Foreign Analyst Following and Forecast Accuracy around. Mandated IFRS Adoptions

Transcription

1 Foreign Analyst Following and Forecast Accuracy around Mandated IFRS Adoptions Hongping Tan University of Waterloo Shiheng Wang Hong Kong University of Science and Technology Michael Welker* Queen s University October, 2009 Tan can be reached at hptan@uwaterloo.ca, Wang can be reached at acwang@ust.hk, and Welker can be reached at mwelker@business.queensu.ca. We are grateful for comments from workshop participants at the University of Wisconsin, the University of Waterloo, and the Hong Kong University of Science and Technology. Welker is grateful for financial support from the KPMG fellowship at Queen s University. We are also grateful for financial support from the Social Sciences and Humanities Research Council of Canada. We are grateful to Thomson Financial for access to their Institutional Brokers Estimate System (I/B/E/S), provided as part of a broad academic program to encourage earnings expectations research. *Correspondence author: mwelker@business.queensu.ca, (613)

2 ABSTRACT This study investigates how accounting harmonization affects one particular group of financial statement users foreign financial analysts through enhanced usefulness of accounting data across countries. We utilize a unique database that identifies analysts locations to assemble a sample of 2,059 distinct foreign analysts from 36 countries covering a total of 2,748 firms that have adopted International Financial Reporting Standard (IFRS) during We find that mandatory IFRS adoption attracts foreign analysts, particularly those who are located in countries that are simultaneously adopting IFRS along with the covered firm s country. We also find that mandatory IFRS adoption improves foreign analysts forecast accuracy. Multivariate regressions show an increase of 26% in the foreign analyst following from all sample countries, 37% for foreign analysts from simultaneously adopting countries, and 46% for foreign analysts from simultaneously adopting countries who have prior experience following firms using IFRS. There is an improvement of around 15% in forecast accuracy from the pre- to the post-ifrs period across all three of these analyst groups. Further analysis shows that the gains in analyst following are greater for firms located in countries that had local GAAP that was more different from IFRS prior to IFRS adoption, and that the change in foreign analyst following is increasing in the extent to which IFRS adoption eliminates GAAP differences between the firm s home country and the analyst s home country. These measures of the extent to which IFRS adoption eliminates GAAP differences are unrelated to increases in foreign analysts forecast accuracy. These results suggest that accounting harmonization in the form of widespread IFRS adoption facilitates cross-border comparisons of financial data and therefore lowers the costs for financial analysts to follow firms from other countries. 2

3 1. Introduction This study investigates how the recent mandatory adoption of International Financial Reporting Standards (IFRS) affects foreign financial analysts. We are interested in foreign analysts since one of the most important motivations for accounting harmonization is to enhance the usefulness of accounting information across countries. Analysts are among the most important and sophisticated users of financial statements, so an examination of how harmonization affects analysts, especially foreign analysts who are most likely to be impacted by the existence of different accounting standards across countries, adds to the literature documenting other effects of harmonization. In addition, analysts forecast earnings and other accounting numbers produced by GAAP, so an examination of analysts forecasts allows an assessment of how accounting standard harmonization affects financial statement users forecasting accuracy. To the extent that widespread IFRS adoption reduces accounting standard induced differences in financial reporting across countries, it should facilitate cross-border comparisons of financial data. This would allow foreign analysts to cover more firms from different countries that have mandated IFRS reporting. The foreign analysts should also be more accurate in their earnings forecasts if accounting harmonization eliminates forecast errors caused by differences in accounting standards. While there have been several studies that examine various financial accounting outcomes surrounding IFRS adoption, we are not aware of any study that has directly examined how accounting harmonization affects foreign financial analysts. Bae, Tan and Welker (2008) (hereafter referred to as BTW ) present evidence suggesting that differences in generally accepted accounting principles (GAAP) across countries impose costs on financial analysts that dissuade them from following firms in other countries with dissimiliar GAAP. In addition, they provide some evidence that analysts who do choose to follow firms from countries with dissimiliar GAAP forecast those firms less accurately than other analysts following those firms. However, it is not clear that accounting harmonization achieved through IFRS adoption will result in a meaningful reduction in observed differences in accounting practice across countries since accounting practices are affected by many factors beyond accounting standards. Hence, it is inappropriate to conclude that mandated IFRS adoption will necessarily attenuate the effects 3

4 documented in BTW, nor is it clear how big an effect IFRS adoption will have if it has any effect. Our study extends their analysis to examine whether convergence in accounting standards decreases these costs and encourages analysts to follow more firms in IFRS adopting countries and forecast earnings for those companies more accurately. We assume that analysts coverage decisions are based on cost-benefit analysis. Accounting harmonization can reduce information acquisition and processing costs for analysts so that the costs of learning a new set of accounting standards is not an impediment to following foreign firms. In addition to reducing information acquisition and processing costs, accounting harmonization can reduce the reputational costs for foreign analysts if they can do a better job of forecasting and providing recommendations for foreign firms due to reduced information asymmetry between the firm and foreign analysts. The benefits may come from increased commissions for the analysts brokerage firms that accompany the expanded coverage and investment recommendations, or investment banking fees. Further, since IFRS requires more comprehensive and timely information than local standards in many cases, foreign analysts could improve their forecasting performance due to reduced information asymmetry between the firm and foreign analysts. We find that IFRS adoption attracts foreign analysts, particularly those who are located in a country that is adopting IFRS at the same time as the firm s country. Prior IFRS experience among analysts located in countries simultaneously adopting IFRS does not appear to significantly influence analyst following decisions. Firms located in countries where local GAAP differed more from IFRS prior to IFRS adoption gain more foreign analysts than do firms located in countries with local GAAP more similar to IFRS. We also find that foreign analyst following increases more when IFRS adoption causes a greater reduction of GAAP differences between analyst home countries and firm home countries. We find that IFRS adoption improves foreign analysts forecast accuracy. The improvements in forecast accuracy occur for most analyst groups we examine and are generally unrelated to our measures of how IFRS adoption affects GAAP differences from IFRS or from other sample countries. These results are 4

5 consistent with the argument that accounting harmonization in the form of widespread IFRS adoption enhances the usefulness of accounting data across borders. The rest of the paper is organized as follows. In the next section we discuss how our study relates to the existing literature. Section 3 presents our sample selection procedures and descriptive statistics. Section 4 describes our research design. Section 5 presents empirical evidence regarding the impact of IFRS adoption on foreign analyst following and earnings forecast accuracy. Section 6 concludes. 2. Related Literature and Motivation The development of a set of globally accepted accounting standards formally began in 1973 with the establishment of the International Accounting Standards Committee (IASC), which aimed to improve and harmonize financial reporting around the world. In April 2001, the International Accounting Standards Board (IASB) succeeded the IASC. This change in structure of the standard setting body was one of many factors in the early 2000s that contributed to an acceleration of IFRS adoption internationally. As of October 2009 there are 117 jurisdictions around the globe that permit or require IFRS for their domestically listed companies 1. The desirability of mandating IFRS adoption has provoked heated debate. Proponents of IFRS adoption argue that a single global set of standards facilitates cross-border comparisons of financial data, allowing investors and other market participants to better use their current expertise to analyze firms from other countries. Furthermore, the more comprehensive disclosure requirements under IFRS relative to some domestic accounting standards make earnings easier to understand and predict, which should improve average analyst forecast accuracy (Ball, 2006). Ernst and Young s survey of the first annual reports prepared under IFRS by major European firms finds that annual report length has increased on average by 20-30%. The survey attributes this to the fact that IFRS disclosure requirements exceed those of most European local GAAP. The survey also finds that while the implementation of IFRS has 1 The website of IAS plus provides a list of jurisdictions where the basis of preparation note and the auditor's report refers to conformity with IFRS. Please refer to 5

6 brought about significantly greater consistency in accounting recognition and measurement IFRS financial statements retain a strong national identity. On the other hand, opponents of accounting harmonization argue that the unique accounting problems, history, culture and institutional frameworks in each country determine the optimal form and content of accounting standards. Hence, an appropriate set of accounting standards in one country is not necessarily an appropriate set of accounting standards in another country. Mandating IFRS adoption also involves transition costs for firms complying with the new standards. In the US, the SEC has estimated this conversion cost at around 0.125% to 0.130% of annual revenue per company. Although the extensive fair value accounting rules under IFRS could possibly incorporate more timely information about economic gains and losses, opponents criticize IFRS for allowing too much judgment in fair value measurements. This managerial subjectivity in the absence of stringent and consistent enforcement may affect the transparency and comparability of reported accounting numbers across firms, industries and countries. If so, then mandating the use of a common set of accounting standards across countries will not achieve accounting harmonization in practice. Our study is closely related to the past literature that examines the impact of accounting standards or their implementation on financial analysts and investors internationally. Based on a sample of 80 non-u.s. firms that voluntarily adopted International Accounting Standards (IAS) in 1993, Ashbaugh and Pincus (2001) show that the extent of difference between local accounting standards and IFRS is negatively associated with analysts forecast accuracy prior to IFRS adoption. Further, they find that forecast accuracy improves after adopting IFRS. Hope (2003a) documents that analysts forecast accuracy is better for firms that disclose their accounting method choice. Hope (2003b) further documents that analyst earnings forecast accuracy is positively associated with legal enforcement in a country. Basu, Hwang and Jan (1998) show that forecast accuracy is lower in countries with less accrual basis accounting, more market value based accounting and fewer accounting choices. Guan, Hope and Kang (2006) find that analysts provide more accurate forecasts for firms from countries with accounting standards that are more similar to US accounting standards, and this effect is stronger in weak disclosure countries and for firms 6

7 followed by few analysts. BTW find that differences in accounting standards dissuade analysts from following firms from foreign countries and impede their ability to accurately forecast earnings for the foreign firms they do follow. There are some studies that examine changes in analyst following or earnings forecast accuracy around the mandatory IFRS adoption in the EU. Horton, Serafeim and Serafeim (2008) document that both voluntary and mandatory adopters experience improved earnings forecast accuracy, decreased analyst forecast dispersion and decreased volatility in earnings forecast revisions. Wang, Young and Zhuang (2008) also find that earnings forecast errors and earnings forecast dispersion decrease after mandatory adoption dates for both voluntary and mandatory adopters in 17 EU countries. The effects, however, do not differ between the two groups of firms. They also show that stock return variance surrounding earnings announcements increases for both sets of firms after mandatory IFRS adoption dates, suggesting IFRS-based earnings are more informative than local-gaap based earnings. Byard, Li and Yu (2008) find that the decrease in earnings forecast errors is greater for firms located in countries where domestic GAAP diverges relatively more from IFRS, and for firms located in countries with better legal enforcement. Unlike the above studies that examine overall analysts earnings forecast characteristics, our study focuses specifically on the effects of IFRS adoption on foreign analysts and we examine both analysts decisions to follow foreign firms and their earnings forecast accuracy. Our study also complements recent research that investigates the relation between firms choice of accounting standards and foreign investor interest, including international investment allocation decisions. Bradshaw, Bushee and Miller (2004) find that conformity to US GAAP helps foreign firms attract US institutional investment since US GAAP is more familiar to these US investors, potentially reducing information processing costs and providing them with what is perceived to be higher quality information. Covrig, DeFond and Hung (2007) examine the holdings of over 25,000 mutual funds from around the world and find that foreign mutual fund holdings are higher for firms adopting IFRS, suggesting that familiar or better GAAP increases foreign investor interest. Davis-Friday and Skaife (2008) find that firms listed on the Stock Exchange Automated Quotation System (SEAQ) International in London that 7

8 adopt IFRS or U.S. GAAP are more likely to become targets in mergers and acquisitions. One interpretation of these results is that higher quality or more familiar financial reporting allows outsiders to better identify takeover targets, leading to more acquisitions of firms using IFRS or U.S. GAAP. Several recent studies examine the impact of mandatory IFRS adoption on foreign investors decisions. Florou and Pope (2009) find that institutional ownership increases by 3.3% and the number of institutional investors increases by around 10 one year after mandatory IFRS adoption. Yu (2009) documents that foreign mutual funds increase their holdings by 1.8% of total outstanding shares for firsttime mandatory IFRS adopters relative to firms that were not required to adopt IFRS. Moreover, the effects are more pronounced when IFRS adoption causes a greater reduction of accounting standard differences between the funds home country and the firms home country, or when the funds home country and the investee firms home country adopt IFRs at the same time. Based on 53 countries portfolio investment in up to 81 foreign countries and territories for the year 1997 and the years from , Amiram (2009) finds that foreign investors have statistically and economically higher holdings of foreign equity portfolio investments in countries that use IFRS. This relationship is stronger if both the investing country and investee country use IFRS at the same time and for investee countries with low corruption and better property rights. Time-series analysis suggests that foreign investment increases after IFRS adoption. Brüggemann et al. (2009) find that individual investors participating in the Open Market at the Frankfurt Stock Exchange increase trading in the adopting companies stock following mandatory IFRS adoption. Using two measures for the IFRS adoption induced convergence in accounting standards within industries, DeFond, Hu, Hung and Li (2009) document that US mutual funds increase their equity holdings of both early voluntary and mandatory IFRS adopters from 14 EU countries around mandatory adoption dates only for industries or industry-country groups with larger improvements in comparability. In general, this stream of studies suggests that accounting standard differences across borders impose economic costs on investors, while convergence in accounting standards potentially reduces such costs. 8

9 Our study also adds to an emerging literature that provides early evidence of the economic consequences of mandating IFRS around the world. Platikanova (2007) analyzes liquidity and information asymmetry for four European countries. She finds heterogeneous liquidity changes for these countries, but the liquidity differences between countries become smaller after the adoption of IFRS. Daske, Hail, Leuz and Verdi (2008) find that market liquidity increases and cost of capital decreases around the time when mandatory IFRS adopters release their first IFRS annual report. However, they find that the capital market benefits occur only in countries where firms have incentives to be transparent and where legal enforcement is strong. The capital market effects are most pronounced for firms that voluntarily switched to IFRS prior to the mandatory adoption date, suggesting that enhanced comparability with their country peers benefits early adopting firms. We utilize this insight in our research design to test for effects not only for mandatory adopters but also for early adopters. We extend this literature by providing time-series evidence on how mandatory IFRS adoption affects foreign analysts decisions to follow both prior voluntary and first-time mandatory adopters. While BTW show that GAAP differences across countries impose costs that dissuade analysts from following foreign firms, we are aware of no evidence that suggests harmonization in the form of widespread mandatory adoption reduces such costs and encourages more foreign analysts to follow IFRS adopting firms and/or to forecast the foreign firms more accurately when they do follow. As discussed earlier, many argue that inconsistent application of IFRS will reduce or eliminate any comparability benefits that could accompany accounting standard harmonization. The Ernst and Young s survey of IFRS financial statements finds both enhanced comparability but continued national identities in mandatory adopters financial statements, so the extent to which costs imposed on foreign analysts are reduced by harmonization remains an empirical issue. We do not expect the effects of IFRS adoption to be similar for all adopting firms, nor do we expect the effects to be similar across all firm-analyst pairs. We hypothesize that IFRS adoption will reduce information acquisition and processing costs and reputational costs to a greater extent for foreign analysts from countries that adopt IFRS at the same time as the home countries of covered firms, and for foreign 9

10 analysts who have already covered firms that voluntarily adopted IFRS before mandatory IFRS reporting. When harmonization causes a greater reduction in GAAP differences it is expected to cause a more dramatic change in the costs of covering foreign firms. We therefore predict that firms in countries where local GAAP differs more from IFRS prior to IFRS adoption will receive greater benefits than firms from countries where local GAAP was already close to IFRS, increasing their foreign analyst following. We also predict that the extent to which IFRS adoption reduces GAAP differences between the analyst home country and the firm home country will be positively associated with changes in the number of analysts from that foreign country. Because lowering costs to analysts is also expected to improve their forecast accuracy, we make similar predictions about foreign analysts forecast accuracy. However, we acknowledge that it is possible that the decision to follow a firm indicates that the benefits of following the firm outweigh the costs of learning the effects of different GAAP on the accounting representation of firm performance. If so, analysts who choose to follow foreign firms using local GAAP might incur those costs and successfully understand the effects of GAAP differences such that eliminating GAAP differences through convergence would have negligible effects on forecast accuracy. Our main hypotheses are stated below in alternative form. Hypothesis 1: IFRS adoption is associated with increased coverage by foreign analysts and an improvement in foreign analysts earnings forecast accuracy. Hypothesis 2: Increases in foreign analyst following and improvements in their forecast accuracy following IFRS adoption are more pronounced for analysts located in countries that adopt IFRS at the same time as the firm s home country and for those with previous IFRS experience. Hypothesis 3: Increases in foreign analyst following and improvements in their forecast accuracy following IFRS adoption are more pronounced for firms from countries with local GAAP that differed more from IFRS prior to IFRS adoption, and are positively associated with the extent to which GAAP differences between the analyst home country and firm home country are reduced by IFRS adoption. 10

11 3. Data and Sample Selection 3.1 Analyst Location We use the annual volumes of Nelson s Directory of Investment Research for to identify the locations of financial analysts. These Nelson s Directories provide information on nearly 1,700 research firms with approximately 40,000 equity analysts covering publicly traded companies located around the world. We obtain the full names of equity analysts and the research firms they were associated with from these annual volumes of Nelson s Directories and follow the same procedure as BTW to match them to the analysts in the I/B/E/S database. We are able to identify the country location for nearly 66 percent of all analysts included in the I/B/E/S database for our sample firms. We are interested in identifying analyst location because analysts' locations are expected to correlate with their GAAP knowledge and familiarity. Implicit in our hypothesis development is the assumption that analysts locations are stable that analysts by and large work in the country where they have been trained and educated such that they are most familiar with the GAAP in the country where they are currently located. Of course, like any profession, there is some international mobility among financial analysts. Consistent with BTW, less than six percent of analysts with multiple years of location data relocate to another country, suggesting that while cross-border relocations by financial analysts do occur, they are not common. 3.2 Identification of Sample Firms We obtain historical information on the use of accounting standards during 1988 to 2007 for a total of 21,723 firms based on the item Accounting standards followed in the Worldscope database. We restrict our sample firms to 25 countries where we can identify the date that a country announced its intention to mandate IFRS and the date IFRS reporting became compulsory in the country. Following the definition of IFRS adopters in Daske, Hail, Leuz and Verdi (2007), we identify 6,859 firms that have used IFRS. Appendix A shows the specific Worldscope accounting standard codes that we classify as indicating IFRS use. We exclude the following dubious IFRS users: (i) 54 firms that adopted IFRS before year 2005 but used IFRS for only one year; (ii) 40 firms that switched from IFRS to local standards; and (iii) 16 firms 11

12 that switched from IFRS to US GAAP and 12 firms that switched between IFRS and US GAAP from time to time. This yields a total of 6,737 firms that have adopted IFRS during the period of We follow the same procedure to exclude 255 dubious IFRS users for the full sample of 21,732 firms in the Worldscope universe so that we can merge with the I/B/E/S database to evaluate analysts experience with the use of IFRS based on the full set of firms they have covered in the past. We require the firm have total asset data available and be followed by at least two analysts covered in I/B/E/S for at least one year during the two years immediately prior to the IFRS mandatory adoption year and for at least another year during the two years after IFRS adoption 2. The two analysts need not be foreign analysts. We lose 3,989 firms due to missing values in I/B/E/S, of which 2,840 firms do not have required data in any year of the sample period and 1,149 firms have required data in either the pre-ifrs or post-ifrs period, but not both. The full sample comprises 10,068 firm-years, covering 2,748 unique firms from 25 countries and 2,051 unique foreign analysts from 36 countries. The sample size varies across different empirical tests, depending on further data restrictions. Appendix B shows our sample selection criteria. Since Daske et al. (2008) find that voluntary adopting firms enjoy economic benefits at the time that the first mandatory IFRS annual reports are disclosed in their country, we follow their empirical design to isolate voluntary adopters to see if they also benefit from greater foreign analyst following and forecast accuracy due to mandatory IFRS reporting in their countries. We classify sample firms into Mandatory or Voluntary adopters depending on their first IFRS adoption dates. Mandatory adopters are those firms that switched to IFRS for fiscal years beginning on or after the earliest mandatory IFRS application date for their countries. Voluntary adopters are firms that applied IFRS before this date. We further classify voluntary adopters into two groups. Early Voluntary indicates those firms that voluntarily switched to IFRS prior to the announcement by local authorities that they intended to require IFRS reporting for their domestic firms. Late Voluntary indicates those firms that voluntarily switched to IFRS after local 2 Our inferences remain qualitatively the same if we drop the requirement of at least two analysts for any firm-year. In the multivariate regression we substitute for the variable diversity with stock return volatility to proxy for analyst forecast difficulty. 12

13 authorities announced their plans to mandate IFRS reporting. Table 1 presents the IFRS adoption announcement and mandatory adoption dates together with the progress of IFRS adoption over time for each sample country. All the European countries in our sample announced on June 4, 2002 their plans to mandate IFRS for fiscal years beginning on or after January 1, Singapore announced its plan on December 7, 2000 and mandated IFRS for fiscal years ending on or after December 31, South Africa, the Philippines and Hong Kong announced their plans after the EU announcement but mandated IFRS compliance commencing on the same date as the European countries. The total number of sample firms that have adopted IFRS ranges from five in the Czech Republic to 578 in the U.K., with Australia, France, Germany and Hong Kong each having more than 200 IFRS adopters. We report the progress of IFRS adoption for each country under the headings voluntary and mandatory. We further classify mandatory IFRS adopters into three groups, depending on whether the fiscal year end of the initial IFRS adoption falls on December 31, 2005, on any date in 2006, or any date in There are 363 early voluntary adopters, 154 late voluntary adopters and 2,231 mandatory adopters in our final sample. Consistent with Barth, Landsman, and Lang (2008) and Daske, Hail, Leuz and Verdi (2007), Germany, Italy and Switzerland have the largest number of voluntary IFRS adopters in our sample (including both the early voluntary and late voluntary adopters) 3. Sixty-three percent of mandatory adopters started to use IFRS for fiscal year ending on December 31, Twenty-nine percent of mandatory adopters initiated IFRS sometime during 2006, the majority of which are in Australia and the U.K. as the fiscal year end for a large number of firms in these two countries falls on March 31 or June 30. Ninety-nine Singaporean firms had adopted IFRS as of December 31, The majority of the firms adopting IFRS in 2007 are from the UK where firms listed on the Alternative Investment Market (AIM) of the London Stock Exchange are not subject to the EU IFRS Regulation. The AIM adopted a rule requiring companies to submit IFRS financial statements starting in Other 3 Following Daske et al. (2007, 2008) definition of IFRS adopters, 100 Italian firms that followed local standards with some IASC guidelines before IFRS was mandated are classified as voluntary IFRS adopters. Our conclusions remain the same if we reclassify these firms as mandatory IFRS adopters. 13

14 countries such as Norway and Germany had rules in place that allowed some firms to adopt in 2007 under specific circumstances. 3.3 Identification of sample analysts After identifying the IFRS adopting firms for our sample, we then identify the analysts from the 36 countries that are home to at least one analyst following at least one sample firm. New analysts enter the profession and existing analysts exit the profession from time to time. An analyst may stop covering a firm not because of the firm's decision to switch to IFRS but because the analyst has left the profession. Alternatively an analyst may cover firms only in the post-ifrs period simply because the analyst had not started her career prior to the IFRS adoption date of the covered firms. To reduce the impact of analyst entry and exit on our measures of analyst following around IFRS adoption we impose the following restriction. We find the first and last forecast date in the I/B/E/S universe for each of our sample analysts. For any particular firm that has mandatorily adopted IFRS we keep only those foreign analysts who are active in the I/B/E/S database from before the firm s mandatory IFRS adoption date (the beginning of the firm s first IFRS fiscal year) to after the end of the firm s first IFRS fiscal year. For example, if a firm adopted IFRS for the fiscal year ending December 31, 2005, we require the analysts that cover this firm be active in I/B/E/S from before December 31, 2004 to after December 31, In this way, we ensure that the analysts are active in the I/B/E/S database during both the pre-ifrs period and post-ifrs period. 4 For firms voluntarily adopting IFRS, we keep those foreign analysts who are active in the I/B/E/S database at least one year before the IFRS adoption implementation date in the firm s country and continue to be active in I/B/E/S database after the end of the first IFRS fiscal year in the firm s country. For example, if a country mandated IFRS for fiscal years ending on December 31, 2005, but a local firm voluntarily adopted IFRS beginning with the fiscal year ending on December 31, 2002, we require analysts be active in I/B/E/S before December 31, 2004 and continue to be active in the I/B/E/S database until after December 31, This reflects the fact that our time-series tests focus on the period around 4 Alternatively we require foreign analysts be present in I/B/E/S from two years before to two years after the firm s first IFRS fiscal year. The results under this alternative specification (reported in Table 8) support the same general inferences. 14

15 mandatory IFRS adoption, even for early adopters. Defining a variable to capture analysts' IFRS experience is complicated by the fact that firms adopt IFRS at various dates before and during our sample period. Clearly the number of analysts without IFRS experience would decline dramatically for firm-years ending after December 31, 2005 if this variable were to be coded on an annual basis. To avoid this problem, we code IFRS experience (which is coded for each firm-analyst pair) as follows. First, analysts who have never followed a firm using IFRS for a fiscal year ending prior to December 31, 2005 (or December 31, 2003 for Singaporean firms) are treated as analysts with no IFRS experience throughout our sample period. Analysts who have followed a firm using IFRS for a fiscal period prior to December 31, 2002 (or December 31, 2000 for Singaporean firms) are coded as analysts with IFRS experience throughout our sample period. In some cases analysts begin following an IFRS firm with a fiscal year ending in (or for Singaporean firms), and this is the first time this analyst follows an IFRS firm. With respect to this initial firm, the analyst is treated as an analyst with no IFRS experience throughout our sample period. If this analyst subsequently begins following any other IFRS firm in our sample period the analyst is then coded as having IFRS experience with respect to that firm thereafter. For the foreign analysts in our sample, we keep the last annual earnings forecast in I/B/E/S for each of the two fiscal years immediately prior to a firm s IFRS adoption date and for each of the two fiscal years immediately after the adoption date. We then count the number of unique foreign analysts that follow each firm-year. If there is no identified foreign analyst covering a firm-year, we set the number of foreign analyst following the firm to zero that year. Table 3 shows the distribution of sample firms, analysts, and firm size by country. Column (1) is the number of unique sample firms located in each country. Column (2) (4) report for the sample firms within the sample period the total number of analysts in I/B/E/S, the number of unique analysts in I/B/E/S for whom we could identify their country location using Nelson s Directory of Investment Research, and 15

16 the corresponding matching rate 5. Column (5) shows the number of sample analysts who follow the sample firms in each country. France attracts the largest number of foreign analysts. Fifteen countries have one hundred or more foreign analysts covering firms in their markets, and nineteen countries have at least sixty foreign analysts. Column (6) shows the number of analysts located in each of the sample countries who follow firms from other sample countries 6. The UK is home to the largest number of sample foreign analysts (910), followed by France and Germany, which house 272 and 197 sample foreign analysts, respectively. Greece and Luxembourg have no analyst covering sample firms in other markets. Column (7) shows the total number of foreign analysts earnings forecasts for firms located in each country. Eight countries have over 1,000 forecasts and twenty-one have more than 100 forecasts. Column (8) presents the mean value of firm size in the sample countries. The covered firms are smallest in the Philippines and largest in Ireland, with average total assets of U.S. $2.71 and U.S. $26.29 billion, respectively. Table 2 lists the definitions for all the variables used in this study. 4. Research Design 4.1 Analyst Following The main dependent variables of this study are foreign analyst following and foreign analysts earnings forecast accuracy surrounding each firm's mandatory IFRS adoption. To test our hypotheses, we divide foreign analysts into five different categories: (i) all, foreign analysts from all of the 36 countries that are home to cover sample firms; (ii) ctry25, foreign analysts who are located in the 25 sample countries that mandated IFRS adoption; (iii) nctry25, foreign analysts who are not located in the 25 sample countries that mandated IFRS adoption (iv) ifrsexp, foreign analysts from the 25 sample countries that mandated IFRS adoption with prior IFRS experience as defined earlier; and (v) nifrsexp, foreign analysts from the 25 sample countries that mandated IFRS adoption who do not have prior IFRS 5 As a sensitivity check on the analysts for whom we cannot identify country location, we mechanically assign them a country location based on where the majority of the firms they follow each year are located as long as these analysts cover at least a total of five firms in I/B/E/S. This procedure increases our analyst matching rate to around 81%, but only adds an additional 150 foreign analysts to our final sample since the majority of these mechanically identified analysts follow only local firms. We rerun our regressions in Table 5 and all of our results remain qualitatively unchanged. 6 There are 111 foreign analysts located in 13 other countries, most of whom are located in the USA. 16

17 experience as defined earlier. As the descriptive statistics discussed below indicate, the sample of analysts following sample firms who are not from the 25 countries mandating IFRS adoption is too small (averaging only around 1 analyst for every 10 sample firms) to allow further partitioning into experienced and inexperienced groups with significant sample sizes. 7 To test hypotheses 1 and 2, we compute the total number of foreign analysts from each of the five analyst categories who cover firm i in year t. To test hypothesis 3 regarding the association between foreign analyst following and the extent to which harmonization reduces accounting differences, we first compute the total number of foreign analysts from country j who cover firm i in year t. We then compute the average annual number of foreign analysts from country j who cover firm i during the two year pre- IFRS period and two year post-ifrs period, respectively. Finally, we use the difference between the post- IFRS and pre-ifrs periods to measure changes in the average number of foreign analysts from country j who cover firm i after the firm's IFRS adoption. We also develop an alternative measure for hypothesis 3. We count the number of unique foreign analysts from country j who cover firm i during the pre-ifrs and post-ifrs periods, respectively. We then use the difference in the number of unique foreign analysts between the two periods as an alternative measure for the changes in the number of foreign analysts from country j who cover firm i around the firm's IFRS adoption. Our first measure assigns a value of 0.5 to an analyst who covers an IFRS adopting firm in only one of the two pre or post IFRS years, while the second measure assigns a value of 1 to this analyst. Table 4 Panel A shows the average number of foreign analyst following sample firms for each year around the mandatory IFRS adoption date. The average number of foreign analysts from all of the 36 analyst countries, from the 25 countries that mandate IFRS adoption and those with prior IFRS experience have all increased from two years before the IFRS adoption date to two years after the IFRS 7 Despite these concerns, we did examine the samples of analysts not from the 25 countries that mandate IFRS adoption with and without IFRS experience. The results indicate a significant increase in coverage from these analysts with experience, and a significant decline in coverage for these analysts without experience. Neither group experiences a change in forecast accuracy, though there are only367 total (combined pre and post periods) forecasts from experienced analysts and 460 forecasts from inexperienced analysts located outside the 25 IFRS adopting countries. 17

18 adoption date. For example, the average number of foreign analysts from all of the 36 analyst countries has increased from in year -2 to the year after the IFRS adoption year. In contrast, the number of foreign analysts without prior IFRS experience has declined from year -1 to the year IFRS is adopted for the first time. 8 Table 4 Panel B compares the difference in analyst following from each group of foreign analysts between the pre-ifrs and post-ifrs periods. For mandatory adopters, the pre-ifrs period refers to the two years before the firm first uses IFRS and the post-ifrs period refers to the two years after the firm first uses IFRS. For early voluntary and late voluntary adopters, the pre-ifrs period refers to the two years before IFRS use becomes mandatory in the firm s home country and the post-ifrs period refers to the two years after IFRS becomes mandatory in the firm s home country. The mean difference tests suggest that the number of foreign analysts from all of the 36 countries, foreign analysts from the 25 countries that mandate IFRS adoption and foreign analysts from these countries with prior IFRS experience increase significantly after IFRS adoption, while the number of foreign analysts from the 25 countries without prior IFRS experience decreases significantly following IFRS adoption. We also show the changes of control variables used in the multivariate regressions. Average firm size has increased from the pre-ifrs period to the post-ifrs period, while earnings forecast difficulty (diversity) declines after IFRS adoption. 4.2 Forecast Accuracy As explained earlier, we keep only those foreign analysts who are present in the I/B/E/S database at least one year before the firm s IFRS adoption date and continue to be present in the I/B/E/S after this adoption date. In addition, we require that a firm have at least one foreign analyst forecast observation during the two years immediately before the firm s IFRS adoption date (i.e., [-2,-1]) and at least one foreign analyst forecast observation during the two years immediately after the IFRS adoption date (i.e., [0,1]). Both of these restrictions are intended to ensure that any changes in forecast accuracy surrounding 8 Recall that experience is coded only once for each firm-analyst pair, so changes in this variable over time are due to changes in the analysts who follow the firm over time, not due to changes in IFRS experience that would occur mechanically over time for our sample analysts. 18

19 IFRS adoption are not due to changes in our sample of firms or analysts between the pre and post IFRS periods. We obtain from I/B/E/S the last annual earnings forecast by foreign analyst j before the annual earnings announcement date for firm i in year t. We then deflate the absolute differences between the earnings forecast and actual earnings by the latest available stock price in the previous year from the I/B/E/S summary file. To facilitate exposition we multiply this variable by -100 so that greater values indicate more accurate earnings forecasts. To test hypothesis 3 regarding the impact of changes in country-pair GAAP differences between the pre and post IFRS periods on forecast accuracy, we require that foreign analysts cover the same firm in both the pre-ifrs and post-ifrs periods. This allows us to compute changes in their earnings forecast accuracy around IFRS adoption. For a given firm we first compute the average earnings forecast accuracy for each foreign analyst in the pre-ifrs and post-ifrs periods, respectively. We then compute the difference between the post-ifrs and pre-ifrs periods to measure the change in earnings forecast accuracy around the firm's IFRS adoption. Alternatively, we require a firm be covered by foreign analysts from a particular country in both the pre-ifrs and post- IFRS period, but not necessarily covered by the same analyst from that foreign country. For a given firm we first compute the average earnings forecast accuracy for all the foreign analysts from a particular country in the pre-ifrs period and post-ifrs periods, respectively. We then compute the difference between the post-ifrs and pre-ifrs periods to measure the changes in earnings forecast accuracy by foreign analysts from the particular country around the firm s IFRS adoption. Table 4 Panel C reports the mean and median foreign analysts earnings forecast accuracy for 976 distinct firms for each of the four years surrounding the IFRS adoption date by five analyst groups. We find that forecast accuracy for all of the foreign analyst groups improves from two years before IFRS adoption to two years after IFRS adoption. The biggest improvement occurs from one year before IFRS adoption to the adoption year. For example, the forecast accuracy of foreign analysts from all of the 36 analyst countries has increased from in year -1 to in the IFRS adoption year. 19

20 Panel D compares the difference in earnings forecast accuracy and other analyst/forecast characteristics between the pre-ifrs and post-ifrs periods. Significant changes in both the mean and median show that earnings forecast accuracy improves substantially after IFRS adoption. There is also an overall increase in all the control variables. 4.3 Changes in GAAP caused by IFRS adoption BTW developed two measures of differences in accounting standards across countries based on GAAP 2001: A Survey of National Accounting Rules Benchmarked against International Accounting Standards (IFAD 2001). In this survey, partners in large accounting firms from more than 60 countries benchmarked the local accounting standards in their countries against IAS, focusing their attention on rules in place as of December 31, BTW s first measure of GAAP differences across countries (gaapdiff1) is based on a list of 21 key accounting items. The 21 items are selected from the past literature and are items that 1) display substantial variation across countries, and 2) are accounting issues where countries not following IAS follow similar methods. From this list of 21 key accounting items, they construct a measure of GAAP differences for each country-pair in their sample as follows. First, for each accounting item in the list, countries that do not conform to IAS receive a score of one for that accounting item. All other countries receive a score of zero for that item. Then, for each item in this 21-item list, they compare each pair of countries. If both countries conform to IAS (i.e., each has a score of zero) for that item, or if both countries differ from IAS (i.e., each has a score of one) for that item, then that pair of countries is deemed to have similar GAAP for that item and the country-pair is assigned a GAAP difference score of zero for that item. If one country conforms to IAS (i.e., a score of zero for that item) and the other country does not (i.e., a score of one for that item), then the two countries are deemed to have different GAAP for that item and the country-pair is assigned a GAAP difference score of one for that item. This procedure is repeated for all 21 accounting items on the list and the total GAAP difference score is simply the sum of the scores for that country-pair across all 21 items. For their second measure (gaapdiff2), BTW use an algorithm based solely on the IAS section 20

21 numbers reported in the GAAP 2001 survey results to (1) identify the important accounting practices to include as individual items in the list, and (2) to code cases where local GAAP does not conform to IAS. For this measure, 52 IAS sections were identified as important and a coding of 1 again indicates nonconformity with IAS for each country that includes that particular IAS section number in the country report. The construction of the country-pair GAAP differences then mirrors the procedures followed for gaapdiff1. 9 We use the BTW measures to construct two related but distinct measures of the extent to which IFRS adoption eliminates accounting standard differences. First, we use the BTW measure of the number of differences between local GAAP and IFRS as a measure of the extent to which the mandatory adoption of IFRS produces more comparable reporting (relative to other IFRS firms internationally) for firms in each country. Since IFRS adoption results in zero differences between local GAAP and IFRS, the number of differences between local GAAP and IFRS prior to IFRS adoption serves as a measure of the number of accounting differences that are eliminated by IFRS adoption in each country. We label these two measures ifrsdiff1 for the measure based on BTW s gaapdiff1 and ifrsdiff2 for the measure based on BTW s gaapdiff2. We predict that firms from countries where a greater number of accounting differences are eliminated by IFRS adoption benefit more from the adoption since these firms overcome greater GAAP differences (i.e., the increase in foreign analyst following and foreign analyst forecast accuracy is increasing in ifrsdiff1 and ifrsdiff2). This prediction applies to all analysts groups we examine, and we expect the greater benefits of IFRS adoption due to greater reduction of differences from IFRS to be manifested in both greater foreign analyst following and greater forecast accuracy among foreign analysts. We also develop an additional measure that captures the extent to which IFRS adoption eliminates (or possibly creates) GAAP differences between the firm s home country and the analyst s home country. If the analyst s home country and the home country of the covered firm adopt IFRS simultaneously, we assume that both measures of GAAP differences for this country-pair decrease to zero after IFRS 9 A detailed list of the 21 accounting items and the differences between local GAAP and IFRS by country used in gaapdiff1 and a list of the 52 accounting items used in gaapdiff2 are provided as appendices in BTW. 21