read article and answer question

Journal of Accounting and Public Policy 25 (2006) 609–625

www.elsevier.com/locate/jaccpubpol

Corporate governance and the quality of financial analysts’ information

Donal Byard *, Ying Li, Joseph Weintrop

Stan Ross Department of Accountancy, Zicklin School of Business, Baruch College – CUNY,

One Bernard Baruch Way, Box B 12-225, New York City, NY 10010-5585, United States

Abstract

We examine the association between corporate governance and the quality of infor- mation available to financial analysts. Our results indicate that the quality of financial analysts’ information about upcoming earnings increases with the quality of corporate governance. Recent research indicates that the quality of firm-provided mandatory and voluntary disclosures increases in the quality of specific corporate governance mecha- nisms. Our results add to this stream of research by showing that better quality corpo- rate governance is associated with a key benefit to the end users of firm-provided financial disclosures: an increase in the overall quality of information possessed by financial analysts, one of the key users of firm-provided financial disclosures. � 2006 Elsevier Inc. All rights reserved.

1. Introduction

In the United States, many recent financial reporting scandals have been attributed to poor corporate governance oversight of the financial reporting pro- cess (e.g., Agrawal and Chadha, 2005). In response to these financial reporting

0278-4254/$ - see front matter � 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jaccpubpol.2006.07.003

* Corresponding author. Tel.: +1 646 312 3187. E-mail address: [email protected] (D. Byard).

610 D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625

scandals, regulators and major stock exchanges have implemented new rules designed to improve the quality of corporate governance, e.g., by requiring audit committees to be fully independent.1 Implicit in these regulatory changes is a belief that such measures will eventually improve the quality of information available to the users of financial reports (e.g., financial analysts).

In this paper, we test this assertion by studying the association between corpo- rate governance and the quality of financial analysts’ information. Specifically, we examine if better quality governance is associated with a higher quality of ana- lysts’ information. We test our research question using four measures of corpo- rate governance quality: the independence of the board, the independence of the audit committee, the size of the board, and the presence or absence of a dual CEO. We use analysts’ forecast accuracy as a proxy for the quality of a firms’ informa- tion environment, because analysts are key users of firms’ financial disclosures.2

Prior studies document that the quality of firms’ mandatory and voluntary disclosures both increase with the quality of firms’ corporate governance. In the case of firms’ mandatory financial reports, better quality governance is asso- ciated with a lower likelihood of financial statement fraud (Beasley, 1996), and less earnings management (e.g., Dechow et al., 1996). In the case of firms’ vol- untary disclosures, better quality governance is associated with a higher overall level of voluntary disclosure (Eng and Mak, 2003). Better governance is also associated with both a higher likelihood that management will issue a voluntary forecast of future earnings and, if made, a greater level of precision in such fore- casts (see Ajinkya et al., 2005; Karamanou and Vafeas, 2005). While these stud- ies focus on the association between corporate governance quality and firms’ mandatory and voluntary disclosures, our study differs in focus: we examine the association between corporate governance quality and the quality of earn- ings information from a user’s (i.e., financial analysts’) perspective.3 Our

1 For example, the Sarbanes Oxley Act of 2002 requires the audit committee to be fully independent. Both the New York Stock Exchange (NYSE) and NASDAQ have adopted new corporate governance rules, which require that audit, compensation, and nominating committees all be fully independent.

2 Accounting researchers frequently use analysts’ information as a proxy for investors’ information. Studies that use analysts’ earnings forecasts as a proxy for investors’ (i.e., the market’s) expectation rely on this assumption.

3 When studying the association between corporate governance quality and the quality of analysts’ information, we treat board and governance structures as exogenous. Our approach is the same as that of Core et al. (1999, footnote 2), where they observe that, ‘‘Following most prior empirical research in this area, we treat the board and ownership structures as exogenous, when economic theory would argue that these variables are endogenous.’’ This well-established approach of treating governance structures as exogenous is reasonable, in the sense that some institutional features of contracting cause governance characteristics to be ‘‘sticky.’’ For example, directors serve for fixed terms, so naturally it takes time to change board members to adjust to a changed operating environment. Consistent with many prior studies, we argue that it is difficult for firms to have optimal governance structures at all times (e.g., see Larcker et al., 2006).

D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625 611

perspective follows from the Financial Accounting Standards Board’s view that the role of financial reporting is to provide information useful to investors and creditors.

In this paper, we examine the association between analysts’ forecast accu- racy and corporate governance quality using a sample of analysts’ forecasts for 2887 firm-years, representing 1279 separate firms, between 1999 and 2002. The sample is compiled by matching corporate governance data from the Investor Responsibility Research Center (IRRC) database with analyst forecast data from the IBES database. Consistent with our conjecture, we find that analysts following better-governed firms possess better information about these firms’ future earnings. Specifically, analysts’ forecast accuracy is signif- icantly positively related to the independence of the board of directors, and significantly negatively associated with the presence of a CEO who is also the chair of the board (i.e., a dual CEO). We also find some evidence that smaller boards are associated with increased analysts’ forecast accuracy. However, we find no evidence that a more independent audit committee improves the quality of analysts’ information above and beyond that explained by the degree of independence of the full board. This does not imply that audit committees have no impact on firms’ information environ- ments, because our measure of the independence of full boards already cap- tures the independence of board committees, including audit committees. In addition, this insignificant relation is probably driven by a lack of variation in the independence of audit committees in our sample: over 75% of our sam- ple firms had voluntarily adopted independent audit committees during our sample period, 1999–2002, which was before the Sarbanes-Oxley Act of 2002 mandated independent audit committees. Finally, our results are robust to using different forecast periods and to the introduction of Regulation FD. Overall our findings indicate that firms with better governance have better quality information environments.

Our study adds to the growing literature on how governance quality is associated with corporate transparency and disclosure (e.g., Beasley, 1996; Ajinkya et al., 2005). While these studies focus on specific disclosures made by firms, we complement such studies by showing that better corporate governance is ultimately associated with better firm information environ- ments from a user’s (e.g., analysts’) perspective. Our study is useful to reg- ulators and policy makers in that it also indicating the specific governance mechanisms critical in promoting transparency that benefits the users of firms’ disclosures.

The paper proceeds as follows. Section 2 outlines the background of our study and our main research question. Section 3 outlines the sample selection and variable measurement. Section 4 provides the empirical analysis, and Sec- tion 5 concludes.

612 D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625

2. Background and research question

2.1. Background: Analyst forecast accuracy

Our study relies on prior research findings that analysts seek to forecast earnings accurately, and that analysts use firm-provided disclosures in making their forecasts.4 For example, Lang and Lundholm (1996) show that analysts’ absolute forecast errors are smaller for firms with higher levels of disclosure. Ashbaugh and Pincus (2001) find that analysts’ absolute forecast errors decrease when firms increase their level of disclosure, indicating that firm-pro- vided disclosures are important to analysts in their forecasting process.5 As a result, we expect that firm characteristics (i.e, governance quality, as discussed below) that affect the quality of firms’ disclosures also affect the quality of ana- lysts’ information.

2.2. Development of our research conjecture

A common rationale for regulators to expand disclosure requirements is that firms with poor quality governance are more likely to insulate managers from external scrutiny. Consistent with this conjecture, prior studies find that better-governed firms provide mandatory and voluntary disclosures of higher quality. In the case of mandatory disclosure, Beasley (1996) shows that better governance is associated with a lower incidence of financial statement fraud. Dechow et al. (1996), Peasnell et al. (2000), and Klein (2002b) also find less earnings management in better governed firms. In the case of voluntary disclo- sures, Eng and Mak (2003) provide direct evidence that firms with better qual- ity governance provide higher overall levels of disclosure to investors through their annual reports. Similarly, Karamanou and Vafeas (2005) document that better-governed firms are more likely to issue voluntary earnings forecasts, while Ajinkya et al. (2005) also document that for firms that do issue voluntary

4 Both observed industry practice and prior research support the notion that analysts seek to forecast accurately. For example, in the annual rankings of financial analysts published by the financial press, forecast accuracy is used as a criterion to determine the best analysts. In fact, the annual rankings of financial analysts published by the Wall Street Journal rely exclusively on forecast accuracy. Further, Mikhail et al. (1999) show that more accurate analysts are less likely to loose their jobs. Taken together, this evidence indicates that individual analysts have a strong incentive to forecast earnings accurately.

5 More direct evidence that analysts rely on firm-provided disclosures is evident from analysts’ forecast revisions around earnings announcements. Stickel (1989) documents an abnormal flurry of forecast revisions immediately after earnings announcements; while Barron et al. (2002b) find that much of analysts’ private information about upcoming annual earnings is triggered by prior earnings announcement, indicating that firm-provided disclosures are a key input in analysts’ forecasting process.

D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625 613

forecasts of future earnings, better-governed firms tend to issue more precise forecasts.

While prior studies indicate that higher governance quality is associated with improvements in firms’ mandatory and voluntary disclosures (e.g., the incidence of fraud, or the incidence of managerial forecasts), existing research does not directly address how governance quality affects the information ulti- mately available to investors. It does not automatically follow that higher gov- ernance quality leads to improved information available to investors, i.e., analysts. Though analysts rely on firm-provided disclosures, the effect of improved governance quality on the quality of information available to ana- lysts may be marginal, as firm-provided public disclosures and analysts’ pri- vately acquired information may be substitutes. For example, high-intangible firms make less informative public accounting disclosures, but they attract a larger analyst following (Barth et al., 2001) which substitutes for the less infor- mative accounting disclosures.6 In the setting of our study, it is possible that analysts following poorly governed firms similarly substitute their own pri- vately acquired information for the lower quality firm-provided disclosures, leaving the overall quality of analysts’ information unrelated to governance quality. Thus, it is important to test if better corporate governance quality is associated with an improvement in the quality of firms’ information environ- ments from a user’s perspective. This leads to our research conjecture: Higher quality corporate governance is associated with financial analysts having access to better quality information about upcoming earnings. Thus, analyst forecasts are more accurate for better-governed firms. This research question is important, because it provides for a direct test of whether better governance is associated with financial analysts possessing better quality information.

3. Sample selection and variable measurement

3.1. Sample selection

We select all firms covered by the Investor Responsibility Research Center (IRRC) database, which provides data on a broad set of board and director characteristics (e.g., board membership, committee membership, director affil- iations, director stock ownership, and the number of board and committee meetings) for the S&P Super 1500. While the IRRC database is available from 1997 through 2002, it only contains data for institutional ownership from 1999 to 2001. Because we need to control for institutional ownership (see below),

6 In addition, Barron et al. (2002a) show that analysts following high-intangible firms put more weight on their privately acquired information and less weight on the commonly observed public information.

614 D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625

this restricts our sample period. Since we expect a firm’s governance mecha- nisms in a given year to possible impact analysts’ forecasts in the following year, we match the IRRC data (i.e., 1999–2001) with the IBES database for the following year (i.e., 2000–2002). IBES provides data to calculate analysts’ forecast accuracy (see Section 3.2 below). Our final sample consists of 2887 firm-years, representing 1279 firms spread over 21 different two-digit NAICS codes. Our sample is spread fairly evenly across our three sample years: 2000 (936), 2001 (945), and 2002 (1006).

3.2. Measuring forecast accuracy

We measure analysts’ forecast accuracy (ACCURACY) similar to Duru and Reeb (2002). Specifically, for each firm-year, we estimate ACCURACY as the absolute difference between the IBES consensus forecast of annual earnings (made 9 months before the earnings announcement date) and the actual annual earnings reported by IBES, scaled by stock price on the day prior to the mea- surement of the IBES consensus forecast.7

ACCURACY ¼ð�1Þ jFCAST � ACTUALj

PRICE ; ð1Þ

where ACTUAL is the actual annual earnings per share reported by IBES; FCAST is the consensus analyst forecast made nine months prior to the end of the fiscal year; and PRICE is the closing stock price one day before the con- sensus analyst forecast.

3.3. Measures of corporate governance quality

Based on disclosures in proxy statements, the IRRC database classifies directors into three categories: employee (i.e., inside) directors, affiliated direc- tors, and independent directors. Employee directors are company executives who also serve as directors. Affiliated directors are outside directors with cer- tain affiliations with management (e.g., an affiliated director can be a former employee; an employee of a service provider, supplier, or a customer; a recipi- ent of charitable funds; an interlocking or designated director; or a family member of a director or executive etc), which may impair their monitoring of the management. Finally, outside directors without such affiliations are independent directors. Based on this classification, we estimate four proxies for a firm’s corporate governance quality below.

7 The consensus forecast here refers to the mean forecast. Our inferences are unchanged using the median forecast. Our results are also similar when we use alternative forecast horizons, i.e., 3-, 6-, and 12-months (see Table 6).

D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625 615

DUAL_CEO is a dummy variable equal to one if the CEO is also the chair of the board, and zero otherwise. Prior studies (Jensen, 1993; Yermack, 1996) argue that the presence of a dual CEO is an indicator of poor gover- nance. Thus, we predict that analysts’ accuracy is negatively related to DUAL_CEO. BOARD_SIZE is the total number of directors serving on the board of directors. Based on Yermack (1996), we expect that larger boards are less effective, and so we predict that analyst accuracy is negatively related to BOARD_SIZE. BOARD_IND% is the percentage of independent directors on the board. Core et al. (1999) indicate that less independent outside directors provide poor monitoring, so we expect that analysts’ forecast accuracy is posi- tively related to the percentage of independent directors on the board (BOARD_IND%). AUDIT_IND% is the percentage of independent directors on the audit com- mittee. Klein (2002b) finds that earnings management decreases with the per- cent of outside directors on the audit committee, so we expect that the percentage of independent directors on the audit committee (AUDIT_IND%) is positively related to analysts’ forecast accuracy.

3.4. Control variables

When estimating the relation between analysts’ forecast accuracy and firms’ governance quality, we include two sets of control variables, along with sets of dummy variables to control for industry and year fixed effects. The first set controls for firms’ ownership structure, and the second set controls for firm characteristics that affect analysts’ forecast accuracy. Note that, since we measure our four proxies for governance quality at the end of the prior year, we also measure the first set of control variables (i.e., measures of firms’ ownership structure) at the same time. We measure the second set of control variables (i.e., determinants of analysts’ forecast accuracy), however, at the end of the current year, which is the time we measure analysts’ forecast accuracy.

3.4.1. The first set of control variables – ownership structure

Our four proxies for governance quality are based on board characteristics. We include control variables for firms’ ownership structure because, in the overall governance system, ownership structure may serve as a complement to or a substitute for board characteristics (Core et al., 1999). We include three controls for ownership structure, BOARD_1%, MGT_OWN%, and INST_ OWN%. BOARD_1% is a dummy variable equal to one if a director is also a blockholder who owns more than 1% of the firm’s outstanding stock, and

616 D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625

zero otherwise.8 MGT_OWN% is the CEOs’ percentage ownership of a firm’s outstanding stock. INST_OWN% is the percentage of a firm’s stock that is owned by institutional investors. We obtain BOARD_1% and INST_OWN% from the IRRC database, and obtain MGT_OWN% from the ExecuComp database. We do not make predictions on the coefficients of these control variables.

3.4.2. The second set of control variables – determinants of analysts’ forecast

accuracy

Based on prior research, we include controls for firm-characteristic determi- nants of analyst forecast accuracy. The first firm-characteristic control variable we include is firm size: LN_SIZE is the natural log of market capitalization (Clement, 1999). FOLLOW is the number of analysts following a firm (Bhu- shan, 1989). LOSS is a dummy variable equal to one for loss years (Abarbanell and Lehavy, 2003). EPS_VOL is a measure of earnings volatility, i.e., the stan- dard deviation of actual EPS reported by IBES over the prior five-year period, scaled by stock price. DISP is the standard deviation of analysts’ forecasts, scaled by stock price. Since the earnings of some firms are harder to forecast than other, we use earnings volatility (EPS_VOL) and analysts’ forecast disper- sion (DISP) as direct controls for forecast difficulty. In addition to these firm- characteristic determinants of forecast accuracy, we also control for the forecast horizon: HORIZ is forecast horizon, i.e., the number of days between the forecast date and the actual earnings announcement date (Clement, 1999).

After including these two sets of control variables, we predict a positive rela- tion between governance quality and analysts’ forecast accuracy. Our study is in the same line of research that relates firms’ disclosure quality to governance structures (e.g., Eng and Mak, 2003; Ajinkya et al., 2005; Karamanou and Vaf- eas, 2005). Note that, by using explanatory variables for governance structures these studies treat governance structure as exogenous, while economic theory would argue that governance structure is endogenous. Implicit in such a research design is a belief that firms in the real world do not always have optimal gover- nance structures (for a discussion, see Larcker et al., 2006); should firms always choose optimal governance structures, no relation should be expected between their governance choices and disclosure choices. In our study, we share this belief that it is difficult for firms to optimize their governance structures at all times.

Our use of governance structures as explanatory variables does not, however, contradict findings from a separate line of research showing that governance structures vary with cross-sectional determinants (e.g., Demsetz and Lehn, 1985; Klein, 2002a). We interpret these findings as indicating that, although it is difficult for firms to choose optimal governance structures at all times,

8 Our inferences are unchanged if we define this blockholder variable in terms of directors who own a minimum of 5% of the firm’s outstand stock.

D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625 617

governance structures will, in part, be determined by firm characteristics. We address this issue in our research design by including control variables that cap- tures the commonly identified cross-sectional determinants of governance struc- tures. Specifically, among our second set firm characteristics control variables for that determine forecast accuracy, the following variables also control for the cross-sectional variation of governance structures: firm size (LN_SIZE), analyst following (FOLLOW), and incidence of losses (LOSS). For example, Klein (2002a) finds that audit committee independence decreases with growth opportunities, varies with firm size, and is lower in firms that have experienced consecutive losses. To capture these determinants identified by Klein, we include direct measures of firm size and losses, but use analyst following (FOL- LOW) as an indirect control for growth opportunities: high growth firms have higher analyst following (FOLLOW) – see Bhushan (1989) and Barth et al. (2001).9 In addition, FOLLOW also helps control for variation in the nature of firms’ assets: analyst following is significantly positively related to the degree to which firms rely on intangible assets (see Barth et al., 2001).

4. Empirical analysis

Table 1 reports descriptive statistics. Our sample firms tend to be large: mean (median) market capitalization is $9292 billion ($1651 billion). Summary statis- tics on board characteristics reveal that the CEO serves as chairman of the board for 68% of our sample. BOARD_SIZE is fairly uniform across our sample firm- years, with a median of 9, and the 25th (75th)-percentile value of 8 (11). The mean (median) of the percentage of independent directors sitting on the audit committee (AUDIT_IND%) is 86.65 (100) percent, higher than the mean (med- ian) of 64.31 (66.67) percent of independent directors sitting on the full board (BOARD_IND%). Further, 28% of our firm-years have a director who owns more than 1% of the firm’s outstanding stock: mean BOARD_1% is 0.28. Since our sample firms tend to be large, managerial ownership is extremely small: mean MGT_OWN% is 0.25%. Average institutional ownership is 61.64%. Finally, the mean of LOSS is 0.08, indicating that 8% of our firm-years report losses.

Table 2 reports a Pearson correlation matrix. Our four measures of corporate governance quality (DUAL_CEO, BOARD_SIZE, AUDIT_IND%, and BOARD_IND%) are significantly positively correlated with one another. In addition, they are also statistically significantly correlated with many of our con- trols variables. Of these four measures of governance quality, only BOARD_ SIZE is significantly correlated with ACCURACY; however, the correlation is significantly positive, which is contrary to our prediction as it indicates that

9 However, when we include the book-to-market ratio as an additional control for growth opportunities, our results are qualitatively the same.

Table 1 Summary statistics [N = 2887]

Mean Standard deviation Q1 Median Q3

Mean consensus forecast 1.99 1.56 1.08 1.77 2.60 Mean forecast error 0.0177 0.0665 �0.0015 0.0057 0.0222 Stock price 34.21 29.04 17.50 28.00 43.38 Size (market capitalization) 9292.57 30,982.32 634.12 1651.30 5408.22

Dependent variable

Price-scaled forecast accuracy (ACCURACY) �0.0271 0.0633 �0.0268 �0.0164 �0.0031

Primary explanatory variables – corporate governance quality

CEO is also chair of board (DUAL_CEO) 0.68 0.46 0.00 1.00 1.00 Size of board (BOARD_SIZE) 9.77 3.04 8 9 11 Percentage of independent directors on the audit committee (AUDIT_IND%) 86.65 20.17 75.00 100.00 100.00 Percentage of independent directors on the board (BOARD_IND%) 64.31 17.81 53.84 66.67 77.78

Control variables – ownership structure

Blockholder with a minimum of 1% ownership on the board (BOARD_1%) 0.28 0.45 0.00 0.00 1.00 Percentage of stock held by management (MGT_OWN%) 0.25 0.61 0.01 0.03 0.14 Institutional holdings (INST_OWN%) 61.64 19.47 50.35 64.15 75.75

Control variables – firm characteristics determining analysts’ forecast accuracy

Log (market capitalization) (LN_SIZE) 7.59 1.60 6.45 7.41 8.59 Analyst following (FOLLOW) 13.83 8.36 7 12 19

6 1

8 D

. B

y a

rd e t

a l.

/ J

o u

rn a

l o

f A

c c o

u n

tin g

a n

d P

u b

lic P

o lic

y 2

5 (

2 0

0 6

) 6

0 9

– 6

2 5

Loss firm-years (LOSS) 0.08 0.27 0.00 0.00 0.00 Forecast horizon (HORIZ) 322 18.44 313 320 329 Earnings volatility (EPS_VOL) 0.0279 0.0491 0.0090 0.0164 0.0302 Forecast dispersion (DISP) 0.0050 0.0095 0.0009 0.0020 0.0051

The sample comprises the intersection of the Investor Responsibility Research Center’s (IRRC) corporate governance database, data for analysts’ forecasts from the IBES database, and data for managerial ownership from the ExecuComp database. The IRRC database contains data for both the board structure and institutional ownership. Data for institutional ownership is only available for the period 1999–2001. Matching IRRC data for this period with analyst forecast data for the following year from the IBES database results in our sample of 2887 firm-years, representing 1279 separate firms spread over 21 different industries (two-digit NAICS codes). These firm-year observations are spread over the three sample years as follows: 936 (2000), 945 (2001), and 1006 (2002). Variable Definitions: ACCURACY is a measure of analyst forecast accuracy, calculated as follows: �1 * (jMean Forecast � Actual EPSj/StockPrice); DUAL_CEO is a dummy variable equal to one when the CEO is also the chair of the board, and zero otherwise; BOARD_SIZE is the number of member of the board; AUDIT_IND% is the percentage of audit committee members that are independent directors; BOARD_IND% is the percentage of board members that are independent directors; BOARD_1% is a dummy variable equal to one if a blockholder owning more than one percent of the firms’ stock sits on the board; MGT_OWN% is the percentage of stock hold by management; INST_OWN% is the percentage of stock hold by institutional investors; LN_SIZE is the natural log of market capitalization; FOLLOW is the number of analysts following a firm; LOSS is a dummy variable equal to one in firm-years where a loss is reported (IBES actual EPS < 0), and equal to zero otherwise; HORIZ is the number of days between the date of the consensus analyst forecast and the eventual earnings announcement date; EPS_VOL is the standard deviation of earning (IBES actual EPS) over the prior five years, scaled by stock price at the start of the fiscal year; and DISP is the standard deviation of analysts’ forecasts, scaled by stock price at the start of the fiscal year.

D .

B y

a rd

e t

a l.

/ J

o u

rn a

l o

f A

c c o

u n

tin g

a n

d P

u b

lic P

o lic

y 2

5 (

2 0

0 6

) 6

0 9

– 6

2 5

6 1

9

Table 2

Pearson correlations [N = 2887]

DUAL_

CEO

BOARD_

SIZE

AUDIT_

IND%

BOARD_

IND%

BOARD_

1%

MGT_

OWN%

INST_

OWN%

LN_

SIZE

FOLLOW LOSS HORIZ EPS_

VOL

DISP

ACCURACY 0.03 0.12*** �0.01 0.02 �0.02 �0.02 0.02 0.28*** 0.16*** �0.50*** �0.06*** �0.54*** �0.48*** DUAL_CEO 0.06*** 0.04** 0.14*** �0.15*** 0.12*** 0.07*** 0.12*** 0.08*** �0.03 �0.06*** �0.04** �0.07*** BOARD_SIZE 0.04** 0.13*** �0.01 �0.18*** �0.13*** 0.45*** �0.07*** �0.14*** �0.14*** �0.07*** �0.12*** AUDIT_IND% 0.59*** �0.13*** �0.09*** 0.06*** �0.01 �0.01 <0.01 �0.03 0.02 0.04** BOARD_IND% �0.21*** �0.27*** 0.11*** 0.07*** 0.07*** 0.02 �0.13*** 0.04** 0.02 BOARD_1% �0.04** �0.13*** �0.17*** �0.18*** 0.01 0.03 <�0.01 0.02 MGT_OWN% �0.18*** �0.14*** �0.14*** <0.01 0.07*** �0.01 0.12*** INST_OWN% 0.10*** 0.11*** �0.06*** 0.01 �0.05** �0.05*** LN_SIZE 0.79*** �0.20*** �0.23*** �0.26*** �0.31*** FOLLOW �0.12*** �0.21*** �0.14*** �0.15*** LOSS 0.12*** �0.12*** 0.10*** HORIZ �0.12*** 0.10*** EPS_VOL 0.49***

***, **, and * denote significant at the one-, five- and ten-percent levels. Variable Definitions are provided at the end of Table 1.

6 2

0 D

. B

y a

rd e t

a l.

/ J

o u

rn a

l o

f A

c c o

u n

tin g

a n

d P

u b

lic P

o lic

y 2

5 (

2 0

0 6

) 6

0 9

– 6

2 5

D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625 621

analyst forecast accuracy is higher for firms with larger boards. This contradic- tion with our prediction is possibly because this univariate analysis fails to con- trol for other factors likely affecting the relation between analysts’ forecast accuracy (ACCURACY) and governance quality.

In the regression below, we test the association between analysts’ forecast accuracy and governance quality after controlling for firms’ ownership struc- ture, other factors that affect analyst accuracy, as well as year and industry fixed effects:

ACCURACY ¼ a0 þ a1DUAL CEO þ a2BOARD SIZE þ a3AUDIT IND% þ a4BOARD IND% þ a5BOARD 1% þ a6MGT OWN % þ a7INST OWN % þ a8LN SIZE þ a9FOLLOW þ a10LOSS þ a11HORIZ þ a12EPS VOL

þ a13DISP þ /t X3

t¼2 YRt þ

X21

i¼2 siINDi þ e: ð2Þ

For ease of exposition, we omit the time subscripts in this equation. As dis- cussed above, we measure governance quality and ownership structure one year prior to other variables.

Table 3 reports the results from estimating Eq. (2). Analysts’ forecast accu- racy (ACCURACY) is related to three out of the four measures of governance quality in the predicted direction. Specifically, the coefficient on DUAL_CEO is significantly negative (p < 0.01, one-tailed), indicating that analysts’ forecast accuracy is lower when the CEO also serves as board chairman. We also find that forecast accuracy increases with BOARD_IND%, the proportion of inde- pendent directors on the board (p < 0.01, one-tailed), and decreases with BOARD_SIZE (p < 0.06, one-tailed). However, ACCURACY is insignificantly related to AUDIT_IND%, indicating audit committee independence does not explain analysts’ forecast accuracy above and beyond that explained by the degree of independence of the full board. This finding is possibly a result of our measure of board independence already capturing the independence of audit committees, and of a lack of variation in the independence of audit com- mittees. The results for the control variables are largely consistent with prior studies.10 Our findings are also robust to using alternative forecast horizons (see additional results in Table 4), or to a WSL regression procedure that

10 The only exception is INST_OWN%, the percentage of stock held by institutions. We find it significantly negatively related to ACCURACY, which seems contrary to evidence in prior studies that analyst forecast accuracy increases with institutional holdings (see O’Brien and Bhushan, 1990). However, the significant negative result reported here is sensitive to the inclusion of some firm-year observations with extremely high values of INST_OWN%. Removing a small number of firm-year observations where INST_OWN% is extremely high (e.g., over 75% ownership), results in a statistically insignificant coefficient on INST_OWN%.

Table 3 Pooled cross-sectional OLS regression analysis of analyst forecast accuracy and corporate governance [N = 2887]

Predicted sign Coefficient estimate (1) T-stat. p-Value (2)

Intercept ? 11.67 0.58 0.56

Primary explanatory variables – corporate governance quality

DUAL_CEO – �4.61 �2.32 0.01*** BOARD_SIZE – �0.57 �1.55 0.06* AUDIT_IND% + �4.24 �0.77 0.78 BOARD_IND% + 18.31 2.77 <0.01***

Control variables – ownership structure

BOARD_1% ? �0.14 �0.07 0.94 MGT_OWN% ? <�0.01 �0.06 0.87 INST_OWN% ? �0.14 �2.91 <0.01

Control variables – firm characteristics determining analysts’ forecast accuracy

LN_SIZE + 4.48 4.19 <0.01***

FOLLOW + �0.31 �1.67 0.95 LOSS – �72.88 �20.12 <0.01*** HORIZ – �0.10 �1.88 0.03** EPS_VOL – �456.95 �20.97 <0.01*** DISP – �1086.82 �9.05 <0.01***

Adj. R2 (%) 44.75

This table shows the results for our main sample of analysts’ forecasts made nine months before the annual earnings announcement date. The sample consists of 2887 firm-year observations from 2000 to 2002. We measure corporate governance quality and ownership structure one year prior to measuring analysts’ forecast accuracy and control variables for analysts’ forecast accuracy. For our main test we regress forecast accuracy on our experimental variables that capture aspects of firms’ board structures (e.g., DUAL_CEO), and control variables for both firms’ ownership structures (e.g., INST_OWN%), and firm-specific determinants of analyst forecast accuracy (e.g., market capitalization – LN_SIZE). Our regression analysis also includes dummy control variables for firm and year fixed effects

ACCURACY ¼ a0 þ a1 DUAL CEO þ a2BOARD SIZE þ a3 AUDIT IND% þ a4 BOARD IND% þ a5 BOARD 1% þ a6 MGT OWN % þ a7INST OWN % þ a8LN SIZE þ a9 FOLLOW þ a10LOSS þ a11HORIZ þ a12EPS VOL

þ a13DISP þ /t X3

t¼2 YRt þ

X21

i¼2 siINDi þ e ð2Þ

Note, the coefficients for the dummy control variables for years and industries are untabulated. ***, **, and * denote significant at the one-, five- and ten-percent levels. Note: (1) Coefficient estimates shown are multiplied by 1000. (2) p-Values are one tailed for directional hypotheses, and two-tailed otherwise. Variable definitions are provided at the end of Table 1.

622 D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625

Table 4

Pooled cross-sectional OLS regression analysis of analyst forecast accuracy and corporate governance: alternative forecast horizons

Predicted sign 3 Month horizon [N = 3087] 6 Month horizon [N = 3105] 12 Month horizon [N = 3053]

Coefficient estimate (1) p-Value (2) Coefficient estimate (1) p-Value (2) Coefficient estimate (1) p-Value (2)

Intercept ? 152.14 <0.01*** 144.40 <0.01*** 176.84 <0.01***

Primary explanatory variables – corporate governance quality

DUAL_CEO – �4.45 0.05** �5.79 <0.01*** �6.73 <0.01*** BOARD_SIZE – �0.16 0.38 �0.53 0.06* �0.94 <0.01*** AUDIT_IND% + 1.76 0.40 �3.29 0.26 �3.18 0.28 BOARD_IND% + 16.42 0.03** 14.18 <0.01*** 12.17 0.03**

Control variables – ownership structure

BOARD_1% ? �2.16 0.45 �0.88 0.65 1.59 0.43 MGT_OWN% ? 0.03 0.12 10.01 0.80 0.01 0.41

INST_OWN% ? �0.10 0.10 �0.07 0.11 �0.05 0.24

Control variables – firm characteristics determining analysts’ forecast accuracy

LN_SIZE + 2.55 0.04** 3.08 <0.01*** 7.17 <0.01***

FOLLOW + �0.75 0.99 �0.39 0.97 �0.64 0.99 LOSS – �32.21 <0.01*** �55.80 <0.01*** �95.69 <0.01*** HORIZ – �1.03 <0.01*** �0.65 <0.01*** �0.52 <0.01*** EPS_VOL – �238.86 <0.01*** �318.20 <0.01*** �231.00 <0.01*** DISP – �2323.18 <0.01*** �1320.38 <0.01*** �841.99 <0.01***

Adj. R2 (%) 36.69 40.98 44.79

This table shows the results of our analysis using alternative samples of forecasts that are selected three, six, and twelve months before the annual earnings announcement

dates. We use the same regression analysis as for our main sample in Table 3: we regress forecast accuracy on our experimental variables that capture aspects of firms’ board

structures (e.g., DUAL_CEO), and firm-specific determinants of analyst forecast accuracy (e.g., market capitalization – LN_SIZE). The analysis also includes dummy

control variables for firm and year fixed effects.

ACCURACY ¼ a0 þ a1 DUAL CEO þ a2 BOARD SIZE þ a3 AUDIT IND% þ a4 BOARD IND% þ a5 BOARD 1% þ a6 MGT OWN % þ a7 INST OWN %

þ a8 LN SIZE þ a9 FOLLOW þ a10 LOSS þ a11 HORIZ þ a12 EPS VOL þ a13 DISP þ /t X3

t¼2 YRt þ

X21

i¼2 si INDi þ e ð2Þ

Note, the coefficients for the dummy control variables for years and industries are untabulated. ** (*) significant at the one (five) percent level two-tailed. Note: (1) Coefficient estimates shown are multiplied by 1000. (2) p-Values are one tailed for directional hypotheses, and two-tailed otherwise. Variable definitions are provided in Table 1.

D .

B y

a rd

e t

a l.

/ J

o u

rn a

l o

f A

c c o

u n

tin g

a n

d P

u b

lic P

o lic

y 2

5 (

2 0

0 6

) 6

0 9

– 6

2 5

6 2

3

624 D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625

addresses a lack of independence in the data from using multiple observations of the same firms. Results for the pre- and post-FD periods are similar.

5. Conclusions

In this paper, we find that analysts’ forecast accuracy is positively related to firms’ governance quality, after controlling for firms’ ownership structure and other determinants of analysts’ forecast accuracy. We show that analyst fore- cast accuracy increase with the independence of the board, decreases with board size, and decreases when the CEO also serves as chairman of the board. Our study adds to the growing literature on the impact of corporate gover- nance quality on firms’ transparency and disclosure. For example, recent research shows that governance affects both the quality of firms’ public accounting disclosures (e.g., Dechow et al., 1996), and voluntary managerial forecasts (Ajinkya et al., 2005). While prior studies focus on how governance quality affects firms’ disclosure practices, we extend this analysis by adopting a user’s perspective. We focus on a main user group of firms’ disclosures, finan- cial analysts, and document that analysts’ information about firms’ earnings is indeed positively related with these firms’ governance quality.

Acknowledgements

We would like to thank conference participants at the 2006 European Accounting Association annual meetings in Dublin. We would also like to thank the contribution of I/B/E/S International Inc. for providing earnings per share forecast data, available through the Institutional Brokers’ Estimate System. These data have been provided as part of a broad academic program to encourage earnings expectation research. This research has also been sup- ported (Byard) by a grant from PSC-CUNY (# 66597-00 35).

References

Abarbanell, J., Lehavy, R., 2003. Biased forecasts or biased earnings? The role of reported earnings in explaining apparent bias and over/underreaction in analysts’ earnings forecasts. Journal of Accounting and Economics 36, 105–146.

Agrawal, A., Chadha, S., 2005. Corporate governance and accounting scandals. Journal of Law and Economics 48 (2), 371–406.

Ajinkya, B., Bhojraj, S., Sengupta, P., 2005. The association between outside directors, institutional investors and the properties of management earnings forecasts. Journal of Accounting Research 43, 343–376.

Ashbaugh, H., Pincus, M., 2001. Domestic accounting standards, international accounting standards, and predictability of earnings. Journal of Accounting Research 39 (3), 417–435.

D. Byard et al. / Journal of Accounting and Public Policy 25 (2006) 609–625 625

Barron, O., Byard, D., Kile, C., Riedl, E., 2002a. High-technology intangibles on analysts’ forecasts. Journal of Accounting Research 40 (2), 289–312.

Barron, O., Byard, D., Kim, O., 2002b. Changes in analysts’ information around earnings announcements. The Accounting Review 77 (4), 821–846.

Barth, M., Kasznik, R., McNichols, M., 2001. Analyst coverage and intangible assets. Journal of Accounting Research 39 (1), 1–34.

Beasley, M., 1996. An empirical analysis of the relation between the board of directors composition and financial statement fraud. The Accounting Review 71 (4), 443–465.

Bhushan, R., 1989. Firm characteristics and analyst following. Journal of Accounting and Economics 11 (2–3), 255–274.

Clement, M., 1999. Analyst forecast accuracy: Do ability, resources, and portfolio complexity matter? Journal of Accounting and Economics 27 (3), 285–303.

Core, J., Holthausen, R., Larker, D., 1999. Corporate governance, chief executive officer compensation, and firm performance. Journal of Financial Economics 51, 371–406.

Dechow, P., Hutton, A., Sloan, R., 1996. Causes and consequences of earnings management: An analysis of firms subject to enforcement actions by the SEC. Contemporary Accounting Research 13 (1), 1–36.

Demsetz, H., Lehn, K., 1985. The structure of corporate ownership: Causes and consequences. Journal of Political Economy 93, 1155–1177.

Duru, A., Reeb, D., 2002. International diversification and analysts’ forecast accuracy and bias. The Accounting Review 77 (2), 415–433.

Eng, L., Mak, Y., 2003. Corporate governance and voluntary disclosure. Journal of Accounting and Public Policy 22, 325–345.

Jensen, M., 1993. The modern industrial revolution, exit, and the failure of internal controls systems. Journal of Finance 48, 831–880.

Karamanou, I., Vafeas, N., 2005. The association between corporate boards, audit committees, and management earnings forecasts: An empirical analysis. Journal of Accounting Research 43 (3), 453–485.

Klein, A., 2002a. Economic determinants of audit committee independence. The Accounting Review 77 (2), 435–452.

Klein, A., 2002b. Audit committee, board of director characteristics, and earnings management. Journal of Accounting and Economics 33 (3), 375–400.

Lang, M., Lundholm, R., 1996. Corporate disclosure policy and analyst behavior. The Accounting Review 71 (4), 467–492.

Larcker, D., Richardson, S., Tuna, I., 2006. Corporate Governance, Accounting Outcomes, and Organizational Performance. Working paper, University of Pennsylvania.

Mikhail, M., Walther, B., Willis, R., 1999. Does forecast accuracy matter to security analysts? The Accounting Review 74 (2), 185–200.

O’Brien, P., Bhushan, R., 1990. Analyst following and institutional ownership. Journal of Accounting Research 28 (Supplement), 55–76.

Peasnell, K., Pope, P., Young, S., 2000. Accrual management to meet earnings targets: UK evidence pre- and post-Cadbury. British Accounting Review 32, 415–445.

Stickel, S., 1989. The timing of and incentives for annual earnings forecasts near interim financial report releases. Journal of Accounting and Economics 11 (2, 3), 275–292.

Yermack, D., 1996. Higher market valuation for firms with a small board for directors. Journal of Financial Economics 40, 185–211.