Capital Structure Decisions in Developing Economies

Transcription

1 Capital Structure Decisions in Developing Economies Master Thesis By Floris P.P. Loermans ANR: Tilburg University Faculty of Economics and Business Administration Department of Finance Supervisor: Dr. M.R.R. Van Bremen

2 Table of Contents Abstract... 3 Section I: Introduction... 4 Section II: The Static trade off theory... 6 Section III: The Agency Problem... 7 Section IV: Asymmetric Information Section V: The difference between a country with a developed and one with a developing economy Section VI: Hypotheses Methodology Section VII: Data Collection Section VIII: The Model Section IX: Data Analysis Discussion Section X: Conclusion References Appendix Part A Part B Part C Part D

3 Abstract This paper investigates the determinants of capital structure decisions in developing countries by analysing recent financial data, of the biggest, non-financial companies, listed on the stock exchanges of nine developing countries. It was found that the same factors, identified in previous studies, to have been important in developed countries, are affecting companies in developing countries with different magnitudes. Factors related to the agency problem have a smaller impact in developing countries, consequently the impact of factors related to the problem of asymmetric information is larger. 3

4 Section I: Introduction The Modigliani Miller theorem states that in an efficient market, where there are no taxes, bankruptcy costs, and asymmetric information, the capital structure of the firm does not influence its value. When, in the real world, these values do have to be taken into account, capital structure decisions become much more complicated. Since the time of the Modigliani Miller paper, a lot of research has been conducted on how companies make these capital structure decisions. The theories derived from research fall into three distinct categories. One category is based on the agency problem, the second on asymmetric information and the third on taxes. From these different concepts, theories can be derived which should have an effect on the capital structure of a firm. These categories will be dealt with more extensively later in this paper. What is known about Capital Structure choices made by firms in the real world is largely based on research done in countries with well developed financial markets. Rajan and Zingales (1995), Bradley, Jarrell and Kim (1984), and Titman and Wessels (1988) are just a few examples of the papers written on this subject. There are many differences between developed and developing countries. The most profound ones are of an economic nature. Developing countries are poorer and most of the wealth is controlled by a small minority. Reinforced by financial law this leads to a much more concentrated ownership structure of major firms than is usual in the developed countries. This in turn has an effect on a firm s capital structure decisions. There are also conflicting views on whether capital structure decisions are more country or more firm specific. Booth, Aivazian, Demirguc-Kunt and Maksimovic (2001) find that: Knowing the country of origin is usually at least as important as knowing the size of the independent variables. While Psillaki and Daskalakis (2009) find that the differences are due to firm specific factors. However there is much less research on capital structure decisions made by companies in developing countries and the research that has been done has yielded conflicting results. Mayer (1990) finds that financial decisions in developing countries are somehow different from those in developed countries. Booth et al. (2001) on the other hand, provide evidence that their decisions are affected by the same variables as in developed countries. These papers, however, have used rather old data. Although the paper by Booth et al. was written in 2001, their findings are based on data from the 4

5 developing countries between 1980 and The world economy of the 80 s looks completely different from the globalized world economy of today, and some of the developing countries back then are not the upcoming economies of our time anymore. For example, South-Korea, one of the countries discussed by Booth et al., is now considered to have a developed financial system and currently is the 15 th biggest economy in the world (IMF 2009). Zimbabwe, one of the other countries Booth et al. (2001) investigated, went in the complete opposite direction. Relying on very recent data, this paper will investigate whether the same factors believed to have an influence on capital structure decisions made by companies operating in developed economies with modern financial systems, are also influencing the capital structure decision of firms operating in countries with developing economies and financial systems. The main research question for this paper therefore will be: Are the determinants of capital structure decisions taken by companies in developed countries the same as the determinants of these decisions in developing countries? The sub questions that will help in answering this main question are: 1 What are the effects of the static trade off theory on capital structure decisions made by firms? 2 What are the effects of the agency problem on capital structure decisions made by firms? 3 What are the effects of asymmetric information on capital structure decisions made by firms? 4 What are the differences between developed and developing countries that could affect the capital structure decisions made by firms? These sub questions will be answered in sections II, III, IV, and V respectively. Section VI will state the hypotheses to be tested in the empirical part of this investigation. They have come forth out of the theory dealt with in the preceding sections. The methodology used will also be discussed in the same section. Section VII will deal with the procedure used for the data collection and in section VIII the model to be used will be discussed. In Section IX this model will be analyzed and the results will be compared to the hypotheses and the theory. Finally section X will provide the conclusions. 5

6 Section II: The Static trade off theory The Modigliani Miller theorem states that in an efficient market, where there are no taxes, bankruptcy costs and asymmetric information, the capital structure of the firm does not influence its value. In the Modigliani Miller case, the capital structure of a firm and the capital structure decisions of the firm s management are irrelevant. In the real world these factors do play a role. This section will show how tax might affect the capital structure decisions of a firm s management. Without tax, the value of the unlevered firm (V U ) is equal to the value of the levered firm (V L ): V U = V L (Modigliani and Miller 1957) When corporate taxes are introduced the value of the unlevered firm is not the same anymore as the value of the levered firm. That is because the interest payments on the debt are tax deductible. This is the tax shield. The value of the tax shield is T C X i B, where T C is the Corporate Tax, i the interest rate and B the amount of debt. When this is discounted by the interest rate the present value of the tax shield is obtained (T C B). To calculate the value of the levered firm the present value of the tax shield has to be added to the value of the unlevered firm: V L = V U + T C B (Modigliani and Miller 1963) The tax shield provides an incentive for firms to take on more debt. However there is a down side for firms taking on additional debt too: The cost associated with the possibility of financial distress. The static trade off theory suggests that firms will borrow up to the point when the advantages of debt due to the tax shield are offset by the disadvantages (Myers 2001). 6

7 Section III: The Agency Problem Agency theory in its simplest form discusses the relationship between two people: a principle, and an agent. The agent is hired by the principle to make decisions for the principle. Because of the nature of the relationship, conflicts of interest can arise between the principle and the agent. Jensen and Meckling (1976) identify two of these conflicts of interest. One between managers of the firm and the firm s shareholders and the second conflict of interest arises between shareholders and debt holders. This section will examine the effects of these agency problems on capital structure decisions. Agency problems between the managers of the firm and its shareholders The agency problem between managers and shareholders exists because the people who are usually in charge of a firm, the firm s managers, control less than 100 percent of the firm s equity and therefore do not get the entire residual claim. The benefit of their profit enhancing activities will be divided equally and paid out mostly to external shareholders. The managers do however have the power to divert part of the company s profits for their personal consumption. This personal consumption can take many forms, e.g. the usage of corporate jets or company paid parties. Everything they are able to divert does not have to be shared with the rest of the shareholders. They receive 100 percent of the value of the resources diverted and it only costs them a fraction as they own less than 100 percent of the firm s equity. These extra benefits are paid for by all the shareholders (Harris and Raviv 1991). To counter the problem mentioned above the firm could issue more debt. Debt forces the managers to pay out a specified amount, if they do not the company goes into bankruptcy. Due to the debt, the pool to extract personal benefits from is smaller. Thus the more debt, the more is left for the firm to pay out. There also can be disagreement between the managers and the investors (in this case the debt holders) of the firm over operating decisions. Managers are more inclined to keep the firm operating even though liquidation is preferred by investors (Harris and Raviv 1991). If the firm stops operating they stop receiving salary, and the other benefits discussed above. Debt mitigates these problems by giving the debt holders the option to force liquidation if the firm s cash flows are poor. Capital structure is determined by trading off the benefits of debt against the costs of debt (Harris and Raviv 1991). The cost of debt here is related to the information that needs to be gathered for the 7

8 liquidation decision. Firms with a higher liquidation value meaning relatively more tangible assets, and low investigation costs will be able to take on more debt (Harris and Raviv 1991). In addition tangible assets can be used as collateral, reducing the interest charged by debt holders. According to Stulz (1990), managers value investment of excess funds over paying out cash to shareholders because their perquisites increase with the investment, even if the NPV of these projects is negative. Shareholders are not able to assess these investment decisions made by managers. As a result there is a possibility that inefficient investment of funds occurs when there is a relatively large amount of free cash flow. When there is not enough free cash flow to fund the positive NPV projects the firm has, shareholders are also reluctant to provide more funds since shareholders are not able to assess whether the projects proposed by management are really positive NPV, or projects which management just wants to take on because it is profitable for the managers themselves. At this point management would be engaging in empire building (Kanniainen 1998), and (Hope and Thomas 2007) The solution proposed by Stulz (1990) is debt. Debt forces management to pay out a certain amount of the cash flow, thereby reducing the funds available for investment. This could be both advantageous and disadvantageous for shareholders. The optimal capital structure according to Stulz (1990), is determined by trading off the benefits of debt, in this case to prevent cash flowing into value decreasing projects, and the cost of debt, when debt could prevent the firm from investing in value increasing projects. Therefore firms, where the shareholders believe it has ample good investment opportunities, could be expected to have relatively low levels of debt compared to firms for which shareholders believe to have less favourable value increasing investment opportunities. A high market to book ratio suggests that shareholders expect the firm to grow in the future and thus have enough good investment opportunities. Market to book value will be used as a proxy for investment opportunities. Agency problems between share- and bondholders. Because of limited liability, the debt holders of the firm bear the consequences when an investment goes bad, but its shareholders get all the benefits when this risky investment does pay off. Therefore the shareholders who control the firm are willing to invest in high risk projects. When these projects pay off, those who control the residual claim get all the benefits. However, if a project does not pay off, the debt holders pay the price. Because of this bondholders are reluctant to invest in companies that take on risky projects, and if they do, they demand higher pay-offs, which make the loans more 8

9 expensive. So the higher the business risk taken on by a firm the less debt it will be expected to have. A good proxy for business risk is the variability of the returns on assets; with a higher variability, there is a higher probability that the firm is not able to pay its debt obligations and go into bankruptcy. Since these costs of financial distress are caused by threatened or actual default, safe firms, firms that do not take on excessive risks, are able to borrow more, before the costs of financial distress become too large and offset the advantages of debt (Myers 1984). Fortunately most firms have other factors besides high risks and high payoffs for their shareholders to consider when they choose their projects. The firm s reputation for example can play a significant role. The older more established firms usually have a good reputation of repaying their debts. Therefore they are offered lower interest rates. They are more inclined to choose less risky projects, because then they will have a better chance of repaying their debts so that they can keep enjoying relatively low interest rates. The relatively younger firms did not have time to have built up such a good reputation yet, and have to borrow at a higher interest rate. These younger firms are more inclined to choose the more risky projects. Bigger firms usually are also more diversified then smaller firms, thereby decreasing the risk of financial distress. Because of this it is expected that older, bigger, well established firms, have more debt than younger, smaller and less well established firms. 9

10 Section IV: Asymmetric Information This section will examine the effects of asymmetric information on capital structure decisions. When two parties engage in a business agreement or transaction, it is often the case that one of the parties has more or better information, about the transaction than the other party. Thus, one of the parties has more power than the other party. This imbalance in power could cause problems. In business firms, managers or insiders are assumed to have better and often private information about the firm s flow of returns and investment opportunities. However, they need financing from outsiders who do not possess this insider information. Potential lenders would benefit greatly from this high quality information. But because of moral hazard the flow of information between borrowers and lenders is not always trustworthy. Hence lenders are unable to recognize the good from the bad firms and projects. The market borrowing rate will reflect the average quality of the firms and projects, and therefore will only attract borrowers from the lower half (for them the market rate is underpriced). Because of the relative poorer quality of the projects, some of them will fail, and the lenders will lose all or part of their investment. The market for borrowing/lending money will not attract the relatively good firms since the market rate is too expensive for them. The money market will cease to exist since lenders lose a lot of money to the poor firms that go into bankruptcy. For the good firms and projects to receive financing though, extra information needs to be transferred in a way which can be trusted by potential lenders. One way to do this is for the insiders to invest in their own firms and projects. So the firm s capital structure can be used to signal to outside investors the information insiders possess (Leland and Pyle 1977), (Ross 1977). Capital structure could also function as a way of mitigating inefficiencies in the firm s investment decisions that are caused by information asymmetries (Myers 1984) and (Myers and Majluf 1984). If outsiders are less well informed about the value of the firm s assets than insiders, the firm s equity might be mispriced. If the firm is required to fund new projects by issuing equity and the equity is underpriced, new investors could get an extra high payoff at the expense of existing shareholders. In such a situation the firm s existing shareholders are much better off if the project can be financed by internal funds or debt, which are not subjected to undervaluation (Myers and Majluf 1984). Since debt is costly compared to internal funds, internal financing is preferred over debt and equity (external financing). 10

11 If external financing is required because a firm does not have enough internal cash flow, debt is preferred over equity. Debt is safer to issue than equity. A firm will issue equity at a time when the market price for its equity is favourable. If the firm issues equity at any other less favourable point in time it will harm the existing shareholders in favour of the new ones. Because of this, if a firm does issue equity, it signals to the market that management thinks the share price is relatively high and consequently the share price drops (Vithessonthi 2008). Since internal funds are preferred over debt and debt is preferred over equity, this theory suggests that the capital structure will be driven by the firms desires to finance new investments. This has been referred to by Myers (1984) as the pecking order theory of financing. In the part discussing agency problems between the managers of the firm and the firm s shareholders it was argued that asset tangibility and debt should have a positive correlation. However, when considering the theory based on asymmetric information, the opposite applies. Harris and Raviv (1991) suggest that firms with relatively more intangible assets will be more subjected to information asymmetries than firms with relatively a lot of tangible assets. The real worth of these intangible assets is much easier to assess by insiders than by outsiders. If more funds are required to finance one of the firm s new projects debt is preferred over equity (as has been discussed above). Since firms with fewer tangible assets are more susceptible to this information asymmetry problem, they will accumulate more debt over time. 11

12 Section V: The difference between a country with a developed and one with a developing economy If capital structure decisions of companies in developing countries are to be studied, a clear definition is needed of what makes a country either developed or developing. How to assess this is not without controversy. There are many different approaches whereby each attempts to define what makes a country s economy developed or not. Economic criteria are both easily measurable and the most objective and are being used by organizations like the IMF and the UN. Those will also be used for this study. The data collection part of this paper will deal more extensively with the selection of the countries needed for this study and which countries are developed or developing. This part will focus on the differences in capital structures between developed and developing countries. In developed countries, and especially in the US, firms are owned by a large set of dispersed shareholders. In contrast, in many developing countries firms are often held in groups of complicated ownership structures often referred to as pyramids (Bertrand and Mullainathan 2003). In Asia for example two-thirds of the public companies have a controlling shareholder (Gilson 2007). What are the reasons for this difference in ownership structure? The biggest and most obvious difference between developed and developing countries is the amount, and the distribution of wealth. Developing countries have a lower GDP s per capita, and a higher Gini-coefficient. Only a small wealthy part of the population is able to invest their wealth on the financial markets which makes the ownership structures of the listed companies more concentrated. This unequal basis is reinforced by the differences in financial law. In developing countries financial law is not as good and efficient in protecting the rights of investors. In those countries it is more difficult for a large group of small shareholders to control the firm s management. A single shareholder with a controlling stake can mitigate this problem. Not only can big shareholders better control management, if there is no effective law in place to protect minority shareholders, controlling shareholders could benefit at the expense of minority shareholders. La Porta et al (1998) finds that the more effective the law is at protecting the rights of individual and minority investors, the less concentrated corporate ownership tends to be. Developing countries tend to have poor minority protection and poor commercial law in general (Gilson(2007). 12

13 Thereby reinforcing the concentrated ownership positions developing countries already have due to the inequality of wealth distribution. Countries with a more equal distribution of wealth and good investor protection laws tend to have companies with more dispersed ownership. Agency problems primarily originate with the managers who control the firm. Companies controlled by large shareholders or the pyramid ownership structures, mentioned by Bertrand and Mullainathan (2003), have agency problems of their own, however they are of a different nature. In studying the capital structure decisions of firms in developing countries, it can be expected that variables related to agency theory have a smaller affect on the decisions made by firms in developing countries. Therefore other theories like the Static Trade-off theory and the problem of asymmetric information should play a bigger role. This will be tested in the following sections. 13

14 Section VI: Hypotheses The theories referred to in previous sections affect the capital structure decisions in different ways. Because of the difference in corporate ownership structure in developing countries, it can be expected that factors related with agency theory are of limited significance. As a consequence, variables related to the problem of asymmetric information and the static trade-off model should be the determinants of capital structure decisions made by firms in developing countries. The main hypothesis for this paper is: Factors related to the problem of asymmetric information and the static trade-off theory, are the main factors that determine the capital structure decisions made by firms in developing countries. To be able to test this hypothesis, the each individual factor has to be tested. How these factors affect capital structure decisions is listed below. These sub hypotheses have come forth out of the theory discussed in previous sections. 1. Debt and income tax are positively correlated. 2. Debt and EBIT are positively correlated. 3. Debt and Assets Tangibility are positively correlated. 4. Debt and the market to book value are negatively correlated. 5. Debt and business risk are negatively correlated. 6. Debt and size are positively correlated. 7. Debt and profitability are negatively correlated. 8. Debt > Equity 9. Debt and assets tangibility are negatively correlated. For an exact definition of the individual factors see section XIII. For a further explanation of how the sub hypotheses came to be, see part A of the Appendix. Methodology To be able to say anything about the main hypothesis, the nine sub hypotheses, listed above and explained in more detail in Part A of the Appendix, should be tested individually. To test all but the 14

15 eighth sub hypothesis, it has to be shown that the variables influence the total debt ratio. This can be done using linear regression. The different countries in the sample use different standards of accounting, therefore they cannot be compared directly. If a separate model is created for each country, then the sub hypotheses can be tested on all countries separately. First it has to be shown that the variables are independent of each other. If multicollinearity is a problem variables have to be taken out of the equation. Next tests have to be performed to check for normality, linearity and independence. Finally, an analysis can be done to establish which variables significantly influence the debt ratio. To be able to test the eighth sub hypothesis, it has to be established that there is a significant difference between the amounts of outstanding debt of the companies in the sample, and the amount of outstanding equity. This can be done by checking the confidence interval of the difference between them. If this confidence interval is completely above zero, there is evidence in support of the second part of the pecking order theory. 15

16 Section VII: Data Collection This section will cover the data collection. First of all countries with upcoming economies have to be selected. Then companies need to be selected from whose balance sheets the data can be obtained to test the hypotheses in the previous section. Selection of Countries: The purpose of this paper is to investigate whether the determinants of capital structure decisions are similar across countries with developed and developing economies. But what is the difference between the two? There are many opinions of what the criteria should be as to which country can be considered to be a developed nation. Economic criteria, like GDP per capita and level of industrialization are dominating this discussion, since they are both objective and can be measured easily. Regardless of which criteria are used, the consensus is that the developed countries across the world include: the USA, Canada, Western Europe, Australia, New-Zealand and Japan. A recent addition to this selective group includes the four Asian Tigers: South Korea, Singapore, Hong Kong and Taiwan. (IMF) All the other countries of the world are considered developing countries. The IMF divides the developing countries into two groups. The countries with the least amount of development have economies and financial systems that are too small to be considered for this study. This group among others includes countries like Afghanistan, Myanmar, Cambodia and various countries in Africa. This paper has to rely on sound data to be successful. The financial systems of these countries are so underdeveloped, and the number of companies that limited, that there simply is not enough data to study. Therefore all the countries used in this investigation belong to the group of developing countries with relatively more development. The Datastream database will be used later in this paper to obtain the necessary data. The following countries are included in the sample: Brazil, Russia, India, China, Indonesia, South- Africa, Poland, Thailand and the Philippines. These countries cover a wide spectrum of legal and economic backgrounds, as well as covering different geographic regions. Asia could be seen as being overrepresented in this sample, but it is justified since over 60 percent of the world s population lives in Asia. One of the major reasons for these countries to be considered was the amount and quality of the data available. Another reason is that they are part of selective groups of countries considered to 16

17 have major economic potential. Brazil, Russia, India and China form the so-called BRIC block. This term was first used by Goldman Sachs, and it is predicted that these four countries will eclipse the combined economies of the richest countries in the world today by The other countries in the sample are also part of different blocks identifying them as countries with high potential, these groups like the BEM economies (Big Emerging Markets) are less well known than the BRIC group. In all countries the stock markets have been hit hard by the financial crisis. The capitalisation of almost all of the markets went down by roughly 50%. India being hit the hardest: its stock market capitalization went down by 65%. The one exception is Russia, in 2008 its stock market capitalization only went down with 15%. In 2008 the real economy was not affected as hard by the crisis as financial markets, and all the countries except for Thailand have a higher real GDP growth rate than the world average of 3.1%. With China having the highest growth rate of them all at 9.6% The selected countries come from a variety of different legal origins. These legal origins come from a paper written by La Porta, Lopez-de-Silanes, Shleifer and Vishny in La Porta et al. (1995) examines the legal rules covering the protection of corporate shareholders and creditors, the origin of these rules and the quality of enforcement. Three of the four different origins identified by La Porta et al. (1995) are represented by the countries selected for this paper: The English (India, Thailand and South-Africa), French (Brazil, Indonesia and The Philippines) and German (possibly China) legal origins. Unfortunately Russia and Poland are not included in this report. China being the only country with the German legal origin is only indirectly identified. La Porta et al. identifies Taiwan as being of German legal origin, China is said to have based its financial laws on the Taiwanese legal system. Their findings show that common law countries, English origin, have the strongest rules for shareholder protection, while French origin countries have the weakest. The German origin is somewhere in the middle. La Porta et al. (1995) argues that investor protection is important since the protection investors receive determines the readiness to finance firms and might help to explain the differences in financing of firms in different countries. This paper will not investigate whether this legal origin has any affect on firm s capital structures, since this paper tests if different theories will also hold using different countries as sources of data. It does not intend to compare the different countries. It plays a significant role in explaining the difference in capital structure across countries and because of this, it is important enough to be mentioned in this paper. However it will not be a source of bias, because whether for example the size of a firm has an effect on its debt level, (as proposed in hypothesis 6) is an effect that can be determined independently of the origin of the legal system the firm operates in. 17

18 Data The data used in this study was collected using the Datastream Database. First a list of companies to be studied needed to be drafted for each individual country. Two criteria were used to select these companies: The sector they operate in, and their size. The structure of financial firms is different from that of non-financial firms. Regulation may even directly influence capital structure. Also their liabilities are different from the liabilities of non-financial firms. Therefore all financial firms were excluded from the sample. The sectors that were used to compile the list of companies are mentioned in Table I. The second criterion is the size of the firm. The attention of this study is limited to the biggest listed firms. When there are enough firms available to be able to do a meaningful study, the largest firms are selected based on the market value of their equity on the 31 st of December This list was used to extract the different kinds of financial data needed, to formulate a regression equation needed for this study, from the abbreviated balance sheets and income statements listed in the Datastream Database. Table I This table shows the sectors used to search for the companies needed for this study in the Datastream Database. This list was used to extract the different kinds of financial data needed, to formulate the regression equation needed to study the factors influencing capital structure decisions in various countries. Using the method described above, 8752 individual observations, over the period , on the financial situation of companies, from nine developing countries, operating in the industries depicted in this table have been extracted from the Datastream Database. This data will be used in the following section to analyze capital structure decisions. The potential biases that could affect the results of this study are dealt with in part B of the appendix. When reviewing the results of this study, those potential biases should be kept in mind. However, the best studies on this topic in the past all had similar biases, so one could conclude that it is not possible to conduct this type of study without having these potential biases. 18

19 Section VIII: The Model This section will set up the empirical part of this report. First the model to be used will be introduced, and the variables will be explained. The variables will be tested for multicollinearity and the model itself will be tested for linearity, independence and normality. A multiple linear regression is used to study the factors influencing the capital structure of firms in different countries. Each individual country has its own regression. The dependent variable will be the firm s total debt ratio, which is defined as total liabilities divided by total assets. The independent variables come forth out of the sub hypotheses stated in part A of the appendix. 1. Income tax rate is obtained by dividing the amount of income tax paid, by the earnings before tax. 2. EBIT (by dividing Earnings Before Interest and Tax by total assets). 3. Asset tangibility, by total assets minus current assets divided by total assets. 4. The market to book ratio, by dividing the market value of equity by the book value of equity. 5. Business risk, by the standard deviation of the return on assets over the five year period being studied. 6. The firm s size, by taking the natural logarithm of its total sales or revenue divided by Return on Assets, as a proxy for profitability, by dividing the earnings before tax by total assets. As has been mentioned in the methodology part the eighth sub hypothesis will be tested separately. The ninth sub hypothesis uses the same independent variable as the third sub hypothesis: asset tangibility. The basic regression estimated will be: Total Debt Ratio (firm i) = α + β 1 Income Tax i + β 2 EBIT i + β 3 Asset Tangibility i + β 4 Market to Book Ratio i + β 5 Business Risk i + β 6 Size i + β 7 Profitability i + ε i To test the regression, the independent variables need to be checked for multicollinearity. For all countries the tolerance levels for the independent variables EBIT and Profitability are too high and EBIT has to be taken out of the equation. 19

20 Now the model has to be tested for linearity, independence and normality. This can be done by examining the residuals of the models. To test for linearity an F-test has to be done. See Table II for each models individual F statistic. With 95% confidence the null hypothesis of a standard F test, that there is no linear relationship between the dependent and independent variables, can be rejected for every single model. With 95% confidence it can be said that the model is linear. For a further explanation of this F test see the Appendix Part D I Next the independence needs to be checked. The Durbin- Watson statistic can be used to see if adjacent observations are correlated. Also see table II for each models individual Durbin- Watson statistic. It appears that there is no correlation between the different observations that make up the data. For a further explanation of this test see the appendix Part D II. Table II This table shows the result of tests for linearity and independence of the model, to determine the affects of various variables on capital structure decisions. The tests checks data, from the period , obtained from the abbreviated balance sheets and financial statements from the largest listed, non-financial companies stored in the Datastream database. The F statistics are used to check for the linearity of the model. All the F Statistics fall above the 95% confidence mark. Therefore, it can be assumed the model is linear. The Durbin Watson statistics are used to check for independence. All statistics fall close enough to two; therefore it is safe to assume independence of the various observations. A more extensive explanation of these tests can be found in part D of the Appendix. 20

21 To be able to check for normality the normal probability plot of each individual model needs to be examined. There was a problem with one of the models, an extreme outlier in the South-African sample. This particular observation had a total debt ratio of 68. This not only distorted the entire model, it also made the model appear non normal. In the subsequent years this particular company had a perfectly normal debt ratio. Therefore, it has to be assumed that there was an error in the first observation. This observation was removed from the model. Now all of the models are relatively normal. The normal probability plots of each individual country can be found in Part C of the Appendix. The only sub hypothesis that cannot be included in the regression is hypothesis 8. It concerns the second part of the pecking order theory, that debt is preferred over equity when outside sources of finance are needed by the firm. The null hypothesis states that total debt should be significantly bigger than total equity. This can be tested by doing a paired sample t-test with H 0 being: µ 1 - µ 2 0, where µ 1 stands for total debt and µ 2 for total equity. The analysis of this test can be found in the next section. 21

22 Section IX: Data Analysis This section deals with the analysis of the results from tests and their implications. See the table below for the results of the regression. This regression was used to analyze seven of the sub hypotheses mentioned in Part A of the Appendix. Then the paired sample t-test which is used to assess the eighth hypothesis is analyzed. Table III This table shows regressions of the total debt ratio on six independent variables. This regression is used to study the effects of the variables on capital structure decisions made by companies in nine different countries from 2004 to The data used was obtained using the abbreviated balance sheets and income statements, stored in the Datastream database, from the largest non-financial companies in each country mentioned. The dependent variable; the total debt ratio is defined as total liabilities divided by total assets. The independent variables are defined as follows: income tax rate is obtained by dividing the amount of income tax paid by the earnings before tax, asset tangibility by total assets minus current assets divided by total assets, the market to book ratio by dividing the market value of equity by the book value of equity, business risk by the standard deviation of the return on assets over the five year period being studied, the firm s size by taking the natural logarithm of its total sales or revenue divided by 100 and the profitability by dividing the earnings before tax by total assets. N stands for the number of observations. Table III shows the regression of the total debt ratio on the various independent variables. There are many significant t-statistics meaning that some of the determinants of capital structure are as is suggested in earlier sections. There are differences too, and some independent variables seem to affect some countries more than others. The most successful independent variable is RoA/ profitability, as it is significant and its coefficient negative in all countries. This means that the seventh null hypothesis: that debt and profitability are not negatively correlated can be rejected and with 99% confidence it can be said that there is a negative relationship between these two variables. This is in line with the first part of the pecking order theory where firms with enough free cash flow have less debt. This does, however, give some problems for the second sub hypothesis that Debt and 22

23 EBIT are positively correlated. This could not be investigated directly because of multicollinearity problems between these two variables, but since there is a negative relationship between profitability and debt it is unlikely there can be a positive relation between EBIT and debt. Therefore it is unlikely that agency problem between the firm s management and its shareholders is a determinant in capital structure decisions. Size is significant in eight of the models. The only country in which it is not significant is Russia. Its coefficient is always positive, except in South Africa where it is negative. This means that in South- Africa the smaller companies have relatively more debt. For the other seven countries the sixth null hypothesis that debt and size are not positively correlated can be rejected, and with 99% confidence it can be said that there is a positive relationship between these two variables. Bigger firms thus have relatively more debt than smaller firms. It could be that this is because it is cheaper for the larger firms to borrow funds. If they take on less high risk projects, or they are more diversified than smaller firms, investors are more willing to issue debt to the bigger firms. Size does seem to matter in this case. Asset tangibility also appears to affect the total debt ratio. It has a negative influence in all countries and is significant in five with significance levels of over 99%. In these five countries the ninth null hypothesis which states that debt and asset tangibility are not negatively correlated can be rejected, and with 99% confidence it can be said there is a negative relationship. This could be because of the existence of asymmetric information between the firm s management and its investors in these countries. The real worth of intangible assets is easier to assess by insiders than by outsiders. When outside funds are required to invest in new projects debt is preferred over equity. Firms with more intangible assets are more susceptible to the problem of asymmetric information and therefore appear to have higher levels of debt. The third null hypothesis also deals with assets tangibility. It states that debt and assets tangibility are not positively correlated. Because this hypothesis cannot be rejected there is no evidence in the data to suggest that firms with more tangible assets, thus having a higher liquidation value, and low investigation costs will be able to take on more debt. This could be the case if there was an agency problem between the firm s management and its debt holders. However it could be that this last affect does, in contrast to the evidence above, influence capital structure decisions. It could have an effect so small that it does not significantly alter the negative influence of the former theory but does limit it. However this cannot be tested in our model but it would be an interesting topic for further research. 23

24 The independent variable of business risk gives mixed results. In China, India, Indonesia and Poland it is not significant in influencing the total debt ratio. In Brazil and The Philippines it is significant, their positive coefficients are remarkable. The data on these countries suggests that firms that are riskier are able to take on more debt. It was expected that countries with lower risk levels would be able to take on relatively more debt. Only Russia and Thailand do have the expected sign for the business risk coefficients and only for these two countries can the fifth null hypothesis that there is no negative correlation between the debt ratio and business risk be rejected. However since this relationship is only established with 95% confidence in two out of nine countries there is no evidence to claim this is a determinant for capital structure decisions everywhere. This is again a possible example of the influence of certain country specific factors affecting capital structure decisions made by firms. The least successful independent variables in this sample are income tax and market to book ratio. Their coefficients are at or very close to zero. Income tax is also insignificant in all cases except for India where it has a slightly negative and South-Africa where it has a positive influence on the total debt ratio. The market to book ratio is significant in 5 out of 8 cases (it was not available for Brazil) but its coefficients are even closer to zero. There is not enough evidence to suggest that these two variables are important determinants in the making of capital structure decisions by firms in the various countries. The only country where income tax is significant and has the expected sign is South-Africa. Paired sample t-test See table IV on the following page for the results of the paired sample t-test used to check the second part of the pecking order hypothesis. The second part of the pecking order hypothesis states that when a firm requires external financing, the issuing of debt is preferred over the issuing of equity. To be able to assess sub hypothesis 8 it has to be established that there is a significant difference between the debt- and equity-levels of the companies in the various countries. First Total Shareholders Equity is subtracted from Total Liabilities, and then a 95% confidence interval is calculated. If this interval falls completely above zero then it can be said (with 95% certainty) that Debt > Equity, and vice versa when this interval falls below zero. See Table IV for the results of this test. 24

25 Table IV This table shows the results of a paired samples test. This test is used to study the effects of the variables on capital structure decisions made by companies in nine different countries from 2004 to Specifically to assess whether a firm prefers to issue debt or equity when external financing is required. The data used to perform this test, was obtained using the abbreviated balance sheets and income statements, stored in the Datastream database, from the largest non-financial companies in each country mentioned. The paired sampled t-test investigates whether there is a difference between two sets of data: Total Liabilities and Total Shareholders Equity. A 95% confidence interval has been calculated using SPSS, if this interval is completely above or below zero it can be said with 95% certainty that the two samples are different. As can be seen from table IV for six out of nine countries, the amount debt companies have standing out appears to be higher than the amount of outstanding equity (the 95% confidence interval is completely positive). For these countries the eighth null hypothesis: Debt Equity can be rejected and with 95% certainty the alternate hypothesis; that debt > equity can be accepted. This means that there is evidence supporting the second part of the pecking order theory: That when external funds are required by the firm debt is preferred over equity. For two countries; China and Poland the eighth null hypothesis cannot be rejected and nothing conclusive can be said about the validity of the pecking order theory for these two countries. Interestingly Russia is a different case: Russian companies appear to have less debt outstanding then equity and the second part of the pecking order theory is disproved in Russia. 25

26 Discussion In the first sections of this paper three capital structure theories were introduced. From these theories several factors could be deduced, each one affecting a firm s capital structure. In the past it has been proven that these factors have a role to play for companies in developing countries. In the empirical part of this paper, a multiple linear regression was used to study the effects of six of these factors using data from companies in developing countries. Evidence was found that the variables Asset Tangibility, Size and Profitability each had a significant effect on a firm s capital structure. When taking these factors into account, there is no evidence to reject the idea, that capital structure theories can be applied universally. However the factors business risk and income tax provide a conflicting view. These two variables, although significant in developed countries, are only significant in a few of the developing countries. They even have opposite effects in different countries. These two variables are certainly not in line with the concept of capital structure theories being universally applicable. A paired sample t-test was used to study the final factor. This factor concerned the second part of the pecking order theory, which states that when external financing is required by a firm, the issuing of debt is preferred over the issuing of equity. Six of the countries: Brazil, India, Indonesia, the Philippines, South- Africa and Thailand behaved exactly as the pecking order theory predicted. Companies in these countries have higher levels of debt than of equity. The data for Poland and China is inconclusive. However, Russian companies have more outstanding equity than debt. This completely contradicts the pecking order theory. It becomes very interesting when it is noted that all three of these countries have a history of communism. While this paper does not have the means to prove the validity of this claim, it does seem that there is a clear feature, possibly related to their communist history, affecting the variation in the data that is beyond the scope of the model. Unlike the paper of Booth et al (2001), claiming country specific factors are very important, while Psillaki and Daskalakis (2009) contradict this by claiming firm specific factors are responsible for capital structure decisions. The evidence presented in this paper suggests it is a combination of the two. I believe that some of the capital structure theories can be applied universally, but that for others, country specific factors are stronger than universal theories. 26

27 Now all the individual factors and the sub hypotheses have been handled, the results can be used to analyse the main hypothesis of this paper: The factors related to the problem of asymmetric information and the static trade-off theory are the main factors that determine the capital structure decisions made by firms in developing countries. The main hypothesis can only be partially confirmed. No evidence was found supporting for the fact that the static trade-off theory has any effect on capital structure decisions. Of the five variables that deal with Agency theory, only the variable size is significant and has the expected sign. While the three variables dealing with asymmetric information all have the expected sign, and all three seem to be significant. The most interesting observation comes from the asset tangibility variable. In this sample, it appears that because of the problem of asymmetric information, the relation between asset tangibility and debt is negative. In previous studies on capital structure decisions like the one done by Rajan and Zingales in 1995 on the countries of the G7, it was found this variable was positively correlated with debt because it reduces the related agency costs. The data used in this paper confirms that in developing countries, factors related to asymmetric information, have a higher influence on capital structure decisions than the factors related to agency theory. 27

28 Section X: Conclusion This paper set out to determine if the same factors that influence the capital structure decisions of companies in countries with developed economies, also apply for companies in countries with developing economies. Three theories were found in the existing literature dealing with these factors: The static trade-off theory, agency theory and the problem of asymmetric information. All try to explain the relation between various variables, and the levels of debt a company has. Because of intrinsic differences between developed and developing countries, this paper hypothesized that: The factors related to the problem of asymmetric information and the static trade-off theory, are the main factors that determine the capital structure decisions made by firms in developing countries. No evidence was found in support of the static trade off theory. All of the factors related to the problem of asymmetric information were significant. In contrast, only one of the factors related to agency theory was found to be significant. This does not mean that companies in developing countries do not face agency problems. It means that these agency problems are of a different nature than the ones companies in developed countries face. What these different agency problems are, and how they affect companies in developing countries, is an interesting topic for further research. In the introduction of this paper the main research question was stated: Are the determinants of capital structure decisions taken by companies in developed countries the same as the determinants of these decisions in developing countries? The answer is: The determinants of capital structure decisions taken by companies in developed countries, being the static trade-off theory, agency theory and the problem of asymmetric information, also seem to be the determinants of these decisions of companies in developing countries. However the various magnitudes of the influence of these determinants are not completely similar. Factors related to agency theory seem to have a smaller effect in the developing countries. Consequently the effect of asymmetric information appears to be larger. 28

29 References Bertrand, M. and Mullainathan, S., 2003, Pyramids, Journal of the European Economic Association. Booth, L., Aivazian, A., Demirguc-Kunt, A. and Maksimovic, V., 2001, Capital Structures in Developing Countries, The Journal of Finance. Bradley, M., Jarrell, G.A. and Kim, E.H., 1984, On the Existence of an Optimal Capital Structure: Theory and Evidence, The Journal of Finance. Daskalakis, N. and Psillaki, M., 2009, Do country or firm factors explain capital structure? Evidence from SMEs in France and Greece, Applied Financial Economics. Gilson, R., 2007, Controlling Family Shareholders in Developing Countries: Anchoring Relational Exchange, Harris, M. and Raviv, A., 1991, The Theory of Capital Structure, The Journal of Finance. Hope, O. and Thomas, W., 2007, Managerial Empire Building and Firm Disclosure, Jensen, M.C. and Meckling, W.H., 1976, Theory of the Firm: Managerial Behaviour, Agency Costs and Ownership Structure, Journal of Financial Economics. Kanniainen, V., 1998, Empire Building by Corporate Managers: The Corporation as a Savings Instrument, Journal of Economic Dynamics & Control La Porta, R., Lopez-de-Silanes, F., and Shleifer, A, 1998, Corporate Ownership Around the World, Journal of Finance La Porta, R., Lopez-de-Silanes, F., Shleifer, A and Vishny, R., 1998, Law and Finance, Journal of Political Economy Leland, H. and Pyle, D., 1977, Informational Asymmetries, Financial Structure, and Financial Intermediation, The Journal of Finance Mayer, C., 1990, Financial Systems, corporate finance and economic development, in R, Glenn Hubbard, ed.: Asymetric Information, Corporate Finance and Investment (University of Chicago Press) Modigliani, F. And Miller, M., 1957, The Cost of Capital, Corporation Finance and the Theory of Investment, The American Economic Review 29

30 Modigliani, F. And Miller, M., 1963, Corporate Income Taxes and the cost of Capital: A Correction, The American Economic Review Myers, S.C., 1984, The Capital Structure Puzzle, The Journal of Finance, July Myers, S.C. and Majluf, N. S., 1984, Corporate financing and investment decisions when firms have information that investors do not have, Journal of Financial Economics. Myers, S.C., 2001, Capital Structure, Journal of Economic Perspectives Rajan, R. and Zingales, L., 1995, What do we know about Capital Structure? Some Evidence from International Data, The Journal of Finance. Ross, S., 1977, The Determination of Financial Structure: The Incentive-Signalling Approach, The Bell Journal of Economics Stulz, R. 1990, Managerial discretion and optimal financing policies, Journal of Financial Economics Titman, S. and Wessels, R., 1988, The Determinants of Capital Structure Choice, The Journal of Finance Vithessonthi, C., 2008, What explains stock market reactions to proposals to increase the authorized common stock?, Journal or International Finance and Economics 30

31 Appendix Part A Hypothesis related to the Static Trade-off Theory: Hypothesis 1 If because of the tax shield the value of the firm is increased by higher amounts of debt, the amount of debt and income tax should be positively correlated. H 0 Debt and income tax are not positively correlated H 1 Debt and income tax are positively correlated Hypotheses related to Agency Theory: Hypothesis 2 Because debt should force the managers to pay out more of the money made, debt and the Earnings Before Tax and Interest (EBIT) should be positively correlated. H 0 Debt and EBIT are not positively correlated H 1 Debt and EBIT are positively correlated Hypothesis 3 The cost of debt here is related to the information which needs to be gathered for the liquidation decision. Firms with a higher liquidation value, meaning relatively more tangible assets, and low investigation costs, will be able to take on more debt. In addition, tangible assets can be used as collateral, reducing the interest charged by debt holders. Debt and asset tangibility should be positively correlated. H 0 Debt and Assets Tangibility are not positively correlated H 1 Debt and Assets Tangibility are positively correlated Hypothesis 4 Firms having shareholders, who believe these firms have ample good investment opportunities, could be expected to have relatively low levels of debt. A high market to book ratio suggests that shareholders expect the firm to grow in the future, and that it thus has enough good investment opportunities. Debt and market to book ratio should be negatively correlated. H 0 Debt and the market to book ratio are not negatively correlated H 1 Debt and the market to book value are negatively correlated 31

32 Hypothesis 5 Relatively safe firms do not take on excessive risks, and are able to borrow more before the costs of financial distress become too large to offset the advantages of debt. Debt and business risk should be negatively correlated. H 0 Debt and business risk are not negatively correlated H 1 Debt and business risk are negatively correlated Hypothesis 6 Bigger firms usually are also more diversified then smaller firms, thereby decreasing the risk of financial distress. Also they have their reputation to think of. Therefore Debt and size should be positively correlated. H 0 Debt and size are not positively correlated H 1 Debt and size are positively correlated Hypotheses related to Asymmetric Information: Hypothesis 7 The more profitable a firm, the more free cash flow it has. Firms with enough free cash flow will have relatively less debt. Debt and profitability should be negatively correlated H 0 Debt and profitability are not negatively correlated H 1 Debt and profitability are negatively correlated Hypothesis 8 If outside financing is required debt is preferred over equity. Firms should have relatively more outstanding debt than equity. H 0 Debt Equity H 1 Debt > Equity Hypothesis 9 Since firms with less tangible assets are more susceptible to the problem of asymmetric information they will accumulate more debt over time. Debt and asset tangibility are negatively correlated. H 0 Debt and asset tangibility are not negatively correlated H 1 Debt and assets tangibility are negatively correlated 32

33 Part B Potential Biases Selection bias I: Listed firms only. The sample only includes listed firms. The main reason for this is the availability and quality of reported data. When reviewing the results one should keep in mind those are based on data from listed companies only. Selection bias II: Size. For countries where there were ample companies available only the biggest were selected. Selection bias III: Survivor bias. The list of companies that was compiled for all years between 2004 and 2008 was done on the basis of their market value on the last day of That means that big companies that might have gone into bankruptcy in this period were not included in the sample. Differences in accounting: The different countries use different kinds of accounting policies. Some use the International Financial Reporting Standards (IFRS) others use their own national systems. There is a potential problem for my research because one cannot compare sets of data if different methods are used to collect this data. However this study investigates whether different theories of Capital Structure will hold in countries with upcoming economies. Since each country has an accounting system the data of each country is selected in the same way and each set can be analyzed separately. Financial crisis: The financial crisis hit hard at the end of However the research being done is unaffected since all financial firms were already excluded from the sample. The debt policy for the non financial firms could have been affected somewhat, but not as severely as one might think. The long-term debt was fixed long before the crisis began. The short term debt was only partially affected since the crisis did not affect entire Although the financial crisis has had a profound effect on companies throughout the world, it does not mean it has turned around the rules of capital structure. It are the determinants of capital structure, that are being investigated in this paper, and those have not changed. 33

34 When reviewing the results of this study, the above mentioned potential biases should be kept in mind. However, the best studies on this topic in the past all had similar biases, so one could conclude that it is not possible to conduct this type of study without having these potential biases. Part C These are normal probability plots of the standardized residuals of the models used to study the determinants of capital structure decisions in the countries mentioned. It is used to check for the normality of the data which was obtained from the datastream database. The dependent variable: the total debt ratio is defined as total liabilities divided by total assets. The expected cumulative probability, on the y-axis, is plotted against the observed cumulative probability. If residuals fall close to the 45 line the normality of the data can be assumed. B1: Brazil B2: China B3: India B4: Indonesia 34

35 B5: The Philippines B6: Poland B7: South Africa B8: Russia B9: Thailand 35