J Account Public Policy 28 (2009) 281–300 Contents lists available at ScienceDirect J Account Public Policy journal homepage: www.elsevier.com/locate/jaccpubpol Family control, board independence and earnings management: Evidence based on Hong Kong firms Bikki Jaggi a,b,*, Sidney Leung c, Ferdinand Gul d,e a Department of Accounting and Information Systems, School of Business, Rutgers University, Levin Building, Piscataway, NJ 08854, United States b The Hong Kong Polytechnic University, Hong Kong c Department of Accountancy, City University of Hong Kong, Hong Kong d School of Accounting and Finance, The Hong Kong Polytechnic University, Hong Kong e The Faculty of Business and Accountancy, University of Malaya, Kuala Lumpur, Malaysia a r t i c l e i n f o JEL classification: G32 G34 M41 Keywords: Corporate governance Earnings management Earnings quality Family ownership concentration Family board members a b s t r a c t In this study, we document that independent corporate boards of Hong Kong firms provide effective monitoring of earnings management, which suggests that despite differences in institutional environments, corporate board independence is important to ensure high-quality financial reporting The findings also show that the monitoring effectiveness of corporate boards is moderated in family-controlled firms, either through ownership concentration or the presence of family members on corporate boards The results based on firms reporting small earnings increases provide additional support for our finding that the monitoring effectiveness of independent corporate boards is moderated in family-controlled firms Ó 2009 Elsevier Inc All rights reserved Introduction Studies based on data for US and UK firms document that corporations with independent boards tend to have less earnings management (see Dechow and Dichev, 2002; Peasnell et al., 2000) The Sarbanes-Oxley Act (2002) especially emphasizes the need for corporate board independence to improve earnings quality by reducing earnings management Another strand of research indicates that the institutional arrangements of a country have a significant impact on the magnitude of earnings man- * Corresponding author Address: Department of Accounting and Information Systems, School of Business, Rutgers University, Levin Building, Piscataway, NJ 08854, United States Tel.: +1 732 445 3539; fax: +1 732 445 32091 E-mail addresses: jaggi@rbsmail.rutgers.edu, jaggi@business.rutgers.edu (B Jaggi) 0278-4254/$ - see front matter Ó 2009 Elsevier Inc All rights reserved doi:10.1016/j.jaccpubpol.2009.06.002 282 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 agement and earnings quality For example, Leuz et al (2003) document higher earnings management in countries with low investor protection Ball et al (2003) argue that institutional factors have a strong influence on the private benefits of control and managerial incentives for financial reporting Recently, some authors have argued that family ownership concentration also influences firm performance and earnings quality (e.g., Anderson and Reeb, 2004; Ali et al., 2007) We extend the existing research by investigating whether independent corporate boards provide effective monitoring of earnings management in firms operating in institutional environments which differ from those of US and UK firms More importantly, we evaluate whether the monitoring effectiveness of independent corporate boards is affected by the family ownership control or the appointment of family members on corporate boards We conduct analyses on the Hong Kong firms, which operate in an environment where family ownership control and appointment of family members to corporate boards are well documented Hong Kong has the third highest percentage of family ownership of listed companies in the region after Indonesia and Malaysia (SCMP, 2002), and the Hong Kong Society of Accountants (HKSA, 1997) documents that members of controlling families are routinely appointed to corporate boards On the other hand, Hong Kong regulations emphasize corporate board independence (HKSE, 2004) In this study, we first evaluate whether the negative association between board independence and earnings management that has been documented in the US and UK (see Dechow and Dichev, 2002; Peasnell et al., 2000) also holds for Hong Kong firms Given the markedly different ownership structure of Hong Kong-based firms, a separate study of this setting is warranted Furthermore, as the leading financial center in the region and one of the largest international financial centers, the Hong Kong financial market attracts local and international investors, which has created demand for high-quality earnings information Investors are interested to know whether board independence, which has received increased attention in the Hong Kong institutional environment, improves earnings quality We argue that despite differences in the institutional environments between Hong Kong and the US and the UK, independent corporate boards are likely to provide stricter monitoring of managerial behavior with respect to earnings management, which will lead to better earnings quality Thus, we expect earnings management to be low and earnings quality to be high in the Hong Kong firms with more independent corporate boards Second, we examine whether family ownership control or family members on corporate boards moderate the monitoring effectiveness of independent boards Two opposing arguments exist as to the effect of family control through ownership or the appointment of family members to corporate boards on the effective monitoring by boards of earnings management On one hand, family control may not result in higher earnings management because founding families will limit the ability of managers to manipulate earnings, and there will be less pressure on management to manage earnings to look good in the short term since the controlling family will have a long-term interest in the firm (e.g., Jiraporn and DaDalt, 2007) This argument is consistent with family control reducing the Type I agency problem (conflict between managers and shareholders) On the other hand, earnings management is higher in countries where family ownership concentration is higher because of weak investor protection (e.g., Leuz et al., 2003) and majority shareholder motivation to expropriate minority shareholders’ interests (e.g., Fan and Wong, 2002; Cheung et al., 2006) This second argument is consistent with firms operating in institutional environments in which the Type II agency problem (conflict between controlling shareholder and minority shareholders) is more common Thus, the overall effect of family control on earnings management depends on whether the Type I or Type II agency problem dominates In this study, we conjecture that the prevalence of family control in Hong Kong is likely to moderate the monitoring effectiveness of independent corporate boards The following three arguments are presented in support of this conjecture First, we argue that when insider ownership is high, the monitoring role of corporate boards decreases (Jensen and Meckling, 1976) Second, controlling families are more likely to appoint independent directors to seek expertise and advice on strategic direction rather than give them the responsibility of monitoring and controlling managerial activities (Anderson and Reeb, 2004; Johnson et al., 1996), thus weakening the monitoring effectiveness of independent directors Third, director independence is likely to be compromised when a family controls a firm either through ownership domination or appointing family members to the board, because family B Jaggi et al / J Account Public Policy 28 (2009) 281–300 283 members will have control over the appointment and reappointment of independent directors Independent directors are less likely to go against the wishes of controlling family members, especially when they sit on corporate boards We use the proportion of independent non-executive directors (INED) on corporate boards as a proxy for corporate board independence Following Anderson and Reeb (2003), we use fractional equity ownership of the family as a measure of ownership control concentration A cutoff of 20% ownership is used to identify firms with family control concentration We also conduct sensitivity analyses on different cut-off points for family ownership concentration Because of interlocking relationship among firms and insufficient disclosure in annual reports about director ownership via corporate pyramids, effective ultimate ownership and the ratio of family voting control over ultimate ownership are not determinable Therefore, we use the appointment of controlling family members on corporate boards as an additional proxy for family control We consider the board as family controlled when two or more members of the controlling family are present on the board on the assumption that two or more family members will exercise a significant influence on the board’s decisions We use two widely adopted proxies for earnings management: the magnitude of discretionary accruals, and the discretionary component of accrual quality (see, for example, Kothari et al., 2005; Francis et al., 2005) Discretionary accruals are measured on the basis of the performance-adjusted discretionary accruals model (PACDA), suggested by Ashbaugh et al (2003) The discretionary component of quality of accrual (AQ) is measured based on the model suggested by Francis et al (2005) A lower magnitude of discretionary accruals and a lower discretionary component of AQ reflect lower earnings management (or higher earnings quality) Our study is based on all firms traded on the Hong Kong Stock Exchange (HKSE) during the period 1998–2000 for which financial data are available on the Global Vantage database The final sample for PACDA analysis consists of 770 firm-year observations Because of non-availability of data, the AQ analysis is based on 309 observations.1 We conduct 2SLS regression with discretionary accruals (PACDA) and discretionary component of accrual quality (AQ) as dependent variables, and use the predicted value of ownership in the analyses to control for the potential endogeneity associated with family ownership We also conduct OLS regression analyses as a sensitivity analysis The 2SLS regression results, which control for the endogeneity problem, show a negative association between earnings management and board independence, suggesting that a higher proportion of INEDs on corporate boards results in more effective monitoring of earnings management that increases earnings quality The results based on family-controlled and non-family-controlled firms show that the negative association between board independence and earnings management is moderated by family ownership and control We test the robustness of our findings by examining whether family ownership and family board control have an impact on the effectiveness of independent boards to control earnings management when firms report small earnings increases The results of this test indicate that more independent boards are associated with a reduced likelihood of reporting small earnings increases The results on the impact of family control show that the negative association between earnings management and board independence is weak for family-controlled firms Sensitivity tests performed on different cut-off points for family ownership concentration for identifying family-controlled firms not change the result Overall, our findings indicate that independent boards tend to be effective in controlling earnings management only in non-family-controlled firms Board independence does not appear to improve earnings quality in family-controlled firms An alternative explanation for the results could be that independent directors and family control are substitutes for controlling earnings management.2 The findings of this study make the following contributions First, the results indicate that, on average, INEDs provide effective monitoring of earnings management in Hong Kong firms This finding suggests that strengthening the independence of boards by appointing more INEDs is a positive step toward improving earnings quality Second, the monitoring effectiveness of independent directors is moderated in family-controlled firms Third, given similarities in the business and institutional The number of AQ observations is much smaller than that of PACDA observations because of non-availability of observations for running the time-series regression model to obtain the residuals that are used as a proxy of AQ We thank a reviewer for suggesting this alternative explanation of our results 284 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 arrangements of Hong Kong and countries in East and Southeast Asia (e.g., Taiwan), the findings based on this study provide useful information for regulators in these countries Fourth, the results suggest the policy makers in Hong Kong and other Southeast Asian countries should be careful in borrowing rules/regulations from the US/UK because they may not work as effectively in a different institutional setting Finally, the findings also provide useful information to investors in evaluating the impact of board independence on earnings quality, especially in family-controlled firms The remainder of the paper is organized as follows: Section introduces the background of the study Section contains discussion on the development of hypotheses, and Section presents research design, including the sample selection procedures and research methodology Discussion of results is provided in Section 5, and Section presents the summary and conclusions Background of the study 2.1 Institutional environments of Hong Kong firms La Porta et al (1998) highlight the differences in the institutional and cultural environments among different countries Though Hong Kong’s legal framework is influenced by English common law, significant differences exist between the business environment of Hong Kong and the business environments of western industrialized countries, especially with regard to corporate governance, ownership and control The corporate governance structure of Hong Kong firms is characterized by a personal networking system (guanxi), which revolves around informal relationships rather than formal written contracts As a result, family ownership concentration in firms and the appointment of family members to corporate boards are common (e.g., Claessens et al., 2000; Mok et al., 1992) The 10 most prominent business families control 32.1% of all the corporate assets in Hong Kong (Tsui and Stott, 2004) Additionally, according to the 1994 statistics, family ownership of Hong Kong firms was worth about US$155 billion, representing 60% of total market capitalization (for example, see Weidenbaum and Hughes, 1996) As a result of family ownership concentration, market control mechanisms are weak in Hong Kong: hostile takeovers and mergers and acquisitions are almost non-existent Moreover, because of family ownership concentration, institutional shareholdings is not very common (Tsui and Stott, 2004) Hong Kong firms also differ from US firms with regard to corporate borrowings: private borrowing through banks rather than issuing public debt is more common in Hong Kong 2.2 Agency problems in the Hong Kong ownership structure Family-controlled firms are likely to face agency problems different from those of non-family-controlled firms The phenomenon of family ownership concentration results in two distinct groups of shareholders, i.e., majority and minority shareholders As a result of these two groups of shareholders, family-controlled firms are more likely to suffer from the Type II agency problem (conflict between majority and minority shareholders) than the Type I agency problem (conflict between managers and shareholders) (see Anderson and Reeb, 2004; Ali et al., 2007) Controlling shareholders have an opportunity to maximize their private benefits by expropriating minority shareholders (e.g., Fan and Wong, 2002; Cheung et al., 2006) Thus, some managerial actions in family-controlled firms may not be in the best interest of outside (minority) shareholders 2.3 Regulations on the presence of INEDs on corporate boards The Hong Kong Stock Exchange’s (HKSE) guidelines at the time of this study required that firms appoint at least two independent non-executive directors on corporate boards To strengthen corporate board independence, the HKSE appointed a committee to improve the Stock Exchange’s operations and strengthen the listing requirements so that corporate boards would assume greater responsibility and accountability in ensuring reliability of reported information The HKSE Committee (2004) recom- B Jaggi et al / J Account Public Policy 28 (2009) 281–300 285 mended that the number of independent directors on Hong Kong corporate boards be raised from two to three members effective for accounting periods starting on or after January 1, 2005 Research questions and hypotheses 3.1 Corporate board independence and earnings management The first research question addressed in this study is whether a higher degree of corporate board independence of Hong Kong firms, proxied by the proportion of INEDs on corporate boards, is associated with lower earnings management The impact of corporate board independence on monitoring effectiveness has been examined in prior studies Most of these studies emphasize the importance of independent corporate boards, proxied by a higher proportion of independent non-executive directors, to monitor managerial activities.3 Fama and Jensen (1983) emphasize the importance of corporate board independence to provide effective monitoring of managerial activities and initiatives Williamson (1981) argues that the independence of corporate boards is needed to protect investor interests Roe (1991) supports the monitoring role of corporate boards on the ground that managerial activities could not be targeted by legislative actions, and argued that effective monitoring by corporate boards prevent the abuse of powers by managers Some studies have examined empirically the impact of board independence on earnings management; these studies have primarily been based on data from the US and UK firms Dechow et al (1996) evaluate the causes and consequences of earnings manipulation based on firms subject to enforcement actions by the Securities and Exchange Commission (SEC) Their findings indicate that ‘‘the likelihood of earnings manipulation is systematically related to weaknesses in the oversight of management” (p 3), and they argue that firms with greater earnings manipulation are more likely to have a board dominated by insiders In another study, Beasley (1996) concludes that inclusion of a larger proportion of outside directors on boards reduces the likelihood of financial statement fraud In a recent study based on a small sample of US firms, Xie et al (2003) find a negative association between corporate board independence and discretionary accruals Peasnell et al (2000) find a similar result using a sample of UK firms for pre- and post-Cadbury periods Their findings show that in the post-Cadbury period, there is less income-increasing accrual management to avoid earnings losses or declines when the proportion of non-executive directors is high We extend the existing empirical research by evaluating whether the negative association between corporate board independence and earnings management is also valid for Hong Kong firms, which operate in a different institutional environment than US or UK Based on the conceptual arguments presented in the literature, we postulate that independent corporate boards provide effective monitoring of managerial behavior in Hong Kong Furthermore, in the absence of audit committees during the study period, responsibility for ensuring high-quality earnings information falls on the boards We test the following hypothesis, stated in the alternative form: H1: Higher independence of Hong Kong corporate boards, proxied by the percentage of INEDs on corporate boards, is associated with lower earnings management 3.2 Impact of family control on the association between board independence and earnings management Our second research question examines whether family control has a moderating effect on the monitoring effectiveness of independent boards There are two opposing theoretical viewpoints on the impact of family control on earnings management On one hand, earnings management is expected to be lower in family-controlled firms The results of studies based on US firms show that family firms are significantly less likely to manage earnings (Jiraporn and DaDalt, 2007; Ali et al., 2007; Wang, 2006) The following arguments are presented based on this finding First, controlling families Among others, these studies include those of Brickley et al (1994), CALPERS (1998), Cadbury (1992), Fama (1980), Dahya and McConnell (2005), American Law Institute (1982), Byrd and Hickman (1992), Mayers et al (1997), and Dalton et al (1998) 286 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 are expected to monitor managerial behavior and actions effectively, which will reduce managerial opportunities to engage in earnings management Second, in accordance with the stewardship theory, earnings are less likely to be manipulated because controlling families would identify their interests more closely with the firms’ wealth (e.g., Tosi and Gomez-Mejia, 1989) Third, there will be less pressure on management to meet short-term earnings expectations because controlling families focus more on the long term On the other hand, family control of a firm is likely to result in the Type II agency problems, i.e., a conflict of interests between majority and minority shareholders (e.g., Anderson and Reeb, 2004) In this case, the majority controlling shareholders may use earnings management to camouflage the reported earnings and hide expropriation from minority shareholders Some studies find that the owners of family-controlled firms extract private benefits at the cost of minority shareholders (e.g., Morck et al., 1988; DeAngelo and DeAngelo, 2000) Type II agency problems are more serious in the East Asia countries, where controlling family ownership is widespread, legal protection of minority shareholders is weaker, and financial reporting is less transparent (Fan and Wong, 2002; Ball et al., 2003) Empirical studies also document higher earnings management in countries with lower investor protection (e.g., Faccio et al., 2001; Leuz et al., 2003) These studies suggest that earnings management may be used to maximize the private benefits of majority shareholders.4 In view of the different expectations regarding the effect of family control on earnings management, it is an empirical question whether family control moderates the monitoring effectiveness of independent boards We conjecture that family control through family ownership concentration or appointment of family members to the board is likely to moderate the monitoring effectiveness of INEDs for the following reasons First, controlling families will appoint INEDs to seek their advice rather than giving them the responsibility to monitor managerial activities (Anderson and Reeb, 2004) Second, consistent with the Type II agency problem, controlling families will have a motivation to expropriate minority shareholders’ interest, and thus they will have an incentive to limit monitoring by INEDs they appoint Third, INEDs’ independence may also be compromised because of their closeness and loyalty to the controlling family that appoints or reappoints them to corporate boards We develop the following hypothesis to test this expectation: H2: The negative association between corporate board independence and earnings management is moderated in firms with family ownership control or corporate board control through family board members Sample selection and research methodology 4.1 Sample selection and data collection procedures We started the sample selection process by searching the Global Vantage database (CD dated December 2002) for Hong Kong firms for the three-year period from 1998 to 2000 The number of Hong Kong firms that have financial data in Global Vantage by year is as follows: 391 firms in 1998, 394 firms in 1999 and 399 firms in 2000 In the second step, we examined the Global Vantage database for financial data to estimate current discretionary accrual (PACDA) and total discretionary accruals (TDA) Because of the requirement of at least 10 observations in a two-digit SIC code per year, we could measure discretionary accruals for 943 of firm-years In the third step, we manually collected data on corporate governance and family control variables from the annual reports of sample firms for each study year As a result of missing reports or missing values in the available statements, the sample size was reduced to 876 firm-year observations As a result of some missing values for the control variables, the final sample consists of 770 firm-year observations The data for measuring the quality of accruals based on the model of Francis et al The controlling shareholders may also use other means to expropriate minority shareholders, such as selling assets, goods, or services to other companies under their control (e.g., Cheung et al., 2006) However, the Type II agency problem can come at a price to the controlling owners and their firms because investors can discount the share prices in response to the agency conflict (Claessens et al., 2000; La Porta et al., 2002) 287 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 Table Sample by industry and year Two-digit SIC group Year Total 1998 1999 2000 15 20 22 23 24–27 28–29 30–31 32–34 35 36 37–39 47 48 50–51 53 58 70 73 79–89 11 14 16 8 17 37 19 37 11 9 13 15 18 13 17 38 21 10 42 13 9 10 13 12 15 13 15 35 21 11 45 11 11 37 23 16 41 49 24 34 22 49 110 61 17 30 124 35 24 26 30 18 Total 244 269 257 770 (2005) were obtained from the PACAP databases.5 Because of the strict data requirement of the AQ estimation that is based on the Fama and French (1997) portfolio approach, which requires residual values for five years, the number of data observations was reduced to 309 for this variable The number of firm-year observations by SIC code and year is provided in Table The sample distribution indicates that there is no significant difference in the number of observations across the sample years The sample distribution by industry classification shows a higher number of observations for SIC code 36 (electronic and other electric equipment), SIC code 50 (wholesale trade – durable goods), and SIC code 51 (wholesale trade – nondurable goods) There is relatively a small number of observations for SIC codes 22 (textile mill products) and 47 (transport services) Distribution of our sample is similar to the industry distribution of Hong Kong firms in the database 4.2 Family board members and family ownership The ‘directors and management profile’ in the annual reports of Hong Kong firms provides a profile of each director According to the listing rules of the Hong Kong Companies Ordinance and Stock Exchange, companies are required to disclose the profile of all directors and senior management and their relationships, if any The annual reports also contain the shareholding information for each director, and disclosure of this information is done at three levels, namely personal interest, family interest and corporate and other interest Information contained in the annual reports enabled us to identify related family members on corporate boards Based on this information, we could identify the following relationships: father/mother and son/daughter, husband and wife, father/mother-in-law, son/daughter-in-law, brothers and sisters, nieces and nephews The board is defined as family controlled when two or more members of the controlling family are appointed as directors Consistent with the literature (e.g Ho and Wong, 2001), we find that controlling family members are routinely appointed as chairman or as executive director to control board decisions (Ho and Wong, 2001) We use the PACAP database for AQ estimation because it covers Hong Kong firms more extensively than does the Global Vantage database The PACAP databases are compiled by the Pacific-Basin Capital Markets Research Center at the University of Rhode Island College of Business, and provide historical research data for Hong Kong, Indonesia, Japan, Korea, Malaysia, Singapore, Taiwan and Thailand 288 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 Our analyses are based on data for 1999 Among 269 firms in 1999, 141 (52.4%) have family members representing family interest on the board, and 72 (51.1%) out of the 141 family-controlled firms have two family members serving as directors The number of firms with three family members on the board is 36 (25.5%), and the number with four or more members is 33 (23.4%) Additionally, we find that in 132 (93.6%) of the family-controlled firms, family members hold the position of Board Chairman, and an overwhelming majority of family board members (86.7% of firms) hold one of the key positions (i.e., chairman, CEO, or executive directors) Also, the family-controlled firms with three or more family members on the board occasionally appoint a family member as a non-executive director Additional analyses indicates that in half of the family-controlled firms (N = 70), a ‘father/mother and son/daughter’ relationship exists among family board members, which shows that family firms have a tendency to keep the business within the family over generations In 26.2% of the 141 family firms, husband and wife jointly represent the family interest on the board Descriptive statistics also indicate that controlling shareholders tend to have family members holding key managerial positions Information contained in annual reports also indicates families’ ultimate share ownership.6 We use a 20% cut-off point for family ultimate ownership control to identify the family-controlled firm.7 4.3 Corporate board independence The corporate board independence is measured by the proportion of INED on corporate boards Directors are considered to be INEDs if they not hold any executive position in the firm, have no relationship to the firm, and have no related-party transactions with the firm Thus, grey directors are excluded from the INED category 4.4 Calculation of discretionary accruals and accrual quality Our first proxy for earnings management is the magnitude of discretionary accruals We calculate both total and current discretionary accruals Total discretionary accruals (TDA) are calculated using the cross-sectional discretionary accruals model suggested by Jones (1991) and modified by Dechow et al (1995) Recently, some researchers have argued that current discretionary accruals are more susceptible to earnings manipulation (e.g., Ashbaugh et al., 2003), and others have argued that firm performance should also be considered in calculating discretionary accruals (Kothari et al., 2005) Our main results are based on the performance-adjusted current discretionary accruals (PACDA) model, which takes both of these factors into consideration, and TDA is used in a sensitivity test Details for estimation of PACDA and TDA are provided in Appendix A.8 As a second proxy, we use the quality of accruals, as suggested by Francis et al (2005) Because of estimation errors9 in calculating discretionary accruals, the quality of accruals has recently been suggested as an alternative proxy for earnings management (Francis et al., 2005) Dechow and Dichev (2002) argue that earnings quality is better if accruals are better associated with realized cash flows in the preceding, current and subsequent periods The model of Francis et al (2005) extends the Dechow and Dichev (2002) accrual quality model by adding two variables, namely the change in revenues (DRev), and property, plant and equipment (PPE), both of which are scaled by the average total assets McNichols Family ownership is calculated as the fractional ordinary shares held by family directors as the sum of beneficial interests at the personal, family and corporate levels It represents the ultimate voting control of the family in the firm Because the majority of footnotes associated with corporate interest of directors are unclear about the effective ownership of family members, we cannot measure the disparity between family voting control and ownership rights A similar ownership cutoff point for concentrated ownership has been used by Morck et al (1988) and by Hermalin and Weisbach (1991) Our results, however, are not sensitive to this cut-off Because the results for PACDA and TDA are qualitatively similar, the results on the magnitude of discretionary accruals are tabulated for PACDA only Discretionary accrual models of earnings management have limitations (see Erickson et al., 2004) The measurement of discretionary accruals is surrounded by the controversy whether discretionary accruals can be isolated from non-discretionary accruals with precisions (Guay et al., 1996) Furthermore, discretionary accruals may not always represent opportunistic earnings management They could be used to signal managers’ private information to investors (Dechow, 1994, p.5) B Jaggi et al / J Account Public Policy 28 (2009) 281–300 289 (2002) argues that in order to ‘‘extract” the estimation error in the form of a residual, it is necessary to control for these two variables The model is as follows: TCAi;t ẳ a0 ỵ a1 CFOi;t1 ỵ a2 CFOi;t ỵ a3 CFOi;tỵ1 ỵ a4 DRevi;t þ a5 PPEi;t þ ei;t ð1Þ where TCA is total current accruals, CFO is cash flow from operations, DRev is change in revenue and PPE is net property, plant and equipment.10 All variables are scaled by the average of total assets Eq (1) is estimated cross-sectionally for all firms (minimum of 10 firms) within each one of the 48 industry groups defined by Fama and French (1997) for each year AQ is defined as the standard deviation of the residual, ei,t for years t À to year t (a minimum of years firm residual data is required) A higher value of AQ means higher standard deviation, meaning higher variation in reported earnings, and this reflects lower earnings quality Francis et al (2005) decompose AQ into discretionary and innate components, where discretionary component is more prone to managerial manipulation We identify the discretionary component of AQ by separating it from the innate component Francis et al (2005) suggest the use of the following five factors to estimate the innate component of AQ: firm size, standard deviation of cash flow from operations (r(CFO)), standard deviation of sales (r(SALES), operating cycle, and incidence of losses We cannot calculate the operating cycle for Hong Kong firms because data on cost of sales (COGS) is not available during the study period As a result, we estimate the innate component of AQ by the following annual estimations: AQ it ẳ bi0 ỵ b1 SIZEit þ b2 rðCFOit Þ þ b3 rðSALESit Þ þ b4 NEGEARNit ỵ eit 2ị where SIZE is the log of total assets, r(CFOit) and r(SALESit) are the standard deviation of cash flow from operations and sales respectively, calculated over the past seven years, NEGEARN is the proportion of loss (negative earnings) years out of the past seven years We require at least four observations in the 7-year window.11 The discretionary component of AQ (DISC_AQ) is the residual eit from Eq (2) 4.5 Regression models We evaluate the association between the proportion of INEDs on corporate boards (PINED) and earnings management, after controlling for the impact of other relevant variables The use of control variables is based on their relevance to earnings management, as discussed in the literature The findings of Xie et al (2003) suggest that board size (BD_SIZE) is related to the extent of earnings management Fama and Jensen (1983) as well as the Cadbury Committee (1992) have argued that corporate boards would be more independent if the board chairman is independent of the firm’s chief executive officer (CEO) To highlight corporate board independence through PINED, we use board chairman/CEO duality as a control variable The variable CEO is coded when the board chairman and CEO positions are held by one individual, and otherwise Because some Hong Kong firms established audit committees voluntarily during the study period, we also include an indicator variable for audit committee in the model because it is likely to improve earnings quality (Klein, 2002) We use log of total assets (F_SIZE) as a control variable to control for firm size The impact of firm performance, firm growth and liquidity are controlled through the use of return on assets (ROA), market-to-book ratio (MB) and debt-to-equity-ratio (DE) respectively In addition, we include a dummy variable for Big5 auditors and we control for the effect of time periods by including year dummy variables The following model is used to evaluate the association between earnings management (EM) and PINED: EMitj ẳ a ỵ b1 PINEDit ỵ b2 BD SIZEit ỵ b3 CEOit ỵ b4 ACit ỵ b5 ROAit ỵ b6 DEit ỵ b7 BIG5it ỵ b8 F SIZEit ỵ b9 MBit ỵ b10 DUM YEAR99 ỵ b11 DUM YEAR00 ỵ eit 10 3ị Data on gross PPE are not available in the PACAP database As a result, we use net PPE in the estimation of Eq (1) We use a 7-year rolling window and require at least four observations in the past years, whereas Francis et al (2005) use a 10-year window and require at least five observations in the past 10 years We use a narrower window because the number of Hong Kong firms covered in the database is much smaller in the earlier years 11 290 N Mean Panel A: descriptive statistics of all variables PACDA 770 AQ 352 DISC_AQ 309 PINED 770 FAMOWN 770 NUMFAM 770 NBD_SIZE 770 ROA 770 DE 770 F_SIZE 770 MB 770 Std dev Minimum Median Maximum 0.117 0.066 À0.000 0.432 0.196 1.508 8.260 À0.079 0.297 4.866 1.116 0.175 0.043 0.038 0.177 0.246 1.708 2.394 0.528 1.133 1.285 2.283 0.000 0.004 À0.099 0.125 0.000 0.000 2.000 À7.975 0.000 0.579 À12.286 0.068 0.057 À0.004 0.400 0.000 2.000 8.000 0.015 0.166 4.765 0.583 2.141 0.294 0.208 1.000 0.932 8.000 20.000 3.366 23.452 9.092 23.194 Dichotomous variables CEO AC BIG5 DUM_YEAR99 DUM_YEAR00 384 (49.87%) 372 (48.31%) 43 (5.58%) 501 (65.06%) 513 (66.62%) 386 398 727 269 257 PACDA AQ DISC_AQ PINED Panel B: Pearson correlation coefficients between variables AQ 0.411** DISC_AQ 0.332** 0.893** PINED À0.105** À0.109* À0.039 À0.055 À0.033 À0.010 FAMOWN À0.071* FAMOWN NUMFAM BD_SIZE CEO AC ROA DE BIG5 F_SIZE MB (50.13%) (51.69%) (94.42%) (34.94%) (33.38%) DUM_ YEAR99 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 Table Descriptive statistics and correlations NUMFAM BD_SIZE CEO AC ROA DE BIG5 F_SIZE MB DUM_YEAR99 DUM_YEAR00 À0.115** À0.056 À0.009 0.029 À0.205** 0.038 0.018 À0.171** À0.025 0.012 0.042 À0.150** À0.004 À0.097 0.069 À0.134* 0.214** 0.041 À0.097 À0.077 0.019 0.131* À0.062 0.028 À0.050 À0.013 À0.115* 0.197** À0.016 À0.027 0.144** 0.000 0.000 0.025 0.152** À0.024 0.041 0.057 À0.028 0.032 0.185** 0.001 À0.004 0.004 0.721** À0.048 0.060 À0.024 0.023 0.007 À0.051 À0.077* À0.055 0.004 À0.024 0.143** 0.026 À0.049 0.053 0.009 À0.027 0.054 À0.074** 0.005 À0.020 À0.213** 0.104** 0.078* À0.040 0.030 0.313** À0.006 À0.013 0.020 À0.073* À0.092* 0.059 À0.005 À0.089* 0.033 À0.005 À0.038 0.077* À0.089* 0.104** 0.128** 0.042 0.170** 0.279** À0.567** 0.004 0.295** 0.007 À0.008 0.071* 0.002 À0.176** À0.069 0.060 À0.056 0.085* 0.055 0.048 0.052 À0.023 0.012 0.005 0.097** À0.035 À0.519** B Jaggi et al / J Account Public Policy 28 (2009) 281–300 PACDA = The absolute value of discretionary current accruals, scaled by lagged total assets AQ = Standard deviation of firm residuals, from years t À to t from annual cross-sectional estimations of the Francis et al.(2005) model DISC_AQ = Discretionary component of AQ PINED = Proportion of non-executive directors on the board of directors FAMOWN = The fraction of shares owned by family members on Corporate board NUMFAM = Number of directors from the same family on the board of directors BD_SIZE = Dummy variable: if the total number of directors on the board is greater than the median value of the sample; if the total number of directors on the board is smaller than or equal to the median value of the sample CEO = Dummy variable: if the CEO and the chairman of the board of directors are the same person and otherwise AC = Dummy variable: for presence of audit committee; otherwise ROA = Ratio of net income before extraordinary items to total assets DE = Ratio of total debt to total assets BIG5 = Dummy variable: for big5 auditor; for non-big5 auditor F_SIZE = Natural log of the total assets MB = Ratio of the firm’s market value of common equity to book value of common equity DUM_YEAR99 = Dummy variable: if year = 1999; if year = 1998 and 2000 DUM_YEAR00 = Dummy variable: if year = 2000; if year = 1998 and 1999 * Correlation is significant at the 0.05 level (two-tailed) ** Correlation is significant at the 0.01 level (two-tailed) 291 292 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 where EMit earnings management of firm i for period t by using different proxies for EM (i.e., PACDA and DISC_AQ) PINED proportion of independent non-executive directors on the board of directors12 BD_SIZE dummy variable: if the total number of directors on the board is greater than the median value of the sample; if the total number of directors on the board is smaller or equal to the median value of the sample CEO dummy variable: if the CEO and the chairman of the board of directors are the same person and otherwise AC dummy variable: for the presence of an audit committee; otherwise ROA ratio of net income before extraordinary items to total assets DE ratio of total debt to total assets BIG5 dummy variable: for big5 auditor; for non-big5 auditor F_SIZE natural log of total assets MB ratio of the firm’s market value of common equity to book value of common equity DUM_YEAR99 dummy variable: if year = 1999; if year = 1998 and 2000 DUM_YEAR00 dummy variable: if year = 2000; if year = 1998 and 1999 eit residual term We evaluate the impact of family control on the association between PINED and earnings management by including an interaction term between PINED and family control EMitj ẳ a ỵ b1 PINEDit þ b2 FAMitj þ b3 ðFAMitj à PINEDit Þ þ b4 BD SIZEit ỵ b5 CEOit ỵ b6 ACit ỵ b7 ROAit ỵ b8 DEit ỵ b9 BIG5it ỵ b10 F SIZEit ỵ b11 MBit ỵ b12 DUM YEAR99 ỵ b13 DUM YEAR00 ỵ eit 4ị where FAMit is family control of firm i for period t, and is measured as the fraction of shares owned by a family (FAM_OWN) We expect that b3 > Discussion of results 5.1 Descriptive statistics Descriptive statistics on the variables used in the regression tests are provided in Table The mean (median) of absolute values of performance-adjusted current discretionary accruals (PACDA) is 0.117 (0.068), and the mean and median values of discretionary AQ are À0.000 and À0.004 respectively On average, 43.2% of directors are INEDs This percentage shows that a significant number of board members are executive directors The median number of directors on corporate boards is The sample firms are almost equally divided between those that have and those that not have a dual board chairman/CEO (50.13% vs 49.87%) The mean and median numbers of directors from the same family are 1.508 and 2, respectively On average, 19.6% of outstanding ordinary shares are owned by one family The results (unreported) of annual regression of Eq (2) show that the coefficient estimates on r(CFO), r(SALES), and NEGEARN are in the expected direction, positive and significant for all three years (the mean coefficient estimates are 0.172, 0.030, and 0.032 respectively) The coefficient estimate on SIZE is positive, but insignificant (mean coefficient estimate is 0.0032) The explanatory power of the equation averages 15.7% across yearly estimations The operating performance (ROA) of sample firms varies significantly, and on average it is negative – possibly because of recessionary conditions during the period The average debt ratio is 29.7%, and 12 It is not uncommon that annual reports of Hong Kong firms in the sample years only label directors as either executive directors or non-executive directors without providing further classification whether they are independent non-executive directors We exclude those grey non-executive directors who have a relationship to the firm or related-party transactions from the INED category based on the information in the annual reports B Jaggi et al / J Account Public Policy 28 (2009) 281–300 293 the median value shows that 50% of the sample observations have less than 17% debt There is a significant variation in the market-to-book (MB) ratio; the median for this ratio is 58.3% The correlation matrix is provided in Panel B of Table The correlations indicate that the absolute value of PACDA is significantly negatively associated with PINED, P_FAM, D_LG_BD, BD_SIZE, F_SIZE, and ROA These results suggest that earnings management is lower when the proportion of INEDs on corporate boards is high, corporate board size and firm size are large, firm’s operating performance is high, and family members are present on corporate boards As expected, DISC_AQ is significantly positively associated with PACDA (0.294), indicating that firms with higher levels of discretionary accruals have lower accrual quality (high DISC_AQ values mean lower accrual quality) 5.2 Regression results 5.2.1 INEDs and earnings management We conduct regression tests to evaluate the association between independent corporate boards (PINED) and absolute values of performance-adjusted current discretionary accruals (PACDA) and discretionary component of AQ as proxies for EM (Eq (3)), The regression results are reported in Model of Table The results based on PACDA show that the coefficient for PINED is negative and statistically significant, suggesting that firms with a higher proportion of INEDs on corporate boards are associated with lower discretionary accruals, i.e., lower earnings management In other words, more independent Hong Kong corporate boards, proxied by a higher proportion of INEDs, are associated with lower magnitude of discretionary accruals The results on the discretionary component of AQ, however, show that the PINED coefficient is negative, but insignificant.13 The PINED coefficients in Eq (4) are significantly negative for PACDA and DISC-AQ (see Model in Table 3) These results support our hypothesis H1 that there is a negative association between independence of Hong Kong corporate boards and earnings management 5.2.2 Family control, INEDs and earnings management To evaluate the impact of family control on the association between earnings management and PINED, we include an interaction variable between PINED and family control (Eq (4)) Prior studies suggest that family ownership and firm performance may be endogenously determined (Anderson and Reeb, 2003; Demsetz and Lehn, 1985), and this may bias the results based on the OLS regression We address this problem by using an instrumental variable in the two-stage least squares (IV-2SLS) regression (e.g., Anderson and Reeb, 2003) Demsetz and Lehn (1985) suggest that ownership is a function of firm size and risk Following Anderson and Reeb (2003), we first regress family ownership (FAM_OWN) on the natural log of total assets, the square of the natural log of total assets and monthly stock return volatility (standard deviation of the 60 monthly stock returns in the previous five years) to obtain the predicted value of family ownership In the second stage, we use the predicted value of the family ownership to replace the actual family ownership The IV-2SLS regression results based on the predicted value of family ownership are presented in Table (Model 2) The results show that the coefficient of PINED is significantly negative for regression tests with PACDA or discretionary AQ as the dependent variables The interaction term on PINED * PFAM_OWN is positive and statistically significant at the 0.05 level for PACDA and the 0.001 level for discretionary AQ These results are consistent with our hypothesis H2, indicating that family control in Hong Kong firms moderates the negative association between independent corporate boards and earnings management These results suggest that higher independent corporate boards in non-family-controlled firms are more effective in controlling earnings management than in family-controlled firms 13 As a result of the requirement of a minimum of 10 observations in each industry (based on Fama and French, 1997) for each year for model estimation and a minimum of years residual values for the computation of the standard deviation (AQ measure), the sample size for this test is significantly reduce This test is based on 309 observations rather than 770 observations 294 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 Table Regression analysis of earning management, board independence and family control based on the predicted value of family ownership PACDA DISC_AQ Model Model Model Model Coeff t-Stat Coeff t-Stat Coeff t-Stat Coeff t-Stat Intercept PINED PFAM_OWN PINED * PFAM_OWN BD_SIZE CEO ROA DE BIG5 F_SIZE MB AC DUM_YEAR99 DUM_YEAR00 0.196*** À0.074*** 3.43 À2.01 1.00 À0.71 À0.17 À0.85 À1.15 À0.96 1.13 À1.48 À1.18 0.58 1.28 2.17 3.81 À2.10 À2.13 2.03 À0.83 À0.64 À2.29 À1.71 0.17 À0.86 À1.03 À0.20 2.09 1.56 0.015 À0.013 À0.002 À0.011 À0.081 À0.019 0.019 À0.016* À0.003 0.008 0.019* 0.026** 0.235*** À0.180** À0.105** 0.182** À0.010 À0.009 À0.149*** À0.017** 0.003 À0.008 À0.003 À0.003 0.034** 0.021* 0.003 À0.004 À0.004 0.024*** À0.004 À0.002 À0.003*** À0.000 0.003 0.002 0.63 À0.82 À0.72 2.65 À0.41 À1.00 À2.24 À0.09 0.62 0.45 0.033* À0.087*** À0.032** 0.100*** 0.003 À0.003 À0.010* 0.023*** À0.010 0.000 À0.003** À0.000 0.004 0.003 1.29 À3.03 À1.66 2.83 0.56 À0.63 À1.44 2.56 À0.93 0.04 À2.18 À0.10 0.82 0.68 N Adj R-SQ F-value 770 0.0657 5.92 627 0.1539 9.76 309 0.0341 1.99 290 0.0688 2.64 The reported t-statistics are white-adjusted (White, 1980) values to control for heteroskedasticity PACDA = The absolute value of discretionary current accruals, scaled by lagged total assets DISC_AQ = Discretionary component of AQ PFAM_OWN = Predicted value of family ownerships is the predicted value of regressing family ownership on the natural log of total assets, the square of the natural log of total assets, and return volatility (standard deviation of 60 month stock market rates of return for the previous years) PINED = Proportion of non-executive directors on the board of directors BD_SIZE = Dummy variable: if the total number of directors on the board is greater than the median value of the sample; if the total number of directors on the board is smaller than or equal to the median value of the sample CEO = Dummy variable: if the CEO and the chairman of the board of directors are the same person and otherwise ROA = Ratio of net income before extraordinary items to total assets DE = Ratio of total debt to total assets BIG5 = Dummy variable: for big5 auditor; for non-big5 auditor F_SIZE = Natural log of the total assets MB = Ratio of the firm’s market value of common equity to book value of common equity AC = Dummy variable: for presence of audit committee; otherwise DUM_YEAR99 = Dummy variable: if year = 1999; if year = 1998 and 2000 DUM_YEAR00 = Dummy variable: if year = 2000; if year = 1998 and 1999 * Designates statistical significance at the 0.1 level, one-tailed test ** Designates statistical significance at the 0.05 level, one-tailed test *** Designates statistical significance at the 0.01 level, one-tailed test As an additional analysis, we conduct regression tests on the sub-samples of family-controlled and non-family-controlled firms separately.14 The sample is divided into high and low family ownership control by the using the cut-off point of 20% ownership15 (40.5% of the sample with family ownership over 20% is defined as the high family ownership sub-sample) The sample is also divided into two groups based on the corporate board control by using two or more family members on the corporate board as the cut-off point The subsample results for PACDA (untabulated) show that the coefficients of PINED for 14 It has been argued that tests based on full sample with an interaction variable may be less precise under certain circumstances if the coefficients on the control variables differ between two groups An analysis based on pooled data forces the coefficients of all variables other than the test variables to be equal for the two groups It has been further suggested that separate regression tests on the two groups may provide better results when the association between the X variable (PINED) and Y variable (PACDA) is hypothesized to be contingent on the moderator variable Z (family control) (e.g., Staw and Oldham, 1978; Wright et al., 1996, p 452) 15 Morck et al (1988) and Hermalin and Weisbach (1991) use a similar ownership cutoff for concentrated ownership B Jaggi et al / J Account Public Policy 28 (2009) 281–300 295 both the family-controlled and non-family-controlled groups are negative, but the coefficient is significant only for non-family-controlled firms, and insignificant for family-controlled firms Similarly, the untabulated results for the discretionary component of AQ show that the PINED coefficient is negative and significant for non-family-controlled firms and is positive and insignificant for family-controlled firms Although the subsample analysis does not provide a direct test on the effect of family control, these results indirectly support that family control moderates the negative association between independent corporate boards, proxied by the proportion of INEDs, and earnings management 5.2.3 Discussion on control variables The results for control variables based on model of Table show that the CEO duality coefficient is negative, but it is statistically insignificant This finding thus indicates that the CEO duality does not have any significant impact on earnings management The results on firm size consistently show lower earnings management by large firms The results, however, not show any significant impact of Big-5 on earnings management The negative coefficient on ROA suggests good firm performance is associated with lower earnings management Results for DE are, however, mixed 5.2.4 Earnings benchmark test We also conduct an earnings benchmark test to evaluate whether family ownership control and family board control have an effect on the effectiveness of independent boards to control earnings management Specifically, we focus on the firms reporting small earnings increases (e.g., Degeorge et al., 1999; Ashbaugh et al., 2003) and examine weather family control weakens the negative association between board independence and the likelihood of small earnings increases We modify Ashbaugh et al.’s (2003) logistic model (p 630) for the earnings benchmark test as follows: INCREASEit ẳ b0 ỵ b1 PINEDit ỵ b2 FAMitj ỵ b3 FAMitj PINEDit ị þ b4 LITIGATIONit þ b5 MBit þ b6 lnMVEit þ b7 BIG5it ỵ b8 LOSSit ỵ b9 PACDAit ỵ eit ð5Þ where INCREASE represents small earnings increases and is coded when the difference between the current year and previous year’s net income, scaled by beginning-of-year market capitalization of equity, falls in the interval [0.00, 0.02], and otherwise We could not examine the impact of independent boards on managing earnings to meet or beat analyst earnings forecasts because analyst forecasts were not commonly available for the sample firms for the study years.16 FAMitj is family control of firm i for period t The value of j is equal to when family control is measured as the fraction of shares owned by a family (FAM_OWN) The value of FAM_OWN is equal to when family ownership is more than 20%, and otherwise The value of j is equal to when family control is measured as the family control on corporate boards (FAM_BD) The value of FAM_BD is equal to when two or more family members are on corporate boards, and otherwise We include PINED, FAM and their interaction in the model as the experimental variables We expect that b3 > Other variables are similar to those of Ashbaugh et al (2003) with two exceptions First, the audit fee variable is not included because audit fee data are not available on the database Instead, we include a BIG auditor variable (BIG5) to control for audit quality Second, institutional shareholding is not included because institutional shareholding in HK firms is generally small and such data are not publicly available LITIGATION is an indicator variable equal to if the company operates in a high-risk industry.17 MB is the market-to-book equity ratio for growth LnMVE is firm size, defined as the log of the market value of equity LOSS is an indicator variable equal to if the firm reports a net loss and PACDA is performanceadjusted discretionary current accruals Due to non-availability of beginning-of-year market values on the database in the INCREASE analysis, the sample is reduced to 593 observations, of which 70 observations report small earnings increases The results of the logistic regression are reported in Table The results clearly indicate 16 We examined whether analyst forecasts are available for the time period of this study Our search indicated that the IBES database contained less than 20% of the sample firms for the study’s time period 17 We find a negative and insignificant coefficient on LITIGATION The result is consistent with the institutional environment in Hong Kong that shareholder litigation is not an important issue 296 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 Table Logistic regression on meeting earning benchmark of small earnings increase INCREASEit ¼ b0 þ b1 PINEDit þ b2 FAMitj þ b3 ðFAMitj à PINEDit ị ỵ b4 LITIGATIONit ỵ b5 MBit ỵ b6 lnMVEit ỵ b7 BIG5it ỵ b8 LOSSit ỵ b9 PACDAit þ eit : Family ownership Family board control Coeff Intercept PINED FAM_OWN PINED à FAM_OWN FAM_BD PINED à FAM_BD LITIGATION MB LnMVE BIG5 LOSS PACDA N Pseudo R-SQ Chi-square stat Coeff À2.706 À1.763 À1.224 2.883 10.28*** 3.02** 2.61* 3.27** À2.602 À1.944 9.57*** 3.41** À1.360 3.126 À0.230 0.011 0.433 À0.088 À1.365 À0.732 593 0.1983 3.29** 3.84** 0.47 0.03 18.74*** 0.02 10.15*** 0.86 À0.182 0.011 0.433 À0.094 À1.342 À0.667 593 0.1966 0.30 0.04 18.39*** 0.02 9.81*** 0.71 Chi-square stat The reported t-statistics are white-adjusted (White, 1980) values to control for heteroskedasticity INCREASE = Dummy variable: when the difference between current year and previous year’s net income, scaled by MVE t À 1, falls in the interval [0, 0.02], and otherwise PINED = Proportion of non-executive directors on the board of directors FAM_OWN = Dummy variable: if the fraction of shares owned by family members on Corporate board is greater than 20%, otherwise FAM_BD = Dummy variable: if the family has two or more members on the broad of directors; otherwise PINED à FAM_OWN = Interaction term between PINED and FAM_OWN PINED à FAM_BD = Interaction term between PINED and FAM_BD LITIGATION = Dummy variable: if the firm operates within a high-litigation industry and otherwise, where high-litigation industries are industries with SIC codes of 2833–2836, 3570–3577, 3600–3674, 5200–5961, 7370–7374 MB = Ratio of the firm’s market value of common equity to book value of common equity LnMVE = Natural log of the firm’s market value of common equity BIG5 = Dummy variable: for big5 auditor; for non-big5 auditor LOSS = Dummy variable: if the firm reports a net loss, otherwise PACDA = Performance-adjusted discretionary current accruals, scaled by lagged total assets * Designates statistical significance at the 0.1 level, one-tailed test ** Designates statistical significance at the 0.05 level, one-tailed test *** Designates statistical significance at the 0.01 level, one-tailed test significantly negative coefficients on PINED and significantly positive coefficients on the interaction variable between PINED and FAM irrespective of whether FAM is represented by family ownership or family board control The evidence suggests that higher board independence reduces the likelihood of reporting small earnings increases, which is a potential outcome of earnings management The evidence also confirms the reduction in the INEDs’ monitoring effectiveness in family-controlled firms The findings are consistent with our earlier reported discretionary accruals and accrual-quality results 5.2.5 Additional robustness tests First, we perform sensitivity checks by varying the family ownership’s cut-off point from 20% to 25%, 30% and 50% The results are robust to different cut-off points in classifying the family- and non-family-controlled firms Second, we perform a robustness check to control for the effect of potential outliers We winsorize all variables that exceed four standard deviations of their respective means and rerun the regression tests The results based on these observations indicate that the results reported in the tables become stronger in some cases, which suggests that the results are not affected by outliers B Jaggi et al / J Account Public Policy 28 (2009) 281–300 297 Conclusion This paper evaluates the association between corporate board independence and earnings management in Hong Kong firms Additionally, it examines whether family control influences the association between INEDs and earnings quality Overall, the findings provide evidence that a higher proportion of INEDs is associated with more effective monitoring to constrain earnings management This suggests that a higher proportion of INEDs on corporate boards is likely to deter managers from manipulating the reported earnings; thus the quality of reported earnings of firms with a higher proportion of INEDs is expected to be high However, the monitoring effectiveness of INED’s is reduced in family-controlled firms, proxied by family ownership concentration or the presence of family members as board directors These results suggest that an increase in the proportion of outside directors to strengthen board monitoring is unlikely to be effective in family-controlled firms Like most studies of this nature, this study is subject to some limitations First, the validity of these findings depends upon discretionary accruals and accrual quality as proper proxies for earnings quality Second, although we have used two proxies for family control, the validity of the findings is also subject to proper estimation of family control of the firm Third, our results are for the period 1998– 2000, and caution should be exercised in extrapolating these results to more recent times Finally, the results may have been influenced by the control variable of ROA which is on average negative, indicating recessionary conditions during the study period Despite their inherent limitations, the findings provide useful insights to regulators for developing appropriate regulations on the corporate governance mechanism of Hong Kong firms, especially the appointment of independent non-executive directors on corporate boards The findings are relevant for countries with an institutional environment similar to that of Hong Kong Investors may also benefit from the findings because they provide insight into the impact of a higher proportion of INEDs on the reliability of reported earnings Last, but not least, the consideration of family board control adds to the corporate governance literature on the role of family ownership Acknowledgement Authors are grateful to the Research Grants Council of the Hong Kong Special Administrative Region, China for financial support (Project No CityU 1133/03H) Appendix A A.1 Calculation of performance-adjusted current discretionary accruals (PACDA) The cross-sectional performance-adjusted current discretionary accruals (PACDA) are calculated by including the lagged variable of ROA, as suggested by Kothari et al (2005) The PACDA are similar to RECDA calculated by Ashbaugh et al (2003) The parameters for calculation of expected current accruals ECA are estimated by using the following equation:     TCAit DREVit ẳ a0 ỵ a1 ỵ a2 ROAit1 ị ỵ eit ATitÀ1 ATitÀ1 ATitÀ1 ð1Þ The expected current accruals (ECA) use the estimated parameters as follows:     ECAit DREVit DARit ỵ a1 ỵ a2 ROAit1 ị ¼ a0 ATitÀ1 ATitÀ1 ATitÀ1 ð2Þ where TCA DREV DAR ROA AT eit total current accruals is net income (earnings before extraordinary items and discontinued operations) plus depreciation and amortization minus operating cash flows for firm i in the year t change in net revenue for firm i in the year t change in accounts receivable for firm i in the year t ratio of net income before extraordinary items to total assets for firm i in the year t À total assets for firm i in the year t error term for firm i in year t 298 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 Consistent with the models developed by Kothari et al (2005) and Ashbaugh et al (2003), current discretionary accruals are defined:  PACDA ¼ TCAit ECAit À ATitÀ1 ATitÀ1  The model is estimated separately for each combination of two-digit SIC code and year to obtain industry-specific estimates of the coefficients in Eq (1) A.2 Calculation of total discretionary accruals (TDA) According to the modified Jones model, the scaled total discretionary accruals are calculated as a difference between total accruals and non-discretionary accruals, scaled by total assets for the beginning period, whereas the total accruals (TA) are the difference between net income and cash flows from operations  Discretionary accruals TDAị ẳ TAit NDAit ATit1 ATit1  3ị The parameters for calculation of non-discretionary accruals (NDA) are estimated by using the following equation:       TAit DREVit PPEit ỵ a1 ỵ a2 ỵ eit ẳ a0 ATit1 ATit1 ATit1 ATit1 4ị The NDA are estimated based on the parameters obtained from Eq (4):       NDAit DREVit À DARit PPEit ỵ a1 ỵ a2 ẳ a0 ATit1 ATit1 ATitÀ1 ATitÀ1 ð5Þ where TA DREV DAR PPE AT eit total accruals, measured as the difference between net income (earnings before extraordinary items and discontinued operations) and operating cash flows for firm i in the year t change in net revenue for firm i in the year t change in accounts receivable for firm i in the year t property, plant and equipment for firm i in the year t total assets for firm i in the year t error term for firm i in year t The model is estimated separately for each of two-digit SIC code and year to obtain industry-specific estimates of the coefficients In case the number of observations for the two-digit SIC is less than 8, we combine the two-digit SIC codes It should also be noted that, as suggested in the literature, change in accounts receivable (DAR) is not included in estimating the parameters, but is included in estimating non-discretionary accruals (e.g., Ashbaugh et al., 2003) References Ali, A., Chen, T., Radhakrishnan, S., 2007 Corporate disclosures by family firms Journal of Accounting and Economics 44, 238– 286 American Law Institute, 1982 Principles of Corporate Governance and Structure: Restatement and Recommendations Tentative draft no 1, American Law Institute, Philadelphia, PA Anderson, R., Reeb, D., 2003 Founding-family ownership and firm performance Evidence from the S&P 500 The Journal of Finance 58 (3), 1301–1328 Anderson, R., Reeb, D., 2004 Board composition: Balancing family influence in S&P 500 firms Administrative Science Quarterly 49 (2), 209–237 Ashbaugh, H., LaFond, R., Mayhew, B., 2003 Do nonaudit services compromise auditor independence? Further evidence The Accounting Review 78 (3), 611–639 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 299 Ball, R., Robin, A., Wu, J., 2003 Incentives versus standards: Properties of accounting income in four East Asian countries Journal of Accounting and Economics 36, 235–270 Beasley, M., 1996 An empirical analysis of the relation between the board of director composition and financial statement fraud The Accounting Review 71 (4), 443–464 Brickley, J.A., Coles, J.L., Terry, R.L., 1994 The board of directors and the enactment of poison pills Journal of Financial Economics 35, 371–390 Byrd, J., Hickman, K., 1992 Do outside directors monitor managers? Evidence from tender offer bids Journal of Financial Economics 32, 195–221 Cadbury Committee Report, 1992 Report of the Cadbury Committee on the Financial Aspects of Corporate Governance (GCC, London) California Public Employees’ Retirement System (CALPERS), 1998 Corporate Governance: Core Principles and Guidelines Cheung, Y.L., Rau, P.R., Stouraitis, A., 2006 Tunneling, propping, and expropriation: Evidence from connected party transactions in Hong Kong Journal of Financial Economics 82, 341–386 Claessens, S., Djankov, S., Lang, L., 2000 The separation of ownership and control in East Asian Corporations Journal of Financial Economics 58, 81–112 Dahya, J., McConnell, J., 2005 Outside directors and corporate board decisions Journal of Corporate Finance 11 (1–2), 37–60 Dalton, D., Daily, C., Johnson, J., Ellstrand, A., 1998 Meta-analytic reviews of board composition, leadership structure, and financial performance Strategic Management Journal 19 (3), 269–290 DeAngelo, H., DeAngelo, L., 2000 Controlling stockholders and the disciplinary role of corporate payout policy: A study of the time mirror company Journal of Financial Economics 56 (2), 153–207 Dechow, P.M., 1994 Accounting earnings and cash flows as measures of firm performance: The role of accounting accruals Journal of Accounting and Economics 18, 3–42 Dechow, P.M., Dichev, I., 2002 The quality of accruals and earnings: The role of accrual estimation errors The Accounting Review 77, 35–60 Dechow, P., Sloan, R., Sweeney, A., 1995 Detecting earnings management The Accounting Review 70 (April), 193–225 Dechow, P., Sloan, R., Sweeney, A., 1996 Causes and consequences of earnings manipulation: An analysis of firms subject to enforcement actions by the SEC Contemporary Accounting Research 13 (1), 1–36 Degeorge, F., Patel, J., Zeckhauser, R., 1999 Earnings management to exceed thresholds The Journal of Business 72 (1), 1–33 Demsetz, H., Lehn, K., 1985 The structure of corporate ownership: Causes and consequences Journal of Political Economy 93, 1155–1177 Erickson, M., Hanlon, M., Maydew, E., 2004 How much will firms pay for earnings that not exist? Evidence of taxes paid on allegedly fraudulent earnings The Accounting Review 79 (2), 387–408 Faccio, M., Lang, L., Young, L., 2001 Dividends and expropriation American Economic Review 91 (1), 54–78 Fama, E., 1980 Agency problems and the theory of the firm Journal of Political Economics 88, 288–307 Fama, E., French, D., 1997 Industry costs of equity Journal of Financial Economics 43, 153–193 Fama, E., Jensen, M.C., 1983 Separation of ownership and control Journal of Law and Economics 26, 301–325 Fan, J., Wong, T.J., 2002 Corporate ownership structure and the informativeness of accounting earnings in East Asia Journal of Accounting and Economics 33, 401–425 Francis, J., LaFond, R., Olsson, P., Schipper, K., 2005 The market pricing of accruals quality Journal of Accounting and Economics 39, 295–327 Guay, W., Kothari, S.P., Watts, R., 1996 A market-based evaluation of discretionary accrual models Journal of Accounting Research 34, 83–115 Hermalin, B., Weisbach, M., 1991 The effects of board composition and direct incentives on firm performance Financial Management 20, 101–112 Hong Kong Society of Accountants (HKSA), 1997 Second Report of the Corporate Governance Working Group Hong Kong Exchange and Clearing Limited (HKSE), 2004 Draft Code on Corporate Governance Practices and Corporate Governance Report, January 2004 Ho, S., Wong, K., 2001 A study of the relationship between corporate governance structures and the extent of voluntary disclosure Journal of International Accounting, Auditing and Taxation 10, 139–156 Jensen, M., Meckling, W., 1976 Theory of the firm: Managerial behavior, agency cost and ownership structure Journal of Financial Economics 3, 305–360 Jiraporn, P., Dadalt, P., 2007 Does founding family control affect earnings management? An empirical note Working paper, SSRN abstract_id=1017856 Johnson, J., Daily, C., Ellstrand, A., 1996 Board of directors: A review and research agenda Journal of Management 22, 409–438 Jones, J., 1991 Earnings management during import relief investigations Journal of Accounting Research 29 (Autumn), 193– 228 Klein, A., 2002 Audit committee, board of director characteristics, and earnings management Journal of Accounting and Economics 33 (3), 375–400 Kothari, S., Leone, A., Wasley, C., 2005 Performance matched discretionary accrual measures Journal of Accounting and Economics 39, 163–197 La Porta, R., Lopez-de-Silanes, F.L., Shleifer, A., Vishny, R.W., 1998 Law and finance Journal of Political Economy 106, 1113– 1155 La Porta, R., Lopez-de-Silanes, F.L., Shleifer, A., Vishny, R.W., 2002 Investor protection and corporate valuation Journal of Finance 57 (3), 1147–1170 Leuz, C., Nanda, D., Wysocki, P.D., 2003 Earnings management and investor protection: An international comparison Journal of Financial Economics 69, 505–527 Mayers, D., Shivdasani, A., Smith, C., 1997 Board composition and corporate control: Evidence from the insurance industry Journal of Business 70, 33–62 McNichols, M., 2002 Discussion of the quality of accruals and earnings: The role of accrual estimation errors The Accounting Review 77 (Suppl.), 61–69 300 B Jaggi et al / J Account Public Policy 28 (2009) 281–300 Mok, H., Lam, K., Cheung, I., 1992 Family control and return covariation in Hong Kong’s common stocks Journal of Business Finance and Accounting 19 (2), 277–293 Morck, R., Shleifer, A., Vishny, R., 1988 Management ownership and market valuation: An empirical analysis Journal of Financial Economics 20, 293–315 Peasnell, K., Pope, P., Young, S., 2000 Accrual management to meet earnings targets: UK evidence pre- and post-Cadbury The British Accounting Review 32 (4), 415–445 Roe, M., 1991 A political theory of American corporate finance Columbia Law Review 91 (1), 10–67 Sarbanes-Oxley Act of 2002, Pub L No 107–204, 2002 South China Morning Post (SCMP), 2002 August 28, 2002, p (ProQuest ID: 155923101) Staw, B.M., Oldham, G., 1978 Reconsidering our dependent variables – A critique and empirical study Academy of Management Journal 21, 539–559 Tosi, H., Gomez-Mejia, L., 1989 The decoupling of CEO pay and performance An agency theory perspective Administrative Science Quarterly 34, 169–189 Tsui, J., Stott, V., 2004 Governance in family-owned Hong Kong corporations In: Ferdinand, A.G., Tsui, J (Eds.), The Governance of East Asian Corporations: Post Asian Financial Crisis Palgrave Macmillan Wang, D., 2006 Founding family ownership and earnings quality Journal of Accounting Research 44 (3), 619–656 Weidenbaum, M., Hughes, S., 1996 The Bamboo Network Martin Kessier Books, The Free Press, New York White, H., 1980 A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity Econometrica 48, 817–838 Williamson, O., 1981 The modern corporation: Origins, evolution, attributes Journal of Economic Literature 91 (14), 1537– 1568 Wright, P., Ferris, S., Sarin, A., Awasthi, V., 1996 Impact of corporate insider, blockholder, and institutional equity ownership on firm risk taking Academy of Management Journal 39 (2), 441–463 Xie, B., Davidson III, W.N., DaDalt, P.J., 2003 Earnings management and corporate governance: The roles of the board and the audit committee Journal of Corporate Finance 9, 295–316  ... increased attention in the Hong Kong institutional environment, improves earnings quality We argue that despite differences in the institutional environments between Hong Kong and the US and the UK,... proportion of INEDs on corporate boards results in more effective monitoring of earnings management that increases earnings quality The results based on family- controlled and non -family- controlled... 281–300 297 Conclusion This paper evaluates the association between corporate board independence and earnings management in Hong Kong firms Additionally, it examines whether family control influences