A cross-country study on the relationship between financial development and earnings management

Transcription

1 A cross-country study on the relationship between financial development and earnings management Masahiro Enomoto * Kobe University, Kobe, Japan Fumihiko Kimura Tohoku University, Sendai, Japan Tomoyasu Yamaguchi Tohoku Gakuin University, Sendai, Japan February 2014 Acknowledgments The authors gratefully acknowledge the financial support from The Japan Securities Scholarship Foundation and Ishii Memorial Securities Research Foundation. * Corresponding author Research Institute for Economics & Business Administration, Kobe University, 2-1 Rokkodaicho, Nada-ku, Kobe , JAPAN. menomoto@rieb.kobe-u.ac.jp. Tel: ; fax: Graduate School of Economics and Management, Tohoku University, Katahira, Aoba-ku, Sendai , JAPAN. fkimura@econ.tohoku.ac.jp. Tel: ; fax: Faculty of Business Administration, Tohoku Gakuin University, Tsuchitoi, Aoba-ku, Sendai , JAPAN. yamaguchi@tscc.tohoku-gakuin.ac.jp. Tel: ; fax:

2 A cross-country study on the relationship between financial development and earnings management Abstract This paper investigates whether the level of financial development influences earnings management in an international setting. We deal with accrual-based and real earnings management. While prior cross-country research on financial accounting treats legal traditions, outside investor protection, and corporate governance as non-accounting institutions, we focus on financial development. Financial development is defined as the factors, policies, and institutions that lead to effective financial intermediation and markets, as well as deep and broad access to capital and financial services (The World Economic Forum, 2012). Given financial accounting s key role of offering information to investors, under the accounting standards of each country, we propose that there is a link between financial development and the resulting outcomes from accounting institutions. We examine the relationship between financial development and both types of earnings management using 54,178 observations in 37 countries from 2009 to The results show that a manager is restrained in both types of earnings management when under a higher level of financial development. We interpret the results as showing (1) higher quality accounting information is needed in countries with more developed financial systems, (2) there is a link between financial development and accounting institutions in each country, and (3) financial development disciplines managers and mitigates their incentives to manage earnings. JEL Classification: M41 Key Words: Financial development, Accounting institutions, Earnings management

3 1. Introduction The purpose of this paper is to investigate whether the level of financial development in a country influences earnings management in that country. Financial development is closely related to the spread of financial accounting. We focus on financial development as a crucial factor of managerial discretion internationally. Our measure for financial development is based on the World Economic Forum. The scores cover a wide range of institutional factors used in prior research. The results of this paper provide some evidence that financial development influences the reporting incentive of a country s manager in an international setting. In other words, we show a relationship between accounting institutions and non-accounting institutions. Leuz and Wysocki (2009) and Wysocki (2011) emphasized new institutional accounting research that focuses on relationships between accounting and institutional factors. They point to two critical issues; that accounting institutions are key economic institutions, and that a linkage between accounting institutions and non-accounting institutions exists. Accounting institutions include accounting standards, disclosure systems, audits, and so forth. Non-accounting institutions include legal systems, corporate governance mechanisms, the existence and enforcement of laws governing investor protection, and disclosure standards (Wysocki, 2011, 312). Both institutions vary from country to country. A useful method in new institutional accounting research is cross-country analysis. For example, Ball, Robin, and Wu (2003) investigate the properties of accounting income in four countries (Hong Kong, Malaysia, Singapore, and Thailand) and show that their financial reporting quality is not higher than under code law, with quality operationalized as timely recognition of economic income (particularly losses). As for non-accounting institutions, Leuz, Nanda, and Wysocki (2003) and Boonlert-U-Thai, Meek, and Nabar (2006) define investor protection as the power to prevent managers from expropriating minority shareholders and creditors within the constraints imposed by law. Leuz, Nanda, and Wysocki (2003) examine the relationships between outside investor protection and earnings management in 31 countries, from 1990 to 1999, and find that earnings management decreases in countries with stronger 1

4 investor protection. Boonlert-U-Thai, Meek, and Nabar (2006) also investigate earnings management in 31 countries, from 1996 to 2002, and they suggest that earnings are smoothed in countries where investor protection has progressed. Bushman and Piotroski (2006) examine conservatism in earnings among the countries, and suggest that various institutions, such as legal systems, ownership structures, and taxation, have an effect on the degree of conservatism. While prior cross-country research on financial accounting treats legal traditions, outside investor protection, and corporate governance as non-accounting institutions, we focus on financial development. Financial development is defined as the factors, policies, and institutions that lead to effective financial intermediation and markets, as well as deep and broad access to capital and financial services (The World Economic Forum, 2012). Given financial accounting s key role of offering information to investors, under the accounting standards of each country, we propose that there is a link between financial development and the resulting outcomes from accounting institutions. In this paper, we measure financial development based on The Financial Development Report from the World Economic Forum. 1 The report ranks 62 countries according to different aspects of complex financial systems. These are based on research and data from organizations such as the World Bank and the IMF including the institutional environment, the business environment, financial stability, banks, capital markets, overall capital availability, and access. They are used frequently in research on financial development, and reflect the various aspects of non-accounting institutions. They are, therefore, suitable measures for our research. In addition, the scores for rank are updated annually, and those that relate to institutional factors are usually, in most previous cross-country research, fixed in the sample period. In fact, these scores should vary with economic circumstances and revisions of laws and regulations. It is, therefore, appropriate for this study to adopt the scores of the World Economic Forum, and it should be noted that almost all scores are changing every year. For example, the score relating to the 1 The World Economic Forum is an independent international organization committed to improving the state of the world by engaging business, political, academic and other leaders of society to shape global, regional and industry agendas ( 2

5 institutional environment in the United Kingdom has risen from 5.54 in 2009 to 6.00 in Looking at previous research, we focus on earnings management as an economic consequence caused by accounting institutions. Earnings management is one of the subjects of capital market-based accounting research and positive accounting theory, and is a crucial factor affecting accounting quality (see Dechow, Ge, and Schrand, 2010). Because accounting information is essential to the development of financial markets and transactions, financial development brings about the expanded use of accounting information. This could lead to two patterns in the relationship between earnings management and financial development. One is that earnings management is more often implemented in countries with higher financial development, because a growing use of accounting information produces a greater benefit from earnings management. The other is that earnings management is restricted in countries with higher financial development, because earnings management is strictly monitored or more severely punished in some cases. Further, while extensive research relating to earnings management has covered accrual-based earnings management (AEM), we focus on real earnings management (REM) that affects a firm s real activities. This paper examines the relationship between AEM, REM, and financial development by using 54,178 observations in 37 countries from 2009 to The results show that managers are restrained from using both AEM and REM when under higher financial development. Our evidence is robust, with the use of multiple financial development scores and the exclusion of observations from U.S. and Japan and the elimination of firm-years in The most important contributions of this paper are as follows. First, we expend on the earnings management research by investigating the relationship between financial development and earnings management. Our paper provides the evidence that financial development restrains earnings management, both AEM and REM, by disciplining a manager s behavior. The results suggest that managers in countries with more developed financial systems are required to provide a higher level of transparency with regard to their financial accounting information. In other words, financial development leads to increased monitoring, scrutiny, and punishment 3

6 from outside of the firm and puts pressure on a manager s ability to manipulate a firm s real activities as well as accruals. Although prior research has already shown negative relationships between AEM and investor protection, there is scant evidence of an association between earnings management and financial development. We added the REM measures of Roychowdhury (2006) as a proxy to capture managers discretionary behavior at the international level while Leuz, Nanda, and Wysocki (2003) and Haw, Hu, Hwang, and Wu (2004) use AEM mainly as a proxy for earnings management. Our results are also in conflict with Francis, Hasan, and Li (2011) who show that the substitution of REM for AEM in countries with stronger investor protection is prevalent. This substitution is frequently observed in prior research that uses data from a single country setting, but it is not found at the international level in our study. Second, we refine the global impact of financial development on accounting at a firm level employing the comprehensive measure of financial development. Further, our score includes many institutional factors which prior research frequently employ as a critical determinant of earnings management at the international level. Therefore, this paper contributes to a growing literature evaluating the factors of international differences in the quality of accounting information. The remainder of the paper is organized as follows. Section 2 considers the relationship between earnings management and financial development. Section 3 delineates the research design of the study. Section 4 contains the results of the econometric analysis. Section 5 concludes the paper with an extensive discussion of suggestions for future research. 2. Hypothesis development 2.1. Financial development and earnings management The circumstances surrounding a firm s financing activities differ from one country to another. A great deal of cross-country research into financial theory suggests that financial development is related to economic growth. For example, Beck and Levine (2004) use a dataset of 40 4

7 countries, from 1976 to 1998, and conclude that fully functional financial systems ease information and transaction costs and thereby enhance resource allocation and economic growth. The World Economic Forum (2012, xiii) defines financial development as the factors, policies, and institutions that lead to effective financial intermediation and markets, as well as deep and broad access to capital and financial services. It also points out that financial development is dictated by seven pillars, specifically (1) Institutional environment, (2) Business environment, (3) Financial stability, (4) Banking financial services, (5) Non-banking financial services, (6) Financial markets, and (7) Financial access. Each pillar aggregates 12 to 26 indices. 2 All pillars have a relationship with financial accounting to a greater or lesser extent. This paper aims to examine the effect of financial development on financial accounting. However, three pillars, (2) Business environment, (3) Financial stability, and (7) Financial access, associate only weakly with financial accounting. Therefore, we have deleted these three pillars and selected the remaining four from the perspective of the importance of financial accounting. The definition of the seven pillars and the reason for eliminating the three is described in Section 3.2. Next, we discuss the relationship between financial accounting and financial development. Financial accounting is expected to play an important role in the financial systems of each country and in the global financial system. The FASB Statement of Financial Accounting Concepts No. 1 states that financial reporting should provide information that is useful to present and potential investors in making rational investment decisions (par. 34), and provide information to help present and potential investors in assessing the amounts, timing, and uncertainty of prospective cash receipts (par. 37). In addition, the conceptual framework of International Financial Reporting Standards (IFRS) states that the objective of general purpose financial reporting is to provide financial information about the reporting entity that is useful to present and potential equity investors, lenders, and other creditors in making decisions in their capacity as capital providers. Information that is decision useful to capital providers may also be 2 See The World Economic Forum (2012, ) for details of the indices. 5

8 useful to other users of financial accounting information who are not capital providers (IASB, 2008, OB2). 3 Wysocki (2011, 311) points out that accounting is one institutional mechanism that can help lower transaction costs, reduce information costs and information asymmetry, lower coordination costs and improve enforcement of property rights, and that accounting is the growing body of evidence about which institutions are the primary mechanisms for financial market development. This suggests that the goal of offering accounting information is to develop and maintain the financial system in a country. However, in order to achieve this goal, the accounting information presented by managers is subject to monitoring and scrutiny by auditors and stakeholders. Managers prepare financial statements under pressure. We, therefore, argue that higher quality accounting information is required in countries with more developed financial systems. The quality of accounting information is affected by earnings management. Earnings management is defined as the choice by a manager of accounting policies, or real actions that affect earnings so as to achieve a specific reported earnings objective (Scott, 2011, 423). Some previous research suggests that the motivation of earnings management is related, to an appreciable degree, to financing activities and related legal issues. Examples discussed are raising stock prices in an initial public offering (Teoh, Welch, and Wong, 1998a), secondary equity offering (Teoh, Welch, and Wong, 1998b), avoiding debt covenant violation (Bikky and Picheng, 2002), meeting or beating analyst expectations (Bartov, Givoly, and Hayn, 2002), and the Sarbanes-Oxley Act of 2002 (Cohen, Dey, and Lys, 2008). Based on the above discussions, there are two scenarios in the relationship between earnings management and the financial development of each country. One is that higher financial development is likely to encourage earnings management because managers acquire the benefits of spreading the use of accounting information about stakeholders. Another is that higher financial development is likely to constrain earnings management. The reason is that, with 3 Examples of other users of financial accounting information that are not capital providers are regulators, securities exchanges, analysts, auditors, and so forth. 6

9 financial development, managers fear monitoring, scrutiny, and punishment from outside of the firm, and new accounting standards, that narrow their discretion to achieve target income, might be adopted The choice between AEM and REM In this subsection, we explain the two types of earnings management, provide an overview of previous research into earnings management in an international setting, and develop our hypotheses. As mentioned in section 1, earnings management is divided into AEM and REM. AEM alters the accrual process to manage earnings. AEM leads to the reversal of accruals in subsequent periods, but does not have a direct effect on cash flow. 5 Unlike REM, managers are allowed to implement AEM after the period-end. REM reflects a firm s real activities and manages earnings through changing the timing or structure of an operating, investment, or financial decision. REM may reduce future cash flow due to non-optimal decisions, such as opportunistically reducing R&D and advertising costs, increasing price discounts, and overproducing to decrease unit costs. Prior research shows that there is the relationship between earnings management and institutional factors in international settings, employing mainly accrual-based measures as a proxy for earnings management. Leuz, Nanda, and Wysocki (2003) look at 31 countries, from 1990 to 1999, and show that AEM decreases in countries with stronger investor protection. 6 Similar findings are also shown in Haw, Hu, Hwang, and Wu (2004) using 31 countries, from 1996 to Boonlert-U-Thai (2006) investigates the relationship between investor protection 4 The former is that financial development (i.e. non-accounting institution) directly influences the manager s incentives to manage earnings. For example, the extent of anti-director rights and/or legal enforcement (Leuz, Nanda, and Wysocki, 2003; Francis, Hasan, and Li, 2011) and the number of analyst followings in countries with high financial development (Degeorge, Ding, Jeanjean, and Stolowy, 2013) have been tested. Investor protection by law and financial intermediation are incorporated into our financial development score (see details in section 3.2.). The latter is the indirect effect of financial development on earnings management via the changes in accounting institutions. Barth, Landsman, and Lang (2008) and Ipino and Parbonetti (2011) examine the impact of the adoption of IFRS on accounting quality. In other words, there are two possible channels where financial development might impact earnings management. We do not distinguish between the direct and indirect effect, but financial development probably influences all the examples above. 5 The reversal of accruals is detailed in Dechow, Hutton, Kim, and Sloan (2012). 6 Three of the four measures of earnings management in Leuz, Nanda, and Wysocki (2003) are based on accruals and can, therefore, be regarded as AEM. Another is loss avoidance. 7

10 and AEM in 31 countries, from 1996 to 2002, and suggests that earnings are smoothed by accruals in countries where investor protection has progressed. Francis and Wang (2008) use 49 countries, from 1995 to 2004, and provide evidence that earnings quality improves at firms audited by brand name auditors in countries with stronger investor protection. 7 Generally, financial development requires stricter investor protection. From the above discussion, we predict that AEM is decreases with financial development. Degeorge, Ding, Jeanjean, and Stolowy (2013) use 21 countries, from 1993 to 2002 and find that analyst following negatively affects AEM in countries with high levels of financial development. However, it is not clear if financial development has a direct effect on AEM. Next, we should review the relationship between AEM and REM in an international setting, but little research has been done on this. Francis, Hasan, and Li (2011) show that REM is encouraged and AEM is discouraged in countries with stronger legal environments, but they do not include critical variables relating to financial development 8 Enomoto, Kimura, and Yamaguchi (2013) also investigate the relationship between AEM, REM, and investor protection. Although they use analyst following as a proxy for financial intermediaries, their discussion and analysis are based on country data rather than firm-year data. A manager may choose AEM and/or REM to manage earnings under the prevailing economic conditions. Many studies have pointed out that managers tend to employ REM rather than AEM to achieve target income since AEM is more likely to incur the scrutiny of auditors, regulators and others (Graham, Harvey, and Rajgopal, 2005; Roychowdhury, 2006; Cohen and Zarowin, 2010; Gunny, 2010). Kim, Lei, and Pevzner (2010) also argue that AEM may lead to litigation, SEC investigation, and criminal liability for managers as well as the scrutiny of auditors. Their research suggests that AEM may be costly when it is revealed. Meanwhile, Kothari, Mizik, and Roychowdhury (2012) state that REM is easier to camouflage as normal activities than AEM. Further, they argue that discretion, relating to 7 We can interpret their finding in relation to AEM, because their evidence is based on abnormal accruals. 8 Ipino and Parbonetti (2011) also show that a substitution between AEM and REM is detected when IFRS becomes mandatory. 8

11 operating and investment activities, is inherently given to managers by shareholders. From surveys and interviews with executives, Graham, Harvey, and Rajgopal (2005) present evidence suggesting that managers prefer REM because they fear overzealous regulators. Prior research has argued that managers may prefer REM to avoid the scrutiny and oversight of stakeholders, in spite of the higher cost to the firm in the future. Graham, Harvey, and Rajgopal (2005) and Cohen and Zarowin (2010) state that REM that reduces maintenance, advertising costs, and positive NPV investments may be more costly to firms than AEM as this behavior may not maximize the future value of the firm. Badertscher (2011) finds that managers compare the future cost of AEM and REM, and employ REM before AEM. One purpose of this paper is to clarify whether REM would be employed more in countries with higher financial development. If financial development heightens the monitoring and scrutiny of accounting figures, then AEM is less prevalent under high financial development, and more prevalent under low financial development. Does the manager, then, substitute AEM with REM, or avoid both AEM and REM in an environment of high financial development? As mentioned in the previous subsection, accounting information under higher financial development plays an important role in a range of decision making in financial markets and firms, as well as in the enforcement of many kinds of laws, regulations, and contracts. Relevant and reliable accounting information is typically required for investors to take risks and make decisions. Accounting standards can, therefore, be regarded as regulations for firms and their revision can result in managers engaging in discretionary behaviors. Ewert and Wagenhofer (2005) analytically show that tight accounting standards restrict managers discretion to manipulate accruals, leading them to prefer REM. In addition, Cohen, Dey, and Lys (2008) provide evidence that managers have shifted away from AEM to REM in the post Sarbanes Oxley Act (SOX) period. The passage of SOX can be seen as strengthening regulations, thus increasing the restraint on employing AEM and leading to an inducement to employ REM. Chi, Lisic, and Pevzner (2011) provide evidence that managers tend to avoid AEM under higher quality audit 9

12 conditions, which in turn leads to the employment of REM. The scrutiny of auditors, therefore, reduces AEM, but increases REM. The research provides evidence of a substitution between AEM and REM as AEM becomes increasingly constrained. In fact, Francis, Hasan, and Li (2011) conclude that managers in countries with stronger legal environment face a higher risk of litigation due to AEM and, therefore, resort to focusing on real activities. If we assume that financial development can be regarded as a type of institutional factor, analogous to investor protection, then it may lead managers to reduce AEM and shift to REM instead. However, Leuz, Nanda, and Wysocki (2003, 506) mention that strong and well enforced outsider rights limit insiders acquisition of private control benefits and, consequently, mitigate insiders incentives to manage accounting earnings because they have little to conceal from outsiders. The mitigation of managers incentives may reduce overall managerial discretionary behaviors, both AEM and REM. We can infer that as managers are disciplined by strengthening regulation and investor protections other than the legal system, they will avoid AEM and REM in the development of financial systems. Monitoring and scrutiny by stakeholders also discipline managers behavior. Wongsunwai (2013) shows that firms backed by higher quality venture capital do not appear to engage in both REM and AEM. This means that high quality venture capitalists monitor their portfolio companies closely and effectively and then behave rationally leading to a constraint of REM. In other words, sophisticated investors understand the future implications of REM and take into consideration in decision-making. Kim and Sohn (2013) find that cost of capital positively relates to REM, when compared with AEM. Based on this evidence, it appears that rational investors see through the adverse effects on future cash flow of REM. These studies also suggest that sophisticated market participants monitor managers value-destroying behavior and effectively restrain managers from earnings management by compelling them to keep stakeholders reactions in mind. What is the effect of financial development on AEM and REM? With higher financial 10

13 development, more accounting information is required. Financially sophisticated stakeholders pay close attention to the accounting numbers. Therefore, we hypothesize that managers tend to avoid AEM where there is higher financial development. For REM, our hypotheses are twofold. First, if the substitution effect between the two types of earnings management occurs, REM is restrained in countries with high financial development since AEM becomes more costly. Second, earnings management, both AEM and REM, reduces with high financial development, because managers are disciplined by higher financial development. In the next section, we provide the research design to test the hypotheses. 3. Research Design 3.1. Earnings management measures Accrual-based earnings management measure Following previous research (e.g., Warfield, Wild, and Wild, 1995; Becker, DeFond, Jiambalvo, and Subramanyam, 1998; Cohen, Dey, and Lys, 2008), we use the absolute value of abnormal accruals as an AEM measure in order to capture both the effects of income-increasing and income-decreasing AEM. To measure abnormal accruals, we use the cross-sectional modified Jones (1991) model (see DeFond and Jiambalvo, 1994; Dechow, Sloan, and Sweeney, 1995; Becker, DeFond, Jiambalvo, and Subramanyam, 1998). Specifically, we estimate the following regression model for each industry-year combination in each country, where industry is identified by a two-digit SIC code. ACC ijt / A ijt-1 = β 0 + β 1 (1 / A ijt-1 ) + β 1 ((ΔS ijt - ΔAR ijt ) / A ijt -1 ) + β 3 (PPE ijt / A ijt -1 ) + ε ijt (1) ACC is accruals that is calculated by net income minus operating cash flow reported in the statement of cash flow; A is the total assets; ΔS is the change in net sales; ΔAR is the change in accounts receivable; PPE is the net property, plant, and equipment; the subscripts refer to firm i, country j, and time t. Abnormal accruals are calculated as the estimated residuals from equation 11

14 (1), and its absolute value is our proxy for AEM ( A_ACC ) Real earnings management measures Following Roychowdhury (2006), Cohen, Dey, and Lys (2008), and Cohen and Zarowin (2010), we developed the proxy for three methods of REM: (1) sales manipulation, (2) reduction of discretionary expenses, and (3) overproduction. Sales manipulation is managers' behavior that tries to increase sales through price discounts or more lenient credit terms. As long as the margins are positive, the additional sales will increase earnings in the current period. However, both price discounts and more lenient credit terms lead to lower margins, resulting in lower cash flow from operations (CFO) and higher production costs compared with the sales. Reduction of discretionary expenses can also be an earnings management method. Managers can increase earnings by the reduction of discretionary expenses such as R&D and advertising costs. Reducing such expenses should lead to low discretionary expenses. Overproduction is the production of greater than expected demand in order to increase earnings. If managers engage in overproduction, fixed overhead costs are allocated to a larger number of units, thereby lowering fixed costs per unit. As long as the decrease in fixed costs per unit is not offset by any increase in marginal cost per unit, the total cost per unit decreases. As a result, this decreases the cost of goods sold and increases earnings. However, overproduction leads to higher production costs and lower CFO than normal production level given the sales, because of additional production and holding costs. In summary, sales manipulation and overproduction lead to abnormally high production costs relative to sales, and abnormally low cash flow from operating activities relative to sales, while the reduction of discretionary expenditures leads to abnormally low discretionary expenses (Roychowdhury 2006, ). 9 To measure the abnormal level of CFO (A_CFO), discretionary expenses (A_DE), and production costs (A_PD), we estimate the following regression models. Similar to equation (1), 9 If the firm paid for discretionary expenses in cash, reduction of discretionary expenses could also lead to abnormally high cash flow (Roychowdhury 2006; Cohen and Zarowin 2008). 12

15 the regression models are estimated for each industry-year combination in each country, where industry is identified by a two-digit SIC code. CFO ijt / A ijt-1 = β 0 + β 1 (1 / A ijt-1 ) + β 2 (S ijt / A ijt-1 ) + β 3 (ΔS ijt / A ijt-1 ) + ε ijt (2) DE ijt / A ijt-1 = β 0 + β 1 (1 / A ijt-1 ) + β 2 (S ijt-1 / A ijt-1 ) + ε ijt (3) PD ijt / A ijt-1 = β 0 + β 1 (1 / A ijt-1 ) + β 2 (S ijt / A ijt-1 ) + β 3 (ΔS ijt / A ijt-1 ) + β 4 (ΔS ijt-1 / A ijt-1 ) + ε ijt (4) CFO represents the operating cash flows reported in the statement of cash flows; DE represents the selling, general, and administrative expenses; PD represents production costs and is calculated as the cost of goods sold plus the change in inventory; and S represents the net sales. 10 A_CFO, A_DE, and A_PD are calculated as the estimated residuals from equations (2), (3), and (4), respectively. Since the three types of REM described above might be implemented to decrease earnings, consistent with Francis, Hasan, and Li (2011) and Kim and Sohn (2013), we convert A_CFO, A_DE, and A_PD to the absolute values and use them as our REM proxies ( A_CFO, A_DE, and A_PD, respectively). 11 In addition, we combine these three measures to capture the total effects of REM. Consistent with Cohen and Zarowin (2010), we multiply A_CFO and A_DE by negative one, and add them to A_PD in order that higher values indicate greater income-increasing earnings management. Again, considering the possibility of income-decreasing REM, we convert the aggregated REM measure to the absolute values and use it as our fourth REM proxy ( REM ). 10 Following Bartov and Cohen (2009) and Gunny (2010), we use selling, general, and administrative expenses as discretionary expenses because they frequently include discretionary expenses such as R&D and advertising costs. 11 For example, Francis, Hasan, and Li (2011) point out the possibility of income-decreasing real earnings management from an income-smoothing perspective. When a firm s performance is good in the current period, managers may choose to spend more on R&D, advertising, employee training, etc. These activities have an income-decreasing effect for the current year but an income-increasing effect for future periods (Francis, Hasan, and Li, 2011, 9). 13

16 3.2. Financial development measures We adopted the financial development score used by the World Economic Forum. The reason for this is that it takes a comprehensive view when assessing the factors that contribute to the long-term development of financial systems (The World Economic Forum, 2012, xiii). It also includes various factors that are used in prior cross-country research on financial accounting, such as corporate governance, legal and regulatory issues, and contract enforcement. The World Economic Forum has provided a score and rank for the breadth, depth, and efficiency of 62 of the world s leading financial systems and capital markets since The index analyzes drivers of financial system development that support economic growth, and thus compares the overall competitiveness of financial systems (The World Economic Forum, 2012, xiii). The World Economic Forum (2012, xiii) defines seven pillars as follows. (1) Institutional environment encompasses financial sector liberalization, corporate governance, legal and regulatory issues, and contract enforcement. (2) Business environment considers human capital, taxes, infrastructure, and costs of doing business. (3) Financial stability captures the risk of currency crises, systemic banking crises, and sovereign debt crises. (4) Banking financial services measure size, efficiency, and financial information disclosure. (5) Non-banking financial services include IPO and M&A activity, insurance, and securitization. (6) Financial markets encompass foreign exchange and derivatives markets and equity and bond market development. (7) Financial access evaluates commercial and retail access. A pillar is scaled from one to seven. The financial development score is the average of the scores relating to the seven pillars. Since all of the pillars cannot be associated with financial accounting (and resulting earnings management), we extracted the scores from four pillars out of the seven (Institutional environment, Banking financial services, Non-banking financial services, and Financial markets) as the factors most closely related to financial accounting. First, as is demonstrated above the definition of (1) Institutional environment defines overall financial development as a fundamental structure. This pillar includes the overall laws, regulations, and supervision of the financial sector, as well as the quality of contract 14

17 enforcement and corporate governance (The World Economic Forum 2012, 5). Financial accounting is essential for all of these to work efficiently. Leuz, Nanda, and Wysocki (2003) claim that strong investor rights and strong legal enforcement discipline managers, making earnings management appear to be lower. Next, (4) Banking financial services, (5) Non-banking financial services, and (6) Financial markets are chosen. It is self-evident that accounting information smooth transactions among firms and stakeholders within the framework of the pillars. A large number of studies have looked at earnings management in these areas. We, therefore, focus on the above four pillars in the main analysis that follows and our financial development score (FD) is the average score of these. 12 Unlike the four pillars, (2) Business environment, (3) Financial stability, and (7) Financial access can be considered to have an indirect linkage to financial accounting. (3) Financial stability is excluded because this pillar focuses on the financial crises that affect economic growth. In a similar vein, (7) Financial access is dropped because it is assumed that greater access to financial services is associated with the usual proxies for financial development and resulting economic growth. At first glance (2) Business environment, including taxation policy and the costs of doing business, seems to relate to financial accounting. However, judging from the definition of the indices that constitute the pillar, major parts of the index are remotely related to the necessity for financial accounting. It is, therefore, eliminated from our financial development score. The World Economic Forum (2012, 5) divides the seven pillars into three categories: Financial intermediation; Financial access; and Factors, policies, and institutions. Financial intermediation includes (4) Banking financial services, (5) Non-banking financial services, and (6) Financial markets. 13 Financial intermediation is defined as the variety, size, depth, and efficiency of the financial intermediaries and markets that provide financial services. Our 12 In the robustness checks in section 4.3, we test using the original financial development score. 13 Factors, policies, and institutions comprise the Institutional environment, Business environment, and Financial stability. It is defined as the foundational characteristics that allow the development of financial intermediaries, markets, instruments, and services. 15

18 financial development score can, therefore, be interpreted as the financial intermediaries and the institutional environment supporting them The models to test the hypotheses To examine the relationship between financial development and earnings management, we estimate the following regression model. EM ijt = β 0 + β 1 FD jt + β 2 Leverage ijt-1 + β 3 Size ijt-1 + β 4 MTB ijt-1 + β 5 ROA ijt + β 6 NOA ijt-1 + Σ β Year_Fixed_Effect + Σ β Firm_Fixed_Effect + ε ijt (5) EM represents the earnings management proxies, that is, either A_ACC, A_CFO, A_DE, A_PD or REM ; FD is the mean value of four pillars (Institutional environment, Banking financial Services, Non-banking financial services, and Financial markets) in the financial development report of the World Economic Forum; Leverage is total debt divided by total assets; Size is the natural logarithm of the market value of equity; MTB is the market to book ratio; ROA is the net income divided by lagged total assets; NOA is the net operating assets divided by sales. 14 When the dependent variable is the proxy for AEM (that is A_ACC ), we predict that the coefficient of FD has a negative sign. On the other hand, when the dependent variable is the proxy for REM (that is A_CFO, A_DE, A_PD, or REM ) the sign of the coefficient of FD should be negative (positive) if REM is more restrained (engaged) in countries with higher levels of financial development. In addition to FD, some variables are included to control other factors likely to affect the earnings management proxies. Leverage is included because previous research finds that it is related to earnings management measures (e.g., DeFond and Jiambalvo, 1994; Becker, DeFond, Jiambalvo, and Subramanyam, 1998; Roychowdhury, 2006). Following Gunny (2010), we include SIZE and MTB to control size effects and growth opportunity 14 Following Roychowdhury (2006), we use the values at the beginning of the year for Size and MTB, and use the values at the end of the year for ROA. 16

19 respectively. Depending on previous research, showing that earnings management measures are correlated with firm performance (e.g., Kothari, Leone, and Wasley, 2005; Cohen, Pandit, Wasley, and Zach, 2011), ROA is included as a control for firm performance. Following Barton and Simko (2002) and Zang (2012), we include NOA as a proxy for the extent of AEM in previous periods. Due to the limited flexibility within GAAP and the reversal of accruals, AEM in previous periods affects managers ability to manipulate accruals, with a consequent impact on REM (Zang, 2012). Finally, according to Degeorge, Ding, Jeanjean, and Stolowy (2013), Year_Fixed_Effect and Firm_Fixed_Effect are also included in our regression to control industry effects and year effects Sample Selection Financial development and relevant data, from 2009 to 2012, are obtained from Global Note. 15 The sample period is chosen because the Financial Development Score (FD), by the World Economic Forum, is available in Global Note from The countries in this paper are based on the 49 in La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1998). Ecuador, New Zealand, Taiwan, Uruguay, Sri Lanka, and Zimbabwe are dropped, as their FD scores are not included in the report of the World Economic Forum. Zimbabwe is also eliminated due to its experience of hyperinflation in the sample period. 16 The sample comprises data from Capital IQ, from which we obtained 81,317 pieces of firm-years data, covering sales and total assets of over 1 million dollars. Next, the data for financial services firms (2,108 firm-years) are eliminated. To calculate earnings management measures, we require at least six firm-year observations for each industry-year combination in each country (13,825 firm-years are excluded). To provide the condition and availability of the relevant measures that we need, Austria, Colombia, Egypt, Kenya, Portugal, and Venezuela (8,326 firm-years) are not included in our sample. Using this sample selection process, we 15 Global Note is the website that collects and provides various kinds of international statistics such as Gross Domestic Product. URL: (in Japanese). 16 We define hyperinflation as over 100% per year. 17

20 obtained 54,178 observations from 37 countries. 4. Empirical results 4.1. Descriptive statistics Panel A of Table 1 shows the number of firm-years in 37 countries, and the mean value of FD. The highest number of firm-years is for the United States. (10,772 observations, 19.8%) and Japan has a similar number. Observations for the United States and Japan occupy approximately 40% of the total. The lowest number of firm-years is Greece (31 observations). 17 The third column of Panel A reports the mean values of the financial development score (FD). FD is the mean value of the four pillars i.e., Institutional environment, Banking financial services, Non-banking financial services, and Financial markets. The United Kingdom has the highest value, 5.45, in the 37 countries, and the United States is second highest. The lowest score is 2.11 for Nigeria. Only the United Kingdom and the United States have scores exceeding five. 18 The ranking is similar to Beck and Levin (2002) and Degeorge, Ding, Jeanjean, and Stolowy (2013). 19 Panel B is the number of firm-years in the sample period. [Insert Table 1 here] Table 2 shows the descriptive statistics for the dependent and independent variables in equation (5). The mean values of A_ACC, A_CFO, A_DE, A_PD, and REM are 6.95%, 7.58%, 9.81%, 11.83%, and 22.74%, respectively. These values are slightly larger than Francis, Hassan, and Li (2011) and smaller than Kim and Sohn (2013) The number of firm-years depends on not only the number of listed firms in each country but also the coverage by Capital IQ. For example, India has more listed firms than Japan. 18 Original financial development scores range from the lowest value of 2.51 for Nigeria to the highest value of 5.17 for the United States, and show the same trends as our financial development score (FD). 19 Beck and Levin (2002) and Degeorge, Ding, Jeanjean, and Stolowy (2013) do not include the United Kingdom in their sample. 20 These studies used the absolute value of abnormal accrual and three REM measures. The former uses international data and the latter uses United States data. 18

21 [Insert Table 2 here] Table 3 is the correlation matrix. Since a high correlation coefficient is not observed, the results in the regressions from this section will not be influenced by multicollinearity. FD negatively associates with A_ACC, but positively correlates with A_CFO, A_DE, A_PD, and REM. [Insert Table 3 here] 4.2. Regression Results Table 4 reveals evidence of the influences of financial development on managerial behavior. 21 Five types of dependent variables in the analyses are provided. Column (1) in Table 4 displays the regression results of A_ACC. The coefficient of FD is significantly negative. This supports our prediction that AEM is restrained in countries with high financial development. When regarding financial development as an institutional factor, this result is consistent with Leuz, Nanda, and Wysocki (2003) and Boonlert-U-Thai, Meek, and Nabar (2006). Financial development serves as an institutional factor that inclines managers to avoid accrual-based discretionary behaviors to manage earnings. 22 [Insert Table 4 here] In the regression, where the dependent variable is A_CFO (column (2) of Table 4), FD also has a significant negative coefficient. Where the dependent variables are A_DE, A_PD, and REM (column (3), (4) and (5) respectively), the coefficients of FD are also significantly negative. From these results, it follows that financial development leads to REM reducing to 21 We winsorize all dependent and independent variable at the 1 percent and 99 percent levels. 22 Another reason for the small abacc of financially developed countries is that it is possible financial development affects accounting institutions. For example, stakeholders pursuing high accounting quality require tighter accounting standards in those countries. 19

22 similar levels as AEM. In economies that are relatively more financially developed, managers would reduce noise in earnings and avoid the decrease in future revenue caused by earnings management, fearing that stakeholders would detect these. Taken together, the results suggest that, as financial development leads stakeholders to focus on accounting numbers, managers may tend to avoid the costs incurred by earnings management. These include the scrutiny of auditors and regulators, litigation, a decline in future sales, and increasing cost of capital. The evidence supports our prediction that both types of earnings management are restrained (e.g., Wongsunwai, 2013). This is not consistent with prior research that shows substitution between AEM and REM (e.g. Francis, Hasan, and Li, 2011). Financial development will discipline managers and mitigate their incentives to engage in earnings management Additional tests FD is designed to capture financial development that affects accounting information. Therefore, FD is composed of the Institutional environment, Banking financial Services, Non-banking financial services, and Financial markets that are extracted as the factors closely relating to earnings quality in the original financial development score. However, the original financial development score has a further three components, in addition to the four mentioned above. These are Business environment, Financial stability, and Financial access. Hence, we provide two additional tests. One is the replacement of FD and another is the decomposition of it. First, FD is extended to include the three components. We replace FD with the original financial development score, Original_FD, and re-estimate equation (5). The coefficient of Original_FD in all regressions is significantly negative (not tabulated) and supports the results of Table 4. Next, FD is made the simple mean value of four components selected from the seven indicators of the financial development score. We are concerned with whether FD faithfully 20

23 represents a true financial development by adding the four components equally. 23 Hence, FD is divided into four components, namely, Institutional_Environment, Banking_Financial_Services, Non-banking_ Financial_Services, and Financial_Markets. The decomposition of FD makes it possible to evaluate the effect of each component on earnings management. Equation (6) includes each component of FD in the place of FD in equation (5). EM ijt = β 0 + β 1 Institutional_Environment jt + β 2 Banking_Financial_Services jt + β 3 Non-banking_Financial_Services jt + β 4 Financial_Markets jt + β 5 Leverage ijt-1 + β 6 Size ijt-1 + β 7 MTB ijt-1 + β 8 ROA ijt + β 9 NOA ijt-1 + Σ β Year_Fixed_Effect + Σ β Firm_Fixed_Effect + ε ijt (6) In column (1) of Table 5, we provide evidence of the effect of each financial development component. Among them, Non-banking_Financial_Services is negatively associated with the level of A_ACC and the absolute value of the coefficient is the largest. By comparison, Financial_ Markets has a significant positive value, but does not have a large effect on A_ACC. For the dependent variables of REM, we find that Institutional_Environment and Financial_Markets negatively affect the REM variable. 24 In particular, the significant negative value of the coefficients of Institutional Environment is not consistent with Francis, Hasan, and Li (2011) Original financial development score by the World Economic Forum is also the simple mean value of seven components. 24 As is well known, the development of a component works for the improvement of other components and entire financial development. Thus, as there is the issue of interdependency, the findings of Table 5 should be carefully interpreted. To mitigate any concern, we replace each component with FD. That is, the following regression is estimated. EM ijt = β 0 + β 1 Component of FD jt + β 2 Leverage ijt-1 + β 3 Size ijt-1 + β 4 MTB ijt-1 + β 5 ROA ijt + β 6 NOA ijt-1 +Σ βfirm_fixed_effect + ε ijt-1, Component of FD = Institutional_Environment, Banking_Financial_Services, Non-banking_Financial_Services, or Financial_Markets. Only the estimated coefficients of Non-banking_Financial_Services are similar to Table 5. None of the coefficients of Institutional_Environment are significant. The coefficients of Banking_Financial_Services are significantly positive only when the dependent variable is REM. In addition, the coefficients of Financial_Markets are consistent with Table 5 in the regression of abacc, abpd, and REM. 25 As described before, Institutional_Environment includes legal enforcement. 21