How Does Corporate Governance Affect Firm Performance? The Mediating Role of Agency Costs

Chiao Da M日nagement Review ;七l. 30 No. 2, 2010 pp. 107-146

公司治理如何影響企業績效?代理成

本之中介效果

How Does Corporate Governance Affect Firm

Performance? The Mediating Role of Agency Costs

黃志仁 Chih-JenHuang 靜宜大學財金系

Department of Finance, Providence University

薛敏正 1 Min-Jeng Shiue 台北大學會計學系

Department of Accountancy, National Taipei University

張瑪珊 Yu-ShanChang

淡江大學會計學系

Department of Accounting, Tamkang University

摘 要:本文嘗試納入代理成本之中介效果，以 LISREL 方法分析公司治理 對企業績效的影響。本文先驗證「公司治理對績效」與「公司治理對代理成 本」的直接影響關餘，結果顯示公司治理與企業績效呈正相闕 ， 而公司治理 則與代理成本呈負相闕，與預期相符 。 在納入代理成本為中介變數後，公司 治理能有效控制代捏成本進而提升績效，顯示代理成本對於 「 公司治理影響 績效」具有部份中介的效果。亦即，公司治理能有效降低代理成本，進而對 績效有所提升，顯示降低代理問題是公司治理提升績效之重要途徑。本文在 以下三方面對於「公司治理影響績效」文獻具有貢獻: (1)點出代理成本在其 中扮演關鍵的中介角色; (2) 支持治理機制同時存在興利與除弊的功能; (3) 在相關議題上提供未來可能的研究方向 。 關鍵詞:公司治理;代理成本;公司績效;中介效果

Abstract: This is the first study on the mediating effect of agency costs on the association between corporate governance and firm performance. In the LISREL setting, we first examine the direct effect of governance on firm performance and I Corresponding author: Department of Accountancy, National Taipei Universi旬，Taipei, Taiwan,

108 How Does COrpOrale Governance AffeCl Firm Pe吶 rmance? The Mediating Ro/e 01 Agency Costs

agency costs. The results indicate a positive (negative) correlation between govemance and firrn perforrnance (agency costs). We further inc1ude agency costs as a mediating variable; the results show that govemance is negatively correlated with agency costs at a conservative level, whereas the impact of govemance on firrn perforrnance remains significantly positive. This evidence indicates that agency costs play a partial interrnediary role in the relationship between govemance and firrn perforrnance. Specifically, decreasing agency costs is an important way for govemance to improve firrn perforrnance. This paper contributes to the literature on the association between govemance and firrn perforrnance by: (1) identi秒ing the mediating effect of agency costs, (2) highlighting both the beneficial and maladministration-mitigating roles of govemance, and (3) initiating a new research aspect for related topics.

Keywords: Corporate govemance; Agency costs; Corporate perforrnance; Mediating effect

1.

Introduction

The Organization for Economic Cooperation and Development (OECD) emphasized the importance of c。中orate govemance in 1999; it proc1aimed that corporate govemance must perforrn perforrnance-enhancing and safeguarding functions that maximize firrn profits and shareholder value. However, the academy has yet to establish the relevance of the influence of c。中orate

govemance on firrn perforrnance. Little research has investigated both the perforrnance-enhancing and safeguarding functions of corporate govemance simultaneously. This study builds on the existing research on the influence of corporate govemance on firrn perforrnance in order to provide a c1ear understanding of the relationship between the two. We consider agency costs to be a mediating variable in the relationship between corporate govemance and firrn perforrnance and investigate whether corporate goveIτlance can simultaneously perforrn both perforrnance-enhancing and safeguarding functions. If corporate govemance has direct perforrnance-enhancing and safeguarding functions, it must be effective in controlling intemal agency problems and improving firrn

Chiao Da Management ReviewVol. 30 No. 2. 2010 109

perfonnance. We argue that researchers may have overlooked the possibility of agency costs serving as a mediator in this relationship and therefore presented mixed results.

The mediating effect refers to the mediating mechanism between the independent and dependent variables. Barron and Kenny (1986) contend that three conditions must be satisfied for a mediating variable in a regression assumption: (1) there must be a significant correlation between the independent and mediating variables; (2) there must be a significant correlation between the mediating variable and dependent variables; and (3) the inc1usion ofthe mediating variable decreases the strength of the direct relationship between the independent and dependent variables. If, after the inc1usion of a mediating variable, the direct effect between the independent and dependent variables remains statistically significant, it is a partial mediating effect; if, however, the direct effect is insignificant, it is a full mediation effect. In a LISREL setting, this paper first employs confinnatory factor analysis (CFA) to obtain the representative observable variables for each latent variable and establish a completely fit model.

lt then proceeds with analyses of direct and mediating effects among latent variables.

Joreskog and Sorbom (1993) point out that with multicollinearity among independent variables, the LISREL analysis is a more valid method for estimating a model than a simple regression method is. Furthennore, when a certain variable (e.g. agency costs) is probably a dependent variable of another variable (e.g. co叩orate govemance) and at the same time, is an independent variable of another variable (e.g. finn perfonnance), LISREL can analyze the complex causal relationships better than ordinary path analysis. Moreover, LJSREL is not constrained by the hypothesized conditions of the regression methods. Thus, this paper aims to adopt LISREL in analyzing the influence of corporate govemance on finn perfonnance and in detennining whether agency costs exists as a mediating variable between them.

Ang et al. (2000) investigate the corporate govemance mechanism from an agency costs perspective and argue that an effective reduction in agency costs helps to increase finn value. Shiue et al. (2007) examine the mediating effect of

110 How Does Corporate Goνernance Aifect Firm Performance? The Mediating Ro/e 01 Agency Costs

eamings management on the relationship between corporate govemance and firm performance; they discover that corporate govemance can effectively contain intended eamings management within a firm and at the same time, can increase

n口n performance.

This study differs from previous studies in that we use a broader view to analyze the mediating role that agency costs play. First, we test the direct effects of c。中orate govemance on firm performance and on agency costs. We then proceed by examining the mediating effect of agency costs on the relationship between corporate govemance and firm performance and discuss whether beneficial and maladministration-mitigating functions can together increase firm performance.

The sample used in this study is a set of listed companies in Taiwan from 2000 to 2006. We find a positive direct effect of c。中orate govemance on firm performance and a negative direct effect of corporate govemance on agency costs, both consistent with the expectations of this study. The evidence also indicates the mediating role played by agency costs. When agency costs are included as a mediating variable, the direct effect of corporate govemance on firm perfo口nance

increases dramatically, showing that a failure to control agency costs wiU suppress the beneficial function in corporate govemance. The results provide evidence that reducing agency costs is an impo此ant way for Taiwan's corporate govemance to increase firm performance.

Since the effect of corporate govemance on firm performance remains significant after the inclusion of agency costs as a mediating variable, agency costs exist as a partial mediator in the relationship between corporate govemance and firm performance, showing that both beneficial and maladministration-mitigating functions exist for the corporate govemance of Taiwanese firms. This study proposes that the neglect of agency costs as a mediating variable could be one of the critical reasons for the inconclusiveness in related li terature

This paper is organized as follows. The next section provides a literature review. Section 3 introduces the research methods, variable definitions, and data Section 4 presents the empirical results. Finally, Section 5 concludes the study.

Chiao Da Management ReviewVol. 30 No. 2. 2010 l l l

2. Literature Review and Hypothesis

Corporate govemance has two functions: (1) helping in the effective implementation of firm policy, thus improving firm performance and (2) helping mitigate entrenching behavior in a firm. This section reviews the literature on the beneficial and mitigating functions of corporate govemance and its mediating effects.

Corporate govemance is a supervising mechanism for agents; it helps in controlling the agency costs brought about by information asymmetry and conflicts of interest (Berle and Means, 1932). Jensen and Meckling (1976) point out that partial shareholding by managers will incur mismanagement, tunneling, and thereby damaging firm value as well as the wealth of stockholders. Based on the job and risk characteristics of managers and shareholders, Fama and Jensen (1983) analyze firms' decision-making processes and contend that managers do not necessarily take shareholders' interests into consideration when making decisions, thereby resulting in agency costs. Agency costs increase with the debt ratio, which further influences firm performance (Jensen, 1986). Myers (2001) points out that debt cost contributes to a higher probability that firms will face a financial crisis， 的 well as higher bondholder risk, which aggravates the debt agency problem and reduces firm operating performance.

Previous research has not found a common ground on the beneficial role of good corporate govemance in improving fmn performance. McKinsey's report in 1999 indicates that, as compared to other firms in the same industry, Japanese and Taiwanese firms with good c。中orate govemance experience a 20% abnormal stock retum (24% for Korean firms; more than 25% for Thai and lndonesian firms), indicating that good govemance mechanisms can increase firm value. Kim and Lee (2003) examine Korean firms during the Asian financial crisis and show that the magnitude of destructiveness resulting from agency costs is inversely dependent upon the firms' corporate govemance. Gompers et al. (2003) find that firms with hetter corporate govemance are associated with higher stock retum, sales growth, and firm value. Black et al. (2006) use Tobin's Q and market-to-hook ratio to measure performance and find similar results as in

112 How Does Corporale Governance Affecl Firm Performance?

The Mediating Ro/e 01 Agency Costs

Gompers et al. (2003). As for the ownership structure, Mehran (1995) finds a positive relationship between firm performance and the stockholding ratio of managers and outside directors. Agrawal and Knoeber (1996) and Steiner (1996) further state that inside directors who hold a high fraction of stocks are more likely to be informed about firm operating performance, which helps in the quality of decision-making on the board, thereby helping to increase firm performance. As for the board structure, Pearce and Zahra (1992) find that the ratio of outside directors has positively affected firm performance. Agrawal and Knoeber (1996) investigate the relationship between intemal control mechanisms and firm performance and find that firm performance increases with the number of inside directors on the board?

Yet, contrary results are also presented in the literature. Based on the entrenchment hypothesis, Jensen and Ruback (1983) propose that the board of directors usually engages in anti-takeover for their own job security. They found that the possibility of anti-takeover behavior and prerequisite consumption is increasing in the concentration of shares ownership held by directors and large shareholders. Yermack (1996), Conyon and Peck (1998), and Eisenberg et al.

(1998) investigate the influence of board structure on firm performance and find a negative relationship between board size and the market value of the firm. Some studies find no relation between corporate govemance and firm performance. Fama (1980) claims that manager' inclination to raise firm performance is dependent upon the competition in the labor market rather than upon the ownership structure. Gillies and Leblanc (2003) aim to veri今 whether a high-quality board helps maintain a firm's excellence, but are unable to identify the relationship between corporate govemance and firm performance. Yermack (1996) and Bhagat and Black (2002) find no significant relationship between the ratio of outside or independent directors and firm performance. Demsetz and Villalonga (2001) and Balatbat et al. (2004) also point out that ownership structure has no impact on firm performance. Finally, Morck et al. (1988) find a nonlinear relationship between c。中orate govemance and firm performance, but

2 Agrawal and Knoeber (1996) also found that outside directors have a negative effect on firm

Chiao Da Management Revi的VJ台1. 30 No. 2, 2010 113

influenced by the shares held by the board. Specifically, there is a positive relationship between corporate governance and firrn perforrnance when shares held by the board are below 5% or above 25%; there is, however, a negative relationship between them when shares held by the board is between 5% and 25%.

The Iiterature on the mediating variables affecting corporate governance and firrn perforrnance is Iimited. Wu (2004) examines the mediating effect of the combinations of executive management officers' salary, R&D expenses, and ownership structure on firrn value. The results show that R&D expenses have a positive mediating effect on the relationship between executive managers' salaries and firrn value. Executive managers' salaries have a positive mediating effect on the relationship between R&D expenses and firrn value; and the salaries of executive managers in high-tech firrns have a negative mediating effect on the relationship between ownership structure and firrn value. Chiang (2004) finds that non-financial perforrnance has a mediating effect on salary strategy and financial perforrnance. Shiue et al. (2007) investigate the mediating effect of earnings management on the relationship between corporate governance and firrn perforrnance and find that governance can improve firrn perforrnance by effectively controlling for intended earnings management.

Previous studies have shown that the existence of agency costs results in a tendency toward mismanagement and tunneling. The self-benefit motivation of agents will harrn the perforrnance and long-terrn development of a firrn. Little research has investigated the mediating role of agency costs in the influence of

c。中orate governance on firrn perforrnance. Building on the findings in related literature, this study attempts to understand whether beneficial and maladministration-mitigating functions can simultaneously arise from corporate governance in Taiwanese firrns.

We believe that the inc1usion of agency costs as a mediating variable can accurately capture the beneficial and maladministration-mitigating effects of corporate governance on firrn perforrnance. The results of this study will be helpful not only in c1arifying the inconc1usive results about the association between co叩orate governance and firrn perforrnance, but also in recommending

114 HowDo自由中orateGovernance Affect Firm Pe柚rmance? The Mediating Ro/e 01 Agency Costs

future research topics. We hypothesize that:

Hypothesis 1: Agency costs have a mediating effect on the relationship between corporate govemance and firm performance.

3. Research Method

We first measured the direct effect of corporate govemance on firm performance, followed by the inf1uence of corporate govemance on agency costs. We then examined the mediating effect of agency costs in the inf1uence of corporate govemance on firm performance, in order to test for the existence of beneficial and maladministration-mitigating functions of corporate govemance in the firms. Figure 1 shows the research 企amework for this study.

Figure 1 Research Framework 一...

...

...一 一... -一一--.. ...一 ...-一一...

...--

--...

...-

-....

...-

-...

白色.'- -0. ..'- -.. .. 已 Corporate Governance

時

Agency Cost

時

_Firm

Performance

There is no generally accepted method for measuring corporate govemance,

agency costs, and firm performance in the literature; 伽thermore， using a proxy variable to measure each latent variable wi11 easily result in bias, and wi11

eventually affect empirical results. We performed a confirmatory factor analysis (CFA) on the relevant variables based on previous literatu間， in order to test the representativeness of the observed variables and to establish the complete fit for the research model. The description and definition of observable variables for each latent variable are given below.

Chiao Da Management ReviewVol. 30 No. 2, 2010 115

3.

1. Definition of Variables 3.1.1 Corporate Governance

Many methods are available for measuring corporate govemance variables, yet the key point of measurement should be focused on effectively protecting different types of investors (Yen et al., 2006). Bhagat and Jefferris (2002) assert that only considering a single govemance mechanism willlead to overlooking the interaction effect between mechanisms, and will further lead to an error in inference. We use three latent variables to measure c。中orate govemance: ownership structure, composition of the board of directors, and related party transactions. In ownership structure, Jensen and Meckling (1976) point out that when the manager's stockholding ratio is higher or when the ownership is more concentrated, the manager's behavior will not deviate far from the goal of maximizing shareholder value. Agrawal and Knoeber (1996) and Steiner (1996) believe that the higher the stock holding ratio for inside directors, the more in tune they are with the firms' operations; this pushes the board toward efficient and high-quality decision-making, thus effectively increasing firm performance. We used manager's stockholding ratio (SHA), family and personal stocks (SHB),

direct stockholding ratio (SHY), ratio of separation of seats and shares (SHK),

large stockholding ratio (SHC), and critical control level (SHO) 的 factors of ownership structure and performed a confirmatory factor analysis.

A common be!ief is that the composition of the board of directors plays an important role in the advantages and disadvantages of corporate govemance. Cadbury (1999) believes that the focal point of corporate govemance is the way in which the board of directors functions, because the board is responsible for changing and compensating executive managers; the board should actively protect or maintain shareholders' rights and should maximize its supervision functions. Fama (1980) maintains that when a director holds an intemal management position within the firm, the board cannot be effective in supervising and decision-making, and this might even result in conspiracy between the manager and the director through misappropriating shareholders' wealth. Dobrzynski (1993) finds that the increased participation of independent directors in board

116 HowDoes Coψorate Goνernance Affect Firm Performance?

The Mediating Role of Agency Costs

meetings leads to more objective decision-making within the board. Pearce and Zahra (1992) find that the ratio of independent and outside directors has a positive effect on finn perfonnance. Combining the works of previous literature, we employ directors' seats ratio (DRA), directors' seats (DRN), supervisors' seats (DRC) , directors' and supervisors' stock holding ratio (DRD), directors and supervisors pledged shares ratio (DRG), independent directors' and supervisors' seats ratio (DRE), and controlling supervisors' seats ratio (DRB) (Warfield et al., 1995; Chen, 2004) in the CF A for the board of directors composition variables.

Finally, Yeh et al. (2002a) state that Taiwanese finns that have a high share of related pa口y transactions will have a negative effect on corporate govemance; thus, we include purchase of related party transactions (P AB), sales of related

pa向， (P AC), revenue from sales and process of related party (P AE), revenue from purchase and outsourcing of related pa此y (P AF), related pa此y accounts receivable ratio (PAQ), and related party accounts payable ratio (PAU) (Yeh et al., 2002b; Ch凹， 2004) to capture the effect of related party transactions on corporate govemance. CF A was carried out in order to observe for the variables, and the variable definitions are included in Table 1.

3.1.2 Agency Costs

Ang et al. (2000) use the rate of operation expense and the asset tumover rate to measure agency costs. Singh and Davidson III (2003) build on the work of

Ang et al. (2000) and use SG&A costs to measure agency costs and capture managers' discretionary expenses. Yafeh and Yosha (2003) use Japanese finns as samples in their study and use discretionary expenses to measure agency costs.

Their findings suggest that when large shareholders act as managers at the same time, they reduce discretionary expenses and in tum lessen agency costs. From the discussions in previous studies, we include discretionary cost ratio (AQA), free cash flows (AQD), asset tumover rate (AQE), SG&A costs (AQW), finn size (AQR), debt ratio (AQV), and operating cost ratio (AQN) as observable variables. Table 2 shows the observable variables obtained from CF A as well as the definition of agency costs.

Chiao Da Management Review 的1.30No. 2. 2010 117

Table 1

Definition and Observed Variables for Corporate GovernanceB Construct Variables SHA SHB SHY SH COM DIR SUP SHK SHC SHO DRA DRN DRC DRD DRG DRE DRB PAB PAC PAE PAR TRA PAF PAQ Definition

Stockholding ratio of internal directors/number of shares outstanding

(end-。ιye訂)

Family and personal stockholding ratio/number of shares outstanding

(end-oιyear)

Final controller (personal shares ratio + stockholding ratio of unlisted family firrn + family mutual fund stockholding ratio )

Seats of con甘ollingrights/shares of controlling rightsb

M句 or stockholders number of shares heldlnumber of shares outstanding ( end-of-year)

Degree of dispersi凹， final con仕oller's stockholding ratio necessary for C油ntrollingthe firrnc

Controlling directors' seats/dir官ctors'seats Directors' seats

Superviso筒， seats

Directors' and supervisors' number of shares held (end-oιyear) Inumber of shares outstanding (end-of-year)

Directors' and supervisors' shareholding pledged shares ratio (end-of-year)/number of shares outstanding (end-of-year)

Independent directors' and supervisors' seats Controlling sup前visors'seats/supervisors' seats

Related pa此y's total import volume (end-of-year)/firrn's to個 import

volume (end-of-year)

Related party's tot耳J sales volume (end-of-year)/firrns' total sales volume ( end-of-year)

Revenue from sales and process of 間\a阻:d party/operating income ( end-of-year)

Revenue from purchase and outsourcing of related pa此y/ op釘'ating cost ( end-of-year)

Re\ated pa叮 total accounts receivable (end-of-year)/firrn 's to個1 accounts receivable (end-of-year)

PAU RElated mrtyωta\ accounts payable (end-oιyear)/firrn冶的tal ac∞un的

payable (end-oιyear)

'Ownership control1ing r也hts， also terrned voting righ紹， are adopted from the La Porta et al (1999) method, assigning the extreme stockholding ratio in the controlling chain as the indirect stockholding. Seats of con仗。 lling rights: Final controlled directo路， and superviso時， seats

loverall directo時， and supervisors' seats

~ecessary controlling stocks held % (concentration percentage of ownership rights) is calωl旭ted

118 How Does Corporate Governance Affect Firm Performance? The Mediating Role 01 Agency Costs

Table 2

Definition and Observable Variables of Agency Costsa Construct Obse且

variables

..'u.o.A

AQA AQD

AGE COS AQE

AQW AQR AQV AQN

Definition

Total costs minus cost of sales, interest cost, and salary expenses/firm's net sales (end-of-year) b

Eamings before depreciation less tax and interest, then subtract cash dividends/end-of-year total assets for previous period

Firm' s net operating income/total 帥的(end-of-year)

Total selling, general and administrative expense/net operating in∞me

( end-of-year)

Total assets (naturallog)

Totalliabilities/total assets (end-of-year)

Operating costs/net operating income (end-of-year)

'The observable variables for ag得n向f ∞sts include discretior>ary cost ratio (AQA)，企ee cash flo啊

(AQD), asset turnover rate (AQE), SG&A costs (AQW), firm size (AQR), liabilities ratio (AQV), band operatlng c。stratio (AQN)

Management can decide on unnecessary 侃pensesthat are sources of privilege consumption. 3.1.3 Firm Performance

This study uses ROA, EPS, ROE, and FPA to measure firm performance. Previous studies recognize that the effect of govemance mechanisms on the perfonnance of the finn will not reflect on the current period, leading to a lagged effect. We will therefore lag the current retums to the next period and observe for the true effect. We lag the ROA into the next period to ROAl, as well as the ROE into the next period to ROEl, as observed variables. Aside 企om these, because financial ratios are easily influenced by management manipulation, we 臼rther include RETURN as a variable to measure finn perfonnance. Overall, we perfonned CFA on FPA, EPS, ROA, ROE, RETURN, lagged ROA (ROAl), and lagged ROE (ROEl) (Yen et al., 2006). Table 3 shows the observable variables obtained from CF A and the definition of firm perfonnance.

Chiao Da Managemenl Re νiewVo l. 30 No. 2, 2010 119

Table 3

Definition and Observable Variables of Firm Performance3

'， aa ‘自 由M 悟 "山 e-M BSVE ba

ov

e •• ρ •• u vz & •• 位， n oo

c

FPA Definition

Interest cover = Eamings before interest and taxlcurrent period

mterest expense

Eamings per share = Past four seasons recurring income/ current

period base plus weighted average stock number stock retums b Ordinary income + interest expense * (1-25%))/average of total assetsC

Income after tax/total equity EPS

RETURN FIN PER

ROA ROE

ROAl Lagged one period ofretum on asset ROEl Lagged one period ofretum on equity

'Firm performance observable variables inc\ude interest cover multiplier (FPA), eamings per share (EPS), retum on stock price (RETURN), retum on assets (ROA), retum on equity (ROE), lagged ROA (ROAl), and lagged ROE (ROEI)

bRt = (P，*(l+u+的+D)/(P'_l+u*C)- I)*IOO(%) and R, = Ln(P，*(l+且+ß)+D)/(Pt_1+u*C))*100(%), with P

,

的 the stock price in t period (index) ， αas purchase rate in the current period exc1uding righ紹， C as the cash price in the current period exc1uding rights, and D as cash dividend payments for the current period

c Retum on assets after tax before interest.

3.2.

Sample Selection

We acquired data from the Taiwan Stock Exchange Corporation (TWSE) and the OTC registered firrn GreTai Securities Market (GTSM) 的 samples in our study. The sampling period spans the years 2000 to 2006. Since OECD promulgated the c。中orate govemance principle in 1999, many firrns have slowly promoted the application of corporate govemance, By using the year 2000 as a base point for our observation, we can examine for the effect of corporate govemance on Taiwanese firrns from the year 2000 to 2006. We excluded the financial industry and other industries that have insufficient data for our study;

“

rtherrnore, we eliminated full-cash delivery stock firrns, Data are collected from the Taiwan Economic Journal (TEJ), with 4,926 observable samples. Table 4 indicates the distribution of samples by year and by industry, It shows that

120 How Does Corpor，αte Governance Affect Firm Performance? The Mediating Role of Agency Costs

total, followed by the chemical industry (7.33%) and the electrical industry (6.33%).

Table 4

Distribution of Sampled Firms by Year and Industry

2000 2001 2002 2003 2004 2005 2006 Total Percentage Cement 2 3 4 5 5 21 _{。 43%} Food 8 12 12 12 13 13 14 84 1.71% P1astic 19 21 19 19 25 26 26 155 3.15% Textiles 31 32 34 41 44 53 50 285 5.79% E1ectrical 26 31 36 47 53 57 62 312 6.33% Wires and 9 7 8 10 10 12 10 66 1.34% Cab1es Chemica1s 29 37 39 52 62 62 80 361 7.33% Glass and 5 4 4 5 4 6 7 35 _{。 71%} Ceramics Paper 4 4 4 5 7 7 6 37 _{。 75%} manufacturing

1ron and stee1 16 17 18 17 16 23 28 135 2.74%

Rubber tyres 5 7 5 7 10 10 12 56 1.14% Transporting 3 4 3 3 3 3 4 23 _{。 47%} equipment Information 136 186 293 445 535 651 708 2954 59.97% and e1ectronics Other 19 21 37 41 45 50 47 260 5.28% industries Construction 3 4 3 5 4 2 5 26 。 53% Transportation 3 4 4 2 3 6 5 27 0.55% Tourism 4 2 4 6 7 10 10 43 。.87% Merchandising 3 4 5 6 8 11 9 46 0.93% Tota1 324 399 529 726 853 1007 1088 4926 100.00% Percentage 6.58% 8.10% 10.74% 14.74% 17.32% 20.44% 22.09% 100.00 %

4.

Results

4.

1. Descriptive Statistics and Data Transformation

Table 5 shows the descriptive statistics for each observable variable after perfonning CFA. We use a Winsorized method to deal with extreme values, with the interest cover multiplier having a larger deviation than the rest. After dealing with extreme values, the average manager's stockholding ratio is 2.1 % and the average number of supervisors' seats is 2.664, while the average directors' and

Chiao Da Management ReviewVol. 30 No. 2, 2010 121

supervisors' shareholding pledged shares ratio is 8.2%.3 Data transformation is a Table 5

Descriptive Statistics for Observable Variables

Variable Mean Standard Min 2S~屯 Median 7S~屯 _Max

Error Ownership Structure Managers' 0.021 0.030 0.000 0.001 0.008 。.029 。.175 Stockholding Ratio Ratio of Separation of 0.268 0.237 -0.305 _{。 109} 0.251 _{。 419 。 998}

Seats and Shares Family and Personal

0.140 _{。 133} 0.000 0.034 0.108 _{。 210 。 633} Stockholding Ratio Composition of Directors and Supervisors Supervisors' Seats 2.664 0.651 1.000 2.000 3.000 3.000 8.000 Directors' and Superviso凹， 0.082 。 174 0.000 0.000 0.000 0.071 0.910 Stockholding Pledged Ratio Directors' and Superviso間， _{。 244} 0.125 。 515 。 153 。 219 0.313 _{。 664} Stockholding Ratio Related Party Transactions

Related Party Total Impo!

0.186 。.282 。 .000 0.000 0.030 _{。 270} 。.990

Ratio

Related Party's Revenue

。 143 。 194 0.000 0.005 0.060 。 207 。.877

from Sales and Processin Related Party's Revenue from 。.189 。.275 0.000 0.000 0.045 0.258 0.972 Purchase and Outsourcing Agency Cost Discretionary Expense 0.137 2.103 -0.420 0.039 。.068 。 110 6.894 Ratio

Free Cash Flow Ratio 0.120 。.142 -0.621 0.038 _{。.104 。 188} 1.360 Asset Tumover 0.932 0.620 0.003 0.541 0.777 1.144 5.559 Firm Performance

Interest Cover

490.716 2537 -147.38 1.74 11.65 51.76 20903.35 Multiplier

Eamings Per Share 1.494 2.638 -6.460 _{。 115} 1.190 2.720 12.290

Lagged One Period on

0.046 。 793 -41.851 0.014 0.088 0.168 3.344

Retumon 包uit且

necessary step in structural equation modeling (SEM) analysis if the normal assumption is violated (Mice剖， 1989). Chou et al. (1991) and Hu and Bentler

3 Aside 仕'om the interest cover multiplier having a larger deviation, the Winsorized method deals with the initial 1 % and the final 99% of the ex甘emevalues

122 How Does Corporate Goνernance Ajjlxt Firm Perjòrmance? The Mediating Ro/e 01 Agency Costs

(1995) further point out that when a multi-normal assumption is found to be

insignificant, bias will occur in the parameter estimation standard error and in the

t-value of the SEM model, which will cause distortion in the significance of

estimation results. Hence, to avoid non-positive definition technical problems in

SEM analysis, we adopt Abarbanell and Bushee's (1998) method; we adjust the

scale and unit of each observable variable in a common method. The raw data

were arranged in deciles, leading to values of 1-10 for the transformed data. Since

each latent variable is measured by multiple observable variables, and due to the

fact that the expected effect of each observable variable on each latent variable is

not the same, when an observable variable has a negative effect on a latent

variable, the variable is arranged in an opposite direction in order to account for

consistency in the expected direction for the latent variable.4

4.2. Direct Effect

A CFA analysis was first performed to measure the fundamental part ofthe SEM model before testing for the direct effects of latent variables. The main purpose was to test for the hypothesized relationships between variables and latent variables; CFA can also be applied independently when assessing reliability or validi句， as well as when testing for the effectiveness of theories (Bentler,

1989).

We first performed CFA on the observable variables of c。中orate

govemance, agency costs, and firm performance latent variables, and we deleted

the inefficient observable variables in order to adapt to the different measures of

each factor. After deleting for inefficient variables, the CFA results for ownership

structure (SH_COM) of corporate govemance (COR_GOV) inc1ude managers'

stockholding ratio (SHA), family and personal stockholding ratio (SHB), and ratio

of separation of seats and shares (SHK).

As for the CFA results for the composition of directors and supervisors, the

final construct inc1udes the variables supervisors' seats (DRC) , directors' and

4 Only the three related party transaction variables in this study (related pa討y total import ratio,

related party's revenue from sales and processing, and related p訂ty's revenue 企om purchase and outsourcing) are arranged in reverse order.

Chiau Da Managemenl ReviewVul. 30 Nu. 2, 2010 123

supervisors' stockholding ratio (DRD), and directors' and supervisors'

stockholding pledged ratio (DRG). The CFA results for related party transactions

show that the final included variables are related party's total import ratio (PAB),

related party's revenue from sales and processing (PAE), and related party's

revenue from purchase and outsourcing (PAF).

The CFA results for agency costs (AGE_COS) include the variables

discretionary expense ratio (AQA), free cash f10w (AQD), and asset tumover

(AQE). The firm performance CFA results indicate that interest cover multiplier

(FPA), eamings per share (EPS), and lagged one period on retum on equity

(ROEI) are included. To measure for the model fit and assist in examining the

direct and indirect eflì凹的， we attempted to find the optimal fit index to conform to

each variable.

Using the CFA results, we proceeded with two tests for examining direct

effects. Figure 2 shows the direct effect of co叩orate govemance inf1uence on

agency costs; this tested whether the corporate govemance of Taiwanese firms can

effectively control for agency costs problems, which shows the

maladministration-mitigating function of co叩orate govemance. The results

indicate that corporate govemance (COR _ GOV) has a negative significant

correlation with agency costs (AGE_COS) (coefficient = -0.68, t-value = -3.55),

which implies that a maladministration-mitigating function for c。中orate

govemance exists in Taiwanese firms.

Table 6 is the evaluation chart for the direct effect of c。中orate govemance

on agency costs. Ifthe chi-square value indicator is not significant, then a good fit

is indicated between the model and the sample; however, since its significance is

easily affected by the number of samples, we proceeded with the absolute,

comparative, and parsimonious model fit tests. There were 12 tests for the model

fit (see Appendix 1 for the formula). Amongst all of the fit measures, aside from

the comparative measures of 缸， which were below the acceptable standards, the

absolute and parsimonious fits all achieved values above the standards.

Figure 3 and Table 7 show the direct effect of co叩orate govemance

inf1uence on firm performance. Figure 3 shows the positive significant correlation

124 How Does Corporate Goνernance AjJect Firm Performance? The Mediating Ro/e

0/

Agency Costs

6.21). Aside from the chi-square value, which is easily influenced by the number of samples, Table 7 shows that the comparative fit index values are close to standards, and the other indices have all achieved accepted standards, which indicates that c。中orate govemance has a beneficial function.

Figure 2

The Direct Effect of Corporate Governance Influence on Agency Costs

1. 00

仁王于0

.

92

~叮叮亡至于。 79

/ ' - - - " -0.31'{王抖。

l 之至多只;:

C

Erog9

一Ò.88 ì 師戶。.BB

mE:

..., ..曹 軒唔'0

可主弘 H

o2

;

:淘

0.91

-1

AQE

1/

Figure 3

The Direct Effect of Corporate Governance Influence on Firm Performance

:::耳

了

:;母

斗吋三己/

~

DlC

f-

O_93

A4毛主:::

ζ王]刮目 1. 00 。。。 o az nunu --SE-'.--n= 圖-個詞自 ooo <

@

89 00 。

ChiaoDa A位magement Re叫的vVol. 30 No. 2, 2010 125

Table 6

Evaluation Chart for the Direct Effect of Corporate Governance influence on Agency Costs

Overall model fit absolute fit index

Overall model fit - comparative fit index

Overall model fit parsimonious fit index

Evaluation Content

Chi-square value is insignificant, indicating a good model fit

GFI ~ 0.90 AGFI 這 0.90 SRMR < 0.05 (excellent fit); 。.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); > 0.10 (poor fit)

RMSEA < 0.05 (excellent fit);

0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); > 0.10 (poor fit) NFI ~三 0.90 NNFI ~三 0.90 IFI 這 0.90 CFI ;三 0.90 RFI 這 0.90 PNFI ~三 0.5 PGFI ;三 0.5 CN>200

4.3.

The Mediating Effect of Agency Costs

Results 845.84 (P = 0.00) 0.97 0.96 0.045 0.054 0.83 。 78 0.84 0.84 0.77 0.63 0.62 444.34

Figure 4 shows the mediating effect of agency costs on the influence of corporate govemance on firm performance. The figure indicates that with the inc1usion of the mediating effect of agency costs, the influence of corporate

126 How Does Corporale Goνernance Affecl Firm Peiformance? The Medialing Role 01 Agency Cosls

govemance on agency costs and of agency costs on finn perfonnance have shown

negative significant correlations (with coe伍的nts of -0.37 and -0.77 and t-values

of -2.91 and -2.82, respectively).

Table 7

Evaluation Chart for the Direct Effect of Corporate Governance Influence

on Firm Performance Overall model fit

- absolute fit index

Overall model fit

- comparative fit

index

Overall model fit - parsimonious fit index

Evaluation Content

Chi-square value is insignificant, indicating

a good model fit

GFI ~ 0.90 AGFI 這 0.90 SRMR < 0.05 (excellent fit); 0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); > 0.10 (poor fit)

RMSEA < 0.05 (excellent fit);

0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); > 0.10 (poor fit) NFI ~ 0.90 NNFI ~三 0.90 IFI ~ 0.90 CFI 益。.90 RFI ~ 0.90 PNFI ~三 0.5 PGFI 三 0.5 CN > 200 Results 482.00 (P = 0.00) 0.98 0.97 0.032 0.042 0.87 0.85 0.89 0.89 。 .83 0.66 。 63 779.56

ChiaoDaManαgement Review 均1. 30 No. 2, 2010 127

Figure 4

The Mediating Effect of Agency Costs on the Influence of Corporate Governance on Firm Performance

iOOZ;:MSR

o 呵呵 AQD 1 .... \0.18

O9fZ/f

:::Ztj;也1.00

i 叫二~/一 仁王]-09虹2

θ

必〈」唔

i

:k〈至~H_COV詗誌川川古古古恣::3::正;正主芷立08;

:

0叫阻 、仁王]-唾0.25

冷E函砂h吐1詰詰;:3;i;;置

;zE主主

E

_:::

_::

_:

、仁王]--0033

Table 8 is the evaluation chart for the mediating effect of agency costs on the influence of c。中orate governance on firm performance. The model fit indicates that aside from the chi-square value significance, which is easily influenced by the number of samples, the comparative fit index is below standards, while the other fit indices have all reached acceptable standards.

Table 9 shows the direct effect and the mediating effect among corporate governance, agency costs, and firm performance. The chart indicates that after inc1uding the mediating effect of agency costs, corporate governance still has a positive significant correlation with firm performance (coefficient -0.3 7* (-0.77) = 0.2849, t-value = 6.95) as compared to the mediating effect of uncontrolled agency costs (coefficient = 0.10) on firm performance.

Since the direct effect of c。中orate governance on firm performance is not significant, agency costs have only a partial mediating effect on the influence of corporate governance on firm performance. The above results indicate that corporate governance can increase performance through direct positive influence on firm performance and through decreasing agency costs. The results show that Taiwanese firms' corporate governance has both beneficial and maladministration-mitigating functions. These provide evidence that in

128 How Does Corporale Goνernance Affecl Firm Performance?

The Mediating Ro/e of Agency Costs

investigating the influence of corporate governance on firrn perforrnance, the mediating role of agency costs must be included. This is consistent with the hypothesis that this study has proposed.

Table 8

Evaluation Chart for the Mediating Effect of Agency Costs on the Influence

。fCorporate Governance on Firm Performance

Overall model fit - absolute fit index

Overall model fit

-comparative fit index

Overall model fit

- parsimonious fit index

Evaluation Content Chi-square value is insignificant,

indicating a good model fit

GFI <三 0.90 AGFI ~ 0.90 SRMR < 0.05 (excellent fit); 0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); >O .l O (p∞r fit)

RMSEA<O肘 (excellent fit);

0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); > 0.10 (poor fit) NFI ~ 0.90 NNFI <三 0.90 IFI 這 0.90 CFI 這 0.90 RFI ~ 0.90 PNFI ~ 0.5 PGFI <三 0.5 CN > 200 Results 990.61 (P =0.00) 0.97 0.96 0.040 0.047 0.83 0.81 0.84 0.84 0.79 0.67 0.69 589.46

Chiao Da Managemenl ReνiewVol. 30 No. 2, 2010 129

4.4.

Equity Structure

,

Composition of Directors and Supervisor

,

and Related Party Transactions

In this sωdy， corporate govemance is made up of three latent variables: ownership structure, composition of directors and supervisors, and related par句F

transactions. Since previous studies found both positive and negative effects among these three variables, we investigated their effects on agency costs separately. The purpose of this is to avoid the offsetting of positive and negative effects from each latent variable, since ownership structure, composition of directors and supervisors, and related party trading have different effects on corporate govemance.

Figure 5 shows the mediating effect of agency costs on the influence of the composition of directors and supervisors on firm performance. After control1ing for the effect of agency cost, the composition of directors and supervisors has increased firm performance (coefficient -0.5*(-0.79)

=

0.395, t-value

=

6.85). Table 10 is the evaluation chart for the mediating effect of agency costs on the inf1uence of the composition of directors and supervisors on firm performance.

The resu1ts show that aside from the easily inf1uenced chi-square values, two comparative fit measures are below standards, while the other fit measures showed an acceptable fit.

Figure 6 shows the mediating effect of agency costs on the inf1uence of equity structure on firm performance, after controlling for the e宜ect of agency costs, ownership structure has helped increase firm performance (coe伍的nt -0.24 * (一0.78) = 0.1872, t-value = 5.07). Table 11 is the evaluation chart for the

mediating effect of agency costs on the inf1uence of equity structure on firm performance. The results show that aside from the easily influenced chi-square value, the comparative fit index measures are all below standards, while the other fit measures show acceptable standard values.

130 Ho帥wD卸oe臼s c.， αor，伊po仰rat仿eGωOν附e凹r鬥阱'n叩lceAj，加伶伽ct F.扔irmPe吃柚治 rma仰n悶Cαe?

The Mediating Ro/e 01 Agency Costs

Table 9

The Direct and Mediating Effects among Corporate Governance

,

Agency Costs

,

and Firm Performance

Direct Effect t-value

Corporate govemance on firm performance 0.10 6.21 *** Corporate govemance on agency costs -0.68 -3.55*** Corporate govemance on agency costs -0.37 -2.91***

Agency costs on firm

performance -0.77 -2.82***

Mediating Effectl t-valuez

0.2849 6.95***

Mediating effect is calcu1ated as (-0.37 * (-0.77) = 0.2849)

* * * Indicates that a 1 % significance is achieved

Figure 5

The Mediating Effect of Agency Costs on the Influence of the Composition of Directors and Supervisors on Firm Performance

.

8

…

0821

"

DRC

I \

0.

92

1

DRD

1_

0

0811DR

cJ

'

l /

有且這

。〈。

可

相…也

Chiao Da Management ReviewVo/. 30 No. 2, 2010 131 Table 10

Evaluation Chart for the Mediating Effect of Agency Costs on the Influence

。fthe Composition of Directorsand Supervisors on Firm Performance

Overal1 model fit

- absolute fit index

Overal1 model fit

一comparative fit index

Overal1 model fit

- parsimonious fit index

Evaluation content Chi-square value is

insignificant ,indicating a good model fit

GFI ~三 0.90 AGFI 這 0.90

SRMR < 0.05 (excel1ent fit);

0.05 - 0.08 (good fit);

0.08 - 0.10 (moderate fit);

> 0 .10 (poor fit)

RMSEA < 0.05 (excel1ent fit);

0.05 - 0.08 (good fit); 0.08 - 0.1 0 (moderate fit); > 0.10 (poor fit) NFI ~三 0.90 NNFI ~ 0.90 IFI ~三 0.90 CFI 去。 90 RFI 三 0.90 PNFI 益。 5 PGFI ~ 0.5 CN>200 Results 159.49(P = 0.00) 0.99 。.99 0.026 0.033 0.91 0.89 0.93 0.92 0.87 。.63 。 55 1380.63

132 How Does Corporate Goνernance AjJect Firm Pe柚rmance?

The Mediating Role 01 Agency Costs

Figure 6

The Mediating Effect of Agency Costs on the Influence of Equity Structure on Firm Performance

#'

1

AQA

1-

1.. 00 0.93 -←-戶。 2

司EE

油田

毛。可

尹@

0.05 O.zs

/

Chi-Square=289.45, df=25, p-value= 口 .000 口 0 ， RMSEA= 口.口 46

Figure 7 shows the mediating effect of agency costs on the influence of related party transactions on firm performance. Related party transactions has a positive but insignificant relationship with agency costs (coefficient = 0.01, t-value = 0.61); thus

,

agency costs has no mediating effect on the influence of related party transactions on firm performance. Table 12 is the evaluation chart for the mediating effect of agency costs on the influence of related party transactions on firm performance. The results show that aside from the easi1y influenced chi-square value, three comparative fit measures are below standards, whi1e the other fit measures were able to reach an acceptable standard fit.

The above results indicate that the composition of directors and supervisors has the highest coefficient (0.395), higher than the overall corporate govemance coefficient value (0.2849), whi1e the ownership structure coefficient is a bit low (0.1872), but is still positive and significant. Although we find that related party transactions through agency costs has a negative effect on firm performance, the effect is not significant; thus, the mediated relationship among the three variables does not exist Nevertheless, we find that related party transactions will reduce firm performance and are positively correlated with agency costs. From the results of this study, we find that both beneficial and maladministration-mitigating

Chiao Da Management ReviewVol. 30 No. 2, 2010 \33

functions of corporate govemance common!y exist for Taiwanese firms, and that there is no !arge counter-effect in the measurement of corporate govemance.

Tab!e 11

Eva!uation Chart for the Mediating Effect of Agency Costs on the Influence of Equity Structure on Firm Performance

Overall model fit

一absolute fit index

Overall model fit comparative fit index

Overall model fit

- parsimonious fit index

Evaluation content

Chi-square value is insignificant, indicating a good model fit

GFI ~三 0.90 AGFI 這 0.90 SRMR < 0.05 (excellent fit); 0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); > 0 .10 (poor fit)

RMSEA < 0.05 (excellent fit); 0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit) > 0.1 0 (poor fit) NFI 這 0.90 NNFI 這 0.90 IFI 這 0.90 CFI 2: 0.90 RFI 這 0.90 PNFI 2: 0.5 PGFI 這 0.5 CN > 200

4.5.

Sensitivity Analysis ResuIts 289.45(P = 0.00) 0.99 0.98 0.035 0.046 0.85 0.80 。.86 0.86 0.78 0.59 0.55 748.09

This study has investigated the direct effects of corporate govemance on firm performance and of corporate govemance on agency costs, as well as the

Chiao Da Management Revi凹v~包1. 30No. 2, 2010 135

Table 12

Evaluation Chart for the Mediating effect of Agency Costs on the Influence of Related Party Transactions on Firm Performance

Overall model fit - absolute fit index

Overall model fit - comparative fit index

Overall model fit

- parsimonious fit index

Evaluati叫 ncontent

Chi-square value is insignificant indicating a good model fit

GFI 這 0.90 AGFI 這 0.90 SRMR < 0.05 (excellent fit); 0.05 - 0.08 (good fit); 0.08 - 0.10 (moderate fit); > 0 .10 (poor fit)

RMSEA < 0.05 (excelle叫tfit); 0.05 - 0.08 (good fit); 0.08 - 0.10 (moderat單位t); 〉0.1。他∞rfit) NFI 逞。.90 NNFl 這 0.90 IFI ~ 0.90 CFI ~三 0.90 RFI ~ 0.90 PNFI ~ 0.5 PGFI ~三 0.5 CN>200 Results 175.13(P = 0.00) 0.99 0.99 0.028 0.035 0.88 0.85 0.90 0.90 0.83 0.61 0.55 1233.86

correlation (coefficient

=

O.紗， t-value

=

1.23).5 The corporate govemance of nor卜electronics firms can also increase firm performance (coefficient O.肘， t-value = 2.96); however, the coefficient is smaller as compared to that of the

5 When the coefficient for the influence of corporate governance on firm performance and the

coefficient for the influence of corporate governance on agency costs are significant at the same time, the mediating effect of agency costs on the influence of corporate governance on firm performance is established. Therefore, a mediating e能cttest need not be carried out.

136 How Does Corporate Governance A，加ctFirm Pe祕rmance? The Mediating Role 01 Agency Costs

electronics industry, with a coefficient ofO.28. The e釘ect of corporate govemance on agency costs cannot be contained, which implies that firrns within the electronics industry have better corporate govemance than non-electronics firrns. Between 2000 and 2002, the samples show that corporate govemance and firrn perforrnance have a positive but insignificant correlation (correlation = 0.28, t-value = 1.91), and the effect of corporate govemance on agency costs cannot be contained. After 2002, results indicate that there is a positive and significant correlation (coefficient = 0.26, t-value = 2.37) between corporate govemance and firrn perforrnance, corporate govemance has a negative and significant correlation with agency costs (coefficient = -0.67, t-value = -3.56), and the mediating effect of agency costs on the influence of corporate govemance on firrn perforrnance signifies that c。中orate govemance can first weaken agency costs and then improve firrn perforrnance (-0.34

*

(-0.80) = 0.272). These results indicate that between 2003 and 2006, the effective role of corporate govemance came 臼lly into play, thereby improving both the beneficial and maladministration-mitigating functions of corporate govemance

5. Conclusions

This study has investigated the beneficial and maladministration-mitigating roles of corporate govemance. We have attempted to include the mediating effect of agency costs in our analysis of the effect of c。中orate govemance on firrn perforrnance. First, we tested for the direct effect of c。中orate govemance on firrn perforrnance (beneficial role) and the direct effect of corporate govemance on agency costs (maladministration-mitigating role). Then, we set out to prove the mediating effect of agency costs in the influence of corporate govemance on firrn perforrnance and examined whether these two functions can together improve firrn perforrnance.

In the examination of direct efiì闊的， the empirical results show that corporate govemance and firrn perforrnance are positively and significantly correlated, whereas corporate govemance and agency costs are negatively and significantly correlated. The results for the mediating role of agency costs indicate

Chiao Da Managemenl Review 均 1. 30 No. 2, 2010 137

that corporate governance can first control for agency costs in a firm and then help improve firm performance. This is consistent with the expectations of our study.

With the inc1usion of agency costs as a mediating variable, corporate governance can control for agency costs and can increase firm performance. This indicates that reducing agency costs is an important way for the corporate governance of Taiwanese firms to increase firm performance, thus supporting the hypothesis proposed in this study. Due to the inc1usion of agency costs as a mediating variable, the inf1uence of corporate governance on agency costs and the inf1uence of agency costs on firm performance are significant. Therefore, agency costs only has a partial mediating effect in the inf1uence of c。中orate goveIτlance

on firm performance, implying that the corporate governance of Taiwanese firms has both beneficial and maladministration-mitigating functions.

The main contribution of this paper lies in c1arifying the ambiguity concerning the relationship between corporate governance and firm performance; for this purpose, agency costs are inc1uded as a mediator and the LISREL analysis was used. We believe that when the mediating effect is large, it can easily lead to an inability to determine whether the functions of corporate governance are effectively manifested or not. Only with the inc1usion of testing and the determination of a mediating variable wil1 the relationships between variables be apparent. By relying on a mediating variable, the beneficial and maladministration-mitigating functions of corporate governance can increase the strength of firm performance.

The m句 or limitation of this study is that LISREL is unable to analyze discontinuous variables; hence, we cannot set dummy variables in order to view a c1earer relationship between the variables in the study. Although we segregated samples before proceeding with the analysis, we could not test for al1 samples and control for industry and year effects at the same time. Future research could move further in this direction and find a framework for further analysis.

138 How Does Corporate Goνernance Affect Firm PerfiJrmance? The Mediating Ro/e 01 Agency Costs

6. References

Abarbanell, 1. S. and Bush間， B. 1. (1998),“Abnorrnal Retums to A Fundamental Analysis Strategy," Journal

01

Accounting Research, 73(1), 19-45.

Agrawal, A. and Knoeber, C. R. (1996),“Firrn Perforrnance and Mechanism to Control Agency Problems between Managers and Shareholders," Journal

01

Financial and Quantitative Analysis, 31(3), 377-397.

Ang, J. S., Cole, R. A., and Lin, 1. W. (2000),“Agency Costs and Ownership Structure," Journal

01

Finance, 55(1), 81-106.

Balatbat, M., Taylor, S., and Walter, T. (ο20∞04你)，

Ownershiψp and Operating Perforrnance of Australian lnitial Offerings,"

Accounting & Finance, 44(3), 299-329.

Baron, R. M. and Kenny, D. A. (1986), “The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic and Statistical Consideration丸" Journal

01

Personality and Social Psychology,

51(6), 1173-1182.

Bent1er, P. M. (1989), EQS Structural Equations Program Manual, Los Angeles,

CA:BMDP.

Bentler, P. M. (1990)λ，

Psychological Bulletin, 107(2),238-246.

Bent1er, P. M. and Bonett, D. G. (1980),“Significance Tests and Goodness of Fit in the Analysis of Covariance Strucωres ，" Psychological Bulletin, 88(3),

588-606.

Ber峙， A. and Means, G. (1932), The Modern C。中oration and Private Property,

New York, NY: MacMillan.

Bhagat, S. and Black, B. (2002),‘'The Non-correlation between Board

lndependence and Long-terrn Firrn PerfoIτnance，" Journal

01

Corporation Law, 27(2), 231-274.

Bhagat, S. and Jefferris, R. (2002), The Econometrics

01

Corporate Governance Studies, Boston，扎1A: MIT Press.

Black, B. S., Jang，旺， Kim, w., and Mark, 1. (2006), “Does Corporate

Chiao Da Management Reνlew 均 1. 30 No. 2, 2010 139

Economics and Organization, 22(2), 366-413.

Bollen, K. A. (1989),“Sample Size and Bentler and Bonetl's Nonnormed Fit Index," Psychometrika, 51 (3), 375-377.

Cadbury, A. (1999),“What are the Trends in Corporate Govemance? How Will They Impact Your Company?" Long Range Planning, 32(1), 12-19.

Chen, L. C. (2004), “The Empirical Analysis of Relationship between Composition of Board of Directors, Ownership Structure, the Related Party Transaction and Firm Performance - Evidence in Textile Industry and Electronics Industry Listed on TSE," Master's thesis, National Cheng Kung University

Chen, R. S. (ρ2004) ，

Shareholders Prot臼ectiωon，" Do巳tωorat臼e t由he臼SI凶s， National Chung Cheng University.

Chiang, C. S. (2004),“The Impacts of Business Strategies and Compensation Strategy on Organizational Performance," Master's thesis, Soochow University.

Chou, C. P., Bentler, P. M., and Satorra, A. (1991),“Scaled Test Statistics and Robust Standard Errors for Non-normal Data in Covariance Structure Analysis: A Monte Carlo Study," British Journal on Mathematical and Statistical Psychology, 44(2), 347-357.

Conyon, M. 1. and Peck, S. 1. (1998),“Board Size and Corporate Performance: Evidence from European Countries," European Journal of Finance, 4(3),

291-304.

Cubbin, J. and Leech, D. (1983),“The Effect of Shareholding Dispersion on the Degree of Control in British Companies: Theory and Measurement," The Economic Journal, 93(4), 35 卜369.

Demse紹， H. and Villalonga, B. (2001), “Ownership Structure and Corporate Performance," Journal ofCorporate Finance, 7(3), 209-233.

Dobrzynski, J. H. (1993),

Business Week， 日， 26.

Eisenberg, T., Sundgren, S., and Wells, M. T. (1998),“Larger Board Size and Decreasing Firm Value in Small Firms," Journal of Financial Economics,

140 How Does Corporate Governance Affect Firm Performance? The Mediating Role of Agency Costs

48(1),35-54.

Fama, E. (1980),“Agency Problems and the Theory of the Finn," Journal 01 Political Economics, 88(2), 288-308.

Fama, E. and Jensen, M. C. (1983), “Separation of Ownership and Control,"

Journal 01 Law & Economics, 26(2), 301-325.

Gillies, 1. and Leblanc, R. (2003), “The Coming Revolution in Corporate

Governance," Ivey Business Journal， 的(1)， 1-1 1. Gompers, P., Ishii, 1., and Metrick, A. (ο20ω03)，

P

昕削n必ce忱s丸，" The Quω4ωαr斤terz秒yJournal 01 Ecωonoωom閒?η衍1IC.臼.s， 118(1), 107-156.

Hoelter, 1. W. (1983),“The Analysis of Covariance Structures: Goodness-of-Fit

Indices," Sociological Method and Research, 11(3), 325-344.

恥， L. T. and Bentler, P. M. (1995), Evaluating Model Fit, In R且 Hoyle (Ed.),

Structural Equation Modeling, Thousand Oaks, CA: Sage.

James, L. R., Mulaik, S. A., and Brett, 1. M. (1982), Casual Analysis:

Assumptions, Models, and Data, Beverly Hills, CA: Sage.

Jensen, M. C. (1986),“Agency Costs of Free Cash Flow, Corporate Finance and

Takeovers," American Economic Review, 76(2), 323-329.

Jens凹， M. C. and Meckling, W. H. (1976),“Theory of the Finn: Managerial

Behavior, Agency Costs, and Ownership Structure," Journal 01 Financial

Economics, 3(4),305-360.

Jensen, M. C. and Ruback, R. (1983), “The Market of Corporate Control,"

Journal 01 Financial Econom帥， 11(1), 5-50.

Joresk嗯， K. G. and Sorbom, D. (1993), Structural Equation Modeling with the SIMPLlS Command Language, Chicago: Scientific Software International.

Kim, B. and Lee, 1. (2003), “Agency Problems and Perfonnance of Korean

Companies during the Asian Financial Crisis: Chaebol vs. Non-chaebol

Finns," Pacific-Basin Finance Journal, 11(3),327-348.

La Porta, R., Lopez-de-Silanes, F., and Shleifer, A. (1999),“Corporate Ownership

around the World," Journal 01 Finance, 54(2),471-517.

Marsh, H. W. and Hocevar, D. (1985),“The Application of Confinnatory Factor

Analysis to the Study of Self-concept: First and Higher Order Factor

Chiao Da Management Review 均l. 30 No. 2, 2010 141

97(3), 562-582

Marsh, H. w., Balla, J. R., and McDona1d, R. P. (1988),“Goodness-of-Fit lndexes

in Confirrnatory Factor Ana1ysis: The Effect of Samp1e Size,"

Psychological Bulletin, 103(3),391-410.

Mehran, H. (1995), “Executive Compensation Structure, Ownership, and Firrn Perforrnance," Journal of Financial Economics, 38(2), 163-184.

Micc凹， T. (μ198紗9)，

Cr昀eatωur昀es丸，" Psychologicα1 Bulletin爪1丸， 105(1η)， 156-165.

Morck R., Sh1eifer, A., and Vishny, R. W. (1988),“Management Ownership and Market Va1uation: An Empirica1 Ana1ysis,'\Journal of Financial Economics,

20(1),293-315.

Myers, S. C. (2001),“Capita1 Structure," Journal ofEconomic Perspectives, 15(2),

81-102

Pearce, 1. A. and Zah悶， S. A. (1992), “Board Composition from a Strategic Contingency Perspective," Journal of Management Studies, 29(4),53-58. Shiue, M. J., Huang, C. 1., and Chang, Y. S. (2007),“Corporate Govemance and

Firrn Perforrnance: The Mediating R01e of Earnings Management," 2007 Annua1 Conference ofthe Taiwan Finance Association.

Singh, M. and Davidson III, W. N. (2003),“Agency Cost, Ownership Structure and Corporate Governance Mechanism," Journal of Banking and Finance,

27(5), 793-816.

Steig仗" J. H. (1990),“Structura1 Model Eva1uation and Modification: An Interva1

Estimation Approach," Multivariate Behavioral Research, 25(2), 173-180 Steiner, T. L. (1996),“A Reexamination of the Re1ationships between Ownership

Structure, Firrn Diversification, and Tobin's 眩， Quarterly Journal of Business and Economics, 35(4), 39-48.

Warfie1d, T.缸， Wi1d, J. 1., and Wi1d, K. L. (1995), “Manageria1 Ownership,

Accounting Choice, and Inforrnativeness of Earnings," Journal of Accounting & Economics, 20(1), 61-91.

Wu, M. Y. (ρ200側4) ，

Management 0 宜伍lce叮rs丸， Research

&

Deve10pment Expenditure, Ownership Structure, and the Corporate Va1ue," Master's thesis, Tamkang University.

142 How Does Corporate Governance Affect Firm Performance?

The Mediating Ro/e of Agency Costs

Yafeh, Y. and Y os恤， O. (2003),“Large Shareholders and Banks: Who Monitors and How?" The Economic Journal, 113(484), 128-146.

Yeh, Y. H., Lee, T. S., and Ko, C. E. (2002a), Corporate Governance and Rating System, Taipei, Taiwan: Sunbrieht Publishing Co. (in Chinese).

Yeh, Y.旺， Lee, T. S., and Ko, C. E. (ο2002b) ，

Govemance on the Related-party Transaction," 2002 Conference ofTa剖lwanl Finance Association.

Yen, C. c., Chao, C. L., and Chang, Y. S. (2006),“An Empirical Investigation of the Association among Corporate Govemance, Eamings Management, and Business Performance," Journal

0/

Kaohsiung University

0/

Applied Sciences, 35(3), 281-308.

Yermack, D. (1996),“Higher Market Valuation for Firms with a Small Board of Directors," Journal

0/

Financial Economics, 40(2), 185-211.