The Association between Management Earnings Forecast Errors and Corporate Governance Structure in IPO Firms

The Association between Management Earnings Forecast Errors and Corporate Governance Structure in IPO Firms Ching-Hsi Yang Department of Accounting, National Taipei University 67, Sec. 3, Minsheng E. Rd., Jhongshan District, Taipei City 104, Taiwan (R.O.C.) yangch@mail.ntpu.edu.tw Hui-Sung Kao Department of Accounting, National Taipei University 13F., 567, Gongjheng Rd., Luodong Township, Yilan County 265, Taiwan (R.O.C.) kaohs@post.savs.ilc.edu.tw ABSTRACT This study investigates the relationship between management earnings forecast errors and corporate governance structure in initial public offerings (IPO) companies. To examining initial public offerings (IPO) companies corporate governance structure whether the good or bad of the structure could reduce the information asymmetry, and promoting management forecast likely to be precise and more accurate. Using the sample of initial public offerings (IPO) companies from 2002 February 19th to 2004 fiscal year. It s including 176 observations to use multiple regression models for empirical analysis. Proxy management forecast s preciseness by management forecast errors absolute vale (ABSFE), positive management forecast errors(fe + ) and negative management forecast errors(fe - ). The empirical results indicate that, the negative relations between the board of directors size (DSIZE), the cash-flow rights(cr) and control right deviating from cash-flow rights (DEV) with management forecast errors absolute vale(absfe). And the positive relations between chair of board acting as CEO (CHAIR) and management forecast errors absolute vale (ABSFE). We can t find significant relations between effectiveness of independent directors and independent supervisors (DSRATIO) (DSEXP) and inside director ratio (IDRATIO) with management forecast errors absolute vale (ABSFE). This research result provides regulator rethink institution of independent directors and independent, and help outside investors to do investment decision for initial public offerings (IPO) companies. They can use the signals of corporate governance structure, and understand management forecast s preciseness. Moreover, they could reduce the information asymmetry and avoid questions of adverse choice. Keywords: Management earnings forecast, Corporate governance, IPO Firms Author for correspondence 1

Ⅰ.Introduction This study aims to investigate the initial public offerings (IPO) company of the connection between the corporate governance structure and accuracy of management earnings forecast. In order to attract the public to make investments on IPO companies, the executives are likely to issue favorable earnings forecast. When collecting complete information set of the capital market, the investors (outsider) may often have less information than the executives (insider). Hence, taking the advantage of in possession of more sufficient information than the investors, the executive can on the one hand make accurate forecast and yet on the hand issue inaccurate forecast by hiding the information. As a result, in the case of asymmetrical information between the executive and investors, the quality of a corporate governance structure is essential to the accuracy of management earnings forecast. In this study, it is argued that IPO companies will tend to exaggerate the earnings forecast. Therefore, deviation is much smaller of the earnings forecast issued by management with well established corporate governance structure which reduce problems of adverse selection by the investors. In contrast, companies with less complete corporate governance structure are likely to make larger deviation which aggravates the issues of adverse selection. For IPO companies, the company value is not highly reliable as the company may increase the offering price in order to help raise capital. Hence, the executives may make over-optimistic forecasts. As is showed in prior studies (Hsu, 1992; Lin and Wang, 1994; Su, 1996; Wu, 2001 ), the reliability of management earnings forecast in Taiwan is not as unbiased and do not possess the same information content as that issued by foreign companies executives (Fama, 1980; Lees, 1981; King, Pownall and Waymire, 1990; Palepu et al., 2000). Management earnings forecast is one of the channels to outside communication which serve as the references for investors in regards to future financial status, operation results, and cash flow and decide the value of the IPO companies. As these information serve as important reference for investors and outsiders. Thus, management earnings forecast is an issue worth of discussion. Since Berle and Mean (1932) consider separation hypothesis between the ownership and controlling right and the investigation of agency theory about theory of the firm by Jensen and Meckling (1976), the method of separation of ownership and controlling right is widely adopted by some modern companies. In addition, in the wake of some largest financial frauds, the well corporate structure has become a highly attentive issue. In fact, this issue has been discussed as early as in the 1930 s in the US. Moreover, such issue is highly valued by some enterprises worldwide after the Asian financial crisis in 1997. As a result, Organization for Economic 2

Co-operation and Development (OECD) published the corporate governance principle in April, 1998 and proposed that issues of corporate governance is the main reason that impedes the growth of some Asian countries which is closely connected to the ethic standards and sincerity principle of the management. Current study proposes that the corporate governance structure will have an influence on the reliability and accuracy of management earnings forecast and on the subsequent performance of its stock price after being listed. In the prior literature mainly focused on the relationship between the management earnings forecast and corporate governance structure which has drawn a conclusion that the efficiency of board of directors and structure of the audit committee will lead the executive to issue and update financial prediction which for a major part lacks accuracy. On the other hand, accurate prediction will result in better market response (Karamanou and Vafeas, 2005). In addition, when institutional investor possess a larger ratio of share and that the majority in the board of directors is outside directors, the executive tends to have higher frequency of forecast issuance which is often accurate and do not induce optimistic errors (Ajinkya et al. 2004). However, some research indicates that executive tend to make more accurate forecast that the analysts as the executive usually have more private information and that the public trust the forecast issued by the executive more. Current study argues that asymmetrical information will derive serious agency problem in IPO companies and the IPO companies credibility is likely to be questioned as the executive has the tendency to spread good signals to the capital market in order to draw attention from the investors. On the other hand, whether the corporate governance structure of a well-established company is able to reduce the forecast error and optimistic biased will be the focus of current study. The data source of this study is from the public offering prospectus of IPO companies in Market Observation Post System of Taiwan Stock Exchange Corporation (TSEC) and TEJ database. The research subjects of current study are companies that were listed between February 19 th, 2002 and December 31st, 2004 regarding the listed and OTC review principle of mandatory appointment of independent directors and independent supervisors which included 176 observations as the sample. We finding that the DSIZE, CR and DEV will effectively inhibit management earnings forecast error. In addition, the chair of board acting as CEO (CHAIR) will progress management earnings forecast error and that DSRATIO, DSEXP and IDRATIO do not show significantly connection with management earnings forecast error. Hence, it is concluded that the corporate governance structure has an supervising impact which reduce management earnings forecast error by a IPO 3

company and increase the executive s sincerity which avoids situations of asymmetrical information of the IPO company. The remainder of the paper is structured as follows. Section Ⅱ discusses literature review and presents the hypothesis development. Section Ⅲ describes the research design. Section Ⅳ discusses the results and additional analyses. Section Ⅴ presents conclusion and suggestions. Ⅱ. Literature Review and Hypothesis Development Agency problem leads to the acquisition of a balance between the agent and principal in order to maximize the value of shareholders. Prior literature has focused on the discussion of corporate governance and relevant hypotheses such as Agency Theory, Signaling Hypothesis, Expectation Adjustment Hypothesis, Litigation Hypothesis and etc; on the part of Agency Theory, a positive relationship was discovered between the earnings forecast and agency cost (Leftwich et al 1981; Gaber 1985); Signaling Hypothesis indicates that the executive tend to release positive news to the investors for the company s affirmative review on the market(peman, 1980; Grossman, 1981; Gaber, 1985). In addition, Expectation Adjustment Hypothesis suggests that executive tend to issue earnings forecast in order to correct the prediction of future value by analysts so that relevant issues derived from asymmetrical information may be improved(lees, 1981); finally, Litigation Hypothesis argues that the company tends to issue negative earnings forecast under the consideration of litigation cost as the cost of releasing negative new is higher than positive news (Skinner, 1994; Kasznik and Lev, 1995). Current study puts a emphasis on the influence of the corporate governance structure on the accuracy of the earnings forecast of a IPO company. The examination of the corporate governance mechanism will focus on the structure of board of directors, cash flow right and the separation of cash flow right and control right. Hypotheses developments in current study are described in the following paragraph. In the formation of board of directors, outside directors ( 1 ) serve as the representatives of the company s independence with the obligation to supervise. 1 Outside directors as indicated in the literature refer to those who do not involved in the company s directorial duties and thereby do not have share interests with the company s controllers or major shareholders, but supervise the company s operation with professional knowledge and experience. In addition, independent directors categorized as Gray directors in literature abroad are merged in the category of outside or inside directors. Current study stressed on the independence and supervising function of outside directors and the spirit of independent directors in Taiwan. The definition of independent directors in this study is pursuant to Article 14-2 in Stock and Exchange Law in which (Independent directors shall possess professional knowledge and there shall be restrictions on their shareholdings and the positions they may concurrently hold. They shall maintain independence within the scope of their directorial duties, and may not have any direct or indirect interest in the company). 4

Outside directors tend to supervise the executive in a more aggressive approach due to the incentives of reputation capital; in addition, their professional advice will enhance the company s operation efficiency (Fama and Jensen, 1983). In previous literature suggested a positive relationship between the ratio of outside directors to the independence of the management. That is, higher ratio of outside directors leads to better quality of financial prediction (Ajinkya et al., 2004). In addition, some research indicates that the role of outside directors serving as a supervisor will increase the company value (Bacon, 1997; Zehra and Pearce, 1989); moreover, in the study of the structure of the board of directors, it was discovered that better corporate governance structure will generate the incentive to improve the problem of asymmetrical information and the consideration of litigation risk result will result in a positive relationship between the accuracy of earnings forecast and the corporate governance structure. Finally, conservative earnings forecasts tend to show higher reliability which is especially obvious in the study of the connection between the corporate governance structure and negative earnings forecast. As a result, the independence (ratio of outside director to the structure of board of directors) plays an essential role in the issuance of forecast by the corporate governance structure. The issue of independent directors and supervisors receives much attention in recent years. Much research has been conducted abroad regarding the independent directors and supervisors ability to supervise which highly reduce the problem between the principals and agents (Weisbach, 1988; Brickley et al., 1994; Beasley, 1996). On the other hand, researches of Taiwan tend to focus on the fact that independent directors and supervisors do not have obvious effects (Lee, Lin and Chen, 2005). This study expects to find in IPO company grievous asymmetrical information phenomenon and that higher independency will enhance internal supervision and that accurate prediction is a result of a sincere executive and thereby leading to smaller prediction errors. The first hypothesis of this study is as the following: H 1 :For IPO companies, the larger the ratio of independent directors and supervisors is, the more accurate the management earnings forecast is. It regards to the influence of independent directions on the independency of board of directions. However, the competence of independent director and supervisors plays a role in the independent directors and supervisors practice of supervision. The capability of board directors highly depend on their professional knowledge (Morck et al., 1988; Stearns and Mizruchi, 1993). In other words, not only wills the professional knowledge of independent directors and supervisors (2) benefit decision making of the 2 Criteria Governing the Review of Listing and OTC company states that the primary qualification of independent directors as a natural person possessing working experience of more than 5 years in 5

management department but will facilitate efficient supervision. Specialties such as professional knowledge, strategy planning, anti-fraud supervision and business practice of the board directors significantly relate to the performance of the board of directors in major positions (Baysinger and Butler, 1985). Current study predicts that for IPO companies with independent directors and supervisors who are equipped with high professional knowledge, their competence for the position, efficiency of the company supervision mechanism, management sincerity as well as accuracy of earning forecast will significantly increase. As a result, this study hypothesize that inaccuracy of earning forecast made by the management sector will decrease. The second hypothesis of current study is as the following: H 2 :For IPO companies, the higher the competence of the independent directors and supervisors, the more accurate the management earnings forecast is. Prior literature indicates that the size of board of director will influence the efficiency of its operation. It was shown that a smaller board of director will enhance the efficiency of supervision (Jensen, 1993) and that the size of board of director and Tobin Q will generate a reverse chance as a large size of board of directors tend to be less efficient. In addition, an oversized board of director is not only less efficient but has limited impact on the corporate governance mechanism (Yermack, 1996; Karamanou and Vafeas, 2005) which will cause more problems of the issue of agents and make the board of directors as nothing more than a symbol (Hermanlin and Weisbach, 2001). On the other hand, the control holders and the size of board of directors in Taiwan tend to show a negative relationship in which the company tends to gain control of the board of directors by scaling down its size which will result in the increase of family-controlled shares and the decrease of the company s operation performance. Hence, a smaller board of directors in size will have less influence on the corporate governance mechanism (Yeh, Lee and Ko, 2002) and have the advantage of low negotiation cost as well as fewer cases of making compromises and reduce free rider s situations. Nevertheless, it will also have other shortcomings such as lacking the ability to interrogate and supervise (Bushman et al., 2004). Therefore, as the size of board of directors has both positive and negative influence on a corporate governance mechanism, we do not tend to predict its potential influence. This study expects that for IPO companies, the size of board of director will have an influence on the performance of a company s supervision mechanism and its credibility which will in turn influence the accuracy of management earnings forecast. business, law, finance, or company operation. 6

The third hypothesis of current study is as the following: H 3 :For IPO companies, the size of board of director will influence the accuracy of the management earnings forecast. According to the agent theory, the concentration level of the ownership will influence its agency cost (Jensen and Meckling, 1976). Hence, the share ratio of insider to the ownership separation is likely to show a negative relationship and that the more widely dispersed of the share, the agency cost will be higher (Karamanou and Vafeas, 2005). In the case of highly ownership concentrated, the influence of interest conflicts (block shareholders is likely to take the advantage of other shareholders) will decrease the accuracy of earnings forecast (Hsu, 1999). In addition, in the aspect of the structure of board of directors, lower ratio of inside directors will result sin higher compensation of the executive (Core et al., 1999). Hence, fewer inside directors in the board will increase the supervision performance of the board of directors (Hermanlin and Weisbach, 2001). On the other hand, outsiders are more independents than the company s management. But as they have relatively less knowledge on the company s general plans, inside directors who are more familiar with the inside operation of the company will be able to more efficiently supervise the board of directors in making proper decisions (Bushman et al., 2004). Hence, the ratio of inside director in the board also shows both positive and negative influences and thereby we will not attempt to predict its potential influence. Current study predicts that the ratio of inside directors will influence the supervision performance of the board of directors which in turn will affect the executive s credibility level and earnings forecast. The fourth hypothesis of current study is as the following: H 4 :For IPO companies, the seat ratio of inside directors will influence the accuracy of the management earnings forecast. The participation of the chair of board in operation executive will lead the management to make decisions favorable to them and thus decrease the supervision performance of the board of directors which in turn reduce the reliability of the earnings forecast issued by the executive and has incentive to process earnings management. In addition, the chair of board acting as CEO will result in potential manipulation of the board and meeting agenda. In other words, the chair of board may make decisions based on his own interests rather than that of the company and will not be able to objectively evaluate the general manager s performance and fulfill his obligation to supervise (Dayton, 1984). Current study predicts that in the case of a joint position of chair of board and general manager in a IPO company, the company 7

is likely to undergo less efficient supervision mechanism and its credibility level will be questioned which will results in inaccurate earnings forecast and larger prediction errors. The fifth hypothesis proposed in current study is as the following: H 5 :For IPO companies, the chair of board acting as CEO will result in less accuracy of the management earnings forecast. La Porta et al. (1999) proposed the concept of cash flow right and control right which has drawn attention to the issues of ultimate controlling shareholder, pyramid structure and cross-holding. These concepts argue that the measure of cash flow right includes direct control and indirect control of the product sum of the shares will truly depicts and control the conflicts of interest among shareholders. As a result, current study has adopted the method by La Porta et al. (1999) and La Porta et al. (2002) for the calculation of cash flow right ( 3 ). In addition, according to the Convergence of Interests Hypothesis, the higher ratio of the manager s shareholding is, the manager is more likely to convergence the same interests with outside investors (Mehran, 1995). Hence, the more cash flow right the controlling shareholders have, the more incentive they have to accept supervision which will increase the firm s value (La Porta et al., 2002; Claessen et al., 2000) and reduce errors of the earnings forecast upon the carry of better signal from the corporate governance mechanism. Current study predicts that for IPO company, the higher the cash flow right is, the better performance the company s supervision mechanism will show which in turn will increase the executive s credibility level and accuracy of the management earnings forecast and reduce the earnings forecast errors. The sixth hypothesis proposed in this study is as the following: H 6 :For IPO companies, the higher the cash flow right, the more accurate is the management earnings forecast. It was argued in prior literature that the deviation level of the cash flow right and control right will balance the quality of the corporate governance mechanism (La Porta et al., 2002; Claessen et al., 2000). In other words, larger deviation level of the cash flow right and control right will provide the controlling shareholders more incentives to encroach the company s asset and damage interests of outside investors worsening the corporate governance mechanism and disallowing the supervision of the earnings forecast issued by the executive. Current study predicts that for IPO companies, larger deviation level of the cash flow right and control right will result in less efficient supervision mechanism, disputed credibility of the executive, less 3 When the controlling shareholders possess shares as unlisted over-the-counter, ultimate controlling shareholders will have full control of the company (the sample company of in this study). Current study refers to the hypothesis of 50% share holding in Yeh, Lee and Ko (2002) in calculating cash flow right. 8

accurate prediction result and larger prediction errors. The seventh hypothesis proposed in this study is as the following: H 7 :For IPO companies, larger deviation level of the cash flow right and control right will lead to less accurate management earnings forecast. Ⅲ. Research Design Cross-section analysis method is adopted in current study to investigate the relationship between the corporate governance mechanism and accuracy of management earnings forecast in the IPO companies. It is argued that a corporate governance mechanism with high performance will lead to accurate financial prediction and reliable information regarding the prediction by management. Data sources and sample selection In consideration of the experimental variable requirement of the empirical model of independent directors and supervisors, the research subjects of current study are companies with mandatory appointment of independent directors and independent supervisors based on to the review principle of listed and OTC companies during the period of February 19 th, 2002 and December 31 st, 2004. The research samples do not include financial industry, insurance, and securities companies and other industries different than those listed in the regulations (a total of 12 companies). In addition, a total 94 companies are deleted which have transferred from listed to OTC companies or from OTC to listed companies. Because the samples of this study just include IPO companies. In addition, the samples do not include companies that were listed or became OTC companies after February 19 th 2002 and have been waived from the regulation to appoint independent directors and independent supervisors (a total of 102 companies). Lastly, observations that are incomplete or insufficient are not included. Based on the above selection criterion, a total of 176 sample companies are studied. The data source of this study is collected from the public offering prospectus of IPO companies in Market Observation Post System of Taiwan Stock Exchange Corporation (TSEC) which includes detailed information of the board of directors, ownership, and the financial prediction of management. The financial data and information of the stock price are collected from TEJ database. China Credit Information Service, LTD which allows us to determine the type of the company s controlling shareholders and the enterprise it belongs to. Lastly, information of the principal underwriter is provided by the database on the websites of Taiwan Stock Exchange Corporation (TSEC) and Taiwan Securities Association. 9

Definition of variables Current study intends to investigate the IPO companies of the relationship between the corporate governance mechanism and prediction accuracy issued by the executive. Hence, the experimental variables of this study include the corporate governance mechanism, which are structure of board of directors, cash flow right, deviation level of the cash flow right and control right and etc. The dependent variables of this study include management forecast errors (FE), optimistic biased (FE + ), and pessimistic biased (FE - ) which serves as the measure of forecast accuracy issued by the executive. The relevant variables of the empirical model are as the following: Dependent Variable The evaluation of accuracy of the management earnings forecast in this study will be valued by FE, FE + and FE -. The measure of dependent variables will be calculated based on the method proposed in Ajinkya et al. (2004). The accuracy of the management earnings forecast will be represented with the management earnings forecast errors; that is, smaller errors indicate higher accuracy. The prediction of net income before tax studied is adopted from nearer the public offering prospectus of IPO companies, as well as the first net income ( 4) before tax recorded. The definition of FE of this study is as the following: MFEARNINGit - AEARNING FE it = AEARNING Where MFEARNING it FE it it it is the management earnings forecast error of i company in t year; is the predicted net income before tax of i company in t year; AEARNINGit is the actual net income before tax i company in t year; AEARNINGit is the absolute value of actual net income before tax of i company in t year. Current study will use the absolute value of FE it (ABS FE it ) as the dependent variable of the empirical model and divide all samples into the groups of FE + and FE - respectively for separate discussion of the relationship between the corporate governance structure of IPO companies and FE + and FE - on the influences of FE + and FE - on the corporate governance structure. FE + represents overly optimistic biased of the earnings forecast whereas FE - overly pessimistic biased. Experimental Variable In view of the corporate governance mechanism and the influence of independent directors and independent supervisors, current study will adopt 4 It is more appropriate to use net income before tax predicted by the management than management EPS forecast which will avoid errors caused by erroneous estimation of outstanding shares. 10

DSRATIO, which is commonly used in prior literature in the calculation of independence of the board of directors, which is the ratio of independent directors and independent supervisors to total directors and supervisors in the board to examine H 1 ; Second, hypothesis 2 will be examined using the variable method of competence of independent directors and independent supervisors proposed by Lee et al, (2005) in reviewing the working experience and background of the independent directors and independent supervisors. In the case that the independent director and supervisors has been or currently is a high-level executive (chair of board, general manager, vice general manager), industrial technician, or personnel with financial background and practical experiences, these research subject will be give 1 and then be added the sum (shown as DSEXP) to examine the overall suitability of the independent directors and independent supervisors of their influence on the accuracy of earnings forecast issued by the executive (H 2 ); third, the size of board of directors will be evaluated by refereeing to the calculation method of Karamanou and Vafeas (2005) in which the total number of directors will be shown by SDIZE to examine hypothesis 3. Forth, for the evaluation of the ratio of seats of inside directors, current study will adopts the ratio of inside directors in the board (shown as IDRATIO) to examine hypothesis 4; fifth, current study has established dummy variables for the evaluation of the chair of board acting as CEO which will be given 1 or 0 if otherwise to examine hypothesis 5; Sixth, current study has referred to the calculation method used by Claessens et al. (2002), La Porta et al. (2002) and Yeh (2002) for the calculation of cash flow right. In addition, hypothesis 6 will be examined using the sum of the direct shares ratio of ultimate controlling shareholders and product indirect shares ratio of each control path (shown as CR); lastly, the deviation level of the cash flow right and control right which calculation method of Claessens et al (2002) and Yeh (2002) is used to examine hypothesis 7 by dividing ratio of cash flow right with control right (shown as DEV). Control Variable Based on prior literature, we will include variables influencing management earnings forecast in to the empirical model, the evaluation method is as the following: Top 4 Audit Firms (Big4) Previous literature indicates that the audit quality of audit firm is influenced by the outside mechanism of corporate governance structure. For IPO companies, the earnings forecast issued by the executive based on current or future scenarios must be reviewed by accountants who will issue a review report including financial forecast comments and present their opinion to earnings forecast information. Hence, the quality of audit is likely to influence supervision efficiency. Moreover, the incentives 11

of reputation and avoidance of litigation risks too play a role in the audit quality and thereby affecting the transparency of the company s information and corporate governance mechanism (Craswell and Taylor, 1992). Lastly, reviews of earnings forecast conducted by high-quality audit firms will obvious have an influence on the earnings management behavior by executive as well as the management earnings forecast errors (Davidson and Neu, 1993). The evaluation method will be similar to that used in Craswell and Taylor (1992) and Davidson and Neu (1993) in which dummy variables will be conducted and 1 represents companies whose financial forecasts are reviewed by the top 4 audit firms (hereinafter referred to as BIG 4) and 0 means otherwise. Growth Opportunity (Growth) A company s growth opportunity is related to the managers risk interest caused by the compensation of bonus plan and a stock option (Smith and Watts, 1992), which becomes the incentive that will affect management earnings forecast and thereby is included in the empirical model of this study. The evaluation method is the growth rate of the average sales revenue of the last three years (hereinafter referred to as Growth). Good News (NEWS) According to Signaling Hypothesis, the company often takes the initiative to release positive news by issuing the earnings forecast which will prevent the degrading of firm value (Baginski et al., 2002; Cairney and Pantzalis, 2002; Karamanou and Vafeas, 2005; Lee,1997). On the other hand, it was discovered that the discrepancy symbol of random walk of the earnings indicated a negative association with the prediction issued (Baginski et al., 2002). The negative association is triggered due to the prevention of potential litigation by issuing management earnings forecast (Bhojraj, 2002 and Ajinkya., 2004). The evaluation method is to represent with 1 if annual EPS of current year larger than EPS of previous year (hereinafter referred to as News) and 0 otherwise. Standard Deviation of Earnings (RSD) As variation of forecast is likely to influence the accuracy of earnings forecast by the executive ( Karamanou and Vafeas, 2005) and that variation of forecast is associated with standard deviation of earnings, the standard deviation of earnings will be affected by the cost of agency and law enforcement. On the other hand, as the litigation hypothesis is not applicable in Taiwan, for agency cost, the larger the standard deviation of earnings is, the higher the agency cost. In addition, the company also tends to issue voluntary earnings forecast (Lee, 1997). In contrast, larger standard 12

deviation of earnings will increase the risk and reduce accuracy of voluntary earnings forecast (Liu and Chen, 1998; Hsu, 2000). The evaluation method used is the relative standard deviation of ROA ( ) 5 in the past five years (RSD). Term of Forecast (TERM) Prior literature indicates that the term of forecast may serve as a reference to the uncertainty of earnings and accuracy of the executive s concepts that can not observe (Baginski and Hassell, 1997; Ajinkya et al., 2004; Karamanou and Vafeas, 2005). The evaluation method is the number of days between the forecast date and end of fiscal accounting period (indicated as TERM). High Technology Industry (HT) For the industry of high technology, the variation and uncertainty of earnings tend to be larger which will affect accuracy of the management earrings forecast. Current study has referred to prior literature (Ajinkya et al., 2004 and Karamanou and Vafeas, 2005) including to control variable of this study. The evaluation method is to represent electric companies (indicated as HT) with 1 and 0 otherwise. Company Size (SIZE) A company that is large in size will have more resources and more accurate earnings forecast (Michael and Smith, 1992; Chang, 1996; Hsu, 2000). On the other hand, a large size company may also issue inaccurate earnings forecast due to accounting complexity. Moreover, as prior studies indicate that the size of a company includes a combination of many factors, we do not attempt to make predict direction; the evaluation method is take log of the company s total asset (indicated as SIZE). Underwriter Reputation (UWR) As IPO companies tend to have serious asymmetrical information, the company may send signal of the company s confidence for future operation situation to outside investors through the underwriter s reputation (Titman and Trueman, 1986; Carter and Manaster, 1990). Relatively, underwriter s reputation will also indicate the reliability of the management earnings forecast for outside investors. The measure of underwriter s reputation (UWR) is based on the market share ratio. A dummy variable is given in which 0 represents the market share of the underwriter is the last 33%, 1 is the middle 33% and 2 is the first 33% (Chen, Ye and Chen, 2003; Chin, Lin and Hong, 2003; Lin and Chen, 2004 ; Lee et al., 2005). First Underwrite Ratio (ISSUER) 5 Calculation of ROA is EBIT/Average of total asset. 13

Based on the consideration of economic incentives, as the executive attempts to raise more funding which will result in a higher underwriting ratio, it will rely more heavily on outside funding and is opted to make optimistic prediction of the company s future performance. Moreover, higher underwriting ratio indicated lower shares by the insider and that public offering is a way for the executive to solve problems of financial predicament (Bartov et al., 2002). As a result, higher underwriting ratio is likely to lead to larger management earnings forecast errors and thereby shall be included in the empirical model of this study. The evaluation for first time underwriting ratio is the ratio of issued shared divide by outstanding shares in the IPO company (which is indicated as ISSUER). Empirical Model The information transparency of IPO company is relatively lower than that of an already listed and OTC company. Under the circumstance of serious asymmetrical information, it is argued in current study that the executive tends to issue more accurate earnings forecast when the company possesses a well-established corporate governance mechanism. Hence, the relationship between the accuracy of management earnings forecast and the corporate governance mechanism is a worth-noting issue for investors who rely on the management earning forecasts as investment decision of IPO companies. As a result, based on controlling the relevant variables of prior literature which affect management earnings forecast, current study attempts to investigate the relationship between the corporate governance mechanism and management earnings forecast errors as well as that between the corporate governance mechanism and optimistic (pessimistic) management earnings forecast errors. The method of ordinary least square (OLS) is conducted in this study. The empirical model is as the following: OLS empirical model 1: the influence of the corporate governance mechanism on the absolute value of management earnings forecast errors. ABSFE it = C + β1dsratio it + β 2DSEXPit + β3dsize it + β 4IDRATIO it + β5chair it + β6cr +β8big4it +β9growth+β10newsit +β11rsdit +β12termit + β13htit β14sizeit +β15uwrit +β16issuerit + εit + (1) it + β 7 DEV it OLS empirical model 2: the influence of the corporate governance mechanism on the biased of optimistic (pessimistic) management earnings forecast. FE + it - ( FE it ) = C + β1dsratio it + β 2DSEXPit + β3dsize it + β 4IDRATIOit + β5chair it + β6cr it + β7devit +β 8BIG4it +β9growth+β10newsit +β11rsdit +β12termit + β13htit 14SIZEit +β15uwrit +β16issuerit + εit + β (2) Definitions of relevant variable in the above models are as the following. About 14

experimental variables, FE it is i company, errors of management earnings forecast in t term; the evaluation method is (the net income before tax of i company in t term net income before tax of i company in t term) / the absolute value of net income before tax of i company in t term; ABSFE it is absolute value of management earnings forecast errors of i company in t term; the evaluation method is to take absolute value of FE it ; DSRATIO it is the ratio of independent directors and independent supervisors divide by total directors and supervisors in the board in t term of i company; DSEXP it is competence of independent directors and supervisors in t term of i company; the evaluation method is to establish dummy variables in which 1 is given to add on the total if the independent directors and supervisors was a former general manager, vice general manager, chair of the board or industrial technician of a listed and OTC company or with financial background and 0 is given otherwise; DSIZE it is the total number of directors in the board of i company in t term; IDRATIO it is the ratio of inside directors divide by the size of the board of i company in t term; CHAIR it is dummy variable; 1 is given is the chair of director of i company in t term also serve as the general manager and 0 is give otherwise; CR it is the cash flow right of i company in t term; the evaluation method is the sum of direct share ratio of ultimate controlling shareholders and the product of the indirect share ratio of each control path; DEV it is the deviation level of cash flow right and control right in t term of i company; the evaluation method is the ratio arrived from dividing control right by cash flow right. About control variables, BIG4 it is dummy variable; 1 will be given for i company in t term if the financial forecast is audited by one of the top 4 audit firms, 0 will be given otherwise; GROWTH it is the growth opportunity of i company in t term; the evaluation method is the growth rate of average sales revenue of the past three years; NEWS it is good news of i company in t term; the evaluation method is to establish dummy variable in which 1 will be given if the EPS of current year is larger than that of the previous year and 0 represents otherwise; RSD it is the standard deviation of ROA ( 6) of i company in t term in the past 5 years; TERM it is the number of days between the forecast date and end of the fiscal accounting period of i company in t term; HT it is dummy variable; 1 will be given if i company is an electric industry in t term, 0 will be given otherwise; SIZE it is take log for the value of total asset of the end of the fiscal accounting period of i company in t term; UWR it is for i company, underwriter reputation of the underwriter selected by IPO companies; dummy variable shall be established; ISSUER it is for i company, first underwrite ratio of the IPO companies; the evaluation method is ratio of issued shares of the IPO company divide by all outstanding shares; C is intercept item; ε is error item. it 6 Calculation of ROA is EBIT/Average of total asset. 15

Ⅳ. Empirical Result Descriptive Statistics Analysis Table 1 represents the descriptive statistics of all variables used in current study. TABLE 1 Descriptive Statistics Variable a Mean Median Maximum Minimum Std. N =176 FE 0.1637-0.010 9.2293-0.6927 1.0348 ABSFE 0.3490 0.1347 9.2293 0 0.9876 DSRATIO 0.3500 0.3167 0.6 0.2143 0.0811 DSEXP 1.7247 2 3 0 0.6356 DSIZE 9.7360 10 14 5 1.5231 IDRATIO 0.3296 0.375 1 0 0.1196 CHAIR 0.6067 1 1 0 0.4899 CR 27.2937 24.805 84.48 0.34 18.2288 DEV 1.1016 1 4.5390 1 0.3817 BIG4 0.9045 1 1 0 0.2947 GROWTH 0.6081 0.2626 29.2993-0.7783 2.4370 NEWS 0.5819 1 1 0 0.4946 RSD 8.8290 6.7249 67.0693 0.5423 8.1394 TERM 229.8034 223.5 485 22 110.448 HT 0.8090 1 1 0 0.3942 SIZE 13.8071 13.6873 17.4415 11.9309 0.9168 UWR 1.6011 2 2 0 0.575781 ISSUER 0.1103 0.1 0.8842 0.01 0.065776 a FE it is i company, errors of management earnings forecast in t term; ABSFE it is absolute value of management earnings forecast errors of i company in t term; DSRATIO it is the ratio of independent directors and independent supervisors divide by total directors and supervisors in the board in t term of i company; DSEXP it is competence of independent directors and supervisors in t term of i company; DSIZE it is the total number of directors in the board of i company in t term; IDRATIO it is the ratio of inside directors to the size of board of director of i company in t term; CHAIR it is dummy variable; 1 is given is the chair of director of i company in t term also serve as the general manager and 0 is give otherwise; CR it is the cash flow right of i company in t term; DEV it is the deviation level of cash flow right and control right in t term of i company; BIG4 it is dummy variable; 1 will be given for i company in t term if the financial forecast is audited by one of the top 4 audit firms, 0 will be given otherwise; GROWTH it is the growth opportunity of i company in t term; NEWS it is good news of i company in t term; RSD it is the standard deviation of ROA of i company in t term in the past 5 years; TERM it is the number of days between the prediction date and end of the fiscal accounting period of i company in t term; HT it is dummy variable; 1 will be given if i company is an electric industry in t term, 0 will be given otherwise; SIZE it is total asset of the end of year in i company in t term take log; UWR it is for i company, underwriter reputation of the IPO company; dummy variable shall be established; ISSUER it is for i company, first underwrite ratio of the IPO company. 16

It is shown in table 1 that the FE average is 16.37% in which fifth of it is forecast error. The average of ABSFE is 34.90% and the average of DSRATIO is 35%. Because it is regarding the listed and OTC review principle of mandatory appointment of independent directors and independent supervisors ( 7 ). The DSEXP of companies independent directors and independent supervisors is 2, indicating they have highly professional competence. The average of DSIZE is ten people and the average of IDRATIO is close to 33%. The probability of CHAIR relatively high and the average for CR is 27.29%. The average of DEV is 1.10, suggesting that the control right is higher than cash flow right, a supportive evidence of the theory proposed by La Porta et al. (1999). Table 2 represents the correlation coefficient of Pearson and Spearman among all variables of the empirical model. It is shown in table 2 that the correlation coefficient among each variable is under 10% ( 8) which is the lowest correlation. Furthermore, the correlation coefficient between DSIZE and DSRATIO is the largest (Spearman correlation coefficient) and the Pearson correlation coefficient between DSIZE and DSRATIO is the second whereas the Pearson correlation coefficient between ABSFE and ISSUER is the third. Moreover, the VIF value of empirical regression model is less than 10. Hence, the problem of collinearity does not exist between empirical variables and control variables (Neter et al. 1990). Analysis of empirical model multiple regression OLS was used in this study to conduct empirical analysis. Based on prior literature, controlling the variables will affect the relevant variables of the management earnings forecast. As a result, multiple regressions will be used to estimate the influence on the corporate governance structure to management earnings forecast error. The research period is between 2002 and 2004 and the research subjects are IPO companies who are enforceable required to appoint independent directors and independent supervisors. The results of the empirical are shown in table 3 (The influence of the corporate governance mechanism on the ABSFE) and talbe 4 (The influence of the corporate governance mechanism on biased of optimistic and pessimistic management earnings forecasts). In Table 3, the Adjusted R 2 of empirical model is 49.35% which possesses explanatory power itself. The F value of the model s goodness to fit is 11.66 (P-value is 0) which is at its significant. As to the empirical results, it was discovered that the DSRATION and DSEXP did not have a significantly impact on the absolute value of 7 Beginning on February 19 th in 2002, all companies that apply for being listed and OTC in accordance with criteria of review for listed and OTC companies shall appoint independent directors not less than two in number and independent supervisors not less than one in number. 8 r 0.3 :low correlation; 0.3 <r< 0.7 :middle correlation; r.7 : high correlation. 17

TABLE 2 Pearson and Spearman Correlations Between Variables Variable a ABSFE FE DS RATIO DSEXP ID GROWT DSIZE CHAIR CR DEV BIG4 NEWS RSD TERM HT SIZE UWR ISSUER RATIO H ABSFE 0.96065*** -0.06563 0.03994 0.05479-0.1456** 0.11235-0.11354-0.03742 0.04491 0.26873*** -0.10409 0.1053 0.08982 0.07158 0.21477*** 0.03527 0.58248*** FE 0.03385-0.0883 0.0278 0.08206-0.17921** 0.10549-0.13317* -0.04049 0.04514 0.25995*** -0.21985*** 0.11704 0.03385 0.06279 0.20059*** 0.02139 0.56265*** DSRATIO 0.08086 0.00179 0.0398-0.616*** 0.07793 0.03209 0.27518*** 0.01925-0.00019-0.08093-0.018-0.07801 0.0712-0.06813-0.15432** 0.06248-0.05007 DSEXP -0.10784-0.10642 0.00424 0.16377 0.07907 0.04954-0.09139 0.02559 0.13028* 0.05364 0.04533-0.02924-0.03329 0.10463 0.04966 0.11509 0.00525 DSIZE -0.11654 0.01464-0.66085*** 0.15667** -0.00524-0.1021-0.18642*** 0.06915 0.03161 0.17285** 0.14554** 0.08239 0.02555 0.13195* 0.31933*** 0.03384 0.06194 IDRATIO -0.07252-0.08493-0.03966 0.08707 0.11936-0.06229 0.20904*** 0.12077* 0.04189 0.02546-0.03077 0.0035 0.07758-0.02082 0.00835 0.07767-0.08074 CHAIR 0.09272 0.03912 0.0092 0.05084-0.11338-0.08885 0.20058*** -0.02324-0.06595 0.08255 0.01721-0.1462** 0.00107-0.04011-0.01761 0.00158 0.0374 CR 0.02358-0.03393 0.27749*** -0.12161* -0.20458*** 0.18754*** 0.17862** -0.10068-0.12943* 0.0339 0.02519-0.1456** 0.07732-0.23072*** 0.07395 0.02964-0.03075 DEV 0.09538-0.12087* 0.05972-0.02155-0.07273 0.14408** 0.12488** 0.04129 0.01522-0.02021-0.00828-0.04382 0.01821 0.09017 0.14864** 0.01331-0.00437 BIG4-0.00525 0.01163-0.04302 0.14383** 0.02712 0.03561-0.06595-0.16387** -0.04501 0.03844 0.07357 0.14668** -0.02332 0.13386* 0.12292* 0.04059-0.01419 GROWTH 0.07775-0.26428*** -0.09736 0.14071* 0.21542*** -0.01249-0.00828-0.18094** -0.05076 0.07049 0.14424* 0.03281 0.05197** 0.04797 0.30402*** -0.09295-0.09959 NEWS 0.11097-0.47413*** -0.0642 0.04456 0.15672** -0.06583 0.01721 0.03273 0.02267 0.07357 0.36944*** -0.03981-0.08394-0.00624-0.00717-0.0053-0.0988 RSD -0.03724 0.03024-0.08724 0.043 0.10061-0.00665-0.20963*** -0.25205*** -0.08611 0.1805** 0.20189*** -0.11993 0.08018 0.06466-0.07823 0.02818 0.00884 TERM 0.34181*** 0.03842 0.06452-0.01449 0.01484 0.05848 0.01544 0.09047-0.08784-0.02976 0.04224-0.1094 0.0323 0.02833 0.06066 0.04798 0.05883 HT 0.08533 0.00561-0.11307 0.09845 0.14148** 0.01527-0.04011-0.21111*** 0.08351 0.13386* 0.09679-0.00624 0.23156*** 0.01488 0.16211*** -0.03887-0.00201 SIZE 0.07568 0.04614-0.1365* 0.02685 0.26468*** 0.13074* -0.07095 0.09749 0.05836 0.0917 0.0376-0.04663-0.0766 0.04561 0.14074* 0.01552 0.00942 UWR 0.03246-0.01406 0.05521 0.10827 0.0115 0.03862-0.02473-0.01481 0.01854 0.06421 0.0178-0.00107 0.13749* 0.06216-0.01695-0.03228 0.0824 ISSUER 0.14504* 0.08811 0.04506-0.02604-0.05614 0.03646 0.03167-0.03257 0.03356-0.07575-0.028-0.10893-0.0072 0.05034-0.03675-0.19378*** 0.09702 a Variables definition as the table 1. b Right-up is Pearson correlation coefficients, left-down is Spearman correlation coefficients. *, **, *** Significant at the 10 percent, and 5 percent, and 1 percent levels, respectively (two-tailed). 18