Do Multiple Directorships Signal Quality? Evidence from

Transcription

1 Do Multiple Directorships Signal Quality? Evidence from Independent Directors Adrian C.H. Lei Jie Deng University of Macau 1

2 Do Multiple Directorships Signal Quality? Evidence from Independent Directors Abstract This paper studies the effect of independent directors' multiple directorships on firm value and examines the countervailing effects of quality and "busyness. Using a unique panel data set covering all Hong Kong-listed firms from 2001 to 2009, we find a strong and positive relation between the number of multiple appointments of independent directors and firm value, consistent with the quality hypothesis. The directors quality comes with a busyness tradeoff, however, as we find a non-linear relationship in various empirical settings. Our results are robust to a range of estimation procedures, including alternative multiple directorships measurements and instrumental variables. Overall, our empirical evidence supports increasing the minimum requirement of independent directors, even under a supply constraint of qualified directors. Keywords: Multiple board appointments, independent directors, busyness, quality JEL classifications: G30, G34, K22 2

3 1. Introduction Corporate-governance codes around the world commonly advocate board independence. 1 One method to enhance board independence is to increase the number or proportion of independent directors on the board of directors. Countries that originally had low requirements for the number of independent directors are trying to advance to international standards. 2 Yet, apparent supply constraints in the market of independent directors may significantly increase the number of multiple board appointments, or so-called overlapping independent directors. A recent consultation paper of the Hong Kong Stock Exchange (HKEx), under the section Directors Duties and Time Commitments, states, We note market concerns that some directors, particularly independent non-executive directors (INEDs), may have taken on too many directorships. This may compromise their ability to devote sufficient time and energy to their duties, as well as currently, there is no Code recommendation for a review of directors time commitments. 3 The focus now is on multiple directorships of independent directors, and yet few studies examine this issue in emerging markets, where authorities are pressured to increase the minimum 1 The New York Stock Exchange (NYSE) requires issuers to have a majority of independent directors, see NYSE Corporate Governance Standards. The London Stock Exchange and the Australian Stock Exchange require issuers to comply or explain whether INEDs form a majority of the board (in the UK, smaller companies are an exception). In the Mainland China stock exchanges, an issuer s board must be made up of at least one-third independent directors, while the Singapore Stock Exchange requires firms to comply or explain whether one-third of an issuer s board are INEDs. 2 In 2003, for example, Hong Kong increased the minimum requirement for the number of independent directors of listed companies from two to three. Recently, the HKEx launched a formal consultation to further increase the minimum requirement. In 2001, the China Securities Regulatory Commission established the independent director institution in listed companies with a minimum of two independent directors before June 30, 2002, and required at least one third of directors to be independent before June 30, After the Asian Financial Crisis in 1997, the Korea stock exchange required at least one-quarter of the board of all listed companies be independent directors; then in 2001, the minimum requirement for companies with a book value of assets exceeding 2 trillion won (about $2 billion) was raised to 1/2. 3 Source: On review of the code on corporate governance practices and associated listing rules. Consultation paper, Hong Kong Stock Exchange, December

4 requirement of the fraction of independent directors while limiting the number of multiple board appointments of these directors. There are abundant yet contradictory opinions on multiple directorships. Fama and Jensen (1983) suggest that multiple board appointments can better signal directors quality, 4 thus setting up the quality hypothesis. This hypothesis takes the view that directors with multiple appointments are of high quality and may have rich experience (Carpenter and Westphal, 2001; Perry and Peyer, 2005) and good reputations (Ferris et al., 2003; Sarkar and Sarkar, 2009). If the investors assess higher value for the firms with better quality directors, multiple directorships may increase firm value. From another perspective, Ferris et al. (2003) argue that directors with multiple board appointments may be incapable of effectively monitoring all the firms they represent, thus suggesting the busyness hypothesis. According to agency theory, the busyness hypothesis implies that a director's multiple appointments decrease firm value because there is less oversight of management, thereby increasing the agency cost, for example, by diluting board members ability (Jiraporn et al., 2009), deepening the diversification discount (Jiraporn et al., 2008), excessively compensating CEOs (Core et al., 1999), and making it easier to commit financial statement fraud (Beasley, 1996). Substantiating this hypothesis, Fich and Schivdasani (2006) find that busy boards are negatively related to firm value. One important criterion that firms use to decide about appointing an independent director is whether the appointment would add value to the firm itself. Yet, the overall effects of busyness and quality on firm value are unclear. We therefore attempt to fill this gap by analyzing independent directors' multiple board appointments. Most existing studies about multiple appointments focus on all directors (Ferris et al., 2003; Jiraporn et al., 2009; Loderer and Peyer, 2002) or inside directors 4 Following Fama and Jensen (1983), Shivdasani (1993), Cotter et al. (1997), and Vafeas (1999) use number of directorships as a proxy for independent director s reputation capital and argue that an independent director with more directorships will be a better monitor. 4

5 (Conyon and Read, 2006; Perry and Peyer, 2005). There are fundamental differences among inside directors, outside directors, and independent directors. In particular, independent directors are outside directors who have no affiliation with the firm other than being on the board. According to the Sarbanes Oxley Act of 2002 (SOX), independent directors may not accept any consulting, advisory, or other compensatory fee from the issuer or be an affiliated person of the issuer or any subsidiary thereof. Also, independent directors should not have material interests in the form of shares of the company on whose board they serve. 5 Ideally, independent directors should be uninterested and indifferent so as to perform their monitoring role and protect minority shareholders rights. Despite their lack of incentives, they play a significant role in modern corporate governance (Fama, 1980; Fama and Jensen, 1983). The environment for independent directors in emerging markets is very different compared to that of developed markets in the aspects of size, quality, and regulations. In the US, the Council of Institutional Investors recommends no more than three directorships per director. 6 Emerging markets, facing a supply constraint of qualified independent directors, allow firms to appoint independent directors with loose restrictions on the number of multiple appointments. 7 So far, few studies investigate the consequence of this important matter in emerging markets. 5 For example, Hong Kong Main Board Listing Rules Chapter 3, Authorized Representative and Directors, suggests less than 1% of total shareholdings of the listed company. Similar regulations apply for Mainland China listed companies. In the US, no explicit percentage is stated, although no material interests applies for Securities and Exchange Commission (SEC) regulations. 6 See Core Policies, General Principles, Positions, and Explanatory Notes, Council of Institutional Investors (CII), Washington, DC, In this file, CII recommends, Absent unusual, specified circumstances, directors with full-time jobs should not serve on more than two other boards. 7 In mainland China, an independent director can take up at most 5 independent director positions in different listed companies. In India, Malaysia, and Pakistan, the limit is much higher, ranging from 10 to 20 (Sarkar and Sarkar, 2009). Currently, there are not any restrictions on the number of appointments by INEDs nor any regulations about directors' time commitment in Hong Kong. 5

6 In a related study, Sarkar and Sarkar (2009) analyze the directors of Indian firms and find that the overall effects of independent directors' multiple appointments on firm value are insignificant. In addition, they show that if firm median directorships held by independent directors are five or more, there is a positive relationship between multiple directorships and firm value. While the majority of independent directors have multiple directorships, most directors have fewer than 5 simultaneous appointments (about 60% for India 8 and 84% for HK). One potential reason they cannot find strong results is that they use a single-year dataset, which is prone to misspecification because of the heterogeneity of firms, and thus they can identify only cross-sectional variations. Furthermore, the effect of multiple directorships on firm value is unlikely to be strictly exogenous. We conjecture that there should be more concrete relationships between multiple directorships and firm value. From 2003, the minimum number of INEDs 9 of Hong Kong-listed companies required by the Listing Rules of the HKEx increases from two to three in order to improve corporate-governance practices, and the HKEx recommends that INEDs form one-third of the board. 10 This recommendation is currently under consultation 11 to determine whether a mandatory requirement is necessary. According to our data, in 2009, more than two-thirds of companies have appointed INEDs with multiple appointments. This relatively high percentage creates a unique opportunity to investigate the issue. Changes to the HKEx listing rules also provide a natural experiment to address the endogeneity of independent directors' multiple appointments. 8 From Sarkar and Sarkar (2009), Table 3. 9 We refer to INED, which is the abbreviation for Independent Non-Executive Director in Hong Kong, as independent director, and these terms are used interchangeably in this paper. 10 Please refer to the consultation conclusion of Proposed Amendments to the Listing Rules Relating to Corporate Governance Issues, HKEx, January Please refer to the consultation paper on Review of the code on Corporate Governance Practices and Associated Listing Rules, HKEx, December

7 The Hong Kong market possesses characteristics similar to those of many emerging markets in various aspects of corporate governance, such as a limited shareholder activism culture and a highly concentrated and family-dominated ownership structure, making Hong Kong an essential representation of emerging markets (Cheung et al., 2007). As an international financial market, Hong Kong has a well-developed market infrastructure and a stable political and legal environment (Porta et al., 1998), and its local accounting standards are harmonized with international standards (Standard & Poor's, 2002). These characteristics ensure that we can access data with high comparability and limited noise. Therefore, Hong Kong provides an ideal testing ground for our study. We find that the number of directorships held by INEDs is positively related to firm value and is statistically and economically significant. A one-standard-deviation increase in the median number of directorships is associated with increases in Tobin s q of (7.6% increase) and MTBV of (8.1% increase). This suggests that INEDs with multiple appointments signal higher quality to the market, which corroborates the quality hypothesis. The signaling effect reduces when INEDs have more directorships, suggesting a non-linear relationship and showing a busyness tradeoff. We find that when the firm median number of directorships is higher than 4, the positive relationships start to dampen. Similarly, in the INED-level regression, the number of appointments is associated with very strong positive effects on firm value, with receding effects similar to the firm-level results. To address endogeneity issues, we apply instrumental variable regressions, and the results are consistent with the fixed-effect results. This paper contributes to the board independence and multiple board appointments literature. There are currently very few related studies in emerging markets, where multiple board appointments of independent directors raise major concerns. Using our unique dataset that spans from 2001 to 2009, we show that director s quality comes with a slight busyness tradeoff. Nevertheless, quality seems to be dominant overall from investors point of view. We also address the 7

8 endogeneity problem using several types of instrumental variables. Our empirical evidences support increasing the minimum requirement for INEDs and suggest that allowing existing independent directors to hold more directorships can be an acceptable solution for the problem of supply constraints in the independent director market. The remainder of this paper is organized as follows. Section 2 describes our data and methodology. We discuss the empirical results in Section 3 and present the robustness tests results in Section 4. Section 5 concludes. 2. Data and Methodology 2.1 Data Source and Sample Description Our dataset consists of all non-financial companies listed on the Mainboard of Hong Kong from 2001 to We retain all observations that have no missing data (firm value, information about INEDs with multiple appointments, and corporate governance data) and generate an unbalanced panel that consists of 10,664 firm-year observations and 611 firms. 12 We develop the unique dataset of INEDs with multiple appointments based on the INED database of Hong Kong-listed companies compiled by Webb-site.com 13 and listed company annual reports. We record the number of INEDs with multiple appointments for each firm in each year, the number of multiple appointments held by each INED in each year, and also the specific firms in which they are simultaneously appointed as INEDs in each year. The multiple appointments are therefore only associated with INED appointments of other listed companies. This enables us to objectively identify multiple 12 Our dataset also consists of 2,953 INED-year observations (for INEDs with multiple directorships only). 13 Webb-site.com provides independent commentaries and a database on corporate and economic governance, business, finance, investments, and regulatory affairs in Hong Kong, run by Mr. David Webb, who is the independent director of Hong Kong Stock Exchange from We also validate the data from their dataset using listed company annual reports. 8

9 appointments, because the job nature of non-listed appointments can be extremely diverse. Accounting and market data are downloaded from Datastream. Corporate-governance data are collected manually from annual reports, and because of the vast amount of data collected, our data cover year 2001, 2002, 2003, 2005, 2007, and We use the latest available data to replace the missing years, i.e., 2004 values are from 2003 values, 2006 values are from 2005 values, etc. As corporate-governance variables usually persist over time, we believe that this is a reasonable approximation and that it should have little impact on the OLS and fixed-effects estimations. Table 1 Panels A and B summarize the frequencies of multiple appointments of INEDs for both firm- and INED-level data in our dataset. [Insert Table 1 Here] From Panel A, we notice that the number of INEDs with multiple appointments in Hong Kong-listed companies grows significantly and increasingly since In 2009, more than two-thirds of companies have INEDs with multiple appointments, up from less than one-fourth in Panel B shows that nearly half of INEDs with multiple appointments hold only two directorships at the same time and over 90% hold two to five directorships. Even though the largest number of directorships held simultaneously by an INED is as high as 18, there are few cases of such a large number of multiple directorships. 2.2 Variable Definitions and Regression Models Effect of Multiple Directorships on Firm Value To examine the relation between multiple directorships held by INEDs (hereafter multiple directorships ) and firm value, we use the following regression model: 9

10 (1) We do not include industry dummies in the model, because we apply a fixed-effects specification when estimating the coefficients. 14 Firm value, multiple directorships, and all the control variables are defined in Appendix I and are explained as follows. A. Measurement of Firm Value Our study uses two alternative measures of firm value, Market-to-Book Value (MTBV) and Tobin s q. Ferris et al. (2003) and Fich and Shivdasani (2006) use asset-based MTBV, while Sarkar and Sarkar (2009) use equity-based MTBV. Considering the comparability with Tobin s q, we choose asset-based MTBV and calculate it as the market value of the firm s equity plus the book value of total liabilities, divided by the book value of total assets. 15 Tobin s q is the ratio of market value to replacement cost. Because it is very costly and difficult to assess the replacement cost of assets, we use the simple approximation of Tobin s q developed by Chung and Pruitt (1994), which is defined as Tobin s q = (market value of the firm s equity + liquidating value of outstanding preferred stocks + book value of total debt) / book value of total assets. B. Measurement of Multiple Directorships We use the median number of directorships held by INEDs and the ratio of INEDs with multiple appointments as firm-level proxies for multiple directorships. The median number of directorships held by INEDs shows how many directorships a majority of INEDs in a firm at least hold. This prevents a biased influence of extreme observations (Sarkar and Sarkar, 2009). Alternatively, the average number of directorships held by INEDs is also used as a proxy for multiple directorships. To investigate the incremental effect of multiple appointments, we further create a 14 Hausman test rejects the null hypothesis (p=0.000), so we should use fixed-effects model. 15 All data use the number at the end of the year. 10

11 series of ratio of INEDs with multiple appointments, which are defined as the number of INEDs with i or more multiple appointments over the total number of INEDs in the firm, i = 2 to We further use one INED-level proxy for multiple directorships, number of directorships, to test the robustness of the empirical relation found in Model (1). Number of directorships shows how many directorships are simultaneously held by the observed INED. In particular, when using the INED-level number of directorships, all other variables in Model (1) are at the INED-level and calculated by the weighted average method based on the market value of each firm where the observed INED simultaneously serves. C. Control Variables Control variables related to corporate governance include return on assets (ROA), log of board size, INED ratio, substantial-shareholder director ratio, ratio of executives variable bonus in total payment, CEO-chair duality (dummy), executives' options (dummy), conflict (dummy), and family (dummy). Other control variables include log of total assets, debt ratio, code of best practice (dummy), and big 4 (dummy). Definitions and computations of all variables used in our regression are displayed in Appendix I Nonlinear Relationship between Multiple Directorships and Firm Value Model (1) sets the linear relation between multiple directorships and firm value, and it is designed to detect whether busyness or quality dominates in the net effect under all levels of multiple directorships. Based on the non-linear relation found in India by Sarkar and Sarkar (2009), however, we conjecture that busyness and quality have a trade-off relation in the effect of multiple directorships and that the balance of their powers may change with an increase of multiple directorships. 16 Over 90% of INEDs with multiple directorships hold two to five directorships, as we state in the sample description. So by using ratio of INEDs with multiple appointments as well as ratios of busy INEDs, we can almost capture the multiple-appointment phenomenon in the Hong Kong market. 11

12 To test the potential non-linear relationship, we introduce a quadratic term of multiple directorships into our model: (2) Using the above quadratic model, we simply assume that the non-linear relation is symmetric, which is not necessarily true. If the assumption is not reasonable, we may fail to find the hidden non-linear relation. To solve this problem, we need to use a more general way to estimate the potential piece-wise relation between multiple directorships and firm value and find out the turning point. Sarkar and Sarkar (2009) apply the normal spline regression method, 17 which allows the relation between multiple directorships and firm value to change at several levels of multiple directorships. Their model, however, can detect only the significant net effect of multiple appointments at different levels of multiple directorships and cannot capture the significant difference between net effects at different levels. For example, say the busyness and quality effects exist at the same time, and the busyness effect increases substantially after a specific threshold of multiple directorships. The change of net effect caused by busyness is important and can 17 Sarkar and Sarkar's (2009) methodology is as follows: Where spline (multiple directorships) is defined as follows: multiple directorships, if multiple directorships < i Spline1 i, if multiple directorships i 0, if multiple directorships < i Spline2 INED multiple directorships - i, if multiple directorships * i - the spline node i 12

13 provide evidence to the busyness hypothesis. Because the change may not be strong enough to turn the net effect in opposite direction with a significant slope, however, we may fail to identify it using the normal spline technique. Therefore, to prevent losing valuable information, we use the marginal spline technique to investigate the marginal effects of various thresholds. The marginal spline technique is more sensitive to changes in the net effect of multiple appointments of independent directors and can offer deeper insight into the trade-off relation of counter-effects. We formulate this method as follows: Where marginal_spline (multiple directorships) is defined as follows: Marginal _ Spline1 multiple directorships 0, if multiple directorships < j Marginal _ Spline2 multiple directorships, if multiple directorships * j - the marginal spline node j (3) Then, using the linear regression of firm value on the two marginal spline variables of multiple directorships, we can see whether there is a negative effect and how it influences the net effect of multiple appointments of INEDs. The estimated coefficient on Marginal_Spline1 represents the effect of multiple directorships on firm value for those companies with a majority of INEDs holding less than the threshold j (the marginal spline node), i.e., multiple directorships less than j, while the coefficient of Marginal_Spline2 represents the change of effect once multiple directorships are across j, compared with less than j. 13

14 3. Empirical Results and Discussion 3.1 Descriptive Statistics and Correlations Table 2 Panel A provides descriptive statistics of variables used in our regression at the firm level and INED level, respectively. Table 2 Panel B displays correlations of variables used in our models. There are several mildly significant correlation coefficients, but the variance inflation factor (VIF) is relatively low, 18 suggesting a minor multicollinearity problem. [Insert Table 2 Here] 3.2 Regression Results Effect of Multiple Directorships on Firm Value The regression results using Model (1) are presented in Table 3. In Panel A, we use the median number of directorships and the average number of directorships held by INEDs as the proxy for firm-level multiple directorships. For MTBV and Tobin s q, the coefficients of both the median number of directorships and the average number of directorships are significantly positive at the 1% level. These results are stronger than those of Sarkar and Sarkar (2009), as we find an overall positive relationship rather than a positive relationship only at a particular threshold. This finding supports the view that multiple appointments of independent directors benefit the firm in terms of market value. This positive relation found in firm value seems to suggest that the market considers INEDs with multiple directorships as high-quality ones and thereby perceives that firms with more directorships perform better in terms of their long-term view and future growth prospect, which lends support to quality hypothesis. 18 In general, a VIF above 5 or 10 indicates a multicollinearity problem (O brien, 2007). Hair et al. (1995) suggest that a VIF less than 10 indicates inconsequential collinearity. 14

15 Next, we investigate the economic importance of multiple directorships on firm value, by estimating the magnitude of the economic effect resulting from changes in the measure of multiple appointments. The coefficient estimate on the median number of directorships in Panel A shows that a one-standard-deviation increase in the median number of directorships is associated with an increase in MTBV of (0.0908*1.245) and Tobin s q of (0.0807*1.245), which account for 7.6% (12.5%) of the mean (median) of MTBV and 8.1% (14.5%) of the mean (median) of Tobin s q. The economic effect of the average number of directorships is similar. In Panel B, we use a series of ratio of INEDs with multiple appointments as a proxy for firm-level multiple directorships, and the results are consistent with median/average number of directorships held by INEDs. These results provide strong evidence that the market perceives multiple appointments of INEDs positively. [Insert Table 3 Here] Nonlinear Relationship between Multiple Directorships and Firm Value In Table 4, Panel A, we add the squared median number of directorships held by INEDs into the regression (Model (2)) to capture the possible non-linear relation between multiple appointments and firm value. For firm value, both MTBV and Tobin s q report a mild quadratic relation. The theoretical turning point of the relation curve is 7.18 (6.68) median (average) number of directorships for MTBV and 6.79 (6.31) for Tobin s q. 19 This result suggests that at a relatively low degree of multiple directorships, the relation of multiple directorships and firm value is negative, and then turns negative above the threshold (6 directorships). [Insert Table 4 Here] 19 A quadratic equation, say y(x) = ax 2 +bx+c, defines a parabola, which has a vertex, say (h, k). Then we can calculate h=-b/2a. 15

16 The symmetry assumption in the quadratic model may be unrealistic, however. We thus further investigate the piece-wise relation using the normal spline and marginal spline (Model (3)) techniques for the median number of directorships. Using the normal spline specification, we fail to find any significant coefficient of spline2, as in Sarkar and Sarkar (2009), for either MTBV or Tobin s q. 20 This may suggest that the quadratic relation we find in Model (2) is relatively mild and that the piece-wise relation is also not strong enough to be captured by the normal spline technique. We further use the marginal spline to generate the results, which are more sensitive to the change of relationship. To save space, we only display the results for Tobin s q on the median number of directorships in Table 4, Panel B; these results are similar to those of MTBV. Marginal_Spline1 shows significantly positive coefficients at all nodes, consistent with the results in Models (1), (2), and (3). When we vary the node from 3 to 8, nodes 5, 6, and 7 show mild negative significance. The second marginal spline detects whether the change of the net effect of multiple directorships is significant, different from the second normal spline, which shows whether the net effect is significant. So the significant Marginal_Spline2 in our results shows that there is a significant and negative change in the net effect of multiple appointments of independent directors on firm value when a majority of INEDs hold 5 or more directorships. In other words, even though the quality effect mostly dominates and multiple directorships overall have a positive influence on firm value, this positive relation declines significantly as the degree of multiple directorships increases. 4. Robustness Test 4.1 INED-level Multiple Directorships 20 We do not report the results of the normal spline to save space. In unreported results, spline1 is significantly positive for firm value and significantly negative for accounting value, while spline2 is insignificant for all firm-value measures. From nodes=3 to 8, the results remain similar. 16

17 In this section, we focus on the INED-level analysis and use the number of appointments to test its effect on firm value. The results are presented in Table 5. In Panel A, with the INED-level number of appointments as the explanatory variable, we obtain results similar to the firm-level median/average number of directorships held by INEDS in Table 3 Panel A and Table 4 Panel A, i.e., the coefficients on number of appointments is significantly positive for the linear regression and at the same time there is a significant quadratic relation. 21 We estimate that if an INED holds one more directorship, the market-value weighted average of MTBV and Tobin s q will increase by (10.7% of mean, 15.6%of median) and (12.1% of mean, 18.6% of median), respectively. The turning point of the estimated quadratic curve is around 12 appointments. Even though the turning point is high, in Table 5 Panel B, the INED-level marginal spline results show negative and significant coefficients for Marginal_spline2 from node=3 to node=10, which gives more explicit evidence than at the firm-level that the "busyness" effect does exist and reduces the positive net effect of multiple directorships on firm value. These results reinforce our argument that the quality effect comes with a tradeoff with the busyness effect. [Insert Table 5 Here] 4.2 Instrumental Variable for Multiple Directorships Most studies in corporate governance are subject to endogeneity problems, because the results can be explained by quality signaling, reverse causation, or optimal differences (Black et al., 2006). Our study could be prone to reverse causation, i.e., INEDs who serve in firms with higher firm value may be more likely to get more extra directorships (Fich and Shivdasani, 2006). Such reverse causality leads to corporate governance being correlated with the error item and results in an 21 The heterogeneity of INEDs is controlled by a fixed-effect specification in our regression. In addition, we also split the sample into different groups by age of INED and redo the regression of Model (1) for each group. The results remain consistent. 17

18 inconsistent OLS parameter estimator. 22. To address this problem, we use the 2SLS regression to deal with the possible endogeneity between multiple appointments and firm value. 2SLS regression requires at least one instrumental variable that is not correlated with the error term of the model for firm performance (i.e., it must be exogenous). At the same time, this instrument should strongly correlate with and thereby be able to predict multiple directorships. The ideal instrument is the result of a natural experiment, an event that changes the endogenous regressors, but leaves the other aspects of the economic system unaffected. It is very hard, however, to disentangle the effect of the specific event from all other events occurring at the same time (Larcker and Rusticus, 2010). Therefore, some studies use a weak instrument that is exogenous but weakly correlated with endogenous variable, but this may result in a larger bias than an OLS estimate (Renders et al., 2010). The other kind of imperfect instrument is a so-called semi-endogenous one, which is partially endogenous but highly correlated with endogenous variable. The endogeneity of such an instrument should be lower than the endogeneity of the original endogenous variable (Larcker and Rusticus, 2010; Renders et al., 2010). We estimate models using three sets of instruments. The first set of instruments is a lagged endogenous variable (Luca and Petrova, 2008; Neanidis and Savva, 2009; Renders et al., 2010). With panel data, we can use one-year lagged multiple directorships as the instrument for itself. Past multiple directorships should be strongly correlated with current multiple directorships. Besides, past multiple directorships are the predetermined variable for current firm value, so there should be no reverse causation with current firm value, and past multiple directorships are expected to influence current firm value mainly through current multiple directorships. The second set is industry mean of endogenous variables, because we conjecture that the endogeneity problem could have existed at the firm level, but not 22 We find the endogeneity of multiple directorships and firm performance using the Durbin-Wu-Hausman test. 18

19 at the industry level (Fisman and Svensson, 2007; Laeven and Levine, 2009; Lin et la., 2010; Lin et la., 2011). The third is the policy-change year dummy. From Table 1 Panel A, we can observe that there is a persistent growth in the number of INEDs with multiple appointments at the firm level from 2001 to In particular, year 2004 encounters a jump in firm-level multiple directorships, because a change in the listing rules in Hong Kong increased the requirement for the number of INEDs from 2 to 3. This creates a natural experimental condition to address the endogeneity issue. We use a dummy that indicates the firm-year observations after 2004 as the instrument to capture the increase in multiple directorships. 23 We use the average number of directorships held by INEDs as the proxy of multiple directorships, which is also the endogenous variable. These three sets of instruments are all highly correlated with the average number of directorships held by INEDs. 24 To be consistent with the OLS regressions, we use a fixed-effect 2SLS when using the first and second set of instruments. But we use random-effect 2SLS for the third set, because the policy-change year dummy is time-invariant. The second-stage results of 2SLS using different sets of instruments are reported in Table 6, and they all show positive associations between average number of directorships and firm value, which confirms our findings in the OLS regressions. [Insert Table 6 Here] 5. Concluding Remarks 23 We also develop a dummy variable as an instrument, which equals 1 if the company has less than 3 INEDs before 2004, and otherwise 0. We define this instrumental variable as the meets-requirement dummy. The 2SLS regressions using the meets-requirement dummy still report significant positive coefficients even though less than OLS regression. The coefficient is much larger than that in the OLS, however, because the predicted multiple directorships explain only a small proportion of original one. So meets-requirement dummy is a weak instrument. Considering that the weak-instrument 2SLS may be more biased than the OLS, we do not report the result using the meets-requirement dummy. 24 The correlation coefficients of average number of directorships with its one-year lagged value, its industry-mean value, and the policy-change year dummy are , , and , respectively. 19

20 This paper examines the effect of multiple appointments of INEDs on firm value using a unique dataset of Hong Kong-listed companies from 2001 to The effect of multiple appointments is mostly positive with some nonlinearity. We use different sets of instruments to address the potential endogeneity issue, and the results remain similar. Consistent with Sarkar and Sarkar (2009), we support the quality hypothesis, i.e., investors perceive INEDs with multiple appointments as being of high quality and thereby highly value firms with such INEDs. Overall, quality comes with a tradeoff with the busyness of INEDs with multiple appointments; yet, quality comes first for investors. Findings from this study offer several policy implications. Higher levels of multiple directorships may benefit the company more than the cost. The results support the argument of Ferris et al. (2003) and Sarkar and Sarkar (2009) that setting a strict limit on the number of directorships held by INEDs is not necessary, and that the market would adjust valuations of firms that may have overly busy INEDs. Our results contrast with recent concerns that multiple appointments may damage firm value. 25 The quality effect of INEDs may also augment firm value. Despite the limited supply of qualified INEDs, our results support stock exchanges increasing the minimum requirement of independent directors for listed companies. A suitable grace period may ease administrative problems for firms with a marginal number of independent directors. 26 References 25 For example, during the consultation period of amendment of Listing Rules in Hong Kong in 2002, a number of respondents do not support raising the minimum requirement from 3 INEDs to one-third of the members of the board by arguing that the supply of qualified INEDs in the market is limited, and similar arguments can be found in the 2010 consultation. 26 Similar changes have been made previously for HKEx in Jan 2003, with a grace period of one year. 20

21 Beasley, M. S. (1996). An empirical analysis of the relation between the board of director composition and financial statement fraud. Accounting Review, 71(4), Black, B. S., Jang, H., and Kim, W. (2006). Does corporate governance predict firms' market values? evidence from korea. Journal of Law, Economics, and Organization, 22(2), 366. Carpenter, M. A., and Westphal, J. D. (2001). The strategic context of external network ties: Examining the impact of director appointments on board involvement in strategic decision making. Academy of Management Journal, Cheung, Y. L., Thomas Connelly, J., Limpaphayom, P., and Zhou, L. (2007). Do investors really value corporate governance? evidence from the hong kong market. Journal of International Financial Management and Accounting, 18(2), Chung, K. H., and Pruitt, S. W. (1994). A simple approximation of Tobin s q. Financial Management, Conyon, M. J., and Read, L. E. (2006). A model of the supply of executives for outside directorships. Journal of Corporate Finance, 12(3), Core, J. E., Holthausen, R. W., and Larcker, D. F. (1999). Corporate governance, chief executive officer compensation, and firm performancevalue. Journal of Financial Economics, 51(3),

22 Cotter, J. F., Shivdasani, A., and Zenner, M. (1997). Do independent directors enhance target shareholder wealth during tender offers? Journal of Financial Economics, 43(2), Fama, E. F., and Jensen, M. C. (1983). Separation of ownership and control. Journal of Law and Economics, 26(2), Ferris, S. P., Jagannathan, M., and Pritchard, A. C. (2003). Too busy to mind the business? monitoring by directors with multiple board appointments. The Journal of Finance, 58(3), Ferris, S. P., Jagannathan, M., and Pritchard, A. C. (2003). Too busy to mind the business? monitoring by directors with multiple board appointments. The Journal of Finance, 58(3), Fich, E. M., and Shivdasani, A. (2006). Are busy boards effective monitors? The Journal of Finance, 61(2), Fisman, R., and Svensson, J. (2007). Are corruption and taxation really harmful to growth? Firm level evidence. Journal of Development Economics, 83(1), Hair Jr, J. F., Anderson, R. E., Tatham, R. L., and Black, W. C. (1995). Multivariate data analysis: With readings Prentice-Hall, Inc. Jiraporn, P., Kim, Y. S., and Davidson III, W. N. (2008). Multiple directorships and corporate diversification. Journal of Empirical Finance, 15(3),

23 Jiraporn, P., Singh, M., and Lee, C. I. (2009). Ineffective corporate governance: Director busyness and board committee memberships. Journal of Banking and Finance, 33(5), Larcker, D. F., and Rusticus, T. O. (2010). On the use of instrumental variables in accounting research. Journal of Accounting and Economics, 49(3), Laeven, L., and Levine, R. (2009). Corporate governance, regulation, and bank risk taking. Journal of Financial Economics, 93(2), Lin, C., Lin, P., and Song, F. (2010). Property rights protection and corporate RandD: Evidence from china. Journal of Development Economics, 93(1), Lin, C., Ma, Y., Malatesta, P., and Xuan, Y. (2011). Ownership structure and the cost of corporate borrowing. Journal of Financial Economics, 100(1), Loderer, C., and Peyer, U. (2002). Board overlap, seat accumulation and share prices. European Financial Management, 8(2), Luca, A., and Petrova, I. (2008). What drives credit dollarization in transition economies? Journal of Banking and Finance, 32(5), Neanidis, K. C., and Savva, C. S. (2009). Financial dollarization: Short-run determinants in transition economies. Journal of Banking and Finance, 33(10), O brien, R. (2007). A caution regarding rules of thumb for variance inflation factors. Quality and Quantity, 41(5),

24 Perry, T., and Peyer, U. (2005). Board seat accumulation by executives: A shareholder's perspective. The Journal of Finance, 60(4), Porta, R. L., Lopez-de-Silanes, F., Shleifer, A., and Vishny, R. W. (1998). Law and finance. Journal of Political Economy, 106(6), Renders, A., Gaeremynck, A., and Sercu, P. (2010). Corporate-governance ratings and company performance: a cross-european study. Corporate Governance: An International Review, 18(2), Sarkar, J., and Sarkar, S. (2009). Multiple board appointments and firm performance in emerging economies: Evidence from India. Pacific-Basin Finance Journal, 17(2), Shivdasani, A. (1993). Board composition, ownership structure, and hostile takeovers. Journal of Accounting and Economics, 16(1-3), Standard & Poor's. (2002). Corporate governance in Hong Kong. Standard & Poor's. Vafeas, N. (1999). Board meeting frequency and firm performance. Journal of Financial Economics, 53(1),

25 Table 1 Frequency tables for multiple directorships measures - This table displays the distributions of firm- and INED-level multiple appointments in our sample. The sample consists of firm-year observations and 2,953 INED-year observations (excluding INEDs without multiple appointments), which cover all Hong Kong-listed non-financial companies from 2001 to Panel A presents the distribution of firm-year observations by year and number of INEDs with multiple appointments. Panel B presents the frequencies of number of appointment for INEDs with multiple directorships for each year. Panel A - Frequencies of No. of INEDs with multiple appointments (firm-year) No. of INEDs with multiple appointments in a firm Year Total Total 5,772 2,773 1, Percent 54.13% 26.00% 13.63% 5.18% 0.96% 0.10% 100% Panel B - Frequencies of No. of directorships held by INED (INED-year) No. of directorships held by INED Frequency Percent Cum. Perc. 2 1, Total 2,

26 Table 2 Summary statistics - This table displays the summary statistics of firm value, measures of firm-level multiple directorships and networks, and other firm-characteristic variables. Panel A provides descriptive statistics for all the variables used in our model. INED-level variables, except for No. of directorships, are all calculated by weighted average method based on the market value of each firm that the observed INED serves in simultaneously. Panel B displays correlations of variables used in our model. MTBV, Tobin s q, ROA, log of total assets, log of board size, INED ratio, debt ratio, substantial-shareholder director ratio, and ratio of executives variable bonus in total payment are winsorized at the 1st and 99th percentiles. See Appendix I for the definitions of variables. Panel A - Descriptive statistics of variables Firm-level 26 INED-level Variable Mean Median S.D. Min Max Mean Median S.D. Min Max Dependent Variables MTBV Tobin s q Explanatory Variables Median No. of directorships held by INEDs Average No. of directorships held by INEDs Number of appointments Control Variables ROA Log of total assets Log of board size INED ratio Debt ratio Substantial-shareholder director ratio Ratio of executives variable bonus CEO-chair duality Code of best practice Executives' options Big Conflict Family Age No. of Observations 10,664 2,953

27 Panel B - Correlations of Variables (firm-level) MTBV = Market-to-book value; Q = Tobin s Q; Med = Median No. of directorships held by INEDs; Avg = Average No. of directorships held by INEDs; ROA = Return on assets; L_A = Log of total assets; L_B = Log of board size; I_R = INED ratio; D_R = Debt ratio; SSD_R = Substantial-shareholder director ratio; EVB_R = Ratio of executives variable bonus in total payment; Dual = CEO-chair duality; CBP = Code of best practice; ExOp = Executives' options. We regress each of the explanatory variables on the remaining explanatory variables, obtain the variance inflation factors (VIF), and report them in the rightmost column. * represents 5% significance level. MTBV Q Med Avg ROA L_A L_B I_R D_R SSD_R EVB_R Dual CBP ExOp Big4 Conflict Family VIF MTBV Q 0.98* Med Avg * ROA -0.33* -0.31* L_A -0.36* -0.35* 0.14* 0.19* 0.25* L_B -0.14* -0.13* 0.06* 0.12* 0.14* 0.52* I_R 0.08* 0.09* -0.13* -0.07* * -0.57* D_R 0.22* 0.20* -0.02* -0.03* -0.24* 0.05* SSD_R 0.03* 0.04* -0.04* -0.04* * -0.33* 0.27* -0.08* EVB_R -0.06* -0.06* 0.04* 0.08* 0.16* 0.36* 0.25* * -0.06* Dual * -0.03* -0.10* -0.20* 0.04* * -0.08* CBP -0.06* -0.05* -0.03* -0.03* 0.04* 0.11* 0.09* -0.11* * 0.04* ExOp 0.06* 0.07* 0.06* 0.07* -0.08* -0.18* -0.09* * -0.05* * Big4-0.15* -0.13* 0.08* 0.09* 0.13* 0.27* 0.16* -0.18* -0.04* -0.14* 0.15* -0.03* 0.11* Conflict -0.06* -0.05* 0.14* 0.18* -0.03* 0.18* 0.19* -0.08* 0.03* -0.07* 0.05* -0.09* * 0.10* Family -0.03* -0.03* -0.07* -0.14* 0.16* -0.04* -0.03* -0.03* -0.06* 0.31* * 0.03* * -0.15*

28 Table 3 Firm-level linear regression on alternative multiple-directorship measures - This table displays the results of fixed effect linear regression according to Model (1). Panel A displays the results of firm value on median number of directorships or average number of directorships held by INEDs, which are proxies for firm-level multiple directorships. Panel B displays the results of firm value on a series of ratio of INEDs with i or more appointments, which is defined as the number of INEDs with i or more multiple appointments over the total number of INEDs, i = 2 to 5, also proxies for firm-level multiple directorships. The rest of variables are firm characteristics and corporate-governance attributes, which are used as control variables and defined in Appendix I. *, **, and *** represent 10%, 5%, and 1% significance levels, respectively. Panel A - Median/Average number of directorships held by INEDs 28 Dependent Variable (1) (2) (3) (4) Independent Variable MTBV MTBV Tobin s q Tobin s q Median No. of directorships held by INEDs *** *** ( ) ( ) Average No. of directorships held by INEDs *** *** ( ) ( ) ROA *** *** *** *** (0) (0) (1.65e-10) (1.58e-10) Log of total assets *** *** *** *** (0) (0) (0) (0) Log of board size 0.820*** 0.759*** 0.696*** 0.643*** (2.21e-06) (9.89e-06) (1.39e-05) (5.15e-05) INED ratio 1.516*** 1.336*** 1.182*** 1.023*** (3.07e-05) ( ) ( ) ( ) Debt ratio 2.106*** 2.126*** 1.821*** 1.839*** (0) (0) (0) (0) Substantial-shareholder director ratio 1.250*** 1.257*** 1.295*** 1.301*** ( ) ( ) (7.51e-05) (6.96e-05) Ratio of executives variable bonus in total payment 0.630*** 0.618*** 0.559*** 0.548*** ( ) ( ) ( ) ( ) CEO-chair duality * (0.106) (0.125) (0.0886) (0.105) Code of best practice (0.327) (0.308) (0.746) (0.713) Executives' options (0.867) (0.921) (0.878) (0.930) Big *** *** ** ** ( ) ( ) (0.0118) (0.0129) Conflict (0.604) (0.535) (0.316) (0.268) Family * (0.251) (0.250) (0.0996) (0.101) Constant 11.22*** 11.40*** 10.01*** 10.16*** (0) (0) (0) (0) Observations 4,754 4,754 4,748 4,748 R-squared Number of firms