Description
Financial Study on Capital Structure and Institutional Setting: A Decomposition and International Analysis:- The term "level of analysis" is a social sciences term that points to the location, size, or scale of a research target. "Level of analysis" is distinct from the term "unit of observation" in that the former refers to a more or less integrated set of relationships while the latter refers to the distinct unit from which data have been or will be gathered.
Financial Study on Capital Structure and Institutional Setting: A Decomposition and International Analysis
The legal and institutional setting is more and more influential in firms' financial decisions. Our article analyses firms' capital structure in an international framework in order to assess the different level of debt use across countries and to identify both common and differential explanatory factors. Although the level of financial leverage is quite different, the factors that have traditionally driven capital structure decisions have much in common in all the legal and institutional settings. The performance and size of the firm, the assets tangibility and the growth opportunities have a relevant but differential effect across the different institutional systems. Consequently, our results suggest that the legal and institutional system of each country does not only affect firms' capital structure but also creates the conditions to explain a differential effect of the common determinants of firms' financial choices.
I. Introduction The capital structure of firms has been the core of an academic debate for a long time. This debate has run parallel with the research about the influence of the legal and institutional setting on firms' financial decisions. Laws and specially investor protection, have been proved to have a great influence on the corporate system. In this sense, the analysis of the origins of the legal system can help to explain institutional factors such as corporate governance, the relative importance of capital markets and the development of some industries (La Porta et al., 2000). This article is based on both fields and aims to analyse the capital structure of firms in the framework of the legal system of each country. It is a suggestive approach because, although most
of the research focuses on developed countries (Rajan and Zingales, 1995), there are notable institutional differences between them which should be taken into account. This is the main contribution of our article, as we aim to study the factors determining capital structure in an international framework. We do not simply wish to check the different financial leverage of the different countries or groups of financial systems; we are also interested in analysing different measures of capital structure in order to elucidate how the impact of the factors that have traditionally explained capital structure is conditioned by that legal and institutional framework. Our results stress the significant differences in financial leverage between countries and between legal frameworks and how those differences are not due to different factors but to their
1851
1852
differential impact. More specifically, we find that the firms belonging to different financial systems show big differences in the level of debt and especially in the maturity structure of debt. In addition, we document the influence of law enforcement and the quality of accounting on firms' capital structure. Our article can be divided into four sections. Section II looks at the main literature on this topic and we introduce the theoretical background on which the empirical analysis is grounded. Section III presents the data, the definition of the variables to be used and the statistical methodology. In Section IV, the results of the empirical estimation are reported and discussed. Section V includes the most relevant conclusions and some directions for future research. show big differences in the extent to which the banks are present in the core of the system, there will also be big differences in the capital structure of the firms (Rajan and Zingales, 1995, 1998). In spite of the wide support for this classification, the limitations of this criterion have been underlined in recent years (Corbett and Jenkinson, 1998; La Porta et al., 2000). The banks vs. markets scheme relies on two differentiated levels of financial leverage and Anglo-Saxon firms are usually less leveraged than their Continental counterparts (Rutherford, 1988; Mayer, 1990). Nevertheless, this assertion has been proved not to be completely exact, especially regarding some countries, such as Germany, which are supposed to belong to the bank oriented model (Rajan and Zingales, 1998). This is why another classification scheme has arisen. Instead of grounding on the importance of markets and financial intermediaries, this new criterion is based on the legal origins of each country (La Porta et al., 1997, 1998; Levine, 1998; Levine et al., 2000). Basically, countries are classified into two main groups: common law countries and civil law countries. While the law is made by judges in common law countries, legal scholars play a prominent role in civil law countries. Civil law countries can be further divided into three origins: the French, Scandinavian and German traditions. Legal origins determine the characteristics of each system. In fact, creditors' and shareholders' rights, law enforcement, the quality of accountancy, ownership concentration and per capita wealth are quite different, conditional upon investor protection. Investors have the best legal protection in common law countries and the worst in the French civil law countries. Similarly, law enforcement and the quality of accountancy are higher in the common law and in the Scandinavian civil law countries than in the others (La Porta et al., 1997). Financial and institutional systems are relevant because investor protection has a positive effect on the development of capital markets (both equity and debt markets) and, consequently, it affects firms' financial strategy. For instance, weak investor protection implies a more concentrated ownership and control structure (Himmelberg et al., 2004), whereas good protection indirectly leads to the growth of production and productivity through a more efficient resources allocation (La Porta et al., 2000). Likewise, the development of the banking system is positively related to the protec- tion of creditor rights (Levine, 1998) and economic development is enhanced by the institutional
II. Capital Structure and the Legal System Capital structure has been one of the most controversial topics in finance and there are plenty of articles which have tried to identify its determining factors (Barnea et al., 1985; Harris and Raviv, 1991; Colombo, 2001). Although not exhaustively, the literature has pointed to some of these factors, such as firm size (Marsh, 1982; Jalilvand and Harris, 1984), firm performance (Kester, 1986; Titman and Wessels, 1988), collateral (Bradley et al., 1984; Berger and Udell, 1995), growth opportunities (McConnell and Servaes, 1995), the ownership structure (Kim and Sorensen, 1986; Bergstro¨m and Rydqvist, 1990), debt tax shields (Titman and Wessels, 1988) and assets structure (Balakrishnan and Fox, 1993), etc. In recent years, the literature has paid special attention to the influence of the legal and institutional framework on corporate finance. The financial decisions of firms are not isolated from the institutional characteristics. In fact, the legal and institutional setting creates a net of relations between firms and financial institutions. From this point of view, financial systems have traditionally been classified into two main groups, depending on theorientation or importance of financial intermediaries (Allen, 1995; Allen and Gale, 2001). There is a Continental or bank-oriented system in which banks play a prominent role as financial channels from the ultimate lenders to the ultimate borrowers. It is the dominating system in Japan and in most Continental European countries such as Germany, France, Italy, Spain, etc. There is also an Anglo-Saxon or marketoriented system (e.g. USA, United Kingdom, etc.) in which banks are not so important and financial functions are directly performed by capital markets. Since both systems
framework
(i.e.
institutional
support
and
Capital structure and institutional setting
economic freedom) as found by Assane and Grammy (2003). Hence, the legal framework of each country especially law enforcement and investor protection - has been proved to affect corporate finance (Fabbri, 2001). For instance, Giannetti (2003) has shown how intangible assets - which could be the most difficult assets to fund are more easily funded when creditor rights are better protected and that a lower development of capital markets forces firms to use more debt. In addition, Storey (1994) has proved that bank financing is affected by the legal status of the firm. These two academic fields - namely, capital structure theory and the international comparison of financial systems - are the backbone of our article since we aim to study the capital structure of an international sample of firms following a decompositional analysis. According to Booth et al. (2001), our research is 2-fold: firstly, in a descriptive approach, we attempt to discover whether capital structure shows significant differences across countries and, secondly, we test whether the factors determining firms' financial decisions have a different influence depending on the legal and institutional framework. Although the classification scheme according to the institutional environment does not necessary imply any prediction concerning financial leverage but about internal and external finance (La Porta et al., 1997), there is evidence of the different level of debt and, more specifically, of the different debt maturities across countries. Broadly speaking, firms in the civil law countries usually have more debt and shorter maturity of debt than their common law counterparts (Demirgu¨c-Kunt and Maksimovic, 1999; Fan et al., 2003). Consequently, we could hypothesize that our results are supposed to show higher financial leverage and shorter maturity of debt in civil-law countries relative to common law countries.
1853
Our methodology follows two steps. The first step is broadly descriptive and aims both to compare the level of debt across countries or across legal systems and to test the existence of possible significant differences through the analysis of variance (hereinafter ANOVA). As stated by Rajan and Zingales (1995), both the level of financial leverage and its sensibility to the influencing factors can critically depend on the kind of measure of financial leverage. Therefore, we propose a set of different measures of capital structure and we decompose them into their basic components. The second methodological stage is mainly explana- tory and aims to test the impact of some factors on capital structure. In this phase, we are interested in knowing to what extent the international differences can be explained by a different impact of these factors. Our sample includes data from Austria and Germany as civil law countries with the German tradition, from Canada, the USA and the United Kingdom as common law countries and from Italy, France, Spain, the Netherlands and Belgium as civil law countries with the French tradition. The final distribution by countries and corporate systems is reported in Table 1. We use three main measures of capital structure as suggested by Rajan and Zingales (1995) and, especially, by Bevan and Danbolt (2002), so that we can compare our results with those of the earlier authors. The first measure is a general indicator of financial leverage and includes any kind of debt (both financial and commercial debt). We defined B1 as the ratio of financial debt (FD), i.e. costly debt, plus commercial debt (CD) to total assets at book value (TA).
B1 ¼ FDTACD þ
III. Data and Methodology Our database set is Compustat. As is widely known, Compustat gathers financial information with high reliability from a large number of firms. Given the high number of countries (and the disparities among them in terms of accounting rules), we have centred on balance sheets and income statements from a sample of 10 countries throughout 1997-2002 (Table 1). These firms can be divided into three of the four aforementioned main institutional settings.
The second measures explicitly focuses on costly debt and excludes commercial debt. Consequently we define B2 as total financial debt to book total assets ratio. B2 ¼ FD TA The last variable is informative of the relation between debt and costly funds (both implicit and explicit cost). These costly funds are costly debt and equity. The main difference between this ratio and previous ones is the exclusion of commer- cial debt and some elements which are quite difficult
1854
Table 1. Sample distribution across countries Number of firms Countries Number of firms
In this equation, i sub-index stands for the individual and t sub-index for the time. .i is the fixed-effects term which is firm specific and " it is the random component which is supposed to introduce all the remaining factors potentially affecting capital structure. The fixed-effects term introduces firm specific factors which can be correlated with the set of independent variables and whose omission could bias the results of the estimation. This fixed effect or unobservable and constant heterogeneity can not only be identified, but also treated by panel data procedures (Arellano, 2003; Baltagi, 2004). Panel data is basically a multivariate regression analysis along with the use of the Hausman test to detect the existence of these underlying individual effects and their correlation with the explanatory variables. When the Hausman test suggests the rejection of the null hypothesis of no correlation between the fixed effects term and the independent variables, the withingroups estimation provides consistent estimators. If the null hypothesis is not rejected, a random model or generalized least squares provides consistent and asymptotically efficient coefficients. Nevertheless, there is a lot of literature suggesting the endogeneity of some right-hand side variables, so we should control for this possible endogeneity (Cho, 1998; Demsetz and Villalonga, 2001). There are a number of procedures to deal with this problem and we will stress the generalized method of moments GMM (Arellano and Bond, 1991; Mairesse and Hall, 1996). The GMM is based on the use of instrumental variables according to the structure of available lagged variables which are supposed to be endogen- ous. By counting on more instruments than variables to be estimated, GMM provides more efficiently estimated coefficients. This is why, in our last stage, we report the results from the GMM estimation for MTB and PROF in order to test the robustness of our previous results.
Countries
United States 2827 United Kingdom 675 Canada 464 4066 Total common law Germany 671 Austria 92 Total German civil law 763
Holland 185 Belgium 102 Italy 209 564 France 103 Spain Total French 1163 civil law Total 5992
to classify such as deferred taxes, minority interests, nontaxable reserves, etc.1 B3 ¼ ðFDFDBVÞ þ Since an all-inclusive explanation of capital struc- ture is beyond the scope of this article and we simply aim to compare the basic issues of corporate finance across different legal and institutional frameworks, we limit our attention to four variables potentially driving the capital structure decisions of firms (Rajan and Zingales, 1995; Bevan and Danbolt, 2002; Bhaduri, 2002): growth opportunities, firm size, firm performance and assets tangibility. Growth opportunities, according to McConnell and Servaes (1995) and Lasfer (1995), are proxied by MTB or the market-to-book ratio (book value of debt plus market equity value to book total assets ratio). The size of the firm is measured through the log of the firm's turnover and the performance of the firm is measured with the EBIDTA (earnings before interest, depreciation, taxes and amortizations) to total assets ratio. Assets tangibility is measured through the assets with a physical existence (PA) to total assets ratio. These definitions can be expressed as follows: MTB ¼ ðTA À BV þ VMÞ PROF ¼ EBITDA TA TA PA LOGSALE ¼ LnðTurnoverÞ TANG ¼ TA The explanatory analysis is run through regression analysis with the panel data method. The model to be tested can be expressed as follows: Lit ¼ ff þ fi1MTBit þ fi2LOGSALEit þ fi3PROFit þ fi4TANGit þ .i þ "it
1
IV. Results The first step is a test of possible significant differences for the measures of capital structure among different legal systems. Results are reported in Table 2 and, although they are perhaps too detailed, they show a common and persistent pattern for B1, B2 and B3 across legal systems: whereas firms in the French civil law countries are the most
We have defined three measures of capital structure analogously to B1, B2 and B3 with market values instead of book values. Results are not reported for simplicity but are fully consistent.
Capital structure and institutional setting
Table 2. Mean value of debt conditional upon the legal system Mean B1 Anglo French Germ Total Anglo Frenc h Germ Total Anglo Frenc h Germ Total 0.3279 0.3894 0.2839 0.3343 0.2368 0.2313 0.1896 0.2302 0.3276 0.3780 0.3186 0.3360 LTDTA Mean
1855
classification scheme seems to lose importance relative to the legal roots and the institutional development criterion for classifying countries and explaining financial decisions (La Porta et al., 2000). We should note that there are some discrepancies in capital structure (for instance, the ranking based on B2 vs. the ranking based on B1 and B3). In order to elucidate this issue, we decompose the total debt to total assets ratio (B1) as a function of the maturity structure (Bevan and Danbolt, 2002; Ozkan, 2002). So, we have defined LTDTA as the proportion of long term debt over total assets and STDTA as the proportion of short term debt over total assets. Even short term debt can be divided into suppliers or commercial debt (COMTA) and the other short term debt (OSTDTA).2 Results exhibit big differences according to the institutional framework. Consistent with our expec-
0.1946 Anglo French 0.1319 Germ 0.0968 Total 0.1714 Anglo 0.1334 B2 STDTA 0.2575 Frenc 0.1871 h 0.1629 Germ 0.0912 Total B3 COMTA 0.1581 Anglo 0.0943 Frenc 0.1041 h 0.0422 Germ OSTTA 0.0994 Total 0.0928 Anglo 0.0588 Frenc h Germ Total Notes: The whole number of observations is 18 003 for common law countries, 4854 for the French tradition of civil law countries and 3029 from the German tradition of the civil law countries. B1 and M1 are all-items including definitions of capital structure, B2 and M2 includes just costly debt and B3 and M3 are scaled by costly liabilities. LTDTA and STDTA stand for long-term debt or short-term debt to total assets, respectively. COMTA stands for commercial debt to total assets ratio and OSTTA for other short-term debt to total assets ratio.
leveraged, their German civil law counterparts are the least prone to debt. Although these results hold for the three measures of financial leverage, they are inconsistent with our expectations, since civil law firms were hypothesized to have more debt than common law firms. Consequently, new analyses are required to solve this conflict. The next stage is an ANOVA to test the extent to which one can assert that different institutional and legal settings have different mean values of capital structure. ANOVA results, reported in Table 3, are quite significant and show that we can reject the equality of means across the three groups with aconfidence level higher than 99%. Nevertheless, this evidence requires a more detailed development with bilateral comparisons between pairs of systems as reported in Table 4. This table shows that, on average, the level of financial leverage is significantly different across groups of countries and corroborates the fact that firms from the French tradition of civil law countries are the most leveraged, whereas their German civil law counterparts are those with the least leverage. Consequently, the market vs. banks
tations and with previous literature, while Anglo-Saxon firms are those with the highest long term debt ratio (19.4%), the German civil law firms are those with the lowest long term debt (9.6%). On the contrary, if we focus on the short term debt, we can see how French civil law firms are the most leveraged companies whereas AngloSaxon firms are the least ones. In turn, different kinds of debt seem to have an asymmetric role: common law firms appear to be more prone to long term debt whereas civil law firms tend to borrow to short term. Among the possible explanations to these results, we could cite that both the legal protection of investors and the quality of legal enforcement foster long term lending relations, as well as institutional investors and the activity in capital markets - more often in common law countries than in civil law ones. Once we have checked the differences between legal systems in terms of capital structure, we can test whether the factors determining firms' financial choices are responsible for those differences. We have made capital structure depend on four of the factors which are most usually supposed to affect a firm's finance: growth opportunities (MTB), firm size (LNSALES), firm performance (EBITDA) and assets tangibility (TANG). The results of the regression analysis with the method of panel data are reported in Tables 5 and 6. For each explanatory variable and for each measure of capital structure four estimations are provided. The first one has been run over the entire sample while the second, the third and the fourth are estimated for the Anglo-Saxon, the French and the German tradition of civil law system respectively. In general terms, the results reported in Table 5 show a common pattern for the four explanatory
2
All these measures have been scaled by total assets.
1856
Table 3. Analysis of the variance depending on the legal system Square mean B1 B2 B3 Between-groups Within-groups Between-groups Within-groups Between-groups Within-groups 11.585 0.034 2.886 0.030 5.376 0.057 F-test 336.06 94.35 93.90 p-value 0.000 0.000 0.000 LTDTA STDTA COMTA OSTTA Square mean 17.048 0.025 30.462 0.013 8.725 0.006 8.238 0.005 F-test 673.608 2273.038 1288.853 1427.425 p-value 0.000 0.000 0.000 0.000
Table 4. Bilateral post hoc tests System I B1 Anglo French Anglo French Anglo French System II French German German French German German French German German Difference À0.061 0.044 0.105 0.005 0.047 0.041 À0.050 0.009 0.059 p-value 0.000 0.000 0.000 0.055 0.000 0.000 0.000 0.056 0.000 LTDTA Difference 0.0062 0.0097 0.0035 À0.124 À0.053 0.070 À0.066 À0.003 0.063 À0.057 À0.050 0.006 p-value 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.052 0.000 0.000 0.000 0.000
B2
STDTA
B3
COMTA
OSTTA
variables in all the legal systems and for all the measures of capital structure. Although we do not aim to explain corporate finance decisions in each institutional framework but simply to show the common and the distinctive features, we should try to provide some justification for these results. Growth opportunities and firm performance are proved to have a negative and significant relation with financial leverage, whereas the size of the firm and the assets tangibility is positively related. These results can be explained by the link between the size of the firm and the asymmetric information in capital markets (Ojah and Manrique, 2005). Since large companies are usually better known in capital markets, there is less asymmetry between a firm'sinformed managers and investors, so that large firms can more easily borrow from capital markets. The negative relation between debt and firm perfor- mance has been widely documented by previous research into the pecking order theory of capital structure (Myers, 1977; Myers and Majluf, 1984). Nevertheless, the effect of growth opportunities and assets tangibility requires further explanation. As far as growth opportunities are concerned, their negative impact - when significant - is noticeable and coherent with previous research (Bevan and Danbolt, 2002). A more detailed analysis
as a function of maturity structure confirms that firms with more growth opportunities rely on commercial debt because this kind of debt does not impose such constraining covenants as other types of credit (Barclay and Smith, 1999). Table 5 also shows how assets tangibility is positively related to debt for all the measures of capital structure and in all the legal systems. This result can be explained on the basis of tangible assets as collateral: the more important the tangible assets are the more collateral the firm puts and, consequently, the lower the interest rate is. Notwithstanding, Table 6 shows remarkable differences conditional upon the kind of debt: whereas TANG is positively related to LTDTA, it is negatively related to STDTA. This means that long term debt is likely to fund long term assets (which can be put as collateral for long term debt) and short term debt will fund current assets. To sum up, our results up to this point show: (1) Noticeable and quite consistent differences in the level of financial leverage across the firms from different legal systems; (2) Financial leverage is affected by the same factors which have traditionally been supposed to explain capital structure. But, if this is the case, one should question how the same factors could produce such large differences across the
Table 5. Factors affecting capital structure
B2 French 0.002 0.065 0.055 0.0657 164.47** 62.57** 377.63** 341.22** 158.53** 0.0805 0.2188 0.2750 0.0890 (2.89)** 0.163 (7.73)** 0.197 (27.12)** 0.192 (22.35)** 0.186 (10.37)** (17.55)** 0.064 (11.93)** 0.028 (17.77)** 0.021 (11.60)** 0.053 (15.37)** 0.044 0.203 0.0710 46.74** (1.91)* À0.002 (À1.84)* À0.001 (À4.04)** À0.001 (À0.15) (À2.50)** À0.000 German Total Anglo French German B3 Total Anglo (À2.76)** À0.001 (8.37)** (9.92)** 0.049 0.215 0.2221 390.14** (22.64)** (21.47)** 0.036 0.207 0.2864 366.13** (À1.63) (14.75)** (18.10)*** French 0.001 0.094 0.137 0.1420 137.88** (0.54) (18.39)** (5.18)** German À0.001 (À0.64) 0.088 0.284 0.1324 56.93** (10.93)** (9.05)**
B1
Total
Anglo
MTB
À0.0007 (À1.62)
À0.0001 (À0.35)
À0.002 (À2.07)** À0.001
LOGSALE
0.036
(22.38)**
0.028
(14.80)**
PROF
À0.081
(22.49)** À0.092 (À20.90)** À0.290 (À16.04)** À0.025 (À3.70)** À0.070 (À20.34)** À0.080 (À18.85)** À0.259 (À15.71)** À0.019 (À2.93)** À0.106 (À22.32)** À0.119 (À20.98)** À0.422 (À17.33)** À0.034 (À3.33)**
TANG
0.123
(16.38)**
0.122
(13.73)**
Adj. R2
0.1443
0.1950
Hausman
509.79**
449.59**
test
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients.
1858
Table 6. Factors determining capital structure
STDTA Anglo À0.001 (À1.74)* 0.018 (10.10)*** À0.053 (À12.69)*** 0.172 (20.17)*** 0.2821 333.62*** 0.001 (0.59) 0.016 (13.91)*** À0.034 (À12.77)*** À0.051 (À9.20)*** 0.0673 742.44*** À0.001 (À0.29) 0.031 (10.12)*** À0.131 (À8.83)*** 0.194 (12.07)*** 0.1695 60.99*** À0.001 (À0.45) 0.011 (2.79)** À0.016 (À3.14)*** 0.149 (9.41)*** 0.1475 30.48*** French German Total Anglo 0.001 (1.95)* 0.009 (7.46)*** À0.038 (À12.76)*** À0.050 (À8.33)*** 0.0734 352.27*** French 0.003 (2.42)** 0.033 (10.12)*** À0.159 (À10.14)*** À0.139 (À8.2)*** 0.0674 110.62*** German À0.002 (À1.67)* 0.053 (10.89)*** À0.009 (À1.48) 0.014 (0.74) 0.0482 70.54***
LTDTA
Total
MTB LOGSALE PROF TANG Adj. R2
Hausman test OSTTA 0.001 (6.12)*** 0.006 (10.50)*** À0.011 (À7.79)*** À0.070 (À23.28)*** 0.1302 391.45** 0.003 (3.76)*** 0.011 (5.73)*** À0.030 (À3.17)*** À0.130 (À12.54)*** 0.1797 59.42*** 0.001 (0.36) 0.020 (8.54)*** À0.006 (À1.91)* À0.039 (À4.22)*** 0.1246 45.04** À0.001 (À2.86)** 0.008 (7.81)*** À0.023 (À10.19)*** 0.022 (4.61)*** 0.0013 285.87***
À0.001 (À2.23)** 0.019 (13.14)*** À0.047 (À14.16)*** 0.174 (25.13)*** 0.2704 412.61***
COMTA
MTB LOGSALE PROF TANG Adj. R2
Hausman test
0.001 (5.89)*** 0.008 (13.69)*** À0.010 (À7.54)*** À0.073 (À25.35)*** 0.1580 617.32**
À0.001 (À1.27) 0.003 (2.56)** À0.026 (À10.24)*** 0.020 (3.80)*** 0.0049 83.56***
0.001 (0.14) 0.021 (7.79)*** À0.128 (À9.67)*** À0.008 (À0.60) 0.0025 82.76***
À0.002 (À2.03)** 0.032 (7.31)*** À0.003 (À0.59) 0.053 (3.10)*** 0.0100 41.34***
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients.
Capital structure and institutional setting
systems or, more precisely, across countries. This is why, in the subsequent analysis, we introduce two country specific characteristics which are related to the legal and institutional framework: law enforce- ment and the quality of accounting. Based on data from La Porta et al. (1998), we have defined two dummy variables (DEL and DQA) that equal 1 if law enforcement or the quality of accounting is above the mean of the sample.3 These dummy variables have interacted with the four explanatory variables in order to test if they have differential effects conditioned by the law enforcement and the quality ofaccounting. Results are displayed in Table 7. For the sake of simplicity, we will just comment on the most general and common features instead of a too detailed explanation of the results. The interacting variables are quite significant, so we can assert that growth opportunities, the size and performance of the firm and the assets tangibility have a different effect depending on those two characteristics. Two results, nevertheless, deviate from this pattern: neither the interaction of MTB and the dummy of quality accounting nor the interaction of TANG and quality accounting seem to have a significant impact. Excluding these exceptions, Table 7 is interesting because it suggests that different levels of leverage are not per se a result of the legal environment, but that the legal setting creates the conditions so that those factors have a differential impact. We have run some additional regressions in order to check the robustness of our results. Firstly, we have defined a dummy variable for each institutional setting. These dummies (DFC and DGC 4) have interacted with the four aforementioned explanatory variables to test possible differential effects depending on the legal system. Results are shown in Table 8 and are consistent with previous ones: the four variables continue to be significant and, in addition, the interacting variables are also statistically significant, with the sole exception of MTB in some estimations. Since a very exhaustive explanation might obscure the general meaning of the regression, just a general comment can be suitable: the high significance of the interacting variables allows us to infer that the four determinants of capital structure have a differential influence in each legal system. Another sensitivity analysis takes into account the possible endogeneity of the explanatory variables.
3
1859
As some authors have pointed out (Demsetz and Lehn, 1985; Jensen, 1986; McConnell and Servaes, 1995), some firms' characteristics that we have assumed exogenous could be affected by the firms' capital structure. Thus, we have replicated the previous regressions with the GMM to control for the potential endogeneity of the explanatory variables (Table 9). Results basically remain the same as the previous ones with two small exceptions: PROF is no longer as significant as it was and MTB coefficients are inconsistent for some estimations. Notwithstanding, the Hansen test, which aims to control the overidentification restrictions - and, consequently, the validity of instruments - does not support the accuracy of the instruments. Additionally, the second order serial correlation would advise some caveat in analyzing GMM results because the weakness of the instruments reduces the efficiency of the estimations and increases the possible bias.
V. Concluding Remarks Recent research has shown the influence of the legal and institutional setting on firms' financial decisions. These decisions are no longer due just to firms' value enhancement strategies but are also affected by the legal framework in which firms operate. Our article joins analyses of firms' capital structure in an international framework in order to test differences across countries or legal systems and to find common points in the factors potentially affecting capital structure. We begin with a division of the countries according to the characteristics of the corporate system. The traditional bank-oriented vs. market-oriented classification scheme is no longer practical enough and we need to use more precise criteria based on the legal origin of institutions. In fact, as a first conclusion of our article, we can assert that the global consideration of the bank oriented system is inexact since it includes a number of countries with fairly different corporate systems. When we decom- pose capital structure in a set of different measures, we find that the firms in the French tradition of civil law countries are more leveraged than their common law counterparts and we also find that the firms in the
The countries with the best law enforcement in our sample are Canada, The United States, Belgium, Holland and Austria, whereas the countries with the best quality of accounting are Canada, The United Kingdom, The United States, France, Holland and Spain. It is interesting to note that this classification differs from the civil vs. common law, so that the dummy variables do not measure the legal tradition but these two features of the legal and institutional setting. 4 Dummy for the French civil and the German civil countries.
Table 7. Factors determining capital structure with law enforcement and the quality of accounting
B2 À0.0017 (À1.35) 0.0008 0.0851 À0.2029 (À16.32)*** 0.2239 À0.002 À0.0145 0.0179 À0.0029 (0.3) (À7.32)*** (À7.62)*** (À1.21) 0.0741 536.43*** COMTA (0.71) 0.052 (14.66)*** À0.0151 (À2.91)*** À0.0068 (À0.47) 0.0007 (0.97) À0.0153 (À6.62)*** 0.0464 (7.14)*** À0.0159 (À1.57) (0.71) 0.0014 À0.0257 À0.0217 À0.024 0.0597 750.67*** (1.6) (À7.82)*** (À4.14)*** (À1.77)* 0.1894 450.07*** 0.0494 731.96*** 0.054 214.98** 0.0585 255.6*** À0.0011 (À1.14) 0.0017 0.0139 (4.82)*** (11.91)*** À0.0132 (À3.66)*** À0.0913 (À18.47)*** 0.0008 (1.72)* 0.0211 (11.39)*** À0.0073 (À2.7)*** À0.0493 (À6.58)*** (À0.2) À0.0002 (À0.18) 0.0019 (0.41) À0.0382 (À5.05)*** 0.0248 0.1135 462.59*** OSTTA À0.0012 0.0188 (À2.08)** (9.67)*** À0.0629 (À10.46)*** 0.0314 (3.82)*** À0.0019 (À2.34)** 0.0309 (10.02)*** À0.0078 (À1.73)* 0.0425 (3.41)*** (1.27) (1.91)** (À4.36)*** (À1.98)** (13.16)*** 0.2419 (9.36)*** À0.0461 (À4.93)*** (21.19)*** 0.0937 (14.66)*** (0.65) 0.0704 (14.63)*** À0.0338 (À4.81)*** À0.1248(À13.83)*** 0.2071 (16.76)*** 0.195 (10.39)*** À0.003 (À2.07)** À0.051 (À10.68)*** 0.1139 (8.46)*** À0.0196 (À0.93) (0.89) À0.0498 À0.0828 À0.0338 0.1207 434.08*** 0.0006 (À1.26) (À8.59)*** (6.58)*** (À1.23) 0.0012 À0.0237 (À4.81)*** À0.0599 (À7.59)*** 0.0008 0.0479 540.37*** STDTA À0.0007 (À0.55) 0.0005 0.0327 (14.6)*** À0.0761(À10.99)*** À0.0599 (À6.33)*** (À1.75)** (À5.31)*** (0.87) (4.36)*** 0.0004 À0.0141 (À7.15)*** À0.0045 (À1.44) À0.0284 (À3.49)*** 0.1432 530.14*** 0.0184 À0.0187 (À2.88)*** 0.1525 À0.0007 À0.0074 0.0034 0.0266 (1.76)* (À10.53)*** (À4.35)*** (À3.36)*** 0.2816 412.65*** (8.47)*** (4.13)*** (0.04) 0.1457 À0.0013 À0.0298 0.0643 À0.0188 (1.14) (À7.39)*** (À7.88)*** (À1.31) 0.0447 635.1*** (7.48)*** À0.0265 (À3.91)*** 0.0489 (16.78)*** 0.0493 (10.63)*** À0.0007 (À0.8) À0.0026 (À2.09)** À0.0017 (À1.02) B3
B1
MTB
0.001
(1.08)
LOGSALE
0.0628
(20.76)***
PROF
À0.138
(À14.74)***
TANG
0.1158
(9.04)***
MTB*DLE
À0.002
(À1.89)*
LOGSALE*DLE
À0.0372 (À10.33)***
PROF*DLE
0.0677
(6.67)***
TANG*DLE
0.0078
(0.49)
MTB*DQA
0.0016
LOGSALE*DQA
À0.0378
PROF*DQA
À0.0645
TANG*DQA
À0.0276
Adj. R2
0.0512
Hausman test
646.91***
LTDTA
MTB
0.0005
(0.6)
LOGSALE
0.0302
(10.78)***
PROF
À0.062
(À7.15)***
TANG
0.1758
(14.82)***
MTB*DLE
À0.0001
(À0.07)
LOGSALE*DLE
À0.0226
(À8.52)***
PROF*DLE
0.0498
(6.64)***
TANG*DLE
0.0107
(0.92)
MTB*DQA
0.0018
LOGSALE*DQA
À0.0397
PROF*DQA
À0.0263
TANG*DQA
À0.0523
Adj. R2
0.0883
Hausman test
494.81***
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients. Independent variables have interacted with a dummy of law enforcement (DLE) and with a dummy of the quality of accounting (DQA).
Table 8. Factors determining capital structure with dummy variables for institutional settings
B2 B3 LTDTA STDTA COMTA OSTTA (À1.21) (2.44)** (À9.76)*** (3.62)*** (0.53) (6.29)*** (À7.6)*** (À1.87)* (À1.88)* (7.85)*** (4.37)*** (2.25)**
Capital structure and institutional setting 1861
B1
MTB LOGSALE PROF TANG MTB*DFC LOGSALE*DFC PROF*DFC TANG*DFC MTB*DGC LOGSALE*DGC PROF*DGC TANG*DGC Adj. R2
Hausman test
À0.0002 (À0.36) 0.0285 (15.35)*** À0.092 (À21.68)*** 0.1222 (14.25)*** 0.0028 (1.67)* 0.0368 (7.82)*** À0.1981 (À9.48)*** À0.067 (À2.81)*** À0.0022 (À1.52) 0.0364 (6.15)*** 0.0666 (8.03)*** 0.041 (1.75)* 0.068 658.3***
À0.0014 (À2.61)*** 0.0216 (12.08)*** À0.0804 (À19.63)*** 0.1926 (23.28)*** 0.0012 (0.74) 0.032 (7.06)*** À0.1793 (À8.9)*** À0.0065 (À0.28) À0.0012 (À0.89) 0.0225 (3.94)*** 0.0608 (7.6)*** 0.0104 (0.46) 0.0905 500.73*** À0.0012 (À1.64)* 0.0367 (14.9)*** À0.1193 (À21.19)*** 0.2079 (18.29)*** 0.0022 (1.00) 0.0581 (9.32)*** À0.3034 (À10.95)*** À0.0707 (À2.24)** À0.0001 (À0.03) 0.0518 (6.6)*** 0.0851 (7.74)*** 0.0763 (2.46)** 0.1096 521.0*** À0.001 (À1.88)* 0.0187 (10.88)*** À0.0538 (À13.67)*** 0.1726 (21.72)*** 0.0006 (0.4) 0.013 (2.99)*** À0.0774 (À4.00)*** 0.0221 (1.00) 0.0005 (0.38) À0.0073 (À1.34) 0.0375 (4.88)*** À0.0235 (À1.08) 0.1802 445.24****
0.0008 (1.87)* 0.0098 (7.14)*** À0.0382 (À12.21)*** À0.0505 (À7.97)*** 0.0022 (1.76)* 0.0237 (6.84)*** À0.1207 (À7.83)*** À0.0891 (À5.07)*** À0.0027 (À2.54)** 0.0437 (10.01)*** 0.029 (4.75)*** 0.0645 (3.73)*** 0.0824 594.38***
0.0012 (5.59)*** 0.0069 (9.59)*** À0.0117 (À7.12)*** À0.0704 (À21.28)*** 0.0016 (2.48)** 0.0048 (2.62)*** À0.0188 (À2.33)** À0.0604 (À6.58)*** À0.001 (À1.72)* 0.0139 (6.08)*** 0.0058 (1.81)* 0.0306 (3.39)*** 0.0430 569.75***
À0.0004 0.0029 À0.0266 0.02 0.0006 0.019 À0.1019 À0.0286 À0.0018 0.0298 0.0233 0.0339 0.0815 233.2***
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients. Independent variables have interacted with dummy variables of the French civil system (DFC) and the German civil system (DGC).
1862
Table 9. Factors determining capital structure (GMM estimation) B2 0.0048 (1.82)* 0.0019 (0.55) 0.0253 (5.66)*** 0.0433 (6.87)*** À0.0902 (À3.06)*** À0.0972 (À2.35)** 0.229 (13.23)*** 0.2502 (10.89)*** 276.9*** 245.5*** 13.28** 10.07 À7.37*** À6.29*** À5.04*** À4.84*** B3 LTDTA STDTA COMTA OSTTA 0.0008 (0.65) 0.0084 (4.37)*** À0.0344 (À2.55)** 0.0071 (0.68) 28.3*** 20.03*** À12.69*** À4.03***
B1
MTB 0.0085 (2.9)*** LSALES 0.0298 (5.56)*** PROF À0.0444 (À1.34) TANG 0.1862 (10.11)*** Wald test 164.6*** Hansen test 19.412*** AR(1) À7.15*** AR(2) À5.59***
0.0025 (1.09) 0.0064 (4.06)*** 0.0053 (5.69)*** 0.0169 (4.06)*** 0.0178 (5.64)*** 0.0037 (1.64)* À0.0504 (À2.46)** À0.0211 (À1.21) 0.0004 (0.08) 0.2133 (14.15)*** À0.0458 (À3.49)*** À0.0424 (À5.84)*** 273.6*** 59.7*** 96.6*** 7.52 14.97** 77.82*** À10.04*** À13.41*** À8.81*** À4.73*** À4.84*** À2.19**
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Wald test is a test of joint significance for all the variables. Hansen test of overidentification restrictions allows controlling the validity of instruments and follows a _ 2 distribution with so many degrees of freedom as the difference between the number of instruments and the number of regressors. AR(1) and AR(2) are tests or first and second order serial correlation.
Capital structure and institutional setting
German tradition of civil law are the least leveraged ones. More important than the level of leverage, our analysis reveals a clear difference in debt maturity so that, the more the rights of investors are protected, the longer the term of the debt becomes. Although the use of debt is different across legal systems, the factors traditionally thought as determi- nants of capital structure have much in common in different financial systems. Although the perfor- mance and the size of the firm, the assets tangibility and growth opportunities have a similar effect in the three scenarios, there are specific effects conditional on the legal systems. Consequently, our results suggest that the effect of the factors that have traditionally been considered determinants of capital structure depends on the legal and institutional setting and that these differential effects can explain international disparities in capital structure. Our research has also shown that the introduction of some variables concerning the legal protection of investors, such as the enforcement of the law and the quality of the financial information, can help to explain firms' financial choices.
1863
Barnea, A., Haugen, R. A. and Senbet, L. W. (1985) Agency Problems and Financial Contracting, Prentice Hall, New Jersey. Berger, A. N. and Udell, G. F. (1995) Relationship lending and lines of credit in small firm finance, Journal of Business, 68, 351-81. Bergstro¨m, C. and Rydqvist, K. (1990) The determinants of corporate ownership, Journal of Banking and Finance, 14, 237-53. Bevan, A. A. and Danbolt, J. (2002) Capital structure and its determinants in the UK - A decompositional analysis, Applied Financial Economics, 12, 159-70. Bhaduri, S. N. (2002) Determinants of capital structure choice: a study of the Indian corporate sector, Applied Financial Economics, 12, 655-65. Booth, L., Aivazian, V., Demirguc-Kunt, A. and Maksimovic, V. (2001) Capital structures in develop- ing countries, Journal of Finance, 56, 87-129. Bradley, M., Jarrell, G. and Kim, E. H. (1984) On the existence of an optimal capital structure: theory and evidence, Journal of Finance, 39, 857-80. Cho, M. H. (1998) Ownership structure, investment, and the corporate value: an empirical analysis, Journal of Financial Economics, 47, 103-21. Colombo, E. (2001) Determinants of corporate capital structure: evidence from Hungarian firms, Applied Economics, 33, 1689-701. Corbett, J. and Jenkinson, T. (1998) German investment financing: an international comparison, in Competition and Convergence in Financial Markets. The German and AngloAmerican Models (Eds) S. W. Black and M. Moersch, Elsevier, Amsterdam, pp. 85-124. Demirgu¨c-Kunt, A. and Maksimovic, V. (1999) Institutions, financial markets, and firm debt maturity, Journal of Financial Economics, 54, 295-336. Demsetz, H. and Lehn, K. (1988) The structure of corporate ownership: causes and consequences, Journal of Political Economy, 93, 1155-77. Demsetz, H. and Villalonga, B. (2001) Ownership structure and corporate performance, Journal of Corporate Finance, 7, 209-33. Fabbri, D. (2001) The legal enforcement of credit contracts and the level of investment. Center for Studies in Economics and Finance, Working paper, No. 57. Fan, J. P. H., Titman, S. and Twite, G. J. (2003) An International comparison of capital structure and debt maturity choices. EFA 2003 Annual Conference Paper No. 769, Available at SSRN: http://ssrn.com/ abstract¼423483. Giannetti, M. (2003) Do better institutions mitigate agency problems? Evidence from corporate finance choices, Journal of Financial and Quantitative Analysis, 38, 185-212. Harris, M. and Raviv, A. (1991) The theory of capital structure, Journal of Finance, 46, 297-355. Himmelberg, C. P., Hubbard, R. G. and Love, I. (2004) Investor protection, ownership, and the cost of capital. World Bank Policy Research Working Paper, No. 2834. Jalilvand, A. and Harris, R. S. (1984) Corporate behavior in adjusting to capital structure and dividend targets: an econometric study, Journal of Finance, 29, 127-44. Jensen, M. C. (1986) Agency costs of free cash flow, corporate finance and takeovers, American Economic Review, 76, 323-9.
Acknowledgements The authors are grateful to Alan Hynds, Giorgio Valente (co-editor of Applied Economics) and an anonymous referee for their comments.
References
Allen, F. (1995) Stock markets and resource allocation, in Capital Markets and Financial Intermediation (Eds) C. Mayer and X. Vives, Cambridge University Press, Cambridge, pp. 81-107. Allen, F. and Gale, D. (2001) Comparing Financial Systems, MIT Press, Cambridge. Arellano, M. and Bond, S. (1991) Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations, Review of Economic Studies, 58, 277-97. Arellano, M. (2003) Panel Data Econometrics, Oxford University Press, Oxford. Assane, D. and Grammy, A. (2003) Institutional framework and economic development: international evidence, Applied Economics, 35, 1811-7. Baltagi, B. H. (2004) Panel data: theory and applications, Physica-Verlag, New York. Balakrishnan, S. and Fox, I. (1993) Asset specificy, firm heterogeneity and capital structure, Strategic Management Journal, 14, 3-16. Barclay, M. J. and Smith, C. W. (1999) The capital structure puzzle: another look at the evidence, Journal of Applied Corporate Finance, 12, 8-20.
1864
Kester, C. W. (1986) Capital and ownership structure: a comparison of United States and Japanese manufacturing corporations, Financial Management, 15, 5-16. Kim, W. S. and Sorensen, E. H. (1986) Evidence on the impact of the agency costs of debt on corporate debt policy, Journal of Financial and Quantitative Analysis, 21, 131-44. La Porta, R., Lo´pez de Silanes, F., Shleifer, A. and Vishny, R. (1997) Legal determinants of external finance, Journal of Finance, 52, 1131-50. La Porta, R., Lo´pez de Silanes, F., Shleifer, A. and Vishny, R. (1998) Law and finance, Journal of Political Economy, 106, 1113-55. La Porta, R., Lo´pez de Silanes, F., Shleifer, A. and Vishny, R. (2000) Investor protection and Journal of Financial corporate governance, Economics, 58, 3-27. Lasfer, M. A. (1995) Agency costs, taxes and debt: the UK evidence, European Financial Management, 1, 265-85. Levine, R. (1998) The legal environment, banks, and longrun economic growth, Journal of Money, Credit and Banking, 30, 596-620. Levine, R., Loayza, N. and Beck, T. (2000) Financial intermediation and growth: causality and causes, Journal of Monetary Economics, 46, 31-77. Mairesse, J. and Hall, B.H. (1996) Estimating the productivity of research and development in French and US manufacturing firms: an exploration of simultaneity issues with GMM methods, in International Productivity Differences and their Explanation (Eds) K. Wagner and B. Van Ark, Elsevier, Amsterdam, pp. 285-315. Marsh, P. (1982) The choice between equity and debt: an empirical study, Journal of Finance, 37, 121-44. Mayer, C. (1990) Financial systems, corporate finance, and economic development, in Asymmetric Information, Corporate Finance and Investment (Ed.) R. G. Hubbard, The University of Chicago Press, Chicago, pp. 307-32. McConnell, J. J. and Servaes, H. (1995) Equity ownership and the two faces of debt, Journal of Financial Economics, 39, 131-57. Myers, S. C. (1977) Determinants of corporate borrowing, Journal of Financial Economics, 5, 147-75. 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, 13, 187-221. Ojah, K. and Manrique, J. (2005) Determinants of corporate debt structure in a privately dominated debt market: a study of the Spanish capital market, Applied Financial Economics, 15, 455-68. Ozkan, A. (2002) The determinants of corporate debt maturity: evidence from UK firms, Applied Financial Economics, 12, 19-24. Rajan, R. and Zingales, L. (1995) What do we know about capital structure? Some evidence from international data, Journal of Finance, 50, 1421-60. Rajan, R. and Zingales, L. (1998) Debt, folklore, and crosscountry differences in financial structure, Journal of Applied Corporate Finance, 10, 102-7. Rutherford, J. (1988) An international perspective on the capital structure puzzle, in New Developments in International Finance (Eds) J. Stern and D. Chew, Basil Blackwell, Nueva York. Storey, D. J. (1994) The role of legal status in influencing bank financing and new firm growth, Applied Economics, 26, 129-36. Titman, S. and Wessels, R. (1988) The determinants of capital structure choice, Journal of Finance, 43, 1-19.
doc_987400384.docx
Financial Study on Capital Structure and Institutional Setting: A Decomposition and International Analysis:- The term "level of analysis" is a social sciences term that points to the location, size, or scale of a research target. "Level of analysis" is distinct from the term "unit of observation" in that the former refers to a more or less integrated set of relationships while the latter refers to the distinct unit from which data have been or will be gathered.
Financial Study on Capital Structure and Institutional Setting: A Decomposition and International Analysis
The legal and institutional setting is more and more influential in firms' financial decisions. Our article analyses firms' capital structure in an international framework in order to assess the different level of debt use across countries and to identify both common and differential explanatory factors. Although the level of financial leverage is quite different, the factors that have traditionally driven capital structure decisions have much in common in all the legal and institutional settings. The performance and size of the firm, the assets tangibility and the growth opportunities have a relevant but differential effect across the different institutional systems. Consequently, our results suggest that the legal and institutional system of each country does not only affect firms' capital structure but also creates the conditions to explain a differential effect of the common determinants of firms' financial choices.
I. Introduction The capital structure of firms has been the core of an academic debate for a long time. This debate has run parallel with the research about the influence of the legal and institutional setting on firms' financial decisions. Laws and specially investor protection, have been proved to have a great influence on the corporate system. In this sense, the analysis of the origins of the legal system can help to explain institutional factors such as corporate governance, the relative importance of capital markets and the development of some industries (La Porta et al., 2000). This article is based on both fields and aims to analyse the capital structure of firms in the framework of the legal system of each country. It is a suggestive approach because, although most
of the research focuses on developed countries (Rajan and Zingales, 1995), there are notable institutional differences between them which should be taken into account. This is the main contribution of our article, as we aim to study the factors determining capital structure in an international framework. We do not simply wish to check the different financial leverage of the different countries or groups of financial systems; we are also interested in analysing different measures of capital structure in order to elucidate how the impact of the factors that have traditionally explained capital structure is conditioned by that legal and institutional framework. Our results stress the significant differences in financial leverage between countries and between legal frameworks and how those differences are not due to different factors but to their
1851
1852
differential impact. More specifically, we find that the firms belonging to different financial systems show big differences in the level of debt and especially in the maturity structure of debt. In addition, we document the influence of law enforcement and the quality of accounting on firms' capital structure. Our article can be divided into four sections. Section II looks at the main literature on this topic and we introduce the theoretical background on which the empirical analysis is grounded. Section III presents the data, the definition of the variables to be used and the statistical methodology. In Section IV, the results of the empirical estimation are reported and discussed. Section V includes the most relevant conclusions and some directions for future research. show big differences in the extent to which the banks are present in the core of the system, there will also be big differences in the capital structure of the firms (Rajan and Zingales, 1995, 1998). In spite of the wide support for this classification, the limitations of this criterion have been underlined in recent years (Corbett and Jenkinson, 1998; La Porta et al., 2000). The banks vs. markets scheme relies on two differentiated levels of financial leverage and Anglo-Saxon firms are usually less leveraged than their Continental counterparts (Rutherford, 1988; Mayer, 1990). Nevertheless, this assertion has been proved not to be completely exact, especially regarding some countries, such as Germany, which are supposed to belong to the bank oriented model (Rajan and Zingales, 1998). This is why another classification scheme has arisen. Instead of grounding on the importance of markets and financial intermediaries, this new criterion is based on the legal origins of each country (La Porta et al., 1997, 1998; Levine, 1998; Levine et al., 2000). Basically, countries are classified into two main groups: common law countries and civil law countries. While the law is made by judges in common law countries, legal scholars play a prominent role in civil law countries. Civil law countries can be further divided into three origins: the French, Scandinavian and German traditions. Legal origins determine the characteristics of each system. In fact, creditors' and shareholders' rights, law enforcement, the quality of accountancy, ownership concentration and per capita wealth are quite different, conditional upon investor protection. Investors have the best legal protection in common law countries and the worst in the French civil law countries. Similarly, law enforcement and the quality of accountancy are higher in the common law and in the Scandinavian civil law countries than in the others (La Porta et al., 1997). Financial and institutional systems are relevant because investor protection has a positive effect on the development of capital markets (both equity and debt markets) and, consequently, it affects firms' financial strategy. For instance, weak investor protection implies a more concentrated ownership and control structure (Himmelberg et al., 2004), whereas good protection indirectly leads to the growth of production and productivity through a more efficient resources allocation (La Porta et al., 2000). Likewise, the development of the banking system is positively related to the protec- tion of creditor rights (Levine, 1998) and economic development is enhanced by the institutional
II. Capital Structure and the Legal System Capital structure has been one of the most controversial topics in finance and there are plenty of articles which have tried to identify its determining factors (Barnea et al., 1985; Harris and Raviv, 1991; Colombo, 2001). Although not exhaustively, the literature has pointed to some of these factors, such as firm size (Marsh, 1982; Jalilvand and Harris, 1984), firm performance (Kester, 1986; Titman and Wessels, 1988), collateral (Bradley et al., 1984; Berger and Udell, 1995), growth opportunities (McConnell and Servaes, 1995), the ownership structure (Kim and Sorensen, 1986; Bergstro¨m and Rydqvist, 1990), debt tax shields (Titman and Wessels, 1988) and assets structure (Balakrishnan and Fox, 1993), etc. In recent years, the literature has paid special attention to the influence of the legal and institutional framework on corporate finance. The financial decisions of firms are not isolated from the institutional characteristics. In fact, the legal and institutional setting creates a net of relations between firms and financial institutions. From this point of view, financial systems have traditionally been classified into two main groups, depending on theorientation or importance of financial intermediaries (Allen, 1995; Allen and Gale, 2001). There is a Continental or bank-oriented system in which banks play a prominent role as financial channels from the ultimate lenders to the ultimate borrowers. It is the dominating system in Japan and in most Continental European countries such as Germany, France, Italy, Spain, etc. There is also an Anglo-Saxon or marketoriented system (e.g. USA, United Kingdom, etc.) in which banks are not so important and financial functions are directly performed by capital markets. Since both systems
framework
(i.e.
institutional
support
and
Capital structure and institutional setting
economic freedom) as found by Assane and Grammy (2003). Hence, the legal framework of each country especially law enforcement and investor protection - has been proved to affect corporate finance (Fabbri, 2001). For instance, Giannetti (2003) has shown how intangible assets - which could be the most difficult assets to fund are more easily funded when creditor rights are better protected and that a lower development of capital markets forces firms to use more debt. In addition, Storey (1994) has proved that bank financing is affected by the legal status of the firm. These two academic fields - namely, capital structure theory and the international comparison of financial systems - are the backbone of our article since we aim to study the capital structure of an international sample of firms following a decompositional analysis. According to Booth et al. (2001), our research is 2-fold: firstly, in a descriptive approach, we attempt to discover whether capital structure shows significant differences across countries and, secondly, we test whether the factors determining firms' financial decisions have a different influence depending on the legal and institutional framework. Although the classification scheme according to the institutional environment does not necessary imply any prediction concerning financial leverage but about internal and external finance (La Porta et al., 1997), there is evidence of the different level of debt and, more specifically, of the different debt maturities across countries. Broadly speaking, firms in the civil law countries usually have more debt and shorter maturity of debt than their common law counterparts (Demirgu¨c-Kunt and Maksimovic, 1999; Fan et al., 2003). Consequently, we could hypothesize that our results are supposed to show higher financial leverage and shorter maturity of debt in civil-law countries relative to common law countries.
1853
Our methodology follows two steps. The first step is broadly descriptive and aims both to compare the level of debt across countries or across legal systems and to test the existence of possible significant differences through the analysis of variance (hereinafter ANOVA). As stated by Rajan and Zingales (1995), both the level of financial leverage and its sensibility to the influencing factors can critically depend on the kind of measure of financial leverage. Therefore, we propose a set of different measures of capital structure and we decompose them into their basic components. The second methodological stage is mainly explana- tory and aims to test the impact of some factors on capital structure. In this phase, we are interested in knowing to what extent the international differences can be explained by a different impact of these factors. Our sample includes data from Austria and Germany as civil law countries with the German tradition, from Canada, the USA and the United Kingdom as common law countries and from Italy, France, Spain, the Netherlands and Belgium as civil law countries with the French tradition. The final distribution by countries and corporate systems is reported in Table 1. We use three main measures of capital structure as suggested by Rajan and Zingales (1995) and, especially, by Bevan and Danbolt (2002), so that we can compare our results with those of the earlier authors. The first measure is a general indicator of financial leverage and includes any kind of debt (both financial and commercial debt). We defined B1 as the ratio of financial debt (FD), i.e. costly debt, plus commercial debt (CD) to total assets at book value (TA).
B1 ¼ FDTACD þ
III. Data and Methodology Our database set is Compustat. As is widely known, Compustat gathers financial information with high reliability from a large number of firms. Given the high number of countries (and the disparities among them in terms of accounting rules), we have centred on balance sheets and income statements from a sample of 10 countries throughout 1997-2002 (Table 1). These firms can be divided into three of the four aforementioned main institutional settings.
The second measures explicitly focuses on costly debt and excludes commercial debt. Consequently we define B2 as total financial debt to book total assets ratio. B2 ¼ FD TA The last variable is informative of the relation between debt and costly funds (both implicit and explicit cost). These costly funds are costly debt and equity. The main difference between this ratio and previous ones is the exclusion of commer- cial debt and some elements which are quite difficult
1854
Table 1. Sample distribution across countries Number of firms Countries Number of firms
In this equation, i sub-index stands for the individual and t sub-index for the time. .i is the fixed-effects term which is firm specific and " it is the random component which is supposed to introduce all the remaining factors potentially affecting capital structure. The fixed-effects term introduces firm specific factors which can be correlated with the set of independent variables and whose omission could bias the results of the estimation. This fixed effect or unobservable and constant heterogeneity can not only be identified, but also treated by panel data procedures (Arellano, 2003; Baltagi, 2004). Panel data is basically a multivariate regression analysis along with the use of the Hausman test to detect the existence of these underlying individual effects and their correlation with the explanatory variables. When the Hausman test suggests the rejection of the null hypothesis of no correlation between the fixed effects term and the independent variables, the withingroups estimation provides consistent estimators. If the null hypothesis is not rejected, a random model or generalized least squares provides consistent and asymptotically efficient coefficients. Nevertheless, there is a lot of literature suggesting the endogeneity of some right-hand side variables, so we should control for this possible endogeneity (Cho, 1998; Demsetz and Villalonga, 2001). There are a number of procedures to deal with this problem and we will stress the generalized method of moments GMM (Arellano and Bond, 1991; Mairesse and Hall, 1996). The GMM is based on the use of instrumental variables according to the structure of available lagged variables which are supposed to be endogen- ous. By counting on more instruments than variables to be estimated, GMM provides more efficiently estimated coefficients. This is why, in our last stage, we report the results from the GMM estimation for MTB and PROF in order to test the robustness of our previous results.
Countries
United States 2827 United Kingdom 675 Canada 464 4066 Total common law Germany 671 Austria 92 Total German civil law 763
Holland 185 Belgium 102 Italy 209 564 France 103 Spain Total French 1163 civil law Total 5992
to classify such as deferred taxes, minority interests, nontaxable reserves, etc.1 B3 ¼ ðFDFDBVÞ þ Since an all-inclusive explanation of capital struc- ture is beyond the scope of this article and we simply aim to compare the basic issues of corporate finance across different legal and institutional frameworks, we limit our attention to four variables potentially driving the capital structure decisions of firms (Rajan and Zingales, 1995; Bevan and Danbolt, 2002; Bhaduri, 2002): growth opportunities, firm size, firm performance and assets tangibility. Growth opportunities, according to McConnell and Servaes (1995) and Lasfer (1995), are proxied by MTB or the market-to-book ratio (book value of debt plus market equity value to book total assets ratio). The size of the firm is measured through the log of the firm's turnover and the performance of the firm is measured with the EBIDTA (earnings before interest, depreciation, taxes and amortizations) to total assets ratio. Assets tangibility is measured through the assets with a physical existence (PA) to total assets ratio. These definitions can be expressed as follows: MTB ¼ ðTA À BV þ VMÞ PROF ¼ EBITDA TA TA PA LOGSALE ¼ LnðTurnoverÞ TANG ¼ TA The explanatory analysis is run through regression analysis with the panel data method. The model to be tested can be expressed as follows: Lit ¼ ff þ fi1MTBit þ fi2LOGSALEit þ fi3PROFit þ fi4TANGit þ .i þ "it
1
IV. Results The first step is a test of possible significant differences for the measures of capital structure among different legal systems. Results are reported in Table 2 and, although they are perhaps too detailed, they show a common and persistent pattern for B1, B2 and B3 across legal systems: whereas firms in the French civil law countries are the most
We have defined three measures of capital structure analogously to B1, B2 and B3 with market values instead of book values. Results are not reported for simplicity but are fully consistent.
Capital structure and institutional setting
Table 2. Mean value of debt conditional upon the legal system Mean B1 Anglo French Germ Total Anglo Frenc h Germ Total Anglo Frenc h Germ Total 0.3279 0.3894 0.2839 0.3343 0.2368 0.2313 0.1896 0.2302 0.3276 0.3780 0.3186 0.3360 LTDTA Mean
1855
classification scheme seems to lose importance relative to the legal roots and the institutional development criterion for classifying countries and explaining financial decisions (La Porta et al., 2000). We should note that there are some discrepancies in capital structure (for instance, the ranking based on B2 vs. the ranking based on B1 and B3). In order to elucidate this issue, we decompose the total debt to total assets ratio (B1) as a function of the maturity structure (Bevan and Danbolt, 2002; Ozkan, 2002). So, we have defined LTDTA as the proportion of long term debt over total assets and STDTA as the proportion of short term debt over total assets. Even short term debt can be divided into suppliers or commercial debt (COMTA) and the other short term debt (OSTDTA).2 Results exhibit big differences according to the institutional framework. Consistent with our expec-
0.1946 Anglo French 0.1319 Germ 0.0968 Total 0.1714 Anglo 0.1334 B2 STDTA 0.2575 Frenc 0.1871 h 0.1629 Germ 0.0912 Total B3 COMTA 0.1581 Anglo 0.0943 Frenc 0.1041 h 0.0422 Germ OSTTA 0.0994 Total 0.0928 Anglo 0.0588 Frenc h Germ Total Notes: The whole number of observations is 18 003 for common law countries, 4854 for the French tradition of civil law countries and 3029 from the German tradition of the civil law countries. B1 and M1 are all-items including definitions of capital structure, B2 and M2 includes just costly debt and B3 and M3 are scaled by costly liabilities. LTDTA and STDTA stand for long-term debt or short-term debt to total assets, respectively. COMTA stands for commercial debt to total assets ratio and OSTTA for other short-term debt to total assets ratio.
leveraged, their German civil law counterparts are the least prone to debt. Although these results hold for the three measures of financial leverage, they are inconsistent with our expectations, since civil law firms were hypothesized to have more debt than common law firms. Consequently, new analyses are required to solve this conflict. The next stage is an ANOVA to test the extent to which one can assert that different institutional and legal settings have different mean values of capital structure. ANOVA results, reported in Table 3, are quite significant and show that we can reject the equality of means across the three groups with aconfidence level higher than 99%. Nevertheless, this evidence requires a more detailed development with bilateral comparisons between pairs of systems as reported in Table 4. This table shows that, on average, the level of financial leverage is significantly different across groups of countries and corroborates the fact that firms from the French tradition of civil law countries are the most leveraged, whereas their German civil law counterparts are those with the least leverage. Consequently, the market vs. banks
tations and with previous literature, while Anglo-Saxon firms are those with the highest long term debt ratio (19.4%), the German civil law firms are those with the lowest long term debt (9.6%). On the contrary, if we focus on the short term debt, we can see how French civil law firms are the most leveraged companies whereas AngloSaxon firms are the least ones. In turn, different kinds of debt seem to have an asymmetric role: common law firms appear to be more prone to long term debt whereas civil law firms tend to borrow to short term. Among the possible explanations to these results, we could cite that both the legal protection of investors and the quality of legal enforcement foster long term lending relations, as well as institutional investors and the activity in capital markets - more often in common law countries than in civil law ones. Once we have checked the differences between legal systems in terms of capital structure, we can test whether the factors determining firms' financial choices are responsible for those differences. We have made capital structure depend on four of the factors which are most usually supposed to affect a firm's finance: growth opportunities (MTB), firm size (LNSALES), firm performance (EBITDA) and assets tangibility (TANG). The results of the regression analysis with the method of panel data are reported in Tables 5 and 6. For each explanatory variable and for each measure of capital structure four estimations are provided. The first one has been run over the entire sample while the second, the third and the fourth are estimated for the Anglo-Saxon, the French and the German tradition of civil law system respectively. In general terms, the results reported in Table 5 show a common pattern for the four explanatory
2
All these measures have been scaled by total assets.
1856
Table 3. Analysis of the variance depending on the legal system Square mean B1 B2 B3 Between-groups Within-groups Between-groups Within-groups Between-groups Within-groups 11.585 0.034 2.886 0.030 5.376 0.057 F-test 336.06 94.35 93.90 p-value 0.000 0.000 0.000 LTDTA STDTA COMTA OSTTA Square mean 17.048 0.025 30.462 0.013 8.725 0.006 8.238 0.005 F-test 673.608 2273.038 1288.853 1427.425 p-value 0.000 0.000 0.000 0.000
Table 4. Bilateral post hoc tests System I B1 Anglo French Anglo French Anglo French System II French German German French German German French German German Difference À0.061 0.044 0.105 0.005 0.047 0.041 À0.050 0.009 0.059 p-value 0.000 0.000 0.000 0.055 0.000 0.000 0.000 0.056 0.000 LTDTA Difference 0.0062 0.0097 0.0035 À0.124 À0.053 0.070 À0.066 À0.003 0.063 À0.057 À0.050 0.006 p-value 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.052 0.000 0.000 0.000 0.000
B2
STDTA
B3
COMTA
OSTTA
variables in all the legal systems and for all the measures of capital structure. Although we do not aim to explain corporate finance decisions in each institutional framework but simply to show the common and the distinctive features, we should try to provide some justification for these results. Growth opportunities and firm performance are proved to have a negative and significant relation with financial leverage, whereas the size of the firm and the assets tangibility is positively related. These results can be explained by the link between the size of the firm and the asymmetric information in capital markets (Ojah and Manrique, 2005). Since large companies are usually better known in capital markets, there is less asymmetry between a firm'sinformed managers and investors, so that large firms can more easily borrow from capital markets. The negative relation between debt and firm perfor- mance has been widely documented by previous research into the pecking order theory of capital structure (Myers, 1977; Myers and Majluf, 1984). Nevertheless, the effect of growth opportunities and assets tangibility requires further explanation. As far as growth opportunities are concerned, their negative impact - when significant - is noticeable and coherent with previous research (Bevan and Danbolt, 2002). A more detailed analysis
as a function of maturity structure confirms that firms with more growth opportunities rely on commercial debt because this kind of debt does not impose such constraining covenants as other types of credit (Barclay and Smith, 1999). Table 5 also shows how assets tangibility is positively related to debt for all the measures of capital structure and in all the legal systems. This result can be explained on the basis of tangible assets as collateral: the more important the tangible assets are the more collateral the firm puts and, consequently, the lower the interest rate is. Notwithstanding, Table 6 shows remarkable differences conditional upon the kind of debt: whereas TANG is positively related to LTDTA, it is negatively related to STDTA. This means that long term debt is likely to fund long term assets (which can be put as collateral for long term debt) and short term debt will fund current assets. To sum up, our results up to this point show: (1) Noticeable and quite consistent differences in the level of financial leverage across the firms from different legal systems; (2) Financial leverage is affected by the same factors which have traditionally been supposed to explain capital structure. But, if this is the case, one should question how the same factors could produce such large differences across the
Table 5. Factors affecting capital structure
B2 French 0.002 0.065 0.055 0.0657 164.47** 62.57** 377.63** 341.22** 158.53** 0.0805 0.2188 0.2750 0.0890 (2.89)** 0.163 (7.73)** 0.197 (27.12)** 0.192 (22.35)** 0.186 (10.37)** (17.55)** 0.064 (11.93)** 0.028 (17.77)** 0.021 (11.60)** 0.053 (15.37)** 0.044 0.203 0.0710 46.74** (1.91)* À0.002 (À1.84)* À0.001 (À4.04)** À0.001 (À0.15) (À2.50)** À0.000 German Total Anglo French German B3 Total Anglo (À2.76)** À0.001 (8.37)** (9.92)** 0.049 0.215 0.2221 390.14** (22.64)** (21.47)** 0.036 0.207 0.2864 366.13** (À1.63) (14.75)** (18.10)*** French 0.001 0.094 0.137 0.1420 137.88** (0.54) (18.39)** (5.18)** German À0.001 (À0.64) 0.088 0.284 0.1324 56.93** (10.93)** (9.05)**
B1
Total
Anglo
MTB
À0.0007 (À1.62)
À0.0001 (À0.35)
À0.002 (À2.07)** À0.001
LOGSALE
0.036
(22.38)**
0.028
(14.80)**
PROF
À0.081
(22.49)** À0.092 (À20.90)** À0.290 (À16.04)** À0.025 (À3.70)** À0.070 (À20.34)** À0.080 (À18.85)** À0.259 (À15.71)** À0.019 (À2.93)** À0.106 (À22.32)** À0.119 (À20.98)** À0.422 (À17.33)** À0.034 (À3.33)**
TANG
0.123
(16.38)**
0.122
(13.73)**
Adj. R2
0.1443
0.1950
Hausman
509.79**
449.59**
test
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients.
1858
Table 6. Factors determining capital structure
STDTA Anglo À0.001 (À1.74)* 0.018 (10.10)*** À0.053 (À12.69)*** 0.172 (20.17)*** 0.2821 333.62*** 0.001 (0.59) 0.016 (13.91)*** À0.034 (À12.77)*** À0.051 (À9.20)*** 0.0673 742.44*** À0.001 (À0.29) 0.031 (10.12)*** À0.131 (À8.83)*** 0.194 (12.07)*** 0.1695 60.99*** À0.001 (À0.45) 0.011 (2.79)** À0.016 (À3.14)*** 0.149 (9.41)*** 0.1475 30.48*** French German Total Anglo 0.001 (1.95)* 0.009 (7.46)*** À0.038 (À12.76)*** À0.050 (À8.33)*** 0.0734 352.27*** French 0.003 (2.42)** 0.033 (10.12)*** À0.159 (À10.14)*** À0.139 (À8.2)*** 0.0674 110.62*** German À0.002 (À1.67)* 0.053 (10.89)*** À0.009 (À1.48) 0.014 (0.74) 0.0482 70.54***
LTDTA
Total
MTB LOGSALE PROF TANG Adj. R2
Hausman test OSTTA 0.001 (6.12)*** 0.006 (10.50)*** À0.011 (À7.79)*** À0.070 (À23.28)*** 0.1302 391.45** 0.003 (3.76)*** 0.011 (5.73)*** À0.030 (À3.17)*** À0.130 (À12.54)*** 0.1797 59.42*** 0.001 (0.36) 0.020 (8.54)*** À0.006 (À1.91)* À0.039 (À4.22)*** 0.1246 45.04** À0.001 (À2.86)** 0.008 (7.81)*** À0.023 (À10.19)*** 0.022 (4.61)*** 0.0013 285.87***
À0.001 (À2.23)** 0.019 (13.14)*** À0.047 (À14.16)*** 0.174 (25.13)*** 0.2704 412.61***
COMTA
MTB LOGSALE PROF TANG Adj. R2
Hausman test
0.001 (5.89)*** 0.008 (13.69)*** À0.010 (À7.54)*** À0.073 (À25.35)*** 0.1580 617.32**
À0.001 (À1.27) 0.003 (2.56)** À0.026 (À10.24)*** 0.020 (3.80)*** 0.0049 83.56***
0.001 (0.14) 0.021 (7.79)*** À0.128 (À9.67)*** À0.008 (À0.60) 0.0025 82.76***
À0.002 (À2.03)** 0.032 (7.31)*** À0.003 (À0.59) 0.053 (3.10)*** 0.0100 41.34***
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients.
Capital structure and institutional setting
systems or, more precisely, across countries. This is why, in the subsequent analysis, we introduce two country specific characteristics which are related to the legal and institutional framework: law enforce- ment and the quality of accounting. Based on data from La Porta et al. (1998), we have defined two dummy variables (DEL and DQA) that equal 1 if law enforcement or the quality of accounting is above the mean of the sample.3 These dummy variables have interacted with the four explanatory variables in order to test if they have differential effects conditioned by the law enforcement and the quality ofaccounting. Results are displayed in Table 7. For the sake of simplicity, we will just comment on the most general and common features instead of a too detailed explanation of the results. The interacting variables are quite significant, so we can assert that growth opportunities, the size and performance of the firm and the assets tangibility have a different effect depending on those two characteristics. Two results, nevertheless, deviate from this pattern: neither the interaction of MTB and the dummy of quality accounting nor the interaction of TANG and quality accounting seem to have a significant impact. Excluding these exceptions, Table 7 is interesting because it suggests that different levels of leverage are not per se a result of the legal environment, but that the legal setting creates the conditions so that those factors have a differential impact. We have run some additional regressions in order to check the robustness of our results. Firstly, we have defined a dummy variable for each institutional setting. These dummies (DFC and DGC 4) have interacted with the four aforementioned explanatory variables to test possible differential effects depending on the legal system. Results are shown in Table 8 and are consistent with previous ones: the four variables continue to be significant and, in addition, the interacting variables are also statistically significant, with the sole exception of MTB in some estimations. Since a very exhaustive explanation might obscure the general meaning of the regression, just a general comment can be suitable: the high significance of the interacting variables allows us to infer that the four determinants of capital structure have a differential influence in each legal system. Another sensitivity analysis takes into account the possible endogeneity of the explanatory variables.
3
1859
As some authors have pointed out (Demsetz and Lehn, 1985; Jensen, 1986; McConnell and Servaes, 1995), some firms' characteristics that we have assumed exogenous could be affected by the firms' capital structure. Thus, we have replicated the previous regressions with the GMM to control for the potential endogeneity of the explanatory variables (Table 9). Results basically remain the same as the previous ones with two small exceptions: PROF is no longer as significant as it was and MTB coefficients are inconsistent for some estimations. Notwithstanding, the Hansen test, which aims to control the overidentification restrictions - and, consequently, the validity of instruments - does not support the accuracy of the instruments. Additionally, the second order serial correlation would advise some caveat in analyzing GMM results because the weakness of the instruments reduces the efficiency of the estimations and increases the possible bias.
V. Concluding Remarks Recent research has shown the influence of the legal and institutional setting on firms' financial decisions. These decisions are no longer due just to firms' value enhancement strategies but are also affected by the legal framework in which firms operate. Our article joins analyses of firms' capital structure in an international framework in order to test differences across countries or legal systems and to find common points in the factors potentially affecting capital structure. We begin with a division of the countries according to the characteristics of the corporate system. The traditional bank-oriented vs. market-oriented classification scheme is no longer practical enough and we need to use more precise criteria based on the legal origin of institutions. In fact, as a first conclusion of our article, we can assert that the global consideration of the bank oriented system is inexact since it includes a number of countries with fairly different corporate systems. When we decom- pose capital structure in a set of different measures, we find that the firms in the French tradition of civil law countries are more leveraged than their common law counterparts and we also find that the firms in the
The countries with the best law enforcement in our sample are Canada, The United States, Belgium, Holland and Austria, whereas the countries with the best quality of accounting are Canada, The United Kingdom, The United States, France, Holland and Spain. It is interesting to note that this classification differs from the civil vs. common law, so that the dummy variables do not measure the legal tradition but these two features of the legal and institutional setting. 4 Dummy for the French civil and the German civil countries.
Table 7. Factors determining capital structure with law enforcement and the quality of accounting
B2 À0.0017 (À1.35) 0.0008 0.0851 À0.2029 (À16.32)*** 0.2239 À0.002 À0.0145 0.0179 À0.0029 (0.3) (À7.32)*** (À7.62)*** (À1.21) 0.0741 536.43*** COMTA (0.71) 0.052 (14.66)*** À0.0151 (À2.91)*** À0.0068 (À0.47) 0.0007 (0.97) À0.0153 (À6.62)*** 0.0464 (7.14)*** À0.0159 (À1.57) (0.71) 0.0014 À0.0257 À0.0217 À0.024 0.0597 750.67*** (1.6) (À7.82)*** (À4.14)*** (À1.77)* 0.1894 450.07*** 0.0494 731.96*** 0.054 214.98** 0.0585 255.6*** À0.0011 (À1.14) 0.0017 0.0139 (4.82)*** (11.91)*** À0.0132 (À3.66)*** À0.0913 (À18.47)*** 0.0008 (1.72)* 0.0211 (11.39)*** À0.0073 (À2.7)*** À0.0493 (À6.58)*** (À0.2) À0.0002 (À0.18) 0.0019 (0.41) À0.0382 (À5.05)*** 0.0248 0.1135 462.59*** OSTTA À0.0012 0.0188 (À2.08)** (9.67)*** À0.0629 (À10.46)*** 0.0314 (3.82)*** À0.0019 (À2.34)** 0.0309 (10.02)*** À0.0078 (À1.73)* 0.0425 (3.41)*** (1.27) (1.91)** (À4.36)*** (À1.98)** (13.16)*** 0.2419 (9.36)*** À0.0461 (À4.93)*** (21.19)*** 0.0937 (14.66)*** (0.65) 0.0704 (14.63)*** À0.0338 (À4.81)*** À0.1248(À13.83)*** 0.2071 (16.76)*** 0.195 (10.39)*** À0.003 (À2.07)** À0.051 (À10.68)*** 0.1139 (8.46)*** À0.0196 (À0.93) (0.89) À0.0498 À0.0828 À0.0338 0.1207 434.08*** 0.0006 (À1.26) (À8.59)*** (6.58)*** (À1.23) 0.0012 À0.0237 (À4.81)*** À0.0599 (À7.59)*** 0.0008 0.0479 540.37*** STDTA À0.0007 (À0.55) 0.0005 0.0327 (14.6)*** À0.0761(À10.99)*** À0.0599 (À6.33)*** (À1.75)** (À5.31)*** (0.87) (4.36)*** 0.0004 À0.0141 (À7.15)*** À0.0045 (À1.44) À0.0284 (À3.49)*** 0.1432 530.14*** 0.0184 À0.0187 (À2.88)*** 0.1525 À0.0007 À0.0074 0.0034 0.0266 (1.76)* (À10.53)*** (À4.35)*** (À3.36)*** 0.2816 412.65*** (8.47)*** (4.13)*** (0.04) 0.1457 À0.0013 À0.0298 0.0643 À0.0188 (1.14) (À7.39)*** (À7.88)*** (À1.31) 0.0447 635.1*** (7.48)*** À0.0265 (À3.91)*** 0.0489 (16.78)*** 0.0493 (10.63)*** À0.0007 (À0.8) À0.0026 (À2.09)** À0.0017 (À1.02) B3
B1
MTB
0.001
(1.08)
LOGSALE
0.0628
(20.76)***
PROF
À0.138
(À14.74)***
TANG
0.1158
(9.04)***
MTB*DLE
À0.002
(À1.89)*
LOGSALE*DLE
À0.0372 (À10.33)***
PROF*DLE
0.0677
(6.67)***
TANG*DLE
0.0078
(0.49)
MTB*DQA
0.0016
LOGSALE*DQA
À0.0378
PROF*DQA
À0.0645
TANG*DQA
À0.0276
Adj. R2
0.0512
Hausman test
646.91***
LTDTA
MTB
0.0005
(0.6)
LOGSALE
0.0302
(10.78)***
PROF
À0.062
(À7.15)***
TANG
0.1758
(14.82)***
MTB*DLE
À0.0001
(À0.07)
LOGSALE*DLE
À0.0226
(À8.52)***
PROF*DLE
0.0498
(6.64)***
TANG*DLE
0.0107
(0.92)
MTB*DQA
0.0018
LOGSALE*DQA
À0.0397
PROF*DQA
À0.0263
TANG*DQA
À0.0523
Adj. R2
0.0883
Hausman test
494.81***
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients. Independent variables have interacted with a dummy of law enforcement (DLE) and with a dummy of the quality of accounting (DQA).
Table 8. Factors determining capital structure with dummy variables for institutional settings
B2 B3 LTDTA STDTA COMTA OSTTA (À1.21) (2.44)** (À9.76)*** (3.62)*** (0.53) (6.29)*** (À7.6)*** (À1.87)* (À1.88)* (7.85)*** (4.37)*** (2.25)**
Capital structure and institutional setting 1861
B1
MTB LOGSALE PROF TANG MTB*DFC LOGSALE*DFC PROF*DFC TANG*DFC MTB*DGC LOGSALE*DGC PROF*DGC TANG*DGC Adj. R2
Hausman test
À0.0002 (À0.36) 0.0285 (15.35)*** À0.092 (À21.68)*** 0.1222 (14.25)*** 0.0028 (1.67)* 0.0368 (7.82)*** À0.1981 (À9.48)*** À0.067 (À2.81)*** À0.0022 (À1.52) 0.0364 (6.15)*** 0.0666 (8.03)*** 0.041 (1.75)* 0.068 658.3***
À0.0014 (À2.61)*** 0.0216 (12.08)*** À0.0804 (À19.63)*** 0.1926 (23.28)*** 0.0012 (0.74) 0.032 (7.06)*** À0.1793 (À8.9)*** À0.0065 (À0.28) À0.0012 (À0.89) 0.0225 (3.94)*** 0.0608 (7.6)*** 0.0104 (0.46) 0.0905 500.73*** À0.0012 (À1.64)* 0.0367 (14.9)*** À0.1193 (À21.19)*** 0.2079 (18.29)*** 0.0022 (1.00) 0.0581 (9.32)*** À0.3034 (À10.95)*** À0.0707 (À2.24)** À0.0001 (À0.03) 0.0518 (6.6)*** 0.0851 (7.74)*** 0.0763 (2.46)** 0.1096 521.0*** À0.001 (À1.88)* 0.0187 (10.88)*** À0.0538 (À13.67)*** 0.1726 (21.72)*** 0.0006 (0.4) 0.013 (2.99)*** À0.0774 (À4.00)*** 0.0221 (1.00) 0.0005 (0.38) À0.0073 (À1.34) 0.0375 (4.88)*** À0.0235 (À1.08) 0.1802 445.24****
0.0008 (1.87)* 0.0098 (7.14)*** À0.0382 (À12.21)*** À0.0505 (À7.97)*** 0.0022 (1.76)* 0.0237 (6.84)*** À0.1207 (À7.83)*** À0.0891 (À5.07)*** À0.0027 (À2.54)** 0.0437 (10.01)*** 0.029 (4.75)*** 0.0645 (3.73)*** 0.0824 594.38***
0.0012 (5.59)*** 0.0069 (9.59)*** À0.0117 (À7.12)*** À0.0704 (À21.28)*** 0.0016 (2.48)** 0.0048 (2.62)*** À0.0188 (À2.33)** À0.0604 (À6.58)*** À0.001 (À1.72)* 0.0139 (6.08)*** 0.0058 (1.81)* 0.0306 (3.39)*** 0.0430 569.75***
À0.0004 0.0029 À0.0266 0.02 0.0006 0.019 À0.1019 À0.0286 À0.0018 0.0298 0.0233 0.0339 0.0815 233.2***
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Hausman test follows a _ 2 distribution with so many degrees of freedom as estimated coefficients. Independent variables have interacted with dummy variables of the French civil system (DFC) and the German civil system (DGC).
1862
Table 9. Factors determining capital structure (GMM estimation) B2 0.0048 (1.82)* 0.0019 (0.55) 0.0253 (5.66)*** 0.0433 (6.87)*** À0.0902 (À3.06)*** À0.0972 (À2.35)** 0.229 (13.23)*** 0.2502 (10.89)*** 276.9*** 245.5*** 13.28** 10.07 À7.37*** À6.29*** À5.04*** À4.84*** B3 LTDTA STDTA COMTA OSTTA 0.0008 (0.65) 0.0084 (4.37)*** À0.0344 (À2.55)** 0.0071 (0.68) 28.3*** 20.03*** À12.69*** À4.03***
B1
MTB 0.0085 (2.9)*** LSALES 0.0298 (5.56)*** PROF À0.0444 (À1.34) TANG 0.1862 (10.11)*** Wald test 164.6*** Hansen test 19.412*** AR(1) À7.15*** AR(2) À5.59***
0.0025 (1.09) 0.0064 (4.06)*** 0.0053 (5.69)*** 0.0169 (4.06)*** 0.0178 (5.64)*** 0.0037 (1.64)* À0.0504 (À2.46)** À0.0211 (À1.21) 0.0004 (0.08) 0.2133 (14.15)*** À0.0458 (À3.49)*** À0.0424 (À5.84)*** 273.6*** 59.7*** 96.6*** 7.52 14.97** 77.82*** À10.04*** À13.41*** À8.81*** À4.73*** À4.84*** À2.19**
Notes: Estimated coefficients and (t-statistics). ***, ** and * For a confidence level higher than 99%, 95% and 90%, respectively. Wald test is a test of joint significance for all the variables. Hansen test of overidentification restrictions allows controlling the validity of instruments and follows a _ 2 distribution with so many degrees of freedom as the difference between the number of instruments and the number of regressors. AR(1) and AR(2) are tests or first and second order serial correlation.
Capital structure and institutional setting
German tradition of civil law are the least leveraged ones. More important than the level of leverage, our analysis reveals a clear difference in debt maturity so that, the more the rights of investors are protected, the longer the term of the debt becomes. Although the use of debt is different across legal systems, the factors traditionally thought as determi- nants of capital structure have much in common in different financial systems. Although the perfor- mance and the size of the firm, the assets tangibility and growth opportunities have a similar effect in the three scenarios, there are specific effects conditional on the legal systems. Consequently, our results suggest that the effect of the factors that have traditionally been considered determinants of capital structure depends on the legal and institutional setting and that these differential effects can explain international disparities in capital structure. Our research has also shown that the introduction of some variables concerning the legal protection of investors, such as the enforcement of the law and the quality of the financial information, can help to explain firms' financial choices.
1863
Barnea, A., Haugen, R. A. and Senbet, L. W. (1985) Agency Problems and Financial Contracting, Prentice Hall, New Jersey. Berger, A. N. and Udell, G. F. (1995) Relationship lending and lines of credit in small firm finance, Journal of Business, 68, 351-81. Bergstro¨m, C. and Rydqvist, K. (1990) The determinants of corporate ownership, Journal of Banking and Finance, 14, 237-53. Bevan, A. A. and Danbolt, J. (2002) Capital structure and its determinants in the UK - A decompositional analysis, Applied Financial Economics, 12, 159-70. Bhaduri, S. N. (2002) Determinants of capital structure choice: a study of the Indian corporate sector, Applied Financial Economics, 12, 655-65. Booth, L., Aivazian, V., Demirguc-Kunt, A. and Maksimovic, V. (2001) Capital structures in develop- ing countries, Journal of Finance, 56, 87-129. Bradley, M., Jarrell, G. and Kim, E. H. (1984) On the existence of an optimal capital structure: theory and evidence, Journal of Finance, 39, 857-80. Cho, M. H. (1998) Ownership structure, investment, and the corporate value: an empirical analysis, Journal of Financial Economics, 47, 103-21. Colombo, E. (2001) Determinants of corporate capital structure: evidence from Hungarian firms, Applied Economics, 33, 1689-701. Corbett, J. and Jenkinson, T. (1998) German investment financing: an international comparison, in Competition and Convergence in Financial Markets. The German and AngloAmerican Models (Eds) S. W. Black and M. Moersch, Elsevier, Amsterdam, pp. 85-124. Demirgu¨c-Kunt, A. and Maksimovic, V. (1999) Institutions, financial markets, and firm debt maturity, Journal of Financial Economics, 54, 295-336. Demsetz, H. and Lehn, K. (1988) The structure of corporate ownership: causes and consequences, Journal of Political Economy, 93, 1155-77. Demsetz, H. and Villalonga, B. (2001) Ownership structure and corporate performance, Journal of Corporate Finance, 7, 209-33. Fabbri, D. (2001) The legal enforcement of credit contracts and the level of investment. Center for Studies in Economics and Finance, Working paper, No. 57. Fan, J. P. H., Titman, S. and Twite, G. J. (2003) An International comparison of capital structure and debt maturity choices. EFA 2003 Annual Conference Paper No. 769, Available at SSRN: http://ssrn.com/ abstract¼423483. Giannetti, M. (2003) Do better institutions mitigate agency problems? Evidence from corporate finance choices, Journal of Financial and Quantitative Analysis, 38, 185-212. Harris, M. and Raviv, A. (1991) The theory of capital structure, Journal of Finance, 46, 297-355. Himmelberg, C. P., Hubbard, R. G. and Love, I. (2004) Investor protection, ownership, and the cost of capital. World Bank Policy Research Working Paper, No. 2834. Jalilvand, A. and Harris, R. S. (1984) Corporate behavior in adjusting to capital structure and dividend targets: an econometric study, Journal of Finance, 29, 127-44. Jensen, M. C. (1986) Agency costs of free cash flow, corporate finance and takeovers, American Economic Review, 76, 323-9.
Acknowledgements The authors are grateful to Alan Hynds, Giorgio Valente (co-editor of Applied Economics) and an anonymous referee for their comments.
References
Allen, F. (1995) Stock markets and resource allocation, in Capital Markets and Financial Intermediation (Eds) C. Mayer and X. Vives, Cambridge University Press, Cambridge, pp. 81-107. Allen, F. and Gale, D. (2001) Comparing Financial Systems, MIT Press, Cambridge. Arellano, M. and Bond, S. (1991) Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations, Review of Economic Studies, 58, 277-97. Arellano, M. (2003) Panel Data Econometrics, Oxford University Press, Oxford. Assane, D. and Grammy, A. (2003) Institutional framework and economic development: international evidence, Applied Economics, 35, 1811-7. Baltagi, B. H. (2004) Panel data: theory and applications, Physica-Verlag, New York. Balakrishnan, S. and Fox, I. (1993) Asset specificy, firm heterogeneity and capital structure, Strategic Management Journal, 14, 3-16. Barclay, M. J. and Smith, C. W. (1999) The capital structure puzzle: another look at the evidence, Journal of Applied Corporate Finance, 12, 8-20.
1864
Kester, C. W. (1986) Capital and ownership structure: a comparison of United States and Japanese manufacturing corporations, Financial Management, 15, 5-16. Kim, W. S. and Sorensen, E. H. (1986) Evidence on the impact of the agency costs of debt on corporate debt policy, Journal of Financial and Quantitative Analysis, 21, 131-44. La Porta, R., Lo´pez de Silanes, F., Shleifer, A. and Vishny, R. (1997) Legal determinants of external finance, Journal of Finance, 52, 1131-50. La Porta, R., Lo´pez de Silanes, F., Shleifer, A. and Vishny, R. (1998) Law and finance, Journal of Political Economy, 106, 1113-55. La Porta, R., Lo´pez de Silanes, F., Shleifer, A. and Vishny, R. (2000) Investor protection and Journal of Financial corporate governance, Economics, 58, 3-27. Lasfer, M. A. (1995) Agency costs, taxes and debt: the UK evidence, European Financial Management, 1, 265-85. Levine, R. (1998) The legal environment, banks, and longrun economic growth, Journal of Money, Credit and Banking, 30, 596-620. Levine, R., Loayza, N. and Beck, T. (2000) Financial intermediation and growth: causality and causes, Journal of Monetary Economics, 46, 31-77. Mairesse, J. and Hall, B.H. (1996) Estimating the productivity of research and development in French and US manufacturing firms: an exploration of simultaneity issues with GMM methods, in International Productivity Differences and their Explanation (Eds) K. Wagner and B. Van Ark, Elsevier, Amsterdam, pp. 285-315. Marsh, P. (1982) The choice between equity and debt: an empirical study, Journal of Finance, 37, 121-44. Mayer, C. (1990) Financial systems, corporate finance, and economic development, in Asymmetric Information, Corporate Finance and Investment (Ed.) R. G. Hubbard, The University of Chicago Press, Chicago, pp. 307-32. McConnell, J. J. and Servaes, H. (1995) Equity ownership and the two faces of debt, Journal of Financial Economics, 39, 131-57. Myers, S. C. (1977) Determinants of corporate borrowing, Journal of Financial Economics, 5, 147-75. 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, 13, 187-221. Ojah, K. and Manrique, J. (2005) Determinants of corporate debt structure in a privately dominated debt market: a study of the Spanish capital market, Applied Financial Economics, 15, 455-68. Ozkan, A. (2002) The determinants of corporate debt maturity: evidence from UK firms, Applied Financial Economics, 12, 19-24. Rajan, R. and Zingales, L. (1995) What do we know about capital structure? Some evidence from international data, Journal of Finance, 50, 1421-60. Rajan, R. and Zingales, L. (1998) Debt, folklore, and crosscountry differences in financial structure, Journal of Applied Corporate Finance, 10, 102-7. Rutherford, J. (1988) An international perspective on the capital structure puzzle, in New Developments in International Finance (Eds) J. Stern and D. Chew, Basil Blackwell, Nueva York. Storey, D. J. (1994) The role of legal status in influencing bank financing and new firm growth, Applied Economics, 26, 129-36. Titman, S. and Wessels, R. (1988) The determinants of capital structure choice, Journal of Finance, 43, 1-19.
doc_987400384.docx