Tax avoidance

Published on March 2017 | Categories: Documents | Downloads: 29 | Comments: 0 | Views: 265
of 15
Download PDF   Embed   Report

Comments

Content

Journal of Banking & Finance 35 (2011) 456–470

Contents lists available at ScienceDirect

Journal of Banking & Finance
journal homepage: www.elsevier.com/locate/jbf

Tax avoidance, cost of debt and shareholder activism: Evidence from Korea
Youngdeok Lim ⇑
Australian School of Business, University of New South Wales, Sydney, NSW, Australia

a r t i c l e

i n f o

a b s t r a c t
This paper examines the impact of tax avoidance on the cost of debt and its interaction effect with shareholder activism. Using Korean firms, I find a negative relationship between tax avoidance and the cost of debt, supporting the trade-off theory. Further tests reveal that the negative relationship becomes stronger when the level of institutional ownership is high. It becomes even stronger after 1998, when the shareholder rights of institutional investors were strengthened. It suggests that the managerial opportunism theory has an additional explanation for tax avoidance activities. My findings indicate that tax avoidance reduces the cost of debt through trade-offs and creates a managerial rent diversion, which is mitigated in firms with larger institutional holdings. Crown Copyright Ó 2010 Published by Elsevier B.V. All rights reserved.

Article history: Received 8 March 2009 Accepted 23 August 2010 Available online 27 August 2010 JEL classification: G30 H20 Keywords: Tax avoidance Cost of debt Trade-off Managerial opportunism Shareholder activism

1. Introduction This paper investigates the impact of tax avoidance1 on the cost of debt and its interaction effect with shareholder activism in publicly-held firms. The Hyundai Automotive Group, a Korean chaebol (business group), faced allegations that it created massive slush funds through the unusual method of deflating its operating profits and inflating losses. In fraudulent accounting practices, the opposite—inflating profits and deflating losses to lure investors—is more common. As a result, a prosecution official stated, ‘‘We’re investigating to determine the overall scale of the slush funds the Hyundai Automotive Group established by adopting such accounting measures. If my probe into the group reveals that it did not pay taxes
⇑ Tel.: +61 2 9385 6081; fax: +61 2 9385 5925.
E-mail address: [email protected] I define tax avoidance as a tax savings that arises from both the general tax reduction methods and tax shelters that are occasionally of questionable legality to minimize tax liability. In other words, the tax avoidance measure conceptually captures the cumulative number of transactions to minimize tax liabilities (Desai and Dharmapala, 2006). Both Korean and American firms are expected to have strong incentives to increase financial income and decrease taxable income, thereby suggesting that the differences in tax laws and accounting standards between the US and Korea could have little effect on the interpretation of tax avoidance. Also Korean firms could avoid tax burden by the tax savings derived from legal tax reduction methods. Korean tax law permits firms to only deduct expenses for tax purposes, not for financial purposes, as follows: depreciation for the special assets class, allowance for depreciation, bad debt losses caused by extinctive prescriptions, and the allowance for retirement pensions.
1

and used the money for business or lobbying purposes, it will face additional charges for tax evasion and appropriation of company funds in the conduct of business or bribery. Investigators are collecting mountains of evidence (Chosun Ilbo, 17 April 2006).” In response, the group announced that its chairman promised to donate $1.1 billion worth of personal assets to society and apologized for causing concerns to the public over the scandal. As a result, the chairman was sentenced to 3 years in prison for creating the large slush funds. The first objective of this paper is to investigate whether tax avoidance as a tax-favored activity reduces the cost of debt. The cost of the debt of a firm is determined by the characteristics of the firm and those of the bond issue that affect default risk, agency costs, and the information asymmetry problem (Bhojraj and Sengupta, 2003). Graham and Tucker (2006) and Lim (2010) suggest that tax-favored activities, such as tax shelters and tax avoidance, are a substitute for the use of debt. Graham and Tucker (2006) examine 44 tax shelter cases that were issued as a Notice of Deficiency by the Internal Revenue Services (IRS). These cases indicate that firms use less debt when they engage in tax sheltering. Using the tax avoidance measure modified from Desai and Dharmapala (2006), Lim (2010) determines the existence of a substitution effect of tax avoidance for the use of debt for a large sample of Korean firms. These results were consistent with the results of Graham and Tucker (2006). If tax avoidance is a substitute for the use of debt (Graham and Tucker, 2006; Lim, 2010), it could increase financial slack, reduce expected bankruptcy costs, enhance credit quality, lower default risk, and therefore reduce the cost of debt.

0378-4266/$ - see front matter Crown Copyright Ó 2010 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jbankfin.2010.08.021

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

457

This supports the trade-off hypothesis that the debt-substitution effect of tax avoidance would reduce the cost of debt, an issue not investigated in Graham and Tucker (2006) and Lim (2010). Second, this paper investigates managerial opportunism theory on the relationship between tax avoidance and cost of debt by examining the effect of shareholder activism on the relationship. From an agency perspective, tax avoidance would reduce the transparency of firms, and permit managers with the opportunity to extract rents from outside investors, creating a shield for managerial opportunism and the diversion of rents (Desai and Dharmapala, 2006; Desai et al., 2007; Desai and Dharmapala, 2009; Wilson, 2009). Desai and Dharmapala (2006) and Desai et al. (2007) argue that corporate tax sheltering and the diversion of rents by managers are interrelated and complementary. Desai and Dharmapala (2009) find that the average effect of tax avoidance on firm value was not significantly different from zero. However, it is found to be positive for well-governed firms, which indicates that a higher quality of corporate governance, measured as a higher level of institutional ownership, leads to a favorable effect of tax avoidance on firm value. Tax avoidance could cause agency conflicts between management and debtholders, since managerial rent diversions induce information asymmetry and create moral hazard problems. However, institutional investors possess greater incentives and capacity to monitor managerial performance (Shleifer and Vishny, 1986; Chung et al., 2002; Hartzell and Starks, 2003; Bhojraj and Sengupta, 2003; Desai and Dharmapala, 2009).2 Thus, the higher the level of institutional ownership, the greater is the degree of scrutiny to which managerial actions are subjected, and the less important is the conflict of interests between the management and debtholders. Institutional investors could reduce the cost of debt by alleviating agency problems, thereby decreasing opportunities for the managerial rent diversion of tax avoidance. Using Korean firms, I examine the impact of tax avoidance on the cost of debt. I further investigate the effect of shareholder activism on the relationship between tax avoidance and the cost of debt. Shareholder activism was defined as a level of ownership by institutional investors (Desai and Dharmapala, 2009) and an exogenous variation of institutional investors in 1998 in Korea. For example, the Sovereign Asset Management Corporation,3 which emerged as the biggest shareholder of the SK Group, one of chaebols in Korea, stated that for the SK to recover, radical reforms were necessary. Sovereign stated that it planned to renovate the corporate governance of the SK to clean up the company, and added that it would work with the management team of SK to achieve such goals. Sovereign expressed the view that the SK must discontinue its old business practices and strive to regain shareholder and market trust, stating that ‘‘Investment in the SK will transform a corporate tragedy into a triumph of corporate governance.” I anticipate that institutional investors reduce the cost of debt by alleviating agency problems, thereby increasing the negative effect of tax avoidance on the cost of debt. Furthermore, I expect that the impact of institutional investors on tax avoidance and the cost of debt increased after 1998, when shareholder rights related to institutional investors, including foreign investors, were strengthened in Korea. I estimate the tax avoidance measure modified from Desai and Dharmapala (2006). My results indicate that there is a negative relationship between tax avoidance and the cost of debt for a large sample of Korean firms, supporting the trade-off hypothesis. Further tests reveal that the relationship becomes stronger when the ownership of institutional investors is high, becoming even stronger after 1998, indicating that managerial opportunism arguments
2 It is also an empirical issue as to whether all type of institutional investors would have an effective monitoring role. I discuss this issue in detail in Section 5.5.3. 3 Sovereign Asset Management possesses over 20 years’ experience in the international capital market.

have an additional explanation for tax avoidance. I find that the influence of shareholder activism on the negative relationship between tax avoidance and the cost of debt was greater when controlling for the other governance effects in 1998. These results are robust for a wide variety of tests, including the relationship between tax avoidance and firm value, changes in the tax rate, the decomposition of institutional investors, the additional control for the probability of bankruptcy, and alternative measures. My findings indicate that tax avoidance reduces the cost of debt through trade-offs and creates a managerial rent diversion, which is mitigated in firms with larger institutional holdings. Also, these suggest that bondholders view institutional investors as a monitoring vehicle that decreases the opportunities for managerial rent diversion by mitigating agency costs. Korea provides a good research setting for exploring these research questions in terms of five aspects. First, unlike in the US, taxable income data necessary to calculate the tax avoidance measure is directly available from annual reports and does not need to be estimated (Desai and Dharmapala, 2006). This reduces the measurement error of tax avoidance and provides a large sample of evidence. Second, traditionally, Korean firms have heavily relied on bank financing (Baek et al., 2004) and experienced the financial crisis that began in 1997. Since chaebols, such as the Daewoo Group, Kia Group, and Hanbo Group, went bankrupt, debtholders suffered from this process and became vulnerable to managerial malfeasance or diversions. Weak investor protection in Korea enabled controlling shareholders to expropriate other investors, including minority shareholders and debtholders. This caused debtholders to become interested in mitigating agency conflicts between controlling shareholders and debtholders through good corporate governance in Korea. Third, Korean corporate governance has characteristics that are particularly suitable to my research questions. In 1998, after the financial crisis, the Korean government reformed shareholder rights. This source of exogenous variation of institutional investors permits the investigation of the causal effect of institutional investors on tax avoidance and the cost of debt. Fourth, Korean companies were required to disclose their ownership in the annual report using standard types of security holders until 2003. These security holders include governments, government-related companies, security firms, insurance firms, banks, individual investors, and foreign investors. This makes it possible to decompose institutional investors and investigate the impact of the types of institutional investors on cost of debt (e.g., Elyasiani and Jia, 2010).4 Finally, the Korean government has shown interest in the increasing number of tax avoidance activities and has designed a policy to effectively prevent such activities in 2006, thereby making tax avoidance an appropriate issue to examine in Korea. This study makes the following contributions to the literature. First, this paper makes significant contribution to previous literature by investigating the relationship between tax avoidance and cost of debt and further applying the agency perspective to the relationship. To my knowledge, this is the first attempt to explicitly examine the impact of tax avoidance on the pricing of corporate debt, which also provides some insight into the bond market reaction to the substitution effect of tax avoidance for the use of debt (Graham and Tucker, 2006; Lim, 2010). Also, this paper considers agency costs for the analysis of corporate tax avoidance on the cost of debt and investigates whether institutional investors, as shareholder activists, alleviate managerial opportunism and affect the negative relationship between tax avoidance and the cost of debt.
4 The requirement was removed in 2004. Since the policy took effect, Korean companies no longer voluntarily disclose this information in their annual report. The problem in obtaining institutional ownership data causes me to define the sample period as 1994–2003.

458

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

Desai and Dharmapala (2009) only test the agency perspective of corporate tax avoidance on firm value and Lim (2010) only tests the trade-off theory by using Korean evidence and show that tax avoidance is negatively associated with the use of debt. Second, this paper illustrates that improved shareholder activism leads to a lower cost of debt, implying that good corporate governance proxied by institutional ownership could reduce the cost of capital from the perspective of enterprise risk management. Even though a substantial amount of research addresses the relationship between corporate governance and firm value (e.g., Joh, 2003; Baek et al., 2004; Black et al., 2006; Desai and Dharmapala, 2009), few studies have examined the relationship between corporate governance and the cost of debt. Daily et al. (2003) identify shareholder activism as an important factor in corporate governance, which has not received adequate attention in the extant literature. This study provides new evidence on the agency conflicts of controlling shareholders and debtholders. Finally, I improve upon the tax avoidance measure used in Desai and Dharmapala (2006). Unlike Desai and Dharmapala (2006), I use the discretionary accruals, rather than total accruals as an earnings management proxy and improve upon the accuracy of the tax avoidance estimation. I also provide evidence that tax avoidance comes from both temporary and permanent differences. The remainder of the paper is organized as follows. Section 2 develops the hypotheses. Section 3 describes the sample selection procedures. Section 4 specifies the regression variables used for testing the impact of tax avoidance and institutional ownership on the cost of debt. Section 5 presents the test methodology and results for the debt pricing tests, and Section 6 concludes the paper.

2. Extant research and development of hypotheses 2.1. Korean institutional backgrounds Chaebol is defined as a family-controlled industrial conglomerate in Korea.5 Although the business group is not unique to Korea and can be found in other countries such as India, Chile, Spain, and Japan, several features distinguish Korean chaebols from those in other countries. First, Korean chaebols operate in many industries and have a substantial economic influence on the Korean economy. For example, the Samsung Group’s market capitalization was almost 21% of total market capitalization in 2006. Second, despite their huge size and economic power, they are controlled by an individual owner-manager. The owner–manager has substantial discretionary power to transfer capital and managerial resources among the affiliated firms. As a result, firms within the same chaebol operate like business units of a large corporation, sharing each other’s resources and risks through an extensive arrangement of reciprocal shareholding agreements and cross-debt guarantees among member firms (Baek et al., 2004). Weak investor protection in Korea facilitates the expropriation of controlling shareholders for other investors (i.e., minority shareholders and debtholders). In Korea, controlling shareholders’ over other shareholders enables the efficient management of a firm, in spite of their low level of ownership. It also takes advantage of managerial opportunism for private benefits. In particular, the extensive arrangement of a pyramidal, or multi-layered, shareholding agreement in chaebols inspires tunneling (Bae et al., 2002; Baek et al., 2006) or propping (Bae et al., 2008).
5 For example, 15 listed firms and 43 unlisted firms belonged to the Samsung Group at the end of 2006. Listed Samsung Group firms included Samsung Corporation, Cheil Industries, Samsung Electronics, Samsung SDI, Samsung Securities, Samsung Techwin, Samsung Electro-mechanics, S1, Samsung Heavy Industries, Cheil Worldwide, Samsung Fire and Marine Insurance, Samsung Fine Chemicals, The Shilla, Samsung Engineering and Credu.

Bae et al. (2002) indicate that when a chaebol-affiliated firm makes an acquisition, its stock price, on average, falls. While the minority shareholders of a chaebol-affiliated firm that has made an acquisition lose money, the controlling shareholders, on average, benefit, because the acquisition enhances the value of the other firms in the group, which is consistent with the tunneling motive. Baek et al. (2006) also report that chaebols, when issuing intra-group equity-linked private securities offerings, determine the offering prices in a manner that benefits their controlling shareholders, thereby suggesting that the equity-linked private securities offerings are used as a mechanism for tunneling among the member firms of the chaebol groups. This suggests that, in Korea, the controlling shareholders benefit at the expense of the other investors (i.e., minority shareholders and debtholders). On the other hand, Bae et al. (2008) find that the announcement of increased (decreased) earnings over the previous year by a chaebol-affiliated firm has a positive (negative) effect on the abnormal returns of the value-weighted portfolio of other non-announcing affiliates in the same group, as well as on those of each individual non-announcing affiliate, suggesting the existence of intra-group propping behavior. In Korea, the importance of corporate governance and the protection of shareholder rights only began to be seriously recognized in 1997, when the Asian economic crisis began. Institutional investors in Korea were deprived of their voting rights through the mandatory shadow voting policy imposed by law, and there was practically no way for shareholders to protect themselves from being exploited by management or controlling shareholders. Joh (2003) suggests that financial institutions and nonfinancial corporations, such as institutional shareholders, did not monitor firm activity prior to 1998, despite the fact that they owned over 40% of the shares.6 The mandatory shadow voting rule forbade all financial institutions from voting on decisions pertaining to firms. Furthermore, nonfinancial institutional investors often protected the incumbent controlling shareholder from potential external threats. A majority of nonfinancial corporation ownership was comprised of cross-holding or interlocking ownership of the affiliated firms in the chaebols. Poor corporate governance, particularly for minority shareholders in Asian countries, including Korea, was considered to be one of the important causes of the crisis. The financial crisis around the end of 1997 led to numerous changes in Korea. The Korean government made various efforts to strengthen its corporate governance (Johnson et al., 2000). In 1998, the Korean government reformed shareholder rights related to institutional investors, including foreign investors, in corporate governance in the following manner: first, the voting rights of the institutional investors were reinstated by abolishing mandatory shadow voting. Second, the exercise of several important shareholder rights, such as the right to file a derivative suit, were made much easier by reducing the minimum shareholding required for exercising those rights. Third, limits on share ownership by foreign investors were abolished in May of 1998. This source of exogenous variation of institutional investors permits the investigation of the causal effect of institutional investors on shareholder activism. Apart from shareholder rights, reforms related to the board of directors, and the audit committee, were initiated after the financial crisis of 1997. According to the Securities and Exchange Act, the appointment of independent external directors to the board and the establishment of an audit committee were mandated for

6 A majority of the banks are virtually controlled by the government and hold approximately 10% of the shares of the listed firms. Other financial institutions, such as insurance companies, security firms, and investment trust companies, own over 10% of the shares. In contrast, nonfinancial corporations hold over 20% of the shares.

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

459

all listed companies with total asset holdings of over 2 trillion Korean Won.7 2.2. Hypothesis development 2.2.1. Tax avoidance, debt-substitution, and cost of debt DeAngelo and Masulis (1980) suggest that firms select a level of debt that is negatively related to the level of non-debt tax shields (hereafter, NDTS), such as depreciation deductions or investment tax credits. Several studies indicate that depreciation and investment tax credits can substitute for debt (MacKie-Mason, 1990; Trezevant, 1992). Graham (2000) provides an empirical measure for the underutilization of debt by corporations, ‘‘the kink”, and argues that the average magnitude of debt usage appeared to be small, relative to the tax benefits of debt, because the ratio of interest deductions to expected income was small for a large number of firms. Graham et al. (2004) analyze corporate stock options and determine that they account for a certain unexplained underleveraged phenomena—the stock options account for approximately 20% of the mean value of the kink (Graham, 2000). Some studies investigate tax avoidance activities as an extension of tax-favored activity. Graham and Tucker (2006) empirically investigate whether tax shelters can be substituted for the use of debt. They construct a sample of firms involved in 44 corporate tax shelter cases over the period of 1975–2000. By comparing these firms with a matched sample of firms that are not involved in such litigation, they find that characteristics, such as size and profitability, are positively associated with the use of tax shelters. They also argue that tax shelters serve as a substitute for interest deductions in determining the capital structure. Lim (2010) examines whether participating in tax avoidance activities is related to the capital structure in a Korean setting and also examine the tax-exhaustion effect in this relationship. Using the tax avoidance measure modified from Desai and Dharmapala (2006), Lim (2010) finds the existence of a substitution effect of tax avoidance for the use of debt for a large sample of Korean firms; Lim (2010) also finds that the substitution effect increases with the tax-exhaustion effect, thereby generalizing the evidence of Graham and Tucker (2006). The cost of the debt of a firm is influenced by the characteristics of the firm and those of the bond issue that affect default risk, agency costs, and the information asymmetry problem (Bhojraj and Sengupta, 2003). If tax avoidance serves as a substitute for the use of debt (Graham and Tucker, 2006; Lim, 2010), it could increase financial slack, reduce expected bankruptcy costs, enhance credit quality, lower default risk, and therefore, reduce the cost of debt. Graham and Tucker (2006) report that the credit ratings of tax shelter firms improved one notch, as compared with matched firms in the years leading up to the inception of the tax shelter, most likely because of falling debt ratios. Using an instrumental variable approach, Molina (2005) determines that the negative impact of leverage on ratings was three times stronger under the endogeniety of leverage. Based on the previous discussion, I propose the trade-off hypothesis as follows: H1. Tax avoidance is negatively associated with the cost of debt.

2.2.2. Tax avoidance, cost of debt, and shareholder activism The agency perspective that emphasizes the relationship between corporate governance and responses to taxes suggests that tax avoidance activities can create a shield for managerial opportunism and the diversion of rents (Desai and Dharmapala, 2006, 2009; Desai et al., 2007; Wilson, 2009; Shackelford and Shevlin,
7

I discuss the control for these governance change effects in Sections 5.4 and 5.5.1.

2001; Chen et al., 2010). Desai and Dharmapala (2006) find that the increased use of incentive compensation for managers reduces tax sheltering activity, which is consistent with the existence of strong complementarities between tax sheltering activity and rent diversion by managers. Desai et al. (2007) develop a model in which corporate tax sheltering and the diversion of rents by managers are interrelated. Strong complementarities may exist between the two activities, because concealing income from tax authorities through complex transactions reduces the ability of shareholders to monitor managerial behavior, thereby making the diversion less costly for managers. Desai and Dharmapala (2009) further examine the effect of corporate tax avoidance on firm valuation in relation to corporate governance. They find that the average effect of tax avoidance on firm value is not significantly different from zero, although it is positive for well-governed firms, which indicates that a higher quality of corporate governance, measured as a higher level of institutional ownership, leads to a favorable effect of tax avoidance on firm value. Wilson (2009) finds that active tax shelter firms with a strong corporate governance exhibited positive abnormal returns. As a result, Wilson argues that tax sheltering is a tool for wealth creation in well-governed firms. Shackelford and Shevlin (2001) point out that the managerial, or insider, control of a firm is potentially an important determinant of tax aggressiveness. Chen et al. (2010) investigate a unique agency conflict between dominant and minority shareholders in firms owned by founding family members. They find that family firms are less tax aggressive than their non-family counterparts, stating that family owners are willing to forgo tax benefits to avoid the non-tax cost of a potential price discount. As discussed previously, during the beginning of the financial crisis in 1997, chaebols, such as the Daewoo Group, Kia Group, and Hanbo Group, went bankrupt in Korea. This event caused debtholders to suffer from bankruptcy or experience a debt-equity swap, thereby causing them to become vulnerable to managerial malfeasance or diversions. Furthermore, because of the weak governance in Korea, which controlling shareholders used to exploit outside investors, debtholders became interested in mitigating agency conflicts between controlling shareholders and debtholders through the implementation of good corporate governance practices in Korea. Thus, tax avoidance could reduce the transparency of Korean firms, which enables controlling shareholders with the opportunity to extract rents from debtholders, thereby resulting in a higher cost of debt. Agency theory suggests that a well-designed corporate governance system can reduce the possibility of information asymmetry and managerial rent diversion, thereby resulting in debtholders discounting the future firm value at a lower rate of return. Ashbaugh-Skaifea et al. (2006) explain debt pricing on the basis of agency theory. They reveal that bondholders, and more generally, debt stakeholders, face two types of agency conflicts that increase the probability of default. First, agency conflict occurs between management and the bondholders, since selfish managerial behavior induces information asymmetry and creates moral hazard problems. The second agency conflict faced by bondholders occurs between shareholders and bondholders, since shareholders in leveraged firms are provided with incentives to make decisions that transfer wealth from the bondholders to themselves. Anderson et al. (2003) indicate that bondholders view establishing family ownership as an organizational structure. This structure results in fewer agency conflicts between equity and debt claimants. However, by using a multi-national sample of firms, Boubakri and Ghouma (2010) find that ultimate ownership and family control have a significantly positive effect on bond yield-spreads, and a significantly negative effect on bond ratings.

460

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470 Table 1 Sample. Panel A: Sample selection summary Number of firm-years Cost of debt sample Number of listed firms in Korea Stock Exchange Less: financial and insurance firms Less: non-December year-end firms Less: firms with unclean audit opinion Less: firms without relevant data Less: firms with less than 10 firms in an industry each year Final sample From 1994 to 1998 From 1999 to 2003 Firm value sample Number of listed firms in Korea Stock Exchange Less: financial and insurance firms Less: non-December year-end firms Less: firms with unclean audit opinion Less: firms without relevant data Less: firms with less than 10 firms in an industry each year Final sample From 1994 to 1998 From 1999 to 2003 Panel B: Industry distribution of cost of debt sample Industry description Manufacture of rubber and plastic products Manufacture of other machinery and equipment Manufacture of other transport equipment Manufacture of electrical machinery and apparatuseses n.e.c. Wholesale trade and commission trade, except of motor vehicles and motorcycles Manufacture of sewn wearing apparel and fur articles Manufacture of other non-metallic mineral products Manufacture of textiles, except sewn wearing apparel Retail trade, except motor vehicles and motorcycles Land transport; transport via pipelines Manufacture of food products and beverages Manufacture of medical, precision and optical instruments, watches and clocks Manufacture of motor vehicles, trailers and semitrailers Electricity, gas, steam and hot water supply Professional, scientific and technical services Manufacture of electronic components, radio, television and communication equipment and apparatuses Manufacture of basic metals Manufacture of fabricated metal products, except machinery and furniture General construction Manufacture of pulp, paper and paper products Manufacture of chemicals and chemical products Total Number of firm-years 88 139 6 115 181 96 193 79 27 54 234 12 226 57 6 247 Percentage 2.9 4.5 0.2 3.7 5.9 3.1 6.3 2.6 0.9 1.8 7.6 0.4 7.3 1.9 0.2 8.0 5530 (432) (444) (195) (953) (426) 3080 1352 1728 5530 (432) (444) (195) (1575) (447) 2437 1031 1406

The primary measure of the quality of shareholder activism is the level of institutional ownership (Desai and Dharmapala, 2009). The basic motivation for this is that institutional investors have greater incentives and a greater capacity to monitor managerial performance (e.g., Shleifer and Vishny, 1986; Chung et al., 2002; Bhojraj and Sengupta, 2003; Hartzell and Starks, 2003; Desai and Dharmapala, 2009).8 Shleifer and Vishny (1986) argue that institutional shareholders, by virtue of their large stockholdings, possess greater incentives to monitor corporate performance. Chung et al. (2002) find that when institutional investors own a large percentage of outstanding shares in a firm, there is less use of discretionary accruals; this indicates that agency problems between managers and shareholders decrease with an increase in institutional investor ownership. Bhojraj and Sengupta (2003) suggest that larger institutional ownership proportions are associated with lower yields on new bond issues. Hartzell and Starks (2003) provide empirical evidence that suggests that institutional investors play a monitoring role in regard to the executive compensation contracts. Desai and Dharmapala (2009) use institutional ownership as the primary quality of corporate governance. I anticipate that institutional ownership has a negative effect on the COD and that it further amplifies the negative effect of tax avoidance on the COD by alleviating the agency costs between controlling shareholders and bondholders, as well as decreasing opportunities for managerial rent diversions related to tax avoidance. This argument leads to the following hypothesis: H2. The negative effect of tax avoidance on the cost of debt increases with the level of institutional ownership. 3. Sample selection Sample firms are selected from companies listed on the Korean Stock Exchange between 1994 and 2003. Initially, a total of 5530 firm-year observations are obtained from the KIS value database.9 I exclude 432 financial and insurance observations from the sample, because of a difference in their financial characteristics. Thereafter, non-December year-end firms are excluded, because their tax change effects were different. Furthermore, I restrict the sample to firms with unqualified audit opinions to enhance the credibility of the financial statement information used in the tests. I also exclude those firms whose financial information was not available during the sample period from 1994 to 2003. Firms with less than 10 member firms each year are also excluded, because discretionary accruals are estimated for each industry and each year using the cross-sectional modified Jones model, as in Dechow et al. (1995) and Kothari et al. (2005). These procedures result in the cost of debt sample, comprised of 3080 firm-years, as reported in Panel A of Table 1. The cost of debt sample are comprised of 1352 firm-years (43.9%) from 1994 to 1998 and 1728 firm-years (56.1%) from 1999 to 2003. The firm value sample are comprised of 2437 firm-years, including 1031 (42.3%) from 1994 to 1998 and 1406 (57.7%) from 1999 to 2003. Panel B of Table 1 describes the industry distribution of the cost of debt sample according to the 2-digit industry classification code of the Korea National Statistical Office. Although there are a greater proportion of firms in industries such as chemicals (20.4%), general construction (9.0%), basic metals (8.2%), and electronic devices (8.0%), firms are relatively evenly distributed across all industries, indicating no significant industry bias.
Brockman and Yan (2009) examine the impact of block ownership on the firm’s information environment and show that block ownership plays a significant role in shaping the firm’s information environment. 9 The KIS value database is provided by the Korean Investors Service Inc., which is affiliated with Moody’s.
8

254 45 278 116 627 3080

8.2 1.5 9.0 3.8 20.4 100

Note: The industry classification is based on 2-digit SIC code from the Korea National Statistical Office.

4. Regression variables This section describes the dependent and explanatory variables used in the empirical tests. I focus on the specification of three key variables—tax avoidance, the cost of debt, and shareholder activism. I also explain the other control variables that affect the cost of debt.

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

461

4.1. Tax avoidance (TA; TA_mod; TA_per) Since book-tax difference (hereafter, BTD) subtracts taxable income from financial income, by definition, BTD can be increased by either (1) the opportunistic increase of financial income (earnings management) or (2) the intentional decrease of taxable income (tax avoidance). Frank et al. (2009) suggest that nonconformity between financial accounting standards and tax law allows firms to manage book income upward and taxable income downward in the same reporting period. I derive my measure of tax avoidance by modifying the formula of Desai and Dharmapala (2006). They use total accruals to isolate the component of the BTD that is attributable to earnings management. The orthogonal component of BTD that cannot be explained by earnings management was inferred to be a measure of tax avoidance activity. Unlike Desai and Dharmapala (2006), I use discretionary, rather than total accruals, since discretionary accruals are a more refined earnings management proxy, according to Dechow et al. (1995) and Kothari et al. (2005). The procedure for calculating tax avoidance is comprised of two steps. The first step is to estimate discretionary accruals. I first calculate total accruals for each firm in my sample for each year over the sample period. Then, I obtain discretionary accruals, DA_modi,t for each firm i in each year t, following Dechow et al. (1995). The discretionary accruals (DA_mod) are calculated as the residuals from

management). This can be interpreted as a measure of tax avoidance.10 I denote this measure as TA_mod (or TA_per):

TA modðor TA perÞi;t ¼ uj þ ei;t

ð4Þ

where TA_mod is obtained using discretionary accruals, as in Eq. (1); and TA_per is obtained using the performance-matched discretionary accruals, as in Eq. (2). In the analysis, I use the TA_mod (or TA_per)i,t estimated from Eq. (4), as the test variable representing tax avoidance.11 4.2. Cost of debt (COD) The cost of debt is the interest rate on the debt of the firm, calculated as the interest expense for the year divided by the average short- and long-term debt during the year (Pittman and Fortin, 2004). 4.3. Shareholder activism: institutional ownership (Insti) The primary measure of shareholder activism is institutional ownership (Insti), calculated as a fraction of the shares of the firm that are owned by the institutional investors (Desai and Dharmapala, 2009). I predict a negative relationship between the cost of debt and the institutional ownership, since institutional investors would reduce the cost of debt by alleviating agency costs. Also, I measure shareholder activism by the exogenous variation of shareholder rights related to institutional investors in 1998.12 4.4. Control variables I include control variables that affect debt pricing, as in Pittman and Fortin (2004). The theory of reputation formation in debt markets, given by Diamond (1989), predicts that interest rates will decline over time as firms compile good credit histories. I measure firm age as the number of years since the firm went public. Auditor quality is specified as Big6, a variable indicating whether the firm retains an auditor affiliated with a Big Six auditor in the US.13 I also predict that interest rates have a positive relationship with leverage, defined as the short- and long-term debt scaled by the market value of the firm (Petersen and Rajan, 1994). The control for profitability is cash flow from operations (Petersen and Rajan, 1994), which is predicted to have a negative coefficient, since firms that can generate more cash from operations are in a better position to service their debts. I predict an inverse relationship between interest rates and firm size, which is measured as the natural logarithm of total assets, since creditors perceive larger firms as less risky and there are economies of scale in debt production costs (Carey et al., 1993). Prior evidence suggests that interest rates have a positive relationship with collateral (John et al., 2003), which is consistent with the perception in the banking industry that riskier borrowers must provide security for their loans (Morsman, 1986). This implies that the coefficient on the control variable for asset structure, the total property, plant, and equipment scaled by assets, will be positive. I then specify a dummy variable,
10 Since I exclude the precise earnings management component in BTD, my tax avoidance measure for the NDTS could be more accurate than the tax spread used in Desai and Dharmapala (2006). 11 The two-stage estimation process of tax avoidance can possibly induce certain measurement errors, even though I estimate tax avoidance more accurately and check the validity of the measure. 12 The Korean government reform to strengthen shareholder rights of institutional investors is explained in Section 2.1. 13 At the end of 2003, the Big six auditors in Korea were Samil (PricewaterhouseCoopers), Younghwa (Ernst and Young), Samjong (KPMG), Anjin (Deloitte Touche Tohmatsu), Ankun (Deloitte Touche Tohmatsu), and Hana (Deloitte Touche Tohmatsu).

Accrualsijt =AssetsijtÀ1 ¼ ajt ð1=AssetsijtÀ1 Þ þ b1jt fðDSALEijt À DA=Rijt Þ=AssetsijtÀ1 g þ b2jt ðPPEijt =AssetsijtÀ1 Þ þ eijt ð 1Þ

where Accrualsijt are the total accruals of firm i in industry j in year t, calculated as ordinary income minus the cash flow from operations; AssetsijtÀ1 are the total assets of firm i in industry j in year t À 1; DSALEijt is the change in the sales of firm i in industry j in year t; DA/Rijt is the change in account receivables of firm i in industry j in year t; and PPEijt represents the property, plant, and equipment (PPE) of firm i in industry j in year t. I also use performance-matched discretionary accruals (DA_per) as an alternative measurement for discretionary accruals. Following Tucker and Zarowin (2006), DA_per is calculated as the residual from Eq. (2), as the regression-based approach (Kothari et al., 2005).

Accrualsijt =AssetsijtÀ1 ¼ ajt ð1=AssetsijtÀ1 Þ þ b1jt fðDSALEijt À DA=Rijt Þ=AssetsijtÀ1 g þ b2jt ðPPEijt =AssetsijtÀ1 Þ þ b3ijt ROAijt þ eijt ð 2Þ
where ROAijt is the net income of firm i in industry j in year t scaled by the lagged total assets. The second step is to isolate the component of the BTD that is not attributable to earnings management and identify the tax avoidance component. As a result, I run the following ordinary least squares (OLS) regression:

BTDi;t ¼ b1 DA modðor DA perÞi;t þ uj þ ei;t

ð 3Þ

where BTDi,t is the book-tax difference for firm i in year t scaled by lagged total assets; DA_mod (or DA_per)i,t represents the discretionary accruals for firm i in year t scaled by the lagged total assets; uj is the average value of the residual for firm i over the sample period; and ei,t is the deviation from the average residual uj of firm i in year t. The residual from Eq. (3) is the component of BTD that cannot be explained by variations in discretionary accruals (earnings

462

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

Negequity, to identify firms with negative book values of common equity. These firms experience financial distress and generally incur higher borrowing costs (Graham et al., 1998). It is not entirely clear whether chaebols in Korea perform a function that is conducive for reducing the cost of debt. Shin and Park (1999) illustrate that, due to their internal capital markets, chaebols belonging to the top 30 business groups are subject to fewer financing constraints than other Korean firms, thereby implying a lower cost of debt.14 However, chaebols are characterized by an extensive arrangement of pyramidal, or multi-layered, shareholding agreements and the existence of cross-debt guarantees among member firms, thereby suggesting a higher cost of debt (Baek et al., 2004). In response to the Korea Fair Trade Commission classification recommendations, I further control for the effect of chaebols on the cost of debt as a chaebol dummy variable, which is determined annually (Bae et al., 2002; Baek et al., 2004), to test the impact of shareholder activism on the relationship between tax avoidance and the cost of debt. 5. Test methodology and results This section explains the research design used for testing the hypotheses and the test results. 5.1. Descriptive statistics Fig. 1 presents the time series data of the cost of debt (COD), Tobin’s Q and tax avoidance from 1994 to 2003. Due to the financial crisis that began in 1997, there was a huge temporary increase in the COD in 1998; however, after 1998, it declined gradually. Tax avoidance activities (TA,15 TA_mod, and TA_per) decreased in 1998 and 2001.16 Tobin’s Q illustrates a slight decline. Panel A of Table 2 presents the descriptive statistics for the variables used in the analysis. The mean (median) BTD is À0.008 (À0.005), suggesting that taxable income is greater than financial income.17 The mean (median) value of COD is 0.062 (0.060). The mean (median) values of TA, TA_mod, and TA_per are 0 (0.002), 0 (0.002), and 0 (0.003), respectively, which confirms that TA_mod (or TA_per) is constrained by the regression procedure to sum to zero over all firms and all years as a residual (Desai and Dharmapala, 2006).18 The mean (median) value of Tobin’s Q is 0.880 (0.836).19 Panel B of Table 2 presents the correlations among COD, tax avoidance, and other control variables. The negative and significant correlation between the COD and tax avoidance (TA, TA_mod, and

Fig. 1. Cost of debt, Tobin’s Q and tax avoidance. Notes: The variables are defined in Table 2.

TA_per) suggests that firms that indulge in tax avoidance activities experience a lower COD, which is consistent with the trade-off hypothesis. Consistent with prior research, I find positive and significant correlations between the COD and the variables Leverage (0.4), Size (0.03), Negequity (0.1), and Chaebol (0.09). In contrast, there are negative and significant correlations between COD and the variables Age (À0.1), CFO (À0.2), Asset (À0.05), and Insti (À0.06). 5.2. Characteristics of tax avoidance Following Desai and Dharmapala (2006), I decompose BTD into two components—earnings management and tax avoidance. As in the US, BTD is comprised of temporary and permanent differences in Korea.20 Since deferred tax accounting was introduced after 1998, I was able to obtain information pertaining to temporary and permanent differences and analyze the characteristics of tax avoidance from 1999 to 2003. The unreported Pearson correlations indicate that there are positive and significant correlations between tax avoidance and the temporary and permanent differences, while there is only a positive and significant correlation between earnings management and temporary differences. Panel B of Table 3 provides the test results using a fixed-effects model, indicating that, for the
20 The difference between the financial and taxable income revenue and expense recognition policy gives rise to temporary differences. In contrast, permanent differences arise when revenue or expenses are recognized under one system, but not the other.

14 The Korea Fair Trade Commission (KFTC) defines a business group as a group of companies with over 30% of the shares owned by the controlling shareholder of the group and its affiliated companies. 15 Following Desai and Dharmapala (2006), I use total accruals as an earnings management proxy. I denoted the tax avoidance measure using total accruals as TA. The results using TA are similar to those based on the TA_mod (or TA_per). 16 Desai and Dharmapala (2006) contend that the TA_mod and TA_per cannot be aggregated across all firms to obtain a measure of aggregate tax sheltering in the economy in year t. However, TA_mod and TA_per are constrained by the regression procedure to sum to zero over all firms and all years as a residual. TA_mod and TA_per can be used to examine whether the aggregate amount of tax avoidance has grown over time. I also estimate the tax avoidance in Eq. (3) for firms in each industry for each year. The results are almost identical to the main results. 17 For robustness, I test 1094 firm-years with a positive BTD. The unreported results for each hypothesis are the same. 18 According to Desai and Dharmapala (2006), TA_mod (or TA_per)i,t is constrained by the regression procedure to sum to zero over all firms and all years as a residual. In the meanwhile, ei,t was constrained to sum to zero for firm i over all years. Therefore, neither TA_mod (or TA_per)i,t nor its components uj and ei,t, can be interpreted as the dollar amount of tax avoidance by firm i in year t. Since I use panel data regressions with firm fixed effects, I only require a measure that adequately proxies for variations in tax avoidance within a firm over time. 19 These figures are consistent with 0.85 (0.81) in 2001 of Black et al. (2006).

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470 Table 2 Summary statistics. Panel A: Descriptive statistics Variable BTD Financial income Taxable income Accruals DA_mod DA_per COD TA TA_mod TA_per Insti Insti_sen Insti_insen Age Big6 Leverage CFO Size Asset Negequity Chaebol Tobin’s Q Ln(Age) Ownership Ownership_squ Size Volatility Tax rate change Bondrating Panel B: Pearson correlation COD COD TA TA_mod TA_per Age Big6 Leverage CFO Size Asset Negequity Insti Chaebol 1.00 TA À0.05ÃÃÃ 1.00 TA_mod À0.05ÃÃÃ 1.00ÃÃÃ 1.00 TA_per À0.06ÃÃÃ 0.99ÃÃÃ 1.00ÃÃÃ 1.00 Age À0.10ÃÃÃ À0.01 À0.02 À0.02 1.00 Big6 À0.02 0.02 0.01 0.01 0.08ÃÃÃ 1.00 Leverage 0.40ÃÃÃ 0.10ÃÃÃ 0.08ÃÃÃ 0.06ÃÃÃ 0.05ÃÃÃ 0.06ÃÃÃ 1.00 CFO À0.20ÃÃÃ 0.13ÃÃÃ 0.09ÃÃÃ 0.05ÃÃÃ 0.06ÃÃÃ 0.07ÃÃÃ À0.14ÃÃÃ 1.00 Size 0.03ÃÃ 0.02 0.01 0.01 0.31ÃÃÃ 0.20ÃÃÃ 0.12ÃÃÃ 0.11ÃÃÃ 1.00 Asset À0.05ÃÃÃ 0.02 0.00 À0.01 0.22ÃÃÃ 0.08ÃÃÃ 0.04ÃÃ 0.13ÃÃÃ 0.18ÃÃÃ 1.00 Negequity 0.10ÃÃÃ À0.04ÃÃ À0.05ÃÃÃ À0.06ÃÃÃ 0.01 0.00 0.34ÃÃÃ À0.09ÃÃÃ À0.03Ã 0.03Ã 1.00 Insti À0.06ÃÃÃ 0.09ÃÃÃ 0.09ÃÃÃ 0.09ÃÃÃ 0.09ÃÃÃ 0.15ÃÃÃ 0.00 0.16ÃÃÃ 0.53ÃÃÃ 0.13ÃÃÃ À0.08ÃÃÃ 1.00 Number of firm-years 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 3080 2437 2437 2437 2437 2437 2437 3080 947 Mean À0.008 0.038 0.045 À0.023 0.000 0.000 0.062 0.000 0.000 0.000 0.389 0.078 0.312 16.579 0.637 0.336 0.053 19.238 0.365 0.019 0.222 0.880 3.533 0.205 0.058 19.287 0.092 0.985 9.001 Median À0.005 0.030 0.037 À0.023 0.001 0.000 0.060 0.002 0.002 0.003 0.371 0.036 0.278 15.001 1.000 0.311 0.052 19.016 0.366 0.000 0.000 0.836 3.529 0.179 0.032 19.071 0.086 1.000 9.000 Standard deviation 0.077 0.103 0.097 0.099 0.078 0.068 0.032 0.077 0.077 0.077 0.238 0.238 0.213 11.134 0.481 0.257 0.089 1.349 0.177 0.135 0.416 0.311 0.330 0.125 0.077 1.334 0.032 0.790 3.126 Min. À0.969 À1.059 À0.892 À0.932 À0.586 À0.355 0.000 À0.983 À0.968 À0.962 0.000 0.000 0.000 0.001 0.000 0.000 À0.765 16.335 0.001 0.000 0.000 0.216 1.391 0.000 0.000 16.518 0.038 0.000 1.000 Max.

463

0.979 1.739 1.353 0.553 0.459 0.461 0.211 1.007 1.003 0.987 1.000 0.888 1.000 105.043 1.000 5.653 0.613 24.757 0.911 1.000 1.000 5.519 4.673 0.886 0.784 24.757 0.269 3.000 21.000

Chaebol 0.09ÃÃÃ 0.03Ã 0.03 0.02 0.12ÃÃÃ 0.18ÃÃÃ 0.13ÃÃÃ 0.03Ã 0.58ÃÃÃ 0.11ÃÃÃ À0.03Ã 0.43ÃÃÃ 1.00

Notes: This table presents the summary statistics for the 3080 cost of debt sample and the 2437 firm value sample over the period of 1994–2003. Panel A presents the distributional statistics for the variables. Panel B reports the Pearson correlations for the regression variables. BTD is the book-tax difference that subtracts taxable income from financial income, scaled by the lagged total assets. Financial income is the pretax book income from the income statement scaled by the lagged total assets. Taxable income is income according to the Korean corporate tax code, reported in the notes of the annual reports scaled by the lagged total assets. Accruals are the total accruals, which were calculated as ordinary income minus cash flow from operations scaled by the lagged total assets. DA_mod is the discretionary accruals measured following Dechow et al. (1995). DA_per is the performance-matched discretionary accruals measured following Kothari et al. (2005). COD is the interest rate on the firm’s debt calculated as its interest expense for the year divided by its average short- and long-term debts during the year. TA is the tax avoidance measure, as in Desai and Dharmapala (2006). TA_mod (or TA_per) is the tax avoidance measure modified, from Desai and Dharmapala (2006). Insti is the percentage of equity shares held by institutional investors at the end of the year. Insti_sen is the percentage of equity shares held by pressure-sensitive shareholders (government, security firms, and foreign investors) at the end of the year. Insti_insen is the percentage of equity shares held by pressure-insensitive shareholders (banks and insurance companies) at the end of the year. Age is the number of years since the firm went public. Big6 is an indicator variable which represents 1 for the auditor affiliated with a Big6 auditor in the US, and 0 otherwise. Leverage is the total short-and long-term debts scaled by the firm market value. CFO is the Cashflow from operations divided by total assets. Size is the natural logarithm of total assets. Asset is the total property, plant, and equipment scaled by total assets. Negequity, a variable representing the book value of common equity, is one when it is negative, and 0 otherwise. Chaebol is a dummy variable that is set to 1 if a firm is affiliated with a Chaebol, and 0 otherwise. Tax rate change represents 3 for 1994, 2 for 1995, 1 for 1996– 2001 and 0 for 2002–2003. Bondrating is measured as AAA = 1, AA+ = 2, AA = 3, AAÀ = 4, A+=5, A = 6, AÀ = 7, BBB+ = 8, BBB = 9, BBBÀ = 10, BB+ = 11, BB = 12, BBÀ = 13, B+ = 14, B = 15, BÀ = 16, CCC+ = 17, CCC = 18, CCCÀ = 19, CC = 20, C = 21 and D = 22 (Francis et al., 2005). Tobin’s Q is estimated as the market value of assets as the (book value of debt + book value of preferred stock + market value of common stock)/book value of assets. Ownership is the ownership by the largest shareholders. Ownership_squ is the square value of Ownership. Ln(Age) is the logarithm of the Age variable. Volatility is the standard deviation of the monthly stock returns calculated over a 60-month period. The superscript asterisks in Panel B indicate the significance at p-values less than 0.10(*), 0.05(**), and 0.01(***).

regression of temporary differences, both tax avoidance and earnings management have positive and significant coefficients, while for the regression of permanent differences, only tax avoidance has positive and significant coefficients. These findings indicate that firms avoid tax liability by using both temporary and permanent differences,

while they manage earnings by primarily utilizing the temporary differences component.21
21 Wilson (2009) finds that five of the eight tax shelters generate permanent differences and the three tax shelters generate temporary differences.

464 Table 3 Characteristics of tax avoidance. Panel A: Summary statistics Variables Temporary difference Permanent difference TA_mod TA_per DA_mod DA_per

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

Number of firm-years 1728 1728 1728 1728 1728 1728

Mean À0.001 À0.004 0.001 0.001 0.000 0.000

Standard deviation 0.083 0.065 0.095 0.095 0.080 0.067

Min. À0.703 À0.628 À0.968 À0.962 À0.586 À0.355

Max. 0.803 0.758 1.003 0.987 0.459 0.461

Panel B: Relationship between temporary and permanent differences and tax avoidance and earnings management Temporary difference Intercept TA_mod TA_per DA_mod DA_per Firm effect Year effect F-value Adj R-squ Number of observations Included Included 3.21ÃÃ 0.469 1728 0.082ÃÃÃ (4.49) 0.029 (1.71) Included Included 3.21ÃÃ 0.468 1728
Ã

Permanent difference À0.001 (À0.30) À0.005Ã (À1.65) 0.301 ÃÃÃ (20.46) À0.005Ã (À1.65)

À0.001 (À0.30) 0.589ÃÃÃ (38.32)

0.591ÃÃÃ (38.61) À0.002 (À0.10)

0.299ÃÃÃ (20.40)

Included Included 5.84ÃÃÃ 0.206 1728

À0.051ÃÃ (À2.46) Included Included 5.85ÃÃÃ 0.207 1728

Notes: The variables, except for temporary and permanent differences, are defined in Table 2. Temporary and permanent differences are measured as the temporary and permanent differences from the notes of auditor’s report scaled by lagged total assets from 1999 to 2003, since the deferred tax accounting was introduced since 1998 in Korea. Each cell exhibits a coefficient (t-statistic), respectively. The t-statistics are computed using the fixed-effects model controlling for firm effect and year effect. The superscript asterisks indicate the explanatory variable coefficient significance at p-values less than 0.10 (Ã), 0.05 (ÃÃ), and 0.01 (ÃÃÃ).

5.3. Does tax avoidance have a negative relationship with the cost of debt? Pooled OLS results may be driven by an omitted variable problem in that omitted firm-specific correlated variables could be a part of the explanatory variables. Thus, I estimate the following fixed-effects model with a correction for unspecified heteroskedasticity to examine the impact of tax avoidance on the borrowing costs of firms22

Cost of debti;t ¼ a1i þ a2 TA modðor TA perÞi;t þ a3 Agei;t þ a4 Big6i;t þ a5 Leveragei;t þ a6 CFOi;t þ a7 Sizei;t þ a8 Asseti;t þ a9 Negequityi;t þ Firm effect þ Year effect ð 5Þ

where COD is the interest rate on the debt of the firm, calculated as its interest expense for the year divided by its average short- and long-term debt during the year; TA_mod (or TA_per) is the tax avoidance measure modified from Desai and Dharmapala (2006); Age is the number of years since the firm went public; Big6 is 1 if the auditor is affiliated with a Big Six auditor in the US, and 0 otherwise; Leverage is the total short- and long-term debt scaled by the market value of the firm; CFO is cashflow from operations divided by the total assets; Size is the natural logarithm of total assets; Asset is total property, plant, and equipment scaled by assets; Negequity is 1 when the book value of common equity is negative, and 0 otherwise. This generates results that are directly comparable to prior research on the determinants of the interest rates of firms. The results in Table 4 are similar to those provided by prior cross22 I also re-estimate the results using standard errors adjusted for clustering at the firm level. The results are qualitatively similar to those reported in the paper, albeit with a decline in the t-statistics.

sectional studies (Petersen and Rajan, 1994; Pittman and Fortin, 2004).23 This test also provides the first cross-sectional evidence on whether tax avoidance affects debt pricing.24 The negative and significant coefficient of À0.023 (À0.026) on the TA_mod (or TA_per) variable implies that higher tax avoidance activities reduce the interest rates of firms by 2.3% (2.6%), on average, which is consistent with the trade-off hypothesis. This suggests that tax avoidance is related, both statistically and economically, to a lower COD, once leverage is explicitly controlled for in the regressions. This means that the tax-favored effect of tax avoidance dominates managerial opportunism in the core relationship with tax avoidance and the COD. The estimate for the Big6 is not statistically significant, indicating that audit quality of the Big6 and the NonBig6 is not significantly different in Korea, which is inconsistent with the evidence pertaining to the US found in Pittman and Fortin (2004). The coefficient of Leverage was positive and significant and the coefficient on CFO is negative and significant, which is consistent with Petersen and Rajan (1994) and Pittman and Fortin (2004). 5.4. Influence of shareholder activism on the relationship between tax avoidance and the cost of debt In this section, I investigate the effects of shareholder activism for tax avoidance activities on COD. I examine whether the effect of tax avoidance on COD increase with the level of institutional ownership. This is tested using an interaction variable between
23 Following Pittman and Fortin (2004), I control for the underlying cost of capital of the risk-free rate and the annual difference in the yield between BBB + rated corporate bonds and 10-year government bonds, as additional control variables in Eqs. (5) and (6) from 2000 to 2003. The unreported results are similar to those previously reported in Tables 4 and 5. 24 The F-test rejects the null hypothesis that there are no fixed effects. The Hausman tests confirm that the fixed-effects model is the appropriate design choice for examining the hypotheses.

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470 Table 4 Regression analysis of tax avoidance on cost of debt. Prediction Intercept TA TA_mod TA_per Age Big6 Leverage CFO Size Asset Negequity Firm effect Year effect F-value Adj R-squ Number of observations ? À À À À À + À + + + 0.000 (1.18) 0.001 (1.44) 0.042ÃÃÃ (22.91) À0.042ÃÃÃ (À8.26) 0.001 (1.55) À0.001 (À0.36) À0.012ÃÃÃ (À3.60) Included Included 153.57ÃÃÃ 0.451 3080 0.000 (1.18) 0.001 (1.44) 0.042ÃÃÃ (22.89) À0.043ÃÃÃ (À8.44) 0.001 (1.53) À0.001 (À0.42) À0.012ÃÃÃ (À3.63) Included Included 153.36ÃÃÃ 0.451 3080 Fixed-effects: Eq. (5) 0.013ÃÃ (1.98) À0.023ÃÃÃ (À4.08) 0.013ÃÃ (2.02) 0.013ÃÃ (2.01)

465

Table 5 Regression analysis of influence of shareholder activism on tax avoidance and cost of debt. Panel A: Cost of debt Prediction Total period 0.004 (0.51) 0.004 (0.35) À0.059ÃÃÃ (À2.92) À0.014ÃÃÃ (À6.10) 0.000 (0.92) 0.002Ã (1.66) 0.042ÃÃÃ (22.70) À0.039ÃÃÃ (À7.65) 0.001ÃÃÃ (2.72) À0.001 (À0.24) À0.013ÃÃÃ (À3.83) 0.003ÃÃÃ (2.58) Included Included 155.92ÃÃÃ 0.460 3080 From 1994 to 1998 0.093ÃÃÃ (7.15) À0.064ÃÃÃ (À2.72) 0.044 (0.82) À0.013ÃÃÃ (À3.37) 0.000ÃÃ (À2.07) 0.002Ã (1.81) 0.057ÃÃÃ (14.18) À0.05ÃÃÃ (À6.64) À0.000 (À0.66) À0.008Ã (À1.94) À0.047ÃÃÃ (À8.09) 0.000 (0.13) Included Included 26.02ÃÃÃ 0.277 1352 From 1999 to 2003 À0.026ÃÃÃ (À2.62) 0.007 (0.80) À0.064ÃÃÃ (À3.84) À0.015ÃÃÃ (À5.70) 0.000ÃÃÃ (2.71) 0.000 (À0.09) 0.038ÃÃÃ (19.45) À0.025ÃÃÃ (À3.87) 0.003ÃÃÃ (4.75) 0.006Ã (1.09) 0.006 (1.51) 0.005ÃÃÃ (3.12) Included Included 88.46ÃÃÃ 0.384 1728

À0.023ÃÃÃ (À4.13) À0.026ÃÃÃ (À4.61) 0.000 (1.16) 0.001 (1.45) 0.042ÃÃÃ (22.92) À0.044ÃÃÃ (À8.68) 0.001 (1.57) À0.001 (À0.42) À0.013ÃÃÃ (À3.70) Included Included 153.40ÃÃÃ 0.452 3080

Intercept TA_mod TA_mod  Insti Insti Age Big6 Leverage CFO Size Asset Negequity Chaebol Firm effect Year effect F-value Adj R-squ Number of observations À À À À À + À + + + +

Notes: All the variables are defined in Table 2. Each cell exhibits a coefficient (tstatistic), respectively. The t-statistics are computed using the fixed-effects model controlling for firm effect and year effect. The superscript asterisks indicate the explanatory variable coefficient significance at p-values less than 0.10 (Ã), 0.05 (ÃÃ), and 0.01 (ÃÃÃ).

Panel B: Control for other governance change effects

tax avoidance and institutional ownership (Insti), as shown in Eq. (6). Institutional ownership and Chaebol are also included as control variables.

Large firms Intercept TA_mod TA_mod  Insti TA_mod  Insti  Dummy for large firms Insti À À N/A À À À + À + + + + À0.007 (À0.66) À0.000 (À0.28) À0.011 (À1.50) 0.071ÃÃà (6.00) À0.055ÃÃà (À2.76) À0.001 (À0.40) 0.023 (2.29) 0.000 (0.31) 0.010à (1.84) Included Included 17.24ÃÃà 0.676 145 0.037 (0.71) 0.224à (1.77) À0.346Ãà (À2.24)

Small firms À0.037ÃÃÃ (À3.05) 0.006 (0.54) À0.057ÃÃ (À2.62)

From 1999 to 2003 À0.027ÃÃÃ (À2.66) 0.008 (0.76) À0.062ÃÃÃ (À3.70) À0.023 (À0.57) À0.015ÃÃÃ (À5.64) 0.000ÃÃÃ (2.70) 0.000 (À0.09) 0.038ÃÃÃ (19.44) À0.025ÃÃÃ (À3.89) 0.003ÃÃÃ (4.78) 0.006Ã (1.89) 0.006 (1.52) 0.005ÃÃÃ (3.10) Included Included 88.25ÃÃÃ 0.386 1728

Cost of debti;t ¼ b1i;t þ b2 TA modðor TA perÞi;t þ b3 TA modðor TA perÞi;t  Instii;t þ b4 Instii;t þ b5 Agei;t þ b6 Big6i;t þ b7 Leveragei;t þ b8 CFOi;t þ b9 Sizei;t þ b10 Asseti;t þ b11 Negequityi;t þ b12 Chaebol þ Firm effect þ Year effect ð 6Þ

Age Big6 Leverage CFO Size Asset Negequity Chaebol Firm effect Year effect F-value Adj R-squ Number of observations

where Insti is the sum of ownership by government, banks, security companies, insurance companies, and foreign investors, except for individual investors, at the end of year t; Chaebol is a dummy variable that is set to 1 if a firm is affiliated with a Chaebol, and 0 otherwise. Column 3 (total period) of Panel A in Table 5 indicates that the coefficient for the interaction between TA_mod and Insti is À0.059 and significant.25 This indicates that the effect of tax avoidance on the COD increases marginally with institutional ownership by 5.9% over the entire period. Tax avoidance creates managerial rent diversions which are mitigated in firms well-covered by institutional investors. Trade-off theory does not predict a negative influence of Insti on the relationship between tax avoidance and the COD, thereby indicating that the managerial opportunism theory reveals an additional explanation for tax avoidance activities. The negative coefficient estimate, À0.014, of the Insti variable is consistent with Bhojraj and Sengupta (2003) that show the lower bond yields and higher ratings for firms with higher institutional
25 I also perform all of the main tests using TA_per as tax avoidance. The results are similar to the reported results.

À0.014ÃÃÃ (À5.29) 0.000ÃÃÃ (2.63) 0.000 (0.21) 0.037ÃÃÃ (18.18) À0.020ÃÃÃ (À2.93) 0.003ÃÃÃ (4.89) 0.003 (0.84) 0.008Ã (1.95) 0.004ÃÃ (2.36) Included Included 71.70ÃÃÃ 0.360 1583

Notes: All the variables are defined in Table 2, except for Dummy for large firms which equals to one if assets >2 trillion Won and zero otherwise. Each cell exhibits a coefficient (t-statistic), respectively. The t-statistics are computed using the fixedeffects model controlling for firm effect and year effect. The superscript asterisks indicate the explanatory variable coefficient significance at p-values less than 0.10 (Ã), 0.05 (ÃÃ), and 0.01 (ÃÃÃ).

466

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

ownership. Table 5 illustrates that the coefficient on the TA_mod for the total period ceases to be significant when Insti and the interaction between Insti and TA_mod are included. This indicates that Insti explains the cost of debt better than TA_mod. Furthermore, the coefficient on TA_mod is negative and significant for the regression from 1994 to 1998, but the coefficient is insignificant for the regression from 1999 to 2003. The test results of the total period are similar to the results for 1999–2003, when the shareholder rights of institutional investors are strengthened. This interpretation is consistent with the argument that Insti has more of an impact on the cost of debt than the TA_mod. The findings support Hypothesis 2 relating to the influence of institutional ownership on the relationship between tax avoidance and the cost of debt. The positive coefficient, 0.003, on Chaebol indicates that chaebols are related with a higher COD (Baek et al., 2004). Furthermore, I anticipate that the monitoring of institutional ownership that decreases opportunities for managerial rent diversion arising from tax avoidance would increase after 1998, when shareholder rights related to institutional investors were strengthened in Korea. As a result, I investigate whether the effect of institutional ownership on the relationship between tax avoidance and the COD is greater after 1998, than prior to 1998. Using an exogenous source of variation in corporate governance rules created by 1998 reforms, I perform separate analyses for the periods before and after the change in shareholder rights related to institutional investors in Korea. In Column 4 (1994–1998 period) of Panel A in Table 5, the coefficient on TA_mod  Insti is insignificant, while it becomes negative and significant, À0.064, in Column 5 (1999– 2003 period) of Panel A. This suggests that the effect of institutional investors on the negative relationship between tax avoidance and the COD is primarily driven by the period after 1998, as expected. Unreported results also illustrate that the TA_mod  Insti  Post-1998 is negative and significant at the 5% level, which is consistent with the above results. In Column 4, the coefficient on the chaebol is insignificant, while it becomes positive and significant in Column 5, thereby suggesting that bondholders became interested in agency conflicts between controlling shareholders and debtholders in chaebols after the financial crisis and require a higher COD. Apart from shareholder rights, reforms related to corporate governance, including the board of directors and audit committee, were also initiated after the financial crisis of 1997 in Korea. According to the Securities and Exchange Act, the appointment of independent external directors to the Board and the establishment of an audit committee were mandatory for all listed companies with total assets over 2 trillion Korean Won. The post-1998 change in the relationship may be affected by factors other than governance that also happened to change around 1998. To control for this, I use smaller firms with assets <2 trillion Won as an experimental group, since these firms were not subject to the reforms. This leads to a comparison of the relationship between tax avoidance and the COD for larger firms after 1998 with the corresponding relationship for smaller firms. It is expected that the negative relationship between tax avoidance and institutional ownership for the COD should hold for both large firms and small firms, since reforms related to shareholder rights covered all firms.

Panel B of Table 5 illustrates the results of regressing institutional ownership on tax avoidance and the COD, after controlling for other governance change effects. It documents negative and significant coefficients of the interaction variable between TA_mod and Insti for both large firms with assets P2 trillion Korean Won (À0.346) and the small firms with assets <2 trillion Korean Won (À0.057). Panel B further illustrates an insignificant coefficient on TA_mod  Insti  Dummy for large firms, as expected.26 These findings indicate that since 1998, shareholder activism increased the negative association between tax avoidance and institutional shareholdings on the COD, while controlling for other governance effects. In summary, the evidence reported in Tables 4 and 5 suggests that both tax avoidance and institutional investors affect the borrowing costs of firms individually, as well as interactively, and that the managerial opportunism theory accounts for tax avoidance activities. 5.5. Sensitivity analysis 5.5.1. Alternative explanation: the relationship between tax avoidance and firm value Finding a negative relationship between tax avoidance and the COD does not necessarily constitute evidence against the managerial opportunism hypothesis. Opportunistic behavior by managers could destroy shareholder value without increasing the probability of bankruptcy or the COD. To exclude the alternative explanation, this section examines the impact of tax avoidance on firm value. Desai and Dharmapala (2009) indicate that the average effect of tax avoidance on firm value was not significantly different from zero, but was positive for well-governed firms, as predicted by the agency perspective on corporate tax avoidance. Wilson (2009) also finds that active tax shelter firms with strong corporate governance exhibited positive abnormal returns and states that tax sheltering was a tool for wealth creation in well-governed firms. As a result, I examine whether tax avoidance tends to be associated with firm value by using Eq. (7), based on the model specification of Desai and Dharmapala (2009) and Black et al. (2006). I then investigate the impact of shareholder activism on the relationship between tax avoidance and firm value by using Eq. (8) in the same way as the test of COD.

Tobin’s Q i;t ¼ c1i;t þ c2 TA modðor TA perÞi;t þ c3 Ownershipi;t þ c4 Ownership squi;t þ c5 Sizei;t þ c6 LnðAgeÞi;t þ c7 Leveragei;t þ c8 Volatilityi;t þ Firm effect þ Year effect Tobin’s Q i;t ¼ d1i;t þ d2 TA modðor TA perÞi;t þ d3 TA modðor TA perÞi;t  Instii;t þ d4 Instii;t þ d5 Ownershipi;t þ d6 Ownership squi;t þ d7 Sizei;t þ d8 LnðAgeÞi;t þ d9 Leveragei;t þ d10 Volatilityi;t þ Firm effect þ Year effect ð8Þ ð7Þ

Large firms Shareholder activism Other governance change Affected Affected

Small firms Affected N/A

The Table 5 illustrates the test design that separates the effect of shareholder activism from other governance changes that occurred in 1998, including the requirement for external directors and the audit committee, by comparing the large and small firms.

All variables in Eq. (8) are defined in Table 2, except for Tobin’s Q, which is estimated as the market value of assets as the (book value of debt + book value of preferred stock + market value of common stock)/book value of assets, Ownership is the ownership of the largest shareholders, Ownership_squ is the squared value of Ownership, Ln(Age) is the logarithm of the Age variable, Volatility is the standard deviation of the monthly stock returns calculated over a 60month period.
26 Also, I find an insignificant coefficient, 0.025 ( t -value = 0.60) on TA_mod  Insti  Dummy for small firms.

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

467

I use Tobin’s Q to measure firm value and define it as discussed in the extensive literature on the determinants of firm value (e.g., Black et al., 2006; Desai and Dharmapala, 2009). Ownership has been found to be an important element of corporate governance, but the relationship between ownership and firm value is unclear and possibly nonlinear. I control for ownership by the largest shareholder, whether an individual or a firm, and ownership_square (Black et al., 2006). I control for size following prior research (Black et al., 2006). Older firms could differ from younger firms both in Tobin’s Q and governance practices (Black et al., 2006). I therefore include Ln(Age) as a control variable. I expect a negative coefficient on firm value, because younger firms are more likely to be faster growing and perhaps more intangible in relation to asset intensives. Leverage is also controlled for within the impact of Tobin’s Q and the firm’s governance practices (Bhojraj and Sengupta, 2003; Black et al., 2006). To control for changes over time in the risk associated with a firm’s stock price, a measure of volatility is also included (Desai and Dharmapala, 2009; Black et al., 2006). Column 3 (total period) of Panel A in Table 6 illustrates that TA_mod has a positive coefficient of 0.149, which is significant at the 10% level on a two-tailed test for the regression of the firm value. This finding implies that higher tax avoidance activities marginally enhance firm value, on average, which does not support the argument that the opportunistic behavior by managers destroys shareholder value without increasing the COD. The coefficient on Ln(Age) is negative and significant, while the coefficient on Leverage and Volatility is positive and significant, which is consistent with Black et al. (2006). Column 4 (total period) of Panel A of Table 6 indicates that the coefficient for the interaction variable between TA_mod and Insti is 0.736, positive and significant as predicted. It indicates that the positive effect of tax avoidance on firm value increases with institutional ownership for the entire period. I also perform separate analyses for the periods before and after the change in shareholder rights related to the institutional investors in Korea. In Column 5 (1994–1998 period) of Panel A, the coefficient on TA_mod  Insti is negative and significant, while it is positive and significant in Column 6 (1999–2003 period). This finding indicates that the institutional investor’s role in the total period primarily comes from the period of 1999 through 2003, in which shareholder rights were strengthened. Furthermore, Panel A in Table 6 illustrates that the coefficient on TA_mod for the total period in Panel A of Table 6 becomes insignificant when Insti and the interaction between Insti and TA_mod are included in the model. The coefficient on Insti is insignificant and the coefficient on the TA_mod is negative and significant for the regression from 1994 to 1998. The coefficient on Insti is positive and significant for the regression from 1999 to 2003. This means that the test results for the total period are similar to the results for 1999 to 2003, when the shareholder rights of institutional investors are strengthened. That could be an indication that Insti also explains firm value better than TA_mod.27 For firms poorly covered by institutional investors, increases in tax avoidance create greater opportunities for managers to divert income from shareholders, rather than raise after-tax firm value. In contrast, the negative effect is less important for firms with larger institutional holdings, since they have stronger institutional restraints on managerial rent diversion. It suggests that the overall effect of shareholder activism on the relationship of tax avoidance and firm value is positive, given the weak governance in Korea.

Table 6 Regression analysis of influence of shareholder activism on tax avoidance and firm value. Panel A: Firm value Prediction Total period From 1994 to 1998 2.482ÃÃÃ (12.65) 0.805ÃÃ (2.07) À2.084ÃÃ (À2.42) 0.017 (0.35) À0.263 (À1.17) 0.364 (0.93) À0.054ÃÃÃ (À6.23) À0.135ÃÃÃ (À4.57) 0.006ÃÃÃ (6.37) À0.799 (À1.43) Included Included 34.29ÃÃÃ 0.238 1031 From 1999 to 2003 1.109ÃÃÃ (7.79) À0.286Ã (À1.87) 1.073ÃÃÃ (3.31) 0.193ÃÃÃ (5.29) À0.404ÃÃ (À2.51) 0.345 (1.39) 0.011Ã (1.68) À0.195ÃÃÃ (À8.88) 0.001ÃÃ (2.34) 1.373ÃÃÃ (5.46) Included Included 8.67ÃÃÃ 0.136 1406

Intercept TA_mod TA_mod  Insti Insti Ownership Ownership_squ Size Ln(Age) Leverage Volatility Firm effect Year effect F-value Adj R-squ Number of observations À + + ± + + ±

1.445ÃÃÃ (12.61) À0.169 (À1.16) 0.736ÃÃ (2.38) 0.121ÃÃÃ (4.10) À0.197 À0.235Ã (À1.50) (À1.78) 0.249 0.216 (1.18) (1.02) 0.003 À0.010Ã (0.58) (À1.91) À0.168ÃÃÃ À0.164ÃÃÃ (À9.34) (À9.14) 0.000ÃÃÃ 0.001ÃÃÃ (3.99) (4.29) 1.040ÃÃÃ 1.019ÃÃÃ (4.43) (4.36) Included Included Included Included 60.98ÃÃÃ 56.54ÃÃÃ 0.232 0.239 2437 2437

1.246ÃÃÃ (11.93) 0.149Ã (1.95)

Panel B: Control for other governance change effects Prediction Large firms À0.077 (À0.09) À2.638 (À1.54) 5.063ÃÃ (2.49) Small firms 1.577ÃÃÃ (9.36) À0.070 (À0.45) 0.411 (1.20) From 1999 to 2003 1.143ÃÃÃ (8.06) À0.140 (À0.89) 0.558 (1.62) 2.209ÃÃÃ (4.12) 0.186ÃÃÃ (5.13) À0.418ÃÃÃ (À2.61) 0.364 (1.48) 0.010 (1.42) À0.194ÃÃÃ (À8.91) 0.001ÃÃ (2.47) 1.370ÃÃÃ (5.48) Included Included 8.85ÃÃÃ 0.147 1406

Intercept TA_mod TA_mod  Insti TA_mod  Insti  Dummy for large firms Insti Ownership Ownership_squ Size Ln(Age) Leverage Volatility Firm effect Year effect F-value Adj R-squ Number of observations À + + ± + ?

+ ±

0.044 (0.25) 0.303 (0.32) À0.726 (À0.36) 0.069Ã (1.76) À0.106 (À1.46) À0.007 (À1.64) À1.304 (À1.53) Included Included 5.07ÃÃÃ 0.238 118

0.188ÃÃÃ (5.07) À0.478ÃÃÃ (À2.91) 0.477Ã (1.91) À0.014Ã (À1.65) À0.196ÃÃÃ (À8.55) 0.001ÃÃÃ (3.14) 1.410ÃÃÃ (5.91) Included Included 18.86ÃÃÃ 0.111 1288

27 Using Spanish firms, Ruiz-Mallorquí and Santana-Martín (2011) show that the presence of other large shareholders affects firm value when a dominant institutional owner controls the firm.

Notes (Panel A): All the variables are defined in Table 2. Each cell exhibits a coefficient (t-statistic), respectively. The t-statistics are computed using the fixedeffects model controlling for firm effect and year effect. The superscript asterisks indicate the explanatory variable coefficient significance at p-values less than 0.10 (Ã), 0.05 (ÃÃ), and 0.01 (ÃÃÃ). Notes (Panel B): All the variables are defined in Table 2, except for Dummy for large firms which equals to one if assets >2 trillion Won and zero otherwise. Each cell exhibits a coefficient (t-statistic), respectively. The t-statistics are computed using the fixed-effects model controlling for firm effect and year effect. The superscript asterisks indicate the explanatory variable coefficient significance at p-values less than 0.10 (Ã), 0.05 (ÃÃ), and 0.01 (ÃÃÃ).

468 Table 7 Sensitivity analysis. Prediction Fixed effects (1) Intercept TA_mod TA_mod  Tax rate change TA_mod  Insti_sen TA_mod  Insti_insen TA_mod  Bondrating Bondrating Insti Insti_sen Insti_insen Age Big6 Leverage CFO Size Asset Negequity Chaebol Firm effect Year effect F-value Adj R-squ Number of observations À À + À + + + + Included Included 154.04ÃÃà 0.452 3080 0.000 (1.20) 0.001 (1.45) 0.043ÃÃà (22.91) À0.041ÃÃà (À8.05) 0.001 (1.51) À0.001 (À0.50) À0.013ÃÃà (À3.65) À + À À À 0.013Ãà (2.04) À0.011 (À1.34) À0.023Ãà (À2.30)

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

governance changes, rather than shareholder activism reforms, compared with the COD results.
(2) 0.007 (0.86) 0.004 (0.44) (3) À0.004 (À0.19) À0.034 (À1.01)

À0.041 (À0.97) À0.065ÃÃÃ (À2.72) À0.001 (À0.38) 0.002ÃÃÃ (5.14) À0.004 (À1.10) À0.004 (À0.83) À0.015ÃÃÃ (À6.51) 0.000 (0.82) 0.002 (1.64) 0.042ÃÃÃ (22.50) À0.039ÃÃÃ (À7.67) 0.001ÃÃ (2.34) À0.000 (À0.02) À0.013ÃÃÃ (À3.87) 0.003ÃÃÃ (2.63) Included Included 152.72ÃÃÃ 0.461 3080

0.000ÃÃÃ (2.90) À0.004ÃÃ (À2.01) 0.063ÃÃÃ (13.25) À0.029ÃÃÃ (À3.07) 0.000 (0.53) 0.005 (1.17) À0.035ÃÃÃ (À3.02) 0.008ÃÃÃ (3.63) Included Included 60.49ÃÃÃ 0.623 947

5.5.2. Changes in tax rate This section investigates exogenous variations in tax avoidance behavior by examining the interaction effect of tax rate changes with tax avoidance on the COD. Firms have stronger incentives for tax avoidance activities when the statutory tax rate is higher, so that tax avoidance for the purpose of reducing their tax liability will change when the statutory tax rate changes. During the sample period, the statutory corporate tax rate changed three times. For the taxable income above $100,000,29 the marginal tax rate in Korea was reduced from 32% to 30%, effective from 1995; in 1996, it was further reduced to 28%; and from 2002 onward, it was reduced to 27%. First, I define the tax rate change variable as the value of 3 for 1994, 2 for 1995, 1 for 1996–2001, and 0 for 2002–2003. Although, percentage-wise, the changes were relatively small, I expect a negative coefficient on the tax avoidance  Tax rate change. In Model 1 of Table 7, the coefficients on the TA mod  Tax rate change are negative and significant, À0.023. This indicates that the effect of tax avoidance on COD is less for the period in which the tax rate was reduced. Next, I use a continuous tax rate change variable, measured as the yearly marginal tax rate, and examine the interaction effect between tax avoidance and the tax rate change variable on the COD. Untabulated results indicate that the coefficient on TA mod  Marginal tax rate is negative and significant, À2.606 (t-value = À3.15), thereby indicating that tax avoidance has a greater negative effect on the COD when the marginal tax rate was higher.

Notes: Model 1 reveals the regression results testing changes in the tax rate. Model 2 provides the regression results for the decomposition of institutional investors over the entire period. Model 3 presents the regression results by controlling for the probability of bankruptcy. Each cell exhibits a coefficient (t-statistic), respectively. The t-statistics are computed using the fixed-effects model controlling for firm effect and year effect. The superscript asterisks indicate the explanatory variable coefficient significance at p-values less than 0.10 (Ã), 0.05 (ÃÃ), and 0.01 (ÃÃÃ).

Furthermore, to control for other governance changes effected in 1998, I use smaller firms with assets <2 trillion Won as an experimental group in the same way as the test for COD. Panel B of Table 6 illustrates the regression of institutional ownership on tax avoidance and firm value after controlling for other governance change effects. It documents a positive and significant coefficient (5.063) for the interaction between TA_mod and Insti, only for the large firms, and further illustrates a positive and significant coefficient (2.209) for the interaction of TA_mod  Insti  Dummy for large firms.28 These findings indicate that the positive and significant coefficient (1.073) of TA_mod  Insti in the total period, and from 1999 to 2003 in Panel A, is primarily driven by large firms, which are subject to other governance reforms, including the external directors and audit committee. Thus, the influence of institutional investors on firm value primarily comes from the other

5.5.3. Decomposition of institutional investors This section examines the effect of the classification of institutional investors by disaggregating institutional investors into pressure-sensitive and pressure-insensitive institutional investors. Only pressure-insensitive (independent) institutional investors such as investment companies and investment advisors monitor the firm’s management effectively (Brickley et al., 1988; Elyasiani and Jia, 2010). Pressure-sensitive investors, e.g., insurance companies, which rely on firm managers to get pension fund management business, are less likely to monitor the management diligently or to vote against its decisions. Following Brickley et al. (1988), I classify pressure-sensitive institutional investors as banks and insurance companies that may have current or potential business relations with a firm, and pressure-insensitive institutional investors as government, security firms, and foreign investors. By separating Insti in Eq. (6) into Insti_sen (pressure-sensitive investors) and Insti_insen (pressureinsensitive investors), I am able to investigate whether the effect of tax avoidance on COD increases more with pressure-insensitive or pressure-sensitive institutional ownership. Model 2 of Table 7 indicates that the coefficient on TA_mod  Insti_insen is negative and significant (À0.065), while the corresponding coefficient on TA_mod  Insti_sen is not significant, (À0.041), thereby indicating that the monitoring role of pressureinsensitive institutional investors on the managerial opportunism relates to tax avoidance and the COD is stronger than that of pressure-sensitive investors, as expected.30
I use the exchange rate, 1$ = 1000 . Untabulated results indicate that while the coefficient estimates on the TSmod– Insti_insen interaction variable are not significantly negative for 1994–1998, those on the TSmod–Insti_sen interaction variable are positive and significant. In addition, for 1999–2003, the coefficients of the TSmod–Insti_insen interactions are more negative and significant than those on the TSmod–Insti_sen interaction variables.
30 29

28 I find a negative and significant coefficient, À2.170 (t-value = À4.04) on TA_mod  Insti  Dummy for small firms.

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470

469

5.5.4. Additional control: probability of bankruptcy When firms experience a higher probability of bankruptcy, they face a higher COD. Tax avoidance would be less valuable if losses were not entirely deductible, thereby suggesting that the impact of tax avoidance on the COD could be mediated through its effect on the probability of bankruptcy. Based on the probability of bankruptcy (Mansi et al., 2004), credit rating agencies, as information intermediaries, provide independent assessments of the credit quality of a firm. I use the bond rating as the probability of bankruptcy and examine its mediation effect on tax avoidance and the COD (Mansi et al., 2004). Following Francis et al. (2005), I assign a numerical value to the bond rating, as follows: AAA = 1, AA+ = 2, AA = 3, AAÀ = 4, A+ = 5, A = 6, AÀ = 7, BBB+ = 8, BBB = 9, BBBÀ = 10, BB+ = 11, BB = 12, BBÀ = 13, B+ = 14, B = 15, BÀ = 16, CCC+ = 17, CCC = 18, CCCÀ = 19, CC = 20, C = 21, and D = 22. It is evident from Model 3 of Table 7, that for the subsample with bond ratings, firms experiencing a higher probability of bankruptcy face a higher COD, however, there is an insignificant mediation effect of the probability of bankruptcy on the relationship between tax avoidance and the COD. 5.5.5. Alternative measures of tax avoidance and cost of debt Recent research suggests that large positive BTD signal tax aggressiveness (Wilson, 2009). For robustness, I use BTD as a measure of tax avoidance for the main tests. Untabulated results are similar to my main results reported in the paper. I also use an alternative measure for the COD calculated as the ratio of firm’s interest expense in next year to average interest-bearing debt outstanding during present and next years (Francis et al., 2005). The results based on this measure are almost identical with those reported in the text. 6. Conclusions This paper examines whether participating in tax avoidance activities is negatively associated with the cost of debt (COD), and whether institutional investors, as shareholder activists, intensify the effect of tax avoidance on the COD. I use the tax avoidance measure modified from Desai and Dharmapala (2006), who decomposed the book-tax difference into earnings management and tax avoidance components. I find a negative relationship between tax avoidance and the COD for a large sample of Korean firms, supporting the trade-off theory. Further tests reveal that the negative relationship becomes stronger when institutional investor ownership is high; and even stronger after 1998, when the voting rights for institutional investors were reinstated. This suggests that the managerial opportunism theory further accounts for the tax avoidance. These results are robust for a wide variety of tests, including the relationship between tax avoidance and firm value, changes in tax rate, decomposition of institutional investors, additional control of probability of bankruptcy, and alternative measures. This suggests that bondholders view tax avoidance as tax savings that reduces the COD and institutional investors as a monitoring vehicle that decreases the opportunities for the rent diversion of tax avoidance by mitigating agency costs with controlling shareholders. This study contributes to a growing line of research on shareholder activism. In particular, the findings complement recent evidence in the US that suggests a greater influence of shareholder proposals and votes on corporate decisions in the post-SOX environment (Del Guercio et al., 2008; Ferri and Sandino, 2009). This study also has a policy implication, i.e., that debtholders favorably view shareholder activism that decreases opportunism for the managerial rent diversion at an early stage in an emerging market given the increasing active role of institutional investors. In addi-

tion, strengthening the role of institutional investors is found to be beneficial to lower the cost of capital for poorly-governed firms. Acknowledgements I am grateful to the editor Ike Mathur and the anonymous referee for useful comments. I would also like to thank Kooyul Jung, Gary Monroe, Philip Brown, Baljit Sidhu, Ahsan Habib, Byungmo Kim, and session participants at the 2010 Accounting and Finance Association of Australia and New Zealand for useful comments. References
Anderson, R.C., Mansi, S.A., Reeb, D.M., 2003. Founding family ownership and the agency cost of debt. Journal of Financial Economics 68, 263–285. Ashbaugh-Skaifea, H., Collins, D.W., LaFond, R., 2006. The effects of corporate governance on firms credit ratings. Journal of Accounting and Economics 42, 203–243. Bae, K.H., Kang, J.K., Kim, J.M., 2002. Tunneling or value added? Evidence from mergers by Korean business groups. Journal of Finance 57, 2695–2740. Bae, G.S., Cheon, Y.S., Kang, J.K., 2008. Intra-group propping: evidence from the stock-price effects of earnings announcements by Korean business groups. Review of Financial Studies 21, 2015–2060. Baek, J.S., Kang, J.K., Park, K.S., 2004. Corporate governance and firm value: evidence from the Korean financial crisis. Journal of Financial Economics 71, 265–313. Baek, J.S., Kang, J.K., Lee, I.M., 2006. Business groups and tunneling: evidence from private securities offerings by Korean chaebols. Journal of Finance 61, 2415– 2449. Bhojraj, S., Sengupta, P., 2003. Effect of corporate governance on bond ratings and yields: the role of institutional investors and outside directors. Journal of Business 76, 455–475. Black, B., Jang, H., Kim, W., 2006. Does corporate governance predict firms’ market values? Evidence from Korea. Journal of Law, Economics, and Organizations 22, 366–413. Boubakri, N., Ghouma, H., 2010. Control/ownership structure, creditor rights protection, and the cost of debt financing: international evidence. Journal of Banking and Finance 34, 2481–2499. Brickley, J., Lease, R., Smith, C., 1988. Ownership structure and voting on antitakeover amendments. Journal of Financial Economics 20, 267–292. Brockman, P., Yan, X., 2009. Block ownership and firm-specific information. Journal of Banking and Finance 33, 308–316. Carey, M., Prowse, S., Rea, J., Udell, G., 1993. The economics of private placements: a new look. Financial Markets Institutions and Instruments 2, 1–66. Chen, S., Chen, X., Cheng, Q., Shevlin, T., 2010. Are family firms more tax aggressive than non-family firms? Journal of Financial Economics 98, 41–61. Ilbo, Chosun, 2006. Hyundai Accused of Fraudulent Accounting. Chung, R., Firth, M., Kim, J., 2002. Institutional monitoring and opportunistic earnings management. Journal of Corporate Finance 8, 29–48. Daily, C., Dalton, D., Cannella, A., 2003. Corporate governance: decades of dialogue and data. Academy of Management Review 28, 371–382. DeAngelo, H., Masulis, R.W., 1980. Optimal capital structure under corporate and personal taxation. Journal of Financial Economics 8, 3–29. Dechow, P., Sloan, R., Sweeney, A., 1995. Detecting earnings management. The Accounting Review 70, 193–225. Del Guercio, D., Wallis, L., Woidtke, T., 2008. Do board members pay attention when institutional investors ‘‘just vote no”? Journal of Financial Economics 90, 84– 103. Desai, M.A., Dharmapala, D., 2006. Corporate tax avoidance and high powered incentives. Journal of Financial Economics 79, 145–179. Desai, M.A., Dharmapala, D., 2009. Corporate tax avoidance and firm value. Review of Economics and Statistics 91, 537–546. Desai, M.A., Dyck, A., Zingales, L., 2007. Theft and taxes. Journal of Financial Economics 84, 591–623. Diamond, D.W., 1989. Reputation acquisition in debt markets. Journal of Political Economy 97, 828–862. Elyasiani, E., Jia, J., 2010. Distribution of institutional ownership and corporate firm performance. Journal of Banking and Finance 34, 606–620. Ferri, F., Sandino, T., 2009. The impact of shareholder activism on financial reporting and compensation: the case of employee stock options expensing. The Accounting Review 84, 433–466. Francis, J., LaFond, R., Olssona, P., Schipper, K., 2005. The market pricing of accruals quality. Journal of Accounting and Economics 39, 295–327. Frank, M.M., Lynch, L.J., Rego, S.O., 2009. Tax reporting aggressiveness and its relation to aggressive financial reporting. The Accounting Review 84, 467–496. Graham, J.R., 2000. How big are the tax benefits of debt? Journal of Finance 55, 1901–1941. Graham, J.R., Tucker, A., 2006. Tax shelters and corporate debt policy. Journal of Financial Economics 81, 563–594. Graham, J.R., Lemmon, M.L., Schallheim, J.S., 1998. Debt, leases, taxes, and the endogeneity of corporate tax status. Journal of Finance 53, 131–161. Graham, J.R., Lang, M.H., Shackelford, D.A., 2004. Employee stock options, corporate taxes, and debt policy. Journal of Finance 59, 1585–1618.

470

Y. Lim / Journal of Banking & Finance 35 (2011) 456–470 Petersen, M.A., Rajan, R.G., 1994. The benefits of lending relationships: evidence from small business data. Journal of Finance 49, 3–37. Pittman, J., Fortin, S., 2004. Auditor choice and the cost of debt capital for newly public firms. Journal of Accounting and Economics 37, 113–136. Ruiz-Mallorquí, M.V., Santana-Martín, D.J., 2011. Dominant institutional owners and firm value. Journal of Banking and Finance 35, 118–129. Shackelford, D., Shevlin, T., 2001. Empirical tax research in accounting. Journal of Accounting and Economics 31, 321–387. Shin, H.H., Park, Y.S., 1999. Financing constraints and internal capital markets: evidence from Korean chaebols. Journal of Corporate Finance 5, 169–191. Shleifer, A., Vishny, R., 1986. Large shareholders and corporate control. Journal of Political Economy 94, 461–488. Trezevant, R., 1992. Debt financing and tax status: tests of the substitution effect and the tax exhaustion hypothesis using firms responses to the economic recovery tax act of 1981. Journal of Finance 47, 1557–1568. Tucker, S.J., Zarowin, P., 2006. Does income smoothing improve earnings informativeness? The Accounting Review 81, 251–270. Wilson, R.J., 2009. An examination of corporate tax shelter participations. The Accounting Review 84, 969–999.

Hartzell, J.C., Starks, L.T., 2003. Institutional investors and executive compensation. Journal of Finance 58, 2351–2374. Joh, S.W., 2003. Corporate governance and firm profitability: evidence from Korea before the economic crisis. Journal of Financial Economics 68, 287–322. John, K., Lynch, A.W., Puri, M., 2003. Credit ratings, collateral, and loan characteristics: implications for yield. Journal of Business 36, 371–409. Johnson, S., Boone, P., Breach, A., Friedman, E., 2000. Corporate governance in the Asian financial crisis. Journal of Financial Economics 58, 141–186. Kothari, S.P., Leone, A., Wasley, C., 2005. Performance matched discretionary accrual measures. Journal of Accounting and Economics 39, 163–197. Lim, Y.D., 2010. Tax Avoidance and Underleverage: Korean Evidence. Working Paper, University of New South Wales. Mackie-Mason, J.K., 1990. Do taxes affect corporate financing decisions? Journal of Finance 45, 1471–1493. Mansi, S.A., Maxwell, W.F., Miller, D.P., 2004. Do auditor quality and tenure matter to investors? Evidence from the bond market. Journal of Accounting Research 42, 755–793. Molina, C.A., 2005. Are firms underleveraged? An examination of the effect of leverage on default probabilities. Journal of Finance 60, 1427–1459. Morsman, J.E., 1986. Commercial loan structuring. Journal of Commercial Bank Lending 58, 2–20.

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close