Supplier-Customer Relationships and Corporate Hedging Policy

Transcription

1 Supplier-Customer Relationships and Corporate Hedging Policy Jun-Koo Kang, Limin Xu, and Lei Zhang * Current draft: December, 2012 * All authors are from the Division of Banking and Finance, Nanyang Business School, Nanyang Technological University, 50 Nanyang Avenue, Singapore Kang can be reached at jkkang@ntu.edu.sg, Xu can be reached at xu0005in@e.ntu.edu.sg, and Zhang can be reached at zhangl@ntu.edu.sg. Please address all correspondence to Limin Xu (corresponding author). All errors are our own. 1

2 Supplier-Customer Relationships and Corporate Hedging Policy Abstract We investigate how a firm determines its hedging policy considering its major customer relationships. We find that the likelihood of a supplier firm using derivatives to hedge interest rate risk is higher when its major customer has high leverage. This result is more pronounced as the supplier economic dependence on its customer increases, that is, when it has a large sales dependence to the customer, when it operates in a durable goods industry, or when it makes high relationship-specific investments. We also find that supplier hedging helps the supplier maintain durable future customer relationships, especially with highly leveraged customers. Further, we find that bond offer announcements by customers and announcements of customers credit downgrades by rating agencies have significant negative effects on the stock prices of non-hedging suppliers, but not on those of hedging suppliers. These results suggest that the major customer s financial distress risk is an important determinant of a supplier s hedging policy and supplier hedging helps increase the perceived viability of the future major customer relationship and alleviate potential negative spillover effects along the supply chain. 2

3 In a Modigliani-Miller (1958) world, corporate hedging policy would be irrelevant to firm value. However, risk management using financial derivatives has become increasingly prevalent. For example, according to the Bank for International Settlements, the outstanding notional value of interest rate derivatives held by nonfinancial firms increased from $6.1 trillion at the end of 2000 to $37.4 trillion at the end of December Finance theory suggests that hedging increases firm value by reducing the expected costs of bankruptcy and financial distress (Smith and Stulz (1985), Stulz (1996)). 1 Consistent with this view, several studies show that hedging reduces stock mispricing (Lin, Pantzalis and Park (2010)), increases firm value (Allayannis and Weston (2001), Zou (2010)), and lowers cost of debt (Campello et al. (2011)). Previous studies further examine the factors that determine a firm s hedging policy and show that corporate tax convexity, financial distress, CEO risk taking incentives, capital market imperfection, and industry competition are important determinants of corporate hedging policies (Smith and Stulz (1985), Froot, Scharfstein and Stein (1993), Nance et al. (1993), Tufano (1996), Mian (1996), Géczy, Minton and Schrand (1997), Graham and Smith (1999), Graham and Daniel (2002), Knopf, Nam and Thornton (2002), Adam, Dasgupta and Titman (2007), and Purnanandam (2008)). 2 However, a firm s hedging policy and hedging benefits can also depend on its stakeholder relation. For example, Smith and Stulz (1985) and Stulz (1996) argue that important stakeholders such as employees, customers, and suppliers cannot diversify away their relationship disruption risks and thus require extra compensations for bearing nondiversifiable risk of their claims. Specifically, if high cash flow uncertainty caused by the changes in interest rates, foreign exchange rates, or commodity prices increases the perceived 1 These costs include both direct and indirect bankruptcy costs due to the potential loss of customers, suppliers, other stakeholders, and growth opportunities (Bris et al. (2006), Purnanandam (2008)). 2 These studies show that firms with a higher debt ratio, those facing more convex tax functions, those whose managers compensation is a concave function of firm value, or those facing more costly external financing are more likely to adopt hedging policies. 1

4 risk of a firm s financial distress, the stakeholders who hold large claims on firm value will have to bear significant non-diversifiable risk due to relationship-specific nature of their claims. Therefore, stakeholders with large claims should have strong incentives to use various hedging techniques to reduce their exposure to this non-diversifiable risk. In spite of this potentially important effect of stakeholder relationship on corporate hedging policies, the literature has paid little attention on this issue and other important questions such as the effect of relationship-specific characteristics on a firm s hedging decision and how its hedging policy affects stakeholders incentives to support their relationships. In this study, we use a large sample of U.S. firms with major customer relationships to shed light on these unexplored issues. We focus on firms with major customer relationships for two reasons. First, the major customer serves as the most important key stakeholder in many firms and its durable trading relationship with a supplier is the prevailing business practice around the world. 3 For example, Fee et al. (2006) show that almost 16 percent of Compustat firms in the U.S. has trading relationships with major customers during the periods. Second, these bilateral durable relationships are known to create significant operating and financial interdependence between trading partners (Titman (1984), Maksimovic and Titman (1991), Kale and Shahrur (2007), Banerjee, Dasgupta and Kim (2008), Johnson et al. (2012)). Therefore, the major customer relationship represents an ideal setting to examine how a firm determines its hedging policy considering its stakeholder relationships and how its hedging policy affects stakeholders incentives to support their relationships. Stulz (1996) argues that corporate hedging has significant impact on the relationship between suppliers and customers. For example, suppliers would be reluctant to enter into 3 A major customer is one that accounts for 10% or more of the supplier s total sales. 2

5 long-term trading relationships with the customers that have uncertain prospects since the major customer relationships involve high levels of asset specificity, which makes it costly for suppliers to switch their major customers (Williamson (1985, 1991)). Alternatively, customers who use very specialized inputs in their production and are concerned about the suppliers ability to fulfill warranty obligations in the future may be reluctant to buy products from highly volatile suppliers since they also suffer significant relationship breakup costs in the event of their suppliers financial distress or bankruptcy (Titman (1984), Titman and Wessels (1988)). 4 Supplier hedging can add value in these cases since it encourages more cooperative behavior among trading partners and allows the suppliers to receive more favorable contract terms from their customers. Therefore, supplier hedging lowers the expected switching costs related to relationship breakup and reduces customers concerns about suppliers potential liquidation risk, which benefits both customers and suppliers. These arguments have several testable predictions. First, they predict that supplier firms that maintain the trading relationships with highly leveraged customers are more likely to pursue hedging strategies than those that do not have such relationships. The risk of financial distress tends to increase with a firm s leverage and thus the leverage can serve as an important financial indicator for the probability of the firm s financial distress (Nance et al. (1993), Opler and Titman (1994), Purnanandam (2008)). Opler and Titman (1994) show that during industry downturns, more highly leveraged firms lose market shares to their less leveraged competitors, suggesting that the suppliers sales to the customers are adversely affected if the customers have high leverage. 4 Titman (1984) argues that customer incentives to make relationship-specific investments depend on the firm s financial condition since customers have to bear switching costs if the firm is liquidated. Since these switching costs are particularly large when the firm produces unique products, it has strong incentives to maintain lower leverage to reduce customers concerns about its potential liquidation risk. Supporting this argument, several papers find that firms that produce unique products and those that have bilateral trading relationships maintain low debt ratio (Titman and Wessels (1988), Kale and Shahrur (2007), Banerjee, Dasgupta, and Kim (2008)). 3

6 Moreover, because of high costs associated with switching to new trading partners, suppliers with a major customer will face difficulty in redeploying their relationship-specific assets when their trading relationships are terminated due to customers in financial distress. In addition, previous studies show that suppliers to customers filing bankruptcy experience negative stock price reactions around bankruptcy filing dates (Hertzela et al. (2008), Kolay, Lemmon and Tashjian (2012)), suggesting that the customer s financial distress risk not only increases the business uncertainty of the suppliers, but also is transferred along the supply chain. 5 Thus, if suppliers are dependent on major customer for their sales, customer leverage, as an indicator for its financial reliability, should be an important consideration for their own risk management policy. To the extent that hedging reduces suppliers overall risk to the stakeholders, customers will perceive suppliers commitment to lower the risk as a value increasing action that benefits both suppliers and customers. Therefore, we expect that supplier firms with highly leveraged customers are more likely to implement hedging policies than other supplier firms. The arguments above also suggest that the positive effect of customer leverage on the likelihood of suppliers adopting hedging policies should be more pronounced when the suppliers dependences to major customers are higher, when their relationship-specific investments (R&D intensity) are larger, or when they operate in durable goods industries in which post-sale guarantees are important (Titman (1984)). The previous literature shows that supplier-customer relationships are largely governed by implicit contracts (Shleifer and Summers (1988)) and that firms are more likely to abide by their implicit contracts when their relationships with trading partners involve high relationship-breakup costs, when they 5 Consistent with the view that there exists the contagion effect along the supply chain, Planning Perspectives Inc., an auto industry consulting firm that conducts the survey of executives for industry suppliers for potential bankruptcy of General Motors, finds that 68% of survey firms would have to downsize in the case of General Motors bankruptcy, while 12% would likely close or would definitely do so ( /business/ article/0,8599, ,00. html). 4

7 are required to make large relationship-specific investments (Williamson (1985)), or when their post-sale product support is important (Titman (1984)). Since switching costs associated with relationship breakup and new relationship establishment are high in these cases, the suppliers concern about their major customers financial health is likely to be strong and thus the adverse effect of the customers financial distress risk on suppliers will be more pronounced. Second, the arguments above predict that supplier hedging help increase the durability of major customer relationships and the supplier s future sales dependence to the major customer, especially when it maintains the relationship with highly leveraged customers. Stulz (1996) argues that customers prefer suppliers characterized by low uncertainty such as hedging suppliers. Supplier hedging tends to increase supply chain stability since it strengthens customer incentives to cooperate and commitment to the trading relationship. In particular, since highly leveraged customers are vulnerable to an unexpected weakening of trading relationships, they are more fragile to risk of supply chain disruptions than low leveraged customers. Thus, high leveraged customers prefer to maintain the trading relationships with suppliers that engage in hedging activities and have strong incentives to continue and strengthen the existing relationships. Third, we expect that the negative contagion effect of customers financial distress risk on supplier is lower for hedging suppliers than for non-hedging suppliers. Negative news about the customer s financial reliability (e.g., bond offer and credit rating downgrade announcements) can trigger negative spillover effects along the supply chain by transmitting adverse information on the financial health of major customer-supplier relationships to the trading partner (Johnson et al. (2012), Hertzela et al. (2007), Kolay, Lemmon and Tashjian (2012)). This negative spillover effect, however, should be partially mitigated for suppliers 5

8 with hedging programs since hedging makes the suppliers less vulnerable to customers financial distress risk. Thus, we expect the negative spillover effects for suppliers induced by bond offer announcements by customers and announcements of customer rating downgrade by rating agencies to be less pronounced for hedging suppliers than for non-hedging suppliers. We also expect this attenuating effect of supplier hedging on the value to be stronger when the hedging supplier s sales dependence on the major customer is higher, when its relationshipspecific investments is larger, or when is operates in the durable goods industry. Using hand-collected data on interest rate hedging for Standard & Poor s (S&P) 1500 firms from 1999 to 2008, 6 we find that suppliers are significantly more likely to use interest rate hedging if their major customers maintain high leverage. This positive relation between customer leverage and the likelihood of supplier hedging is more pronounced as the economic interdependence between the suppliers and the customers rises, the relationshipspecific investment increases, or if the suppliers operate in durable goods industries. We also find that compared with non-hedging suppliers, the major customer relationships are less likely to be terminated and the suppliers sales dependence on the major customers are more likely to increase for hedging suppliers. These results are particularly evident when the customers have high leverage, suggesting that the suppliers hedging policies play an important role in supply chain stability when the customers perceived financial reliability is lower. Moreover, we find that non-hedging suppliers experience significant negative abnormal returns around the announcements of customer bond offerings and credit rating downgrades, while there is no significant impact on hedging suppliers. Consistently, this effect is more 6 We do not consider foreign exchange rate hedging in our analysis because we require that both the supplier and its major customer be U.S. firms. Given that the supplier s sales to its major U.S. customer accounts for a significant portion of its total sales, its exposure to foreign exchange risk is likely to be small. Indeed, we find that 73.4% of our sample suppliers do not have any foreign sales. 6

9 pronounced when hedging suppliers sales dependence on the major customers are high, when they make large relationship-specific investments, or when they operate in durable goods industries. Finally, our results are robust to controlling for potential endogeneity concerns that arise because of reverse causality and omitted variable bias. For example, it could be the case that the supplier's hedging decision affects customer leverage, not the other way around. It is also possible that some omitted variables affect both the supplier s hedging decision and customer leverage at the same time, resulting in a spurious correlation. We address these endogeneity problems by examining the changes in the hedging policies of supplier firms after they establish the relationships with new customers. Since the new customers leverage before relationship establishment is unlikely to be affected by the suppliers hedging policies after relationship establishment, the reverse causality concern is likely to be largely alleviated in this case. In addition, since we use the changes in the suppliers hedging policies around new relationship establishment, the potential endogeneity bias caused by omitted unobservable and invariant supplier characteristics would be mitigated. Consistently, we find that non-hedging suppliers are more likely to adopt hedging policies after they establish the relationships with highly leveraged new customers, particularly when they are more dependent upon their new customers, when they make larger relationship-specific investments, or when they operate in a durable goods industry. 7 Overall, these results suggest that suppliers with major customer relationships have an ex-ante incentive to consider their customers financial distress risk as an important factor in setting their hedging policies. The results also suggest that supplier hedging lowers the expected switching costs associated with relationship breakup and reduces customers concerns about suppliers potential supply chain disruption risk. 7 In an unreported test, we find that hedging suppliers establishing trading relationships with a high leveraged new customer are less likely to discard their hedging policies in the future. 7

10 Our study contributes to the literature in several ways. First, our study is the first to examine how firms hedging policies are affected by their stakeholder relation. Prior studies find that several factors such as tax convexity (Smith and Stulz (1985), Graham and Smith (1999), Graham and Daniel (2002)), financial distress risk (Smith and Stulz (1985), Nance et al. (1993), Purnanandam (2008)), costly external financing (Froot, Scharfstein and Stein (1993)), CEO risk taking incentives (Knopf, Nam and Thornton (2002)), and industry competition (Adam, Dasgupta and Titman (2007)) affect firms hedging policies. We show that firms hedging policy also depend on their relationships with important stakeholders such as major customers. Second, our paper extends the product market literature by examining how customers financial distress risk and relationship-specific characteristics affects suppliers incentives to engage in hedging activities. Titman (1984) and Maksimovic and Titman (1991) argue that stakeholders incentives to make firm-specific investments depend on the firm s financial condition. Kale and Shahrur (2007) and Banerjee, Dasgupta and Kim (2008) show that firms that produce unique products and those with major customer relationship have low leverage. We contribute to this literature by showing that suppliers incentives to implement hedging policies depend on the customers financial condition. We also find that the relation between hedging probability and customer s financial condition is stronger when their sales is more dependent on major customers, when their major customer relationships involve larger relationship-specific investments, and when their post-sale guarantees are important to customers. Third, our study provides new evidence on the benefit of hedging by showing that hedging makes the firm less vulnerable to customer financial distress and thus mitigates the negative spillover valuation effect caused by a weakening in major customer relationships. 8

11 The rest of the paper is organized as follows. In Section II, we describe the data and sample characteristics. Section III investigates the impact of customer leverage on the supplier s hedging decision. Section IV studies the effect of the supplier s hedging policy on the durability and the strength of the supplier-customer relationship. Section V analyzes abnormal returns for suppliers around their largest customers bond offer and rating downgrade announcements. Section VI discusses the results from robustness tests. Section VII summarizes and concludes the paper. II. Data and Summary Statistics A. Sample Our initial sample includes all firms covered in S&P ExecuComp for the period from 1999 to We restrict our sample to those firms whose stock return and financial data are available in CRSP and Compustat, respectively. We also exclude firms in the financial industries (primary SIC codes ) since their hedging policies are known to be more complicated than those of industrial firms and they may use derivatives contracts for both trading and hedging purposes. 9 We obtain each firm s information on the usage of interest rate derivatives by extensively searching its annual report on Form 10-K filed electronically in the SEC s Electronic Data Gathering and Retrieval (EDGAR) database. 10 Specifically, we search Item 8 We require our sample firms to be included in ExecuComp because we need CEO compensation data to control for the effect of CEO compensation on a firm s hedging decision. Previous studies show that the firm s hedging decision is significantly affected by the risk attitude of its CEO (Smith and Stulz (1985), Tufano (1996), Knopf, Nam and Thornton (2002)), suggesting that CEO compensation, particularly CEO option compensation, should be controlled for in the analysis of corporate hedging policy. 9 In untabulated tests, we exclude 14 firms in the utilities industries (primary SIC codes ) from the analysis and obtain results that are almost identical to those reported in the paper. 10 The Financial Accounting Standards Board (FASB) issued new accounting rules for derivatives and hedging transactions (FAS 133) in June 1998 that were effective after FAS 133, which requires firms to disclose the fair market values of derivatives contracts, replaces FAS 119 in 1994 that allows them to disclose the notional values of derivatives contracts. This change in accounting rules during our sample period prevents us from using the values of derivatives contracts in our analyses. 9

12 7A of Form 10-K, Quantitative and Qualitative Disclosures About Market Risk, to examine whether the firm uses the interest rate derivative for hedging. If we are not able to obtain information on the usage of interest rate derivatives from Item 7A, we use the following keywords to locate the firm s use of interest rate derivative in Form 10-K: derivative, hedge, swap, market risk, forward contract, option contract, and risk management. When we find at least one of these keywords, we carefully read its surrounding text to determine the firm s use of interest rate derivative. If the firm s annual report on Form 10-K makes no reference to the use of interest rate derivatives, we treat the firm in that year as a non-hedging firm. We are able to identify 2,037 firms (12,082 firm-years) interest rate hedging policy with derivatives instruments. We use the Compustat Segment Customer file to identify whether a firm discloses its major customers. 11 To determine if the major customer is a privately held or publicly traded firm, we manually match the customer names to the universe of firms in Compustat by closely following the approach in Fee et al. (2006). If a firm is not covered by the Compustat Segment Customer file, it is assumed to have no major customer. Out of 12,052 firm-year observations covered in S&P ExecuComp, CRSP, and Compustat from 1999 to 2008, 60.28% have major customers (25.73% are publicly traded customers, 31.32% are privately held customers, and 6.97% are government customers). To be included in our final sample, we require a firm to have at least one publicly traded customer covered in the Compustat Segment Customer file. Our final sample consists of 641 supplier firms (2,616 supplier firm-year observations), of which 263 supplier firms (855 firmyear observations) use interest rate derivatives for hedging. B. Summary Statistics 11 SFAS 14 requires that public firms disclose the names of their principal customers that account for at least 10 percent of their total sales or whose purchase has a material impact on their businesses. The Compustat Segment Customer file includes customers that account for more than 10% of the firms' sales revenue or those whose purchases have a material impact on their businesses. 10

13 Panel A (B) of Table I reports the distribution of the sample of 2,616 supplier firms according to year (industry) and whether supplier firms use interest rate derivatives for hedging. Supplier firms using an interest rate derivative account for 32.7% of the sample suppliers. This number is comparable to that of 35.6% in Campello et al. (2011) who examine the cost of bank loans of hedging firms. The fraction of supplier firms using an interest rate derivative does not vary much across years, highest in 2001 (35.8%) and lowest in 2006 (29.7%). It is highest in the wholesale and retail trade industries (51.2%), followed by the transportation and utilities (48.6%), manufacturing (33.1%), mining and construction (32.7%), and service (20%) industries. Table II presents summary statistics for the sample suppliers (Panel A) and their publicly listed customers (Panel B). Several observations are noteworthy. First, customers are much larger than the suppliers. The average total asset for publicly traded customers is $46.3 billion, while the corresponding values for hedging and non-hedging suppliers are $6.8 billion and $2.4 billion, respectively. Second, compared to non-hedging suppliers, hedging suppliers are significantly larger and have higher leverage, higher profitability (return on assets), and more tangible assets (net property, plant, and equipment / total assets). They also hold smaller cash (sum of cash and short-term investments / total assets) and have a lower market-to-book ratio, lower R&D intensity, and lower industry competition as measured by industry Herfindahl index. Third, compared to non-hedging suppliers, hedging suppliers have longer durations of the major customer relationships (mean durations of 4.86 years versus 4.07 years) and depend more on major customers for their sales, as measured by the ratio of sales to the major customer to total assets (mean ratios of 20.71% versus 19.56%). Finally, the CEOs of hedging suppliers receive higher total compensation than the CEOs of non-hedging suppliers, 11

14 and their compensation package consists of more of stock compensation but less of option and cash compensation. Detailed definitions of each variable are provided in the Appendix. III. Impact of Customer Leverage on Supplier Hedging A. Major Customer Relationships and Suppler Hedging To investigate whether a firm s hedging decision is affected by its relationships with major customers, we estimate a logit regression of the supplier s hedging policy in which the dependent variable is an indicator that takes the value of one if the supplier uses an interest rate hedging and zero otherwise. Following Nance et al. (1993) and Purnanandam (2007), we control for firm size, book leverage, cash holdings, profitability, market-to-book ratio, R&D intensity, tangibility, and industry competition. According to Tufano (1996) and Knopf, Nam and Thornton (2002), we control for the structure of CEO compensation, that is, CEO total compensation, CEO option compensation ratio, CEO stock compensation ratio, and CEO cash compensation ratio (In the robustness tests, we replace the CEO s compensation variables with the CEO s risk taking incentive measures, pay-performance sensitivity (delta) and the sensitivity of CEO wealth to stock volatility (vega)). Finally, we control for year and industry fixed effects. We estimate p-values using robust standard errors to adjust for heteroskedasticity (White 1980) and cluster the errors by supplier firms 12. The results are reported in Table III. In column (1), we include an indicator for having a major customer as our key explanatory variable. We find that the coefficient estimate on this indicator is not significant. 12 The sample used in this analysis consists of all firms (12,052 firm-year observations) covered in S&P ExecuComp, CRSP, and Compustat (financial industries are excluded) from 1999 to

15 In column (2), we decompose the major customers into major public, private, and government customers and find that only the coefficient estimate on the major public customer indicator is positive and significant at the 10% level. This result suggests that firms with a major public customer are more likely to use interest rate derivatives for hedging than those without a major customer. Its average marginal effect has a coefficient estimate of 0.03, suggesting that the probability of using derivatives to hedge interest rate risk is approximately 3.0% higher for firms with a major public customer than for firms without a major customer. Given the unconditional probability of interest rate hedging of32.6% in our sample, this effect seems to be economically significant. In column (3), we further decompose the major public customers into high- and lowleveraged major public customers according to the sample median of the major public customer leverage. If the firm has multiple major public customers, we calculate the major public customer leverage as the weighted average book leverage of all major public customers based on their purchase amounts from the supplier. We find that the coefficient estimate on the high-leveraged major public customer indicator is positive and significant at the 1% level while that on the low-leveraged major public customer indicator is negative and insignificant. The marginal effect of the high-leveraged major public customer indicator has a coefficient estimate of 0.073, suggesting that all else being equal, the probability of using derivatives to hedge interest rate risk is about 7.3% higher for firms whose major public customer has higher leverage than for firms without a major public customer. This result is consistent with our hypothesis that if suppliers are dependent on major customers for their sales, the customers financial reliability is an important consideration for their risk management policy. B. Relationship-specific Characteristics and Supplier Hedging 13

16 We next examine the determinants of a supplier firm s hedging decision in greater detail by focusing on customer leverage and relationship-specific characteristics as key variables of interest. Since these variables are only available for public customers, our analysis in this subsection is conducted with a small sample of 641 suppliers (4,154 supplier-customer-year observations) that have at least one major public customer. There are 401 unique public firms that serve as the major customer of these 641 suppliers. The results are reported in Panel A of Table IV. The dependent variable is an indicator that takes the value of one if a supplier firm engages in interest rate derivate hedging in a given year and zero otherwise. Our key independent variable of interest is customer leverage. The regressions control for other customer characteristics including firm size, profitability, market-to-book ratio, R&D intensity, and tangibility to mitigate the concern that customer leverage measures firm characteristics other than its financial distress risk, we also include other variables used in Table III as additional controls. Finally, we control for two relationship-strength variables, that is, the length of the supplier-customer relationship and the supplier s sales dependence on its major customers (supplier s sales to major customers / supplier s total assets). Column (1) reports the results based on the full sample of 4,154 supplier-customer-year observations. We find that the probability that the supplier uses an interest rate hedging policy increases significantly with customer leverage. A one-standard-deviation increase in customer leverage is associated with a 2.82% increase in the likelihood of the supplier engaging in hedging policy. In columns (2) and (3), we decompose the sample into two subgroups according to the sample median (12.06%) of the supplier s sales dependence on its major customers. Consistent with our hypothesis that suppliers facing high switching costs are more concerned about their customers financial health, we find that the positive and significant relation 14

17 between customer leverage and the likelihood of suppliers employing interest rate derivative hedging exists only in the high-sale dependence group. Economically, we find that for high- (low-) sale dependence group, a one-standard-deviation increase in customer leverage is associated with a 4.62% (1.26%) increase in the likelihood of suppliers employing interest rate derivative hedging. In columns (4) and (5), we classify the sample into two subgroups according to whether the supplier operates in durable goods industries (SIC codes 34-39). The suppliers in durable goods industries are more likely to provide unique products to the customers, thus facing higher switching costs when the customers fail. We therefore expect that suppliers in durable goods industries are more concerned about the customers financial distress risk and thus the effect of customer leverage on the likelihood of suppliers employing interest rate derivative hedging is stronger when suppliers operate in durable goods industries. Consistent with this prediction, we find that the coefficient estimate on customer leverage is positive and significant only for the subgroup of suppliers in durable goods industries. In columns (6) and (7), we use the extent of the supplier s R&D intensity as the measure of switching costs and divide the sample into two subgroups according to the sample median (0.031) of the supplier s R&D intensity. We find that the coefficient estimate on customer leverage is positive and significant only for the subsample of high R&D intensity suppliers, further supporting our hypothesis that suppliers facing high switching costs are more concerned about their customers financial health. In Panel B of Table IV, to avoid a potential cross-sectional dependence problem in estimating the standard errors when the supplier has multiple public customers, in estimating the regressions, we use only customer-supplier pairs in which the customer purchases the largest amount of raw materials and intermediate products from the supplier in a given year. We find results that are qualitatively similar to those presented in Panel A of Table IV. The 15

18 largest customer leverage has a significant positive impact on suppliers hedging and this impact is more pronounced when the suppliers have high sales dependence on their largest customers, when they operate in durable goods industries, or when their R&D intensity is high. C. Correcting endogeneity bias: Suppliers hedging decision after new customer relationship establishment Thus far, we have not explicitly considered endogeneity bias caused by omitted variables or reverse causality. To address the potential endogeneity problems discussed in the introduction section, in this subsection, we examine how the suppliers choose their hedging policies after they establish the relationships with the new major customer (major customer that makes the first time purchase from the supplier). 13 Since the new customer s leverage before relationship establishment is unlikely to be affected by the supplier s future hedging policy after relationship establishment, this test should mitigate the potential reverse causality problem that exists in our previous tests. Moreover, we expect this test to mitigate the omitted variable problem since we use the change in the supplier s hedging policy after relationship establishment. The results are reported in Table V. The dependent variable is an indicator that takes the value of one if a non-hedging supplier adopts interest rate hedging policy during the first three years of the new customer-supplier relationship establishment and zero otherwise. Our key variable of interest is the leverage ratio of the new customer before relationship establishment. Our control variables are the same as those used in Table IV except that we drop the length of the supplier-customer relationship. 13 The sample used in this subsection consists of 520 supplier-customer-year observations (293 suppliers and 232 new customers) from 1999 to We end the sample period in 2005 because we examine the changes in a supplier s hedging policy during the first three years of the new customer-supplier relationship establishment. 16

19 Column (1) reports the results based on the full sample. We find that a non-hedging supplier is more likely to adopt a hedging policy when its new customer has high leverage prior to the relationship establishment. A one-standard-deviation increase in the new customer s leverage is associated with a 3.29% increase in the likelihood of the supplier adopting hedging policy within next three years. Columns (2) through (7), show that this positive relation between customer leverage and supplier hedging policy is more pronounced when non-hedging suppliers have high sale dependence on their new customers (marginal effect 6.6%), when they operate in durable goods industries (marginal effect 4.98%), and when they make high relationship-specific investments (marginal effect 2.96%). In untabulated tests, we also find that after relationship establishment, suppliers that have adopted interest rate hedging are less likely to abandon their hedging policy if new customers have high leverage. Overall, these results show that the likelihood of suppliers engaging in hedging activities increases with customers leverage are robust to controlling for endogeneity concerns, providing further support to our hypotheses. IV. Impact of Suppliers Hedging Policy on the Strength of Major Customer Relationships In this section, we analyze the impact of the supplier s hedging policy on the strength of customer-supplier relationships. With a sample of 851 customer-supplier relationships (477 suppliers and 318 customers) that were newly established between 1999 and 2007, 14 we examine whether supplier hedging helps increase relationship strength, as measured by the 14 We do not include firms in 2008 since we measure the change in relationship strength one year after the supplier s hedging policy is observed. 17

20 duration of customer-supplier relationships and the increase in a supplier s sales dependence on the new customer. Specifically, we estimate the duration of customer-supplier relationships using COX model at the time when the relationship is newly established, and measure the change in a supplier s sales dependence on the new customer as an indicator, which takes the value of one if the change of supplier s sales dependence on the new customer is higher than the sample median one year after the relationship establishment (we also require that the supplier-customer relationship would still exist in the next year). We find that 851 customersupplier relationships on average last for 3.34 years and the average change of the supplier s sales dependence on new customer is around zero, median change is around %. Table VI shows the results. In columns (1) through (3), we apply Cox model to estimate the duration of the supplier-customer relationships and in columns (4) through (6), the dependent variable is the indicator for the increase in a supplier s sales dependence to the new customer. Our key variable of interest is an indicator that takes the value of one if the supplier uses interest rate hedging and zero otherwise. Columns (1) and (4) report the results using the full sample and the other columns report the results using the subsamples classified according to the sample median of customer leverage. We find that the supplier s hedging indicator is strongly positively related to the relationship duration at the 5% level (column (1)). We also find that the supplier s hedging indicator is strongly positively related to the likelihood of a higher change in the supplier s sale dependence on a new customer at the 5% level (column (4)). Compared to non-hedging suppliers, all else being equal, hedging suppliers experience a 13% higher probability that their change of sales dependence on the new customer is higher than sample median. These results suggest that the effect of supplier hedging on relationship strength is both statistically and economically significant. 18

21 Moreover, we find that the positive effect of supplier hedging on relationship strength is particularly pronounced when the customer has high leverage. The positive relation between supplier hedging and the likelihood of an increase in the supplier s sale dependence on the new customer is also stronger in the high-leverage customer subsample than that in the lowleverage customer subsample (marginal effects of 16.9% compared with 2.0%). Overall, these results provide strong support to our hypothesis that supplier hedging helps increase the durability of major customer relationships and the supplier s sales dependence on the major customer, especially when its major customer has high leverage. V. Abnormal Returns for Suppliers around Customer Bond Offer and Rating Downgrade Announcements A. Univariate analysis In this section, we examine how supplier hedging helps mitigate the negative spillover effect of customers financial distress risk on supplier value. We use bond offer announcements by major customers and the announcements of customer credit downgrades by rating agencies as the news that triggers the negative spillover effect since these events are likely to convey information about customers deteriorating in financial health. To the extent that hedging makes the suppliers less vulnerable to customers financial distress, we expect that the abnormal returns for suppliers around these announcement dates are less negative when the suppliers already use interest rate hedging, particularly when their sales dependence on the large customers is higher, when their relationship-specific investments are larger, or when they operate in durable goods industries. We obtain information on customer bond offerings from Thompson s SDC New Issues database for the period and customer bond rating downgrades from Mergent Fixed 19

22 Investment Securities Database (FISD) for the same period. Following Johnson et al. (2012), we define large customers as those representing 10% or more of a supplier s total sales. When the supplier has multiple large customers, the customer purchasing the largest amount is identified as the large customer. For each bond issue, we use its filing date as the announcement date. If the bond has only the issue date (i.e., private placement), we use its issue date as the announcement date. 15 Following Jorion, Liu and Shi (2005), we use the effective date of the bond rating downgrade reported in FISD as the rating downgrade announcement date. In addition, we search Factiva for major confounding corporate news (earnings announcements, annual report filings, quarterly report filings, CEO turnover, and mergers and acquisitions) within one trading day before and after the announcement and exclude observations associated with such news. These sample criteria yield final samples of 168 bond offerings (including 96 private placements under rule 144A) and 202 bond rating downgrades of the large customers. The abnormal returns are calculated using a market model. We obtain our estimates of the market model by using one year (255 trading days) of return data, beginning 301 days before and ending 46 days before the announcement date. We use as the market portfolio the CRSP equally weighted index. We aggregate the daily abnormal returns to obtain the cumulative abnormal return (CAR) from one day before the announcement date to one day after the announcement date. Panel A of Table VII reports the mean and median CARs (-1, 1) for supplier firms around the bond offer announcements by large customers. We find that the supplier firms on average earn negative but statistically insignificant CARs (-1, 1) of approximately 0.14% 15 To check the accuracy of using the filing date as the announcement date, we randomly select 20 bonds and search the dates that their issues are first disclosed in Factiva and find that approximately 80% of the news announcement dates fall within one day before and after the filing date. Carayannopoulos and Nayak (2010) use the issue date as the announcement date for private placements under rule 144a. 20

23 around the customer bond offer announcement date. 16 The average CAR (-1, 1) for the hedging suppliers is 0.59%, which is significant at the 5% level. The corresponding CAR for the non-hedging suppliers is -0.57%, which is significant at the 5% level. The differences in mean CARs (-1, 1) between the two groups is statistically significant at the 1% level. The median shows a similar pattern. These results suggest that the stock market perceives the increase in the customers financial distress risk as a value-decreasing event only for nonhedging suppliers, supporting our hypothesis that hedging makes the suppliers less vulnerable to customers financial distress risk. Panel B of Table VII reports the mean and median CARs (-1, 1) for suppliers around the announcements of customer credit downgrades by rating agencies. The mean CAR (-1, +1) for the full sample of suppliers is a significant -0.43%. In untabulated tests, we find that the mean CAR (-1, +1) for the customers that experience rating downgrades is a significant %. These results are consistent with those of Hertzela et al. (2007) and Kolay, Lemmon and Tashjian (2012) who document the negative contagion effect of customers distress risk on supplier value. Dividing suppliers into hedging and non-hedging suppliers, we find that the negative CAR (-1, +1) for the full sample is entirely driven by non-hedging suppliers. Thus, the negative spillover effects for suppliers induced by the increase in customers financial distress risk exists only when the suppliers do not engage in interest rate risk hedging, further supporting our hypothesis. B. Multivariate analysis To better understand the cross-sectional variations in abnormal returns for supplier firms around bond offer announcements by large customers (announcements of customer credit 16 The mean CAR (-1,+1) for the large customers is an insignificant %. In comparison, Eckbo (1986) finds that the mean CAR (-1, 0) for his sample of straight bond offerings is an insignificant -0.06%. 21

24 downgrades by rating agencies), we now turn to the multivariate regressions in which the dependent variable is the supplier CAR (-1, +1) around these announcement dates. Our key variable of interest is the supplier s hedging indicator that takes the value of one if the supplier uses derivatives instruments to hedge interest rate risk and zero otherwise. We control for several bond offerings or credit rating downgrades related variables following Eckbo (1986), Carayannopoulos and Nayak (2010) and Jorion, Liu and Shi (2005), which may affect the announcement returns. Table VIII presents the estimates from multivariate regressions. All regressions are estimated using ordinary least squares (OLS). The dependent variable in columns (1) through (4) is the supplier CARs (-1, +1) around the bond offer announcements by large customers. In column (1), we find that the coefficient estimate on the supplier s hedging indicator is with a p-value of This result indicates that all else constant, the hedging suppliers realize a 1.1% higher abnormal announcement return than the non-hedging suppliers. Therefore, the effect of hedging on suppler returns seems to be both statistically and economically significant, suggesting that hedging plays an important role in mitigating adverse vertical spillover effect of customers deteriorating financial health on supplier value. To more closely examine the effect of hedging on supplier value, in next three regressions, we include interaction terms between the supplier s hedging indicator and the indicators that measure the extent of switching costs borne by suppliers (indicator for the supplier s high sales dependence on the large customer, indicator for whether the supplier operates in the durable goods industry, and indicator for the supplier s high R&D intensity). We find that the coefficient estimates on the interaction terms between the supplier s hedging indicator and the indicator for durable goods industries and between the supplier s hedging indicator and the indicator for high R&D intensity are positive and significant (columns (3) and (4)). 22

25 Thus, hedging suppliers experience smaller wealth losses than do non-hedging suppliers, particularly when they operate in durable goods industries in which post-sale guarantees are important (Titman (1984)) or when they make larger relationship-specific investments, suggesting that hedging benefits are particularly larger for suppliers that face greater switching costs. These results support our hypothesis. In columns (5) through (8), we reestimate the regressions in columns (1) through (4) by using as the dependent variable the supplier CARs (-1,+1) around the announcements of customer credit downgrades by rating agencies. The results echo those in columns (1) through (4). The only difference is that the coefficient estimates on the interaction term between the supplier hedging indicator and the indicator for high sales dependence now becomes positive and significant at the 10% level (column (6)). Overall, these results show that customers deteriorating financial health has significantly smaller negative spillover effects on the value of hedging suppliers than on the value of nonhedging suppliers, particularly when hedging suppliers depend on the major customers for a significant proportion of their revenues, when they produce relationship-specific products that cannot easily be redeployed to other customers or other uses, or when their relationshipspecific investments are larger. To the extent that these suppliers face higher switching costs, our results suggest that the value-enhancing role of hedging in customer-supplier relationships is particularly important when these relationships involve greater asset specificity and product customization. VI. Robustness Tests To check the robustness of our key results, we conduct several additional tests. Below, we briefly summarize the results of these untabulated tests. 23