The Impact of Mandatory IFRS Adoption on International Cross-listings

Transcription

1 The Impact of Mandatory IFRS Adoption on International Cross-listings Presented by Dr Jeff Ng Assistant Professor The Chinese University of Hong Kong # 2014/15-03 The views and opinions expressed in this working paper are those of the author(s) and not necessarily those of the School of Accountancy, Singapore Management University.

2 The Impact of Mandatory IFRS Adoption on International Cross-listings Long Chen School of Management George Mason University Jeff Ng School of Accountancy The Chinese University of Hong Kong Albert Tsang School of Accountancy The Chinese University of Hong Kong April 2014 Abstract Using comprehensive international cross-listing data collected from 34 (50) home (target) countries, we examine whether mandatory IFRS adoption facilitates firms cross-listing activities. Our results show that following mandatory IFRS adoption, firms exhibit significantly higher cross-listing propensity and tend to cross-list their securities in more countries. We also find that cross-listing firms from countries with mandatory IFRS adoption are more likely to cross-list their securities in IFRS-adopting countries and countries with larger and more liquid capital markets following IFRS adoption. Further corroborating our results, we find that IFRS adoption has a greater impact on mandatory IFRS adopters from countries with larger accounting changes, lower disclosure requirement and less access to external capital prior to the IFRS adoption. Collectively, our findings are consistent with the notion that mandatory IFRS adoption facilitates firms cross-listing activities and highlight the importance to consider the change in firms crosslisting incentives in examining the capital market consequences of mandatory IFRS adoption. Keywords: Mandatory IFRS Adoption; Cross-listing; Accounting Difference We acknowledge the helpful comments of Shiheng Wang, Shijun Cheng, Hans Christensen, Angela Davis, Nick Dopuch, Ron King, Gerald Lobo, Xiumin Martin, Min Shen, Hollis Skaife, Gnanakumar Visvanathan, Terry Warfield, Fei Xie, and seminar participants at National University of Singapore, University of Wisconsin-Madison, University of Maryland, University of Houston, University of Texas at Dallas, National Taiwan University, Nankai University, Central University of Finance and Economics, University of International Business and Economics, 2013 DC Area Accounting Symposium, 2014 International Conference on Accounting and Technology, 2013 Chulalongkorn Accounting and Finance Symposium, and 4 th International Conference on Corporate Governance in Emerging Markets. We also gratefully acknowledge the financial support of Research Grants Council of HKSAR. 1

3 The Impact of Mandatory IFRS Adoption on International Cross-listings INTRODUCTION Differences in disclosure requirements and financial reporting standards across countries create barriers to cross-border capital flows, and thus have led to a demand for the globalization of accounting standards, i.e., International Financial Reporting Standards (IFRS). One primary assumption held by major international organizations such as the International Organization of Securities Commissions (IOSCO) 1 in seeking to converge global accounting standards is that a single set of high-quality global accounting standards, such as IFRS, will reduce cross-border information and regulatory barriers, thereby facilitating international capital-raising and security listing activities. For example, IOSCO believes that cross-border offerings and listings would be facilitated by high-quality, internationally accepted accounting standards that could be used by incoming multinational issuers in cross-border offerings and listings (IOSCO 2000). With similar belief, accounting standard setters of non-ifrs-adopting countries such as the U.S. and Japan also actively seek to converge local accounting standards with IFRS in recent years (SEC 2000). 2 Although IFRS have generated a high level of worldwide interest (e.g. see FASB 2013) and a growing body of literature examining the various economic consequences of mandatory IFRS adoption (e.g. Daske et al. 2008), surprisingly, the question of whether mandatory IFRS adoption facilitates firms cross-border stock listing activities one of the major motives behind the convergence of accounting standards across different jurisdictions (IOSCO 1998) remains unaddressed. Given the growing importance of foreign financing and the fact that by the end of 1 Formed in 1983, the IOSCO (of which the SEC is a member) comprises securities regulators from more than 200 organizations around the world. Together, they represent over 95% of the world s stock markets. 2 For example, in the U.S., Section 509 of the National Securities Markets Improvement Act of 1996 states that establishment of a high-quality comprehensive set of generally accepted international accounting standards (i.e. IFRS) in cross-border securities offerings would greatly facilitate international financing activities. 2

4 2010 over 120 countries around the world had either required or permitted IFRS (ACCA 2011; Horton et al. 2013), it is important to examine whether and how mandatory IFRS adoption affects firms cross-listing activities. Ex ante, it is unclear how mandatory IFRS adoption would affect firm s cross-listing activities because a firm s cross-listing decision depends on the relative costs and benefits brought out by the listing (Piotroski and Srinivasan 2008). On the one hand, evidence suggests IFRS adoption increases the quantity of required disclosures and the quality of financial information, thereby enhancing financial statement comparability across countries(e.g. Ashbaugh and Pincus 2001; Hong et al. 2013; Florou and Pope 2012; DeFond et al. 2011) 3. As a result, mandatory IFRS adoption can have the potential to increase firms cross-listing incentives by reducing firms perceived disclosure cost concerns associated with cross-listing activities, while at the same time increasing the attractiveness of firms securities to foreign investors. On the other hand, prior studies identify various capital market benefits associated with mandatory IFRS adoption, such as improved financial analysts information environment (Tan et al. 2011; Horton et al. 2013), higher cross-border investment from foreign funds (Covrig et al. 2007; Florou and Pope 2012), and lower cost of capital (Li 2010). To the extent that these increased capital market benefits improve firms access to external capital in their home countries and other cross-listing related costs are non-trivial 4, the need for firms to cross-list in order to access foreign capital 5 (see e.g. 3 Prior studies argue that significant differences in accounting standards across countries result in higher information asymmetry between a firm and its foreign investors, increasing the disclosure and transaction cost concerns of crosslisting firms, which in turn impede firms foreign cross-listing incentives (Biddle and Saudagaran 1989; Karolyi 1998, 2006; Saudagaran and Biddle 1992, 1995; Huddart et al. 1999). 4 In addition to the additional disclosure costs concern, there are also a variety of other costs associated with firms cross-listing decisions. For example, monetary expense, time, and administrative effort in initial registration of cross-listed securities, recurring costs of compliance and foreign regulatory agency s jurisdiction over worldwide business practices (See Table 1 of Biddle and Saudagaran (1991) for a more comprehensive list of major accounting and regulatory costs associated with cross-listings identified from various sources including the literature on securities listing, articles in the financial press, and from onsite interviews with officials from the SEC, stock exchanges, legal consultants and managers of multinational firms). 3

5 Pagano et al. 2002) may be reduced. As such, the primary objective of this study is to empirically examine how mandatory IFRS adoption affects firms cross-listing activities. Specifically, in this study, we seek answers for the following questions: (1) after mandatory IFRS adoption, do firms become more likely to cross-list their securities in foreign countries? (2) do firms cross-list their securities in more foreign countries? (3) do firms change their cross-listing venue choices? Finally, in additional analyses, we also exploit cross-country differences to identify certain country features with which the adoption effect on cross-listings is likely to be greater. To ensure that any observed changes in cross-listing activities during the post adoption period are not attributable to general trends or contemporaneous effects from other concurrent confounding events, we follow previous IFRS studies to use a difference-in-difference research design in our tests (e.g. Hong et al. 2013; DeFond et al. 2013; Byard et al. 2011). Specifically, using a comprehensive global cross-listing dataset from 34 countries (17 IFRS- and 17 non-ifrs countries as treatment and control samples, respectively), we compare the changes in firms cross-listing activities - namely firms cross-listing propensity, the total number of countries where firms cross-list their securities, and firms cross-listing venue choice - of our treatment sample from the pre-mandatory-adoption period to the post-mandatory-adoption period using different benchmark samples including all non-ifrs-adopters and Propensity-Score-Matched (PSM) non-ifrs adopters. We find that following mandatory IFRS adoption, firms undertake significantly more cross-listing activities (i.e. mandatory IFRS adopters have higher cross-listing propensity and tend to cross-list in more foreign countries) and also are more likely to cross-list their securities in IFRS-adopting countries and countries with larger and more liquid security 5 Prior studies suggest that cross-listing is one of the most important factors in attracting foreign investments. For example, Ammer et al. (2012) show that whether a firm cross-lists on a U.S. stock exchange is the most important determinants of investment from U.S. investors. 4

6 markets after controlling for an array of firm-, industry-, and country-level variables capturing firms various cross-listing incentives. In additional cross-sectional tests, we further find that the impact of mandatory IFRS adoption on cross-listings is greater for firms domiciled in IFRSadopting countries with greater changes in accounting standards, lower disclosure level, and less access to external capital before the IFRS adoption. These findings are consistent with the notion that the effect of mandatory IFRS adoption on firms cross-listings is stronger for firms experiencing greater changes in cross-listing incentives following mandatory IFRS adoption. We contribute to the literature in several ways. First we directly test one primary motive underlying the global convergence of accounting standards across different jurisdictions, and document an important, yet untested consequence of mandatory IFRS adoption: facilitation of firms cross-listing activities. To the best of our knowledge, our study is the first to directly examine how mandatory IFRS adoption alters managers incentives to undertake cross-listings, and our findings suggest that mandatory IFRS adoption has real consequences on corporate behavior. 6 In addition, by showing that mandatory IFRS adoption is associated with changes in the geography of cross-listings around the world, this study also provides important implication for security regulators and stock exchanges in evaluating the effectiveness of mandatory IFRS adoption in facilitating cross-border capital flows and equity listings. Second, we also add to the literature on determinants of cross-listings by investigating the effect of IFRS adoption on firms cross-listing decisions using a large dataset of international cross-listings. Adding to the findings of Sarkissian and Schill (2004) that proximity appears to be the dominant factor affecting firms cross-listing decisions, such as cross-listing venue selection, we find that the accounting proximity (i.e. the reduced accounting difference) resulting from mandatory IFRS adoption 6 Han and He (2011) provide country-level summary statistics suggesting that IFRS adoption in a country is associated with changes in stock exchange listings, but do not examine the changes in firms cross-listing incentives. 5

7 among many countries around the world fosters cross-border capital raising and stock-listing activities. Finally, prior studies document that cross-listings are associated with various capital market benefits (see for example, Biddle and Saudagaran 1991; Karolyi 1998; Pagano et al for a review). We show empirically that mandatory IFRS adoption is followed by a significant change in firms cross-listing activities, and thus our findings suggest that it is important to include cross-listing variable as an endogenous control when examining the capital market consequences of mandatory IFRS adoption. 7 The remainder of this paper is organized as follows. The next section reviews the related literature and develops our hypotheses. We discuss our empirical analysis and sample data in Sections III and IV, respectively. Empirical results and additional analyses are presented in Sections V and VI, respectively. In Section VII we conclude the study. BACKGROUND AND HYPOTHESIS DEVELOPMENT As international capital markets become more integrated and the domestic sources of external finance become increasingly limited in many countries (Leuz et al. 2010), the number of companies choosing to raise capital in foreign capital markets has increased substantially (Karolyi 2006; Gagnon and Karolyi 2010). 8 However firms who trade their securities in foreign stock exchanges face many obstacles, such as the cost of navigating a different set of GAAP and disclosure requirement and the perceived reluctance of foreign investors to own their securities due to the higher information disadvantages faced by foreign investors. For example, prior 7 Consistent with our argument, Ammer et al. (2012) show empirically that controlling for the effect of cross-listing might alter inferences in the international investment literature. 8 For example, according to the World Federation of Exchanges (WFE), an international association of 57 publicly regulated exchanges, the number of foreign listings on major member exchanges increased by 27.7 percent between 2005 and See the WFE publication Focus: The monthly newsletter of regulated exchanges available at The numbers provided by the WFE apply to all its members including both IFRS and non-ifrs mandating countries. 6

8 studies suggest that the higher costs incurred in acquiring and processing information about foreign companies relative to domestic firms have a negative effect on investors cross-border equity holding decisions (Kang and Stulz 1997; Lundholm et al. 2013; Hong et al. 2013), and thus investors tend to hold fewer stocks issued by foreign firms with lower accounting conformity and comparability (Bradshaw et al. 2004; Khanna et al. 2004). 9 The considerable accounting diversity across countries further exacerbates the information asymmetry between a firm and its foreign investors (Kang and Stulz 1997), and increases firms cost of disclosing additional information to deal with such information asymmetry during cross-listings. Consistent with this view, prior studies suggest that higher financial disclosure requirements in foreign countries and/or higher demand on information from foreign investors are one of the major factors deterring firms from cross-listing decisions (Saudagaran and Biddle 1992, 1995; Biddle and Saudagaran 1989, 1991). The disclosure costs concern associated with cross-listings is further echoed by the survey evidence from a managerial perspective. For example, in his monograph, Karolyi (1998) states that when making cross-listing decisions, though the direct legal and accounting costs can be substantial and listing requirements for overseas companies quite stringent, managers universally cite the additional disclosure requirements as the greatest challenge. As such, a key motivation behind the initiative to converge global accounting standards is to reduce the significant variation across domestic accounting standards and disclosure requirements that impede cross-border security listing and capital raising activities. Proponents of IFRS claim that mandatory IFRS adoption increases the level of required disclosures and imposes higher quality measurement and recognition rules, thereby mitigating information asymmetry between firms and their shareholders (Florou and Pope 2012). In 9 There is extensive evidence that foreign investors greatly underweight foreign equity in their investment portfolio, a phenomenon commonly known as the home bias (French and Poterba 1991). 7

9 addition, IFRS adoption also improves financial statement comparability by minimizing differences among domestic generally accepted accounting principles across countries (Ashbaugh and Pincus 2001), and reduces foreign investors information acquisition costs (DeFond et al. 2011). Thus a major benefit firms may receive after mandatory IFRS adoption is the reduced financial reporting costs across equity markets and the increased foreign investors demand for cross-listing firms equities. Supporting this argument, Ashbaugh (2001) finds that firms are more likely to use International Accounting Standards (IAS) when their securities are traded on more foreign exchanges. Also in analogy to this argument, Hong et al. (2013) find that with increased disclosure and enhanced comparability, mandatory IFRS adoption has a positive effect in reducing the information asymmetry between the firms and their foreign participants in the IPO process and helping IPO firms raise more proceeds from foreign markets after mandatory IFRS adoption. Additionally, accounting convergence following the widespread mandatory IFRS adoption around the world also adds a new proximity component (i.e. accounting proximity), which arguably could play a role that is similar to, or even more influential than other proximity variables identified by prior studies (e.g. the geographic, economic, cultural, and industrial proximity identified by Sarkissian and Schill 2004) in facilitating firms cross-listing decisions. 10 Following these arguments, to the extent that mandatory IFRS adoption provides an effective mechanism to reduce firms disclosure and transaction cost concerns associated with crosslisting decisions, we expect that firms have greater incentives to cross-list their securities on 10 For example, following IFRS adoption, firms seeking cross-listings have a larger selection of cross-listing venues. 8

10 foreign stock markets after mandatory IFRS adoption. 11 This leads to our first two hypotheses (both stated in alternative form). H1 Firms are more likely to cross-list their securities in foreign countries following mandatory IFRS adoption. H2 Firms tend to cross-list their securities in more foreign countries following mandatory IFRS adoption. Alternatively, it is also possible that mandatory IFRS adoption may reduce firms cross-listing incentives. The capital market benefits brought out by mandatory IFRS adoption, such as higher analyst following and improved analyst forecast precision (Tan et al. 2011; Horton et al. 2013), reduced information asymmetry and cost of capital (Li 2010), and increased cross-border investment from foreign funds (Covrig et al. 2007; Florou and Pope 2012; DeFond et al. 2011) 12, provide firms domiciled in IFRS-adopting countries with easier access to external capital in their home markets, thus reducing the need of incurring any additional costs associated with raising capital across borders. 13 In addition, cross-listing activities post IFRS adoption may impose additional costs on controlling shareholders because the decrease in information asymmetry after mandatory IFRS adoption is expected to help minority shareholder monitor and constrain controlling shareholders from expropriating corporate resources. Consistent with this view, Hong 11 Anecdotal evidence lends strong support to our prediction. For example, a global survey of 500 CFOs conducted in late 2008 by the Association of Chartered Certified Accountants (ACCA, a leading international accountancy body), reports that a majority of respondents agreed that IFRS has a positive effect on cross-border equity listing activities (ACCA 2011). 12 Consistent with the various capital market benefits associated with mandatory IFRS adoption identified by prior studies, a review conducted by the European Commission on the effect of mandatory IFRS concludes that there is an overall reduction in the cost of capital for companies supplying IFRS accounts (EU 2008, p. 7). 13 Recent anecdotal evidence provides support to this argument. For example, Daimler AG (the first Germany luxury carmaker listed on the NYSE since 1993) discontinued its stock-exchange listing in NYSE during 2010, stating that the objective of NYSE delisting is to reduce the complexity of financial reporting as well as administrative expenses and fees because many international investors of Daimler tend to trade in Daimler Shares in Germany (i.e. the firm s home country) in later years rather than through NYSE (i.e. the firm s cross-listing country) (see the announcement of Daimler plans to discontinue stock-exchange listing in New York posted on Daimler s website on May 14, 2010). Several other major companies, such as Bayer AG, Deutsche Telekom, E.On (formerly known as Veba), and EPCOS have also decided to delist their foreign cross-listings during recent years using similar reasons. 9

11 (2013) shows that mandatory IFRS adoption provides an effective mechanism to constrain the private benefits of control of controlling shareholders. Therefore findings of a decrease or no change in cross-listing activities will reject our first two hypotheses. Doidge et al. (2009b) argue that the attractiveness of a cross-listing to a firm could be affected by either the change of the firm itself (for example, in our study, firms mandatorily adopting IFRS) or the change at the listing location (for example, foreign countries mandating IFRS adoption). Prior research shows that firms tend to cross-list in countries where a firm s presence is supported by sufficient information (e.g. countries sharing similar proximity) to overcome the information asymmetry between firms and foreign investors (Pagano et al. 2002; Sarkissian and Schill 2009; Hung et al. 2012). Consequently, given the important role of proximity in affecting firms cross-listing decisions, another important and related question worth exploring is whether firms change their cross-listing venue choices after IFRS adoption. We predict that following mandatory IFRS adoption, firms are more likely to choose IFRSadopting countries as their cross-listing venues because not only the costs of cross-listings (e.g. disclosure costs) could be minimized, but the benefit of cross-listings (e.g. attractiveness to foreign investors) could be maximized due to the minimized accounting and disclosure distance between the firm s home and target country. Thus, our last hypothesis is: H3 Firms are more likely to cross-list in an IFRS-adopting country following mandatory IFRS adoption. EMIPIRICAL ANALYSIS Testing the Effect of IFRS Adoption on Cross-listing Propensity and Intensity (H1 & H2) To test our first two hypotheses, we estimate the following model using Panel data (firm and 10

12 year subscripts omitted for parsimony): (1) Mandatory Adopter is an indicator variable that equals 1 (0) if a firm s home country is a mandatory (non-) IFRS adopting country. Post is an indicator variable that takes the value of 1 (0) for post- (pre-) mandatory-adoption window, i.e. years ( ). The coefficient on the interaction term,, captures the change in firms cross-listing propensity or the number of cross-listing countries for mandatory IFRS adopters around IFRS adoption in 2005 relative to the corresponding change for the benchmark group. We use logistic (Poisson) regression when the indicator variable CL_Firm (count variable CL_NMarket) is the dependent variable in testing H1 (H2). 14 To adjust for possible cross-sectional and time-serial correlations, we cluster all standard errors by both year and home country. 15 We draw on the literature to identify and control for a wide range of firm-, industry-, and country-level control variables that may influence firms cross-listing incentives. The firm-level control variables include: (1) Relative Size, i.e. a firm s market value of equity as a percentage of its domestic equity market, to control for a firm s capital demand in a country as obtaining 14 The number of market measure (CL_NMarket) is a count variable ranging from zero to four (i.e., the maximum number of countries in which a firm cross-lists in a single year) and is zero for a large portion of sample (non-crosslisted) observations. To incorporate the fact that many sample firms do not cross-list, we also estimate the model by employing a zero-inflated Poisson specification and Tobit model for our intensity tests. Our results remain qualitatively unchanged. 15 In addition, we also alternatively conduct all of our tests with standard errors clustered by both year and firm and our results remain qualitatively unchanged. 11

13 capital in a firm s home country is constrained when a firm already represents a greater proportion of market capitalization (Biddle and Saudagaran 1991; Ashbaugh 2001); (2) ROA, to control for differences in profitability (Fuerst 1998); (3) Accruals, to control for earnings opacity (Bhattacharya et al. 2003); (4) Analyst Following, to control for the information environment of the firm (Saudagaran and Biddle 1995); (5) Big4Auditor, to control the quality of a firm s auditor; (6) Geographic Segment, to control for foreign exposure (Saudagaran 1988); (7) Sales Growth, (8) Capital Expenditure, (9) Market-to-Book ratio, to control for growth prospects (Coffee 2002; Doidge et al. 2004; Foucault and Gehrig 2008); (9) Leverage, (10) Interest, (11) Cash, to control for the need to access external capital (Lins et al. 2005); (12) Institutional Holding, and (13) Insider Holding, to control for the stock ownership and the private benefits of control (Coffee 1999, 2002; Doidge 2004; Doidge et al. 2004, 2009a). 16 Industry-level controls including (1) Industry Competition, to control for level of competition within a country; (2) HiTech, to control for high tech industry; and (3) World Competition, to control for industry competition at the global level as more companies now compete in global product markets. Finally, we also include country, industry and year indicators to control for possible fixed effects. We obtain all of the firm-level variables from Capital IQ-Compustat to reduce the loss of observations with the exception of Analyst Following, which is obtained from I/B/E/S and assumed to be 0 when missing. All continuous variables are winsorized at the 1 st and 99 th percentiles. 17 To isolate the effects of the mandatory IFRS adoption from other possible confounding concurrent events that might also affect firms cross-listing incentives, we include an array of 16 Presumably, some of the explanatory variables could be affected by the IFRS adoption, or recorded in different ways (e.g. recorded using different GAAP) for firms from treatment and benchmark samples. In robustness test, we find that the results are robust to allowing the coefficients on the control variables to vary across the pre- and postmandatory-adoption periods, and across the treatment and benchmark firms. 17 Our regression results remain qualitatively similar with or without the winsorization of continuous variables included in our model. 12

14 country-level variables in addition to all the firm- and industry-level variables. These variables include (1) Investor Protection, (2) Regulatory Quality, (3) Market Development, (4) Rule of Law, (5) Access to Capital, and (6) an indicator variable indicating whether there is Substantive Enforcement bundled with mandatory IFRS adoption in a country (see Christensen et al. 2013). 18 With the exception of Substantive Enforcement, all variables (i.e. variable (1) to (5)) are measured by year to capture the possible time-series changes in country-level institutional and regulatory environment. We include detailed variable definitions in Appendix 1. Testing the Effect of IFRS Adoption on Cross-listing Venue Choices (H3) We use the following model to test the effect of mandatory IFRS adoption on firms cross-listing venue choice (CL_Venue). To examine the change in cross-listing venue choices (CL_Venue) post mandatory IFRS adoption, we only focus on cross-listing samples in this test. To test H3, we define CL_IFRSMarket to indicate whether a firm cross-listed its security in an IFRS-mandating country in a given year. Furthermore, we also use two alternative venue measures, i.e. CL_LargerMarket, indicating whether a firm cross-listed its security in a country with larger stock market, and CL_MoreLiquidMarket, indicating whether a firm cross-listed its security in a country with more liquid stock market to test the possible changes in firms cross-listing venue 18 Countries with both mandatory IFRS adoption and substantive enforcement changes are provided by Christensen et al. (2013) and include Finland, Germany, Hong Kong, the Netherlands, Norway, Sweden, Turkey, and the U.K. 13 (2)

15 choices post IFRS adoption. In addition to all the country-level controls described in model (1), we also include in model (2) several variables of proximity between a firm s home and crosslisting target countries to control for the geographical, political, economic and cultural proximity which might influence firms cross-listing target choices (e.g. Pagano et al. 2002; Sarkissian and Schill 2004). These variables include (1) the physical distance between the capital cities of the home and target countries (Distance), (2) an indicator variable indicating whether the home and target countries are contiguous (Contiguous), (3) an indicator variable indicating whether the home and target countries adopt the same official language (Common Language), (4) an indicator variable indicating whether the home and target countries ever had a colonial link (Colony), (5) an indicator variable indicating whether the stock exchange of a firm s home country is merged or cooperated (e.g. through joint venture) with the stock exchange of target country (Stock Market Connection), and (6) an indicator variable indicating whether the home and target countries use the same currency (Common Currency). Furthermore, since any changes in the geography of equity listings could also be related to the characteristics of the venue countries concerned (Pagano et al. 2002; Piotroski and Srinivasan 2008; Doidge et al. 2009b), in model (2) we also include additional controls for the characteristics of each cross-listing venue country. These variables include all the country-level characteristics described in model (1), plus two additional variables controlling for the competition among different stock exchange venues: Stock Market Competition and CL Firms Market Share Stock Market Competition is measured by the Stock Market Herfindahl index (-1), where the Stock Market Herfindahl index is calculated as the sum of the squares of percentage of cross-listers in a country, where the percentage of cross-listers is defined as the total number of cross-listers in a country divided by the total number of cross-listers of all countries in a given year. Thus a higher value of Stock Market Competition suggests higher level of stock markets competitiveness across venue countries. CL Firms Market Share is the market share of cross-listers in a given venue country and year, which is calculated as the total number of cross-listers in a host country divided by the total number of cross-listed firms hosted by all countries in a given year. Thus, a higher percent suggests the venue country is more successful in attracting foreign firms to list securities on its stock markets. 14

16 SAMPLE AND DATA Sample Selection Our treatment sample includes all mandatory IFRS adopters from countries that mandated IFRS adoption in year We exclude 2005 (i.e. the mandatory IFRS adoption year) from the sample period as the effect in the transition year is less clear. To reduce the potential effect of other concurrent confounding events (e.g. the Sarbanes Oxley Act in 2002) 20, we use a relatively short event window and focus on the last two years before (i.e. pre-mandatory-adoption window) and the first two years after the mandatory IFRS adoption (i.e. post-mandatory-adoption window). 21 Following prior studies (e.g. DeFond et al. 2013; Hong et al. 2013), we also use different benchmark groups to facilitate better comparison, i.e. all non-ifrs adopters from non- IFRS-adopting countries, and non-ifrs adopters matched using Propensity-Score-Matching (PSM) methodology based on observable firm-level characteristics (Dehejia and Wahba 2002). 22 To construct the international cross-listing sample, we directly collect data on foreign equity listings for all firms covered by Capital IQ-Compustat (hereafter CIQ). 23 Specifically, we construct our sample as follows: First, beginning with the universe of publicly traded firms 20 SOX could potential play a role in our setting because prior studies (e.g. Piotroski and Srinivasan (2008)) show that smaller foreign firms may have shied away from the U.S. stock markets and are more likely to be cross-listed in the U.K. stock market following the enactment of SOX. In robustness test, our conclusion remains the same after excluding firms listed in either the U.S. or the U.K. 21 We use actual GAAP used by each firm in each year to identify mandatory IFRS adopters. In addition, we also vary the number of years included in the analysis (e.g. one to three year before and one to three year after the mandatory IFRS adoption) to determine and evaluate the length of time which firms take to make cross-listing decisions after IFRS adoption. Using one-year or three-year periods before and after IFRS adoption as the pre- and post-mandatory-adoption windows does not change our results. We thank our reviewer for making this suggestion. 22 In additional analyses, we also discuss the results using voluntary IFRS adopters and US GAAP adopters as alternative benchmarks. 23 In contrast to previous international cross-listing studies (e.g. Sarkissian and Schill 2004, 2009, 2012) which collected cross-listings data from major stock exchanges, our study uses directly collected data on foreign equity listings from each firm, which provides us with more complete information about each firm s security listing status over time. In addition, stock exchanges also differ in how they define foreign firms. Thus, another major advantage of our data is that we can ensure a consistent definition of cross-listing across different firms and stock exchanges. For example, several major stock exchanges we contacted (e.g. exchanges of Hong Kong, Singapore, the U.S. and the U.K.) confirmed that country of incorporation is used to classify whether a firms is domestic or foreign. Other exchanges, however, indicated that they use a firm s headquarter location to define the country origin (e.g. Luxembourg Stock Exchange). 15

17 covered by CIQ, we eliminate firms without primary identifier codes, firms that are investment funds or trusts, and firms listed in tax havens such as Bermuda and the Channel Islands. Next, for firms with secondary securities listed in foreign countries, we collect information on all of their publicly traded tickers together with the stock exchange on which each ticker is listed. Firms with at least one secondary security listed and traded in a country different from the primary listing country are defined as cross-listing firms. 24 We further remove securities listed over-thecounter such as pink sheets, over-the-counter bulletin boards, Norway OTC, and Deutsche Börse 25, because the disclosure requirements of these exchanges are generally substantially lower than those of the major stock exchanges in the same country. Our final sample consists of a total of 1,181 cross-listing firms by 2007, among which 608 firms are from IFRS-adopting countries, and 14,761 non-cross-listing firms as of 2007, among which 5,750 firms are from IFRS-adopting countries. 26 We start by providing a broad picture on the distribution and shifts in the geography of cross-listings for firms from IFRS- versus non-ifrsadopting countries in Table 1. Panel A reveals that during the first two years subsequent to mandatory IFRS adoption (i.e. from 2006 to 2007), there is a substantially larger increase in the number of cross-listing firms originating from countries with mandatory IFRS adoption (a 96.8 percent increase) relative to the last two years right before the adoption year, i.e. from 2003 to 2004 (a 12.7 percent increase), while the increase in the number of cross-listing firms from non- 24 We define a firm s home country based on its primary listing stock exchange because a firm s disclosure is affected mainly by the exchange in which the firm is primarily listed. To ensure that our results are not sensitive to varying home country definitions, we redefine a cross-listed firm s home country based on its headquarters or incorporation country rather than its primary listing country and repeat all of our analyses. All of the results are robust to this alternative home country definition (untabulated). 25 Deutsche Börse (DB) consists of both a regulated and an over-the-counter (OTC) market/segment. However, since we cannot separate regulated listings from OTC listings based on a security s ticker and OTC market/segment makes up to 99% of the DB observations, we consider the entire DB as an OTC market. 26 The fact that only a relatively small percentage of firms around the world (about 4-5 percent based on our sample) have cross-listed securities on foreign markets is consistent with the argument of non-trivial cross-listing costs associated with foreign cross-listing decisions (Biddle and Saudagaran 2001). 16

18 IFRS countries remain similar across two periods (increases of 10.1 percent and 16.7 percent for pre- and post-ifrs periods, respectively). This substantially larger increase is heavily represented by a larger increase in cross-listings by firms from IFRS-adopting countries that cross-list in other IFRS-adopting countries 27 (a percent increase during the post-ifrs period vs. an 18.3 percent increase during the pre-ifrs period). Together, Panel A suggests that mandatory IFRS adopters not only appear to become more outward oriented, IFRS-adopting countries also seem to become more popular cross-listing venues following IFRS adoption. In Panel B, we compare the total number of firms with U.S. vs. without U.S. cross-listings between firms from countries with mandatory IFRS adoption and those without. Again, results suggest that mandatory IFRS adopters are more likely to cross-list in countries other than the U.S. (which include a large number of IFRS-adopting countries) in the post-mandatory-adoption period. Overall, our findings in Table 1 provide preliminary support to the assertion that mandatory IFRS adoption influences firms cross-listing decisions, with IFRS adopting countries becoming more attractive as cross-listing venues. [Insert Table 1 about here] Summary Statistics Our benchmark samples include all non-ifrs adopters and PSM matched non-ifrs adopters. All non-ifrs adopters are drawn from the 17 non-ifrs-adopting countries in our sample. For PSM non-ifrs adopters, we match each treatment firm to one benchmark firm from non-ifrsadopting countries with replacement using the PSM approach (Dehejia and Wahba 2002). Specifically, in the first step, we model the probability of being a mandatory IFRS adopter using the full sample. In the second step, we estimate the propensity score for each firm from both 27 We define a firm as cross-listed in an IFRS-adopting country (the U.S.) if it has at least one cross-listing in an IFRS-adopting country (the U.S.). 17

19 IFRS- and non-ifrs-adopting countries using the predicted probability from the first step and match each treatment firm with a benchmark firm by year using the nearest propensity score. Table 2 presents cross-listing distribution and comparative statistics during the pre- and postmandatory-period by country. Evidence from Table 2 indicates that in general, both mandatory IFRS adopters (Panel A) and non-ifrs adopters (Panel B and C) increase their cross-listing propensity following mandatory IFRS adoption. On average, 3.55 percent of mandatory IFRS adopters are cross-listed in the pre-mandatory-adoption period, a ratio that increases to 7.82 percent in the post-mandatory-adoption period (representing a 120 percent increase). However the change in the percentage of cross-listing firms for non-ifrs adopters is much smaller- a 15 or 12 percent increase for all non-ifrs adopters or PSM non-ifrs adopters, respectively. Furthermore, while 11 out of 17 mandatory IFRS countries (64.7 percent) experience a statistically significant increase in cross-listing propensity following IFRS adoption, only a few non-ifrs mandating countries experience a significant increase in cross-listing propensity during the post-mandatory-adoption period. 28 [Insert Table 2 about here] Descriptive statistics (means and standard deviations) of main test variables for both the treatment and benchmark samples are reported in Table 3. Our main dependent variables are CL_Firm, CL_NMarket, CL_IFRSMarket, for H1, H2 and H3 respectively. Firms cross-listing status, CL_Firm equals 1 if a firm has at least one secondary security actively listed and traded in 28 We obtain similar conclusion by repeating the same comparative analysis on the number of cross-listing market during pre- and post-mandatory-adoption periods (untabulated). On average, mandatory IFRS adopters cross-list their securities in foreign countries during the post-mandatory-period, compared with in the premandatory-period (a statistically significant increase of in the number of cross-listing countries after the IFRS adoption). However, for all non-ifrs adopters and PSM non-ifrs adopters, the increase in the number of crosslisting countries is only (significant) and (insignificant), respectively. 18

20 a foreign country in a given year, and 0 otherwise; CL_NMarket is defined as the total number of unique foreign countries where a firm s securities are cross-listed in a given year; CL_IFRSMarket equals 1 if a firm cross-listed its security in an IFRS-mandating country in a given year, and 0 otherwise. Table 3 indicates that the differences in CL_Firm, CL_NMarket, and CL_IFRSMarket, are significant between mandatory IFRS adopters and the benchmark of all non-ifrs adopters, and such differences decrease in magnitude or become insignificant or opposite when using the benchmark of PSM non-ifrs adopters. This result suggests that the differences of our variables of interest between the treatment and benchmark samples are minimized through the application of PSM method. In particular, the results of Table 3 show that among the firm- and industry- level control variables, Sales Growth, Leverage, Cash, and Industry Competition are statistically different between the treatment firm and all non-ifrs firms before Propensity Score Matching, but none remains statistically different after matching, and only Accruals change from insignificantly different before PSM to significant after matching. Furthermore, the magnitude of difference between the treatment and benchmark samples is reduced for 11 of the 17 firm- and industry-level variables. [Insert Table 3 about here] Table 4 presents the Pearson correlation matrix for our variables of interest and all firm- and industry-level controls. As Table 4 suggests, our sample firms are more likely to be cross-listed (CL_Firm) and cross-list in more foreign countries (CL_NMarket) when they are from mandatory-ifrs-adopting countries (Mandatory Adopter) and in the post-adoption period (Post). The cross-listing measures (CL_Firm, CL_NMarket) are also positively correlated with relative firm size (Relative Size), number of analyst following (Analyst Following), presence of Big 4 auditor (Big4Auditor), number of geographic segments (Geographic Segment), capital 19

21 expenditure (Capital Expenditure), institutional ownership (Institutional Holding), and negatively correlated with cash (Cash), insider ownership (Insider Holding), and industry competition within the country (Competition) and around the world (World Competition). [Insert Table 4 about here] RESULTS Test of H1 and H2 Regression Results Table 5 (Table 6) report tests of H1 (H2) using logistic (Poisson) regression estimates. Across all of the tests in both tables, the coefficient on Mandatory Adopter Post is consistently positive and significant, supporting H1 and H2. These results suggest that relative to either all non-ifrs or PSM non-ifrs adopters, more mandatory IFRS adopters tend to cross-list their securities in foreign stock markets during the post- than pre-mandatory-adoption period (Table 5), while these mandatory IFRS adopters also tend to undertake cross-listing activities in more foreign countries on average (Table 6). Specifically, the odds ratio of firms being cross-listed in post-ifrs period relative to pre-ifrs period is 3.3 or 4.0 times as high for IFRS-adopters as for all non-ifrs or PSM non-ifrs adopters, respectively (Table 5), and the incidence rate ratio of number of foreign countries that host a firm s cross-listings in the post- relative to pre-ifrs period for IFRS-adopters is 2.1 or 2.6 times that for all non-ifrs or PSM non-ifrs adopters, respectively (Table 6, columns I and II) 2930 In columns III and IV of Table 6, we also re-estimate model (1) to test H2 after limiting the sample to firms with cross-listings only and our inference 29 For logistic (Poisson) regressions with an interaction term, i.e. Mandatory Adopter Post, exponentiating the coefficient on interaction implies how the odds ratio of being cross-listed (the incidence rate ratio of cross-listing market counts) in the post-ifrs relative to pre-ifrs period, differs between mandatory IFRS-adopters and non IFRS-adopters. 30 In robustness tests (untabulated), we partition the sample into treatment and benchmark samples and estimate the coefficient on the variable Post separately for each subsample and then test the differences in estimated coefficients using a Chow-test (Chow 1960). Consistent with our results reported in the interaction model, we find that the estimated coefficient of Post for mandatory IFRS adopters is statistically larger than that for either non-ifrs adopters or PSM non-ifrs adopters. 20

22 remains unchanged. [Insert Table 5 & 6 about here] Test of H3 Regression Results The results on changes of cross-listing venue choices are presented in Table After controlling for all firm-, industry-, and country-level control variables and a series of proximity measures which may potentially affect firms cross-listing venue choices as discussed above, we consistently find positive and significant coefficients on Mandatory Adopter Post in all three models for both benchmark samples as reported in Table 7. Column I in Panel A (Panel B) shows that in general, compared to all non-ifrs (PSM non-ifrs) adopters, cross-listing firms from countries with mandatory IFRS adoption are significantly more likely to cross-list their securities in other IFRS-adopting countries after the mandatory adoption year. The economic significance of such difference is also large: the odds ratio of firms choosing an IFRS-adopting country as cross-listing venue in post-ifrs period relative to pre-ifrs period is 7.3 or 17.5 times as high for IFRS-adopters as for all non-ifrs adopters and PSM non-ifrs adopters, respectively. Thus, our findings in Table 7 provide strong support for H3, that mandatory IFRS adoption alters the venue preference of cross-listing firms from IFRS-adopting countries towards countries with proximate accounting standards. In addition, using both benchmarks, we also observe a higher increase in likelihood of firms from IFRS-adopting countries to cross-list their 31 Four countries (Malaysia, Morocco, Philippines, and Sri Lanka) are included in the main test sample (34 countries) while excluded from the venue test sample (30 countries). We drop Malaysia, Philippines, and Sri Lanka as these countries have no new cross-listings within the sample period ( & ), and drop Morocco due to missing value of firm-level variables. 32 Untabulated univariate results show that in general 92.5 percent of cross-listings by mandatory IFRS adopters are targeted at other IFRS-mandating countries during post-mandatory-adoption period, compared with a significantly lower number of 48.0 percent before IFRS adoption. In additions, mandatory IFRS adopters are on average 59.5 percent and 44.8 percent more likely to cross-list their securities in countries with larger and more liquid stock markets after mandatory IFRS adoption (all differences are significant at the 0.01 level). 21

23 securities in countries with larger stock market (Column II), and in countries with more liquid stock market (Column III) post IFRS adoption. These findings are consistent with prior findings that firms in general have higher incentive to cross-list their securities in larger and more liquid foreign markets (e.g. Pagano et al. 2002; Karolyi 2006), and furthermore support the conjecture that the wider choices of cross-listing venue provided by mandatory IFRS adoption facilitate firms cross-listing venue decisions. [Insert Table 7, Panel A & Panel B about here] ADDITIONAL ANALYSES AND SENSITIVITY TESTS Cross-Sectional Tests on the Effect of IFRS on Cross-Listing Activities To provide a better understanding of the effect of mandatory IFRS adoption on crosslistings, we further exploit cross-country differences to identify countries where the adoption effect on cross-listings is likely to be greater. We predict that holding other incentives and the regulatory environment constant, mandatory IFRS adoption has a greater effect on firms crosslisting incentives for mandatory IFRS adopters from countries experiencing larger accounting changes after switching from domestic GAAP to IFRS, countries with lower level of disclosure requirement, and with less access to external capital before the IFRS adoption because of the change in firms cross-listing incentives are likely to be greater for firms from these countries. To test these predictions, we first partition mandatory IFRS adopters into two groups based on whether they are domiciled in home countries with higher versus lower than median values of (1) GAAP Absence, (2) GAAP Divergence, (3) Disclosure Requirement, and (4) Access to Capital in the pre-ifrs adoption period. GAAP Absence captures the accounting change through the number of additional disclosures required by IFRS relative to a country s domestic GAAP. 22