Jersey 3 36 0,3332 0,9049 0,6813 0,7457 Japan 174 2088 0,0329 3,4786 0,6791 0,6695 South Korea 1 12 0,3771 0,6589 0,5597 0,5725 Luxembourg 2 24 0,2915 0,7894 0,5294 0,5074 Mexico 19 228 0,1052 2,2168 0,7070 0,6203 Malaysia 18 216 0,0632 3,3640 0,7733 0,6980 Netherlands 18 216 0,0298 1,8735 0,7622 0,7172 Norway 4 48 0,2158 1,0856 0,5763 0,6427 New Zealand 5 60 0,1295 1,3459 0,6727 0,7544 Peru 1 12 0,5978 0,8277 0,7149 0,7347 Philippines 4 48 0,3173 2,7020 0,8898 0,7829 Portugal 2 24 0,4461 2,7260 1,1187 0,8350 Russia 1 12 0,5533 0,7584 0,6323 0,6130 Singapore 16 192 0,0817 2,3641 0,6099 0,5973 Sweden 15 180 0,0825 1,9981 0,6896 0,6279 Taiwan 2 24 0,0586 0,9840 0,5982 0,7459 Venezuela 1 12 0,7365 1,0354 0,9189 0,9193
British Virgin Islands 1 12 0,5170 0,9229 0,6655 0,6704
South Africa 7 84 0,1669 1,0112 0,6118 0,6365
The median of the short-term and long-term debt ratio are given for each country in Graph 1, which is stated below. The median leverage ratio is a good measure for the leverage ratio of the sample. In this research, the maturity of debt is used as a robustness test. Debt can exist out of short-term debt that has to be paid back in one year and long-term debt that has to be paid back over a longer period than one year. In this respect, short-term debt is more risky than long-term debt. Debt maturity refers to the point in time the principal is due to be paid. Thus, using more short-term debt means that there is more risk of going bankrupt and using more long-term debt means that there is less risk of going bankrupt.
The total of term and long-term leverage ratio is relatively constant, but the short-term and long-short-term leverage ratios are not constant as can be seen in Graph 1. Venezuela has the lowest median short-term and long-term leverage ratio. Looking at the Static Trade-off Theory this would mean the lowest risk of going bankrupt, but also gaining little from tax benefits. However, there is only one company in the sample from Venezuela, therefore this cannot be generalized to the whole country. Jersey has the highest total of short-term leverage and long-term leverage ratio, which is mainly because the short-term leverage ratio is very high. This means high bankruptcy risk and higher gains from tax benefits. However, there are only three companies from Jersey, therefore this also cannot be generalized to the whole country. The countries with a lot of companies in the sample are more in the middle of the graph.
Graph 1: Median short term and median long-term leverage ratios for each country
0 0,2 0,4 0,6 0,8 1 1,2 1,4
Looking at the average leverage ratio for all the firms in the sample, see Graph 2, it can be seen that the leverage ratio was increasing in the years before the financial crisis, but dropped significantly after the financial crisis. This is exactly what we expected to happen with the leverage level through the years. It can also be seen that the leverage level stays stable over the years after the financial crisis.
Graph 2: Mean leverage ratio through the years
Long-term debt is less risk full as short-term debt as we mentioned before. When looking at the long-term debt to total debt ratio over the years, see Graph 3, it can be seen that the long-term debt is increasing through the years. This indicates that the debt that is used is becoming less risky.
Graph 3: Mean long term debt to total debt through the years
0,62 0,64 0,66 0,68 0,7 0,72 0,74 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
leverage ratio
0,23 0,24 0,25 0,26 0,27 0,28 0,29 0,3 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
5. Results
Below the results of the different models can be found in Table 3 till Table 5. Furthermore, the results for each model are discussed. Besides, the result of the robustness test is given.
5.1 Results
The results of the random effects model for model 1 are shown below in Table 3. Model 1 tests for theory of Modigliani and Miller (1958), which states that the tax rate has a positive relation with the leverage ratio. The results are corrected for autocorrelation by adding a trend.
Table 3: Random effects model for Model 1 corrected for autocorrelation
Leverage ratio Short term leverage ratio Long term leverage ratio
Coefficient Std.
error Coefficient Std. error Coefficient Std. error
Tax rate -1.34E06*** 2.79E-07 -7.63E-07 4.28E-06 -4.83E-08 3.79E-07
F statistic 12.74305 0.016525 0.224739
Adjusted R2 0.003381 0.000005 -0.000224
Significance levels: *** :1% **: 5% *:10% Number of observations: 6,960
The results of the fixed effects model for model 2 are shown below in Table 4. Model 2 tests for all the factors of the firms that can have an influence with the leverage ratio. These results are also corrected for autocorrelation by adding a trend.
Table 4: Fixed effects model for Model 2 corrected for autocorrelation
Leverage ratio Short term leverage ratio
Coefficient Std. error Coefficient Std. error
Tax rate -1.36E-06*** 2.78E-07 1.19E-06 4.60E-06
Size 8.20E-07*** 2.93E-07 2.05E-05*** 4.85E-06
Age 0.005885 0.006326 0.007375 0.104680
Profit -2.88E-06*** 7.80E-07 -5.14E-06 1.29E-05
F statistic 46.73830 1.206748
Adjusted R2 0.796906 0.017428
Significance levels: *** :1% **: 5% *:10% Number of observations: 6,960
The results of the pooled ordinary least squares model for model 3 are shown in Table 5. Model 3 tests for all the factors, just as in model 2. In addition, Model 3 will test for the influence of the financial crisis and for the influence presented by the level of development. These results are also corrected for autocorrelation by adding a trend.
Table 5: Pooled ordinary least squares model for Model 1 corrected for autocorrelation
Leverage ratio Short term leverage ratio Long term leverage ratio
Coefficient Std. error
Coefficient Std. error Coefficient Std. error
Tax rate -1.34E-06*** 2.79E-07 -1.81E-06 4.30E-06 -2.14E-08 3.80E-07 Size 7.92E-07*** 2.63E-07 2.82E-06 1.80E-06 1.22E-07 3.21E-07
Age -6.92E-06 0.000262 0.000231 0.000685 9.20E-05 0.000229
Profit -2.16E-06*** 7.49E-07 1.55E-06 6.34E-06 1.63E-06 9.69E-07 Crisis 0.019157*** 0.003894 -0.060190 0.064060 0.016265*** 0.005312 Tax rate * Crisis -2.93E-05 1.96E-05 -8.16E-05 0.000280 1.93E-06 2.65E-05 Development 8.23E-07 8.47E-07 -2.85E-06 2.22E-06 -2.01E-07 7.40E-07
F statistic 8.357797 1.410384 1.949056
Adjusted R2 0.009575 0.001629 0.001095
Significance levels: *** :1% **: 5% *:10% Number of observations: 6,960
Model 3 shows the relationship between all the factors that have a direct influence on the corporate financing decisions of companies, just as model 2, but in model 3 the influence of the crisis and the level of development are also tested. As stated in the methodology, the fixed effects model yields an error in Eviews. The Hausman test is rejected at a 5 percent significant level for the leverage ratio and the short-term leverage ratio, but is not rejected for the long-term leverage ratio. However, the results can only be compared when using the same model, therefore the pooled ordinary least squares model is used. The tax rate is also in model 3 significantly negative. The size is significantly positive and the profit is significantly negative, just as was expected and which is the same outcome as in model 2. Also, as in model 2, the age is not significant. The crisis dummy is significantly positive, again as expected. However, the moderating effect of the crisis is not significant. Furthermore, the level of development is not significant. Considering the relationship between the factors and the short-term leverage ratio, none of the outcomes are significant and considering the relationship between the factors and the long-term leverage ratio, only the crisis is significantly positive. This positive relation between the long-term leverage ratio and the crisis is what was expected by the theory. The effect of the debt maturity remains unclear in Model 3, because the results for the short-term leverage ratio are not significant and can therefore not be compared to the long-term leverage ratio.
5.2 Robustness tests
6. Conclusion
Here the conclusions of this research are given by answering the main question and examining the theories and the outcomes of the hypotheses. Too, any limitations of this research are disclosed with recommendations for further research.
6.1 Conclusions
Corporate financing choices can be determined by different factors that are related to firm and country specific characteristics. Since the paper of Modigliani and Miller (1958), there has been a lot of research about corporate financing choices. Studies such as Rajan and Zingales (1995), Wald (1999), and De Jong, Kabir and Nguyen (2008) examined the corporate financing choices of firms. However, these papers are inconclusive concerning which firm and country specific factors influence the financing choices of companies. Furthermore, recent papers of Graham, Leary and Roberts (2015) and Turk Ariss (2016) suggest that further research about corporate financing choices is still needed. In this research the firm and country specific factors that could have an influence on the corporate financing choices of firms are examined. Therefore, the main question developed for this research was:
“Which firm and country specific factors can explain the capital structure choice of companies?”
In the literature review different theories that could explain this question were discussed and these theories were used to develop hypotheses. The hypotheses that were developed tested if the factors: tax rate, size, age, profit, crisis and level of development had a relationship with the leverage ratio. Unique to this research is that it tests if the global financial crisis of 2008 has influenced the leverage ratio of firms. Furthermore, the debt maturity was tested in this research. The developed hypotheses were therefore split into three hypotheses: 1) for the leverage ratio, 2) for the short-term leverage ratio and 3) for the long-term leverage ratio. This testing for debt maturity was used as a robustness test.
Thus, the Static Trade-off Theory made it clear that holding more debt has besides the tax benefits also costs, for example costs of financial distress which is often stated as bankruptcy costs. As explained in the literature review, the paper of Gruber and Warner (1977) states that the costs of holding an extra amount debt, for example bankruptcy costs, are lower for bigger firms. Therefore, there was expected to be a positive relationship between the size of the firm and the leverage ratio. The outcome of the statistical tests in Model 2 and in Model 3 are that there is indeed a significant positive relationship between the size of the firm and the leverage ratio. The second hypothesis is therefore confirmed.
Besides the Static Trade-off Theory, Myers and Majluf (1984) develop the so-called Pecking Order Theory which is explained in the literature review. The conclusion of the pecking order theory is that firms finance new project first with retained earnings and when outside investment is needed debt financing is preferred over equity financing. When taking the Pecking Order Theory into consideration, it is expected that profit has a negative relationship with leverage ratio. When profit is higher, there are more retained earnings for financing new projects and therefore less debt is needed. Model 2 and Model 3 give both a significant negative relationship between profit and leverage ratio, just as expected. The third hypothesis is therefore also confirmed.
Furthermore, it was expected that the firms switch from the Pecking Order Theory in the short run to the Static Trade-off Theory in the long run and therefore age would be negatively related to the leverage ratio. However, in this research age is not significant and it is not clear what effect age has on the leverage level. The insignificance could be created because a sample period of 12 years is used. The sample is balanced, thus the firms that went bankrupt in this long period of time or the new firms where deleted. Therefore, the final sample consists of mainly older firms.
However, taking the significant factors into consideration, the Static Trade-off Theory should be used regarding the choice of firms for the capital structure instead of the theory of Modigliani and Miller. Furthermore, not only the Static Trade-off Theory can explain the choice of firms for the capital structure, also the Pecking Order Theory can explain this choice. These theories are not mutually exclusive.
For companies these outcomes mean that they should apply the Static Trade-off Theory for their company to find an optimal balance between the costs and benefits of taxes. However, when the company becomes more profitable, the company should rely less on debt.
The level of development is not significant, therefore we cannot conclude that there is a significant difference between developed and developing countries with respect to the leverage ratio.
Taking the debt maturity into consideration and looking at the short-term and long-term leverage ratios we cannot give clear conclusions, because the tests did not give significant results. Therefore, debt maturity cannot be explained in this research and future research should explain what the effect of debt maturity is.
6.2 Limitations and further research
In this research, debt maturity is measured by examining the difference between the short-term leverage ratio and the long-short-term leverage ratio suggested by the research of Serrasqueiro and Caetano (2015). However, this yields insignificant outcomes, which could be caused by the way it is measured. In future research it would be better to add the debt maturity as a factor to the econometric model.
Besides the debt maturity, the age was also not significant. This insignificance could be created because a sample period of 12 years is used. The sample is balanced, thus the firms that went bankrupt in this long period of time or the new firms were deleted. Therefore, the final sample consists of mainly older firms. This period of 12 years was needed to find the effect of the financial crisis, however when using a smaller sample period, the age could be significantly related. Further research about the influence of age on capital structure without the effect of the global financial crisis of 2008 should give a clearer outcome. However, insignificance can also mean that the factors tested really do not have a significant effect on the dependent variable.
References
Booth, L., Aivazian, V., Demirguc-Kunt, A., & Maksimovic, V. (2001). Capital structures in developing countries. The Journal of Finance, 56(1), 87-130.
Cotei, C., & Farhat, J. B. (2009). The trade-off theory and the pecking order theory: Are they mutually exclusive?
De Jong, A., Kabir, R., & Nguyen, T. T. (2008). Capital structure around the world: The roles of firm-and country-specific determinants. Journal of Banking & Finance, 32(9), 1954-1969.
Demirgüc-Kunt, A., Martinez Peria, M., & Tressel, T. (2015). The impact of the global financial crisis on firms' capital structure.
Faccio, M., & Xu, J. (2015). Taxes and capital structure. Journal of Financial and
Quantitative Analysis, 50(3), 277-300.
Fan, J. P., Titman, S., & Twite, G. (2012). An international comparison of capital structure and debt maturity choices. Journal of Financial and Quantitative Analysis, 47(1), 23-56.
Graham, J. R., Leary, M. T., & Roberts, M. R. (2015). A century of capital structure: The leveraging of corporate america. Journal of Financial Economics, 118(3), 658-683.
Gruber, M. J., & Warner, J. B. (1977). Bankruptcy costs: Some evidence. The Journal of
Finance, 32(2), 337-347.
Hovakimian, A., Opler, T., & Titman, S. (2001). The debt-equity choice. The Journal of
Financial and Quantitative Analysis, 36(1), 1-24. 10.2307/2676195 Retrieved from http://www.jstor.org/stable/2676195
Kanatani, M., & Yaghoubi, R. (2017). Determinants of corporate capital structure in new zealand. Amr, , 1.
Kraus, A., & Litzenberger, R. H. (1973). A state-preference model of optimal financial leverage. The Journal of Finance, 28(4), 911-922. 10.2307/2978343 Retrieved from
http://www.jstor.org/stable/2978343
Miller, M. H. (1977). Debt and taxes. The Journal of Finance, 32(2), 261-275.
Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance and the theory of investment. The American Economic Review, 48(3), 261-297.
Mouton, M., & Smith, N. (2016). Company determinants of capital structure on the JSE ltd and the influence of the 2008 financial crisis. Journal of Economic and Financial
Sciences, 9(3), 789-806.
Myers, S. C., & Majluf, N. S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics,
13(2), 187-221.
Pfaffermayr, M., Stöckl, M., & Winner, H. (2013). Capital structure, corporate taxation and firm age. Fiscal Studies, 34(1), 109-135.
Reinhart, C. M., & Rogoff, K. S. (2009). This time is different: Eight centuries of financial
folly princeton university press.
Serrasqueiro, Z., & Caetano, A. (2015). Trade-off theory versus pecking order theory: Capital structure decisions in a peripheral region of portugal. Journal of Business Economics and
Management, 16(2), 445-466.
Shyam-Sunder, L., & Myers, S. C. (1999). Testing static tradeoff against pecking order models of capital structure1. Journal of Financial Economics, 51(2), 219-244.
Sogorb-Mira, F., & Lopez-Gracia, J. (2003). Pecking order versus trade-off: An empirical approach to the small and medium enterprise capital structure.
Titman, S., & Wessels, R. (1988). The determinants of capital structure choice. The Journal
of Finance, 43(1), 1-19.
Turk Ariss, R. (2016). Legal systems, capital structure, and debt maturity in developing countries. Corporate Governance: An International Review, 24(2), 130-144.
Vermaelen, T. (1981). Common stock repurchases and market signalling: An empirical study.
Journal of Financial Economics, 9(2), 139-183.
Wald, J. K. (1999). How firm characteristics affect capital structure: An international comparison. Journal of Financial Research, 22(2), 161-187.
