Firmcharacteristics

321

Managerial FinanceVol. 33 No. 5, 2007

pp. 321-331# Emerald Group Publishing Limited

0307-4358DOI 10.1108/03074350710739605

How firm characteristics affectcapital structure: an empirical

studyNikolaos Eriotis

National and Kapodistrian University of Athens, Athens, Greece, and

Dimitrios Vasiliou and Zoe Ventoura-NeokosmidiAthens 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 dataprocedure for a sample of 129 Greek companies listed on the Athens Stock Exchange during 1997-2001. The number of the companies in the sample corresponds to the 63 per cent of the listed firms in1996. The firm characteristics are analyzed as determinants of capital structure according to differentexplanatory theories. The hypothesis that is tested in this paper is that the debt ratio at time tdepends on the size of the firm at time t, the growth of the firm at time t, its quick ratio at time t andits interest coverage ratio at time t. The firms that maintain a debt ratio above 50 per cent using adummy variable are also distinguished.Findings – The findings of this study justify the hypothesis that there is a negative relation betweenthe debt ratio of the firms and their growth, their quick ratio and their interest coverage ratio. Sizeappears to maintain a positive relation and according to the dummy variable there is a differentiationin the capital structure among the firms with a debt ratio greater than 50 per cent and those with adebt ratio lower than 50 per cent. These results are consistent with the theoretical backgroundpresented in the second section of the paper.Originality/value – This paper goes someway to proving that financial theory does provide somehelp in understanding how the chosen financing mix affects the firm’s value.

Keywords Corporate finances, Financial flexibility, Capital structure, Greece

Paper type Research paper

1. IntroductionThe various financing decisions are vital for the financial welfare of the firm. A falsedecision about the capital structure may lead to financial distress and eventually tobankruptcy. The management of a firm sets its capital structure in a way that firm’svalue is maximized. However, firms do choose different financial leverage levels intheir effort to attain an optimal capital structure. Although theoretical and empiricalresearch suggests that there is an optimal capital structure, there is no specifiedmethodology, yet, that financial managers can use in order to achieve an optimal debtlevel. However, financial theory does provide some help in understanding how thechosen financing mix affects the firm’s value.

This paper shed some light on the determinants of the capital structure of the majorGreek firms listed on the Athens Stock Exchange (ASE). We examine the cross-sectional variation in leverage among the Greek firms for the time period 1997-2001.We include variables that are based on different capital structure theories and havenever been investigated for the Greek market before, such as the interest coverage ratioand the quick ratio. We also differentiate the firms that heavily use debt capital (i.e. adebt ratio more than 50 per cent) using a dummy variable. Thus, the conclusions of thispaper are expected to enlighten the darksome scientific area of the capital structuredetermination for the Greek firms.

The current issue and full text archive of this journal is available atwww.emeraldinsight.com/0307-4358.htm

MF33,5

322

The paper is organized as follows. In the next session, we review some of thetheoretical and empirical literature concerning the determinants and effects ofleverage. In section 3, we describe our data and we justify the choice of the variablesused in our analysis. The fourth section presents the result of the empirical analysisand a discussion of the conclusions that can be derived from the results. Finally, wesummarize our findings in the last section.

2. Theoretical backgroundModigliani and Miller (1958) were the pioneers in theoretically examining andalgebraically demonstrating the effect of capital structure on firm value. Assumingperfect capital markets[1], they concluded to the broadly known theory of ‘‘capitalstructure irrelevance’’ which means that the capital structure that a firm chooses doesnot affect its value. Thereafter, many researchers, including Modigliani and Miller,examined the effects of less restrictive assumptions on the relationship between capitalstructure and the firm’s value. For example, Modigliani and Miller (1963) took taxationunder consideration and they proposed that firms should employ as much debt capitalas possible in order to achieve the optimal capital structure. Along with corporatetaxation, researchers were also interested in analyzing the case of personal taxesimposed on individuals. Miller (1977) discerns three tax rates in the tax legislation ofthe 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 theincome of interest inflows. According to Miller, the value of the firm depends on therelative height of each tax rate, compared with the other two.

As researchers moved on examining deeper the notion of capital structure, severaltheories emerged[2], all of which conclude on the existence of an optimal capitalstructure based on balancing the benefits and costs of debt financing. The main benefitof debt financing is the fact that interest payments are deducted in calculating taxableincome, allowing a ‘‘tax shield’’ for the firms. This ‘‘tax shield’’ allows firms to paylower taxes than they should, when using debt capital instead of using only their owncapital. The costs of debt can be viewed mainly from two different aspects. First, thereis an increased probability that a firm may not be able to successfully deal with itsdebt obligations (i.e. interest payments); thus, there is an increased probability ofbankruptcy. Second, there are agency costs of the lender’s monitoring and controllingthe firm’s actions. There are additional costs concerning the notion of capital structureof the firm that arise from the fact that managers possess more information about thefirm’s future prospects than do investors.

The effect of taxation on capital structure has been thoroughly investigated as adeterminant of capital structure. Except for the tax aspects there are also some otherapproaches that attempt to explain the determination of the capital structure. Theseapproaches examine the debt level determination from the perspective of asymmetricinformation and agency costs, as already mentioned above. Jensen and Meckling (1976)identify the existence of the agency problem which arises due to the conflicts eitherbetween managers and shareholders (agency costs of equity) or between shareholdersand debtholders (agency costs of debt).

Managers of firms typically act as agents of the owners. The owners hire themanagers and give them the authority to manage the firm for the owners’ benefit.However managers are mainly interested in accomplishing their own targets whichmay differ from the maximization of the firm value which is the maximization of theowners’ benefit. They will act in their own interests seeking higher salaries,

Firmcharacteristics

323

perquisites, job security and in some cases even direct exploitation of the firm’s cashflows. It is obvious that the interests of the manager not only differ but in many casesthey even oppose to those of the owners. Thus, a conflict of interests between theshareholders and the managers is inevitable. However, the managers have attained theauthority to manage the firm. Thus, the owners may only try to discourage these valuetransfers through monitoring and control, such as supervision by independentdirectors; these monitoring and control actions presuppose costs, the so-called agencycosts. Perfect control is however extremely costly and therefore, shareholders seek torely on solutions that would not extract large amounts of value from the firm andwould also monitor and control managers’ operations. A reliable tool can be the use ofdebt capital which even adds value to the firm. Leverage will force managers togenerate and pay out cash, simply because interest payments are compulsory. Interestpayments will reduce the amount of remaining cash flows – the so-called free cashflows[3] – after the investment decisions, at the disposal of the managers. Thus, debtcan be viewed as a smart device to reduce the agency costs. In this case, the optimalcapital structure will be derived by the balance between the costs of debt against thebenefits of debt; the firm will choose this amount of debt which will minimize its totalagency costs.

Examining the agency costs of debt from the debtholders’ point of view we have toanalyze the lender – borrower relationship. When a lender provides funds to a firm, theinterest rate charged is based on the lender’s assessment of the firm’s risk. Thisarrangement creates incentives for the firm to increase its risk without increasingcurrent 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 bankor through the sale of bonds, the firm can increase its risk. Managers may be temptedto take actions that transfer value from the firm’s creditors to its shareholders. Forinstance, managers could borrow more and pay out cash to shareholders or may investin risky projects. To avoid this situation lenders impose certain monitoring andcontrolling techniques on borrowers. Debtholders typically protect themselves byincluding provisions that prohibit the management of the firm to significantly alter itsbusiness or financial risk. These provisions mainly refer to the level of net workingcapital, asset acquisitions, executive salaries and dividend payments. These protectivecovenants allow the lender to monitor and control the firm’s risk. Alternatively, if noprotective covenants are accepted by the firm, creditors may demand higher returns, inthe form of higher interest rates. However all these actions enclose some direct orindirect costs that the firm is subject to; these are the agency costs of debt, from thedebtholders’ point of view. In exchange for incurring agency costs by agreeing to copewith the restrictions placed by the lenders, the firm and its owners benefit by obtainingfunds at a lower cost. The optimal capital structure of the firm will be formed at thisparticular level where the benefits of the debt that can be received by the shareholdersbalance with the costs of debt imposed by the debtholders.

The notion of asymmetric information in determining the optimal capital structureis 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 ofexisting shareholders. Therefore, they avoid issuing undervalued stock unless thevalue transfer from ‘‘old’’ to new shareholders is more than offset by the net presentvalue of the growth opportunity. This leads to the conclusion that new shares will onlybe issued at a lower price than that imposed by the real market value of the firm.

MF33,5

324

Therefore, an announcement of new equity issue is directly interpreted as a negativesignal, in the sense that current investors possess overvalued shares. This negativesignal results in the stock price decline. Indeed, several studies[4] have confirmed thatthe announcement of a stock issue have resulted in a decline of the stock price. That iswhy several firms tend to follow the ‘‘pecking order’’ financing pattern. The ‘‘peckingorder’’ theory suggests that firms will initially use internally generated funds, i.e.undistributed earnings, where there is no existence of information asymmetry, thenthey will draw debt capital if additional funds are needed and finally they will turn tonew equity issue to cover any remaining capital requirements. Thus, highly profitablefirms that generate high earnings are expected to use less debt capital than those thatare not very profitable. Several researchers have tested the effects of profitability onfirm leverage. Kester (1986) and Friend and Lang (1988) conclude that there is asignificantly negative relation between profitability and debt/asset ratios. Rajanand Zingales (1995) and Wald (1999) find a significantly negative relation betweenprofitability and debt/asset ratios for the USA, the UK and Japan.

At this point we should mention that the notion of information asymmetry impliesthat firms should maintain some reserve borrowing capacity which will allow themto take advantage of good investment opportunities by issuing debt capital ifnecessary. The notion of asymmetric information is also used to combine the growthopportunities of a firm with its capital structure. Growth causes variations in the valueof 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 arisky firm and will face difficulties in raising debt capital with favorable terms. Thus, itwill employ less debt in its capital structure. On the other hand, the cash flows of a firmwhich value is most likely to remain stable in the future are predictable and its capitalrequirements can be financed with debt more easily than these of a firm with growthpotential. Myers (1977) argues that firms with growth potential will tend to have lowerleverage.

To sum up, there is no universal theory of the debt-equity choice. There are severaluseful conditional theories that attempt to approach the determination of capitalstructure, each from different aspect. In this paper, we examine some specific factorsthat determine the capital structure of the Greek firms.

3. Data and measurement of variablesIn this paper, we investigate the determinants of capital structure for the firms listed inthe ASE market during the period 1997-2001. All the companies included in the samplefulfil the following two criteria; they were all listed in the market in 1996 and none ofthem was expelled during the period 1997-2001. These criteria were imposed to ensurethat the capital structure was not distorted by the effects of a recent official listing. Weform our variables using data derived from the financial statements contained in theASE database. The final sample, after considering any missing data, consists of 129firms. This figure represents the 63 per cent of the listed companies on the ASE in 1996.Thus, our sample consists of a significant proportion of the listed firms in the ASEduring the five-year-period 1997-2001.

Our dependent variable is the debt ratio (variable: DRi,t) which is defined as the ratioof total debt divided by the total assets of the firm. Total debt contains both long-termand short-term liabilities. Although the strict notion of capital structure refersexclusively to long-term leverage, we have decided to include short-term debt aswell, mainly because Greek firms use either very little – less than 10 per cent – or no

Firmcharacteristics

325

long-term capital. Banks in Greece are hesitant in providing long-term financing withattractive terms. Therefore, Greek firms turn to short-term borrowing even whenfinancing their long-term investments. That is why we also consider short-termfinancing as a measure of gearing.

The next variable we consider refers to the size of the firm. Size can be considered asa potential explanatory determinant of differences in leverage among the firmscontained in the samples. Size is closely related to risk and bankruptcy costs. Largerfirms are usually more diversified and thus bear less risk. Therefore, they have a lowerprobability of default. Furthermore, larger firms will more easily attract a debt analystto provide information to the public about the debt issue. Banks are more willing tolend their funds to larger firms partly because they are more diversified and partlybecause larger firms usually request larger amounts of debt capital than smaller firms.As a consequence, larger firms are usually able to reduce transaction costs associatedwith long-term debt issuance and can arrange a lower interest rate. Examining theeffect of size in the determination of capital structure, Marsh (1982) and Bennett andDonnelly (1993) found that larger firms are likely to use more debt. Thus, we expectthat there will be a positive relation of size to leverage. We proxy the size of the firmconsidering its sales (variable: SIZEi,t). The higher sales revenue a firm has, the biggerit 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 astrong influence in the debt ratio. Furthermore, the short-term leverage coverage is anindication of the liquidity of the firm. That is why we also consider the relationbetween the liquidity of the firm and its capital structure. We use the quick, or acid test,ratio (variable: LIQi,t) which is equal to current assets minus inventories divided bycurrent liabilities. This ratio shows the ability of the firm to cover its short-termliabilities and it measures the liquidity of the firm. We expect that there will be anegative relation between the debt ratio of the firm and its liquidity simply because themore debt the firm uses the more current liabilities this will imply and the fewercurrent assets will remain after dealing with the liabilities. Nevertheless, the fact that afirm employs more current assets implies that it can generate more internal inflowswhich can then use to finance its operating and investment activities. Thus if thenegative relation will be confirmed, there is an implication that firms finance theiractivities following the financing pattern implied by the ‘‘pecking order’’ theory.

Another variable that we consider is the interest coverage ratio which is expressedas net income before taxes divided by interest payments (variable: INCOVi,t). Theinterest coverage ratio has already been theoretically investigated as a determinant ofcapital structure. Harris and Raviv (1990) propose that leverage is negativelycorrelated with the interest coverage ratio. They argue that an increase in debt resultsin a higher default probability. Assuming that interest coverage ratio is a measurementof default probability, this implies that a higher interest coverage ratio indicates a lowerdebt ratio.

Next, we also investigate if there is a relation between the growth of the firm and itscapital structure (variable: GROWTHi,t). We proxy our growth measurement as theannual change on earnings. As already mentioned in the previous section, there shouldbe a negative relation between this regressor and our dependent variable.

Since the capital structure of the firm is actually the relation between the total debtand the assets of the firm, we expect that firms that employ more debt than equity willmaintain a different capital structure than the market as a whole. In order to capture

MF33,5

326

and isolate that difference made by those firms we introduce a dummy variable, whichdistinguish them from the whole (variable: DUMMYDRi,t). More specifically, thedummy is set equal to one for firms which debt ratio is more than 50 per cent, and zerootherwise. The impact of this dummy variable is that the estimated model describesthe behavior of the market as a whole and provides information concerning the extravolume of debt that those firms use compare to the market.

4. The modelIn order to combine cross-sectional with time series data and formulate thecharacteristics of the market, we use pooling methods for our panel data. The modelsfor panel data are powerful research instruments, which give the researcher the abilityto take in to account any kind of effect that the cross-sectional data may have, andfinally to estimate the appropriate empirical model. A general model for panel data thatallows the researcher to empirically estimate the relation between dependent andindependent variables with great flexibility and formulate the differences in thebehavior of the cross-section elements is theoretically as follows[5]:

yit ¼ x 0it� þ z0itaþ "it

where yit is the dependent variable, xi the matrix with the independent variables and zia matrix which contains a constant term and/or a set of individual or group specificvariables (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 themodel can be estimated as a classical linear model and provide the researcher withunbiased coefficient matrix. The method to perform the analysis is the pooled leastsquare. On the other hand, if the observations have individual or group effects thenthose 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, therandom effects model estimates the coefficient matrix under the assumption that theindividual and/or group effects are uncorrelated with the other independent variablesand can be formulated and the second one, the fixed effects model, which relaxes thesetwo restrictions. Since, there is not justification that the effects should be treated asuncorrelated with the other regressors, the random effects model may suffer frominconsistency due to omitted variables[6]. In order to have an indication aboutthe 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) canbe seen as a function of the size of the firm, its ability to successfully fulfil its short rundebt, the interest rate coverage ratio, the growth of the firm and the proportion of theextra 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 previoussection, we estimate the following model:

DRi;t ¼ �0 þ �1SIZEi;t þ �2LIQi;t þ �3INCOVi;t

þ �4GROWTHi;t þ �5DUMMYDRi;t þ "i;t ð1Þ

where DRi,t is the debt ratio of the firm i at time t, SIZEi,t the size of the firm i at time t,LIQi,t the quick ratio of the firm i at time t, expressed as current assets minusinventories divided by current liabilities, INCOVi,t the interest coverage ratio of firm i attime t, expressed as net income before taxes divided by interest payments, GROWTHi,t

Firmcharacteristics

327

the percentage change in earnings of the firm i between time t and t� 1, DUMMYDRi,t

the dummy variable for DRi,t greater than 50 per cent and �i,t the error term.

5. Empirical resultsIn order to estimate the effect of the independent variables on the dependent and toimprove our results we consider the three different econometric approaches presentedin the previous section. Under the hypothesis that there are no group or individualeffects among the firms included in our sample we estimate the total model. The resultsare presented in Table I. The diagnostics provide us with useful results concerning thetheoretical model presented in equation 1. All the variables proved to be significant inconfidence level of 5 per cent. The power of the model is given by the high F-statistic of1,352.4. According to adjusted R2 the independent variables explain the 92 per cent ofthe 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. Thereare two procedures to deal with those effects and each of them has already presented inthe beginning of section 4. These two approaches are the random and the fixed effectsmodels for panel data. The case where all the effects are uncorrelated with theregressors and can be formulated as constant terms for each individual or group offirms in the known matrix z, is presented in Table II. The diagnostics from the randomeffects model suggest that the variable of growth is not statistically significant anddoes not affect the debt ratio. The adjusted R2 is lower than that of the total model at83.5 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 areuncorrelated with the independent variables and the third that the effects can beformulated. 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 stayconsistent with the major assumption, there are effects, and relax the last tworestrictions concerning them. The results from the fixed effect model are presented inTable III. According to Table III all the independent variables of our model are

Table I.The effect of the

independent variables onthe dependent using the

total model

Variable Coefficient Std. error t-statistic Prob.

C 0.302464 0.008713 34.71596 0.0000SIZE 6.21� 10�11 1.71� 10�12 36.35105 0.0000LIQ �0.011476 0.004888 �2.347688 0.0192INCOV �1.37� 10�6 2.40� 10�7 �5.702233 0.0000GROWTH �4.59� 10�6 1.92� 10�7 �23.87668 0.0000DUMMYDR 0.341478 0.007514 45.44843 0.0000

Weighted statisticsR2 0.919998 Mean dependent var. 0.502023Adjusted R2 0.919318 SD dependent var. 0.426139SE. of regression 0.121043 Sum squared resid. 8.615047F-statistic 1,352.366 Durbin–Watson stat. 1.294846Prob (F-statistic) 0.000000

Notes: Dependent variable: DR; Method: GLS (cross-section weights); White heteroskedasticity-consistent standard errors and covariance

MF33,5

328

statistically significant at 5 per cent. The F-statistic proves the high explanatory powerof the estimated model and the high R2 (adjusted) indicates that the estimated modelexplain the 97.2 per cent of the size in the dependent variable.

According to our findings there is a contradictory result concerning the variable ofgrowth. The total and the fixed effects model accept this variable but the random effectsmodel 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 effectsare uncorrelated with the independent variables. In consequence, the random effectsmodel may suffer from inconsistency as a result of omitted variables, something thatdoes not happen with the fixed effects model. On the other hand, the fixed effects modeluses the individual effects as given by the sample. In order to see if the individual effectsare uncorrelated with the regressors we perform a Hausman test. The test statistic is565.3 and the critical value of the chi–square table with five degrees of freedom, at 95per cent, is 11.7, which is lower than the test’s value. Hence, the hypothesis that theindividual effects are uncorrelated with the regressors can be rejected. The randomeffects model estimates suffer from inconsistency probably due to omitted variables (see

Table II.The effect of theindependent variables onthe dependent using therandom effects model

Variable Coefficient Std. error t-statistic Prob.

C 0.307623 0.008897 34.57707 0.0000SIZE 7.57� 10�11 1.63� 10�11 4.644296 0.0000LIQ �0.004909 0.000834 �5.888968 0.0000INCOV �1.12� 10�6 4.41� 10�7 �2.530628 0.0116GROWTH �2.49� 10�6 3.06� 10�6 �0.812516 0.4168DUMMYDR 0.268865 0.011178 24.05228 0.0000

GLS transformed regressionR2 0.836422 Mean dependent var. 0.382953Adjusted R2 0.835031 SD dependent var. 0.217482SE of regression 0.088333 Sum squared resid. 4.588026Durbin–Watson stat. 1.391236

Notes: Dependent variable: DR; method: GLS (variance components)

Table III.The effect of theindependent variables onthe dependent using thefixed effects model

Variable Coefficient Std. error t-statistic Prob.

SIZE 4.35� 10�11 1.03� 10�11 4.235927 0.0000LIQ �0.003551 0.000756 �4.699572 0.0000INCOV �1.11� 10�6 1.15� 10�7 �9.631313 0.0000GROWTH �1.45� 10�6 5.78� 10�7 �2.509531 0.0124DUMMYDR 0.199252 0.003510 56.75966 0.0000

Weighted statisticsR2 0.977800 Mean dependent var. 0.543577Adjusted R2 0.971871 SD dependent var. 0.470460SE of regression 0.078904 Sum squared resid. 2.913721F-statistic 5,153.334 Durbin–Watson stat. 2.076985Prob (F-statistic) 0.000000

Notes: Dependent variable: DR; method: GLS (cross-section weights)

Firmcharacteristics

329

section 4). Hence, according to our sample and findings, the appropriate model toexplain 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 debtratio, 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 moredebt. The short-term leverage coverage is an indication of the liquidity of the firm. Aswe expected there is a negative relation between the debt ratio of the firm and itsliquidity. The negative relation confirms that firms finance their activities followingthe financing pattern implied by the ‘‘pecking order’’ theory. As we expected thenegative relation between debt and growth has been confirmed from our data. Thestatistical significance of the dummy variable and its positive sign indicate that there isa distinction in the capital structure between firms who have debt ratio greater than 50per cent and those that do not have. According to our results from the fixed effectsmodel these firms use, compared to the market, an extra debt of 19 per cent.

6. ConclusionsIn this study, we conduct our analysis in order to investigate how some specific firmcharacteristics determine the firm’s capital structure. We use the panel data derived bythe financial statements of 129 Greek firms listed in the ASE. In our calculations weconsider 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 totalassets. The debt ratio includes both long-term and short-term liabilities mainly becauseGreek firms use either very little or no long-term debt capital at all. According to theresults, the debt ratio of the firm is positively related to its size which is measured bythe sales figure. Thus, larger firms employ more debt capital in comparison withsmaller firms, a finding which is consistent with the theoretical background mentionedin the second section of the paper.

On the other hand, our findings show that the liquidity of the firm is negativelyrelated 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 currentliabilities. 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 anindication that firms generally finance their activities following the financingprocedure implied by the pecking order theory. Firms with high liquidity maintain arelatively high amount of current assets, which means that they maintain high cashinflows. This means that they also generate high cash inflows. As a consequence, theyare 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 soprofitable because they prefer to use these funds rather than debt capital; this is anindication of pecking order financing.

This finding is further supported by the result of the negative relation between theinterest coverage ratio of the firms and their capital structure. The interest coverageratio is expressed as net income before taxes divided by interest payments. Thus, firmsthat maintain a relatively high interest coverage ratio prefer to use less debt capital. If afirm has a high interest coverage ratio, this means that it has the ability to generaterelatively high earnings. The negative relation implies that firms probably prefer touse these earnings to finance their activities and thus use less debt capital; this is alsoan implication of the pecking order financing.

MF33,5

330

The negative relation between the growth of the firm and its capital structure showsthat firms with high growth potential employ less debt in their capital structure. Weproxy our growth measurement as the annual change on earnings. Thus, high growthmeans high variation in earnings which can be interpreted as higher risk. Firms thatare risky generally find it difficult to raise debt capital, simply because the lenders willdemand higher returns making debt capital more expensive.

According to the results of the dummy variable, we find strong evidence that thereis 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 aspresented in the second section of the present paper. All the three models conclude inthe same remarks except for the situation of the growth variable. The growth variableis not statistically significant in the random effects model, but it is in the other twomodels. However, the Hausman test indicates that the fixed effects model fits better toour 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 notaxes, (b) there are no transaction costs, (c) there is symmetrical information, (d) thereare homogenous expectations and (e) investors can borrow at the same rate ascorporations.

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 FinancialEconomics, Vol. 15, pp. 61-89.

Bennett, M. and Donnelly, R. (1993), ‘‘The determinants of capital structure: some UK evidence’’,British Accounting Review, pp. 43-59.

Chamberlain, G. (1978), ‘‘Omitted variable bias and panel data: estimating the returns toschooling’’,Annales de L’Insee, Vol. 30/31, pp. 49-82.

Friend, I. and Lang, H.P. (1988), ‘‘An empirical test of the impact of managerial self-interest oncorporate capital structure’’, Journal of Finance, Vol. 43, pp. 271-81.

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 ofFinance, Vol. 45, pp. 321-49.

Hausman, J. (1978), ‘‘Specification tests in econometrics’’, Econometrica, Vol. 46, pp. 1251-71.

Hausman, J. and Taylor, W. (1981), ‘‘Panel data and unobservable individual effects’’,Econometrica, Vol. 49, pp. 1377-98.

Jensen, M. (1986), ‘‘Agency costs of free cash flow, corporate finance and takeovers’’, AmericanEconomic Review, Vol. 76, pp. 323-39.

Jensen, M. and Meckling, W. (1976), ‘‘Theory of the firm: managerial behavior, agency costs andownership structure’’, Journal of Financial Economics, Vol. 2, pp. 305-60.

Firmcharacteristics

331

Kester, W.C. (1986), ‘‘Capital and ownership structure: a comparison of United States andJapanese manufacturing corporations’’, Financial Management, Vol. 15, pp. 5-16.

Marsh, P. (1982), ‘‘The choice between equity and debt: an empirical study’’, Journal of Finance,Vol. 37, pp. 121-44.

Miller, M.H. (1977), ‘‘Debt and taxes’’, Journal of Finance, Vol. 32, pp. 261-76.

Modigliani, F. and Miller, M. (1958), ‘‘The cost of capital, corporate finance and the theory ofinvestment’’,American Economic Review, Vol. 48, pp. 261-97.

Modigliani, F. and Miller, M. (1963), ‘‘Corporate income taxes and the cost of capital: a correction’’,American Economic Review, Vol. 53, June, pp. 443-53.

Myers, S.C. (1977), ‘‘Determinants of corporate borrowing’’, Journal of Financial Economics, Vol.5, pp. 147-75.

Myers, S.C. (1984), ‘‘The capital structure puzzle’’, Journal of Finance, Vol. 39, July, pp. 575-92.

Myers, S.C. and Majluf, N.S. (1984), ‘‘Corporate financing and investment decisions when firmshave information that investors do not have’’, Journal of Financial Economics, Vol. 13,pp. 187-221.

Rajan, R.G. and Zingales, L. (1995), ‘‘What do we know about capital structure: some evidencefrom international data’’, Journal of Finance, Vol. 50, pp. 1421-60.

Wald, J. (1999), ‘‘How firm characteristics affect capital structure: an international comparison’’,The Journal of Financial Research, Vol. XXII No. 2, pp. 161-187.

Corresponding authorNickolaos Eriotis can be contacted at: neriot@econ.uoa.gr; nikolaos.eriotis@aueb.gr

To purchase reprints of this article please e-mail: reprints@emeraldinsight.comOr visit our web site for further details: www.emeraldinsight.com/reprints

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