The Determinants of Corporate Hedging and Firm Value: An Empirical Research of European Firms

Transcription

1 The Determinants of Corporate Hedging and Firm Value: An Empirical Research of European Firms Ying Liu S882686, Master of Finance, Supervisor: Dr. J.C. Rodriguez Department of Finance, School of Economics and Management, Tilburg University 1

2 The Determinants of Corporate Hedging and Firm Value: An Empirical Research of European Firms Abstract This paper seeks to identify 1) the determinants of corporate hedging and 2) test the hypothesis that whether corporate hedging by using derivatives can add value to European firms. In the sample of 253 European non-financial firms on the FTS Euro First Index ranked by the market capitalization as of year-end 2003, we find strong evidences in support for the relationship between financial distress and corporate hedging. We also find that firm size has significantly positive relation with corporate hedging. When we measure the relation between liquidity and corporate risk management, the empirical results provide weak evidence to support that liquidity reduce costs of bankruptcy and financial distress. Finally, to examine the debt capacity and tax shield effects of corporate hedging for European firms, we follow Graham and Rogers (2002) and use two-stage estimation technique, the empirical results show that corporate hedging does create value for European firms by increasing their debt capacity leading to 2.10% increase in firm value. 2

3 1. Introduction According to the classic proposition of Modigliani and Miller (1958, 1961) on capital structure, corporate hedging does not add value to firm when there is no asymmetric information, taxes, or transaction costs, that is to say, corporate risk management is irrelevant to the firm since shareholders can replicate it on their own or create a well-diversified portfolio. If capital markets are perfect, shareholders have the necessary information about a firm s systematic risk exposures to create their desired risk profiles; therefore, it is not necessary for a firm to hedge in this condition. However, this conclusion may not hold because the assumptions that the theorem is based on are often violated in the real world. The positive theory of corporate hedging developed by Smith and Stulz (1985) presents that imperfect capital markets can create conditions where corporate hedging becomes economically justified because it can add value to the firm. Previous empirical studies have tested the corporate decision of whether and how to hedge and why firms employ derivatives for corporate risk management. Several recent theories indicate that corporate hedging is a value-enhancing strategy for the firm. Most of these theories focus on the introduction of some frictions (taxes or costly access to external finance) into the Modigliani and Miller paradigm. The key question for shareholders is whether corporate hedging can add value to firm. The empirical studies on this question are relatively recent and have generally focused on US firms. However, there are significant differences between US firms and Non-US firms. This paper seeks to identify a) the determinants of corporate hedging for European firms and test b) whether corporate hedging by using 3

4 derivatives can add value to European firms. Based on a sample taken from the top 253 non-financial European firms on the FTS Euro First index ranked by the market capitalization for the year ended 2003, we use parametric (Two Sample T-test) and non-parametric (Wilcoxon Rank Sum Test) univariate tests and multivariate logistic analysis to examine what determines corporations level of derivative use. To examine the value effect of European firms, we follow Graham and Rogers (2002) and use two-stage estimation technique to test the hypothesis that corporate hedging can add value to European firms by enhancing their debt capacity and thus increase their tax benefit. European countries provide a particular valuable focus for empirical investigation for corporate risk management. First of all, European Union has developed a single market through a standardized system of laws which ensures the free movement of people, goods, services, and capital, enact legislation in justice and home affairs, and maintain common policies on trade. Most countries are in the system of Monetary Union with the exception of UK, Sweden and Denmark. Furthermore, the economy of European Union was open, developed and highly industrialized, which has unrestricted financial capital markets and beneficial trading policies. Therefore, the corporate financing and hedging decisions by the firms in the sample are more likely to reflect economic and financial criteria rather than the result of constraints imposed by shallow domestic capital markets, bureaucratic controls and the like (Clark, Judge and Belghitar, 2008). The EU countries have a large number of firms with foreign 4

5 sales, foreign operation and foreign currency debt, which are facing continuous foreign exchange rate risk and interest rate risk. Therefore, the data for European countries is well adapted to the empirical analysis for the determinants and value effect of corporate risk management. This paper makes two major contributions to the literature. First, we find that leverage ratio, interest cover ratio, natural log of firm size, tax loss carry forward, foreign sales, foreign operation and foreign currency debt are important determinants for corporate hedging of European firms. The empirical results show that there are strong evidences in support for the relationship between costs of bankruptcy / financial distress and corporate hedging. We also find that firm size has significantly positive relation with corporate hedging. This indicates that transaction costs and economies of scale are essential determinants of corporate hedging. However, this result is not consistent with the financial distress costs argument, which suggests that small firms are faced with substantial information asymmetries and are more likely to make external finance more expensive therefore have stronger incentive to hedge. When it comes to the exposure to financial price risks, we find that the exposure to foreign exchange rate risk is positively related to corporate hedging measured by foreign sales, foreign operation and foreign currency debt. When we measure the relation between liquidity and corporate risk management, the empirical results provide weak evidence to support that liquidity reduce costs of bankruptcy and financial distress, which is partly consistent with the substitutes for hedging hypothesis. Interestingly, the empirical 5

6 results show that the potential investment growth opportunities for firms have no relation with corporate hedging and thus there is no evidence linking underinvestment and asset substitution costs to corporate hedging. Secondly, we find strong evidence that corporate hedging does create value for European firms by enhancing their debt capacity and thus increasing their tax benefit, leading to the increase in firm value of 2.10%. The reminder of the paper is organized as follows. Section 2 reviews the empirical theories and evidences of value creation through corporate risk management and defines the proxies of determinants of corporate hedging as explanatory variables. Section 3 provides the sample formation process and describes the sample s characteristics. Section 4 presents the empirical evidences of both univariate and multivariate analysis. Section 5 provides the empirical results on the debt capacity effect of corporate hedging. Section 6 presents the limitation and challenge of corporate risk management. Section 7 concludes the paper. 2. Overview of Theories and Evidences on the Use of Derivatives Standard economic theory seems to imply that hedging at the corporate level does not contribute to firm value (Dufey and Srinivasulu, 1983). For example, according to Modigliani-Miller (1958) propositions, when capital markets are perfect, the total value of a firm is equal to the market value of the total cash flows generated by its assets and not affected by its choice of capital structure. In other words, in the perfect 6

7 capital market, corporate financial decisions are irrelevant and cannot influence firm value. Since corporate hedging is regarded as a financing policy, it does not add value to firm and cannot be justified (Bartram, 2002; Stulz, 2000; Smith, 1995). Consequently, when the capital market is less than perfect, the benefit of corporate risk management should arise and can be justified (Stulz, 2001; Smith et al., 1990). That is to say, one or more of the assumptions of MM propositions such as no taxes, transaction costs or issuance costs, must be violated for corporate hedging to enhance firm value. As discussed below, capital market imperfection such as agency conflicts, financial distress costs, costs of external finance and taxes can provide positive rationales for corporate hedging. Empirical research on corporate hedging theories has been limited due to the general unavailability of reliable data on corporate hedging activities. Until the beginning of the 1990s, the position of the firm in the derivative risk management was not disclosed in their annual reports, which was considered an essential component of strategic competitiveness. It is only since then that companies are obliged and required to report the notional amount of derivatives they are employing in the footnotes of annual reports. Therefore, most of the earlier empirical studies used questionnaires or other survey data to obtain information on hedging activity in order to examine the determinants of corporate hedging. For example, Nance, Smith, and Smithson (1993) used survey data on Fortune 500 firms' use of derivatives and found that firms that hedge face more convex tax functions, have less coverage of fixed 7

8 claims, are larger, and have more growth opportunities (Allayannis and Weston, 2001). Most empirical research have tried to examine the relation between proxies of determinants for corporate risk management and firm, industry or country specific characteristics in order to identify the determinants of corporate hedging and test whether firms with particular properties should benefit from corporate hedging and thus have great need to hedge (Haushalter, 2000; Tufano, 1996). Moreover, more recent empirical research have distinguished among the types of corporate hedging such as foreign exchange rate risk or interest rate risk (Bartram, Brown and Fehle, 2009; Graham and Rogers, 2002), commodity price risk (Géczy et al., 2006; Haushalter, 2000), or alternative risk management without derivatives (Petersen and Thiagarajan, 2000), which indicates that different factors may be important for each type of hedging. Most of these studies are also typically tested by using binary variables indicating whether a firm uses derivatives or not for the reason that notional amounts disclosed in annual report were less reliable (Bartram, Brown and Fehle, 2009; Mian, 1996; Nance et al., 1993). In addition, most empirical work employs a large number of different proxies to examine the hypothesis such as financial distress costs hypothesis, costs of external finance hypothesis, underinvestment and asset substitution costs hypothesis, transaction costs and economies of scale hypothesis, substitutes for risk management hypothesis and the exposure to financial price risk hypothesis. In addition, several studies examine whether the determinants for the 8

9 decision to hedge and the extent of hedging are different (Graham and Rogers, 2002; Allayannis and Ofek, 2001). 2.1 Underinvestment and Asset Substitution Problems A company can be seen as a nexus of contracts between different parties, such as managers, shareholders, creditors, and employees (Jensen and Meckling, 1976). The manager who runs the company, acting as the agent for the shareholder, is supposed to make decisions that will maximize shareholder wealth. Since managers have more information on daily activities of the company, it is in the manager's own best interest to maximize his own wealth. Consequently, the agency conflict arises where the two parties have different interests and asymmetry information. Even though managers share common goals with shareholders, they still might have incentive to reject positive NPV projects if the gain from the project will accrue to debtholders rather than shareholders (underinvestment or debt overhang problem), which might have the potential to increase firm value. This situation arises especially when a firm is highly levered and firm value is low, as shareholders are reimbursed after debtholders, and thus shareholders might not benefit from valuable projects. While the decrease of debt outstanding, which will sacrifice the tax benefit of debt, can alleviate the underinvestment problem, corporate hedging have the same effect without giving up the tax benefit of debt. 9

10 The agency conflicts between shareholders and debtholders might also arise, if management of a leveraged firm has an incentive to undertake highly risky projects instead of safe projects, even if the projects have negative NPV (asset substitution or risk shifting problem) (Smith, 1995; Campbell and Kracaw, 1990). This is because if the project is successful, shareholders will get all the benefits, in comparison if it is unsuccessful, debtholders have to bear all the cost. If the projects are undertaken, the firm value might decrease and shareholders can transfer wealth form debtholders. Since the debt is highly risky, debtholders demand higher yields on the debt or protective covenants, which will increase more costs and thus reduce the total value of the company (Smith and Warner, 1979). Corporate hedging can stabilize the firm value and reduce the chance that the management shifts towards highly risky investment projects. Thus, agency costs could be reduced through corporate risk management. Underinvestment and asset substitution problems are much more essential for firms which have high investment growth opportunities for the reason that these firms will lose most of the value if they fall to invest into profitable projects, which have high leverage for the reason that there is high possibility that shareholders-debtholders conflicts can occur. Consequently, high investment growth opportunities and high leverage should have positive relationship with corporate hedging. Proxies such as research and development (R&D) expenditures (scaled by total assets), 10

11 property, plant, and equipment (PP&E) expenditures (scaled by total assets), acquisition activities and asset growth ratio could directly measure the existence and magnitude of available investment growth opportunities. Since these proxies directly measure the resources invested into investment growth activities, these variables should be positively related to growth opportunities and, in turn, also related to corporate hedging. Other proxies such as price-earnings and market-to-book ratio are more indirect measurement for high investment growth opportunities. A higher market-to-book ratio indicates larger availability of investment growth opportunities, which suggests that there is a great chance for firms to hedge. Similarly, firms with higher price-earnings ratio are expected to have higher forecast earnings growth. Therefore, market-to-book and price-earnings ratio should have positive relationship with corporate risk management. In addition, the interaction variables such as leverage multiplied the market-to-book ratio or R&D expenditures (financial constraints concurrent with growth opportunities) can also lead to a higher propensity for corporate hedging. Since high growth opportunities and high financial leverage can increase the value of this interaction term, there is a positive relationship between the interaction variables and corporate hedging. When it comes to the empirical evidences of direct measure of available investment growth opportunities, empirical results are strongly in support of the hypothesized relation between R&D expenditures and corporate hedging (Allayannis and Ofek, 2001; Gay and Nam, 1998; Géczy et al., 1997; Nance et al., 1993). However, the 11

12 coefficients of property, plant, and equipment (PP&E) expenditures, acquisition activities and asset growth ratio never have the correct direction at any conventional significance levels (Haushalter, 2000; Tufano, 1996), which is due to the fact that the relation between growth options and corporate risk management is more complicated than conjectured in the empirical research (Morellec and Smith, 2007). With regards to the more indirect measures of investment growth opportunities, the price-earnings ratio provides some support for the underinvestment and asset substitution hypothesis in multivariate tests (Gay and Nam, 1998; Berkman and Bradbury, 1996). Moreover, the sign of the coefficient of market-to-book ratio is often correct with the exception of Mian (1996), which finds that a firm s growth opportunities is negatively and statistically related to corporate risk management. 2.2 Costs of External Finance Corporate risk management can add value to firm by coordinating corporate investment and financing policies. Internal funds that are invested into investment growth activities often vary dramatically because of the uncertainty of future free cash flow. Consequently, if firms do not have sufficient internal funds to finance available positive NPV projects, they have to abandon their current investment plan or to raise more money through external finance such as equity and debt which are much more costly due to agency conflicts. Because of high expectations of bankruptcy and financial distress, debtholders will 12

13 demand higher yields on the firm s debt considering the probability of default into their lending decisions. While this can guarantee that they can get a fair value for their investment, the debt is much more costly for companies and thus the value of firm will be largely decreased. On the other hand, debt covenant is another way to protect their claims and help them ensure that the risk attached to the debt does not unexpectedly deteriorate prior to maturity, which has limitation on decision-making of management with respect to corporate investment and financing policies. Therefore, the firm value is reduced if the debt covenant restricts managers from following sound investment projects. Since external finance is too costly, firms usually choose to abandon their current investment plan than turn to financial capital markets. Corporate hedging can make sure that internal funds are sufficient to finance available and profitable investment growth opportunities instead of raising capital through costly external finance. Similar to underinvestment and asset substitution problem, the costs of external finance hypothesis depends on the existence of cash flow constraints and available investment growth opportunities in the situation of imperfect capital market. Consequently, the proxies for underinvestment and asset substitution hypothesis can also be applied to test this hypothesis. The existing empirical evidence strongly supports the hypothesis that companies with a greater correlation between available and needed funds have lower incentive to 13

14 hedge than companies with a lower correlation (Gay and Nam, 1998). Likewise, there is strong empirical evidence in support for the hypothesis that derivative hedging firms have a lower sensitivity of investment to pre-hedging cash flows (Allayannis and Mozumdar, 2004). In comparison, there are also evidences to support the hypothesis that the effect of derivatives use on cash flows is small (Guay and Kothari, 2003). Direct evidence supporting this hypothesis can be found in Minton and Schrand (1999), while indirect evidence can be found in Géczy et al. (1997). 2.3 Costs of Bankruptcy and Financial Distress Firms with high leverage run the risk of having less sufficient free cash flow and failing to pay off all financial payment obligations prior to maturity, which increases with higher leverage or high uncertainty of free cash flow. A firm that cannot repay the debts it owes to creditors is forced into bankruptcy that is associated with some direct and indirect costs to the company. Indirect costs of bankruptcy are costs of creditors, loss of customers, employers and suppliers, loss of receivables, fire sale of asset and insufficient delay of liquidation. Whereas, direct costs of bankruptcy accrue in the actual bankruptcy procedure and pertain mainly to lawyers charges, administration and accounting fees, and expenses for expert witnesses. On average, direct costs of bankruptcy are often only in the order of 1% to 3% of shareholder value (Weiss, 1990; Warner, 1977b). In comparison, indirect costs tend to be much larger than direct costs and can approach 20% of firm value (Cutler and Summers, 1988). 14

15 The chance of financial distress increases when a firm has high fixed costs, illiquid assets, or revenues that are sensitive to economic downturns and the expected costs of financial distress stem mostly from low levels of firm value. However, corporate risk management reduces the volatility of company value, thus it can add value to firm by decreasing the expected costs of financial distress. That is to say, corporate hedging enables firms to carry more debt and enjoy greater tax shields (Graham and Rogers, 2002; Leland, 1998). Firms with high default rate and high possibility of bankruptcy should have higher chance to hedge for the reason that corporate hedging can lower the costs of bankruptcy and financial distress. The most common proxy of costs of financial distress hypothesis is leverage or debt ratio, which can indicate the possibility of bankruptcy and financial distress. High leveraged firms with less sufficient free cash flow might have higher possibility of failing to pay off all payments obligations prior to maturity, thus they are more likely to engage in corporate hedging. However, the use of leverage ratio as the measure of bankruptcy and financial distress is not without controversy, the leverage ratio fails to consider the level of firm s cash flow or negative debt because a firm with high levels of free cash flow is less financially constrained than a firm with a lack of available cash. Therefore, a modified financial distress proxy is net debt ratio (debt net of cash and short-term investments). To account for the fact that a firm with high leverage does not necessarily imply a higher 15

16 probability of bankruptcy and financial distress, the interest cover ratio is used as alternative proxy to measure the expected costs of bankruptcy and financial distress. Since a higher interest coverage ratio suggests more pre-tax income to satisfy payment obligations, it should have negative relation with corporate hedging. Firms with cash or cash equivalent and marketable securities particularly higher short-term liquidity, which can be used to pay off financial payment obligations in a very short time, have very low incentives to hedge. The most popular proxies for short-term liquidity are cash ratio and quick ratio. By the same token, highly liquid firms that are paying out high dividends regularly to their shareholders have lower need to hedge. In comparison, if firms exhaust their liquidity by paying dividends, they should be more inclined to hedge. Preferred stock and convertible debt (both scaled by total assets) should be positively related to corporate hedging because they constrain a company s access to external finance and create more financial distress costs (Géczy et al., 1997). Alternatively, firms can maintain the tax benefit of debt and control the firm s agency conflicts by issuing preferred stock and convertible debt acting as substitutes for corporate risk management (Nance et al., 1993). Therefore, the direction of the relation of these variables with derivatives use is still an empirical question. In addition, asset tangibility might also help to proxy for the probability of bankruptcy and financial distress. As tangible assets could be easily sold off to increase liquidity, firms that have large proportion of tangible assets are more liquid and have less cost of bankruptcy and financial distress than firms with more 16

17 intangible assets, thus they have lower propensity to hedge (Howton and Perfect, 1998). A company s profitability which is negatively related to corporate hedging might also be a determinant of its corporate risk management, since unprofitable or less profitable firms might have insufficient free cash flow and thus experience more difficulties in paying off their financial payment obligations. Therefore, unprofitable or less profitable companies might have much stronger incentives to hedge than profitable companies. The common measures of profitability, which are also the proxies for financial distress costs hypothesis, are return on assets (ROA), gross margin and sales. With regards to the size of financial distress costs, all empirical studies examine the relation between firm size and corporate risk management. However, a positive or negative relation between firm size and corporate hedging is still an empirical question. The negative relationship between firm size and financial distress costs indicates that there is a greater chance for small firms to hedge (Allayannis and Ofek, 2001; Haushalter, 2000). Small firms do not have significant economies of scale and face higher financing transaction costs and more expensive external finance and thus are more likely to engage in corporate hedging. In contrast, large firms are more inclined to hedge than small firms, and thus seems to suggest that corporate hedging 17

18 exhibits significant economies of scale. The general proxy for firm size is the natural log of total assets. To control for endogeneity problems, the corporate hedging and capital structure decisions can be modeled by using simultaneous equations in an effort to test whether and to what extent the increased debt capacity and leverage due to hedging can create value for firm. The results of simultaneous equation models provide strong evidences that leverage ratio is significantly and positively related to corporate hedging (Bartram, Brown and Fehle, 2009; Graham and Rogers, 2002; Géczy et al., 1997). Moreover, while the sign is sometimes in the predicted direction, the interest coverage ratio is often insignificant at conventional levels (Howton and Perfect, 1998; Berkman and Bradbury, 1996). In addition, the proxies of profitability generate mixed results. In univariate analyses, all three profitability measures such as ROA, gross margin and sales indicate that firms with more profit have a greater chance to hedge, which is opposite to the costs of financial distress hypothesis. In contrast, these variables are sometimes positively and sometimes negatively associated with derivatives use in multivariate tests. The empirical evidence strongly indicates that derivative hedgers have lower short-term liquidity than non-hedgers (Allayannis et al., 2003; Géczy et al., 1997; Tufano, 1996). Finally, when it comes to cash dividend or dividend yield, the sign of dividend yield is still not clear in the empirical research. Although many empirical studies obtain significant results, the relation varies across analyses (Bartram, Brown and Fehle, 2009; Haushalter, 2000; Mian, 1996). Similarly, the 18

19 direction of preferred stock and convertible debt is still an empirical question, as these proxies are often insignificant at conventional levels (Gay and Nam, 1998). 2.4 Exposure to Financial Price Risk Anther potential factor in the corporate risk management is the magnitude of financial price risks particularly the level of exposure to foreign exchange rate risk and interest rate risk. Firms with greater fluctuation in cash flows or accounting earnings due to exposure to financial price risks have stronger incentives to hedge. Furthermore, the costs of hedging are likely to be lower for firms with greater financial price exposure due to economies of scale (Clark, Judge and Ngai, 2006). Thus, the exposure to financial price risks is positively related to corporate risk management. The measures of exposure to foreign exchange rate risk are foreign sales (scaled by total assets), foreign currency debt (dummy) and foreign operation (dummy). The proxy for exposure to interest rate risk is interest cover ratio. 2.5 Corporate Taxes If the tax schedule is convex, that is to say, if taxes increase more than proportionally with taxable income, volatile taxable income generates a higher tax burden than stable pre-tax income. However, corporate hedging can stabilize taxable income by creating more value for firm, since savings from higher income states exceed additional taxes from lower income states, thus lowering the average corporate tax burden (Stulz, 2001; Bartram, 2000; Graham and Smith, 2000; Smith and Stulz, 1985). 19

20 The direct factor that causes convexity in the effective tax function is progressivity in the statutory tax code. However, statutory progressivity is relatively limited in most tax systems (Mayers and Smith, 1990). In addition, the indirect factors, which also can give rise to convex tax function, come through special tax preference items such as tax loss carry-forwards, investment tax credits and foreign tax credits (MacKie-Mason, 1990). Therefore, The more convex the tax schedule or the more special tax items that firms have, the greater the incentive for them to hedge. The empirical evidence provides strong support for the tax hypothesis by using a tax code progressivity dummy, which indicates income in the convex tax region (Haushalter, 2000; Howton and Perfect, 1998; Nance et al., 1993). In comparison, marginal (or average) tax rate proxies are significantly but positively related to the corporate risk management, which is not consistent with the tax hypothesis (Haushalter, 2000). There are, however, some evidences to support the tax hypothesis when the tax savings from volatility reductions are considered (Dionne and Triki, 2005). This might be explained by the fact that the tax incentive to hedge in order to increase leverage is larger than the tax incentive of progressivity (Graham and Rogers, 2002; Graham and Smith, 1999). The dummies for special tax preference items generally lead to the similar findings. In the majority of studies, there is no relation between tax loss carry forwards dummy and corporate hedging. In contrast, the empirical studies provide strong evidence to support that tax credits is positively and 20

21 statistically related to corporate risk management (Bartram, Brown and Fehle, 2009). Table 1 summarizes the variable s definition and the source of data for the variable. 3. Sample Formation The sample is taken from the largest 600 European firms on the FTS Euro First index ranked by the market capitalization as of year-end 2003 and the final sample consists of 253 European non-financial firms and comes from 17 different countries in Europe. The data on corporate hedging was obtained from qualitative and quantitative derivative risk management disclosures contained in the annual report of company for the year ended This study classifies firms as derivative hedgers as those that have specific quantitative disclosures in their annual report only for hedging purposes rather than speculative purposes. Table 2 provides data on the number of derivative users for the sample of 253 European non-financial firms that are supposed to have exposure to financial price risks for the year ended It also shows the number of derivative hedgers and the types of derivative used for each European country. Panel A presents data on the number of derivative users and non-users. Panel B provides data on combinations of exposures hedged between foreign exchange derivative, interest rate derivative and commodity price derivative. Panel C and panel D present data on the types of foreign exchange derivatives used and interest rate derivatives used respectively. Foreign exchange derivatives include currency swap, currency option and foreign exchange 21

22 rate forward. Interest rate derivatives include forward rate agreement, interest rate swap and interest rate option (cap, floor and collar). Part A of Table 2 shows that 83.8 percent of firms in the sample are classified as derivative users, whereas only 16.2 percent are non-users. Part B presents that more than half of firms in the sample are both foreign exchange rate and interest rate derivative users. More than one fourth of firms are three types (foreign exchange rate, interest rate and commodity price) derivative users. Part C and Part D shows that 93 percent of foreign currency hedgers use forward exchange rate contract and 97.4 percent of interest rate derivative users choose interest rate swap. Table 3 provides summary statistics of explanatory variables. It is necessary to point out that the mean level of foreign sales to total sales is 58% percent for our sample. This level of foreign sales is at least triple that reported for US firms. Allayannis and Weston (2001) indicate that foreign sales as proportion of total sales were on average 18% percent of for 720 US firms. Graham and Rogers (2002) reports foreign sales to total sales were 10% percent for their sample of US firms. Table 4 shows the correlation matrix for each variable used in the univariate and multivariate analysis. 4. Empirical Results 4.1 Univariate Analysis Table 5 reports the parametric (Two Sample T-test) and non-parametric (Wilcoxon 22

23 Rank Sum Test) tests results of comparisons of independent variables between derivative users and non-users. Panel A shows the tests of differences between the means of independent variables for users and nonusers of derivatives; Panel B reports the tests of differences in location between the derivative hedgers and non-hedgers. The number of observations may vary dramatically due to general unavailability of reliable data. Derivative user firms are not significantly different from non-users firms with respect to variables that are proxies of determinants for investment growth opportunities. The results show that the market-to-book and price-earnings ratio are insignificantly higher for derivative users than that for non-users, which are proxies that indirectly measure the existence and magnitude of available investment growth opportunities. Derivative hedgers have insignificantly higher property, plant & equipment and research & development than non-hedgers, which are more direct measures of growth opportunities. Hedging firms are statistically different from non-hedging firms with regard to the interaction variable (the leverage ratio multiplied with the market-to-book ratio) based on the Wilcoxon rank sum test, partially supporting the hypothesis that financial constraints concurrent with growth opportunities lead to a higher probability for corporate hedging. Collectively, these results provide very weak evidence in support of the underinvestment or asset substitution costs hypothesis. 23

24 Derivative hedging firms are statistically different from non-hedging firms with respect to variables that are proxies of determinants for bankruptcy and financial distress costs. All measures of leverage (gross leverage and net leverage) are significantly higher for derivative users than for non-users. The net leverage ratio is positive and significant at 1% percent level, whereas the gross leverage ratio is statistically positive but at 5% percent level. Interest cover ratio is statistically lower for hedgers than that for non-hedgers based on the Two Sample T-test. Hedging firms are more likely to have tax credit and tax loss carry forward. Collectively, these results are consistent with the costs of financial distress hypothesis. Convertible debt, preferred stock, cash dividend (dividend yield) and the liquidity of firm s assets are proxies employed to examine the substitutes for hedging hypotheses. The results of univariate analysis show that derivative users have significantly higher dividend yield than that for non-users. The difference of the means for derivative hedgers and non-hedgers indicates that hedging firms have significantly lower quick ratio relative to non-hedging firms in the two sample T-test. Tangible assets are more important for non-hedgers than that for hedgers based on the two sample T-test. These suggest that hedging firms have statistically lower levels of liquidity relative to non-hedging firms for both liquidity measures. However, there is no significant difference with respect to preferred stock and convertible debt. Collectively, the univariate evidence provides weak support for the substitutes for hedging hypotheses. 24

25 In addition, hedging firms are statistically larger than non-hedging firms, which is consistent with the hypothesis of transaction costs and economies of scale. Finally, derivative users have significantly greater exposure to foreign exchange rate risk than non-users, as measured by foreign sales, foreign operation and foreign currency debt. 4.2 Multivariate Analysis The univariate tests in the previous section present the results for difference in means between derivative users and non-users for various firm-level characteristics. However, the univariate analysis does not consider the correlation between different firm-level characteristics and hence, the multivariate approach is employed to investigate the simultaneous effects of the independent variables on firm s corporate risk management. We employ the multivariate logistic analysis where firms have qualitative and quantitative disclosures on hedging activity contained in annual reports. Hedging firms are assigned a value of one for the binary variable, and non-hedging firms are assigned a value of zero. We use equation (1) to obtain the predicted probability of hedging. Log P! 1 P! = α! + α! Tax! + α! FD! + α! UC! + α! Exp! + α! Sub! + α! Tcosts! + ε! (1) Where Tax = Tax loss carry forward or tax credits; FD = Financial distress; UC = Underinvestment costs or costs of external finance; Exp = Financial price exposure; Sub = Hedging substitutes; Tcosts = Transaction costs or economies of scale. 25

26 The categorical variable (the binary variable) is regressed on independent variables that measure tax function convexity or special tax preference items, costs of bankruptcy and financial distress, potential investment growth opportunity for company, the exposure to financial price risks, the substitutes for hedging and transaction costs and economies of scale. According to results of the multivariate logistic analysis, the coefficients of the independent variables such as market-to-book ratio, price-earnings ratio, property, plant & equipment, research & development, market-to-book*leverage, preferred stock, convertible debt, dividend yield, quick ratio, tangible asset and tax credits are not statistically significant. Therefore, these results in the logistic regression are excluded and not reported in this paper, which will not affect the final conclusion. Table 6 reports the multivariate logistic regression estimates of relationship between the probabilities that the European firms hedge and proxies for incentives to corporate risk management. We employ several models that consist of different combinations of proxies, because the problem of multi-collinearity might exist if we include all the independent variables simultaneously. Derivative hedgers are significantly different from non-hedging firms with respect to variables that are proxies of determinants for financial distress and financial contracting costs. The estimated coefficients on gross leverage are statistically positive at the 10% level whereas the coefficient on net leverage is positive and significant at the 5% level. Moreover, the estimated 26

27 coefficient on the interest cover ratio is negative and significant at the 10% level. In addition, the tax loss carry forward dummy has a statistically positive coefficient. These results provide strong evidence in support of the financial distress costs hypothesis. Therefore, firms with higher leverage ratio, lower interest cover ratio and more special tax preference items are more inclined to corporate hedging. Interestingly, the return on assets (ROA) ratio has positive and significant coefficient, which suggests that firms with more profitability have a stronger need to hedge. In this case the observed relationship between corporate hedging and profitability is opposite to the predicted direction of financial distress costs hypothesis. The firm size is positively and significantly related to corporate hedging that is consistent with the transaction costs and economies of scale hypothesis. Finally, the estimated coefficients on foreign sales to total sales, foreign currency debt dummy and foreign operation dummy are positive and significant, which confirms the argument that firms with higher exposure to foreign exchange rate risk have higher propensity to hedge. It is necessary to point out that the multivariate tests provide no evidence in support of the underinvestment or asset substitution hypothesis and the substitutes for hedging hypothesis. The unreported results for these hypotheses are not statistically significant even though they have correct direction. 5. Corporate Hedging, Debt Capacity and Firm Value In the previous section, the results of multivariate logistic regression show that 27

28 leverage increases the likelihood of corporate hedging for European firms. In this section, we jointly model the corporate hedging decision using simultaneous equations in an effort to test whether and to what extent the increased debt capacity and leverage due to hedging can create value for firm. The value creation for company mainly stems from the reduction of financial distress costs and enhanced debt capacity due to corporate hedging. To examine whether the corporate risk management can increase debt capacity, we follow Graham and Rogers (2002) and estimate the determinants of the capital structure and corporate hedging decisions simultaneously with a two-stage least square technique. In the first stage, the multivariate logistic regression is performed using derivatives hedging (binary variables) as dependent variables. In the second stage, structural equations are estimated using the predicted values from the first-stage regressions as explanatory variables. For the first stage, the hedging specification is estimated with a logistic model and the dependent variable assigns hedgers the value of 1 if firms use derivatives for hedging in their annual reports; other non-hedgers are assigned the value of 0. The independent variables in logistic model are listed in Table 1. The binary variable is regressed on independent variables that measure tax function convexity or special tax preference items, costs of bankruptcy and financial distress, potential investment growth opportunity for company, the exposure to financial price risks, the substitutes for hedging and transaction costs and economies of scale. We employ equation (1) to 28

29 obtain the maximum likelihood of corporate hedging: Log P! 1 P! = α! + α! Tax! + α! FD! + α! UC! + α! Exp! + α! Sub! + α! Tcosts! + ε! (1) Where Tax = Tax loss carry forward or tax credits; FD = Financial distress; UC = Underinvestment or asset substitution costs or costs of external finance; Exp = Financial price exposure; Sub = Hedging substitutes; Tcosts = Transaction costs or economies of scale. For the second stage, we employ the capital structure decision equation following Rajan and Zingales (1995) and add the estimated hedging probabilities from the first-stage regressions as explanatory variables to measure the sensitivity of leverage to hedging. The capital structure decision equations are: Leverage(Firm! ) = β! + β! Tangible Assets! + β! R&D! + β! LogSize! + β! Profitability! + β! Hedging +ε! (2) Where Hedging* is endogenous and presents the estimated likelihood of corporate hedging obtained form first stage estimation (equation 1) of the logistic regression. The first row of Table 7 presents the predicted coefficient and its P-value for the hedging* variable in the second stage capital structure decision equation. The 29

30 corporate hedging is associated with a increase in the leverage ratio. The estimated probability of hedging is significantly and positively related to leverage. This suggests that the corporate hedging can add value to European firms by enhancing their debt capacity and thus increase their tax benefit. Then, the size of the tax benefit could be quantified by the increased debt capacity for each corporate hedging firm scaled by the total value of company. Specifically, taking the product of the predicted coefficient on the corporate hedging variable, the firm s average tax rate and book value of total debt, which generates the incremental debt tax benefit from hedging, and then scaled by the market value of the firm s assets lagged by one year (Clark, Judge and Ngai, 2006). For all derivative hedgers the increase in leverage of converts into a mean (median) predicted increase in firm value of 2.10% (1.59%). The column 2 of Table 7 reports that for all Europeans derivative hedgers the increase in leverage of converts into a mean (median) predicted increase in firm value of 2.10% (1.59%), which is similar to the corresponding results for UK firms reported by Clark, Judge & Belghitar (2008). As a matter of cross-country comparison, the column 3, 4 and 5 present the results of US derivative users, Hong Kong and Chinese users and UK derivative users respectively. The value effect of corporate hedging for European derivative hedgers is larger than 1% percent reported by Graham and Rogers (2002) for US derivative hedgers in the Column 3. This indicates that the value creation due to corporate hedging is much higher for European firms than for 30

31 US firms. The mean increase of tax benefit for Chinese and Hong Kong firms is 0.84% reported by Clark, Judge and Ngai (2006) in Column 5, which is much lower than European and US firms due to the fact that the Hong Kong and Chinese firms have lower leverage ratio and thus lower tax benefit. 6. Caveats and Limitation of Empirical Analysis Empirical analysis of positive theory of corporate hedging decision has various caveats and limitation, which are essential to be considered when interpreting the results or findings of empirical studies. In the first place, most empirical work does not consider the endogeneity problems of variables, which have different characteristics of corporate financing policy. To illustrate, almost all empirical work typically employ proxies such as leverage or debt ratio, cash dividends or dividend yields, Altman s Z-score, ownership concentration, executive stock options and so on, which are all endogenous to some extent. Consequently, most of empirical findings suffer from a serious simultaneous equations bias (Guay, 1999). Secondly, the identification problem is another major issue that many empirical studies fail to address. Therefore, the conclusions that are drawn form these empirical findings are seriously limited. To illustrate, it is difficult to find appropriate proxy variables of determinants for corporate risk management that are not simultaneous as determinants of other corporate hedging hypothesis such as leverage ratio, convertible debt, preferred stock and firm size. For example, convertible debt and preferred stock 31

32 could be the proxies for the substitutes for hedging hypothesis, which are also important for costs of financial distress hypothesis (Bartram and Artez, 2009). Most empirical studies fail to recognize the identification problem and only few recent empirical works have tried to solve this issue. Finally, almost all empirical analysis frequently focus only on one single dimensions of risk management, that is, the use of financial derivatives, to proxy for hedging at the firm level. In comparison, firms typically employ a range of coordinated risk management instruments, such as foreign currency debt, operational hedging measures, and pass-through. To illustrate, a firm may have a strong incentive to hedge, but might do so using hybrid debt instruments instead of stand-alone derivatives (Bartram, 2006; Gay et al., 2003; Brown and Toft, 2002). In summary, almost all empirical studies in the field of corporate risk management are faced with significant limitation and challenges that need to be put into consideration when interpreting the results. To illustrate, the endogeneity problems, the empirical modeling of structural relations, the identification problems of appropriate proxy variables for corporate hedging beyond the use of derivatives are major issues that most empirical studies fail to explain and thus potentially limit the conclusions that can be drawn from their results (Bartram and Artez, 2009). 32

33 7. Conclusion In this paper, we employ the unique data for European non-financial firms to examine the determinants and value effect of corporate hedging. Using data obtained from DataStream and Annual Report, the empirical proxies of determinants of corporate hedging identified in this study are independent variables that are related to financial distress costs hypothesis, underinvestment and asset substitution hypothesis or costs of external finance hypothesis, financial price exposure hypothesis, the substitutes for hedging hypothesis and the transaction costs or economies of scale hypothesis. The empirical evidence in support of financial distress costs hypothesis is especially strong, which shows that the relation between all measures of leverage and the corporate hedging decision is positive and significant and the interest cover ratio is significantly and negatively related to corporate risk management. We also provide strong empirical evidence linking the transaction costs, economies of scale and corporate risk management, which presents that firm size is statistically and positively related to corporate hedging. Our results also show that the exposure to foreign exchange rate risk is positively related to corporate risk management. In addition, when we look at the relationship between liquidity and corporate risk management, the results of univariate analysis provide some evidences to support that the liquidity reduces costs of bankruptcy and financial distress. However, according to the results of multivariate analysis, there is no relation between liquidity and corporate hedging, which is not consistent with the substitutes for hedging hypothesis. Interestingly, the empirical results provide that the potential investment growth opportunities have no 33