Review of Quantitative Finance and Accounting

, Volume 31, Issue 2, pp 121–145 | Cite as

Board size and firm performance: the moderating effects of the market for corporate control

  • Shijun Cheng
  • John H. EvansIII
  • Nandu J. Nagarajan
Original Paper

Abstract

We examine whether takeover threats affect the importance of board size using the passage of state antitakeover laws enacted in mid-to-late 1980s as our empirical setting. While the Complement Hypothesis predicts that board size matters more before the passage of the laws, the Substitute Hypothesis predicts the opposite. For a sample of 350 Forbes 500 firms over the period 1984–1991, we find a significant association between smaller boards and better firm performance before passage of antitakeover laws, but a much weaker relation (reduced by more than one-third) after the takeover restrictions were in place. Consistent with the Complement Hypothesis, this finding suggests that decreasing board size is more valuable when the market for corporate control is more active.

Keywords

Antitakeover laws Board of directors Firm performance 

1 Introduction

The importance of board size in the corporate governance process is well-recognized. The negative association between board size and firm performance, first documented by Yermack (1996), is one of the prominent empirical regularities of the role of boards (Hermalin and Weisbach 2003). This negative board-size effect potentially arises because coordination and communication costs, as well as director free-riding costs, increase with board size. These costs impair prompt and effective board decisions. They also highlight the critical role of directors’ incentives, which depend on a variety of governance mechanisms, including the market for corporate control (Jensen 1993; Williamson 1983). Our study focuses on the impact of antitakeover laws (ATLs), an important determinant of the market for corporate control, on the importance of board size as reflected in the negative association between board size and firm performance.

The literature suggests two competing hypotheses about how takeover threats affect the importance of board size (e.g., John and Senbet 1998). The Complement Hypothesis predicts that by boosting the importance of prompt board decisions and directors’ incentive to monitor top management, a more active market for corporate control likely increases the importance of board size in the governance process. When the market for corporate control is more active, more information about firm and managerial performance is available (e.g., Mikkelson and Partch 1997), the likelihood of a firm making acquisitions or being acquired is greater, and the timeliness of board decisions becomes more important. Likewise, directors have stronger incentives to monitor top management because takeover markets discipline directors as well as managers. In particular, directors who fail as monitors lose current and future board positions (e.g., Harford 2003; Kini et al. 1995; Morck et al. 1989).

Nevertheless, as board size increases, directors’ free-riding will eventually outweigh this increased monitoring incentive. Therefore, when the market for corporate control is more active, the Complement Hypothesis predicts a significant, negative relation between board size and firm performance. That is, when takeover intensity is stronger, small, highly motivated boards will make more effective, timely decisions, while large boards will be burdened by coordination costs and free-riding, leading to slower, less effective decisions. In contrast, when takeover intensity is weaker, e.g., following passage of ATLs, the Complement Hypothesis predicts the negative association between board size and firm performance to be significantly weaker. Reduced takeover intensity diminishes both the importance of timely, effective board decisions and the incentives of directors to invest the time and effort to make such decisions.

On the other hand, the Substitute Hypothesis predicts that by reducing the demand for internal monitoring, a more active market for corporate control is likely to decrease the importance of board size (e.g., Williamson 1983). Prior studies examining board structures (e.g., Brickley et al. 1987; Kini et al. 1995; Mayers et al. 1997) and managerial compensation and dismissal (e.g., Bertrand and Mullainathan 2000; Huson et al. 2001; Mikkelson and Partch 1997), document mixed findings regarding the descriptive validity of the Complement versus SubstituteHypotheses. In distinguishing these competing hypotheses, our study differs from these prior studies by focusing on the well-documented negative association between board size and firm performance.

Specifically, we test the two competing hypotheses involving takeover intensity, board size, and firm performance using a sample of 350 unregulated Forbes 500 firms for the years 1984–1991. The mid-to-late 1980s witnessed the introduction of second-generation antitakeover laws in various states in the US Passage of the ATLs produced a series of staggered, relatively exogenous shocks that increased the bidder’s cost in a hostile takeover (Comment and Schwert 1995; Karpoff and Malatesta 1989).1 Following passage of the ATLs, managerial pay increased, while managerial ownership and firm leverage declined (Bertrand and Mullainathan 2000; Cheng et al. 2005; Garvey and Hanka 1999), suggesting a post-ATL decrease in takeover threat. We treat passage of the ATLs as a natural experiment enabling us to test our hypotheses concerning how takeover intensity affects the role of board size.

In addition to confirming the inverse relation between board size and firm performance, our results demonstrate that this association becomes weaker following the passage of the ATLs. Specifically, we relate firm performance, based on Tobin’s Q and accounting returns on assets, to the natural log of firm board size, an indicator variable for the passage of ATLs, and the interaction between board size and ATLs, while controlling for industry- and firm-level takeover activities and other determinants of firm performance. Consistent with the Complement Hypothesis, we find that the passage of ATLs is associated with a significant decline in the negative association between board size and firm performance. To illustrate, our results indicate that reducing board size by one member would be associated with an estimated 4.18% increase in Q before the passage of the ATLs versus only a 2.62% increase in Q after the passage. These results remain unchanged after fully controlling for the effect of an important internal antitakeover mechanism—poison pills.

We also provide some supporting evidence for the Complement Hypothesis on two related issues. First, the Complement Hypothesis predicts that smaller boards’ superior monitoring increases CEO pay-performance sensitivity (PPS), particularly when the board’s monitoring incentives are reinforced by higher levels of takeover intensity. This prediction is modestly confirmed because prior to passage of ATLs, smaller boards are associated with marginally greater CEO PPS, but this relation no longer holds after the passage of ATLs. Second, we find that before the passage of the ATLs, boards tend to be smaller for our sample firms, consistent with increased importance of board size when takeover intensity is higher.

Our findings contribute to the literature in two ways. First, our results demonstrate that when takeover intensity is greater, the advantages of smaller, more agile boards are more pronounced. Conversely, when takeover intensity is reduced by the passage of ATLs, the advantages of smaller boards in encouraging improved firm performance are significantly reduced. As such, we add new results for each of the three questions that Hermalin and Weisbach (2003) identify as motivating previous empirical research on the role of boards in the governance process. With respect to how board characteristics such as size affect profitability, we document that smaller boards exert more positive influence on profitability when takeover intensity is high. Regarding how board characteristics affect observable board actions, we demonstrate that smaller boards’ ability to establish stronger CEO PPS is enhanced by the intensity of the market for corporate control. Finally, in terms of factors affecting the makeup of boards, we provide evidence that legislation reducing takeover intensity is associated with larger boards.

Second, our results extend prior research results on ATLs and board composition by providing new results on the dynamics of the role of board size in the governance process. Previous studies document post-ATL changes in managerial pay, ownership, and firm’s capital structures (Bertrand and Mullainathan 2000; Cheng et al. 2005; Garvey and Hanka 1999). Our results add to these studies by establishing that the importance of board size to firm performance declined after the passage of the ATLs. Likewise, to examine the impact of takeover threats on board monitoring, prior studies focus on board composition (e.g., Brickley and James 1987; Mayers et al. 1997) and directors’ incentives (e.g., Harford 2003; Kini et al. 1995). By focusing on board size, another important characteristic of boards, and its relation to firm performance, our study makes the previous analysis more complete.

The next section discusses the literature and develops the hypotheses. The third section describes the data and method. The fourth section presents results on the relation between board size and firm performance. The fifth section reports results on the relation between board size and CEO PPS, as well as on the impact of ATLs on board size. The sixth section concludes the paper.

2 Literature and hypotheses

2.1 The importance of board size

Jensen (1993) and Lipton and Lorsch (1992) argue that larger boards are generally less effective than smaller boards for two reasons. First, it is harder for larger boards to arrange meetings, reach consensus, and react rapidly because of communication and coordination costs. Second, the ability and incentives of the board to control management decrease with board size. Board directors hesitate to criticize the policies of top managers or to candidly discuss corporate performance (Lipton and Lorsch 1992; Magnet 1992).2 These problems become more acute as board size increases because the cost to any individual director of not exercising diligence in monitoring management falls in proportion to the total number of board directors. That is, larger boards encourage directors to free-ride. While reducing board size helps mitigate the communication and coordination costs of the board, managers have incentives to maintain a larger board, which is easier for management to control.

The empirical evidence confirms that larger boards are generally less effective. Using data from 452 large US industrial corporations between 1984 and 1991, Yermack (1996) documents an inverse association between board size and firm performance, as measured by Tobin’s Q, after controlling for other factors that are likely to affect Q. He also finds that board size is negatively associated with accounting profitability, the sensitivity of CEO cash compensation (including salary and annual bonus) to stock returns, and the sensitivity of CEO turnover to abnormal stock returns. Using data from a random sample of approximately 900 small Finish firms, Eisenberg et al. (1998) provide evidence that the inverse relation between board size and firm performance holds for small firms as well. These findings confirm the importance of board size to firm performance as well as to the quality of board actions.3

Because the board of directors is only one of a number of governance mechanisms, the importance of board size is likely to depend on the performance of other governance mechanisms (Shleifer and Vishny 1997; Williamson 1983). We focus on the market for corporate control because it is well-recognized in the literature as an effective external governance mechanism (Jensen 1993; Morck et al. 1989). Based on previous studies, the market for corporate control can complement or substitute for board monitoring (e.g., Jensen 1993; Williamson 1983).

2.2 The complement hypothesis

Under the Complement Hypothesis, the importance of board size increases with takeover intensity. In particular, the takeover market can affect the importance of board size by influencing the two sources of the board-size effect, as discussed above. First, takeover threats may increase the importance of prompt and effective decisions, and hence, make larger boards more costly to the firms. When the takeover market is more active, more information about firm and managerial performance is likely to become available, reinforcing the importance of rapid and effective board leadership (Mikkelson and Partch 1997). Similarly, the firm’s business environment changes due to the increased opportunities to acquire or to be acquired in times of active takeover intensity. The increased information about firm and managerial performance and the changes in business environment both add to the importance of prompt and effective actions by the board, while the ability of the board to make such decisions decreases with board size.

Second, greater takeover intensity strengthens directors’ incentives to monitor, reflecting the disciplinary power of the market for corporate control over managers and directors. Harford (2003) finds that directors are rarely retained following a completed takeover bid and target directors hold fewer directorships in the future. Morck et al. (1989) find that in poorly-performing industries where boards apparently fail as monitors, takeovers, especially hostile ones, are more active. Kini et al. (1995) conclude that “the discipline associated with corporate takeovers extends beyond top management to effect restructuring of the entire board”. These findings suggest that takeover threats trigger financial and reputational considerations for the directors, resulting in stronger directors’ incentives to monitor management.

Whether this heightened incentive for directors to monitor actually results in more effective board monitoring depends on board size. Because the costs of communication, coordination, and free-riding increase with board size, the benefits of this heightened incentive are likely to be dominated by the increased costs as the board becomes sufficiently large. As a result, we expect that the increased incentives to monitor in periods of greater takeover intensity will lead to more effective monitoring primarily for smaller boards.

In summary, the Complement Hypothesis predicts that board size matters more as takeover intensity increases because that intensity enhances the importance of prompt board decisions and provides directors with sufficient incentive to supply the monitoring necessary to support good decisions.4 Following prior studies (e.g., Bertrand and Mullainathan 1999, 2003; Garvey and Hanka 1999), we use the introduction of the ATLs as our proxy for a reduction in the level of takeover intensity. According to the Complement Hypothesis, the negative association between board size and firm performance will be weaker after the passage of the ATLs.

2.3 The substitute hypothesis

Under the Substitute Hypothesis, the importance of board size decreases with takeover intensity. When the market for corporate control is more active, internal governance mechanisms could actually become less important (Mayers et al. 1997; Williamson 1983). This reasoning suggests that board size should become more important after the passage of the ATLs.

Besides the relation between board characteristics and firm performance, Hermalin and Weisbach (2003) identify the relation between board characteristics and the quality of observable board decisions, and the determinants of board characteristics as prominent issues in the previous literature. We examine the impact of ATLs on these two issues to provide a more complete perspective on how ATLs influence the importance of board size. Consistent with Yermack (1996), we use CEO compensation policy as an observable board decision. Yermack (1996) documents a negative association between board size and CEO PPS. Given the Complement Hypothesis that board size becomes more important when the market for corporate control is more active, the passage of ATLs should weaken the negative association between board size and CEO PPS. Following the same hypothesis, board size is smaller before the passage of the ATLs, reflecting the increased potential for small boards to monitor effectively in the presence of the discipline from the takeover market. Once again, the Substitute Hypothesis predicts the opposite, and our empirical results provide a basis for choosing between these competing explanations.

3 Data and methods

3.1 Sample and data

We start with the sample in Yermack (1996), which contains 452 Forbes 500 firms over 1984–1991. This sample is appropriate for this study because the passage of the second-generation ATLs in various states in the US in the mid-to-late 1980s produced a major shift in the intensity of the market for corporate control (Jensen 1993; Mikkelson and Partch 1997), providing a natural experimental setting to test the moderating effects of the market for corporate control.

An advantage of this setting is that ATLs are generally exogenous to firm performance and internal governance mechanisms (Garvey and Hanka 1999), and endogenous measures of takeover threats such as firm level merger and acquisition bids make it hard to infer causal effects of the market for corporate control on firm performance and internal governance (e.g., Bertrand and Mullainathan 2003). In addition, our sample period witnessed the passage of ATLs, allowing before- versus after-passage comparisons and enhancing the power of the statistical tests. For similar reasons, several recent studies have used data from this period to examine the role of the market for corporate control (e.g., Bertrand and Mullainathan 2003; Cheng et al. 2005; Garvey and Hanka 1999). Although our sample period is the mid-to-late 1980s, our inquiry continues to be relevant today, because after a brief decline in the early 1990s, there has been a recent resurgence in merger and takeover activity (Holmstrom and Kaplan 2001).

The Yermack (1996) sample excludes utility and financial firms. Because takeover threats and board effectiveness are likely to operate differently in regulated industries (Mitchell and Mulherin 1996; Yermack 1996), we further exclude firms in communications, which is also heavily regulated.5 All firms in our sample appeared at least four times between 1984 and 1991 in the Forbes 500 rankings based on sales, assets, net income, or market capitalization. In addition, we require that all of our sample firms be publicly traded for at least four consecutive fiscal years. The 4-year criterion represents a trade-off between the data requirements for panel data analysis, which favors including only firms with several consecutive years of data, and a potential survivorship bias, which argues for retaining more firms (Yermack 1996).

We use data on board structures and other governance variables from the dataset in Yermack (1995, 1996), as obtained from annual firm proxy statements, 10-Ks, and Forbes annual surveys of CEO compensation. Bertrand and Mullainathan (1999, 2003) are our sources for the ATLs and their passages in various states in the US. We use Comment and Schwert’s (1995) dataset to identify the timing of poison pill adoptions. Additionally, we collect stock price and takeover data from CRSP, and obtain data on takeover attempts by searching the Wall Street Journal Index. We use COMPUSTAT for financial data and for state incorporation data (following Bertrand and Mullainathan (1999)), supplemented by information from Compact D/SEC and proxy statements. To control for firm-specific fixed effects, we retain only those firms with the required data for at least two consecutive years. This process yields 350 firms in 35 states (of which 192, or 54.86%, are incorporated in Delaware), representing 48 two-digit SIC code industries. These 350 firms experienced 274 CEO turnovers in the sample period.

3.2 Antitakeover laws and other takeover characteristics

Many states passed second-generation ATLs in the mid-to-late 1980s. These takeover restrictions typically limit the form of business combinations, constrain prices in takeover-related transactions, and reduce large shareholders’ voting rights. Business combination laws impose a 3 to 5-year moratorium on specified transactions between the target and a bidder unless the target board approves the transaction. Fair price laws require that, for shares purchased for takeover purposes, the acquirer pay no less than the maximum price the acquirer paid for shares within the preceding 2 years. Finally, control share laws give non-interested shareholders the right to decide whether a large shareholder should retain voting rights. Prior research documents that such restrictions do inhibit takeover activity and encourage management entrenchment (e.g., Bertrand and Mullainathan 2000; Jensen 1993; Mikkelson and Partch 1997).

In the absence of a clear theoretical basis for ranking the relative strength of the three forms of ATLs described above, we follow Barnhart et al. (2000) and Garvey and Hanka (1999) in treating the ATLs as equivalent. Thus, we take no a priori position on the relative effectiveness of the different forms of ATLs. Instead, we focus on the timing of the passage of the first ATL in a given state, reasoning that this is likely to be the most significant shift in the potential for takeover activity in that state. Accordingly, for each firm we define the indicator variable AfterLAW as equal to one for all years following passage of the first ATL in the state in which the firm is incorporated.

We also consider industry- and firm-level takeover activities. Mitchell and Mulherin (1996) demonstrate that takeover activities vary by industry, reflecting the effect of differential external shocks. We measure industry-level takeover activities using the frequency of takeovers by industry, as well as by the percentage of total value taken over in the industry. Specifically, for every two-digit SIC code industry, we define the indicator Ind_THREAT as equal to one, indicating high industry takeover activity, if at least one of the following industry-specific ratios exceeds the corresponding median ratio for all two-digit SIC code industries, and zero otherwise. First, we use the ratio of the number of firms taken over to the total number of the firms in that two-digit SIC code industry; second, the ratio of market value of firms taken over to the total market value of the firms in that two-digit SIC code industry; and third, the ratio of book value of assets of firms taken over to the total book value of all firms in that two-digit SIC code industry.6 Next, we consider the following two firm-level takeover threat variables: BID_TKN (an indicator equal to one if the firm experienced an open bid or was taken over within 2 years, and zero otherwise) and AfterPIL (an indicator equal to one if the firm had adopted a poison pill provision in the previous year or earlier, and zero otherwise).

3.3 Descriptive statistics

Table 1 presents descriptive statistics for the ATLs, industry- and firm-level takeover activities, and governance variables. Panel A presents means and standard deviations by year and for the entire sample period. The mean of AfterLAW is 0.448, indicating that 44.8% of the sample firm-years follow passage of the first ATL in the state in which the firm is incorporated. The percentages of sample firm-years that follow the passage of the Business Combination (PostBC), Fair Price (PostFP), and Control Shares (PostCS) Laws are 36.6, 21.2, and 8.4%, respectively. The mean of Ind_THREAT is 0.702, indicating that 70.2% of the firm-years are classified as experiencing a high industry takeover threat. 31.4% of the sample firm-years follow the firm’s adoption of a poison pill provision, but fewer than 5% of all firm-years are subject to takeover attempts. Finally, the coverage of the ATLs and poison pills increased steadily over the sample period.
Table 1

Descriptive statistics of state antitakeover laws and governance characteristics

Panel A. State antitakeover laws and other takeover characteristics

Variables

1984–1991

Mean by year

Mean

Standard deviation

1984

1985

1986

1987

1988

1989

1990

1991

AfterLAW (dummy = 1 after the passage of the first second-generation antitakeover law by the state in which the firm is incorporated, and 0 otherwise)

0.448

0.497

0.029

0.074

0.201

0.261

0.295

0.860

0.900

0.963

PostBC (dummy = 1 if after the passage of the Business Combination laws, and 0 other wise)

0.366

0.482

0.000

0.000

0.072

0.140

0.158

0.759

0.868

0.931

PostFP (dummy = 1 if after the passage of the Fair Price laws, and 0 other wise)

0.212

0.409

0.029

0.060

0.178

0.246

0.252

0.261

0.304

0.364

PostCS (dummy = 1 if after the passage of the Control Shares laws, and 0 other wise)

0.084

0.277

0.000

0.014

0.023

0.037

0.069

0.149

0.189

0.189

Ind_THREAT (dummy = 1, if industry % of number, market value, or book value of assets of firms taken over is above median, and 0 otherwise)

0.702

0.458

0.740

0.754

0.820

0.814

0.731

0.857

0.674

0.680

BID_TKN (dummy = 1, if having an open bid or receiving a bid within a year, or taken over within two years, 0 otherwise)

0.048

0.214

0.026

0.037

0.060

0.083

0.080

0.060

0.023

0.017

AfterPIL (dummy = 1, after adoption of poison pills, 0 otherwise)

0.314

0.464

0.000

0.006

0.051

0.329

0.397

0.526

0.597

0.609

Panel B. Correlations among takeover threat proxies

Variables

Pearson correlation coefficients

AfterLAW

PostBC

PostFP

PostCS

Ind_THREAT

BID_TKN

AfterLAW (dummy = 1 after the passage of the first second-generation antitakeover law by the state in which the firm is incorporated, and 0 otherwise)

      

PostBC (dummy = 1 if after the passage of the Business Combination laws, and 0 other wise)

0.844***

     

PostFP (dummy = 1 if after the passage of the Fair Price laws, and 0 other wise)

0.576***

0.376***

    

PostCS (dummy = 1 if after the passage of the Control Shares laws, and 0 other wise)

0.336***

0.175**

0.350***

   

Ind_THREAT (dummy = 1, if industry % of number, market value, or book value of assets of firms taken over is above median, and 0 otherwise)

−0.028

0.024

0.014

−0.032

  

BID_TKN (dummy = 1, if having an open bid or receiving a bid within a year, or taken over within two years, 0 otherwise)

−0.031

−0.036

−0.016

−0.007

−0.002

 

AfterPIL (dummy = 1, after adoption of poison pills, 0 otherwise)

0.366***

0.390***

0.138**

0.125**

−0.002

0.067**

Panel C. Governance characteristics and their correlations with AfterLAW

Variables

Mean

Standard deviation

25th Percentile

Median

75th Percentile

Correlation with AfterLAW

Number of directors

12.211

3.261

10.000

12.000

14.000

−0.015

% of outside directors

53.088

19.075

41.667

55.556

66.667

0.110**

Stock holdings of directors and officers (as % of shares outstanding)

8.839

13.470

1.100

2.900

9.900

−0.034

Indicator of stock option plans for outside directors (1 = yes, 0 = no)

0.084

0.278

0.000

0.000

0.000

0.152***

% of directors leaving the board before next annual meeting

8.184

9.637

0.000

7.143

13.333

0.037

CEO is not the chairman of the board (1 = true, 0 = false)

0.238

0.426

0.000

0.000

0.000

−0.034

Years as CEO

9.729

8.426

3.000

7.000

14.000

−0.024

CEO stock holdings (as % of shares outstanding)

2.743

7.860

0.040

0.170

0.965

−0.020

CEO cash pay (in thousands)

1,008.038

698.359

634.243

873.644

1,194.974

0.115**

CEO total pay (in thousands)

2,071.685

7,394.297

845.941

1,308.564

2,075.970

0.062**

CEO turnover (1 = yes, 0 = no)

0.102

0.303

0.000

0.000

0.000

0.019

Block dummy (= 1, if there exists a block of at least 5% of shares outstanding that is not directly owned by a director of the board, 0 otherwise)

0.384

0.486

0.000

0.000

1.000

0.029

*, **, and *** indicate significance at 5, 1, and 0.1% level (two-tailed), respectively

Data are from 350 Forbes 500 firms in unregulated industries over 1984–1991. Governance and ownership data are from proxy statements, 10-Ks, supplemented by Forbes annual surveys on executive compensation. Data on firms’ takeover activities are obtained by searching the Wall Street Journal Index. Data on poison pills are from Comment and Schwert’s (1995) dataset. Stock price and takeover data are from CRSP. Accounting data and states of incorporation of the firms are from COMPUSTAT and proxy statements. All monetary items are restated into constant 1991 dollars using the Consumer Price Index at the end of the fiscal year. AfterLAW is a dummy variable equal to one starting one year after the passage of the first second-generation antitakeover law by the state in which the firm is incorporated, and zero otherwise. CEO cash pay includes salary and annual bonus. CEO total pay includes salary, annual bonus, stock options, fringe benefits and cash payouts from other long-term incentive plans. Stock options are valued at the end of the fiscal year using the Black-Scholes (1973) model adjusted for dividends. The calculations are based on 2,566–2,800 observations

Table 1, Panel B, shows the correlations among the takeover threat proxies. As expected, AfterLAW is significantly positively associated with PostBC, PostPF, PostCS, and AfterPIL. However, AfterLAW is not significantly correlated with Ind_THREAT or BID_TKN.

Panel C of Table 1 presents results about governance characteristics and their correlations with takeover threat. The mean (median) values of board size and percent of outside directors are 12.11 (12) and 53.1% (55.56%), respectively. Outside directors cannot be current or former officers of the firm or relatives of corporate officers, nor can they have substantial business relationships with the firm, either personally or through their main employers. The results indicate that AfterLAW is significantly positively correlated with the percentage of outside directors, but not with board size. The result on outside directors is consistent with Mayers et al. (1997), who find that lower takeover threats are associated with a higher percentage of outside directors.

In addition, the results suggest that stock option plans for directors and director turnover (defined as the percentage of directors leaving the board before the next annual meeting) are more likely after the passage of the ATLs. The positive correlation between AfterLAW and CEO cash and total compensation is consistent with Bertrand and Mullainathan’s (2000) finding that managers receive greater compensation when they are protected by the ATLs. CEO cash pay includes salary and annual bonus. Our measure of CEO total pay is all direct compensation to the CEO, including salary, bonus, the value of annual stock option grants, cash payouts from long-term incentive plans, and other fringe benefits. We value stock options at the end of the fiscal year using the Black-Scholes (1973) model adjusted for dividends. We next extend the univariate results above to control for other factors that may influence the association between board size and firm performance.

4 Board size and firm performance

To examine the relation between board size and firm performance, prior studies (e.g., Eisenberg et al. 1998; Yermack 1996) regress firm performance on board size. Following these studies, we measure firm performance using Tobin’s Q and the accounting return on assets (ROA). Consistent with Hartzell and Starks (2003) and Agrawal and Knoeber (1996), we calculate Tobin’s Q as the ratio of the sum of market value of equity plus book value of total liabilities to the book value of total assets. We calculate ROA as income before extraordinary items divided by book value of total assets at the beginning of the fiscal year, and use Ln(Number of directors) to measure board size, both consistent with Yermack (1996). We measure board size at the end of the last fiscal year to mitigate endogeneity.

4.1 Empirical model

We are interested in the medium-run effects of state antitakeover laws on the association between board size and firm performance, because changes in internal governance mechanisms in response to exogenous shocks are likely to be gradual (Kole and Lehn 1999). To test such medium-run effects, we extend Yermack (1996) by including AfterLAW and its interaction with Ln(Number of directors). This interaction term captures the effects of the ATLs on the relations between board size and firm performance. To the degree that the ATLs exacerbate agency problems by enabling management entrenchment, we expect AfterLAW to have a negative impact on firm performance, consistent with many prior studies (e.g., Karpoff and Malatesta 1989). Following Eisenberg et al. (1998) and Yermack (1996), we expect the coefficient on Ln(Number of directors) to be negative. The Complement Hypothesis predicts a positive coefficient for AfterLAW × Ln(Number of directors)t−1because passage of the ATLs weakens the negative effect of board size on firm performance. In contrast, the Substitute Hypothesis predicts a negative coefficient for the interaction term because passage of the ATLs strengthens the negative effect of board size on firm performance.

Among the other factors likely to affect the association between board size and firm performance is board composition. We include % of outside directors, which we expect to have a positive sign, reflecting the greater independence of outside directors.7 Historical performance is likely related to current performance (Eisenberg et al. 1998; Yermack 1996), so we include lagged ROA in the regressions. Prior studies show that firm size is typically associated with board size and firm performance (e.g., Eisenberg et al. 1998; Lehn et al. 2003; Mayers et al. 1997; Yermack 1996), so we include Ln(Market value of common equity), measured at the beginning of the fiscal year, and expect it to have a positive effect on firm performance. The firm’s growth opportunities can affect both board size and firm performance (e.g., Eisenberg et al. 1998; Yermack 1996). We use the ratio of R&D expenditures to total assets of the previous fiscal year to control for the effects of growth opportunities and expect it to have a positive effect on firm performance.8 The inclusion of this ratio is important especially when Tobin’s Q is used to measure firm performance, because Tobin’s Q is also a widely used proxy for growth opportunities.

Next, firm diversification may affect not only board size, but also firm performance (Eisenberg et al. 1998; Yermack 1996). To control for diversification effects, we include Ln(Number of business segments) and expect it to have a negative effect on firm performance. Director and officer ownership may affect firm performance and directors’ incentive to monitor managers. Consistent with Yermack (1996), we include the percentage of common stock owned by directors and officers and expect it to have a positive effect on firm performance. Because CEO turnover may accompany board restructuring and weak firm performance (e.g., Kini et al. 1995; Warner et al. 1988), we include an indicator variable for whether or not there is CEO turnover in the fiscal year. However, the effect of CEO turnover on the current firm performance is unclear. To control for industry- and firm-level takeover threats, we include Ind_THREAT, BID_TKN, and AfterPIL, but we make no predictions for the effects of these variables. Finally, we include firm and year indicators to control for firm fixed effects and secular trends, which further helps mitigate the potential endogeneity in the specified relation between board size and firm performance.9

4.2 Main results

Table 2, Columns (1)–(2) present the results when firm performance is measured by Tobin’s Q. Consistent with Eisenberg et al. (1998) and Yermack (1996), the results in Column (1) of Table 2 indicate that after controlling for other determinants of firm performance, Ln(Number of directors) has a significantly negative effect on the current Q, with a coefficient of −0.573 (p = 0.000).10 However, AfterLAW does not have a significant effect on the current Q. Consistent with prior studies (Hermalin and Weisbach 2003), % of outside directors is not significantly associated with the current Tobin’s Q. As expected, Ln(Market value of common equity) and stock holdings of directors and officers are positive and statistically significant, while Ln(Number of business segments) is negative and statistically significant.
Table 2

Regressions of firm performance on state antitakeover laws and board size

Dependent variable

 

(1) Qt

(2) Qt

(3) ROAt

(4) ROAt

(5) Adj_Qt

(6) Adj_ROAt

Independent variable

Predicted sign

Estimated coefficient [p value]

AfterLAWt

−0.009 [0.798]

−0.629 [0.009]**

−0.001 [0.700]

−0.066 [0.005]**

−0.591 [0.001]***

−0.059 [0.009]**

Ln(Number of directors)t−1

−0.573 [0.000]***

−0.667 [0.000]***

−0.029 [0.031]*

−0.039 [0.008]**

−0.313 [0.002]**

−0.021 [0.093]

AfterLAWt × Ln(Number of directors)t−1

+

 

0.25 [0.007]**

 

0.026 [0.005]**

0.223 [0.002]**

0.023 [0.009]**

% of outside directorst−1

+

0.002 [0.329]

0.002 [0.359]

0 [0.282]

0 [0.306]

0.001 [0.704]

0 [0.581]

ROAt-1

+

0.133 [0.838]

0.11 [0.864]

0.049 [0.777]

0.046 [0.789]

  

Adj_ROAt-1

+

    

−0.253 [0.691]

0.008 [0.964]

Ln(Market value of common equity)t-1

+

0.291 [0.000]***

0.294 [0.000]***

0.026 [0.039]*

0.027 [0.036]*

0.191 [0.000]***

0.021 [0.045]*

R&D expenditures divided by total assetst-1

+

−0.228 [0.922]

−0.145 [0.950]

0.437 [0.421]

0.446 [0.409]

0.513 [0.826]

0.477 [0.385]

Ln(Number of business segments)t

−0.103 [0.062]

−0.1 [0.074]

0.004 [0.487]

0.005 [0.452]

0.043 [0.386]

0.003 [0.571]

Stock holdings of directors and officers (as % of shares outstanding)t

+

0.008 [0.025]*

0.008 [0.029]*

0 [0.718]

0 [0.664]

0.008 [0.018]*

0 [0.599]

CEO turnover (1 = yes, 0 = no)t

?

−0.075 [0.022]*

−0.075 [0.021]*

−0.021 [0.000]***

−0.021 [0.000]***

−0.066 [0.038]*

−0.017 [0.000]***

Ind_THREAT (dummy = 1, if industry % of number, market value, or book value of assets of firms taken over is above median, and 0 otherwise)t

?

0.017 [0.544]

0.013 [0.645]

0.001 [0.626]

0.001 [0.740]

0.02 [0.376]

0 [0.895]

BID_TKN (dummy = 1, if having an open bid or receiving a bid within a year, or taken over within two years, and 0 otherwise)t

?

−0.114 [0.078]

−0.102 [0.117]

−0.006 [0.595]

−0.005 [0.670]

−0.056 [0.388]

−0.005 [0.628]

AfterPIL (dummy = 1, after adoption of poison pills, and 0 otherwise)t

?

−0.015 [0.709]

−0.017 [0.683]

−0.005 [0.245]

−0.005 [0.229]

0.011 [0.776]

−0.004 [0.295]

Intercept

?

2.279 [0.000]***

2.461 [0.000]***

−0.084 [0.102]

−0.065 [0.206]

0.732 [0.092]

−0.112 [0.024]*

Fixed effects

 

Firm; Year

Firm; Year

Firm; Year

Firm; Year

Firm; Year

Firm; Year

Sample size

 

2199

2199

2253

2253

2199

2253

Adjusted R2

 

74.31%

74.43%

55.56%

55.74%

71.64%

47.34%

*, **, and *** indicate significance at 5%, 1%, and 0.1% level (two-tailed), respectively, based on standard errors corrected for heteroskedasticity and auto-correlations

Data are from 350 Forbes 500 firms in unregulated industries over 1984–1991. Governance and ownership data are from proxy statements, 10-Ks, supplemented by Forbes annual surveys on executive compensation. Stock price and takeover data are from CRSP. Accounting data and states of incorporation of the firms are from COMPUSTAT and proxy statements. All monetary items are restated into constant 1991 dollars using the Consumer Price Index at the end of the fiscal year. AfterLAW is a dummy variable equal to one starting one year after the passage of the first second-generation antitakeover law by the state in which the firm is incorporated, and zero otherwise. Tobin’s Q is calculated as the ratio of the sum of market value of equity and book value of liabilities to book value of assets. ROA is income before extraordinary items divided by book value of total assets at the beginning of the fiscal year. Adj_Q (Adj_ROA) is the difference between the firm’s Q (ROA) and the median value of Q (ROA) for all firms in the same two-digit SIC code industry. Firm and year fixed effects are controlled for, but not reported in the table

Consistent with the Complement Hypothesis, the results in Column (2) of Table 2 indicate that the coefficient on AfterLAW × Ln(Number of directors)t−1 is positive and statistically significant (0.25, p = 0.007), suggesting that the negative coefficient on Ln(Number of directors)t−1 decreased significantly in absolute magnitude following ATLs. The coefficient on Ln(Number of directors)t−1 is negative and statistically significant (−0.667, p = 0.000). Together, these results confirm the inverse relation between board size and Q, but establish that passage of the ATLs significantly weakened that relation. Specifically, we estimate that before passage of the laws, the combined estimated coefficient for Ln(Number of directors)t−1 for the current Q was −0.667, but following passage of the ATLS, the combined effect fell to −0.667 + 0.25 = −0.417. This change represents a 37.5% reduction in the magnitude of the negative association between board size and the current Q. The inverse relation between board size and Q remains significantly different from zero after the passage of the ATLs.

To assess the economic significance of this change in the relation between board size and the firm’s current Q, consider a firm with a median-sized board of 12 directors and a median Q of 1.387. Given this situation, decreasing the board size by one director would reduce Ln(Number of directors) by 0.087. Using the estimates in Column (2) of Table 2, this decrease leads to an increase in Q of (−0.087)*(−0.667) = 0.058, which constitutes a 4.18% improvement from the median Q of 1.387, before the passage of the state antitakeover laws. However, after passage of the ATLs, the same decrease in board size leads to an increase of only (−0.087)*(−0.417) = 0.036, a 2.62% improvement over the median Q. Thus, the change in Q associated with one less director declines by 1.56 basis points, or a decrease of (4.18−2.62%)/(4.18%) = 37.2% from 4.18% before the passage of the laws. These results suggest that the importance of board size to firm performance declines after the passage of the state ATLs.

The results for ROA, reported in Columns (3) and (4) of Table 2, are consistent with those for Q. The results in Column (3) show that the coefficient on Ln(Number of directors)t−1 for ROA is significantly negative (−0.029, p = 0.031), consistent with smaller boards being associated with better firm performance. Consistent with the Complement Hypothesis, the results in Column (4) indicate that the coefficients on AfterLAW × Ln(Number of directors)t−1 is significantly positive (0.026, p = 0.005). The coefficient on Ln(Number of directors) continues to be significantly negative (−0.039, p = 0.008). These results imply that the combined estimated coefficients for Ln(Number of directors) before versus after the passage of the laws are −0.039 and −0.039 + 0.026 = −0.013, respectively. The combined coefficient for Ln(Number of directors) is no longer significant after the passage of the ATLs. These results confirm the earlier finding that the importance of board size to firm performance is weakened by the antitakeover laws.

The results for industry-adjusted firm performance, reported in Columns (5) and (6) of Table 2, remain consistent with our predictions. Following Eisenberg et al. (1998), we use industry-adjusted performance to control for the effect of industry and general economic conditions. Specifically, Adj_Q (Adj_ROA) is the difference between the firm’s Q (ROA) and the median value of Q (ROA) for all firms in the same two-digit SIC code industry. Similar to those results reported in Columns (2) and (4), the coefficient on Ln(Number of directors) is significantly negative for Adj_Q (−0.313, p = 0.002), while the coefficient on AfterLAW × Ln(Number of directors) is significantly positive for both Adj_Q (0.223, p = 0.002) and Adj_ROA (0.023, p = 0.009). Overall, the results in Table 2 are consistent with the Complement Hypothesis that the passage of the ATLs weakens the negative association between board size and firm performance.11

4.3 Sensitivity checks

We now describe the results (not reported in the paper) of sensitivity checks which indicate that the results in Table 2 are generally robust. First, the effectiveness of ATLs likely depends on other antitakeover mechanisms in place. Prior studies suggest that poison pills are a powerful antitakeover provision (Comment and Schwert 1995). Indeed, Karpoff and Malatesta (1989) find that stock values decrease following ATLs, but this decrease is significant only for firms without poison pills, suggesting that other ATLs may be redundant in the presence of poison pills. To fully control for this potential effect of poison pills, we extend our main empirical model above to include two new terms: AfterPIL × Ln(Number of directors) and AfterPIL × AfterLAW × Ln(Number of directors). If poison pills affect the impact of the ATLs on the importance of board size, the three-way interaction term should capture this effect. The results show that the coefficient on AfterLAW × Ln(Number of directors) is marginally significantly positive (0.198, p = 0.067), while the coefficients on AfterPIL × Ln(Number of directors) (0.122, p = 0.285) and AfterPIL × AfterLAW × Ln(Number of directors) (0.023, p = 0.201) are both insignificant. These results suggest that the effect of the ATLs on the importance of board size is not contingent on the existence of poison pills.12

The Table 2 results assume that the ATLs became effective in the year following passage of the first law. To consider alternative timing assumptions, we first exclude the firm-years in which the first law was passed and find similar results. For example, the coefficient on AfterLAW × Ln(Number of directors) remains significantly positive for Q (0.298, p = 0.008). Next, we recode AfterLAW to be one in the year of the law’s passage and thereafter. After this recoding, the coefficient on AfterLAW × Ln(Number of directors) continues to be significantly positive for Q (0.278, p = 0.008). Finally, Bertrand and Mullainathan (1999) argue that Business Combination (BC) laws were the most powerful of the three forms of second-generation ATLs. We replicate our tests replacing AfterLAW by PostBC, an indicator equal to one after the passage of the BC laws. We find that the coefficient on PostBC × Ln(Number of directors) is significantly positive (0.254, p = 0.006) for Q, similar to the results in Column (2) of Table 2.

Because firms in any particular state are subject to the same ATLs, observations from such firms are likely to be correlated, leading to inflated significance of the independent variables. This potential problem may be particularly severe for Delaware, where more than half of our sample firms are incorporated. We take two steps to address this issue. First, we implement generalized linear models (GLM) clustering on incorporate states. We find that our results are robust to correcting state group effects. For example, the coefficients on AfterLAW × Ln(Number of directors), as reported in Column (2) of Table 2, are still significant at the 1% level. Second, we exclude all firms incorporated in Delaware from the analyses and find similar results. The coefficient on AfterLAW × Ln(Number of directors) remains significantly positive (0.322, p = 0.017) for Q.13

Because we are primarily interested in the medium-run effects of the ATLs, our previous analyses focused on several years before versus after the passage of the laws. As an alternative, we use only 1 year immediately before and 1 year immediately after the passage of the laws. This approach identifies the transient effect of the laws, while at the same time eliminating repeated observations from any one firm, further mitigating potential auto-correlation issues. The results again remain similar to those reported in Table 2. For example, the coefficient on AfterLAW × Ln(Number of directors) is significantly positive for ROA (0.026, p = 0.022) based on 552 observations.

5 Additional results

The results discussed above show that passage of the ATLs is associated with a weakened inverse relation between board size and firm performance, suggesting that the importance of board size declined following the passage of the ATLs. These results are consistent with the Complement Hypothesis. We next examine whether the ATLs affect the importance of board size in determining firm performance by changing the relation between board size and board actions. We also examine whether changes in the importance of board size following the passage of the ATLs are accompanied by different board sizes. We address these issues to provide additional empirical evidence on the impact of the ATLs.

5.1 CEO compensation

Prior studies examine observable board actions, such as the design of CEO compensation and CEO turnover, to infer board effectiveness (e.g., Weisbach 1988; Yermack 1996). In the same spirit, we examine the design of CEO compensation as an action controlled by the board, and we adopt Yermack’s (1996) focus on the association between board size and CEO PPS. As in prior studies (e.g. Jensen and Murphy 1990), we measure CEO PPS by regressing changes in CEO compensation on unexpected firm performance, measured as annual stock returns (RET).14 Since CEO compensation is skewed as shown in Table 1, Panel B, we use the logarithmic transformation of CEO compensation as the dependent variable (e.g., Sloan 1993). Following Yermack (1996), we include the interaction between board size and stock returns, i.e., Ln(Number of directors)t-1 × RETt, to capture the effect of board size on the relation between CEO pay and firm performance. Consistent with the notion that smaller boards are more effective in monitoring top management, we expect the coefficient on Ln(Number of directors)t-1 × RETt to be negative, meaning that CEO PPS decreases with board size.

Our regressions include several control variables. We include % of outside directors, which we expect to have a negative effect on the changes in CEO compensation. Consistent with Hartzell and Starks (2003), we also include lagged Ln(Market value of common equity) and lagged Q to control for the effects of firm size and growth opportunities. We expect Ln(Market value of common equity) to have a positive coefficient, but we have no prediction for the sign of the lagged Q. We include Ind_THREAT, BID_TKN, and AfterPIL to control for the effects of industry- and firm-level takeover activities (Agrawal and Knoeber 1998). In addition, we use year indicators to control for secular trends.

To examine the impact of the passage of the ATLs on the relation between board size and CEO pay-performance sensitivity, we split the sample into two sub-samples based on the passage of the first ATL. This approach facilitates interpreting three-way interaction terms. Columns (1) and (2) of Table 3 report the results before and after the passage of the ATLs, respectively. The results in Column (1) show that before the passage of the laws, the coefficient on Ln(Number of directors)t-1 × RET t is −0.259 (p = 0.072). After passage of the ATLs, this coefficient becomes 0.478 (p = 0.197), as shown in Column (2) of Table 3. According to Columns (3) and (4), the results are similar when we use abnormal stock returns (XRET) to measure the unexpected firm performance, where XRET is defined as the difference between the firm’s annual stock return and the return predicted by the capital asset pricing model (CAPM). Largely consistent with the Complement Hypothesis, these results provide modest evidence that board size influences CEO PPS only before the passage of the laws, confirming our earlier conclusion that passage of the ATLs reduced the importance of board size.15
Table 3

Regressions of CEO compensation on board size and firm performance: before versus after the passage of state antitakeover laws

Dependent variable

 

Ln(CEO total pay)t − Ln(CEO total pay)t−1

Independent variable

Predicted sign

Estimated coefficient [p value]

(1) AfterLAW = 0

(2) AfterLAW = 1

(3) AfterLAW = 0

(4) AfterLAW = 1

Annual stock returns (RET)t

+

1.014 [0.014]*

−0.726 [0.276]

  

Abnormal stock returns (XRET)t

  

1.289 [0.012]*

−0.579 [0.412]

Ln(Number of directors)t−1 × RETt

+

−0.259 [0.072]

0.478 [0.197]

  

Ln(Number of directors)t−1 × XRETt

-

  

−0.324 [0.063]

0.449 [0.142]

Ln(Number of directors)t-1

-

0.057 [0.576]

−0.073 [0.420]

−0.022 [0.808]

0.002 [0.982]

% of outside directorst−1

-

0 [0.631]

0.001 [0.155]

0 [0.567]

0.001 [0.167]

Ln(Market value of common equity)t-1

+

0.008 [0.709]

0.007 [0.740]

0.013 [0.565]

0.011 [0.608]

Qt−1

?

−0.045 [0.108]

−0.007 [0.779]

−0.054 [0.064]

−0.013 [0.609]

Ind_THREAT (dummy = 1, if industry % of number, market value, or book value of assets of firms taken over is above median, and 0 otherwise)t

?

−0.006 [0.890]

0.014 [0.766]

−0.009 [0.824]

0.003 [0.940]

BID_TKN (dummy = 1, if having an open bid or receiving a bid within a year, or taken over within two years, and 0 otherwise)t

?

−0.099 [0.361]

0.084 [0.545]

−0.098 [0.361]

0.09 [0.521]

AfterPIL (dummy = 1, after adoption of poison pills, and 0 otherwise)t

?

−0.061 [0.233]

0.041 [0.323]

−0.055 [0.285]

0.047 [0.258]

Intercept

?

−0.144 [0.536]

0.207 [0.320]

0.141 [0.512]

0.116 [0.569]

Fixed effects

 

Year

Year

Year

Year

Sample size

 

974

994

974

994

Adjusted R2

 

6.33%

4.06%

7%

3.84%

*, **, and *** indicate significance at 5, 1, and 0.1% level (two-tailed), respectively, based on standard errors corrected for heteroskedasticity and auto-correlations

Data are from 350 Forbes 500 firms in unregulated industries over 1984–1991. Governance and ownership data are from proxy statements, 10-Ks, supplemented by Forbes annual surveys on executive compensation. Stock price and takeover data are from CRSP. Accounting data and states of incorporation of the firms are from COMPUSTAT. All monetary items are restated into constant 1991 dollars using the Consumer Price Index at the end of the fiscal year. AfterLAW is a dummy variable equal to one starting one year after the passage of the first second-generation antitakeover law by the state in which the firm is incorporated, and zero otherwise. CEO total pay includes salary, annual bonus, stock options and fringe benefits and cash payouts from other long-term incentive plans. Stock options are valued at the end of the fiscal year using the Black-Scholes (1973) model adjusted for dividends. Tobin’s Q is calculated as the ratio of the sum of market value of equity and book value of liabilities to book value of assets. XRET is the firm’s annual stock return minus the return predicted by the capital asset pricing model (CAPM). Firm-years are excluded if the previous year’s compensation data relates to a different CEO than the CEO of the current year. Year fixed effects are controlled for, but not reported in the table

5.2 Board size

To examine the impact of the ATLs on board size, we regress board size on AfterLAW with the expectation that the indicator variable will have a positive sign according to the Complement Hypothesis and a negative sign according to the Substitute Hypothesis. We next describe several other control variables that we include in the analysis of board size.

Because board size exhibits path dependence (e.g., Lehn et al. 2003), we control for the lagged board size and expect it to have a positive effect on the current board size. Eisenberg et al. (1998), Lehn et al. (2003), and Mayers et al. (1997) all find that firm size is positively associated with board size. Based on these studies, we include lagged Ln(Market value of common equity) and expect it to have a positive effect on board size. Consistent with Yermack (1996) and Lehn et al. (2003), we expect a negative association between board size and growth opportunities, measured by Tobin’s Q of the last fiscal year. We include Ind_THREAT, BID_TKN, and AfterPIL to control for industry- and firm-level takeover threats. We also control for firm and year fixed effects.

Table 4, Columns (1) and (2) present the results for board size. Consistent with the Complement Hypothesis, the results in both columns indicate that AfterLAW has a positive coefficient of 0.015, with p-values of 0.052 and 0.047, respectively. This implies that on average, board size is 1.5% smaller before versus after the passage of the ATLs. A reduction of 0.015 in Ln(Number of directors) before the passage of the laws translates to an estimated increase of 0.015*0.667 = 0.01 basis point in the firm’s current Q, which reflects an increase of 0.72% above the median Q of 1.387. This result suggests that the 1.5% change in board size leads to a non-trivial effect on economic performance.
Table 4

Regressions of board size and composition on state antitakeover laws

Dependent variable

Ln(Number of directors)t

% of outside directorst

Independent variable

Predicted sign

Estimated coefficient [p value]

Predicted sign

Estimated coefficient [p value]

(1)

(2)

(3)

(4)

AfterLAWt

0.015 [0.052]

0.015 [0.047]**

+

0.247 [0.587]

0.248 [0.585]

Ln(Number of directors)t-1

+

0.4 [0.000]***

0.399 [0.000]***

   

% of outside directorst-1

   

+

0.543 [0.000]***

0.543 [0.000]***

Ln(Market value of common equity)t−1

+

0.036 [0.000]***

0.035 [0.000]***

?

−1.449 [0.019]*

−1.481 [0.027]*

Tobin’s Qt−1

−0.023 [0.000]***

−0.023 [0.000]***

?

0.688 [0.046]*

0.691 [0.049]*

Return on assets (ROA)t−1

?

 

0 [0.990]

?

 

0.472 [0.845]

Ln(Number of business segments)t

?

 

−0.011 [0.462]

?

 

0.26 [0.759]

CEO turnover (1 = yes, 0 = no)t

?

 

0.013 [0.115]

?

 

−0.161 [0.737]

Ind_THREAT (dummy = 1, if industry % of number, market value, or book value of assets of firms taken over is above median, and 0 otherwise)t

?

−0.006 [0.244]

−0.006 [0.273]

?

−0.072 [0.815]

−0.065 [0.834]

BID_TKN (dummy = 1, if having an open bid or receiving a bid within a year, or taken over within 2 years, and 0 otherwise)t

?

−0.006 [0.667]

−0.007 [0.575]

?

0.432 [0.522]

0.457 [0.498]

AfterPIL (dummy = 1, after adoption of poison pills, and 0 otherwise)t

?

−0.021 [0.002]**

−0.022 [0.001]***

?

0.262 [0.537]

0.287 [0.499]

Intercept

?

1.359 [0.000]***

1.345 [0.000]***

?

44.415 [0.000]***

46.283 [0.000]***

Fixed effects

 

Firm; Year

Firm; Year

 

Firm; Year

Firm; Year

Sample size

 

2220

2217

 

2220

2217

Adjusted R2

 

90.08%

90.08%

 

91.82%

91.8%

*, **, and *** indicate significance at 5, 1, and 0.1% level (two-tailed), respectively, based on standard errors corrected for heteroskedasticity and auto-correlations

Data are from 350 Forbes 500 firms in unregulated industries over 1984–1991. Governance and ownership data are from proxy statements, 10-Ks, supplemented by Forbes annual surveys on executive compensation. Stock price and takeover data are from CRSP. Accounting data and states of incorporation of the firms are from COMPUSTAT and proxy statements. All monetary items are restated into constant 1991 dollars using the Consumer Price Index at the end of the fiscal year. AfterLAW is a dummy variable equal to one starting one year after the passage of the first second-generation antitakeover law by the state in which the firm is incorporated, and zero otherwise. Tobin’s Q is calculated as the ratio of the sum of market value of equity and book value of liabilities to book value of assets. Firm and year fixed effects are controlled for, but not reported in the table

Consistent with prior studies (e.g., Mayers et al. 1997; Lehn et al. 2003), the results in both columns show that the coefficients on lagged Ln(Number of board directors) and Ln(Market value of common equity) are significantly positive, and the coefficient on lagged Q is significantly negative. Moreover, as shown in Column (2) of Table 4 the results are robust to the addition of several other variables, including proxies for historical performance (lagged ROA), firm diversification (Ln(Number of business segments)), and management change (CEO turnover).

Finally, for completeness, we use similar regressions to estimate the impact of the antitakeover laws on the % of outside directors. The results in Columns (3) and (4) of Table 4 indicate that the passage of the ATLs do not have a significant effect on board composition. Similarly, we find no evidence that the ATLs affect board director turnover or the likelihood of the adoption of stock option plans for outside directors.

6 Conclusion

This study examines whether the importance of board size changed after the introduction of the second-generation antitakeover laws in the mid-to-late 1980s. These laws represent exogenous shocks that reduced the takeover threats facing directors and managers. Our results from a sample of 350 unregulated Forbes 500 firms over 1984–1991 indicate that the negative association between board size and firm performance becomes weaker after passage of the ATLs. The same pattern largely holds for the association between board size and CEO pay-performance sensitivity, which moves from being negative and marginally significant before the ATLs to being not significantly different from zero after passage of the laws. We also document that boards become larger after the passage of the laws.

These results are consistent with the Complement Hypothesis that takeover threats increase the importance of board size in corporate governance. The results reflect an active market for corporate control increasing the importance of rapid board decisions and directors’ incentives to monitor effectively. However, as board size increases, the timeliness of board decisions is likely to decline, and directors’ free-riding incentives grow stronger.

Our findings contribute to the literature in two ways. First, we provide empirical evidence that the intensity of the market for corporate control increases the importance of board size. Our results advance our understanding of three important empirical issues regarding boards that motivated prior empirical studies (Hermalin and Weisbach 2003). These are, (1) the relation between board characteristics and firm profitability, (2) the relation between board characteristics and the quality of observable board decisions, and (3) the determinants of board characteristics. For the first issue, we show that small size boards have a greater positive influence on profitability when takeover intensity is high. For the second issue, we demonstrate that smaller boards’ ability to establish stronger CEO pay-performance sensitivity is enhanced by the intensity of the market for corporate control. For the third issue, we provide evidence that larger boards are associated with the ATLs that reduced takeover intensity.

In addition, our results extend prior studies of antitakeover legislation and board composition. Prior studies examine the effect of the antitakeover laws on firms’ capital structure (Garvey and Hanka 1999) and managerial ownership (Cheng et al. 2005). Prior studies also examine the relation between takeover threats and board composition (e.g., Brickley and James 1987; Mayers et al. 1997) or directors’ incentives (e.g., Harford 2003; Kini et al. 1995). We document the effect of the ATLs on board size and its importance to firm performance and the quality of board decisions, shedding light on the dynamics of the role of board size in corporate governance.

A limitation of our analysis is our exclusive focus on takeover threats as the mechanism that determines directors’ incentives to monitor. The literature suggests that the labor market for directors and the product market also discipline directors (Fama 1980; Jensen 1993). Further studies may refine our results by including additional mechanisms that affect board effectiveness. Second, to the extent that the negative association between board size and firm performance is an equilibrium phenomenon, some further unidentified factors must account for this association (Hermalin and Weisbach 2003). Our results appear to suggest that the market for corporate control may play a role in this process, but like prior studies, we are unable to distinguish between the equilibrium and the out-of-equilibrium interpretations.

Footnotes

  1. 1.

    Comment and Schwert (1995) find that ATLs are associated with a significant increase in takeover premiums. Karpoff and Malatesta (1989) find that ATLs significantly reduce stock values of firms covered by the ATLs.

  2. 2.

    Magnet (1992, p. 86): The corporate governance system doesn't need rebuilding from the ground up; the existing machinery just needs to be switched on. Even so, getting boards to stop snoozing, and even if need be to turn activist, isn't easy. After all, why should they? A culture of quietism reigns in many boardrooms. Says Calpers General Counsel Richard Koppes, "One of the problems is, it's not polite to ask questions".

  3. 3.

    Hermalin and Weisbach (2003) describe how investors’ perspectives on board size represent another important indication of its significant role in the governance process.

  4. 4.

    The impact of takeover threats on the importance of board composition is unclear, however, as reflected in the mixed evidence of prior studies (e.g., Brickley and James 1987; Mayers et al. 1997).

  5. 5.

    The results are robust to this exclusion.

  6. 6.

    The results are the same if we define Ind_THREAT as equal to one if all of the industry specific ratios exceed the corresponding median ratio.

  7. 7.

    The inclusion of the interaction term AfterLAW × % of outside directorst-1, which is insignificant, does not change the results.

  8. 8.

    The results remain unchanged when we use capital expenditures divided by total assets to measure growth opportunities.

  9. 9.

    The results are similar if we control for industry fixed effects rather than firm fixed effects.

  10. 10.

    The reported p-values are based on standard errors corrected for heteroskedasticity and auto-correlations.

  11. 11.

    The results reported in Table 2 are robust to adding indicator variables for the presence of stock option plans for directors and non-director blockholders of at least 5% of shares outstanding. Likewise, the results remain qualitatively unchanged when the models are estimated using median regressions, which are less sensitive to outliers.

  12. 12.

    We also include the interactions of Ind_THERAT and BID_TKN, with Ln(Number of directors) and find these interactions to be insignificant as well. A possible reason for this insignificance is that BID_TKN, and AfterPIL are likely endogenous. While Ind_THREAT is exogenous to the firm, Panel B of Table 1 shows that this variable remained stable over the sample period. Thus, the insignificance of these interaction terms further justifies our approach of using passage of the ATLs as an exogenous shock to takeover threats. Furthermore, the inclusion of interactions between the indicator that a state passed an ATL over the sample period and Ln(Number of directors), or between year indicators and Ln(Number of directors) does not change the main results.

  13. 13.

    Generalized linear models clustering on incorporation states generate similar results. Furthermore, adding an indicator variable for firms incorporated in Delaware to the model in Column (2) of Table 2 yields very similar results.

  14. 14.

    We find similar results when unexpected performance is measured as the change in the accounting rate of return on assets (ΔROA). In addition, the changes specifications control for firm-fixed effects.

  15. 15.

    We find no evidence that antitakeover laws affect the importance of board size on the relation between CEO turnover and firm performance. One possible explanation for the different finding for CEO compensation versus CEO turnover is that the two serve as substitute CEO incentive mechanisms.

Notes

Acknowledgements

We thank Jerry Davis, DJ Nanda, Cindy Schipani, and participants at the University of Michigan Interdisciplinary Workshop on Corporate Governance and the 2005 American Accounting Association Management Accounting Research Conference. We are very thankful to David Yermack for sharing his governance data and to William Schwert for making his poison pill dataset publicly available.

References

  1. Agrawal A, Knoeber C (1996) Firm performance and mechanisms to control agency problems between managers and shareholders. J Financ Quant Anal 31:377–397CrossRefGoogle Scholar
  2. Agrawal A, Knoeber C (1998) Managerial compensation and the threat of takeover. J Financ Econ 47:219–239CrossRefGoogle Scholar
  3. Barnhart S, Spivey M, Alexander J (2000) Do firm and state antitakeover provisions affect how well CEOs earn their pay? Manage Decis Econ 21:315–328CrossRefGoogle Scholar
  4. Bertrand M, Mullainathan S (1999) Is there discretion in wage setting? A test using takeover legislation, RAND J Econ 30:535–554CrossRefGoogle Scholar
  5. Bertrand M, Mullainathan S (2000) Agents with and without principals. Am Econ Rev 90(2):203–208CrossRefGoogle Scholar
  6. Bertrand M, Mullainathan S (2003) Enjoying the quiet life? Managerial behavior following antitakeover legislation. J Pol Econ 111:1043–1075CrossRefGoogle Scholar
  7. Black F, Scholes M (1973) The pricing of options and corporate liabilities. J Pol Econ 81:637–654CrossRefGoogle Scholar
  8. Brickley J, James C (1987) The takeover market, corporate board composition, and ownership structure: the case of banking. J Law Econ 30:161–181CrossRefGoogle Scholar
  9. Cheng S, Nagar V, Rajan M (2005) Identifying control motives in managerial ownership: evidence of antitakeover legislation. Rev Finan Stud 18:637–672CrossRefGoogle Scholar
  10. Comment R, Schwert G (1995) Poison or placebo? Evidence on the deterrent and wealth effects of modern antitakeover measures. J Finan Econ 39:3–43CrossRefGoogle Scholar
  11. Eisenberg T, Sundgree S, Wells M (1998) Larger board size and decreasing firm value in small firms. J Finan Econ 48:35–54CrossRefGoogle Scholar
  12. Fama E (1980) Agency problems and theory of the firm. J Pol Econ 88:288–307CrossRefGoogle Scholar
  13. Garvey G, Hanka G (1999) Capital structure and corporate control: the effect of antitakeover leverage statutes. J Finance 54:519–546CrossRefGoogle Scholar
  14. Harford J (2003) Takeover bids and target directors’ incentives: the impact of a bid on directors’ wealth and board seats. J Finan Econ 69:51–83CrossRefGoogle Scholar
  15. Hartzell J, Starks L (2003) Institutional investors and executive compensation. J Finance 58:2351–2374CrossRefGoogle Scholar
  16. Hermalin B, Weisbach M (2003) Boards of directors as an endogenously determined institution: a survey of the economic literature. Econ Pol Rev 9:7–26Google Scholar
  17. Holmstrom B, Kaplan S (2001) Corporate governance and merger activity in the US: making sense of 1980s and 1990s. J Econ Perspect 15:121–144CrossRefGoogle Scholar
  18. Huson M, Parrino R, Starks L (2001) Internal monitoring mechanisms and CEO turnover: a long-term perspective. J Finance 56:2265–2297CrossRefGoogle Scholar
  19. Jensen M (1993) The modern industrial revolution, exit and the failure of internal control systems. J Finance 48:831–880CrossRefGoogle Scholar
  20. Jensen M, Murphy KJ (1990) Performance pay and top management incentives. J Pol Econ 98:225–264CrossRefGoogle Scholar
  21. John K, Senbet L (1998) Corporate governance and board effectiveness. J Bank Finance 22:371–403CrossRefGoogle Scholar
  22. Karpoff J, Malatesta P (1989) The wealth effects of second-generation state takeover legislation. J Financ Econ 25:291–322CrossRefGoogle Scholar
  23. Kini O, Kracaw W, Mian S (1995) Corporate takeovers, firm performance, and board composition. J Corp Finance 1:383–412CrossRefGoogle Scholar
  24. Kole S, Lehn K (1999) Deregulation and the adaptation of governance structure: the case of the U.S. airline industry. J Financ Econ 52:79–117CrossRefGoogle Scholar
  25. Lehn K, Patro S, Zhao M (2003) Determinants of the size and structure of corporate boards: 1935–2000. Working paper. University of Pittsburgh and Bentley CollegeGoogle Scholar
  26. Lipton M, Lorsch J (1992) A modest proposal for improved corporate governance. Bus Lawyer 48:59–77Google Scholar
  27. Magnet M (1992) Directors, wake up! Fortune 15:85–92Google Scholar
  28. Mayers D, Shivadashani A, Smith C (1997) Board composition and corporate control: evidence from the insurance industry. J Business 70:33–62CrossRefGoogle Scholar
  29. Mikkelson W, Partch M (1997) The decline of takeovers and disciplinary managerial turnover. J Financ Econ 44:205–228CrossRefGoogle Scholar
  30. Mitchell M, Mulherin JH (1996) The impact of industry shocks on takeover and restructuring activity. J Financ Econ 41:193–229CrossRefGoogle Scholar
  31. Morck R, Shleifer A, Vishny R (1989) Alternative mechanisms for corporate control. Am Econ Rev 79:842–852Google Scholar
  32. Shleifer A, Vishny R (1997) A survey of corporate governance. J Finance 52:737–783CrossRefGoogle Scholar
  33. Sloan RG (1993) Accounting earnings and top executive compensation. J Account Econ 16:55–100CrossRefGoogle Scholar
  34. Warner J, Watts R, Wruck K (1988) Stock prices and top management changes. J Financ Econ 20(1–2):461–492CrossRefGoogle Scholar
  35. Weisbach M (1988) Outside directors and CEO turnover. J Financ Econ 20:413–460CrossRefGoogle Scholar
  36. Williamson OE (1983) Organizational form, residual claimants, and corporate control. J Law Econ 26:351–366CrossRefGoogle Scholar
  37. Yermack D (1995) Do corporations award CEO stock options effectively? J Financ Econ 39:237–269CrossRefGoogle Scholar
  38. Yermack D (1996) Higher market valuation of companies with a small board of directors. J Financ Econ 40:185–211CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2007

Authors and Affiliations

  • Shijun Cheng
    • 1
  • John H. EvansIII
    • 2
  • Nandu J. Nagarajan
    • 2
  1. 1.Robert H. Smith School of BusinessUniversity of MarylandCollege ParkUSA
  2. 2.Joseph M. Katz Graduate School of BusinessUniversity of PittsburghPittsburghUSA

Personalised recommendations