Corporate finance in Europe: confronting theory with practice.
We present the results o fan international survey of 313 European CFOs on capital budgeting, cost of capital, capital structure, and corporate governance. We find that although large firms often use present value techniques and the capital asset pricing model to assess the feasibility o[an investment
**********
In this article, we conduct a survey on how professionals deal with different dilemmas within modern financial management. We measure the extent to which theoretical concepts have been adopted by professionals from a wide range of firms from the UK, the Netherlands, Germany, and France.
Recent studies have documented fundamental differences between the financial markets and systems when comparing the United States with Europe. La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1997, 1998) focus on the underlying disparities between the legal systems encompassing both continents, as well as on the relation between legal systems and the development of capital markets. Rajan and Zingales (2003) stress the continental differences by comparing the polar forms of financial systems: the institution-heavy relationship-based, more prevalent in Europe, and the market-intensive arms' length, more prevalent in the United States. Finally, from a corporate governance perspective, Chew (1997) shows how the Anglo-Saxon marked-based corporate governance system differs significantly from the relation-based or insider system, which is most widespread in Europe. In this study, we investigate the effect of the corporate governance system on individual firms and include this important issue in our overall analysis of European corporate finance practices. We conclude that the US and European financial markets and firms differ considerably. We contribute to the debate in the current literature by comparing the corporate finance practice of individual firms in both continental markets. We test whether the apparent differences in institutional settings translate into significantly different financial management practices.
To address theory with the behavior of financial managers in practice, we apply survey research. (1) We analyze many corporate finance issues, ranging from capital budgeting techniques to capital structure and corporate governance. Doing so allows us to link the different issues and to deepen our analysis.
Furthermore, we analyze the responses in our survey conditional on firm-specific characteristics. This approach enables us to test whether these factors drive the results. We sample a cross-section of 6,500 companies from the UK, Netherlands, France, and Germany. We collect 313 responses. The size of our sample represents one of the largest survey samples in the financial literature.
Survey research is relatively rare within the empirical corporate finance literature, in which most studies are based on large samples of financial observations. Although these large samples offer cross-sectional variations and the statistical power to analyze these variations, they are limited in their ability to deal with non-quantifiable issues. Our approach combines a relatively large sample with the ability to ask qualitative questions.
Survey research is also associated with some limitations. We measure beliefs rather than actions. In doing so, we implicitly assume that managers "do what they say they do." To test this assumption, we consider the consistency of the answers and where possible, compare our survey evidence with other research. Moreover, the anonymity of our survey stimulates frank responses. Another limitation of survey research is potential respondent bias. We take this drawback into consideration when we compose our samples and construct our questionnaire. Thus, we are able to limit this bias to the minimum.
By using an international sample we are able to assess whether existing insights on corporate finance practices documented by Graham and Harvey (2001) also hold outside the US. Furthermore, we address the corporate governance policy of firms, which enables us to investigate whether corporate governance differences influence the way in which firms organize their financial management. Finally, we extend the univariate results of Graham and Harvey (2001) by using multivariate regression analysis.
This study complements and adds to Bancel and Mittoo's (2004) survey of European CFOs (published in this same issue). We complement Bancel and Mittoo's work as we include questions on capital budgeting and cost of capital estimation and we study both public and private firms. Bancel and Mittoo focus exclusively on the debt policies of publicly listed corporations. Our data set facilitates cross-country comparisons, while Bancel and Mittoo cluster countries into four legal systems (their study covers 87 observations from 16 countries). We complement Bancel and Mittoo's analysis of legal systems and country-level governance characteristics, because we include firm-level corporate governance characteristics. We limit our discussion of the capital structure results to a comparison of the relative importance of the static trade-off and pecking-order theories.
Our results on capital budgeting show that European firms are still remarkably keen on applying the payback criterion, instead of discounting their cash flows by using the internal rate of return (IRR) or the net present value (NPV). Similar to their US colleagues, European CFOs determine their cost of capital using the capital asset pricing model (CAPM), rather than applying arithmetic-average historic returns or the dividend discount model.
Overall, we find that firm size is positively related to the use of the discounted cash flow method and the application of the CAPM. Smaller firms and firms less oriented towards maximizing shareholder value are more likely to evaluate their investment opportunities by using the payback period criterion and setting their cost of capital at whatever level their investors tell them.
For capital structure, we find smaller disparities between corporate debt policies. In all four national samples, respondents report that financial flexibility is the key factor when determining their debt structure. This result corroborates previous studies from the US.
Our main results show that corporate financial management practices are predominantly determined by firm size, to a lesser extent by shareholder orientation, and least by country of origin. Interestingly, we can relate our findings on the role of shareholder orientation to the international differences in legal systems and capital markets documented by La Porta et al. (1997, 1998), Rajan and Zingales (2003) and Chew (1997). We confirm that shareholder orientation prevails in the UK and in the Netherlands, but in the German and French firms shareholders are less important. We also find that in capital budgeting, the orientation towards shareholders induces managers to apply techniques that are based on maximizing the wealth of these stakeholders. However, in capital structure choice, we find neither a role for shareholder orientation nor strong country differences. Apparently the fundamentals of capital structure choice are independent of legal system and capital market development.
The article is organized as follows. In the next section, we present the sample collection procedures and sample statistics. Section II offers a comprehensive overview of our results on capital budgeting. Section III deals with the common practices regarding the cost of capital. Section IV focuses on our capital structure results. Section V concludes.
I. Data and Method
This section details the procedures we used to obtain our data and the robustness tests we performed. We also present our sample statistics.
A. Sample Collection Procedures
Our survey comprises four groups of questions. First, we use several questions to describe the firm and its CEO. Next, we pose questions on the firm's capital budgeting techniques and the ways in which the firm estimates its cost of capital. We continue by focusing on capital structure policy. We finish our questionnaire by asking firms about their goals and their perception of the importance of different stakeholders.
The starting point for our questionnaire is the survey of Graham and Harvey (2001). To facilitate a fair comparison of both sets of survey results, we ask exactly the same questions. In addition we add questions on the firm's goals and stakeholders.
We surveyed firms in the UK, the Netherlands, Germany, and France. The survey of Graham and Harvey (2001) has been translated into German and French by a certified translation agency and into Dutch by the authors. Next, in order to test whether the translations were correct and whether the wording was understood, we conducted several interviews in each of the four countries. In these interviews, potential respondents first filled out the questionnaire, and then discussed each question. We learned that the average time to fill out the questionnaire was about 15 minutes. We adjusted some of the wording and added brief explanations, based on the interviewees' feedback.
We use the Amadeus data set of Bureau Van Dijk as our sample universe. This data set covers public and private firms in Europe. From it we select all firms with 25 or more employees. We also use the Kompass database, which gives us the names and positions of the high-ranking officials. We search for the name of the CFO in the Kompass data for each firm in the Amadeus data. Our goal is to select 2,000 firms in the UK, Germany, and France, and 500 firms in the Netherlands. We first select all public firms in each country. Then, we complement our sample sets with randomly chosen private firms for which we know the name of the CFO.
The questionnaire was sent out by a third party. This approach ensures that the results are handled anonymously, which stimulates respondents to answer frankly. Between November 1 to 8, 2002, the mailing firm sent the questionnaires and mail to the sample firms. Each firm received a cover letter, the four-page questionnaire, a pre-stamped envelope, and a response form for requesting a free report of the results. The latter served as an incentive to fill in the questionnaire. The respondents could return their questionnaire and form by mail or by fax.
About two weeks after the firms received the questionnaire, our survey partner contacted all non-respondents by phone by native speakers, and reminded them to return the questionnaire. During the phone conversation, the respondents could either answer the questions over the phone immediately, or receive an email link to a web page for filling in the questionnaire. This telephonic and email effort lasted until January 7, 2003. We received our last response on January 30, 2003.
In total, we received 313 responses, 68 in the UK, 52 in the Netherlands, 132 in Germany, and 61 in France. (2) We received 50.5% of the questionnaires by mail or fax, 19.2% by telephonic interviews, and 30.3% through the web page.
In analyzing our results we paid particular attention to potential response biases, which threaten survey research.
We first investigate whether the questionnaires show a bias caused by the type of response medium or by the sequence of questions. We cluster our results according to the way in which the responses were received (mail, fax, telephone, or Internet) and analyze both the average responses and the distributions within each cluster. We obtain a total of 146 items and four response clusters. Using a standard mean-test for all six comparisons between the four clusters, we find 87 differences significant at the 10% level. This finding implies that the results are not biased, because for a random set we expect 88 significant differences (10% of 146 times six). At the 5% level, we find 66 differences and expected 44. We find no distinct patterns in the differences between clusters.
For our second test we sent out two versions, this time interchanging questions 1-4 and 11-14. We find five differences between the two sets at the 10% significance level, where a random set would yield 15 differences (10% of 146). At the 5% level we expected seven differences and find four. Thus, we detect no significant differences in responses based on the questionnaire structure.
We also perform an experiment to investigate whether our results are affected by nonresponse bias. We follow the example of Moore and Reichert (1983) by comparing characteristics such as firm size, industry, and public status of the responding firms to the nonrespondents. We find no statistically significant differences between the two groups at a 5% confidence level.
Overall we find that our sample is representative of the overall universe of firms, and we detect only a small variation in answers based on the response technique. The overall response rate is 5%, which is somewhat lower than studies such as those by Trahan and Gitman (1995) and Graham and Harvey (2001), which obtained 12% and 9% response rates, respectively. However, given the length and depth of our questionnaire and the vast size of our sample, we feel confident when analysing our results.
B. Corporate Governance Characteristics
La Porta et al. (1998) describe institutional details for 49 different countries, including the five countries that are part of our study. Their results clearly show that external capital is most important in the US, UK, and the Netherlands. The importance of the capital markets in the US and UK is further stressed by the large number of listed firms and IPOs per million inhabitants. Furthermore, La Porta et al. (1998) report cross-national statistics on the power of shareholders and creditors, using a anti-director index. This anti-director index measures the power of shareholders, which is much higher in the Anglo-Saxon countries. Finally, creditor rights also differ substantially across countries in our sample and illustrate the large variation in institutional settings. To correctly incorporate the fundamental differences in national market characteristics, we include questions on corporate governance, i.e., important stakeholders and company goals. By doing so we test whether the individual firms in our sample reflect institutional variations, and we can control for these variations in our further analysis.
First we ask our respondents which goals their companies aim to achieve. Panel A of Figure 1 shows that in all countries firms aim at maximizing their profits, sustainable growth, and market position, and that they give leverage and dividends lower priorities.
[FIGURE 1 OMITTED]
The most prominent distinction is that of the goal of maximizing shareholder wealth. Dutch and British firms declare to consider shareholder wealth as one of their top priorities, but French and German firms consider this goal even less important than optimizing their leverage.
To extend our analysis of this phenomenon we ask an additional question on stakeholder importance. The outcomes, which are presented in Figure 1, Panel B, show high customer importance in each country. The results on the importance of shareholders and debtholders are more scattered and clearly show that CFOs in the UK and the Netherlands consider their suppliers of capital to be much more important than do their colleagues from France and Germany. Since our sample contains relatively high fractions of private firms in the French and German samples (88% and 77%, respectively), this proportion of private firms may explain the lower scores on shareholder importance in these countries. However, the scores on the importance of shareholders for public firms are 2.63 in Germany and 1.43 in France, while the UK and the Netherlands score 3.38 and 3.1 (results not tabulated). This result corroborates our conclusion that shareholder orientation is lower in France and Germany. The French and German firms consider the general public to be more important to them than are their financiers.
C. Firm Statistics
Figure 2 presents summary information on the characteristics of the firms in our European samples and compares these firms with the US firms in Graham and Harvey (2001). The companies in our European sample are smaller, on average, compared to Graham and Harvey's US firms. Although 51% of all firms in their US sample have sales exceeding US$500 million, this number is less than 25% in each of our European samples. In our analysis we refer to firms with sales exceeding US$1 billion as "large firms."
[FIGURE 2 OMITTED]
We observe the opposite disparity in the proportion of foreign sales. This component exceeds a quarter of total sales for at least 40% of each European sample, but the US firms exhibit remarkably lower involvement in foreign sales.
The distribution across industry types is similar in all countries. In each sample most firms are engaged in manufacturing. Like Graham and Harvey (2001) for their US sample, in our European samples we find that nonmanufacturing firms are spread evenly across other industries.
We find slightly lower average values for the price-earnings ratios in our European sample. Only 40% of the US firms reported a price-earnings ratio below 15, but our European firms reported this response more frequently, ranging between 56% in France to 67% in Germany. However, this result needs to be treated with caution, given the time difference between both surveys.
Panel E of Figure 2 displays information on corporate debt policy. The long-term debt ratios show that about a quarter of the firms in the UK and France possess no long-term debt at all. These firms are financed completely with equity and short-term liabilities. The German firms are over-represented in the 10% to 19% interval. Many Dutch firms are in the 20% to 29% interval. Therefore, we define low-levered firms as firms with leverage below 30%, and high-levered firms as those with a debt ratio above 30%.
The fractions of firms with low and high leverage are not very different between the countries. The only exception is France, which is under-represented in the highest interval. This international pattern in leverage complies with previous studies of Rajan and Zingales (1995) and De Bondt (1998), who document similar national differences and explain them by emphasizing the institutional differences.
The next component of our summary statistics concerns the CEO's background. On average, our results indicate that European CEOs in our sample are slightly younger than are their US counterparts. The information on their tenure the variation is less strong. The most remarkable result on tenure stems from France, where CEOs appear to stay with their firms for significantly longer time periods than do their colleagues from other markets. However, the French sample has a relatively high fraction of private firms, which may explain the longer tenures in France. Additional unreported analysis shows that the percentages of CEOs with the maximum tenure in public firms is 57% in France and 32%, 31%, 24%, and 17% in the US, UK, Netherlands, and Germany, respectively. Therefore, we conclude that this result is not driven by the high fraction of private firms. (3)
Our results show comparable patterns in the level of education of the surveyed CEOs. Compared to their US colleagues, a smaller portion of European CEOs quit their studies after their undergraduate, and a slightly higher portion acquires an MBA. The exception is in the UK, where MBAs are rare.
Our results show very little evidence of cross-national patterns in executive stock ownership. In each sample the majority of firms respond that their executives own less than 5% of the firms' shares.
Panel J of Figure 2 presents the key results that we obtain when we gather some summary statistics on the public listing of the company, dividend policy, and credit rating. Like the US firms in Graham and Harvey's (2001) study, most of our European firms are not utilities,
do pay dividends, and do have an investment grade rating.
II. Capital Budgeting Techniques
The evaluation of new investment projects requires capital budgeting techniques. This section investigates the capital budgeting practices.
A. Design
To link the results to differences in, for example, firm size and CEO education, we consider the underlying firm characteristics. As do Graham and Harvey (2001) we use a variety of capital budgeting techniques: discounted cash flow techniques like the IRR, NPV, adjusted present value (APV) (see Brealey and Myers, 2003), discounted payback period, profitability index, and hurdle rates; next to price earnings multiples, book rates of return; and then more advanced methods like sensitivity analysis, real options, and value at risk. We asked respondents to score how frequently they use the capital budgeting techniques on a scale of zero to four (zero meaning never, four meaning always). We display the main results in Table I.
B. Results
Most European respondents select the payback period as their most frequently used capital budgeting technique. In the UK, Netherlands, Germany, and France, 69.2%, 64.7%, 50%, and 50.9%, respectively, of CFOs use the payback period as their favorite tool. Of the US firms cited by Graham and Harvey (2001), 56.7% said they used this payback rule, but it was the third most popular tool after the IRR and NPV. In Europe, following the payback period criterion are the NPV and IRR methods. In the UK, Netherlands, Germany, and France 53.1%, 56%, 42.2%, and 44.1%, respectively, of all CFOs use the IRR method, while 47%, 70% 47.6%, and 35.1% of all CFOs in these countries rely on the NPV method.
The relative popularity of the payback period in Europe is surprising. The payback period ignores the time value of money and cash flows beyond the cut-off date. Some researchers argue that the payback approach is rational for severely capital-constrained firms: if an investment project does not pay positive cash flows early on, the firms will close the operation and therefore cannot receive positive cash flows that occur in the distant future.
Like Graham and Harvey (2001), we do not find any evidence to support this claim. When taking firm characteristics into account, we see that the use of the payback criterion is more popular among smaller firms (except for the UK) and among firms with management that falls in the highest age cluster. Large firms and firms managed by a CEO with an MBA (except for the UK) use NPV significantly more often. This finding might also explain the difference between our European results and the US outcomes of Graham and Harvey, since we have already noted that our European firms tend to be somewhat smaller on average.
The payback criterion is also more popular among private companies than among public corporations. (4) When we account for the cross-sectional variation in shareholder orientation, we find that firms that report that they maximize shareholder value are the same firms that prefer to use discounting techniques rather than the plain payback criterion. Theory shows that this indeed enhances shareholder wealth.
III. Cost of Capital
When applying discounting techniques firms need to identify a proper cost of capital. This section investigates the ways in which the cost of capital is derived and applied.
A. Results
The first question we ask on the cost of equity capital is whether firms explicitly compute this cost. The response to this question shows a very limited cross-national variation, i.e., 64.2% in the US, 57.4% in the UK, 59.6% in the Netherlands, 59% in France, and 53% in Germany.
We continue our analysis by focusing on the firms that responded positively by first asking them how they compute their cost of equity capital. We explore whether firms use the CAPM, a multi-beta CAPM (with the market beta plus extra risk factors), average historical returns, or a dividend discount model, or whether they apply the average historic return on common stocks or whatever the firm's investors tell them they require.
The results in Table II indicate that the CAPM is the most popular method of estimating the cost of equity capital in Europe: in the UK, Netherlands, Germany, and France, 47.1%, 55.6%, 34%, and 45.2%, respectively, of CFOs rely on the CAPM for estimating the cost of equity. Although the CAPM is a popular method in Europe, our results also show that its popularity is lower than in the US. Graham and Harvey (2001) report that almost 73.5% of US CFOs rely to some extent on the CAPM when estimating the cost of equity capital. In Europe, this percentage is considerably lower and averages around 45%. (5)
The second and third most popular methods in the European countries are, respectively, the use of average historical returns and the use of some version of a multi-beta CAPM. Again, the percentages for the European countries are substantially lower than are those in the US. An explanation for this discrepancy might be the public or private status of a firm. (6) In our cross-sectional analysis we find that in each national sample, public firms are more likely to use the CAPM for deriving their cost of equity capital, but private firms use whatever their investors tell them. This difference is rational, since public firms have stock prices at their disposal, which they will need to run the CAPM properly. Due to the absence of public stock returns, private firms prefer to use whatever investors tell them when discounting their cash flows. (Private firms can use "beta books," so it is possible that private firms will use CAPM as much as will public firms.) In the Netherlands, Germany, and France, the percentages for this category vary between 44.8%, 39.2%, and 34.4%, respectively. In the Netherlands and France, "whatever our investors tell us they require" is the second most popular after the CAPM. In Germany, this method is the most popular method and outperforms the CAPM as a way to obtain an estimate for the cost of capital.
When we look at the underlying firm characteristics, we find that the CAPM is consistently more popular among large firms, and among firms with relatively high proportions of foreign sales. The same holds for the more advanced CAPM alternatives that use additional risk factors. These alternatives are used mostly by large companies and by firms with relatively high leverage. This result indicates that large, public firms are more inclined to apply more sophisticated techniques when setting their cost of capital, but small firms rely on rules of thumb. However, this difference is not due to a lack of familiarity with the theoretical concepts, since there appears to be no relation between the age and education of the CEO and the use of theoretical tools like the CAPM. We do find though that CEOs with long tenures use CAPM more frequently. Apparently CEOs learn to appreciate the use of CAPM while they are in charge of the company. This on-the-job learning implies that CAPM requires a critical mass (size), a public listing, and a CEO with long tenure.
B. Specific Risk Factors
We now explicitly analyze individual projects. We identify a variety of specific risk factors that might be important when evaluating an individual project. These factors are interest-rate risk, foreign-exchange risk, business-cycle risk, unexpected inflation, commodity-price risk, term-structure risk, and distress risk. We follow Fama and French (1992) and Jegadeesh and Titman (1993) by including the fundamental factors size, value, and momentum. We ask our respondents whether they take these individual factors into account when valuing projects, and if so whether they adjust their discount rate, the cash flow estimations, or both. We display our key results in Table III.
Overall, we find a strong tendency to omit most of the specific risk factors. This result complies with the average response of US companies on the same issue. The majority of firms do not take specific risk factors into account when evaluating individual investment projects. If they consider any risk factors, they will look at interest-rate and currency risk. In most of those cases, firms acknowledge these risks by adjusting either the discount rates or the cash flows. Only a small minority of firms consider momentum. The exception is France, where 27.8% of our respondents say they adjust their discount rate based on recent stock price performance. We also note that 26.3% of German companies and a remarkable 46.6% of French companies are likely to amend cash flow estimations according to their perception of commodity price risk.
C. Project Compared to Firm Risk
We concentrate on the use of discount rates when firms consider new projects in overseas markets. Graham and Harvey's (2001) US results on the use of discount rates are surprising. The majority of firms in their sample claim to use the "plain vanilla" firm discount rate when they evaluate new foreign projects. This finding implies that most companies do not properly incorporate differences in project risks and foreign markets. By posing the same questions to our respondents we determine whether the same management approach prevails in Europe.
Table IV presents the main results of this exercise. We find that, except for the French, European CFOs, like their US counterparts, prefer to use a plain company-wide discount rate rather than more sophisticated risk-matched rates. We find that the level of popularity of the discount rate of the entire company is comparable to the US results reported by Graham and Harvey (2001), and that in each country, large, public firms are more likely to use the companywide rate than are their small, private competitors.
For the second most popular discount rate, the risk-matched project rate, we see that the European firms are remarkably less keen to apply this rate compared to the US firms. In the US, 50.9% of the firms said they applied a risk-matched rate always or almost always. In our European sample, the percentage ranges between only 23.7% and 27.3%. Again we find that large firms are more likely to choose the risk-matched rates. Similar to the US, the remaining three alternatives (the use of a discount rate for the overseas market, a divisional discount rate, or a different rate for each component cash flow that has a different risk characteristic) are almost never used in each European sample. Large firms are more likely to apply these theoretically more sound risk-matched rates.
Except for the Netherlands, we also find that the CEOs' education tends to increase the likelihood of using risk-matched projects rates. Contrary to Graham and Harvey (2001), our results show that highly educated CEOs are more likely to use the more complicated alternatives of discounting new projects. This result is consistent throughout all our national samples.
We note that older CEOs are more likely to use the simple company-wide discount rate, and that younger CEOs are more likely to use the more complicated project specific rates. (7) We find a similar variation when we examine the shareholder orientation of the firms involved. Firms that say they maximize their shareholders' wealth tend to use more complicated firm-specific rates. We also find that public firms are more likely to opt for risk-matched rates.
Since we ask questions on new projects in overseas markets, we also explicitly take foreign sales into account when interpreting the responses. As do Graham and Harvey (2001), we find no evidence that firms with foreign sales use more sophisticated discounting schemes. These international firms also tend to rely on company discount rates in most cases.
D. Multivariate Regression Analysis
To deepen our analysis we run a set of multivariate probit regressions in which we compare the impact of various explanatory variables on the four most important capital budgeting issues. In doing so we discover which factors determine whether firms apply DCF techniques, which types of firms compute the cost of capital, and which firms use the CAPM to do so, and which types of firms use sophisticated discount rates. These questions are framed by using three sets of model specifications.
In the first model, we analyze the significance of national variations of our full sample, which includes the US observations of Graham and Harvey (2001). We use a set of country dummies in which the Netherlands serves as the omitted variable. In the second model we extend this country analysis by using dichotomous variables to control for the cross-sectional variation in firm size, the educational level of the CEO, and the public listing of the firm. In our third model we extend the second model by including the level of shareholder orientation of each firm. This variable is only available for our European sample. (8)
First, we analyze the use of capital budgeting techniques. We distinguish between the DCF techniques (NPV, IRR, APV, and the discounted payback period) and the non-DCF techniques (the DCF variable has the value of one if the response to at least one of the four DCF technique questions exceeds two, and zero otherwise). In the US sample, we classify 96.5% of the firms as frequent DCF users, compared to 68.2% in the UK, 55% in the Netherlands, 59.5% in Germany, and 78.4% in France.
Table V shows our regression results. Model 1 shows that the country dummies reveal significant national differences. German and French firms make little use of DCF techniques, but US firms use these methods far more often than do the omitted Dutch firms. When we include the control variables (size, CEO education, and exchange listing), we find that size and public status contribute significantly to explaining the cross-sectional variation, and that both these variables have significant, positive effects on the dependent variable.
In Table V, Model 3, which focuses on our European sample, exhibits stability of the coefficient estimates across continents and shows that shareholder orientation is significantly and positively related to the use of DCF methods. This conclusion confirms our results in Section III, that large firms and firms that have a strong shareholder orientation are more likely to use DCF capital budgeting techniques.
In our second set of models, we explain which firms compute their cost of equity capital. A majority (64.2%) of the US sample reported that they always, or almost always, compute their cost of capital. These percentages in our European samples are lower: 57.4% in UK, 59.6% in the Netherlands, 53% in Germany, and 59% in France. The cross-national analysis does not yield any common results. Apparently, computing the cost of equity capital is not dictated by nationality. Firm size, exchange listing, and shareholder orientation tend to significantly increase the likelihood of cost of capital calculations.
Among the firms that compute their cost of capital, we differentiate between those that apply CAPM (or an extended CAPM) from the group that uses another technology. Within the US sample 81% of all firms indicate that they use a type of CAPM when estimating their cost of capital. This level is lower in our European samples: 73.5% in UK, 81.5% in the Netherlands, 43.4% in Germany, and 67.7% in France. We find that German firms use CAPM significantly less (a 10% confidence level), and that US firms turn out to be the most frequent users of CAPM. Much of this cross-national variation disappears when we extend Model 1 with control variables. These variables show that both firm size and public listing appear to significantly drive the use of CAPM.
To examine which discount rate firms use when evaluating new projects in overseas markets, we split our samples into two groups. The first group utilizes a sophisticated approach in which they apply risk-matching (on a project, division, or component level). The second group uses only a company- or country-wide discount rate. Firms can respond in such a way that they fall into either group. We choose to assign these firms to the risk-matched group, because the measure they use is more sophisticated.
In the US, 59.2% of the sample firms belong to the first category. This percentage is significantly lower in Europe, ranging between 17.4% in Germany and 31.2% in France. The results from our first model specification show that US firms are more likely to use a risk-matched discount rate compared to their European competitors. However, this difference diminishes after we include the variation in firm size, exchange listing, and CEO education. In combination with our results from Model 3, we conclude that larger, shareholder-oriented firms are significantly more likely to apply a risk-matched discount rate.
Overall, our results highlight the importance of multivariate regressions, because this approach enables us to isolate the impact of variables conditional on other influences. Continental and cross-national variations appear to be present, but lose much of their impact when we control for the underlying variation in firm size, exchange listing, CEO education, and shareholder orientation. Capital budgeting and cost of capital dilemmas are influenced most by firm size and the degree of shareholder orientation, both have a significantly positive impact on the quality of the financial decisions. Adding these variables significantly helps to increase the fit of the models.
IV. Capital Structure
Here, we explore the key determinants of capital structure choices. We focus on two competing theories. The static trade-off theory predicts that firms trade off the costs and benefits of leverage associated with tax effects, bankruptcy, and agency costs. This trade-off yields a firm-specific target capital structure. In pecking order theory, no such target exists, and firms raise outside capital based on a pecking order, preferring to finance first from internal sources. We add to their evidence in two ways.
We measure the relevance of these two theories by including in our questionnaire the same six questions and 65 items on capital structure choice as in Graham and Harvey (2001) and Bancel and Mittoo (2004). We add to their evidence for two reasons. First, we provide an extensive and selective test of the trade-off versus the pecking-order theories. Second, Bancel and Mittoo have 87 publicly listed European firms, while our tests are for both listed and unlisted firms.
A. Summary Statistics
In Figure 2, our firm statistics indicate that the overall debt levels in our European and US samples are comparable. The exception is the French firms, which have relatively low leverage.
The first test of the static trade-off compared to the pecking order theory is the presence of a target capital structure. Panel A of Table VI describes this structure. In the US, 83.2% of the firms have a target debt ratio. Because the target ratio is only relevant in the trade-off theory, this evidence favors this theory. In the Netherlands and Germany, 75% and 71.2%, respectively, of the respondents have a target debt ratio. In the UK and especially in France the percentages are much lower, 60.3% and 42.6%, respectively. For firms that claim to have a target debt ratio, we make a distinction of flexible, somewhat tight, and strict. We find in all five countries that most firms have a flexible target, and that a strict target is least popular.
B. Univariate Results
Panel B of Table VI reports the results for the seven most important factors that determine the firm's capital structures. Myers and Majluf's (1984) pecking-order model hypothesizes a hierarchy in financing: firms prefer internal over external financing, and prefer debt over equity. The degree of asymmetric information determines the relative costs of each financing source.
Panel B demonstrates that financial flexibility is the most important factor that influences the amount of debt in each of the five countries. This finding seems to be evidence of pecking-order behavior. Bancel and Mittoo (2004) also show that financial flexibility is the most important factor across the 16 countries in their samples. Our survey includes additional unreported questions related to pecking-order behavior. First, we ask if a debt issue is triggered by insufficient recent profits, the results on a scale of zero to four are weak and scattered, ranging between 1.24 for France and 2.3 for Germany. Second, we ask whether firms consider debt issues when equity is undervalued. This behavior would be consistent with the pecking order theory. Compared to the 1.56 score in the US, our European firms score relatively low, i.e., between 0.45 and 1.02. These findings cast doubt on the pecking-order interpretation of the results for financial flexibility.
The static trade-off theory predicts a trade-off between tax advantages and the bankruptcy costs of debt. We test this theory by asking about the importance of both factors. We find that CFOs consider the tax advantages of interest deductibility as the fourth most important factor in this context, after financial flexibility, credit ratings, and earnings volatility. The cross-national variation in this result is modest, and indicates that both European and US firms consier tax advantages are equally important.
The results in Table VI indicate that bankruptcy costs, which represent the negative effects of debt financing, are less important. The costs of bankruptcy scores range between 7.1% for Germany and 30.2% for the UK
Table VI also shows that the volatility of earnings, which increases the probability of bankruptcy and thus the expected costs, is more important. In the US, the UK, and Germany this factor is the third most important. In the Netherlands and France, volatility is the second most important factor. We find no compelling variation across countries or continents. Firms in all countries consider bankruptcy costs and tax advantages as important.
C. Multivariate Regression Analysis
Table VI demonstrates that financial flexibility is the most important factor that influences the amount of debt in each of the five countries. On the one hand, this seems to be evidence in favor of the pecking order model, since flexibility increases the possibility to choose between different financing alternatives. On the other hand, Opler, Pinkowitz, Stulz, and Williamson (1999) show that flexibility may be important for other reasons than the pecking order.
In Table VII, we report a regression test in which country dummies explain a dummy for a high score (three and four) on flexibility. To avoid perfect multicollinearity, we omit the Netherlands. The results yield no significant country dummies, which confirms our earlier findings.
In Model 2, we add firm characteristics. We note that flexibility is significant (at the 10% level) and more important in firms with a target debt ratio. This finding suggests that the pecking-order and the static trade-off theories are complements.
A more detailed test of the pecking order investigates the relation between asymmetric information and the desire for flexibility. Graham and Harvey (2001) use size and dividends as proxies for information problems, i.e., larger and dividend-paying firms have less asymmetry. Therefore, larger firms and dividend-payers should score lower on flexibility. We find that the opposite is true, as both size and dividends have positive coefficients (the coefficient for dividends is significant at the 1% level). The result is similar to the univariate comparisons in the US, in which larger firms score (nonsignificantly) higher. Dividend payers also score higher (significant at the 1% level). (9) We also find that flexibility is more important in public firms. This result can be explained by the higher information problems in public firms. All these results corroborate Graham and Harvey's conclusion that the desire for financial flexibility may represent pecking-order behavior, but that desire is not driven by the asymmetric information-based rationale that underlies the pecking-order theory. Our results complement the evidence documented by Bancel and Mittoo (2004), who mainly relate the capital structure preferences to legal systems and country characteristics. An interesting theoretical model that seems to be in line with our findings, and which deserves future empirical investigation, is Heaton (2002). The author derives a pecking order for financing choices, based on managerial optimism. Because managers attach higher probabilities to good future prospects than the capital market does, the market undervalues risky securities.
Firms that adopt the static trade-off model do so in two steps. First they decide to set a target capital structure. Then they choose factors that are included in the trade-off for the optimal capital structure. In Table VII, we investigate which factors motivate firms to set a target capital structure. Model 1 contains country dummies and shows that French firms are significantly less likely to set a target. This result supports the patterns displayed in Panel A of Table VI.
After including firm characteristics, we find that leverage, size, and dividends have a positive impact on the probability of aiming for a target. All these factors are significant at the 1% level. Theoretical models are not explicitly concerned with the prediction of which firms have a target and which firms do not. Additional regression analysis shows that the relation between firm size and the probability of aiming at a capital structure target is significantly stronger for our European firms than it is among Graham and Harvey's (2001) US firms. As far as we know, we are the first to document that targets are most likely set by large, highly levered, dividend-paying firms. Adding the cross-sectional variation in shareholder orientation and the public status do not explain the target-setting dilemma.
Under the static trade-off theory, firms trade off tax advantages against bankruptcy costs. We estimate the relation between country dummies plus firm characteristics and these factors. Our regression results show that in Germany, the tax advantage is less important. Of the German responses, only 21.1% cited tax advantages as an important factor when determining their capital structure. For the other samples these percentages were much higher (see Table VI). However, the next regression in Model 2 illustrates that the result is driven by cross-country differences in firm characteristics. After including more variables, we find that the coefficient for Germany loses most of its significance. Evidently the presence of a target significantly increases the probability that tax issues are important. Also, large firms appear to find tax advantages more relevant. Bankruptcy costs are the cost of leverage in the static trade-off and are measured as the probability that a firm considers the costs of bankruptcy or the likelihood important. Bankruptcy costs as a driving factor shows little variation across countries, ranging between 55% for the US and 37.2% among German firms. In the UK, the variation is 48.4%, in the Netherlands 55.3%, and in France 52.2%. Again we find that these costs are less relevant in Germany, but this difference reduces after extending the estimation model. (10)
As expected, we also find a significantly positive effect for firms with a target. This result is reassuring, because we would expect that firms with a target would attach the greatest value to the factors that are relevant in the trade-off theory.
The regression results in Table VII also include a dummy variable for shareholder orientation, which is insignificant in each of the four models. This finding contrasts with the capital budgeting results. Although both decisions have implications for shareholder wealth, the link is much more direct in capital budgeting. When we perform the same robustness checks as in the capital budgeting section, we find that our results are robust.
A comparison between the variables for countries and the firm characteristics shows that countries matter, but that variable tells only part of the story. We find that for flexibility, countries have no effect, and that for tax advantages the single significant coefficient becomes nonsignificant when we include other variables. Judging from the [R.sup.2]'s, compared to the other firm-level characteristics the country effects explain much less of the variation.
D. Comparison with Empirical Literature
Our survey evidence describes the opinions of the CFOs. Several international empirical studies have investigated capital structure decisions. It is interesting to confront this evidence with the survey results. Rajan and Zingales (1995) study US, German, French and UK firms, among others. The results indicate that the tangibility of assets, the market-to-book ratio, firm size and profitability are relevant determinants of capital structure. The authors relate these results to agency theory, distress costs, and the pecking order. Wald (1999) tests the determinants of leverage for the US, France, Germany, the UK (and Japan). The author notices several international differences, which are attributed to differences in agency problems and tax policies. De Bondt (1998) includes all four European countries and finds that internal financing is dominant, which is consistent with the pecking order theory. The empirical studies all concern public firms. The survey evidence in this study, as well as in Graham and Harvey (2001) and Bancel and Mittoo (2004), is in line with the evidence in favour of the static theories. Pecking order behaviour is found in all studies. However, the survey evidence adds that this behaviour is not driven by asymmetric information mechanisms.
In Europe, we document the following conclusions regarding capital structure practices. We find moderate support for the static trade-off theory, which predicts that firms have a target debt ratio, based on tax and bankruptcy considerations. In the US, the strongest evidence is found, both for the existence of targets and for the role of corporate taxes. The pecking order theory is rejected in each of the countries. However, the result of this theory, the desire for financial flexibility and pecking-order behavior, are important considerations in all countries. We find, however, that the asymmetric information problems do not drive this pecking order behavior. These result corroborate the findings of Graham and Harvey (2001) and Bancel and Mittoo (2004).
We conclude that the static trade-off theory faces at least moderate confirmation. Financial flexibility is important, but not driven by the asymmetric information as in Myers and Majluf's (1984) pecking order theory. Despite their institutional variations, we document strong similarities across the live very different countries when comparing capital structure policies.
V. Conclusion
In this article, we examine corporate finance practices in tour European countries, the UK, Netherlands, Germany, and France. We investigate capital budgeting, cost of equity capital estimation, and capital structure choice for both public and private firms. We compare these practices with previous results of Graham and Harvey (2001) for US firms and findings on capital structure decisions of Bancel and Mittoo (2004) for public European firms.
We observe a remarkable cross-national pattern in corporate governance. Firms in the UK and the Netherlands consciously strive to maximize their shareholder's wealth, while German and French firms attach a low priority to this corporate goal.
In corporate finance practices we find remarkably few differences across countries. When we examine capital budgeting techniques, we discover a strong preference for the simple payback criterion among our European firms. Although this preference is stronger in Europe, it does not differ significantly from the capital budgeting policies of US firms. We find that this preference for payback criteria is consistently stronger among small firms and among firms that are less oriented towards shareholder wealth maximization. Of the firms that do calculate their cost of capital, most CFOs told us they use the CAPM when computing their cost of equity capital. This preference of the CAPM over more intuitive alternatives is comparable to the way in which US firms compute the cost of capital. The use of the CAPM tends to rise with firm size, CEO tenure, exchange listing, and the importance of shareholder wealth maximization, while the educational background of the CEO appears to be irrelevant. We also control for the public listing of firms, which shows that our main conclusions also hold for only the public firms in our European sample.
When we examine capital structure policy, we find surprisingly few international differences. Financial flexibility, i.e., pecking-order behavior, appears to be the most important factor when firms are determining the proper of amount of corporate debt. However, the pecking-order behavior is not driven by asymmetric information considerations. Generally, we also find evidence for the static trade-off theory. Our evidence is based on bankruptcy costs and tax advantages.
When we analyze corporate finance practices, we find fundamental differences between large and small firms. Our results show that large firms are likely to use more sophisticated techniques for evaluating risky projects. In all samples we find that large firms are more likely to use NPV criteria and the CAPM for calculating the proper discount rate. Moreover, our results show that large firms are apt to utilize more sophisticated, risk-matched discount rates instead of a standard cost of capital. This consistent difference in corporate finance practice along the size dimension is an intriguing result, one which might help us to understand the size anomalies in the asset pricing literature.
We also document shareholder orientation as a second explanatory variable for the corporate finance practices in our samples. Firms that strive to maximize shareholder wealth are more likely to apply modern discounting techniques when considering investment projects. Such firms are also more likely to use the CAPM to derive their proper cost of capital.
Recent studies by La Porta et al. (1997, 1998) and Rajah and Zingales (2003) have illustrated the institutional variation that is present within our international sample. As expected, shareholder orientation prevails in the UK and in the Netherlands, but in the German and French firms we see that shareholders are less important. Moreover, in capital budgeting decisions, the orientation towards shareholders motivates managers to apply techniques that maximize shareholder wealth. However, the institutional differences seem to have no visible effect on firm's capital structure practice.
In both the US and European markets, professionals tend to adopt some of the same theoretical models and theories and neglect others when managing their corporate finances. The gap between science and practice appears to be consistent across borders. Further, although institutional differences are large and significant, they do not seem to dominate the way firms are run financially.
Table I. Survey Responses to the Question: "How Frequently Does Your
Firm Use the Following Techniques when Deciding which Projects or
Acquisitions to Pursue?"
US
Size CEO MBA
% always
or
most
always Mean Small Large Yes No
(b) Internal rate 75.61 3.09 2.87 3.41 *** 3.17 3.03
of return
(a) Net present 74.93 3.08 2.83 3.42 *** 3.17 3.00 *
value
(f) Payback 56.74 2.53 2.72 2.25 *** 2.48 2.55
period
(c) Hurdle rate 56.94 2.48 2.13 2.95 *** 2.57 2.42
(j) Sensitivity 51.54 2.31 2.13 2.56 *** 2.41 2.25
analysis
(d) Earnings 38.92 1.89 1.89 2.01 * 1.98 1.86
multiple
approach
(g) Discounted 29.45 1.56 1.56 1.55 1.68 1.49
payback
period
(l) We 26.56 1.47 1.47 1.57 1.49 1.39
incorporate
the "real
options" of a
project when
evaluating it
(i) Accounting 20.29 1.34 1.34 1.25 1.42 1.29
rate of
return
(k) Value at risk 13.66 0.95 0.95 1.22 *** 0.99 0.88
(e) Adjusted 10.78 0.85 0.85 0.72 * 0.74 0.91 *
present value
(h) Profitability 11.87 0.85 0.83 0.75 0.83 0.85
index
UK
Size CEO MBA
% always
or
most
always Mean Small Large Yes No
(b) Internal rate 53.13 2.31 3.33 2.15 *** 1.70 2.50
of return
(a) Net present 46.97 2.32 2.12 3.56 *** 2.18 2.45
value
(f) Payback 69.23 2.77 2.77 2.75 2.73 2.74
period
(c) Hurdle rate 26.98 1.35 1.07 3.00 *** 0.80 1.49
(j) Sensitivity 42.86 2.21 2.02 3.50 *** 1.60 2.35
analysis
(d) Earnings 39.06 1.81 1.78 2.00 1.90 1.90
multiple
approach
(g) Discounted 25.40 1.49 1.56 1.00 2.20 1.31 *
payback
period
(l) We 29.03 1.65 1.67 1.50 2.09 1.49
incorporate
the "real
options" of a
project when
evaluating it
(i) Accounting 38.10 1.79 1.82 1.63 1.30 1.90
rate of
return
(k) Value at risk 14.52 0.85 0.72 1.75 * 0.80 0.94
(e) Adjusted 14.06 0.78 0.71 1.22 1.20 0.76
present value
(h) Profitability 15.87 1.00 1.15 0.00 *** 1.60 0.92
index
Netherlands
Size CEO MBA
% always
or
most
always Mean Small Large Yes No
(b) Internal rate 56.00 2.36 2.25 2.80 2.73 2.07 *
of return
(a) Net present 70.00 2.76 2.53 3.70 *** 2.86 2.68
value
(f) Payback 64.71 2.53 2.56 2.40 2.86 2.28
period
(c) Hurdle rate 41.67 1.98 1.74 2.90 * 2.36 1.65
(j) Sensitivity 36.73 1.84 1.74 2.20 1.91 1.78
analysis
(d) Earnings 26.53 1.61 1.56 1.80 1.82 1.44
multiple
approach
(g) Discounted 25.00 1.25 1.32 1.00 1.27 1.23
payback
period
(l) We 34.69 1.49 1.62 1.00 1.57 1.43
incorporate
the "real
options" of a
project when
evaluating it
(i) Accounting 25.00 1.40 1.45 1.20 1.27 1.50
rate of
return
(k) Value at risk 4.26 0.51 0.47 0.67 0.71 0.35
(e) Adjusted 8.16 0.78 0.74 0.90 0.73 0.81
present value
(h) Profitability 8.16 0.78 0.82 0.60 0.77 0.78
index
Germany
Size CEO MBA
%
always
or most
always Mean Small Large Yes No
(b) Internal rate 42.15 2.15 2.04 3.08 ** 2.40 1.97 *
of return
(a) Net present 47.58 2.26 2.08 3.64 *** 2.70 1.93 ***
value
(f) Payback 50.00 2.29 2.31 2.08 2.40 2.21
period
(c) Hurdle rate 28.81 1.61 1.52 2.31 1.59 1.62
(j) Sensitivity 28.07 1.65 1.58 2.15 2.04 1.37 ***
analysis
(d) Earnings 20.51 1.25 1.18 1.77 1.47 1.09
multiple
approach
(g) Discounted 30.51 1.59 1.50 2.31 * 1.57 1.61
payback
period
(l) We 44.04 2.24 2.28 1.92 2.22 2.25
incorporate
the "real
options" of a
project when
evaluating it
(i) Accounting 32.17 1.63 1.76 0.62 *** 1.46 1.76
rate of
return
(k) Value at risk 23.68 1.45 1.36 2.15 ** 1.73 1.24 *
(e) Adjusted 7.83 0.71 0.63 1.38 * 0.96 0.54 **
present value
(h) Profitability 16.07 1.04 1.00 1.31 0.98 1.08
index
France
Size CEO MBA
% always
or
most
always Mean Small Large Yes No
(b) Internal rate 44.07 2.27 2.18 2.88 2.82 1.95 **
of return
(a) Net present 35.09 1.86 1.63 3.25 *** 2.30 1.62
value
(f) Payback 50.88 2.46 2.51 2.13 2.52 2.42
period
(c) Hurdle rate 3.85 0.73 0.80 0.17 ** 0.76 0.71
(j) Sensitivity 10.42 0.79 0.85 0.43 0.50 0.97
analysis
(d) Earnings 33.33 1.70 1.73 1.50 1.84 1.63
multiple
approach
(g) Discounted 11.32 0.87 0.91 0.57 1.11 0.74
payback
period
(l) We 53.06 2.20 2.27 1.88 2.05 2.30
incorporate
the "real
options" of a
project when
evaluating it
(i) Accounting 16.07 1.11 1.16 0.71 1.15 1.08
rate of
return
(k) Value at risk 29.79 1.68 1.66 1.83 2.00 1.50
(e) Adjusted 14.55 1.11 1.12 1.00 1.53 0.89 *
present value
(h) Profitability 37.74 1.64 1.63 1.71 2.00 1.46
index
The column labeled Mean shows the mean score based on a 0 (never) to
4 (always) scale. The letters in the parantheses at the left of each
column correspond to the questions originally used by Graham and
Harvey (2001). Averages marked with *, **, *** are significantly
different at a 10%, 5%, and 1% confidence level from the average in
the preceeding column, using a standard differences of means test.
Table II. Survey Responses to the Question: "Does Your Firm Estimate
the Cost of Equity Capital? If 'Yes,' How do you Determine Your Firm's
Cost of Equity Capital?"
US UK
% %
always always
or or
almost almost
always Mean always Mean
(b) Using the Capital 73.49 2.92 47.06 2.06
Asset Pricing Model
(CAPM, the beta
approach)
(a) With average 39.41 1.72 31.25 1.47
historical returns on
common stock
(c) Using the CAPM 34.29 1.56 27.27 1.45
but including some
extra "risk factors"
(f) Back out from 15.74 0.91 10.00 0.73
discounted
dividend/eamings
model, e.g.: price =
div./(cost of cap.
growth)
(d) Whatever our 13.93 0.86 18.75 1.19
investors tell us they
require
(e) By regulatory 7.04 0.44 16.13 0.94
decisions
Netherlands Germany
% %
always always
or or
almost almost
always Mean always Mean
(b) Using the Capital 55.56 2.37 33.96 1.36
Asset Pricing Model
(CAPM, the beta
approach)
(a) With average 30.77 1.42 18.00 1.06
historical returns on
common stock
(c) Using the CAPM 15.38 1.08 16.07 0.89
but including some
extra "risk factors"
(f) Back out from 10.71 0.79 10.42 0.58
discounted
dividend/eamings
model, e.g.: price =
div./(cost of cap.
growth)
(d) Whatever our 44.83 1.86 39.22 1.98
investors tell us they
require
(e) By regulatory 3.70 0.33 0.00 0.27
decisions
France
%
always
or
almost
always Mean
(b) Using the Capital 45.16 1.90
Asset Pricing Model
(CAPM, the beta
approach)
(a) With average 27.27 1.30
historical returns on
common stock
(c) Using the CAPM 30.30 1.39
but including some
extra "risk factors"
(f) Back out from 10.34 0.69
discounted
dividend/eamings
model, e.g.: price =
div./(cost of cap.
growth)
(d) Whatever our 34.38 1.66
investors tell us they
require
(e) By regulatory 16.13 0.87
decisions
The letters in the parantheses at the left of each column correspond
to the questions originally used by Graham and Harvey (2001).
Table III. Survey Responses to the Question: "When Valuing a Project,
Do You Adjust Either the Discount Rate or Cash Flows for the Following
Risk Factors?"
US
Disc. Cash
Rate Flow Both Neither
(b) Interest rate risk 15.30 8.78 24.65 51.27
(change in general level
of interest rates)
(f) Foreign exchange risk 10.80 15.34 18.75 55.11
(d) GDP or business cycle 6.84 18.80 18.80 55.56
risk
(a) Risk of unexpected 11.90 14.45 11.90 61.76
inflation
(h) Size (small firms being 14.57 6.00 13.43 66.00
riskier)
(e) Commodity price risk 2.86 18.86 10.86 67.43
(c) Term structure risk 8.57 3.71 12.57 75.14
(change in the long-
term vs. short-term
interest rate)
(g) Distress risk 7.41 6.27 4.84 81.48
(probability of
bankruptcy)
(i) "Market-to-book" ratio 3.98 1.99 7.10 86.93
(j) Momentum (recent 3.43 2.86 4.86 88.86
stock price
performance)
UK
Disc. Cash
Rate Flow Both Neither
(b) Interest rate risk 20.97 27.42 27.42 24.19
(change in general level
of interest rates)
(f) Foreign exchange risk 12.50 32.81 17.19 37.50
(d) GDP or business cycle 16.13 24.19 8.06 51.61
risk
(a) Risk of unexpected 17.74 25.81 12.90 43.55
inflation
(h) Size (small firms being 21.88 12.50 7.81 57.81
riskier)
(e) Commodity price risk 19.05 19.05 7.94 53.97
(c) Term structure risk 17.19 17.19 12.50 53.13
(change in the long-
term vs. short-term
interest rate)
(g) Distress risk 14.52 9.68 6.45 69.35
(probability of
bankruptcy)
(i) "Market-to-book" ratio 17.74 9.68 4.84 67.74
(j) Momentum (recent 16.95 5.08 6.78 71.19
stock price
performance)
Netherlands
Disc. Cash
Rate Flow Both Neither
(b) Interest rate risk 20.41 8.16 20.41 51.02
(change in general level
of interest rates)
(f) Foreign exchange risk 6.00 26.00 18.00 50.00
(d) GDP or business cycle 8.33 6.25 10.42 75.00
risk
(a) Risk of unexpected 8.00 12.00 16.00 64.00
inflation
(h) Size (small firms being 17.02 14.89 14.89 53.19
riskier)
(e) Commodity price risk 2.13 19.15 10.64 68.09
(c) Term structure risk 10.64 0.00 10.64 78.72
(change in the long-
term vs. short-term
interest rate)
(g) Distress risk 14.58 4.17 8.33 72.92
(probability of
bankruptcy)
(i) "Market-to-book" ratio 4.26 2.13 19.15 74.47
(j) Momentum (recent 4.35 0.00 8.70 86.96
stock price
performance)
Germany
Disc. Cash
Rate Flow Both Neither
(b) Interest rate risk 26.72 14.66 22.41 36.21
(change in general level
of interest rates)
(f) Foreign exchange risk 13.27 19.47 18.58 48.67
(d) GDP or business cycle 6.19 9.73 11.50 72.57
risk
(a) Risk of unexpected 18.80 9.40 9.40 62.39
inflation
(h) Size (small firms being 9.91 9.01 12.61 68.47
riskier)
(e) Commodity price risk 4.39 26.32 16.67 52.63
(c) Term structure risk 17.12 7.21 8.11 67.57
(change in the long-
term vs. short-term
interest rate)
(g) Distress risk 8.77 14.04 13.16 64.04
(probability of
bankruptcy)
(i) "Market-to-book" ratio 4.63 8.33 12.96 74.07
(j) Momentum (recent 5.66 0.94 3.77 89.62
stock price
performance)
France
Disc. Cash
Rate Flow Both Neither
(b) Interest rate risk 23.21 26.79 21.43 28.57
(change in general level
of interest rates)
(f) Foreign exchange risk 16.36 20.00 5.45 58.18
(d) GDP or business cycle 15.79 22.81 12.28 49.12
risk
(a) Risk of unexpected 17.54 24.56 26.32 31.58
inflation
(h) Size (small firms being 23.64 16.36 10.91 49.09
riskier)
(e) Commodity price risk 8.62 46.55 12.07 32.76
(c) Term structure risk 22.81 12.28 17.54 47.37
(change in the long-
term vs. short-term
interest rate)
(g) Distress risk 12.50 23.21 14.29 50.00
(probability of
bankruptcy)
(i) "Market-to-book" ratio 20.00 12.73 12.73 54.55
(j) Momentum (recent 27.78 3.70 7.41 61.11
stock price
performance)
The letters in the parantheses at the left of each column correspond
to the questions originally used by Graham and Harvey (2001). Averages
marked with *, **, *** are significantly different at a 10%, 5%, and
1% confidence level from the average in the preceeding column, using a
standard of means test.
Table IV. Survey Responses to the Question: "How Frequently Would
Your Company Use the Following Discount Rates when Evaluating a
New Project in an Overseas Market? To Evaluate this Project We
Would Use ..."
US
Size
%
always
or
almost
always Mean Small Large
(a) The discount rate for 58.79 2.50 2.50 2.50
our entire company
(d) A risk-matched 50.95 2.09 1.86 2.36 ***
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 34.52 1.65 1.49 1.82 **
the overseas market
(country discount rate)
(c) A divisional discount 15.61 0.95 0.82 1.09 **
rate (if the project line
of business matches a
domestic division)
(e) A different discount 9.87 0.66 0.68 0.64
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
US
CEO MBA
Yes No
(a) The discount rate for 2.49 2.51
our entire company
(d) A risk-matched 2.20 1.99
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 1.77 1.60
the overseas market
(country discount rate)
(c) A divisional discount 0.88 0.98
rate (if the project line
of business matches a
domestic division)
(e) A different discount 0.59 0.67
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
UK
Size
%
always
or
almost
always Mean Small Large
(a) The discount rate for 40.98 1.97 1.87 2.63
our entire company
(d) A risk-matched 23.73 1.17 1.04 1.89
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 20.00 0.97 0.88 1.44
the overseas market
(country discount rate)
(c) A divisional discount 17.24 0.91 0.82 1.44
rate (if the project line
of business matches a
domestic division)
(e) A different discount 10.53 0.58 0.61 0.38
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
UK
CEO MBA
Yes No
(a) The discount rate for 1.80 1.88
our entire company
(d) A risk-matched 1.78 1.04
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 2.33 0.77 **
the overseas market
(country discount rate)
(c) A divisional discount 1.33 0.89
rate (if the project line
of business matches a
domestic division)
(e) A different discount 1.33 0.47
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
Netherlands
Size
%
always
or
almost
always Mean Small Large
(a) The discount rate for 64.58 2.48 2.37 2.90
our entire company
(d) A risk-matched 27.08 1.27 1.13 1.80
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 14.89 1.09 0.92 1.70
the overseas market
(country discount rate)
(c) A divisional discount 17.02 0.96 0.68 2.11 **
rate (if the project line
of business matches a
domestic division)
(e) A different discount 2.13 0.26 0.22 0.40
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
Netherlands
CEO MBA
Yes No
(a) The discount rate for 2.55 2.43
our entire company
(d) A risk-matched 1.05 1.44
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 1.38 0.85
the overseas market
(country discount rate)
(c) A divisional discount 1.40 0.63 *
rate (if the project line
of business matches a
domestic division)
(e) A different discount 0.38 0.15
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
Germany
Size
%
always
or
almost
always Mean Small Large
(a) The discount rate for 41.96 2.00 1.89 2.79 *
our entire company
(d) A risk-matched 25.00 1.16 1.00 2.31 **
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 14.85 0.92 0.84 1.69 *
the overseas market
(country discount rate)
(c) A divisional discount 12.00 0.69 0.67 0.85
rate (if the project line
of business matches a
domestic division)
(e) A different discount 7.14 0.51 0.47 0.83
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
Germany
CEO MBA
Yes No
(a) The discount rate for 2.15 1.89
our entire company
(d) A risk-matched 1.58 0.85 **
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 0.98 0.88
the overseas market
(country discount rate)
(c) A divisional discount 0.77 0.63
rate (if the project line
of business matches a
domestic division)
(e) A different discount 0.83 0.27 ***
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
France
Size
%
always
or
almost
always Mean Small Large
(a) The discount rate for 24.14 1.03 0.88 1.89
our entire company
(d) A risk-matched 27.27 1.16 1.06 1.75
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 16.36 0.76 0.53 2.13 **
the overseas market
(country discount rate)
(c) A divisional discount 12.50 0.70 0.60 1.25
rate (if the project line
of business matches a
domestic division)
(e) A different discount 11.32 0.62 0.54 1.14
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
France
CEO MBA
Yes No
(a) The discount rate for 1.36 0.83
our entire company
(d) A risk-matched 1.57 0.91
discount rate for this
particular project
(considering both
country and industry)
(b) The discount rate for 0.90 0.69
the overseas market
(country discount rate)
(c) A divisional discount 1.24 0.37 **
rate (if the project line
of business matches a
domestic division)
(e) A different discount 0.79 0.53
rate for each
component cash flow
that has a different
risk characteristic
(e.g. depreciation
compared to
operating cash flows)
The letters in the parantheses at the left of each column correspond
to the questions originally used by Graham and Harvey (2001). Averages
marked with are significantly different at a 10%, 5%, and 1% confidence
level from the average in the preceeding column, using a standard
differences of means test.
Table V. Multivariate Probit Regression Output for Capital Budgeting
Models 1 and 2 employ a pooled data set in which our European sample
is merged with the US sample of Graham and Harvey (2001). Model 3
exclusively focuses on the European sample. The dummy for capital
budgeting has value one if at least one response to the questions a,
b, e or g of Table I exceeds 2, or zero otherwise. Calculate Cost of
Capital equals one if respondents indicate that they calculate the
cost of equity capital, and zero otherwise. Use CAPM has value one if
at least one response to the questions b or c of Table II exceeds 2,
and zero otherwise. Use Risk-Matched Discount Rate equals one if at
least one response to the question c, d, or e of Table IV exceeds 2,
or zero otherwise. The dummy control variables are Size (=1 if sales
over US $1 Billion), MBA (=1 if CEO has MBA), Public, Shareholder
Orientation (=1 if score over two in Panel B of Figure 1). The McFadden
[R.sup.2] is the likelihood ratio index and is an analogue to the
R-squared reported in linear regression models. The Akaike info
criterion provides a measure of information that strikes a balance
between this measure of goodness of fit and parsimonious specification
of the model; the lower the value the better the fit of the model.
Use DCF Capital Budgeting
Full Sample EU
Mode 1 Model 2 Model 3
Constant 0.79 *** 0.49 ** 0.21
(4.00) (2.23) (0.82)
Germany-Dummy -0.55 ** -0.47 * -0.36
(-2.41) (-1.96) (-1.47)
France-Dummy -0.66 *** -0.57 ** -0.49 *
(-2.59) (-2.14) (-1.76)
UK-Dummy -0.31 -0.35 -0.34
(-1.24) (-1.29) (-1.24)
US-Dummy 1.03 *** 0.89 *** --
(4.43) (3.53)
Size -- 1.14 *** 1.31 ***
(3.95) (3.62)
MBA -- 0.10 0.09
(0.69) (0.51)
Public -- 0.47 ** 0.30
(2.86) (1.50)
Shareholder -- -- 0.38 **
Orientated (EU) (2.11)
N 678 648 303
McFadden [R.sup.2] 0.22 0.28 0.11
Akaike Info 0.75 0.71 1.22
Criterion
Calculate Cost of Capital
Full Sample EU
Mode 1 Model 2 Model 3
Constant 0.07 -0.08 -0.39
(1.11) (-0.41) (-1.63)
Germany-Dummy -0.25 -0.15 -0.03
(-1.22) (-0.69) (-0.14)
France-Dummy -0.05 0.09 0.20
(-0.21) (0.35) (0.75)
UK-Dummy -0.16 -0.26 -0.23
(-0.68) (-1.07) (-0.93)
US-Dummy 0.18 -0.11 --
(0.97) (-0.54)
Size -- 0.83 *** 0.99 ***
(5.99) (3.91)
MBA -- -0.11 -0.11
(-0.94) (-0.67)
Public -- 0.54 *** 0.28
(5.51) (1.55)
Shareholder -- -- 0.45 ***
Orientated (EU) (2.64)
N 672 645 313
McFadden [R.sup.2] 0.02 0.12 0.09
Akaike Info 1.35 1.22 1.32
Criterion
Use CAPM
Full Sample EU
Mode 1 Model 2 Model 3
Constant -0.56 *** -1.04 *** -1.23 ***
(-3.03) (-4.83) (-4.33)
Germany-Dummy -0.38 * -0.24 -0.17
(-1.69) (-0.97) (-0.69)
France-Dummy 0.07 0.30 0.35
(0.27) (1.11) (1.25)
UK-Dummy -0.03 0.01 0.02
(-0.11) (0.03) (0.08)
US-Dummy 0.79 *** 0.49 ** --
(4.00) (2.23)
Size -- 0.99 *** 1.09 ***
(7.27) (5.02)
MBA -- 0.11 0.10
(0.91) (0.57)
Public -- 0.44 *** 0.34 *
(3.31) (1.76)
Shareholder -- -- 0.23
Orientated (EU) (1.18)
N 624 599 313
McFadden [R.sup.2] 0.10 0.22 0.13
Akaike Info 1.24 1.09 1.01
Criterion
Discount Rates
Full Sample EU
Mode 1 Model 2 Model 3
Constant -0.34 * -0.56 *** -1.00 ***
(-1.85) (-2.83) (-3.76)
Germany-Dummy -0.13 -0.08 0.04
(-0.60) (-0.34) (0.19)
France-Dummy -0.22 -0.16 -0.02
(-0.87) (-0.61) (-0.10)
UK-Dummy -0.12 -0.07 -0.06
(-0.49) (-0.30) (-0.24)
US-Dummy 0.48 ** 0.33 --
(2.45) (1.61)
Size -- 0.42 *** 0.51 **
(3.24) (2.28)
MBA -- 0.16 0.25
(1.10) (1.47)
Public -- 0.14 0.11
(1.10) (0.61)
Shareholder -- -- 0.44 **
Orientated (EU) (2.18)
N 603 574 283
McFadden [R.sup.2] 0.04 0.06 0.06
Akaike Info 1.33 1.31 1.25
Criterion
Coefficient estimates marked with *, **, *** are statistically
significant at a 10%, 5%, and 1% confidence level.
Table VI. Summary Statistics on Capital Structure
Panel A of this table displays the percentage of firms which use a
Target Debt ratio in general and specific types of target ratio for
each national sample. Panel B reports the average response to our
survey question on what factors affect how firms choose the
appropriate amount of debt.
US UK NL
Panel A. Summary Statistics (n=392) (n=68) (n=52)
Target Debt Ratio 83.16% 60.29% 75.00%
- Flexible Target Debt Ratio 34.18% 29.41% 44.23%
- Somewhat Tight Target Debt Ratio 31.63% 16.18% 15.38%
- Strict Target Debt Ratio 8.93% 11.76% 9.62%
Panel B. Question: "What factors affect how you choose the
appropriate amount of debt for your firm?"
(% always or almost always)
(g) Financial flexibility (we restrict 59.38 50.00 51.06
debt so we have enough internal
funds available to pursue new
projects when they come along)
(d) Our credit rating (as assigned by 57.10 27.42 34.04
rating agencies)
(h) The volatility of our earnings and 48.08 35.48 42.55
cash flows
(a) The tax advantage of interest 44.85 30.16 37.50
deductibility
(e) The transactions costs and fees 33.52 25.40 15.22
for issuing debt
(c) The debt levels of other firms in 23.40 16.13 26.53
our industry
(b) The potential costs of bankruptcy, 21.35 30.16 27.08
near-bankruptcy, or financial
distress
GER FR
Panel A. Summary Statistics (n=132) (n=61) N
Target Debt Ratio 71.21% 11.76% 620
- Flexible Target Debt Ratio 36.36% 9.62% 235
- Somewhat Tight Target Debt Ratio 19.70% 9.85% 173
- Strict Target Debt Ratio 6.56% 8.20% 66
Panel B. Question: "What factors affect how you choose the
appropriate amount of debt for your firm?"
(% always or almost always)
(g) Financial flexibility (we restrict 47.83 37.25 341
debt so we have enough internal
funds available to pursue new
projects when they come along)
(d) Our credit rating (as assigned by 38.60 30.19 305
rating agencies)
(h) The volatility of our earnings and 30.97 34.78 305
cash flows
(a) The tax advantage of interest 21.05 29.63 289
deductibility
(e) The transactions costs and fees 26.32 21.15 276
for issuing debt
(c) The debt levels of other firms in 14.04 12.96 258
our industry
(b) The potential costs of bankruptcy, 7.08 24.07 265
near-bankruptcy, or financial
distress
The letters in the parantheses at the left of each column correspond
to the questions originally used by Graham and Harvey (2001).
Table VII. Multivariate Probit Regression Output for Four
Important Factors Affecting Capital Structure
Models 1 and 2 use a pooled data set in which we merge our European
sample with Graham and Harvey's (2001) US sample. Model 3 focuses
exclusively on the European sample. The dummy variable Leverage
equals one if the reply to question (g) of Table VI B was "always
or almost always", or zero otherwise. Target Leverage equals one if
respondents indicate that they use some sort of target debt ratio,
and zero otherwise. The dummy for tax advantage of debt has the value
of one if the reply to question (a) of Table VI B was "always or
almost always", and zero otherwise. The dummy for bankruptcy costs
of debt equals one if the reply to question (b) of Table VI B was
"always or almost always", or zero otherwise. The dummy control
variables are Size (sales over US $1 billion), Dividend, Public, and
Shareholder Orientation (score of over two in Panel B of Figure 1).
The McFadden Rz is the likelihood ratio index and is an analog to the
R-squared reported in linear regression models. The Akaike information
criterion measures information that strikes a balance between this
measure of goodness of fit and a parsimonious specification of the
model in which the lower the value, the better the fit of the model.
Financial Flexibility
Full Sample EU
Model 1 Model 2 Model 3
Constant 0.03 -0.31 -0.39
(0.15) (-2.42) (-1.34)
Germany-Dummy -0.07 0.11 0.16
(-0.33) (0.51) (0.67)
France-Dummy -0.37 -0.24 -0.21
(-1.45) (-0.91) (-0.74)
UK-Dummy -0.03 0.06 0.11
(-0.11) (0.23) (0.41)
US-Dummy 0.21 0.27 --
(1.08) (1.31)
Leverage -- -0.15 -0.13
(-1.29) (-0.73)
Target Leverage -- 0.23 * 0.33 **
(1.67) (1.79)
Size -- 0.02 0.15
(0.16) (0.63)
Dividend -- 0.44 *** 0.57 ***
(3.67) (3.31)
Public -- 0.26 0.32
(2.00) (1.64)
Shareholder -- -- -0.06
Orientated (EU) (-0.32)
N 632 560 275
McFadden [R.sup.2] 0.02 0.06 0.07
Akaike Info Criterion 1.38 1.34 1.35
Target Debt Ratio
Full Sample EU
Model 1 Model 2 Model 3
Constant 0.67 *** 0.09 0.19
(3.57) (0.40) (0.73)
Germany-Dummy -0.11 0.02 -0.04
(-0.52) (0.10) (-0.16)
France-Dummy -0.86 *** -0.75 ** -0.81 ***
(-3.46) (-2.88) (-3.01)
UK-Dummy -0.41 * -0.33 -0.32
(-1.70) (-1.31) (-1.22)
US-Dummy 0.21 0.27 --
(1.03) (1.21)
Leverage -- 0.60 *** 0.73 ***
(4.56) (3.98)
Target Leverage -- -- --
Size -- 0.44 *** 0.84 ***
(2.65) (3.08)
Dividend -- 0.42 *** 0.40 **
(3.34) (2.43)
Public -- 0.11 0.14
(0.82) (0.73)
Shareholder -- -- -0.22
Orientated (EU) (-1.21)
N 674 613 313
McFadden [R.sup.2] 0.06 0.15 0.14
Akaike Info Criterion 1.11 1.01 1.18
Tax Advantage of Debt
Full Sample EU
Model 1 Model 2 Model 3
Constant -0.32 * -0.57 *** -0.59 *
(-1.73) (-2.55) (-1.89)
Germany-Dummy -0.49 ** -0.37 -0.33
(-2.14) (-1.57) (-1.32)
France-Dummy -0.23 -0.03 0.12
(-0.90) (-0.10) (0.41)
UK-Dummy -0.20 -0.11 -0.11
(-0.81) (-0.41) (-0.40)
US-Dummy 0.19 0.03 --
(0.97) (0.16)
Leverage -- 0.07 -0.05
(0.60) (0.29)
Target Leverage -- 0.39 ** 0.65 ***
(2.55) (3.18)
Size -- 0.72 *** 0.69 **
(5.10) (2.93)
Dividend -- 0.14 -0.02
(1.14) (-0.10)
Public -- 0.10 -0.02
(0.74) (-0.08)
Shareholder -- -- 0.33
Orientated (EU) (1.55)
N 639 567 280
McFadden [R.sup.2] 0.03 0.11 0.11
Akaike Info Criterion 1.30 1.20 1.12
Bankruptcy Costs of Debt
Full Sample EU
Model 1 Model 2 Model 3
Constant 0.13 0.25 0.15
(0.73) (1.15) (0.51)
Germany-Dummy -0.46 *** -0.47 ** -0.40 *
(-2.10) (2.11) (1.67)
France-Dummy -0.08 0.06 0.20
(-0.31) (0.21) (0.71)
UK-Dummy -0.17 -0.12 -0.10
(-0.72) (-0.47) (-0.42)
US-Dummy -0.01 -0.01 --
(-0.05) (-0.03)
Leverage -- 0.04 -0.04
(0.31) (-0.24)
Target Leverage -- 0.41 *** 0.62 ***
(2.99) (3.42)
Size -- -0.10 0.07
(-0.73) (0.31)
Dividend -- 0.02 -0.03
(0.16) (-0.18)
Public -- 0.03 0.13
(-0.27) (0.67)
Shareholder -- -- 0.13
Orientated (EU) (0.67)
N 632 559 268
McFadden [R.sup.2] 0.01 0.03 0.06
Akaike Info Criterion 1.38 1.38 1.37
Coefficient estimates marked with *, **, *** are statistically
significant at a 10%, 5%, and 1% confidence level.
The authors thank Mark Flood. Campbell Harvey, and an anonymous referee for their helpful comments on a previous version of this paper; and also thank the Vereniging Trustfonds Erasmus Universiteit Rotterdam for their financial support. All remaining errors are the responsibility of the authors.
(1) The most famous survey study in the recent financial literature is by Graham and Harvey (2001), which was awarded the Jensen Price for the best corporate finance paper published in the Journal of Financial Economics in 2001. Other seminal survey papers in the field of corporate finance are Gitman and Forrester (1977) on capital budgeting and Pinegar and Wilbricht (1989) on capital structure.
(2) The Graham and Harvey (2001) study analyses a sample of 392 US CFOs, the second largest published survey after the Graham, Harvey, and Rajgopal (2004) paper on corporate financial reporting in which they survey 401 financial executives. Moore and Reichert (1983) analyze a dataset on 298 large firms from the US. A recent survey by Bray, Graham, Harvey, and Michaely (2003) on payout policy in the US includes 384 respondents.
(3) This striking result is also reported by consulting firm DBM (2002) in the study "Turnover at the top; research highlights from a global study". Out of 25 countries, the French CEOs have the longest average tenure, which is twice as long as the world-wide average.
(4) We analyse sample splits based on several firm characteristics. Table I presents the results for size and CEO education. We present additional comparisons for public compared to private firms and on shareholder orientation on our website: (http://web.eur.nl/fbk/dep/dep5/research).
(5) See also Jagannathan and Meier (2002), who summarize 13 US surveys on the use of the CAPM and discounting methods in general.
(6) We provide the complete results of the comparisons for subsamples on our website.
(7) We provide the complete results of the comparisons for subsamples on our website.
(8) On our website we provide additional regression results for two robustness checks. First, we do not know the exact degree of shareholder orientation of US firms, because Graham and Harvey (2001) did not survey their firms on this issue. Therefore, we performed a robustness check on Models 2 and 3 by applying the assumption that all US firms maximize shareholder wealth. We reran Regression 3 including the US dummy variable. Second, we ran regressions for the European public firms to test whether our results are consistent in this subsample. Our results are robust to these checks.
(9) Brav et al. (2003) argue that the levels of dividends are nearly untouchable. This implies that paying dividends reduces the flexibility of firms, which explains the positive relation we report between the importance of flexibility and dividends.
(10) This finding contradicts Rajan and Zingales' (1995, p.1444) conclusion based on a description of the institutional setting that Germany's bankruptcy code is creditor friendly. However, their empirical test confirms our findings. Firm size, as proxy for the inverse of the probability of default, has a negative effect on debt, which is not found in the other countries.
References
Bancel, F. and U.R. Mittoo, 2004, "The Determinants of Capital Structure Choice: A Survey of European Firms," Financial Management 33, 103-132.
Brav, A., Graham, J.R., Harvey, C.R., Michaely, R., 2003, "Payout Policy in the 21st Century", Duke University Working Paper.
Brealey, R.A. and S.C. Myers, 2003, "Principles of Corporate Finance," 7th Ed., New York, NY, McGraw-Hill.
De Bondt, G.J., 1998, "Financial Structure: Theories and Stylized Facts for Six EU Countries," De Economist 146, 271-301.
Chew, D., 1997, Studies In International Corporate Finance and Governance Systems: A Comparison of the US, Japan, and Europe, Oxford University Press.
Fama, E.F. and F.R. French, 1992, "The Cross-Section of Expected Stock Returns," Journal of Finance 47, 427-465.
Gitman, L.J. and J.R. Forrester Jr., 1977, "A Survey of Capital Budgeting Techniques Used By Major US Firms," Financial Management 6, 66-71.
Graham, J.R. and C.R. Harvey, 2001, "The Theory and Practice of Corporate Finance: Evidence From The Field," Journal of Financial Economics 61, 187-243.
Graham, J.R., Harvey, C.R., and S. Rajpogal, 2004, "The Economic Implications of Corporate Financial Reporting," Working Paper.
Heaton, J.B., 2002, "Managerial Optimism and Corporate Finance," Financial Management 31, 33-45.
Jagannathan, R. and I. Meier, 2002, "Do We Need CAPM for Capital Budgeting?" Financial Management 31, 55-77.
Jegadeesh, N. and S. Titman, 1993, "Returns To Buying Winners and Selling Losers: Implications For Stock Market Efficiency," Journal of Finance 48, 65-91.
LaPorta, R., Lopez-de-Silanes, F., Shleifer, A., and R.W. Vishny, 1997, "Legal Determinants of External Finance," Journal of Finance 52, 1131-1150.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and R.W. Vishny, 1998, "Law and Finance," Journal of Political Economy 106, 1113-1155.
Myers, S.C. and N. Majluf, 1984, "Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have," Journal of Financial Economics 13, 3-46.
Moore, J.S. and A.K. Reichert, 1983, "An Analysis Of The Financial Management Techniques Currently Employed By Large US Corporations," Journal of Business Finance and Accounting 10, 623-645.
Opler, T.C., Pinkowitz, L., Stulz, R., and R. Williamson, 1999, "The Determinants And Implications of Corporate Cash Holdings," Journal of Financial Economics 52, 3-46.
Pinegar, J.M. and L. Wilbricht, 1989, "What Managers Think of Capital Structure Theory: A Survey," Financial Management 18, 82-91.