Mfi

VF

a

ARRAA

KMDC

1

dadtridasviliet

mticamtt

1h

The Quarterly Review of Economics and Finance 54 (2014) 70– 81

Contents lists available at ScienceDirect

The Quarterly Review of Economics and Finance

j ourna l ho me p age: www.elsev ier .com/ locate /qre f

anagement entrenchment and the valuation discount of dual classrms

ishaal Baulkaran ∗

aculty of Management, University of Lethbridge, 4401 University Drive, Lethbridge, Alberta T1K 3M4, Canada

r t i c l e i n f o

rticle history:eceived 4 September 2012eceived in revised form 1 August 2013

a b s t r a c t

Prior studies provide empirical evidence that dual class firms are discounted compared to single class firmsdue to the extraction of private benefits. This study examines the link between managerial entrenchmentand the dual class discount. Using propensity score matching and conditioning for past underperformance,

ccepted 4 August 2013vailable online 14 August 2013

eywords:anagement entrenchmentual class discount

the paper shows that investors apply a greater discount to the value of dual class firms as the degree ofmanagerial entrenchment increases. The impact of entrenchment on dual class discount is more pro-nounced when the CEO is the controlling shareholder compared to when the controlling shareholder is adirector or the chairman of the board.

© 2013 The Board of Trustees of the University of Illinois. Published by Elsevier B.V. All rights reserved.

lcsfii

scmhcuwmd

oncentrated ownership and control

. Introduction

Dual class structure is arguably the most effective anti-takeoverevice ever invented (Jarrell & Poulsen, 1988; Ruback, 1988). As

result, it is likely that entrenchment is particularly strong inual class firms since the market for corporate control is vir-ually nonexistent.1 With dual class structure, it is difficult toemove a controlling shareholder-manager even when the firms performing poorly. Therefore, these managers face little or noiscipline from the market for corporate control due to the sep-ration of voting and cash flow rights which allows a controllinghareholder with minority equity stake to control majority of theoting rights. Hence, long tenure in firms with poor performances an indication of entrenchment. In addition, since the control-ing shareholder in dual class firms is not always the CEO, it is

mportant to examine directors’ tenure as a further measure ofntrenchment. Entrenched managers and directors are more likelyo extract private benefits of control and therefore, investors are

∗ Tel.: +1 403 329 2074.E-mail address: vishaal.baulkaran@uleth.ca

1 Managerial entrenchment, broadly defined, occurs when managers gain souch power that they are able to use the firm to further their own interests rather

han the interests of shareholders (Weisbach, 1988). The degree of entrenchments the extent to which managers fail to experience discipline from the full range oforporate governance and control mechanisms, including monitoring by the boardnd the threat of dismissal or takeover (Berger et al., 1997). In this research, I defineanagerial entrenchment more precisely as occurring when the CEO remains on

he job longer than an industry peer or a matching company CEO, especially whenhe CEO’s company is performing relatively poorly.

ifi

hatta

osa1

062-9769/$ – see front matter © 2013 The Board of Trustees of the University of Illinoisttp://dx.doi.org/10.1016/j.qref.2013.08.001

ikely to discount the value of dual class firms relative to singlelass concentrated control companies.2 Hence, the paper addresseseveral important empirical questions. Do managers in dual classrms display characteristics of entrenchment? Does entrenchment

n turn, lead investors to discount the value of dual class firms?Using a sample of S&P 1500 dual class firms and a propen-

ity matched sample of single class companies with concentratedontrol, I show that CEOs and directors in dual class firms areore entrenched. CEOs and directors of dual class firms tend to

ave longer tenure compared to their counterparts in single classompanies. As entrenchment is most concerning in the context ofnderperformance by management, I show that dual class firmshich have poor past performance and management entrench-ent as measured by excess CEO tenure, excess E-index or excess

irector tenure are discounted by investors. This implies thatnvestors are aware of the impact of managerial entrenchment inrms with dual class ownership structure.

Studies often assume that managers are also controlling share-olders in firms with concentrated ownership. However, this is notlways the case. Hence, examining dual class firms where the con-

rolling shareholder or a member of his or her family is the CEO andhose where the controlling shareholder is not the CEO but is either

member of the board or the Chairman provides evidence that the

2 I define concentrated control as a firm with an individual, a family or a groupf related individuals controlling at least 15% of the total votes. This definition isimilar to La Porta, Lopez-de-Silanes, and Shleifer (1999) and Claessens, Djankov,nd Lang (2000). Both studies utilise two alternative definitions of control of a firm:0% and 20% control of voting rights.

. Published by Elsevier B.V. All rights reserved.

f Econ

dcpaotm

tdt

2

abpclctahpl

csvtcfitrf

babametha2&

btb

r

t2lt

loett

n1ttar

mittptvaaettmeo

ahBoa(c1

rtfiorec

Hct

V. Baulkaran / The Quarterly Review o

ual class discount is more sensitive to entrenchment when theontrolling shareholder is the CEO. Excess CEO tenure in firms withoor past performance and a controlling shareholder-CEO leads ton 18.9% discount relative to mean value for a matched samplef single class concentrated control firms. The results are robusto endogeneity, alternative definition of dual class discount and a

ore traditional matching technique.The rest of the paper is organised as follows: Section 2 reviews

he relevant literature and develops testable hypothesis, Section 3escribes the methodology and data, Section 4 contains analyses ofhe results and discussion, and Section 5 concludes the research.

. Literature review and hypothesis development

The separation of voting and cash flow rights in dual class firmsllows the controlling shareholder-manager to become entrenchedy insulating managers from the discipline of the market for cor-orate control. In the absence of the threat of hostile takeover,ontrolling shareholders-managers in dual class firms face veryimited threat of job dismissal.3 Ruback (1988) argues that duallass ownership structure may be the most effective universal anti-akeover device ever invented. Similarly, Jarrell and Poulsen (1988)rgue that dual class structure provides an effective defense againstostile takeovers. Therefore, we can argue that entrenchment isarticularly strong in dual class firms. As a result, investors are

ikely to discount the value of dual class firms.4

Similarly, concentrated equity ownership in single class firmsan also serve to entrench managers. Managers who control a sub-tantial fraction of a single class firm’s equity may have enoughoting power or influence to guarantee their employment withhe firm (Morck, Shleifer, & Vishny, 1988). However, compared toontrolling managers of dual class firms, managers of single classrms with the same equity stake control less vote and face a greaterhreat of removal for poor performance from the market for corpo-ate control. Thus, the ability to remain entrenched is not as strongor managers of single class firms with concentrated control.

Entrenchment can have adverse effects on managementehaviour and incentives (Bebchuk, Cohen, & Ferrell, 2009). Man-gerial entrenchment may result in agency costs to shareholdersy allowing poorly performing managers to remain on the jobnd allow for empire-building.5 Entrenching mechanisms allowanagers to pursue their own interests and extract wealth at the

xpense of outside shareholders. Several studies provide evidencehat firms with entrenched managers significantly underperform,old large amounts of cash, pay lower dividends, and are less lever-ged (Berger, Ofek, & Yermack, 1997; Gompers, Ishii, & Metrick,003; Harford, Mansi, & Maxwell, 2008; Morck et al., 1988; Ozkan

Ozkan, 2004).Studies have argued that entrenchment can also produce

enefits to shareholders by reducing the extent to which thehreat of a takeover distorts investments in long-term projects andy allowing managers to extract higher acquisition premiums in

3 An exception is when a company defaults on its debt and creditors exercise theiright to replace management.

4 Previous studies provide evidence that dual class firms are discounted comparedo single class firms (Bennedsen & Nielsen, 2010; Gompers et al., 2003; King & Santor,008; Smith et al., 2009). In a sample of Canadian dual class firms, family ownership

ead to a 17% discount relative to single class firms (King & Santor, 2008). Researchersend to explain the dual class discount in terms of extraction of private benefits.

5 Florackis and Ozkan (2009) provide evidence that managerial entrenchmenteads to higher agency costs. They used the inverse of asset turnover as a measuref agency costs and interpret this ratio as an asset utilisation ratio which shows howffectively managers deploy firm assets. Bebchuk, Cohen, and Wang (2012) showhat G-index and E-index were no longer associated with abnormal returns duringhe period 2000–2008.

o

ewcSmLlb

oioaomt

omics and Finance 54 (2014) 70– 81 71

egotiated transactions (Bebchuk & Stole, 1993; Stein, 1988; Stulz,988). Prior studies such as Wilcox (2002) and Faleye (2007) arguehat staggered election of directors is an entrenching mechanismhat encourages board independence by reducing the threat that

director who refuses to succumb to management will not beenominated each year.

Managerial entrenchment can occur in several ways, includinganager-specific investment, concentrated control as well as var-

ous anti-takeover provisions. By making corporate investmentshat fit the expertise of a particular CEO that CEO can reducehe probability of being replaced, extract higher wages and largererquisites from shareholders (Shleifer & Vishny, 1989).6 Longenured managers, because of manager-specific assets, are morealuable to shareholders compared to an alternative manager. As

result, these managers can negotiate for higher compensationnd increase their latitude in running the firm. Managers can alsontrench themselves through holding a controlling share of theotal votes. In a single class company, they can be entrenchedhrough concentrated ownership whereas in dual class companies

anagers may control a company with a small percentage of thequity and majority of the total votes. Thus, it is easier to entrenchneself as a manager in a dual class firm.

Furthermore, anti-takeover provisions such as staggered boardsnd poison pills may serve to entrench managers and therefore,ave a negative impact on firm value (Bebchuk & Cohen, 2005;ebchuk et al., 2009; Faleye, 2007).7 A staggered board is onlyne of several provisions which may serve to entrench man-gers. Bebchuk et al. (2009) construct an entrenchment indexE-index) and provide evidence that the E-index level is monotoni-ally associated with reduction in firm valuation during the period990–2003.8

Given that dual class ownership structure can lead to manage-ial entrenchment, the more entrenched dual class managers are,he more likely it is that such managers will extract private bene-ts. As a result, investors are expected to apply a greater discountn the value of dual class firms. Several characteristics of manage-ial entrenchment such as CEO tenure, and directors’ tenure arexpected to be related to the observed valuation discount of duallass firms. Given this argument, the hypothesis being tested is:

1. The greater the managerial entrenchment, the larger the duallass discount, after controlling for other relevant firm-specific fac-ors.

6 Shleifer and Vishny (1989) argue that excessive growth in sales in the directionf the CEOs’ talents and experience is a means of entrenchment.7 Bebchuk and Cohen (2005) show that staggered boards are associated with an

conomically meaningful reduction in the firm value. This result is stronger for firmshich establish staggered boards through corporate charter which shareholders

annot amend compared to staggered boards established in the company’s by-laws.imilarly, Faleye (2007) shows that staggered boards destroy value by entrenchinganagement and reducing the likelihood of forced CEO turnover. Bates, Becher, and

emmon (2008), on the other hand, argue that staggered boards do not change theikelihood that a firm, once targeted, is ultimately acquired. However, staggeredoards reduce the likelihood of receiving a takeover bid.8 Bebchuk et al. (2009) excluded dual class firms from their sample. They use six

f the 24 provisions followed by the Investor Responsibility Research Centre whichs used to construct the G-index. Four of the six provisions (staggered boards, limitsn shareholder amendments of the by-law, supermajority requirement for mergersnd supermajority requirements for charter amendments) set constitutional limitsn shareholders’ voting power. The remaining two provisions reduce the impact ofarket for corporate control (poison pill and golden parachute). They argue that

he remaining provisions were uncorrelated with firm valuation.

7 f Econ

3

3

ttwmtd

dsaQfiretoti(tdttveFamd

tda

w

uIStccs(o(

iufit

to

c(EhtE

L

wypvaApvbtaeeteTb

tFeva(ptaao

2 V. Baulkaran / The Quarterly Review o

. Methodology and data

.1. Methodology

To examine the effects of managerial entrenchment on valua-ion discount of dual class firms I estimate Eq. (1). In estimatinghe models, I adjust the standard errors for firm-level clusteringhich is usually associated with panel data. I expect entrenchmenteasures to be negatively related to dual class discount. That is,

he higher the level of managerial entrenchment, the greater theiscount.

Dual class discountj,t+1 = ̨ + � Entrenchmentj,t

+ˇ1 Excess compensationj,t + ˇ2 Mgmt. votej,t

+ˇ3(Mgmt. vote × excess cash)j,t

+ˇ4 Excess cashj,t + ˇ5 Financial leveragej,t

+ˇ6 Conversion rightj,t + ˇ7 Sizej,t + ˇ8 Div. preferencej,t

+ˇ9 CEO-Chairman dualityj,t + εj,t

(1)

The dependent variable, dual class discount, is computed as theifference in Tobin’s Q ratio of dual class firms and their propen-ity score matched single class concentrated control firms.9 Anlternative definition for dual class discount is based on industry

ratio. It is the difference between Tobin’s Q ratio of dual classrms and the industry average Q ratio. There are several manage-ial entrenchment measures utilised in the specification (1). First,xcess CEO tenure is the difference in tenure for dual class CEOs andheir matching counterpart in single class firms with concentratedwnership. Second, industry adjusted CEO tenure is computed ashe difference between tenure of dual class CEOs and the medianndustry CEO tenure.10 Third, industry adjusted directors’ tenureIADirectors tenure) is the difference in the median tenure per direc-or less the median industry tenure per director. Entrenchment ofirectors is likely to be more pronounced in dual class firms becausehe controlling shareholders of these firms have a stronger incen-ive to elect directors willing to support the interests of controllingersus outside shareholders. Also, Morck et al. (1988) suggest thatven outside board members are capable of becoming entrenched.inally, Excess E-index, utilises the E-index which consists of severalnti-takeover provisions that may result in managerial entrench-ent. Excess E-index is the difference between the E-index value of

ual class firms and matching single class firms.Amoako-Adu, Baulkaran, and Smith (2012) provide evidence

hat excess compensation and excess cash holding is related to theual class discount.11 Therefore, I control for excess compensationnd excess cash holdings in Eq. (1). To measure excess CEO

9 In order to estimate Tobin’s Q ratio I follow the approach used by Lins (2003)

here Q =(

Market value of equity+Total assets−Book value of equityTotal assets

). Gompers et al. (2010)

sed a similar definition except they subtract deferred taxes from the numerator.n order to calculate the market value equity of the dual class companies, I followmith et al. (2009). For dual class companies with both classes of shares trading,he market value equity is the sum of the market value of the restricted votinglass plus the superior voting class. For companies with only the restricted votinglass shares trading, I add a premium of 5.98% to the price of the restricted votinghares in order to estimate the market value of equity for the superior voting sharesFollowing Zingales (1995) definition for voting premium for firms with both classesf shares trading, the mean (median) premium over the period 1998–2007 is 5.10%5.98%)).10 Using average industry CEO tenure and mean tenure per director produces sim-lar results. Following Villalonga (2004), I compute industry means and medianssing 4-digit and 3-digit SIC codes for which there is a minimum of five single classrms in the industry for a given year. In 9% of the cases, I rely on 2-digit SIC codeso compute the industry mean and median.11 Management voting leverage is the percentage of total votes controlled byhe management and directors divided by the percentage of total equity stakewned by management and directors as a group. Excess cash is defined as:

3

3

olf

(

ca

is

at

o

mv

omics and Finance 54 (2014) 70– 81

ompensation, I follow Zingales (1995), Masulis, Wang, and Xie2009), and Amoako-Adu, Baulkaran, and Smith (2011). I estimateq. (2) using firm characteristics and governance variables thatave been proven to explain executive compensation and extracthe residuals as a measure of CEO total excess compensation usingq. (2).

n(compensation)j,t = ˛i +5∑

k=1

�i,k Firm characteristicsk,j,t−1

+8∑

k=1

ıi,k Governancek,j,t−1 + εi,j,t (2)

here Ln(compensation) is the CEO total compensation for firm j atear t. Total compensation is defined as salary + bonus + other com-ensation + stock options.12 Firm characteristics and governanceariables are based on prior studies such as Core, Holthausen,nd Larcker (1999), Chalmers, Koh, and Stapledon (2006), andmoako-Adu et al. (2011). Firm characteristics include: size,rofitability, growth, risk and financial leverage. The governanceariables are as follows: board size, percentage of independent,usy and grey directors, percentage of institutional ownership, CEOenure, percentage of family members on the board of directorsnd CEO-Chairman duality dummy variable.13 It is possible thatntrenchment and excess compensation are correlated becausentrenched managers have the ability to extract higher compensa-ion from their firm. However, I examined the correlation betweenxcess compensation and the various proxies for entrenchment.he correlation is relatively low with the highest being 0.07etween excess compensation and industry adjusted CEO tenure.

There are other factors that reduced the difference in pricing ofhe two classes of shares.14 Therefore, I controlled for these factors.ollowing Zingales (1995), conversion right is an indicator variablequal to 1 if superior voting shares can be converted into restrictedoting shares and 0 otherwise, market value of equity (size) is useds a proxy for the probability of acquisition and dividend preferenceDiv. preference) is an indicator variable equal to 1 if the dividendaid or payable to restricted voting shares is greater than that ofhe superior voting shares and zero otherwise.15 Following Goyalnd Park (2002), I define CEO-Chairman duality as an indicator vari-ble equal to 1 if the CEO is also the Chairman of the board and 0,therwise.

.2. Data

.2.1. Sample constructionThe dataset used in this research is constructed from a variety

f sources. To construct a sample of U.S. dual class firms, I retrieve aist of firms with dual class share structure from Corporate Libraryor 2005–2007. During this period, Corporate Library identifies all

�cashi,tTotal assetst−1

)Dual class

−(

�cashi,tTotal assetst−1

)Single class

where the change in corporate

ash holdings is cash and marketable securities at the end of year t minus cashnd marketable securities at the end of year t − 1.12 Other compensation includes the value of restricted stock grants, long termncentive payouts, contributions to pension plans, life insurance premiums, con-ulting fees and awards under charitable award programmes.13 A busy director is defined as a director with more than four board membershipsnd grey directors are defined as outside directors who are related to the companyhrough a transactional relationship.14 Zingales (1995) examined price differential of dual class firms with both classesf shares traded.15 In 13.5% of the dual class firms, holders of restricted voting (RV) shares are paidore or will receive more dividends in the future relative to holders of superior

oting (SV) shares.

f Econ

fiadcodfdCfippsd

smcoscasatiofda

rltdntifctfE

g

T

RStfTmAa2

(wms

4

4

ddptaaatiictecsg

aamcsocpaCkfi

ccynlifcti

V. Baulkaran / The Quarterly Review o

rms with dual class common shares as a takeover defense mech-nism for 2005–2007. I further extend the data back to 2001 usingata from Gompers, Ishii, and Metrick (2010).16 Since tenure andompensation data are needed, the list of 1910 dual class firm-yearsver the period 2001–2007 is merged with Execucomp database toetermine whether compensation and tenure data are availableor these firms.17 For each dual class company with compensationata, I retrieve proxy statements from the Securities and Exchangeommission (SEC) website and check the proxy statement for eachrm in the sample to ensure that they are indeed dual class com-anies. Next, using proxy statements, I hand collect voting rightser share and the number of superior voting and restricted votinghares owned by the largest shareholders and management andirectors as a group.

The list of dual class companies is matched with a list ofingle class concentrated control firms using propensity scoreatching.18 I estimate a probit model of the determinants of dual

lass structure and compute a propensity score for each firm basedn several firm and governance characteristics. The propensitycore is then used to match each dual class firm with a similar singlelass company. The following firm and governance characteristicsre used in the propensity score matching exercise: equity owner-hip of the largest shareholder, sales, industry, return on asset,nnual stock return, beta, standard deviation of returns, market-o-book, debt-to-asset, sales growth, board size, proportion ofndependent directors, busy directors, grey directors, institutionalwnership, company age, R&D-to-sales, capex-to-total asset andamily firms. This matching exercise results in a final sample of 792ual class firm-years over the period of 2001–2007. This representsn average of 113 dual class firms per year.19

For each firm, I collect accounting data from Compustat. Ietrieve annual firm-level information such as total assets, sales,ong-term debt, common equity and operating income. In addi-ion, I collect several governance variables and equity ownershipata from Corporate Library and Execucomp. These include theumber of directors, outside related directors and unrelated direc-ors. I use proxy statements filed with the SEC, firm websites andnternet search engines such as Lexus Nexus and Google to identifyamily executives and family directors. I then calculate the per-entage of family members who are directors of the board. In ordero complete the set of control variables, I collect stock return datarom CRSP. Finally, I collect CEO total compensation (TDC1 fromxecucomp).

16 A list of dual class IPOs is available on Jay Ritter’s IPO website. Andrew Metrickenerously provided the list of dual class companies used in their study.17 This represents, on average, 272 dual class firms per year over the sample period.his is consistent with the sample used by Howell (2008).18 Propensity score matching methods were developed by Heckman (1979),osenbaum and Rubin (1983), Heckman and Robb (1986), and Heckman, Ichimura,mith, and Todd (1998). One of the major benefits of propensity score matching ishat it can accommodate a larger number of matching variables which can corrector the bias due to systematic differences between the treated and control groups.he greater the overlap in all characteristics of the treated and control groups, theore comparable the groups are and the smaller the bias (Heckman et al., 1998).s a result, propensity score matching has become a popular matching techniquepplied to studies of financial markets (see Hillion & Vermaelen, 2004; Villalonga,004).19 The sample is approximately one-third of the sample used by Gompers et al.2010) due to the following: (1) dual class firms not covered by Execucomp databaseere eliminated due to lack of tenure and compensation data and (2) unification,ergers and bankruptcy as well as the matching procedure further reduce the

ample.

fipaaebmuhe

icacfibh

omics and Finance 54 (2014) 70– 81 73

. Results and discussion

.1. Descriptive statistics

Table 1, Panel A, reports the descriptive statistics of severalifferent characteristics of managerial entrenchment. There is noifference in mean and median CEO age in dual class firms com-ared to single class concentrated control companies. Similarly,here is no difference in the mean (median) age of directors in dualnd single class companies. The median age of directors in both dualnd single class companies is 59 years. The first measure of man-gerial entrenchment, CEO tenure, indicates that dual class CEOsend to remain on the job longer than their matching counterpartsn single class firms. The difference is positive and significant asndicated by the tests for difference in the mean and median. Duallass CEOs, on average, retain their position for 5.32 years longerhan CEOs in single class firms. This is an indication of managerialntrenchment. Alternatively, since dual class firms have a higheroncentration of family involvement, the longer tenure may reflectuch involvement. Family CEOs remain longer in their position toive the next generation time to mature enough to succeed them.

In Table 1, Panel A, I report summary statistics for industrydjusted CEO tenure. The results indicate that dual class CEOs have

longer tenure relative to single class CEOs after adjusting foredian industry CEO tenure. CEOs in dual class firms serve in this

apacity for 8.28 years longer than their industry peers. In compari-on, CEOs in single class concentrated control firms serve in this rolenly 3.08 years longer than their industry peers. This implies thatontrolling shareholders of dual class firms are using their votingower to remain on the job longer or keep in place a CEO who actsccording to the interests of the controlling shareholder. Therefore,EOs of dual class firms are more likely to be entrenched. Investors,nowing this, are more likely to discount the value of dual classrms relative to single class firms.

Using the directors’ tenure as a measure of entrenchment, duallass directors have longer tenure compared to directors in singlelass firms. The mean tenure per director in dual class firms is 1.90ears longer than those in single class concentrated control compa-ies. Also, directors of dual class firms serve as directors 1.74 years

onger than their industry counterparts. In comparison, directorsn single class concentrated control firms retain their directorshipor 1.09 years longer than the industry median. This suggests thatontrolling shareholders use their voting power to elect and main-ain a board of directors who will stay on longer and potentially actn their interest.

E-index as an entrenchment measure is lower for dual classrms than for single class concentrated control companies. Oneossible explanation is that dual class structure is the most effectiventi-takeover defense and therefore, dual class firms do not needdditional anti-takeover defenses such as poison pills. Bebchukt al. (2009) argue that holders of superior voting rights mighte sufficient to provide incumbents with a powerful entrenchingechanism that renders other entrenching provisions relatively

nimportant. Nevertheless, dual class firms on average (median)ave 1.80 (2.0) entrenching provisions which makes up Bebchukt al. (2009) E-index.20

20 Using data available from Lucian Bebchuk’s website the mean (median) E-ndex for all single class firms is 2.4 (2) compared to 1.52 (1) for all duallass companies over the period 1990–2006. In general, dual class firms have

lower E-index compared to single class firms. Hence, it is possible that duallass share structure is enough to entrench managers and therefore, dual classrms are less likely to use other entrenching mechanisms. E-index data foroth dual and single class firms can be accessed at Lucian Bebchuk’s website:ttp://www.law.harvard.edu/faculty/bebchuk/data.shtml.

74V

. Baulkaran

/ The

Quarterly

Review

of Econom

ics and

Finance 54 (2014) 70– 81

Table 1Descriptive statistics.

Dual class Single class Mean test Median test

Mean Median Std dev. Skewness Kurtosis Mean Median Std dev. Skewness Kurtosis t-Stat. z-Stat.

Panel A: Entrenchment characteristicsCEO age (years) 55.88 56.00 7.41 0.57 4.01 55.38 56.00 6.91 0.68 3.56 1.37 1.19Directors’ age (years) 59.33 59.00 4.68 −2.21 28.27 59.43 59.00 3.79 −0.32 3.81 −0.43 −0.31CEO tenure (years) 14.12 10.00 12.13 1.07 3.20 8.80 6.00 8.55 1.91 6.68 9.98*** 9.13***

IACEO tenure 8.28 4.25 12.30 2.73 13.89 3.08 0.50 8.56 1.02 4.56 9.68*** 8.87***

Directors’ tenure 8.58 8.00 5.00 0.84 4.20 6.68 7.00 4.14 0.80 3.51 4.14*** 3.88***

IADirectors’ tenure 1.74 1.00 4.25 7.67 10.7 1.09 0.00 3.93 1.70 3.54 3.13*** 3.65***

E-index 1.80 2.00 1.42 0.38 2.21 2.66 3.00 1.28 −0.11 2.42 −12.5*** −11.9***

G-index 6.09 6.00 1.75 0.07 3.09 6.80 7.00 1.88 0.16 3.14 −7.69*** −7.09***

Perdum2 × IACEO tenure 10.45 8.50 12.76 3.35 14.85 1.82 1.00 8.68 0.35 2.56 4.64*** 7.55***

Perdum2 × IADirectors’ tenure 1.74 1.50 4.15 1.79 6.60 1.07 0.50 4.75 1.05 4.05 2.94*** 2.32**

Panel B: Ownership characteristicsCash flow rights of the largest shareholder % 22.50 18.20 16.50 1.27 4.44 23.60 19.20 13.20 1.63 7.12 −0.71 −0.62Voting rights of the largest shareholder % 57.80 54.80 25.70 −0.03 2.11 23.60 19.20 13.20 1.63 7.12 34.3*** 15.7***

Cash flow rights of management and directors 24.90 19.30 15.60 1.18 3.77 17.10 16.50 14.10 1.52 5.61 14.2*** 4.47***

Voting rights of management and directors 58.30 57.30 25.20 −0.14 2.06 17.10 16.50 14.10 1.52 5.61 44.2*** 15.82***

Management voting leverage (mgmt. vote) 3.01 2.42 2.15 3.34 6.56 1.00 1.00 0.00 − − 58.6* 82.55***

Panel C: Firm characteristicsTobin’s Q ratio 1.86 1.47 1.24 1.35 3.91 2.04 1.62 1.25 0.98 2.39 −3.02*** −4.53***

Industry adjusted Q −0.59 −0.41 1.73 1.41 2.44 −0.39 −0.35 2.26 1.02 2.35 −2.00** −1.98**

Size (sales, $ million) 4917.45 1523.90 14,597.2 0.28 1.67 4653.8 1629.0 15,302.4 0.36 2.03 0.78 −0.89Financial leverage (D/A) 21.61 20.34 18.10 0.98 4.24 21.81 20.82 18.96 0.76 4.51 −0.58 −0.28Performance (Ret) % 12.61 8.35 34.87 1.12 6.58 10.80 7.45 38.52 1.03 4.05 2.18** 1.58Performance (ROA)% 9.75 8.87 8.81 0.26 5.84 9.57 9.06 9.30 −0.02 5.68 0.86 −0.04Risk (Beta) 0.99 0.77 0.81 2.10 8.84 1.06 0.85 0.81 1.54 6.12 −1.46 −1.63Growth % (Total Asset) 9.98 6.67 15.61 4.11 35.07 9.37 6.46 16.64 4.12 36.29 1.07 1.08Institutional ownership % 16.92 13.60 15.66 2.11 5.46 21.49 18.71 14.96 −0.61 1.37 −13.2*** −15.65***

Directors’ age is the average age of a company’s director. Directors’ tenure is the median number of years a company’s director serves in this role. IACEO tenure and IADirectors’ tenure are defined as the industry adjusted tenurefor CEOs and Directors, respectively based on industry median. E-index is the sum of the 6 entrenching provisions identified by Bebchuk et al. (2009) and G-index is the sum of the 24 governance provisions used in Gomperset al. (2003). Perdum2 is a dummy variable equal to 1 if the firm’s 3-year ROA is less than the 3-year industry average ROA. In most cases, 3 and 4 digit SIC industry codes are used to calculate industry mean and median.Cash flow rights is defined as the percentage of equity owned by the largest shareholder or management and directors. Voting rights is the percentage of votes held by the largest shareholder or management and directors.Management voting leverage (Mgmt. vote) is the percentage of total votes controlled by management and directors divided by the percentage of total equity held by management and directors, size (natural logarithm ofsales), financial leverage (total debt divided by total assets), performance (ROA = EBIT divided by total assets and Ret is measured as the annual stock returns), risk (beta is estimated using the CRSP equally weighted indexand the previous 5-year monthly stock returns, when 60 months are not available I required at least 36 months of returns), growth is the geometric mean growth in total assets over the previous three years (five years whenavailable), and institutional ownership (percentage of shares held by institutional investors). The test for difference in mean is the t-test and the test for difference in median is the Wilcoxon signed-rank test.Notes: The number of observations for each of the dual and single class samples is 792 firm-years. Industry adjusted Q-mean is defined as the difference between the firm’s Q ratio and the average SIC Industry Q ratio. Onlynon-zero observations are used to calculate the summary statistics for the interaction terms Perdum2 × IACEO tenure and Perdum2 × IADirector tenure. For the Perdum2 × IACEO tenure, there are 230 observations for thedual class sample and 157 for the single class sample. For the Perdum2 × IADirector tenure, there are 200 observations for the dual class sample and 142 for the single class sample. The median test for the interaction terms isthe Wilcoxon rank sum test.

* Significance level at 10% is indicated.** Significance level at 5% is indicated.

*** Significance level at 1% is indicated.

f Econ

irCscifiow

ft2Mvloflaac

iwattKaevaTv

4

oprfitotidr

ota

si

a

cl

ir

naiemais

eii(ticibp1

pAwode

4

amYaeatpf

pHiamfirip

V. Baulkaran / The Quarterly Review o

Longer tenure during periods of significant underperformances a stronger indication of managerial entrenchment. Therefore, Ieport results conditional on past underperformance. Dual classEOs and directors typically remain on the job longer than theiringle class counterparts even when the firm is underperformingompared to its industry peers. The univariate tests for differencen mean and median are statistically significant. CEOs of dual classrms with poor past performance, relative to the industry, remainn the job 8.63 years longer than similar CEOs in single class firmsith concentrated ownership (Table 1, Panel A).

The last two columns in Table 1, Panel B, show the test statisticsor the difference in means and medians for the two samples. Inhe dual class sample, the largest shareholder owns, on average,2.5% of the equity stake compared to 57.8% of the voting rights.anagement and directors as a group, control 58.3% of the total

otes compared to 24.9% of the equity stake. In comparison, theargest shareholder in single class firms owns, on average, 23.6%f the equity outstanding.21 The disparity between voting and cashow rights in dual class firms is at the heart of the agency problemsssociated with this type of ownership structure. It can allow man-gers to become entrenched with a small proportion of the equityapital.

Table 1, Panel C, reports descriptive statistics for firm character-stics. Dual class firms are valued less compared to single class firms

ith concentrated control. Using mean (median) Q ratio, there isn 8.8% (9.3%) discount for a sample of dual class firms relativeo a sample of single class firms.22 This result is consistent withhe findings of King and Santor (2008), Smith, Amoako-Adu, andalimipalli (2009), and Gompers et al. (2010). Based on the meannd median tests, it is evident that the propensity score matchingxercise produces samples of dual and single class firms that areery similar. There is no difference in size (sales), financial lever-ge (D/A), performance (ROA), risk (beta) and growth (total assets).he tests for mean (median) difference are insignificant for theseariables.

.2. Regression analysis and discussion

Table 2 reports the effects of various entrenchment measuresn the dual class discount. In this table, dual class discount is com-uted as the difference in the Q ratio of dual class firms and the Qatio of propensity score matched single class concentrated controlrms.23 In model (1), the coefficient for excess CEO tenure is nega-ive and statistically significant. As expected, the greater the degreef entrenchment, the larger the discount of dual class firms relativeo single class companies. The results are economically significantn explaining the dual class discount. For example, a one standardeviation change in excess CEO tenure results in a 6.1% discountelative to the average value of a sample of single class firms.24

Furthermore, Goyal and Park (2002) show that the probability

f CEO turnover is significantly lower when the CEO also serves ashe Chairman of the board. Therefore, I include an indicator vari-ble equal to 1 if the CEO is also the Chairman. The coefficient is

21 In the single class matching sample, there are a few cases where the largesthareholder is an institution and hence, mean ownership of the largest shareholders greater than the mean ownership of management and directors (insiders).22 The correlation between dual class discount and excess CEO tenure and industrydjusted CEO tenure is −0.055 and −0.087, respectively.23 An alternative measure of dual class discount is the difference in Q ratio of duallass firms and the industry average Q ratio. The results are similar in significanceevels and magnitude to those reported in Table 2.24 Economic significance is calculated as (0.011 × 11.32)/2.04 = 0.061 or 6.1%, thats standard deviation of excess CEO tenure (not reported) is 11.32 and the mean Qatio for the single class sample is 2.04.

etoeaatDiH

d

omics and Finance 54 (2014) 70– 81 75

ot statistically significant. This implies that the dual role of CEOnd Chairman does not seem to affect dual class discount. Follow-ng Amoako-Adu et al. (2012), I include several other variablesxplaining the dual class discount such as excess compensation,anagement voting leverage and interaction between excess cash

nd management voting leverage. These variables serve as prox-es for the extraction of private benefits. All of these variables areignificant with the expected sign.25

In model (2), excess CEO tenure is measured as the differ-nce between CEO tenure in dual class firms and the medianndustry CEO tenure. As expected, this entrenchment measures negative and statistically significant at the 1% level. In model3), entrenchment is measured using industry adjusted directors’enure (IADirectors’ tenure). The coefficient is negative and signif-cant indicating the greater the entrenchment, the larger the duallass discount. The entrenchment proxy in model (4) is excess E-ndex. The excess E-index is the difference in the E-index valueetween a dual class firm and its matching single class counter-art. The coefficient is negative and statistically significant at the0% level.

Finally, since the correlation among the various entrenchmentroxies are relatively low, I include all the proxies in model (5).26

ll of the entrenchment variables are negative and significanthich confirms my expectation that investors discount the value

f dual class firms which appear to have entrenched managers andirectors. In summary, all of the measures of entrenchment areconomically significant in explaining the dual class discount.

.3. Entrenchment conditional on past performance

The prior literature on managerial entrenchment often utilisednti-takeover provisions, executive tenure, and age to proxy foranagerial entrenchment (Bebchuk et al., 2009; Berger et al., 1997;

ermack, 2006). However, age and tenure can also proxy for valu-ble experience (Norburn & Birley, 1988). Salas (2010) argues thatxecutive tenure conditional on firm performance is a more suit-ble measure of managerial entrenchment as managers are onlyruly entrenched when they are not removed in the face of poorerformance. In light of this argument, it is important to controlor firms’ prior performance when examining entrenchment.

Studies which examine concentrated ownership often sim-ly assume that managers are also the controlling shareholders.owever, this is not always the case. In this study, I am able to

dentify firms where the controlling shareholder is an executivend those where the controlling shareholder is a director or Chair-an of the board. This is important especially in poor performing

rms because the controlling shareholder-CEO is less likely to beeplaced following several years of underperformance. Accord-ngly, I construct two dummy variables based on prior operatingerformance. The first performance dummy variable (Perdum1) isqual to 1 if dual class firms’ previous 3-year average ROA is lesshan the 3-year average ROA of matching single class firms and 0,therwise. The second performance dummy variable (Perdum2) isqual to 1 if the firm’s 3-year ROA is less than the 3-year industryverage ROA and 0, otherwise. The 3-year average ROA is utiliseds a performance measure in order to eliminate the impact of anyransitory effects of operating performance. In addition, Denis and

enis (1995) show that firms with 3-years of prior poor operat-

ng performance are more likely to replace their CEO. Similarly,uson, Parrino, and Starks (2001) provide evidence that executive

25 I conducted multicollinearity diagnostics and there is no multicollinearity evi-ent among the independent variables.26 The highest correlation is 0.40 between IACEO tenure and IADirectors tenure.

76 V. Baulkaran / The Quarterly Review of Economics and Finance 54 (2014) 70– 81

Table 2Effects of entrenchment on discount of dual class Q ratio to that of matching single class firms.

(1) (2) (3) (4) (5)

Excess CEO tenure −0.011*

0.006IACEO tenure −0.014*** −0.012**

0.004 0.005IADirectors’ tenure −0.031** −0.020**

0.014 0.009Excess E-index −0.054* −0.063**

0.032 0.032Excess G-index 0.050 0.052

0.031 0.035Excess compensation −0.228*** −0.223*** −0.236*** −0.231*** −0.227***

0.067 0.068 0.067 0.067 0.068Mgmt. vote × excess cash −0.294 −0.241** −0.241** −0.262 −0.252**

0.245 0.124 0.122 0.247 0.125Mgmt. vote −0.109*** −0.102*** −0.110*** −0.113*** −0.108***

0.031 0.030 0.030 0.031 0.029Excess cash 0.828 0.735 0.742 0.773 0.715

0.805 0.818 0.823 0.803 0.804Financial leverage −0.571 −0.871*** −0.874*** −0.937*** −0.901***

0.373 0.278 0.277 0.278 0.278Conversion rights −0.265* −0.204 −0.183 −0.255 −0.215

0.154 0.152 0.149 0.155 0.150Size 0.142*** 0.123*** 0.123** 0.133*** 0.107**

0.048 0.047 0.048 0.049 0.050Dividend preference −0.333 −0.334 −0.296 −0.233 −0.289

0.252 0.253 0.253 0.251 0.251CEO-Chairman duality 0.143 0.148 0.105 0.124 0.132

0.129 0.126 0.125 0.125 0.126Intercept −0.689 −0.643 −0.577 −0.757* −0.447

0.450 0.420 0.437 0.432 0.466

Industry and year effects Yes Yes Yes Yes YesAdjusted R2 0.242 0.237 0.256 0.267 0.289Obs. 792 792 792 792 792

The dependent variable, dual class discount, is computed as the difference between the Q ratio of dual class firms and the Q ratio of matching single class concentratedcontrol firms. Excess CEO tenure is the difference in tenure for dual class CEOs and their matching counterparts in single class firms. IACEO tenure is the difference betweenthe tenure of dual class CEOs and the median industry CEO tenure. Excess E-index is the difference between the E-index value of dual class firms and matching single classfirms. In constructing the excess G-index, I subtract the E-index value for each firm from the G-index value and then take the difference between dual class firms’ G-indexand matching single class firms’ G-index. IADirectors’ tenure is computed as the median tenure per director less the median industry tenure per director. Mgmt. vote is theratio of total votes divided by total equity ownership by management and directors as a group, excess cash is defined as change in cash holdings from year t − 1 to t dividedby total assets for a dual class firm minus that of a matching single class firm, financial leverage is defined as total debt/total assets, size is the natural logarithm of sales,conversion rights is an indicator variable equal to 1 if superior voting shares can be converted into restricted voting shares and 0 otherwise, and dividend preference is anindicator variable equal to 1 if the dividend paid or payable to restricted voting shares is greater than that of the superior voting shares and zero, otherwise. CEO-Chairmanduality is an indicator variable equal to 1 if the CEO is also the Chairman of the board and 0, otherwise. The standard-errors are adjusted for firm-level clustering and arereported below the estimated coefficients.

* Significant at the 10% level.

tmHp

euirma((

atceai

cr

fttac(sa

controlling shareholder is not a member of the executive team.In these firms, the controlling shareholders are chairs or directorsof the board. The results are similar to those presented in Table 3

** Significant at the 5% level.*** Significant at the 1% level.

urnover tends to occur when industry adjusted accounting perfor-ance has declined and stock returns have recently been negative.ence, CEOs with long tenure during periods of underperformancerovides strong evidence of managerial entrenchment.

Table 3 presents the results of the analysis of managerialntrenchment when the controlling shareholder is also an exec-tive conditioned on prior operating performance. I include an

nteraction term between a measure of prior underperformanceelative to a benchmark and various measures of entrenchment. Inodels (1) and (4), the benchmark is based on the performance of

matching group of control firms (Perdum1). In models (2), (3) and5), the benchmark is based on the industry average performancePerdum2).

In models (1) and (3), the performance dummy variable is inter-cted with excess CEO tenure. The coefficient of the interactionerm is negative and statistically significant. This indicates that dual

lass firms with previous poor performance and greater managerialntrenchment are valued less when the controlling shareholder islso the CEO. The results are economically significant in explain-ng the dual class discount. For example, a one standard deviation

ir

hange in excess CEO tenure for firms with poor past performanceesults in an 18.9% discount.27

In model (2), I used industry as the benchmark for both per-ormance (Perdum2) and entrenchment (industry adjusted CEOenure). The results show that investors apply a larger discounto dual class firms with worse performance relative to the industrynd when the CEO remains on the job longer than their industryounterpart. In model (4), I used industry adjusted directors’ tenureIADirectors’ tenure) as the proxy for entrenchment. The resultshow that firms with poor performance and longer directors’ tenurere valued less.

Table 4 presents the results for dual class firms where the

27 Economic significance is calculated as (0.034 × 11.32)/2.04 = 0.189 or 18.9%, thats standard deviation of excess CEO tenure (not reported) is 11.32 and the mean Qatio for the single class sample is 2.04.

V. Baulkaran / The Quarterly Review of Economics and Finance 54 (2014) 70– 81 77

Table 3Entrenchment effect on dual class discount for a sub-sample of firms with controlling shareholders-executives.

(1) (2) (3) (4) (5)

Perdum1 × excess CEOtenure

−0.034***

0.010Excess CEO tenure 0.012

0.008Perdum1 0.124

0.223Perdum2 × IACEO tenure −0.015**

0.007IACEO tenure −0.011

0.017Perdum2 0.158

0.289Perdum2 × excess CEOtenure

−0.022**

0.009Excess CEO tenure 0.006

0.009Perdum2 0.059

0.021Perdum1 × IADirectors’tenure

−0.103***

0.026IADirectors’ tenure 0.003

0.022Perdum1 0.202

0.220Perdum1 × excess E-index 0.025

0.071Excess E-index −0.101*

0.057Excess G-index 0.045

0.048Perdum1 −0.028

0.193Excess compensation −0.236** −0.133 −0.265*** −0.137 −0.242**

−0.092 0.101 0.092 0.096 0.096Mgmt. vote × excess cash 0.089 0.041 0.068 −0.023 −0.007

0.330 0.331 0.329 0.343 0.300Mgmt. vote −0.048** −0.049* −0.064 −0.072*** −0.063**

0.024 0.026 0.044 0.027 0.027Excess cash −0.452 −0.304 −0.400 −0.075 −0.198

1.013 0.959 1.003 0.950 1.032Financial leverage −0.034 −0.063 −0.03 −0.051 −0.048

0.049 0.051 0.049 0.046 0.050Conversion rights −0.163 −0.161 −0.171 −0.154 −0.263

0.227 0.221 0.229 0.227 0.254Size 0.291*** 0.252*** 0.280*** 0.280*** 0.287***

0.062 0.063 0.060 0.062 0.063Dividend preference −0.900** −0.866* −0.853* −0.871** −0.733*

0.455 0.446 0.458 0.431 0.421CEO-Chairman duality −0.071 −0.115 −0.108 −0.169 −0.081

0.174 0.172 0.172 0.170 0.169Intercept −1.941*** −1.472** −1.804*** −1.561*** −1.883***

0.579 0.581 0.587 0.600 0.594

Industry and year effects Yes Yes Yes Yes YesAdjusted R2 0.102 0.106 0.120 0.113 0.103Obs. 486 486 486 486 486

The dependent variable, dual class discount, is computed as the difference between the Q ratio of dual class firms and the Q ratio of matching single class concentrated controlfirms. Excess CEO tenure is the difference in tenure for dual class CEOs and their matching counterparts in single class firms. IACEO tenure is the difference between the tenureof dual class CEOs and the median industry CEO tenure. Excess E-index is the difference between the E-index value of dual class firms and matching single class firms. Inconstructing the excess G-index, I subtract the E-index value for each firm from the G-index value and then take the difference between dual class firms’ G-index and matchingsingle class firms’ G-index. IADirectors’ tenure is computed as the median tenure per director less the median industry tenure per director. Perdum1 is a dummy variableequal to 1 if dual class firms’ previous 3-year average ROA is less than the ROA of matching single class firms. Perdum2 is a dummy variable equal to 1 if the firm’s 3-yearaverage ROA is less than the 3-year industry average ROA. Mgmt. vote is the ratio of total votes divided by total equity ownership by management and directors as a group,excess cash is defined as change in cash holdings from year t − 1 to t divided by total assets for a dual class firm minus that of a matching single class firm, financial leverageis defined as total debt/total assets, size is the natural logarithm of sales, conversion rights is an indicator variable equal to 1 if superior voting shares can be converted intorestricted voting shares and 0, otherwise and dividend preference is an indicator variable equal to 1 if the dividend paid or payable to restricted voting shares is greater thanthat of the superior voting shares and zero, otherwise. CEO-Chairman duality is an indicator variable equal to 1 if the CEO is also the Chairman of the board and 0, otherwise.The standard-errors are adjusted for firm-level clustering and are reported below the estimated coefficients.

* Significant at the 10% level.** Significant at the 5% level.

*** Significant at the 1% level.

78 V. Baulkaran / The Quarterly Review of Economics and Finance 54 (2014) 70– 81

Table 4Entrenchment effect on dual class discount for a sub-sample of firms with non-controlling shareholders-executives.

(1) (2) (3) (4) (5)

Perdum1 × excess CEOtenure

−0.0240.015

Excess CEO tenure 0.0020.100

Perdum1 −0.0140.233

Perdum2 × IACEO tenure −0.052***

0.015IACEO tenure 0.004

0.009Perdum2 0.160

0.203Perdum2 × excess CEOtenure

0.0220.015

Excess CEO tenure −0.023**

0.010Perdum2 −0.146

0.214Perdum1 × IADirectors’tenure

−0.060**

0.028IADirectors’ tenure −0.037

0.029Perdum1 −0.006

0.200Perdum1 × excess E-index −0.044

0.083Excess E-index 0.020

0.058Excess G-index 0.009

0.034Perdum1 −0.043

0.209Excess compensation −0.271*** −0.365*** −0.294*** −0.361*** −0.283***

0.095 0.106 0.097 0.106 0.097Mgmt. vote × excess cash −0.683** −0.696*** −0.642** −0.640** −0.693**

0.284 0.265 0.252 0.260 0.280Mgmt. vote −0.079*** −0.075*** −0.157*** −0.083*** −0.086***

0.025 0.023 0.044 0.023 0.025Excess cash 2.468*** 2.343** 2.320*** 2.191** 2.433***

0.934 0.915 0.882 0.876 0.939Financial leverage −0.041 −0.035 −0.020 −0.044 −0.015

0.064 0.059 0.063 0.070 0.065Conversion rights 0.163 0.161 0.166 0.231 0.167

0.204 0.200 0.204 0.202 0.206Size 0.159** 0.134** 0.161** 0.123** 0.152**

0 .068 0.069 0.070 0.065 0.069Dividend preference 0.212 0.221 0.198 0.420 0.255

0.287 0.314 0.274 0.293 0.290CEO-Chairman duality 0.511*** 0.560*** 0.481*** 0.432*** 0.511***

0.163 0.165 0.160 0.166 0.164Intercept −0.620 −0.458 −0.405 −0.321 −0.562

0.601 0.611 0.636 0.583 0.621

Industry and year effects Yes Yes Yes Yes YesAdjusted R2 0.203 0.224 0.205 0.224 0.194Obs. 306 306 306 306 306

The dependent variable, dual class discount, is computed as the difference between the Q ratio of dual class firms and the Q ratio of matching single class concentrated controlfirms. Excess CEO tenure is the difference in tenure for dual class CEOs and their matching counterparts in single class firms. IACEO tenure is the difference between the tenureof dual class CEOs and the median industry CEO tenure. Excess E-index is the difference between the E-index value of dual class firms and matching single class firms. Inconstructing the excess G-index, I subtract the E-index value for each firm from the G-index value and then take the difference between dual class firms’ G-index and matchingsingle class firms’ G-index. IADirectors’ tenure is computed as the median tenure per director less the median industry tenure per director. Perdum1 is a dummy variableequal to 1 if dual class firms’ previous 3-year average ROA is less than the ROA of matching single class firms. Perdum2 is a dummy variable equal to 1 if the firm’s 3-yearaverage ROA is less than the 3-year industry average ROA. Mgmt. vote is the ratio of total votes divided by total equity ownership by management and directors as a group,excess cash is defined as change in cash holdings from year t − 1 to t divided by total assets for a dual class firm minus that of a matching single class firm, financial leverageis defined as total debt/total assets, size is the natural logarithm of sales, conversion rights is an indicator variable equal to 1 if superior voting shares can be converted intorestricted voting shares and 0, otherwise and dividend preference is an indicator variable equal to 1 if the dividend paid or payable to restricted voting shares is greater thanthat of the superior voting shares and zero, otherwise. CEO-Chairman duality is an indicator variable equal to 1 if the CEO is also the Chairman of the board and 0, otherwise.The standard-errors are adjusted for firm-level clustering and are reported below the estimated coefficients.

* Significant at the 10% level.** Significant at the 5% level.

*** Significant at the 1% level.

V. Baulkaran / The Quarterly Review of Economics and Finance 54 (2014) 70– 81 79

Table 5Predicted entrenchment effect on dual class discount for a sub-sample of firms with controlling shareholders-executives.

(1) (2) (3) (4) (5) (6)

Perdum × predicted excessCEO tenure

−0.136*** −0.100***

0.024 0.019Predicted excess CEOtenure

0.094*** 0.059**

0.032 0.026Perdum 0.015 0.123

0.125 0.132Perdum × predicted IACEOtenure

−0.105*** −0.073***

0.017 0.013Predicted IACEO tenure 0.044 0.088***

0.038 0.034Perdum 0.065 0.120

0.133 0.130Perdum × predictedIADirectors’ tenure

−0.497*** −0.381***

0.083 0.061Predicted IADirectors’tenure

−0.524 0.960**

0.452 0.374Perdum 0.057 0.149

0.139 0.132Excess compensation −0.186*** −0.183*** −0.198*** −0.384*** −0.382*** −0.376***

0.069 0.069 0.070 0.065 0.065 0.065Mgmt. vote × excess cash −0.275** −0.261* −0.223** −0.266** −0.268** −0.303**

0.139 0.144 0.112 0.113 0.113 0.154Mgmt. vote −0.133*** −0.117*** −0.170*** −0.035 −0.054 0.014

0.034 0.034 0.045 0.035 0.035 0.038Excess cash 0.641 0.623 0.530 0.720 0.703 0.826

0.772 0.779 0.803 0.632 0.628 0.640Financial leverage −0.039 −0.029 −0.025 −0.026 −0.027 −0.026

0.035 0.035 0.035 0.036 0.036 0.035Conversion rights −0.323 −0.323 −0.296 −0.194 −0.220 −0.226

0.155 0.151 0.147 0.117 0.116 0.118Size 0.155*** 0.148*** 0.144*** 0.228*** 0.217*** 0.225***

0.048 0.049 0.048 0.050 0.049 0.049Dividend preference −0.419* −0.43* −0.44* 0.304* 0.306* 0.343*

0.245 0.244 0.240 0.180 0.178 0.183CEO-Chairman Duality 0.165 0.154 0.148 0.062 0.060 0.057

0.127 0.127 0.126 0.105 0.105 0.104Intercept −0.95** −0.744 0.733 −1.613*** −1.849*** −3.02***

0.448 0.465 1.078 0.399 0.449 0.839

Sagan test p-value 0.198 0.523 0.148 0.216 0.369 0.214Adjusted R2 0.112 0.124 0.128 0.163 0.169 0.173Obs. 486 486 486 486 486 486

The dependent variable in models (1)–(3) is computed as the difference between the Q ratio of dual class firms and the Q ratio of matching single class concentrated controlfirms and the dependent variable in models (4)–(6) is the difference between the Q ratio of dual class firms and industry average Q ratio. Predicted Excess CEO Tenure is thedifference in predicted tenure for dual class CEOs and their matching counterparts in single class firms. Predicted IACEO tenure is predicted industry adjusted tenure of dualclass CEOs based on industry median. Predicted IADirectors’ tenure is the predicted median industry adjusted tenure per director. Perdum is a dummy variable equal to 1 ifdual class firms’ previous 3-year average ROA is less than the ROA of matching single class firms. The standard-errors are adjusted for firm-level clustering and are reportedbelow the estimated coefficients.

fetitchat

4

erben

fafmtprimhffounder-CEO (correlation = 0.22). Similarly, corporate governanceprovisions which make up the E-index such as poison pills mayalso render the market of corporate control ineffective and hence,

* Significant at the 10% level.** Significant at the 5% level.

*** Significant at the 1% level.

or models (2) and (4). This finding suggests that managerialntrenchment is less of a concern when the CEO is not the con-rolling shareholder. It is interesting to note however, that theres weak evidence to indicate that entrenchment is still a concerno shareholders even when the CEO of a dual class firm is not aontrolling shareholder. Executives who are not controlling share-olders can also be entrenched because controlling shareholdersre more likely to hire executives who are less likely to act againsthe interest of the controlling shareholder.

.4. Two-stage least squares

Potential endogeneity concerns are common in corporate gov-rnance literature. Simultaneity and reverse causality can bias our

esults. Managerial entrenchment can lead to dual class discountut it is unlikely that dual class discount influences managerialntrenchment. Nevertheless, I utilise a two stage least square tech-ique as a robustness check. In the first stage, I estimate a model o

or the determinants of entrenchment (excess CEO tenure, industrydjusted CEO tenure and Industry adjusted directors’ tenure). Theollowing regressors are included: CEO ownership, age, manage-

ent voting leverage, founder-CEO, E-index, G-index (excludinghe E-index value) and dual class discount.28 Ownership and age areositively correlated with CEO tenure (correlation of 0.42 and 0.64,espectively). Dual class ownership structure (management vot-ng leverage) is arguably the most effective anti-takeover defense

echanism which can allow managers to become entrenched andence correlated with CEO tenure (correlation = 0.37). In addition,

ounder-CEO tends to have longer tenure compared to non-

28 Founder-CEO is a dummy variable equal to 1 if the founder is the CEO and 0,therwise.

8 f Econ

laen

ppcCdfrAio

4

detamwfot

mocrbSti(cd

sbmpoiopt

5

tc

ats

a4

vcaadcpoaaaddw

acjTHrcamLfiewy

eotadcc

siacam

A

AmMs

0 V. Baulkaran / The Quarterly Review o

ead to entrenchment. Ownership, age, management voting lever-ge and founder-CEO are statistically significant (not tabulated) inxplaining CEO tenure. Dual class discount, e-index and g-index areot statistically significant in the first stage regression.

Using the coefficient from the first stage, I construct the variousredicted measures of entrenchment. In the second stage, theseredicted variables are then used to explain the documented duallass discount. I utilised three measures of entrenchment including:EO tenure, industry adjusted CEO tenure and industry adjustedirectors’ tenure. I created interaction terms between the per-ormance dummy and the predicted entrenchment variables. Theesults reported in Table 5 are similar to those presented in Table 3.lso, based on the p-values for the Sagan test for overidentify-

ng restrictions which are reported in Table 5, I fail to reject theveridentification restrictions hypothesis.29

.5. Robustness checks

In order to exclude the potential bias in estimating dual classiscount using a matching firm approach, an alternative method ofstimating discount is utilised. Dual class discount is estimated ashe difference between Q ratio of a dual class firm and its industrydjusted Q ratio. Following Villalonga (2004), I compute industryean and median Q ratio using 4-digit and 3-digit SIC codes forhich there is a minimum of five single class firms in the industry

or a given year from the Compustat universe.30 Using this measuref discount, I re-estimate the models presented in Tables 2–4 andhe results are similar in significance and magnitude.

As an additional robustness check, I utilised a more traditionalatching technique that is, using size (sales), equity ownership

f the largest shareholder and SIC industry code to match duallass firms with single class firms. I re-estimate the models and theesults are similar to those presented above. In addition, it is possi-le that the information environment among S&P Large 500 firms,&P Mid 500 firms and S&P small firms are different and are likelyo affect the results presented above.31 Hence, I split the samplento S&P 500 and S&P Mid and Small Cap. Even though the resultsnot tabulated) are slightly stronger for the Mid-cap and Small-ap group, it does not appear that the results presented above areriven by the smaller firms.

In addition, instead of a ratio of voting rights to equity owner-hip (Mgmt. vote), I construct a wedge variable as the differenceetween voting rights and equity ownership and re-estimate theodel. The results (not tabulated) are qualitatively similar to those

resented in the paper. I also re-examined the valuation discountf dual class firms using only those with both classes of shares trad-ng on the exchange. There are 18 dual class firms (126 firm-yearbservations) that have both classes of shares trading in the sam-le. The regression results (not reported) are qualitatively similaro those presented above.

. Conclusions

Dual class ownership structure is one of the most effective anti-akeover defense mechanisms. It can lead to entrenchment sinceontrolling shareholders can maintain a voting block of superior

29 I check the correlation between the residuals from models (1) to (3) in Table 5nd the variables used in first stage. The correlations are low ranging from −0.06o 0.096 and not statistically significant at the 5% level. I thank the referee for thisuggestion.30 In 9% of the sample, I relied on 2-digit SIC codes to compute the industry aver-ges. In 33% of the sample, I used 3-digit SIC codes and in 68% of sample, I used-digit SIC codes to estimate the mean and median industry Q ratio.31 I thank the referee for making this suggestion.

c

R

A

A

B

B

omics and Finance 54 (2014) 70– 81

oting shares and still raise additional equity capital. Entrenchmentan be viewed as a benefit to controlling shareholders but imposesgency costs on outside shareholders. Therefore, outside investorsre likely to attach lower value to dual class firms with a greateregree of entrenchment. In this paper, I investigate whether duallass firms have a greater degree of managerial entrenchment com-ared to single class firms and whether investors discount the valuef dual class companies which tend to have entrenched managersnd directors. The evidence provided in the research shows thatfter using a propensity score matching technique, dual class CEOsnd directors are entrenched compared to single class CEOs andirectors. After adjusting for industry median, dual class CEOs andirectors have longer tenure than CEOs in single class firms evenhen the firm is making losses.

Univariate tests indicate that dual class CEOs and directors have longer tenure than their counterparts in single class firms withoncentrated control. Dual class CEOs, on average, remain on theob for 5.32 years longer than CEOs in similar single class firms.his evidence can be interpreted as managerial entrenchment.owever, longer tenure may also indicate experience and supe-

ior performance in running the firm. Therefore, I construct tests,onditional on poor past firm performance, using the industry and

matching sample as benchmarks. When firms perform poorly,anagers who are not entrenched are more likely to lose their job.

onger tenure of CEOs and directors in dual class firms when theserms have been performing poorly is consistent with managerialntrenchment. The univariate test shows that in dual class firmsith poor past performance, CEOs have a longer tenure (8.63 more

ears) compared to CEOs in similar single class firms.In a panel regression analysis, I use several measures of

ntrenchment and provide evidence that the greater the degreef entrenchment, the larger the dual class discount. Excess CEOenure (compared to matching firms and industry median) leads to

greater discount of dual class firms. Also, investors apply a greateriscount to dual class firms with longer directors’ tenure. The dis-ount is more pronounced when the measures of entrenchment areonditional on poor past performance.

I examine sub-samples of dual class firms with controllinghareholders as CEOs and those where the controlling shareholders a director or Chairman of the board. It appears that shareholdersre more sensitive to managerial entrenchment when the CEO is aontrolling shareholder. In summary, the paper finds that investorsre concerned with the negative effects of managerial entrench-ent and therefore, attach a lower value to dual class companies.

cknowledgements

I would like to thank Ben Amoako-Adu, Brian Smith, Ebenezersem, Jessica Johnson, Massimo Guidolin (Co-Editor), two anony-ous referees, the participants of the Eastern Finance Associationeeting (2012), and faculty members of Wilfrid Laurier Univer-

ity and University of Lethbridge for their suggestions and helpfulomments.

eferences

moako-Adu, B., Baulkaran, V., & Smith, B. F. (2011). Executive compensation infirms with concentrated control: The impact of dual class structure and familymanagement. Journal of Corporate Finance, 17, 1580–1594.

moako-Adu, B., Baulkaran, V., & Smith, B. F. (2012). Dual class discount, agencyproblems and wealth expropriation. In Wilfrid Laurier University Working Paper

ates, T. W., Becher, D. A., & Lemmon, M. L. (2008). Board classification and man-agerial entrenchment: Evidence from the market for corporate control. Journalof Financial Economics, 87, 656–677.

ebchuk, L., & Cohen, A. (2005). The costs of entrenched boards. Journal of FinancialEconomics, 78, 409–433.

f Econ

B

B

B

B

B

C

C

C

D

F

F

G

G

G

H

H

H

H

H

H

H

J

K

L

L

M

M

N

O

R

R

S

S

S

S

S

V

W

W10, 1–5.

V. Baulkaran / The Quarterly Review o

ebchuk, L., & Stole, L. (1993). Do short-term managerial objectives lead to under-oroverinvestment in long-term projects? Journal of Finance, 48, 719–729.

ebchuk, L., Cohen, A., & Ferrell, A. (2009). What matters in corporate governance?Review of Financial Studies, 22, 783–827.

ebchuk, L., Cohen, A., & Wang, C. C. Y. (2012). Learning and the disappearing asso-ciation between governance and returns. Journal of Financial Economics, 108,323–348.

ennedsen, M., & Nielsen, K. M. (2010). Incentive and entrenchment effects in Euro-pean ownership. Journal of Banking and Finance, 34, 2212–2229.

erger, P., Ofek, F., & Yermack, D. L. (1997). Managerial entrenchment and capitalstructure decisions. Journal of Finance, 52, 1411–1438.

halmers, K., Koh, P. S., & Stapledon, G. (2006). The determinants of CEO com-pensation: Rent extraction or labour demand? British Accounting Review, 38,259–275.

laessens, S., Djankov, S., & Lang, L. (2000). The separation of ownership and controlin East Asian corporations. Journal Financial Economics, 58, 81–112.

ore, J., Holthausen, R., & Larcker, D. (1999). Corporate governance, chief executiveofficer compensation and firm performance. Journal of Financial Economics, 51,371–406.

enis, D. J., & Denis, D. K. (1995). Performance changes following top managementdismissals. Journal of Finance, 50, 1029–1057.

aleye, O. (2007). Classified boards, firm value and managerial entrenchment. Jour-nal of Financial Economics, 83, 501–529.

lorackis, C., & Ozkan, A. (2009). The impact of managerial entrenchment on agencycosts: An empirical investigation using UK Panel Data. European Financial Man-agement, 15, 497–528.

ompers, P. A., Ishii, J., & Metrick, A. (2003). Corporate governance and equity prices.Quarterly Journal of Economics, 118, 107–155.

ompers, P. A., Ishii, J., & Metrick, A. (2010). Extreme governance: An analysis ofdual-class firms in the United States. Review of Financial Studies, 23, 1051–1088.

oyal, V. K., & Park, C. W. (2002). Board leadership structure and CEO turnover.Journal of Corporate Finance, 8, 49–66.

arford, J., Mansi, S. A., & Maxwell, W. F. (2008). Corporate governance and firmcash holdings in the US. Journal of Financial Economics, 87, 535–555.

eckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47,153–161.

eckman, J., & Robb, R. (1986). Alternative methods for solving the problem of selec-tion bias in evaluating the impact of treatments on outcomes. In H. Wainer(Ed.), Drawing inference from self-selected samples (pp. 63–107). Berlin: Springer-Verlag.

eckman, J., Ichimura, H., Smith, J., & Todd, P. (1998). Characterizing selection bias

using experimental data. Econometrica, 66, 1017–1098.

illion, P., & Vermaelen, T. (2004). Death spiral convertibles. Journal of FinancialEconomics, 71, 381–415.

owell, J. W. (2008). No more share classes: A study of U.S. dual class stock unifica-tion. In University of Georgia Working paper.

Y

Z

omics and Finance 54 (2014) 70– 81 81

uson, M. R., Parrino, R., & Starks, L. T. (2001). Internal monitoring mecha-nisms and CEO turnover: A long-term perspective. Journal of Finance, 56,2265–2297.

arrell, G., & Poulsen, A. (1988). Dual-class recapitalization as antitakeover mecha-nisms: The recent evidence. Journal of Financial Economics, 20, 129–152.

ing, M. R., & Santor, E. (2008). Family values: Ownership structure, performanceand capital structure of Canadian firms. Journal of Banking and Finance, 32,2423–2432.

a Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (1999). Corporate ownership aroundthe world. Journal of Finance, 54, 471–517.

ins, K. (2003). Equity ownership and firm value in emerging markets. Journal ofFinancial and Quantitative Analysis, 38, 159–184.

asulis, R. W., Wang, C., & Xie, F. (2009). Agency problems at dual-class companies.Journal of Finance, 64, 1697–1727.

orck, R., Shleifer, A., & Vishny, R. (1988). Management ownership and marketvaluation: An empirical analysis. Journal of Financial Economics, 20, 293–315.

orburn, D., & Birley, S. (1988). The top management team and corporate perfor-mance. Strategic Management Journal, 9, 225–237.

zkan, A., & Ozkan, N. (2004). Corporate cash holdings: An empirical investigationof UK companies. Journal of Banking and Finance, 28, 2103–2134.

osenbaum, P., & Rubin, D. (1983). The central role of the propensity score in obser-vational studies for causal effects. Biometrika, 70, 41–55.

uback, R. S. (1988). Coercive dual-class exchange offers. Journal of Financial Eco-nomics, 20, 153–173.

alas, J. M. (2010). Entrenchment, governance, and the stock price reaction to suddenexecutive deaths. Journal of Banking and Finance, 34, 656–666.

hleifer, A., & Vishny, R. W. (1989). Management entrenchment: The case ofmanager-specific investments. Journal of Financial Economics, 25, 123–139.

mith, B. F., Amoako-Adu, B., & Kalimipalli, M. (2009). Concentrated control andcorporate value: A comparative analysis of single and dual class structures inCanada. Applied Financial Economics, 19, 955–974.

tein, J. (1988). Takeover threats and managerial myopia. Journal of Political Economy,96, 61–80.

tulz, R. (1988). Managerial control of voting rights: Financing policies and themarket for corporate control. Journal of Financial Economics, 20, 25–54.

illalonga, B. (2004). Does diversification cause the “diversification discount”?Financial Management, 33, 5–27.

eisbach, M. S. (1988). Outside directors and CEO turnover. Journal of FinancialEconomics, 20, 431–460.

ilcox, J. C. (2002). Two cheers for staggered boards. Corporate Governance Advisor,

ermack, D. (2006). Flights of fancy: Corporate jets, CEO perquisites, and inferiorshareholder returns. Journal of Financial Economics, 80, 211–242.

ingales, L. (1995). What determines the value of corporate votes? Quarterly Journalof Economics, 110, 1047–1073.