Effects of corporate governance and board equity ownership on earnings quality.

Pergola, Teresa M. ; Joseph, Gilbert W. ; Jenzarli, Ali 等

INTRODUCTION

Earnings quality is critical to the efficient allocation of resources in capital markets. Reports of earnings that do not reflect the underlying economic performance of a firm inflicts losses on individual investors, employees, other companies, and the economy as a whole. The General Accounting Office, the investigative office for Congress, reports that earnings restatements rose approximately 145 percent from 1997 through June of 2002. During that period, approximately 10 percent of all listed companies announced an accounting correction or restatement costing investors approximately $100 billion in market capitalization (Wells 2002).

One of the most important functions of corporate governance is to ensure the quality of the financial reporting process. The Sarbanes-Oxley Act of 2002 and the major stock exchanges now require boards to have a majority of outside directors and audit committees to have three independent directors. The intent is to limit the ability of management to engage in opportunistic behavior by increasing the ability of both the board and the audit committee to monitor management. The intended outcome is to restore investor confidence in the quality of reported earnings.

However, evidence suggests that management can overcome monitoring mechanisms, even those that meet the new legislative and exchange standards. Chief Executive Magazine (Niskanen 2003) reports that all major firms charged with accounting scandals through 2002 were in full compliance with the standards for board and audit committee independence now required by the Sarbanes-Oxley Act. This implies that monitoring by independent directors is not always sufficient and/or effective in controlling management malfeasance.

There is clearly a need for further research to explore why corporate governance monitoring mechanisms are less effective in some cases. One possible reason is that board members become entrenched. Board members are in agency relationship with owners to monitor management behavior. Entrenchment occurs when a manager (or other party in an agency relationship), has enough firm equity, influence, or power to overcome governance constraints on their behavior (Fama and Jensen, 1983). When this occurs, insiders gain effective control and are able to engage in non-value-maximizing behavior. Effective control can occur through structural and ownership power (Dunn, 200), and from weak oversight by the board (Beasley, 1996; Dechow, Sloan, and Sweeney, 1996), attributed to lack of incentives for board members to monitor management.

In this study, we examine the incentive and power effects of the equity ownership of board groups (independent and insider board members) on the earnings quality of publicly traded companies for the year 2002. The year 2002 is selected because wide-scale accumulation of governance data is not available prior to 2002 and the effects of the Sarbanes-Oxley Act are not fully reflected until after that year. We examine the effect of governance and combined effects of governance and equity ownership to determine if entrenched board members allow management to overcome governance constraints resulting in lower earnings quality.

Earnings quality refers to persistence of earnings, i.e., its ability to forecast future earnings. Prior research has shown that accruals are better estimates of future earnings than cash flows (Sloan, 1996; Dechow et. al. 2002) and that the accuracy of the accruals is more explanatory than the size or level of accruals. An accurate accrual equals its related cash flow indicating that no errors in estimates occurred, i.e., earnings quality is perfect.

While many earnings quality studies focus on the size or discretionary nature of accruals to measure intentional manipulation of earnings, we define the quality of earnings by measuring accrual estimation errors, which we use as a proxy for earnings quality (Dechow and Dichev, 2002). We choose this approach because it does not differentiate between errors and irregularities in accruals and does not rest on the assumption that accrual errors only occur because of intentional manipulation. Accrual quality is related to firm and industry characteristics, which are observable, and can result from management incompetence, shirking, and/or honest estimation errors. Effective oversight by board members should influence management and result in a reduction of the size and incidence of estimation errors resulting in higher earnings quality.

We examine earnings quality in relation to the proportionate ownership of board insiders to total board equity, consistent with Dechow, Sloan, and Sweeney (1996), and more recently, Dunn (2004). While many entrenchment studies have measured insider ownership of total equity to assess insiders' incentives, we wished to measure both the power and incentive of the board groups. We measure power through the proportionate ownership of the board groups to total board equity; consistent with prior research studies, we measure incentive as the total ownership of each board group to total equity. These measures allow us to assess the power of each board group across different ownership levels, in addition to each group's incentive to protect shareholders' interests.

The major contributions of this paper can be summarized as follows.

1. We find strong evidence to support the entrenchment hypothesis as opposed to the uniformity proposed under the convergence of interest theory.

2. We use a relatively new measure of earnings quality to test for evidence of entrenchment (Dechow and Dichev, 2002). This approach allows us to provide evidence of entrenchment related to earnings quality, not just extreme forms of earnings management such as fraud.

3. We find that both inside management serving on the board and independent board members can become entrenched.

4. We find governance mechanisms can be effective to control management behavior.

5. We also find evidence that, for firms experiencing entrenchment, the strength of governance mechanisms is reduced but still effective enough to reduce the impact of entrenchment on earnings quality.

In summary, the incidence of earnings restatements in firms that were in compliance with recent governance reforms implies that there are factors present that allow management to overcome governance constraints and continue to report low quality earnings. Although we find that both independent and insider board members became entrenched, we also find that effective governance moderates the negative effect of entrenchment on earnings quality.

THEORETICAL BASES AND HYPOTHESIS DEVELOPMENT

Recent financial scandals have raised the issue of whether public companies are being run in the best interests of the shareholders. Management may have too much power and not enough supervision and accountability, particularly in companies with widely dispersed ownership. Agency conflicts, conflicts that arise from the separation of ownership and control, may not be effectively resolved through corporate governance systems. Corporate governance is a set of mechanisms put in place by stakeholders to exercise control over management. The way in which these mechanisms are set up and fulfill their role defines the governance structure of the firm (Farinha, 2003).

Agency theory discusses the types of monitoring and bonding costs that can be employed to reduce agency conflicts. According to Fama and Jensen (1983), the most critical monitoring mechanism is that of the board of directors. The board is charged with monitoring the decisions and actions of management to ensure that management acts in the best interest of shareholders. It is important for the board to have both inside and outside directors as each brings different qualifications and motivations to the table. However, there is clear evidence that board independence results in more effective monitoring of management, particularly with respect to the board's ability to monitor the earnings process.

The assertion is that a board's relative independence from management is negatively related to earnings manipulation. Klein (2002a) found a negative association between a firm's discretionary accruals (proxy for earnings management) and the percent of outside directors on the board. Klein also found that as boards became more independent by changing the board composition, the level of discretionary accruals declined. Xie, Davidson III, and DaDalt (2003) found evidence that more independent boards were associated with lower levels of discretionary current accruals.

In fraud studies, researchers found that firms with higher levels of independent directors were less likely to commit fraud (Beasley, Carcello, Hermanson, and Lapides, 2000; Beasley, 1996) In addition to the number of independent directors, Beasley (1996) also found that as total levels of stock ownership by outside or independent board members increased, the likelihood of fraud decreased. This suggests that board structure and stock ownership of the board have an impact on the board's monitoring effectiveness.

Increased ownership by independent board members may provide more incentive for monitoring but may also imply that ownership affects the power of the board to monitor management. An alternative interpretation of Beasley's findings may have to do with the relative power of the two board groups, insiders and independents, over each other, though this was not measured in Beasley's study. Other studies, discussed in the following paragraphs, have measured the proportionate ownership of board groups as a proxy for power.

Dechow, Sloan, and Sweeney (1996) used a proportionate measure of managerial ownership (defined as number of shares of stock owned by officers on the board) as a proportion of total board member ownership in their analysis of firms subject to SEC enforcement action. Like Beasley (1996), they found that earnings management was less likely to occur if outside or independent directors dominated the board. However, they found that low oversight by the board was the predominant variable impacting the likelihood of earnings management. They also found that for firms subject to SEC enforcement action, insiders held a significantly greater proportion of total board equity relative to outsider board members. This implies that even though insiders were not dominant in terms of numbers of board members, they may have been able to dominate independent board members through increased ownership. An alternative interpretation could be that independent board members did not own enough stock to align their interests with owners. This interpretation is consistent with Beasley's finding that increased ownership by independent board members reduced the likelihood of fraud.

These findings suggest that weak monitoring is a catalyst for earnings management and that there is a greater likelihood of earnings management if management can dominate the board, through greater numbers of directors or higher stock ownership percentages, relative to independent board members. Recent legislative reform has addressed the proportion of board members that can be insiders. The legislation now requires a majority of independent directors on public boards. What is unclear is the ability of management to overcome the effectiveness of board oversight when the number of outside directors is in the majority.

Dunn (2004) attempted to address the power issue by comparing two groups of firms convicted of financial statement fraud: firms with large concentrations of ownership power and firms lacking ownership power. He found that financial statement fraud was more likely to occur in firms where insiders had high ownership power. Like previous researchers, the number of insider board members was positively correlated with the incidence of fraud. However, the relative stock ownership of insiders to outsiders was more significant and had a large impact on the likelihood of fraud. Sensitivity tests suggested that alternative forms of corporate governance could be overcome when insiders had a concentration of relative ownership power. This is why we chose the relative measure of equity ownership used in this study.

Recognizing that there is an agency relationship between owners, independent board members, and insiders, stock ownership can have a positive incentive effect to align the interests of both groups with shareholders (Jensen and Meckling, 1976). However, if the level of board equity ownership allows board groups to concentrate their power and incentives are low, entrenchment may set in.

Entrenchment occurs when a manager (or other party in an agency relationship), has enough firm equity, influence, or power to overcome governance constraints on their behavior (Fama and Jensen, 1983). Entrenchment studies have shown that this can occur at very low levels of absolute ownership (Morck, Shlefier, and Vishny , 1988; Short and Keasy, 1999; Griffith, Fogelberg, and Weeks, 2002; Peasnell, Pope, and Young, 2003; Warfield, Wild, and Wild, 1995; Yeo, Tan, and Chen, 2002). These studies defined entrenchment levels as the percent of total equity owned and tested for management entrenchment using a variety of measures, such as firm value, performance, demand for outside directors, discretionary accruals, and the information content of earnings.

We test for entrenchment of board groups to measure the power and incentive effects of independent and insider board members. We use the proportionate ownership of the independent and insider board members to total board equity as a proxy for power and incentives consistent with Dechow et. al., 1996 and Dunn, 2004.

We measure entrenchment by assessing the persistence of earnings as measured by the quality of accruals. The quality of accruals refers to working capital accruals and how well those accruals match their related cash flows. Prior studies have shown low earnings quality, proxied by levels of discretionary accruals, when entrenchment has occurred. We choose to measure earnings quality as opposed to earnings management for several reasons.

First, discretionary accrual models used to measure earnings manipulation can contain specification and measurement errors that can bias results and result in erroneous research findings (McNichols, 2002). Second, the power of the models to find earnings manipulation is low. Dechow, Sloan, and Sweeney (1995) find that earnings management of less than 5% of total assets is likely to go undetected. In addition, Dechow and Dichev (2002) point out that traditional measures of earnings manipulation suggest that managerial intent affects the size and incidence of accrual estimation errors. They argue that accrual quality will be related to firm and industry characteristics, absent earnings manipulation. For example, the volatility of operations will be related to the likelihood of estimation errors, which will be recurring and observable, as opposed to earnings management, which is often unobservable. Finally, while poor earnings quality can result from earnings management, it can also result from management shirking, incompetence, or just honest estimation errors. The important factor is the quality of earnings, not the exact nature of the cause.

The research on the composition and effectiveness of boards raises issues with respect to the power of insider and independent board members. Agency conflicts are shown for both groups and the relative equity ownership of both groups is relevant, first to their incentive to act in owners' interests and second, to their ability to exert control.

We hypothesize that when board insider ownership is low, insiders lack sufficient power to act in their own self-interest. Independent board members have the power to keep insiders from reporting low quality earnings. The absence of a negative correlation in this range indicates that independent board members are performing their monitoring function.

H1: In public companies, when relative board insider ownership is low, earnings quality is not correlated with or is positively correlated with insider ownership.

At very high relative levels of board insider ownership, insiders have more power than independent board members and as absolute ownership increases, more incentive to act in owners' interests. The absence of a negative correlation between earnings quality and insider ownership is indicative of convergence-of-interests for insider board members.

H2: In public companies, when relative board insider ownership is high, earnings quality is not correlated with or is positively correlated with insider ownership.

Entrenchment sets in when board insider ownership is high enough for insiders to dominate but total equity ownership is not high enough to provide incentive to protect owners' interests.

H3: In public companies, earnings quality is negatively correlated with relative board insider ownership in the entrenchment range.

While stock ownership is meant to align the interests of management and owners, it is not the only mechanism used by owners to protect their interests. Boards of directors and audit committees are two of the primary monitoring mechanisms employed by owners to monitor management behavior. Effective monitoring should improve earnings quality as board members keep insiders from engaging in behavior inconsistent with owners' interests. Consistent with the requirements of the Sarbanes-Oxley Act of 2002, research has shown that the independence of the full board and audit committee may contribute to the effectiveness of monitoring (Weisbach 1988; Dahya et al. 2002; Klein 2002a; Xie et al. 2003; Krishnan 2005; Archambeault and DeZoort 2001).

However, there are additional factors that impact the effectiveness of the board such as board size (Anderson et al. 2004; Klein 2002b; Brown and Caylor 2004), director attendance (Allen et al. 2004), number of board appointments (Young et al. 2003; Fich and Shivdasani 2004) and power (Dunn 2004). Having the firm's Chief Executive officer also serve as the Chairperson of the Board (called CEO/COB duality) can also compromise the independence of the board and the audit committee causing both mechanisms to be less effective (Farber 2005; Brown and Caylor 2004).

Recognizing that single mechanisms can be compromised, firms employ many controls to define their governance structure. Some mechanisms act as substitutes for others and some are complementary (see Farinha 2003 for a survey of corporate governance literature), implying that the exact combination of mechanisms employed is firm-specific. Recognizing the unique nature of each firm's governance structure, more recent research (Gompers, Ishii, & Metrick, 2003: Brown & Caylor, 2004) incorporates governance composite indexes that represent the combined strength of all mechanisms employed. Early results show that firms with higher governance indexes exhibit higher market returns, higher firm value (Tobin's Q), and better operating performance than firms with lower indexes (Gompers et al. 2003). We hypothesize that effective governance will also have a positive effect on earnings quality because stronger oversight of management reduces management's likelihood to shirk or commit fraud.

H4: In public companies, the effectiveness of governance is positively correlated with earnings quality.

However, Dunn (2004) showed that when insiders gain effective control, they might be able to overcome governance mechanisms put in place to control their behavior.

H5: In public companies, earnings quality is negatively correlated with relative board insider ownership for firms that are entrenched, irrespective of all other governance mechanisms.

RESEARCH DESIGN AND VARIABLE DEFINITIONS

We examine earnings quality across relative board insider ownership (BIO) levels. To test for entrenchment, we divide BIO into intervals of five percent and run piecewise linear regressions at different BIO intervals to estimate the coefficient on BIO. Changes in the slope, or in its sign, at two different thresholds indicate the range of entrenchment for BIO. We use the following empirical model to test Hypotheses H1, H2, and H3 in each of three ranges: below entrenchment, at entrenchment, and above entrenchment.

[|SEE|.sub.[Florin],t] = [b.sub.0] + [b.sub.1][BIO.sub.[Florin],t] + [e.sub.[Florin],t] (1)

where:

[|SEE|.sub.[Florin],t] = absolute value of standardized estimation errors found in accruals for firm [Florin] at time t using the Dechow and Dichev Model(2002)(discussed below);

[BIO.sub.[Florin],t] = proportion of insider (management and affiliated directors) equity to total board equity for firm [Florin] at time t;

[e.sub.[Florin],t] = the error term for firm [Florin] at time t.

Governance mechanisms are put in place to monitor and control management behavior. If effective, governance and earnings quality should be positively correlated. Model (2) is used to test H4:

[|SEE|.sub.[Florin],t] = [b.sub.0] + [b.sub.1][GI.sub.[Florin],t] + [e.sub.[Florin],t] (2)

where all variables are as defined above and:

[GI.sub.[Florin],t] = a discrete governance index variable indicating the effectiveness of a composite of governance mechanisms for firm [Florin] at time t (discussed below).

We regress BIO and GI on |SEE| for firms in the entrenchment range to assess the combined effect of both BIO and GI on earnings quality. We use following multivariate regression model to test H5:

[|SEE|.sub.[Florin],t] = [b.sub.0] + [b.sub.1][GI.sub.[Florin],t] + [b.sub.2][GI.sub.[Florin],t] + [e.sub.[Florin],t] (3)

where all variables are as defined above.

We use the Dechow and Dichev model (2002) of accrual estimation errors as a proxy for earnings quality. The model uses the following equation to measure earnings quality:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

where:

[Delta][WC.sub.t] = change in accounts receivable (Compustat #302), plus the change in inventory (Compustat #303), minus the change in accounts payable (Compustat #304), minus the change in taxes payable (Compustat #305), plus the change other assets (Compustat #307) at time t;

[CFO.sub.t-1] = cash flows from operations for the prior period;

[CFO.sub.t] = cash flows from operations for the current period;

[CFO.sub.t+1] = cash flows from operations for the next period;

[TA.sub.t] = average total assets for a firm in the current period.

The model captures the extent to which accruals map into cash flow realizations, measuring any estimation errors using the error term ([e.sub.t]). We scale all variables by average total assets ([TA.sub.t]) (Compustat #6) to account for differences in firm size. The intercept ([b.sub.0]) is included to measure positive working capital accruals related to firm growth.

We estimate the model cross-sectionally, using a 3-year period to derive [CFO.sub.t-1], [CFO.sub.t], and [CFO.sub.t+1]. We used the model for t = 2002 because wide-scale accumulation of governance data is not available prior to 2002. In addition, 2002 was the year that Sarbanes-Oxley was enacted, but not fully effective. We wanted to examine earnings quality prior to governance reform.

Following Dechow and Dichev (2002), we ran the regression in (4) for all sample firms in a 2-digit SIC code (We also classify firms using the Fama-French industry classification for comparison purposes. See Table 1, panel B).

We required a minimum of 30 observations for each 2-digit industry group to obtain sufficient power for a .05 significance level (Hair et. al. 1998, 165). We expected the size of earnings, cash flows, and accruals to vary across industry, be larger for firms with large absolute accruals, and estimation errors to be higher for firms in difficult forecasting environments (McNichols 2002). To address these concerns, we standardized the measure of earnings quality derived from equation (4) for each firm (ef,t) to remove any industry effects as follows:

[SEE.sub.[Florin],t] = ([e.sub.[Florin],t] - [x.sub.i]) / [s.sub.i] (5)

where:

[SEE.sub.[Florin],t] = the standardized estimation error for firm [Florin] at time t;

[e.sub.[Florin],t] = the estimation error as measured by equation (4) for firm [Florin] at time t;

[x.sub.i] = the mean of the estimation error using equation (4) for a 2-digit SIC code industry portfolio of firms;

[s.sub.i] = the standard deviation of estimation errors for the same 2-digit SIC code industry portfolio of firms.

Both overestimation and underestimation errors impact earnings quality adversely. As such, and consistent with other earnings quality studies (Klein 2002a; Warfield et al. 1995), we use the unsigned (absolute value of) standardized estimation errors (|SEE|) as the dependent variable.

Board insider ownership (BIO) is the proportion of insider common stock ownership (managers and affiliated board members) to total board ownership. We defined affiliated board members as those board members that are not independent of management, as defined by the Securities and Exchange Commission standards (SEC 2003). Sources of data for this measure were provided by The Investor Responsibility Research Center (IRRC 2005), which obtains the data from firm proxy statements.

Effective governance is a calculated governance index using governance data provided by Institutional Shareholder Services (ISS 2003). ISS compiles 61 different variables encompassing eight corporate governance categories: board of directors, audit, charter/bylaws, director education, executive and director compensation, ownership, progressive practices, and state of incorporation.

In this study, we used a calculated governance index (GI) of 51 of the total 61 variables (See Appendix 1 for a detailed listing of the rating criteria based on the minimum governance standard defined by ISS). We awarded one point for each acceptable variable and summed them for a firm-specific index with a maximum value of 51. We used this index as the measure of governance effectiveness, where a high score was considered more effective than a low score. We do not control for individual governance characteristics, such as independent boards or audit committees for example, as they are included in the computed index. In addition, some governance factors act as substitutes for other factors such that the absence of a factor for a firm may not

indicate a weakness if there is a substitute mechanism in place.

SAMPLE AND VARIABLE MEASUREMENTS

We performed our study on a sample of publicly traded firms taken from the S&P 500 (large cap), S&P 400 (mid cap), and S&P 600 (small cap) indices, for the year 2002. Each index contains 500 firms totaling an initial sample of 1,500 firms; Panel A in Table 1 summarizes the sample selection. We restricted the sample to firms with complete data for calculation of working capital accruals and governance indices. We eliminated firms in regulated industries (SIC 4900--utilities, 6000-6999--financial institutions), non-domestic firms, outliers, and firms that had a fiscal year change. In addition, we required at least 30 firms within each SIC grouping. The final sample was comprised of 499 firms representing seven two-digit SIC codes, as shown in Panel B of Table 1.

The measure of earnings quality we used is the absolute value of the standardized residual of industry regressions using regression model (4) repeated here:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

We defined change in working capital accruals ([Delta][WC.sub.t]) as the change in accounts receivable, plus the change in inventory, minus the change in accounts payable, minus the change in taxes payable, plus the change in other assets. We used net cash provided by operating activities from the statement of cash flows as cash flows from operations (CFO). The intercept ([b.sub.0]) was included to measure positive working capital accruals related to firm growth. The year [sub.t] is 2002, [sub.t-1] is 2001, and [sub.t+1] is 2003. We scaled all variables by average total assets, consistent with the Dechow and Dichev Model (2002). All values were obtained from Compustat as defined in the variable definition section of the paper. Table 2 presents the descriptive statistics for the variables used in the model as well as statistics for other variables related to cash flows and accruals to validate the relationships predicted in the model.

An examination of Panel A in Table 2 reveals that the descriptive statistics are in line with Dechow and Dichev (2002). CFO is consistent across lagged ([CFO.sub.t-1]), current ([CFO.sub.t]), and future periods ([CFO.sub.t+1]) and exceeds earnings in time t (Earnt), implying that short-term accruals are negative. Average total accruals (Acct) are negative (-.0887), primarily because of depreciation. The size of the firms, in terms of average total assets, varies significantly (standard deviation of 21,809) and the mean (5,419) is well in excess of the upper quartile indicating a skewed distribution. The average operating cycle of the sample firms is 151 days, consistent with the Dechow and Dichev (2002) model assumption that changes in working capital accruals relate to cash flows within a one-year time frame. Frequency distributions (not shown) reveal that 97.6 percent of the sample firms have an operating cycle of less than one year.

Pearson correlations in Panels B and C of Table 2 show the sign and strength of the relationship between variables that underlie the model design.

Prior research provides the predictions for the relationships between the variables (Barth et al. 2001; Dechow and Dichev 2002). Our results confirmed these relationships. Note that the correlations between the change in working capital (?[CFO.sub.t]) and past and future cash flows ([CFO.sub.t-1] and [CFO.sub.t+1]) are -.035 and .019 and are insignificant. These were expected to be significant and positive. Panel C shows that after controlling for the effect of current cash flows ([CFO.sub.t]), both correlations are positive and significant. We also find positive correlations between earnings (Earnt) and past cash flows ([CFO.sub.t-1]) (.752), and accruals (Acct) and past cash flows ([CFO.sub.t-1]) (.086), significant at the .05 level.

Table 3 presents the results of regressions of working capital accruals on past, present, and future cash flows by two-digit SIC groupings. Results from the industry-specific regressions show that the model is well specified for these SIC groups (F statistics are all significant at the .05 level). The variability of the change in working capital accruals (?[CFO.sub.t]) explained by the model ranges from a low of 32.4 percent (SIC code 38) to a high of 67.8 percent (SIC code 13). These results are consistent with Dechow and Dichev (2002) who reported a mean adjusted [R.sup.2] on an industry basis of 34 percent.

The correlations, signs, and sizes of the coefficients between the dependent and independent variables are also consistent with Dechow and Dichev (2002). Changes in working capital accruals (?[CFO.sub.t]) are negatively correlated to current cash flows ([CFO.sub.t]), positively correlated to future cash flows ([CFO.sub.t+1]), and positively correlated to past cash flows ([CFO.sub.t-1]) except in SIC codes 35 and 38, where the coefficients on past cash flows are negative but not statistically significant. The absolute value of the coefficients on the cash flow variables is less than 1 and smaller for the coefficients on past and future cash flows ([CFO.sub.t-1] and [CFO.sub.t+1]) than for current cash flows ([CFO.sub.t]), as expected.

Dechow and Dichev (2002) and McNichols (2002) both show that the cash flow coefficients can contain measurement error related to sales growth. To test this premise, we reran the model with a sales growth term (results not shown). The results of the regressions indicate that change in sales did not have a significant impact on the model. Examination of the descriptive statistics for ?Sales (not shown) supported this result. The mean was -100.081 and the 5 percent trimmed mean was -2.49. The frequencies (not shown) showed that 45.7 percent of the sample firms have decreases in sales, and the distribution is normal after adjusting for two outliers with large increases in sales (?Sales = 8,861 and 11,362). The results showed that changes in sales do not bias the cash flow variables.

The firm-specific residuals from the SIC regressions are the estimation errors present in the accruals. However, they can contain measurement error if there are differences in the volatility of the industries with respect to forecasting ability. We normalized these differences by standardizing the measure for testing purposes as described previously. We used the unsigned (absolute) value of standardized estimation errors (|SEE|) as the dependent variable.

Board insider ownership (BIO) is the percentage of shares owned by insiders to total board shares. All ownership data was for the year 2002. To get the proportion of total board stock owned by insiders, we combined employee and affiliated shares and divide by the total board shares. The proportion of independent board ownership is, by definition, 1--(the insider proportion).

Governance index (GI) was a calculated firm-specific index representing the sum of points awarded for 51 governance variables. Descriptive statistics for the |SEE|, BIO, and GI appear in Table 4. (See Appendix 1 for the percentage of firms that meet the minimum governance standard for each of the 51 variables.)

The variable, |SEE|, ranged from 0 to just over 3, with a mean of .7691, a median of .6156, and a standard deviation of .6287. The mean of BIO is 79.427. The median, however, was 88.205 indicating that the distribution was skewed. A total of 313 firms (75 percent of the sample) had BIO higher than the mean and 28 percent were in the 95 to 100 percent range. The distribution showed that although the board may be "independent" in terms of the number of directors, insiders owned significantly more stock than independent outsiders. Governance (GI) ranged from 13 to 41 with a mean and median of 25.76 and 25.00 respectively and a standard deviation of 5.318.

PRESENTATION AND INTERPRETATION OF RESULTS

Hypotheses H1--H3 test the relationship between relative board insider ownership (BIO) and earnings quality (|SEE|). To examine the relationship between BIO and |SEE| at different levels of ownership, we ran regression curve estimates for each BIO increment (increments of 5% ownership). Examination of the best fitting line, significance of the fit for each increment, and the slope of the line indicated break points of 30 percent and 49 percent (points where the slope significantly changes). Table 5 summarizes the results of the curve estimations and shows the descriptive statistics for the IO increments.

Panel A of Table 5 shows that most firms (442) had BIO of 60 percent or higher, 24 firms from 30 to 49 percent, and 33 firms with less than 30 percent. The mean percent of absolute equity ownership owned by insiders rose as BIO rose but did not exceed a mean 7.76 percent (See Panel B; Table 5). The mean level of total equity ownership for independent board members fell as their board ownership proportion fell indicating that increases (decreases) in the proportion of board ownership for both groups followed increases (decreases) in total equity ownership. Each inside ownership group contained firms from each industry and from all three indices, as shown in Panel C and Panel D of Table 5, respectively. As such, neither industry nor S&P index should bias any reported results.

Hypothesis Testing

H1--H3

Panel A in Table 5 shows the relationships between BIO and |SEE| for the sample firms. The lack of a significant relationship between BIO and |SEE| when insiders had a low relative ownership proportion of board stock (0--29 percent; below entrenchment) was consistent with agency theory. In this range, insiders' absolute equity (.86 percent) and relative equity ownership (0 to 29 percent) were low indicating that their interests were not aligned with owners but that they did not have the power to act in their own self-interest. Independent board members relative ownership in this range was from 71 to 100 percent and their absolute equity ownership was higher (7.48 percent) than in any other range and also exceeded insider ownership (.86 percent). The relative levels of ownership indicated that independent board members could control insiders and that they had more incentive to monitor in this range than in any other ownership range. These test results provided support for hypothesis H1, that earnings quality is not correlated with levels of board insider ownership when insider ownership is low. These results show evidence of convergence-of-interests for independent board members.

When insider relative board equity levels were high (50 to 100 percent; above entrenchment), total absolute insider equity ownership was also higher than in any other range (7.76 percent). Insiders had equity control of the board and more incentive to protect shareholder interests. We did not find a significant effect for BIO on earnings quality, as expected. The results at this level of insider ownership provided support for hypothesis H2, that earnings quality is not correlated with insider ownership when relative insider ownership is high.

We did find a significant relationship between BIO and |SEE| in the ownership range of 30 to 49 percent. The relationship was linear, with a positive slope, significant at the .05 level. A positive slope indicated that estimation errors rose (earnings quality fell) as board insider ownership rose within the range. This result was inconsistent with convergence-of-interests for both insider ownership and independent board ownership and provides support for hypothesis H3 as follows.

Insiders did not have a majority relative ownership position in this range. Both insiders and independent board members had very low absolute ownership (1.38 percent and 2.21 percent, respectively; see Table 5, panel B), indicating that neither group had the incentive to act in shareholders' interests. Insiders' entrenchment is evidenced by the deterioration of earnings quality; independents' entrenchment is evidenced by the lack of control of insiders' actions even though they held the dominant board ownership position.

The negative effect of insider ownership on earnings quality indicated that neither group was acting in the owners' interests i.e., both groups were entrenched. These findings are consistent with Morck et al. (1988), who found that independent and grey board members with stock holdings responded to financial incentives in a similar fashion to management and could also become entrenched. The results provide support for hypothesis H3 that earnings quality is negatively correlated with relative board insider ownership in the entrenchment range.

H4

Hypothesis 4 posits that the effectiveness of governance is positively correlated with earnings quality. To test this premise, we modeled the relationship between GI and |SEE| using SPSS curve estimation. The curve estimation procedure produces curve estimation statistics and related plots for eleven different curve estimation regression models, such as linear, logarithmic, quadratic, inverse, etc..., to model |SEE| as a function of GI. The best fitting model, as determined by the highest [R.sup.2], showed a logarithmic relationship between the variables, significant at the .0510 level (Table 6, Panel A). The log model regresses the log of |SEE| on GI. The coefficient for GI was negative, as expected, and significant (p=.0510). These results provided support for H4, that governance is positively correlated with earnings quality. However, the explanatory power of the model was low ([R.sup.2] = .00764, the percent of variation in the log of |SEE| that can be explained by GI) indicating that governance does not explain very much of the variation in |SEE| when considering all firms.

Although entrenchment was found in the board insider ownership range of 30 to 49 percent, a closer examination of the descriptive statistics in Table 5, Panel A shows that the mean value of |SEE| is not significantly different across the ownership ranges. Although there was a significant slope change for BIO on |SEE| within the entrenchment range, parametric ANOVA tests and nonparametric Kruskal-Wallis tests (not shown) failed to show any differences in |SEE| across the three different ownership ranges. These results could occur if governance was effective and able to overcome the negative effect of entrenched board members on earnings quality. To examine whether GI differed across ownership groups, we conducted an ANOVA test of means of GI for below entrenchment, entrenchment, and above entrenchment groups. Results, show the mean GI for firms below the entrenchment range (27.7879) was significantly higher than the GI for firms above the entrenchment range (25.5181). There were no significant differences between entrenched firms and non-entrenched firms although the mean for entrenched firms (27.4167) was closer in size to the below entrenchment group (Descriptive statistics in Table 6, Panel B; ANOVA test results in Panel C and D). These results indicate that as BIO rises, insiders are able to resist imposition of additional governance mechanisms.

To examine the impact of governance on entrenchment, we divided the firms in the entrenchment range into high and low governance groups based on the median governance index. ANOVA tests of differences in means for governance between the above median and below median groups, shown in Table 7, were significant (F=35.267, p=.000). Tests for differences in BIO between the above and below median governance groups also showed significant differences (F=14.438, p=.000). Examination of the group differences showed that entrenched firms with lower governance had significantly higher relative board insider ownership.

This implied that as relative board insider ownership rose to higher levels within the entrenchment range, insiders were able to resist imposition of additional governance constraints. This result, however, did not address whether GI was effective at reducing |SEE| (improving earnings quality) in the entrenchment range.

To examine this possibility, we ran a univariate regression of GI on |SEE| for entrenched firms. The coefficient on GI was -.051 indicating that as governance rose, estimation errors fell (See Table 8; Panel A). The relationship was statistically significant (F = 5.083, p = .034, [R.sup.2] = .188) indicating that governance did significantly impact earnings quality for entrenched firms. This evidence is consistent with the results from the full sample and provides support for hypothesis H4 that effective governance is positively correlated with earnings quality.

H5

Hypothesis 5 tests the relationship of earnings quality in entrenched firms to board insider ownership and predicts that they are negatively correlated, irrespective of all other governance mechanisms. In univariate tests of entrenched firms, BIO had a significant negative effect on earnings quality and GI had a significant positive effect on earnings quality. Results of a multivariate regression of BIO and GI on |SEE|, shown in Table 8, Panel B, showed a more significant model (F = 4.654, p = .021, adjusted [R.sup.2] = .241) than the univariate results.

The signs on both coefficients were consistent with the univariate results but the size and significance of both coefficients was smaller indicating that the presence of each variable had a moderating effect on the other (BIO = .035, t=1.903, p=.071; GI = -.042, t=-1.936, p=.066). When considered together, more of the variation in earnings quality is explained and while both factors have less of an effect individually, governance has a stronger effect on earnings quality than BIO. These results may explain why there were no significant differences in earnings quality among ownership groups, even though earnings quality declined in the entrenchment range. This finding did not provide support for hypothesis H5, that earnings quality is negatively correlated with relative board insider ownership in the entrenchment range, irrespective of all other governance mechanisms. The effectiveness of governance was able to moderate the negative effect of entrenchment.

SUMMARY AND CONCLUSIONS

In this study, we examined the incentive and power effects of the equity ownership of board groups on earnings quality. Agency conflicts are shown for both groups and the relative equity ownership of both groups is relevant, first to their incentive to act in owners' interests and second, to their ability to exert control. We found that all board members, both independents and insiders, can become entrenched if they have low incentives to act in owners' interests, as reflected by the varying levels of power and incentives. When relative ownership for independent board members was high (power) but absolute ownership was low (incentive), we found a negative correlation with earnings quality. In the entrenchment range, independent board members had the power to act in owners' interests, but neither independent nor insider board members had the incentive to do so. The negative effect on earnings quality indicated that both groups were entrenched. Conversely, when relative board ownership was high (for either group) in conjunction with absolute equity ownership, we found no effect for ownership on earnings quality. Our findings supported an agency relationship between independent board members and owners, consistent with the research results of Beasley (1996) and Dechow et al., (1996). This is significant in two ways: 1) we provided evidence of entrenchment related to earnings quality, not just for extreme forms of earnings management such as fraud, and 2) we showed that earnings quality was directly affected by entrenched board members, both independent and insider board members. Our findings implied that relying on the independence of the board, or board committees, to ensure high quality earnings and/or high levels of internal control may be inadequate if board members are entrenched.

We also examined the effect of governance on earnings quality to determine if entrenched board members could overcome governance constraints. Test results showed that as relative insider board ownership rose, governance was lower indicating that insiders were able to avoid the imposition of additional governance mechanisms. Governance, however, still had a positive effect on earnings quality, even on entrenched firms. When insider ownership and governance were considered together for entrenched firms, we found that negative entrenchment effects could be moderated by effective governance structures. These results were inconsistent with Dunn (2004) who found that entrenched insiders could overcome governance constraints in fraud firms. However, Dunn limited the governance mechanisms considered in his study to board and audit committee characteristics while our study incorporated a broader measure of the governance structure as a whole. These results implied that stakeholders should consider the relative level of board ownership and the total governance structure in tandem when assessing risk.

There are several caveats to consider when interpreting these results that suggest areas for future research. First, because our research design was cross sectional, we were forced to use industry measures of earnings quality (as opposed to firm-specific measures), which are less precise and may contain more measurement error. We also eliminated many industries from the sample because there were not enough observations to apply the Dechow and Dichev Model (2002). As a result, the ability to generalize these results to other years and other industries is constrained.

Second, we did not consider the composition of the board, except as it related to independence. It is possible that certain board members represent blockholders and/or institutional investors, groups that can have interests that differ from both management and owners, which may have affected the results. We did not consider the selection of board members and management's role in the selection process, which may also impact the results.

Finally, we did not identify which specific combinations of governance variables were effective in controlling earnings quality. There were no significant differences in independent boards or audit committees between groups indicating that other governance variables, in combination with these, are important in designing an effective governance structure. Future research, aimed at identifying the best combination of governance factors, would significantly add to the literature.

APPENDIX

Appendix 1: Strength of Governance Structure Rating Criteria and Percent of Sample Firms That Meet Minimum Governance Standards

ISS compiles 61 different factors encompassing the eight corporate governance categories shown below. We omitted four provisions of poison pills and six provisions related to the state of incorporation. Poison pills require shareholder approval. The specific provisions were not considered. We assumed that shareholders in the approval process considered these items and that shareholder approval implied that they deemed them in their best interests. As to the state of incorporation, the only consideration is whether the firm is incorporated in a state with or without stakeholder laws. The specific laws are not necessary for this evaluation and only served to subdivide the sample needlessly. We eliminated the provision that measures officer and director ownership as "at least 1 percent but not over 30 percent of total shares outstanding" because we used a specific measure of ownership for each director in the testing.

Consistent with Brown and Caylor (2004), we separated one provision into two components: poison pill and blank check preferred stock. All these adjustments resulted in consideration of 51 separate provisions of governance.

We determined whether or not a firm's governance was acceptable (coded 1) or unacceptable (coded 0) by comparing the data for each firm with governance best practices as provided in the ISS Corporate Governance: Best Practices User Guide and Glossary (2003) and with empirical research results (similar to Brown and Caylor, 2004). The factors for minimally acceptable governance standards, by category, and the percentage of firms meeting those standards follow. The chart includes the percentages for all firms, firms below entrenchment ("Below E"), firms that are entrenched ("E"), and firms above entrenchment ("Above E") that meet the minimum governance standards.

Minimum Governance Standard All Firms Below E

Category 1: Board of Directors

Board is controlled by more than 50 % 89.6% 93.9%
 independent directors
Compensation committee is composed solely 82.6% 78.8%
 of independent directors
Nominating committee is composed solely 61.7% 57.6%
 of independent directors
Governance committee exists and meets at 70.1% 66.7%
 least once during the year
Size of the board is at least six but not 94.6% 90.9%
 greater than fifteen members
Shareholder approval is required to change 12.8% 12.1%
 board size
Shareholders have cumulative voting rights 10.0% 12.1%
 to elect directors
Board members are elected annually 39.5% 33.3%
CEO serves on no more than two other boards 92.6% 90.9%
 of public companies
Outside directors serve on no more than five 4.0% 6.1%
 addn'l boards of public companies
No former CEO's sit on the board 76.2% 81.8%
The CEO and chairman duties are separated and a 56.5% 54.5%
 lead director is specified
Board guidelines are published in the firm 36.3% 39.4%
 proxy statement
Managers respond to all shareholder proposals 98.2% 97.0%
 within twelve months of the last
 shareholder meeting

Directors attend at least 75 % of board meetings 97.8% 93.9%
 or have a valid excuse for non- attendance
Shareholders vote on directors selected to 39.9% 30.3%
 fill vacancies
CEO is not listed as having a related party 84.2% 97.0%
 transaction in the proxy statement

Category 2: Audit

Audit committee consists solely of independent 83.8% 93.9%
 outside directors
Consulting fees paid to auditors are less than 77.2% 87.9%
 audit fees paid to auditors
Company has a formal policy of audit rotation 18.6% 18.2%
Auditors were ratified at the most recent 60.9% 63.6%
 shareholder meeting

Category 3: Charter/Bylaws

Company either has no poison pill or a pill 33.7% 36.4%
 that has shareholder approval
A simple majority is required to approve a 64.1% 57.6%
 merger (not a supermajority)
A majority vote is required to amend 4.4% 36.4%
 charter/bylaws (not a supermajority)
Shareholders are allowed to call special meetings 37.7% 36.4%
Shareholders may act by written consent and the 24.8% 24.2%
 consent is non-unanimous
Company is not authorized to issue blank check 10.2% 15.2%
 preferred stock (stock over which the
 board has broad authority to determine voting,
 dividend, conversion and other rights)
Board cannot amend bylaws without shareholder 40.7% 6.1%
 approval or can only do so under limited
 circumstances
Company has a single class of common stock 92.8% 100.0%

Category 4: Director Education

One or more directors have participated in an 13.6% 15.2%
 ISS-accredited director education program

Category 5: Executive and Director Compensation

No interlocks exist among directors on the 98.8% 97.0%
 compensation committee
Non-employees do not participate in the 97.00 100.0%
 company pension plans
Option re-pricing did not occur within the 93.8% 97.0%
 last three years
Stock incentive plans were adopted with 80.4% 84.8%
 shareholder approval
Directors receive all or a portion of their 94.4% 84.8%
 fees in stock
Company does not provide any loans to 90.2% 90.9%
 executives for exercising options
The last time shareholders voted on a pay plan, 58.3% 69.7%
 ISS did not deem its costs to be excessive

The average options granted in the past three 11.0% 15.2%
 years as a %age of basic shares outstanding
 did not exceed 3 % (option burn rate)
Option re-pricing is prohibited 44.5% 48.5%
Company expenses stock options 5.8% 12.1%
Category 6: Ownership
All directors with more than one year of 91.8% 87.9%
 service own stock
Executives are subject to stock ownership 25.3% 33.3%
 guidelines
Directors are subject to stock ownership 21.0% 24.2%
 guidelines

Category 7: Progressive Practices

Mandatory retirement age for directors exists 31.3% 33.3%
Performance of the board is reviewed regularly 36.9% 36.4%
A board-approved CEO succession plan is in place 31.5% 36.4%
Board has outside advisors 29.9% 33.3%
Directors are required to submit their 21.4% 30.3%
 resignation upon a change in job status
Outside directors meet without the CEO and 36.7% 45.5%
 disclose the number of times they meet
Director term limits exist 4.0% 3.0%
Category 8: State of Incorporation
Incorporation in a state without any 2.6% 0.0%
 anti-takeover provisions

Minimum Governance Standard E Above E

Category 1: Board of Directors

Board is controlled by more than 50 % 85.1% 86.0%
 independent directors
Compensation committee is composed solely 86.2% 82.0%
 of independent directors
Nominating committee is composed solely 71.3% 59.7%
 of independent directors
Governance committee exists and meets at 74.5% 69.4%
 least once during the year
Size of the board is at least six but not 93.6% 95.2%
 greater than fifteen members
Shareholder approval is required to change 10.6% 13.4%
 board size
Shareholders have cumulative voting rights 9.6% 9.9%
 to elect directors
Board members are elected annually 39.4% 40.1%
CEO serves on no more than two other boards 94.7% 92.2%
 of public companies
Outside directors serve on no more than five 3.2% 4.0%
 addn'l boards of public companies
No former CEO's sit on the board 80.9% 74.5%
The CEO and chairman duties are separated and a 61.7% 55.4%
 lead director is specified
Board guidelines are published in the firm 31.9% 37.1%
 proxy statement
Managers respond to all shareholder proposals 98.9% 98.1%
 within twelve months of the last
 shareholder meeting

Directors attend at least 75 % of board meetings 98.9% 97.8%

 or have a valid excuse for non- attendance
Shareholders vote on directors selected to 31.9% 42.7%
 fill vacancies
CEO is not listed as having a related party 90.4% 81.5%
 transaction in the proxy statement

Category 2: Audit

Audit committee consists solely of independent 87.2% 82.0%
 outside directors
Consulting fees paid to auditors are less than 76.6% 76.3%
 audit fees paid to auditors
Company has a formal policy of audit rotation 23.4% 17.5%
Auditors were ratified at the most recent 55.3% 62.1%
 shareholder meeting

Category 3: Charter/Bylaws

Company either has no poison pill or a pill 27.7%% 34.9%
 that has shareholder approval
A simple majority is required to approve a 68.1% 63.7%
 merger (not a supermajority)
A majority vote is required to amend 33.0% 43.0%
 charter/bylaws (not a supermajority)
Shareholders are allowed to call special meetings 30.9% 39.5%
Shareholders may act by written consent and the 21.3% 25.8%
 consent is non-unanimous
Company is not authorized to issue blank check 10.6% 9.7%
 preferred stock (stock over which the
 board has broad authority to determine voting,
 dividend, conversion and other rights)
Board cannot amend bylaws without shareholder 5.3% 4.0%
 approval or can only do so under limited
 circumstances
Company has a single class of common stock 94.7% 91.7%

Category 4: Director Education

One or more directors have participated in an 13.8% 13.4%
 ISS-accredited director education program

Category 5: Executive and Director Compensation

No interlocks exist among directors on the 98.9% 98.9%
 compensation committee
Non-employees do not participate in the 98.9% 96.2%
 company pension plans
Option re-pricing did not occur within the 95.7% 93.0%
 last three years
Stock incentive plans were adopted with 84.0% 79.1%
 shareholder approval
Directors receive all or a portion of their 97.9% 94.4%
 fees in stock
Company does not provide any loans to 90.4% 90.1%
 executives for exercising options
The last time shareholders voted on a pay plan, 54.3% 58.3%
 ISS did not deem its costs to be excessive

The average options granted in the past three 9.6% 11.0%
 years as a %age of basic shares outstanding
 did not exceed 3 % (option burn rate)
Option re-pricing is prohibited 41.5% 44.9%
Company expenses stock options 3.2% 5.9%
Category 6: Ownership
All directors with more than one year of 92.6% 91.9%
 service own stock
Executives are subject to stock ownership 24.5% 24.7%
 guidelines
Directors are subject to stock ownership 21.3% 20.7%
 guidelines

Category 7: Progressive Practices

Mandatory retirement age for directors exists 23.4% 33.1%
Performance of the board is reviewed regularly 35.1% 37.4%
A board-approved CEO succession plan is in place 28.7% 31.7%
Board has outside advisors 25.5% 30.6%
Directors are required to submit their 21.3% 20.7%
 resignation upon a change in job status
Outside directors meet without the CEO and 35.1% 36.3%
 disclose the number of times they meet
Director term limits exist 5.3% 3.8%
Category 8: State of Incorporation
Incorporation in a state without any 2.1% 3.0%
 anti-takeover provisions

REFERENCES

Allen, M., C. Renner, and D. English. 2004. Evaluating the Corporate Board. Strategic Finance 85(7), 37-43.

Anderson, R.C., S.A. Mansi, and D.M. Reeb. 2004. Board Characteristics, Accounting Report Integrity, and the Cost of Debt. Journal of Accounting and Economics 37: 315-342.

Archambeault, D., and F.T. DeZoort. 2001. Auditor Opinion Shopping and the Audit Committee: An Analysis of Suspicious Auditor Switches. International Journal of Auditing 5: 33-52.

Barth, M.E., D.P. Cram, and K.K. Nelson. 2001. Accruals and the Prediction of Future Cash Flows. The Accounting Review 76(January): 27-58.

Beasley, M.S. 1996. An Empirical Analysis of the Relation between Board of Director Composition and Financial Statement Fraud. The Accounting Review 71(4): 443-465.

Brown, L.D., and M.L. Caylor. 2004. Corporate Governance and Firm Performance. Available at: Social Science Research Network http://ssrn.com/abstract=586423

Dahya, J., J.J. McConnell, and N.G. Travlos. 2002. The Cadbury Committee, Corporate Performance, and Top Management Turnover. The Journal of Finance LVII(1): 461-483.

Dechow, P.M., and I.D. Dichev. 2002. The Quality of Accruals and Earnings: The Role of Accrual Estimation Errors. The Accounting Review 77: 35-59.

Dechow, P.M., R.G. Sloan, and A.P. Sweeney. 1996. Causes and Consequences of Earnings Manipulation: An Analysis of Firms Subject to Enforcement Actions by the SEC. Contemporary Accounting Research 13(1): 1-36.

Dunn, P. 2004. The Impact of Insider Power on Fraudulent Financial Reporting. Journal of Management 30(3): 397-412.

Fama, E.F. and M.C. Jensen, 1983. Separation of Ownership and Control. Journal of Law and Economics 26, 301-325.

Farber, D.B. 2005. Restoring Trust After Fraud: Does Corporate Governance Matter? The Accounting Review 80(2): 539-561.

Farinha, J. 2003. Corporate Governance: A Survey of the Literature. Paper presented at the meeting of the Universidade do Porto Economia. Discussion Paper No. 2003-06. Available at: http://ssrn.com/abstract=470801

Fich, E.M., and A. Shivdasani. 2004. Are Busy Boards Effective Monitors? European Corporate Governance Institute. Available at: http://ssrn.com/abstract=607364

Gompers, P., J. Ishii, and A. Metrick. 2003. Corporate Governance and Equity Pricing. The Quarterly Journal of Economics 118(1): 107-155.

Griffith, J.M., L. Fogelberg, and H.S. Weeks. 2002. CEO Ownership, Corporate Control, and Bank Performance. Journal of Economics and Finance 26(2): 170-183.

Hair, Jr., J.H., R.E. Anderson, R.L. Tatham, and W.C. Black. 1998. Multivariate Data Analysis (5th ed.). Upper Saddle River, New Jersey: Prentice Hall.

IRRC, Investor Responsibility Research Center. 2005. Available at: http://www.irrc.org

ISS, Institutional Shareholder Services. 2003. Best Practices User Guide & Glossary. Rockville, MD.

Jensen, M.C., and W.H. Meckling. 1976. Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure. Journal of Financial Economics 3: 305-360.

Klein, A. 2002a. Audit Committee, Board of Director Characteristics, and Earnings Management. Journal of Accounting and Economics 33: 375-400.

Klein, A. 2002b. Economic Determinants of Audit Committee Independence. The Accounting Review 77(2): 435-452.

Krishnan, J. 2005. Audit Committee Quality and Internal Control: An Empirical Analysis. The Accounting Review 80(2): 649-675.

McNichols, M.F. 2002. Discussion of the Quality of Accruals and Earnings: The Role of Accrual Estimation Errors. The Accounting Review 77: 761-69.

Morck, R., A. Shleifer, and R.W. Vishny. 1988. Management Ownership and Market Valuation. Journal of Financial Economics 20: 293-315.

Niskanen, W.A. 2003, December. The Real Governance Challenges. Chief Executive 194: 46.

Peasnell, K.V., P.F. Pope, and S. Young. 2003. Managerial Equity Ownership and the Demand for Outside Directors. European Financial Management 9(2): 231-250.

SEC, U.S. Securities and Exchange Commission. November 4, 2003. Release No. 34-48745. Available at: http://www.sec.gov/rules/34-48747.htm

Warfield, T.D., J.J. Wild, and K.L. Wild. 1995. Managerial Ownership, Accounting Choices, and Informativeness of Earnings. Journal of Accounting and Economics 20: 61-91.

Weisbach, M.S. 1988. Outside Directors and CEO Turnover. Journal of Financial Economics 20: 431-460.

Wells, R. Oct 24, 2002. Restatements of Profits Prove Costly to Investors. The Wall Street Journal: D.2.

Xie, B., W.N. Davidson III, and P.J. DaDalt. 2003. Earnings Management and Corporate Governance: The Role of the Board and Audit Committee. Journal of Corporate Finance 9: 295-316.

Yeo, G.H.H., P.M.S. Tan, and S. Chen. 2002. Corporate Ownership Structure and the Informativeness of Earnings. Journal of Business Finance & Accounting 29(7): 1023-1046.

Young, M.N., A.K. Buchholtz, and D. Ahlstrom. 2003. How Can Board Members Be Empowered If They Are Spread Too Thin? S.A.M. Advanced Management Journal 68(4): 4-11.

Teresa M. Pergola, The University of Tampa

Gilbert W. Joseph, The University of Tampa

Ali Jenzarli, The University of Tampa

Table 1: Sample Used in the Analysis

Panel A: Sample reconciliation

Initial S&P indices sample for 2002 1,500
Missing Compustat data (7)
Utility firms (SIC code 4900) (91)
Financial Service firms (SIC codes 6000--6999) (233)
Non-US firms (3)
Firms with a fiscal year change in 2002 (3)
Outliers (42)
Firms with missing governance data (70)
SIC code 20 (30)
Firms with less than thirty observations (522)
Final Sample 499

Panel B: SIC (Fama-French Codes) and
Corresponding Number of Firms in the Sample

SIC Code and Industry Name # of firms

13 (3) Minerals--oil and gas extractions 31
28 (14) Manufacturing--chemicals and allied products 80
35 (17) Manufacturing--industrial machinery and equipment 73
36 (18) Manufacturing--electrical and electrical equipment 103
37 (19) Manufacturing--transportation equipment 35
38 (20) Manufacturing--Instruments and related products 72
73 (32) Service--business services 105
Final Sample 499

Table 2: Descriptive Statistics and Correlations for 2002
Earnings Quality Using the Dechow and Dichev Model

Panel A: Descriptive Statistics (n=499)

Model and Related Variables Mean Std Dev Lower
 Quartile

Change in working capital -.0081 .0404 -.0286
 ([Delta] [WC.sub.t])
Cash flow from operations .1032 .1031 .0542
 ([CFO.sub.t-1])
Cash flow from operations .1100 .0958 .0628
 ([CFO.sub.t])
Cash flow from operations .1181 .1148 .0580
 ([CFO.sub.t+1])
Earnings before long-term .1019 .0899 .0618
 accruals ([Earn.sub.t])
Earnings before long-term .1195 .1198 .0606
 accruals ([Earn.sub.t+1])
Earnings before extraordinary .0213 .1374 -.0028
 items ([Prof.sub.t])
Accruals ([Prof.sub.t] -.0887 .1011 -.1112
 - [CFO.sub.t] = [Acc.sub.t])
Average total assets 5,419 21,809 442
 (in dollars million)
Average Operating Cycle 151 84 91
 (in days)

Model and Related Variables Median Upper
 Quartile

Change in working capital -.0055 .0165
 ([Delta] [WC.sub.t])
Cash flow from operations .0999 .1500
 ([CFO.sub.t-1])
Cash flow from operations .1083 .1579
 ([CFO.sub.t])
Cash flow from operations .1136 .1738
 ([CFO.sub.t+1])
Earnings before long-term .0991 .1417
 accruals ([Earn.sub.t])
Earnings before long-term .1096 .1746
 accruals ([Earn.sub.t+1])
Earnings before extraordinary .0408 .0778
 items ([Prof.sub.t])
Accruals ([Prof.sub.t] -.0711 -.0412
 - [CFO.sub.t] = [Acc.sub.t])
Average total assets 1,092 2,917
 (in dollars million)
Average Operating Cycle 132 200
 (in days)

Panel B: Pearson Correlations

 [[Delta]WC.sub.t] [CFO.sub.t-1] [CFO.sub.t]

[[Delta]WC.sub.t] 0 -.353 **
[CFO.sub.t-1] -0.035 .720 **
[CFO.sub.t] -.353 ** .720 **
[CFO.sub.t+1] 0.019 .690 ** .797 **
[Earn.sub.t] 0.073 .752 ** .907 **
[Prof.sub.t] 4.70% .566 ** .678 **
[Acc.sub.t] .399 ** .086 * -.027
[Earn.sub.t+1] 0.031 .678 ** .813 **

 CFO.sub.t+1] [Earn.sub.t] [Prof.sub.t]

[[Delta]WC.sub.t] 0.019 0.073 0.047
[CFO.sub.t-1] .690 ** .752 ** .566 **
[CFO.sub.t] .797 ** .907 ** .678 **
[CFO.sub.t+1] .858 ** .610 **
[Earn.sub.t] .858 ** .744 **
[Prof.sub.t] .610 ** .744 **
[Acc.sub.t] .074 * .151 ** .717 **
[Earn.sub.t+1] .915 ** .881 ** .630 **

 [Acc.sub.t] [Earn.sub.t+1]

[[Delta]WC.sub.t] .399 ** 0.031
[CFO.sub.t-1] .086 * .678 **
[CFO.sub.t] -0.027 .813 **
[CFO.sub.t+1] .074 * .915 **
[Earn.sub.t] .151 ** .881 **
[Prof.sub.t] .717 ** .630 **
[Acc.sub.t] .085 *
[Earn.sub.t+1] .085 *

Panel C: Pearson Correlations (controlling for the effect
of [CFO.sub.t]

Pearson Correlation [CFO.sub.t-1] [CFO.sub.t+1]

[Delta][WC.sub.t] .338 * .531 *

** Significant at the .01 level

* Significant at the .05 level

Variable definitions:
Cash flow from operations (CFO) = item 308 from the
Compustat Statement of Cash Flows;

Change in working capital ([Delta]WC) = [Delta]AR
+ Inventory - [Delta]AP - [Delta]TP + Other
Assets (net), where AR is accounts receivable,
AP is accounts payable, and TP is taxes payable;

Earnings before long-term accruals
(Earn) = CFO + [Delta]WC;

Earnings before extraordinary items
(Prof) = Compustat item 123; and

Accruals = Prof - CFO.

Average operating cycle = 360/([Sales.sub.f]/Avg [AR.sub.f])
+ 360/([CGS.sub.f]/Avg [Inv.sub.f])

All variables are scaled by average total assets.

Table 3: Regressions of the Change in Working Capital on
Past, Current, and Future Cash Flow From Operations
for the Year 2002

[Delta][WC.sub.t]/[TA.sub.t] = [b.sub.0][1/[TA.sub.t]] +
[b.sub.1][[CFO.sub.t-1]/[TA.sub.t]] + [b.sub.2][[CFO.sub.t]/[TA.sub.t]]
+ [b.sub.3][[CFO.sub.t+1]/[TA.sub.t]]+ [e.sub.t]

SIC Intercept

 [b.sub.0] [b.sub.1] [b.sub.2]

13 Mean .019 .096 -.489
 (t-statistic) (3.257) (2.443) (-7.904)
28 Mean .009 .272 -.562
 (t-statistic) (2.414) (5.873) (-9.392)
35 Mean .011 -.021 -.512
 (t-statistic) (2.054) (-.412) (-6.501)
36 Mean .005 .095 -.613
 (t-statistic) (0.858) (1.963) (-10.501)
37 Mean -.001 .196 -.663
 (t-statistic) (-.157) (2.201) (-7.241)
38 Mean .009 -.056 -.326
 (t-statistic) (1.414) (-1.011) (-4.868)
73 Mean .000 .185 -.499
 (t-statistic) (-.022) (3.905) (-8.292)

SIC [b.sub.3] Adj [R.sup.F Stat

 .168 .678 22.035
 (4.923)
13 Mean .242 .525 30.102
 (t-statistic) (6.061)
28 Mean .226 .445 20.211
 (t-statistic) (2.698)
35 Mean .308 .525 38.636
 (t-statistic) (6.116)
36 Mean .430 .613 18.946
 (t-statistic) (4.621)
37 Mean .256 .324 12.356
 (t-statistic) (5.357)
38 Mean .241 .389 23.112
 (t-statistic) (5.002)
73 Mean
 (t-statistic)

Table 4: Descriptive Statistics for |SEE|, BIO, and GI (N = 499)

Variable Mean Standard Median Minimum Maximum
 Deviation

|SEE| .7691 .6287 .6156 .00066 3.038
BIO 79.4270 22.8080 88.2050 0.00000 99.900
GI 25.7600 5.3180 25.0000 13.00000 41.000

Variable definitions:
|SEE| = absolute value of standardized estimation errors
found in accruals using the Dechow and Dichev Mode l (2002);

BIO = proportion of insider (management and affiliated
directors) equity to total board equity;

GI = a continuous governance index variable indicating
the effectiveness of a composite of governance mechanisms

Table 5--Descriptive Statistics for BIO Groups

Panel A: Curve Estimation Results for IO Groups

Board Insider Board n Mean Mean
Own % Independent BIO % |SEE|
 Own %

0-29 71-100 33 12.76 0.7992
30-49 51-70 24 41.66 0.7262
50-100 0-50 442 85.70 0.7682

Board Insider Board n Relationship p value
Own % Independent of IO to
 Own % |SEE|

0-29 71-100 33 Flat 0.8177
30-49 51-70 24 Positive 0.0370
50-100 0-50 442 Flat 0.1684

Panel B: Total Equity Ownership for BIO Groups

Board Insider Mean Insider Board Mean
Own % % of Total Independent Independent
 Equity Own % % of Total
 Equity

0-29 .86 71-100 7.48
30-49 1.38 51-70 2.21
50-100 7.76 0-50 .67

Panel C: Breakdown of BIO Groups By SIC Code

SIC 0 to % of 30 to % of
Code 29% Total 49% Total

13 2 6% 1 5%
28 7 21% 2 16%
35 4 12% 10 27%
36 6 18% 4 22%
37 1 3% 1 7%
38 8 24% 2 5%
73 5 15% 4 18%
Total 33 100% 24 100%

SIC 50 to % of Total % of
Code 100% Total Total

13 28 6% 31 6%
28 71 16% 80 16%
35 59 13% 73 15%
36 93 21% 103 21%
37 33 7% 35 7%
38 62 15% 72 14%
73 96 22% 105 21%
Total 442 100% 499 100%

Panel D: Breakdown of BIO Groups by Index

Index Firm Type 0 to % of 30 to % of
 29% Total 59% Total

S&P 500 Large Cap 15 46% 15 27%
S&P 400 Mid Cap 9 27% 14 25%
S&P 600 Small Cap 9 27% 27 48%
Total 33 100% 56 100%

Index Firm Type 60 to % of Total % of
 100% Total Total

S&P 500 Large Cap 150 37% 180 36%
S&P 400 Mid Cap 102 25% 125 25%
S&P 600 Small Cap 158 38% 194 39%
Total 372 100% 499 100%

Variable definitions:

Board Insider Ownership (BIO) = proportion of employee
and affiliated board members' stock ownership to total
board stock ownership as determined by SEC independence
guidelines;

Independent Ownership = proportion of independent
board members' stock ownership to total board stock
ownership;

|SEE| = absolute value of standardized estimation
errors found in accruals using the Dechow and
Dichev Model(2002)

Panel B: Total Equity Ownership for BIO Groups

Board Insider Mean Insider Board Mean
Own % % of Total Independent Independent
 Equity Own % % of Total
 Equity

0-29 .86 71-100 7.48
30-49 1.38 51-70 2.21
50-100 7.76 0-50 .67
Panel C: Breakdown of BIO Groups By SIC Code

SIC 0 to % of 30 to % of
Code 29% Total 49% Total

13 2 6% 1 5%
28 7 21% 2 16%
35 4 12% 10 27%
36 6 18% 4 22%
37 1 3% 1 7%
38 8 24% 2 5%
73 5 15% 4 18%
Total 33 100% 24 100%

SIC 50 to % of Total % of
Code 100% Total Total

13 28 6% 31 6%
28 71 16% 80 16%
35 59 13% 73 15%
36 93 21% 103 21%
37 33 7% 35 7%
38 62 15% 72 14%
73 96 22% 105 21%
Total 442 100% 499 100%

Panel D: Breakdown of BIO Groups by Index

Index Firm Type 0 to % of 30 to % of
 29% Total 59% Total

S&P 500 Large Cap 15 46% 15 27%
S&P 400 Mid Cap 9 27% 14 25%
S&P 600 Small Cap 9 27% 27 48%
Total 33 100% 56 100%

Index Firm Type 60 to % of Total % of
 100% Total Total

S&P 500 Large Cap 150 37% 180 36%
S&P 400 Mid Cap 102 25% 125 25%
S&P 600 Small Cap 158 38% 194 39%
Total 372 100% 499 100%

Variable definitions:

Board Insider Ownership (BIO) = proportion of employee
and affiliated board members' stock ownership to total
board stock ownership as determined by SEC independence
guidelines;

Independent Ownership = proportion of independent
board members' stock ownership to total board stock
ownership;

|SEE| = absolute value of standardized estimation
errors found in accruals using the Dechow and
Dichev Model(2002)

Table 6--Panel A--All Firms

Descriptive Statistics for Governance Index (GI)

Panel A: SPSS Curve Estimation of GI on |SEE|

Coefficient Significance T F Stat [R.sup.2] Model
on GI

-.265869 .0510 3.82806 .00764 Logarithmic

Table 6--Panel B--Descriptive Statistics--By Group

BIO Groups N Mean Std. Std. Error Min Max
 Deviation

0 to 30 33 27.7879 5.49294 .95620 18.00 41.00
31 to 49% 24 27.4167 5.09831 1.04069 17.00 35.00
50 to 100% 442 25.5181 5.28000 .25114 13.00 41.00
Total 499 25.7595 5.31822 .23808 13.00 41.00

Table 6- Panel C - ANOVA

 Sum of Df Mean F Sig.
 Squares Square

Between Groups 227.439 2 113.719 4.070 0.018
Within Groups 13857.704 496 27.939
Total 14085.142 498

Variable definitions:

|SEE| = absolute value of standardized estimation errors found
in accruals using theDechow and Dichev Model(2002);

BIO = proportion of insider (management and affiliated
directors) equity to total board equity;

GI = a continuous governance index variable indicating
the effectiveness of a composite of governance mechanisms

Table: Panel D--Differences Between Groups

BIO BIO Groups Mean Std. Sig.
 Difference Error

< 30% > or = 30 % & < 50% 0.37121 1.41801 1.000
 > or = 50% 2.26978 * .95386 .053
> or = 30% & < 50% < 30% -0.37121 1.41801 1.000
 > or = 50% .1.89857 1.10785 .262
> or = 50% < 30% -2.26978 * .95386 .053
 > or = 30 % & < 50% -1.89857 1.10785 .262

* The mean difference is significant at the .05 level.
Variables as defined in Panel C

Table 7" ANOVA on Above and Below Median
GI for Entrenched Firms

Panel A: GI Within the Entrenchment
Range For Firms Above and Below Median GI

 Sum of df Mean F Sig.
 Squares Square

Between Groups 368.167 1 368.167 35.267 .000 *
Within Groups 229.667 22 10.439
Total 597.833 23

* Indicates significant differences in the
means of GI between low and high governance
within the entrenchment group

Panel B: BIO Within the Entrenchment
Range For Firms Above and Below Median GI

 ANOVA

 Sum of df Mean F Sig.
 Squares Square

Between Groups 763.904 1 763.904 1444% .000 *
Within Groups 2857.043 54 52.908
Total 3620.947 55

* Indicates significant differences in the mean
BIO between high and low governance firms
within entrenchment

Table 8: Regression Results for Entrenched Firms N = 24

Panel A: GI on |SEE|

Variable Expected Sign Coefficient (p-value)
Intercept 2.125 (.003)
GI - -.051 (.034)

Panel B: GI and BIO on |SEE|

Variable Expected Sign Coefficient (p-value)
Intercept .445 (.680)
GI - -.042 (.066)
BIO 0 .035 (.071)

All variables as defined in Table 4