Cross-Country Determinants of Capital Structure Choice: A Survey of European Firms

Cross-Country Determinants of Capital Structure Choice: A Survey of European Firms Franck Bancel (ESCP-EAP) Usha R. Mittoo (University of Manitoba) Forthcoming in Financial Management Journal Abstract We survey managers in 16 European countries on the determinants of capital structure. Financial flexibility and earnings per share dilution are primary concerns of managers in issuing debt and common stock, respectively. Managers also value hedging considerations and use windows of opportunity when raising capital. We find that although a country s legal environment is an important determinant of debt policy, it plays a minimal role in common stock policy. We find that firms financing policies are influenced by both their institutional environment and their international operations. Firms determine their optimal capital structures by trading off costs and benefits of financing. Keywords: Capital Structure, European Managers, Survey, Debt, Equity JEL Classification: G32, G15, F23. We are grateful to all chief financial officers who have participated in this study, and to John Graham, two anonymous reviewers, and corporate finance teams of BNP Paribas and Merrill Lynch for their valuable comments and suggestions on an earlier version of our paper. We also thank rence Booth, Amitabh Dutta, and participants at the 2002 European Financial Management Association meetings, the 2002 Multinational Financial Society meetings, the 2002 Northern Finance Association meetings, the 2002 CGA Conference (University of Manitoba), and the 2003 American Finance Association meetings for helpful comments, and Zhou Zhang for research assistance. Mittoo acknowledges financial support from the Bank of Montreal Professorship. Corresponding authors: Franck Bancel Professor ESCP-EAP 79 avenue de la République 75 543 Paris cédex 11 - France Phone: (33) 1 49232076, Fax: (33) 1 49232036, Email: bancel@escp-eap.net Usha R. Mittoo Bank of Montreal Professor in Finance I.H. Asper School of Business University of Manitoba Winnipeg, Manitoba, R3T 5V4 - Canada Phone: (204) 474-8969, Fax: (204) 474-7545, Email: umittoo@ms.umanitoba.ca

Cross-Country Determinants of Capital Structure Choice: A Survey of European Firms How firms make their capital structure decisions has been one of the most extensively researched areas in corporate finance, yet there is little consensus among these studies. In a recent paper, Graham and Harvey (2001) examine the theory and practice of corporate finance by surveying U.S. managers. Our study does the same in the European context, but differs in its scope and focus. Unlike Graham and Harvey who examine several aspects of corporate finance in a single country, we focus on the cross-country comparisons of managerial views on determinants of capital structure in a sample of 16 European countries: Austria, Belgium, Greece, Denmark, Finland, Ireland, Italy, France, y, Netherlands, Norway, Portugal, Spain, Switzerland, Sweden, and the U.K. Our study examines whether European and U.S. managers views on capital structure are driven by similar factors. We also examine the role of legal institutions in explaining the financing policies of firms across countries. We investigate whether these policies are determined largely by the legal institutions of the home country or are the result of a complex interaction of several institutions in a country. We also study the sensitivity of different determinants of capital structure to the country s institutional environment. It is possible that factors underlying debt or equity policies may be influenced differently by various institutions. Thus, to understand the impact of a country s institutions on leverage, we need to analyze determinants of different components of leverage across countries, which is an onerous task. We examine this issue by asking managers about the determinants of the debt, equity, convertible, and foreign capital-raising policies of their firms. The paper is organized as follows. In Section I we develop our hypotheses and describe our method. Section II compares European managers views with those of U.S. managers. Section III presents our cross-country analysis. Section IV summarizes and concludes. I. Hypotheses and Survey Design We first develop the hypothesis on the literature and then discuss our sample and data. A. Hypotheses Graham and Harvey (2001) test the implications of different capital structure theories through a survey of U.S. managers. They find moderate support that firms follow the trade-off theory and target their debt ratios. They also find some support for the pecking-order theory. Their results show that firms value financial flexibility 1

but its importance is not related to information asymmetry or growth options in the manner predicted by the pecking-order theory. They find little evidence that other factors including agency costs, signaling, asset substitution, free cash flow and product market concerns affect capital structure choice. They also report that managers use many informal criteria, such as credit rating and earnings per share dilution, in making their financing decisions. An important issue is whether U.S. managers views are influenced largely by the U.S. institutional environment or are also shared by their peers in other countries. We examine the following hypothesis: H1: European and U.S. managers make their capital structure decisions using similar factors, all else equal. Several studies have examined the role of different institutions in explaining differences in leverage across countries. La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998) stress that the legal system is the primary determinant of the availability of external financing in a country. They argue that common-law system provides better quality of investor protection than civil-law systems, and among the civil-law systems, and Scandinavian systems provide better protection than system. They show that the size and breadth of capital markets vary systematically and positively with the quality of legal systems across countries. Demirgüç Kunt and Maksimovic (2002), however, argue that a country can partially compensate for the effect of the deficiency of the legal systems on banks through a combination of administration and regulation of the banking system, and that the legal systems in different countries can have different comparative advantages in supporting a quality banking system or quality securities markets. Demirgüç Kunt and Maksimovic (1999) show that debt maturity is affected by both financial and legal institutions in their study of debt policies across 30 countries. Rajan and Zingales (2003) examine the evolution of European financial system over time and conclude that it reflects a complex interaction of several country and global factors including economic and political. Rajan and Zingales (1995) compare leverage and its determinants across G-7 countries and find that although leverage and its correlations with variables, such as firm size and profitability, are fairly similar across their sample countries, the theoretical underpinnings of the observed correlations are different. Booth, Aivazian, Demirgüç Kunt A. and Maksimovic (2001) find that although the capital structure decisions in their sample of ten developing countries are affected by the same variables as in the developed countries, there are persistent differences across countries. Other studies show that the increased accessibility to global capital markets, through firms foreign listing or multinational operations, also influences capital structure (see for e.g., Pagano, Röell, and Zechner (2002), and Doukas and Pantzalis (2003)). 2

The evidence on whether financing policies of firms are determined primarily by the legal system of its home country or whether other country institutions also play a major role is unclear. A major problem in the cross-country research is that differences in accounting and disclosure practices make it difficult to compare and interpret financial data across countries. Further, different components of leverage, such as debt or equity, are likely to be influenced differently by various institutions. For instance, debt financing is likely to be more sensitive to the bankruptcy law, but equity financing might be influenced more by the stock market regulation in a country. Moreover, it is almost impossible to collect data on the evolution of the financial system in a country over time. To the extent a country s institutions affect its financing structure, their impact should be reflected in managerial policies and practices in that country. We examine the following hypotheses on the sensitivity of the important determinants of debt and equity, as identified by managers in our survey, to legal systems: H2: Cross-country differences in managerial views on the determinants of capital structure are influenced primarily by the legal system of the home country, all else equal. H3: Cross-sectional differences in managerial rankings of major determinants of debt and equity policies differ systematically according to the quality of legal systems, all else equal. B. Survey The design of our questionnaire is similar to that in the Graham and Harvey (2001) study, but we add or modify several questions to facilitate cross-country analysis. For example, we ask managers questions on the ownership structure, the influence of different stakeholders on their firm s financial decisions and foreign sales, foreign listings, and foreign capital-raising activities. We had academics and financial executives test the first draft of the survey questionnaire in summer 2001 and revised it after incorporating their suggestions. Our final questionnaire is structured around nine topics and comprises about 100 questions. 1. Sample We select our sample of firms to maximize representation and to minimize firmspecific differences across European countries. We construct our initial sample from two sources. First, we include all non- European firms for which the Financial Journal La Tribune provides daily trading information. These firms represent different industries and most of them are either part of the national stock index of their country or of other market indexes such as European Nasdaq. We obtain a total of 621 firms from this source. We add another 116 firms that 3

comprise the SBF 120 index. We delete 17 firms because of non-availability of their addresses. Our final mailing group totals 720 firms. The data characteristics of our initial mailing group are for the year ending December 2001. We mailed the survey to the chief financial officers (CFOs) of all sample firms. 1 We kept the survey anonymous to facilitate honest responses. We made three mailings. The first mailing was done in September 2001, the second in November 2001, and the third in January 2002. We included a letter with each mailing that explained the objectives of the study and promised to send a summary of our findings to those who wished to receive it. We received 87 responses by mail or by fax, representing a 12% response rate. Insert Table I Table I compares the percentage of responses across countries and across,,, and Scandinavian law countries. The largest number of sample firms (about 45 %) belong to the law countries followed by the (21 %), (19 %), and Scandinavian (15 %) legal systems. France, y, and the U.K. have the largest number of respondents. This response rate is not surprising, because these countries also represent about half of the initial mailing group. The univariate and multinomial tests in Table I show that the proportions of our respondents across countries and legal systems are similar to those in our initial mailing group. We also find that the mean (median) market capitalization and P/E ratios of respondents, and the proportion of dividend paying respondents, are similar to those in the initial mailing group across all legal systems (not reported for brevity). Thus, nonresponse bias does not appear to be a major problem in our survey. 2. Summary Statistics of Respondent Firms Figure I presents the characteristics of respondent firms. Most of the respondent firms are large; over 75 % have sales and market capitalization over 1 billion. These firms represent several industries, and are concentrated in the manufacturing, mining, energy, and transportation (about 37 %), high technology (18 %), and financial sectors (18 %). High-growth firms, which we define as firms with P/E ratio greater than 14, comprise 65 % of the sample. 66 % of the firms are also widely held public firms, and 36 % have multiple classes of shares. Over 90 % are non-utility firms and pay regular dividends. Insert Figure I 1 We obtained the names and addresses of the CFOs from the Bloomberg database provided by BNP Paribas. We mailed the survey to the chief executive officer (CEO) if the name of the CFO was not available. A copy of the survey instrument is available on: http://www.escp-eap.net/faculty_research/publications/publications/fbancel/mdq6.pdf 4

About 75 % of the firms have a target debt-to-equity ratio. The estimated cost of equity for most firms ranges between 9 % and 15 %, and over 60 % of the sample firms use the capital asset pricing model (CAPM) to calculate this cost. The average long-term debt to total debt ratio is 66 %, but it varies from a minimum of 4 % to a maximum of 96 %. The average (median) debt ratio (total debt to market value of equity) is 113 % (35 %) and the average (median) long-term debt ratio is 44 % (24 %). Most of the respondents report that the financial policy of their firm is influenced primarily by their stockholders and to a far lesser degree by other stakeholders. Most of our respondents are also internationally oriented. 58 % have the majority of their sales in foreign countries, and have raised foreign capital during the last ten years. 45 % of the respondents are also cross-listed on foreign exchanges, and 16 % are listed on both European and U.S. stock exchanges. We also collect information on characteristics of the CEOs of the firms. The majority of the managers (87 %) own less than 5 % of their firm s stock (Figure 1-M), 58 % are between 50 and 59 years old, and 68 % have a Masters degree (40 % have an MBA) (untabulated). Table II presents the correlations among the demographics of the respondent firms. Panel A shows the variables used in previous studies on capital structure. The variables in Panel B are more relevant for analyzing the cross-country differences. The correlations among the variables are largely as predicted in the literature. For example, high P/E ratio firms are likely to have a lower D/E ratio, higher managerial stock ownership, a higher percentage of free-float shares, and a lower estimated cost of equity. However, the average strength of correlations among these variables is much lower than that in the Graham and Harvey (2001) study. This finding could reflect the differences in the demographic variables of firms in the two samples. For example, over 90 % of our sample firms pay regular dividends compared to less than 55 % in the Graham and Harvey study. These differences are likely to be more pronounced across different countries. Insert Table II II. Comparison of European and U.S. Managers Views Tables III through VI present the European and U.S. managers responses on debt, equity, convertible debt, and foreign capital-raising policies.. We ask managers to rank the importance of different factors on a scale of zero to four (with zero as not important and four as very important). A. European Views We summarize European managers views on their capital structure decisions and discuss their implications. 5

1. Debt Policy We ask European managers three questions on the debt policy of firms. Table III (Panel A) summarizes the responses on how firms choose the appropriate amount of debt. Executives cite financial flexibility as the most important factor (mean rank 3.39), followed closely by credit rating (mean rank 2.78). 2 Firm managers also care about the tax deductibility of interest (mean rank 2.59), volatility of earnings (mean rank 2.33), concerns of customers/suppliers about firm s financial stability (mean rank 1.97), and the potential costs of bankruptcy (mean rank 1.76). These concerns are consistent with the trade-off theory. Managers consider the transaction costs of debt and debt levels of industry peers as less important. Table III (Panel B) presents the managerial rankings of other factors that influence the debt policy. A majority of managers (70 %) try to minimize the weighted average cost of capital (mean rank 2.8). Over 40 % of managers issue debt when interest rates are low (mean rank 2.1) or when the firm s equity is undervalued by the market (mean rank 2.08). These findings suggest that managers use windows of opportunity to raise capital. Insert Table III Table III (Panel C) summarizes the responses on factors that drive a firm s choice between short- and long-term debt. To decide the maturity of debt most managers (77 %) use the matching principle, matching the maturity of debt with that of the assets financed (mean rank 3.1). 70 % of respondents also state that they issue long-term debt to minimize the risk of refinancing in bad times (mean rank 2.83) and about one third report that they issue short-term debt when they are waiting for the long-term interest rates to decline (mean rank 1.85). 2. Common Stock Policy Table IV presents managers views on the determinants of their firm s common stock policy. Over 77 % of the respondent firms have issued equity during the last ten years. A majority of the managers rank earnings per share dilution as the most important factor in their equity issuance decision (mean rank 2.72). Fifty-nine percent of managers (mean rank 2.67) also rank maintaining a target debt-to-equity ratio as important. This finding is consistent with the trade-off theory. Managers also appear to be actively involved in selecting the timing of equity issues. 59 % of managers report that issuing stock after a rise in the firm s stock price is an important factor (mean rank 2.61), and about 50 % also agree that the amount of stock over- or undervaluation is important in issuing equity (mean rank 2.44). 2 Executives we met to present our results explained us that it is really important to "negotiate financing when you don't need it". 6

Insert Table IV Many respondent firms issue common stock for employee stock option plans (mean rank 2.07). Managers also tend to issue equity when there are insufficient profits to finance investment activities and not when they are unable to obtain funds from other sources. This finding is contrary to the implications of the pecking-order theory. There is also little support for the signaling theory, as very few firms issue stock to give a better impression of the firm. 3. Convertible Debt Policy More than half of the respondent firms in our sample have issued convertible debt during the last ten years. Table V shows that managers value convertible debt highly as an inexpensive way to issue delayed common stock (mean rank 2.45) and for the ability to call or the flexibility to force conversion of convertible debt when they want to (mean rank 2.43). This result is consistent with the asymmetric information theory. Managers also cite the option to issue convertible debt when equity is undervalued (mean rank 2.4), and avoiding short-term equity dilution, as important advantages of issuing convertible debt (mean rank 2.16). This finding supports the responses on common stock policy. Few managers issue convertible debt because it is less expensive than straight debt, or to attract investors who are unsure about the riskiness of the firm. Factors that relate to agency theory, such as protecting bondholders against the actions of stockholders or managers, are considered unimportant. Insert Table V 4. Foreign Debt or Equity Policy 58 % of the respondent firms have raised capital in foreign markets. Managers of these firms cite hedging considerations as the most important factors (Table VI). Sixty-seven percent of managers (mean rank 2.7) cite providing a natural hedge and matching sources and uses of funds as important or very important. Favorable tax treatment relative to Europe and better market conditions are also ranked modestly important (mean ranking of about 2). We are surprised to find that managers consider the level of interest rates as important when issuing debt on the domestic market, but not when issuing it in foreign markets (mean rank 1.48). In summary, our evidence provides medium support for the trade-off theory but less support for the pecking-order and agency theories. For example, few firms use debt when recent profits are not sufficient to support firm s activities as predicted by the pecking order. Factors relating to the agency costs, such as to motivate managers 7

to work hard or to borrow short-term to reduce the chance that firm will undertake risky projects, are considered unimportant (Table III). Insert Table VI B. Univariate Tests: Ranking of Different Factors We compare European and U.S. managers rankings of different factors and across legal systems. 1. Comparison of European and U.S. Managers Rankings In Tables III-VI, columns 1 through 4, we compare European and U.S. managers rankings of different factors. Two observations follow from this comparison. First, that the relative rankings of most determinants of capital structure are strikingly similar between the two groups. This finding holds for all components of the capital structure choice. For instance, both the European and U.S. groups rank financial flexibility and credit rating as the most important factors of their debt policy, and earnings per share dilution the most important factor of their equity policy. Similar patterns emerge when we compare convertible debt and foreign capital raising policies. We test the equality of the paired relative rankings for all factors across the two groups, using the parametric t-test and nonparametric Wilcoxon Signed Rank test. The values of both test statistics (about 0.16) are not significant at any reasonable level. This result supports Hypothesis 1, that European and U.S. managers use similar factors in making their financing decisions. Our results confirm Graham and Harvey (2001) conclusions that there is modest support for the trade-off and pecking order theories, but less for the asset substitution, transaction costs, and free cash flow hypotheses. Second, although the relative rankings of different factors are almost identical across European and U.S. managers, the mean ratings for some factors, such as financial flexibility, are significantly different between the two groups (3.39 compared to 2.59). These differences could be attributed to both demographic and country-specific differences.. 2. Comparisons of Rankings Across Legal Systems The last ten columns in Tables III-VI present the t-tests for the equality of the mean ratings for each factor across the,,, and Scandinavian law systems. The t-tests show significant differences for several factors. Some of these differences appear to be consistent with the quality of the country s legal system. For instance, we would expect that concern for financial flexibility or for matching maturity would be higher in civil law systems compared to that in common law systems, because of less availability of external financing in civil systems (see La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1997, 1998). Our evidence supports this contention. 8

However, there are also some notable differences on some dimensions, even across systems of similar quality. For instance, although the quality of and Scandinavian systems is similar, managers concerns about the potential cost of bankruptcy are significantly different across systems (rating of 1.35 compared to 2.23). The Scandinavian managers appear to have significantly different views than their peers on common stock policy. For instance, all non-scandinavian managers (mean rank about 3) rank earnings per share dilution as the most important determinant of common stock policy, but Scandinavian managers consider it unimportant (mean rank 1.56).These differences might reflect the influence of firm- and country-specific factors. We need to control for these factors if we are to draw meaningful conclusions. III. Cross-Country Analysis We examine whether the cross-sectional differences in managerial rankings of different factors in our sample can be explained by the country- and firm-specific factors that are often used in the capital structure literature. A. Country-level and Firm-specific Variables We discuss country-level and firm-specific variables that we use in the cross-sectional regression analysis. 1. Country-level variables We use eight institutional variables as our proxies for the institutional environment of a country (see Table VII for the detailed description of the variables). From La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998) we obtain two variables that measure the quality of investors rights in the country. The creditors rights index (CRRHT) is an index that ranges between zero and four and measures the rights creditors have in bankruptcy. The shareholders rights index (SHRHT) is an index that ranges between zero and five and measures the rights of minority shareholders of the firm. Three variables measure the developments of capital markets in a country. DEBT/GNP is the ratio of the sum of bank debt of the private sector and outstanding non financial bonds to gross national product in 1994. We obtain this measure from La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997). TOVER is the ratio of the total value of shares traded on the stock exchange divided by the market capitalization in each country (based on 1989-1996 averages). MARKET is a dummy variable that takes the value of one for market-based financial systems and zero for bank-based systems. We obtain both MARKET and TOVER from Demirgüç-Kunt and Maksimovic (2002). We use three variables as our proxies for the corporate tax rate, cost of capital, and corporate ownership structure in a country. The firm s corporate tax rate (TAX) is the tax rate on undistributed corporate profits. The cost of capital (COC) is the pre-tax rate of return 9

required in each country on a hypothetical investment project with a post-tax required real rate of return 5 %. However, COC is determined using different sources of financing and thus reflects the average effect of corporate tax rates. We obtain this variable from Devereux, Spengel, and Lammersen (2003). They calculate the average for different types of assets purchased, such as intangibles, industrial buildings, machinery, financial assets, and inventories, by using the following weights for different financing sources: retained earnings (55%), new equity (10%), and debt (35%). Data are available for 14 of the 16 countries in our sample. We measure corporate ownership concentration by using the average percentage of common shares owned by the three largest shareholders in the ten largest nonfinancial corporations (OWN10) in the country, as compiled in La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1998). Table VII, Panel A shows that the mean (median) values for the country-specific variables are significantly different across legal systems for almost all variables. Further, many of the variables vary substantially across countries even within legal systems (not tabulated). For example, in the Scandinavian law countries. The creditors rights index varies from a low of one (Finland) to a high of three (Denmark), but the shareholders rights index varies from a low of two (Denmark) to a high of four (Norway). The corporate tax rates vary from a minimum of 25 % (Finland) to a maximum of 54 % (y) and the cost of capital from a minimum of 4.8 % (Italy) to a maximum of 7.5 % (France) in our total sample of countries. The mean ownership concentration in our total sample is 42 % but it varies from a minimum of 19 % (U.K.) to a maximum of 67 % (Greece). Insert Table VII 2. Firm-Specific Variables The capital structure literature suggests that leverage increases with firm size and decreases with a firm s growth opportunities, volatility of earnings, probability of bankruptcy, and managerial ownership (Harris and Raviv, 1991). To capture these effects we use five firm-specific variables for which the firms self-report data. SIZE is the natural log of the market capitalization of the firm and measures the firm size. P/E ratio measures the growth opportunities of a firm, and INDUSTRY (manufacturing, energy, and transportation compared to others) measures the industry effects. MGMTOWN is a dummy variable that takes the value of zero for firms with managerial ownership less than 5 %, and one otherwise. We also use two variables to examine the impact of international operations of a firm on its capital structure: FRNLST is an indicator variable for firms listed on foreign exchanges. FRNSALE is an indicator variable that takes a value between one and four, based on the percentage of the firm s foreign sales. Higher scores indicate a higher level of foreign sales. 10

Myers (1977) argues that firms with higher agency costs can either limit their total debt or use short-term debt to minimize underinvestment problems in the presence of agency costs. We use the percentage of short-term debt to total debt (%SHRTDEBT) as our proxy for agency costs. Firms in our sample that have a higher %SHRTDEBT are also more likely to say that they borrow short-term because it reduces the chance to undertake risky projects, consistent with the agency theory. However, we use caution in interpreting this variable, since about one third of our sample also use short-term debt as a temporary financing while they are waiting for the long-term interest rates to come down. Table VII, Panel B, shows that means (medians) of firm size, P/E ratios, and industry composition are similar across different legal systems. However, there are some notable differences for some variables. For instance, the legal system countries have a much higher percentage of foreign listed firms relative to the countries in other legal systems. On the other hand, the average percentage of short-term debt, is lower in and law countries (about 29 %) compared to and Scandinavian law countries (about 44 %). B. Cross-Sectional Regression Analysis For this analysis, we focus on the 11 factors that at least 50% of the respondents identify as major determinants of debt and common stock policies. For each factor, we estimate the following cross-sectional regression: [ Response Rank ] ij = α i + β 1i + β 3i [ Legal Systems ] j + β 2 [ Country - Specific i ] [ Firm - Specific ] + where Response Rank (ij) is the rating of factor i by respondent j and varies between zero and four,, with four as very important and zero as not important. The legal system variables are the three dummy variables representing the,, and Scandinavian legal systems, and the country and firm-specific variables are those just discussed. We perform the analysis in two steps. First, we run univariate regressions for each of the country and firm-specific variable and then run two multivariate regressions. The first multivariate regression includes only the three dummy variables representing the,, and Scandinavian legal systems. The second multivariate regression is the same as the first one but also includes those countryand firm-level variables that have significant coefficients in the univariate regressions at less than 0.10 level. For brevity, we do not report the results of the multivariate regressions but summarize our main findings and their implications in Table VIII. We first discuss the results for each of the factors examined and then summarize our findings. j i j (1) 11

Insert Table VIII 1. Debt Policy Factors Financial Flexibility: The need for financial flexibility should be inversely related to the size of capital markets and the quality of legal system of a country. Our results show that while flexibility is negatively related with creditors rights (CRRHT) in individual regressions, neither DEBT/GNP nor TOVER is significant in the individual regressions. However, financial flexibility is less of a concern for managers in countries with market-based financial systems. In fact, the coefficient of MARKET remains robust to the inclusion of other country- and firm-specific variables in the regression, but CRRHT loses significance in this regression. Firms with a larger managerial ownership are also more concerned with financial flexibility. This finding is consistent with Jensen s (1986) agency theory. Financial flexibility is also positively related to the degree of foreign sales of the firm. This result could reflect the higher growth opportunities and, consequently, a higher need for flexibility for such firms. Credit Rating: Since credit rating provides a signal to investors about the quality of a firm, the value of and need for this signal is likely to be higher in countries with weak shareholders rights (SHRRHT) and creditors rights (CRDRHT). Contrary to our expectations, neither of the two variables is significant in the individual regressions. Credit rating is also valued more by firms in countries with a relatively higher cost of capital, which suggests that its ranking could be driven by the need of firms in such countries to tap international capital markets. To examine this issue, we run a multivariate regression by including the three legal system variables and an indicator variable (FRNISSUE) for firms that have issued debt or equity in foreign markets during the last ten years. The coefficient of FRNISSUE is positive and is the only significant variable in this regression (coefficient 1.06, t-value 3.95) that supports the idea that the concern for credit rating is influenced by the firm s need to access international capital markets. Firm size is also positively related to this factor, but most of its significance comes from firms that have a credit rating. Tax Advantage of Debt: We expect to find that the cross-country variation in tax advantages of debt is positively related to tax and cost of capital variables. Our analysis shows that the coefficient of cost of capital is positive and significant at the 0.1 level in the individual regression, as predicted, but that the coefficient of TAX has a negative sign, opposite to what we expected. The results are qualitatively similar when we use the other proxies of corporate tax rates discussed in La Porta, Lopez-de- Silanes, Shleifer, and Vishny (2000). 12

The nonsignificance of tax variables could be because of the difficulty in estimating the effective tax rates for firms. Effective tax rates are likely to be different from the statutory tax rates (see Graham, 1996). We note that multinational firms and firms with a higher percentage of short-term debt are also more concerned about the tax advantage of debt. This result support the idea that firms decide their optimal capital structures by using different trade-offs of costs and benefits. Earnings Volatility: Since firms with high earnings volatility have a higher probability of default, investors are less likely to provide financing to such firms, especially in countries with weak creditors rights. The results support this prediction. We find that this factor is negatively correlated with the creditors rights index, and is the only country-specific variable that is significant in the multivariate regression. This result is also consistent with the agency theory, because agency costs of debt are likely to be higher in countries with weaker investor protection. Weighted Average Cost of Capital (WACC): This factor is likely to be positively related to the cost of capital and tax variables, but neither of the two is statistically significant in any regression. This result could be partly attributed to the problems of measuring the appropriate tax rates for different firms. However, firms with a higher percentage of short-term debt have significantly lower concerns about this factor. This lack of concern suggests that in deciding their optimal capital structures firms may trade off the reduction of agency costs against the advantages of debt. Matching Maturity: To manage the risk of refinancing in bad times, managers commonly match the maturity of assets with liabilities. The need for and value of this tool are likely to be higher for firms located in countries with a lower level of the quality of their legal systems. Our results support this prediction, as managers in civillaw countries value this factor much more highly than do their peers in common-law countries. Financing in Bad Times:This factor is also likely to be inversely related to the quality of the legal system. Our results support this prediction. The creditors rights index is the only significant country-level variable. It has a negative coefficient, as expected, and also remains robust to the inclusion of firm-specific variables in the regression. Firms with a higher percentage of short-term debt are also less concerned about financing in bad times, a finding that is consistent with their views on matching maturity and minimizing the cost of capital. 2. Equity Policy Factors Earnings Per Share Dilution (EPS): The results show that managers in Scandinavian legal system countries are significantly less concerned about the EPS dilution than are their peers in other countries. Two country-specific variables, DEBT/GNP and OWN10, are also significant in separate individual regressions, with a positive and 13

negative sign, respectively. However, neither is significant when we include both variables in the regression. We find similar results when we analyze the factor pertaining to the use of debt because of close relationships with a bank that is also ranked significantly higher by the Scandinavian managers compared to other managers (untabulated). This result suggests that DEBT/GNP and OWN10 may be proxies for the effect of some omitted factors that differ across Scandinavian and non-scandinavian countries, or they could just reflect random chance. For example, the ranking of this factor may depend on whether the managerial compensation in the firm is based on the earnings per share or not. Target Debt-to-Equity: The trade-off theory predicts that this factor should be positively related to the tax and cost of capital variables. Alternatively, managers may use the target debt-to-equity ratio primarily to manage the riskiness of debt. In that case it should be negatively related to the quality of legal systems, because of the potentially higher cost of bankruptcy in systems with poor creditor protection. The results support the latter explanation, because managers in civil-law systems have significantly higher concerns for maintaining target ratios relative to their commonlaw peers. Managers of multinational firms and foreign listed firms are also more concerned, and managers of firms that rely more on short-term debt are less concerned about maintaining target ratios. This finding implies that firms may decide their capital structure by trading off different costs and benefits. High Stock Price: This factor is related to managers use of windows of opportunity in issuing common stock and is likely to be a universal factor. The results generally support this prediction, because none of the country-specific variables are significant when we include the firm-specific factors in the regression. The price to earnings ratio (P/E) is the only significant explanatory variable in this regression. Growth firms are likely to rely more on equity financing and may be more concerned with the stock price at the time of the equity issue. This finding is consistent with the asymmetric information theory. Under- or Overvalued Stock: This factor is also related to managers beliefs about valuation of the firm s stock. We expect it to be portable across countries. The results support this prediction. High-growth firms are also more concerned about stock valuation when issuing equity. This finding supports the asymmetric information theory. The evidence shows that cross-sectional differences in the rankings of several factors including earnings volatility, matching maturity, financing in bad times and target ratios are explained primarily by the legal system variables in a predictable 14

manner. For example, firms in countries with lower protection of creditors rights are more concerned about earnings volatility and financing in bad times, as expected. This evidence provides moderate support for hypothesis 2 but less support for hypothesis 3 because most of these factors relate to the debt policy and not the equity policy. Our evidence also supports that increased accessibility to global capital markets also influences a firm s ranking of some factors. B. Robustness Checks We conduct several robustness checks to examine the sensitivity of our results. First, we redo our analysis by replacing legal system dummies in the regressions with the country-level indexes for effectiveness of law and rule-of-law as compiled and discussed in La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998). The results are qualitatively similar to those presented above when we use country-level indexes although the significance of creditors rights index increases in most regressions. We use an ordered probit model instead of an ordinary least squares model for analysis. The estimated coefficients and their significance in all regressions are similar in both models. We also examine the sensitivity of our results to the CEO characteristics and other demographic variables such as influence of shareholders (INFLSHR) discussed in Table II. Our results remain essentially the same. We also compare our results with those in Brounen, Jong, and Koedijk (2004) who survey managers from four European countries, U.K., The Netherlands, y and France on the theory and practice of corporate finance. They analyze cross-country variation in managerial rankings of four determinants of capital structure in their sample: flexibility, target ratios, tax advantage and bankruptcy costs. It is reassuring that they also report that flexibility and tax advantage factors do not vary significantly across countries, consistent with our findings. However, they find that firm size is the dominant explanatory variable and national differences play only a weak role in explaining cross-sectional variation in factor rankings in their sample. In contrast, we find strong evidence that debt policy factors vary systematically with the quality of the legal system in the country. One plausible explanation for these differences could be that their sample comprises of both large and small firms while our sample consists mainly of large and well-known European firms. In addition, our conclusions are based on the analysis of eleven factors compared to four examined in their study. We also examine factors that are ranked as unimportant (minor) by managers, but find that their rankings differ significantly across legal systems (Tables III-IV). Our analysis (not tabulated for brevity) shows that the firm-specific factors explain most of the variation in the minor factors, and that the institutional variables have very little explanatory power. Firm size and growth opportunities are the major 15

explanatory variables for the factors relating to debt and equity policies, respectively. Further, the correlations of the firm-specific variables with the factors are consistent with the predictions of the capital structure theories. For instance, large firms care less about the potential cost of bankruptcy, a finding that is consistent with their lower agency costs. Firm size also explains the differences in bankruptcy costs for the and Scandinavian countries that we noted in Table III, Panel A (b). Larger firms with insufficient recent profits also care less about issuing debt. This result is consistent with their lower agency costs and asymmetric information relative to their smaller peers (see Table III, Panel B (a)). Table IV (d) shows that high-growth firms are more likely to consider common stock is the cheapest source of funds, which is consistent with the asymmetric information theory. Industry and managerial ownership also have explanatory power in some cases, which supports the theoretical literature IV. Summary and Conclusions In this study, we compare managerial views on the determinants of capital structure across 16 European countries. Our goal is to gain some insights into the following questions: Are financing policies of European and U.S. firms driven by similar factors? Are cross-country differences in managerial views explained primarily by the legal institutions, or do other institutions also play a major role? And which factors underlying debt and equity policies are more sensitive to the institutional environment and which are portable across countries? Our conclusions are as follows. First, European managers use factors similar to those used by their U.S. peers for their financing decisions. However, there are differences across countries on several dimensions, especially between Scandinavian and non-scandinavian countries. Second, the quality of the country s legal system explains cross-country variations in the rankings of several major factors, but so do other country-specific factors such as cost of capital. In addition, although differences in debt policy factors vary systematically with the quality of a country s legal system, firm-specific factors such as the firm s growth opportunities strongly influence the common stock policy factors. Our results should be interpreted with some caution because of potential biases and measurement problems that are normally associated with survey data. Surveys measure beliefs and may not represent the reality in the field. Another concern is that managers responses may not appropriately reflect their views on the implications or assumptions of the theories. In addition, the problems of endogenity and high correlations among some of the variables used in our analysis weaken the statistical power of our tests. We find that the Scandinavian managers views on capital structure differ significantly from their other civil-law peers, especially on equity, convertible debt, and raising foreign capital. These differences 16

could reflect random chance, or differences in the population of firms, or the effect of other institutions, such as moral and ethical norms not accounted for in our study. However, we are unable to pinpoint the source of these differences because there are several factors, such as the effective tax rates and ownership structure for firms that we cannot measure. To examine some of these biases, we present our results to two different groups, business executives and senior investment bankers. Both groups agreed with our major conclusions. They also provided some examples based on their experience that illustrated how country-specific institutions and regulations could impact the financing choices of firms across countries that have the same legal system. Their feedback indicates the robustness of our findings and also suggests future research, in that cross-country research requires more refined country-level variables for more informative analysis. Nevertheless, our study provides insights into cross-country determinants of leverage that are difficult to obtain through traditional empirical studies. We show that factors related to debt are influenced more, and those related to equity are influenced less, by the country s institutional structure, especially the quality of its legal system. This evidence strengthens arguments of La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998) that the availability of external financing in a country is influenced primarily by its legal environment. Since agency costs of debt are likely to be higher in countries with lower quality of legal systems, this evidence is also consistent with theories of capital structure such as agency theory that assign a central role to debt contracts and bankruptcy law (Harris and Raviv, 1991). However, we also show that firms can adopt strategies to mitigate the negative effects of the quality of the legal environment in their home country. For instance, firms in civil-law countries have significantly higher concerns for maintaining target debt to-equity ratios and matching maturity than do their peers in the common-law countries. Further, we find that firms operating internationally have significantly different views than do their peers in several ways. For example, firms that have issued foreign debt or equity in our sample during the last ten years are more concerned about credit ratings. We also show that firm-specific variables that are commonly used in the capital structure literature to explain leverage also explain cross-country differences in managerial rankings of several factors. For example, large firms are less concerned about bankruptcy costs, and high growth firms consider common stock as the cheapest source of funds and use windows of opportunity to issue common stock. These results support that firms capital structures are the result of a complex interaction of several institutional features as well as firm characteristics in the home country, as argued by Rajan and Zingales (1995, 2003). Overall, our results support that most firms determine their optimal capital structure by trading off factors such as tax advantage of debt, or bankruptcy costs, agency costs, and 17

accessibility to external financing. We also confirm the conclusions of Titman (2001): Corporate treasurers do occasionally think about the kind of trade-offs between tax savings and financial distress costs that we teach in our corporate finance classes. However, since this trade-off does not change much over time, the balancing of the costs and benefits of debt financing that we emphasize so much in our textbooks is not their major concern. They spend much more time thinking about changes in market conditions and the implications of these changes on how firms should be financed. 18