|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 002020934
007 cr mnu|||uuuuu
008 090723s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002494
Two essays on corporate governance are local directors better monitors, and directors incentives and earnings management
h [electronic resource] /
by Hong Wan.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 80 pages.
Dissertation (Ph.D.)--University of South Florida, 2008.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
ABSTRACT: Previous literature have documented that the independent directors play a crucial goal in corporate governance but the research on the firm value and board independence remains inconclusive. In my dissertation, I examine the impact of independent directors' geographic proximity to corporate headquarters on the effectiveness of corporate boards and the motivations of board directors. Using a large sample of directors trading, I show that independent directors who live close to headquarters ("local director") earn higher abnormal returns on their trades than other directors, and that this advantage is stronger in small firms. Further, I find an inverse relationship between the number of local independent directors on the board and firm value. Companies with fewer local independent directors also have higher ROA ratios, lower abnormal CEO compensations, and higher CEO incentive compensations. Collectively, the findings suggest that local independent directors are more informed but less effective monitors. I also provided evidence that firms with a higher proportion of directors' incentive compensation are more likely to manage earnings. Directors are more likely to exercise options in the year following the firms' earnings management being in the top tercile of the sample. The results are robust after controlling for self-selection bias. Taken together, the evidence suggests that director incentive pay is more likely to align directors' interest with the CEO's, rather than to induce the directors to act in the best interest of the shareholders.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Advisor: Chris Pantzalis, Ph.D.
t USF Electronic Theses and Dissertations.
Two Essays on Corporate Governance Are Local Directors Better Monitors, and Directors Incentives and Earnings Management by Hong Wan A dissertation submitted in partial fulfillment of the requirements for the degree of Doctoral of Philosophy Department of Finance College of Business University of South Florida Major Professor: Christos Pantzalis, Ph.D. Delroy Hunter, Ph.D. Ninon Sutton, Ph.D. Jianping Qi, Ph.D. Date of Approval: May 20 2008 Keywords: Distance, CEO compensation, Insider trading, Agency problem, Director compensation Copyright 2008, Hong Wan
Dedication This is dedicated to my beloved pare nts, Renyi Wan and Fenge Zhang, my wife, Yingxia, and my children, Richard and Serena, for their everlasting support and love.
Acknowledgments I want to thank my committee, Dr. Chri s Pantzalis, Dr. Delroy Hunter, Dr. Ninon Sutton and Dr. Jianping Qi for their invaluab le support and insightful advices during the process. I can not find a right word to express my appreciation to their help. Without my committee, and especially my advisor, Dr. Pant zalis, there will have no dissertation of mine. I also want to thank Susan Albring, Dong Bai, Alex Butler, Lily Qiu, Lei Wedge, Tina Yang, Ken Weiand for their helpful comme nts and insightful discussions. I am also grateful to all friends in the finance PhD program at the University of South Florida, for their support and friendship during last five years. Last, my thanks go to my wife, Yingxia, fo r what she has done to the family while I am writing the essays for the dissertation a nd my son, Richard, and my daughter, Serena, whose gentle hearts remind me how joyful a nd beautiful my life is while I am writing the dissertation. This dissertation is written for them, for their unconditional support and everlasting love.
i Table of Contents List of Tables iii Abstract vi Essay 1 Are Local Directors Better Monitors Introduction 1 Geography and Director Independence: Background Distance and information 7 Distance and monitoring 8 Director locality and CEO compensation 9 Data and Variable Construction 10 Sample selection 10 Variables 11 Empirical Results 18 Directors trades return: local vs. non-local 18 Multivariate regression tests: director locality and trade return 20 Firm value and director locality 25 Director locality and ROA 31 Director locality and CEO compensation 34 Robustness Tests 38 Conclusion 39 Essay 2 Director Incentives and Earnings Management Introduction 41 Literature Review 45 Earnings Management 45 Director Incentive 47 Data and Variables 49 Sample description 49 Measures of director incentive 50 Measures of earnings management 50 Measures of options sale 52 Descriptive statistics 53 Empirical Results Univariate analysis 57 Regressing earnings management on director incentives 58
ii Board characteristics and earnings management 60 CEO incentives and earnings management 62 Controlling self selection biases 64 Earnings management and director option sale 68 Additional robustness tests 69 Conclusion 72 References 74 About the Author End Page
iii List of Tables Table 1 Distribution of director by distance 12 Table 2 Descriptive statistics 15 Table 3 Trading performance of directors: local vs. non-local 19 Table 4 Trading performance of directors by distance and firms information characteristics 20 Table 5 Independent directors trades return: local vs. non-local 24 Table 6 Independent directors trades return: local versus non-local after controlling for asymmetry information 25 Table 7 Fixed effect regression of Q on director locality characteristics 28 Table 8 Two Stage IV regression of Q on director locality characteristics 30 Table 9 Fixed effect regression of ROA on director locality characteristics 32 Table 10 Two Stage IV regression of ROA on director locality characteristics 33 Table 11 Regressing CEO total compensation on director locality characteristics 36 Table 12 Regressing CEO incentives on directors characteristics 37 Table 13 Descriptive statistics of earnings management and incentive measure by years 54 Table 14 Summary statistics and correlations of earnings management and incentive measures 56 Table 15 Univariate analysis 57 Table 16 Director incentives and earnings management 59 Table 17 Director incentives and earnings management: controlling for board characteristics 61 Table 18 Univariate test: directors incentives and earnings management: Controlling for CEO incentives 62
iv Table 19 Directors incentives and earnings management: Controlling for CEO incentives 64 Table 20 Earnings management for incentive vs. no incentive paid companies 65 Table 21 Earnings management for incentive vs. no incentive paid companies: self selection analysis 67 Table 22 Earnings management and insider sell 69 Table 23 Cumulative change of director incentives and earnings management 70 Table 24 Director incentives and earnings management: sub-samples 71 Table 25 Directors incentives and earnings management: alternative measures 72
v Two Essays on Corporate Governance Are Local Directors Better Monitors, and Directors Incentives and Earnings Management Hong Wan ABSTRACT Previous literature have documented that the independent directors play a crucial goal in corporate governance but the research on the firm value and board independence remains inconclusive. In my dissertation, I examine the impact of independent directors geographic proximity to corpor ate headquarters on the effectiveness of corporate boards and the motivations of board directors. Using a large sample of directors trading, I show that independent directors who live close to headquarters (local director) earn higher abnormal returns on their trades than other directors, and that this advantage is stronger in small firms. Further, I find an inverse relationship between the number of local independent directors on the board and firm value. Companies with fewe r local independent dir ectors also have higher ROA ratios, lower abnormal CEO compensations and higher CEO incentive compensations. Collectively, the findings suggest that local inde pendent directors are mo re informed but less effective monitors. I also provided evidence th at firms with a higher proportion of directors incentive compensation are more likely to mana ge earnings. Directors are more likely to exercise options in the year following the firms earnings management being in the top
vi tercile of the sample. The results are robust after controlling for self-selection bias. Taken together, the evidence suggests that director in centive pay is more likely to align directors interest with the CEOs, rather than to induce th e directors to act in the best interest of the shareholders.
1 Essay 1 Are Local Directors Better Monitors Introduction Board independence is an important mechanism of corporate governance. Independent directors are elected to oversee the managers and act in the best interests of stockholders. Fama (1980) and Fama and Jensen (1983) focus on the outside directors role as monitors and emphasize that independent directors have incentives to build their reputation as expert monitors. Since then, a series of corporate scandals have led to changes in laws and regulations that are aimed at enhancing board independence. According to the National Association of Corporate Directors, in 2005, 83% of boards consisted of a majority of independent directors, up from 54% in 2000. However, in spite of the crucial function of independent directors as monitors, the evidence on the relationship between board independence and firm value remains inconclusive. For example, Aggrawal and Willamson (2006) find that the fr action of independent directors on the board has a positive impact on firms Q. However, Hermalin and Wissbach (1991) show an insignificant relationship between board i ndependence and firm performance, while Yermack (1996), Klein (1998), and Bhagat and Black (2002) find a negative relationship. Two possible reasons for the mixed evidence have been suggested in the literature. First, it is unclear what constitutes director independence. The literature defines independent directors by their affiliation. However, Hermalin and Weisbach (1998) and others argue that being unaffiliated does not necessarily mean being independent. Second, while the boards function is to monitor the CEOs, the CEO most always determines the agenda of board meetings and the information given to the board (Jensen (1993)). Consequently he may be
2 able to keep the board in the dark whenever he wants to make decisions that may harm the shareholders. Adams and Ferreira (2007) also present a model wherein the CEO is less likely to share information with directors when the board is monitoring intensively. To shed new light on this issue, I examin e board effectiveness by considering a new dimension of board independence: proxim ity of board members to the corporate headquarters, and hence to the CEOs. My primar y focus in this paper is on the relationship between geographic proximity of independent directors and board effectiveness. I propose two contending hypotheses to explain this a ssociation. The first proposes that board effectiveness rises with the pr oportion of the board that is made up of local independent directors (local director). I call this the e fficiency hypothesis. An alternative hypothesis is that local directors, because of their proxim ity to the firms CEO and other executives, may be less effective monitors. In other words, mo re geographically proximate directors may be less objective and hence reduces board effectivene ss. I call this the entrenchment hypothesis. In support of the first hypothesis, many st udies have documented that geographic proximity is an important component of m onitoring activities. For example, geographic proximity determines the effectiveness of in ternal control mechanisms within bank holding companies (Berger and DeYoung (2001, 2002)), and the representation of venture capitalists on the boards of U.S. private firms (Lerner (1995 )). Physical closeness to the firm further influences activities of equity analyst and auditors. Malloy (2005) shows that equity analysts forecast local stocks more accurate ly and that their forecast revisions for the local stock have a strong effect on the market, suggesting that local equity analysts have information advantage over distant analysts. Moreover, Malloy (2005) documents that the underwriter affiliated analyst biases are only observed for the distant affiliated analysts. Choi et al. (2007) find that local auditors provide higher quality auditing services while charging lower
3 auditing fees. Like in the case of external mon itors, such as banks, aud itors and analysts, the director locality or geographic proximity to the fi rm is more likely to be associated with an information advantage that could lead to a more effective board. Local directors may, for example observe more firm operations, be aware of more news about the firm in the local media, have first-hand knowledge about the firm from their local sources, or have a better understanding of the local industry conditions. On the other hand, local directors may be w eak monitors. Proximity to the firms is more likely to be associated with more persona l ties with the firm mana gers. This is because both CEO and directors are, for example, likel y to serve on other local boards, charitable institutions and to attend the same country clubs. These interactions may lead to the development of social and/or personal bonds between the directors and CEOs and may compromise local directors objectivity. Consequently, one can argue that the pres ence of local independent directors on the board could make the board either more eff ective or more entrenched. The information advantage, if available, would allow the lo cal independent directors to monitor more effectively, and thus lead to better firm perfor mance. On the other side of the coin, more frequent social interactions with the CEO ma y impair the independence of local directors. This, in turn, may lead to a misalignment of directors and shareholders interest. It is reasonable for local independent dir ectors to be more concerned about local issues and persons with whom they interact more frequently. Additionally, CEOs are more visible in the local community and, consequentl y, local independent directors are also more likely to weigh the implications of board decisions on their social standing. Social considerations of local independent director s may lead to a conflict of interests with shareholders. Local independent directors ar e therefore more likely to place emphasis on
4 politeness and candor at the expense of fr ankness and truth (Jensen (1993)) in the functions as members of the board. Therefore, the presence of local independent directors could lead to a weaker board and poor firm performance. Given the two opposite hypothesized relationships between local independent directors and firm performance, there are seve ral natural empirical questions that can be addressed: are local independent directors more informed than other directors? Do firms that have more local independent directors perform better or worse than firms that have more geographically distant independent director s? Does the geographic dispersion of the independent directors affect board decisionmaking, such as setti ng up CEO compensation? In this study, I find addresses of direct ors for S&P 1500 firms for the years 1996 to 2004 and compute the distance between the hom e address and corporate headquarters for each director. The director is identified as local if they live within a 50 miles radius from corporate headquarters. First, I use the performa nce of directors trades in the stock of the firm on whose board they serve as a proxy of potential information advantage. I collect all the director trades from Thomas Financial In sider Trading Database and compute the return for each director trade by mimicking their positions. Among all trades, I only investigate purchase trades as they are more informa tive compared with sales trades (Lakonishok and Lee (2001)). By comparing the profits from the trades of local directors and non-local directors, I reveal that the local directors trades overall outperform the non-local directors trades. Specifically, over a one-year horizon, the lo cal independent director s trades have an average cumulative return of 25.3% versus 21.3% for the non-local direct ors. After adjusting by the value-weighted market index, the differe nce between the local directors trades and
5 non-local director trades is about 3.8%1. I further explore the issue by sorting the tr ades by firm characteristics. If the local directors have an information advantage over th eir distant counterparts, it is reasonable to expect that the advantage will be stronger for firms with high levels of information asymmetry. Indeed, I find the difference on th e trades by local and non-local independent directors is strongest for the smallest size terc ile or the group of firms that have the fewest analysts following. Local directors of firms belonging to the smallest tercile earn a 7.72% higher cumulative market adjusted return ove r a one-year horizon compared to the non-local counterparts. However the return difference drops to 1.56% in large size tercile and becomes statistically insignificant at the 5% level. Next, I investigate whether the majority of local independent directors on the board enhance board effectiveness, i.e. leads to be tter firm performance. The analysis shows an inverse relationship between the proportion of lo cal independent directors and firm value and ROA, respectively. This suggests that the pr esence of local independent directors has a negative impact on board effectiveness and lends support to the idea that local directors are less effective monitors and more likely to side with the CEOs. Finally, I investigate whether the proportion of local independent directors affect the pay-performance sensitivity of CEO comp ensation and the total compensation CEO received. I find that CEOs of firms with 100% non-local independent directors receive a lower total compensation than the median number in industry. This paper contributes to the literature in at least three research areas: (1) To the best of my knowledge, this is the first paper that links directors geographic proximity to the firm 1 It is possible that this reflects only the advantage of being local investors, rather than being more informed local directors with access to the CEOs.
6 (i.e. the CEOs), and their monitoring role. Pr evious literature has focused on various board characteristics and firm performance, but has fa iled to identify director locality as a factor that affects board effectiveness. (2) This study sheds light on the tradeoff between having more geographically proximate and more poten tially informed independent directors and having directors that are less informed but mo re objective due to their distance. Recognizing that even unaffiliated, but geographically proxi mate, directors may not exercise sufficient independence has important implications for both corporate and the mutual fund boards in light of recent legal proposals that have sought to make boards more independent2. The classification of directors as independent ba sed on affiliation alone has been previously criticized. This study provides novel evidence that independent, i.e. local but unaffiliated, directors behavior is not always in line with the best interest of the shareholders. The evidence suggests that adding geographic proximity to the CEO to the selection criteria for director may improve shareholder values. (3) This study also contributes to the growing literature on the importance of geography on economics by demonstrating that local independent director trades are more profitabl e than those of their non-local counterparts. The reminder of the paper is organized as follows. Hypotheses are developed in section I. Section II contains the data selection process and descriptive statistics. Section III provides the results. Section IV presents additional robustness tests and section V concludes. 2 For example, in 2002 NYSE and NASD have a rule change to require that a majority of a listed firms board be made up of independent directors Detail of this rule can be found at http://www.sec.gov/rules/sro/34-48745.htm.
7 Geography and Director Independence: Background Distance and Information Previous research has documented that ge ographical proximity is of importance for investors. The literature has shown that lo cal investors earn, on average, higher returns compared to remote investors. For instan ce, Ivkovic and Weisbenner (2005) examine the returns of a large number of individual investors using data from 1991 to 1996 obtained from a discount broker and find that households exhib it a strong preference fo r local investments. The average household earns an additional annuali zed return of 3.2% from its local holdings relative to its non-local holdings, suggesting that local investors can exploit local knowledge. Excess returns to investing locally are even larger, about 6%, among stocks not in the S&P 500 index. Mutual funds also earn a higher return on holding local stocks than remote stocks. Coval and Moskowitz (2001) separate mutual fund holdings into local and distant stocks. Local stocks that are held by the mutual funds earn a higher return than the local stocks that are not held by mutual funds. More over, turnover of the local stocks is less frequent than that of non-local stocks. Local firms held by mutual funds tend to be sma ll and highly leveraged. Coval and Moskwitz (2001) suggest that stocks of firms with these kinds of characteristics are the ones in which local investors have a gr eater information advantage. Further evidence of the local information advantage comes from equity analyst and investment bank studies. For example, Malloy (2005) finds that local anal ysts provide more accura te earnings forecast and that their forecast revisions have a greater impact on the market. Butler(2007) shows that local investment banks have better access to soft information and have absolute and comparative advantage to place low-rated bonds. By employing a large sample of municipal
8 bond offerings from 1997 to 2001, he finds that for per credit rating notch of the bond, the increase in all-in cost (yield plus investment banking fee) is approximately 4.4 basis points if the bond is underwritten by local investment banks compared with 18.7 basis points if underwritten by non-local investment banks. In the same vein, local directors might have their own information advantage over directors who are residing far away from the co rporate headquarters. If the local directors possess better information, I could expect that on average, their trades will yield higher positive abnormal returns than that of their remote counterparts. Furthermore, it would also be reasonable to expect that the local info rmation advantage would decrease with firm transparency. Large firms with greater analys ts following and /or firms in the urban area would be less likely to exhibit local information advantages. Based on the above, I propose the following hypothesis on the information advantage associated with local directors. H1: Local directors have an information advantage over non-local directors and the advantage is stronger in firms with higher information asymmetry. Distance and Monitoring Naturally, the next question to ask is how geographic proximity of independent directors may affect firm value. Many other st udies have documented other attributes of the board that are related to firm value. For inst ance, Yermack (1996) finds that the size of the board is negatively related to firm value. Vaf eas(1999) finds that board meeting frequency is negatively related to Tobins Q. Ferris, Ja gannathan and Pritchard (2003) find there is a positive link between firm performance and the num ber of board seats a director holds while Fich and Shivdasani (2006) show that firms with busy boards, i.e. those in with a majority of outside directors holding three or more director ships, are associated with weak corporate
9 governance, exhibiting lower market-to-book ratio s and weaker profitability. Over a wide range of issues, all management has to do to cap ture the board is to present information in a way that is likely to generate support for its pe rspectives, or in a selective way, to achieve effective capture of the board. In a model developed by Adams and Ferreira (2007), the CEO is less likely to share information with director s when the board is m onitoring intensively. If the managers want to act against the share holders interest, they can simply keep the directors in the dark. Geographic proximity enables the local directors to have more information advantage over the remote directors. In this sense, the board could become more effective if the majority of outside director s are local. Therefore, I want to examine the interplay between the effectiveness of the board and local directors. In contrast, local independent directors are mo re likely to have greater levels of social interaction with the managers. This friendly in teraction makes local directors less likely to challenge managers. Landier, Nair and Wulf (2006) have shown that the proximity to headquarter affects the managerial concerns to their employees at different divisions. By using a firm division level data, they show that divisions that are closer to firm headquarter are less likely to experience layoffs. Thus, under this alternative view, geographic proximity could cause local independent directors to work less effectively in the board. Based on the above, the second hypothesis is as follows: H2: Geographic proximity does not affect firm value, other things equal. The firm that has a greater proportion of local independent directors is not systematically different from the firms that have few local independent directors. Director Locality and CEO Compensation The board of directors is the primary internal corporate governance mechanism
10 responsible for setting management compen sation and monitoring senior management (Jensen (1993)). It has been considered to play an important role in setting an effective incentive contract structure that alleviates ag ency problems arising from the separation of ownership and control (Murpey(1999), Cory, Guay and Larker(2001)). If boards with more local independent directors act more effectiv ely because locals are better informed, then companies with such boards will favor lower non-incentive pay (such as cash compensation) and higher pay-performance se nsitivity of CEO compensation. However, the opposite effect may hold if the proximity of the directors re sidence compromises their objectivity. In that case, the local independent directors may be less likely to challenge the CEO in compensation matter in the board. Therefore, I propose the following hypothesis. H3: The director geographic proximity does not systematically affect CEO compensation and there is no systematic di fference in CEO total compensation and CEO incentive compensation and performance betw een firms with board dominated by local directors and other things being equal. Data and Variable Constructions Sample Selection The databases used in this study are CRSP, Compustat, IRRC, ExecuComp, I/B/E/S, and Insider Trading. The IRRC database cove rs board director information for S&P 1500 firms for the period 1996-2004. For each director in the database, I identify her home address from her report on their insider trading to the SEC. For the directors who change their home address at a given year, I pick the address that is closer to the annual board meeting date. The stock return of the firm is retrieved from CRSP database while the accounting information is
11 from Compustat database. Variables Distance is defined as the number of mile s from the directors home address to corporate headquarters. I computed it based on th e zip codes of these two addresses. I then classify a director as local if the distance be tween the directors home address and corporate headquarters is within 50 miles and non-local if the distance is greater than 50 miles. To obtain a clearer local variable, I exclude direct or observations where the distance is between 50 and 100 miles for the tests of directors trades3. I identify around 11,300 distinct zip codes in the data. I obtain the latitude and longitude for each of the zip codes from the U.S. Census Bureaus Gazetteer Place and Zip Code Database. The corresponding company lo cation comes from Compact Disclosure, which contain information about the company h eadquarters zip code. Finally, I compute the distances based on the combination of firm h eadquarters zip code and directors zip code4. Table 1 presents the distributions of directors by distance from their firms headquarters. For the employee directors, 18,560 out of 22,879 director-year observations are within 50 miles. This is not surprising si nce most of employee directors are working in the corporate headquarters. For the employ ee directors, only 3,966 out of 22,879 directoryear observations live over 100 miles away, accounting for about 17% of all insider directoryear observations. For the independent di rectors, there are 33,724 out of 68,751 outside director-year observations are within 50 miles radius, accounting for 49% of all observations for independent directors. Over 48% or 32,751 of independent director year observations are 3 There are 696 buy trade observations from the direct ors who live between 50 and 100 miles away from headquarters. 4 A detailed explanation about the method can be found in Ivkovic and Weisbenner (2005).
12 over 100 miles away from the corporate headqua rters. Looking at the geographic distribution by the city or state, there are 3,214 employee directors living in the same city as the headquarters and 18,764 living in the same state. This is compared with 3,908 independent directors who live in the same city and 36,075 w ho live in the same state as the corporate headquarters. In this study, my main variable for geographic proximity is the local variable classified by the distance. The drawback of th e same-state or same-city variable is that headquarter and director home addresses might be located close to state boundaries or city boundaries leading to incorrect classification of proximity. Since my focus is on the social interaction between independent directors and CEOs, the distance measure would better capture the proximity between the CEO and directors. For example, the independent director who lives in Miami would not necessary have more social interaction with the CEO in Jacksonville than the independent directors living in Savannah, Georgia. Table 1 Distribution of Director by Distance Director Type Distance SameCity SameState Total < 50 miles 50-100 miles > 100 miles Employee 18,560 353 3,966 3,214 18,764 22,879 Independent 33,724 2,456 32,571 3,908 36,075 68,751 Gray 9,440 506 6,293 1,439 9,872 16,239 Table 2 presents the descriptive statistics for the major variables. Q is defined as market value of the firm divided by the replacement cost, as in Chung and Pruitt (1994)5. It 5Alternatively, Q has been proxied by the mmarket-to-book ra tio, as the market value of the firms equity at the end of the year plus the difference between the book value of the firm's assets and the book value of the firm's equity at the end of the year, divided by the book va lue of the firm's assets at the end of the year. In
13 is computed as the sum of market value of equity, liquidation value of preferred stock(Compustat item 10), net value of debt, which is short term debt liability (Compustat item 72 ) net current asset (Compustat ite m 68), and book value of long term debt (Compustat item 9), and then scaled by book value of total assets. In the sample, firms have a mean Q of 1.43 with a standard deviation of 1.84. ROA is computed, following Fich and Shivda sani (2006), as operating income before depreciation (Compustat item 13) plus the decrease in receivables (Compustat item 2), the decrease in inventory (Compustat item 3), the increase in current lia bilities (Compustat item 72), and the decrease in other current assets (Compustat item 68). I then scared it by the average of beginningand ending-year book value of total assets (Compustat item 6). Following Shivdasani and Yermack (1999) I also create a variable for CEO involvement to measure the influence of CEO in the selection of directors. It equals to 1 when the CEO sits on the nominating committee or CEO sits on the board when the firm does not have a nominating committee. It has a mean value of 0.3, indicating that in about 30% of the firm whose director selection procedure is influenced by the CEOs. I also use urban and rural variables as documented in Loughran and Schultz (2005). A stock is defined as an urban stock if the co mpany headquarters is in one of the ten largest metropolitan areas of the United States accord ing to the 2000 census. These include New York City, Los Angeles, Chicago, Washingt on-Baltimore, San Francisco, Philadelphia, Boston, Detroit, Dallas, and Houston. A company is defined as rural if its headquarters is 100 miles or more from the center of any of the 49 U.S. metropolitan areas of one million or more people according to the 2000 census. The sample has 40% director-year observations with headquarter in urban areas, 8% in rural area and the remaining 52% in the suburbs. the tests, using this metric, not reported for the sake of brevity, yield similar results with the ones reported
14 I created two variables to identify the independent directors geographic characteristics based on the distance of their residence from firm headquarters: a) an indicator variable for 100% non-local independent director s on the board, and b) the number of local independent directors on the board. In the sample, about 16% of the available firmyear observations have a board with 100% nonlocal independent directors. The sample firms have an average 2.79 local independent di rectors with a median 2 local independent directors on the board. Overall, 50% of i ndependent directors on the board are local. In Panels B and C of Table 2, I also sepa rate the sample firms into groups based on the geographic characteristics: urban and rural, and based on the board locality characterisitics: zero local independent directors board and all others. As shown in Panel B, urban firms are, on average, larger than rura l firms. Urban firms also have more local independent directors on the board. On averag e, urban firms have 2.9 local independent directors (55% of the total independent directors) compared with 2.2 for the rural firms (39% of the total independent directors). In Panels B and C of Table 2, I also sepa rate the sample firms into groups based on the geographic characteristics: urban and rural, and based on the board locality characterisitics: zero local independent directors board and all others. As shown in Panel B, urban firms are, on average, larger than rura l firms. Urban firms also have more local independent directors on the board. On averag e, urban firms have 2.9 local independent directors (55% of the total independent directors) compared with 2.2 for the rural firms (39% of the total independent directors). When we compare the firms with 100% nonlocal independent director boards with all other firms, we observe that the former gr oup, on average, consists of small firms in terms here.
15 of assets value. Also, in contrast to firms w ith local independent directors on the board, those firms are more likely to be headquartered in a ru ral area, rather than in an urban area. For the firms with zero local independent directors 28% are in urban and 15% are in rural areas, while 42% (7%) of firms with all other types of board structure are in (rural) areas. Table 2 Descriptive Statistics Panel A Firm characteristics Mean Median SD N Sales ($ million) 4527 1218 12491 11416 Assets ($ million) 11711 1568 55096 11416 Q 1.43 0.98 1.84 11416 Return on assets 0.13 0.13 0.12 10483 Age 27.55 22 18.73 11416 # of business segments 2.7 2 1.79 10691 Capital expenditure/assets 0.06 0.04 0.06 10569 Governance structure and CEO compensation Mean Median SD N Board size 8.74 8 2.97 11416 Number of independent directors 6.17 6 2.72 11416 CEO ownership (%) 2.54 0.33 6.22 9978 Director ownership (%) 8.62 1.3 19.51 11416 CEO compensation ($ million) 11.83 6.59 18.67 10439 CEO involvement dummy 0.3 0 0.46 11416 CEO-Chair dummy 0.66 1 0.47 11416 Board and firm geographic characteristics Mean Median SD N Urban dummy 0.4 0 0.49 11416 Rural dummy 0.08 0 0.27 11416 100% non-local independent director board 0.16 0 0.36 11416 # of local independent directors 2.79 2 2.29 11416 % of independent directors are local 50% 50% 33% 11416
16 Table 2 Descriptive Statistics (continue) Panel B Firm Characteristics Urban Firms Rural Firms Mean Median SD N Mean Median SD N Difference of Mean Pvalue Sales ($ million) 5416.97 1271.65 15316. 8 4570 4771.82 979.69 18568.02 930 645.15 0.26 Assets ($ million) 17993.14 1889.06 76853.02 4570 7234.28 1459.01 15680.82 930 10758.86 0.00 Q 1.55 1 2.37 4570 1.09 0.88 0.9 930 0.46 0.00 Return on assets 0.17 0.16 0.18 4082 0.17 0.16 0.17 828 0 0.25 Age 27.49 21 20.14 4570 28.34 28 16.21 930 -0.85 0.27 # of business segments 2.8 3 1.84 4284 3.2 3 2.04 848 -0.4 0.00 Capital expenditure/assets 0.06 0. 04 0.06 4224 0.07 0.05 0.07 849 -0.01 0.00 Governance structure and CEO compensation Board size 8.59 8 2.96 4570 8.97 9 2.96 930 -0.38 0.00 Number of independent directors 6 6 2.56 4570 6.35 6 2.71 930 -0.35 0.00 CEO ownership (%) 2.48 0.31 5.79 3919 3.42 0.41 8.34 816 -0.94 0.00 Director ownership (%) 8.56 1.3 18.02 4570 7.62 1.4 14.74 930 0.94 0.13 CEO compensation ($ million) 14824.2 8236.72 21989.99 4159 7290.66 4494.46 8148.3 856 7533.54 0.00 CEO involvement dummy 0.29 0 0.45 4570 0.32 0 0.47 930 -0.03 0.03 CEO-Chair dummy 0.67 1 0.47 4570 0.62 1 0.49 930 0.05 0.00 Board geographic characteristics 100% non-local independent director board 0.11 0 0.31 4570 0.29 0 0.45 930 -0.18 0.00 # of local independent directors 2.98 3 2.19 4570 2.23 2 2.26 930 0.75 0.00 % of independent directors are local 0.55 0.6 0.32 4570 0.39 0.33 0.36 930 0.16 0.00
17 Table 2 Descriptive Statistics (continue) Panel C Firm Characteristics Board w/ 100% non-Local Director s Board w/ Local Directors Mean Median SD N Mean Median SD N Difference of mean P-value Sales ($ million) 4511.17 1040.21 13896.85 1779 4533.15 1259.6 12223.13 9623 -21.98 0.94 Assets ($ million) 6784.46 1300.03 35458.51 1779 12634.13 1646.31 57995.98 9623 -5849.67 0.00 Q 1.43 1.01 2.28 1779 1.43 0.97 1.75 9623 0 0.93 Return on assets 0.17 0. 17 0.18 1707 0.17 0.17 0.17 8666 0 0.23 Age 26.33 19 19.32 1779 27.78 23 18.62 9623 -1.45 0.00 # of business segments 2.67 2 1.82 1757 2.71 3 1.78 8920 -0.04 0.39 Capital expenditure/assets 0.07 0.05 0.07 1742 0.06 0.04 0.06 8814 0.01 0.00 Governance structure and CEO compensation Board size 7.89 8 2.79 1779 8.91 9 2.97 9623 -1.02 0.00 Number of independent directors 5.27 5 2.61 1779 6.34 6 2.7 9623 -1.07 0.00 CEO ownership (%) 2.69 0.39 6.46 1491 2.52 0.32 6.18 8473 0.17 0.34 Director ownership (%) 10.52 1. 7 23.67 1779 8.28 1.3 18.63 9623 2.24 0.00 CEO compensation ($ million) 10483.37 6258.78 16344.29 1560 12079.53 6670.28 19051.08 8865 -1596.16 0.00 CEO involvement dummy 0.34 0 0.47 1779 0.3 0 0.46 9623 0.04 0.00 CEO-Chair dummy 0.64 1 0.48 1779 0.67 1 0.47 9623 -0.03 0.00 Firm Geographic Characteristics Urban dummy 0.28 0 0.45 1779 0.42 0 0.49 9623 -0.14 0.00 Rural dummy 0.15 0 0.36 1779 0.07 0 0.25 9623 0.08 0.00
18 Empirical Results Directors Trades Return: Local vs. Non-local In order to test the information advantag e of the local directors versus non-local directors, I collect all insider buys for all independent directors from Insider Trading database6. For each director-purchase trade, I mimi c it by going long the firms stock and at same time short the value-weighted CRSP ma rket index. I compute cumulative market adjusted returns and buy and hold market adjust return for each trade. The cumulative market adjusted return for the horizon H is computed as the sum of difference of daily stock return and daily value-weighted market index return. 11 HH iHitIndexH ttCARRR The buy and hold market adjusted return (BHAR) for the horizon H is the sum of compounded daily stock return minus the co mpounded daily value weighted CRSP market index return. 1212(1)(1)(1)(1)(1)(1)iHiiiHIndexIndexIndexHBHARRRRRRR Each director-purchase trade is then classi fied into local or non-local trades based on the geographic location of the directors. Tabl e 3 presents the directors trades cumulative return and BHR return before and after ad justed by the value weighted CRSP index for periods from 3 months up to 2 years. As shown in the Table 3, the cumulative retu rn for the local independent directors purchase trades ranges between 0.085 for a th ree months period to 0.44 for two years, compared with 0.078 to 0.375 for non-local inde pendent directors trades. The difference between their trades stands 0.008 to 0.065, which ar e statically significant at the 5% level.
19 After adjusting by the CRSP value-weighted ma rket return, the difference of the trades profit are similar to those of cumulative returns and are still significant, in both economic and statistic levels. Panel C and D show the buy and hold return for the director trades. For all periods from 3 months to 2 years, the return for local independent director purchase trades are higher than non-local director trades. The differen ces between these two are still statistically significant. Table 3 Trading Performance of Directors: Local vs. Non-local Panel A: Cumulative return 3 months 6 months 1 year 2 years Local independent 0.085 0.140 0.253 0.440 Non-local independent 0.078 0.110 0.213 0.375 Difference(Local-Non-local) 0.008 0.030 0.040 0.065 P-values (0.038) (0.000) (0.000) (0.000) Panel B: Market adjusted cumulative return Local independent 0.053 0.080 0.142 0.268 Non-local independent 0.045 0.053 0.104 0.193 Difference(Local-Non-local) 0.007 0.027 0.038 0.074 P-values (0.038) (0.000) (0.000) (0.000) Panel C: Buy and hold return Local independent 0.082 0.137 0.302 0.495 Non-local independent 0.072 0.104 0.209 0.400 Difference(Local-Non-local) 0.009 0.034 0.092 0.095 P-values (0.026) (0.000) (0.000) (0.000) Panel D: Market adjusted buy and hold return Local independent 0.049 0.078 0.188 0.305 Non-local independent 0.041 0.048 0.097 0.198 Difference(Local-Non-local) 0.009 0.030 0.091 0.107 P-values (0.026) (0.000) (0.000) (0.000) In Table 4, I sort the firms in the sample into the size terciles and report CARs and BHARs by size terciles. Malloy (2005) shows th at the local information advantage is stronger in small and highly levered firms. If local directors, as other monitors, also have an
20 information advantage, the advantage would be stronger when the firm has less external information which the non-local directors could use. On the other hand, the local information advantage can be diminished if the firms are larg er or more transparent. To account for this, each firm is assigned into small, middle and t op size tercile based on the market value at the end of previous calendar year. As shown in the Table 4, the difference between the returns for local directors trades and non-local director trades appears to be strongest in the small tercile and gradually declines for the medium and large size terciles. For example, the difference for the 6 month period is 4.3% between the local director trades and non-local director trades in small size tercile, while for the top size tercile, the difference is only 1.99%. Moreover, it becomes insignificant for the 1 year and 2 year horizon for the firms in top size tercile. This result indicates in small firms local directors have a clearer information advantage while for the large firms; the local information advantage b ecomes weaker as the large firms have more external information availability. This finding c ould be driven by the fact that analysts are more likely to follow the large firms, large firms have greater media coverage and more voluntary disclosures. This is consistent with Malloy (2005)s finding. Table 4 Trading Performance of Directors by Distance and Firms Information Characteristics Market-Adjusted Return by Size Tercile Small Size Tercile Cumulative Return Buy and Hold Return 6 month 1 year 2 years 6 month 1 year 2 years Local Independent 16.65% 28.57% 49.07% 18.21% 41.64% 68.61% Non-local independent 12.35% 20. 85% 30.30% 13.11% 22.14% 40.96% Difference(Local-Nonlocal) 4.30% 7.72% 18.77% 5.10% 19.50% 27.65% P-values (0.00) (0.00) (0. 00) (0.00) (0.00) (0.00)
21 Table 4 Trading Performance of Directors by Distance and Firms Information Characteristics (continue) Middle Size Tercile Local Independent 6.02% 11.63% 23.99% 5.42% 14.76% 21.92% Non-local independent 4.02% 8. 71% 20.29% 2.97% 7.44% 17.35% Difference(Local-Nonlocal) 2.00% 2.92% 3.70% 2.45% 7.32% 4.57% P-values (0.02) (0.03) (0. 04) (0.01) (0.00) (0.08) Top Size Tercile Local Independent 3.25% 5.74% 12.48% 2.18% 5.19% 9.71% Non-local independent 1.26% 4. 18% 10.68% 0.22% 2.36% 6.50% Difference(Local-Nonlocal) 1.99% 1.56% 1.80% 1.96% 2.83% 3.21% P-values (0.00) (0.10) (0. 15) (0.00) (0.01) (0.05) Multivariate Regression Tests: Director Locality and Trades Return Though the univariate tests show that there is a significant difference between local versus non-local director trades, these tests do not control for other effects that could explain the trading differences between local and non-local directors. Table 5 presents the multivariate regression tests for the following models: ,,,, ,,,,,, ,,,,,,, iHijtijt iHijtijtitit iHijtijtitititBHARLocaldummyDirectorHolding BHARLocaldummyDirectorHoldingSizeBM BHARLocaldummyDirectorHoldingSizeBMPoorGovernance where DirectorHolding i,j,t is the percentage of the shares held by each director j in firm i at time t Localdummyi,j,t equals to 1 if director js home address is within 50 miles from firm is corporate headquarters or 0 otherwise. Size is the market value of the firm at the end of the previous calendar year; book to market is defined as the book value at fiscal year scaled by the market value of the firm; PoorGovernance is equal to 1 if the G-index is greater than 10 and 0 otherwise. As shown in Table V, the dependent variables are the compounding abnormal market return at the di fferent horizons. The independent variables
22 are local director indicator (0,1) and director holding. The local director indicator, which is the variable of main concern, is statistically significant at the 1% level or better for the six monthsand one year horizons. The result indicates that on average the local director trades yield 0.01 or 1% more than non-local director tr ades for the three monthsand the difference goes up to 0.03 for the six months horizon. In the second model, I include size and bookto-market as additional controls in the main regressions. Size and book-to-market have been documented as main explanatory variables for cross-sectional returns. Adding th ese two variables does not alter the effect of the local director indicator. The coefficients for the local director indicators remain pretty stable in all three regressions in the second m odel. Size has a negative coefficient, indicating that director trades have a lower return when the size goes up. In the third model, I control for the qua lity of the fi rms governance by adding a poor governance dummy. The evidence shows that if firms have poor govern ance (i.e., a G-Index of 10 or more), their directors trades yield a higher return. Directors earn 2.6% higher return on their trades over a one year horizon if th e firm has poor governance. This evidence implies that in poor governance firms, insiders are more likely to enjoy an informational advantage. The results in Table 3 to 5 show that th e local directors trades yield a higher abnormal return than those of the non-local directors. However, this evidence cannot answer the question of where this return difference comes from and how other governance mechanisms, such as analysts, board meetings and board memberships, could alleviate this advantage. In Table 6, I add the following va riables to the model: number of analysts following the stock at the quarter when th e trade made; audit committee member dummy; less than 75% board meeting dummy, which equals to one if the director attends over 75% of
23 the board meetings; interaction terms with the local director variables. If board attendance or audit committee membership is substitute channels for accessing firms information, then they could also explain the performance of director trades. As shown in all regressions, the significance of local director trades is not affected after adding these two controls: an indicator of board attendance which equals to 1 if the director attends less than 75% board meetings and a dummy for audit committee membership, and it is still significant statistically at 0.000 level. Board attendance has a negative coeffici ent, implying that attending less board meeting would reduce the trade profit. Audit me mbership also has a positive coefficient but insignificant for the horizons over six months and one year, indicating being a member of the audit committee will result in more profitable trades. The inclusion of interaction terms serves the purpose of disentangling the effects of the informative monitor and social ally ro les the local directors may play on the board. Conditional on analyst following or size, the margin al effect of the local indicator variable would be smaller if local directors play an informative monitor role. However, conditional on the poor governance, I would expect the marginal effect of the local director indicator would be greater if local directors assume the role of the CEOs social ally. Note that, as shown in Table 5, if the firm has poor governance, both local and distant directors would expect to have earn highe r returns on their trades consistent with the notion that they both enjoy some information advantage. As shown in Table 6, the coefficients for the interaction terms between the local director indicator and analyst following a nd poor governance respectively, are significant and negative when performance is measured over a one year horizon.
24 Table 5 Independent directors trades return: local vs. non-local Buy and Hold Market Adjusted Return 3-month 6-month 1-year 3-month 6-m onth 1-year 3-month 6-month 1-year Local director indicator (0,1) 0.011 0.033 0. 103 0.011 0.034 0.103 0.011 0.034 0.103 (0.013)** (0.000)*** (0.000)*** (0.012)** (0. 000)*** (0.000)*** (0.011)** (0.000)*** (0.000)*** Director holding -0.001 -0.000 0.001 -0.001 -0.000 0.002 -0.001 0.000 0.002 (0.301) (0.721) (0.571) (0.447) (0. 999) (0.308) (0.519) (0.959) (0.270) Size -0.000 -0.001 -0.001 -0.000 -0.001 -0.001 (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** Book to market 0.024 0.046 0.094 0.025 0.047 0.096 (0.003)*** (0.005)*** (0.011)** (0.003)*** (0.005)*** (0.011)** Poor governance(Gindex>10) 0.018 0.013 0.026 (0.000)*** (0.073)* (0.089)* Constant -0.013 0.016 -0.078 -0.026 -0.010 -0.128 -0.030 -0.013 -0.135 (0.293) (0.388) (0.004)*** (0.050)* (0. 652) (0.000)*** (0.023)** (0.549) (0.000)*** Observations 16038 16038 16038 16038 16038 16038 16038 16038 16038 Adjusted R-squared 0.03 0.04 0.03 0.04 0.05 0.03 0.04 0.05 0.03 This suggests that the informational advantage of the lo cal directors over the distant directors decreases as analyst following increases. This is consistent with the notion that th e local directors information advantage weakens in the more transparent firms. The coefficient of the interaction term betw een local director indicator and audit committee membership indi cator is positive, indicating that local directors enjoy and even great er information advantage if they are also sitting on the audit committee.
25 The negative coefficient of the interaction term between local director indicator and poor governance implies that the local directors information advantage relative to distant directors becomes smaller when governance is poor. Table 6 Independent directors trades return: local versus non-local after controlling for asymmetry information Buy and Hold Market Adjusted Return 3-month 6-month 1-year Local director 0.004 0.061 0.137 (0.700) (0.000)*** (0.000)*** Size -0.000 -0.000 -0.001 (0.003)*** (0.110) (0.000)*** Book to market 0.034 0.062 0.130 (0.000)*** (0.000)*** (0.001)*** Director holding -0.004 -0.004 -0.002 (0.000)*** (0.034)** (0.468) Number of analysts -0.001 -0.002 -0.001 (0.038)** (0.003)*** (0.487) Less than 75% attendance indicator (0,1) -0.004 -0.033 -0.038 (0.771) (0.111) (0.306) Committee membership audit -0.008 0.011 0.005 (0.269) (0.303) (0.771) Poor governance(G-Index >10) 0.026 0.061 0.086 (0.001)*** (0.000)*** (0.000)*** Local X Size 0.000 -0.000 0.000 (0.035)** (0.884) (0.669) Local X Analysts -0.000 -0.000 -0.006 (0.595) (0.861) (0.013)** Local X Audit Membership 0.014 0.010 0.116 (0.174) (0.522) (0.000)*** Local X Poor governance -0.004 -0.073 -0.083 (0.675) (0.000)*** (0.008)*** Constant -0.024 -0.021 -0.184 (0.100)* (0.371) (0.000)*** Observations 13542 13542 13542 Adjusted R-squared 0.04 0.05 0.04 Firm Value and Director Locality Previous research has documented that firm value is related to the board
26 characteristics. An effective board would lik ely enhance the firm value. For example, Yermack (1996) finds a negative link between board size and firm value, Fich and Shivdasani (2006) shows that busy boards have a negative impact on firms Q. Vafeas (1999) presents evidence that board meetings become more frequent after bad firm performance. However, how the distribution of director affects board effectiveness and hence the firm value is not yet resolved. In the previous section, I have shown that local independent directors seem to have more info rmation on their firms, as evidenced by their more informative trades. According to our e fficiency hypothesis, more local independent directors sitting on the board would benefit th e firm by increasing efficiency and, hence, improve the Q as the local directors are more informed and thus better monitors. On the other hand, the likelihood of social interacti on between board independent directors and corporate executives increases with geographic proximity and, thus, creates a weaker board when the board is dominated by the local directors. To address this issue, I collect data from 1996 to 2004 from IRRC and find the distance information for each director in the board. The number of firms per year with nonmissing distance information ranges about 800 to 1100 firms per year. The dependent variable Q is defined following Chung and Pru itt (1994). The main explanatory variable is director locality. I create two different director locality variables. The first one is the number of local independent directors on the board, a nd the second one is the zero local independent directors indicator, which takes the value of 1 if there is none of the independent directors who are local, and 0 otherwise. To control the board characteristics, I include several main variables previously used in th e literature. The independent variables include the log form of board size (Yermack(1996)), stock ownershi p of CEOs, directors ownership(Morck,
27 Shleifer and Vishny(1988)), busy board dummy, which is equal to 1 if the firm has at least 1 busy director(Fich and Shivdasai(2006)), l og form of number of board meetings (Vafeas(1999), firm age, number of different business segments(Lang and Stultz(1994)), growth opportunity as proxied by capital e xpenditure (Smith and Watts(1992)), firm leverage and size proxied by the log form of sales. Table 7 shows the result for th e fixed-effect panel regressions. The year dummies are also included but not reported. In the first colu mn of Table 7, the locality variable is the number of local independent directors. The co efficient for the number of local independent directors is negative and significant. This s hows that for each additional local independent director on the board, the Q on average decreases by 0.045. The board size has a negative and slightly significant coefficient in the first regression. This is consistent with Yermack (1996) and indicates a large board would be less effective and thus decrease the firms value. CEO ownership and di rector ownership both have a positive impact on Q but only CEO ownershi p has a statically significant effect. This result is consistent with the notion that owne rship helps align the interests of the CEO and stockholders. Busy board has a negative coefficien t, in support of the view that the directors who are busy would exert less effort. Consiste nt with Vafeas(1999), the number of board meetings is associated with lower Qs, possibl y because firms tend to increase the frequency of board meetings after they have experienced poor performance. In the second column, I create a strong local board dummy to explore the effect of non-local independent directors on the board. The base group is that the firm with at least one local independent directors. As shown in the table 7, the zero local independent director dummy has a significant positive impact on the firm value. Compared with base group, on
28 average, the Q is 0.317 higher for the firm w ith zero local independent directors or 100% non-local independent directors. Given that th e standard deviation of Q is 1.84, it indicates that firm with no local directors on the board would have about a 17% higher Q than other firms. The results do not support the notion that th e presence of local independent directors helps improve board effectiveness, i.e. increase the firm value. Instead the finding in Table 7 implies that firm value decreases with the pr esence of more local independent directors on the board. Q mostly reflects the firms market value or the value of growth opportunities. In the next section, I discuss how accounting perfor mance measures are affected by the board locality. Table 7 Fixed effect regression of Q on director locality characteristics Q # of local independent directors -0.045 (0.011)** Zero local independent directors indicator (0,1) 0.317 (0.000)*** Log(board size) -0.206 -0.246 (0.080)* (0.031)** CEO ownership 0.021 0.020 (0.000)*** (0.000)*** Director ownership 0.002 0.002 (0.258) (0.229) Busy directors on the board indicator (0,1) -0.030 -0.034 (0.476) (0.429) Log(number of board meetings) -0.158 -0.161 (0.011)** (0.010)*** Age 0.001 0.002 (0.946) (0.891) Number of different business segments -0.012 -0.013 (0.454) (0.430) Capital expenditure 2.982 2.962 (0.000)*** (0.000)***
29 Table 7 Fixed effect regression of Q on director locality characteristics (continue) Leverage -1.474 -1.482 (0.000)*** (0.000)*** Log(sales) -0.221 -0.229 (0.000)*** (0.000)*** Constant 4.115 4.065 (0.000)*** (0.000)*** Observations 8972 8972 R-squared 0.03 0.03 The corporate governance literature has documented extensively that firm performance and board composition are endogenous. It is possible, in the context of this study, that firms with weak governance choose a board that is tilted toward greater local director representation, which in turn leads to poor performance. To address this issue, we conduct a two stage instrumental variables (IV ) test to for the endogenous nature of the relationship between performance and board geographic structure. In the first stage of the regression, we us e the number of local independent directors or the zero local independent di rectors dummy as the dependent variables. The independent variables are: a CEO involvement dummy, which is takes the value of one if the CEO is sitting in the nomination committee or if the CEO is chair when the company does not have a nomination committee, and the value of zero ot herwise; an urban dummy, which takes the value of one if the headquarters of the firm are located if the firms headquarter is located in one of the ten largest metropolitan areas of the United States according to the 2000 census; the interaction term between the CEO involvement dummy and the urban dummy; the interaction term between the CEO involvement dummy and the natural logarithm of the number of independent directors; and, the inte raction term between the urban dummy and the
30 natural logarithm of the number of independe nt directors. The remaining independent variables are all control variables used to pred ict firm value. As shown in the table 8, the interaction term between CEO involvement du mmy and the natural logarithm of the number of independent directors ha s a coefficient of 0.118, indicating that conditioning on the number of independent directors, the board will have on average 0.118 more local directors when the CEO is involved in the nomination. In addition, when the firm is located in an urban area, the firm has 1.47 more local direct ors on the board. The same result is obtained when the dependent variable used is th e zero local independent directors dummy: conditioning on the number of independent director s, firms when the CEO is involved in the nomination and firm headquarted in the urban ar ea, are less likely to have a board with zero local independent board. Our second stage regression uses the control variables used in our previous tests of firm value together with the pr edicted board locality variable from the first stage. The results we obtain remain qualitatively similar to the ones obtained from the OLS regressions. As shown in Table 8, the firm valu e exhibit a negative association with the boards dominated by local directors. An additional local indepe ndent director would reduce Tobins Q by 0.5. Compared with a zero local independent director board, the other boards lead to 4.5 lower in terms of Q. Table 8 Two Stage IV regression of Q on director locality characteristics Q First Stage Second Stage First Stage Second Stage # of local independent directors -0.583 (0.000)*** Zero local independent directors indicator (0,1) 4.878 (0.000)***
31 Table 8 Two Stage IV regression of Q on director locality characteristics (continue) log(board size) 1.553 0.747 -0.108 0.318 (0.000) (0.000)*** (0.000) (0.046)** CEO ownership 0.005 0.020 0.000 0.016 (0.201) (0.001)*** (0.915) (0.019)** Director Ownership -0.005 -0.001 0.000 -0.000 (0.000) (0.491) (0.424) (0.983) Whether the firm has any busy directors -0.036 0.033 0.001 -0.005 (0.253) (0.465) (0.911) (0.924) Log(Number of board meetings) -0.031 -0.153 0.014 -0.202 (0.431) (0.020)** (0.131) (0.007)*** Firm Age -0.026 -0.015 0.000 -0.000 (0.025) (0.404) (0.998) (0.986) Number of different business segments that a firm has 0.010 0.031 -0.004 0.030 (0.387) (0.083)* (0.142) (0.139) Capital Expenditure Adj 0.160 1.940 0.126 1.419 (0.638) (0.001)*** (0.102) (0.031)** Book Leverage -0.113 -1.567 0.034 -1.713 (0.381) (0.000)*** (0.253) (0.000)*** Log(sales) -0.127 -0.094 0.032 -0.191 (0.001) (0.107) (0.000) (0.003)*** CEO involved in nomination indicator (0,1) -0.237 0.096 (0.014) (0.000) Urban dummy -2.313 0.141 (0.000) (0.295) CEO involvement X Urban 0.021 0.007 (0.682) (0.521) Log(# of independent directors) X CEO involvement 0.118 -0.065 (0.019) (0.000) Log(# of independent directors) X Urban 1.478 -0.083 (0.000) (0.000) Constant 3.051 2.068 (0.000)*** (0.013)** Observations 8972 8972 R-squared Number of spc permanent number 1804 1804 Directors Locality and ROA Table 9 presents multivariate regressions of ROA on director locality. The dependent
32 variable, ROA, is created following Fich and Shivdasani (2006) as previously described.7 The other independent variables are those incl uded in the Q regressions. As shown in Table VIII, the results of the ROA regressions are cons istent with those of the Q regressions. Local director presence has a negative impact on the ROA. The coefficient of the number of local independent director variable is -0.004, indicating that for an additional local director in the board, the firm ROA decreases by 0.004. The imp act of locality becomes stronger when I compare the board with 100% no local directors and those with at least 1 local director in the board, the difference is 0.029. It implies that th e board with no local directors on the board has a 0.029 higher ROA than other firms. Consider ing that the standard deviation of ROA is about 0.12, the firm with zero local independent directors has a about 0.23 standard deviation higher ROA than other counterparts. Table 9 Fixed effect regression of ROA on director locality characteristics Return on Assets # of local independent directors -0.004 (0.025)** Zero local independent directors indicator (0,1) 0.029 (0.000)*** Log(board size) -0.020 -0.024 (0.075)* (0.034)** CEO ownership 0.003 0.003 (0.000)*** (0.000)*** Director ownership -0.000 -0.000 (0.409) (0.437) Busy directors on the board indicator (0,1) -0.017 -0.017 (0.000)*** (0.000)*** Log(number of board meetings) -0.017 -0.017 (0.005)*** (0.005)*** Age -0.001 -0.001 (0.471) (0.509) 7 For the sake of brevity, we also have two different de finitions of ROA. But the results are basically same. So we omitted it to report in the tables.
33 Table 9 Fixed effect regression of ROA on director locality characteristics (continue) Number of different business segments -0.013 -0.013 (0.000)*** (0.000)*** Capital expenditure 0.509 0.507 (0.000)*** (0.000)*** Leverage -0.121 -0.121 (0.000)*** (0.000)*** Log(sales) 0.069 0.068 (0.000)*** (0.000)*** Constant -0.167 -0.171 (0.009)*** (0.007)*** Observations 8740 8740 R-squared 0.06 0.06 Following the IV technique in the previous section, we ran the same test tor the ROA regression, the result remains also qualitatively similar to the fixed effect regression. On average, a local independent director is a ssociated with a reduction in ROA by 0.07, while a zero local independent board would cause a 0.7 increase in ROA. Table 10 Two Stage IV regression of ROA on director locality characteristics ROA First Stage Second Stage First Stage Second Stage # of local independent directors -0.078 (0.000)*** Zero local independent directors indicator (0,1) 0.702 (0.000)*** log(board size) 1.543 0.109 -0.108 0.060 (0.000) (0.000)*** (0.000) (0.001)*** CEO ownership 0.005 0.003 0.000 0.002 (0.143) (0.000)*** (0.834) (0.004)*** Director Ownership -0.005 -0.001 0.000 -0.000 (0.000) (0.002)*** (0.310) (0.047)** Whether the firm has any busy directors -0.038 -0.008 0.002 -0.014 (0.229) (0.094)* (0.743) (0.016)** Log(Number of board meetings) -0.029 -0.016 0.010 -0.021 (0.463) (0.018)** (0.264) (0.016)**
34 Table 10 Two Stage IV regression of ROA on director locality characteristics (continue) Firm Age -0.026 -0.003 0.000 -0.001 (0.026) (0.066)* (0.996) (0.533) Number of different bus segments that a firm has 0.008 -0.007 -0.004 -0.007 (0.491) (0.000)*** (0.092) (0.004)*** Capital Expenditure Adj 0.149 0.351 0.156 0.255 (0.677) (0.000)*** (0.054) (0.001)*** Book Leverage -0.115 -0.133 0.024 -0.148 (0.379) (0.000)*** (0.417) (0.000)*** Lsales -0.126 0.086 0.032 0.072 (0.001) (0.000)*** (0.000) (0.000)*** CEO involved in nomination indicator (0,1) -0.219 0.102 (0.025) (0.000) Urban dummy -2.308 0.128 (0.000) (0.344) CEO involvement X Urban 0.023 0.006 (0.657) (0.624) Log(# of independent directors) X CEO involvement 0.107 -0.068 (0.036) (0.000) Log(# of independent directors) X Urban 1.474 -0.077 (0.000) (0.000) Constant -0.312 -0.461 (0.000)*** (0.000)*** Observations 8740 8740 R-squared Number of spc permanent number 1759 1759 Director Locality and CEO Compensation The previous sections have shown that the presence of local independent directors is negatively related to firm performance. This evidence lends support to the entrenchment hypothesis that more social interaction between local independent di rectors and firm CEO leads to a weaker board. To further explore this question, I am testing another crucial function the directors play on the board: setti ng up the CEO compensation. If local directors behave in accordance with the entrenchment hypothesis, the presence of local directors would more likely result in higher CEO compensation and more importantly a weaker link
35 between CEO compensation and firm performan ce. Agency theory advocates that strong CEO incentive pay is more likely to reduce ag ency costs. Jensen and Murphy (1990) and Yermack (1996) show that strong boards are associated with higher CEO incentive pay. Westphal and Zajac(1995) illustrate that the pow erful CEO seeks to recruit new directors with similar demographic characteristics and, ther efore, those directors are more friendly to the CEOs when setting up the compensation. In Table 11, I assess the relationship betw een abnormal CEO compensation and local directors. The abnormal CEO compensation is de fined as the difference between current total CEO compensation, sum of salary and bonus and the median industry CEO total compensation. The industry is classified by the fi rst two digits of the firms SIC code. As control variables, I include size (log of sale s), CEO-Chairman dummy, profitability (industry adjusted ROA), and growth opportunities (depreciation over sales). ()itititititititCEOCompensationLocalLogsalesCEOChairdummyROADepreciation Table 11 has two columns, each correspondi ng to a different director locality variable. As shown in the first column in Ta ble 11, the number of local independent director has a significant coefficient of 0.024, implying that for an additional local directors sitting on the board, the abnormal CEO compensation increases by 0.024 million dollars. In the second column, the coefficient for the zero local di rector dummy has an opposite sign and with a coefficient of -0.04, implying 100% non-local dir ectors would compensate CEO less than the median CEO receives in the same industry. Over all, the result shows that local director presence would increase the CEO total compensation.
36 Table 11 Regressing CEO total compensation on director locality characteristics Industry-adjusted CEO compensation # of local independent directors 0.024 (0.014)** Zero local independent directors indicator (0,1) -0.040 (0.411) Log(sales) 0.429 0.440 (0.000)*** (0.000)*** Depreciation/Sales 0.067 0.068 (0.022)** (0.021)** Chair-CEO indicator(0,1) 0.091 0.091 (0.001)*** (0.001)*** Return on assets 1.079 1.052 (0.000)*** (0.000)*** Constant -3.023 -3.025 (0.000)*** (0.000)*** Observations 9917 9917 Adjusted R-squared 0.28 0.28 In addition, the board of di rectors not only set up the level of CEO compensation, but more importantly, they also set up the link between CEO compensation and firm performance. Previous research has documente d that the link is stronger when firms have more effective boards (for instance, Ye rmack(1997)). In Table 12, I create the CEO incentive variable in the spirit of Core and Guay (1999), measured by the dollar change in the value of stock and option CEO holds from a one percentage change of firms stock price. Following Bergstresser and Philippon (2006), I create the CEO incentive ratio as follows: 0.01()ititititOnepctPriceSharesOptions /()itititititCEOIncentiveOnepctOnepctSalaryBonus Where shares is the number of shares the CEO holds, options is the sum of the stock options the CEO granted, number of unexercised options and number of exercisable options. Each regression includes the same control variab les as in Table X, with the addition of the log of board size. I then regress the CEO incen tive pay on the director locality variables and
37 the other controls. According to agency theory, an effective board acting in the best interest of stockholders would create a strong li nk between CEO compensation and firm performance. As shown in the column and sec ond column, the director locality variable has a negative sign, which implies that adding the lo cal directors on the board would weaken the link between CEO incentive pay and firm perfor mance. For example, for additional local directors on the board, the CEO incentive pay decreases by 0.003. The second column further shows that those firms with no local di rectors have an increased CEO incentive pay, with a positive coefficient of 0.037 which is significant at the 5 percent level. In conclusion, Tables 11 and Table 12 indicate that firms with more local directors on their boards are more likely to pay their CEOs higher levels of total compensation, thus providing a weaker link between CEO pay and firm performance. Overall, these findings are consistent with the entrenchment hypothesis. Th e potential social interactions between local directors and CEOs make local directors le ss likely to play the role of watchdog on the board. They are more likely to return favors b ack to CEOs by helping them to set up higher compensation and weaken the link between incentive pay and performance. Table 12 Regressing CEO incentives on directors characteristics CEO Incentive # of Local Independent Directors -0.003 (0.001)** Zero local independent directors indicator (0,1) 0.037 (0.00)* Log(sales) 0.057 0.058 (0.000)*** (0.000)*** Depreciation/Sales 0.002 0.003 (0.453) (0.400) Chair-CEO indicator(0,1) 0.038 0.037 (0.000)*** (0.000)***
38 Table 12 Regressing CEO incentives on directors characteristics (continue) Return on assets 0.056 0.049 (0.005)*** (0.014)** Log(number of board meetings) -0.079 -0.079 (0.000)*** (0.000)*** Log(board size) -0.109 -0.11 (0.000)*** (0.000)*** Constant 0.282 0.265 (0.000)*** (0.000)*** Observations 9933 9933 Adjusted R-Square 0.29 0.29 Robustness Tests In this study, I investigate the effect of pr oximate directors on board effectiveness. I rely on the measure of distance between the home address and corporate headquarters to determine whether the directors are local or distant. I decide the directors are local when they are live within a 50 miles radius from the h eadquarters. This measure is reasonable given that the actual travel time could be with in 1 hours drive. However, for the robustness check, I excluded the observations for the director s who are living between 50 and 100 miles as their information role can be ambivalent. The re sults, of the trades return test do not change. Additionally, I also repeated the te sts using a different measure, SameState, instead of the local director indicator. The SameState variable takes the value of one if the directors live in the same state as the corporate headquarter. Using SameState rather than distance measure, yields qualitatively similar results. This can be partly attributed to the fact that most directors in the same state as the headquarters are more likely to be classified as local as well in the sample. These results are not reported here for the sake of brevity. In addition to using a variety of local di rector measures, I have also used an
39 alternative measure for Q. Q is a proxy of firm value widely used in the literature. Besides the Chung and Pruitt (1994) version of Q, I also used the market-to-book measure, defined as sum of market value of equity, book value of assets, minus common equ ity and then divided by the book value of assets. Te sts based on market-to-book yield qualitatively similar results with the ones reported in the paper. Conclusion While many studies have examined geographic characteristics in the context of monitoring commercial banks, venture capitalists and security analysts, to our knowledge, this is the first study to investigate the role of geographic characteristics in the context of board effectiveness. I extend the literature of geographic proximity and information advantage by examining the difference in performance between trades of local independent directors and distant directors. The empirical tests results support the notion that the local advantage previously recorded in the context of anal ysts, mutual fund investors, and individual investors also appears to exist in the context of the board of directors. On average, local director trades significantly outperform thos e by distant directors. Moreover, the local directors advantage becomes stronger in smalle r firms and firms followed by fewer analysts. I further analyze the effect of director lo cality on board effectiveness. The efficiency hypothesis predicts that the presence of local directors would enhance board effectiveness and hence improve the firms Q and ROA. On the other hand, the entrenchment hypothesis posits that more social interaction between local directors and CEO would make local
40 directors weak monitors. Thus, more local di rectors would weaken the board and lower Q and ROA. The presence of local directors significantly reduces both Q and ROA. Our empirical tests support the notion that the local directors play a weak role on the board. Finally, I test how directors locality affects the CEO compensation level and incentive pay. Agency theory suggests that highe r incentive pay would help align the interest of CEO and stockholders. An effective board w ould be associated with a strong link between CEO pay and their performance, i.e., a highe r proportion of incentive pay. The results show when there are more local directors on the boar d, firms are inclined to pay a higher level of compensation and lower proportion of incentive pay. Overall, my results show strong relati ons between director locality and firm performance and CEO compensation, suggesting that director geographic characteristics play an important role in the effectiveness of co rporate boards. These findings are of special interest to both law makers as well as research ers by highlighting the effect of the geographic characteristics of corporate boards on board effectiveness and firm performance.
41 Essay 2 Director Incentives and Earnings Management Introduction A well-functioning corporate board is genera lly considered an important disciplinary mechanism that monitors managers and helps align the scope of their decision making with the interests of stockholders. Therefore, monitoring by the board of directors has an important effect on the economic performance of organizations (Jensen (1989)). However, how to motivate board directors to monitor inst ead of forming alliances with the managers has been the subject of ongoing debate. Previous research has focused on the importance of directors reputation concerns8. In this study, my primary focu s is on directors incentive compensation. According to Compustat ExecuComp Database, there were 274 firms that paid directors with stock options in 1992, wh ile the figure climbs to 1,214 in 2002 but then drops again to 1,055 in 2004. During the same pe riod, about 660 firms never paid directors with stock options and another 102 firms stopped awarding stock options to their directors. These changes in director compensation stru cture pose an interesting empirical question. Does the incentive pay matter in terms of motiva ting directors to monitor managers? If yes, how do changes in the form of compensation affect directors behavior? Earlier research on the impact of equity-based incentive compensation, such as stocks and stock option, on managerial behavior remains inconclusive. Jensen and Meckling(1976) suggested that firms suffering from the agency problem resulting fro m the separation of management and control can use incentive compensation to help align the interests of the CEO and stockholders.
42 Several subsequent empirical studies have provided support for this notion. For example, Morck, Shelifer and Vishny (1988), and Mc Connell and Servaes (1990) find a positive relation between Tobins Q and inside director shareholdings. Warfield et al. (1995) show a negative link between managerial stockholdings a nd the absolute value of abnormal accruals. They interpret their results as being consiste nt with managerial shareholdings acting as a disciplinary mechanism. However, incentive compensation may also induce executives to become short-term oriented, causing more seve re agency problems. For example, Burns and Kedia (2006) find that the firms are more likel y to misreport accounting information if the sensitivity between CEOs compensation and stock prices is high. Bergstresser and Philippon (2006) show that when the CEOs in centive pay ranks near the top decile, the CEO is likely to sell more shares and exercise more stock options in the current year. In the same vein, Aboody and Kasznik (2000) and Ye rmack (1997) show that CEOs manage investors earnings expectations downward prio r to scheduled stock option awards in order to increase the future value of their awards. Taken together, the empirical evidence suggests that if the CEO manages earnings to increase his overall compensation, then there will be a positive relation between CEO incentive and earnings management. While a large body of research on executive equity incentives exists, there is less research on the effect of the form of compensation on directors behavior. Unlike CEOs, the directors compensation contract that govern their continued participation in board activities are relatively less important than their other sources of income. In addition, directors investment portfolios are more diversified than that of CEOs since director compensation is rarely the only source of income. Nevertheless, directors compensation is sizeable (Yermack(2004)), especially for directors th at sit on multiple corporate boards (Fich and 8 For example, Fama(1980), Fama and Je nsen (1983), Ofek and Yermack(2000).
43 Shivdasani (2006) and Ferri s and Jagannathan et al. (2003)). This suggests that as is the case with managers, the directors compensation st ructure might also provide a mechanism that fosters an alignment of interests with shareholders. In this study I empirically address how elem ents of director compensation structure affect directors monitoring behavior. Specifica lly, I test whether directors incentive pay affects the level of firms earning management. If incentive pay induces directors to act as monitors, then earnings management is likely to be negatively related to the directors incentive pay. This study is in the strand of res earch of Fich and Shivdasani(2005) and Byard and Li(2005). Fich and Shivdasani(2005) show th at firms that offer stock options to their directors exhibit higher market to book ratios. Moreover, outside di rectors appointments produce near zero abnormal returns for firms w ith option plans but significantly negative abnormal returns for firms without them. Fich and Shivdasani (2005) indicate that stockoption plans help to align the interests of outsi de directors and shareholders. However, Byard and Li(2005) argue that when stock options are used as a common component of the compensation to the CEOs and directors, they can compromise directors independence and ability to monitor CEOs option timing opportunities. Consequently, one can argue that awarding stock options to directors may redu ce their incentives to monitor and therefore allow CEOs to manipulate earnings. Prior studies have found that board character istics are associated with the level of firms earnings management. For example, Klein (2002), among others, shows a negative relation between the independence of the a uditing committee of the board and abnormal accruals. However, her study does not account fo r the potential conflict arising from stock option grants to board directors and how it affects the relationship between directors
44 compensation and earnings management. In this paper, I empirically investigate whether directors incentive pay affects the firms earnings management using a samp le of S&P 500, S&P middle cap 400 and S&P small cap 600 firms. Following previous res earchers (i.e. Jones (1991), Teoh, Welch and Wong (1998)), I estimate the abnormal discretiona ry accruals from a group of cross sectional regressions estimated by year and industry. I find that earnings management is more severe among firms paying their directors with a highe r proportion of stock options. This result still holds when I control for firm fixed effects. I further examine how board characteristics and CEO incentives pay jointly determine earni ngs management. Conditioning on director incentives the firm paid, I show earnings manage ment is significantly larger for firms with high level of CEO incentive pay than for firm s with low CEO incentive pay, but only when directors incentive compensation is high as well. In the tests, I also control for self-selection bias. Specifically, it is possible that firms that have never paid stock options to their directors during the entire sample period could also be more likely to have a specific govern ance mechanism in place that would affect the level of firms earnings management. After c ontrolling for self selection bias, I find firms that awarded their board with stock options engage in more severe earnings management than those that never paid their directors w ith stock options. Moreover, after controlling for bias, the degree of earnings management is still positively related to the proportion of incentive pay the directors received. These results suggest that the interests of directors who receive higher levels of incentive pay tend be closely aligned with those of the CEO rather than the shareholders. I also examine directors option sales for th e years following severe earnings management.
45 The findings show large and significant options sales by directors following the years when the firms earnings management is in the top tercile of the sample firms. This is additional evidence in line with the notion that directors may join the CEO in an effort to manipulate the earnings so that they could maximize their benefit from exercising their options for the purpose of profiting or diversification. This paper sheds new light on the re lationship between director incentive compensation and earnings manage ment. Specifically the results show a positive relationship between director incentive pay and earnings ma nagement, suggesting that incentive pay may compromise the boards independence. Second, it provides an interpretation for the recent years phenomenon wherein firms decide to pay less incentive pay to the board of directors.9 The paper proceeds as follows. Section 1 pr esents related literature and hypotheses. Section 2 describes data and presents descrip tive statistics. Section 3 discusses the results. Section 4 presents robustness tests and Section 5 concludes. Literature Review Earnings Management Healey and Wahlen (1999) define earni ngs management as a process where managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influen ce contractual outcomes that depend on reported accounting numbers. Many researchers have shown evidence that earnings management is used to make firms financial statement l ook more optimistic. For example, Teoh, Welch and Wong (1989) show that firms are more likely to increase accruals before they go public or
46 issue seasoned equity offerings. It is argued that the managements use of earnings management has both costs and benefits. Benefits include potential improveme nts in managements credible communication of private information to external stockhol ders (Healey and Wahlen (1999)). A similar statement was also appeared in the financial press, the CEOs know that investors hate surprises, so they try to keep net income trending up a nice straight slop (Fortune 1989, 196). Goel and Thakor (2003) propose that uninformed liquidity investors are more likely to hold stocks with less volatile earnings as the informed investors make profits from their inside information when the earnings are more volatile. In this context, smoothing earnings by borrowing earnings from the next year when current earnings are actually low, or lending earnings when the previous year earnings ar e exceptionally high may provide a manipulative way to convince investors to hold on to the stock. This practice could also be beneficial to the firm because it reduces the variance of the firms observed earnings and thus reduces the cost of borrowing (Trueman and Titman ( 1999)). However, this argument relies on unrealistic economic and behavior assumptions. Watts (1982) argues that the investors are sophisticated enough to be able to undo such manipulations. Thus, the usefulness of CEOs manipulation of earnings for the sake of reducing earnings variance is suspect. Recent empirical research has also highli ghted the costs of earnings management. Beneish and Vargus (2002) find that abnormally high accruals are associated with increases in insider sales of shares but after the event period, stock returns tend to be poor. Recent work focusing on how CEO compensation is re lated to earnings management shows that CEOs are more likely to use earnings management to inflate stocks prices and thus affect their own wealth. For example, Cheng and Warfield (2005) and Bergstresser and Philippon 9 See Wall Street Journal Feb 24, 2003. pg. R. 4 for a special report on Corporate Governance
47 (2006) provide evidence that the use of discretionary accruals to manipulate reported earnings is more pronounced among firms where the CEOs potential total compensation is more closely tied to the value of the stock and option holdings. Francis, Nanda and Olsson (2007) use discretional accruals as a proxy for earnings quality and show that higher abnormal discretional accruals are positively related to the firms cost of capital, and that cost of capital has no effect on the voluntar y disclosure in the presence of earnings management measures. Their evidence further shows that higher accruals are more likely to be associated with insider sales of shares or options. Director Incentives Previous research documented that career and reputation concerns are constitute an important disciplinary mechanism that induces di rectors to act in the best interest of the shareholders. For example, Fama and Jensen (1983) posit that labor market pressure and concerns for reputation will lead directors to fulfill their duty. Consistent with this argument, a handful of studies show that vigilant direct ors that establish reputations as good monitors are expected to be rewarded with additiona l board seats in the labor market, while lax monitors are expected to be penalized with a reduction in board seats. For example, outside directors hold fewer board seats after serv ing in dividend-reducing firms (Kaplan and Reishus (1990), in companies th at experience financial distress (Gilson (1990)), and in firms that perform poorly (Yermack(2004)). In contra st, CEOs from firms that have performed well receive outside directorships after retirement (Brickley, Coles, and Linck (1999)). Using a sample of 111 public firms that e ither filed for bankruptcy or privately restructured their debt between 1979 and 1985, Gilson (1990) finds that, on average, only
48 46% of incumbent directors remained on the board at the end of bankruptcy or debt restructuring period. Director s who resign hold significantly fewer seats on other boards following their departure. Harford (2003) reports that all directors, and outside directors in particular, are unlikely to be retained following a completed ta keover offer. All target directors hold fewer directorships in the future than a control gr oup, suggesting that the target board seat is difficult to replace. If we view takeover as a disciplinary action, his findings support the notion that ineffective directors are rewarded with fewer directorship. However, this also implies that the incumbent directors including out side directors have an incentive to resist a takeover offer in order to keep their hard-to-replace director seats. Yermack (2004) finds a positive relation betw een the companys performance in the previous two years and the net acquisition of new board seats by outside directors over the four years after their appointments. Additiona lly, if a shorter time period is used, such relation disappears, suggesting that the mark et for directors takes time to assess and assimilate the monitoring ability of newly appointed directors. Another strand of research posits that direct or compensation is an important tool for motivating directors. The typical compensation c ontract of a director who typically works an average of 150 hours a year and sometimes sits on multiple boards include an annual retainer, board meeting fees, and restricted st ock and option awards. Th e rationale behind the recent trend of including stock award as part of director pay resembles the one for granting stock options to a CEO: giving an ownership st ake to the agent helps align his/her interests with those of the owners. Ther e is some evidence that directors who own more equity are better monitors. Perry (1999) finds that when directors of independent boards receive
49 incentive compensation, the likelihood of CEO turnover following poor performance increases. He also shows that the likelihood of a firm adopting a stock-based incentive plan for directors is positively related to the fraction of independent directors on the board. In spite of the benefits of improved monitoring, director incentive compensation could be a double edge sword, especially in light of the recent dramatic increase of the director incentive compensations. The exis ting research shows that stock based compensation can cause managers to manipulat e information in order to increase their compensation by transferring wealth from the stockholders to the managers. Byard and Li (2005) show that when directors receive a lo wer portion of compensati on from stock options, the CEOs are less likely to set up the option grant date before (after) the good news (bad news) and vise versa.. In centive pay leads to directors interest being more closely tied to the interests of the executives and reduces their incentives to monitor. Data and Variables Sample description The primary sample is retrieved from I RRC. The sample peri od is from 1996 to 2004. The information reported in the IRRC is then combined with data in Compustat and Compustat ExecuComp. I exclude all financia l firms with SIC 6000-6999 and require that firms granted directors stock options in at least one year during the entire sample period. The resulting final sample contains 6837 firm-year observations. Measures of director incentive
50 The director incentive measure is defined as the proportion of incentive pay over the directors total compensation during the give n year. IRRC reports the directors committee membership, director classification, shar es owned, pension and annual compensation. Directors receive compensation in the form of cash, stock and options. Since most company disclosures for directors equity pay are less detailed when compared with those for executives, the basic terms of these awards such as the date, the stock price when awarded, vesting, or restrictions on sale are often unava ilable for directors. I therefore follow Yermack (2006) and make a range of assumptions for valuing equity compensation and its incentive features. I assume all stock op tions are awarded at-the-money with 10 year lives. I first value options using the BlackScholes method. Vola tility and dividend yields for each firm-year are obtained for the vast majority of observa tions from the ExecuComp database. Then I compute total compensation by summing up the option value, annual retainer, board meeting fee and value of restricted stock. The direct or incentive variable is finally obtained by dividing the option value from the total compensation of the director. Measures of earnings management I use data from firms reported income st atements to compute accrual measures. My method closely follows that of Dechow et al (1995), which estimates discretionary accruals from regressions of total accruals on changes in sales and on property, plant, and equipment (PPE) within industries. I obtained accounting information from the Compustat Annual Industrial, Research and Full Coverage files. All firm-year observations should satisfy the following criteria: (1) domestic firms; (2) firms with non missing values for sales, total assets, net
51 income before extraordinary items, and cash from operations; (4) non-financial firms. Ultimately, I am able to estimate discretionary accruals for 57,903 firm-year observations over the period from 1996 and 2004. To determine discretionary accruals, I fi rst run the following cross-sectional OLS regression for each combination of calendar year and two-digit SIC code with a minimum of 8 observations to estimate coefficients b0, b1, and b2. 012 11111jtjtjt j t jtjtjtjtTACSALESPPE bbb TATATATA,,,, where j indexes firms, t indexes time, Total accr uals(TAC) equals Net Income (Compustat item #172) minus Cash Flow from Operations (#308). Sales is the changes in sales revenues (#12), PPE is gross property, plant, and equipment (#7) All variables used here are scaled by total assets(TA) at the beginni ng of the period (#6) to reduce heteroskedasiticity. I estimate the earning using cross-sectional models. I then use the estimated coefficients to calculate nondiscretionary accruals (NDTAC) as follows: 012 1111ititit it itititSALESARPPE NDTACbbb TATATA,,, where, AR is the dollar value of Accounting Receivable (#2) Thus, I can derive discretionary accruals (DTAC) as: 1 it itit itTAC DTACNDTAC TA, Since earnings manipulation involves both positiv e and negative discretional accruals, I use the absolute value of discretional accruals to measure the level of earnings management of the firm. To reduce possible problems from outliers, I also winsorize the absolute
52 discretional accrual variable at the 1% level. 10 As all the variables are scaled by total assets at the beginning of the period, the magnitude of a firms discretionary accruals is indicated as a percentage of the firms assets. Measures of option sale I also collect the data on board directors option sales from Thomas Financial Insider Trading database. I gather option sales by direct ors in each calendar year and scaled it by the firms shares of outstanding. Other control variables Other control variables I retrieve to explai n the firms earnings management behavior are profitability, size, market to book, financial leverage, institutional ownership and current growth. Profitability is proxied by ROA, which is obtained as the net income scaled by the total assets at the beginning of fiscal year. Fi rms that are more profitable have less incentive to adjust the earnings upward to cover a fi nancial problem and thus suggest a negative relationship between ROA and abnormal accruals. Large firms are followed by more external capital markets and receive more analyst coverage while small firms are less like to r eceive scrutiny about their accounting statement. This indicates a negative relationship between earnings management and size (Dechow et al.(1995)). Institutional investors serve as the exte rnal monitors. The firms with large institutional ownership would be less likely to hide their earnings with abnormal accruals 10 The results are consistent even when we do not winsorize the accrual measures.
53 (Chung et al. (2002)). Similar to institutional investors, the control of the book leverage, proxied by the total long term debt over the to tal assets, implies a negative relationship as firms will receive closer scrutiny from lenders when the debt level is high. But on the other hand, firms will have more incentive to adjust ear nings because they do not want to lead to a debt-covenant violation when the performance is poor and thus miss earnings target set by the lenders (DeFond and Park (1997)). Previous studies also argue that earnings management is an interaction between the current stockholders and future stockholde rs (Bolton, Scheinkman, and Xiong (2005)). Although its effect on the earnings management is not clear, I control it with the current asset growth, which is defined as change of total assets scaled by the total assets at the beginning of fiscal year Firms with higher growth rate are more lik ely to overinvest in current assets in anticipation of future growth of sales. This practice leads to a positive relationship between earnings management and growth rate. The gr owth rate is proixed by the market to book ratio. Descriptive statistics Table 13 provides descriptive statistics fo r the director compensation, director incentive and earnings management measures. Pa nel A presents details of the director compensation variable. Across 6,839 firm-year observations, the mean annual retainer is $22,260, with a standard deviation of $14,700. Th e Annual meeting fee, the product of number of board meetings and meeting fee, has a mean value of $7,740 with a standard deviation of $7,380. On average, restricted stoc k, while reflects the value of the restricted
54 stock newly issued when the board meeti ng holds, is $2,840 per year. The option awards accounts for the largest part of the direct or compensation. Based on the Black-Scholes formula, the option value stands at $103,160 for an average director in the sample. The variation of the option pay is substantial, as indicated by a standard deviation of $262,180. To further assess how the average leve l of director incentive and earnings management varies over time and with one another, I also provide their mean values for each year from 1996 to 2004 in Panel B and C, respectively. As shown in Panel B, the option portion of the payment in director compensa tion climbs each year from 1996 until it peaked in 2002 and then started to decline. Panel C shows the yearly distribution of the mean absolute discretionary accrual. In 2000 and 2004, the evidence indicates a severe earnings management problem on average with absolute discretional accruals of 0.21 and 0.28 respectively. However, the median shows a more consistent pattern of earnings ma nagement over time, thus suggesting that the mean values could be reflective of extreme outliers. Table 13 Descriptive statistics of earnings management and incentive measure by years Panel A: Board director Compensation (in $ thousand) Sample size Mean Median Standard deviation Annual Retainer 6837 22.26 20 14.7 Annual Meeting Fee 6837 7.74 7 7.38 Restricted Stock 5605 2.84 0 8.45 Option 6694 103.16 41.08 262.18
55 Table 13 Descriptive statistics of earnings management and incentive measure by years(continue) Panel B: Director Incentive 1996 544 0.39 0.37 0.33 1997 670 0.47 0.52 0.33 1998 694 0.49 0.55 0.33 1999 700 0.51 0.57 0.32 2000 756 0.53 0.58 0.34 2001 806 0.56 0.62 0.33 2002 865 0.59 0.67 0.31 2003 862 0.55 0.61 0.3 2004 797 0.53 0.61 0.33 Total 6694 0.52 0.58 0.33 Panel C: Absolute Discretional Accrual 1996 544 0.06 0.04 0.06 1997 677 0.07 0.04 0.09 1998 730 0.07 0.04 0.10 1999 729 0.09 0.04 0.17 2000 785 0.20 0.05 0.38 2001 848 0.08 0.05 0.12 2002 865 0.08 0.04 0.12 2003 862 0.08 0.04 0.26 2004 797 0.23 0.04 0.59 Total 6837 0.11 0.04 0.28 Table 14 provides summary statistics on main variables and a first look at whether measures of director incentive and earnings ma nagement are correlate d. Averaging across all firm-year observations, the mean (median) ab solute discretionary accrual is 0.12 (.04) and the sample standard deviation is 0.39. The mean (median) director incentive is 52% (58%) of the directors total compensation, with a sta ndard deviation of 0.33. The correlation between director incentive and absolute discretionary accruals is 0.08, significantly at the 1% level, which suggests that a higher portion of incentive pay in the director compensation is related with a more severe earnings management probl em. The absolute discretionary accruals are
56 also negatively correlated with ROA, which indi cates that profitable firms have less severe earnings manipulation problems. The director incentive measure is negatively correlated with the book to market ratio. Table 14 Summary statistics and correlations of earnings management and incentive measures Panel A Sample size Mean Median Standard deviation Absolute discretionary accrual 6837 0.12 0.04 0.39 Director incentive 6694 0.52 0.58 0.33 ROA 6836 0.03 0.05 0.18 Book to market 6830 0.53 0.43 0.52 Total asset(in million dollars) 6837 4690.38 1149.74 15747.08 Asset growth 6836 0.14 0.07 0.38 Institutional ownership 6837 0.62 0.63 0.17 Panel B Correlation(p-value) Director incentive ROA Book to market Total asset Asset growth Institutional ownership Absolute discretionary accrual 0.08 -0.18 -0.02 -0.02 0.08 0.01 (0.00) (0.00) (0.11) (0.12) (0.00) (0.55) Director incentive 1.00 -0.02 -0.16 -0.12 0.10 0.17 (0.14) (0.00) (0.00) (0.00) (0.00) ROA 1.00 0.09 0.09 0.14 0.12 (0.00) (0.00) (0.00) (0.00) Book to market 1.00 -0.04 -0.13 -0.09 (0.00) (0.00) (0.00) Total asset(in million dollars) 1.00 0.06 0.08 (0.00) (0.00) Asset growth 1.00 0.09 (0.00) Institutional ownership 1.00
57 Empirical Results Univariate Analysis In Table 15, I test for differences in earnings management acro ss groups of firms formed after sorting on different variables of interest. More specifically, I sort by director incentives, ROA, market to book ratio, log(ma rket value), institutional owne rship, and book leverage, and test for differences of mean values of earnings management across the highest tercile and lowest tercile groups. For example, after sorting firms in the top and bottom tercile of director incentive, the mean earnings management is s hown separately for the sub-samples of high-director incentive firms and low-director incentive firms. Table 15 Univariate analysis Director incentive ROA Market to book Log(market value) Asset growth Institutional ownership Book leverage High Group 0.144 0.107 0.139 0.111 0.121 0.102 0.098 Low Group 0.083 0.133 0.079 0.121 0.117 0.125 0.135 Difference 0.061 -0.026 0.060 -0.009 0.005 -0.024 -0.037 P-value (0.000) (0.000) (0.000) (0.289) (0.558) (0.000) (0.000)
58 Not surprisingly, firms with high market to book ratio have significantly higher earnings management than firms with low market to book ratio (0.139 versus 0.079). This difference is also statistically significant (p -value=0.000). Earnings management is also larger for high director incentive than for low director incentive stocks and the difference is also statistically significant at 1% level. This finding provides preliminary evidence that stock option pay for directors ma y provide encouragement, or at least tolerance, for earnings management. Earnings management is much higher for low than for high institutional ownership firms (0.125 versus 0.102), consistent with the view that instit utional investors act as external monitors, thereby reducing earnings management. Overall, the evidence from table 3 is consistent with that of the corre lation evidence found in Table 2. However, this univariate analysis is still not sufficient enough to draw definitive conclusions as it does not control for many other variables simultaneously. Thus, the next section proceeds with a series of multivariate tests to further exam ine the relationship between directive incentive pay and earnings management. Regressing earnings management on director incentives Table 16 provides multivariate regression anal ysis of the effect of director option compensation on earnings management. I c ontrol for ROA market-to-book, size (log of market value), asset growth, institutional ownership, book leverage, year dummies (not reported), and Fama-French 48 industry dummies (not reported). The Whites (1980) method is used to control for heteroskedasticity, and all p-values are reported with robust standard errors. The regression is clustered at the indivi dual firm level to reduce the serial correlation across each firm over different years. The first column contains the OLS regressi on and the second column the fixed effects
59 regression results. The findings are consistent with the univariate evidence in Tables 2 and 3. Specifically, the coefficient for director incen tive is 0.023, and statistically significant with a p-value of 0.017. This result shows that when director incentives increase by 1, earnings management will increase by 0.023. Or put anot her way, if the percentage of option compensation to total compensation increases fr om 50% to 80%, which is about 1 standard deviation change of director incentive, th e earnings management will increase by 0.69%. Given that the absolute discretionary accrual is scaled by total assets at the beginning of each year, 0.69% represents a $3.236 million increase in discretionary accruals on average. In line with the univariate analysis ROA, institutional ownership and leverage have significant negative effects on the firms earnings manage ment, while market to book and asset growth have a positive impact on earnings management. Overall, the result suggests that the firms are more likely to manipulate their earnings when the directors are paid with higher incentive compensation. Table 16 Director incentives and earnings management Dependent variable: absolute discretionary accrual OLS Fixed Effect Director incentive 0.023 0.080 (0.017)** (0.000)*** ROA -0.216 -0.283 (0.000)*** (0.000)*** Market to book 0.005 0.006 (0.035)** (0.006)*** Log(market value) -0.001 -0.014 (0.715) (0.091)* Asset growth 0.070 0.096 (0.000)*** (0.000)*** Institutional ownership -0.043 0.014 (0.043)** (0.775) Book leverage -0.034 -0.066 (0.066)* (0.020)** Observations 6665 6665 Adjusted R-squared 0.22 0.08 Number of standard and poor's identifier 1281
60 Board characteristics and earnings management Klein (2005) and Xie et al. ( 2003) show that a more independent board may lead to lower earnings management and that an i ndependent auditing committee may also reduce earnings manipulations. Cornett, Marcus and Tehr anian (2008) further show that an effective governance structure, such as more independent board, will decrease the earnings management. Thus, the analysis in Table 17 cont rols for different board characteristics. In particular, the following variables are added: a board independence indicat or that is equal to 1 if over 51% of board members are independen t; the percent of independent directors on the auditing committee; an auditing committee indepe ndence indicator variable; and an indicator variable equal to 1 if the auditing committee is 100% independent. In all four regressions shown in Table 5, director incentive pay s hows a significant positive relationship with earnings management. The coefficients for di rector incentives range from 0.023 to 0.027 and are statistically significant at the 5% level. The board characteristics variables are not statistically significant. Thus, these findings provide further evidence that stock option pay for directors is associated with higher earnings management, even after controlling for other important board characteristics.
61 Table 17 Director incentives and earnings management: controlling for board characteristics Dependent variables: Absolute discretional accrual Director incentive 0.023 0.027 0.028 0.027 (0.017)** (0.019)** (0.017)** (0.020)** ROA -0.216 -0.203 -0.204 -0.203 (0.000)*** (0.000)*** (0.000)*** (0.000)*** Market to book 0.005 0.004 0.004 0.004 (0.035)** (0.053)* (0.054)* (0.053)* Log(market value) -0.001 -0.002 -0.002 -0.002 (0.732) (0.429) (0.433) (0.444) Asset growth 0.070 0.074 0.074 0.074 (0.000)*** (0.000)*** (0.000)*** (0.000)*** Institutional ownership -0 .042 -0.050 -0.048 -0.050 (0.046)** (0.038)** (0.044)** (0.036)** Book leverage -0.034 -0.032 -0.032 -0.032 (0.067)* (0.127) (0.133) (0.128) Independent board (0,1) indicator -0.002 (0.715) % independent directors on the audit committee 0.016 (0.309) Independent auditing committee (0,1) indicator -0.001 (0.919) 100% independent directors on the auditing committee (0,1) indicator 0.008 (0.269) Observations 6665 5454 5456 5456 Adjusted R-squared 0.22 0.24 0.24 0.24
62 CEO incentive and earnings management Previous studies, such as Peasnell et al. (2005), have documented that board directors help constrain earnings management, while Bergstresser and Philippon (2006) show that firms with high CEO incentive pay are likely to be involved in high levels of earnings management. Table 18 presents univariate analysis of CEO incentives on earnings management conditioning on the director incentiv e. I assign firm into three terciles after sorting on director incentive. Within each dir ector incentive tercile, I further segmented the data into three sub-terciles after sorting on the CEO incentive pay, which is measured as the proportion of CEO option compen sation over the CEO total compensation. The result shows that when director incentive pay is in the low or median tercile, there is no significant difference in earnings management between th e high CEO tercile and the low CEO tercile subgroups. However, when the director incentive is in the high tercile, the differences in the earnings management between the high CEO incentive tercile and low CEO incentive tercile subgroups is about 0.053, which is statistically significant with a p-value of 0.000. This evidence that CEO incentive pay has an impact on earnings management only when director incentive is high, lends support to the notion that the directors are more likely to form an alliance with CEO in manipulating earnings. Table 18 Univariate test: directors incentives and earnings management: Controlling for CEO incentives Absolute Discretional Accruals Low Director Incentive (1) (2) Low Tercile CEO Incentive 0.089 0.078 High Tercile CEO Incentive 0.081 0.084 Difference 0.008 -0.006 P-Value (0.778) (0.280)
63 Table 18 Univariate test: directors incentives and earnings management: Controlling for CEO incentives (continue) Median Director Incentive Low Tercile CEO Incentive 0.104 0.077 High Tercile CEO Incentive 0.094 0.084 Difference 0.010 -0.007 P-Value (0.755) (0.202) High Director Incentive Low Tercile CEO Incentive 0.115 0.091 High Tercile CEO Incentive 0.168 0.149 Difference -0.053 -0.058 P-Value (0.001)*** (0.000)*** To further address the concern on the eff ect of director incentive on constraining CEOs earnings manipulation. I sort the sample into three terciles by the CEO incentive and then run the main regressions discussed in S ection 3.2. The result is not reported for the sake of brevity, the coefficients of director incen tives are only statistically significant for the subsample where CEO incentive is at the top terc iles. When CEO incentives are at lower or middle tercile, director incentives are not significant. This confirms that the director incentive affects earnings management only when CEO incentive and director incentive are high. Further, I add two control variables: high director incentive high CEO incentive dummy and low director incentive high CEO incentive dummy. As shown in table 18, the result indicates that although the director in centive is still significantly positive, but the group of the company year having high direct ors incentive and high CEO incentives have significantly larger absolute abnormal discre tional accruals while the group with high CEO incentive low director incentive are significantly indifferent from the other groups with low CEO incentives in terms of earnings management. This result suggests is consistent with our universal test. The CEO incentive does affect th e firms abnormal discretional accruals when
64 coupled with high directors incentives award to the board directors. Table 19 Directors incentives and earnings management: Controlling for CEO incentives Dependent variables: Absolute discretional accrual Director incentive 0.036 (0.007)*** Low Director Incentive X High CEO Incentive 0.002 (0.748) High Director Incentive X High CEO Incentive 0.019 (0.064)* ROA -0.273 (0.000)*** Market to book 0.007 (0.016)** Log(sales) -0.001 (0.839) Asset growth 0.085 (0.000)*** Institutional ownership -0.046 (0.083)* Book leverage -0.07 (0.000)*** Constant 0.243 (0.000)*** Observations 6670 Adjusted R-squared 0.19 Controlling for self selection bias The tests thus far show that director incentive pay has a positive, significant relationship with the degree of earnings mana gement. However, the sample only includes those firms that pay stock options to their di rectors and, thus, exclude s firms that do not pay their directors any stock options. Therefore, the previous findings could be attributed to the lack of control for selection bias, which may ex ist, because the decision to pay the directors stock options is not random. For instance, a firm that is involved in high earnings
65 management would be likely to pay its direct ors stock options and therefore, based on the sample criteria, it would be included in the sa mple. The director incen tive effect on earnings management could be distorted without controlling for self-selection biases To address this issue, I first conduct a univariate test by comparing earning management across the subsample of firms that pay their directors w ith stock options and firms that do not award any options at all. Table 20 Earnings management for incentive vs. no incentive paid companies Absolute discretional accrual N Mean Median Incentive paid companies 6837 0.109 0.042 Non-incentive paid companies 1894 0.084 0.034 Difference 0.026 0.008 (0.000)*** (0.000)*** As shown in Table 20, there are 1894 firm-year observations with no incentive pay to directors. When I compare the average earni ngs management between these two groups, I see that non-incentive paying companies have mean absolute discretional accruals of 0.088 in contrast to 0.119 for the incentive paying comp anies. The difference of the mean absolute discretional accruals is statis tically significant as supported by a t-statistic of 3.144. Looking at the median value of the absolute discreti onal accrual also leads to the same conclusion. The median absolute discretional accrual is 0.042 for incentive paying firms and the median absolute discretional accrual for non incentive paying companies is 0.034. To address the self-selection bias issue, we use a two-stage Heckman selection model that controls for the probability of a firm paying directors incentive compensation. The first stage is a probit regression to estimate the probability of the firm awarding the directors with
66 a stock option plan. The predicted probability of paying incentive options obtained from the first stage regression will be added as a control variable in the second stage regression. In the first stage probit regression, the depende nt variable is an incentive paid dummy that is equal to one if the firm pays dir ectors incentive pay, or zero otherwise. Following Fich and Shivdasani (2005), I in clude the following control variables that could help explain the firms decision to award directors with stock options: board independence dummy, the natural logarithm of one plus the annual re tainer, new CEO dummy, CEO of retirement age dummy, an indicator of whether the directors r eceive a pension plan, percentage of directors who are CEOs in other firms, CEO-chairman dummy, dividend yield, ma rket adjusted stock return for the last 12 months and the natural l ogarithm of sales. The results, shown in Table 21, indicate that the firm is more likely to pay directors with stock options when it has a more independent board, the director annual re tainer is high, and the CEO is new and at retirement age. This result is consistent with the findings of Fich and Shivdasani (2005). Directors are reluctant to adopt stock option pl an in their compensati on package as the stock option will make their portfolio less diversifie d and thus more risky. When the cash-based annual retainer is high, the board is more lik ely to adopt the stock option plan. The more independent the board, implying less insider ownership, the more likely it is to adopt the stock option plan. But when the CEO also chairs the board, the board is less likely to adopt the stock option plan. This result is not in conflict with the relationship between board effectiveness and earnings management. It s hows that more independent boards are more likely to adopt the stock option pl an, but we fail to conclude that independent boards lead to severe earnings management. Thus, our study is not inconsistent with some other studies have shown that independent boards can help reduce earnings management (for example, see Cornett, Marcus and Tehranian (2008)).
67 The second stage regression test includes previously employed control variables and the predicted probability to pay directors incentive compensation from the first stage regression as a new control variable. The coeffici ent of the fitted value of director incentive is still strongly significant. Moreover, compared to our previous results, the coefficient has increased from 0.023 to 0.043, which is about 100% larger in magnitude. The predicted probability of incentive pay is significant but w ith a positive sign, which is consistent with the univariate analysis findings. The results show that firms without incentive pay, on average, have lower earnings management than those firms which paid directors with incentive compensation. In addition, the lambda statistics from the selection model has a negative coefficient of -0.095 and significant at 0.000 level, which indicates that selection biases exists and the result has a downward bi ases without controlling the firms probability to pay directors incentive compensation. This self-selection test supports our result that the earnings management is more severe when the directors receive incentive compensation even after we control the factors that aff ect the firms decision to provide incentive compensation, the coefficient of director in centive compensation variable moves up from 0.02 to 0.043. Table 21 Earnings management for incentive vs. no incentive paid companies: self selection analysis Dependent variable: Absolute discretional accrual First stage (Incentive paid dummy) Second stage Director incentives 0.043 (0.000)*** ROA -0.266 (0.000)***
68 Table 21 Earnings management for incentive vs. no incentive paid companies: self selection analysis (continue) Market to book 0.007 (0.000)*** Log(market value) 0.001 (0.685) Asset growth 0.073 (0.000)*** Institutional ownership -0.032 (0.066)* Book leverage -0.053 (0.000)*** Predicted incentive paid (0,1) indicator 0.143 (0.000)*** Independent board (0,1)indicator 0.247 (0.000)*** Log(1+annual retainer) 0.124 (0.000)*** New CEO 0.147 (0.017)** CEO of retirement age 0.185 (0.000)*** Pension plan to director -0.063 (0.279) Percentage of directors who are CEOs of other firm 0.280 (0.052)* CEO chairs the board -0.111 (0.002)*** Dividend yield -0.133 (0.615) Market adjust 12-month stock return 0.023 (0.337) Log(sales) -0.067 (0.000)*** Lambda statistics -0.095 (0.000)*** Observations 8533 8533 Earnings management and directors option sale The evidence thus far is consis tent with the notion that director incentive pay inhibits the directors monitoring effectiveness a nd thus leads to higher level of earnings management. Next, I investigate whether direct ors understand and take advantage of the high
69 earnings management activity of the firm. I coll ect information on directors options sale for the year t+1 for all sample firms. The dependent variable is total director options realized in year t+1. The control variables are a high-earni ngs management dummy that takes the value of 1 if the firms earnings management is in th e top tercile in year t, change of cash flow (#308), market return and log(sales). In the fi rst column of Table 10, the coefficient for the high-earnings management dummy is 0.067, signifi cant at the 5% level. When all other control variables are included, the coefficient is 0.059, significant at the 10% level. This finding indicates that for firms with earnings management level in the top tercile in the previous year, the current year directors options realization increases by approximately 0.06 to 0.07. Table 22 Earnings management and insider sell Directors Option Sell Directors Option Sell Absolute discretional accrua ls (Top Tercile) 0.059 0.067 (0.060)* (0.049)** Stock market return -0.008 (0.701) Change of cash flow 0.288 (0.095)* Log(sales) -0.127 (0.000)*** Observations 2194 2196 Adjusted R-squared 0.11 0.05 Additional robustness tests Given the small time variation of the direct or incentive, it is reasonable to assume that firms are less likely to change the stru cture of the board compensation. I therefore employ in my analysis the cumulative change of director incentives over a three year period from year t-3 to year t. As shown in Table 23, the dependent variable is the change of earnings management from year t-3 to year t, and all independent variables are also
70 measured as changes over the same period. As in Table 4, I use year and industry dummies to control for time and industry effects, but thei r coefficients are not reported for the sake of brevity. The number of observations drops to 3195 and the overall R-square drops to 17% from 22% in Table 4. The coeffi cient of director incentives is 0.07, which is about 3.5 times of the corresponding coefficient in Table 4 a nd strongly significant at the 1% level. The significance for the other variables is retained except for the coefficient of the change in institutional ownership, which becomes insignificant. Table 23 Cumulative change of director incentives and earnings management Dependent variable: change of absolute discretional accrual Change of director incentive 0.072 (0.002)*** Change of ROA -0.172 (0.211) Change of market to book 0.008 (0.079)* Change of log(market value) -0.032 (0.005)*** Change of asset growth 0.156 (0.000)*** Change of institutional ownership -0.026 (0.645) Change of book leverage -0.103 (0.011)** Observations 3195 Adjusted R-squared 0.17 The results are robust to other sample a nd model specifications. For example, Table 24 focuses on a subsample of firms with indepe ndent boards, i.e., firms for which 51% or more of the board directors are independent. Th e regression result, presented in column 1 in Table 11 shows that director incentive pay a ffects earnings management even when the
71 majority of the board are independent directors. This relationship is significant at the 1% level. Second, I include in the sample only fi rms with an independent auditing committee, i.e., firms for which 51% or more of th e auditing committee members are independent directors. The purpose of excluding all firms without an independent auditing committee is to re-test the relationship for a subset of fi rms which potentially have a less severe earnings management problem Again, the director incentive variable is positive and significant at the 1% level. Thus, these results indicate that whether a board is independent or not does not materially change the relationship between di rector incentive pay a nd earnings management. Table 24 Director incentives and earnings management: sub-samples Dependent variables: Absolute discretional accrual (1) (2) Director incentive 0.043 0.029 (0.000)*** (0.003)*** ROA -0.213 -0.210 (0.000)*** (0.000)*** Market to book 0.005 0.005 (0.036)** (0.036)** Log(market value) -0.000 -0.001 (0.874) (0.572) Asset growth 0.049 0.063 (0.008)*** (0.000)*** Institutional ownership -0.046 -0.048 (0.064)* (0.020)** Book leverage -0.022 -0.035 (0.316) (0.066)* Observations 5026 6262 Adjusted R-squared 0.22 0.21 I consider alternative measures of earni ngs management and directors incentive pay. In the Table 25, column 1, I use the absolute di scretional accrue estimated as Jones model. In column 2 and 3, I used a standardized di rector incentive measure instead of measuring director incentive in percentage terms. For the sake of brevity, I do not report the coefficients
72 for other control variables. Table 12 shows that the result remains robust. For one standard deviation change of director incentive, the absolute discretional accrual would change by 0.8%. Economically, this number is not small given that the average total asset is about 4,690 million dollars in the sample. A 0.8% percentage change means that 37.52 million dollars change of absolute discretional accruals. Table 25 Directors incentives and earnings management: alternative measures Dependent variables: Absolute discretional accrual (1) (2) (3) Director incentive 0.024 (0.017)** Director incentive (standardized) 0.008 0.008 (0.017)** (0.017)** Observations 6665 6665 6665 Adjusted R-squared 0.22 0.22 0.22 Conclusion This paper finds that director incentiv e compensation has a positive and significant relationship with the level of the firms earni ngs management. To identify the effect of the director incentive on firms earnings management, I employ a series of tests that control for self selection and provide robust evidence using alternative earnings management and directors compensation measures. This paper cont ributes to two strands of research. First, while some policymakers and researchers belie ve that incentive compensation for directors may lead to an alignment of their interest with that of the stockholders this paper shows that high incentive pay may lead to a higher degr ee of earnings management. Second, previous studies show that high CEO incentive pay woul d lead to high earning management, but they
73 fail to control for director incentives. I find th at after controlling for director incentives, it becomes apparent that CEO incentives lead to higher earning manipulation levels only if directors incentive pay is high as well.
74 References Aboody, D., Kasznik, R., 2000. CEO stock option awards and the timing of corporate voluntary disclosures. Journal of Accounting & Economics 29, 73-100 Adams, R.B., Ferreira, D., 2007. A theory of friendly boards. Journal of Finance 62, 217250 Aggarwal, R., Williamson, R., 2006. Did New Regulations Target the Relevant Corporate Governance Attributes? SSRN. Beneish, M.D., Vargus, M.E., 2002. Insider trading, earnings quality, and accrual mispricing. The Accounting Review 77, 755-791 Berger, A.N., DeYoung, R., 2001. The Effects of Geographic Expansion on Bank Efficiency. Journal of Financial Services Research 19, 163-184 Bergstresser, D., Philippon, T., 2006. CEO incentives and earnings management. Journal of Financial Economics 80, 511-529 Bhagat, S., Black, B.S., 2002. The non-correlation between board independence and long-term firm performance. Journal of Corporation Law 27, 232-273 Bolton, P., Scheinkman, J., Xiong, W., 2006. Executive Compensation and Short-termist Behavior in Speculative Markets. Review of Economic Studies 73, 577-610 Brickley, J.A., Linck, J.S., Coles, J.L., 1999. What happens to CEOs after they retire? New evidence on career concerns, horizon problems, and CEO incentives. Journal of Financial Economics 52, 341-377 Bryan, S., Hwang, L., Lilien, S., 2000. CEO Stoc k-Based Compensation: An Empirical Analysis of Incentive-Intensity, Relative Mix, and Economic Determinants. Journal of Business 73, 661-693 Burns, N., Kedia, S., 2006. The impact of performance-based compensation on misreporting. Journal of Financial Economics 79, 35-67 Butler, A.W., 2007. Distance Still Matters: Evidence from Municipal Bond Underwriting. Review of Financial Studies Forthcoming Byrd, J.W., Hickman, K.A., 1992. Do outside directors monitor managers? : Evidence from tender offer bids. Journal of Financial Economics 32, 195-221 Chan, K., Covrig, V., Ng, L., 2005. What determines the domestic bias and foreign bias? Evidence from mutual fund equity allocations worldwide. Journal of Finance 60, 1495-1534
75 Cheng, Q., Warfield, T., 2005. Equity incentives and earnings management. The Accounting Review 80, 441-476 Choi, J., Kim, J.-B.V., Qiu, A., Zang, Y., 2007. Auditor Locality, Audit Quality and Audit Pricing. SSRN. Chung, K.H., Pruitt, S.W., 1994. A Simple Approximation of Tobins-Q. Financial Management 23, 70-74 Chung, R., Firth, M., Kim, J.B., 2002. Institutional Monitoring and Opportunistic Earnings Management. Journal of Corporate Finance 8, 29-48 Conyon, M.J., Peck, S.L., 1998. Board control, remuneration committees, and top management compensation. Academy of Management Journal 41, 146-157 Core, J.E., Guay, W.R., 2001. Stock option plans for non-executive employees. Journal of Financial Economics 61, 253-287 Core, J.E., Guay, W.R., 2002. Estimating the Value of Employee Stock Option Portfolios and Their Sensitivities to Price and Volatility. Journal of Accounting Research 40, 613-630 Core, J.E., Holthausen, R.W., Larcker, D.F., 1999. Corporate governance, chief executive officer compensation, and firm performance. Journal of Financial Economics 51, 371-406 Cornett, M.M., Marcus, A.J., Tehranian, H., 2008. Corporate governance and pay-forperformance: The impact of earnings management. Journal of Financial Economics 87, 357-373 Coval, J.D., Moskowitz, T.J., 1999. Home bias at home: Local equity preference in domestic portfolios. Journal of Finance 54, 2045-2073 Coval, J.D., Moskowitz, T.J., 2001. The geography of investment: Informed trading and asset prices. Journal of Political Economy 109, 811-841 Dechow, P.M., Sloan, R.G., Sweeney, A.P., 1995. Detecting Earnings Management. The Accounting Review 70, 193-225 DeFond, M.L., Park, C.W., 1997. Smoothing income in anticipation of future earnings. Journal of Accounting and Economics 23, 115-139 DeFusco, R.A., Johnson, R.R., Zorn, T.S., 1990. The Effect of Executive Stock Option Plans on Stockholders and Bondholders. Journal of Finance 45, 617-627 Degeorge, F., Patel, J., Zeckhauser, R., 1999. Earnings Management to Exceed Thresholds. Journal of Business 72, 1-33 Degryse, H., Ongena, S., 2005. Distance, le nding relationships, and competition. Journal
76 of Finance 60, 231-266 Del Guercio, D., Dann, L.Y., Partch, M.M., 2003. Governance and boards of directors in closed-end investment companies. Journal of Financial Economics 69, 111-152 Desai, H., Hogan, C.E., Wilkins, M.S., 2006. The reputational penalty for aggressive accounting: Earnings restatements and management turnover. Accounting Review 81, 83-112 DuCharme, L.L., Malatesta, P.H., Sefcik, S. E., 2004. Earnings management, stock issues, and shareholder lawsuits. Journal of Financial Economics 71, 27-49 Dvorak, T., 2005. Do domestic investors have an information advantage? Evidence from Indonesia. Journal of Finance 60, 817-839 Fama, E.F., 1980. Agency Problems and the Theory of the Firm. Journal of Political Economy 88, 288-307 Fama, E.F., Jensen, M.C., 1983. Agency Problems and Residual Claims. Journal of Law & Economics 26, 327-349 Fama, E.F., Jensen, M.C., 1983. Separation of Ownership and Control. Journal of Law and Economics 26, 301-325 Farrell, K.A., Whidbee, D.A., 2003. Impact of firm performance expectations on CEO turnover and replacement decisions. Journal of Accounting & Economics 36, 165196 Fenn, G.W., Liang, N., 2001. Corporate payout policy and managerial stock incentives. Journal of Financial Economics 60, 45-72 Ferris, S.P., Jagannathan, M., Pritchard, A.C., 2003. Too busy to mind the business? Monitoring by directors with multiple board appointments. Journal of Finance 58, 1087-1111 Fich, E.M., 2005. Are Some Outside Directors Better than Others? Evidence from Director Appointments by Fortune 1000 Firms. Journal of Business 78, 19431971 Fich, E.M., Shivdasani, A., 2005. The Impact of Stock-Option Compensation for Outside Directors on Firm Value. Journal of Business 78, 2229-2254 Fich, E.M., Shivdasani, A., 2006. Are busy boards effective monitors? Journal of Finance 61, 689-724 Fisk, G., 1971. Geography and Retailing Scott,P. Journal of Business 44, 235-235 Fortune 1989, Cute tricks on bottom line, (April, 24): 193-200
77 Francis, J., Nanda, D., Olsson, P., Voluntary Disclosure, Earnings Quality, and Cost of Capital. Journal of Accounting Research Forthcoming Gerety, M., Hoi, C.K., Robin, A., 2001. Do shareholders benefit from the adoption of incentive pay for directors? Financial Management 30, 45-61 Gilson, S.C., 1990. Bankruptcy, boards, banks, and blockholders: Evidence on changes in corporate ownership and control when firms default. Journal of Financial Economics 27, 355-387 Goel, A.M., Thakor, A.V., 2003. Why Do Firms Smooth Earnings? The Journal of Business 76, 151-192 Grinblatt, M., Keloharju, M., 2001. How di stance, language, and culture influence stockholdings and trades. Journal of Finance 56, 1053-1073 Harford, J., 2003. Takeover bids and target directors' incentives: the impact of a bid on directors' wealth and board seats. Journal of Financial Economics 69, 51-83 Hartzell, J.C., Starks, L.T., 2003. Institutional Investors and Executive Compensation. Journal of Finance 58, 2351-2374 Hau, H., 2001. Location matters: An examination of trading profits. Journal of Finance 56, 1959-1983 Haugen, R.A., Senbet, L.W., 1981. Resolving the Agency Problems of External Capital through Options. Journal of Finance 36, 629-647 Hauswald, R., Marquez, R., 2006. Competition and strategic information acquisition in credit markets. Review of Financial Studies 19, 967-1000 Healy, P.M., Wahlen, J.M., 1999. A review of the earnings management literature and its implications for standard setting. Accounting Horizons 13, 365-383 Hermalin, B.E., 2005. Trends in corporate governance. Journal of Finance 60, 2351-2384 Hermalin, B.E., Weisbach, M.S., 1991. The Effects of Board Composition and Direct Incentives on Firm Performance. Financial Management 20, 101-112 Hermalin, B.E., Weisbach, M.S., 1998. Endogenously Chosen Boards of Directors and Their Monitoring of the CEO. The American Economic Review 88, 96-118 Hong, H., Kubik, J.D., Stein, J.C., 2005. Thy ne ighbor's portfolio: Word-of-mouth effects in the holdings and trades of money managers. Journal of Finance 60, 2801-2824 Huberman, G., 2001. Familiarity breeds investment. Review of Financial Studies 14, 659-680 Inderst, R., Mueller, H.M., 2006. Informed lending and security design. Journal of
78 Finance 61, 2137-2162 Ivkovic, Z., Weisbenner, S., 2005. Local does as local is: Information content of the geography of individual investors' common stock investments. Journal of Finance 60, 267-306 Jensen, M.C., 1989. Eclipse of the Public Corporation. Harvard Business Review 67, 6174 Jensen, M.C., 1993. The Modern Industrial Revolution, Exit, and the Failure of Internal Control Systems. The Journal of Finance 48, 831-880 Jensen, M.C., Meckling, W.H., 1976. Theory of Firm Managerial Behavior, Agency Costs and Ownership Structure. Journal of Financial Economics 3, 305-360 Jones, J.J., 1991. Earnings Management during Import Relief Investigations. Journal of Accounting Research 29, 193-228 Kaplan, S.N., Reishus, D., 1990. Outside Dir ectorships and Corporate Performance. Journal of Financial Economics 27, 389-410 Klein, A., 1998. Firm Performance and Board Committee Structure. Journal of Law and Economics 41, 275-303 Klein, A., 2002. Audit committee, board of director characteristics, and earnings management. Journal of Accounting & Economics 33, 375-400 Lakonishok, J., Lee, I., 2001. Are insider trades informative? Review of Financial Studies 14, 79-11 Landier, A., Nair, V.B., Wulf, J.M., 2006. Geographic Dispersion and Corporate Decision-Making. Review of Financial Studies Forthcoming Larcker, D.F., Richardson, S.A., 2004. Fees paid to audit firms, accrual choices, and corporate governance. Journal of Accounting Research 42, 625-658 Lerner, J., 1995. Venture Capitalists and the Oversight of Private Firms. Journal of Finance 50, 301-318 Leuz, C., Nanda, D., Wysocki, P.D., 2003. Earni ngs management and investor protection: an international comparison. Journal of Financial Economics 69, 505-527 Loughran, T., Schultz, P., 2005. Liquidity: Urban versus rural firms. Journal of Financial Economics 78, 341-374 Malloy, C.J., 2005. The geography of equity analysis. Journal of Finance 60, 719-755 Mcconnell, J.J., Servaes, H., 1990. Additional Evidence on Equity Ownership and Corporate Value. Journal of Financial Economics 27, 595-612
79 Mehran, H., 1992. Executive Incentive Plans, Corporate Control, and Capital Structure. Journal of Financial and Quantitative Analysis 27, 539-560 Mian, A., 2006. Distance constraints: The limits of foreign lending in poor economies. Journal of Finance 61, 1465-1505 Morck, R., Shleifer, A., Vishny, R.W., 1988. Management ownership and market valuation : An empirical analysis. Journal of Financial Economics 20, 293-315 Nagaoka, S., 2005. Determinants of the Introduction of Stock Options by Japanese Firms: Analysis from the Incentive and Selection Perspectives. Journal of Business 78, 2289-2315 Ofek, E., Yermack, D., 2000. Taking Stock: Equity-Based Compensation and the Evolution of Managerial Ownership. Journal of Finance 55, 1367-1384 Pagano, M., Roell, A.A., Zechner, J., 2002. The geography of equity listing: Why do companies list abroad? Journal of Finance 57, 2651-2694 Peasnell, K.V., Pope, R.F., Young, S., 2005. Board monitoring and earnings management: Do outside directors influence abnormal accruals? Journal of Business Finance & Accounting 32, 1311-1346 Petersen, M.A., Rajan, R.G., 2002. Does distance still matter? The information revolution in small business lending. Journal of Finance 57, 2533-2570 Pirinsky, C., Wang, Q.H., 2006. Does corporate headquarters location matter for stock returns ? Journal of Finance 61, 1991-2015 Raheja, C.G., 2005. Determinants of Board Size and Composition: A Theory of Corporate Boards. Journal of Financial and Quantitative Analysis 40, 283-306 Ryan, J., Harley E., Wiggins, I., Roy A., 2004. Who is in whose pocket? Director compensation, board independence, and barriers to effective monitoring. Journal of Financial Economics 73, 497-524 Shivdasani, A., Yermack, D., 1999. CEO involve ment in the selection of new board members: An empirical analysis. Journal of Finance 54, 1829-1853 Teoh, S.H., Welch, I., Wong, T.J., 1998. Earnings Management and the Long-Run Market Performance of Initial Public Offerings. Journal of Finance 53, 1935-1974 Teoh, S.H., Welch, I., Wong, T.J., 1998. Earnings management and the underperformance of seasoned equity offerings. Journal of Financial Economics 50, 63-99 Vafeas, N., 1999. Board meeting frequency and firm performance. Journal of Financial Economics 53, 113-142
80 Vafeas, N., 2005. Audit committees, boards, and the quality of reported earnings. Contemporary Accounting Research 22, 1093-1122 Warfield, T.D., Wild, J.J., Wild, K.L., 1995. Managerial Ownership, Accounting Choices, and Informatives of Earnings. Journal of Accounting & Economics 20, 61-91 Westphal, J.D., Zajac, E.J., 1995. Who Shall Govern? CEO Board Power, Demographic Similarity, and New Director Selecti on. Administrative Science Quarterly 40 Westphal, J.D., Zajac, E.J., 1997. Defections from the Inner Circle: Social Exchange, Reciprocity, and the Diffusion of Boar d Independence in US Corporations. Administrative Science Quarterly 42 Xie, B., Davidson, W.N., DaDalt, P.J., 2003. Earnings management and corporate governance: the role of the board and the audit committee. Journal of Corporate Finance 9, 295-316 Yermack, D., 1995. Do corporations award CEO stock options effectively? Journal of Financial Economics 39, 237-269 Yermack, D., 1996. Higher market valuation of companies with a small board of directors. Journal of Financial Economics 40, 185-211 Yermack, D., 1997. Good Timing: CEO Stock Option Awards and Company News Announcements. Journal of Finance 52, 449-476 Yermack, D., 2004. Remuneration, Retention, and Reputation Incentives for Outside Directors. Journal of Finance 59, 2281-2308
About the Author Hong Wan is a Ph.D. candidate in finance at the Department of Finance, University of South Florida. His primary research inte rests focus on empirical corporate finance, mergers and acquisitions, and corporate governance. His papers have been presented at Western Finance Association annual conference, Financial Management Association annual conference, Southern Finance Association annual conference and China International Conference in Finance. While in the Ph.D. program at the Univers ity of South Florida, Mr. Wan has taught International Finance and Principles of Finance to undergraduate level class. Prior to coming to the Ph.D. program, Mr. Wan studied at Vi rginia Tech at Blacksburg, VA and received a M.B.A. in 2003.