Unintended Consequences of the Sarbanes-Oxley Act s Timing For the U.S. and Foreign Rule 144A Debt Issuers

Transcription

1 Unintended Consequences of the Sarbanes-Oxley Act s Timing For the U.S. and Foreign Rule 144A Debt Issuers Usha R. Mittoo University of Manitoba usha.mittoo@umanitoba.ca Zhou Zhang University of Regina zhou.zhang@uregina.ca This version: January 10, 2016 JEL Classification: G01; G15; G32 Keywords: Sarbanes-Oxley Act; Rule 144A; Registration Rights; Regulation S; U.S. and Foreign Issuers; Market and Security Choices; Yield Spreads; Credit Ratings 1

2 Unintended Consequences of the Sarbanes-Oxley Act s Timing For the U.S. and Foreign Rule 144A Debt Issuers This version: January 10, 2016 Abstract The timing of the Sarbanes-Oxley Act (SOX) and its associated reforms coincided with major global regulatory reforms (GLOBAL) implemented in many countries that increased the competition and lowered the cost of debt in international bond markets. We distinguish between the impacts of SOX and GLOBAL on yield spreads and security market choices using Rule 144A debt issues with registration rights (RR) that allow conversion of 144A issues into public debt issues as a treatment sample and non-rr as a control sample. SOX predicts a decline in yield spreads for RR issues but no effect on non-rr issues whereas GLOBAL predicts a decline for both RR and non-rr issues. SOX also predicts a similar effect for U.S. and foreign issuers whereas GLOBAL predicts a larger decline for foreign issuers due to home bias and country-specific factors. We find a decline of more than 50 basis points in yield spread after SOX for both RR and non-rr issues and the decline is significantly larger for foreign issuers (68 basis points) that eliminated their pre-sox yield premium relative to their U.S. peers. The post-sox proportion of RR issues also declined by 50% whereas that of combined 144A and Regulation S (RS) offerings that placed simultaneously with U.S. and offshore investors increased over 100% after SOX, with a much larger increase for foreign issuers. The post-sox median credit quality of U.S. RR issuers increased from B to BB+ but was unchanged for foreign issuers (BB+). We also find a post- SOX yield spread decline of 30 basis points in the U.S. public debt market and a much larger decline for foreign issuers. Our evidence supports the notion that GLOBAL, not SOX, was the major driver of the yields spread decline in both 144A and public debt markets and its timing with SOX resulted in a more adverse impact on U.S. speculative credit quality issuers relative to their foreign peers as GLOBAL eliminated their cost of debt advantage and reduced their fast access to financing through RR issues. JEL Classification: G01; G15; G32 Keywords: Sarbanes-Oxley Act; Rule 144A; Registration Rights; Regulation S; U.S. and Foreign Issuers; Market and Security Choices; Yield Spreads; Credit Ratings 2

3 1. Introduction The 2002 Sarbanes-Oxley Act (hereafter, SOX) is considered as the most important security legislation in the U.S. since the Securities Act of 1933 and the Securities Exchange Act of Enacted in response to several high-profile corporate scandals, SOX increased disclosure requirements, tightened internal controls, introduced legal liability for corporate executives, and also reversed the tradition of accommodating foreign firms by making it mandatory for them. SOX was also accompanied by stricter listing and corporate governance standards by stock exchanges and a tougher enforcement by the U.S. Security and Exchange Commission (SEC). The impact of SOX and its associated reforms on shareholders has been studied extensively but the evidence is mixed on the net benefits for shareholders. 1 Although SOX was aimed primarily at protecting shareholders, its provisions could affect bondholders as well. For instance, the tightened corporate governance and internal controls in SOX regulations could reduce agency costs and cost of debt whereas the closer alignment between shareholders and managers interests could increase the agency problems between shareholders and bondholders, implying a higher cost of debt. Several recent studies examine the impact of SOX on bondholders and report an insignificant effect but a more adverse impact on low credit quality bondholders - with one exception, Nejadmalayeri et al. (2013), concluding that bondholders reaped major benefits of SOX. Using a structural model and municipal bonds as a control, Nejadmalayeri et al. (2013) show that the U.S. public debt holders experienced a decline of 27 basis points from 2001 to 2006, with a much larger decline for low credit quality firms. They conclude that this decline is due to the increased transparency and tougher internal controls under SOX. The time-period in their study, however, also 1 See Coates and Srinivasan (2014) for a review of SOX related literature and Linck et al. (2011) and Bradley and Chen (2015) on SOX s related changes and their impact on firms. 3

4 coincided with several major regulatory reforms in many countries, such as the Financial Services Action Plan (FSAP) that harmonized the functioning of financial intermediaries throughout the European Union, the adoption of International Financial Reporting Standards (IFRS) and of SOX-type regulation in many countries that integrated international bond markets and lowered the cost of debt. 2 These global regulatory reforms (hereafter, GLOBAL) dismantled cross-border financing and investment barriers, transformed and integrated the European bond markets, and enhanced the competition between the U.S. and European bond markets that resulted in lower underwriting costs and yield spreads in both U.S. and non-u.s. countries. For example, Peristiani and Santos (2010) show that yield spreads declined from 2002 to 2006 in both U.S. and European public bond markets and converged by Resnick (2011) finds no significant difference in the yield spreads between the U.S. domestic bonds, or Yankee bonds, or Eurodollar bonds in recent years. Thus, whether SOX or GLOBAL is the dominant driver of the yield spread decline documented in Nejadmalayeri et al. (2013) is a debatable issue. In this paper, we contribute to this debate by providing evidence that GLOBAL, not SOX, is the major driver of the post-sox yield spread decline in the U.S. public debt markets. Due to the lack of a control group, disentangling SOX and GLOBAL effects in the public bond markets is challenging because all public debt issuers were impacted by both regulations. 3 Our empirical strategy aims to alleviate this concern and involves two steps. In the first step, we use Rule 144A debt offerings with registration rights (hereafter, RR) that allow issuers to exchange their 144A issues with public debt issues subsequent to the issuance as a control group 2 Several studies have examined the impact of GLOBAL reforms on cost of equity and debt across countries(e.g., Barth et al., 2008; Horton et al., 2008; Daske et al., 2008; Christensen et al., 2009; Li, 2010; Gao, 2011; Kalemli- Ozcan et al., 2010; Peristiani and Santos, 2011; Resnick, 2012; Akyol et al., 2014; Bris et al., 2014). 3 Leuz (2007) and Hochberg et al. (2009) also emphasize the lack of a control group of publicly traded firms that were not affected by SOX. 4

5 for SOX to generate and test empirical predictions on yield spreads that distinguish between SOX and GLOBAL hypotheses. In the second step, we extend the analysis to the public bond market. Despite that 144A debt market was not directly subject to SOX, RR issues were essentially delayed public debt issues because over 90% of them were converted into public debt and registered with the SEC within a few months after issuance (Huang and Ramirez, 2010). The yield spread differential between RR and the public debt issues, though positive initially, vanished over time reflecting their high liquidity and likelihood of conversion into public debt (Fenn, 2000). SOX is likely to shift costs and benefits for RR issues similar to a public debt issue but will have no effect on non-rr issues, providing us treatment and control samples for SOX. More importantly, because post-sox issuers of RR offerings voluntarily choose to subject themselves to SOX, they are likely to perceive positive net benefits of SOX that will reflect in a lower yield spread relative to the pre-sox period, all else equal. GLOBAL hypothesis also predicts a decline in yield spreads but for both RR and non-rr issues, all else equal, because it arises as a result of increased competition between U.S. and non-u.s. debt markets. This leads to the first implication that distinguishes between SOX and GLOBAL hypotheses: all else equal, a post-sox decline in the yield spreads of RR issues will be consistent with GLOBAL, not with SOX. Another prediction that differs between SOX and GLOBAL hypotheses relates to their relative effects on foreign versus U.S. issuers. Because SOX was mandatory for both U.S. and foreign issuers in the public markets, it predicts a similar effect for U.S. and foreign RR issuers. In contrast, GLOBAL predicts a larger yield spread decline for foreign 144A issuers (both RR and non-rr) because they are likely to pay a higher yield spread premium relative to their U.S. peers due to country-specific factors such as home bias, and differences in disclosure and 5

6 accounting standards (Chaplinsky and Ramchand, 2004). 4 GLOBAL is likely to diminish this premium which leads to the second implication that distinguishes between SOX and GLOBAL hypotheses: a larger post-sox decline in yield spreads of foreign RR and non-rr issuers relative to their U.S. peers will be consistent with GLOBAL, not with SOX. We examine these predictions in a sample of 144A debt issues from 1997 to 2007, dividing the sample period into pre- and post-sox periods. We find strong support for GLOBAL hypothesis and show (1) the average post-sox yield spreads decline is more than 50 basis points for both RR and non-rr debt issues, and (2) the decline is significantly larger for foreign issuers and mostly eliminates their pre-sox yield premium relative to their U.S. peers. To confirm whether the post-sox yield spread decline in 144A debt market is indeed driven primarily by GLOBAL and not by SOX, we examine whether 144A debt issuers security market choices also changed consistently with the yield spread declines. First, If SOX indeed had little net benefits for RR issuers as concluded above, the number of both U.S. and foreign RR issues will decline post-sox, all else equal. Our evidence supports this consistency check; the proportion of RR issues declines from 55% in the pre-sox period to 22% post-sox for both U.S. and foreign issuers. Second, if GLOBAL is the main driver of yield spread declines, the proportion of 144A issuers who place their debt with offshore investors should increase and the increase will be larger for foreign issuers due to their home bias and investor recognition advantage in offshore markets, all else equal. While 144A issuers can use several channels to reach offshore investors, 144A issues conducted simultaneously with Regulation S offerings (hereafter, RS) provide a more convenient channel to reach both U.S. and offshore investors 4 Huang et al. (2013) find that foreign 144A debt issues have a significantly higher transaction spreads relative to U.S. 144A debt, Yankee debt and U.S. public debt. 6

7 simultaneously with minimum incremental costs due to the economy of scale. 5 Thus, GLOBAL hypothesis predicts that the proportion of RS offerings will increase post SOX and the increase will be larger for foreign issuers. Our evidence supports this conjecture: the proportion of RS issues increases from 28% in pre-sox period to 53% in post-sox period, and the increase almost doubles for foreign RS issuers. We further conduct several robustness checks to confirm our findings. Because SOX is likely to have disproportionately higher costs for low credit quality firms who are generally small firms, we expect that few of them will undertake RR issues post-sox. 6 Our evidence supports this prediction; the median (average) credit quality of U.S. RR issuers increased from B to BB+ after SOX whereas it is unchanged for foreign issuers who had an average of BB+ ratings prior to SOX. We also conduct subsample analysis separately in RR, non-rr, and RS issues and examine the sensitivity of results to different time periods. The results remain essentially the same. In particular, we should not see any significant decline in yield spreads or shift security market choices in the pre-sox period. Our evidence supports this prediction. In the next step, we examine whether GLOBAL is also the major driver of the yield spread decline in the U.S. public debt market. We document a post-sox decline of about 30 basis points in yield spreads similar to that in Nejadmalayeri et al. (2013). While we cannot test the first implication that distinguishes between SOX and GLOBAL hypotheses because we do not have a control group, we can still test the second implication that GLOBAL predicts a larger yield spread decline for foreign issuers whereas SOX predicts a similar effect for U.S. and foreign issuers, all else equal. Our evidence is consistent with GLOBAL hypothesis; the decline is significantly larger for foreign issuers and eliminated their pre-sox yield premium relative to 5 Regulation S also adopted by the SEC in 1998 that allowed both U.S. and foreign issuers to reach offshore investors without the SEC registration. 6 Faulkender and Petersen (2006) show a strong positive relation between firm size and credit ratings. 7

8 their U.S. peers similar to that in the 144A debt market. We also find a more adverse effect for U.S. speculative grade issuers. In sum, our evidence supports the notion that GLOBAL, not SOX, is the major driver of the yield spread declines in both R144A and public debt markets. Our study makes several contributions to the SOX literature. First, despite a preponderance of research, there is little consensus on the net benefits of SOX, primarily because the difficulty of disentangling SOX effects from other confounding effects. In a review of over 100 SOX-related studies, Coates and Srinivasan (2014, page 55) summarize these challenges Studying such changes is much harder than rocket science, which after all models relatively simple inert objects moving through space, and not large groups of independent agents interacting in a complex regulatory and market environment. Coates and Srinivasan suggest that given the complexity of the task and the low power of tests, there is need to find better quasi-experimental designs to adequately control for omitted variables and identify causal effects. We construct a novel quasi-experiment research design exploiting the unique features of SOX exempt 144A debt market to distinguish rather than disentangle SOX and GLOBAL effects. Our work complements that of Bradley and Chen (2015) and differs from prior SOX studies in at least three ways. First, we use a pre- and post-sox windows rather than only pre-sox or post- SOX period (see e.g., DeFond et al., 2011; Nejadmalayeri et al., 2013). Second, unlike Nejadmalayeri et al. (2013) and Andrade et al. (2014), we have a treatment and a control sample for SOX and examine the relative impact on U.S. and foreign debt issuers to distinguish between SOX and GLOBAL effects. Finally, we also conduct consistency checks using post-sox shifts in issuers security market choices and credit quality to confirm our results in yield spreads. Our study contributes to growing literature on the impact of SOX on bondholders that shows mixed results. Our evidence suggests that the reduction in cost of debt of 27 basis points reported in 8

9 Nejadmalayeri et al. (2013) or of 17.7 basis points in Andrade et al. (2014) is driven primarily by GLOBAL. Our findings also support DeFond et al. (2011) that SOX had little net benefits for bondholders and had a more adverse effect on low credit quality firms. Second, our study also extends the literature on the SOX effects on foreign firms security market choices. Most prior studies examine SOX s impact on foreign firms decisions to delist from the U.S. stock exchanges but find mixed results because of endogeniety issues and confounding effects (Coates and Srinivasan, 2014). Gao (2011) is one of the few studies that examine the impact of SOX on foreign issuers bond market choices and finds that foreign issuers rely less on the U.S. public market and more on the Eurobond and 144A debt market after SOX. She acknowledges that this shift could be partly driven by other contemporaneous changes, such as the development of Eurobond market and the adoption of IFRS in many countries. Our study complements and extends Gao s work by first distinguishing between SOX and GLOBAL effects on yield spreads for both U.S. and foreign issuers and then analyzing the shift in security market choices as consistency checks. Our evidence supports the notion that GLOBAL is the dominant factor in explaining the shift in foreign firms preferences from the U.S. to European bond markets after SOX, although SOX exacerbated this effect. Third, there is some consensus that SOX had several unintended consequences, especially for small U.S. firms who faced disproportionately higher SOX implementation costs. For example, Gao et al. (2007) find that SOX-exempt firms adopted several strategies to remain small, including making more cash payouts to shareholders and undertaking less investment. Linck et al. (2009) show that smaller firms faced difficulties in recruiting outside board directors mandated by SOX. Our study extends this literature by showing that the coincidence of SOX timing with GLOBAL resulted in more adverse impact on the access and cost of debt for U.S. 9

10 low credit quality firms relative to their high quality peers. Unlike its long tradition of accommodating foreign issuers, SOX was mandatory for both U.S. and foreign firms. Prior to SOX, a vast majority of U.S. low credit quality issuers who intended to make public debt offerings used RR issues for quick financing by circumventing the delay due to the SEC registration. The dramatic post-sox decline in the number of RR issues and the increase in the median credit quality of U.S. RR issuers from B to BB+ supports the notion that SOX had more adverse effect on the U.S. poor credit quality firms. More importantly, GLOBAL had a more positive effect on foreign issuers access and cost of debt relative to the U.S. issuers. The net effect of GLOBAL and SOX could diminish the access to quick financing and the lower cost of debt advantages that U.S. low credit quality issuers used to enjoy prior to SOX, which was not anticipated by regulators. Finally, our study contributes to the growing literature on the impact of major regulatory reforms in European and international financial markets on the cost of debt for U.S. and foreign issuers. Kalemli-Ozean et al. (2010) and Bris et al. (2014) examine the impact of the adoption of the Financial Services Action Plan (FSAP) on cost of debt among European countries and conclude that FASP increased the access to financing and lowered the cost of debt and these effects were stronger in countries with weak legal systems. Peristiani and Santos (2010) show that the differences in underwriting costs between the U.S. and Eurobond markets have diminished substantially. Resnick (2011) finds that there is no significant difference in the yield spreads between the U.S. dollar global bonds and U.S. domestic bonds, or Yankee bonds, or Eurodollar bonds in recent years. Our study extends this stream of work by investigating the impact of SOX and Global on yield spreads and security market choices for U.S. and foreign 144A and public debt issuers. Our findings in 144A debt market are consistent with Peristiani 10

11 and Santos s findings that smaller and less frequent issuers that did not keep a constant presence in the market were particularly sensitive to cheaper financial costs in the Eurobond market and were more likely to issue abroad in The remainder of the paper is organized as follows. Section 2 provides an overview of Rule 144A, Registration Rights and Regulation S, and develops testable hypotheses. Section 3 describes the data, sample and control variables used in multivariate regressions. Sections 4 and 5 present the empirical results on the borrowing costs and security choices. Section 6 summarizes and concludes the paper. 2. Background of Rule 144A, SOX, GLOBAL, Related Literature and Hypotheses In this section, we provide an overview of Rule 144A market, discuss the time-line of SOX and GLOBAL regulations and related literature to distinguish between SOX and GLOBAL predictions for the 144A debt issuers Rule 144A Debt Market Prior to 1990, the U.S. had the world s largest public and private debt markets but few foreign issuers tapped these markets. Foreign firms regarded the SEC registration and disclosures requirements for public debt as onerous, and shunned the private debt market because investors were required to hold the securities for at least two years before resale. 7 In April 1990, the SEC adopted Rule 144A to attract more foreign issuers to the U.S. markets. Rule 144A removed the restriction on immediate resale of 144A securities among Qualified Institutional Buyers (QIBs) who are investors with large investment portfolios and include institutional investors, such as insurance companies, pension and mutual funds. Because most QIBs are the major investors in 7 In addition, the liability risk is generally perceived to be more significant in the U.S. than in most other jurisdictions and the SEC could also bring civil actions against issuers and other persons for registration violations and disclosure deficiencies (Zoubek and Rosen, 2008). 11

12 foreign debt, foreign issuers viewed 144A debt as an effective substitute for public debt without having to meet strict SEC registration requirements. As a result, the proportion of foreign 144A debt relative to total foreign debt issues grew rapidly, from 11% in 1991 to 65% by 1997, leading Chaplinsky and Ramchand (2004) to predict that 144A could soon eclipse the public debt market for international firms. Although 144A was aimed primarily at foreign issuers, a growing number of U.S. public bond issuers also actively used this market for the purpose of speedy issuance. Firms can issue 144A with registration rights (RR) that allowed them to convert the 144A issue to a public debt issue after issuance, which was then registered with the SEC and could be freely traded among all investors rather than only among QIBs. This innovative use of 144A market combined the best features of the private (speedy issuance) and public (high liquidity) debt markets because it eliminated the delay of about three to six months due to the SEC registration. The process of underwriting and distributing RR securities is also similar to that of public debt issues and is generally done on a firm commitment basis (Johnson, 2000). RR issues became a popular channel of financing among U.S. high-yield debt issuers who generally need to access financing quickly (Fenn, 2000; Livingstone and Zhou, 2002). By 2001, U.S. domestic 144A debt issues comprised over 70% of the total 144A debt issuances and the vast majority of them were RR issues that were subsequently registered with the SEC within 30 to 90 days after issuance (Huang and Ramirez, 2010). The yield spread differential between public debt and 144A issues, although positive initially, vanished overtime reflecting their high likelihood of conversion into public debt (Fenn, 2000). Another innovative use of 144A debt offerings was to combine them with Regulation S (RS). Regulation S, also adopted by the SEC in April 1990, governs the registration of offshore 12

13 placements by both U.S. and foreign issuers, and their subsequent secondary market resale in the U.S. Unlike Rule 144A, Regulation S prohibits pre-selling securities in the U.S. but allows the placements and trading of these issues on many international exchanges, recognized by the SEC as Designated Offshore Securities Markets (DOSM). 8 Rule 144A also provided an alternate method for RS securities issued by foreign firms to reach the U.S. capital markets and investors. 9 Although Rule 144A and RS offerings are subject to different jurisdictions, they involve minimum incremental cost due to the same underwriter and economy of scale. RS issues became very popular with foreign firms since it provided a convenient channel for them to simultaneously reach U.S., offshore, and home country investors. For example, in 1994, 40% of the total foreign 144A debt issues comprised RS issues but only 1% of the U.S. issuers used RS offerings The Time Line of SOX and GLOBAL Regulations, and Related Literature The main objectives of the Sarbanes Oxley Act are: 1) to strengthen the independence of auditing firms, (2) to improve the quality and transparency of financial statements and corporate disclosure, (3) to enhance corporate governance, (4) to improve the objectivity of research, and (5) to strengthen the enforcement of the federal securities laws (Linck et al., 2009). Although SOX became law on July 30, 2002, the timing of many regulations associated with SOX, such as Section 404 and the stricter listing and corporate governance requirement adopted by the U.S. stock exchanges, were implemented from 2002 to For example, NYSE and NASDAQ submitted their proposals for tightening their governance listing standards to the 8 Rule 144A and Regulation S offerings are governed by SEC Rule 12g3-2(b) which requires only home country accounting statements with adequate English translation. The SEC recognizes sixteen exchanges as DOSM and trades on these DOSM are settled through European Clearing Agencies CEDEL (Pinegar et al., 2002). 9 This method involves foreign issuers selling securities offshore pursuant to Regulation S to such parties as foreign institutions or to U.S broker-dealer affiliates located overseas. Using Rule 144A as an exemption from the Securities Act, such parties resell the securities to QIBs in the United States. By treating Rule 144A as an exemption, the party reselling does not violate either the direct selling efforts or offshore transaction requirement of the Regulation S. 13

14 SEC in 2002, these were approved only in November The exchanges required firms to adopt the new governance standards during their first annual meeting after January 15, 2004, and firms with classified boards were given until the second annual meeting to comply, but no later than December 31, 2005 (Linck, 2009).The SEC also delayed the implementation of the SOX Section 404 that requires a public company s top management and the external auditor to assess and attest the adequacy of internal controls until the fiscal years ending on or after November 15, 2004 (Schneider et al., 2009). In addition, the SEC also gave exemptions to comply with Section 404 to small companies with a public float less than $75 million until 2007 and foreign private issuers from July 15, The time line of SOX implementation from 2002 to 2006 coincided with several regulatory reforms in many countries that led to the development and integration of the international bond markets, enhanced the competition for the U.S. bond markets and resulted in a decline in the cost of debt in both U.S. and non-u.s. markets. Although there were many reforms, we highlight three reforms that had a major impact on the bond market integration. The first regulation is the Financial Services Action Plan (FSAP), an ambitious initiative launched by the European Union (EU) Commission with the goal of harmonizing the financial intermediary market within the EU. The FSAP included various items for new EU-wide legislation regarding securities settlements, rules for bank reorganizations, money laundering, consumer protection in financial services, cross-country mergers, insider dealing, market manipulation and transparency in securities markets. Several studies show that FSAP had a major impact on European firms cost of debt, access to financing and financing choices. Kalemli-Ozcan et al. (2010) show that 10 Firms with a public float of under $700 million were given a break and did not have to provide an auditor s attestation until July 15, Foreign firms that had to comply with the auditor s attestation requirement in 2006 were larger in terms of market size, assets, and sales and are less likely to be Nasdaq listed than firms that did not have to provide an auditor s attestation. 14

15 cross-border banking activities increased significantly among European countries that quickly adopted the financial services directives of the FSAP and conclude that while the euro has been a major force in driving financial integration, legal-regulatory harmonization policies in financial services have contributed crucially to the deepening of European financial markets. Bris et al. (2014) examine the effect of FSAP on European firms financing activities in pre-fasp ( ) and FASP ( ) periods and find that the increase in external financing is concentrated in the FASP period and that cost of financing has decreased, supporting the notion that changing financial institutions as a result of FASP is the main driver of these results. 11 They conclude that corporate bond markets have become more integrated and credit spreads have converged across the euro area as a result of FSAP. Their firm-level evidence is consistent with the aggregate-level results of Lane (2013) who shows that capital flows in the euro area were dominated by debt investments prior to the global financial crisis but also are consistent with the view that firms have increased financing because cost of financing decreased and access to financing increased has been a major contributing factor in explaining increased use of external financing because firms that were likely to be financially constrained prior to the introduction of the euro experience the largest increases in external financing. Peristiani and Santos (2011) compare the gross spreads in the U.S. and Eurobond markets in and time periods. They find that the gross spreads were significantly higher in the Eurobond market than in the U.S. bond market but during 2002 to 2006 the financial deregulation enabled the Eurobond market to catch up and erase the competitive 11 Bris et al. (2006, 2009) shows that the introduction of the euro resulted in increases in corporate valuations and investments for euro area firms compared to other European firms and attribute these changes to a decrease in cost of equity and debt capital. They find that euro area financial markets have become less segmented since the introduction of the euro, as manifested by the increase in cross-border portfolio investments, in particular for bonds (see, e.g., De Santis and Gerard, 2006; Lane, 2006; Lane and Milesi-Ferretti, 2007). Thus, firms have become less dependent on domestic investors when raising external financing. 15

16 advantage of the U.S. bond market. They also find that these spreads have declined continuously in the U.S. market in their sample period, but they declined at a faster rate in the Eurobond market to the point of eliminating the difference that use to exist between the two markets and the competitive wedge that the U.S. bond market used to enjoy has to a large extent disappeared. Furthermore, they also show that U.S. corporate issuers are increasingly seeking to borrow in the Eurobond market and that smaller and less frequent issuers were particularly sensitive to cheaper financial costs in the Eurobond market and were more likely to issue abroad in Furthermore, European firms also appeared to be moving back to their home market as the share of European issuers borrowing in the U.S. bond market dropped from more than 20 percent in the late 1990s to approximately 9 percent in The second major regulatory regulation is the adoption of uniform financial reporting standards by stock exchanges and accounting standards bodies in over 100 countries from 2002 to These uniform standards, labeled as International Financial Reporting Standards (IFRS), aim to achieve global harmonization and convergence of financial reporting rules and regulations that are also major barriers to cross-border financing. For example, all EU-listed companies were required to adopt IFRS in the beginning of 2005 while some firms adopted voluntarily before 2005 as well as the London and Luxembourg stock exchanges recognized IFRS reporting. Since Eurodollar bonds are usually listed on these two exchanges that will lower the cost of listing of Eurodollar bonds. Research show that the introduction of IFRS facilitates cross-border investment, resulting in higher liquidity and equity valuations in countries adopting IFRS (e.g., Covrig et al. 2007, Wu and Zhang 2010, Daske et al., 2008). Gao (2011) argues that the informational benefits and governance implications associated with the adoption of IFRS 16

17 lower the cost of SOX compliance and finds that firms adopting IFRS are more likely to choose the U.S. bond market in the post-sox period Hypotheses on the Impact of SOX and GLOBAL on 144A Debt Market In this section, we develop predictions that distinguish between the impact of SOX and GLOBAL on the yield spreads and issuance choices of 144A debt issuers. As discussed in Section 2.1, although SOX was not applicable to 144A issuers, it could affect the costs and benefits for RR issues because most of them were converted into public debt after issuance and thus became subject to SOX. For example, improved disclosure requirements and strengthened internal controls, and tougher anti-fraud provisions and penalties under SOX are likely to provide better protection for bondholders and lower agency costs of debt. Consequently, bondholders might be willing to accept a lower yield from issuers who subject themselves to increased scrutiny under SOX (e.g., DeFond et al. 2011; Gao, 2011). SOX is also likely to impose several indirect costs on the bond issuers and investors. For example, better alignment of managers and shareholders interests under SOX s stronger corporate governance provisions could provide incentive to managers to increase shareholder value at the expense of bondholders (e.g., Cremers et al., 2007; Chava et al., 2009; DeFond et al., 2011). More importantly, SOX induced costs and benefits are likely to vary across firms and RR issuers will reevaluate their net benefits and many will switch to non-rr issues to avoid SOX as documented in the case of public debt issuers (Gao, 2011). We expect that only those issuers who perceive positive net benefits of SOX (e.g., lower yield spread) will undertake RR issues in the post-sox period, all else equal. Because SOX was mandatory for both U.S. and foreign public issuers, we also predict a similar effect between them, all else equal. Finally, SOX also predicts that the proportion of RR issues will decrease and that of non-rr issues will increase, all else equal. In sum, the SOX hypothesis 17

18 makes three empirical predictions in post-sox period compared to pre-sox period: (1) all else equal, the average yield spreads for RR issues will be lower, but no effect for non-rr issues; (2) all else equal, the proportion of RR (non-rr) issues will be lower (higher); and (3) all else equal, the decline in yield spreads and the proportion of RR issues will be similar for U.S. and foreign 144A issuers. The GLOBAL effect, on the other hand, is due to the increased market competition between U.S. and international bond markets, and hence, is likely to impact average yield spreads and security market choices of all 144A issues, regardless RR or non-rr. As discussed in Section 2.2, a significant decline in cost of debt and an increase in access to financing in public debt markets in post-sox period could be due to GLOBAL effects. We expect a similar and even stronger GLOBAL effect on the yield spreads of private debt issues for two reasons. First, lowering of FASP and other regulations lowered the barriers to entry of non-european institutional investors in the European countries which will fuel the expansion of hedge funds and private equity funds in Europe, spurring the demand for European and U.S. private and speculative debt issues (Pagano and Von Thadden, 2004). Second, the simultaneous 144A and RS issues provide a direct and convenient channel for 144A issuers to reach European and home investors. Consequently, we expect a decrease in yield spreads for both RR and non-rr issues, and an increase in the proportion of RS issues in the post-sox period, all else equal. However, in contrast to SOX, the GLOBAL hypothesis predicts a stronger effect on yield spreads and security market issuance of foreign 144A issuers because of two reasons. First, GLOBAL is likely to reduce the pre-sox yield premium that foreign 144A issuers paid relative to their U.S. peers because of home bias and investor recognition factors, all else equal. Second, foreign 18

19 144A issuers also have a home bias and visibility factor advantage among international and home investors. In sum, the GLOBAL hypothesis makes three empirical predictions: (1) all else equal, post-sox average yield spreads will be lower for both RR and non-rr issues; (2) all else equal, the post-sox decline in yield spreads will be larger for foreign 144A issuers relative to U.S. issuers; and (3) all else equal, the proportion of RS issues will increase in post-sox period and the effect will be stronger for foreign issuers. Based on the above discussion, the following three implications distinguish between SOX and GLOBAL hypotheses: (1) a post-sox decline in yield spreads for non-rr issues will be consistent with GLOBAL, not with SOX, all else equal; (2) a larger post-sox decline in foreign 144A issuers (RR and non-rr) relative to the U.S. issuers will be consistent with GLOBAL but not with the SOX hypothesis; and (3) a post-sox increase in the proportion of RS issues and a larger increase for foreign issuers will be consistent with the GLOBAL hypothesis, not with the SOX hypothesis, all else equal. 3. Data, Sample Characteristics and Control Variables In this section, we discuss our data, sample characteristics, and control variables used in the empirical analysis The data and sample We use Thomson Reuters SDC Global New issues database as the primary data source to indentify all Rule 144A debt issued by the U.S. and foreign non-financial firms during the

20 2007 period. 12 Following prior 144A studies (Livingston and Zhou, 2002; Chaplinsky and Ramchand, 2004; Huang and Ramirez, 2010), we restrict the sample to non-convertible, fixed or zero coupon rate debt, and exclude perpetuity debt. We label RR and RS issues based on the indicator variables Registration Rights Issue Flag and Regulation S Issue Flag in the SDC database, respectively. To adjust for the effect of inflation rate, we convert the nominal dollar proceeds into the constant dollar, as of January 2000, using monthly consumer price index values. Since SOX was implemented on July 30, 2002, we define the pre-sox period as January 1, 1997 to July 29, 2002, and the post-sox period as July 30, 2002 to December 31, Our sample has 5,569 Rule 144A observations, including 2,113 issues with registration rights and 2,081 issues with Regulation S. 13 [Insert Table 1 about here] Table 1 provides descriptive statistics for 144A issues in the pre- and post-sox periods. Among 5,569 issues, 73% of them were issued by the U.S. firms. Panel A shows that the number of 144A issuances declined by 15% from 3,008 issues in the pre-sox period to 2,561 after SOX. The total amount raised through 144A issues remained almost the same but the amount raised by the U.S. issuers declined by $54 billion whereas that for foreign issuers increased by $47 billion. Furthermore, the increase in median proceeds is much larger for foreign issuers (30%) relative to the U.S. issuers (11%). 14 More information is revealed when we compare the differences in RR and RS subsamples. While the number of RR issues declined from 1,568 in pre-sox to 545 after SOX(a decline of 65%), the number of RS issues almost doubled from 732 in pre-sox to 1, We restrict our sample to because SDC data for yield spread and issuers credit ratings are not available for many 144A issues prior to We exclude post-2007 data because of the global financial crisis. 13 RR and RS are not mutually exclusive in 144A issues. In our sample, about 13% of 144A issues are both RR and RS. We also check for outliers and remove four issues with unreasonable yield spreads. 14 We test the subsamples in Panel A separately for pre- and post-sox difference and find that the difference is significant by using the mean and median tests. 20

21 after SOX and the effect is similar when we compare the total raised proceeds. These results indicate that foreign firms have increased their 144A debt offerings whereas the U.S. issuers decreased their use of 144A debt market. We conduct a multivariate analysis of the shift in security choices in Section 5. Panel B reports the distribution of 144A debt by quintiles based on the proceeds and exhibits significant heterogeneity and changes in post-sox period. From 1997 to 2007, the top quintile issuers raised 49% of the total proceeds whereas bottom quintile issuers raised only 5%. The proportion of proceeds in quintiles 1 (smallest size) and 2 (second smallest) experienced substantial decline after SOX, implying that SOX may have adversely affected smaller firms access to the 144A market. Panel C compares the industrial distribution of 144A issues using Kenneth French s 10 industry classification. 15 Telecommunication, manufacturing, and utility firms are the most active issuers, accounting for 22%, 14% and 12% of the total proceeds respectively but their proportions vary in the pre-and post-sox periods. For example, telecommunication companies account for 30% of the total proceeds in the pre-sox period but only 13% in the post-sox period whereas the percentage of utility firms increases from 9% to 15%, suggesting industry may drive some of the differences in pre-and post SOX issuances Control variables We draw from prior studies on 144A and public debt markets to include several variables related to issue and issuer characteristics and market conditions. Appendix A provides detailed definitions of the variables with references to the prior studies. We use the same control variables in both yield spread and security choice regressions. The first variable is issuer s credit rating 15 We collect data from 21

22 which is considered as one of the most important variables in capital raising literature (e.g., Blackwell and Kidwell, 1988; Denis and Mihov, 2003; Perraudin and Taylor, 2004; Kisgen, 2006; Faukender and Petersen, 2006; Mittoo and Zhang, 2010; Peristiani and Santos, 2010). We use S&P and Moody s rating information to measure issuer s credit risk, and translate the ratings into numerical scores - where credit rating takes a value of 22 if the S&P rating is AAA or Moody s rating is Aaa. The value declines by one to 21 if the S&P rating is AA+ or Moody s rating is Aa1, and so on. The lowest value is 2, corresponding to the S&P or Moody s rating of C. If a firm is rated by both rating agencies, we take the average of the corresponding numerical value. Dennis and Mihov (2003) find that debt issuers credit quality is strongly associated with their market choice decisions - the lowest credit rating issuers tend to use 144A market, medium credit quality firms borrow from banks, and the highest credit quality firms prefer public bond markets. Gilson and Warner (1998) show that firms with high growth but experiencing declining operating performance issue junk rated bonds to pay down bank loans in order to maintain their financial flexibility. Because higher credit quality firms will have lower financial risk, we expect a negative relation between yield spread and credit ratings and a positive relation between credit ratings and the probability of public debt and RR offerings. The next two control variables measure the degree of information asymmetry for issuers. Public equals one if the firm is a publicly traded firm, and First Time Issuer equals one if the firm has never issued fixed income securities in the SDC database since 1970, and both equal zero otherwise. Firms with public status and seasoned issuers will have lower asymmetric information compared to private firms and first time issuers. Lu et al. (2010) find that investors charge a high risk premium for both information uncertainty and information asymmetry, after controlling for credit ratings and other determinants of yield spread. We will expect Public (First 22

23 Time Issuer) to be negatively (positively) related to the yield spread, and positively (negatively) related to the propensity of RR issues. The next three variables, Log (Years to Maturity), Log (proceeds) and Senior, measure issue characteristics. Log (Years to Maturity) is the natural logarithm of years to maturity and is likely to be negatively related to the yield spread since riskier firms are less likely to be able to issue long maturity debt (Diamond, 1991). Log (proceeds) proxies for issuer size since large issue size tends to be offered by large firms and is likely to be negatively related to the yield spread because large issuers have more publicly available information, and more likely to have a credit rating (Faulkender and Petersen, 2006). Senior is a dummy variable that equals one for a senior debt issue and zero for a subordinate debt issuance and is likely to be less risky, and therefore, have a lower yield spread compared to subordinated debt. Finally, we use Default Premium to control for market conditions. Default Premium is computed as the yield difference between Moody s Baa and Aaa bond based on the U.S. Federal Reserve historical data on interest rates (H.15) at the month end when a 144A debt was issued. We expect that Default Premium will be positively associated with the yield spread because a low default premium implies a better issuance environment due to low market opaqueness for low quality debt issuers and evidence shows that managers can probably time the market for debt issuances (Iannotta et al., 2012; Barry et al., 2009). [Insert Table 2 about here] Table 2 provides descriptive statistics for key variables and their comparisons in the preand post-sox periods and shows significant changes in several variables. The median firm size of 144A issuers, measured by total assets before issuance increases from $1.4 billion in pre-sox to $2.5 billion after SOX. The relative issue size, measured as the dollar amount of proceeds 23

24 adjusted by the total assets, on the other hand, decreases from 0.14 to 0.10 that may be due to the larger firms using the market post-sox. The gross underwriter spread (%), also declines in post- SOX period consistent with Peristiani and Santos (2010) but may also reflect more issuances by larger and higher quality firms in post-sox period. 16 The average (median) yield spread declines from 399 (388) basis points in the pre-sox period to 395(372) basis points after SOX. The average credit rating an important determinant of the yield spread declines significantly from before SOX to 9.97 and the share of speculative grade issues increases from 62% in pr- SOX to 72% after SOX. The proportions of public firms or first time issuers are approximately the same in the pre- and post-sox periods. In sum, the univariate results provide mixed results, suggesting a need for multivariate analysis to control for different effects. 4. The Evidence on SOX and GLOBAL Effects in 144A Market 4.1. Distinguishing SOX and GLOBAL Effects on 144A Debt Issuers Yield Spreads In this section, we test the implications developed in Section 2 that distinguish between SOX and GLOBAL effects on yield spreads. Recall that both SOX and GLOBAL predict a post- SOX decline in yield spread but for different subsamples. SOX attributes this to a better protection and a lower yield spread only for RR issuers whereas GLOBAL is due to the competitiveness forces in international bond markets that reduce cost of debt for both RR and non-rr issuers. SOX also predicts a similar decline for U.S and foreign firms whereas GLOBAL predicts a larger decline for foreign issuers. To test these predictions, we specify the following regression: 16 Note we do not include total assets before issuance, relative issue size and gross underwriter spread (%) in our multivariaite analysis because about 50% of 144A issues in the SDC database do not report information on those variables. We report this information in the univariate section to enhance readers understanding about our sample characteristics. 24

25 YLDSPD i,t = α + β 1 SOX + β 2 RR i,t + β 3 (SOX*RR) i,t + f (Issue and issuer s characteristics) i,t + f (Market conditions) t + σ t (industry dummies) t + t (country dummies) t + ε i,t (1) where the dependent variable is yield spread (YLDSPD), and i and t represent i th issuance and t th year, respectively. SOX is an indicator variable that equals on for issuances before July 30, 2002, and zero otherwise. RR and RS are dummy variables that equal one for a RR or RS issue respectively and zero otherwise. The remaining control variables are defined in Section 3. We estimate regression (1) using Ordinary Least Squares (OLS) with robust standard errors corrected for heteroskedasticity (White, 1980) and clustering at the firm level (Rogers, 1993). We also include industry dummies (based on French s 10-industry classification) and country dummies to control for the fixed industry and country effects, respectively. [Insert Table 3 about here] We report the regression results in Table 3. In the first column, we include SOX and the remaining control variables. The coefficient on SOX in this regression is negative and significant (-0.58, t=12.21), indicating a 58 basis points decline in average yield spread for all 144A issues after SOX. In column 2, we add RR and RR*SOX and find that both coefficients are not significant but the coefficient on SOX remains negative (-0.60) and significant (t=9.09). Since SOX predicts a decline in yield spread only for RR issues, our finding of a similar decline in yield spread for RR and non-rr issues is consistent with GLOBAL, not SOX. To test the predictions of SOX and GLOBAL about the effects on foreign versus U.S. issuers yield spreads, we add FRN, FRN*RR, FRN*SOX and FRN*SOX*RR in column 3. The coefficient on FRN is positive (0.43) and significant at the 1% level, which suggests that foreign 144A issuers paid a yield premium of 43 basis points compared to their U.S. counterparts prior to SOX. However, the coefficient on SOX*FRN is negative (-0.24) and significant at the 10% level, which suggests that this premium declined by 24 basis points after SOX. This result is also 25