Académique Documents
Professionnel Documents
Culture Documents
www.elsevier.com/locate/jbf
a,*
c,2
Department of Accounting, Economics and Finance, Texas A&M International University, Laredo,
TX 78041, United States
Department of Economics and Finance, Northern Kentucky University, Highland Heights, KY 41099,
United States
c
Department of Finance, College of Business and Administration, Southern Illinois University,
Carbondale, United States
d
Atkinson Graduate School of Management, Willamette University, Salem, United States
Received 22 September 2004; accepted 29 August 2005
Available online 7 December 2005
Abstract
Grounded in agency theory, this study investigates how the strength of shareholder rights inuences the extent of rm diversication and the excess value attributable to diversication. The empirical evidence reveals that the strength of shareholder rights is inversely related to the probability to
diversify. Furthermore, rms where shareholder rights are more suppressed by restrictive corporate
governance suer a deeper diversication discount. Specically, we document a 1.11.4% decline in
rm value for each additional governance provision imposed on shareholders. An explicit distinction
is made between global and industrial diversication. Our results support agency theory as an explanation for the value reduction in diversied rms. The evidence in favor of agency theory appears to
be more pronounced for industrial diversication than for global diversication.
2005 Elsevier B.V. All rights reserved.
948
1. Introduction
Considerable research has explored the issue of corporate diversication. One critical
question is whether corporate diversication enhances or destroys value. Early researchers
argued in favor of diversication citing factors such as greater operating eciency, the
presence of an internal capital market, greater debt capacity, and lower taxes (for example,
Fluck and Lynch, 1999; Bradley et al., 1998; Kaplan and Weisbach, 1992; Porter, 1987;
Ravenscraft, 1987, among others).
On the contrary, several academic studies in the 1990s provide evidence on the destructive eect on rm value of corporate diversication (for example, Comment and Jarrell,
1995; Liebeskind and Opler, 1995; Lang and Stulz, 1994; Servaes, 1996; Berger and Ofek,
1995; Denis et al., 2002, among others). More recently, arguments have been advanced
and new evidence presented that diversication may be benecial or, at the minimum,
not value-destroying (Villalonga, 2004; Whited, 2001; Campa and Kedia, 2002; Mansi
and Reeb, 2002). Others have suggested that it may be the acquisition of poorly performing units (Graham et al., 2002) or miscalculations of Tobins q (Whited, 2001) that explain
the diversication discount. Hence, the debate on the impact of diversication still continues in the literature.
Motivated by agency theory, we contribute to the literature in this area by exploring the
role of the agency costs in explaining the value discount (or premium?) caused by diversication. In so doing, we examine the relation between rm value, corporate governance,
shareholder rights and the propensity to diversify. We employ the governance index developed by Gompers et al. (2003) to represent the strength of shareholder rights. Gompers
et al. (2003) construct a governance index on the basis of how many corporate governance
provisions exist that restrict shareholder rights, with a higher index indicating weaker
shareholder rights.
This study examines the inuence of shareholder rights both on the extent of diversication and on the excess value arising from diversication. First, we investigate the relation between the propensity to diversify and the strength of shareholder rights. We nd
evidence that rms where shareholder rights are weak are more likely to be industrially
diversied. This evidence is in favor of the explanation that managers exploit the weak
shareholder rights and diversify the rm unwisely. As a result, industrially diversied rms
exhibit a reduction in value. The evidence on global diversication, however, is more
ambiguous. We nd no relation between the strength of shareholder rights and the propensity to be diversied globally. Hence, global diversication does not appear to be motivated by managers taking advantage of weak shareholder rights. The value reduction
aliated with global diversication (Denis et al., 2002), therefore, may not be explained
by the agency cost perspective.
Second, we investigate the impact of shareholder rights on rm value. To measure the
valuation eects, we use the concept of excess value, rst developed by Berger and Ofek
(1995). We document that more restrictive corporate governance is associated with lower
949
excess value in diversied rms. Apparently, where shareholder rights are weaker, rms
suer a more severe reduction in value. The detrimental eect on rm value of weak shareholder rights is found in all of the diversication categories except for global diversication. This evidence is consistent with an agency cost explanation. Diversied rms
where shareholder rights are weak (and, therefore, management powers are strong) are
expected to suer from acute agency costs created by the separation of ownership and control. More specically, we document that each additional restrictive governance provision
imposed on shareholders diminishes the excess value by approximately 1.11.4% on
average.4
The study is organized as follows. We discuss our hypotheses in Section 2. The sample
selection criteria and data are discussed in Section 3. Then, Section 4 displays the empirical
evidence and, nally, Section 5 concludes.
2. Hypothesis development
2.1. Propensity to diversify and shareholder rights
Jensen (1986) argues that, when managers have access to free cash ow, they tend to
spend it unwisely reducing shareholder wealth. There are several ways in which the free
cash ow could be wasted. One possibility might be for managers to consume extra perquisites that are unnecessary. Another possibility could be managers attempting to expand
the rm through acquisitions in unrelated business segments that may not supply adequate
returns to shareholders. We make an explicit distinction between global and industrial
diversication and argue that weak shareholder rights enable managers to diversify the
rm (perhaps, unwisely) either globally or industrially or both. Hence, we hypothesize
an inverse relation between the strength of shareholder rights and the propensity for diversication, the weaker the shareholder rights, the more diversied the rm is expected to be.
Our rst three hypotheses are related to the strength of shareholder rights and the propensity to diversify globally, industrially or both.
H1: When shareholder rights are more restricted, rms are more likely to be globally
diversied.
H2: When shareholder rights are more restricted, rms are more likely to be industrially
diversied.
H3: When shareholder rights are more restricted, rms are likely to be both globally and
industrially diversied.
950
Industrial Diversification
Domestic
Single-segment
Multi-segment
Single-segment Domestic
Multi-segment Domestic
(SD)
(MD)
Focused
Single-segment Global
Multi-segment Global
(SG)
(MG)
Global
Diversification
Global
diversified
Fig. 1. Global and industrial diversication classication. Single-segment rms operate in only one industrial
segment whereas multi-segment rms operate in more than one industrial segment. Domestic rms operate only
in the US while global rms operate in, at least, one country outside the US. Single-segment Domestic rms
operate in only one segment and only in the US. Multi-segment Domestic rms operate in more than one
industrial segment but only in the US. Single-segment Global rms operate in only one industrial segment but
have a presence aboard. Multi-segment Global rms operate in more than one segment and also outside the US.
document a diversication discount whereas others oer evidence of a diversication premium. Motivated by agency theory, we argue that diversication that results from agency
conicts is likely to be value-destroying. Since agency conicts are likely more severe in
rms with weaker shareholder rights, we hypothesize a positive association between the
strength of shareholder rights and rm value.
H4: When shareholder rights are more restricted, rms experience a deeper diversication discount.
951
1862
2000
1800
1600
1400
1200
1993
1000
1995
653
423
776
507
1998
800
Total
279
600
609
176
312
180
108
400
141
477
187
96
200
Total
185
106
75
140
1998
156
1995
SD
MD
1993
SG
MG
Total
Fig. 2. Year distribution of the sample rms. SD stands for single-segment domestic, SG for single-segment
global, MD for multi-segment domestic, and MG for multi-segment global.
Firms are excluded that have segments in the nancial industry (SIC codes 60006999)
and the utility industry (SIC codes 49004999) because these industries are subject to regulations, rendering the characteristics of their nancial information incomparable to those
in other industries.7 We further reduce the sample by excluding those observations that do
not have data on the governance index in the Investor Responsibility Research Center
(IRRC). The IRRC collects data on corporate governance. However, the IRRC collects
data only periodically and our sample is, therefore, restricted to the years in which the
IRRC has data. For our study, we use data from 1993, 1995 and 1998.
Each rm in the nal sample is, then, classied into one of the four diversication categories. Fig. 1 shows the four diversication regimes classied along two diversication
Additional constraints are imposed as in Berger and Ofek (1995). We exclude rms with sales less than $20
million and rms where the dierence between the sum of the segment sales and total sales exceeds 1%.
952
dimensions, industrial and global. The nal sample consists of a total of 1862 rm-year
observations. Fig. 2 presents the year distribution of rms in the sample.
3.2. The governance index (GINDEX)
To measure the strength of shareholder rights, we employ the governance index (GINDEX) developed by Gompers et al. (2003)henceforth GIM. They use data from the
Investor Responsibility Research Center (IRRC), which publishes detailed listings of corporate governance provisions for individual rms in Corporate Takeover Defenses (Rosenbaum, 1993, 1995, and 1998). The data on governance provisions are derived from various
Table 1
Individual governance provisions employed in the construction of the governance index
Percentage of rms with governance provisions in
1993
1995
1998
Full sample
Delay
Blank check
Classied board
Special meeting
Written consent
82.60
63.52
31.66
34.59
86.70
64.04
35.47
36.45
87.37
58.89
33.12
33.51
85.93
61.76
33.51
34.75
Protection
Compensation plans
Contracts
Golden parachutes
Indemnication
Liability
Severance
67.92
16.98
53.03
40.04
72.12
4.61
73.89
14.45
51.89
38.42
67.98
87.03
61.73
12.11
51.42
25.39
48.32
12.50
67.29
14.12
51.99
33.44
60.85
9.24
Voting
Bylaws
Charter
Cumulative voting
Secret ballot (condential voting)
Supermajority
Unequal voting
15.93
2.52
14.05
14.47
21.80
2.94
15.11
2.63
13.46
14.29
20.03
2.46
16.75
2.96
10.70
9.41
15.21
1.93
16.00
2.74
12.46
12.30
18.47
2.36
Other
Anti-greenmail
Directors duties
Fair price
Pension parachutes
Poison pill
Silver parachutes
8.18
8.18
38.57
7.55
63.30
17.34
7.88
7.39
34.98
5.25
60.92
4.60
5.67
6.19
2.60
2.96
56.44
2.96
7.04
7.09
3.22
4.89
59.67
4.62
State
Anti-greenmail law
Business combination law
Cash-out law
Directors duties law
Fair price law
Control share acquisition law
16.56
91.20
3.56
4.82
32.49
2.47
15.44
90.80
3.12
3.95
31.52
24.96
13.27
92.27
2.45
3.74
29.38
22.68
14.82
91.51
2.95
4.08
30.88
23.95
9.63
9.59
8.87
9.30
Governance index
The detailed explanation for each governance provision is available in the Appendix of Gompers et al. (2003).
953
sources, such as corporate bylaws, charters, proxy statements, annual reports, as well as
10-K and 10-Q documents led with the Security and Exchange Commission (SEC).
The individual governance provisions included in the construction of the governance index
are displayed in Table 1. They classify provisions into 5 categories: tactics for delaying
hostile bidders (Delay); voting rights (Voting); director/ocer protection (Protection);
other takeover defenses (Other); and state laws (State).8 Table 1 shows the percentage
of rms in our sample that have each provision in each sample year.
The governance index is constructed as follows; for every rm, GIM add one point for
every provision that restricts shareholder rights (increases managerial power). While this
index does not accurately reect the relative impacts of the various provisions, it has
the advantage of being transparent and easily reproducible. The index does not require
any judgments about the ecacy or wealth eects of any of these provisions; GIM only
considers the impact on the balance of power.
To clarify the logic behind the construction of the governance index, GIM use the following example; consider classied boards, a provision that staggers the terms and elections of directors and, thus, can be employed to slow down a hostile takeover. If
management uses this power judiciously, it could possibly lead to an increase in overall
shareholder wealth; if management, however, uses this power to maintain private benets
of control, then this provision would diminish shareholder wealth. Either way, it is apparent that classied boards enhance the power of managers and weaken the control rights of
large shareholders. Hence, the governance index captures the balance of power between
management and shareholders.
Most provisions other than classied boards can be viewed with the same logic. Almost
every provision enables management to resist dierent types of shareholder activism, such
as calling special meetings, changing the rms charter or bylaws, suing the directors, or
replacing them all at once. GIM note, however, that there are two exceptions, secret ballots (condential voting) and cumulative voting. A secret ballot or condential voting designates a third party to count proxy votes and, therefore, prevents management from
observing how specic shareholders vote. Cumulative voting enables shareholders to concentrate their directors votes so that a large minority shareholder can ensure some board
representation. These two provisions are usually proposed by shareholders and opposed
by management because they enhance shareholder rights and diminish the power of management. Thus, for each one, GIM add one point to the governance index when rms do
not have it. For all other provisions, GIM add one point when rms do have each of the
provisions. In summary, the governance index is simply the sum of one point for the presence (or absence) of each provision.
3.3. Excess value
To measure the value of globally and industrially diversied rms, we use the excess
value measure following the modied version of Berger and Ofek (1995). We use the single-segment domestic rm as the benchmark to compute excess value.9 The excess value
8
The detailed explanation for each governance provision is available in the appendix of GIM.
We argue that the excess value measure is superior to Tobins q (Lang and Stulz, 1994) because Tobins q
requires replacement cost in the denominator and neither foreign ination rates nor the exchange rate eect are
taken into account.
9
954
measure is computed following Berger and Ofek (1995) with the modication in Bodnar
et al. (1999), and is dened as follows:
EVi;t logMVi;t =Imputed Valuei;t ;
X
SSalei;t Multiplier;
Imputed Valuei;t
where: EVi,t is the excess value for rm i in year t; MVi,t is the rms market capitalization
(market value of common equity plus book value of debt) for rm i in year t; and Imputed
Value is the sum of segment sales multiplied by the sales multiplier. The Multiplier is measured as the median total market capitalization to sales for the single-segment domestic
rms in the same industry in the same year. A positive excess value indicates that the entire
rm is worth as a whole more than the sum of its segments whereas a negative excess value
shows that the rm as a whole is worth less than the sum of its segments. Thus, a positive
excess value implies a diversication premium while a negative excess value indicates a
diversication discount.
4. Empirical evidence
4.1. Summary statistics
Descriptive statistics for selected rm characteristics for our sample appear in Table 2
(Panel A). The multi-segment domestic rm is not signicantly larger than its singleTable 2
Summary statistics
Domestic
SD
Panel A: Descriptive statisticsa
Sales
2605.62
Total assets
2021.95
21.85%
Debt ratiob
EBIT/sales
7.73%
CAPX/sales
10.72%
R&D/sales
3.89%
Advertising/sales
10.85%
% Managerial ownership
14.13%
MD
t-Statistics
Global
SG
MG
t-Statistics
Full
sample
2846.33
2575.76
26.02%
9.50%
6.90%
0.63%
8.26%
12.69%
0.60
1.66*
2.87***
1.55
3.81***
3.39***
1.35
0.92
2910.84
2725.20
19.43%
10.52%
9.93%
6.06%
13.10%
10.74%
6093.02
6524.77
26.24%
10.03%
7.51%
3.01%
14.20%
8.46%
5.51***
5.13***
7.35***
0.67
2.64***
6.98***
0.52
2.39**
3947.82
3875.55
23.36%
9.56%
8.80%
3.69%
12.25%
11.03%
Mean (%)
Median (%)
0.34
10.18
7.66
24.65
Total
11.39%
12.28%
SD
279
507
653
423
49.24
62.34
53.89
50.74
1862
55.38
955
Table 3
Univariate tests for the governance index
Single-segment
mean (median)
Multi-segment
mean (median)
t-Test
Z-score
Domestic vs.
global
8.53
8.00
(423)
9.74
10.00
(279)
5.51***
5.31***
9.01
9.00
(702)
9.87
10.00
(653)
5.53***
5.65***
8.96
9.00
(507)
9.47
10.00
(1160)
t-Test
Z-score
2.24**
2.14**
0.65
0.93
Single-segment vs.
Multi-segment
N
8.73
9.00
(930)
9.83
10.00
(932)
Domestic
N
Global
*
**
***
3.37***
3.55***
8.23***
8.21***
segment counterpart in terms of sales but signicantly larger in terms of total assets (2605.62
vs. 2846.33 in sales and 2021.95 vs. 2575.76 in total assets). Multi-segment global rms,
however, are considerably larger than single-segment rms both in terms of sales and total
assets (2910.84 vs. 6093.77 in sales and 2725.20 vs. 6524.77 in total assets). Multi-segment
rms have larger debt ratios, smaller capital expenditures, and smaller R&D expenditures.
Likewise, global rms have larger debt ratios, lower capital expenditures and smaller R&D
expenditures. In terms of managerial (executives and board members) ownership, there is
no signicant dierence between single-segment domestic (SD) rms and multi-segment
domestic (MD) rms (14.13% vs. 12.69%). However, the dierence is signicant between
single-segment global (SG) rms and multi-segment global (MG) rms (10.74% vs. 8.46%).
Panel B of Table 2 displays the excess value10 by diversication type. It should be noted
that the excess value for single-segment domestic rms is the highest, suggesting that
focused rms are more valuable. As mentioned earlier, we do not concentrate on ascertaining whether the excess value is, on average, positive or negative. Rather, our focus is on
determining the association between the excess value and the strength of shareholder
rights.
4.2. Propensity to diversify and shareholder rights
Table 3 presents the univariate analysis for the governance index and the extent of rm
diversication. In the last column, we test domestic rms against global rms regardless of
10
The excess value is constructed as described in Berger and Ofek (1995). The benchmark rm is the singlesegment domestic rm. In computing the excess value, we use the entire universe of rms with available data on
COMPUSTAT. Then, we retain only the observations where the governance index is available. The summary
statistics shown here are only for the observations that remain in the nal sample. Firms that are followed by the
IRRC tend to be large and well established. Hence, rms with low excess value are, perhaps, less likely to be
included. This may explain why both the mean and the median are positive although diversication is found to be
value-destroying.
956
Table 4
Logistic regressions predicting global and industrial diversication with the governance index and controls
Dependent dichotomous
variable
Model 1
(Wald statistics)
Global
Model 2
(Wald statistics)
Industrial
Model 3
(Wald statistics)
Global
Intercept
1.100***
(34.20)
0.100
(0.26)
0.693***
(236.99)
2.137**
(5.81)
2.111***
(126.17)
0.099***
(30.82)
0.457***
(149.27)
0.756
(1.19)
0.967***
(15.20)
0.015
(0.34)
0.711***
(125.88)
3.136
(6.92)
0.000
(0.89)
No. of observations
Pseudo R2
1862
22.4%
1862
16.2%
1006
22.5%
Governance index
Relative log (total assets)
Relative R&D to sales
Model 4
(Wald statistics)
Industrial
2.113***
(73.25)
0.112***
(20.70)
0.449***
(72.97)
0.689
(0.66)
0.000
(0.48)
1006
16.2%
The governance index is as dened in Gompers et al. (2003). Excess value is calculated based on Berger and Ofek
(1995).
*
Statistically signicant at the 10% level.
**
Statistically signicant at the 5% level.
***
Statistically signicant at the 1% level.
whether they are industrially diversied or not. The average number of governance provision for the domestic rm is 9.01 (median 9.00) and that for the global rm is 9.47 (median
10.00). The dierent is statistically signicant at the 1% level both in the t-test and in the
distribution-free non-parametric test (Z-score). The evidence reveals that weaker shareholder rights (more restrictive governance) are associated with global diversication.
In the last row of Table 3, we compare single-segment rms with multi-segment rms,
regardless of whether they operate globally or just domestically. The average number of
governance provisions of the single-segment rm is 8.73 (median 9.00) while that for
the multi-segment rm is 9.83 (median 10.00). The dierence is statistically signicant at
the 1% level. Firms where shareholder rights are weak seem to be industrially diversied.
We also divide the sample further into single-segment domestic (SD), single-segment
global (SG), multi-segment domestic (MG) and multi-segment global (MD) rms. The
multi-segment global rm (MG) has 9.87 governance provisions on average (median
10.00). This is higher than the numbers of governance provisions in the other three groups.
The results imply that weaker shareholder rights are associated with a combination of
both global and industrial diversication.
We enhance the univariate analysis with a logistic regression analysis presented in Table
4. A number of control variables are included.11 First, several prior studies show that rm
size impacts the extent of corporate diversication. For instance, Denis et al. (1997) provide evidence that the number of business segments in which a rm operates is positively
related to rm size. Similarly, Singh et al. (2004) provide evidence that rm size is a
11
Alternative control variables are used and produce qualitatively similar results. To capture the potential
industry eects, we industry-adjust the control variables by subtracting the industry median from the value of
each variable for a given rm. The rst two digits of the SIC codes are employed to identify the industry.
957
Table 5
Multinomial logistic regression predicting rm diversication with the governance index and controls
Intercept
Governance index
Relative log (total assets)
Relative R&D to sales
Pseudo R2
MD (Wald statistics)
SG (Wald statistics)
MG (Wald statistics)
2.361***
(66.75)
0.115***
(16.37)
0.466***
(45.52)
1.397
(1.56)
1.252***
(26.81)
0.002
(0.00)
0.700***
(127.23)
2.650**
(6.07)
3.108***
(131.51)
0.088***
(11.64)
1.121***
(298.16)
2.664
(5.14)
26.7%
The governance index is as dened in Gompers et al. (2003). Excess value is calculated based on Berger and Ofek
(1995).
*
Statistically signicant at the 10% level.
**
Statistically signicant at the 5% level.
***
Statistically signicant at the 1% level.
12
Our managerial ownership variable measures the direct stock ownership by managers and board of directors
and is computed as the number of shares of stock held by the executives and board members as a percentage of
total shares outstanding.
13
The data on managerial ownership are available for only 1006 observations in the sample. Hence, we include
managerial ownership in a separate set of regressions.
958
14
An alternative regression is run where managerial ownership is included (results omitted). The results remain
similar.
15
It can be argued that governance structure and diversication are endogenously determined. If this is the case,
then, the simultaneous equations framework may be more suitable for testing H1, H2, and H3. As a robustness
check, we run Hausmans specication test to check for the presence of simultaneity. The Hausman tests are all
statistically insignicant. There is no evidence of simultaneity. We conclude that endogeneity does not seem to
impact the results.
16
These control variables are employed in a number of prior studies on diversication (Denis et al., 2002; Berger
and Ofek, 1995; among others).
17
The coecient of the global dummy is signicant at the 15% level, however.
959
Table 6
Regressions of the excess value on the governance index, global and industrial diversication dummies and controls
Model 1
(t-statistics)
Model 2
(t-statistics)
Model 3
(t-statistics)
Model 4
(t-statistics)
Model 5
(t-statistics)
Model 6
(t-statistics)
0.022
(0.662)
0.036
(1.11)
0.079**
(2.10)
0.123**
(2.22)
0.011*
(1.94)
0.168***
(2.87)
0.011**
(1.98)
Multi-segment
0.051
(1.42)
Multi-segment domestic
0.120
(3.58)
0.135**
(2.38)
0.014**
(2.45)
0.052
(1.45)
Single-segment global
Multi-segment global
Entrenchment index
(Bebchuk et al., 2004)
Relative log (total assets)
Intercept
Governance index
Global
***
***
Model 7
(t-statistics)
0.117***
(2.73)
0.199***
(3.74)
0.101**
(2.17)
0.168***
(3.54)
0.110
(3.26)
0.189***
(3.55)
0.103**
(2.22)
0.160***
(3.36)
0.048***
(3.83)
1.098***
(8.43)
0.326***
(3.64)
1.054***
(7.23)
1.490***
(6.19)
1.458***
(2.64)
0.000*
(1.82)
0.055***
(4.51)
1.060***
(8.18)
0.335***
(3.79)
1.028***
(7.09)
1.471***
(6.15)
1.403**
(2.55)
0.000**
(1.97)
0.060***
(4.67)
1.065***
(8.21)
0.334***
(3.75)
1.018***
(7.01)
1.447***
(6.05)
1.450***
(2.64)
0.000**
(2.02)
0.055***
(4.31)
1.074***
(8.24)
0.329***
(3.68)
1.035***
(7.10)
1.479***
(6.16)
1.389**
(2.51)
0.000*
(1.67)
0.060***
(4.81)
1.043***
(8.04)
0.339***
(3.83)
1.017***
(7.02)
1.464***
(6.13)
1.349***
(2.46)
0.000*
(1.85)
0.065***
(4.96)
1.047***
(8.07)
0.337***
(3.79)
1.006***
(6.93)
1.440***
(6.03)
1.400**
(2.54)
0.000*
(1.90)
0.193***
(3.62)
0.102**
(2.18)
0.162***
(3.41)
0.022***
(1.82)
0.062***
(4.80)
1.050***
(8.10)
0.330***
(3.71)
1.021***
(7.04)
1.431***
(5.99)
1.403**
(2.56)
0.000*
(1.90)
971
21.55***
14.5%
971
23.14***
15.4%
971
19.10***
15.7%
971
19.92***
15.7%
971
21.04***
16.5%
971
17.77***
16.9%
971
17.71***
16.9%
Excess Value is computed as described in Berger and Ofek (1995). The governance index is constructed based on Gompers et al.
(2003). Global is equal to one if the rm is globally diversied regardless of whether it is industrially diversied or not, zero
otherwise. Multi-segment is equal to one if the rm is industrially diversied regardless of whether it is globally diversied or not,
zero otherwise. Multi-segment domestic is equal to one if the rm is diversied industrially but not globally, zero otherwise.
Single-segment global is equal to one if the rm is diversied globally but not industrially, zero otherwise. Multi-segment global is
equal to one if the rm diversied both industrially and globally, zero otherwise.
*
Statistically signicant at the 10% level.
**
Statistically signicant at the 5% level.
***
Statistically signicant at the 1% level.
960
Table 7
Regressions of the excess value on the interaction terms and controlsa excess value is computed as described in
Berger and Ofek (1995)
Intercept
Governance index global
Governance index multi-segment
Governance index
multi-segment domestic
Governance index
single-segment global
Governance index
multi-segment global
Governance index
% Managerial ownership
multi-segment domestic
% Managerial ownership
single-segment global
% Managerial ownership
multi-segment global
Managerial ownership (%)
Control variables included
No. of observations
F-statistics
Adjusted R2
Model 1
(t-statistics)
Model 2
(t-statistics)
Model 3
(t-statistics)
Model 4
(t-statistics)
0.111***
(2.45)
0.009***
(3.41)
0.005*
(1.94)
0.104**
(2.32)
0.074
(1.27)
0.003
(0.12)
0.010***
(3.00)
0.070
(1.21)
Model 5
(t-statistics)
0.058
(0.93)
0.008*
(1.69)
0.018***
(4.33)
0.020***
(5.22)
0.018***
(4.61)
0.002
(0.33)
0.005
(1.13)
0.019***
(3.90)
0.011*
(1.88)
0.014***
(2.71)
0.001
(0.33)
0.000*
(1.61)
0.00*
(1.69)
0.016**
(2.51)
0.007
(1.15)
0.011*
(1.82)
0.003
(0.44)
0.002
(1.11)
0.003
(1.19)
0.003
(0.97)
0.002
(1.13)
Yes
1862
32.70***
13.3%
Yes
1862
31.24***
14.0%
Yes
971
18.60***
15.4%
Yes
971
17.41***
15.7%
Yes
971
13.81***
16.8%
The governance index is constructed based on Gompers et al. (2003). Global is equal to one if the rm is globally
diversied regardless of whether it is industrially diversied or not, zero otherwise. Multi-segment is equal to one
if the rm is industrially diversied regardless of whether it is globally diversied or not, zero otherwise. Multisegment domestic is equal to one if the rm is diversied industrially but not globally, zero otherwise. Singlesegment global is equal to one if the rm is diversied globally but not industrially, zero otherwise. Multi-segment
global is equal to one if the rm diversied both industrially and globally, zero otherwise.
*
Statistically signicant at the 10% level.
**
Statistically signicant at the 5% level.
***
Statistically signicant at the 1% level.
a
We also run additional regressions by diversication type and obtain qualitatively similar results except for
the multi-segment global rm.
results. The results suggest that rms with weak shareholder rights (restrictive governance)
suer a deeper diversication discount.18 The predictions of H4 are, therefore, supported.
18
As a robustness check, we replace the global dummy with foreign sales as a continuous alternative measure of
global diversication and re-run the regression (results omitted). The results are similar. Likewise, we replace the
multi-segment dummy with the Herndahl index as a continuous alternative measure of industrial diversication
and re-run the regression (results not shown). The results are, again, consistent with the results arrived at using
dummy variable specication.
961
The estimated coecient of the governance index varies from 0.011 to 0.014, indicating that for each additional governance provision imposed on shareholder rights, the
excess value declines by approximately 1.11.4%. Because the average rm in the sample
has about 9 governance provisions, the average discount on rm value is about 9.912.6%
that can be attributed to the suppression of shareholder rights through strict corporate
governance. A discount of this magnitude is both statistically and economically signicant.
Finally, a recently study by Bebchuk et al. (2004) constructs an Entrenchment Index
based on 6 of the 24 governance provisions19 in Gompers et al. (2003). They contend that
this index can better explain rm value (represented by Tobins q) and stock returns than
the governance index- specically, the higher the Entrenchment Index, the lower the rm
value. Following Bebchuk et al. (2004), we create the Entrenchment Index and include it in
the regression in Model 7. The coecient of the Entrenchment Index is negative and signicant, suggesting that managerial entrenchment is associated with a value reduction.
Therefore, our results agree with those in Bebchuk et al. (2004).
We now examine whether the detrimental eect is uniform across dierent diversication categories. We accomplish this by constructing a number of interaction terms between
the governance index and the various diversication dummies. The results of the regressions with the interaction terms appear in Table 7. Model 1 includes two interaction terms,
one that combines the governance index with the global dummy variable and another that
combines the governance index with the multi-segment dummy variable. Both of these
interaction terms produce negative and signicant estimated coecients. The results reveal
that restrictive governance reduces rm value both in the presence of global diversication
and industrial diversication.20
In Model 2, we include three interaction terms in the regression, each term corresponding to the governance index interacting with each diversication dummy. All of the interaction terms in Model 2 display negative and statistically signicant estimated coecients,
implying that restrictive governance provisions destroy value in all of these diversication
categories.
We repeat the previous regressions in Models 3 and 4 but add managerial ownership as
a control variable. Interestingly, in Model 3, when managerial ownership is included, the
estimated coecient of the interaction term between the governance index and global
diversication becomes insignicant. The results for the rest of the interaction terms are
qualitatively similar to those in Models 1 and 2. The results in Model 3 indicate that, after
controlling for managerial ownership, restrictive governance (weak shareholder rights)
does not reduce the excess value in rms that are globally diversied. It does so, however,
in the other diversication regimes.
Finally, in Model 5, we control for the interaction eects between managerial ownership and diversication. The three variables that interact ownership with the diversication dummies are added as controls in Model 5. The results in Model 5 are similar to
19
The six provisions included in the Entrenchment Index are staggered boards, limits to shareholder bylaw
amendments, supermajority requirements for mergers, supermajority requirements for charter amendments,
poison pills, and golden parachutes.
20
In addition, the estimated coecient for the governance index is negative and signicant. This negative sign
suggests that weak shareholder rights may, perhaps, exacerbate the agency problem and is associated with
reduced value. However, this variable becomes insignicant in later models (and is only signicant at the 0.10
level in Model 1) so the evidence that weak shareholder rights reduce value may be marginal, at best.
962
those in Model 4 except for the interaction term between the governance index and the
single-segment global dummy, which loses its signicance. This is hardly surprising, however, given that the governance index does not explain global diversication as much as
industrial diversication in the previous analyses.
In summary, there is a positive (negative) relationship between the strength of shareholder rights (corporate governance) and rm value. The diversication discount is more
severe when the rm has more governance provisions that restrict shareholder rights. This
relation does hold in all of the diversication categories except for global diversication
even after controlling for ownership structure.
5. Concluding remarks
Because of the on-going debate on the costs and benets of diversication, we contribute to the literature by empirically examining the potential connections between corporate
governance, shareholder rights, rm value, and the propensity for a rm to be diversied.
The governance index developed by Gompers et al. (2003) is employed as the measure of
the strength of shareholder rights. There is evidence that when shareholder rights are more
restricted, the rm is more likely to be diversied. We argue that weak shareholder rights
allow management to diversify the rm unwisely, resulting in a decline in value.
The excess value developed by Berger and Ofek (1995) is used as a proxy for rm value.
The evidence in our study reveals that rms where shareholder rights are more suppressed
by restrictive governance provisions suer a deeper diversication discount. This is true for
all diversication categories except for global diversication. When shareholder rights are
weak, agency costs created by the separation of ownership and control are likely to be
more acute. As a result, the diversication discount is more severe.
Our study contributes to the literature both in corporate diversication and agency theory. In corporate diversication, our results complement ndings of those studies that
identify agency costs as responsible for the value reduction (Denis et al., 1997; Hyland
and Diltz, 2002, among others). Consistent with the agency theory perspective, we demonstrate that restrictive corporate governance provisions may enable management to pursue
strategies that are not necessarily consistent with shareholders wealth maximization (in
this particular instance, suboptimal diversication that destroys rm value). Our study
also contributes by considering the relative eects of global and industrial diversication
separately whereas most other studies take into account only industrial diversication
(with a notable exception of Denis et al., 2002).
In conclusion, our results complement those of Denis et al. (1997). Both studies nd
empirical support for agency theory. Unlike Denis et al. (1997), however, our focus is
on the strength of shareholder rights and corporate governance whereas theirs is on managerial ownership. The results of both studies, nevertheless, are remarkably similar in the
sense that they provide support for agency conict as responsible for the diversication
discount.
References
Bebchuk, L., Cohen, A., Ferrell, A., 2004. What matters in corporate governance? Working Paper, Harvard
University and National Bureau of Economic Research.
Berger, P.G., Ofek, E., 1995. Diversications eect on rm value. Journal of Financial Economics 37, 3965.
963
Bodnar, G.M., Tang, C., Weintrop, J., 1999. Both sides of corporate diversication: The value impacts of
geographic and industrial diversication. Working Paper, Johns Hopkins University.
Bradley, M., Desai, A., Kim, E., 1998. Synergistic gains from corporate acquisitions and their division between
the stockholders of target and acquiring rms. Journal of Financial Economics 21, 340.
Campa, J.M., Kedia, S., 2002. Explaining the diversication discount. Journal of Finance 57, 17311762.
Comment, R., Jarrell, G., 1995. Corporate focus and stock returns. Journal of Financial Economics 37, 6788.
Denis, D.J., Denis, D.K., Sarin, A., 1997. Agency problems, equity ownership, and corporate diversication.
Journal of Finance 52, 135160.
Denis, D.J., Denis, D.K., Yost, K., 2002. Global diversication, industrial diversication, and rm value. Journal
of Finance 57, 19511979.
Fluck, Z., Lynch, A., 1999. Why do rms merge and then divest? A theory of nancial synergy. Journal of
Business 72, 319346.
Gompers, P., Ishii, J., Matrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of
Economics 118, 107155.
Graham, J.R., Lemmon, M., Wolf, J., 2002. Does corporate diversication destroy value? Journal of Finance 59,
695720.
Hyland, D., Diltz, D., 2002. Why rms diversify? An empirical examination. Financial Management 31, 5182.
Jensen, M., 1986. Agency costs of free cash ow, corporate nance, and takeovers. American Economic Review
76, 323329.
Kaplan, S., Weisbach, M., 1992. The success of acquisitions: Evidence from divestitures. Journal of Finance 48,
107138.
Lang, L.H.P., Stulz, R., 1994. Tobins q, corporate diversication and rm performance. Journal of Political
Economy 102, 12481280.
Liebeskind, J., Opler, T., 1995. The causes of corporate refocusing: Evidence from the 1980s. Working Paper,
University of Southern California.
Mansi, S., Reeb, D.M., 2002. Corporate diversication: What gets discounted? Journal of Finance 57, 21672184.
McConnell, J., Servaes, H., 1990. Additional evidence on equity ownership corporate value. Journal of Financial
Economics 27, 595612.
Morck, R., Shleifer, A., Vishny, R., 1988. Management ownership and market valuation: An empirical analysis.
Journal of Financial Economics 20, 293316.
Porter, M., 1987. From competitive advantage to corporate strategy. Harvard Business Review 65, 4359.
Ravenscraft, D., 1987. The 1980s mergers wave: An industrial organization perspective. In: Browne, L.,
Rosengren, E. (Eds.), The Merger Boom: An Overview. Federal Reserve Bank of Boston, Boston, pp. 1737.
Servaes, H., 1996. The value of diversication during the conglomerate merger wave. Journal of Finance 51,
12011225.
Singh, Manohar, Mathur, I., Gleason, K.C., 2004. An analysis of interrelationship among corporate governance,
ownership structure and diversication strategies. Financial Review 39, 489526.
Villalonga, B., 2004. Does diversication cause the diversication discount. Financial Management 33, 528.
Whited, T., 2001. Is it ecient investment that causes the diversication discount? Journal of Finance 56, 1667
1692.