The Impact of International Financial Reporting Standards on Comparability: A Test using IPO Underpricing

Transcription

1 The Impact of International Financial Reporting Standards on Comparability: A Test using IPO Underpricing Jangwon Suh University of Massachusetts, Dartmouth Donal Byard Masako Darrough Baruch College The City University of New York April 2015 ABSTRACT Using a sample of IPO firms from countries that mandated IFRS adoption in 2005, we compare IPO underpricing for firms that went public after 2005 using IFRS with that of matched IPOs that went public prior to 2005 using domestic GAAP. While mandatory IFRS adoption increases the number of listed industry peers that report in the same accounting standards (i.e., potentially comparable firms), we find that only those IFRS-reporting industry peers listed in jurisdictions with high-quality enforcement environments are associated with a reduction in IPO underpricing: only peer firms based in high quality enforcement jurisdictions provide enhanced comparability. Further, in the pre-2005 period, we find that IPOs that voluntarily adopt IFRS also experience a reduction in IPO underpricing via enhanced comparability. Our results highlight that IFRS adoption can generate a positive externality by enhancing comparability, but only when coupled with high-quality enforcement (or strong firm-level reporting incentives). Keywords: International Financial Reporting Standards (IFRS); IPO underpricing; comparability; enforcement

2 I. INTRODUCTION We examine if the adoption of International Financial Reporting Standards (IFRS) increases comparability. By greatly increasing the number of potentially comparable firms, IFRS adoption is likely to increase comparability that will benefit investors (see McCreevy 2005). Comparability is the quality of information that enables users to identify similarities in and differences between two sets of economic phenomena (Financial Accounting Standards Board, 1980). Using initial public offering (IPO) underpricing as a setting, we test whether investors benefit from enhanced comparability under IFRS compared to domestic generally accepted accounting principles (GAAP). Further, focusing on the quality of countries enforcement environments, we examine if all of the potentially comparable firms available under IFRS actually contribute to providing enhanced comparability. IPO underpricing refers to the phenomenon whereby the offer prices of the newly issued shares of firms going public tend to be lower than their first trading-day closing prices. 1 IPO underpricing has been attributed to information asymmetry among participants in the IPO process regarding the value of an IPO firm (Rock 1986; Beatty and Ritter 1986; Benveniste and Spindt 1989), and has been documented throughout the world (Loughran, Ritter, and Rydqvist 1994). Information asymmetry about firm value is particularly high before a firm goes public. To value IPO firms, investors typically use comparison measures such as industry peers priceearnings ratios. 2 By using IFRS, an IPO firm can generate a greater number of potentially comparable firms throughout the world. IFRS may therefore enable investors to extract more 1 Following prior studies, we measure the degree of IPO underpricing by the first trading-day closing price relative to the offer price. For example, when LinkedIn went public on May 19, 2011, the offer price and closing price were $45 and $94.25, respectively, giving underpricing of 109% (= $94.25/$45 1). 2 For example, in its Form S-1 registration statement filed with the Securities and Exchange Commission (SEC), Groupon, which went public on November 4, 2011, observed that among the factors to be considered in determining the initial public offering price will be. the price-earnings ratios, price-sales ratios, market prices of securities, and certain financial and operating information of companies engaged in activities similar to ours (Groupon 2011, p. 135, emphasis added). 1

3 useful information by clarifying commonalities and differences among a much larger number of industry peers. Such a potential increase in the overall level of comparability (i.e., enhanced comparability) may reduce IPO underpricing by reducing investors uncertainty and information asymmetry about IPO firms. 3 The global IPO market therefore provides an ideal setting to examine the effects of IFRS on comparability. Broadly speaking, there are two settings: (1) countries that mandated domestic GAAP reporting prior to 2005 but mandated IFRS from 2005 onward; and (2) countries that allowed voluntary IFRS adoption prior to 2005 but mandated IFRS from 2005 onward. We use both settings to examine whether the potential for enhanced comparability under IFRS is associated with a reduction in IPO underpricing. The adoption of IFRS may change IPO underpricing via two potential channels: (1) a change in IPO firms' reporting quality, and (2) an increase in the overall level of comparability for investors. Hong, Hung, and Lobo (2014) provide evidence regarding the first channel: they find that IPO firms based in countries with a larger number of accounting changes (from domestic GAAP to IFRS) and stronger implementation credibility (i.e., stronger enforcement) in their home countries experience a greater decrease in IPO underpricing following mandatory IFRS adoption. In contrast, our study focuses on the role of comparability under IFRS in mitigating IPO underpricing. By focusing on the second channel, our analysis extends and complements that of Hong et al. (2014). This paper first tests if enhanced comparability under mandatory IFRS adoption reduces IPO underpricing. Following DeFond, Hu, Hung, and Li (2011), we use listed industry peers that report in the same accounting standards as an IPO firm as a proxy for comparable firms. While mandatory IFRS adoption will obviously increase the number of potentially comparable firms, it 3 We use the term overall comparability to denote the totality of all comparability arising from comparable firms (proxied by all listed industry peers using the same accounting standards), as opposed to the marginal effect from one additional comparable firm. 2

4 may also adversely impact the level of comparability per listed peer. This is because IFRSreporting comparable firms are by definition based in different countries, which might have different operating, regulatory, cultural, and institutional environments. As a result, it is not a foregone conclusion that IFRS will lead to an increase in overall comparability that will benefit investors (by reducing IPO underpricing). While IFRS will unambiguously increase the number of potentially comparable firms, not all of these firms may provide genuine comparability that benefits investors. The financial information of some peer firms may be unreliable. Prior studies identify the quality of countries enforcement environments as a key determinant of firms compliance with accounting standards. As our second question, we test if the quality of countries enforcement environments affects comparability under IFRS. Specifically, we test if the quality of the enforcement environments in the countries where potentially comparable firms are domiciled affects comparability under IFRS. This is a central focus of our paper. While it is widely acknowledged that any increase in overall comparability from mandatory IFRS adoption is likely to depend on whether IFRS are effectively and consistently enforced across jurisdictions (e.g., McCreevy 2005), to the best of our knowledge, our study provides the first direct evidence of this possible positive externality (enhanced comparability) arising from high-quality enforcement. For our analysis of mandatory IFRS adoption, our test sample consists of IPO firms from countries that mandated domestic GAAP prior to 2005 but IFRS after Our focus is on the treatment IPO firms that went public using IFRS after 2005 (specifically, in 2006 and 2007). 4 We compare these IFRS-reporting IPO firms with benchmark IPO firms that went public using domestic-gaap prior to 2005 (specifically, in 2003 and 2004). We match our treatment firms with our benchmark firms by industry, country, and IPO proceeds. Thus, our mandatory IFRS 4 We focus our analysis on IPOs before the financial crisis period. 3

5 adoption test sample consists of a matched-pair sample of IPO firms reporting in IFRS (in the post-period) and domestic GAAP (in the pre-period); this controls for country-level institutional factors that may affect IPO underpricing. 5 Using this matched-pairs sample, we first test if the increase in the total number of potentially comparable firms following mandatory IFRS adoption is associated with a decrease in IPO underpricing, i.e., whether there is enhanced comparability under IFRS. We then test if enhanced comparability under IFRS is associated with the quality of the enforcement environments of the countries where the listed industry peers are domiciled. For our analysis of the effects of mandatory IFRS adoption, we use both: (1) change models using the mandatory adoption test sample of matched pairs of IPOs (based in countries that mandated IFRS adoption in 2005) outlined above; and (2) difference-in-differences (DID) models using both the mandatory adoption test sample and a control sample. The control sample consists of a similar matched-pair sample of IPOs drawn from countries that did not mandate the adoption of IFRS over our sample period. We find consistent results using both sets of tests. Specifically, we find that only IFRS-reporting industry peers domiciled in strong enforcement jurisdictions are associated with a decrease in IPO underpricing following mandatory IFRS adoption. Industry peers domiciled in weak enforcement jurisdictions do not appear to provide genuine comparability that benefits investors. These results are robust to various alternative specifications, including the EU s adoption of the Prospectus Directive, a concurrent regulatory change (Christensen, Hail, and Leuz 2013, 2014). 5 We exclude two groups of IPO firms from this comparison. First, some countries (e.g., Germany) permitted voluntary IFRS adoption prior to We exclude these voluntary IFRS adopters from our mandatory adoption test sample, but include them in our voluntary adoption sample (see Section V below). Second, the EU-wide mandatory adoption of IFRS in 2005 applied only to public companies listed on EU regulated exchanges (e.g., London Stock Exchange in the UK). Secondary exchanges or exchange-regulated markets such as AIM (UK), Alternext (Continental Europe), FirstNorth (Scandinavia), and Marche Libre (France) were not required to mandate the adoption of IFRS in 2005, so we exclude listings on these exchanges from our mandatory adoption test sample. 4

6 For our second set of analyses, we examine voluntary IFRS adoption in the pre-2005 period, and find that: (1) voluntary IFRS adopters are more likely to choose to report in IFRS when they have relatively fewer comparable firms reporting in domestic GAAP; and (2) controlling for self-selection, there is enhanced comparability following voluntary IFRS adoption that is associated with a reduction in IPO underpricing. This paper makes a number of contributions to several strands of literature regarding the effects of both voluntary and mandatory IFRS adoption, as well as IPO underpricing. We show that IPO underpricing is affected by enhanced comparability following mandatory IFRS adoption. Our results also show that the quality of the enforcement environments where potentially comparable firms are domiciled plays a key role in determining the level of comparability under mandatory IFRS adoption. Thus, our results highlight that rigorous implementation/enforcement is critical in generating the positive externality of enhanced overall comparability under IFRS. For voluntary IFRS adoption, our results suggest that comparability is both a factor in firms choice of accounting standards, and a benefit of voluntary IFRS adoption. Finally, our results provide new evidence that accounting comparability affects IPO underpricing. These results may be of interest to regulators and policymakers. The remainder of this paper is organized as follows. Section II reviews the prior literature and develops our hypotheses regarding mandatory IFRS adoption. Sections III and IV outline our study design, sample selection, and results related to mandatory IFRS adoption. Section V presents our second set of analyses of voluntary IFRS adoption. Additional analysis and robustness are reported in Section VI, while Section VII concludes. 5

7 II. PRIOR STUDIES AND HYPOTHESES IPO Underpricing Theoretical studies attribute IPO underpricing to uncertainty about firm value and the associated information asymmetry among the participants in the IPO market (Rock 1986; Beatty and Ritter 1986; Benveniste and Spindt 1989). Prior to an IPO, investors in public capital markets have access to only limited financial information about the firm, and therefore IPO firms are subject to a heightened level of information asymmetry. Underpricing is a mechanism to attract (or retain) asymmetrically informed investors to the IPO market. It might also incentivize investment bankers to exert effort in the underwriting process. If IFRS provide investors with more useful information than domestic GAAP, then IPO firms using IFRS may have a lower level of IPO underpricing. As discussed earlier, IFRS may provide investors with more useful information than domestic GAAP through two channels: (1) potentially higher reporting quality; and (2) greater overall comparability. We examine the second channel, i.e., how IFRS may reduce IPO underpricing via enhanced overall comparability. Mandatory IFRS Adoption and Comparability Mandatory IFRS adoption can affect overall comparability through two effects. First, when a firm goes public in the post-mandatory IFRS adoption period, it will (1) automatically acquire a greater number of potentially comparable firms: its industry peers in its home country plus IFRS-using industry peers throughout the world. While the number of industry peers reporting in the same accounting standards will increase, it is possible that (2) comparability per industry peer could be lower for IFRS than for domestic GAAP. We expect that the first effect of the far greater number of comparable firm will dominate (see below) the second effect, resulting 6

8 in enhanced overall comparability and a reduction in IPO underpricing under mandatory IFRS adoption. Our first hypothesis, stated in the alternative form, is: H1: Following mandatory IFRS adoption, enhanced overall comparability under IFRS will reduce IPO underpricing relative to that under domestic GAAP. It is possible, however, that the increased number of comparable firms under IFRS do not provide genuine comparability that benefits investors, with the result that IFRS adoption does not lead to enhanced overall comparability. While effect (1) is straightforward, effect (2) is not. The average level of comparability per listed industry peer might be lower for IFRS than for domestic GAAP, partially due to the fact that IFRS may not be consistently interpreted and enforced across countries. 6 Furthermore, as a more principles-based set of standards, IFRS may be more open to interpretation and managerial discretion (Ahmed, Neel, and Wang 2010). Thus, industry peer firms from outside the IPO firm s home country may not provide genuine comparability. However, an IPO firm has, at a minimum, the same number of comparable firms in its home country before and after mandatory IFRS adoption. These domestic listed peers (now reporting in IFRS rather than domestic GAAP) will still be subject to the same firm-level reporting incentives and are in the same enforcement environment. As a result, it is unlikely that the overall effect on comparability from the two effects could be negative, so our alternative hypothesis is that mandatory IFRS adoption will cause no change in overall comparability. Evidence so far is mixed on whether or not IFRS provide greater overall comparability than domestic GAAP. On the one hand, consistent with enhanced comparability following IFRS 6 Variation in institutions across countries may cause variation in firm-level reporting incentives (Leuz, Nanda, and Wysocki 2003; Ball 2006; Cascino and Gassen 2015), with the result that IFRS may be applied differently across countries. Firm-level reporting incentives are shaped by country-level institutions such as enforcement quality (Leuz et al. 2003; Ball 2006) and investor protection (Djankov, LaPorta, Lopez-de-Silanes, and Shleifer 2008). As a result, firms' incentives to rigorously implement the requirement of IFRS are likely to vary across countries, possibly leading to lower per firm comparability. In addition, the process whereby different jurisdictions "endorse" IFRS for use in these jurisdictions creates the possibility of carve-ins and carve-outs, potentially leading to the use of different versions of IFRS across different countries. 7

9 adoption, prior studies document that: foreign mutual fund ownership increases with the increase in the number of listed industry peers (DeFond et al. 2011); there is an increase in transnational information transfer (Yip and Young 2012; Wang 2014); and, consistent with enhanced comparability, there is an improvement in the information environment of UK firms (Brochet et al. 2013). On the other hand, Lang et al. (2010) find no evidence that mandatory IFRS adoption increases accounting comparability. 7 Using a similar approach, Cassino and Gassen (2014) conclude that the overall impact of mandatory IFRS adoption on comparability is only marginal. Comparability under IFRS: The Role of Enforcement The quality of enforcement environments is critical in determining firms actual reporting practices (Ball 2006; Ball, Kothari, and Robin 2000; Ball, Robin, and Wu 2003). Strong enforcement requires enhanced auditing and regulation, thereby limiting managers exercise of discretion and opportunism (Leuz et al. 2003). Consistent with this argument, prior studies show that the capital market benefits (e.g., increased liquidity, decreased cost of capital, and more accurate analyst forecasts) associated with mandatory IFRS adoption are concentrated among firms domiciled in stronger enforcement environments (Daske et al. 2008; Li 2010; Byard, Li, and Yu 2011). The results of these studies are consistent with the idea that the quality of the enforcement environment where mandatory IFRS adopting firms are domiciled is critical in determining how mandatory IFRS adoption affects these firms' reporting quality. We hypothesize that the quality of enforcement environments is also likely to affect the level of comparability under IFRS. However, our hypothesis is about the quality of the enforcement environments of the countries where an IPO firm s potentially comparable firms are domiciled, not where an IPO firm itself is listed. While IFRS provides a far greater number of 7 They employ a measure of accounting comparability developed by De Franco, Kothari, and Verdi (2011), which is based on how closely the earnings of two firms represent the same economic events captured by stock returns 8

10 potentially comparable firms than domestic GAAP, this will only lead to enhanced comparability if these listed industry peers faithfully and consistently implement IFRS (McCreevy 2005). The quality of the enforcement environments of the jurisdictions where an IPO firm s industry peers are domiciled will determine whether or not these industry peers faithfully and consistently implement IFRS. 8 Our second hypothesis therefore examines how the quality of the enforcement environments of the countries where an IPO firm s listed peers are domiciled affects the degree to which these listed peers enhance overall comparability and, thus, affect IPO underpricing. Stated in the alternative form, our second hypothesis is: H2: Following mandatory IFRS adoption, only those listed industry peers domiciled in high quality enforcement jurisdictions have an effect in reducing IPO underpricing. The Institutional Setting III. STUDY DESIGN AND SAMPLE SELECTION The institutional setting of the global IPO market is complex, posing a number of challenges for our study. First, the mandatory EU-wide adoption of IFRS in 2005 applied only to firms filing consolidated financial statements that were listed on EU regulated exchanges (the primary exchanges in each country). Firms listed on secondary exchanges (e.g., the AIM market in London) also known as exchange-regulated markets were not subject to mandatory IFRS adoption in Second, different countries had different regulatory policies with respect to IFRS prior to For example, while the UK and Sweden forbade the use of IFRS for domestic listed firms, Germany allowed voluntary IFRS adoption. Third, we can only observe a 8 For example, if an IFRS-reporting IPO's listed industry peers are all located in countries with poor enforcement environments, then on average these peer firms will not have incentives to faithfully and consistently apply IFRS. So it is unlikely that they will provide genuine comparability for the IPO firm. In the absence of strong enforcement environments, industry peers are unlikely to have incentives to credibly implement IFRS (Daske, Hail, Leuz and Verdi 2008) and may merely adopt IFRS "as a label" (Daske, Hail, Leuz, and Verdi 2013). 9 While these exchange-regulated markets did not have to adopt IFRS in 2005, they could choose to do so later. See Vismara, Paleari and Ritter (2012) for a discussion of Europe s exchange-regulated secondary markets. 9

11 firm going public once in its life, so we can t use a firm as its own control. Lastly, since we are using a global sample of international IPOs, we need to be cognizant of the fact that important country-level institutional differences (e.g., investor protection) may affect the IPO process, and hence the degree of IPO underpricing, differently in different countries. Study Design: Identifying the Effect of Mandatory IFRS Adoption on IPO Underpricing A number of countries mandated IFRS adoption in Our mandatory adoption test sample consists of IPOs from these countries over the period 2003 through 2007, and consists of a matched-pair sample with: (1) a treatment group of IPOs that went public in 2006 and 2007 using IFRS, and (2) a matched benchmark group of IPOs that went public using domestic GAAP in 2003 and We exclude 2005, the transition year, from our sample. 10 We match the treatment and benchmark IPO firms one-to-one by country, industry, and closest proceeds, with replacement. For our first set of analyses we estimate two change models using this (matched-pairs) test sample. A change model is estimated using the differences between each matched pair: the treatment (IFRS-reporting) IPO vs. the benchmark (domestic GAAP-reporting) IPO. A change model can be specified as a fixed effects model with a dummy variable for each matched pair (so the model is estimated using only the variation within pairs). Thus, we test H1 using the following model: IniRet = X β + β 1 IFRS + β 2 log(1+comp) + β 3 IFRS log(1+comp) + Controls + υ (1a) where IniRet is our measure of IPO underpricing and is calculated as the first trading-day return; X is a matrix of 493 dummy variables, one for each pair of matched observations (see sample below). Within each matched pair of IPOs, another dummy variable IFRS is coded one (zero) for 10 Nevertheless, our results are robust to including observations from 2005 in our sample. 10

12 the IFRS- (domestic GAAP-) reporting IPO in the pair. log(1+comp) is the natural log of one plus Comp, the number of comparable firms. For a domestic GAAP-reporting IPO, Comp is the number of domestic GAAP-reporting industry peers; for an IFRS-reporting IPO, Comp is the number of worldwide industry peers reporting in IFRS. We use Fama and French's (1997) industry classification to identify a firm s industry peers. The control variables are a set of controls for IPO characteristics (e.g., proceeds size, cross listing, etc.) which are explained in Section IV below (see also the endnotes of Table 3). Using a matched pair sample (matched by country and industry) with fixed effects for each pair of matched observations (X above) is equivalent to including both country and industry fixed effects. The model therefore controls for variation in the intercept arising from variation in country-level institutional characteristics such as investor protection (Djankov et al. 2008) and enforcement quality (Kaufmann, Kraay, and Mastruzzi 2007) in IPO firms home countries. 11 This is important since variation in country-level institutional characteristics may: (1) have a direct impact on the level of IPO underpricing (Boulton, Smart, and Zutter 2010), and (2) affect IPO firms reporting incentives and, hence, reporting quality (Leuz et al. 2003). However, while matching in this way provides a control for country-level institutional variation and industry effects, it results in a smaller sample size, 12 potentially affecting the generalizability of our results. Interpreting the coefficient on the IFRS variable, β 1, is problematic, however. Given the matched pair design with fixed effects for matched pairs, it is clear that β 1 picks-up the average 11 This also controls for industry-wide risk factors that affect investor uncertainty, which may determine IPO underpricing (Ritter 1984). 12 IFRS-reporting IPO firms that are not matched with domestic GAAP-reporting IPO firms (by country and by industry) are eliminated from the sample. This matching substantially reduces the sample size. However, the matching balances the sample with respect to the country and industry composition of the IFRS- and domestic GAAP-reporting IPO firms in the sample. 11

13 effect of variation in reporting quality between the matched domestic GAAP and IFRS-reporting firms in each pair. Nevertheless, it is critical to realize that the paired firms are from different time periods: the domestic GAAP (IFRS) IPOs are from ( ), so the "IFRS" dummy variable is also a time period variable. It is well known that IPO markets go through "hot" and "cold" period cycles (e.g., Helwege and Liang 2004; Ljungqvist, Nanda and Singh 2006; Yung, Colak and Wang 2008). Such time period shocks will be reflected in β 1 ; the effects of any macro-economic shocks which will, of course, vary by time will also be reflected in β 1. These time period effects render any interpretation of β 1 as a measure of the effect of differences in reporting quality arising from mandatory IFRS adoption problematic. Equation (1a) is used to test H1. Because IFRS typically allow for a far greater number of potentially comparable firms than domestic GAAP, H1 predicts that IPO underpricing decreases with the increase in overall comparability under IFRS. For any set of standards, since the marginal effect is likely to decline as the total number of comparable firms increases, we use a logarithmic transformation of Comp. The coefficients β 2 and β 3 approximate the incremental effect on IPO underpricing of additional comparable firms. More specifically, β 2 approximates how a 1% increase in the number of comparable firms affects the degree of IPO underpricing for a domestic GAAP-reporting IPO; β 3 approximates how the effect of a 1% increase in the number of comparable firms on IPO underpricing changes following mandatory IFRS adoption. H1 therefore predicts that β 3 < 0. Note, we are interested in estimating the effect on IPO underpricing of overall comparability in either domestic GAAP or IFRS. Thus, we interpret β 2. log(1+comp) and β 2. log(1+comp) + β 3. IFRS log(1+comp) as estimates of the impact on IPO underpricing of the level of overall comparability in domestic GAAP and IFRS. 12

14 To test H2 we modify Equation (1a) to accommodate variation in the quality of the enforcement environments of the countries where an IPO firm s listed industry peers are domiciled. To measure the quality of countries enforcement environments, we follow the prior literature (e.g., Daske et al. 2008) and use the 2005 values of the Rule of Law index developed by Kaufmann et al. (2007). If a country s Rule of Law index is greater (less) than the median value across all countries (0.74), then this country is classified as a strong (weak) enforcement jurisdiction. We find consistent results using Transparency International s Corruption Perception Index. 13 To differentiate between the level of enforcement of the countries where an IPO firm s potentially comparable firms are domiciled, we partition Comp into Comp Weak and Comp Strong and test H2 using the following model: IniRet = X β + β 1 IFRS + β 4 log(1+comp Weak ) + β 5 log(1+comp Strong /(1+Comp Weak )) + β 6 IFRS log(1+comp Weak ) + β 7 IFRS log(1+comp Strong /(1+Comp Weak )) + Controls + υ. (1b) As in Equation (1a), X is a matrix of 493 dummy variables, one for each pair of matched IPOs. In partitioning Comp into Comp Strong and Comp Weak, we normalize (1+Comp Strong ) by (1+Comp Weak ) to ensure that Equation (1b) is nested within Equation (1a). 14 This allows for the comparability of results across models. For a domestic GAAP-reporting firm, all of its comparable firms are domiciled in the same country. Thus, for a domestic GAAP-reporting firm 13 Several other indices have been proposed that measure different aspects of countries' legal systems such as the level of creditor rights or investor protection. In the context of our study, we use the Rule of Law index from Kaufmann et al. (2007) as a proxy for the overall institutional quality of a country that ensures compliance with laws, rules, and, in particular, accounting standards. Since we are interested in the overall quality of the institutional environment of a country as it pertains to the enforcement of all rules and laws, we believe the measures by Kaufmann et al. (or the Corruption Perception Index of Transparency International) are the best proxy for the overall quality of the institutions rather than a more limited index that focuses on just the enforcement of creditor right or the level of investor protection. 14 This follows from: log(1+comp)=log(1+comp Strong +Comp Weak )=log(1+comp Weak ) (1+Comp Strong / (1+Comp Weak )) = log (1+Comp Weak ) + log(1+comp Stong / (1+Comp Weak )). 13

15 domiciled in a weak (strong) country, β 4 (β 5 ) approximates how a 1% increase in the number of comparable firms (domiciled in the same country as the IPO firm) affects the degree of IPO underpricing. For an IFRS-reporting IPO, β 6 (β 7 ) approximates how a 1% increase in the number of comparable firms domiciled in weak (strong) enforcement countries affects the degree of IPO underpricing, incremental to β 4 (β 5 ). For an IFRS-reporting IPO, comparability is potentially provided by all IFRS-reporting listed industry peers, which can be located in either weak or strong enforcement jurisdictions. Recall that H2 predicts that only those listed peers domiciled in strong enforcement jurisdictions will provide genuine comparability for IFRS-reporting IPOs, resulting in an increase in overall comparability under IFRS compared to that under domestic GAAP; therefore H2 predicts that β 7 < 0. Difference-in-Differences Our mandatory adoption test sample consists of 493 matched pairs of IPOs: each pair consists of a treatment IPO firm that reports in IFRS (from 2006 to 2007) and a matched benchmark IPO firm that reports in domestic GAAP (from 2003 to 2004). Using this sample, we estimate Equations (1a) and (1b), both of which are change models. As already discussed, any time-period specific shocks that affect the IPO market (e.g., IPO hot/cold period cycles, or macroeconomic shocks) and IPO underpricing will be correlated with the treatment effect (i.e., the IFRS dummy variable) in our estimates of Equations (1a) and (1b). One way to try to control for such time-period shocks is to use a control sample and a Difference-in-Differences (DID) specification. However, since mandatory IFRS adoption applies to all firms in a given country, and firms only go IPO once in their life, we cannot use a control sample of IPO firms from the same countries as the test firms. Therefore, by necessity, we use a control sample of non-ifrs-using IPOs from countries that did not mandate IFRS adoption over 14

16 our sample period. Similar to the test sample, the control sample also consists of matched pairs of IPO firms (matched by industry, country, and closest proceeds size, with replacement), one from the period 2003 to 2004, and the other from the period 2006 to Only IPO firms in the same industries as our mandatory adoption test sample firms are included in the control sample. For our DID estimates, we use a dummy variable to distinguish between the test and control samples; this dummy variable is interacted with each variable shown in Equation (1a) and (1b). Note, to control for the effect of time period shocks, the DID specification assumes that the test and control samples are similarly exposed to any (time period) shocks that are uncorrelated with the treatment effect. 15 However, in our setting the test and control samples are, by necessity, drawn from different countries. If IPO market hot and cold periods are not perfectly synchronized across countries, or if our test and control firms are subject to different macroeconomic shocks, then there may be residual time period effects. 16 The average effect of any such time period shocks that are not perfectly identical across the test and control samples will be reflected in the difference in β 1 across the two samples. Therefore, as in our change models, it is also problematic to interpret the difference in β 1 across the test and control samples in our DID model estimates as a measure of the effect of changes in reporting quality arising from mandatory IFRS adoption. Sample Selection 15 This is the so called parallel assumption of the DID specification. It is assumed that, in the absence of the treatment effect, the test and control samples would have moved in parallel because they are equally exposed to other uncorrelated shocks (see Abadie 2005). 16 It seems likely that hot and cold IPO markets are not perfectly synchronized across all countries. International capital markets are, to some degree, segmented between countries as evidenced by the widespread prevalence of home bias (see Karolyi and Stulz, 2003; Ahearne, Griever, and Warnock 2004). Such capital market segmentation between countries means that it is unlikely that hot and cold IPO markets are perfectly synchronized across all countries. 15

17 Panel A of Table 1 describes the selection of the mandatory adoption (test and control) samples; Panel B describes the selection of the voluntary adoption sample (discussed in Section V below). For the mandatory adoption test sample, we first use the Security Data Corporation (SDC) database to select a global sample of IPOs over the period 2006 to To avoid the effects of the global financial crisis, we end our sample period in We retain only observations for ordinary common shares. If there are multiple observations for the same firm on the same day, then only one observation is used. This yields an initial global sample of 3,380 IPOs. We delete observations that are missing critical data, such as first day returns or accounting standards. Accounting standards data are from Datastream or Compustat Global. 17 We also delete IPOs in countries that did not mandate the adoption of IFRS in 2005 and IPOs on secondary exchanges (e.g., AIM in UK), which still allowed the use of domestic GAAP for some time after This gives a sample of 749 test IPOs in from countries that mandated the adoption of IFRS in We then match these IFRS-reporting IPOs, one-to-one (by industry, country, and proceeds size with replacement), with benchmark IPOs that went public using domestic GAAP in 2003 and While we are able to match 493 IFRS and domestic GAAP-reporting IPOs in this manner, we are unable to match the remaining 256 IFRSreporting IPOs. Our final mandatory adoption test sample thus consists of 493 matched-pairs of IPOs (in total, 986 IPOs). Because of differences in sample selection procedures, our mandatory adoption test sample is quite different from that of Hong et al. (2014); for example, they do not match firms across the pre- and post-ifrs adoption periods. 17 We use data for accounting standards from Compustat only when this data item is unavailable from Datastream. In classifying Datastream and Compustat Global data on firm s accounting standards between domestic GAAP and IFRS we follow Table A1 of Daske et al. (2013). 18 Since firms on the AIM market in the UK were not required to adopt IFRS until 2007, we drop 2006 listings on AIM. Of the 284 IPOs listed on AIM in 2005 and 2006, 163 reported using domestic GAAP. The other secondary European exchanges that are identified in the SDC database are: Alternext, FirstNorth, and Marche Libre. 16

18 < Insert Table 1 about Here > For our DID tests, we construct a control sample using the same procedures and sample period, but drawing IPOs from countries that did not mandate the adoption of IFRS over our sample period. Our control sample consists of 628 matched pairs of IPOs (1,256 IPO firms) from the periods 2003 to 2004 and 2006 to Panel A of Table 2 shows the country distribution of our mandatory adoption test and control samples; Panel B shows the country distribution of the voluntary adoption sample (again, discussed in Section V below). Panel A shows that Australian firms constitutes about 68-percent (672 of the 986 firms) of the mandatory adoption test sample. As a robustness check, we exclude Australian firms from our sample and find that our inferences are unchanged (see Section VI below). Panel A also shows that the largest number of control firms is from the US. < Insert Table 2 about Here > IV. RESULTS Descriptive Statistics Table 3 provides descriptive statistics for the mandatory adoption test sample of 493 matched-pairs of IPOs: Panel A (B) shows data for the 493 IFRS-reporting (domestic GAAP) IPOs. We winsorize the data at the 2% and 98% levels. Consistent with prior studies (e.g. Loughran et al. 1994; Boulton et al. 2011), mean IniRet, the degree of IPO underpricing, is 23.9% (20.1%) for IFRS- (domestic GAAP-) reporting IPOs. The number of potentially comparable firms increased under IFRS: mean Comp, the raw number of listed industry peers using the same accounting standards is (32.3) for IFRS- (domestic GAAP-) reporting IPOs. For IFRS, 17

19 Comp equals the total number of listed industry peers using IFRS around the world. 19 However, for IFRS-reporting IPOs, not all comparable firms are domiciled in strong-enforcement jurisdictions: mean Comp Strong (Comp Weak ) is (36.6). We use Fama and French s (1997) industry classification. < Insert Table 3 about Here > Table 3 also provides data for our control variables. While our matched-pairs design controls for country and industry effects, there could still be differences at the firm level between the two matched IPOs. Hence, we include controls for important IPO firm characteristics. Proceeds is the IPO proceeds size expressed in millions of U.S. dollars (Habib and Ljungqvist, 1998), adjusted by the U.S. Consumer Price Index. The degree of Partial Adjustment indicates the market s interest in an IPO. Hanley (1993) shows that partial adjustment, the deviation of the final offer price from the initial offer price range, is positively associated with the degree of underpricing; Cross Listing equals one when an IPO firm lists in a foreign country, and zero otherwise. Following Cliff and Denis (2004), we include Market Return, Market Volatility, and #IPO to control for market sentiment. Market Return is the mean daily market return, Market Volatility is the standard deviation of market returns (both measured within three weeks prior to the offer date), and #IPO is the number of other IPOs within 12 months prior to the offer date. Secondary Market equals one if an IPO firm is listing on a secondary market and zero otherwise. The descriptive statistics reported in Table 3 are consistent with prior IPO studies (e.g. Loughran et al. 1994). Regression Analysis 19 When counting the number of potentially comparable firms for an IPO firm, we include only those listed industry peers for which sales data (from Datastream) are available for the fiscal period ending in the three months prior to the IPO offer date. 18

20 Table 4 reports the results of estimating Equations (1a) and (1b). Note, because our mandatory adoption test sample consists of matched pairs of IPOs, matched by country and industry, and we include fixed effects for each pair of matched IPOs, we effectively control for country- and industry-level fixed effects. In Equation (1a) the variable of interest is IFRS log(1+comp). H1 predicts that IFRS will generate enhanced overall comparability compared to domestic GAAP, resulting in reduced IPO underpricing, i.e., β 3 < 0. As can be seen in Table 4, β 2 is significantly negative (p<0.05, two-tailed) showing that, in general, a greater number of comparable firms is associated with a significant decrease in IPO underpricing. However, β 3 is not statistically different from zero, so we reject H1 and find no evidence of a difference in overall comparability between domestic GAAP and IFRS. As can be seen in Panel B, the coefficient estimate of total comparability for IFRS firms, (β 2 + β 3 ), is negative, but is not significant at conventional levels. However, this seems to be due to greater noise in the Comp variable for IFRS firms (see discussion below). < Insert Table 4 about Here > H2 predicts that only IFRS-reporting peers in high enforcement jurisdictions provide genuine comparability for IFRS-reporting IPOs. Equation (1b) tests H2. The main variable of interest is IFRS log(1+comp Strong /(1+Comp Weak )). Comp Strong is the raw number of comparable firms domiciled in strong enforcement countries, so H2 predicts that β 7 < 0. As can be seen in Table 4, consistent with this prediction, β 7 is significantly negative (p<0.01, one-tailed). Panel B of Table 4 shows the results of additional statistical tests that are relevant to H2. For IFRS-reporting IPOs, the impact on IPO underpricing of listed industry peers domiciled in weak enforcement countries is measured by the coefficients on log(1+comp Weak ) and IFRS log(1+comp Weak ), i.e., (β 4 + β 6 ). As can be seen in Panel B, (β 4 + β 6 ) is positive but not 19

21 statistically significant, indicating that IFRS-reporting IPOs do not benefit from having listed industry peers located in weak enforcement countries. On the other hand, for IFRS-reporting IPOs, the impact on IPO underpricing of listed industry peers domiciled in strong enforcement countries is measured by the coefficients on log(1+comp Strong /(1+Comp Weak )) and IFRS log(1+comp Strong /(1+Comp Weak )), i.e., (β 5 + β 7 ). Panel B shows that (β 5 + β 7 ) is significantly less than zero (p<0.01, one-tailed). In summary, consistent with H2, the results show that IFRSreporting IPOs benefit only from having listed industry peers located in strong enforcement countries: only these peers seem to provide genuine comparability. 20 For a domestic GAAP-reporting IPO, the industry peers are domiciled in the same country. If a domestic GAAP-reporting IPO is from a weak enforcement country, then β 4 approximates the impact (on IPO underpricing) of the overall level of comparability derived from listed industry peers in the same country. For IFRS-reporting IPOs (β 4 + β 6 ) approximates the impact (on IPO underpricing) of the overall level of comparability derived from listed industry peers in all countries. While (β 4 + β 6 ) is not statistically significant, we find that β 6 is significantly positive (p<0.05, two-tailed). The significantly positive β 6 indicates that listed peers based in weak enforcement jurisdictions provide less comparability under IFRS than under domestic GAAP. 21 The lack of statistical significance we found for H1 seems to be due to greater noise in the Comp variable for IFRS firms: when we re-estimate Equation (1a) replacing Comp with 20 Our results speak to the average level of overall comparability. Cascino and Gassen (2015) study compliance incentives at the country, region, and firm levels. They find evidence that firm-level reporting incentives play a role in determining the level of comparability. Thus, within IFRS, it is possible that some IFRS-reporting firms in weak enforcement jurisdictions but with high firm-level reporting incentives will provide meaningful comparability. 21 Taken at face value this result suggests that, when coupled with a weak enforcement environment, the principlesbased IFRS actually provide less comparability than domestic GAAP, i.e., in weak enforcement jurisdictions per unit comparability from domestic listed peers actually decreases following mandatory IFRS adoption. However, in our mandatory adoption test sample only 10 IPO firms (from Greece and the Philippines) are domiciled in weak enforcement countries (see Panel A of Table 2). As a result, we caution readers as to the generalizability of this result. 20

22 Comp Strong we find that (β 2 + β 3 ) is now significantly negative (p=0.02, one-tailed). Since Comp Strong is the total number of listed peers based in strong enforcement jurisdictions, it provides for a more homogenous set of comparable firms than Comp. Also, this suggests that IFRS do provide enhanced overall comparability compared to domestic GAAP, but only if one restricts the set of comparable firms to those listed industry peers domiciled in high enforcement jurisdictions. < Insert Table 5 about Here > Table 5 shows the results of our DID tests. These tests parallel our estimates of Equations (1a) and (1b) reported in Table 4, which use just the mandatory adoption test sample. To estimate our DID models, we pool the mandatory adoption test sample of 493 matched pairs of IPOs from mandatory IFRS adopting countries with a control sample of 986 matched pairs of IPOs from non-ifrs adopting countries. All the IPO firms in our control sample report in domestic GAAP. Panel A of Table 5 reports the results for our DID estimates of Equation (1a) and show that, contrary to the prediction of H1, β 3 is not significantly different across our test and control samples. Panel B of Table 5 reports the results for our DID estimate of Equation (1b). Consistent with our prediction for H2, and our earlier results reported in Table 4, we find that β 7 is significantly more negative for the test sample than for the control sample (p<0.05, one-tailed). As discussed earlier, interpreting the difference in the β 1 coefficient between the test and control samples is problematic. This is because the difference in β 1 between the test and control samples reflects not only any difference from any changes in reporting quality arising from mandatory IFRS adoption, but also any differences in macroeconomic shocks between our test and control sample firms (which, by necessity, are drawn from different countries). 21

23 The results for our analysis of the test sample alone (Table 4) and our DID analysis of the pooled test and control samples (Table 5) show a consistent result: IFRS-reporting IPOs do (do not) benefit from having more listed industry peers located in strong (weak) enforcement jurisdictions. These results suggest that enforcement quality is critical in generating enhanced comparability following mandatory IFRS adoption. To evaluate the economic significance of our results, we substitute the mean values of the variables shown in Table 3 into the estimated equations shown in Table 4. We find that, relative to a firm with no comparable firm, the reduction in IPO underpricing from the average number of listed peers is 3.3% (7.9%) for domestic GAAP- (IFRS-) reporting IPOs, indicating that, on average, IFRS provides incremental comparability that reduces underpricing by 4.6%. For an average IPO in our sample, this translated into a saving of over $2 million. V. VOLUNTARY IFRS ADOPTER SAMPLE Our first set of analysis suggests that enforcement quality is critical in determining the extent to which IPO firms benefited from enhanced comparability following mandatory IFRS adoption. A number of prior studies suggest that the effects of mandatory IFRS adoption depend on firms reporting incentives that are shaped by the quality of countries enforcement environments (Daske et al. 2008; Li 2010; Byard et al. 2011). Prior studies also suggest that firms reporting incentives are critical in mediating the effects of IFRS adoption, in particular, in the context of voluntary IFRS adoption (Leuz and Verrecchia 2000; Ashbaugh and Pincus 2001). A number of countries (e.g., Germany) allowed voluntary IFRS adoption prior to 2005 Using a sample of IPOs from these countries, in this section we test if IPO firms that voluntarily adopted IFRS also benefited from enhanced overall comparability (relative to IPO firms that chose to report in domestic GAAP). The overall level of comparability from listed industry peers 22

24 could be different for IFRS IPOs vs. domestic GAAP IPOs because: (1) the total number of comparable firms is different, and (2) the level of comparability per listed industry peer is different. In contrast to mandatory IFRS adoption, it is unlikely that IFRS provides for a far greater number of listed industry peers for voluntary IFRS adopters than would domestic GAAP. However, in the pre-2005 voluntary adoption setting, it is important to recognize that any listed industry peers that report in IFRS are themselves voluntary IFRS adopters. As such, these peers are likely to have strong firm-level incentives for transparency (Leuz and Verrecchia 2000), so they can be expected to rigorously implement the reporting requirements of IFRS. We expect these (voluntary adopter) peer firms to provide a higher level of comparability per listed industry peer. In summary, in the pre-2005 period, we expect that IPO firms that voluntarily report in IFRS will benefit from enhanced comparability relative to IPO firms that chose to report in domestic GAAP. We expect this because, in IFRS, the industry peers are themselves voluntary IFRS adopters with strong firm-level reporting incentives. It follows that there is no need to split the listed industry peers into Comp Stong and Comp Weak ; essentially we assume that, in the voluntary IFRS adoption setting, for IFRS-reporting IPOs, all of listed industry peers are Comp Strong type firms. 22 For our voluntary IFRS adoption sample, we select a sample of IPO firms that went public in 2003 and 2004 in countries where (1) IFRS were mandated in 2005, and (2) there is at 22 Rigorous implementation of IFRS can come from a strong country-level enforcement environment, or strong firmlevel reporting incentives (see Byard et al. 2011; Cascino and Gassen 2015). In a strong enforcement environment, all firms are expected to rigorously implement the ascribed standards; but, in a weak enforcement environment there is more scope for firm-level reporting incentives to affect variation in reporting quality and, hence, comparability. In IFRS, because all of the comparable firms have also voluntarily adopted IFRS, these comparable firms are all likely to have strong firm-level reporting incentives. There is therefore less scope for the quality of country-level enforcement environments to affect the quality of the implementation of IFRS among these comparable firms. So for our voluntary adoption sample, we do not separate comparable firms into those domiciled in strong enforcement countries (Comp Strong ) and those domiciled in weak enforcement countries (Comp Weak ). 23