J. Account. Public Policy

Transcription

1 J. Account. Public Policy 30 (2011) Contents lists available at ScienceDirect J. Account. Public Policy journal homepage: Do earnings reported under IFRS tell us more about future earnings and cash flows? T.J. Atwood a, Michael S. Drake b, James N. Myers c, Linda A. Myers c, a Accounting Department, College of Business, Florida State University, United States b Department of Accounting and MIS, Fisher College of Business, The Ohio State University, United States c Department of Accounting, Sam M. Walton College of Business, University of Arkansas, United States article info abstract Article history: We contribute to the debate about the relative benefits and costs of International Financial Reporting Standards (IFRS) adoption by examining whether earnings persistence and the association between current accounting earnings and future cash flows differ for firms reporting under IFRS versus firms reporting under United States Generally Accepted Accounting Principles (U.S. GAAP) and firms reporting under non-u.s. domestic accounting standards (DAS). Using samples comprised of 58,832 firm-year observations drawn from 33 countries from 2002 through 2008, we find that positive earnings reported under IFRS are no more or less persistent than earnings reported under U.S. GAAP but losses reported under IFRS are less persistent than losses reported under U.S. GAAP. Moreover, we find that earnings reported under IFRS are no more or less persistent and are no more or less associated with future cash flows than earnings reported under non-u.s. DAS. However, we find that earnings reported under U.S. GAAP are more closely associated with future cash flows than earnings reported under IFRS. This is important if a key role of reported earnings is to help investors form expectations about future cash flows. These results should be of interest to academics and standard-setters as they debate the merits of transitioning to IFRS, and to parties who use reported earnings to form expectations about future earnings and cash flows. Ó 2010 Elsevier Inc. All rights reserved. Corresponding author. Tel.: ; fax: address: [email protected] (L.A. Myers) /$ - see front matter Ó 2010 Elsevier Inc. All rights reserved. doi: /j.jaccpubpol

2 104 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Introduction In this paper, we examine whether earnings persistence and the association between current accounting earnings and future cash flows differ for firms reporting under International Financial Reporting Standards (IFRS) versus domestic accounting standards (DAS). We consider two forms of DAS United States Generally Accepted Accounting Principles (U.S. GAAP) and non-u.s. DAS. Our study is motivated by the initiative to establish a uniform set of international accounting standards, which has gained considerable momentum in recent years. Firms in 117 countries, including all publicly listed European Union (E.U.) firms, currently prepare their financial statements in conformity with IFRS. Moreover, Canada, India, and Korea are set to adopt IFRS by 2011, the G-20 target date (Heffes, 2009) and Japan began allowing listed companies to file under IFRS for fiscal years ending after March 2010 (FSA, 2009). In the U.S., the move towards adopting a global set of accounting standards gained momentum in 2007 and 2008 when the Securities Exchange Commission (SEC) announced its decision to no longer require foreign corporations listed in the U.S. to reconcile between IFRS and U.S. GAAP (SEC Release ) and proposed a roadmap for the mandatory adoption of IFRS by U.S. firms (SEC Releases ; ). However, the SEC subsequently delayed making a final decision on the proposed roadmap. Although the momentum towards the adoption of IFRS in the U.S. has slowed, the Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) have recommitted to converging U.S. GAAP and IFRS by 2011 (Defelice, 2009). The movement toward global acceptance of IFRS has generated considerable attention and debate. Proponents maintain that requiring U.S. firms to report under IFRS will enhance the comparability of financial reports across countries and will bring greater efficiency to firms reporting across multiple jurisdictions (Covrig et al., 2007; Kim et al., 2007; Turley, 2007; Barth et al., 2008). Furthermore, some proponents also maintain that since IFRS relies on a principles-based system, it is more likely to deter fraud (Carmona and Trombetta, 2008). Opponents suggest that requiring U.S. firms to adopt IFRS will be costly and that the benefits of comparability may not be realized due to disparities in the application of IFRS across countries (Ciesielski, 2007; FFSA and AFG, 2007; Herz, 2007; Soderstrom and Sun, 2007; Turner, 2007; Holthausen, 2009; Sunder, 2009; Kvaal and Nobes, 2010; Hail et al., 2010, forthcoming). In addition, in 2009, SEC Chairman Mary Schapiro expressed reservations about the IASB and about the quality of IFRS (Forgeas, 2008; Cohn, 2009; Leone, 2009). More recently, Howell (2010) reports that plans for the U.S. to move to global standards are vulnerable to major delays and that chief executive officers have expressed concerns about incurring costs to convert to IFRS when the benefits of conversion are not clear. Finally, existing research has led some academics to suggest that competition between U.S. GAAP and IFRS is preferable to convergence of the standards or to the mandatory adoption of IFRS by U.S. firms (Jamal et al., 2008, 2009; Sunder, 2009; Bradshaw et al., 2010; Kothari et al., 2010). We contribute to this debate by examining whether earnings persistence and the association between current accounting earnings and future cash flows differ between IFRS versus DAS (both U.S. GAAP and non-u.s. DAS). Our focus on earnings persistence is motivated by prior research (discussed below) suggesting that earnings management and analyst forecast errors differ for firms reporting under IFRS versus DAS (Ashbaugh and Pincus, 2001; Barth et al., 2008; Ernstberger et al., 2008; Jeanjean and Stolowy, 2008; Chen et al., 2010) and that earnings are more value-relevant under IFRS than under DAS in some countries (Bartov et al., 2005; Ndubizu and Sanchez, 2006; Barth et al., 2008). These differences may be due to differences in earnings persistence for firms reporting under IFRS. Thus, we test whether earnings reported under IFRS are more persistent than earnings reported under DAS. Our focus on the association between current reported earnings and future cash flows is motivated by the conceptual frameworks issued by both the FASB and the IASB, which suggest that financial reporting should provide information helpful to users in predicting future cash flows (e.g., FASB, 1978, Con 1; IASB, paragraph 10). Thus, the association between current earnings and future cash flows is one measure of earnings quality that is consistent with the objectives set forth in both the FASB and IASB conceptual frameworks.

3 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Using a sample of 58,832 firm-year observations drawn from 33 countries from 2002 through 2007, we conduct empirical tests that examine earnings persistence and the association between earnings and future cash flows for firms reporting under IFRS relative to two control samples. The first control sample (the U.S. GAAP sample) consists of all firm-year observations reporting under U.S. GAAP. The second control sample (the non-u.s. DAS sample) consists of all firm-year observations reporting under non-u.s. DAS. When we compare earnings reported under IFRS versus U.S. GAAP, we find no difference in the persistence of positive earnings across firms reporting under the two standards, but find that losses reported under IFRS are less persistent than losses reported under U.S. GAAP. Moreover, future cash flows have a lower association with current earnings reported under IFRS than under U.S. GAAP. However, we find no evidence of a systematic difference between IFRS and non-u.s. DAS in terms of earnings persistence and/or the association between current earnings and future cash flows. These findings are robust to controls for cross-country factors including legal structure, investor rights, ownership equity, the importance of equity markets, the strength of legal enforcement, book-tax conformity, country fixed effects, and year fixed effects. The results of our study are important for a number of reasons. First, they provide information to standard-setters and legislators around the world as they weigh the costs and benefits of transitioning to IFRS. Our results suggest that earnings reported under IFRS are no more closely associated with future earnings but are less closely associated with future cash flows than are earnings reported under U.S. GAAP. Hail et al. (2010, forthcoming) suggest that the direct effect of IFRS adoption on the quality of U.S. financial reporting is likely to be small because U.S. GAAP are of high quality. However, our results suggest that along at least one dimension the ability to predict future cash flows the quality of U.S. financial reporting may actually decline with the adoption of IFRS. We suggest that this difference in the ability to predict future cash flows supports calls for allowing competition between IFRS and U.S. GAAP rather than requiring all U.S. firms to adopt IFRS or converging IFRS and U.S. GAAP (Jamal et al., 2008, 2009; Sunder, 2009; Bradshaw et al., 2010; Kothari et al., 2010). Second, our results add to the literature on analyst forecast accuracy. We find that earnings persistence is not significantly different for firms reporting under IFRS versus non-u.s. DAS. Therefore, previously documented differences in analyst forecast accuracy for IFRS versus non- U.S. DAS firms (Ashbaugh and Pincus, 2001; Ernstberger et al., 2008) do not appear to be the result of differences in the underlying persistence of those earnings. Instead, these differences may be the results of increased disclosures that occur along with the adoption of IFRS (Daske and Gebhardt, 2006). Finally, our results should be of interest to investors, analysts, creditors, and other parties who use reported earnings information to form expectations about future earnings and cash flows. For example, prior research suggests that lenders charge lower interest rates, extend larger loans, and impose fewer restrictive covenants on firms that switch from reporting under non-u.s. DAS to reporting under IFRS (Kim et al., 2007); however, our results show that earnings reported under IFRS are no more closely related to future earnings and future cash flows than are those reported under non-u.s. DAS. Furthermore, financial statement users should take into account the lower association between current earnings and future cash flows under IFRS as compared to U.S. GAAP in their prediction models. The remainder of this paper is organized as follows. Section 2 discusses prior literature and develops our hypotheses. Section 3 describes our empirical models, sample selection, and descriptive statistics. Section 4 discusses the results of our empirical tests, and Section 5 concludes. 2. Prior literature and hypotheses development Two committees of the American Accounting Association the Financial Accounting Standards Committee (see Jamal et al., 2008, 2009) and the Financial Reporting Policy Committee of the Financial Accounting and Reporting Section (see Hopkins et al., 2008; Bradshaw et al., 2010) have reviewed

4 106 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) the extensive academic literature on IFRS. 1 Hopkins et al. (2008) suggest that the decision to eliminate the previously-required Form 20-F reconciliation between IFRS and U.S. GAAP for foreign registrants is premature because material differences exist between the two standards, because information in the reconciliation is currently impounded into stock prices, and because differential implementation and enforcement of standards across countries may reduce the comparability of financial reports across jurisdictions. 2 Jamal et al. (2008, 2009) assert that there is no conclusive evidence that financial reports prepared using U.S. GAAP are better than reports prepared using IFRS, and recommend that the SEC allow U.S. companies to choose between IFRS and U.S. GAAP rather than mandate the adoption of IFRS. Similarly, Kothari et al. (2010) provide a survey and economic analysis of the properties of GAAP and conclude that, rather than converging U.S. GAAP with IFRS, competition between the FASB and the IASB would allow GAAP to better respond to market forces. Soderstrom and Sun (2007) review the literature on accounting quality and IFRS adoption. They point out that the majority of studies focus on value-relevance, information content, timeliness, and other stock price-related measures, and suggest that these studies do not provide a complete view of usefulness because they focus solely on how information is reflected in stock market investors expectations. Finally, Bradshaw et al. (2010) conclude that it is not clear that IFRS reflect a set of accounting standards that are of equivalent or greater quality relative that U.S. GAAP. They suggest that although both IFRS and U.S. GAAP represent high-quality accounting standards, material reconciling items continue to exist. We extend the literature on the attributes of financial reports prepared under IFRS by examining whether current earnings are more or less persistent and/or more or less closely associated with future cash flows for firms reporting under IFRS versus those reporting under U.S. GAAP or non-u.s. DAS. In a related study, Gordon et al. (2008) examine a sample of foreign firms that filed Form 20-F Reconciliations from IFRS to U.S. GAAP over the period They find that accrual quality, earnings predictability, and cash predictability are not significantly different overall for the IFRS and U.S. GAAP earnings measures but do vary with financial reporting incentives. 3 In addition, although they present data suggesting that earnings persistence, cash persistence, and earnings smoothness are higher for the IFRS measure, they do not perform statistical tests of significance on these differences. 4 These results are suggestive but, as the authors state, their sample is composed solely of foreign firms choosing to list on U.S. stock exchanges, so the inferences may not generalize to a broader sample of firms. In contrast, we directly examine whether IFRS or U.S. GAAP provides earnings that are more persistent and more closely associated with future cash flows across a broad spectrum of firms in 33 countries. Our results should be of interest to many stakeholders (e.g., creditors, customers, suppliers, as well as investors). While some of these stakeholders are primarily interested in valuation, others are not. Holthausen and Watts (2001) conclude that contracting, litigation, political considerations, and tax considerations are all factors affecting the nature of GAAP and that inputs to equity valuation is not 1 This extensive academic literature includes a large number of papers addressing various aspects related to the use of IFRS for financial reporting. For example, Tarca (2004) and Francis et al. (2008) study factors leading firms to voluntarily adopt IFRS. Leuz and Verrecchia (2000), Dargenidou et al. (2006), Daske (2006), Kim and Shi (2007, 2008), Kim et al. (2007), Daske et al. (2008), Platikanova (2009), and Wu and Zhang (2009) study the economic benefits and costs of adopting IFRS. Kinnunen et al. (2000), Ashbaugh and Olsson (2002), Bartov et al. (2005), Lin and Chen (2005), LaPointe-Antunes et al. (2006), Ndubizu and Sanchez (2006), Barth et al. (2008), and Gjerde et al. (2008) study the value-relevance and/or information content of earnings under IFRS versus DAS. Harris and Muller (1999), Niskanen et al. (2000), Lin and Chen (2005), Christensen et al. (2009), Schadewitz and Vieru (2007), Chen and Sami (2008), and Henry et al. (2009) study the properties and value-relevance of reconciliations from IFRS to DAS, and Chen et al. (1999), Ashbaugh and Pincus (2001), Daske and Gebhardt (2006), Haverty (2006), LaPointe-Antunes et al. (2006), Soderstrom and Sun (2007), Van der Meulen et al. (2007), Barth et al. (2008), and Christensen et al. (2009) study the attributes of financial statements prepared using IFRS. 2 Consistent with this, Henry et al. (2009) find that reported net income was 59 (29) percent higher in 2004 (2005) under IFRS than under U.S. GAAP for 83 E.U. companies cross-listed in the U.S. and providing Form 20-F reconciliations. 3 Gordon et al. (2008) measure accrual quality as the standard deviation of the residual from a regression of accruals on future year, current year, and previous year cash flows from operations, earnings predictability as the standard deviation of the residual from a regression of current year net income on prior year net income, and cash predictability as the standard deviation of the residual from a regression of current year operating cash flows on prior year operating cash flows and accruals. 4 Gordon et al. (2008) measure earnings persistence as the estimated coefficient on previous year net income from a regression of current year net income on previous year net income, cash persistence as the estimated coefficient on previous year operating cash flows from a regression of current year operating cash flows on previous year operating cash flows and accruals, and earnings smoothness as the standard deviation of net income divided by the standard deviation of operating cash flows.

5 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) a dominant force. Moreover, Barth et al. (2001) focus on future cash flows rather than share prices because cash flow is a primitive valuation construct and share prices fail to accurately reflect the differential persistence of accruals and cash flows. For these reasons, we focus directly on earnings persistence and the association between current earnings and future cash flows rather than on the relation between current earnings and stock prices or changes in stock prices. Prior research suggests that there are significant differences between earnings reported under IFRS and DAS, including U.S. GAAP (Haverty, 2006; Ding et al., 2007; Hopkins et al., 2008; O Connell and Sullivan, 2008; Henry et al., 2009). Earnings reported under IFRS may differ from earnings reported under DAS in terms of persistence or in the usefulness of earnings for predicting future cash flows because of differences in reporting flexibility. Specifically, IFRS are commonly viewed as being more principles-based, while DAS, at least in some countries (and especially in the U.S.), are more rulesbased (Ijiri, 2005; Bennett et al., 2006; Reilly, 2007). 5 Carmona and Trombetta (2008) argue that rules-based standards impose uniform accounting treatment on firms operating in different circumstances and this uniformity imposes an informational cost because it reduces the information that can be extracted from observing the firm s accounting policy choices. However, Bradshaw and Miller (2008) argue that harmonizing standards does not necessarily result in a harmonization of accounting practices. Sunder (2009) further argues that standards become more rules-based over time in response to business and political pressures and that applying a single set of principles-based standards to all companies worldwide will not necessarily make financial reports more comparable or assist financial statement users in making better decisions. 6 If managers use the increased reporting flexibility under IFRS to convey private information, earnings reported under IFRS may be more persistent and/or more closely associated with future cash flows than earnings reported under U.S. GAAP or non-u.s. DAS. However, if managers use their discretion to report earnings optimistically or opportunistically, earnings reported under IFRS may be less persistent and/or less closely associated with future cash flows than earnings reported under U.S. GAAP or non-u.s. DAS. Results from prior studies of analyst forecast errors suggest that earnings persistence may differ for firms reporting under IFRS versus U.S. GAAP or non-u.s. DAS (although these studies do not explicitly examine earnings persistence). For example, Ashbaugh and Pincus (2001) find that analyst forecast errors are smaller after the adoption of IFR. However, Ernstberger et al. (2008) find that analyst forecasts are more accurate when firms report under U.S. GAAP versus IFRS, and find that analyst forecast accuracy improves for German firms switching from German GAAP to IFRS or to U.S. GAAP, but not from U.S. GAAP to IFRS, or from IFRS to U.S. GAAP. We suggest these differences may be due to differences in the persistence of earnings reported under IFRS, U.S. GAAP, and non-u.s. DAS. Prior studies of earnings properties also suggest that earnings persistence may differ for firms reporting under IFRS versus U.S. GAAP or non-u.s. DAS. Barth et al. (2008) find evidence of less earnings management, more timely loss recognition, and more value-relevance for firms reporting under IFRS versus matched samples of firms reporting under non-u.s. DAS. 7 In contrast, Chen et al. (2010) find more earnings smoothing and less timely recognition of large losses following the mandatory adoption of IFRS in 15 E.U. countries. Similarly, Jeanjean and Stolowy (2008) find that earnings management 5 Proponents of more principles-based standards argue that they allow managers more discretion to use their judgment to convey information (PwC 2007a, 2007b, 2007c). Consistent with this, a report written by the Big Four audit firms plus Grant Thornton and BDO Siedman suggests that virtually all stakeholders favor the idea of one set of principles-based standards and express concerns about countries, including the U.S., that rely on overly prescriptive rules-based standards (Leone, 2008; Taub, 2008). 6 For example, U.S. GAAP and IFRS often differ with respect to revenue recognition. The more principles-based nature of IFRS often allows technology companies to recognize revenues much earlier than under U.S. GAAP. However, a survey by BDO Siedman found that 60% of the technology company chief financial officers surveyed agree that revenue recognition rules for technology firms are better under U.S. GAAP than under IFRS (Johnson, 2008). 7 More specifically, Barth et al. (2008) find the variability of the change in net income (scaled by total assets) is significantly larger and the magnitude of the negative correlation between accruals and cash flows is significantly smaller for firms reporting under IFRS than for firms reporting under non-u.s. DAS. In addition, firms reporting under IFRS recognize large losses more frequently than do firms reporting under non-u.s. DAS. They find no significant difference in the ratio of the variability of net income changes to the variability of cash flow changes or in the frequency of small positive earnings numbers across the two groups.

6 108 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) did not decline following the mandatory adoption of IFRS in Australia, France, and the United Kingdom. Although these studies suggest that the properties of earnings differ across reporting standards, they do not test for differences in earnings persistence. In summary, both Ashbaugh and Pincus (2001) and Ernstberger et al. (2008) find differences in analyst forecast accuracy across firms reporting under IFRS, U.S. GAAP, and non-u.s. DAS, and Barth et al. (2008) and Chen et al. (2010) find differences in earnings properties, including smoothing, timeliness of loss recognition, and value-relevance, across reporting standards. These findings are consistent with differences in earnings persistence across standards; however, they provide conflicting predictions as to whether earnings persistence is expected to be higher or lower under IFRS versus U.S. GAAP or non- U.S. DAS. Thus, we examine whether, on average, earnings persistence is higher or lower under IFRS, as compared to U.S. GAAP or non-u.s. DAS, but we make no directional predictions. We also examine the associations between current earnings and future cash flows. The conceptual frameworks of the FASB and the IASB uniformly suggest that financial reports should provide information helpful to users in evaluating the firm s ability to generate future cash flows. For example, the U.S. FASB s conceptual framework states that financial reporting should provide information helpful to users in assessing the amounts, timing, and uncertainty of prospective net cash inflows (FASB, 1978, Con 1). The IASB s conceptual framework states that financial statements should provide information helpful to users in evaluating an enterprise s ability to generate cash and cash equivalents, and should provide information about the timing and certainty of those future cash flows (IASB, paragraph 10). Moreover, the IASB s framework states that the objective of financial reporting is to provide information about the financial position, performance, and changes in financial position of an enterprise (IASB, paragraphs 12 14). Managers may also use the increased reporting flexibility available under IFRS (as discussed above) to convey more information about future cash flows. Badertscher et al. (2010) find that for a sample of U.S. firms that meet or beat analyst forecasts and subsequently restated earnings, the initially reported (i.e., misstated) earnings were less positively associated with one-period-ahead cash flows than were the restated earnings. However, for firms that did not meet or beat analyst forecasts, initially reported earnings were more positively associated with one-period-ahead cash flows than were restated earnings. Thus, managers can use their discretion over reported earnings to provide more or less information about future cash flows. Because of this, greater reporting flexibility under IFRS may result, on average, in a higher or lower association between current earnings and future cash flows, so we examine whether current earnings reported under IFRS are, on average, more or less closely associated with future cash flows than are earnings reported under U.S. GAAP or non-u.s. DAS, but again make no directional predictions. 3. Empirical models, sample selection, and descriptive statistics 3.1. Empirical models We test for differences in earnings persistence and associations between current earnings and future cash flows across reporting regimes using the following models (country and firm subscripts are suppressed) 8 : EARN tþ1 ¼ IFRS ½a 0 þ a 1 EARN t þ a 2 LOSS t þ a 3 EARN LOSSŠþGAAP=DAS ½c 0 þ c 1 EARN þ c 2 LOSS t þ c 3 EARN LOSSŠþh country þ h year þ e t CFO tþ1 ¼ IFRS ½b 0 þ b 1 EARN þ b 2 LOSS þ b 3 EARN LOSSŠþGAAP=DAS ½x 0 þ x 1 EARN t þ x 2 LOSS þ x 3 EARN LOSSŠþw country þ w year þ k t ð1þ ð2þ where EARN = net income before extraordinary items (Compustat Global data Item #32); CFO = cash flows from operations, measured as EARN accruals, where accruals are calculated as the change 8 We assess the statistical significance of all of our models using Roger s standard errors, clustered by firm.

7 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) in noncash current assets (Item #75 Item #60) less the change in current liabilities (Item #104) plus the change in the current portion of long-term debt (Item #94) plus depreciation (Item #11), following Pincus et al. (2007); LOSS = an indicator set to one if EARN is negative in year t, zero otherwise; IFRS = an indicator set to one for firms in the IFRS sample, zero otherwise; GAAP = an indicator set to one for firms in the U.S. GAAP sample, zero otherwise; DAS = an indicator set to one for firms in the non-u.s. DAS sample, zero otherwise; h country = country fixed effects in the future earnings model 9 ; h year, year fixed effects in the future earnings model; w country = country fixed effects in the future cash flows model; and w year = country fixed effects in the future cash flows model. We scale both EARN and CFO by average total assets (Item #89). We include LOSS and the interaction EARN LOSS in our regressions to allow for differences in earnings persistence and in the association between current earnings and future cash flows across firms reporting current year profits versus losses. 10 Hayn (1995) and Collins et al. (1999) conclude that losses are less informative than profits about a firm s future prospects. Since the publication of these studies, the partitioning of profit and loss firms has become common in empirical research on firms reporting under U.S. GAAP. 11 Moreover, Bartov et al. (2005) find that the relation between earnings and returns differs for profit and loss firms reporting under IFRS and Ndubizu and Sanchez (2006) conclude that losses reported by Chilean firms under U.S. GAAP are more timely, conservative, and informative about expected future normal earnings than are losses reported by Peruvian firms under IFRS. These differences lead us to believe that the relation between current and future earnings and between current earnings and future cash flows may differ for firms reporting losses versus profits. Moreover, the extent of these differences may be unequal across firms reporting under U.S. GAAP versus IFRS. We include year and country variables to control for fixed cross-country and time effects. As part of our sensitivity analyses, we also conduct our tests using the 2 year average future earnings and cash flows. In addition, we conduct analyses including five country-level variables from La Porta et al. (1998) which control for cross-country differences in institutional structures and accounting structures, specifically, code- versus common-law, investor rights, ownership concentration, importance of equity markets, legal enforcement, and book-tax conformity. We describe these results in the robustness tests section (Section 4). In models (1) and (2), we stack two regressions into one model: the first regression is estimated using IFRS observations only, and the second regression is estimated using GAAP or DAS observations only. 12 Estimating the regressions in a stack allows us to test for statistically significant differences in the coefficients. The coefficients on EARN estimate earnings persistence in model (1) and the association between earnings and future cash flows in model (2) for firms that report non-negative earnings in year t. Thus, we are interested in whether the coefficients on EARN are equal for IFRS and U.S. GAAP or non-u.s. DAS firms (i.e., we test whether a 1 = c 1 in model (1) and whether b 1 = x 1 in model (2)). If a 1 (or b 1 )is significantly greater than c 1 (or x 1 ), this suggests that earnings persistence is higher (or current earnings are more closely associated with future cash flows) for firms reporting under IFRS than for firms reporting under U.S. GAAP or non-u.s. DAS when those firms report non-negative earnings. Similarly, if a 1 (or b 1 ) is significantly less than c 1 (or x 1 ), this suggests that earnings persistence is lower (or current earnings are less closely associated with future cash flows) for firms reporting under IFRS than for firms reporting under U.S. GAAP or non-u.s. DAS when those firms report non-negative earnings We follow Hope (2003) in including country indicator variables to control for fixed cross-country differences. 10 Note that allowing the persistence coefficients in our models to vary across profit and loss firms is less restrictive than forcing the coefficients to be the same. That is, if the persistence coefficients for profit and loss firms are equal, then our models will allow for that relation (i.e., the coefficients will be equal). However, if the persistence coefficients differ for profit versus loss firms, our models allow for the observed coefficients to differ. 11 See, for example, Joos and Plesko (2005), Klein and Marquardt (2006), Darrough and Ye (2007), Franzen and Radhakrishnan (2009), Balakrishnan et al. (2010), Dhaliwal et al. (2010), and Li (2010). 12 Our use of stacked regressions is similar to that in Riedl (2004). 13 Note that we estimate the average persistence coefficient by reporting standard (i.e., for all firms reporting under U.S. GAAP, IFRS, or domestic standards) rather than estimating a separate persistence coefficient by reporting standard for each country. Modeling the later would require that we include an interactive effect for each country-standard and would make estimation of the model non-tractable. Because of this, we focus our analyses on the average persistence level for IFRS, Domestic Standards, and US GAAP firms. In addition, we acknowledge that accounting quality may differ across countries within these sub-samples. However, we suggest that this may induce noise rather than bias.

8 110 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) We are also interested in the summed coefficients on EARN and on the interaction term EARN - LOSS (i.e., in a 1 + a 3 and c 1 + c 3 ), which estimate earnings persistence and the association between earnings and future cash flows respectively for firms that report losses in year t. Thus, we perform similar tests of equality across reporting standards to investigate whether earnings persistence and the association between current earnings and future cash flows differ for loss firms (i.e., we test whether a 1 + a 3 = c 1 + c 3 for model (1) and whether b 1 + b 3 = x 1 + x 3 for model (2)) Sample selection To select our sample, we obtain all firm-year observations from the Compustat Global Industrial/ Commercial file from 2002 through 2008 with sufficient data to calculate our accounting variables. 14 We begin our sample period in 2002 to correspond with the European Parliament and Council s endorsement of IFRS, which required E.U. firms to prepare their financial statements in conformity with IFRS starting in fiscal We impose four additional data requirements. First, we follow Bartov et al. (2005) and delete observations that do not provide fully consolidated financial data. Second, we delete any observations where the firm changed from domestic standards to IFRS in the current or future year. 15 Third, we delete observations in the top or bottom 1/2% of the distributions of our continuous accounting variables in each year to remove potential outliers. Finally, we require countries to have at least 40 usable observations with sufficient data in the year to be included in the sample, 16 and include only observations from countries with La Porta et al. (1997, 1998) proxies for country-level institutional structures and accounting structures (i.e., legal origin, investor rights, ownership concentration, the importance of the equity market, and the strength of the legal enforcement). These restrictions result in a final sample of 58,832 firm-year observations drawn from 33 countries. Table 1 presents a list of the countries in our sample and partitions on whether the country allows firms to report under IFRS during at least part of our sample period. Table 1 also provides the number of sample observations from each country. The majority of countries included in our sample allow the use of IFRS (25 versus 8); however, more observations are drawn from countries that do not allow IFRS (N = 31,601) than from countries that allow IFRS (N = 27,231). Furthermore, for those observations drawn from countries that do not allow IFRS (N = 31,601), more observations are drawn from countries other than the U.S. (N = 20,243) than from the U.S. (N = 11,358) Sub-sample construction We use the Compustat Global accounting standard variable (data item = ASTD) to classify observations into two groups. Consistent with Daske et al. (2008), the IFRS group consists of observations with ASTD codes of DI, DA, or DT. The U.S. GAAP group consists of observations with ASTD codes of DU, MU, or US. The non-u.s. DAS group consists of observations with ASTD codes of DD, DO, DR, DS, MI, or LJ. Our tests investigate the earnings persistence and the associations between current earnings and future cash flows for IFRS versus U.S. GAAP firms and for IFRS versus non-u.s. DAS firms Descriptive statistics In Table 2, Panel A, we present descriptive statistics for the test variables in models (1) and (2) for our full sample. Approximately 10% of all observations report under IFRS (so 20% of firm-year observations from countries that allow IFRS report under IFRS) and 27% of sample observations report a loss. In Panel B, we present Spearman and Pearson correlations. The correlations between current earnings 14 Since the dependent variables in models (1) and (2) are future earnings and future cash flows, respectively, the dependent variables are from 2003 through Our empirical tests require 2 years of data so this requirement ensures that the standards used are consistent between years t and t We require 40 usable observations in a year to ensure that our results are not driven by extremely small samples in particular countries.

9 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Table 1 Sample countries and reporting practices under IFRS. Country N Countries having resident firms reporting under IFRS Australia 3939 Austria 204 Belgium 275 Brazil 482 Denmark 380 Finland 366 France 1847 Germany 2004 Greece 257 Hong Kong 589 Indonesia 834 Italy 713 Korea 1059 Malaysia 3018 Mexico 293 Netherlands 440 Norway 382 Philippines 331 Singapore 1750 South Africa 590 Spain 408 Sweden 776 Switzerland 736 Thailand 1291 U.K Total 27,231 Countries having no resident firms reporting under IFRS Canada 1855 Chile 391 India 555 Ireland 84 Japan 12,917 New Zealand 342 Taiwan 4099 U.S. 11,358 Total 31,601 and future earnings and between current earnings and future cash flows are positive and significant (p ). In Table 3, we present descriptive statistics (i.e., means and medians) by reporting regime and test for differences across reporting regimes. We find that observations reported under IFRS have significantly higher mean and median future earnings, future cash flows, and current earnings than do observations reported under U.S. GAAP or non-u.s. DAS. We also find that firms reporting under IFRS have a significantly lower incidence of losses than firms reporting under either U.S. GAAP or non-u.s. DAS. 4. Empirical results 4.1. IFRS versus U.S. GAAP In this section, we investigate whether earnings are more or less persistent and/or more or less closely associated with future cash flows under IFRS versus U.S. GAAP. These comparisons are especially important given that foreign firms are no longer required to reconcile earnings reported under IFRS to

10 112 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Table 2 Descriptive statistics and correlations. Variable Mean Std. dev. Q 1 Median Q 3 Panel A: Descriptive statistics EARN t CFO t EARN t LOSS t 27% IFRS 10% GAAP 20% DAS 70% N 58,832 EARN t+1 CFO t+1 EARN t LOSS t IFRS GAAP DAS Panel B: Pearson (above the diagonal) and spearman (below the diagonal) correlations EARN t CFO t EARN t LOSS t IFRS GAAP DAS Notes: Bolded text indicates that the correlation is significant at p < 0.10; EARN = net income before extraordinary items (Item #32); CFO = cash flows from operations; IFRS = one for firms in the IFRS sample, zero otherwise; GAAP = one for firms in the GAAP sample, zero otherwise; and DAS = one for firms in the GAAP sample, zero otherwise. Table 3 Univariate comparisons between observations reporting under IFRS versus non-u.s. domestic accounting standards (non-u.s. DAS) and under IFRS versus U.S. domestic accounting standards (U.S. GAAP). IFRS U.S. GAAP Non-U.S. DAS IFRS versus U.S. GAAP IFRS versus. non-u.s. DAS Non-U.S. DAS versus. U.S. GAAP N ,946 41,162 EARN t+1 Mean (t-stat) (9.33) *** (11.71) *** (1.19) Median (Z-stat) (6.10) *** (18.36) *** (13.26) *** CFO t+1 Mean (t-stat) (8.40) *** (12.59) *** (4.44) *** Median (Z-stat) (5.64) *** (16.79) *** (13.93) *** EARN t Mean (t-stat) (10.39) *** (10.88) *** ( 1.97) ** Median (Z-stat) (7.28) *** (17.71) *** (10.47) *** LOSS t Mean (t-stat) 21% 29% 27% ( 11.27) *** ( 8.97) *** ( 5.23) *** Notes: All variables are defined as in Table 2; IFRS = the IFRS sample; non-u.s. DAS = the non-u.s. domestic accounting standards sample; and U.S. GAAP = the U.S. domestic accounting standards sample. ** Significance at 5%. *** Significance at 1%. U.S. GAAP and because U.S. firms may be transitioning from U.S. GAAP to IFRS over the next few years. We first test for differences across the IFRS and U.S. GAAP samples using all sample observations re-

11 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Table 4 Tests of associations between current earnings and future earnings for observations reporting under IFRS versus U.S. GAAP. EARN tþi ¼ IFRS ½a 0 þ a 1 EARN t þ a 2 LOSS t þ a 3 EARN t LOSS t ŠþGAAP ½c 0 þ c 1 EARN t þ c 2 LOSS t þ c 3 EARN t LOSS t Šþh country þ h year þ e t ð1þ Panel A: 1-year-ahead (i = 1) Panel B: 2-year-ahead avg. ((i =1+i = 2)/2) IFRS sample U.S. GAAP sample IFRS sample U.S. GAAP sample IFRS (3.53) *** (2.45) ** GAAP (3.60) *** (2.69) *** EARN (23.21) *** (33.35) *** (15.09) *** (25.85) *** LOSS ( 4.61) *** ( 4.95) *** ( 2.61) *** ( 1.93) * EARN LOSS ( 3.85) *** ( 1.16) ( 2.33) ** (0.57) Country indicators Yes Yes Yes Yes Year indicators Yes Yes Yes Yes Adjusted R N 17,670 17,670 12,731 Test of coefficient equality Diff Diff EARN (test a 1 = c 1 ) ( 0.18) (0.03) EARN LOSS (test a 3 = c 3 ) ( 2.84) *** ( 2.20) ** EARN + EARN LOSS (test a 1 + a 3 = c 1 + c 3 ) ( 3.49) *** ( 2.49) ** Notes: All variables are defined as in Table 2; t-stats (in parentheses) are based on standard errors clustered by firm. * Significant at 10%. ** Significant at 5%. *** Significant at 1%. ported under IFRS regardless of the country of origin and all sample observations reported under U.S. GAAP regardless of the country of origin. Here, N = 17,670. Table 4, Panel A presents the results for model (1), comparing 1-year-ahead earnings persistence for observations reported under IFRS versus U.S. GAAP. The earnings persistence parameters are significantly positive at approximately 0.71 for both samples (a 1 = for IFRS firms and c 1 = for U.S. GAAP firms) and a test of equality reveals that they do not differ across accounting standards (i.e., for the test of a 1 c 1, p > 0.10, so we do not reject the null hypothesis that a 1 = c 1 ). Thus, positive earnings reported under IFRS are no more or less persistent than positive earnings reported under U.S. GAAP. For the IFRS sample, the persistence of current reported losses (a 1 + a 3 = 0.470) is lower than the persistence of profits (a 1 = 0.707) and this difference is significant (a 3 = 0.237, p ). In contrast, for U.S. GAAP firms, the persistence of losses (c 1 + c 3 = 0.676) and the persistence of profits (c 1 = 0.714) are not significantly different (i.e., for the test of c 3 0, p > 0.10, so we do not reject the null hypothesis that c 3 = 0). Furthermore, a joint test of equality reveals that the persistence of losses reported under IFRS (a 1 + a 3 = 0.470) is lower than the persistence of losses reported under U.S. GAAP (c 1 + c 3 = 0.676) and this difference is significant (i.e., for the test a 1 + a 3 c 1 + c 3, p , so we reject the null hypothesis that a 1 + a 3 = c 1 + c 3 ). This may be due to firms ability to report significantly higher earnings under IFRS in general (O Connell and Sullivan, 2008; Henry et al., 2009), which makes reported losses less likely to occur, or due to more frequent recognition of large losses under IFRS (Barth et al., 2008) Barth et al. (2008) compares IFRS firms to non-u.s. DAS firms and conclude that the non-u.s. DAS firms smooth earnings by delaying the effects of negative outcomes, resulting in less timely loss recognition.

12 114 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Table 5 Tests of associations between current earnings and future cash flows for observations reporting under IFRS versus under U.S. GAAP. CFO tþi ¼ IFRS ½b 0 þ b 1 EARN t þ b 2 LOSS t þ b 3 EARN t LOSS t ŠþGAAP ½x 0 þ x 1 EARN t þ x 2 LOSS t þ x 3 EARN t LOSS t Šþw country þ w year þ e t ð2þ Panel A: 1-year-ahead (i = 1) Panel B: 2-year-ahead avg.((i = 1 + i = 2)/2) IFRS sample U.S. GAAP sample IFRS sample U.S. GAAP sample IFRS (5.52) *** (4.62) *** GAAP (4.92) *** (4.48) *** EARN (13.29) *** (24.25) *** (10.48) *** (21.54) *** LOSS ( 0.95) ( 2.20) ** (0.40) (0.97) EARN LOSS ( 1.14) ( 0.25) ( 0.49) (0.24) Country indicators Yes Yes Yes Yes Year indicators Yes Yes Yes Yes Adjusted R N 17,670 12,602 Test of coefficient equality Diff Diff EARN (test b 1 = x 1 ) ( 1.96) ** ( 1.74) * EARN LOSS (test b 3 = x 3 ) ( 0.87) ( 0.55) EARN + EARN LOSS (test b 1 + b 3 = x 1 + x 3 ) ( 3.01) *** ( 2.12) ** Notes: All variables are defined as in Table 2; t-stats (in parentheses) are based on standard errors clustered by firm. * Significant at 10%. ** Significant at 5%. *** Significant at 1%. Table 5, Panel A presents the results for model (2), comparing the association between current earnings and 1-year-ahead cash flows for observations reported under IFRS versus U.S. GAAP. Here, we find that current earnings are significantly positively associated with future cash flows for firms reporting profits under IFRS (b 1 = 0.568) and under U.S. GAAP (x 1 = 0.668) but, importantly, tests of equality reveal that the association between current earnings and future cash flows is significantly more positive for profit firms reporting under U.S. GAAP than for profit firms reporting under IFRS (b 1 x 1, p ). For both IFRS and U.S. GAAP firms, the association between current year losses and future period cash flows is not significantly different from the association between current year profits and future period cash flows (b 3 =0,x 3 = 0). Similarly, current earnings are significantly positively associated with future cash flows for firms reporting losses under IFRS (b 1 + b 3 = 0.495) and under U.S. GAAP (x 1 + x 3 = 0.659) but, importantly, tests of equality reveal that the association between current earnings and future cash flows is significantly more positive for loss firms reporting under U.S. GAAP than for loss firms reporting under IFRS (b 1 + b 3 x 1 + x 3, p ). Overall, Tables 4 and 5 reveal that, on average, losses reported under U.S. GAAP provide more information for predicting future earnings than do losses reported under IFRS and both profits and losses reported under U.S. GAAP provide more information for predicting future cash flows than do profits and losses reported under IFRS IFRS versus non-u.s. DAS We next compare earnings persistence and the association between earnings and future cash flows for the IFRS and non-u.s DAS samples, using all available sample observations reporting under IFRS

13 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) regardless of country of origin and under non-u.s. DAS regardless of country of origin. Here, N = 46,886. Table 6, Panel A presents the estimation results for model (1), which measures 1-year-ahead earnings persistence. The earnings persistence coefficients are positive and significant for earnings reported under both IFRS (a 1 = 0.709) and non-u.s. DAS (c 1 = 0.736), and a test of equality reveals that these coefficients are not significantly different across the two samples (i.e., for the test a 1 c 1, p > 0.10, so we do not reject the null hypothesis that a 1 = c 1 ). While earnings persistence is also lower for loss firms than for profit firms in both samples (a 1 + a 3 = for IFRS firms and c 1 + c 3 = for non-u.s. DAS firms), the persistence of reported losses is not significantly different for firms reporting under IFRS versus non-u.s. DAS. Thus, we find no evidence that earnings are more or less persistent under IFRS than under non-u.s. DAS. Table 7, Panel A presents the estimation results for model (2), which tests the associations between current earnings and 1-year-ahead cash flows. As is the case for the earnings model, we find that the association between current earnings and future cash flows is significant and positive under both reporting regimes (for profit firms, b 1 = under IFRS and x 1 = under non-u.s. GAAP; for loss firms, b 1 + b 3 = under IFRS and x 1 + x 3 = under non-u.s. GAAP), and these associations are not significantly different across the two reporting regimes. Moreover, the association between current earnings and future cash flows is significantly lower for firms reporting losses than for firms reporting profits for the non-u.s. DAS sample (x 3 = 0.094) but this association is not significantly different for profit versus loss firms reporting under IFRS. Finally, joint tests of equality reveal that there Table 6 Tests of associations between current earnings and future earnings for observations reporting under IFRS versus non-u.s. domestic accounting standards (non-u.s. DAS). EARN tþi ¼ IFRS ½a 0 þ a 1 EARN t þ a 2 LOSS t þ a 3 EARN t LOSS t ŠþDAS ½c 0 þ c 1 EARN t þ c 2 LOSS t þ c 3 EARN t LOSS t Šþh country þ h year þ e t ð1þ Panel A: 1-year-ahead (i = 1) Panel B: 2-year-ahead avg. ((i =1+i = 2)/2) IFRS sample U.S. GAAP sample IFRS sample U.S. GAAP sample IFRS (5.22) *** (5.21) *** DAS (3.59) *** (4.86) *** EARN (23.86) *** (50.42) *** (15.63) *** (38.67) *** LOSS ( 4.72) *** ( 12.54) *** ( 2.75) *** ( 8.65) *** EARN LOSS ( 4.09) *** ( 9.14) *** ( 2.49) ** ( 5.18) *** Country indicators Yes Yes Yes Yes Year indicators Yes Yes Yes Yes Adjusted R N 46,886 37,867 Test of coefficient equality Diff Diff EARN (test a 1 = c 1 ) ( 0.80) ( 0.48) EARN LOSS (test a 3 = c 3 ) ( 0.72) ( 0.37) EARN + EARN LOSS (test a 1 + a 3 = c 1 + c 3 ) ( 1.32) ( 0.77) Notes: All variables are defined as in Table 2; t-stats (in parentheses) are based on standard errors clustered by firm. Significant at 10%. ** Significant at 5%. *** Significant at 1%.

14 116 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Table 7 Tests of associations between current earnings and future cash flows for observations reporting under IFRS versus non-u.s. domestic accounting standards (non-u.s. DAS) CFO tþi ¼ IFRS ½b 0 þ b 1 EARN t þ b 2 LOSS t þ b 3 EARN t LOSS t ŠþDAS ½x 0 þ x 1 EARN t þ x 2 LOSS t þ x 3 EARN t LOSS t Šþw country þ w year þ k t ð2þ Panel A: 1-year-ahead (i = 1) Panel B: 2-year-ahead avg. ((i =1+i = 2)/2) IFRS sample U.S. GAAP sample IFRS sample U.S. GAAP sample IFRS (9.66) *** (9.07) *** DAS (9.90) *** (10.55) *** EARN (13.58) *** (33.73) *** (10.85) *** (29.47) *** LOSS ( 1.06) ( 7.18) *** (0.30) ( 4.91) *** EARN LOSS ( 1.25) ( 3.75) *** ( 0.69) ( 2.09) ** Country indicators Yes Yes Yes Yes Year indicators Yes Yes Yes Yes Adjusted R N 46,886 37,271 Test of coefficient equality Diff Diff EARN (test b 1 = x 1 ) ( 1.19) ( 0.81) EARN LOSS (test b 3 = x 3 ) (0.25) (0.11) EARN + EARN LOSS (test b 1 + b 3 = x 1 + x 3 ) ( 0.73) ( 0.55) Notes: All variables are defined as in Table 2; t-stats (in parentheses) are based on standard errors clustered by firm. Significant at 10%. ** Significant at 5%. *** Significant at 1%. is no significant difference in the association between current losses and future cash flows for firms reporting under IFRS versus firms reporting under non-u.s. DAS. Together, the results presented in Tables 6 and 7 suggest that earnings reported under IFRS are generally no more or less closely associated with future earnings and/or future cash flows than are earnings reported under non-u.s DAS Robustness test We perform four robustness tests to examine whether our main results are sensitive to alternative variable measurements, to additional control variables, to alternative econometric models, or to sample selection. First, we estimate our models using a more long-term measure of future earnings and cash flows. Specifically, we calculate the average future earnings and cash flows using 1- and 2-year-ahead data. 18 Our sample size decreases to 12,602 observations for the IFRS versus U.S. GAAP sample and to 37,271 observations for the IFRS versus non-u.s. GAAP sample. We find that the results using the 2-year-ahead average (see Panel B of Tables 4 7) are qualitatively similar to those reported using 1-year ahead measures (i.e., Panel A of Tables 4 7). 18 We do not use 3 years of future data because this severely limits our sample. Here, fiscal 2005 would be the last base year in our sample, with future earnings and cash flows being calculated using data from fiscal 2006, 2007, and Given that E.U. firms are required to prepare their financial statements in conformity with IFRS starting in fiscal 2005, this methodology would result in only one potential observation per IFRS firm for those adopting in 2005.

15 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Second, we test the robustness of our results to additional control variables. To control for crosscountry differences in institutional and accounting structures, we add variables to models (1) and (2) that capture cross-country differences in legal systems (code law versus common-law), investor rights, ownership concentration, importance of equity markets, and legal enforcement from La Porta et al. (1998) and book-tax conformity from Atwood et al. (2009). The results (untabulated) are qualitatively similar to those reported. We also estimate the model with industry fixed effects (untabulated) and find similar results as well. Third, we use an alternative econometrical model. Recall that we estimate models (1) and (2) using a stacked regression approach in order to facilitate the interpretation and comparison of the estimated coefficients. An alternative is to model the associations using triple interactions as follows: EARN tþ1 ¼ d 0 GAAP=DAS þ d 1 EARN t þ d 2 LOSS t þ d 3 EARN LOSS t þ d 4 GAAP=DAS EARN t þ d 5 GAAP=DAS t LOSS t þ d 6 GAAP=DAS t EARN t LOSS t þ h country þ hyear þ e t ð3þ CFO tþ1 ¼ r 0 GAAP=DAS þ r 1 EARN t þ r 2 LOSS t þ r 3 EARN LOSS t þ r 4 GAAP=DAS EARN t þ r 5 GAAP=DAS t LOSS t þ r 6 GAAP=DAS t EARN t LOSS t þ w country þ w year þ k t ð4þ where all variables are defined previously. In model (3), the coefficients of interest are d 4 and d 6. The coefficient d 4 measures the difference in earnings persistence when earnings are positive for IFRS versus U.S. GAAP or non-u.s. firms, and d 4 + d 6 measures the difference in earnings persistence when earnings are negative for IFRS versus U.S. GAAP or non-u.s. firms. In model (4), the coefficients of interest are r 4 and r 6. The coefficient r 4 measures the difference in the association between earnings and future cash flows when earnings are positive for IFRS versus U.S. GAAP or non-u.s. firms, and r 4 + r 6 measures the difference in the association between earnings and future cash flows when earnings are negative for IFRS versus U.S. GAAP or non-u.s. firms. We estimate models (3) and (4) (untabulated) and find similar results. 19 Finally, to investigate whether our results are a function of those firms that voluntarily elect to early adopt IFRS, we identify the early adopters in our sample and remove them when we run our tests. To identify the early adopters, we hand-collect mandatory IFRS adoption dates from the IASB website. If a firm reports under IFRS in a reporting period before the mandatory adoption date, we code it is a voluntary adopter and remove it from the sample for this robustness test. We also remove from the sample any non-u.s. firm that voluntarily elects to report under U.S. GAAP. We find that 3.4% of the firms in our sample are removed based on these two additional criteria. The removal of these firms eliminates unobservable factors associated with the decision to adopt IFRS or U.S. GAAP which potentially bias our results. We find that all of our results hold when these firms are omitted. 5. Conclusion Motivated by the current debate regarding the adoption of IFRS in the U.S., we investigate whether the use of domestic accounting standards (DAS) versus IFRS is associated with differences in earnings persistence and the association between current earnings and future cash flows. We use a sample of firm-year observations from 33 countries to examine differences for firms reporting under IFRS relative to two DAS control samples a sample consisting of all firms reporting under U.S. GAAP (the U.S. GAAP sample) and a sample consisting of all non-u.s. firms reporting under their local domestic standards (the non-u.s. DAS sample). We find that current and future earnings are positively associated 19 Specifically, when we estimate the models using IFRS and U.S. GAAP firms, we find that d 4 is insignificant (d 4 = 0.007, p > 0.10) and that (d 4 + d 6 ) is positive (d 4 + d 6 = 0.205) and significant (p < 0.01). Using this sample, we also find that r 4 is positive (r 4 = 0.100) and significant (p < 0.05) and that (r 4 + r 6 ) is positive (r 4 + r 6 = 0.164) and significant (p < 0.01). When we estimate the models using IFRS and non-u.s. GAAP firms, we find that the coefficient estimates are as follows: d 4 = 0.026, (d 4 + d 6 ) = 0.071, r 4 = 0.053, and (r 4 + r 6 ) = 0.037, but that d 4,(d 4 + d 6 ), r 4, but (r 4 + r 6 ) are all insignificantly different from zero.

16 118 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) and that losses are generally less persistent than profits in all three samples. We further find no significant difference in the persistence of positive earnings reported under IFRS versus under U.S. GAAP or under IFRS versus non-u.s. DAS. However, we find that losses reported under IFRS are less persistent than losses reported under U.S. GAAP but the persistence of losses reported under IFRS is not significantly different from the persistence of losses reported under non-u.s. DAS. We also examine the association between current reported earnings and future cash flows for firms reporting under IFRS versus U.S. GAAP and non-u.s. DAS. We find that current earnings are positively associated with future cash flows across all reporting regimes and find no difference in the association between current earnings and future cash flows across profit and loss firms reporting under IFRS and U.S. GAAP; however, losses reported under non-u.s. DAS are less closely associated with future cash flows than are profits. More importantly, current earnings and losses reported under U.S. GAAP are more closely associated with future cash flows than are current earnings and losses reported under IFRS. These differences suggest that current earnings and losses are more informative about future cash flows for firms reporting under U.S. GAAP than for firms reporting under IFRS. These findings are robust to controls for year and country fixed effects and for cross-country factors including legal structure, investor rights, ownership equity, the importance of equity markets, and the strength of legal enforcement. We find no significant differences for IFRS versus non-u.s. DAS firms in terms of associations between current profits or losses and future cash flows. Our results provide useful information to U.S. standard-setters as they consider the costs and benefits of adopting IFRS because they imply that accounting earnings prepared under IFRS tell us less about future cash flows than accounting earnings prepared under U.S. GAAP even though earnings persistence is similar across all reporting regimes. Our results suggest that, while IFRS and U.S. GAAP are both high quality sets of accounting standards, U.S. GAAP is superior with respect to the prediction of future cash flows. We acknowledge that versions of IFRS have emerged throughout the world and thus, our tests compare U.S. GAAP to the average of all firms reporting under (some version of) IFRS. Since we cannot know which version the U.S. will adopt (or whether a new version will emerge), our statements refer to the average quality of financial statements currently issued under IFRS. Acknowledgements We thank workshop participants at the University of Arkansas for helpful comments and suggestions. James Myers gratefully acknowledges financial support from the Ralph L. McQueen Professorship at the University of Arkansas, and Linda Myers gratefully acknowledges financial support from the Garrison/Wilson Chair at the University of Arkansas. References Ashbaugh, H., Olsson, P., An exploratory study of the valuation properties of cross-listed firms IAS and U.S. GAAP earnings and book values. The Accounting Review 77, Ashbaugh, H., Pincus, M., Domestic accounting standards, international accounting standards, and the predictability of earnings. Journal of Accounting Research 39, Atwood, T.J., Drake, M.S., Myers, L.A., Book-tax conformity, earnings persistence and the association between earnings and future cash flows. Journal of Accounting and Economics 50, Badertscher, B., Collins, D., Lys, T., Earnings Management and the Predictive Ability of Accruals with Respect to Future Cash Flows. Working Paper, University of Notre Dame, University of Iowa, and Northwestern University. Balakrishnan, K., Bartov, E., Faurel, L., Post loss/profit announcement drift. Journal of Accounting and Economics 50, Barth, M.E., Cram, D., Nelson, K., Accruals and the prediction of future cash flows. The Accounting Review 76, Barth, M.E., Landsman, W., Lang, M., International accounting standards and accounting quality. Journal of Accounting Research 46, Bartov, E., Goldberg, S.R., Kim, M., Comparative value relevance among German, U.S., and international accounting standards: a German stock market perspective. Journal of Accounting, Auditing and Finance 20, Bennett, B., Bradbury, M., Prangnell, H., Rules, principles and judgments in accounting standards. Abacus 42, Bradshaw, M.T., Miller, G.S., Will harmonizing accounting standards really harmonize accounting? Evidence from non- U.S. firms adopting U.S. GAAP. Journal of Accounting, Auditing and Finance 23, Bradshaw, M., Callahan, C., Ciesielski, J., Gordon, E., Hodder, L., Hopkins, P., Kohlbeck, M., Laux, R., McVay, S., Stober, T., Stocken, P., Yohn, T., Response to the SEC s proposed rule roadmap for the potential use of financial statements prepared in accordance with international financial reporting standards (IFRS) by U.S. issuers. Accounting Horizons 24,

17 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Carmona, S., Trombetta, M., On the global acceptance of IAS/IFRS accounting standards: the logic and implications of the principles-based system. Journal of Accounting and Public Policy 27, Chen, L.H., Sami, H., Trading volume reaction to the earnings reconciliation from IAS to U.S. GAAP. Contemporary Accounting Research 25, Chen, C.J.P., Gul, F.A., Su, X., A comparison of reported earnings under Chinese GAAP vs. IAS: evidence from the Shanghai stock exchange. Accounting Horizons 13, Chen, H., Tang, Q., Jiang, Y., Lin, Z., The role of international financial reporting standards in accounting quality: evidence from the European Union. Journal of International Financial Management and Accounting 21, Christensen, H.B., Lee, E., Walker, M., Do IFRS/UK-GAAP reconciliations convey information? The effect of debt contracting. Journal of Accounting Research 47, Ciesielski, J., International Accounting Standards, Opportunities, Challenges and Global Convergence. Testimony before the Subcommittee on Securities, Insurance, and Investment, Committee on Banking, Housing, and Urban Affairs, U.S. Senate, October 24. Cohn, M., IFRS Roadmap Slams into Reverse. WebCPA, March 11. Collins, D., Pincus, M., Xie, H., Equity valuation and negative earnings: the role of book value of equity. The Accounting Review 74, Covrig, V.M., DeFond, M.L., Hung, M., Home bias, foreign mutual fund holdings, and the voluntary adoption of international accounting standards. Journal of Accounting Research 45, Dargenidou, C., McLeay, S., Raonic, I., Expected earnings growth and the cost of capital: an analysis of accounting regime change in the European financial market. Abacus 42, Darrough, M., Ye, J., Valuation of loss firms in a knowledge-based economy. Review of Accounting Studies 12, Daske, H., Economic benefits of adopting IFRS or US GAAP: has the expected cost of equity capital really decreased? Journal of Business Finance and Accounting 33, Daske, H., Gebhardt, G., International financial reporting standards and expert s perceptions of disclosure quality. Abacus 42, Daske, H., Hail, L., Leuz, C., Verdi, R., Mandatory IFRS reporting around the world: early evidence on the economic consequences. Journal of Accounting Research 46, Defelice, A., FASB, IASB leaders boost collaboration efforts to meet 2011 convergence goal. Journal of Accountancy (October 30). 2 p. Dhaliwal, D., Kaplan, S., Laux, R., Weisbrod, E., The Information Content of Tax Expense for Firms Reporting Losses. Working Paper, University of Arizona, Arizona State University, The Pennsylvania State University, Arizona State University. Ding, Y., Hope, O.K., Jeanjean, T., Stolowy, H., Differences between domestic accounting standards and IAS: measurement, determinants and implications. Journal of Accounting and Public Policy 26, Ernstberger, J., Krotter, S., Stadler, C., Analysts forecast accuracy in Germany: the effect of different accounting principles and changes of accounting principles. BuR Business Research 1, Financial Accounting Standards Board (FASB), Objectives of Financial Reporting by Business Enterprises, Statement of Financial Accounting Concepts No. 1, Stamford, CT. Fédération Française des Sociétés d Assurances (FFSA) and Association Française de la Gestion Financiére (AFG), Investor Perspectives on IFRS Implementation, Collection of Essays, December. Forgeas, R., SEC s Leap Toward IFRS, Has the Momentum Gone? AICPA, February 23. < Francis, J.R., Khurana, I.K., Martin, X., Pereira, R., The role of firm-specific incentives and country factors in explaining voluntary IAS adoptions: evidence from private firms. European Accounting Review 1, 30. Franzen, L., Radhakrishnan, S., The value relevance of R&D across profit and loss firms. Journal of Accounting and Public Policy 28, Financial Services Agency, Publication of the Revised Capital Office Ordinances, etc. for the Voluntary Application of International Financial Reporting Standards in Japan. Press Release, December 11. Gjerde, O., Knivsfla, K.H., Saettem, F., The value-relevance of adopting IFRS: evidence from 145 NGAAP restatements. Journal of International Accounting, Auditing and Taxation 17, Gordon, E., Jorgensen, B., Linthicum, C., Could IFRS Replace US GAAP? A Comparison of Earnings Attributes and Informativeness in the US Market. Working Paper, Temple University, Columbia University, and University of Texas, San Antonio. Hail, L., Leuz, C., Wysocki, P., Global accounting convergence and the potential adoption of IFRS by the U.S. (Part I): conceptual underpinnings and economic analysis. Accounting Horizons 24, Hail, L., Leuz, C., Wysocki, P., forthcoming. Global accounting convergence and the potential adoption of IFRS by the U.S. (Part II): political factors and future scenarios for U.S. accounting standards. Accounting Horizons 24. Harris, M.S., Muller III, K.A., The market valuation of IAS versus US-GAAP accounting measures using Form 20-F reconciliations. Journal of Accounting and Economics 26, Haverty, J.L., Are IFRS and U.S. GAAP converging? Some evidence from People s Republic of China companies listed on the New York Stock Exchange. Journal of International Accounting, Auditing and Taxation 15, Hayn, K., The information content of losses. Journal of Accounting and Economics 20, Heffes, E.M., Sir David Tweedie asks: what s the U.S. going to do about IFRS?. Financial Executive 25, 14. Henry, E., Lin, S., Yang, Y., The European-U.S. GAAP gap: IFRS to U.S. GAAP Form 20-F reconciliations. Accounting Horizons 23, Herz, R., International Accounting Standards, Opportunities, Challenges and Global Convergence. Testimony before the Subcommittee on Securities, Insurance, and Investment, Committee on Banking, Housing, and Urban Affairs, U.S. Senate, October 24. Holthausen, R.W., Accounting standards, financial reporting outcomes, and enforcement. Journal of Accounting Research 47, Holthausen, R.W., Watts, R.L., The relevance of the value-relevance literature for financial accounting standard setting. Journal of Accounting and Economics 31, 3 75.

18 120 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Hope, O.K., Disclosure practices, enforcement of accounting standards, and analyst forecast accuracy: an international study. Journal of Accounting Research 41, Hopkins, P.E., Botosan, R.L., Bradshaw, M., Callahan, C., Ciesielski, J., Farber, D., Kohlbeck, M., Hodder, L., Laux, B., Stober, T., Stocken, P., Yohn, T., Response to the SEC release acceptance from foreign private issuers of financial statements prepared in accordance with international financial reporting standards without reconciliation to U.S. GAAP. American Accounting Association s Financial Accounting and Reporting Section of the Financial Reporting Policy Committee. Accounting Horizons 22, Howell, M Global Accounting Rules may Face Delays. Reuters.com, February 4. International Accounting Standards Board (IASB), Framework for the Preparation and Presentation of Financial Statements. Ijiri, Y., US accounting standards and their environment: a dualistic study of their 75-years of transition. Journal of Accounting and Public Policy 24, Jamal, K., Benston, G., Carmichael, D., Christensen, T., Colson, R., Moehrle, S., Rajgopal, S., Stober, T., Sunder, S., Watts, R., A perspective on the SEC s proposal to accept financial statements prepared in accordance with international financial reporting standards (IFRS) without reconciliation to U.S. GAAP. American Accounting Association s Financial Accounting Standards Committee. Accounting Horizons 22, Jamal, K., Bloomfield, R., Christensen, T., Colson, R., Moehrle, S., Ohlson, J., Penman, S., Previts, G., Stober, T., Sunder, S., Watts, R., A Research Based Perspective on SEC s Proposed Rule on Roadmap for Potential Use of Financial Statements Prepared in Accordance With International Financial Reporting Standards (IFRS) by U.S. Issuers. American Accounting Association s Financial Accounting Standards Committee. Jeanjean, T., Stolowy, H., Do accounting standards matter? An exploratory analysis of earnings management before and after IFRS adoption. Journal of Accounting and Public Policy 27, Johnson, S., The Revenue-Recognition Rules Paradox. CFO.com, February 6. Joos, P., Plesko, G., Valuing loss firms. The Accounting Review 80, Kim, J-B., Shi, H., International Financial Reporting Standards, Institutional Infrastructures and Costs of Equity Capital around the World. Working Paper, Concordia University and The Hong Kong Polytechnic University. Kim, J-B., Shi, H., International Financial Reporting Standards, Analyst Following, Institutional Infrastructure, and Stock Price Synchronicity around the World. Working Paper, The Hong Kong Polytechnic University. Kim, J., Tsui, S., Yi, C., The Voluntary Adoption of International Accounting Standards and Loan Contracting around the World. Working Paper, The Hong Kong Polytechnic University. Kinnunen, J., Niskanen, J., Kasanen, E., To whom are IAS earnings informative? Domestic versus foreign shareholders perspectives. The European Accounting Review 9, Klein, A., Marquardt, C., Fundamentals of accounting losses. The Accounting Review 81, Kothari, S.P., Ramanna, K., Skinner, D., Implications for GAAP from an Analysis of Positive Research in Accounting. Working Paper, Massachusetts Institute of Technology, Harvard University, and University of Chicago. Kvaal, E., Nobes, C., International differences in IFRS policy choice: a research note. Accounting and Business Research 40, La Porta, R., Lopez-De-Silanes, F., Shleifer, A., Vishny, R.W., Legal determinants of external finance. The Journal of Finance 52, La Porta, R., Lopez-De-Silanes, F., Shleifer, A., Vishny, R.W., Law and finance. Journal of Political Economy 106, LaPointe-Antunes, P., Cormier, D., Magnon, M., Gay-Angers, S., On the relationship between voluntary disclosure, earnings smoothing and the value-relevance of earnings, the case of Switzerland. The European Accounting Review 15, Leone, M., Audit Firm Bigs Cite Lust for Global Standards. CFO.com, January 16. Leone, M., Will a Tortoise Pace Win the Global Accounting Race? CFO.com, January 23. Leuz, C., Verrecchia, R., The economic consequences of increased disclosure. Journal of Accounting Research 38, Li, K., How well do Investors Understand Loss Persistence? Working Paper, University of Toronto. Lin, Z.L., Chen, F., Value relevance of international accounting standards harmonization: evidence from A- and B-share markets in China. Journal of International Accounting, Auditing and Taxation 14, Ndubizu, G.A., Sanchez, M.H., The valuation properties of earnings and book value prepared under US GAAP in Chile and IAS in Peru. Journal of Accounting and Public Policy 25, Niskanen, J., Kinnunen, J., Kasanen, K., The value-relevance of IAS reconciliation components: empirical evidence from Finland. Journal of Accounting and Public Policy 19, O Connell, V., Sullivan, K., The impact of mandatory conversion to IFRS on the net income of FTSEurofirst 80 firms. The Journal of Applied Research in Accounting and Finance 3, Pincus, M., Rajgopal, S., Venkatachalam, M., The accrual anomaly: international evidence. The Accounting Review 82, Platikanova, P., Market Liquidity Effects of the IFRS Introduction in Europe. Working Paper, University Pompeu Fabra. PricewaterhouseCoopers (PwC), 2007a. 10 Minutes on IFRS, October. PricewaterhouseCoopers (PwC), 2007b. IFRS, The Right Step for U.S. Business, White Paper. PwC, 2007c. Similarities and Differences: A Comparison of IFRS and U.S. GAAP, October. Reilly, D., What s better in accounting, rules or feel? The Wall Street Journal (April 30), C1. Riedl, E., An examination of long-lived asset impairments. The Accounting Review 79, Schadewitz, H., Vieru, M., How Markets Valuate and Response to IFRS Reconciliations Adjustments in Finland. Working Paper, Turku School of Economics and University of Oulu. Soderstrom, N.S., Sun, K.J., IFRS adoption and accounting quality: a review. European Accounting Review 16, Sunder, S., IFRS and the accounting consensus. Accounting Horizons 23, Tarca, A., International convergence of accounting practices: choosing between IAS and U.S. GAAP. Journal of International Financial Management and Accounting 15, Taub, S., IFRS Gets a Push from Accounting Giants. CFO.com, January 16. Turley, J., Mind the GAAP. The Wall Street Journal, November 9.

19 T.J. Atwood et al. / J. Account. Public Policy 30 (2011) Turner, L., International Accounting Standards, Opportunities, Challenges and Global Convergence. Testimony before the Subcommittee on Securities, Insurance, and Investment, Committee on Banking, Housing, and Urban Affairs, U.S. Senate, October 24. Van der Meulen, S., Gaeremynck, A., Willekens, M., Attribute differences between U.S. GAAP and IFRS earnings: an exploratory study. The International Journal of Accounting 42, Wu, J.S., Zhang, I., The voluntary adoption of internationally recognized accounting standards and firm internal performance evaluation. The Accounting Review 84,