Firm characteristics. The current issue and full text archive of this journal is available at

Transcription

1 The current issue and full text archive of this journal is available at How firm characteristics affect capital structure: an empirical study Nikolaos Eriotis National and Kapodistrian University of Athens, Athens, Greece, and Dimitrios Vasiliou and Zoe Ventoura-Neokosmidi Athens University of Economics and Business, Athens, Greece Abstract Purpose The aim of this study is to isolate the firm characteristics that affect capital structure. Design/methodology/approach The investigation has been performed using panel data procedure for a sample of 129 Greek companies listed on the Athens Stock Exchange during The number of the companies in the sample corresponds to the 63 per cent of the listed firms in The firm characteristics are analyzed as determinants of capital structure according to different explanatory theories. The hypothesis that is tested in this paper is that the debt ratio at time t depends on the size of the firm at time t, the growth of the firm at time t, its quick ratio at time t and its interest coverage ratio at time t. The firms that maintain a debt ratio above 50 per cent using a dummy variable are also distinguished. Findings The findings of this study justify the hypothesis that there is a negative relation between the debt ratio of the firms and their growth, their quick ratio and their interest coverage ratio. Size appears to maintain a positive relation and according to the dummy variable there is a differentiation in the capital structure among the firms with a debt ratio greater than 50 per cent and those with a debt ratio lower than 50 per cent. These results are consistent with the theoretical background presented in the second section of the paper. Originality/value This paper goes someway to proving that financial theory does provide some help in understanding how the chosen financing mix affects the firm s value. Keywords Corporate finances, Financial flexibility, Capital structure, Greece Paper type Research paper Firm characteristics Introduction The various financing decisions are vital for the financial welfare of the firm. A false decision about the capital structure may lead to financial distress and eventually to bankruptcy. The management of a firm sets its capital structure in a way that firm s value is maximized. However, firms do choose different financial leverage levels in their effort to attain an optimal capital structure. Although theoretical and empirical research suggests that there is an optimal capital structure, there is no specified methodology, yet, that financial managers can use in order to achieve an optimal debt level. However, financial theory does provide some help in understanding how the chosen financing mix affects the firm s value. This paper shed some light on the determinants of the capital structure of the major Greek firms listed on the Athens Stock Exchange (ASE). We examine the crosssectional variation in leverage among the Greek firms for the time period We include variables that are based on different capital structure theories and have never been investigated for the Greek market before, such as the interest coverage ratio and the quick ratio. We also differentiate the firms that heavily use debt capital (i.e. a debt ratio more than 50 per cent) using a dummy variable. Thus, the conclusions of this paper are expected to enlighten the darksome scientific area of the capital structure determination for the Greek firms. Managerial Finance Vol. 33 No. 5, 2007 pp # Emerald Group Publishing Limited DOI /

2 MF 33,5 322 The paper is organized as follows. In the next session, we review some of the theoretical and empirical literature concerning the determinants and effects of leverage. In section 3, we describe our data and we justify the choice of the variables used in our analysis. The fourth section presents the result of the empirical analysis and a discussion of the conclusions that can be derived from the results. Finally, we summarize our findings in the last section. 2. Theoretical background Modigliani and Miller (1958) were the pioneers in theoretically examining and algebraically demonstrating the effect of capital structure on firm value. Assuming perfect capital markets[1], they concluded to the broadly known theory of capital structure irrelevance which means that the capital structure that a firm chooses does not affect its value. Thereafter, many researchers, including Modigliani and Miller, examined the effects of less restrictive assumptions on the relationship between capital structure and the firm s value. For example, Modigliani and Miller (1963) took taxation under consideration and they proposed that firms should employ as much debt capital as possible in order to achieve the optimal capital structure. Along with corporate taxation, researchers were also interested in analyzing the case of personal taxes imposed on individuals. Miller (1977) discerns three tax rates in the tax legislation of the USA that determine the total value of the firm. These are the corporate tax rate, the tax rate imposed on the income of the dividends and the tax rate imposed on the income of interest inflows. According to Miller, the value of the firm depends on the relative height of each tax rate, compared with the other two. As researchers moved on examining deeper the notion of capital structure, several theories emerged[2], all of which conclude on the existence of an optimal capital structure based on balancing the benefits and costs of debt financing. The main benefit of debt financing is the fact that interest payments are deducted in calculating taxable income, allowing a tax shield for the firms. This tax shield allows firms to pay lower taxes than they should, when using debt capital instead of using only their own capital. The costs of debt can be viewed mainly from two different aspects. First, there is an increased probability that a firm may not be able to successfully deal with its debt obligations (i.e. interest payments); thus, there is an increased probability of bankruptcy. Second, there are agency costs of the lender s monitoring and controlling the firm s actions. There are additional costs concerning the notion of capital structure of the firm that arise from the fact that managers possess more information about the firm s future prospects than do investors. The effect of taxation on capital structure has been thoroughly investigated as a determinant of capital structure. Except for the tax aspects there are also some other approaches that attempt to explain the determination of the capital structure. These approaches examine the debt level determination from the perspective of asymmetric information and agency costs, as already mentioned above. Jensen and Meckling (1976) identify the existence of the agency problem which arises due to the conflicts either between managers and shareholders (agency costs of equity) or between shareholders and debtholders (agency costs of debt). Managers of firms typically act as agents of the owners. The owners hire the managers and give them the authority to manage the firm for the owners benefit. However managers are mainly interested in accomplishing their own targets which may differ from the maximization of the firm value which is the maximization of the owners benefit. They will act in their own interests seeking higher salaries,

3 perquisites, job security and in some cases even direct exploitation of the firm s cash flows. It is obvious that the interests of the manager not only differ but in many cases they even oppose to those of the owners. Thus, a conflict of interests between the shareholders and the managers is inevitable. However, the managers have attained the authority to manage the firm. Thus, the owners may only try to discourage these value transfers through monitoring and control, such as supervision by independent directors; these monitoring and control actions presuppose costs, the so-called agency costs. Perfect control is however extremely costly and therefore, shareholders seek to rely on solutions that would not extract large amounts of value from the firm and would also monitor and control managers operations. A reliable tool can be the use of debt capital which even adds value to the firm. Leverage will force managers to generate and pay out cash, simply because interest payments are compulsory. Interest payments will reduce the amount of remaining cash flows the so-called free cash flows[3] after the investment decisions, at the disposal of the managers. Thus, debt can be viewed as a smart device to reduce the agency costs. In this case, the optimal capital structure will be derived by the balance between the costs of debt against the benefits of debt; the firm will choose this amount of debt which will minimize its total agency costs. Examining the agency costs of debt from the debtholders point of view we have to analyze the lender borrower relationship. When a lender provides funds to a firm, the interest rate charged is based on the lender s assessment of the firm s risk. This arrangement creates incentives for the firm to increase its risk without increasing current borrowing costs. Agency costs of debt only arise when there is a risk of default. If debt is totally free of default risk, debtholders are not concerned about the income, value or the risk of the firm. After obtaining a loan at a certain, locked rate from a bank or through the sale of bonds, the firm can increase its risk. Managers may be tempted to take actions that transfer value from the firm s creditors to its shareholders. For instance, managers could borrow more and pay out cash to shareholders or may invest in risky projects. To avoid this situation lenders impose certain monitoring and controlling techniques on borrowers. Debtholders typically protect themselves by including provisions that prohibit the management of the firm to significantly alter its business or financial risk. These provisions mainly refer to the level of net working capital, asset acquisitions, executive salaries and dividend payments. These protective covenants allow the lender to monitor and control the firm s risk. Alternatively, if no protective covenants are accepted by the firm, creditors may demand higher returns, in the form of higher interest rates. However all these actions enclose some direct or indirect costs that the firm is subject to; these are the agency costs of debt, from the debtholders point of view. In exchange for incurring agency costs by agreeing to cope with the restrictions placed by the lenders, the firm and its owners benefit by obtaining funds at a lower cost. The optimal capital structure of the firm will be formed at this particular level where the benefits of the debt that can be received by the shareholders balance with the costs of debt imposed by the debtholders. The notion of asymmetric information in determining the optimal capital structure is primarily expressed by Myers (1984) and Myers and Majluf (1984). Myers and Majluf (1984) assumed that managers make decisions with the goal to maximize the wealth of existing shareholders. Therefore, they avoid issuing undervalued stock unless the value transfer from old to new shareholders is more than offset by the net present value of the growth opportunity. This leads to the conclusion that new shares will only be issued at a lower price than that imposed by the real market value of the firm. Firm characteristics 323

4 MF 33,5 324 Therefore, an announcement of new equity issue is directly interpreted as a negative signal, in the sense that current investors possess overvalued shares. This negative signal results in the stock price decline. Indeed, several studies[4] have confirmed that the announcement of a stock issue have resulted in a decline of the stock price. That is why several firms tend to follow the pecking order financing pattern. The pecking order theory suggests that firms will initially use internally generated funds, i.e. undistributed earnings, where there is no existence of information asymmetry, then they will draw debt capital if additional funds are needed and finally they will turn to new equity issue to cover any remaining capital requirements. Thus, highly profitable firms that generate high earnings are expected to use less debt capital than those that are not very profitable. Several researchers have tested the effects of profitability on firm leverage. Kester (1986) and Friend and Lang (1988) conclude that there is a significantly negative relation between profitability and debt/asset ratios. Rajan and Zingales (1995) and Wald (1999) find a significantly negative relation between profitability and debt/asset ratios for the USA, the UK and Japan. At this point we should mention that the notion of information asymmetry implies that firms should maintain some reserve borrowing capacity which will allow them to take advantage of good investment opportunities by issuing debt capital if necessary. The notion of asymmetric information is also used to combine the growth opportunities of a firm with its capital structure. Growth causes variations in the value of a firm. Larger variations in the value of the firm are often interpreted as greater risk. That is why a firm that has considerable growth opportunities will be considered as a risky firm and will face difficulties in raising debt capital with favorable terms. Thus, it will employ less debt in its capital structure. On the other hand, the cash flows of a firm which value is most likely to remain stable in the future are predictable and its capital requirements can be financed with debt more easily than these of a firm with growth potential. Myers (1977) argues that firms with growth potential will tend to have lower leverage. To sum up, there is no universal theory of the debt-equity choice. There are several useful conditional theories that attempt to approach the determination of capital structure, each from different aspect. In this paper, we examine some specific factors that determine the capital structure of the Greek firms. 3. Data and measurement of variables In this paper, we investigate the determinants of capital structure for the firms listed in the ASE market during the period All the companies included in the sample fulfil the following two criteria; they were all listed in the market in 1996 and none of them was expelled during the period These criteria were imposed to ensure that the capital structure was not distorted by the effects of a recent official listing. We form our variables using data derived from the financial statements contained in the ASE database. The final sample, after considering any missing data, consists of 129 firms. This figure represents the 63 per cent of the listed companies on the ASE in Thus, our sample consists of a significant proportion of the listed firms in the ASE during the five-year-period Our dependent variable is the debt ratio (variable: DR i,t ) which is defined as the ratio of total debt divided by the total assets of the firm. Total debt contains both long-term and short-term liabilities. Although the strict notion of capital structure refers exclusively to long-term leverage, we have decided to include short-term debt as well, mainly because Greek firms use either very little less than 10 per cent or no

5 long-term capital. Banks in Greece are hesitant in providing long-term financing with attractive terms. Therefore, Greek firms turn to short-term borrowing even when financing their long-term investments. That is why we also consider short-term financing as a measure of gearing. The next variable we consider refers to the size of the firm. Size can be considered as a potential explanatory determinant of differences in leverage among the firms contained in the samples. Size is closely related to risk and bankruptcy costs. Larger firms are usually more diversified and thus bear less risk. Therefore, they have a lower probability of default. Furthermore, larger firms will more easily attract a debt analyst to provide information to the public about the debt issue. Banks are more willing to lend their funds to larger firms partly because they are more diversified and partly because larger firms usually request larger amounts of debt capital than smaller firms. As a consequence, larger firms are usually able to reduce transaction costs associated with long-term debt issuance and can arrange a lower interest rate. Examining the effect of size in the determination of capital structure, Marsh (1982) and Bennett and Donnelly (1993) found that larger firms are likely to use more debt. Thus, we expect that there will be a positive relation of size to leverage. We proxy the size of the firm considering its sales (variable: SIZE i,t ). The higher sales revenue a firm has, the bigger it is considered to be. As mentioned before, we also include short-term debt in our dependent variable. Thus, it is expected that the rate with which the firm covers its short-term debt has a strong influence in the debt ratio. Furthermore, the short-term leverage coverage is an indication of the liquidity of the firm. That is why we also consider the relation between the liquidity of the firm and its capital structure. We use the quick, or acid test, ratio (variable: LIQ i,t ) which is equal to current assets minus inventories divided by current liabilities. This ratio shows the ability of the firm to cover its short-term liabilities and it measures the liquidity of the firm. We expect that there will be a negative relation between the debt ratio of the firm and its liquidity simply because the more debt the firm uses the more current liabilities this will imply and the fewer current assets will remain after dealing with the liabilities. Nevertheless, the fact that a firm employs more current assets implies that it can generate more internal inflows which can then use to finance its operating and investment activities. Thus if the negative relation will be confirmed, there is an implication that firms finance their activities following the financing pattern implied by the pecking order theory. Another variable that we consider is the interest coverage ratio which is expressed as net income before taxes divided by interest payments (variable: INCOV i,t ). The interest coverage ratio has already been theoretically investigated as a determinant of capital structure. Harris and Raviv (1990) propose that leverage is negatively correlated with the interest coverage ratio. They argue that an increase in debt results in a higher default probability. Assuming that interest coverage ratio is a measurement of default probability, this implies that a higher interest coverage ratio indicates a lower debt ratio. Next, we also investigate if there is a relation between the growth of the firm and its capital structure (variable: GROWTH i,t ). We proxy our growth measurement as the annual change on earnings. As already mentioned in the previous section, there should be a negative relation between this regressor and our dependent variable. Since the capital structure of the firm is actually the relation between the total debt and the assets of the firm, we expect that firms that employ more debt than equity will maintain a different capital structure than the market as a whole. In order to capture Firm characteristics 325

6 MF 33,5 326 and isolate that difference made by those firms we introduce a dummy variable, which distinguish them from the whole (variable: DUMMYDR i,t ). More specifically, the dummy is set equal to one for firms which debt ratio is more than 50 per cent, and zero otherwise. The impact of this dummy variable is that the estimated model describes the behavior of the market as a whole and provides information concerning the extra volume of debt that those firms use compare to the market. 4. The model In order to combine cross-sectional with time series data and formulate the characteristics of the market, we use pooling methods for our panel data. The models for panel data are powerful research instruments, which give the researcher the ability to take in to account any kind of effect that the cross-sectional data may have, and finally to estimate the appropriate empirical model. A general model for panel data that allows the researcher to empirically estimate the relation between dependent and independent variables with great flexibility and formulate the differences in the behavior of the cross-section elements is theoretically as follows[5]: y it ¼ xit 0 þ z0 it a þ " it where y it is the dependent variable, x i the matrix with the independent variables and z i a matrix which contains a constant term and/or a set of individual or group specific variables (depending on the sample), which may be observed or unobserved. In case where, in the original model the matrix z includes only a constant term the model can be estimated as a classical linear model and provide the researcher with unbiased coefficient matrix. The method to perform the analysis is the pooled least square. On the other hand, if the observations have individual or group effects then those effects must be taken in to account and have to be included into the z matrix. There are two ways to estimate the model that includes those effects. The first one, the random effects model estimates the coefficient matrix under the assumption that the individual and/or group effects are uncorrelated with the other independent variables and can be formulated and the second one, the fixed effects model, which relaxes these two restrictions. Since, there is not justification that the effects should be treated as uncorrelated with the other regressors, the random effects model may suffer from inconsistency due to omitted variables[6]. In order to have an indication about the correlation between the effects and the independent variables, Hausman (1978) perform a test concerning the relation between the effects and the regressors. The hypothesis that will be tested is that total debt (short- and long-term debt) can be seen as a function of the size of the firm, its ability to successfully fulfil its short run debt, the interest rate coverage ratio, the growth of the firm and the proportion of the extra debt that the firm, with equity less than the 50 per cent of the total assets, uses. Modeling the Greek market according to the variables described in the previous section, we estimate the following model: DR i;t ¼ 0 þ 1 SIZE i;t þ 2 LIQ i;t þ 3 INCOV i;t þ 4 GROWTH i;t þ 5 DUMMYDR i;t þ " i;t ð1þ where DR i,t is the debt ratio of the firm i at time t, SIZE i,t the size of the firm i at time t, LIQ i,t the quick ratio of the firm i at time t, expressed as current assets minus inventories divided by current liabilities, INCOV i,t the interest coverage ratio of firm i at time t, expressed as net income before taxes divided by interest payments, GROWTH i,t

7 the percentage change in earnings of the firm i between time t and t 1, DUMMYDR i,t the dummy variable for DR i,t greater than 50 per cent and i,t the error term. 5. Empirical results In order to estimate the effect of the independent variables on the dependent and to improve our results we consider the three different econometric approaches presented in the previous section. Under the hypothesis that there are no group or individual effects among the firms included in our sample we estimate the total model. The results are presented in Table I. The diagnostics provide us with useful results concerning the theoretical model presented in equation 1. All the variables proved to be significant in confidence level of 5 per cent. The power of the model is given by the high F-statistic of 1, According to adjusted R 2 the independent variables explain the 92 per cent of the size in the debt ratio. In the analysis of panel data, where cross-section combined with time series data, there might be cross-section effects on each firm or on a set of groups of firms. There are two procedures to deal with those effects and each of them has already presented in the beginning of section 4. These two approaches are the random and the fixed effects models for panel data. The case where all the effects are uncorrelated with the regressors and can be formulated as constant terms for each individual or group of firms in the known matrix z, is presented in Table II. The diagnostics from the random effects model suggest that the variable of growth is not statistically significant and does not affect the debt ratio. The adjusted R 2 is lower than that of the total model at 83.5 per cent. In random effects model there are three assumptions about the cross-section effects. The first is that there exist group or individual effects, the second that those effects are uncorrelated with the independent variables and the third that the effects can be formulated. The case where the major assumption about the effects does not hold (i.e. there are no effects) has already presented in Table I. The next step is to stay consistent with the major assumption, there are effects, and relax the last two restrictions concerning them. The results from the fixed effect model are presented in Table III. According to Table III all the independent variables of our model are Firm characteristics 327 Variable Coefficient Std. error t-statistic Prob. C SIZE LIQ INCOV GROWTH DUMMYDR Weighted statistics R Mean dependent var Adjusted R SD dependent var SE. of regression Sum squared resid F-statistic 1, Durbin Watson stat Prob (F-statistic) Notes: Dependent variable: DR; Method: GLS (cross-section weights); White heteroskedasticityconsistent standard errors and covariance Table I. The effect of the independent variables on the dependent using the total model

8 MF 33,5 328 Table II. The effect of the independent variables on the dependent using the random effects model Variable Coefficient Std. error t-statistic Prob. C SIZE LIQ INCOV GROWTH DUMMYDR GLS transformed regression R Mean dependent var Adjusted R SD dependent var SE of regression Sum squared resid Durbin Watson stat Notes: Dependent variable: DR; method: GLS (variance components) statistically significant at 5 per cent. The F-statistic proves the high explanatory power of the estimated model and the high R 2 (adjusted) indicates that the estimated model explain the 97.2 per cent of the size in the dependent variable. According to our findings there is a contradictory result concerning the variable of growth. The total and the fixed effects model accept this variable but the random effects model does not. These controversial results indicate that further analysis has to be done. As we mention in section 4 the random effects model assumes that the individual effects are uncorrelated with the independent variables. In consequence, the random effects model may suffer from inconsistency as a result of omitted variables, something that does not happen with the fixed effects model. On the other hand, the fixed effects model uses the individual effects as given by the sample. In order to see if the individual effects are uncorrelated with the regressors we perform a Hausman test. The test statistic is and the critical value of the chi square table with five degrees of freedom, at 95 per cent, is 11.7, which is lower than the test s value. Hence, the hypothesis that the individual effects are uncorrelated with the regressors can be rejected. The random effects model estimates suffer from inconsistency probably due to omitted variables (see Variable Coefficient Std. error t-statistic Prob. Table III. The effect of the independent variables on the dependent using the fixed effects model SIZE LIQ INCOV GROWTH DUMMYDR Weighted statistics R Mean dependent var Adjusted R SD dependent var SE of regression Sum squared resid F-statistic 5, Durbin Watson stat Prob (F-statistic) Notes: Dependent variable: DR; method: GLS (cross-section weights)

9 section 4). Hence, according to our sample and findings, the appropriate model to explain the market is the one that includes the GROWTH variable. According to our findings the SIZE of the firm has a positive relation with the debt ratio, something that has been confirmed by Marsh (1982) and Bennett and Donnelly (1993), which found similar results with us. This suggests that larger firms use more debt. The short-term leverage coverage is an indication of the liquidity of the firm. As we expected there is a negative relation between the debt ratio of the firm and its liquidity. The negative relation confirms that firms finance their activities following the financing pattern implied by the pecking order theory. As we expected the negative relation between debt and growth has been confirmed from our data. The statistical significance of the dummy variable and its positive sign indicate that there is a distinction in the capital structure between firms who have debt ratio greater than 50 per cent and those that do not have. According to our results from the fixed effects model these firms use, compared to the market, an extra debt of 19 per cent. Firm characteristics Conclusions In this study, we conduct our analysis in order to investigate how some specific firm characteristics determine the firm s capital structure. We use the panel data derived by the financial statements of 129 Greek firms listed in the ASE. In our calculations we consider the total model, the fixed effects model and the random effects model. Our dependent variable is the debt ratio expressed as total liabilities divided by total assets. The debt ratio includes both long-term and short-term liabilities mainly because Greek firms use either very little or no long-term debt capital at all. According to the results, the debt ratio of the firm is positively related to its size which is measured by the sales figure. Thus, larger firms employ more debt capital in comparison with smaller firms, a finding which is consistent with the theoretical background mentioned in the second section of the paper. On the other hand, our findings show that the liquidity of the firm is negatively related to its financial leverage. We consider the liquidity of the firms using the quick, or acid test ratio which is equal to current assets minus inventories divided by current liabilities. This ratio shows the ability of the firm to deal with its short-term liabilities. Firms with high liquidity tend to use less debt. This finding can be considered as an indication that firms generally finance their activities following the financing procedure implied by the pecking order theory. Firms with high liquidity maintain a relatively high amount of current assets, which means that they maintain high cash inflows. This means that they also generate high cash inflows. As a consequence, they are able to use these inflows in order to finance their operating and financing activities. Thus, they do not use much debt capital in comparison with firms that are not so profitable because they prefer to use these funds rather than debt capital; this is an indication of pecking order financing. This finding is further supported by the result of the negative relation between the interest coverage ratio of the firms and their capital structure. The interest coverage ratio is expressed as net income before taxes divided by interest payments. Thus, firms that maintain a relatively high interest coverage ratio prefer to use less debt capital. If a firm has a high interest coverage ratio, this means that it has the ability to generate relatively high earnings. The negative relation implies that firms probably prefer to use these earnings to finance their activities and thus use less debt capital; this is also an implication of the pecking order financing.

10 MF 33,5 330 The negative relation between the growth of the firm and its capital structure shows that firms with high growth potential employ less debt in their capital structure. We proxy our growth measurement as the annual change on earnings. Thus, high growth means high variation in earnings which can be interpreted as higher risk. Firms that are risky generally find it difficult to raise debt capital, simply because the lenders will demand higher returns making debt capital more expensive. According to the results of the dummy variable, we find strong evidence that there is a capital structure differentiation among the firms which heavily use debt capital (more than 50 per cent of their total assets) and those that use less debt capital. The results and conclusions are consistent with the theoretical background as presented in the second section of the present paper. All the three models conclude in the same remarks except for the situation of the growth variable. The growth variable is not statistically significant in the random effects model, but it is in the other two models. However, the Hausman test indicates that the fixed effects model fits better to our specific set of variables and thus prevails over the random effects model. Thus, growth does affect the determination of capital structure. Notes 1. Perfect capital markets means that the following assumptions hold: (a) there are no taxes, (b) there are no transaction costs, (c) there is symmetrical information, (d) there are homogenous expectations and (e) investors can borrow at the same rate as corporations. 2. Harris and Raviv (1991) refer to various theories of capital structure. 3. For more information about the free cash flows hypothesis see Jensen (1986). 4. See for example Asquith and Mullins (1986). 5. For more information, see Greene (2003). 6. For further analysis see Hausman and Taylor (1981) and Chamberlain (1978). References Asquith, P. and Mullins, W. (1986), Equity issues and offering dilution, Journal of Financial Economics, Vol. 15, pp Bennett, M. and Donnelly, R. (1993), The determinants of capital structure: some UK evidence, British Accounting Review, pp Chamberlain, G. (1978), Omitted variable bias and panel data: estimating the returns to schooling, Annales de L Insee, Vol. 30/31, pp Friend, I. and Lang, H.P. (1988), An empirical test of the impact of managerial self-interest on corporate capital structure, Journal of Finance, Vol. 43, pp Greene, W.H. (2003), Econometric Analysis, 5th ed., Prentice Hall, New Jersey, NJ. Harris, M. and Raviv, A. (1990), Capital structure and the informational role of debt, Journal of Finance, Vol. 45, pp Hausman, J. (1978), Specification tests in econometrics, Econometrica, Vol. 46, pp Hausman, J. and Taylor, W. (1981), Panel data and unobservable individual effects, Econometrica, Vol. 49, pp Jensen, M. (1986), Agency costs of free cash flow, corporate finance and takeovers, American Economic Review, Vol. 76, pp Jensen, M. and Meckling, W. (1976), Theory of the firm: managerial behavior, agency costs and ownership structure, Journal of Financial Economics, Vol. 2, pp

11 Kester, W.C. (1986), Capital and ownership structure: a comparison of United States and Japanese manufacturing corporations, Financial Management, Vol. 15, pp Marsh, P. (1982), The choice between equity and debt: an empirical study, Journal of Finance, Vol. 37, pp Miller, M.H. (1977), Debt and taxes, Journal of Finance, Vol. 32, pp Modigliani, F. and Miller, M. (1958), The cost of capital, corporate finance and the theory of investment, American Economic Review, Vol. 48, pp Modigliani, F. and Miller, M. (1963), Corporate income taxes and the cost of capital: a correction, American Economic Review, Vol. 53, June, pp Myers, S.C. (1977), Determinants of corporate borrowing, Journal of Financial Economics, Vol. 5, pp Myers, S.C. (1984), The capital structure puzzle, Journal of Finance, Vol. 39, July, pp 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, Vol. 13, pp Rajan, R.G. and Zingales, L. (1995), What do we know about capital structure: some evidence from international data, Journal of Finance, Vol. 50, pp Wald, J. (1999), How firm characteristics affect capital structure: an international comparison, The Journal of Financial Research, Vol. XXII No. 2, pp Firm characteristics 331 Corresponding author Nickolaos Eriotis can be contacted at: neriot@econ.uoa.gr; nikolaos.eriotis@aueb.gr To purchase reprints of this article please reprints@emeraldinsight.com Or visit our web site for further details:

12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.