MACRO CORPORATE GOVERNANCE FACTORS AND THE INFORMATIVENESS OF ACCOUNTING EARNINGS. Juana Aledo Martinez Complutense University of Madrid

Transcription

1 MACRO CORPORATE GOVERNANCE FACTORS AND THE INFORMATIVENESS OF ACCOUNTING EARNINGS Juana Aledo Martinez Complutense University of Madrid David Hillier University of Strathclyde Abstract March 2011 Despite the widespread assumption that International Financial Reporting Standards (IFRS) are expected to contribute to the efficient functioning of the capital markets and achieve a high degree of transparency and comparability of financial accounts (Regulation (EC) No 1606/2002 on the application of IFRS at the EU level), prior research on the economics effects of IFRS is mixed. This paper investigates the effects of the IFRS adoption on the informativeness of accounting earnings, measured by the earnings-return relation, and examines the influence of country level institutional factors in this relation across 14 Member States from 1999 to We find that companies exhibit higher earnings-return sensitivity during the pre-ifrs adoption period than do companies adopting IFRS on or after It proxies for an incentive to improve earnings quality. In addition, results are consistent with the rationale that earnings quality increases in strong governance environments (characterized by strong investor protection and corporate control mechanisms, and developed financial systems) and this effect is strengthened by IFRS. Key words: International Financial Reporting Standards (IFRS), Earnings quality, Country level institutional factors, Panel data. 1

2 1. INTRODUCTION Despite the widespread assumption that International Financial Reporting Standards (IFRS) 1 are expected to contribute to the efficient functioning of the capital markets and achieve a high degree of transparency and comparability of financial accounts (Regulation (EC) No 1606/2002 on the application of international accounting standards at the EU level), prior research on the economics effects of IFRS is mixed (Bartov et al. 2005; Hung and Subramanyam, 2007; Christensen et al. 2008; Paananen and Lin, 2009). Furthermore, preceding cross-countries studies do not deeply address whether markets react differently to macro corporate governance factors in this scenario. The legal origin and the quality of its enforcement are identified as determinant variables highlighting differences across countries in previous literature measuring the impact of the IFRS adoption (Daske et al. 2008; Hail et al. 2009; Armstrong et al. 2010). This paper investigates the effects of the adoption of the IFRS on the informativeness of accounting earnings, measured by the relation between earnings and return, across Member States. Because the IFRS implementation process is still in its earlier stages and multiple institutional factors may affect the expected results (i.e., an improvement of the quality of the accounting numbers and the comparability of the reporting practices), we examine how country level corporate governance factors influence earnings quality. Our main hypothesis is that, even applying a common set of high quality accounting standards, institutional factors may negatively affect the earningsreturn relation. Furthermore, we argue that institutional features and market forces dominate accounting quality across countries beyond accounting standards (Ball, 2001; Ball et al. 2003; Soderstrom and Sun, 2007; Christensen et al. 2008). We predict that a stronger corporate governance structure will attenuate this negative effect. Using 1,575 listed companies in 14 countries belonging to the EU from 1999 to 2007, we test the impact of the IFRS adoption on the informativeness of accounting earnings for countries with different institutional structures. We use the relation between earnings and returns as a measure of accounting quality (Fang and Wong, 2002). Following Hillier et al. (2010), we focus on three summary corporate governance factors that represent a global definition of corporate governance and may affect this relation: investor protection, financial system development and corporate control mechanisms. Our main findings revel that the adoption of IFRS increases earnings quality in that period. Differences between the pre-adoption ( ) and the post-adoption ( International Financial Reporting Standards (IFRS) aggregate international accounting standards (IAS), international financial reporting standards (IFRS) and related interpretations (SIC-IFRIC interpretations) issued by the International Accounting Standards Board and adopted by the European Union in conformity with its Regulation (EC) No 1606/

3 2007) periods indicate that companies exhibit higher earnings-return sensitivity during the period than do companies adopting IFRS on or after Additionally, results are consistent with the rationale that the informativeness of accounting earnings increases in strong governance environments (characterized by strong investor protection and corporate control mechanisms and developed financial systems) and this effect is strengthened by IFRS. These findings contribute to the literature on earnings quality in three ways. First, individual-country studies generally examine if companies applying IFRS for the firsttime exhibit higher accounting quality than they do when apply local GAAP (Leuz et al. 2003; Bartov et al. 2005; Hung and Subramanyam, 2007; Paananen, 2008; Paananen and Lin, 2009; Horton and Serafeim, 2009; Cormier et al. 2009). This study complements these individual-country studies by analyzing the relation between earnings and return across 14 different countries. Second, our study complements first cross-country studies on the mandatory IFRS reporting effects on earnings management, timeliness and value relevance metrics (Barth et al. 2008; Daske et al. 2008; Armstrong et al. 2010; Devalle et al. 2010) by addressing the impact of macro corporate governance factors on the earnings-return relation. Because the extent to which accounting standards are followed and implemented is affected by country level institutions, and they vary internationally (Ball et al. 2000), we explore how countries legal and judicial structure, financial system development and corporate control mechanisms infer our analysis. Third, countries different to those belonging to the EU have recently adopted IFRS (actually, more than 100 countries are applying IFRS around the world (Deloitte, 2010) or are evaluating the required use of IFRS in a shortterm period (i.e., Canada and United States (2011 and 2014, respectively), Australia, New Zealand, India and most of the East Asian countries). Early evidence on the effects of the mandatory IFRS adoption at the EU level could be useful to those new jurisdictions that decide to adopt IFRS. The results have implications for investors, national/international standard setters and regulators and also contribute to the literature on corporate governance, including La Porta et al. 1998, 2006, 2008; Pagano and Volpin, 2005; Djankov et al. 2008; Spamann, The paper is organized as follows. In Section 2, we review prior research on the effects of the IFRS adoption and present the hypotheses development based on country level corporate governance factors. Sections 3 and 4 describe the research design, the sample selection, and provide summary statistics. In Section 5, we present the empirical results. Finally, section 6 summarizes the study and outlines the conclusions. 2. RELATED LITERATURE AND HYPOTHESES DEVELOPMENT The EU s movement towards IFRS is expected to contribute to increase the quality of accounting numbers (i.e., earnings) providing capital markets with more value relevant performance measures and helping investors to make better decisions. 3

4 However, this scenario of convergence works depending on how Members States implement IFRS (i.e., allowing Member States legal discretion over the extent to which IFRS are applied in respect of annual accounts and of non-publicly traded companies (see Table 1), or even deferred until 2007, and companies choose among the range of options that are contemplated under IFRS, may exacerbate differences across countries). Additionally, because institutional factors characterizing individual countries differ significantly, incentives and accounting reporting practices vary (Hail et al. 2009) and thus make more complicated to isolate the net effects of the IFRS adoption on the informativeness of accounting earnings. Consequently, these factors need to be considered together. To justify this issue and develop our hypotheses, we first review empirical studies from the international accounting and corporate governance areas. Studies supporting the importance of institutional factors, including those determining the institutional investor protection (i.e., legal origin and enforcement), the financial system development, ownership concentration and board structure, are particularly relevant in this scenario (La Porta et al. 1997, 1998; Ball et al. 2000; Leuz et al. 2003; Haw et al. 2004; Burgstahler et al. 2006). TABLE 1 ABOUT HERE Much of the international accounting evidence on the effects of the IFRS adoption has focused on the role of the accounting standards per se and is generally classified into two groups, based on whether countries allow (do not allow) the voluntary application of IFRS prior to its forced introduction. Studies analyzing the EU s leadership as a consequence of its recent movement towards mandatory IFRS adoption shape the second group and keep the focus of attention. In both cases (voluntary vs. mandatory IFRS adoption), the key argument focuses on the net benefits of the change. Studies on the first group compare IFRS to local GAAP accounting numbers and examine the impact of IFRS. For example, Hung and Subramanyam (2007) examine the voluntary IFRS adoption between by German firms and find that total assets and book value of equity are significantly higher under IFRS. They also find that the adjustments to book value are value relevant but those to earnings are not. However, they do not report evidence suggesting that IFRS have improved the relative value relevance of book value of equity and earnings. Using German stocks markets, Bartov et al. (2005) examine the quality of earnings under different reporting regimes and, opposite to Hung and Subramanyam (2007), find that value relevance is higher for earnings prepared under IFRS than earnings reported under German GAAP. Barth et al. (2008) analyze whether the adoption of IFRS during the period is associated with higher accounting quality. They find that companies applying IFRS from 21 countries present less earnings management, more timely loss recognition and more value relevant accounting figures than do matched firms of the sample applying local GAAP. Studies on reporting quality after the mandatory introduction of IFRS either analyze the market reaction to the IFRS reconciliation adjustments or compare its effects across countries between voluntary and mandatory adopters. Paananen (2008) examine 4

5 whether the accounting quality increases in Sweden after the adoption of IFRS in 2005 and find that there has been no increase in financial reporting quality over the first two year after the introduction. Similarly, Paananen and Lin (2009) compare the characteristics of accounting numbers using a sample of German firms reporting under IFRS during the 2000 through 2006 period. Their results indicate that accounting quality has not improved but worsened over time. Horton and Serafeim (2009) investigate the market reaction to the IFRS reconciliation adjustments following its mandatory introduction in United Kingdom and find that positive earnings adjustments are value relevant both before and after disclosure but the market responds negatively to firms disclosing lower earnings under IFRS compared to UK GAAP. Cormier et al. (2009) examine the value relevance of the optional equity adjustments at transition date that French companies recognize as a result of the IFRS mandatory adoption and also examine how managerial incentives influence the decision to elect optional exemptions that determine these adjustments. They find that mandatory equity adjustments are more value relevant than equity under French GAAP, suggesting an increase in the quality of the financial statements. Related, the evidence in Ball and Shivakumar (2005), Burgstahler et al. (2006) and Holthausen (2009) documents that changes in accounting standards alone do not determine the quality of financial reporting and highlights that additional institutional features of the economy should be taken into account. Daske et al. (2008) examine the economic consequences of mandatory IFRS adoption around the world from 2001 to 2005 and provide evidence suggesting that capital market benefits occur only in countries where firms find incentive to be transparent and the legal enforcement system is strong. In addition, they also find that these effects are most pronounced for firms that voluntarily adopt IFRS. Armstrong et al. (2010) find a positive reaction to the adoption of IFRS for European companies with lower preadoption information quality and higher pre-adoption information asymmetry. This reaction is also positive for companies with high quality pre-adoption information environments. However, the authors find an incrementally negative reaction for firms domiciled in code law countries, which are likely to have weaker enforcement mechanisms. Leuz (2006) supports the view that, as long as there is discretion in financial reporting, even strict enforcement will not successfully eliminate all of the variation in reporting quality. He highlights that there are a number of factors, aside from legal enforcement, which could be responsible for the observed differences in accounting properties. Devalle et al. (2010) also find mixed evidence of an increase in value relevance after the introduction of IFRS in Germany, Spain, France, Italy and UK. Among the five countries considered, they find that value relevance has increased as a consequence of the IFRS adoption in France and UK (code-law vs. common-law countries, respectively). They suggest that differences associated with the consistency of 5

6 implementation and enforcement of IFRS in these countries still remain and impact the cross-border comparability of financial statements 2. In sum, findings related to these studies are mixed. Initially, individual country studies on the first group suggest that there is considerable heterogeneity in the effects of IFRS across different systems. As companies could choose the transition date to IFRS in this scenario (for related evidence, see German literature), it is likely that there are also other factors associated with the country institutional environment and its financial reporting system that could alternatively explain these results. Evidence on the effects on crosscountry studies is in line with this argument and indicates that these factors play an important role in interpreting the results Country Level Corporate Governance Factors If previous literature from the international accounting area concludes that using a common set of high quality accounting standards (i.e. IFRS) does not determine per se the value relevance of accounting numbers and highlights the notion of institutional factors characterizing individual countries appear to have important effects on the quality of the financial information reported (we consider this issue as our starting point), we discuss several hypotheses about the influence of countries legal and judicial structure, financial systems development and corporate control mechanisms on this scenario of global convergence. We use those three summary country-specific factors from corporate governance literature, based on Hillier et al. (2010), because they represent a wide definition for shareholder protection, the development of capital markets and the orientation of the financial systems, and the ownership patterns across countries Investor Protection Prior evidence in the corporate governance literature on the role of legal rules protecting investors and the quality of law enforcement helps to understand why companies have very different financial structures and are owned differently across countries (Jensen and Meckling, 1976; Demirgüç-Kunt and Maksimovic, 1999; La Porta et al., 1997, 1998, 2000, 2006, 2008; Djankov et al. 2008). La Porta et al. (1998) document that legal rules that protect investors vary among legal origins, with the laws 2 The findings for other alternative measures of accounting quality, earnings smoothing, and timely loss recognition, do not suggest that accounting quality improved after the IFRS adoption. The most recent evidence appears to be sensitive to different research designs, especially with regards to alternative empirical metrics used to measure earnings quality. With the same purpose, Chen et al. (2010) examine the effects of IFRS on accounting quality in the EU. Using earnings management and timely loss recognition to assess accounting quality, they find an increase in accounting quality after the mandatory adoption of IFRS. All else equal, they show that the observed improvement is attributable to IFRS, rather than changes in managerial incentives or other environmental factors. 6

7 of common law countries being more protective than the laws of code law countries. Their evidence presents legal investor protection as a strong predictor of financial development. Subsequent literature examining the legal protection influence on the agency conflict has presented evidence suggesting that the level of investor protection determines the quality of financial information reported to outsiders (Ali and Hwang, 2000; Ball et al., 2000; Hung, 2001; Leuz et al., 2003; Holthausen, 2009). Leuz et al. (2003) argue that weak legal protection contributes to a poor-quality financial reporting and, therefore, makes it difficult to contribute to the efficient and competitive functioning of financial markets, even applying the same set of accounting standards. Consequently, countries with strong legal investor protection (i.e., common law systems and countries with high anti-director rights and a strong legal enforcement) are expected to increase the sensitivity of the earnings-return relation under the accounting framework of the IFRS. Hypothesis 1 formally states that companies operating in countries with a strong investor protection exhibit a higher sensitivity of returns to accounting earnings (other alternative performance measures). Based on La Porta et al. (1997, 1998) s measures of investor protection, including revised anti-director rights indices that summarize the minority shareholders protection (Djankov et al., 2008; Spamann, 2010), we construct a proxy that covers the following three components: (1) legal origin, (2) anti-director rights, and (3) legal enforcement. The legal origin, ORIGIN it, equals one if the company belongs to a common-law country (which are English in origin) and zero otherwise (civil-law countries). The second component, the anti-director rights, is defined following the original antidirector rights index from La Porta et al. (1997, 1998), which measures how strongly the legal system protects minority shareholders in the corporate decisionmaking process. We use the Spamann (2010) s corrected version 3 to present a dummy variable, R_ANTIDIR it, which equals one if the company belongs to a country with antidirector rights higher than the sample median, and zero otherwise. Thirdly, we also follow La Porta et al. (1998) to define the legal enforcement component, L_ENFORCEMENT it, which is formed by adding two sub-indices: the judicial system s efficiency and the law and order tradition in the country. It equals one if the company belongs to a country with a higher than the median legal enforcement index, and zero otherwise. Finally, we add up these three components into an aggregated index we labeled as INV_PROTECTION it that takes a value of one for companies belonging to countries with higher than the median investor protection, and zero otherwise Financial System Development Another issue of interest in the international accounting literature is the relationship between accounting and countries economic and financial development, and how that relationship has changed in the last decades. Historically, accounting 3 Spamann (2010, p ) recollects the legal data for 46 of the initial 49 countries included in La Porta el al. (1998) to accurate the original anti-director rights index. The revision of the original index (Djankov et al. 2008) is also discussed in Spamann (2010). 7

8 practices have closely followed the flow of the international investments (Mueller, 1965), originating different patterns in the geography. In fact, differences between countries of Continental Europe and Anglo-Saxon countries are well known in related international literature and have been always pertinent to the debate on accounting harmonization. Since 1995, among the priority objectives of the EU accounting harmonization strategy has been highlighted the role of financial reporting to ensure the transparency and comparability of companies financial information, protect investors and guarantee the financial development 4. Research on financial structure and economic growth identifies the comparative advantages of the different financial systems and classifies them as bank-oriented and market-oriented. In bank-oriented economies, companies form a long-run relationship with those banks that provide the major percentage of their capital needs. This relationship involves private firm-specific information available to only the banks and, thus, restricts the demand for additional public financial information (Ali and Hwang, 2000; and Levine, 2002). In contrast, market-oriented economies populated by multiple individual investors focus more on transactions rather than relationships. Companies financial information is captured by capital markets and transferred to investors, who rely on that information to operate (Boot and Thakor, 2000; Levine, 2002). Therefore, market-oriented countries are expected to display greater financial reporting quality. Hypothesis 2 states that companies operating in countries with a high level of financial system development exhibit a higher sensitivity of returns to accounting earnings (other alternative performance measures). Based on Beck and Demirgüç-Kunt (2009) s financial structure dataset, we use two subindices to examine the level of financial system development: banking and market development indices. The banking development index is defined as the average of three ratios: (1) liquid liabilities/gdp, (2) bank assets/gdp, and (3) claims of deposit money banks on private sector/gdp. The first measure, liquid liabilities/gdp: equals the ratio of liquid liabilities on bank and nonbank financial intermediaries to GDP, which is an indicator of the size of financial intermediaries relative to the size of the economy. Secondly, bank assets/gdp: equals the ratio of the total domestic assets of deposit money banks to GDP, which is a measure of the size of the banking sector. The third measure, claims of deposit money banks on private sector/gdp: equals deposit money bank credits to the private sector as a share of GDP, which is an indicator of bank activity in the private sector. The market development index is defined as the average of two ratios, which are measures of the market size and its activity, respectively: (1) stock market capitalization as a share of GDP, and (2) total value traded as a share of GDP. The first measure equals the ratio of the value of domestic equities to GDP. The second one equals the value of trades of domestic equities on domestic exchanges divided by 4 See strategies of the European Commission on international accounting harmonization: COM 95 (508), 11/14/1995: Accounting harmonization: A new strategy vis-à-vis international harmonization ; and COM (2000) 359 final, 06/13/2000: EU financial reporting strategy: The way forward. 8

9 GDP 5. Finally, we aggregate these two sub-indices and obtain a dummy variable, F_SYSTEM it, which captures the level of financial system development. The F_SYSTEM it variable takes a value of one for companies belonging to countries with higher than the median financial system development, and zero otherwise Control Mechanisms Following Fang and Wong (2002), Ball et al. (2003), Leuz et al. (2003), and Leuz (2006), we also give particular attention to the effects that various control mechanisms of corporate governance have on firms financial reporting quality and characterize the properties of accounting earnings. Our contribution is to analyze its interaction under the case where IFRS are the formal basis for financial reporting from 2005 onwards, and it is expected an increase of earnings quality across EU countries. Interestingly, prior evidence in East Asian countries (with a historical common-law influence on their regulatory requirements, included the influence of IFRS) has provided a useful setting to study the subject of firms financial reporting quality in a different institutional framework from those of the EU or USA. In this sense, Fang and Won (2002) and Ball et al. (2003) s studies show that concentrated ownership structure in East Asia is associated with opacity and low earnings informativeness. Although the accounting standard-setting structure predominantly follows the common-law model in the region, the authors identify code-law properties in preparers incentives. Again, the notion that earnings quality is not ultimately determined by accounting standards alone is present in East Asian economies. Although the literature highlights the role of corporate ownership structure and emphasizes the association between ownership concentration and financial disclosure quality, we also consider two additional corporate governance mechanisms such as the board structure of the company and the market for corporate control in our analysis. The relation between corporate ownership structure and the quality of firms financial reporting that we discuss in this study is based on the quality of accounting earnings reported to market participants. It is argued that concentrated ownership structures allow controlling owners to use their effective control over the firms to influence the information disclosure. Among the different benefits attributed to that behavior in corporate governance literature on financial reporting incentives, the power of the owners to decide how profits are recognized and distributed is crucial. On the one hand, the quality of firms reported earnings depends on their incentives (for example, incentives that determine the use of alternative accounting methods, and thus the recognition of profits and losses, or incentives that are linked to tax and dividends policies). On the other hand, owners incentives are simultaneously influenced by institutional factors discussed above, in previous sub-sections. Ball et al. (2003) suggest that the demand for high quality financial reporting is higher in market-oriented common-law countries due to role played by external shareholders and market analysts, 5 These five ratios are defined following Demirgüç-Kunt and Levine (2001: 84-92). 9

10 and the high frequency of shareholder litigation. Conversely, in code-law countries, particularly those with a high level of book-tax conformity, accounting earnings are usually affected by firm s tax strategy to align financial and tax reporting 6. As a consequence, non-controlling shareholders and market participants pay less attention to the reported figures and reduce the demand for financial reporting quality. We then believe these arguments suggest that high ownership concentration negatively affects the informativeness of accounting earnings. The role played by members of the board (which is determined depending on the structure of the company s board as well as its functions and responsibilities) is also essential to good corporate governance practices and investor relations. Independently of the board system (unitary versus dual board structure), members of the board should ensure that financial reporting and control systems operate properly and both comply with disclosure requirements (Mallin, 2010). Board balance and independence are corporate control mechanisms that promote transparency in corporate reporting and avoid information asymmetries among investors 7. Overall, the evidence on the association between characteristics of the board of directors and internal control procedures considers these mechanisms as monitors of the reporting system that constrain management s use of discretion and determine earnings quality (Dechow et al. 2010). With the exception of Austria, Denmark, Germany and the Netherlands, the majority of EU Member States of the sample has a unitary board of directors, which is characterized by a unique board that is involved in daily transactions and the execution of decisions. In contracts, in a dual board structure there is a clear distinction between executive directors who integrate the executive board, and non-executive directors that constitute the supervisory board and are responsible for the strategic control and supervision 8. Here, the role and effectiveness of non-executive directors help us to better follow the flow of corporate information, and avoid concentrated power situations in management groups or earnings management behaviors. Thus, an effective board structure will lead its members to ensure a high level of financial disclosure quality that facilitates the alignment of executives and shareholders interests and the efficient allocation of resources. With regard to the latter hypothesis, Petra (2007) finds a positive relation 6 There is a vast literature that discusses the effects of tax laws on those countries where conformity between financial and corporate tax reporting is required (its majority is focused on Europe). Goncharov and Werner (2009) investigate the effects of this link on accounting quality, addressing for the first time those cases where conformity applies only to companies that prepare individual financial statements (vs. consolidated financial statements which are not affected by book-tax provisions). After controlling for country and firm institutional differences, they conclude that the effect of book-tax alignment remains robust for individual financial statements and disappears for consolidated accounts. However, alternative non-tax reporting incentives still affect both individual/consolidated accounts. 7 Bushman and Smith (2001) adopt the classic agency perspective to define corporate control mechanisms as those means by which managers are disciplined to act in the investors interest. 8 In practice, certain industries or large companies in countries with a unitary board (for example, Belgium, Italy or Spain) implement a dual system (Van den Berghe, 2002: 71). 10

11 between the percentage of non-executive independent directors serving on the corporate board and earnings informativeness. Lastly, the market for corporate control is an additional external mechanism that may affect the earnings-return relation when management s opportunistic behavior is identified during takeover activities. In those cases, it plays a key role in disciplining managers, resolving extreme agency conflicts (Manne, 1965; Jensen and Ruback, 1983; Franks and Mayer, 1996) and protecting investors. For example, Williamson (1983) argues that the market for corporate control can act as a substitute for board monitoring. In the context of financial reporting research, Armstrong et al. (2010) highlight the central role of the accounting system in reducing the information-related agency costs that appear among managers, directors, and equity and debt capital providers. The discussion in Armstrong et al. (2010) raises an elementary question about how financial reporting facilitates activity in the market for corporate control. Toward this end, Zhao and Chen (2008) provide evidence suggesting that more-protected managers from discipline from market for corporate control reduce their exposition to pressures for earnings management. They find that staggered boards (defined as a strong antitakeover device) are significantly associated with a lower likelihood of financial reporting fraud. In summary, countries with strong control mechanisms are expected to reveal more value relevant financial disclosure. Formally, our hypothesis 3 is: companies operating in countries with strong internal and external control mechanisms exhibit a higher sensitivity of returns to accounting earnings (other alternative performance measures). Following Hillier et al. (2010) we construct an index which proxies for control mechanisms and is defined as the average of the three mechanisms presented previously: (1) ownership concentration, (2) board system, and (3) market for corporate control. Based on Djankov et al. (2008) s measure of ownership concentration, which is defined as the average percentage of common shares owned by the top three shareholders in the ten largest non-financial privately-owned domestic firms in a particular country, we create a dummy variable, O_CONCENTRATION it, and assign it a one if the company is located in a country with a level of ownership concentration higher than the sample median, and zero otherwise. Next, following Mallin (2010) we define a dummy variable that proxies for board structure: B_STRUCTURE it. We assign the board system a value of one if it is dual, and zero if unitary. Similarly, MARK_CONTROL it, equals one if the company belongs to a country with an active market for corporate control, and zero otherwise. Lastly, we add up these three components into an aggregated index, C_MECHANISMS it, which takes a value of one for companies belonging to countries with higher than the median control mechanisms, and zero otherwise. 3. SAMPLE SELECTION AND DATA 11

12 Consistent with Regulation (EC) No 1606/2002 on the application of IFRS at the EU level (more specifically, articles 4 and 5), we identify listed companies belonging to the initial EU Member States from Worldscope from 1999 to In selecting, we use the following criteria: First, companies must elaborate consolidated financial statements during the sample period. Second, we must have enough accounting and financial data from annual reports, and also data on country level institutional factors to implement the analysis. Data on financial development and structure are provided by Beck and Demirgüç-Kunt (2009), which updates financial structure metrics with data through Third, we exclude companies operating in finance and insurance industries from the empirical study due to regulation particularities. As a result, the final sample is integrated by 1,575 listed companies with 11,951 firm-year observations from 1999 to 2007 from Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Portugal, Spain, Sweden, and United Kingdom (UK). Table 2 lists the distribution of companies and firm-year observations by country, showing a great representation from France, Germany and UK. The sample breakdown by country ranks from 21 companies in Ireland or Portugal (181 and 178 firm-year observations, respectively) to 414 in UK (3,131 firm-year observations). Cross-country differences are explained by both the dimension of their capital markets and data availability. Similarly, Table 3 presents the industry breakdown. Companies are ranked by Worldscope Database according to the Standard Industry Classification that covers all the economic activities. There is a significant variation across sectors, with most companies within the industrial groups. Oppositely, the number of listed companies (firm-year observations) belonging to primary activities (i.e. agriculture, foresting, fishing and mining) is very low, representing less than five percent of the total sample. TABLE 2 ABOUT HERE TABLE 3 ABOUT HERE Panels A, B and C in Table 4 provide descriptive statistics on the institutional factors of each country in the sample described in section 2: investor protection, financial system development and control mechanisms. As indicated in Panel A, all countries except Ireland and UK have a code-law tradition. In Panels A, B, and C, Denmark, the Netherlands, Sweden, and UK appear characterized by strong investor protection (widespread anti-directors rights and strong legal enforcement), high financial system development, and strong control mechanisms (i.e., low ownership concentration). Oppositely, Belgium, Greece, Italy, and Portugal exhibit weaker investor protection, lower financial system development, and weaker control mechanisms (high ownership concentration, unitary board structure, and non-active markets for corporate control) than countries in the first group. TABLE 4 ABOUT HERE 4. RESEARCH DESIGN 12

13 In this section, we discuss the method we use to measure the informativeness of accounting earnings and test the hypotheses in section 2. We then present model specifications to examine the link between institutional factors and earnings informativeness across EU Member States. Lastly, we also define a set of variables which are included in the analysis to control for possible variations in the earningsreturn relation that could be explained by sources other than country level corporate governance factors Informativeness of accounting earnings Following prior studies on accounting quality, we focus on the relation between earnings and returns as a measure of accounting quality. As in Hung (2001), this approach allows us to examine the ability of the accounting performance measures to capture information that affects the value of the firm 9. Our aim is to first explore the earnings-return relation, and then to analyze how institutional factors influence accounting quality. Earnings are used as a summary accounting performance measure. Model (1) and models specified in section 4.2 are estimated using panel data analysis. Over cross sections and aggregate time series, the econometric analysis based on panel data models presents two major benefits: controlling for the unobservable variation of individual behavior (unobservable individual heterogeneity) and also its dynamics (Arellano 2003: 7-19), and second, the ability to eliminate the bias of aggregation that arises when time series models are applied. In panel data models, the error term is split into two components: the firm-specific effect that characterizes each individual company over time, denoted by η i, and the random disturbance term (ε it ). Accordingly, we estimate model (1) as follows: A_RETURN it = β 0 + (β 1 + α 1 D_IFRS it )NI it + β 2 D_IFRS it + η i + ε it (1) where, for each individual company i, A_RETURN it is the cumulative 12-month stock returns three months after the company s fiscal year-end t; NI it the net earnings at year t divided by the market value of equity at the beginning of year t; D_IFRS a dummy variable that equals one if the company applies IFRS, and zero otherwise; η i the individual or firm-specific effect; and ε it the error term at year t. Panel A of Table 5 presents a summary of the descriptive statistics of the regression variables. The median A_RETURN it is 6.11% and the median NI it is 5.63%. Additional information of A_RETURN it, due to the large dispersion in values, is also presented in Panel B of Table 5. The 25 th and 75 th percentiles A_RETURN it are % and 34.65%, respectively. We firstly estimate model (1) pooling all of the years and countries from 1999 to This regression examines the earnings-return relation in the pre-adoption period (labeled 9 Hung (2001) s portfolio-returns approach measures value relevance of accounting numbers as the total return that could be earned from a portfolio based on perfect foresight of earnings. 13

14 as the voluntary IFRS adoption period ). When examining the impact of IFRS on net earnings in the pre-adoption period, β 1 + α 1 is the coefficient of NI it for voluntary IFRS adopters (since D_IFRS it equals one); whereas β 1 is the coefficient for companies applying a different set of accounting standards (D_IFRS it = 0). Secondly, we pool all sample years from 2005 to 2007 ( the mandatory IFRS adoption period ) The influence of country level institutional factors We next examine the effects of institutional factors on the earnings-return relation. We estimate models (2)-(4) pooling all sample years from 1999 to 2007, and countries: A_RETURN it = β 0 + β 1 NI it + β 2 SIZE it + β 3 Q it + β 4 LEVERAGE it + (β 5 + α 1 INV_PROTECTION it )D_IFRS it + η i + ε it (2) A_RETURN it = β 0 + β 1 NI it + β 2 SIZE it + β 3 Q it + β 4 LEVERAGE it + (β 5 + α 1 F_SYSTEM it )D_IFRS it + η i + ε it (3) A_RETURN it = β 0 + β 1 NI it + β 2 SIZE it + β 3 Q it + β 4 LEVERAGE it + (β 5 + α 1 C_MECHANISMS it )D_IFRS it + η i + ε it (4) where, A_RETURN it, NI it, and D_IFRS it are as described in model (1); SIZE it is the natural logarithm of the market value of equity at the beginning of year t; Q it is the market value of equity divided by the book value of total assets at the beginning of year t; LEVERAGE it is the total liability divided by total assets at the beginning of year t. SIZE it, Q it, and LEVERAGE it control for possible variations in the earnings-return return relation that could be explained by sources other than institutional factors (Panel A of Table 5 presents the summary descriptive statistics of the variables). INV_PROTECTION it is a dummy variable that equals one for companies belonging to countries with higher than the median investor protection, and zero otherwise; F_SYSTEM it is a dummy variable that captures the level of financial system development, taking a value of one for companies located in countries with higher than the median financial system development, and zero otherwise; C_MECHANISMS it is a dummy variable that equals one for companies belonging to countries with higher than the median control mechanisms, and zero otherwise; η i is the individual or firm-specific effect; and ε it is the error term at year t. TABLE 5 ABOUT HERE Models (2), (3), and (4) assume that INV_PROTECTION it, F_SYSTEM it, and C_MECHANISMS it are exogenous variables. Each individual model includes an interaction term between the dummy variable D_IFRS it and the institutional factor, and tests the impact of the IFRS adoption on the informativeness of accounting earnings for countries with different corporate governance structures (i.e., in model (2), the coefficient of INV_PROTECTION it *D_IFRS it, α 1, represents the incremental effect when changing from an environment characterized by a weak shareholder protection to 14

15 one with strong shareholder protection. In countries with a strong investor protection, (β 5 + α 1 ) represents this effect 10 ). 5. EMPIRICAL RESULTS Table 6 provides results on the basic earnings-return relation comparing the informativenss of accounting earnings in the pre-adoption and post-adoption periods. Column 2 in Table 6 summarizes the results of model (1) for the first period. The estimated coefficient of NI it reveals that stock returns are significant and positively associated with accounting earnings at the 0.01 level. As predicted, the coefficient of D_IFRS it is significantly positive at the 0.01 level, indicating that in the pre-adoption period companies adopting IFRS exhibit a higher level of earnings quality than do nonvoluntary adopters. These findings are consistent with related literature on the effects of the voluntary transition from previous GAAP to IFRS. Additional analysis testing the sum of the coefficients of NI it and D_IFRS it *NI it, (β 1 + α 1 ), suggests that under IFRS companies exhibit annual stock returns which are more sensitive to net earnings (β 1 + α 1 = 1.135, t=26.35) than do companies not applying IFRS (β 1 = 0.439) from 1999 to Findings related to the post-adoption period are reported in column 3 in Table 6. Consistent with predictions, the coefficients of NI it and D_IFRS it remain both statistically significant and positively associated to the dependent variable after the mandatory adoption of the IFRS in However, the coefficient of the interaction term D_IFRS it *NI it, α 1, is significantly negative at the 0.01 level. Although there is an increase in the quality of reported earnings in that period, the results indicate that those companies that were forced to apply IFRS under the EU Regulation do not experience the improvement in earnings quality (β 1 + α 1 = 0.907, t=2.85) than do voluntary adopters during the period Overall, findings in Table 6 provide direct evidence on the important role that NI it play in the earnings-return relation. However, annual stock returns are more sensitive to accounting earnings in the pre-adoption period across EU Member States. TABLE 6 ABOUT HERE Table 7 summarizes the results of models (2)-(4). The estimated coefficients of NI it and D_IFRS it are significantly positive at the 0.01 level across all the models (as in Table 6). It highlights the variables explanatory power when we re-estimate the initial model including control variables that may affect variations of NI it. The results in model (2), which incorporates INV_PROTECTION it variable as a country level corporate governance factor and tests hypothesis 1, are presented in column 1 of Table 7. These results show that the coefficient of INV_PROTECTION it is significantly positive at the 10 In this case, if β 5 and α 1 are significant, a linear restriction test must be used to determine whether (β 5 + α 1 ) is significantly different from zero. 15

16 0.01 level and support our view that investor protection is a fundamental factor influencing earnings quality across sample countries. Additional analysis testing the sum of the coefficients of D_IFRS it and INV_PROTECTION it *D_IFRS it, (β 5 + α 1 ), suggests that under more effective protection environments (INV_PROTECTION it = 1) companies exhibit stock returns which are more sensitive to the IFRS adoption (β 5 + α 1 = 0.378, t=47.39) than do companies operating in countries with weak investor protection (β 1 = 0.168). Column 2 of Table 7 presents the results in model (3), which tests the effect of the F_SYTEM it variable on the informativeness of accounting earnings as stated in hypothesis 2. This factor exhibits a significant positive coefficient at the 0.10 level. A further analysis examining the sum of the coefficients of D_IFRS it and F_SYSTEM it *D_IFRS it, (β 5 + α 1 ), indicates that companies operating in countries with a high level of financial system development (F_SYSTEM it = 1) report a higher sensitivity of annual stock returns to the adoption of IFRS (β 5 + α 1 = 0.327, t=45.51) than do companies operating in countries with a low level of financial system development (β 1 = 0.203). The results in model (4), which incorporates C_MECHANISMS it variable and tests hypothesis 3, are presented in column 3 of Table 7. These results show that the coefficient of C_MECHANISMS it is significantly positive at the 0.01 level, indicating that earnings quality and corporate control mechanisms are positively associated as predicted by our third hypothesis. Finally, additional analysis testing the sum of the coefficients of D_IFRS it and C_MECHANISMS it *D_IFRS it, (β 5 + α 1 ), suggests that under more effective control mechanisms (C_MECHANISMS it = 1) companies exhibit stock returns which are more sensitive to the IFRS adoption (β 5 + α 1 = 0.397, t=48.85) than do companies belonging to countries with less effective control mechanisms (β 1 = 0.165). The coefficients of the additional control variables, SIZE it, Q it, and LEVERAGE it, are statistically significant. The estimated coefficient of Q it is significantly positive at the 0.01 level, consistent with the idea that growth opportunities tend to be positively related to future earnings (Collins and Kothari, 1999). The coefficients of LEVERAGE it and SIZE it are significantly negative at the 0.10 and 0.01 level, respectively. Taking together, it suggests that large companies and companies with a high level of debt are associated to higher risk and as a consequence exhibit a weak earnings-return relation. However, on the other hand, it could be argued that large companies (and/or large companies with high leverage) are expected to spend more resources to increase earnings quality and exhibit a positive effect on the earnings-return relation. In summary, hypotheses 1, 2, and 3 are supported and consistent with the notion that the informativeness of accounting earnings increases in strong governance environments (strong investor protection and corporate control mechanisms, and 16

17 developed financial system) and this effect is strengthened by IFRS across EU Member States. TABLE 7 ABOUT HERE 6. CONCLUSIONS This article investigates the effects of the adoption of IFRS on the informativeness of accounting earnings, measured by the earnings-return relation, across EU Member States. We use 1,575 listed companies in 14 countries belonging to the EU from 1999 to 2007 and focus our attention on three summary corporate governance factors that represent a global definition of corporate governance and may affect this relation: (1) investor protection, (2) financial system development, and (3) corporate control mechanisms. Consistent with prior literature on the effects of the IFRS adoption, we firstly show that the use of IFRS increases earnings quality during the sample period. A further analysis examining differences between the pre-adoption and post-adoption periods ( vs ) indicates that companies exhibit higher earnings-return sensitivity in the period than do companies adopting IFRS on or after In this scenario of international convergence, we also find that different corporate governance structures affect the basic earnings-return relation across countries. The results highlight the positive association between institutional factors such as investor protection, financial system development and corporate control mechanisms, and the quality of accounting earnings reported to capital markets. Overall, the results are consistent with the rationale that the informativeness of earnings increases in strong governance environments and this effect is strengthened by IFRS. Our research results have important implication for investors, national/international standard setters and regulators and also contribute to the literature on corporate governance. Countries different to those belonging to the EU have recently adopted IFRS or are evaluating the required use in a short-term period (i.e. Canada, U.S.A., Australia, New Zealand, India, and most of the East Asian countries) and early evidence on the effects of the IFRS at the EU level could be useful. Future research on the effects of country level corporate governance factors on the informativeness of accounting earnings of countries adopting this set of high-quality standards for the first-time would contribute to the future of global accounting convergence. 17