Managers in the familiar_20130326x - College of Business

positiveriverΔιαχείριση

20 Νοε 2013 (πριν από 3 χρόνια και 11 μήνες)

255 εμφανίσεις


1

Does CEOs’ familiarity with business segments affect their
divestment decisions?
*


James Ang

College of Business

FloridaStateUniversity

e
-
mail:
jang@cob.fsu.edu


Abe de Jong
+

R
otterdam
S
chool of
M
anagement

Erasmus University

e
-
mail:
ajong@rsm.nl


Marieke van der Poel

Rotterdam School of Management

Erasmus University

e
-
mail:
mpoel@rsm.nl


March
201
3






*

We are grateful to Malcolm Baker, Guillermo Baquero, John Doukas, Marie Dutordoir, Denys Glushkov, Nancy
Huyghebaert, Ulrike Malmendier, Gerard Mertens, Peter Roosenboom, Len Rosenthal,
Frederik Schlingemann,
Jack
Stecher, Jeroen Suijs, Mathijs van Dijk,
an anonymous reviewer,
participants at the 2005 ERIM Conference
Financial Management Track, participants at the 2006 EFMA conference (Madrid), participants at the 2007 EFA
conference (New Orleans), participants at the 2007 EFA conference (Ljubljana), parti
cipants at the 2007 FMA
conference (Orlando), seminar participants at RSM Erasmus University, at the University of Antwerp, at the
Norwegian School of Economics and Business Administration, at the Catholic University of Leuven, at Florida State
University,

and at Tilburg University for providing helpful comments and suggestions. We are also grateful to the
Vereniging Trust Fonds for providing financial sup
port and to Sandra Sizer for

excellent editing.

+

Corresponding author:Department of Financ
e

(Room T0
9
-
55
), RSM Erasmus University, PO Box 1738, 3000 DR
Rotterdam, The Netherlands; Phone: +31 10 408 1
022
; Fax: +31 10 408 9017


2

Does CEOs’ familiarity with business
segments affect their
divestment decisions?


Abstract


We examine
the impact of
CEOs’ familiarity with business segments
on
their divestiture decisions. We
find that CEOs are less likely to divest assets from familiar segments, when compared to non
-
familiar
segments.
Our evidence suggests
that
CEOs favor their familiar segments, because they are more
informed about these
segments relative to non
-
familiar segments.
It does not support
CEO selection and
managerial entrenchment
as
main
expla
nations

for
the familiarity effect
.
Even though divestitures from
familiar segments are least
likely to occur
,
these divestitures generate t
he highest abnormal announcement
returns
.




Keywords: Divestitures; Familiarity; Managerial entrenchment; CEO selection


3

1
. Introduction

Divestment decisions in corporations
represent
major restructurings of the business. In this paper, we
examine whether

divestment decisions by CEOs are affected by their career paths
. Recent research in
corporate finance has emphasized that managers’ characteristics play an important role in their
investment, financial and strategic decisions (
e.g.,
Bertrand and Schoar (
2003
)
, Malmendier and Tate
(2005, 2008),
Billet and Qian (2008),
and
Chang, Dasgupta, and Hilary (2010)
). In addition, substantial
empirical evidence shows that investors and managers tend t
o have preferences for choosing the familiar.
For instance, investors tend to invest in familiar companies, ignoring optimal portfolio principles (
e.g.,
French and Poterba (1991), Huberman (2001), Li (2004)
,
Parwada (2008)
, and Graham, Harvey, and
Huang (20
09)
). Firms show a similar geographic proximity effect in their cross
-
listing decisions
(Sarkissian and Schill (2004)) and banking choices (Berger, Dai, Ongena, and Smith (2003)).

We
inquire

whether the likelihood
to divest assets differs between segments

that are familiar to
the CEO and other, non
-
familiar segments. CEOs’ familiarity with segments comes from their

career
path, i.e.,
prior work
experience in these or related segments.
We
analyze
business segments of
multi
-
segment

firms that announce a dive
stment

for a
sample of
608

multi
-
segment firm years, comprising
768

segments with
divestments,
and
1233

segments that are fully retained.
We find that
CEO familiarity

is
associated with a
reluctance

to divest

assets
;
CEOs divest
3
5
% less
often in segments
in which they
work
ed prior to being promoted as CEO
(also referred to
as
the CEO’s
direct
-
experience segment
)
.
The
familiarity effect holds for different transaction sizes.

After documenting the
familiarity

effect on divestment decisions, we examine
three
n
on
-
mutually
exclusive
explanations. Our first explanation is that
boards
specifically
select
CEOs
with work experience
in a particular business segment
to grow and focus on
this
segment as a part of the firm’s strategy.
We test
two predictions of this CEO selection explanation.
First
, if the

CEO’s
direct
-
experience

segment was
desirable to be
fully retained
, the segment should have a lower
ex ante

probability to divest

assets
relative
to
the firm’s
other
business segments
in

the year
prior to the CEO’s appointment
.
Second
, the effects of a
mandate from the board to the new CEO to grow the
direct
-
experience

segment should be particularly
strong in the first years of the CEO’s tenure. Our empirical evidence does not lend suppor
t to either
prediction.
W
e find that
direct
-
experience

segments have
a
higher
rather than

a lower
ex ante

probability
to divest

assets
.
In addition,
our findings indicate that
, instead of
favoring familiar segments
at the
beginning of
their
tenure
, CEOs
favor familiar segments and divest

assets from non
-
familiar segments

later in their tenure
.

Our
second
explanation

for the
familiarity

effect is CEO
entrenchment (Shleifer and Vishny
(1989)). To increase the proportion of assets that are complementary to their
own
skills, CEOs
would
divest assets from non
-
familiar segments. This divestment strategy
to favor familiar segments
facilitates

4

CEO
entrenchment

by making them less dispensable.
As

e
ffective governance mechanisms
should reduce
managerial entrenchment, we
hypothesize but
fail to find a significant impact
of governance mechanisms
on the
familiarity effect and therefore
cannot support the entrenchmen
t hypothesis.

Our
third
explanation
is that
CEOs
are more informed about familiar segments than about non
-
familiar segments.
We approach this explanation from the perspective of an internal capital market where
we view divestitures as a negative capital al
location.
In allocating capital to segments
,

CEOs account for
information asymmetry between them and
segment
managers by scaling
down
the
segments’
allocated
share of the budget
(
Harris and Raviv (1996, 1998); Bernando, Cai, and Luo (2001))
.
We argue that CEOs
are more likely to divest assets from non
-
familiar segments, because
the information asymmetry
is greater
between CEOs and managers of non
-
familiar segments than between CEOs an
d managers of familiar
segments
.
Our evidence is in line wit
h the information explanation.
We distinguish between direct work
experience, industry work experience and no work experience in a segment and find that
CEOs are least
likely to divest assets from segments
of which
CEOs have most detailed
information
, whic
h is
of
their
direct
-
experience segment
.

We further investigate the role of information and explore
a particular version of CEO familiarity
with their
direct
-
experience segment

that relates to their relationship with existing personnel in the
busi
ness
segment. The CEO

may be
reluctant to sell of
f parts of

segment
s

with persons that

previously
worked under t
hem out of a sense of loyalty and

friendship
.
We postulate that
CEOs’ information and
the
strength of
their
personal relations should fade over time
as these particular persons would retire, resign,
or reassign over time, and thus, the CEOs’ partiality to protect the
personnel in the
segment from
divestments

would decline with the time since they left the segment. We do
not
find the predicted
diminishi
ng

familiarity
effect
.I
nstead, CEOs still exhibit
the
familiarity
effect

even after they left their
direct
-
experience segment

over
nine
years

ago
.
The
persistent

familiarity effect
suggests
that,
because the
CEOs accumulated years of knowledge concerning t
he
ir
direct
-
experience

segment, and having scaled the
learning curve, they would find
, after becoming CEO,

the costs to keep up with th
is

segment and
its
respective industr
y

relatively easier than with segments in which they have had no previous experience.

In our final analyses
, w
e conduct an event study around divestiture announcements
.

Consistent
with CEOs’ superior information on
direct
-
experience

segments, we show that t
hey generate higher
abnormal returns with divestitures from
these
segments.
Remarkably, t
he

g
reater returns are particularly
pronounced
among divestitures that
CEOs are least likely to

make
, i.e.
,
direct
-
experience

divestitures of
longer
-
tenured CEOs. The
d
if
ference in abnormal returns
amount
s

to
1.
6
%.

The main contribution of this paper is that we demonstrate
significant effects
of the professional
career paths of CEOs
on corporate strategy and value
.
Even though
we find that
CEOs are predicted to
divest assets from their
direct
-
experience

segment
s

at
the
time they are hired and that
direct
-

5

experience
divestitures generate the highest returns, the variation in
CEOs’
information sets among
different segments
decreases the likelih
ood
to divest from their
direct
-
experience segment
.

We consider alternative explanations for the docume
nted familiarity effect and perform several
tests to establish the robustness of our results.
For instance, a

CEO’s direct experience in a segment could

be a proxy for the segment being core and hence our documented familiarity effect may be a mere
reflection of a firm’s decision to focus on their core business.
However, we find no evidence of
a
mechanical relation. We distinguish between core segments, w
hich are the largest segments that operate
in the same industry as the primary industry of the firm, and non
-
core segments, and
show that CEOs tend
to be
particularly
reluctant to divest assets from
the non
-
core
direct
-
experience
segments
.

Our
paper
contributes

to r
ecent
studies
on
executives’ career paths that distinguish

between
generalist and specialist CEOs.

For instance,

Murphy and Zabojnik (2004) describe the general
management skills have become more important for CEO pay, when compared to firm
-
specific
experience and skills. Because bargaining power in the labor market may explain this effect, Custódio,
Ferreira and Matos (
forthcoming
) investigate the effects of the composition of general skills of CEOs.
They find that CEOs have become more gen
eralist and that a premium i
n

pay is provided, particular
during periods of restructurings and acquisitions.
A

CEO who is familiar with a particular segment can
both be a specialist and a generalist, leading to the interesting question whether our result i
s stronger for
either

type of CEO.
W
e study
this question for divestiture decisions

and find
there is
no
differen
ce in the

familiarity effect for

generalist versus specialist CEOs.

Our paperadds to
the
stream of literature about
capital allocation
decision
s

(see Stein (2003) and
Maksimovic and Philips (2007) for a

review of

the literature)
.
The study that has the most direct
connection to
ours
is Xuan (2009)
, who
shows that
newly
-
hired CEOsincrease investments in non
-
familiar segments
. He
argues that these CEOs
find it expedient to increase investments in
non
-
familiar
segments to induce cooperation

with managers of
those
segments
.
When divestitures are viewed as a
negative capital budgeting problem, our analysis is a mirror image of Xuan (2
009) and yields parallel
results in that new CEOs
also find it expedient to
favor non
-
familiar segments in the first two years of
their tenure.
However, we
extend our analysis beyond two years

of CEO tenure and find that CEOs show
their true preference for

familiar segments when they become more established.

Our paper also

contributes

to the divestment literature.

One related paper is that of
Huang (2010)
,
who

shows that, in order to create a better match between the CEO and his firm, CEOs divest complete
s
egments that operate in industries in which they do not have work experience
, which
improve
s

the firm’s
performance.
Our
paper
includes both complete and partial
divestments of segment, which allows us to
examine CEOs’ restructuring decisions within segments. We show that divestitures from the CEO’s
direct
-
experience

segments generate the highest abnormal returns, suggesting that CEOs are able to pick

6

winners and d
iscard losers (as in Stein (1997)) within their
direct
-
experience

segments, which is a
consequence of their
deeper
knowledge of the
segment
.

A complementary paper to our
s is
Landier, Nair,
and Wulf (2009), who
study the effects of geographic location on co
rporate decisions.
Interestingly, i
n
th
eir

paper proximity

of a segment

to
its

headquarter
s is
geographical distance (i.e.,
whether a segment
and headquarter are in the same state
)
,

while our study measures
social

distance, i.e. the CEOs’ personal
ties wit
h segments. The authors s
how that CEOs are less likely to divest divisions
and lay off employees
of divisions
that are more proximate to their headquarters

and

that information constraints explain why
CEOs
favor
more proximate employees
, which complements
our information explanation
.

The paper is organized as follows. Section
2

documents the familiarity effect. In Section
3

we
investigate
three

explanations for the familiarity effect. In Section
4

we describe the valuation e
ffects of
familiarity. Section 5

concludes.


2
. Does CEO familiarity affect divestiture decisions?

2.1
. Sample and

data sources

We construct our initial sample from the Compustat Business Information File and the Securities Data
Corporation (SDC) file. We select data for firms with at
least two business (or operating) segments for
our sample period, 1996
-
2004. As in Schlingemann, Stulz, and Walkling (
2002
), we select firms with
sales of over $20 million or assets above $100 million, and we exclude American Depository Receipts
(ADRs) and firms that are not incorporated in the U.S. We
also omit firm years with segments that
operate in regulated industries (SIC 4900


4999). Like Berger and Ofek (
1995
) and Schlingemann, Stulz,
and Walkling, we require that the sum of segment sales does not deviate more than 1% from total firm
sales. These selection criteria result in a sample of 5,251 firm years for 1,009 firms for our sample period.

Next
, we use the SDC database for all completed divestments for the 1996
-
2004 period made by
these

1,009

multi
-
segment firms, providing us a data set comprising 1,317 firm years (530 firms) with
divestitures.
Similar to Xuan (2009,
p.4921
)
,

we measure divestitures at the level of segments and use the
terms segment and division interchangeably. It should be noted that segments or divisions may consist of
several business units and the firm
can

divest a full division or one or more
business
un
its.
Taken from
SDC provided information,

we require that more than 95% of the unit’s assets be acquired by the buying
firm after the transaction (as in
McNeil and M
oore (2005
)).We manually link the divested
business units
with the business segments reported in Compustat by using the SDC synopsis on the divestiture, the SDC
SIC codes, and the SDC business description of the divested assets. In addition, we search t
he SEC 10
-
K
filings for descriptions of segments and discontinued operations. Since
the

unit of observation
of our main

7

analysis
is a segment in a given year, we treat multiple divestitures within one segment year as
a single
segment
observation.
1

We requi
re segments to have at least two years of data prior to the divestment. During our sample
period, which includes the introduction of SFAS 131 in 1997, several firms change their segment
reporting (see Berger and Hann (
2003, 2007
)).
2
Compustat provides revised historical financial
information for the new segments for the two years prior to the new segment reporting, based on firms’
annual rep
orts.
We exclude 589 firm years from 192 firms due to incomplete historical Compustat data
.

For our familiarity variables we require detailed information about CEOs’ working experience
inside and outside their current firm. We start with the Marquis
Who’s
Who database and Hoover’s
database. These sources provide
summar
ies

of top executives’ previous positions. In
the
case we cannot
reconstruct the CEO’s career from
these sources, we check details in the SEC 10
-
K and proxy filings
.
When
these sources a
lso lack sufficient data to reconstruct a CEO’s work experience, we exclude the
observation (in total, we exclude 49 firm years for 21 firms).We would like to add that these sources are
imperfect and we may miss
an executive’s prior
experience because of i
ncomplete reporting. However, in
our view, this would lead to a bias against finding economically and statistically significant effects for
variables based on experience. We do not further distinguish the levels of previous positions.

We
further
exclude
71
firm years
(15 firms)

in case the firm has

two different CEOs
,
the
classification of a divestiture is ambiguous
,

or the divestiture occurs in a corporate segment

where

firms
divest assets acquired from a merger in the previous year.
This selection procedur
e provides us 608 firm
years with 2,394 segment years from which
we exclude
393
segment

years

with
incomplete
corporate
financial information
or
that
are

tagged as “elimination”

(as in

Lamont (
1997
))
, resulting in 2,001
segment years
.
Of these 2,001 segment years, 768 are divest
ing segment years that in total divest 1,032
business units.


2.2
. Familiarity measure

We construct three proxy variables for familiarity

at the segment level

that differentiate a
mong the

th
ree
levels of CEOs’ relevant work experience prior to being appoi
nted as CEO. Because understanding our
measures is crucial for our analysis, we illustrate the definitions with an example.
In

the fiscal year 1998
,



1

We note that i
f multiple divestitures occur in different segments, we classify more than one segment within one
firm year as divest
ing segment.

2

We note that under SFAS 131, firms do not always report segments based on industry, but can also report their
segments based on vertical integration. We believe that this change in segment reporting does not affect our
conclusions. If anythi
ng, it would add noise to our analyses; hence work against finding significant results.


8

Bausch & Lomb
disclosed

four segments: Vision Care (two
-
digit SIC 28 and 38), Eyewear (two
-
digit SIC
38), Ph
armaceutica
ls (two
-
digit SIC 28), and Healthcare (two
-
digit SIC 2 and 28). In 1998 t
he CEO of
Bausch & Lomb Inc was William Carpenter
,

who was
previously

employed as a global business manager
in the Bausch
&

L
o
mb’s

Eyewear segment

from March 1995 until December 1995. Before joining
Bausch & Lomb in 1995,
he had

various positions at Johnson & Johnson in
the period starting from
1977until 1991 and
was

president and CEO of
Reckitt & Coleman from 1991 until 1995.

Our first measure, wh
ich is the strongest form of familiarity, is
direct work experience

within a
segment. Because Carpenter worked in the
Bausch
&

L
o
mb’s
Eyewear segment as a global business
manager, we classify this segment as the segment with which the CEO has direct work ex
perience. With
this form of familiarity, a CEO has knowledge of the segment’s industry and is informed about the
segment’s internal operations. He also had direct personal relations with its personnel and management.

The second level of familiarity is
ins
ide
-
industry work experience

in a related segment. With this
level of familiarity, although CEOs did not directly gain work experience in the reported segment itself,
they gained experience in the same industry as the reported segment. We specifically refe
r to inside
-
industry experience, because the CEO gained the industry experience within his firm by working for a
segment that operated in the same two
-
digit SIC industry as the reported segment.
3

In our example, the
Vision Care segment operates in the same

two
-
digit industry (SIC 38) as the Eyewear segment for which
Carpenter worked as global business manager, which gives him inside
-
industry experience in the related
Vision Care segment.

The third level of familiarity is
outside
-
industry work experience

i
n a segment. With this level of
familiarity, CEOs gained industry experience in the same two
-
digit SIC industry in other firm(s) as the
reported segment before joining their present firm. To classify a segment as a segment in which the CEO
has outside
-
indu
stry work experience, we first identify manually the SIC codes of all firms for which the
CEO
had
worked prior to joining the current firm and then relate those SIC codes to the SIC codes of the
reported segments. In our example, one of Carpenter’s former
employers, Johnson & Johnson, operates in
the two
-
digit SIC industries of all segments of Bausch & Lomb, making Carpenter familiar with all
segments, based on his outside
-
industry experience. Carpenter’s other former employer, Reckitt
&Coleman, does not op
erate in any of the segment’s industries.

Please note that the latter two proxies represent industry knowledge about the segment,
suggesting that CEOs are less likely to have personal connections in these segments and will have less
specific knowledge on a
ssets, procedures, and developments within the
se

seg
ment
s

compared to segments



3

Throughout the paper we use two
-
digit SIC industry codes. We conducted robustness analyses with three
-
digit SIC
codes and find qualitatively similar results.


9

in which they have direct work experience. Therefore, we maintain that direct work experience is the
strongest form of familiarity, followed by inside
-
industry experience and
outside
-
industry experience.
The weakest form of familiarity is having no direct or industry work experience with a segment

(“unfamiliar segment”)
.
We treat the classification of the CEO’s familiarity with each segment as
mutually exclusive, where the segm
ents are assigned the strongest form of familiarity.
I
n our example,
that means that we classify the Eyewear
segment as a direct work experience segment, the Vision Care
segment as inside
-
industry work experience segment and both Pharmaceuticals and Health
care segments
as outside
-
industry work experience segments. Carpenter has no segments he is unfamiliar with.


2.3
. Control variables

To examine the type of segment that is selected for divestiture, we include several control variables. For
each business segment we obtain
information on
sales, assets, net capital expenditures (calculated as gross
capital expenditures minus depreciation a
nd amortization), cash flows (which we calculate as operating
profit plus depreciation and amortization), and primary and secondary SIC codes from the Compustat
Business Information File.
4

We obtain financial firm
-
level variables; variables to calculate th
e segments’
Tobin’s
q
, segment industry
-
adjusted measures, and the firm’s primary SIC code from the Annual
Compustat File. We obtain governance information from the IRRC

and share price information from
CRSP.

We classify a segment as a core segment if the
primary two
-
digit SIC code of the segment
corresponds with the primary two
-
digit SIC code of the firm

and the segment is the largest segment in the
firm in terms of reported segment sales
. To facilitate comparability, we apply the same method as
Schlingema
nn, Stulz, and Walkling (
2002
) for the industry
measures. We calculate these measures as the
median of all Compustat firms with the same two
-
digit SIC code in the fiscal year prior to the divestiture
announcement. For more reliable industry measures, we require that at least five firms operate in the sa
me
industry. The Tobin’s
q
of a segment
is the industry ratio of the market value of assets to the book value
of assets, for which we use similar data items as Malmendier and Tate (
2005,
2008).
5

We use item 12 for
the calculation of median industry sales, item 13 for medi
an industry cash flows, and items 128 and 14 for
median industry net capital expenditures. As in Ahn and Denis (
2004
), we estimate cross
-
subsidization as
the segment’s industry
-
adjusted investment minus the firm’s sales weighted sum of indu
stry
-
adjusted



4

We follow the recommendation of Kahle and Walkling (1996) to use Compustat’s industry classification and not
that of CRSP and to use the historical industry classifications throughout our analyses.

5

We calculate the market value of assets as book value o
f total assets (item 6) plus market equity minus book
equity.The market equity=(item 25 * item 199); the book equity=(item 216
-

item 10 + item 35
-

item 336).


10

investment. We follow Schlingemann, Stulz, and Walkling
(2002)
in calculating segment liquidity: we
divide the total value of acquisition transactions by the

total assets in that industry, e
xclud
ing

values
higher than one and industries with l
ess than ten firms from the sample.

At the firm level, we calculate leverage as total debt (item 181) divided by total assets (item
6).
6
The firm’s Tobin’s
q

is the market to book value of assets, for which we use the same data items as
the Tobin’s
q
of a
segment.
We calculate net capital expenditures as gross capital expenditures (item 128)
minus depreciation and amortization (item 14). We use item 1 for the firm’s cash and short
-
term
equivalents.

We follow Berger and Ofek (
1995
) in calculating excess value, which is the percentage difference
between a firm’s total value and the sum of imputed values of its

segments as stand
-
alone firms.We
define excess value as equal to ln(V/I(V)), where V is the total firm value calculated as the market value
of equity (item 199 * item 25) plus book value of debt (item 181) and I(V) the imputed value of the sum
of a firm’s

segments as stand
-
alone firm.




n
i
mf
i
i
AI
V
Ind
AI
V
I
1
)
)
/
(
(
*
)
(
,





(
1
)

whereAI
i

is segment
i
’s sales,
n

the number of segments and Ind
i
(V/AI)
mf

the multiple of total capital to
sales for the median single
-
segment firm with at least $20 million sales in segment
i
’s industry. We
follow Berger and Ofek (
1995
) and Schlingemann, Stulz, and Walkling (
2002
)
by
basing the industry
median ratios on the narrowest SIC gro
uping with at least five firms within that industry, and by
excluding from our sample and considering as outliers any values larger than 1.386 or smaller than
-
1.386.

To calculate the firm’s diversity in
q
, we follow Rajan, Servaes, and Zingales (2000) and

Schlingemann, Stulz, and Walkling (2002):

q
q
q
Sales
Sales
Diversity
n
i
i
n
i
i
i
/
)
(
)
/
(
1
2
1







,



(2)

where Sales
i

refers to segment i’s sales,
n

to the number of segments,
q
i

to the median
q

of all Compustat
firms with the same two
-
digit SIC code as segment
i
, and
q

to the sales
-
weighted average imputed

q

across the
n

segments of the firm.




6

We note that our sample contains 1
5

observations in which the firm’s leverage exceeds one, wh
ich is theoretically
not possible. Therefore, we set these values to one. Not setting these observations to one does not influence our
results.


11

We use the
entrenchment
index constructed by
Bebchuk, Cohen, and Ferrell
(
200
9
), which is a
score
for anti
-
shareholder
charter
provisions
.
The percentage of independent directors is the number of
independent directors divided by the total number of director
s
.

We estimate the abnormal returns to divestiture announcements using the market model as
described by MacKinlay (1997)
.
Our estimation window runs from day
-
160 to
-
41 relative to the
announcement date. We aggregate the abnormal returns over the day pri
or to the divestiture
announcement until the day after the divestiture announcement.


2.4
. Summary statistics


Table 1 provides the statistics for our sample of divesting firm years.



Please insert Table 1 here


We first show the firm statistics. The sa
mple firms have average sales of
over
$
7

billion and average
assets of
over
$
9

billion.
The leverage level is relatively high (6
4
%), which may result from the
select
ionof divesting firm years.
Our average sample firm performs well, with positive cash flo
ws

prior to
the divestment year and an

average Tobin’s
q
of
1.
665
. The average
number of segments in which the
firms operate is 3.291 and the average
diversity in
q
across segments equals 0.1
04
. Our sample of
divesting firm years further shows a positive aver
age excess value of 0.0
04
, indicating that the average
firm in our sample does not underperform its single
-
segment counterparts. The
entrenchment
index ranges
from
zero
to
five
, with an average of
2.436
. On average, the firms in our sample
have 6
6
.
8
%indepe
ndent
directors.

Next,
Panel B
report
s

information on
the divestment statistics at the firm year level.

With an
average of 1.263,
the average divesting firm
in our sample divest
s

assets from more than one segment in a
single year.
Only 15.1% of our divestin
g sample firms divest
s

one or two full segments. When
aggregating the deal value of all the divested business units within a firm year, we find that the average
divesting firm receives $401 million

for its divested assets
, accounting for 11.7% of
these
firms’ total
assets.

Panel C shows
CEO

specific statistics
.
We find that 56% of the CEOs of the 608 divesting firms
have
had
at least one direct
-
experience segment within a firm year, which could reach to a
maximum of
five direct
-
experience segments.
We
note that, since external hires cannot have a direct
-
experience
segment by construction, this percentage rises to 83% for the sub sample of internally
-
hired CEOs.
In
terms of the industry work experience, 26% the CEOs have inside
-
industry experience segment
s and 28%
have outside
-
industry experience segments.
The average CEO is 5
7

years old, and has been employed by
the firm for
18
.5 years and as CEO for
7
.6 years. Of our sample of CEOs,
68
% are hired from inside the

12

firm

and 10% are founders CEOs
.

CEOs hold
on average
around
two titles (
holding the
president and/or
chairman
title
next to their CEO position
).

Table 2 provides statistics for
familiar and non
-
familiar
segments in Panel A and for
segments
from which firms divest
versus
fully retained
segments in P
anel B.



Please insert Table 2 here


Panel A shows that our total sample of
2
,
001

segments consists of
1,178

familiar segments and
823

non
-
familiar segments. CEOs have an average of
12.8

years of direct work experience in
574
segments, 8.9 years of
inside
-
industry exper
ience in 234 segments, and
15
.
4

years of outside
-
industry
experience in
370

segments.
On average, CEOs left their
direct
-
experience segment
5
.
8

years ago.
W
e
classify all segments of founder
CEOs
as familiar segments in the form or dire
ct experience, having an
increasing effect on the number of years of
their
direct work experience in a segment and a reducing
effect on the recency of their direct work experience.
Our results indicate that familiar segments are

significantly

larger
and ar
e more often core segments
compared to non
-
familiar segm
ents.
O
ne could
argue that
boards are more likely to promote managers from core segments (which are typically the
largest segments operating in the same industry as the firm’s core industry) to a CEO
position
. The
statistics

confirm that, indeed, a
relatively
large proportion of direct
-
experience segments are core
segments, but we still have 63.1% of non
-
core direct
-
experience segments
.
In
Section
2.5
, we

discuss

this
issue in
further
detail.
We
further find familiar segments to perform significantly better than non
-
familiar
segments in terms of industry
-
adjusted cash flows

and to operate in industries with higher
q
. F
amiliar
segments receive more capital expenditures relative to the firm’s budget
. Finally,
the asset liquidity of
familiar segments
do
es

not differ from that of non
-
familiar segments
.

Panel A

further shows
that t
he percentage of
divesting segments
does not

significantly

differ
for
familiar compared to
non
-
familiar segments.
A more deta
iled analysis of the sub sample of divesting
segment years indicates that, on average, CEOs divest 1.33 business units from familiar segments and
1.36 from non
-
familiar segments. In addition, the absolute value of the aggregate transaction value of
these d
ivestitures seem somewhat higher for familiar segments compared to non
-
familiar segments, but in
relative terms, a greater proportion of a segment’s assets gets divested from non
-
familiar segments. This
difference is not statistically significant though.

P
anel
B

shows that in
608

firm years, firms divest assets from
768

segments and fully retain
1,233
segments.
T
he proportion of familiar segments
does
not significantly diffe
r
for the retained segments
(
58.8
%)
compared to
the divested segments (
59
%). This result
also applies to
the different levels of
familiarity
.
In contrast to the results of studies that focus exclusively on fully divested segments (e.g.,
Schlingemann, Stulz, and Walkling
(
2002
)
; Dittmar and Shivdasani
(
2003
)
), we show that partially
divested segments are larger in terms of absolute size and

relative size, indicating that they are too

13

important to be ignored. Larger segments often consist of a collection of smaller
business units
, thus
increasing the likelihood that a separable portionof a segment gets divested.

We also find the proportion o
f core segments, which are the largest segments in a firm year
operating in the same industry as the firm’s core industry, to be higher for the divested segments sample
compared to the retained segments sample. Although this result runs counter the
motive
for divesting
assets to increase the focus of the firm’s business (
John and Ofek (1995
))
, it is consistent with a greater
likelihood to divest assets from la
rger segments
.
7
Furthermore, divested segments have lower
industry
-
adjusted
cash flows compared to retained segments
, providing evidence in favor of
two motives for
divesting assets: firms divest to reallocate assets to higher
-
valued users (
Jain (1985
)) and to obtain funds
when
external financing is too expensive and internal financing is insufficient (Lang, Poulsen, and Stulz
(
1995
)).


2.5
. Regression results

To examine whether CEO familiarity influences their decisions
from

which segments to divest assets, we
estimate binary logit regressions in which the

dependent variable takes the value of one for divested
segments and zero for
fully
retained segments

and
T
obit regressions in which the dependent variable
equals the transaction value of all divested assets within a segment year to segment assets (winsori
zed
at
the 1
-
percentile
and 99
-
percentile
)
. We follow Schlingemann, Stulz, and Walkling (
2002
) for the
specification of the economic factors.
We include
segment
performance, investment, cross
-
subsidization,
and
size, whether it is a core segment,
8

segment
q
, whether the segment is less than 10% of total s
ales
,
and segment industry’s

liquidity. We also include year dummies and dummies for the number of
segments in which the firm operates as reported by the firm.
The
robust standard errors
are
clustered by
firm.
Table 3 presents the results.



Please insert

Table 3 here


The results of Regression (1)
in Panel A
largely corroborate the results of Schlingemann, Stulz,
and Walkling (
2002
). The table shows that CEOs are more likely to divest assets from segments with



7

We note that the focus argument applies to fully divested segments as the proportion of core segments in this
sub
sample of fully divested segments drops to 6.1%.

8

In our tests we control for core versus non
-
core segments, based on

the size of the segment and

two
-
digit SIC
classification. In unreported robustness analyses we investigate several alternative proxi
es to distinguish segments
which may have a special status within the company from
other

segment
s: core segment based on
two
-
digit SIC
only; core segments based on
three
-
digit SIC

only
;
and
the largest segment

based on sales
. These alternative proxies
do n
ot affect our results for the familiarity effect.


14

lower cash flows. The negative coefficient for cash flows supports both the
efficiency and
f
inancing
rationales to divest

(
Jain (1985) and
Lan
g, Poulsen, and Stulz (1995)
, respectively)
.
O
ur coefficient of
industry median cash flows is significantly positive
,
which also suggests
that firms divest assets when
their industry peers can manage these
assets more efficiently.
The probability of a divestiture is higher for
larger segments

and for segments with higher imputed Tobin’s
q
.
Our results f
urther
indicate
the CEO’s
choice of divestment is not influenced by
a segment’s capital expenditures, cross
-
s
ubsidization,
whether a
segment
is
the firm’s core
segment
, or whether a segment has sales of less than 10% of the firm’s
consolidated sales.

In the subsequent regressions, w
e examine the familiarity effect
. We are interested in whether
CEOs’ level of familiarity with segments affects their selection
of
segment
s

to divest

assets from
. A
preference for one segment over another


based on familiarity


can only appear when a CEO
h
as
variation in his or her opt
ions, and thus variation in the levels of familiarity within a firm year.
Therefore,
we generate an indicator variable that equals one for firm years with variation in familiarity, and zero
otherwise.
9
We
add an
interact
ion term between
this
variation v
aria
ble
and
our familiarity proxies

and
should
find its coefficient to be significantly different from zero i
n the case of a familiarity effect.
As
familiarity proxy,
Regression (2)
uses
the aggregate familiarity dummy, which is a dummy for CEOs’
direct or indu
stry work experience in a segment.
We find a positive, but statistically
in
significant
coefficient of the aggregated familiarity dummy of no
-
variation firms. More importantly, the interaction
term between the variation and familiarity dummy is significantly negative, indicating that
CEOs are less
likely to divest assets from fami
liar segments

than from non
-
familiar segments
.
Applying
the Delta
method (as described inAi and Norton (2003)) on the interaction term
suggests that the average
interaction effect is also
statistically
significant
at a one
-
percent level
.
We also test whethe
r the
type
s

of
experience makes CEOs more likely to divest from non
-
familiar segments
.

Regression (
3
) includes the
three dummy variables for
direct
-
experience segment
, inside
-
industry experience, and outside
-
industry
experience instead of a single dummy fo
r familiarity

and interacts those three dummy variables with our



9

Of our sample of 608 firm years, 270 firm years have no variation in familiarity. They miss variation for the
following reasons:
61 firm years
have
founders (including CEOs who started as execu
tive office
rs after a spin
-
off);

14 firm years
have
internal hires and all segments are
direct
-
experience segment
s
;

26 firm years
have
internal hires
and all segments are outside
-
industry experience segments;

77 firm years
have
internal hire
s without any familiar
seg
ments;

41 firm years
have
external hires and all segments are outside
-
industry experience segments
;

and 51 firm
years
have
external hires without any familiar segments.


15

variation
-
in
-
familiarity dummy
.
10
Since the strongest form of familiarity is the CEO’s
direct

experience,
followed by inside
-
industry experience and outside
-
industry experience, we expect
direc
t
experience to
induce the strongest effect.
T
he results indicate that the
interaction term between the variation and direct
-
experience
dummy is the only statistically significant dummy
with the predicted

negative impact on the
segment selection
choice.
The
Delta method confirms the significance of this negative impact
with a
p
-
value of the average interaction effect
equal to 0.015
.

Because the interpretation of interaction coefficients
in binary logit regressions
is not trivial (see
Ai and Norton (2003)) and we need to segregate familiar segments from non
-
familiar segments within
firm years, the subsequent regressions exclude the 270 firms years without variation.
As in the regressions
with
the full sample, the res
ults in R
egression
s

(4)
and (5)
confirm

that CEOs are significantly less likely
to divest assets from familiar segments compared to non
-
familiar segments and
that
this effect is mainly
present for the direct
-
experience segments.
The
familiarity
effect is a
n economically significant finding.
The odds ratio of 0.6
45
for the direct
-
experience coefficient
indicates that divestments occur only 6
4
.
5
%
as often among
direct
-
experience

segments, when compared to non
-
familiar segments.
The coefficients of
the industry
-
experience variables are also negative, but not significant.
11
,
12

It is interesting to note that for the sub sample of firms with variation in familiarity,
CEOs are
more likely to divest assets from segments that receive less capital expenditures, but opera
te in industries
with higher capital expenditures, possibly to economize on cash flows. Divestitures are also more likely
to occur in segments with greater cross subsidization.

Panel B of Table 3 reports several robustness checks.
The first set of robustness checks relates to
the literature that focusses on the impact of CEOs’ background on corporate decisions, where a distinction
is made between generalist
and

specialist CEOs (e.g.,
Custódio, Ferreira, and Matos (
forthcoming
)
).
The



10

Regression (3) in Table 3 does not provide a coefficient for inside
-
industry experien
ce for firm years without
variation in familiarity, because this type of experience only appears in firm years with variation in familiarity.
Specifically, inside
-
industry experience segments are always related to a direct
-
experience segment, creating
vari
ation in familiarity within a firm year. If a CEO does not have a direct
-
experience segment, he does not have an
inside
-
industry experience segment by construction.

11

In unreported analyses, we find that our direct
-
experience effect also holds for a regre
ssion in which we only
include the direct
-
experience dummy, instead of the three mutually exclusive familiarity dummies. In this test, the
direct
-
experience coefficient equals
-
0.334 with a
p
-
value of 0.041.

12
We also investigate a related question, which
is whether the presence of familiar segments affects the proportion of
assets that firms divest. In regression models explaining total transaction value over assets at a firm
-
year level, we
find that none of our firm
-
level familiarity measures has a signif
icant effect. Results are available upon request from
the authors.


16

familiarity effect in our study can occur both in firms ran by generalist CEOs (as part of their broad
background) and specialist CEOs (when the CEO is specialist in the to
-
be
-
divested segment). However,
the effect can also be absent in both groups, when g
eneralist
s

have general experience, but not
per se
specifically related to the segment or when specialists’ skills and experience do not relate to the divesting
segment. Therefore, the overlap between our definitions and their specialist
-
generalist distinc
tion is not
predetermined and remains an


interesting


empirical question.

To test whether our documented familiarity effect is more (or less) apparent in generalist or
specialist firms, Regressions (1) and (2) interact our direct
-
experience dummy with a

specialist and,
respectively, a generalist dummy.
We use two different definitions to distinguish between specialists and
generalists. First, w
e define specialist CEOs as CEOs that have direct work experience in one
(and not
more than one)
segment

within
a firm year
.

53% of all segments are managed by specialist CEOs under
this definition
.
Second, w
e define g
eneralist CEOs
as
CEOs that
are familiar with all segments within a
firm year, but, since we need variation in familiarity within a firm year, in diff
erent levels of familiarity

(26% of all segments are run by generalist CEOs under this condition)
.
The regression results show
that
t
he interaction terms in both regressions are not significantly
different from zero, suggesting that the
familiarity effect
d
oes not show up differently
in specialist
or
generalist firms.

In the second set of robustness checks, we
account for the size of the divested business
unitsrelative to the size of the segments.
Regressions (3) until (5)
estimate
T
obit regressions with as
dependent variable the

sum of the

transaction value of all divested business units
in

a segment year

to
beginning of the year segment assets
, winsorized at a the 1
-

and 99
-
percentile. The relative transaction
size equals zero for segment years without dive
stiture.
Because
SDC does not report the transaction value
of all the divestitures
, we
retain
the divesting segment years with at least one available transaction value
in the analysis, but we
eliminate

divesting segment years in which SDC reports transacti
on values for
none of the divested business units. We account for
divesting
segment years
with some, but not all
missing
transaction value
s
, by adding to Regression (5) an indicator variable that equals one in such
instances and zero otherwise.
13
In line wit
h a smaller likelihood to divest from familiar segments, the
results in Regressions (3) to (5) suggest that CEOs also tend to divest a smaller proportion of assets from



13

It is interesting to note that the coefficient of this indicator variable is positive, suggesting that firms are less likely
to disclose the transaction value of a divestiture when they d
ivest another relatively large business unit from the
same segment in that same year.


17

familiar segments relative to non
-
familiar segments. Again, this result is mainly drive
n by a CEO’s direct
work experience in the segment.
14

In a

final set of robustness tests (unreported), we provide a more detailed
analysis of the role of
core segments in the familiarity effect. If boards mainly promote managers with work experience in core

segments to a CEO position, the negative coefficient of the direct experience dummy in our
reported
regressions might
reflect

firms’ decision to focus on their core business,
suggesting that our documented
familiarity effect
is
a mechanical relation.
We d
eal with this issue in two ways. First,
we add an
interaction term between the direct
-
experience segment dummy and core
-
segment dummy to our basic
regression as reported in Model 5 of Table3

(Panel A)
. We find that the coefficient of the direct
-
experience
segment dummy remains significantly negative. Moreover, the interaction term is significantly
positive with a p
-
value equal to 0.013(the
average
interaction effect is also significant at the five
-
percent
level when applying the Delta method)
, contrary to t
he prediction of a mechanical relation
. According to
the mechanical relation explanation, we would expect the familiarity effect to prevail in core segments.
However, these regression results suggest that the familiarity effect is mainly present in
non
-
core
segments, while it is not present in the core segments.

As
a second
test for the mechanical relation explanation, we construct two separate samples of
firm years: (1) firm years in which CEOs got promoted from the core segments (i.e., firm years i
n which
the direct
-
experience segments are also core segments); and (2) firm years in which CEOs got promoted
from non
-
core segments (i.e., firm years in which the direct
-
experience segments are non
-
core
segments).For the mechanical relation explanation to

hold, the familiarity effect should prevail in the first
sample and not the second.

Although we find a negative direct
-
experience coefficient in both sub
samples, the coefficient is only significant in the sub sample of firm years with CEOs that got promo
ted
from non
-
core segments (with a p
-
value below one percent).
Overall, these extra tests
do not

support a
significant mechanical relation between familiarity and a segment’s likelihood to divest.


3
. What explains the familiarity effect?




14

In unreported robustness tests, we estimate the same binary logit regressions as Regressions (4) and (5) of Table 3,
Panel A, but for different sub samples of firm yea
rs dependent on different transaction sizes: i.e., firm years with
divestitures of complete segments (as in Schlingemann, Stulz, and Walkling (2002); firm years in which the divested
assets in a segment reach a transaction value of at least $10 million; fi
rm years in which the divested assets in a
segment reach a transaction value of at least $50 million. The familiarity effect holds in all three sub samples.


18

So far, our analysis
demonstrates

an economically and statistically significant effect: CEOs are less likely
to divest
assets from
their
direct
-
experience

segments. In this section we investigate
three
possible
explanations for this effect.


3.1
.
T
he boar
d’s CEO selection

Considerations in the CEO selection process
may be the underlying explanation for the familiarity effect.
In the selection process for a new CEO, the board of directors takes into account a candidate’s prior work
experience. For instance,
the board may choose a
CEO to grow and focus on the familiar business
segment as a part of the firm’s long
-
term strategy, w
here the non
-
familiar segments are strategically less
important
. As a result,
CEOs

are less likely to divest assets from their familiar segments.
Huang (20
10
)
investigates divestitures
of complete segments
in relation to a
n

industry
-
based measure for the m
atch
between CEO expertise and
retained
assets and finds evidence suggesting that the appointment of CEO
s

improve firm performance in the case of a match between the CEO’s expertise and the firm’s assets.

We

test
in two ways
whether
CEO
selection
explains ou
r familiarity effect. The first test
is
based
on a CEO’s tenure.
CEOs are
least
likely to divest assets

representing their skill sets shortly after their
appointment

(Weisbach (1995))
.
Moreover,
within two or three years of tenure, CEOs have acquired
considerable knowledge from different sources and gained more political leeway to deviate from their
original mandate

(Hambrick and Fukutomi (1991))
.
Therefore, i
f the board’s CEO selection explains the
familiarity effect, we expect
this effect
to
occur
ea
rly in the CEO’s tenure.
We test
this hypothesis
by
splitting the sample into newly
-
hired CEOs with tenure of up to two years and longer
-
tenured CEOs with
tenure of three or more years.
We choose the two
-
year cutoff point to
remain consistent with the
argu
ments of Hambrick and
Fukutomi (1991)
.
Changing the threshold to three years does not change our
conclusions.
Regression (
1
) of Table
4

shows the results for newly
-
hired CEOs and
Regression (2
) for
longer
-
tenured CEOs.



Please insert Table 4 here


Contrary

to the CEO selection hypothesis, we find that n
ewly
-
hired CEOs do not show a familiarity
effect
. Rather, our results suggest that
longer
-
tenured CEOs
exhibit
a familiarity effect
.

The odds ratio of
0.51
2

suggests that
direct
-
experience

segments
of longer
-
tenured CEOs
experience 48.
8
%
fewer

divestitures than do non
-
familiar segments.

For the second test of the selection explanation,
we take two steps to
distinguish between the
board
’s selection

decision and the CEO’s divestment decision. In the first step
we estimate a model to
predict which segment is likely to be divested. We use our standard model as presented in Regression (1)
of Table 3

(Panel A)

to estimate the likelihood of a segment to be divested
.
We take the full sample of
firm years rather than the sub sample of firm years with variation in familiarity to
minimize

the likelihood

19

that any form of familiarity would affect the
determinants
of segment divestment
.
For a stronger
specification to test o
ur selection explanation,
Regression (2)
only
incorporates
segments of
firm years
with
CEOs
who succeed to that position one or two years prior to the divestiture
. Regression (3)
goes one
step further by
u
sing

the sub set of divesting firm years with exter
nally hired CEOs with tenure up to two
years.
15
The
skills
of the lat
ter

group of CEOs
should relate to the remaining assets and less correlated with

recently divested assets.

In the second step, we use the first
-
step coefficients to predict the probability
of a segment
to
divest assets using information from one year
prior

to the CEO appointment.
If CEOs
we
re hired to grow
their
direct
-
experience
segment, the predicted probability
of a
segment to divest should be the lowest
(highest) for the
direct
-
experienc
e

segments (non
-
direct
-
experience

segments).
Because the year of a
CEO's appointment can occur prior to 199
6

and our estimation period is from 199
6

until 2004, we do not
control for the year in which the divestiture takes place in the first
-
step regression
.
Our sample contains 88

CEO
-
firm combinations

and 196 segment years.
16

Table
5
shows
the results.



Please insert Table
5

here


Panel A
provides the coefficients for our
three regression models that
estimat
e

the predicted probability
of
a segment
to divest
. Panel B
shows

the statistics of the predicted probability per level of familiarity

in the
year prior to the CEO’s appointment
, based on all three regression models in Panel A
.
With an average
predicted probability of 41.5% and a median of 40.
0
%

(using Re
gression (1) from Panel A)
, our results
suggest that
, contrary to the hypothesis,

the board appoints CEOs
whose
direct
-
experience
segments that
have the highest predicted probability to be divested.

Moreover, the
direct
-
experience
predicted
probabilities are
significantly higher than that of non
-
familiar segments (
p
-
value equals 0.001)
, which is
also
contrary to the selection explanation.
Our results are robust to using different regression specifications
(as documented in Panel A)
;as
with

none of the specifica
tions
the average or median segments’ predicted



15

Our results also hold when estimating a regression based on a sub set of divesting firm years wit
h all externally
hired CEOs (so, both newly
-
hired and longer
-
tenured external CEOs) (results are unreported).

16
The sample of firm years with variation in familiarity contains 177 CEO
-
firm combinations.
We remove
89
combinations for the following reasons:

for 41 CEO
-
firms no
financial
data is available or we observe a change in
segment reporting (we lose firm years even when we miss information of one segment within that whole firm year);
for 19 CEO
-
firms, the firms report one segment when the CEO was appoi
nted; for three CEO
-
firms, we cannot
download that information from
Compustat
, because the CEO was appointed prior to 1979; for five CEO
-
firms no
Compustat

coverage exists in the year prior to appointment;
for
five CEO
-
firms
we
lose variation in familiarit
y
between segments
; and for 16 cases the CEO has been appointed in the year of the divestment (when including these
observations
theestimation
of the model
and predicted probability would be based on the same data
)
.


20

probability to divest is significantly larger for direct
-
experience segments compared to non
-
familiar
segments.We further find a significantly higher
average
predicted probability
to divest from
outside
-
indus
try experience segments
,and a similar predicted probability to divest from inside
-
industry experience
segments, compared to non
-
familiar segments.

Overall,
our evidence is not in line
with the CEO selection explanation for
the

familiarity effect.
Instead
,

o
ur results suggest that the board selects CEOs to restructure their
direct
-
experience

segments.
But

e
ven
though CEOs are predicted to divest assets from their
direct
-
experience

segments when
appointed as CEO,
we show that,
shortly

after their appointment,

they do not
act as predicted
. Moreover,
later in their tenure CEOs act in the opposite way
than predicted
by
divesting
assets
from non
-
familiar
segments
, while retaining the familiar segments
.


3.2
.
M
anagerial entrenchment

CEOs might favor familiar segme
nt
s

based on agency theory. One way for CEOs to maximize their utility
is by facilitating their entrenchment. CEOs can entrench themselves by investing in assets that are
complementary to their skills, thus becoming more valuable to the shareholders and ma
king it more costly
to replace them (Shleifer and Vishny (
1989
)). CEOs can
achieve the same end by divesting assets from
non
-
familiar segments and increasing the share of familiar assets.

Because newly
-
hired CEOs are the agents who experience the highest marginal benefits from
actions
to

entrench themselves
,
our finding that newl
y
-
hired
CEOs do not show a significant familiarity
effect is preliminary evidence against the entrenchment explanation.
We perform
two
additional
tests for
managerial entrenchment
, where we consider entrenchment
related to

external and internal governance
.
First,
as external governance measure,
we use the entrenchment index of Bebchuk, Cohen and Ferrell
(2009), which contains six provisions that isolate executives from being acquired and from shareholders’
ability to impose their will on them.
If entrenchmen
t explains the familiarity effect, we expect a stronger
familiarity effect for more entrenched CEOs. We note that, although
one could argue that already
entrenched CEOs do not need to divest from non
-
familiar segments to facilitate their entrenchment, we
e
xpect entrenched managers, having to choose which assets to divest, not to
take actions
that could undo
their entrenchment.
To test this hypothesis,
we
include an interaction term with
the
direct
-
experience
dummy
and
a dummy
for an
above sample median

entre
nchment index

(
i.e.,
equal to three or
higher)
.

Regression (1) of
Table 6

shows the results.



Please insert Table 6 here




21

CEOs
of firms
with
a
high
entrenchment index
do not show a stronger familiarity bias
, which is not in
line with the entrenchment
explanation
.

The Delta method confirms this non
-
significant interaction
effect.O
ur
direct
-
experience

dummy remains significant and negative.
17

For our second test of the entrenchment explanation we use
a measure

for
internal
corporate
governance. We expect
good corporate governance to induce CEOs to make value
-
maximizing decisions
for their firm, rather than for themselves. CEOs of firms with more independent directors have less power
over the board; hence, they have less discretion over their decisions (e.g
., Ryan and Wiggins III (2004),
Moeller (2005)
,

and Paul (2007)). If CEOs exhibit a familiarity effect due to their desire to entrench, they
would not be able to exhibit that bias in well
-
governed firms. Regression (2) adds an interaction term
consisting o
f the
direct
-
experience segment

dummy and a dummy that takes the value of one for firm
years with an above
-
median percentage of independent directors in the board
(
a dummy for
the highest
quartile of independent directors

provides similar results
)
. The inte
raction term does not show
up
significant

(neither does the average interaction effect, as estimated with the Delta method of Ai and
Norton (2003))

and our
direct
-
experience

dummy remains significant.
18

These results
are not in line with
managerial entrenchment
as
expla
nation for
the
direct
-
experience

effect for longer
-
tenured CEOs.


3.3
.
CEOs


information
environment

Another explanation for CEOs’ reluctance to divest from their
direct
-
experience segment

is that their
information sets among

the segments vary depending on their familiarity with the segments. For this
explanation, we view asset divestitures from segments as a negative capital allocation decision within an
internal capital market in which segment managers negotiate with CEOs ab
out their share of the budget.
From the point of view of segment managers, their empire
-
building tendencies (as in Jensen (1986) and
Stulz (1990)) give them incentives to bargain for a greater share of the budget and to overstate their



17

Using the CEO’s number of titles as a
proxy for
entrenchment (as in Morck, Shleifer and Vishny

(1989),

Adam
s,
Almeida, and Ferreira (2005),

and Fracassi and Tate (2012))

provides similar results.

18
We note that although an independent board is a better monitor than a more dependent board,
its

e
ffective
ness

depends on the information that the CEO provides (
Song and Thakor
(
2006
) and
Adams and Ferreira
(
2007
)
). A
CEO may be especially reluctant to share relevant information on divestment decisions that are motivated by the
CEO’s desire to entrench,

which could explain the non
-
significant result.
As additional robustness tests, we use an
alternative governance measure, which is based on
CEO ownership.CEOs with a higher stake in their firm have
more decision
-
making power (Finkelstein
(
1992
) and
Adams, Almeida, and Ferreira
(
2005
)
). On the other hand, a
CEO’s incentives are more aligned with shareholders’
incentives

when they own a higher percentage of the firm’s
shares (Jensen and Meckling
(
1976
)
). We find that
both
above
-

and

below
-
median CEO ownership
does not
influence
the home
-
base effect
(results are
available on request)
.


22

investment opportuni
ties (e.g., Milgrom and Roberts (1988), Jensen (2003) and Wulf (200
9
)). They are
more likely to succeed when they are
better

informed than the CEOs. However, CEOs account for this
information asymmetry by scaling down t
he capital allocation to
s
egments whe
re segment managers have
a greater information advantage (e.g., Harris and Raviv (1996, 1998); Bernando, Cai, and Luo (2001)
;
D
uchin and
Sosyura (2010)
)
. In our case,
there is
less asymmetric information between managers
of
familiar segments
and their CEOs
and
thus the CEOs are less inclined to divest familiar segments where
they are more confident of the valuation
.

The impact of the different levels of CEOs’ familiarity with segments on the likelihood to divest
from these segments
is
in
line with
the inform
ation asymmetry explanation

for our familiarity effect
,
because

CEOs are most informed about their
direct
-
experience

segments. CEOs’
previous employment in
the
direct
-
experience

segment provide
s

them with detailed knowledge of its industry (as with the indu
stry
-
experience segments), but also of its operations, procedures and employees. In addition, in the period they
worked in their
direct
-
experience segment
, they came to know the value and growth opportunities of its
assets. Regression (
5
) in Table 3

(Panel

A)

shows that, although all three familiarity coefficients are
negative, the only significant coefficient is that of the
direct
-
experience

segment. This finding is in line
with the conjecture that CEOs are more likely to divest from segments with greater
information
asymmetry between segment managers and CEOs.

We
perform
three

additional tests to

further

investigate the information explanation. The first test
is based on the assumption that CEOs


knowledge
of a business segment cumulates with time, i.e.
in
creases with the number of
years
in their
direct
-
experience segment
. If information asymmetry between
CEOs and segment managers explains the familiarity effect, the relation between the number of years of
work experience in their
direct
-
experience

segment should reduce asymmetric information; hence
negatively affect the probability of divestment.
W
e replace the
direct
-
experience

dummy
with th
re
e

variables for ter
c
iles of
the number of years the CEO has experience in
the
direct
-
experience

segment
,
i
.e
.,

up to three years, from four to
seven
years and
eight
years and longer
.

Table 7 presents the results twofold.
I
n Panel A
we
regress the segment
-
divestment dummy on the
two highest terciles of the direct experience dummy and only include the sub sample o
f direct
-
experience
segments. This way, we can test whether the likelihood to divest from direct
-
experience segments is
significantly stronger for segments in which CEOs have worked a longer period of time. In Panel B, we
report the logit regressions of th
e full sampleas in
Regression (5) of
Table 3
, but replace the direct
-
experience dummy for the three terciles of the length of CEOs’ direct experience.



Please insert Table 7 here


The results in the first r
egression
of
Panel A
imply
that the familiarity

effect does not depend on
the years of direct experience in a segment.
In Panel B, Regression (1) shows negative coefficients of all

23

three direct
-
experience terciles (though the first and third tercile have a p
-
value of 0.152 and 0.154,
respectively). T
he

effect of direct work experience
seems to be
strongest for CEOs with four to seven
years of experience in the segment,
but
a Wald test rejects the hypothesis that the coefficient for the
second tercile significantly differs from
that of
the first or third

tercile (with p
-
values greater than 0.664).
This result
suggests
that
CEOs’ accumulated knowledge on
direct
-
experience

segments
does not affect
the
probability to divest from these segments
.

In our
second
additional
test
of the information explanation, we use the number of years since the
CEO left the
direct
-
experience

segment. Over the years, CEOs’ superior information on their
direct
-
experience

segments is likely to fade, while they gain more knowledge of the non
-
familia
r segments
. This
change has a diminishing effect on the difference in the degree of information asymmetry between CEOs
and non
-
familiar segment managers and between CEOs and familiar segment managers
.

This test also
measures the relevance of a particular v
ersion of CEO familiarity, where a CEO may be disinclined to sell
from segments with personnel that previously worked under them out of a sense of loyalty and friendship.
As we expect that these relations should fade over time due to retirements and other
personnel turnover,
we predict that the familiarity effect should decline with the time since they left the segment.
In
Regression
s

(2
) of Table 7

(Panel A and B)

we
consider
the number of years since the longer
-
tenured
CEO left the segment,
again
based on

ter
c
iles: up to
four
years ago,
five
to
nine
years ago, and over
ten
years ago.
Our results suggest that the familiarity effect does not fade over time.
In fact
, the coefficients
tend to become more negative the longer ago the CEO left the segment
, though
a Wald test rejects a
significant
diffe
rence between the three coefficients

(
p
-
values greater than 0.309)
.
Perhaps, t
he familiarity
effect
is
persistent,
because CEOs’ store of knowledge on their
direct
-
experience

segments makes it
relatively less costly for them
to keep up with the
direct
-
experience
segments and their respective
industries
,
in comparison to what they
ha
ve

to
do
to acquire

knowledge on
other
non
-
familiar segments
.

In our third
additional
test for the

information explanation, we account for the number of years a
CEO has corporate work experience within the firm prior to
his
appointment as CEO.
With
so
-
called
corporate work experience, we refer to the number of years a CEO has work experience in the fir
m’s
headquarters.
The underlying rationale is that the longer managers have corporate work experience, the
more time they have to learn about their non
-
familiar segments. As information asymmetry between
CEOs and non
-
familiar segment managers reduces over
time, we expect the familiarity effect to
diminish

the longer CEOs have corporate experience

prior to their appointment
.
In
R
egression
s
(3) of Table 7
, we
split the
direct
-
experience

segments into three groups:
direct
-
experience

segments where CEOs do not
have corporate work experience
prior to their appointment as CEO
(so they got promotion to CEO
directly from their
direct
-
experience segment
)

or have one year of corporate work experience
;
direct
-
experience

segments where CEOs h
ave corporate experience
between two and four
years; and
direct
-

24

experience

s
egments where CEOs
have
five
or more years of corporate experience.
Although we find all
three coefficients
to be
negative, the familiarity effect is significantly stronger for CEOs with two to four
years of corporate experience compared to less than two years (p
-
value of the Wald test equals 0.040), or
more than four years of corporate experience (p
-
value equals 0.04
2). In addition, the coefficient of
third
tercile with most years of corporate experience lost significance, which
is weak evidence consistent with a
diminishing familiarity effect after
several
years of corporate experience.

Overall, our results suggest
that the variation in information

sets among different segments
induces CEOs to div
est from non
-
familiar segments.
However,
the lower level of CEOs’ effort necessary
to keep up with their
direct
-
experience

segments relative to other segments might explain
our finding
s

that the familiarity effect
is not related to
the
number of years of
work experience in the segment,
does not
fade over time
,

and is
weakly
related to the CEO’s corporate work experience.
19


4
. The value
-
relevance of familiarity

Our results so far suggest that
neither selection nor

entrenchment
drives the familiarity effect
, but the
variation in CEOs’ information sets among segments
.
We find that the familiarity effect is significant for
longer
-
tenured CEOs and pertains to their
d
irect
-
experience

segments.
This section investigates how the
stock market perceives the
se

familiarity effect
s

by means of an event study.
On the one hand, divesting
from non
-
familiar segments can generate higher abnormal returns, because managers have less
knowledge how to best manage and improve the performance of these assets and therefore
there is greater
potential gain
to higher
-
valued users
, who may be willing to pay a higher price
.
On the
other
hand, we
expect returns of divestitures from
direct
-
experi
ence segment
s to be
higher
than
the returns of divestitures
from
non
-
familiar segments, as CEOs can use their superior information to pick winners
(Stein (1997)) by



19

A possible alternative explanation for
this persistent familiarity effect is that, instead of being more informed
about their familiar segments,
CEOs may have a perception of
being informed
about
their
familiar segments, which
is based on an illusion of control.
According to Langer (1975),
fami
liarity enhances the illusion of control. This

illusion leads individual
s

to overestimate the likelihood of a successful outcome of their decisions (Langer (1975))
and to be too optimistic about the likelihoo
d of both positive and negative events (Weinstei
n (1980)). With respect
to
our
familiarity effect, CEOs’ familiarity with segments
can make them prone to the illusion of control, inducing
them to underestimate the familiar segments’ risks and overestimate their future returns
. As a result, they are less

likely
to divest from familiar segments.
Over the years, CEOs’ superior information
relative to the non
-
familiar
segments is likely to
fade, while their illusion of control
is likely to remain. The illusion of control may therefore be
an explanation for t
he home
-
base effect that CEOs exhibit even though they left their direct
-
experience segment a
long time ago or have at least four years of corporate work experience. Although this story is a plausible explanation
for our findings, we cannot design a specif
ic empirical test to provide evidence for this illusion of control.


25

divest
ing underperforming

assets from their
direct
-
experience segment

or exploit an overval
uation of
these assets
. In addition, CEOs
that divest from their
direct
-
experience segment
s
have a comparative
advantage in locating and b
argaining with potential buyers
.
Previous studies provide empirical evidence in
line with these arguments
by showing that CEOs generate higher abnormal returns when acquiring firms
that are in close proximity to the acquirer (Uysal, Kedia, and Panchapagesan (2008)) or that
operate in
industries in which
CEOs
have work experience

(Custódio and Metzger (201
2
))
.

Our sample has in total
1
,
032
divested business units
. Of those announcements, we have
909
observations with available return data

on a unique divestiture date
.
20
To avoid
the possibility
that
outliers drive our results, we
winsorize
CAR
at a one
-

and 99
-
perce
ntile
.
In Table 8, Panel A provides
the
abnormal return
statistics for
th
is

sample

of 909 observations
, but a
lso for
a
sample of
420

observations
for which we have complete information.
We end up with
420

observations, because w
e lose
424

observations due to missing transaction values and another
65

observations due to missing values for
industry
-
adjusted capital expenditures and CEO equity ownership.

Panel A
further reports statistics for
the
subsamples of
divestitures from
direct
-
experie
nce
, inside
-
industry
-
experience, outside
-
industry
-
experience
,

and non
-
familiar
segments
. Since longer
-
tenured CEOs mainly exhibit the familiarity effect,
we also split the sample into longer
-
tenured and newly
-
hired CEOs.



Please insert Table 8 here


Our
results show
a
n average

positive
market response
to divestiture announcements
, which
varies across
different types of divestitures.
In the sample with
fully
available
information, w
e find that
on average
the
direct
-
experience

divestitures generate
0
.
55
%
abnormal returns
,
while the
non
-
familiar
divestitures
generate higher abnormal returns, amounting 0.7
9%. The difference is not significantly different though
(p
-
value equals 0.
656).
Splitting our sample into divestitures announced by newly
-
hired and longer
-
tenured CEOs
provides a different view
; firms with longer
-
tenured CEOs that announce a divestiture from
their
direct
-
experience
segment
generate a
positive average abnormal return of
0
.
70
%
, while
this
percentage is 0.
50
% for
divestitures from their
non
-
familiar segments
.
Again, this difference is not
statistically significant at a less than ten
-
percent level. With 1.96%, non
-
familiar divestitures announced
by
newly
-
hired CEOs

generate the highest abnormal returns.




20
By unique divestiture date, we

refer to the
cases in which SDC reports two or three separate divestiture
announcements taking place at the same date in the same segment

(but for separate
business units). In such
instances
, we aggregate the information to one observation
. This aggregation
reduces 62 divestiture observations to
27 divestiture announcements for the event study. Of the remaining 997 observations, we lose 88 observations due
to

missing abnormal returns values.


26

Panel B of Table 8 shows
estimate
s o
f
ordinary least squares regression
s

where we regress the
three
-
day abnormal returns on the
familiarity dummies

and several control variables. We follow Bates
(2005) for the specification of the control variables, which are: the relative transaction size,
the firm’s
Tobin’s
q
, industry
-
adjusted capital expenditu
res, industry
-
adjusted leverage and

industry
-
adjusted cash,
and the percentage of stock owned by the CEO.
We
use
the 420 observations with available information,
because it is crucial to know the rel
ative transaction size
(the variable responsible for the biggest drop in
observations)
to account for the economic significance of the divestiture
s
. The average (median)
transaction value of the divestitures amounts 6.0% (1.5%) of the book value of the firm
’s total assets
,
while the 25
-

and 75
-
percentiles are, respectively, 0.3% and 6.0%
.

Regression (1) shows a positive
direct
-
experience

coefficient
, suggesting that
direct
-
experience

divestitures generate 0.
3
% higher abnormal returns than non
-
familiar divestitures.
However, the
coefficient is not significant.

The effects of inside
-

and
outside
-
industry experience are also insignificant,
with very small coefficients.
Since divestitures that increase a firm’s foc
us
are
positively
related to
firm
performance (Berger and Ofek (1995)), we add the variables excess value and a core dummy to
Regression (2).
We
do not
find
a significant
core
-
segment
or excess
-
value effect on abnormal returns.
The
direct
-
experience

effect

remained unaffected.
Regression (3)
add
s

two indicator variables for
divestitures
announced by
founder CEOs and
by internally
-
hired
CEOs, because we assume that founder CEOs have
direct work experience in all the segments of their firm

and thus all divest
itures by founder CEOs are
direct
-
experience divestitures, and
CEOs can have
divestitures from
direct
-
experience and inside
-
industry
experience segments

only when they are hired internally
, possibly affecting the insignificant direct
-
experience coefficient
.

We find that the abnormal returns of divestiture announcements by founder CEOs
are 2.8% lower and that of internally
-
hired CEOs 1.5% lower.
More importantly, after adding these two
dummy variables, the direct
-
experience coefficient increases to 1.0% and,

with a p
-
value equal to 0.114,
almost significant at the ten
-
percent level
.

Because we find that longer
-
tenured CEOs exhibit a familiarity effect, we split

the
direct
-
experience
dummy into
a
direct
-
experience

dummy
of
longer
-
tenured CEOs and
one
of
newly
-
hired CEOs

in Regression (3)
. We find that
shareholders reward
direct
-
experience
sales
made by longer
-
tenured CEOs
with an economically and statistically significant 1.
6
%.
Even though longer
-
tenured CEOs tend to be
reluctant to sell assets from their
d
irect
-
experience

segments, their superior information helps them to
generate higher abnormal returns if they act against their disinclination.
The
direct
-
experience

dummy is
not significant
for the
newly
-
hired CEOs.
21




21

We did an additional robustness analysis on the positive direct
-
experience effect on abnormal returns (results are
not
reported
). We add to Regression (3) an interaction term between the direct
-
experience and the
core
-
segment

27

In separate (unreported) regression ana
lyses, we examine the distribution of the transaction
surplus between divesting firms and buyers.
If superior information
gives
familiar CEOs
more
bargain
ing
power
, we should observe the divesting firms to get a bigger proportion of the takeover surplus. We
follow Kale, Kini, and Ryan (2003) and construct a measure for the fraction of the wealth gain received
by the divesting firm.
We note that, for the calculation
of the transaction surplus, we can only focus on
deals with US publicly listed buyers with available return information, restricting our sample to 303
divestitures.
However, this sub

sample
shrinks
to 181 observations after deleting deals for which we miss
values
of
the
independent variables
(we lose 102 observations due to missing values on the

relative
transaction size)
.
We first measure the dollar wealth gains for the divesting firm (WD) and the buyer
(WB), by multiplying their respective three
-
day abnormal

returns and their market capitalizations four
weeks prior to the divestiture announcement. Next, we define the total wealth gain to be the sum of the
gains for buyer and divesting firm (WT=WD+WB). In the case the total wealth gain is positive, we define
t
he proportion of gain received by the divesting firm to be WD/WT, and otherwise 1
-
WD/WT. The
proportion of the gains is winsorized at the 1
-
percentile
and 99
-
percentile

and has an average (median)
value of 0.55 (0.50) for the 181 observations with availabl
e information.
We then perform the same
regressions as in Table 8, but replace CAR for this measure of the divesting firm’s proportion of the
synergy gains.
We find that, consistent with
the information advantage of familiar CEOs
,
longer
-
tenured
CEOs
gain
by
bargain
ing for

a greater share of the takeover surplus
when divesting assets from direct
-
experience segments
than from non
-
familiar segments
(coefficient equals 0.622, p
-
value equals 0.056)
.


5
.
Conclusion

Our paper examines the impact of CEOs

career
paths

on corporate decisions.
We
focus on
how
CEOs


familiarity with segments
influences
their
divestment decisions
.
CEOs are familiar with a segment due to





dummy and find no significant interaction term. However, when running the same regression for the sub sample of
divestitures announced by longer
-
tenured CEOs, we find that the positive familiarity effect can be attributed to the
divestitures f
rom non
-
core direct
-
experience segments (coefficient equals 0.019, p
-
value equals 0.076), which are
exactly the divestitures that CEOs are less likely to make. Second
,
we replace the internally
-
hired
-
CEO and founder
-
CEO dummy in Regressions (3) and (4) for

an
interaction term
between the direct
-
experience dummy and a dummy
for firm years with variation in familiarity. The interaction term in Regression (3) has a coefficient equal to 0.025,
which is significant at the five
-
percent level, suggesting that CEOs

with variation in familiarity generate 2.5%
higher abnormal returns when announcing a direct
-
experience divestiture then when they announce a non
-
familiar
divestiture. The two interaction terms in Regression (4) indicate that this significant interaction
effect can be
attributed to the divestiture announcements of longer
-
tenured CEOs.


28

their work experience in the segment or
in
its industry.
We empirically show that
CEOs are
less
like
ly to
divest assets from
familiar
segments relative to
non
-
familiar segments.
The familiarity effect mainly
manifests itself among segments in which CEOs have direct work experience.

After documenting the familiarity effect, we explore three potential
non
-
mutually exclusive
explanations: the board’s CEO selection, CEO entrenchment, and
the CEO’s
information
environment
.
We find
no evidence consistent with the
CEO selection

and

CEO entrenchment
explanations
. Rather, our
evidence
is in line with the
conjecture that CEOs are more likely to divest assets from non
-
familiar
segments

than familiar segments
, because
the information asymmetry
between
CEOs
and non
-
familiar
segment

managers

is greater
.

Our

event study suggests that, on average,

firms are capab
le of
creating value for shareholders by
divesting assets.
Moreover, CEOs’ superior information about their
direct
-
experience

segments provides
them a comparative advantage in finding higher
-
valued users for assets that belong to their
direct
-
experience se
gment

and
in
negotiating about the deal,
because
direct
-
experience

divestitures generate the
highest abnormal returns.
However,
the

familiarity
effect
can
be
costly to shareholders,
as these
divestitures
are the least likely to occur.



29

References

Adams, R
.

B., andD
.
Ferreira
. 2007. A theory of friendly boards.

Journal of Finance

62
:
217
-
250.

Adams, R
.

B.
,
H
.
Almeida,

and D.

Ferreira
.
2005
.

Powerful CEOs and their im
pact on corporate
performance
.
Review of Financial Studies

18:
1403
-
1432.

Ahn, S
.
,

and D
.

J.

Denis
.

2004
.

Internal capital markets and investment policy: Evidence from corporate
spinoffs
.
Journal of Financial Economics

71:
489
-
516.

A
i, C., and E.C. Norton. 2003. Interaction terms in logit and probit models.
Economics Letters
80: 123
-
129.

Bates, T
.

W.

2005.
Asset sales, investment opportu
nities, and the use of proceeds.
Journal of Finance

60
:
105
-
135
.

Bebchuk, L.A., A. Cohen, and A. Ferre
ll.
200
9. What matter in corporate governance?
Review of
Financial Studies
22:
783
-
827.

Berger, A.N., Q. Dai, S. Ongena, and D.C. Smith. 2003, To what extent will the banking industry be
globalized? Evidence of bank nationality and reach in 20 European nations.
Journal of Banking &
Finance

27: 383
-
415.

Berger, P
.
G.
,

and R
.

N.
Hann
.

20
03
.

The impact of SFAS no.13
1 on information and monitoring.
Journal
of Accounting Research

41
:
163
-
223.

Berger, P
.
G.
,

and R
.

N.
Hann.

200
7
.

Segment profitability and the proprietary and agency costs of
disclosure
,
The Accounting Review

82: 869
-
906
.

Berger, P
.

G
.,

and E.
Ofek
.

1995
.

Diversi
fication’s effect on firm value
.
Journal of Financial Economics

37
:
39
-
65.

Bernardo, A. E., H. Cai, and J. Luo. 2001. Capital budgeting and compensation with asymmetric
information and moral hazard.
Journal of Financial
Economics

61: 311
-
344.

Bertr
and, M
.
,

and A
.
Schoar
.

2003
.

Managing with style: The effe
ct of managers on firm
policies.
Quarterly Journal of Economics

143:
1169
-
1208.

Billet, M.T., and Y. Qian. 2008. Are overconfident CEOs born or made? Evidence of self
-
attr
ibution bias
from frequent acquirers.
Management Science

54: 1037
-
1051.

Chang, Y. Y., S. Dasgupta, and G. Hillary. 2010. CEO ability, pay, and firm performance.
Management
Science

56: 1633
-
1652.

Custódio, C., M.A. Ferreira, and P. Matos.Forthcoming. Genera
lists versus specialists: Lifetime work
experience and chief executive officer pay.
Journal of Financial Economics.

Custó
dio, C., and D. Metzger. 201
2
.
How do CEOs matter? I
ndustry expertise

on

acquisition

returns
.
Working paper,
Stockholm
School of Econom
ics.


30

Dittmar, A
.
,
and A
.Shivdasani.

2003
.

Divestitures and

divisional investment policies
.
Journal of Finance

58
:
2711
-
2743.

Duchin, R., and D. Sos
y
ura. 2010. Divisional managers and internal capital markets. Working paper,
University of Michigan.

Finkelstein,
S.

1992
.

Power in top management teams: Dimensio
ns, measurement, and validation.
The
Academy of Management Journal

35
:
505
-
538.

FracassiC., and G. Tate. 2012. External networking and internal firm governance.
Journal of
Finance

67: 153
-
194.

Fren
ch, K.R., and J.M. Poterba. 1991. Investor diversification and international equity markets.
American
Economic Review

81: 222
-
226.

Graham, J.R., C.R. Harvey, and H. Huang. 2009. Investor competence, trading frequency, and home bias.
Management Science

55:1
094
-
1106.

Hambrick, D
.
C.
,

and
G
.
D.S.
Fukutomi
.

1991
.

The seasons of a CEO’s tenure
.
The Academy of
Management Review

16
:
719
-
742.

Harris, M., and A. Raviv. 1996. The capital budgeting process: Incentives and information.
Journal of
Finance

51: 1139
-
1174.

Harris, M., and A. Raviv. 1998. Capital budgeting and delegation.
Journal of Financial Economics

50:
259
-
289.

Huang, S. 2010
. CEO characteristics,
corporate decisions and firm value: Evidence from corporate
refocusing
. Working paper,
Singapore Management
University
.

Huberman, G. 2001. Familiarity breeds investments.
Review of Financial Studies

14:659
-
680.

Jain, P
.

C.

1985
.
The effect of voluntary sell
-
off anno
uncements on shareholder wealth
.
Journal of Finance

40
:
209
-
224.

Jensen, M. C. 1986. Agency costs of free cash flow, corporate finance, and takeovers.
The American
Economic Review
76: 323
-
329.

Jensen, M. C. 2003. Paying people to lie: The truth about the budgeting process.
European Financial
Management
9: 379
-
406.

Jensen
, M
.
C.
,

andW
.
H.
Meckling
.

1976
.

Theory of the firm: Managerial behavior, agen
cy cost and
ownership structure
.
Journal of Financial Economics

3
:
305
-
360.

John, K
.
,
and E.
Ofek
.

1995
.

Asset sales and increase in focus
.
Journal of Financial Economics

37:
105
-
126
.

Kahle, K.M., and Walkling, R.A. 1996. The impact of industry classifications on financial
research.
Journal of Financial and Quantitative Analysis

31: 309
-
335.


31

Kale, J.R., O. Kini, and H.E. Ryan Jr. 2003.Financial advisors and shareholder wealth gains in
corporate
takeovers.
Journal of Financial and Quantitative Analysis

38: 474
-
501.

Lamont, O
.

1997
.

Cash flow and investment: Evidenc
e from internal capital markets
.
Journal of Finance

52
:
83
-
109.

Landier, A
.
,

V
.
B.

Nair,
and J
.
Wulf
.

200
9
.

Tradeoffs in staying close: Corporate decision
-
m
aking and
geographic dispersion.
Review of Financial Studies

22: 1119
-
1148.

Lang, L
.
,
A
.
Poulsen,
and R
.
M.
Stulz
.

1995
.

Asset sales, firm performance, and the agency

costs of
managerial
discretion.
Journal of F
inancial Economics

37
:
3
-
37.

Langer, E
.

J.

1975
.
The illusion of control
.
Journal of Personality and Social Psychology

32
:
311
-
328.

Li, K. 2004. Confidence in the familiar: An international perspective.
Journal of Financial and
Quantitative Analysis

39: 47
-
68.

MacKinlay, A. G
.

1997
.

Event studies in economics and finance
.
Journal of Economic Literature

35
:
13
-
39.

Maksimovic, V., and G. Philips. 2007. Conglomerate firms and internal capital markets
, in B.
EspenEckbo (ed.),
Handbook of corporate finance:Empirical
corporate finance
, Vol. 1. Elsevier: North
Holland.

Malmendier, U
.
,

and G
.
Tate
.

2005
.

CEO overconf
idence and corporate investment
.
Journal of Finance

60
:
2661
-
2700.

Malmendier, U
.
,

and G
.
Tate
.

200
8
.

Who makes acquisitions? CEO overconfi
dence and the marke
t’s
reaction
.
Journal of Financial Economics

89: 20
-
43.

Milgrom, P., and J. Roberts.
1988. An economic approach to influence activities in organizations.
American Journal of Sociology

94, 154
-
179.

McNeil, C
.

R.
,

and W
.

T.

Moore
.

2005
.

Dismantling internal capital markets via spinoff: Effects on
capital allocation efficiency and firm valuation
.
Journal of Corporate Finance

11:
253
-
275.

Moeller, T
.

2005
.

Let’s make a deal! How shareholder control impacts merger payoffs
.
Journal of
Financial

Economics

76
:
167
-
190.

Morck,
R., A.
Shleifer
,

and
R.W.
Vishny
. 1989. Do managerial objectives drive bad
acquisitions?
Journal of Finance

45: 31
-
48.

Murphy, K.J., and J. Zabojnik. 2004. CEO pay and appointments; a market
-
based explanation for
recent trends
.
American Economic Review Papers and Proceedings

94: 192
-
196.

Parwada, J.T. 2008.The genesis of the home bias? The location and portfolio choices of investment
company start
-
ups.
Journal of Financial and Quantitative Analysis

43: 245
-
266.

Paul, D. 2007. Board composition and corrective action: Evidence from corporate responses to bad
acquisition bids.
Journal of Financial and Quantitative Analysis

42: 759
-
784.


32

Rajan, R
.
,
H
.
Servaes,

and L
.
Zingales
.

2000
.

The cost of diversity: The diversific
ation discount and
inefficient investmen
t.
The Journal of Finance

55:
35
-
80.

Ryan

Jr.
, H
.

E
.
,

and R
.

A.

Wiggins III.

2004
.

Who is in the pocket? Director compensation, board
independence, and b
arriers to effective monitoring.
Journal of Financial Economics

73:
497
-
524.

Sarkissian, S., and M.J. Schill. 2004. The overseas listing decision: New evidence of proximity
preference.
Review of Financial Studies

17: 769
-
809.

Schlingemann, F
.

P.
,

R
.

M.
Stulz,
and R
.

A.
Walkling
.

2002
.

Divestitures and the liquidity of
the market
for corporate assets
.
Journal of Financial Economics

64
:
117
-
144.

Shleifer, A
.
,

and R
.

W.
Vishny.

1989
.

Management entrenchment: The case
of manager
-
specific
investments
.
Journal of Financial Economics

25:
123
-
139.

Song, F
.
,

and A
.

V.
Thakor
.

2006
.

In
formation control, career con
cerns, and corporate governance
.
Journal
of Finance

61
:
1845
-
1896.

Stein, J
.

C.

1997
.
Internal capital markets and the comp
etition for corporate resources
.
Journal of Fi
nance

52
:
111
-
133.

Stein, J
.

C.

2003
.
Agency, infor
mation and
corporate investment,i
n Constantinides, G.,
M.
Harris, and
R
.

M.
Stulz
(
ed.
),
Handbook of the Economics of Finance
.Elsevier: North Holland
.

Stulz, R. M. 1990. Managerial discretion and optimal financing policies.
Journal of Financial Economics

26: 3
-
27.

Uysal
, V. B., S. Kedia, and V. Panchapagesan. 2008. Geography and acquirer returns.
Journal of
Financial Intermediation

17: 256
-
275.

Weinstein, N
.

D.

1980
.
Unrealistic optimism about future life events
.
Journal of
P
ersonality and Social
Psychology

39
:
806
-
820.

Weisbach, M. S. 1995. CEO turnover and the firm’s investment decisions.
Journal of Financial
Economics

37: 159
-
188.

Wulf, J. 2009. Influence and inefficiency in the internal capital market.
Journal of Economic Behavior &
Organization

72: 305
-
321.

Xuan, Y
.

2
00
9
.
Empire
-
building or bridge
-
building? Evidence from new CEOs’ internal capital allocation
decisions
,
Review of Financial Studies
22: 4919
-
4948.



33

Table 1: Firm and CEO summary statistics

The table shows the means, standard deviations, minimum,

and maximum values
,

and the number of
observations

of firm
, divestiture,

and
CEO variables.
Leverage is debt divided by total assets. Net capital
expenditures are the gross capital expenditures minus depreciation and amortization.
We define cash
flows as o
perating profit plus depreciation and amortization.
Tobin’s
q

is the ratio of the market
-
to
-
book
value of assets, as calculated in Malmendier and Tate (2005, 200
8
). We calculate

the excess value
measure as in Berger and Ofek (1995); the diversity in
q

as i
n Rajan, Servaes, and Zingales (2000)
; and
the
entrenchment
index as in
Bebchuk, Cohen, and Ferrell (200
9
).
Independent directors is the percentage
of independent directors relative to all directors.
Fully divested segments are segments that
are completely
divested and
firms
cease to report the
se

segment post divestiture
. The aggregate transaction value is the
sum of the transaction values of all divested business units within a firm year. Direct
-
experience segments
are segments in which the CEO worked prior

to his appointment as CEO. Inside
-
industry experience
segments are segments that operate in the same two
-
digit SIC industry as the direct
-
experience segments.
Outside
-
industry experience segments are segments that operate in the same two
-
digit SIC industr
y as the
firms for which the CEO used to work prior to joining his current firm.



Mean

SD.

Minimum

Maximum

N

Panel A. Firm summary statistics






Sales
t
-
1

($M)

7,249

13,869

41

153,627

608

Assets
t
-
1

($M)

9,401

23,952

109

279,097

608

Leverage

0.638

0.155

0.190

1.000

608

Capx
t
-
1
/sales
t
-
2

0.014

0.111

-
0.593

2.008

608

Cash flow
t
-
1
/sales
t
-
2

0.174

0.120

-
0.087

1.716

608

Tobin's
q

1.665

0.864

0.241

7.302

608

Number of business segments

3.291

1.105

2.000

7.000

608

Diversity in
q

0.104

0.109

0.000

0.492

595

Excess value

0.004

0.539

-
1.386

1.386

555

Entrenchment
index

2.436

1.364

0

5

608

Independent directors

0.6
68

0.1
77

0.
083

0.923

528







Panel B. Divestment statistics






Number of segments with a divestment

1.263

0.523

1

4

608

Dummy for

firm years with a full
y

divest
ed segment

0.151

0.359

0

1

608

Number of segments with a full
y

divest
ed segment

0.163

0.400

0

2

608

Aggregate transaction value of all divested business units

($M)

401

96
7

0

10,83
7

401

Aggr.

transaction value of all divested business units/assets
t
-
1

0.117

0.266

0

3.508

401







Panel C. CEO summary statistics






Dummy for
presence of
a direct
-
experience segments

0.559

0.497

0

1

608

Number of direct
-
experience segments

0.939

1.094

0

5

608

Dummy for
presence of
inside
-
industry experience segments

0.262

0.440

0

1

608

Number of inside
-
industry experience segments

0.385

0.753

0

5

608

Dummy for
presence of
outside
-
industry experience segments

0.280

0.449

0

1

608

Number of outside
-
industry experience segments

0.609

1.159

0

6

608

Years employed as CEO

7.610

9.090

0

54

608

Years worked for firm

18.495

13.010

0

54

594

Number of titles (CEO, president, chairman)

2.219

0.520

1

3

526

Internally hired
CEO

0.676

0.468

0

1

608


34

Founders

0.100

0.301

0

1

60
8

CEO age

57.072

7.376

36

90

608















35

Table 2: Characteristics of sub samples

The table presents means, standard deviations,
number of observations,
and mean differences for segments of the fiscal year prior to the
divestiture
announcement. Panel A
compares
familiar and non
-
familiar segments and Panel B compares
segments

from which CEOs decide to divest and fully
retained segments
.
Proxies for familiarity are the CEOs’ direct experience, inside
-
industry experience, and

outside
-
industry experience

segments
.
The size<10% dummy indicates that segment sales are less than 10% of firm sales.
The

core segment

dummy equals one for
the largest
segments
within a firm year based on sales
with the same primary two
-
digit SIC code as

the primary two
-
digit SIC code of the firm. Cash flows are the
segment’s operating profit plus depreciation and amortization. We define net capital expenditures as the gross capital expend
itures minus
depreciation and amortization.
We define the cross
-
sub
sidization variable as in Ahn and Denis (2004).
The segment’s Tobin’s
q

represents the
median industry
q

of all Compustat firms with the same two
-
digit SIC code as the segment (
q

is the ratio of the market
-
to
-
book value of assets, as
calculated in Malmendi
er and Tate (2005, 2008)). Segment’s industry liquidity is the liquidity index at the two
-
digit SIC code level, as calculated
by Schlingemann, Stulz, and Walkling (2002). We define the industry
-
adjusted variables as the segment variable minus the median of

all
Compustat firms with the same two
-
digit SIC code.
Fully divested segments are segments that are completely divested and firms cease to report
these segment post divestiture. The aggregate transaction value is the sum of the transaction values of all d
ivested business units within a firm year.
The aggregate transaction value to segment sales and the aggregate transaction value to segment assets are winsorized at the
one
-

and 99
-
percentile.
The subscripts refer to the year relative to the year in which f
irms announce their divestment.
W
e truncate ratios at
-
1 and +1.
***, **,
and * denote that the means of familiar (divested) segments are significantly different from the means of non
-
familiar (retained) segments at the 1
percent, 5 percent, and 10 percent

level, respectively.

Panel
A
: Familiar versus unfamiliar segments



Familiar






























Not familiar



Direct experience



Inside industry exp.



Outside industry exp.



All


All


Mean

SD.

N


Mean

SD
.

N


Mean

SD.

N


Mean

SD.

N


Mean



SD.

N

Aggregate familiarity dummy

1.000

0.000

574


1.000

0.000

234


1.000

0.000

370


1.000

0.000

1178


-


-

823

Years direct experience

12.843

13.440

574


-

-

234


-

-

370


6.258

11.366

1178


-


-

823

Years since left direct exp. segment

5.828

7.441

558


.

-

-


.

-

-


-

-

-


.


-

-

Years inside
-
industry experience

-

-

574


8.910

8.891

234


-

-

370


1.770

5.319

1178


-


-

823

Years outside
-
industry experience

-

-

574


-

-

234


15.351

9.281

370


4.822

8.822

1178


-


-

823










































Sales
t
-
1
($
-
mln)

3,335

9,764

572


2,244

4,225

234


1,699

2,263

370


2,603

7,212

1176


1,693

***

3,647

822

Sales
t
-
1
/ firm sales

t
-
1

0.371

0.247

574


0.218

0.153

234


0.315

0.216

370


0.323

0.229

1178


0.277

***

0.218

822

Size<10% dummy

0.136

0.343

574


0.261

0.440

234


0.157

0.364

370


0.167

0.373

1178


0.204

**

0.403

822

C
ore segment

0.369

0.483

574


0.103

0.304

234


0.268

0.443

370


0.284

0.451

1178


0.202

***

0.402

823

Cash flow
t
-
1
/sales
t
-
2

0.205

0.177

565


0.171

0.181

231


0.160

0.208

366


0.184

0.189

1162


0.187


0.208

815


36

Industry
-
adj. cash flow
t
-
1
/sales
t
-
2

0.090

0.196

565


0.088

0.197

231


0.081

0.224

363


0.087

0.205

1159


0.066

**

0.206

809

Capx
t
-
1
/sales
t
-
2

0.013

0.166

565


0.003

0.100

231


0.011

0.150

366


0.010

0.150

1162


0.009


0.130

815

Industry
-
adj. capx
t
-
1
/sales
t
-
2

0.008

0.164

565


0.004

0.098

231


0.010

0.151

363


0.008

0.149

1159


0.004


0.131

809

Cross
-
subsidization

-
0.001

0.144

517


-
0.005

0.096

216


-
0.002

0.147

327


-
0.002

0.137

1060


-
0.014

*

0.150

747

Segment's Tobin's
q

1.635

0.571

574


1.704

0.578

234


1.688

0.572

368


1.665

0.573

1176


1.547

***

0.485

817

Segment's industry liquidity

0.119

0.112

572


0.122

0.114

234


0.106

0.103

364


0.116

0.110

1170


0.124


0.113

812











































Divestment dummy

0.383

0.487

574


0.346

0.477

234


0.411

0.493

370


0.385

0.487

1178


0.383


0.486

823

Fully divested segment dummy

0.047

0.212

572


0.039

0.193

233


0.057

0.233

367


0.049

0.215

1172


0.051


0.220

822






















Sub sample of segments with divestments only
















Number of
divested business units

1.468

1.040

220


1.185

0.422

81


1.214

0.647

152


1.332

0.844

453


1.360


0.922

315

Aggr
egate

transaction value of
divest
ed business units ($
-
mln)

386.2

934.
9

126


481.8

965.
6

46


318.
1

632.2

88


380.
1

849.
3

260


322.0


940.
4

193

Aggr.
transaction

value/
sales
t
-
1

0.396

0.698

125


0.510

0.854

46


0.466

0.757

88


0.440

0.746

259


0.558


0.891

193

Aggr.
transaction

value/assets
t
-
1

0.417

0.724

126


0.479

0.723

46


0.492

0.740

88


0.454

0.727

260


0.540


0.786

192











































Panel B: Divested versus retained segments



Divested (1)



Retained (2)


Mean

SD
.

N



Mean



SD
.

N

Aggregate familiarity dummy

0.590

0.492

768


0.588


0.492

1233

Direct experience

0.286

0.452

768


0.287


0.453

1233

Inside
-
industry experience

0.105

0.307

768


0.124


0.330

1233

Outside
-
industry experience

0.198

0.399

768


0.177


0.382

1233










Sales
t
-
1

2,671

7,242

767


1,953

**

5,102

1231

Sales
t
-
1
/ firm sales

t
-
1

0.349

0.241

768


0.276

***

0.211

1232

Size<10% dummy

0.143

0.351

768


0.207

***

0.405

1232

C
ore segment

0.298

0.458

768


0.221

***

0.415

1233

Cash flow
t
-
1
/sales
t
-
2

0.178

0.196

764


0.190


0.198

1213

Industry
-
adj. cash flow
t
-
1
/sales
t
-
2

0.065

0.199

760


0.086

**

0.210

1208

Capx
t
-
1
/sales
t
-
2

0.013

0.137

764


0.008


0.145

1213

Industry
-
adj. capx
t
-
1
/sales
t
-
2

0.008

0.137

760


0.005


0.145

1208


37

Cross
-
subsidization

-
0.004

0.127

714


-
0.009


0.152

1093

Segment's Tobin's
q

1.643

0.539

765


1.600

*

0.543

1228

Segment's industry
liquidity

0.120

0.106

761


0.118


0.114

1221




















38

Table 3:
Segment divestment binary logit r
egressions

This table presents the results of binary logit regressions that explain from which type of segments firms
choose

to divest assets

(Panel A, Regressions (1) to (5)

and Panel B, Regressions (1) and (2)
)
. The
dependent variable takes the value of one for divested segments and zero for retained segments.
Regressions (3) to (5)
of Panel B
report the results of Tobit regr
essions that explain the relative size of the
divestitures, measured as the total sum of transaction values of all divested business units to the beginning
of the year segment’s assets. Segments without divestitures get the value of zero. We exclude
divest
ing
segment years of which no
transaction value

is available
, and we winsorize this variable at the one and 99
-
percentile.
Regressions (1) to (3) of Panel A incorporates the full sample of 608 firm years. All the other
regressions
use

the sub sample of firm

years with variation in familiarity. Variation in familiarity
stands
for
firm years in which the CEO’s familiarity varies

across the firm’s segments
.
We measure
familiarity
by means of
the CEOs’
work

experience, which we split into
direct
-
experience
experi
ence, inside
-
industry
work experien
ce, and outside
-
industry work

experience.
The specialist dummy equals one for CEOs that
have one direct
-
experience segment in their firm, zero otherwise. The generalist dummy equals one for
CEOs that are familiar with all

the segments, but in different levels of familiarity
, zero otherwise
.

The
dummy missing transaction value equals one for segment years with a divestiture for which the transaction
value is missing.
All other variables are self
-
explanatory or defined more c
ompletely in Table 2. The
subscripts refer to the year relative to the year in which firms announce their divestment.
W
e truncate
ratios at
-
1 and +1. All regressions include year dummies and dummies for the number of segments

(unreported)
.
P
-
values appear

in parentheses and are based on

robust standard errors clustered by firm
.

Panel A:
Segment divestment



Full sample



Variation in familiarity
sample



(1)

(2)

(3)



(4)



(5)



Aggregated familiarity dummy



0.139





-
0.313

*






(0.125)





(0.063)




Variation in familiarity



0.344

***

0.345

***









(0.004)


(0.004)







Variation in familiarity * aggr. familiarity dummy



-
0.461

***











(0.009)









Direct experience





0.089





-
0.439

**






(0.340)





(0.027)


Variation in familiarity * direct experience





-
0.497

**











(0.015)







Inside
-
industry experience










-
0.238












(0.279)


Variation in familiarity * inside
-
industry exp.





-
0.282












(0.191)







Outside
-
industry experience





0.192





-
0.177







(0.143)





(0.457)


Variation in familiarity * outside
-
industry exp.





-
0.387












(0.136)







Cash flow
t
-
1
/sales
t
-
2

-
0.606

**

-
0.632

**

-
0.624

**


-
0.966

**

-
0.958

**


(0.037)


(0.028)


(0.030)



(0.012)


(0.013)


Industry median cash flow
t
-
1
/sales
t
-
2

1.483

***

1.407

***

1.451

***


1.438

**

1.505

**


(0.008)


(0.010)


(0.007)



(0.042)


(0.032)


Capx
t
-
1
/sales
t
-
2

-
0.276


-
0.302


-
0.308



-
1.290

***

-
1.301

***


39


(0.498)


(0.471)


(0.468)



(0.003)


(0.003)


Industry median capx
t
-
1
/sales
t
-
2

1.556


1.577


1.616



6.463

**

6.591

**


(0.458)


(0.452)


(0.450)



(0.021)


(0.020)


Cross
-
subsidization

0.409


0.469


0.474



0.925

**

0.939

**


(0.330)


(0.256)


(0.257)



(0.034)


(0.031)


Sales
t
-
1
/ firm sales

t
-
2

1.060

**

1.100

**

1.125

**


1.707

***

1.775

***


(0.029)


(0.025)


(0.021)



(0.007)


(0.004)


Core
segment

-
0.058


-
0.047


-
0.037



-
0.062


-
0.033



(0.774)


(0.818)


(0.858)



(0.807)


(0.897)


Segment's Tobin's
q

0.304

***

0.307

***

0.309

***


0.262

*

0.268

*


(0.004)


(0.004)


(0.004)



(0.082)


(0.073)


Size<10% dummy

-
0.098


-
0.104


-
0.107



0.122


0.111



(0.542)


(0.517)


(0.510)



(0.534)


(0.572)


Liquidity

0.189


0.119


0.145



1.202


1.241



(0.747)


(0.841)


(0.806)



(0.128)


(0.117)














Number of observations

1,805


1,805


1,805



1,049


1,049


McFadden
R
-
squared

4.18%



4.45%



4.50%





5.26%



5.39%



* significant at 10%; ** significant at 5%, *** significant at 1%













40

Panel

B: Alternate specifications

of segment divestment



Likelihood to divest



Transaction value
to segment assets



(1)

(2)





(3)

(4)

(5)

Aggregated familiarity dummy







-
0.224

**












(0.017)






Direct experience

-
0.592

***


-
0.436

*




-
0.227

**

-
0.204

*


(0.007)



(0.059)





(0.035)


(0.052)


Specialist

0.136













(0.336)












Specialist * direct experience

0.299













(0.336)












Generalist




-
0.059













(0.860)









Generalist * direct
experience




0.050













(0.911)









Inside
-
industry experience

-
0.262



-
0.203





-
0.214

*

-
0.145



(0.227)



(0.563)





(0.089)


(0.260)


Outside
-
industry experience

-
0.114



-
0.171





-
0.233


-
0.210



(0.655)



(0.477)





(0.122)


(0.157)


Missing transaction value











0.781

***












(0.000)


Cash flow
t
-
1
/sales
t
-
2

-
0.994

**


-
0.953

**


-
0.659

**

-
0.659

**

-
0.712

***


(0.011)



(0.014)



(0.011)


(0.011)


(0.005)


Industry median cash flow
t
-
1
/sales
t
-
2

1.503

**


1.501

**


1.060

***

1.066

***

0.991

***


(0.033)



(0.034)



(0.001)


(0.001)


(0.003)


Capx
t
-
1
/sales
t
-
2

-
1.362

***


-
1.305

***


-
0.277


-
0.275


-
0.129



(0.003)



(0.003)



(0.613)


(0.620)


(0.821)


Industry median capx
t
-
1
/sales
t
-
2

6.231

**


6.553

**


3.003

**

3.009

**

2.360

*


(0.026)



(0.020)



(0.014)


(0.016)


(0.073)


Cross
-
subsidization

0.954

**


0.938

**


-
0.192


-
0.192


-
0.252



(0.029)



(0.032)



(0.755)


(0.755)


(0.685)


Sales
t
-
1
/ firm sales

t
-
2

1.651

***


1.773

***


0.639

**

0.646

**

0.482



(0.009)



(0.005)



(0.046)


(0.041)


(0.133)


Core
segment

-
0.040



-
0.034



-
0.192


-
0.192


-
0.176



(0.878)



(0.896)



(0.204)


(0.208)


(0.252)


Segment's Tobin's
q

0.265

*


0.268

*


0.250

**

0.250

**

0.249

***


(0.075)



(0.072)



(0.011)


(0.011)


(0.010)


Size<10% dummy

0.076



0.113



0.279

**

0.280

**

0.289

**


(0.697)



(0.569)



(0.033)


(0.033)


(0.029)


Liquidity

1.314

*


1.254



0.819

*

0.816

*

0.729



(0.091)



(0.116)



(0.076)


(0.080)


(0.132)















Number of observations

1,049



1,049



875


875


875


McFadden
R
-
squared

5.62%





5.39%





5.56%



5.56%



8.09%



* significant at 10%; ** significant at 5%, *** significant
at 1%












41

Table 4:
Segment divestment logit r
egressions
:
subsamples
split according to CEO tenure

This table presents the results of binary logit regressions that explain from which type of segments firms
choose to divest assets

for the sub sample of firm years with variation in familiarity
. The dependent
variable takes th
e value of one for segments
with a divestiture
and zero for
fully
retained segments.
Regression (1) contains firm years with CEOs
that have

tenure up to two years. Regression (2) contains
firm years with CEOs
that have
tenure of
three or more
years.
We mea
sure familiarity by means of the
CEOs’ work experience, which we split into
direct
experience, inside
-
industry experience, and outside
-
industry experience.
All other variables are self
-
explanatory or defined more completely in Table 2. The
subscripts refer

to theyear relative to the year in which firms announce their divestment.
W
e truncate
ratios at
-
1 and +1. All regressions include year dummies and dummies for the number of segments

(unreported)
.
P
-
values appear in parentheses and are based on robust
standard errors clustered by firm
.



(1)



(2)





Newly hired

Longer tenured

Direct experience

0.105


-
0.669

***


(0.775)


(0.006)


Inside
-
industry experience

0.139


-
0.370



(0.720)


(0.190)


Outside
-
industry experience

-
0.049


-
0.186



(0.915)


(0.464)


Cash flow
t
-
1
/sales
t
-
2

-
1.233

*

-
0.862

*


(0.063)


(0.078)


Industry median cash flow
t
-
1
/sales
t
-
2

0.753


1.853

**


(0.647)


(0.021)


Capx
t
-
1
/sales
t
-
2

-
1.851

**

-
1.655

**


(0.015)


(0.026)


Industry median capx
t
-
1
/sales
t
-
2

3.522


7.703

**


(0.417)


(0.017)


Cross
-
subsidization

0.068


1.967

**


(0.892)


(0.016)


Sales
t
-
1
/ firm sales

t
-
2

2.341

**

1.588

*


(0.034)


(0.063)


Core
segment

-
0.297


0.104



(0.515)


(0.745)


Segment's Tobin's
q

0.355


0.117



(0.268)


(0.518)


Size<10% dummy

0.169


0.068



(0.649)


(0.787)


Liquidity

1.743


1.210



(0.167)


(0.205)







Number of observations

360


689


McFadden
R
-
squared

6.47%



7.02%



* significant at 10%; ** significant at 5%, *** significant at 1%




42

Table 5
: CEO
selection

Panel A presents the results of a binary logit regression that explains from which type of segment
s firms
choose to divest assets.
Model (1) consists of the full sample of divesting firm years, Model (2) the sub
sample of divesting firm years wit
h CEOs
having

tenure up to two years, and Model (3) the sub sample of
divesting firm years with externally hired CEOs
having

tenure up to two years.
The dependent variable
takes the value of one for segments
with a divestiture
and zero for
fully
retained segments.
Table 2 defines
a
ll variables.

The subscripts refer to the year relative to the year in which firms announce their divestment.
W
e truncate ratios at
-
1 and +1. All regressions include dummies for the number of segments

(unreported)
.
P
-
va
lues appear in parentheses and are based on
robust standard errors clustered by firm
.
Panel B presents the relation between the predicted probability of a segment to be divested, based on the
regression
s

in
Panel A
, for firm years
prior

to CEO
s’appointment
.
The p
anel
presents

the

mean, median
and standard deviation
of this predicted probability
per level of familiarity

for the sub sample of firm years
with variation in familiarity
.
We measure familiarity by means of the CEOs’ work experience, which we
split
into
direct
experience, inside
-
industry experience, and outside
-
industry experience.

Panel A:
Segment divestment binary logit regression



(1)



(2)



(3)





Full sample

Newly hired

Newly and externally
hired

Cash flow
t
-
1
/sales
t
-
2

-
0.598

**

-
1.083

**

-
1.533



(0.037)


(0.031)


(0.137)


Industry median cash flow
t
-
1
/sales
t
-
2

1.386

***

0.436


-
1.622



(0.008)


(0.689)


(0.390)


Capx
t
-
1
/sales
t
-
2

-
0.279


-
0.657


-
0.606



(0.468)


(0.220)


(0.501)


Industry median capx
t
-
1
/sales
t
-
2

1.789


4.637


7.222



(0.354)


(0.182)


(0.454)


Cross
-
subsidization

0.405


0.363


1.541



(0.326)


(0.338)


(0.407)


Sales
t
-
1
/ firm sales

t
-
2

1.058

**

1.738

**

1.288



(0.029)


(0.049)


(0.451)


Core
segment

-
0.062


-
0.262


-
0.497



(0.755)


(0.446)


(0.505)


Segment's Tobin's
q

0.272

***

0.39
0

**

0.787

**


(0.005)


(0.032)


(0.027)


Size<10% dummy

-
0.099


0.071


-
0.212



(0.531)


(0.812)


(0.725)


Liquidity

0.288


0.341


-
0.081



(0.576)


(0.686)


(0.954)









Number of observations

1805


578


180


McFadden
R
-
squared

4.01%



4.12%



9.16%



* significant at 10%; ** significant at 5%, *** significant at 1%








43

Panel B: Predicted probability to divest per type of familiarity



Direct exp.

Inside ind. exp.

Outside ind. exp.

Non
-
familiar



P
-
value of

differences



(1)

(2)

(3)

(4)



(1)
-

(4)

(2)
-

(4)

(3)
-

(4)

N

91

54

32

119














Based on Model (1), the full sample







Mean

0.415

0.341

0.397

0.366


0.001

0.052

0.116

Median

0.400

0.331

0.383

0.339


0.000

0.197

0.039

SD.

0.104

0.069

0.090

0.099














Based on Model (2), the sample with newly hired CEOs






Mean

0.411

0.348

0.402

0.366


0.004

0.173

0.073

Median

0.401

0.341

0.409

0.353


0.004

0.368

0.024

SD.

0.116

0.068

0.086

0.105














Based on Model (3), the
sample with newly and externally hired CEOs





Mean

0.351

0.311

0.395

0.334


0.460

0.401

0.016

Median

0.371

0.332

0.415

0.334


0.311

0.447

0.043

SD.

0.160

0.158

0.110

0.169






44

Table 6:
Segment divestment logit r
egressions

This table presents the
results of binary logit regressions that explain from which type of segments firms
choose to divest assets

for the sub sample of firm years with variation in familiarity
. The dependent
variable takes the value of one for divested segments and zero for reta
ined segments.
We measure
familiarity by means of the CEOs’ work experience, which we split into
direct

experience, inside
-
industry
experience, and outside
-
industry experience.
The high
entrenchment
dummy equals one
for firm years in
which the entrenchment

index as constructed by Bebchuk, Cohen and Ferrel (2009) equals three or higher
.
The good internal governance dummy equals one for firm years where the percentage of inside directors is
above the median of our sample. All other variables are self
-
explanat
ory or defined more completely in
Table 2. The subscripts refer to the year relative to the year in which firms announce their divestment.
W
e
truncate ratios at
-
1 and +1. All regressions include year dummies and dummies for the number of
segments

(unrepor
ted)
.
P
-
values appear in parentheses and are based on
robust standard errors clustered
by firm.



(1)



(2)



Direct experience

-
0.545

**

-
0.513

*


(0.047)


(0.054)


Direct exp
erience
* high entrenchment dummy

0.106





(0.727)




High entrenchment
dummy

-
0.120





(0.398)




Direct exp
erience

* good internal governance dummy



0.021





(0.943)


Good internal governance



-
0.119





(0.337)


Inside
-
industry experience

-
0.238


-
0.200



(0.283)


(0.371)


Outside
-
industry experience

-
0.159


-
0.150



(0.514)


(0.550)


Cash flow
t
-
1
/sales
t
-
2

-
0.932

**

-
1.015

**


(0.018)


(0.013)


Industry median cash flow
t
-
1
/sales
t
-
2

1.621

**

1.632

**


(0.021)


(0.028)


Capx
t
-
1
/sales
t
-
2

-
1.354

***

-
1.348

***


(0.003)


(0.002)


Industry median capx
t
-
1
/sales
t
-
2

6.592

**

7.274

**


(0.021)


(0.011)


Cross
-
subsidization

1.029

**

0.995

**


(0.023)


(0.021)


Sales
t
-
1
/ firm sales

t
-
2

1.753

***

1.973

***


(0.006)


(0.003)


Core
segment

0.004


-
0.045



(0.989)


(0.865)


Segment's Tobin's
q

0.28

*

0.268

*


(0.061)


(0.078)


Size<10% dummy

0.079


0.081



(0.689)


(0.683)



45

Liquidity

1.365

*

1.401

*


(0.097)


(0.097)







Number of observations

1023


1004


McFadden
R
-
squared

5.36%



5.74%



* significant at 10%; ** significant at 5%, ***
significant at 1%







46

Table 7:
Segment divestment logit r
egressions

This table presents the results of binary logit regressions that explain from which type of segments firms
choose to divest assets
.
The dependent variable takes the value of one for divested segments and zero for
retained segments.
Panel A only incorporates direct
-
experience segments of the sub sample of firm years
with variation in familiarity and Panel B incorporates all segments of
firm years with variation in
familiarity
.
We measure familiarity by means of the CEOs’ work experience, which we split into
direct
experience, inside
-
industry experience, and outside
-
industry experience.
All other variables are self
-
explanatory or de
fined m
ore completely in Table 2
. The subscripts refer to the year relative to the year in
which firms announce their divestment.
W
e truncate ratios at
-
1 and +1. All regressions
in Panel B
include
year dummies and dummies for the number of segments

(unreported)
.

P
-
values appear in parentheses and
are based on
robust standard errors clustered by firm
.

Panel A:
Sample of direct
-
experience segments



(1)



(2)



(3)



Direct experience, 4
-

7 years work experience

-
0.151







(0.663)






Direct experience, > 7

years work experience

0.062







(0.862)






Direct experience, left 5
-
9 years ago



-
0.163







(0.608)




Direct experience, left > 9 years ago



-
0.269







(0.318)




Direct experience, 2
-
4 yrs corporate experience





-
0.605

**






(0.035)


Direct experience, >4 yrs corporate experience





0.061







(0.808)


Constant

-
0.445


-
0.384

**

-
0.336

**


(0.102)


(0.029)


(0.048)









Number of observations

323


317


317


McFadden R
-
squared

0.14%



0.23%



1.57%



* significant
at 10%; ** significant at 5%, *** significant at 1%







Panel
B
:
Full sample



(1)



(2)



(3)



Direct experience, < 4 years work experience

-
0.415







(0.152)






Direct experience, 4
-

7 years work experience

-
0.519

**






(0.023)






Direct experience, > 7 years work experience

-
0.384







(0.154)






Direct experience, left < 5 years ago



-
0.382

*






(0.088)




Direct experience, left 5
-
9 years ago



-
0.465







(0.128)




Direct experience, left > 9 years ago



-
0.630

***






(0.006)




Direct experience, < 2 yrs corporate experience





-
0.386

*


47






(0.083)


Direct experience, 2
-
4 yrs corporate experience





-
0.913

***






(0.000)


Direct experience, >4 yrs corporate experience





-
0.332







(0.168)


Inside
-
industry experience

-
0.240


-
0.260


-
0.289



(0.274)


(0.231)


(0.186)


Outside
-
industry experience

-
0.177


-
0.192


-
0.240



(0.456)


(0.415)


(0.305)


Cash flow
t
-
1
/sales
t
-
2

-
0.952

**

-
0.987

**

-
1.081

***


(0.014)


(0.011)


(0.005)


Industry median cash flow
t
-
1
/sales
t
-
2

1.467

**

1.494

**

1.456

**


(0.040)


(0.032)


(0.039)


Capx
t
-
1
/sales
t
-
2

-
1.301

***

-
1.278

***

-
1.375

***


(0.003)


(0.004)


(0.001)


Industry median capx
t
-
1
/sales
t
-
2

6.651

**

3.345


3.781



(0.018)


(0.283)


(0.210)


Cross
-
subsidization

0.934

**

0.831

*

1.033

**


(0.033)


(0.068)


(0.020)


Sales
t
-
1
/ firm sales

t
-
2

1.765

***

1.869

***

1.917

***


(0.005)


(0.002)


(0.002)


Core
segment

-
0.036


-
0.027


-
0.013



(0.891)


(0.917)


(0.961)


Segment's Tobin's
q

0.265

*

0.265

*

0.245



(0.075)


(0.076)


(0.106)


Size<10% dummy

0.110


0.160


0.159



(0.579)


(0.414)


(0.425)


Liquidity

1.237


1.086


0.918



(0.118)


(0.173)


(0.260)









Number of observations

1049


1043


1037


McFadden R
-
squared

5.40%



5.11%



5.70%



* significant at 10%; ** significant at 5%, *** significant at 1%









48

Table 8:
Analysis of
CA
Rs to divestiture announcements

Panel A presents the means, standard deviations, and mean differences of the
cumulative abnormal returns
over days
-
1 to +1 relative to the divestiture announcement. We estimate the abnormal returns by means of
the market model with an estimation window running from day
-
160 to day
-
41 relative to the
announcement date.

The abnorma
l returns are winsorized at the one
-

and 99
-
percentile values.
We measure
familiarity by means of the CEOs’ work experience, which we split into
direct
experience, inside
-
industry
experience, and outside
-
industry experience. Newly
-
hired CEOs are CEOs with t
enure up to two years
and l
onger
-
tenured CEOs are CEOs with tenure of three
or more
years. Panel B presents the results of
ordinary least squares regressions of three
-
day CARs to divestiture announcements.
The relative
transaction size is the transaction v
alue divided by the book value of the firm’s total assets.
We calculate
the firm’s industry adjusted capital expenditures, leverage, and cash as the firm variable minus the median
of all Compustat firms with the same two
-
digit SIC code.
All other variables

are defined in Table
s1 and 2
.
P
-
values appear in parentheses and are based on
Huber
-
White standard errors.

Panel A
: CARs in familiarity and tenure subsamples





Direct exp.

Inside ind. exp.

Outside ind. exp.

Non
-
familiar





(1)



(2)



(3)



(4)



Full sample

Mean

0.42%

*

0.39%


0.69%

**

0.74%

***


SD.

(4.30%)


(4.64%)


(4.19%)


(3.97%)



N

286


84


166


373












Sample with available
information

Mean

0.55%


0.27%


0.84%


0.79%

*
*

SD.

(4.64%)


(4.64%)


(4.63%)


(4.35%)



N

122


40


81


177






















Longer
-
tenured CEOs










Full sample

Mean

0.51%

*

0.76%


0.40%


0.56%

*
*


SD.

(4.25%)


(4.63%)


(4.05%)


(3.95%)



N

193


46


90


278












Sample with available
information

Mean

0.70%


0.42%


0.51%


0.50%


SD.

(4.50%)


(3.49%)


(3.93%)


(4.23%)



N

80


23


46


142












Newly
-
hired CEOs










Full sample

Mean

0.23%


-
0.06%


1.04%

**

1.28%

*
**


SD.

(4.43%)


(4.68%)


(4.36%)


(4.00%)



N

93


38


76


95












Sample with available
information

Mean

0.28%


0.06%


1.28%


1.96%

*
*

SD.

(4.94%)


(5.96%)


(5.44%)


(4.70%)




N

42



17



35



35




49

Table 8: Analysis of CARs to divestiture announcements (cont.)

Panel B
: Regression analysis explaining CARs



(1)



(2)



(3)



(4)



Direct experience

0.003



0.002



0.010








(0.512)


(0.780)


(0.114)




Direct experience,
long tenured







0.016

*








(0.051)


Direct experience,
newly hired







0.003









(0.669)


Excess value



-
0.001


-
0.001


-
0.001





(0.822)


(0.931)


(0.832)


C
ore segment



0.004


0.003


0.002





(0.417)


(0.528)


(0.710)


Internally hired CEO





-
0.015

**

-
0.014

**






(0.039)


(0.041)


Founders





-
0.028

**

-
0.032

**






(0.027)


(0.021)


Inside
-
industry experience

-
0.001


-
0.002


0.001


0.001



(0.857)


(0.773)


(0.872)


(0.868)


Outside
-
industry experience

-
0.000


-
0.003


-
0.006


-
0.006



(0.966)


(0.653)


(0.373)


(0.402)


Relative transaction size

0.102

***

0.106

***

0.107

***

0.106

***


(0.001)


(0.001)


(0.001)


(0.001)


Tobin's
q

-
0.006

***

-
0.006

***

-
0.007

***

-
0.007

***


(0.002)


(0.005)


(0.004)


(0.005)


Industry
-
adj. capx
t
-
1
/sales
t
-
2

0.130


0.149


0.143


0.148



(0.107)


(0.127)


(0.146)


(0.137)


Industry
-
adj. leverage

0.020


0.021


0.018


0.018



(0.149)


(0.140)


(0.229)


(0.233)


Industry
-
adj, cash
t
-
1
/sales
t
-
2

0.008


0.006


-
0.002


0.005



(0.727)


(0.838)


(0.956)


(0.858)


Percentage of stock owned

-
0.038


-
0.042


-
0.028


-
0.030



(0.313)


(0.282)


(0.477)


(0.440)


Intercept

0.010

*

0.010

*

0.021

***

0.022

***


(0.051)


(0.054)


(0.005)


(0.005)











Number of observations

420


386


386


386


Adjusted R
-
squared

6.30%



6.20%



7.50%



7.70%



* significant at 10%; ** significant at 5%, *** significant at
1%