The Value of Takeover Defenses and the Interaction Effects of Firm Characteristics

Abstract

The value of takeover defenses changes with different firm characteristics. In the paper, I examine the interactions of firm characteristics with the value of takeover defenses. The results show that the relationship between firm value and takeover defenses differs in firm age, monitoring costs, advising needs, and their interactions. Takeover defenses are not necessarily more detrimental for older firms. Instead, takeover defenses are more harmful for older firms with higher monitoring costs, but the adverse impact is positively moderated for older firms with higher advising needs. Thus, the influence of firm age on the value of takeover defenses depends on a firm’s monitoring costs and advising needs. The findings of the paper present consistent evidence that takeover defenses have a heterogeneous impact for firms with different firm characteristics.

1. Introduction

Many commentators and shareholder activists claim that takeover defenses insulate firms from hostile takeover attempts, thereby entrenching management. The consequent shirking, empire building, and extraction of private benefits by incumbents reduce firm value. Policy makers and ESG advocates thus support the notion to abolish some takeover defenses, especially classified board structure, to enhance shareholders’ value.

In theory, takeover defenses either destroy or enhance firm value. On one side, takeover defenses entrench managers by insulating them from the external disciplinary activities and increasing the agency costs (DeAngelo and Rice 1983; Harford 1999; Gompers et al. 2003; Masulis et al. 2007). Conversely, takeover defenses are beneficial by encouraging long-term investment and bonding with important customers (Stein 1988; Adams et al. 2009; Cremers and Ferrell 2014; Cremers et al. 2017; Field and Lowry 2022). Previous empirical tests provide a mixed evidence regarding the impact of takeover defenses on firm value (see for the summary of the literature, Burkart and Panunzi 2008; Straska and Waller 2014; Karpoff and Wittry 2024).

More recently, studies suggest that the benefits and costs of takeover defenses differ in firm specific characteristics, including monitoring costs, advising needs, operational complexity, and outside bonding needs (Ahn and Shrestha 2013; Cremers et al. 2017; Field and Lowry 2022). Johnson et al. (2022) further argue that the costs and benefits of takeover defenses change as a firm ages.

Following the recent arguments, I explore the heterogeneous nature of the relationship between takeover defenses and firm value by examining the interactions of firm age, monitoring costs, and advising needs. The setting of the paper differs from previous studies in that it examines the potential interactions and the moderating effects of all three firm characteristics factors. In addition, I utilize the panel data set and include the entire universe of the US firms available in the RiskMetrics directors and governance databases from 1996 to 2006. This contrasts with Johnson et al. (2022), who examine a subset of firms that went through IPOs recently.

The existing literature lacks conclusive evidence on the valuation effect of takeover defenses. It is also less clear about the nature and shape of the interactions with firm characteristics. This paper fills the gap and sheds a light on the issue by providing additional evidence on the value of takeover defenses with the interactions of firm characteristics. In this paper, I examine the heterogeneous effects of takeover defenses on firm value. Specifically, this paper tests that the value of takeover defenses changes with firm age and a firm’s monitoring costs and advising needs. The entrenchment index (the E-index) is used to measure the level of takeover defenses adopted by a firm. Proposed by Bebchuk et al. (2009), the E-index includes six takeover defenses that are particularly important to shareholders’ value: classified boards, poison pills, golden parachutes, limits on shareholder amendments of firm bylaws, supermajority requirements to change the firm charter, and supermajority rules to approve mergers. Firm value is estimated with Tobin’s q. A firm’s monitoring costs are measured by asset intangibility, and its advising needs are measured by firm size.

In multivariate tests, the relationship between Tobin’s q and the E-index is examined, including industry- and year-fixed effects, and a set of control variables that represent investment opportunities and profitability. The results show that the E-index is negatively associated with firm value, which is consistent with the findings in the previous studies. The main focus of this paper is the interaction effects of firm age, monitoring costs, and advising needs. The coefficient on firm age is negative and significant.

More importantly, the E-index and firm age interact positively, indicating that takeover defenses become less detrimental as the firm ages. This contrasts with the findings in Johnson et al. (2022), who report that the relationship between Tobin’s q and a firm’s use of defenses decreases as the firm ages. The discrepancy is attributable to the difference in sample firms used. The firm age in my sample ranges from 5 years to 72 years in the 5th and 95th percentiles, while most of the sample firms in Johnson et al. (2022) are less than 5 years old. Long-survived firms had sufficient time to optimally adjust their governance structures, and thus the impact of firm age is less negative and could even be positive. Johnson et al. (2022) argue that the value-reversal pattern persists in fixed samples of surviving firms and thus is robust to selection effects. Considering the maximum firm age used in their sample is only 17 years, the positive coefficient on the interaction of the E-index and firm age suggests that it takes a really long time for a firm to adjust its governance structure.

Further analysis shows that the impact of firm age is detrimental for firms with high monitoring costs, but it is positively moderated for firms with high advising needs. These findings suggest that the value of takeover defenses changes in a complex manner as firm age, monitoring costs, and advising needs interact each other. One might argue that a firm could adjust takeover defenses to optimize firm value. In that case, I would not expect any significant relationship between takeover defenses and firm value and the interaction effects of firm characteristics. The presence of the costs associated with altering existing governance structures suggests that some firms maintain suboptimal governance structures in the foreseeable periods.

This paper also utilizes a dummy variable approach and divides sample firms based on the median values of firm age (Old = 1 if firm age is greater than the median value and 0, otherwise), monitoring costs (Monitor = 1 if asset intangibility is greater than the median value and 0, otherwise), and advising needs (Advise = 1 if firm size is greater than the median value and 0, otherwise). The multivariate tests with these dummy variables and their interactions yield results similar to those with continuous variables. The interaction of the E-index and Old is positive and significant in all models. Examining the triple interactions shows that the value of takeover defenses is negative for firms with the above medians of monitoring costs and firm age. The value of takeover defenses is positive for firms with the above medians of advising needs and firm age. The result confirms that firm age interacts with the firm’s monitoring costs and advising needs.

An endogeneity concern is addressed with two additional tests. First, the OLS regression is re-estimated with additional control variables including CEO ownership and its squared term, institutional shareholdings, and CEO incentives with delta and vega. Second, a two-stage least-squared regression (2SLS) is used to address the endogeneity concern with the econometric modeling. Following Ahn and Shrestha (2013), in the first stage regression, the E-index is estimated with an instrument, Massachusetts incorporation dummy variable, along with other control variables. The estimated value of the E-index (Instrumented E-index) is introduced in the second-stage estimation regression. Although the signs of the coefficients estimated are in opposite directions in some cases, I continue to find the interaction effects of the E-index and the three firm characteristics variables.

In sum, the findings of this paper present a consistent evidence that the value of takeover defenses changes with firm age, monitoring costs, advising needs, and their interactions. The impact of takeover defenses on firm value is beneficial for some firms and is detrimental for other firms. The heterogeneous nature of the relationship implies that multiple factors intertwined to influence the value of takeover defenses. The evidence complements a growing body of literature that asserts “one size does not fit all” when it comes to the corporate governance structure (Bhagat and Black 2001; Boone et al. 2007; Coles et al. 2008; Lehn et al. 2009; Chi and Lee 2010; Duchin et al. 2010; Straska and Waller 2014; Johnson et al. 2022).

The remainder of this paper is organized as follows. Section 2 discusses the related literature and develops testable hypotheses. Section 3 explains the sample selection procedure and describes key variables used in the empirical analysis. Section 4 presents the empirical tests and interprets the results. Section 5 concludes.

2. Previous Literature and Hypothesis

2.1. Previous Studies

Arguably, takeover defenses affect firm value in either way. Gompers et al. (2003), among others, showed that firm value is negatively related to takeover defenses. This negative relationship is consistent with the view that takeover defenses entrench managers and thus destroy firm value (DeAngelo and Rice 1983; Gompers et al. 2003; Masulis et al. 2007; Bebchuk et al. 2009).

The opposing view holds that takeover defenses encourage long-term investment and bond relationships with important counterparties. Thus, takeover defenses help firms to enhance value and performance (Cen et al. 2016; Cremers and Ferrell 2014; Cremers et al. 2017).

Netting the potential benefits and costs of takeover defenses, some argue that the impact of takeover defenses on firm value differs in each firm depending on the firm’s characteristics (Ahn and Shrestha 2013; Cremers et al. 2017; Johnson et al. 2022). Ahn and Shrestha (2013) find that the effect of classified boards on firm value differs in firms with different monitoring costs and advisory needs. Cremers et al. (2017) also document a positive effect of classified boards by promoting long-term investments in the relationship-specific and innovative projects. Because classified boards are the most powerful takeover deterrent of all, as Bates et al. (2008) and Catan and Kahan (2016) argue, their findings can be extended to other takeover defenses.

Recently, Johnson et al. (2022) show that while the bonding benefits provided by takeover defenses are important for young firms, the potential costs of takeover defenses are low for young firms with large managerial ownership. Johnson et al. (2022) argue that the costs of takeover defenses increase as firms get old, and the costs are associated with lower cash value, diversifying acquisitions, and CEO-board chair duality. These arguments imply that takeover defenses are associated with lower firm value for old firms.

Conversely, the complexity of a firm’s operations increases with firm age. As a result, the advisory role of outside directors becomes more important in old firms (Boone et al. 2007). Takeover defenses may enhance managerial incentives to invest in firm-specific human capital. In addition, takeover defenses dissuade opportunistic bidding for firms with high opacity (DeAngelo and Rice 1983; Stein 1988). As a result, takeover defenses provide directors with a degree of permanence to invest in firm-specific human capital. This line of arguments suggests that the value of takeover defenses is beneficial for older firms.

For the relationship of firm value and corporate boards, Fama and Jensen (1983) and Hermalin and Weisbach (1998) argue that boards play the important role of monitoring and advising managers. Takeover defenses could affect the efficacy of monitoring and advisory functions of boards. Coles et al. (2008) document that the relationship between firm value and boards depends on firm characteristics, such that a firm’s asset complexity and R&D intensity affect the optimal board size and independence. Outsider-dominant boards are less efficient in monitoring for firms with high R&D intensity and high asset intangibility (Raheja 2005; Harris and Raviv 2008). These arguments imply that takeover defenses are costly for firms with high monitoring costs, while they are beneficial for firms with greater advising needs.

The empirical tests rely on the measures of firm age, monitoring costs, and advising needs. Firm age is estimated with the years since incorporation, as reported in CRSP. The monitoring costs are measured with asset opacity, the ratio of intangible assets to total assets. Firms with largely intangible assets are hard to monitor, and thus, takeover defenses are detrimental to the firm value in these opaque firms.

The advising needs are measured with firm complexity, as in Coles et al. (2008). Larger firms have more complex contracting relationships and a greater dependence on the external resources. Thus, firm size serves as a proxy for complexity and the advisory needs. As mentioned earlier, firm age might also represent firm complexity. In complex firms with higher advisory needs, takeover defenses are beneficial.

Taken together, the previous studies suggest that the value of takeover defenses changes with firm age, monitoring cost, advisory needs, and their interactions.

2.2. Testable Hypothesis

Previous research implies that the net effect of takeover defenses is dependent on firm characteristics such as firm age, asset tangibility, and the firm complexity. These firm characteristics represent the costs and benefits of takeover defenses. Equation (1) represents the testable model, where TD is takeover defenses, Old is firm age, Intan is asset intangibility, and Comp is the firm complexity.

$$\begin{array}{cc}\hfill \mathrm{Firm} \mathrm{Value}= \mathsf{\alpha }& + {\beta }_{1}TD+{\beta }_{2}TD\times Old+{\beta }_{3}TD\times Intan+{\beta }_{4}TD\times Comp\hfill \\ & + {\beta }_{5}TD\times Old\times Intan+{\beta }_{6}TD\times Old\times Comp+Controls\hfill \\ & + \epsilon \hfill \end{array}$$

(1)

Based on the theoretical arguments discussed in the previous section, the testable hypotheses are as follows:

H1. 

Takeover defenses are more negatively associated with firm value for older firms. Previous studies document that β₁ is negative and statistically significant, i.e., firm value is negatively associated with takeover defenses. From Equation (1), H1 predicts that ${\beta }_2$ is negative.

As a firm gets old, managerial ownership decreases, and thus, the agency cost increases (Helwege et al. 2007; Johnson et al. 2022). If the costs of takeover defenses outweigh the benefits, the adverse impact of takeover defenses is expected to increase with firm age. Conversely, if firm age represents the complexity of a firm’s operations, the protection provided by takeover defenses is value-enhancing, and thus, firm age positively moderates the adverse impact of takeover defenses.

H2. 

Takeover defenses are more negatively associated with firm value for firms with higher monitoring costs. The monitoring costs are measured by asset intangibility. Thus, ${\beta }_3$ is expected to be negative and statistically significant.

H3. 

The adverse impact of takeover defenses is positively moderated in firms with higher advisory needs. The advisory needs are measured by firm complexity proxied by firm size. Thus, it is expected that ${\beta }_4$ is positive.

The presence of the costs and benefits of takeover defenses implies that the net effects are dependent on the relative magnitude of a firm’s monitoring costs and advising needs. For firms with higher monitoring costs, the costs outweigh the benefits of takeover defenses. The opposite is true for firms with higher advising needs. Thus, a firm’s monitoring costs and advising needs moderate the impact of takeover defenses.

H4. 

The value of takeover defenses is determined by complex interactions among firm age, monitoring costs, and advisory needs. Thus, ${\beta }_5$ and ${\beta }_6$ will be significantly different from zero.

Hypothesis 4 is the main hypothesis of the paper. I argue that the costs and benefits of takeover defenses change with firm age, monitoring costs, and advisory needs. The signs of coefficients are less clear in theory. Thus, it is important to examine the interaction of these variables to infer the nature of the relationship between firm value and takeover defenses.

3. Data and Variables

3.1. Sample Selection

The initial sample consists of the universe of firms covered in the RiskMetrics directors and governance databases for the period from 1996 to 2006. The RiskMetrics data include the entire S&P 1500 firms in the US. The RiskMetrics governance database issued six volumes of data in the years 1996, 1998, 2000, 2002, 2004, and 2006. RiskMetrics has changed its data collection method in year 2007, and thus, to maintain the consistency, the data period ends as of 2006. In addition, Karthaus et al. (2021) suggest that the post-2006 data appear biased toward finding a less negative relationship between firm value and takeover defenses. Nonetheless, the generalization of the findings in the paper beyond the sample period is limited.

Following Gompers et al. (2003), firms are assumed to maintain the same governance provisions as in the previous publication year between two consecutive publications. From the initial sample, I exclude firms with sales revenues less than $20 million, firms in Real Estate Investment Trusts (SIC 6798), and those lacking the required financial data from Compustat annual files. In addition, I collect the state incorporation data from COMPUSTAT quarterly files, the number of segments from Compustat Segment files, and firm age information from the listing dates in CRSP. The final sample consists of 11,904 firm-year observations by 2172 unique firms.

3.2. Variables

Key test variables and control variables are described in this section. To examine the main hypothesis of the paper, I need to define the measure of firm performance, the adoption of takeover defenses, and the interactions between firm age, monitoring costs, and advising needs, along with other control variables. Appendix A summarizes the detailed definition of variables and data sources.

3.2.1. Firm Performance and the E-Index

Firm performance is measured by Tobin’s q and is approximated by the ratio of the market value of assets to the book value of assets. The market value of assets is defined as the market value of equity and the book value of debt. The book value of debt is the book value of assets minus the book value of equity and deferred taxes. I later use the industry-adjusted Tobin’s q to further exclude the industry effect. Industry is defined at the two-digit SIC levels.

Takeover defenses could affect the effectiveness of the external control activities. To measure the deterrence effect of takeover defenses, Gompers et al. (2003) use the G-index by summing the 24 anti-takeover provisions adopted by a firm. Bebchuk et al. (2009) propose the E-index, as some provisions are particularly important to shareholders’ value. The E-index is composed of the six takeover defenses, which are classified boards, poison pills, golden parachutes, limits on shareholder amendments of firm bylaws, supermajority requirements to change the firm charter, and supermajority requirements to approve mergers. I use the E-index to measure the degree of a firm’s reliance on takeover defenses.

3.2.2. Firm Age, Monitoring Costs, and Advising Needs

The main hypothesis of the paper is that the effect of takeover defenses differs across firms with different firm age, monitoring costs, and advising needs. Furthermore, I examine the interaction effects of these three factors on the value of takeover defenses. Firm age is the number of years since incorporation recorded in the listing dates of the CRSP database. This is different from Johnson et al. (2022), who used years after the IPO dates. As long as the value of takeover defenses changes over time, I expect that my measure of the firm age will provide the same intuition as the years since the IPOs.

Boards play two distinct roles, monitoring and advising managers. Takeover defenses affect the efficacy of board roles in the two dimensions. A board’s monitoring ability depends on the nature of a firm’s assets, such that board monitoring is less effective when information costs are high (Raheja 2005; Harris and Raviv 2008). Following Ahn and Shrestha (2013), a firm’s monitoring costs are measured by asset intangibility. Asset intangibility is the ratio of net property, plant, and equipment to the market value of assets. The market value of assets is assets minus common equity plus market value of equity. While Ahn and Shrestha (2013) estimate monitoring costs with R&D intensity and asset tangibility by creating a dummy variable that is generated with factor scores from the factor analysis, I use asset intangibility alone for three reasons. First, the two variables are highly and positively correlated to each other, and thus, asset intangibility is a good proxy for the combined measure of R&D intensity and asset intangibility. Second, I can use the asset intangibility as a continuous variable instead of using a dummy variable. Third, the R&D variable is lacking in many firms, while asset intangibility provides a more extensive coverage in the sample.

The advising role of the board increases in large firms with complex operations (Coles et al. 2008). Firm complexity increases in larger firms because their operations rely on extensive large amounts of contracts and external resources. Therefore, firm size is a good proxy for a firm’s complexity and thus the advisory needs of the firm. Firm size is measured by the log of the book value of assets. For simplicity, I use the firm size alone to proxy a firm’s advising needs.

In addition, I utilize a dummy variable approach. Firms with above-median firm age are referred to as old firms, having a value of 1 (Old = 1), and those with below-median firm age are considered young firms and are assigned a value of 0 (Old = 0). Similarly, firms with above-median asset intangibility are referred to as firms with high monitoring costs, having a value of 1 (Monitor = 1), and those with below-median asset intangibility are considered firms with low monitoring costs and are assigned a value of 0 (Monitor = 0). Finally, firms with above-median firm size are referred to as firms with high advising needs, having a value of 1 (Advise = 1), and those with below-median firm size are considered firms with low advising needs and are assigned a value of 0 (Advice = 0).

In summary, I expect the relationship between takeover defenses and firm value to change with firm age, monitoring costs, advising needs, and their interactions.

3.2.3. Control Variables

The analysis includes several control variables that are expected to affect Tobin’s q. Following previous studies that examine Tobin’s q regressions, the regression analysis includes a set of control variables that affect Tobin’s q other than the main test variables: the number of segments, R&D intensity, leverage, ROA, Capex, Delaware incorporation, year-fixed effect, and industry-fixed effect. The number of segments is the number of business segments from Compustat segment files. R&D intensity is the ratio of research and development expense to book assets. R&D intensity is set to zero if R&D expenditure is missing. Leverage is the ratio of total debt (debt in current liabilities plus long-term debt) to book assets. The return on assets (ROA) controls for firm profitability, and it is measured by the ratio of income before extraordinary items to book assets. Capex is used to control for the investment opportunities, and it is measured by the ratio of capital expenditures to assets. Delaware dummy variable is used to control for the valuation impact of Delaware incorporation. The variable has a value of 1 if a firm is incorporated in Delaware, and zero otherwise.

All regression specifications include year- and industry-fixed effects. Industry is defined at the two-digit SIC levels.

4. Empirical Analysis

4.1. Descriptive Statistics

Table 1. Key statistics for main variables and control variables.

	Variables	Mean	Median	5%	95%	Min.	Max.
Main Variables	Tobin’s q	1.94	1.52	0.88	4.61	0.71	8.04
	E-index	2.25	2.00	0.00	4.00	0.00	6.00
	Age	25.61	19.50	5.00	72.00	3.00	79.00
	Intan	0.78	0.85	0.38	0.98	0.23	0.99
	Size	7.26	7.08	5.06	10.02	2.59	12.92
Controls	Segments	2.60	2.00	1.00	6.00	1.00	8.00
	R&D	0.05	0.00	0.00	0.22	0.00	0.66
	Leverage	0.30	0.30	0.00	0.68	0.00	1.19
	ROA	0.03	0.05	−0.16	0.16	−0.66	0.24
	Capex	0.06	0.04	0.01	0.16	0.00	0.27
	Delaware	0.58

shows descriptive statistics for main test variables and control variables. The average Tobin’s q is 1.94, and the median is 1.52. These numbers are similar to those reported in the previous literature. The Tobin’s q ranges from 0.88 to 4.61 in its 5 and 95 percentiles. The E-index has the lowest value of 0 and the highest value of 6, but the mean and median are around 2, and most firms have takeover defenses from 0 to 4.

The average firm age is 26 years, while the median is 20 years. This is quite different from the firm age ranging from 1 to 17 years, reported in Johnson et al. (2022), who defined firm age as years since the IPO. I use firm age as years since incorporation reported in CRSP. Provided that the value of takeover defenses changes over time, I expect the impact of firm age will be similar whether it is defined as years since IPOs or years since incorporation. Table 1 also reports key statistics for asset intangibility (Intan) and firm size (Size).

The lower part of Table 1 reports the statistics of the control variables used in the empirical analysis. There are no unusual outliers, and they are similar in magnitude to those reported in the previous literature.

Table 2. Pearson correlations among main variables.

	Tobin’s q	E-Index	Age	Intan	Size	Old	Monitor	Advise
Tobin’s q	1.00
E-index	−0.13	1.00
Age	−0.12	0.07	1.00
Intan	0.48	−0.06	−0.19	1.00
Size	−0.06	0.04	0.45	−0.22	1.00
Indicator Variables:
Old	−0.14	0.14	0.77	−0.19	0.28	1.00
Monitor	0.46	−0.08	−0.18	0.78	−0.18	−0.16	1.00
Advise	−0.05	0.08	0.35	−0.20	0.80	0.24	−0.15	1.00

Note: All Pearson correlations estimated above are significant at the 1% significance level.

reports Pearson correlation matrix among variables. Tobin’s q is negatively correlated with E-index, which confirms the previous findings of the adverse impact of takeover defenses on firm value. Tobin’s q is also negatively correlated with firm age and firm size, but it is positively correlated with asset intangibility. Therefore, it is necessary to conduct multivariate tests to examine the effect of takeover defenses on firm value after considering the interactions of firm age, monitoring costs, and advising needs.

Utilizing an indicator variable approach, the correlations of Tobin’s q and Old, Monitor, and Advise are similar to those of continuous variables. Old firms and firms with high advising needs have negative correlations with Tobin’s q, while firms with high monitoring costs have a positive correlation. Similarly, the correlation between Old and Advise is positive, and the correlation between Old and Monitor is negative. Thus, the correlation analysis in Table 2 suggests that the impact of takeover defenses on firm value may change after considering the interactions among firm age, monitoring costs, and advising needs.

4.2. Univariate Test Results

Table 3. Univariate tests of main variables.

Variable	Sort	Below-Median Firms	Above-Median Firms	t-Test	Mann–Whitney z-Test
	Old	2.12 [1.65]	1.76 [1.41]	15.57 ***	15.32 ***
Tobin’s q	Monitor	1.35 [1.24]	2.52 [2.02]	−56.01 ***	−58.16 ***
	Advise	2.00 [1.56]	1.88 [1.48]	5.10 ***	4.69 ***
	Old	2.07 [2.00]	2.43 [3.00]	−15.96 ***	−16.15 ***
E-index	Monitor	2.34 [2.00]	2.15 [2.00]	8.71 ***	8.37 ***
	Advise	2.15 [2.00]	2.35 [2.00]	−8.93 ***	−8.66 ***

Note: *** denotes statistical significance at the 1% level. The mean values are reported with median values in the blanket. In the last two columns of the table, t-test results for the mean difference and Mann–Whitney Wilcoxon signed-rank test for the median difference are reported.

presents the two-way sorts for the mean and median differences in Tobin’s q and E-index between firms with different firm age, monitoring costs, and advising needs. The mean values are reported with median values in the blanket. In the last two columns of the table, t-test results for the mean difference and Mann–Whitney Wilcoxon signed-rank test for the median difference are reported.

The first part of Table 3 examines the differences in Tobin’s q between firms with the above and below-median values of firm age, monitoring costs, and advising needs. The results show that older firms, firms with low monitoring costs, and firms with high advising needs tend to trade at lower firm value. Next, the second part of Table 3 examines the differences in E-index for firms with different characteristics. The result indicates that older firms, firms with low monitoring costs, and firms with high advising needs tend to have more takeover defenses.

The results suggest the endogenous nature of the relationship between firm value and takeover defenses, such that certain firm characteristics are endogenously related to the firm’s adoption of takeover defenses. Provided that the value of takeover defenses changes as a firm becomes older and is also dependent on the firm’s monitoring costs and advising needs, multivariate tests shall include the interactions of firm age, monitoring costs, and advising needs.

4.3. Multivariate Test Results

This section examines the relationship between Tobin’s q and E-index for firms with different firm age, asset intangibility (a proxy for monitoring costs), and firm size (a proxy for advising needs). I expect that the impact of takeover defenses on firm value changes as firms get older. In addition, a firm’s monitoring costs and advising needs play an important role in determining the value of takeover defenses as well. Finally, I expect the interactions and the moderating effects among firm age, monitoring costs, and advising needs.

4.3.1. The Relation between Firm Value and E-Index with the Interactions of Firm Age, Monitoring Costs, and Advising Needs

Table 4. The relationship between the firm value and E-index with interactions.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
E-index	−0.057 ***	−0.067 ***	0.185 ***	0.134 ***	0.062 *	0.183 ***	0.496 ***
E-index × Age		0.0004	−0.0002	0.002 **	0.001 ***	−0.004 ***	−0.003 ***
E-index × Intan			−0.310 ***	−0.244 ***			−0.306 ***
E-index × Intan × Age				−0.003 ***			−0.002 *
E-index × Size					−0.020 ***	−0.036 ***	−0.045 ***
E-index × Size × Age						0.001 ***	0.001 ***
Age	−0.001 *	−0.002 *	−0.001	−0.001	−0.004 ***	−0.004 ***	−0.003 ***
Intan	3.879 ***	3.878 ***	4.559 ***	4.556 ***	3.881 ***	3.852 ***	4.618 ***
Size	0.016 *	0.017 *	0.017 *	0.018 *	0.060 ***	0.062 ***	0.076 ***
Intercept	−2.302 ***	−2.278 ***	−2.886 ***	−2.887 ***	−2.584 ***	−2.539 ***	−3.317 ***
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Industry-fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year-fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Adj. R2	0.4455	0.4455	0.4489	0.4491	0.4461	0.4469	0.4513
N	11,904	11,904	11,904	11,904	11,904	11,904	11,904

Note: ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Controls include the number of operating segments, leverage ratio, R&D-to-Assets ratio, ROA, Capex-to-Assets ratio, and Delaware incorporation dummy. Industry-fixed effects are indicator variables for 3-digit SIC code. Year-fixed effects are indicator variables for each calendar year. Adj. R² is adjusted R-squared. N is the number of observations.

presents the baseline regression models. The dependent variable is the measure of firm performance, Tobin’s q. All regression specifications include a set of control variables, including the number of segments, R&D intensity, leverage, ROA, Capex, and Delaware incorporation. The regression models are also estimated with industry dummy variables defined at the 2-digit SIC levels, and year dummy variables to account for the industry-fixed effects and year-fixed effects, respectively.

In model 1, the coefficient of the E-index is −0.057 and significant at the 1% significance level. This result is consistent with the previous finding that documents the negative relationship between the firm value and the E-index.

From model 2 of Table 4, the interaction term between E-index and firm age (E-index × Age) is added to the model. This result is not altered materially. The coefficient on E-index continues to be negative and statistically significant. The coefficient on (E-index × Age) is 0.0004, and it is insignificant. Thus, hypothesis 1 (H1) is not supported. This contrasts Johnson et al. (2022), who find a strong negative effect of the interaction term of the E-index and firm age. The reasons for the inconsistency are due to the differences in sample firms and the definition of firm age. Nonetheless, firm age represents the altering value of takeover defenses and thus is expected to have some moderating effects on other explanatory variables.

In model 3, the interaction term between E-index and asset intangibility (E-index × Intan) is introduced. The coefficient on the interaction of (E-index × Intan) is significantly negative −0.310, while the coefficient on E-index itself turns to positive 0.185. This result indicates that the adverse impact of takeover defenses is relevant only to those firms with high asset intangibility. Thus, this result is consistent with the prediction of hypothesis 2 (H2). The value of takeover defenses is actually beneficial for firms with low asset intangibility. This supports the view that the value of takeover defenses differs in firms with different monitoring costs.

In model 4, the triple interaction term of E-index, asset intangibility, and firm age (E-index × Intan × Age) is added to examine the moderating effect of firm age. The coefficient of the triple interaction term is negative −0.003, and it is significant at the 1% significance level. The coefficient on the interaction term of (E-index × Intan) continues to be negative −0.244. This result indicates that the adverse impact of takeover defenses is more evident in firms with higher asset intangibility and older firms. In the opposite case, for firms with lower asset tangibility and younger firms, the takeover defenses are associated with higher firm value.

Model 5 examines the interaction effect of E-index and firm size (E-index × Size). Similar to the result in model 3, the coefficient on the interaction of (E-index × Size) is significantly negative −0.020, while the coefficient on E-index itself becomes positive 0.001. This result indicates that takeover defenses are detrimental for firms with bigger firms, but not so for smaller firms. This supports the view that the value of takeover defenses differs in firms with different advising needs. The sign of the interaction term (E-index × Size) though was expected to be positive in theory. Thus, hypothesis 3 (H3) is not supported.

In model 6, the triple interaction term of E-index, firm size, and firm age (E-index × Size × Age) is added and the coefficient on the triple interaction term is positive, 0.001. The coefficient on the interaction term of (E-index × Size) continues to be negative, −0.036. The coefficient on the interaction of (E-index × Age) becomes −0.004. This result indicates that the interactions among E-index, firm age, and firm size is complex and heterogeneous to firm characteristics.

Finally, in model 7, all interaction terms are included in the model. This result shows that the value of takeover defenses is detrimental for older firms, bigger firms, and firms with high asset intangibility. Conversely, takeover defenses add value for younger firms, smaller firms and firms with low asset intangibility. I also find some moderating effect of firm age for bigger firms with high advising needs, as the triple interaction term (E-index × Size × Age) is positive and significant.

Taken together, the regression analysis in Table 4 reinforces the fact that those firm characteristics variables have moderating effects, i.e., the relationship between takeover defenses and firm value is complex and moderated by firm characteristics, including firm age, asset intangibility, and firm size. These results provide strong evidence that supports the main hypothesis of the paper, hypothesis 4 (H4).

The statistical significance of the interaction terms indicates that the marginal effect of the E-index on firm value changes at the different levels of firm age, asset tangibility, and the firm complexity. To compute the marginal effect, several input values of interaction variables are required (Busenbark et al. 2022). Because I examine the triple interactions of firm age, asset tangibility, and the firm complexity, three-dimensional simulation is required to gauge the marginal effect. This is an extensively complicated task with continuous variables. However, it can be simpler under the indicator variable approach. Therefore, the marginal effect of the E-index on firm value at the different levels of firm age, asset tangibility, and the firm complexity are explored in the next section.

4.3.2. The Relationship between Firm Value and E-Index with Indicator Variables

In this section, indicator variable approach is used by dividing sample firms into two sub-groups based on the median values of firm age, asset intangibility, and firm size. Old has a value of 1 (Old = 1) if firm age has an above-median value, and 0 otherwise. Monitor equals to 1 if asset intangibility has an above-median asset intangibility, and 0 otherwise. Advise equals to 1 if firm size is above-median, and 0 otherwise.

Table 5. Multivariate tests with indicator variables.

Variables	Model 1	Model 2	Model 3	Model 4
E-index	−0.085 ***	−0.021 **	−0.056 ***	0.024 *
E-index × Old	0.060 ***	0.057 ***	0.051 ***	0.047 ***
E-index × Monitor		−0.115 ***		−0.129 ***
E-index × Monitor × Old		−0.041 ***		−0.035 **
E-index × Advise			−0.069 ***	−0.091 ***
E-index × Advise × Old			0.038 ***	0.040 ***
Old	−0.223 ***	−0.175 ***	−0.249 ***	−0.209 ***
Monitor	1.033 ***	1.346 ***	1.033 ***	1.372 ***
Advise	0.087 ***	0.090 ***	0.197 ***	0.248 ***
Intercept	0.607 ***	0.359	0.552 **	0.259
Controls	Yes	Yes	Yes	Yes
Industry-fixed effects	Yes	Yes	Yes	Yes
Year-fixed effects	Yes	Yes	Yes	Yes
Adj. R2	0.4180	0.4228	0.4187	0.4240
N	11,904	11,904	11,904	11,904
F-tests for Interactions:
Partial Effect: E-index × Old + E-index × Monitor + E-index × Monitor × Old				−0.117 ***
Partial Effect: E-index × Old + E-index × Advise + E-index × Advise × Old				−0.004
Total Effect: E-index + E-index × Old + E-index × Monitor + E-index × Monitor × Old				−0.093 ***
Total Effect: E-index + E-index × Old + E-index × Advise + E-index × Advise × Old				0.020 **

Note: ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Controls include the number of operating segments, leverage ratio, R&D-to-Assets ratio, ROA, Capex-to-Assets ratio, and Delaware incorporation dummy. Industry-fixed effects are indicator variables for 3-digit SIC code. Year-fixed effects are indicator variables for each calendar year. Adj. R² is adjusted R-squared. N is the number of observations.

reports the results.

These results are, in general, consistent with the findings with continuous variables used in Table 4. In model 4, E-index itself has neutral impact on firm value. Takeover defenses are costly for firms with high monitoring costs and also for firms with high advising needs. The adverse impact of takeover defenses in firms with high monitoring costs is further exacerbated in old firms, i.e., the coefficient on the triple interaction term of (E-index × Monitor × Old) is negative −0.035. Conversely, the negative impact of takeover defenses in firms with high advising needs is positively moderated in old firms, i.e., the coefficient of the triple interaction term (E-index × Advise × Old) is positive 0.040.

Examining the differential effect of classified boards on firm value, Ahn and Shrestha (2013) posit that the benefits of a classified board outweighs its costs for firms with high monitoring costs but with low advising needs, and vice versa. Extending the idea, the main interest in hypothesis 4 (H4) is the moderating effect of firm age in two sub-groups of firms, i.e., firms with high monitoring costs but low advising needs (Monitor = 1 and Advise = 0) and, on the opposite side, firms with high advising needs but with low monitoring costs (Advise = 1 and Monitor = 0).

At the bottom of model 4, the partial effects and total effects of interaction terms are reported. The partial effects are measured for two groups of firms: the first partial effect represents the moderating effect for old firms with high monitoring costs but low advising needs (Old = 1 and Monitor = 1 and Advise = 0). The second partial effect represents the moderating effect for old firms with high advising needs but with low monitoring costs (Old = 1 and Advise = 1 and Monitor = 0).

The first partial effect of Old, Monitor, and the interaction (E-index × Old + E-index × Monitor + E-index × Monitor × Old) is −0.117, and it is significant at the 1% significance level. This indicates that the relationship between the E-index and firm value is negatively moderated for old firms with high monitoring costs but low advising needs.

The second partial effect of Old, Advise, and the interaction (E-index × Old + E-index × Advise + E-index × Advise × Old) is smaller −0.004, and it is insignificant. This indicates that the relationship between firm value and the E-index is not moderated for old firms with high advising needs but low monitoring costs.

The total effects are reported in the next two lows. The total effect represents the relationship between the E-index and firm value for old firms with high monitoring costs but low advising needs (Old = 1 and Monitor = 1 and Advice = 0) and for old firms with high advising needs but low monitoring costs (Old = 1 and Advise = 1 and Monitor = 0).

The total effect of the E-index on firm value for old firms with high monitoring costs but low advising needs is −0.095, which implies that, from the mean value, the E-index decreases firm value by 4.9% (=−0.095/1.94) than for young firms with low monitoring costs. The total effect of the E-index on firm value for old firms with higher advising needs but low monitoring costs is 0.020, which implies that, from the mean value, the E-index increases firm value by 1% (=0.020/1.94) than for young firms with low advising needs.

In all regressions, the interaction of E-index and Old (E-index × Old) is positive and significant at the 1% significance level. Again, this contrasts with the findings in Johnson et al. (2022), who report the negative coefficient on the interaction term of E-index and firm age. The discrepancy can be attributable to the difference in sample firms where I use the entire universe of firms with older firms, while their sample firms include only IPOs firms and relatively younger firms. As firms slowly adjust their behavior, older firms in my sample tend to adopt optimal takeover defenses over time. Nonetheless, the implication of the results is similar to that of Johnson et al. (2022), i.e., the value of takeover defenses is dependent on the benefits and costs of takeover defenses that change over time and differ in firm characteristics.

4.3.3. Endogeneity Regressions

The results in the previous section are subject to some endogeneity issues. One is the reverse causality that low quality firms tend to adopt more takeover defenses to insulate themselves from the hostile takeover threats. However, the reverse-causality argument is less relevant in explaining the interaction effects among firm age, monitoring costs, and advising needs, which are the main focus of the paper.

It is also possible that the influence of unknown missing variables might influence the results in the paper. Although this type of endogeneity issue is impossible to completely resolve, I attempt to alleviate this issue by introducing additional control variables and utilizing two-stage least-squared regressions (2SLS).

Additional control variables include stock ownership by pension funds, largest block ownership, CEO share ownership and its squared term, and a log of CEO delta and vega. Institutional blockholdings exert additional monitoring on a firm’s decision-making and affects the board structure. I used pension fund ownership and the largest block ownership to control for the influence of institutional ownership. The data on institutional blockholdings are collected from 13F filings.

Following Morck et al. (1988) and McConnell and Servaes (1990), who propose that the relationship between firm value and managerial ownership is nonlinear, I include CEO ownership and its squared term in the regression. When CEO ownership data are missing, it is assumed that it has a value of 0. In addition, CEO incentives are measured by delta and vega. Following the method of Core and Guay (2002) and Low (2009), delta is computed as the dollar changes in the value of a CEO’s stocks and stock options for a 1% change in the stock price. Vega is the dollar changes in a CEO’s stock option value for a 1% change in the standard deviation of the stock returns. CEO compensation and equity portfolio holdings data are from the Execucomp database.

In two-stage least-squared (2SLS) regression, I firstly estimate the probability of adopting takeover defenses with the Massachusetts Classified Board Law, an instrument proposed by Ahn and Shrestha (2013), and other control variables. The predicted value of E-index is plugged in the second-stage regression to estimate coefficients. The dependent variable is the measure of firm performance, Tobin’s q. In the last two columns, industry-adjusted Tobin’s q is used.

In

Table 6. Regression analysis with additional controls and 2SLS.

	Tobin’s q		Industry-Adjusted q
Variables	Model 1	Model 2	Model 3	Model 4
	OLS	2SLS	OLS	2SLS
E-index	0.025 *		0.028 **
E-index × Old	0.039 **		0.029 *
E-index × Monitor	−0.102 ***		−0.096 ***
E-index × Monitor × Old	−0.038 ***		−0.044 ***
E-index × Advise	−0.073 ***		−0.073 ***
E-index × Advise × Old	0.039 ***		0.041 ***
Instrumented E-index		−0.078 ***		−0.116 ***
Instrumented E-index × Old		0.009		0.029
Instrumented E-index × Monitor		0.176 ***		0.248 ***
Instrumented E-index × Monitor × Old		−0.063 ***		−0.088 ***
Instrumented E-index × Advise		0.057 ***		0.090 ***
Instrumented E-index × Advise × Old		−0.002		−0.037
Old	−0.159 ***	−0.016	−0.110 ***	−0.028
Monitor	1.172 ***	0.608 ***	1.016 ***	0.658 ***
Advise	−0.045	−0.261 ***	−0.019	−0.161 ***
Intercept	0.415 **	0.557 ***	−0.875 ***	−0.772 ***
Controls	Yes	Yes	Yes	Yes
Additional Controls	Yes	Yes	Yes	Yes
Industry-fixed effects	Yes	Yes	Yes	Yes
Year-fixed effects	Yes	Yes	Yes	Yes
Adj. R2	0.4677	0.4743	0.3441	0.3531
N	11,904	11,904	11,882	11,882
F-tests for Interactions:
Partial Effect: E-index × Old + E-index × Monitor + E-index × Monitor × Old	−0.101 ***	0.122 ***	−0.111 ***	0.189 ***
Partial Effect: E-index × Old + E-index × Advise + E-index × Advise × Old	0.005	0.064 ***	−0.003	0.082 ***
Total Effect: E-index + E-index × Old + E-index × Monitor + E-index × Monitor × Old	−0.076 ***	0.044 ***	−0.083 ***	0.073 ***
Total Effect: E-index + E-index × Old + E-index × Advise + E-index × Advise × Old	0.030 ***	−0.014 ***	0.025 ***	−0.034 ***

Note: ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Controls include the number of operating segments, leverage ratio, R&D-to-Assets ratio, ROA, Capex-to-Assets ratio, and Delaware incorporation dummy. Industry-fixed effects are indicator variables for 3-digit SIC code. Year-fixed effects are indicator variables for each calendar year. Additional control variables include stock ownership by pension funds, largest block ownership, CEO share ownership and its squared term, and a log of CEO delta and vega. Adj. R² is adjusted R-squared. N is the number of observations.

, the result is consistent with the view that the value of takeover defenses differs in firms with different age, monitoring costs, and advising needs. At the bottom of each model, the partial effects and total effects of the interaction terms are included.

From model 1, the value of takeover defenses is significantly negative in firms with high monitoring costs and also in firms with high advising needs. The adverse impact of E-index on firm value is exacerbated in old firms when the firm has high monitoring costs but low advising needs. Conversely, the effect of E-index on firm value is positively moderated in old firms when the firm has high advising needs but low monitoring costs. Thus, with additional controls, I continue to find the heterogeneous and complex effects of takeover defenses on firm value.

In the 2SLS estimation, the predicted value of the E-index from the first-stage regression is used in the second-stage estimation, and it is entitled as an instrumented E-index. In model 2, the result shows that the value of takeover defenses is negative and statistically significant. The partial effect and the total effect of E-index for old firms with high monitoring costs but low advising needs are significant and positive, 0.122 and 0.044, respectively. The partial effect and the total effect of E-index for old firms with high advising needs but with low monitoring costs are significant, 0.064 and −0.014, respectively. Thus, takeover defenses are value-enhancing for old firms with high monitoring costs but with low advising needs. Conversely, takeover defenses are value-destroying for old firms with high advising needs but with low monitoring costs.

Contrary to OLS estimation, the signs of certain interaction terms are in the opposite direction. The flip of signs appears driven by the instruments used in the 2SLS. In instrumental variable (IV) methods such as 2SLS, the use of a weak instrument, which is weakly correlated with the endogenous variables, causes poor properties in the estimation and yields no different result with the OLS estimation. Conversely, a strong instrument could cause over-identification and increase the bias and distortion of the estimated coefficients in the second stage estimation (Keane and Neal 2023). Nonetheless, the primary objective of the paper is to demonstrate the differential effect of takeover defenses on firm value. Thus, the result of Table 6 is largely consistent with the prediction in hypothesis 4 (H4), i.e., the differential effect of takeover defenses on firm value in different dimensions of firm age, monitoring costs, and advising needs persists with additional controls and 2SLS.

In model 3 and model 4, industry-adjusted Tobin’s q is used instead of Tobin’s q, and the results are virtually identical to the previous tests.

The E-index is possibly measured with errors and thus less representative of the degree of a firm’s takeover defenses (Catan and Kahan 2016). Among takeover defenses in the E-index, classified boards presumably have a strong deterrence effect. To address concerns about the measurement errors in the E-index, I conduct all the tests using just the presence or absence of classified boards, excluding other provisions in the E-index. In an un-tabulated test, I found qualitatively similar results.

In addition, Karthaus et al. (2021) report that the post-2006 governance data tend to generate a more negative relationship between firm value and the E-index. The data period used in the paper ends in year 2006 and is thus less subject to the negative bias and data quality issue.

5. Conclusions

In this paper, I document additional evidence that the value of takeover defenses depends on firm characteristics, including firm age, monitoring costs, and advising needs. The empirical tests show that the relationship between Tobin’s q and the E-index is negative for firms that are older and have higher asset tangibility, a proxy for monitoring costs. Conversely, the relationship is less negative for firms that are older, but bigger in firm size, a proxy for complexity. Thus, firm age has a positive moderating effect for firms with higher advising needs. I conducted additional tests using dummy variables, including additional control variables and addressing the endogeneity concern with the two-step procedures (2SLS). The results presented in this paper are consistent with the notion that the value of takeover defenses is heterogeneous to each firm having different firm characteristics. Unlike Johnson et al. (2022), in some regressions, the interaction of the E-index and firm age is positive to firm value, which suggests that it takes a long time for a firm to optimally adjust its corporate governance structure.

The findings taken together suggest that the relationship between firm value and takeover defenses depends on the interactions of the firm characteristics that represent the costs and benefits of takeover defenses. The evidence complements a growing body of literature that asserts “one size does not fit all”. The complex interactions of multiple factors suggest that takeover defenses have a heterogeneous impact on firm value for each firm. Thus, the one-size-fits-all approach is potentially misleading in the corporate governance structure policy. Policy makers and ESG activists attempting to regulate a firm’s takeover defenses should carefully examine the benefits and costs of takeover defenses for each firm.