International Review of Economics and Finance 22 (2012) 208221

Contents lists available at SciVerse ScienceDirect

International Review of Economics and Finance

journal homepage: www.elsevier.com/locate/iref

Capital structure and corporate governance quality: Evidence from the Institutional Shareholder Services (ISS)
Pornsit Jiraporn a, b, 1, Jang-Chul Kim c, 2, Young Sang Kim c,, Pattanaporn Kitsabunnarat d, 3
a b c d

Great Valley School of Graduate Professional Studies, Pennsylvania State University, Malvern, PA 19355, United States Thammasat Business School, Thammasat University, Bangkok, Thailand Haile/US Bank College of Business, Northern Kentucky University, Highland Heights, KY 41099, United States SASIN Graduate Institute of Business Administration, Chulalongkorn University, Bangkok, Thailand

a r t i c l e

i n f o

a b s t r a c t
Grounded in agency theory, this study explores how capital structure is influenced by aggregate corporate governance quality. We measure governance quality using broad-based comprehensive governance metrics provided by the Institutional Shareholder Services (ISS). The empirical evidence reveals a robust inverse association between leverage and governance quality. Firms with poor governance are significantly more leveraged. It appears that leverage substitutes for corporate governance in alleviating agency conflicts. Further, we utilize empirical methods that control for endogeneity and show that poor governance quality likely brings about, and does not merely reflect, higher leverage. Our results are important as they show that the overall quality of corporate governance has a material impact on critical corporate decisions such as capital structure choices. 2011 Elsevier Inc. All rights reserved.

Article history: Received 18 March 2010 Received in revised form 18 October 2011 Accepted 31 October 2011 Available online 7 November 2011 JEL classification: G32 G34 Keywords: Capital structure Corporate governance Agency costs Leverage

1. Introduction The literature in capital structure began with the seminal work by Modigliani and Miller (1958) on the irrelevance of capital structure. Since then, capital structure continues to be a topic of interest in financial economics and has produced an enormous volume of research. Parsons and Titman (2008) offer a thorough and timely review of the literature in this area. Several theories have been advanced to explain capital structure decisions. One theory that has garnered strong empirical support is agency theory. Agency theory posits that capital structure is determined by agency costs due to conflicts of interest. The literature in this area has been built on the early work by Fama and Miller (1972) and Jensen and Meckling (1976). Motivated by agency theory, this study is related to agency costs as an explanation for capital structure decisions. Specifically, this paper explores the association between capital structure and corporate governance quality. Corporate governance exists to provide checks and balances between shareholders and management and thus to mitigate agency problems. Hence, firms with better governance quality should suffer less agency conflicts.

Corresponding author. Tel.: + 1 859 572 5160; fax: + 1 859 572 6177. E-mail addresses: pxj11@psu.edu (P. Jiraporn), kimj1@nku.edu (J.-C. Kim), kimy1@nku.edu (Y.S. Kim), pattanaporn.kitsabunnarat@sasin.edu (P. Kitsabunnarat). 1 Tel.: + 1 610 725 5342. 2 Tel.: + 1 859 572 1486. 3 Tel.: + 66 2 218 4034. 1059-0560/$ see front matter 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.iref.2011.10.014

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

209

Like corporate governance, leverage has been argued to alleviate agency costs as well. Agency problems can be mitigated by leverage in several ways. First, one way to reduce agency conflicts is to cause managers to increase their ownership in the firm (Jensen & Meckling, 1976). By increasing the use of debt financing, effectively displacing equity capital, and firms shrink the equity base, thereby increasing the percentage of equity owned by management. Second, the use of debt increases the probability of bankruptcy. This additional risk may further motivate managers to decrease their consumption of perks and increase their efficiency (Grossman & Hart, 1982). Finally, the obligation of interest payments resulting from the use of debt helps resolve the free cash flow problem (Jensen, 1986). Because leverage imposes constraints on managerial discretion, agency theory suggests that managers may be motivated to adopt sub-optimal leverage that does not maximize shareholders' wealth. The extent to which managers can take on suboptimal leverage should hinge critically on the strength of corporate governance as corporate governance is specifically designed to combat agency conflicts. Hence, we argue that there ought to be a significant relation between leverage and corporate governance quality. This is the central premise of our study. To measure corporate governance quality, we employ the governance standards provided by the Institutional Shareholder Services (ISS). The ISS governance standards include 51 factors encompassing eight corporate governance categories: audit, board of directors, charter/bylaws, director education, executive and director compensation, ownership, progressive practices, and state of incorporation. The ISS governance standards are the most all-inclusive data on corporate governance ever collected. Governance matrices constructed from the ISS governance standards are especially interesting because prior literature shows that they are significantly related to several crucial corporate outcomes. For instance, using the ISS governance data to gauge the strength of corporate governance, Brown and Caylor (2006) find that firms with better governance quality are more profitable and more valuable (higher Tobin's q). Their results imply that firms with better governance quality experience lower agency costs and, hence, exhibit better performance and higher firm value. In addition, Chung et al. (2010) show that firms with better governance quality have narrower spreads, higher market quality index, smaller price impact of trades, and lower probability of information-based trading. 4 Also, Charoenwong et al. (2011) investigate the relationship between the quality of governance structure and adverse selection component with stocks listed on Singapore Exchange. They show that corporate governance has an inverse relationship with the adverse selection components of bidask spreads Our empirical evidence shows a robust inverse relationship between leverage use and governance quality. In other words, firms where corporate governance is weaker are found to be significantly more leveraged. We argue that, due to the role of debt in mitigating agency costs, higher leverage substitutes for weaker governance mechanisms in alleviating agency conflicts. The negative association documented in this study is consistent with the substitution hypothesis. First advanced by La Porta et al. (2000), the substitution hypothesis posits that firms with weak governance, in an attempt to raise capital on attractive terms, need to establish a reputation for not expropriating wealth from shareholders. One way to do so is to carry more debt as fixed interest payments reduce what is left for expropriation; the weaker the firm's governance, the stronger the need for the reputation mechanism, and, thus, the more debt the firm should carry. The substitution effect we document is not only statistically significant but also economically meaningful. As firms improve their governance index from the 25th to 75th percentile, their leverage would decrease by as much as 12.88%. Furthermore, it can be argued that leverage and corporate governance are endogenously determined. We relate corporate governance quality in an earlier time period to subsequent leverage to minimize endogeneity and find that the direction of causality is much more likely to run from governance quality to leverage than vice versa. In addition, using the two-stage least squares (2SLS) approach, which is less vulnerable to endogeneity, we find that the inverse association remains robust. Finally, our analysis based on year-to-year changes in corporate governance quality also reveals that an increase in governance quality is associated with a reduction in leverage. Overall, the evidence suggests that endogeneity does not appear to pose a serious threat in our sample. We also compare our governance quality metrics with those employed in prior studies. In particular, we examine both the Governance Index developed by Gompers et al. (2003) and the Entrenchment Index invented by Bebchuk et al. (2009). We argue that our governance metrics are broader and capture far more information beyond what is impounded in the Governance Index or the Entrenchment Index. The empirical evidence strongly reinforces our argument, showing that our ISS-based governance metrics can explain variation in capital structure far better than the other two indexes. The results of this study enrich the literature in several ways. First, the literature in corporate governance benefits because we show that governance quality is a significant determinant of capital structure. 5 Prior studies have examined the impact of Gompers et al.'s (2003) index on capital structure decisions (Jiraporn & Gleason, 2007; John & Litov, 2009). Our ISS governance metrics, however, provide much broader measures of governance quality than Gompers et al.'s (2003) Governance Index. Moreover, a recent study by Bebchuk et al. (2010) finds that the effect of Gompers et al.'s index on stock returns does not exist after the year 2000. They suggest that market participants learn over time about the difference between well-governed and poorly

4 A number of recent studies also make use of the governance data provided by the ISS. For example, Aggarwal et al. (2009) employ the ISS governance data to compare governance quality between U.S. and international rms. They nd that minority shareholders benet as governance quality improves. 5 A substantial number of studies seek to understand the impact of corporate governance on various corporate outcomes such as rm value (Brown & Caylor, 2006; Cremers & Nair, 2005; Gompers et al., 2003; Park et al., 2008), costs of debt nancing (Cremers et al., 2007), corporate diversication (Jiraporn et al., 2006), cash holding (Dittmar & Mahrt-Smith, 2007), debt maturity structure (Harford et al., 2008), CEO compensation (Fahlenbrach, 2009), institutional ownership (Ciceksever et al., 2006), and liquidity (Chung et al., 2010).

210

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

governed firms. Thus, abnormal returns cannot be generated in the more recent years. This study highlights the need to reexamine the role of governance in the more recent period, which is precisely what we do. Second, we contribute to the literature on capital structure. 6 The results in Bebchuk et al. (2010) raise the question of whether leverage choices are affected by governance quality after 2000. Prior studies in this area are based on samples largely from the 1990s. Our results demonstrate that, even after 2000, capital structure choices continue to be influenced by corporate governance quality. Finally, our study makes a contribution to the literature that investigates the usefulness and effectiveness of the ISS recommendations. This literature is broad, spanning several areas such as accounting, finance, and law (Aggarwal et al., 2011; Brown & Caylor, 2006; Choi et al., 2008; Chung et al., 2010; Cotter et al., 2009). 7 Our results show that the governance features promoted by the Institutional Shareholder Services do influence capital structure decisions. The paper is organized as follows. Section 2 discusses the motivation, and develops two competing hypotheses. Section 3 shows the sample selection procedure and discusses the relevant data. Section 4 discusses the methods and displays the empirical results and the final section concludes. 2. Motivation and hypothesis development 2.1. Motivation The finance literature is replete with studies that attempt to explain the determinants of capital structure. Several hypotheses have been advanced in the past couple of decades, for instance, the signaling hypothesis (Leland & Pyle, 1977; Ross, 1977), the pecking order hypothesis (Myers, 1984), and agency theory (Jensen, 1986). Agency theory argues that capital structure is determined by agency costs, which arise from conflicts of interests. Motivated by agency theory, this study is related to the agency costs as an explanation of capital structure; using the newly developed broad-based governance metrics, we investigate how capital structure is influenced by corporate governance quality. Capital structure decisions are shown to be affected by certain governance mechanisms such as executive compensation (Berger et al., 1997), anti-takeover provisions (Jiraporn & Gleason, 2007; John & Litov, 2009), board structure (Harford et al., 2008), and anti-takeover statutes (Garvey & Hanka, 1999). Nevertheless, our study is the first to investigate the impact of aggregate governance quality on capital structure, using the broadest metrics available in the literature. It is imperative to use broad measures of corporate governance as governance attributes likely interact with one another (Agrawal & Knoeber, 1996; Bowen et al., 2008). 2.2. Capital structure and corporate governance quality Agency theory suggests that there are several ways in which debt can help mitigate agency conflicts between shareholders and managers. Holding constant the manager's absolute investment in the firm, increases in the fraction of the firm financed by debt increase the manager's share of the equity, thereby bringing the manager's and the shareholders' interests into better alignment. Moreover, as argued by Jensen (1986), since debt commits the firm to pay out cash, it reduces the amount of free cash available to managers to engage in excessive perquisite consumption. Corporate governance is put in place specifically to ensure that managers act in the interest of shareholders. Therefore, corporate governance is designed to minimize agency conflicts. 8 Due to agency costs, managers may adopt leverage choices that enhance their own private benefits rather than maximize shareholder wealth. The degree to which managers can deviate from the optimal leverage depends critically on the strength of corporate governance in the firm. As a result, we hypothesize that a relation should exist between leverage and corporate governance quality. It is not entirely clear, however, what the association should be between leverage and governance quality, we advance two competing hypotheses that may explain the influence of governance quality on capital structure. 2.3. The Outcome hypothesis This view argues that capital structure is determined as an outcome of corporate governance quality. Firms with low governance quality suffer more severe agency problems. Managers of these firms are better able to exploit shareholders, placing their private benefits ahead of those of the shareholders. As argued by agency theory and shown by empirical evidence, debt plays a
6 Two recent studies by Jiraporn and Gleason (2007), and John and Litov (2009) also examine the impact of corporate governance on capital structure. However, they employ only the Governance Index developed by Gompers et al. (2003) to represent corporate governance. The scope of our study is much more exhaustive as our study encompasses many facets of corporate governance, including boards, executive compensation, audit committees, director education, executive ownership, and state of incorporation. Each governance category is important by itself and has attracted a great deal of attention in the literature (for instance, for executive compensation, see Murphy, 1998; Coughlan & Schmidt, 1985; Jensen & Murphy, 1990a, 1990b, for state of incorporation, see Daines, 2001; Lipton & Rowe, 2002; Gilson, 2002; Macey, 1998; Sitkoff, 2002, for executive ownership, see Demsetz & Lehn, 1985; Morck et al., 1988; McConnell & Servaes, 1990; Hermalin & Weisbach, 1991). Because various governance mechanisms can complement or substitute for one another, it is critical to examine them collectively (Agrawal & Knoeber, 1996; Bowen et al., 2008). 7 The ISS governance recommendations are not merely for informational purpose. Several institutional shareholders have actively encouraged their rms to adopt or repeal a number of governance provisions or practices that are included in the ISS recommendations. So, these ISS recommendations have led to changes in corporate governance in a number of rms. 8 The impact of corporate governance and ownership structure on the agency conict has been examined in several prior studies such as Ang et al. (2002), Singh and Davidson (2003), McKnight and Weir (2009), and Henry (2010).

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

211

role in controlling agency costs, making it more difficult for opportunistic managers to misbehave. Managers are thus inherently likely to carry debt at a sub-optimal level because they do not want to impose additional constraints on themselves in the form of fixed interest payments or be deprived of the free cash flow that they have control over. Therefore, this view predicts that poor governance quality is associated with low leverage. In other words, there is a positive relationship between governance quality and leverage. 2.4. The Substitution hypothesis This perspective contends that leverage acts as a substitute for corporate governance. Debt helps alleviate agency costs. Likewise, corporate governance is installed to mitigate agency conflicts. Thus, debt and governance play the same role and may substitute for each other. In firms with weak governance, the need for debt to act as a tool for controlling agency costs may be greater than in firms with strong governance. Hence, firms with poor governance quality should be more leveraged. There is also another rationale for the substitution hypothesis. This argument relies critically on the need for firms to raise money in the external capital markets, at least occasionally. To be able to raise external funds on attractive terms, a firm must establish a reputation for moderation in expropriating shareholders. One way to establish such a reputation is by carrying debt and making interest payments, which reduces what is left for expropriation. 9 A reputation for good treatment of shareholders is worth the most for firms with weak corporate governance. As a result, the need for debt to establish a reputation is the greatest for such firms. By contrast, for firms where governance quality is high, the need for a reputation mechanism is weaker, and, thus, so is the need for leverage. This view, therefore, posits that, all else equal, leverage should be higher in firms with weaker governance quality. In other words, an inverse relationship should be observed. 3. Sample selection and data 3.1. Sample selection The original sample includes all firms reported by the Institutional Shareholder Services (ISS) from 2001 to 2004 (16,013 firm year observations). ISS collects data on governance standards for a large number of firms (2400+ firms in 2001 and 5000+ firms in 2004). Then, we narrow down our sample by eliminating firms whose financial and accounting data do not exist on COMPUSTAT. Financial and utility firms are excluded because they are regulated and because their leverage cannot be interpreted in the same manner as for industrial firms. 10 The data for firm characteristics including credit ratings are obtained from COMPUSTAT. The final sample consists of 7557 firmyear observations from 2001 to 2004. 3.2. Corporate governance metrics To gauge corporate governance quality, we employ year-end data on governance standards provided by the Institutional Shareholder Services (ISS) in a manner comparable to Brown and Caylor (2006) and Chung et al. (2010). The scope of the governance data is very broad, encompassing fifty-one governance standards in eight categories as defined by ISS. The eight categories include audit issues, board structure and composition, other charter and bylaw provisions, director education, executive and director compensation, director and officer ownership, progressive practices, and laws of the state of incorporation related to takeover defenses. 11 The governance standards reported by ISS capture various dimensions of corporate governance. For instance, the Audit category includes four governance standards associated with auditor independence (composition of the audit committee, ratification at the annual meeting, consulting fees paid to auditors, and the company's policies on auditor rotation). The Charter category consists of seven governance standards related to provisions for delaying or impeding takeovers. The Board category is composed of seventeen governance standards related to the composition and other characteristics of the board. Finally, two categories, Director education and State, consist of a single governance factor. We employ two metrics to gauge the aggregate quality of corporate governance. First, similar to Brown and Caylor (2006) and Chung et al. (2010), we construct an index for each firm by assigning one point for each governance standard that is satisfied. We label this index Gov-score. We ascertain whether a specific governance standard is met using the minimum standard provided in the ISS Corporate Governance: Best Practices User Guide and Glossary (2003). Second, we employ the metric computed by ISS to measure governance quality. We refer to this metric as ISS-score. 12 Although constructed based on the same governance standards, ISS-score is different from Gov-score because ISS-score allows interaction terms that occur in combination with others. For
A similar argument is made by La Porta et al. (2000) and Jiraporn and Ning (2006) on dividend policy and shareholder protection. SIC codes 60006999 and 49004999 respectively. 11 The fty-one governance standards and their eight categories are shown in the Appendix. 12 ISS-score, converted to a relative index, is now publicly available on nancial websites under the registered trademark Corporate Governance Quotient. To compute a Corporate Governance Quotient (CGQ) for each rm, ISS analysts employ publicly available documents and website disclosure to gather data on ftyone different issues in the following four broad rating categories: board of directors, audit, anti-takeover, and compensation/ownership. On the basis of this information and a scoring system developed by an external advisory panel and ISS, they compute a CGQ for each company. While each variable is assessed on an individual basis, some variables are also evaluated in combination under the supposition that corporate governance is improved by the presence of selected combinations of favorable governance provisions.
10 9

212

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

instance, ISS assigns more weight to a firm whose board consists of a majority of independent directors and whose key board panels (audit, nominating, and composition) are all composed of independent directors, than it assigns to each of those standards individually. It is important to note that Gov-score and ISS-score are more comprehensive measures of corporate governance quality than the Governance Index, compiled by Gompers et al. (2003), for several reasons. First, the ISS data are available for a much larger number of firms in more recent years. Second, the ISS data are available annually, rather than biannually. Third, the ISS data are much broader, and still encompass about half of the standards incorporated into the Governance Index by Gompers et al. (2003). The Governance Index is a less comprehensive corporate governance measure because it is based on the existence of antitakeover amendments. 13 Finally, the ISS data include five of the six standards that are identified as most relevant for firm value (Bebchuk et al., 2009). We utilize Gov-score as our primary metric for the quality of corporate governance and use ISS-score simply as a supplemental variable to confirm the results. There are two reasons why we focus on Gov-score rather than ISS score. First, Gov-score has been used in a number of previous studies such as Brown and Caylor (2006). Thus, to make our results comparable to those in prior studies, we use Gov-score as our principal variable. Second, the way Gov-score is constructed is much more transparent than the way ISS-score is computed. ISS does not disclose how it assigns weights to the interaction effects among governance devices, potentially making it more difficult to interpret the results. Throughout this study, we show the results using Gov-score in all empirical tests and discuss ISS-score only in certain selective tests. 3.3. Descriptive statistics Table 1 shows the industry distribution of the sample based on Fama and French classification. We report both market and book leverage ratios along with Gov-score and ISS-score. The business equipment industry has the lowest level of market and book leverage and represents 25.20% of the total sample, while the telecommunication industry has the highest level of leverage. As far as governance scores, the chemicals and allied products industry has the strongest governance as measured by both Govscore and ISS-score. It is noteworthy that the median Gov-score value is 22.89. Based on an indicated maximum of 51, this suggests that the average sample firm has quite poor corporate governance structure based on the ISS standards. This average Govscore, however, is similar to the average in Chung et al. (2010). Table 2 summarizes the descriptive statistics for salient firm characteristics. The average market debt ratio for the sample is 14.31% whereas the average book debt ratio is 19.74%. 14 On average, the sample firms have $5113 million [$663 million median] in assets. The EBIT ratio averages 8.78% [8.81% median]. Tobin's q, which proxies for growth opportunities, averages 1.46. The fixed-assets ratio averages 4.75%. The ratio of non-debt tax shields to total assets averages 4.29%. 15 The average Gov-score is 22.95 [22.0 median], ranging from 8 to 44. The ISS-score averages 56.21 [56.4 median], ranging from 17 to 96.95. We also show summary statistics for each governance category; Board, Ownership, Charter, Audit, State, Compensation, Progressive, and Director education. 4. Empirical results 4.1. Regression analysis To ascertain the impact of aggregate governance quality on leverage, we perform regression analysis, where leverage is the dependent variable. We employ two alternative definitions of leverage, book leverage and market leverage. 16 The independent variables of interest are Gov-score and ISS-score. We include a number of control variables in our empirical analyses based on past empirical research on capital structure. Following Titman and Wessels (1988) and Johnson (1997), we use the logarithm of total assets to proxy for firm size. The composition of the firm's assets has been found to affect capital structure decisions (Mehran, 1992; Titman & Wessels, 1988). Hence, we include the fixed assets ratio in the regression analysis. As in Johnson (1997), the fixed asset ratio is property, plant, and equipment to total assets. We control for growth opportunities by using Tobin's q, which is computed following Chung and Pruitt (1994), because Myers (1977) and Rozeff (1982) show that growth opportunities of a firm are one of the significant determinants of capital structure choices. Profitability may be relevant to capital structure decisions. Myers (1984) suggests that managers have a pecking order in which retained earnings represent the first choice, followed by debt and equity financing. Thus, the pecking order hypothesis would imply a negative relationship between profitability and leverage. We employ the EBIT ratio to control for profitability. DeAngelo and Masulis (1980) contend that non13 The IRRC database provides annual data for the years 1990, 1993, 1995, 1998, 2000, 2002, and 2004. Following Gompers et al. (2003) and Bebchuk et al. (2009), we ll in observations in the missing years by assuming no change in the Governance Index in the interim years. When we drop the missing year observations, the results are not qualitatively different from the reported results. 14 There are many rms that have no forms of debt in our sample (for example, we have 437 observations in year 2003). This sample observation is also consistent with Byoun et al. (2009). They show that on average, about 12% of COMPUSTAT U.S. rms operate with no debt in any given year during the 19712006 periods. In addition, Byoun (2008) shows the average [median] of debt to market ratio in year 2003 is 13.73% [7.97%]. 15 We dene non-debt tax shields as investment tax credits using the sum of depreciation and amortization. 16 The market value of leverage is total debt divided by the market value of assets. The market value of assets is total assets minus the book value of equity minus deferred taxes and investment tax credits plus the market value of common equity plus preferred stock. The book value of leverage is computed as total debt divided by the book value of total assets.

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

213

Table 1 Sample descriptive statistics by Fama-French 12 industries. This table shows the sample descriptive statistics of debt ratio and governance quality by Fama-French 12 industries. Market debt ratio is total debt (data9 + data34) divided by market value. The market value of asset is the total assets minus book value of equity minus deferred taxes and investment tax credit plus market value of common equity (number of shares outstanding times closing price on the fiscal-year's ending date) plus preferred stock liquidating value. Book debt ratio is the ratio of the total debt to book value of asset. We create an index for each firm by awarding one point for each governance standard that is met, and denote this index Gov-score. We determine whether a particular governance standard is met using the minimum standard provided in ISS Corporate Governance: Best Practices User Guide and Glossary (2003). ISS-score is the governance index provided by Institutional Shareholder Service (ISS). The number of firms, percentage of sample, and mean value of variables are reported. Fama-French 12 industries 1.Consumer nondurables 2. Consumer durables 3. Manufacturing 4. Energy oil, gas, and coal 5.Chemicals and allied products 6. Business equipment 7. Telecommunication 9. Shops (whole, retail) 10. Healthcare 12. Other Total N 439 224 1132 362 245 1871 169 1085 793 1105 7425 % of sample 5.91% 3.02% 15.25% 4.88% 3.30% 25.20% 2.28% 14.61% 10.68% 14.88% 100.0% Market debt ratio 15.14% 15.16 18.54 22.25 19.59 6.46 26.68 15.56 7.51 20.62 14.31 Book debt ratio 22.42% 20.73 23.62 28.40 28.05 10.56 35.14 20.50 14.28 26.08 19.74 Gov score 22.80 23.06 23.14 23.06 24.80 22.47 22.94 22.92 22.71 22.97 22.89 ISS score 54.72 57.11 57.29 58.19 62.10 54.84 55.34 55.35 54.64 55.69 55.87

Table 2 Descriptive statistics of the sample. We create an index for each firm by awarding one point for each governance standard that is met, and denote this index Govscore. We determine whether a particular governance standard is met using the minimum standard provided in ISS Corporate Governance: Best Practices User Guide and Glossary (2003). ISS-score is the governance index provided by Institutional Shareholder Service (ISS). Board, Audit, Charter, State, Ownership, Compensation, Progressive, and Director education denote governance score for each of the following categories: board, audit, charter/bylaws, state of incorporation, ownership, executive and director compensation, progressive practices, and director education. Variable Mean Standard deviation 15.61 19.62 17,281 21.52 1.36 5.29 3.44 1.80 1.25 6.41 12.84 2.79 0.83 1.43 1.18 0.18 1.27 1.97 0.10 Percentile Min. Debt to market ratio (%) Debt to book ratio (%) Total assets ($ millions) EBIT ratio (%) Tobin's q Fixed-asset ratio (%) Non-debt tax shields (%) Credit rating Dividend ratio (%) Gov-score ISS-score Board Ownership Charter Audit State Compensation Progressive Director education 14.31 19.74 3118 3.71 1.60 5.11 4.97 1.20 0.57 22.89 55.87 9.18 1.90 2.35 2.02 0.03 6.05 1.42 0.01 0 0 1.665 0.94 0.14 0.17 0.00 0.00 0.00 8.00 17.00 2.00 0.00 0.00 0.00 0.00 2.00 0.00 0.00 25th 0.61 1.46 119.68 1.45 0.80 1.82 2.76 0.00 0.00 18.00 46.68 7.00 1.00 1.00 1.00 0.00 5.00 0.00 0.00 50th 9.71 17.02 434.60 6.69 1.20 3.39 4.18 0.00 0.00 22.00 56.40 9.00 2.00 2.00 2.00 0.00 6.00 0.00 0.00 75th 22.68 31.61 1547.72 12.52 1.91 6.33 6.15 3.00 0.56 28.00 65.60 11.00 2.00 3.00 3.00 0.00 7.00 2.00 0.00 Max. 83.35 99.95 750,507 78.86 8.41 30.10 21.26 7.00 7.15 44.00 96.95 17.00 4.00 6.00 4.00 1.00 10.00 7.00 1.00

debt tax deductions substitute for the tax shield benefits of debt. As a result, firms with greater non-debt tax shields would be expected to have lower levels of debt. We define non-debt tax shields as the ratio of the sum of depreciation and amortization to total assets. Credit ratings affect the ability of the firm to borrow and therefore should be relevant to capital structure decisions. We include credit ratings for long-term debt as a control variable. Cash distribution in the form of dividends likely influences the firm's leverage choices as well. We hence include the dividend payout ratio in our analysis. Frank and Goyal (2009) show that industry leverage is an important determinant of market-based leverage. Therefore, we include industry-median leverage as a control variable. Finally, we add year dummies to capture leverage variation over time. The regression results are reported in Table 3. In Model 1, we use the market leverage ratio as the dependent variable. The independent variable of interest is the logarithm of Gov-score. 17 The coefficient of Gov-score is negative and significant, suggesting that firms with weaker governance adopt higher leverage. In Model 2, we replace Gov-score with ISS-score. Again, the coefficient of ISS-score is negative and significant, corroborating the result based on Gov-score. In Model 3, we control for possible industry effects by including industry dummies. Gov-score continues to exhibit a negative and significant coefficient in Model 3, implying

17

Following Chung et al. (2010), we use the log form to reduce the potential impact of outliers.

214

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

Table 3 Regression results for governance indices. The market value of leverage is total debt divided by the market value of assets. The market value of assets is total assets minus the book value of equity minus deferred taxes and investment tax credits plus the market value of common equity plus preferred stock. The book value of leverage is computed as total debt divided by the book value of total assets. Gov-score is computed by awarding one point for each governance standard that is met for each firm. ISS-score is the governance index provided by the Institutional Shareholder Service (ISS). Log (total-assets) is the logarithm of total assets. Profitability (EBIT ratio) is measured by earnings before interest, and taxes divided by total assets. Growth opportunities (Tobin's q) are measured by Tobin's q as computed by Chung and Pruitt (1994). The fixed asset ratio is measured as the ratio of fixed assets to total assets. Non-debt tax shields are defined as the ratio of the sum of depreciation and amortization to total assets. Credit ratings are calculated by scoring 7 categories of AA, A, BBB, BB, B, C, D and others. The dividend ratio is the ratio of cash dividends to total assets. Year and Industry dummy (Fama-French 12 industries) variables are included. Standard errors are adjusted for 2725 clusters in firms and robust variance estimators. t-statistics are reported in parentheses. (1) Dependent Variable Intercept Log (Gov-score) Log (ISS-score) Log (total assets) EBIT ratio Tobin's q Fixed-asset ratio Non-debt tax shields Credit rating Dividend ratio Year dummies Industry dummies Adjusted R2 0.007 (0.002) 0.010 (0.012) 0.029 (0.001) 0.111 (0.058) 0.300 (0.087) 0.019 (0.002) 1.949 (0.168) Yes No 0.350 Market Leverage 0.195 (0.034) 0.058 (0.013) (2) Market Leverage 0.264 (0.040) (3) Market Leverage 0.375 (0.035) 0.073 (0.013) (4) Book Leverage 0.201 (0.044) 0.061 (0.017) (5) Book Leverage 0.262 (0.050)

0.058 (0.010) 0.007 (0.002) 0.012 (0.012) 0.028 (0.001) 0.106 (0.057) 0.290 (0.086) 0.019 (0.002) 1.936 (0.167) Yes No 0.353

0.008 (0.002) 0.003 (0.012) 0.029 (0.001) 0.104 (0.065) 0.227 (0.083) 0.019 (0.002) 2.041 (0.178) Yes Yes 0.317

0.006 (0.003) 0.026 (0.017) 0.013 (0.002) 0.101 (0.069) 0.430 (0.107) 0.031 (0.003) 1.664 (0.251) Yes No 0.279

0.058 (0.013) 0.006 (0.002) 0.027 (0.017) 0.013 (0.002) 0.092 (0.069) 0.420 (0.106) 0.031 (0.003) 1.646 (0.250) Yes No 0.279

Statistically significant at the 10% level. Statistically significant at the 5% level. Statistically significant at the 1% level.

that our results are robust to industry variation. In Model 4 and Model 5, we utilize book leverage instead of market leverage. Both Gov-score and ISS-score continue to show negative and significant coefficients. In summary, our regression results are consistent with the substitution hypothesis. As leverage can function as a device for controlling agency costs, firms with better governance (hence, less agency costs) do not need leverage as much as those with poor governance. This is why there is an inverse relation between governance quality and leverage. Our results are robust to alternative measures of governance quality (Gov-score and ISS-score) and alternative measures of leverage (book and market leverage).

4.2. Possible endogeneity To further corroborate the results, we explore the issue of endogeneity. First, endogeneity can arise from simultaneity (i.e., both leverage and governance quality may be simultaneously influenced by a third variable that is unobservable). If this is the case, the association documented earlier might be spurious. Following the literature, we employ fixed-effects modeling, which can partially alleviate these problems, as fixed effects control for unobserved attributes that are constant over time. In any event, bias can still result from omitted variables that are non-constant. Table 4 shows the results of the firmyear fixed-effects regressions. In Model 1, the dependent variable is book leverage. Gov-score exhibits a negative and significant coefficient, consistent with our previous findings. In Model 2, we use market leverage as the dependent variable. Again, Gov-score retains a negative and significant coefficient. In Model 3 and Model 4, we replace Gov-score with ISS-score. The results remain similar, i.e. better governance quality is associated with lower leverage. Second, endogeneity may be attributed to reverse causality. So far, it has been assumed that governance quality affects leverage choices. However, it can be argued that the direction of causality might be reverse, i.e. capital structure choices lead to governance quality. This reverse argument is less plausible for several reasons. First, there is no theoretical model in the literature suggesting that leverage decisions lead to changes in governance quality. Second, it is rather difficult for managers to modify the firm's corporate governance, an act that usually requires shareholders' approval and a lengthy period of execution. 18 On
18

Although some of the ISS governance standards require shareholders' approval, others do not.

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

215

Table 4 Fixed-effects regression results for governance indices. The market value of leverage is total debt divided by the market value of total assets. The market value of assets is total assets minus the book value of equity minus deferred taxes and investment tax credits plus the market value of common equity plus preferred stock. The book value of leverage is computed as total debt divided by the book value of total assets. Gov-score is computed by awarding one point for each governance standard that is met for each firm. ISS-score is the governance index provided by the Institutional Shareholder Service (ISS). Log (total-assets) is the logarithm of total assets. Profitability (EBIT ratio) is measured by earnings before interest, and taxes divided by total assets. Growth opportunities (Tobin's q) are measured by Tobin's q as computed by Chung and Pruitt (1994). The fixed asset ratio is measured as the ratio of fixed assets to total assets. Non-debt tax shields are defined as the ratio of the sum of depreciation and amortization to total assets. Credit ratings are calculated by scoring 7 categories of AA, A, BBB, BB, B, C, D and others. The dividend ratio is the ratio of cash dividends to total assets. The industry debt ratio is the industry median market (book) debt ratio based on two-digit SIC code. Standard errors are adjusted for 2725 clusters in firms and robust variance estimators. t-statistics are reported in parentheses. Model 1 Dependent Variable Intercept Log (Gov-score) Log (ISS-score) Log (total assets) EBIT ratio Tobin's q Fixed-asset ratio Non-debt tax shields Credit rating Industry debt ratio Adjusted R2 0.034 (0.005) 0.047 (0.014) 0.003 (0.002) 0.195 (0.039) 0.119 (0.075) 0.011 (0.003) 0.243 (0.046) 0.053 0.042 (0.004) 0.079 (0.011) 0.013 (0.001) 0.236 (0.031) 0.128 (0.060) 0.006 (0.002) 0.269 (0.036) 0.158 Book Leverage 0.066* (0.035) 0.035 (0.005) Model 2 Market Leverage 0.020 (0.028) 0.048 (0.004) Model 3 Book Leverage 0.092** (0.042) Model 4 Market Leverage 0.021 (0.034)

0.026 (0.006) 0.024 (0.005) 0.048 (0.014) 0.004 (0.002) 0.153 (0.039) 0.070 (0.075) 0.010 (0.003) 0.371 (0.041) 0.044

0.029 (0.005) 0.030 (0.004) 0.078 (0.011) 0.015 (0.001) 0.183 (0.031) 0.055 (0.060) 0.006 (0.002) 0.460 (0.033) 0.133

Statistically significant at the 10% level. Statistically significant at the 5% level. Statistically significant at the 1% level.

the contrary, capital structure decisions are much more subject to managerial discretion and hence can be executed with much more ease. For these reasons, it is much more probable that the direction of causality runs from governance quality to leverage rather than vice versa. This assumption is also consistent with a large number of prior studies that make a similar argument (Berger et al., 1997; Garvey & Hanka, 1999; Harford et al., 2008; Jiraporn & Gleason, 2007; John & Litov, 2009). Finally, Ciceksever et al. (2006), utilizing simultaneous equations, find robust empirical evidence showing that managers regard governance quality as pre-determined when they make capital structure decisions. In any case, we employ two alternative estimation methods that explicitly take into account potential endogeneity. First, we focus on the earliest year in the sample for each firm and replace the values of Gov-score and ISS-score in each of the following years by their values in the earliest year. The idea is that the firm's corporate governance in the earliest year could not have resulted from leverage in subsequent years. The concern for endogeneity should thus be minimized. Using this alternative specification, we find that the results remain similar, still showing an inverse association between governance quality and leverage. This approach suggests that the direction of causality is much more likely to run from governance quality to leverage than vice versa.19 Second, to further confirm the results, we use the two-stage least squares (2SLS) analysis. This method requires instrumental variables that are related to corporate governance quality but cannot be correlated with leverage, except through governance quality. We consult the literature and identify two instrumental variables. First, the SarbanesOxley Act of 2002 (SOX) mandates several governance requirements aimed at improving governance quality. SOX can be regarded as an exogenous shock that influences governance quality. We construct a dichotomous variable equal to one for the period after the passage of SOX and zero otherwise. This variable is labeled Post-SOX and serves as our first instrument. This variable is also employed as an instrument in Knyazeva (2009) and other recent studies both in finance and accounting. Second, we employ industry-median Gov-score (or ISS-score) as our second instrument. Although the capital structure of a given firm might influence the same firm's governance, it is unlikely to be related to industry-level governance quality. Managers may have influence over their own firm's governance but they should have little influence, if any, on other firms' governance. This is why

19

The results are not shown but available upon request.

216

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

industry-level governance quality is likely exogenous and should function as a valid instrument. This approach of using industry-level governance is employed by a number of recent studies including John and Knyazeva (2006), and John and Kadyrzhanova (2008). Table 5 exhibits the two-stage least squares (2SLS) results. Model 1 is the first-stage regression, where Gov-score is the dependent variable. Industry-median Gov-score carries a significantly positive coefficient. As expected, industry-level governance significantly explains firm-level governance quality, consistent with the findings in John and Kadyrzhanova (2008). The coefficient of Post-SOX is also significantly positive, indicating an improvement in governance quality after the enactment of SOX, consistent with our expectations. Model 2 is the second-stage regression, where market leverage is the dependent variable. 20 We replace Gov-score with predicted Gov-score from the first-stage regression. Predicted Gov-score shows a negative and significant coefficient, corroborating our earlier findings. We also compute Shea's (1997) partial R 2, which is 63.31%, suggesting that our instruments are not weak instruments. Furthermore, to ensure that our instrumental variables are acceptable, we perform Sargan's (1958) test of over-identifying restrictions. The Sargan Chi-squared statistic is not significant at any conventional level. We are thus unable to reject the null hypothesis that our instruments are uncorrelated with the residuals in the second-stage regression. In other words, our instrumental variables are valid. In Model 3 and Model 4, we run a similar 2SLS analysis but replace Gov-score with ISS-score. The results remain consistent. 21 In conclusion, after subjecting the results to a battery of tests to account for both reverse causality and simultaneity, we still obtain consistent results, i.e. firms with higher governance quality are significantly less leveraged. The empirical tests in this section enhance our confidence that our conclusion is not unduly influenced by endogeneity. 4.3. Regression results based on changes in the variables As a robustness test, we estimate regression models using changes in both the dependent and independent variables. It is advantageous to analyze changes because regressions using changes are less likely to show spurious relations between variables than those using only levels. 22 Changes in leverage are regressed on changes in Gov-score. The results show a negative relation, implying that a reduction in corporate governance quality is associated with an increase in leverage. 23 The results based on yearto-year changes reinforce the previous results based on the levels of variables. The inverse association between governance quality and leverage thus seems to be robust. 4.4. Further robustness tests and other additional results A recent study by Bharath et al. (2009) reports that information asymmetry is a significant determinant of capital structure decisions. To ensure that our results are robust, we execute additional regressions with additional control variables for information asymmetry. In particular, we add the probability of informed trading (PIN) as a control. Additionally, we also include the ratio of intangible assets to total assets. Firms with more intangible assets possess more information asymmetry for it is harder to value intangible assets. The results are shown in Table 6. Model 1 and Model 2 include PIN and the intangible assets ratio as control variables. Note that ISS-score and Gov-score exhibit negative and significant coefficients in Model 1 and Model 2 respectively. Thus, even when we expand our set of control variables to account for information asymmetry, we continue to find consistent results. We also attempt to replicate the results in Jiraporn and Gleason (2007). Model 3 includes Gompers et al.'s (2003) Governance Index. The coefficient of the Governance Index, however, turns out to be insignificant. It is not surprising however that we do not obtain the same result as Jiraporn and Gleason (2007) because their sample period goes from 1992 to 2002, whereas our sample period goes from 2001 to 2004. There is only a slight overlap between the two samples, the fact that may explain our inability to reproduce their results. Model 4 replaces the Governance Index with Bebchuk et al.'s (2009) Entrenchment Index. The coefficient of the Entrenchment Index is not significant either. 24 It is worth noting that our results are consistent with those in Bebchuk et al. (2010), who report that Gompers et al.'s (2003) Index shows no significant correlation with stock returns after 2000. They provide evidence consistent with the hypothesis that the correlation and its subsequent disappearance were due to market participants' gradually learning to appreciate the difference between firms scoring well and poorly on the governance indices. Furthermore, we explore the explanatory power of Gov-score, relative to that of the Governance Index. We begin by determining the extent to which Gov-score and the Governance Index are related. We regress Gov-score on the Governance Index and find a significantly negative coefficient. This result makes sense. A higher Gov-score indicates better governance quality whereas a higher Governance Index shows poor governance quality. Thus, the two indexes are inversely related. We find, however, that the adjusted-R 2 from the regression is only 0.3%, implying that Gov-score captures other information beyond what is explained by the Governance Index. This is hardly surprising, given the substantially broader scope of Gov-score. Then, we re-estimate the regressions on capital structure, including both Gov-score and The Governance Index as independent variables. The result is shown in Model 5 (Table 6). The coefficient of Gov-score is negative and significant, consistent with our
To conserve space, we show only the results based on market leverage. When we use book leverage, the results remain similar. In addition, we also employ the generalized method of moment (GMM) to estimate the model. We obtain similar results using GMM. Year-to-year changes in variables provide a more useful test of causal relations than do levels of these variables for the levels of many variables are crosssectionally correlated without any direct causal link. While correlations in changes do not necessarily imply causality either, a failure to nd correlation in changes is likely to indicate no causal relation (Chung et al., 2010). 23 Results are not shown but available upon request. 24 John and Litov (2009) report that the Entrenchment Index is positively related to leverage. However, their sample period (19902006) is different from ours (20012004). Thus, it is not surprising that we do not report the same ndings.
21 22 20

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

217

Table 5 Two-stage least squares (2SLS) regression results for governance indices. The market value of leverage is total debt divided by the market value of total assets. The market value of assets is total assets minus the book value of equity minus deferred taxes and investment tax credits plus the market value of common equity plus preferred stock. The book value of leverage is computed as total debt divided by the book value of total assets. Gov-score is computed by awarding one point for each governance standard that is met for each firm. ISS-score is the governance index provided by the Institutional Shareholder Service (ISS). Log (total-assets) is the logarithm of total assets. Profitability (EBIT ratio) is measured by earnings before interest, and taxes divided by total assets. Growth opportunities (Tobin's q) are measured by Tobin's q as computed by Chung and Pruitt (1994). The fixed asset ratio is measured as the ratio of fixed assets to total assets. Non-debt tax shields are defined as the ratio of the sum of depreciation and amortization to total assets. Credit ratings are calculated by scoring 7 categories of AA, A, BBB, BB, B, C, D and others. The dividend ratio is the ratio of cash dividends to total assets. The Industry debt ratio is the industry median market (book) debt ratio based on two-digit SIC code. t-statistics are reported in parentheses. Model 1 First stage Dependent variable Intercept Industry-median Gov-score Industry-median ISS-score Post-SOX Predicted Log (Gov-score) Predicted Log (ISS-score) Log (total assets) EBIT ratio Tobin's q Fixed-asset ratio Non-debt tax shields Credit rating Dividend ratio Industry debt ratio Shea's (1997) Partial R2 Sargan's (1958) Statistics Adjusted R2 F-statistics Log (Gov-score) 0.647 (11.59) 0.683 (34.82) 0.170 (18.37) 0.039 (24.26) 0.003 ( 0.27) 0.005 (3.57) 0.124 ( 2.81) 0.008 (0.11) 0.008 (5.22) 0.602 (3.64) 0.171 ( 9.25) 0.6331 0.654 1354.97 Model 2 Second stage Market leverage 0.058 (7.46) 0.062 ( 6.92) 0.004 (3.05) 0.012 (1.51) 0.029 ( 25.12) 0.110 ( 3.26) 0.298 (5.82) 0.018 (14.95) 1.997 ( 15.92) 34.44 (7.46) 0.958 0.349 426.36 Model 3 First stage Log (ISS-score) 0.844 (8.98) 0.715 (30.95) 0.032 (5.82) 0.045 (21.69) 0.008 (0.59) 0.011 (5.65) 0.025 ( 0.45) 0.152 ( 1.76) 0.003 (1.39) 0.693 (3.29) 0.169 ( 7.16) 0.1191 0.260 252.94 Model 4 Second stage Market leverage 0.058 (7.58) 0.060 ( 8.62) 0.004 (2.92) 0.013 (1.59) 0.029 ( 25.21) 0.110 ( 3.28) 0.301 (5.90) 0.018 (15.00) 1.993 ( 15.99) 0.484 (34.65) 2.55 0.352 432.16

Statistically significant at the 10% level. Statistically significant at the 5% level. Statistically significant at the 1% level.

earlier findings. The Governance Index, however, does not carry a significant coefficient. It appears that Gov-score plays a more important role than the Governance Index in explaining leverage choices. Because Gov-score and the Governance Index are correlated, this collinearity may create a bias against finding significance for either one of the variables. Nevertheless, Gov-score still shows a significant coefficient, despite the stacking of the deck against such a finding. 25 Overall, the evidence suggests that Gov-score can explain capital structure variation substantially better than the Governance Index. Thus, our study represents a significant improvement over prior studies and sheds further light on the impact of overall governance quality on leverage choices. 4.5. Relations between leverage and governance categories The empirical evidence so far demonstrates a significant association between leverage and composite governance scores (Govscore and ISS-score). To gain further insights, we investigate which governance standards drive the association between leverage
25

In any event, in unreported regressions, we orthogonalize Gov-score and the Governance Index to minimize collinearity. The results remain similar.

218

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

Table 6 Regression results for Gov-score, the governance index, and the entrenchment index. The market value of leverage is total debt divided by the market value of assets. The market value of assets is total assets minus the book value of equity minus deferred taxes and investment tax credits plus the market value of common equity plus preferred stock. Gov-score is computed by awarding one point for each governance standard that is met for each firm. ISS-score is the governance index provided by the Institutional Shareholder Service (ISS). The Governance Index is the Investor Responsibility Research Center (IRRC) database and is constructed by 24 antitakeover provisions (Gompers et al., 2003). The Entrenchment Index is developed by Bebchuk et al. (2009) and consists of six most effective provisions (i.e., staggered board, limits to shareholder bylaw amendments, supermajority requirement for mergers, supermajority requirements for charter amendments, poison pills, and golden parachutes) from 24 antitakeover provisions. Log (total-assets) is the logarithm of total assets. Profitability (EBIT ratio) is measured by earnings before interest, and taxes divided by total assets. Growth opportunities (Tobin's q) are measured by Tobin's q as computed by Chung and Pruitt (1994). The fixed asset ratio is measured as the ratio of fixed assets to total assets. Non-debt tax shields are defined as the ratio of the sum of depreciation and amortization to total assets. Credit ratings are calculated by scoring 7 categories of AA, A, BBB, BB, B, C, D and others. The dividend ratio is the ratio of cash dividends to total assets. Intangible ratio is the ratio of intangible assets to total assets. PIN is the probability of information-based trading and proxy for information asymmetry measure. The industry debt ratio is the industry median market (book) debt ratio based on two-digit SIC code. Robust and clustered by firm adjusted t-statistics are reported in parentheses. Model 1 Intercept Log (ISS-score) Log (Gov-score) Log (Gindex) Log (Eindex) Log (total assets) EBIT ratio Tobin's q Fixed-asset ratio Non-debt tax shield Credit rating Dividend ratio Intangible PIN Industry debt ratio Adjusted R2 0.007*** (3.862) 0.007 (0.631) 0.027*** ( 18.904) 0.046 ( 0.809) 0.273*** (3.236) 0.018*** (9.653) 1.898*** ( 11.575) 0.058*** (4.217) 0.104*** (5.476) 0.474*** (17.645) 0.359 0.008*** (3.833) 0.006 (0.504) 0.028*** ( 18.968) 0.053 ( 0.913) 0.282*** (3.297) 0.018*** (9.720) 1.915*** ( 11.595) 0.056*** (4.106) 0.109*** (5.675) 0.473*** (17.523) 0.357 0.014*** (3.936) 0.009 ( 0.322) 0.035*** ( 10.811) 0.133* ( 1.665) 0.178 (1.404) 0.012*** (5.381) 2.025*** ( 9.671) 0.073*** (3.501) 0.163*** (5.795) 0.411*** (12.012) 0.386 0.225*** (5.608) 0.057*** ( 5.620) Model 2 0.158*** (4.502) Model 3 0.028 ( 0.822) Model 4 0.007 ( 0.262) Model 5 0.035 (0.884)

0.058*** ( 4.518) 0.009 (0.815) 0.009 (1.294) 0.013*** (3.579) 0.003 (0.107) 0.037*** ( 10.468) 0.130 ( 1.538) 0.141 (1.092) 0.013*** (5.494) 1.905*** ( 8.720) 0.071*** (3.233) 0.151*** (5.217) 0.399*** (11.505) 0.376

0.023*** ( 3.425) 0.007 (0.642)

0.014*** (4.088) 0.008 ( 0.277) 0.035*** ( 10.917) 0.140* ( 1.752) 0.181 (1.442) 0.013*** (5.572) 2.009*** ( 9.611) 0.071*** (3.399) 0.165*** (5.847) 0.409*** (11.990) 0.387

*, **, *** statistically significant at the 10%, 5%, and 1% level respectively.

and aggregate governance quality. Although finance theory does not provide clear guidance on this issue, we can make certain conjectures on which categories of Gov-score and ISS-score should have more influence on leverage. For instance, as detailed in the Appendix, Director education, consists of just a single governance standard. In 2004, fewer than 2% of firms satisfied the single governance standard under Director education. This governance category is therefore unlikely to explain much variation in leverage. To ascertain which governance categories drive the results, we run a regression analysis, where market leverage is regressed on the eight governance standards shown in the Appendix (the eight standards are Board, Audit, Charter, State, Ownership, Compensation, Progressive, and Director education) and the control variables. The regression results are shown in Table 7. We find that five governance categories are negatively associated with leverage; Board, Ownership, Audit, State, and Compensation. 26 It is worth noting that the coefficient on Charter is not significant. This is somewhat unexpected because Charter is the category that is most closely related to the Governance Index developed by Gompers et al. (2003). Each of the seven standards in Charter also appears in the Governance Index. The results are surprising, given that both John and Litov (2009) and Jiraporn and Gleason (2007) report a significant association between leverage and the Governance Index. In any event, Charter is merely

26 We do not include all the governance categories in a single regression due to multi-collinearity. Nevertheless, when all of them are included, their coefcients generally retain the signs and signicance levels except for Compensation.

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

219

Table 7 Regression results for governance categories. The dependent variable is the market value of leverage, which is calculated as total debt divided by the market value of total assets. The market value of assets is total assets minus the book value of equity minus deferred taxes and investment tax credits plus the market value of common equity plus preferred stock. Gov-score is computed by awarding one point for each governance standard that is met for each firm. ISS-score is the governance index provided by the Institutional Shareholder Service (ISS) that consists of eight standards are Board, Audit, Charter, State, Ownership, Compensation, Progressive, and Director education. Models 18 include each governance standards and control variables used in Table 3. Detailed description of each standard is shown in Appendix. Standard errors are adjusted for 2725 clusters in firms and robust variance estimators. t-statistics are reported in the parenthesis. Model 1 Board Ownership Charter Audit State Compensation Progressive Director education Control variables Year dummies Adjusted R2 Included Included 0.349 Included Included 0.348 Included Included 0.345 Included Included 0.345 Included Included 0.346 Included Included 0.345 Included Included 0.350 0.005*** ( 6.76) 0.010*** ( 5.17) 0.000 ( 0.11) 0.004** ( 2.00) 0.028*** ( 3.18) 0.001 (0.42) 0.008*** ( 6.94) 0.015 ( 0.94) Included Included 0.345 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

*, **, *** statistically significant at the 10%, 5%, and 1% level respectively.

similar to the Governance Index, containing only a subset of the governance provisions included in the Governance Index. The results therefore may not be directly comparable.

5. Conclusion We relate capital structure to aggregate corporate governance quality. Agency theory argues that capital structure decisions are influenced by agency costs. Corporate governance is designed to alleviate agency problems. We thus argue that capital structure and corporate governance are related through their association with agency costs. Two hypotheses are advanced to explain the relationship; the outcome hypothesis and the substitution hypothesis. Employing broad-based comprehensive governance indices, we empirically test the link between leverage and governance quality. The empirical evidence demonstrates a robust inverse relation, which is consistent with the prediction of the substitution hypothesis. Furthermore, we also demonstrate that our results are unlikely influenced by endogeneity. Stronger governance appears to bring about, not merely reflect, lower leverage. When we compare our governance metrics with those previously employed, we find that our governance metrics can explain variation in leverage choices much better. The results of our study are important for they demonstrate that the overall quality of corporate governance has a palpable impact on crucial corporate decisions like capital structure choices.

Appendix. Governance standards included in the construction of the governance score

Governance standards A. Audit 1. Audit committee consists solely of independent outside directors 2. Auditors were ratified at the most recent annual meeting. 3. Consulting fees paid to auditors are less than audit fees paid to auditors 4. Company has a formal policy on auditor rotation B. Board of Directors 1. Managers respond to shareholder proposals within 12 months of shareholder meeting 2. CEO serves on no more than two additional boards of other public companies. 3. All directors attend at least 75% of board meetings or had a valid excuse for non-attendance.

220 Appendix (continued ) (continued ) Governance standards

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

4. Size of board of directors is at least 6 but not more than 15 members. 5. No former CEO serves on board. 6. CEO is not listed as having a related party transaction in proxy statement. 7. Board is controlled by more than 50% independent outside directors. 8. Compensation committee is comprised solely of independent outside directors. 9. The CEO and chairman duties are separated or a lead director is specified. 10. Shareholders vote on directors selected to fill vacancies. 11. Board members are elected annually. 12. Shareholder approval is required to change board size. 13. Nominating committee is comprised solely of independent directors. 14. Governance committee meets at least once during the year. 15. Shareholders have cumulative voting rights to elect directors. 16. Board guidelines are in each proxy statement. 17. Policy exists requiring outside directors to serve on no more than five additional boards. C. Charter/bylaws 1. A simple majority vote is required to approve a merger (not a supermajority). 2. Company either has no poison pill or a pill that was shareholder approved. 3. Shareholders are allowed to call special meetings. 4. A majority vote is required to amend charter/bylaws (not a supermajority). 5. Shareholders may act by written consent and the consent is non-unanimous. 6. Company is not authorized to issue blank check preferred stock. 7. Board cannot amend bylaws without shareholder approval or can only do so under limited circumstances. D. Director education 1. At least one member of the board has participated in an ISS-accredited director education program. E. Executive and Director compensation 1. No interlocks exist among directors on the compensation committee. 2. Non-employees do not participate in company pension plans. 3. Option re-pricing did not occur within last three years. 4. Stock incentive plans were adopted with shareholder approval. 5. Directors receive all or a portion of their fees in stock. 6. Company does not provide any loans to executives for exercising options. 7. The last time shareholders voted on a pay plan, ISS did not deem its cost to be excessive. 8. The average options granted in the past three years as a percentage of basic shares outstanding did not exceed 3% (option burn rate). 9. Option re-pricing is prohibited. 10. Company expenses stock options. F. Ownership 1. All directors with more than one year of service own stock. 2. Officers' and directors' stock ownership is at least 1% but not over 30% of total shares outstanding. 3. Executives are subject to stock ownership guidelines. 4. Directors are subject to stock ownership guidelines. G. Progressive practices 1. Mandatory retirement age for directors exists. 2. Performance of the board is reviewed regularly. 3. A board-approved CEO succession plan is in place. 4. Board has outside advisors. 5. Directors are required to submit their resignation upon a change in job status. 6. Outside directors meet without the CEO and disclose the number of times they met. 7. Directors term limits exist. H. State of incorporation 1. Incorporation in a state without any anti-takeover provisions.

References
Agrawal, A., & Knoeber, C. (1996). Firm performance and mechanisms to control agency problems between managers and shareholders. Journal of Financial and Quantitative Analysis, 31, 377397. Aggarwal, R., Erel, I., Ferreira, M., & Matos, P. (2011). Does governance travel around the world? Evidence from international investors. Journal of Financial Economics, 100, 154181. Aggarwal, R., Erel, I., Stulz, R., & Williamson, R. (2009). Differences in governance practices between U.S. and foreign firms: Measurement, causes, and consequences. Review of Financial Studies, 22, 31313169. Ang, J., Cole, R. A., & Lin, J. W. (2002). Agency costs and ownership structure. Journal of Finance, 55, 81106. Bebchuk, L. A., Cohen, A., & Ferrell, A. (2009). What matters in corporate governance? Review of Financial Studies, 22, 783827. Bebchuk, L. A., Cohen, A., & Wang, C. (2010). Learning and disappearing association between governance and returns. Working paper, Harvard University. Berger, P., Ofek, E., & Yermack, D. (1997). Managerial entrenchment and capital structure decisions. Journal of Finance, 52, 14111438. Bharath, S., Pasquariello, P., & Wu, G. (2009). Does asymmetric information drive capital structure decisions? Review of Financial Studies, 22, 32113243.

P. Jiraporn et al. / International Review of Economics and Finance 22 (2012) 208221

221

Bowen, R., Rajgopal, S., & Venkatachalam, M. (2008). Accounting discretion, corporate governance and firm performance. Contemporary Accounting Research, 25, 351405. Brown, L., & Caylor, M. (2006). Corporate governance and firm valuation. Journal of Accounting and Public Policy, 25, 409434. Byoun, S. (2008). How and when do firms adjust their capital structures toward targets? Journal of Finance, 63, 30693095. Byoun, S., Moore, W., & Xu, Z. (2009). Why do some firms become debt-free? Working paper, Baylor University. Charoenwong, C., Ding, D., & Siraprapasiri, V. (2011). Adverse selection and corporate governance. International Review of Economics and Finance, 20, 406420. Choi, S. J., Fisch, J. E., & Kahan, M. (2008). Director elections and the role of proxy advisors. Working paper, New York University. Chung, K., Elder, J., & Kim, J. C. (2010). Corporate governance and liquidity. Journal of Financial and Quantitative Analysis, 45, 265291. Chung, K., & Pruitt, S. (1994). A simple approximation of Tobin's q. Financial Management, 23, 7075. Ciceksever, B., Kale, J., & Ryan, H. (2006). Corporate governance, debt, and activist institutions. Working paper, Georgia State University. Cotter, J. F., Palmiter, A., & Thomas, R. (2009). The effect of ISS recommendations on mutual voting. Working paper, Vanderbilt University. Coughlan, A., & Schmidt, R. (1985). Executive compensation, management turnover, and firm performance: An empirical investigation. Journal of Accounting and Economics, 7(13), 4366. Cremers, K. J. M., & Nair, V. (2005). Governance mechanisms and equity prices. Journal of Finance, 60, 28592894. Cremers, K. J. M., Nair, V., & Wei, C. (2007). Governance mechanisms and bond prices. Review of Financial Studies, 20, 13591388. Daines, R. (2001). Does Delaware law improve firm value? Journal of Financial Economics, 62, 559571. DeAngelo, H., & Masulis, R. (1980). Leverage and dividend irrelevancy under corporate and personal taxation. Journal of Finance, 35, 453464. Demsetz, H., & Lehn, K. (1985). The structure of corporate ownership: Causes and consequences. Journal of Political Economy, 93, 11551177. Dittmar, A., & Mahrt-Smith, J. (2007). Corporate governance and the value of cash holdings. Journal of Financial Economics, 83, 599634. Fahlenbrach, R. (2009). Shareholder rights, boards, and CEO compensation. Review of Finance, 13, 81113. Fama, E., & Miller, M. (1972). The theory of finance. New York: Holt, Rinehart, and Winston. Frank, M., & Goyal, V. (2009). Capital structure decisions: Which factors are reliably important? Financial Management, 38, 137. Garvey, G., & Hanka, G. (1999). Capital structure and corporate control: The effect of antitakeover statutes on firm leverage. Journal of Finance, 54, 519548. Gilson, R. (2002). Unocal fifteen years later (and what we can do about it). Delaware Journal of Corporate Law, 26, 491513. Gompers, P., Ishii, J., & Metrick, A. (2003). Corporate governance and equity prices. Quarterly Journal of Economics, 118, 107155. Grossman, S., & Hart, O. (1982). Corporate financial structure and managerial incentives. In John McCall (Ed.), The economics of information and uncertainty. Chicago, IL: University of Chicago Press. Harford, J., Li, K., & Zhao, X. (2008). Corporate boards and the leverage and debt maturity choices. International Journal of Corporate Governance, 1, 327. Henry, D. (2010). Agency costs, ownership structure, and corporate governance compliance: A private contracting perspective. Pacific-Basin Finance Journal, 18, 2446. Hermalin, B., & Weisbach, M. (1991). The effects of board composition and direct incentives on firm performance. Financial Management, 20, 101112. Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance and takeovers. American Economics Review, 76, 323339. Jensen, M. C., & Meckling, W. (1976). Theory of the firm: Managerial behavior, agency costs, and capital structure. Journal of Financial Economics, 3, 305360. Jensen, M., & Murphy, K. J. (1990). Performance pay and top-management incentives. Journal of Political Economy, 98, 225264. Jensen, M., & Murphy, K. J. (1990). CEO incentives: It's not how much, but how. : Harvard Business Review (May/June).. Jiraporn, P., & Gleason, K. (2007). Capital structure, shareholder rights, and corporate governance. Journal of Financial Research, 30, 2133. Jiraporn, P., Kim, Y. S., Davidson, W. N., & Singh, M. (2006). Corporate governance, shareholder rights and firm diversification: An empirical analysis. Journal of Banking and Finance, 30, 947963. Jiraporn, P., & Ning, Y. (2006). Dividend policy, shareholder rights, and corporate governance. Journal of Applied Finance, 2436 Fall/Winter,. John, K., & Kadyrzhanova, D. (2008). Spillover effects in the market for corporate control. Working paper, New York University. John, K., & Knyazeva, A. (2006). Payout policy, agency conflicts, and corporate governance. Working paper, New York University. John, K., & Litov, L. P. (2009). Corporate governance and financing policy: New evidence. Working paper. St. Louis: Washington University. Johnson, S. A. (1997). The effect of bank debt on optimal capital structure. Financial Management, 28, 4756. Knyazeva, D. (2009). Analysts and journalists: Firm policies and governance design. Working paper, University of Rochester. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. (2000). Agency problems and dividend policy around the world. Journal of Finance, 55, 133. Leland, H., & Pyle, D. (1977). Information asymmetries, financial structure, and financial intermediation. Journal of Finance, 32, 371388. Lipton, M., & Rowe, P. (2002). Pills, polls, and professors: A reply to Professor Gilson. Delaware. Journal of Corporate Law, 27, 155. Macey, J. R. (1998). The legality of the shareholder rights by-law in Delaware: Preserving the market for corporate control. Journal of Applied Corporate Finance, 4, 6368. McConnell, J., & Servaes, H. (1990). Additional evidence on equity ownership and corporate value. Journal of Financial Economics, 27, 595612. McKnight, P. J., & Weir, C. (2009). Agency costs, corporate governance mechanisms and ownership structure in large UK publicly quoted companies: A panel data analysis. The Quarterly Review of Economics and Finance, 49, 139158. Mehran, H. (1992). Executive incentive plans, corporate control, and capital structure. Journal of Financial and Quantitative Analysis, 27, 539560. Modigliani, F., & Miller, M. (1958). The cost of capital, corporation finance, and the theory of investment. American Economic Review, 48, 261297. Morck, R., Shleifer, A., & Vishney, R. (1988). Management ownership and market valuation: An empirical analysis. Journal of Financial Economics, 20, 293316. Murphy, K. (1998). Executive compensation. Working paper, University of Southern California. Myers, S. (1977). Determinants of corporate borrowing. Journal of Financial Economics, 5, 147176. Myers, S. (1984). The capital structure puzzle. Journal of Finance, 39, 575592. Park, Y. W., Selvili, Z., & Song, M. H. (2008). Large outside blockholders as monitors: Evidence from partial acquisitions. International Reviews of Economics and Finance, 17, 529545. Parsons, C., & Titman, S. (2008). Empirical capital structure: A review. Foundation and Trends in Finance, 3, 193. Ross, S. (1977). The determination of financial structure: The incentive signaling approach. Bell Journal of Economics, 8, 2340. Rozeff, M. (1982). Growth, beta and agency costs as determinants of dividend payout ratios. Journal of Financial Research, 5, 249259. Sargan, J. D. (1958). The estimation of economic relationships using instrumental variables. Econometrica, 26, 393415. Shea, J. (1997). Instrument relevance in multivariate linear models: A simple measure. The Review of Economics and Statistics, 79, 348352. Singh, M., & Davidson, W. N. (2003). Agency costs, ownership structure, and corporate governance mechanisms. Journal of Banking and Finance, 27, 793816. Sitkoff, R. (2002). Corporate political speech, political exhortation, and the competition for corporate charters. University of Chicago Law Review, 69, 11031166. Titman, S., & Wessels, R. (1988). The determinants of capital structure choice. Journal of Finance, 43, 119.