The motivation and impact of pension fund...

Journal of Financial Economics 52 (1999) 293}340

The motivation and impact of pension fundactivismq

Diane Del Guercio!,*, Jennifer Hawkins"

!Lundquist College of Business, University of Oregon, Eugene, OR 97403, USA"Hawkins Pound Consulting, Brookline, MA 02446, USA

Received 7 August 1997; received in revised form 19 May 1998; accepted 1 January 1999

Abstract

Pension funds have pursued an active role in corporate governance, although somequestion their e!ectiveness and motivations. We examine the impact and motivation ofpension fund activism by studying the shareholder proposals of the largest, most activefunds from 1987 through 1993. We "nd signi"cant heterogeneity across funds in activismobjectives, tactics, and impact on target "rms, consistent with di!ering investmentstrategies. We "nd the funds are more successful at monitoring and promoting change intarget "rms than previously recognized. We also "nd no evidence to support motivationsother than fund value maximization. ( 1999 Elsevier Science S.A. All rights reserved.

JE¸ classi,cation: G23; G34

Keywords: Shareholder activism; Pension funds; Corporate governance; Proxy contest;Control activity

*Corresponding author. Tel.: #1-541-346-5179; fax: #1-541-346-3341.

E-mail address: [email protected] (D. Del Guercio)qEarlier versions of this paper were circulated under Jennifer Hawkins' previous surname of van

Heeckeren. Most of the work on this paper was completed while Hawkins was a faculty member atthe University of Oregon and a corporate governance specialist at McKinsey & Co. We are gratefulfor helpful comments from John Chalmers, Larry Dann, Stu Gillan, Jarrad Harford, DavidHaushalter, Andrew Karolyi, Jon Karpo!, Wayne Mikkelson, Megan Partch, John Pound, RobertaRomano, Paula Tkac, Michael Weisbach (the referee), and participants in seminars at the 1997 FMAConference (Honolulu), the Paci"c Northwest Finance Conference, the University of Notre Dame,the University of Western Ontario, and the University of Wisconsin. We appreciate the time ofindividuals from the pension funds in our sample who kindly agreed to interviews. We thank JohnBlease, Elisa Canum, Wayne Lottinville, Marc Goettel, Mark Simonson, and especially Todd Perryfor valuable research assistance.

0304-405X/99/$ - see front matter ( 1999 Elsevier Science S.A. All rights reserved.PII: S 0 3 0 4 - 4 0 5 X ( 9 9 ) 0 0 0 1 1 - 2

1. Introduction

Large public pension funds pursue a highly active role in the governance ofcompanies principally through the submission of shareholder proposals. Theseproposals request changes ranging from altering the structure of board gover-nance or management incentives to the removal of takeover defenses. Severalstudies conclude that shareholder proposals are generally ine!ective. Someargue that this is not surprising given the agency problems within the fundsthemselves. Romano (1993) argues that public pension funds are subject topressures to take actions that are politically popular, but that harm the funds'investment performance. Murphy and Van Nuys (1994) maintain that publicpension funds are run by individuals who do not have the proper incentives tomaximize fund value. Both studies argue that public fund managers may useproposals to generate publicity or enhance their reputations in order to gainfuture employment or political opportunities. In support of these arguments,Wahal (1996) and Karpo! et al. (1996) "nd little evidence that operatingperformance or share price improves for companies that are the targets ofa shareholder proposal. In their survey articles, Karpo! (1998) and Black (1997)summarize the empirical evidence and conclude that proposals are ine!ective.

We revisit the question of the motivation and impact of pension fund activismby studying the shareholder proposals of the largest and most active funds from1987 through 1993. Although these funds appear to be quite similar, we showthat recognizing important sources of heterogeneity aids in our understandingof the funds' underlying motivation and their impact on target "rms. In particu-lar, it is important to take into account any constraints dictated by theirinvestment strategies. For example, a fund that is heavily indexed might pursueactivism tactics aimed at boosting the performance of the stock market overall.A reasonable goal for an index fund might be to a!ect the behavior andmanagement not only of the companies it targets, but also of many othercompanies that proactively make changes to avoid con#ict and public scrutiny.Thus, a broad-based and highly publicized activism program might be inter-preted as optimal from a fund value maximization perspective, rather thana source of perquisites to the fund managers. Richard Koppes, former chiefcounsel of the heavily indexed CalPERS fund, notes, &It makes sense for us to tryto raise the ocean in order to lift our boat' (remarks at Stanford University`Directors' Collegea, March 21, 1996). Similarly, whether investment and trad-ing decisions are made by internal fund managers, or delegated to outsidemanagers, a!ects a fund's ability to pro"tably coordinate trading and activismdecisions.

Our study contributes to ongoing research on the motivation and impact ofshareholder activism in three ways. First, we document signi"cant heterogeneityin fund objectives and activism tactics that are closely linked to di!erences in thefunds' investment strategies. Second, we present evidence that shareholder

294 D. Del Guercio, J. Hawkins / Journal of Financial Economics 52 (1999) 293}340

proposals have a signi"cant impact on company policies, and that variation inimpact across target "rms is related to fund heterogeneity. Finally, we examinethe target selection criteria of the funds, along with their trading patterns aroundthe submission of proposals, and "nd no evidence that they are pursuingobjectives other than bene"ciary wealth maximization. We "nd that the fundsdi!er in their tendency to buy and sell target shares around proposals, but thesedi!erences are consistent with their investment strategies. For example, themore heavily indexed funds show little variation in holdings around shareholderproposals, while the most actively managed fund exhibits signi"cant and pro"t-able movements in its holdings.

We focus on assessing the e!ectiveness of the funds in generating signi"cantchanges in target company policies. Overall, relative to a performance-, size- andindustry-matched control sample, we "nd that companies receiving publicpension fund proposals subsequently experience a higher frequency of gover-nance events such as shareholder lawsuits as well as responsive corporatepolicies such as asset sales, restructurings, and layo!s. However, we "nd di!er-ences in target managements' response across the sample funds. Proposalssponsored by CalPERS, which has the largest ownership stakes among theexternally managed indexed funds, appear to have the broadest and mostsubstantial impact on subsequent events at target "rms. Proposals sponsored bythe other externally managed indexed funds, California State Teachers Retire-ment System (CalSTRS) and the New York City funds (NYC), also appear togenerate signi"cant change. In contrast, consistent with their narrower statedobjectives, proposals sponsored by the internally managed College RetirementEquities Fund (CREF) and State of Wisconsin Investment Board fund (SWIB)are not associated with general increases in governance-related events. Rather,these two funds are generally successful in convincing corporate management toadopt the speci"c changes requested in the proposal.

We also "nd that "rms that receive antitakeover proposals, those targeted byCalPERS, and those with proposals that garner enough votes to pass havea signi"cantly higher probability of a takeover attempt, after controlling forvariables related to takeover probability. This evidence suggests that proposalsare low-cost mechanisms that can be fruitfully used to further a number of goals,such as putting pressure on management, signaling to the market the views ofthe fund regarding target company management, and building shareholdersupport for more costly governance activity such as takeovers. Similar to otherstudies, we "nd no evidence that this activity has signi"cant e!ects on stockreturns or accounting measures of performance in the three years following aninitial targeting, and only sketchy evidence of positive e!ects in the short term.Overall, we conclude that the activist funds in our sample are generally success-ful in furthering their stated objectives. Furthermore, any potential agencyproblems within the funds do not appear to be responsible for the lack ofmeasurable wealth e!ects.

D. Del Guercio, J. Hawkins / Journal of Financial Economics 52 (1999) 293}340 295

We begin our study by providing some perspective on pension fund activismin general and our sample funds in particular. In Section 3, we describethe heterogeneity in investment strategies as well as activism objectivesand tactics based on our interviews with key decision makers at the funds.Section 4 presents our main evidence on the corporate events that followpension fund proposal submissions. Section 5 discusses important issues rel-evant to interpreting event study results in this context and presents evidence onaccounting and long-run stock return measures of target performance. Section6 examines whether the funds' target selection and trading behavior appear to beconsistent with fund value maximization, and Section 7 presents our con-clusions.

2. Historical perspective on shareholder activism: emergence of pension funds asactivists

Shareholder proposals are a governance mechanism created by the Securitiesand Exchange Commission (SEC) under Section 14 of the Securities andExchange Act of 1934. First used in 1942, proposals are brief statementssubmitted by a shareholder requesting a speci"c action by management. Theyare included in the annual proxy statement at the expense of the "rm, along withmanagements o$cial response and voting recommendations regarding theproposal. Proposals are almost always advisory due to provisions in state law.Even if the proposal passes with a majority of votes, management is not requiredto take the requested action. As intended by the SEC, however, a passing votewould e!ectively communicate the consensus views of dispersed shareholders tomanagement. For a more complete description of rules regarding shareholderproposals see Gordon and Pound (1993) and John and Klein (1995).

Until the emergence of large institutional investors in the late 1980s, share-holder proposals were used almost exclusively by individual gad#y investorsand social activist groups. During this time proposals never received enoughvotes to pass, and it was rare for a proposal to garner more than 10% of thevotes in its favor. In 1987, institutional investors began to submit proposals oncorporate governance topics. Around this time, corporate governance share-holder proposals began to gain signi"cant support, despite management opposi-tion. In our sample of 266 proposals sponsored by pension funds, the averagepercent of votes in favor is 34%, with 15 proposals receiving a majority.

The 1990s brought a few key developments. Partly in response to requestsfrom activist funds, in 1992 the SEC relaxed restrictions in the proxy rules ondisclosure of communications among shareholders, signi"cantly lowering thecosts and potential legal liability associated with shareholder activism. Shararaand Hoke-Witherspoon (1993) report that these were the "rst major changes tothe proxy rules in nearly 40 years. They argue that the changes were in &response


to institutional investors such as CALPERS' (pp. 327 and 337). Less publicforms of activism such as private letters and phone calls to management becameincreasingly common. The funds reached the point at which few corporatemanagements ignored requests to meet with them, and many times managementmade the requested change without a proposal being formally "led. Thus, inmany instances "ling a proposal on the proxy statement became unnecessaryand was used only as a &last resort'. Of course, this would not work if the crediblethreat of the proposal did not exist. As Kurt Schacht of SWIB puts it, &everyonce in a while the junkyard dog has to bite'.

We study the proposals submitted in 1987 through 1993 by the largest andmost activist pension funds: CREF, CalPERS, CalSTRS, SWIB, and NYC.These funds represent 16% of the assets of the top 200 pension funds and 12% ofthe 1000 largest pension funds, according to data from the January 1995 issue ofPensions and Investments. Our "ve sample funds are also the most active fundson governance issues, representing 18% of all corporate governance proposalsmade in our sample period, according to data from the Investor ResponsibilityResearch Center (IRRC).

Table 1 summarizes the distribution by topic and year of the 266 proposalssubmitted to 125 "rms by our sample funds. Proposal topics fall into three maincategories: voting issues, such as a request that voting be con"dential; antitake-over issues, such as rescinding the poison pill or opting out of state antitakeoverlegislation; and board of directors issues, such as requesting that a majority ofthe board consist of independent directors. Voting issues are a popular proposaltopic throughout the sample period, with antitakeover proposals becoming lesscommon and board issues becoming more common in the 1990s. This trend inproposal topics mirrors the trend among non-pension fund proposal sponsors.Despite some overlap, however, some issues popular with individual gad#ieswere notably absent from the pension fund agenda. For example, according tothe IRRC Bulletin, capping the level of CEO compensation was the second mostpopular proposal topic in 1994, but these were not sponsored by any of oursample funds.

The ownership statistics in Table 2 show that a signi"cant portion of thesample funds' total portfolio value is devoted to activism, with relatively largestakes in individual targets. The dollar amount invested in portfolio "rmstargeted to receive proposals ranges from $422 million at CalSTRS to $2.2billion at CREF. Although CalPERS is perceived to be the most activist fund,CREF has nearly double the dollar investment, and almost the same number oftarget "rms, as CalPERS. The average percentage stake in target "rms owned bythe funds ranges from 0.4% for CalSTRS to 2.3% for SWIB. In contrast, Wahal(1996) reports that the average percentage stake in target "rms by inactiveinstitutional investors is only 0.3%. Public pension funds not only have largestakes in individual targets, they also tend to retain sole voting authority overtheir externally managed shares. According to Brancato (1993), public funds


Tab

le1

Fre

quen

cydistr

ibution

ofsh

areh

old

erpro

posa

lsby

pen

sion

fund

spons

oran

dby

year

This

tabl

eco

ntai

ns

all

shar

ehol

der

pro

pos

als

subm

itte

dto

port

folio"rm

sby

the

sam

ple

pens

ion

fund

sfrom

1987}19

93,in

cludin

gth

ose

that

wer

esu

bseq

uen

tly

withdr

awn

by

the

pen

sion

fund.

Inth

era

reca

seofa

pro

posa

lth

atis

co-s

pons

ore

dby

two

sam

ple

pen

sion

funds

,itis

coun

ted

only

once

under

the

prim

ary

spons

or

Pro

posa

lty

peFund

spon

sor

Yea

rT

otal

SWIB

CR

EF

CA

LP

ER

SC

ALSTR

SN

YC

8788

8990

9192

93

Vot

ing

issu

esC

on"de

ntia

lvo

ting

216

105

850

1128

2822

1514

118

One

shar

eon

evo

te0

00

10

10

00

00

01

Don't

countab

sten

tion

sas

vote

sag

ainst

00

20

00

00

11

00

2

Discl

ose

shar

ehol

der

spon

sor

00

00

50

00

00

05

5

Ant

itak

eove

ris

sues

Res

cind

poison

pill

2026

1817

228

1715

139

01

83O

ptout

ofst

ate

antita

keo

ver

law

00

51

40

04

33

00

10A

nti-g

reen

mai

l0

01

20

02

10

00

03

Gold

enpa

rach

ute

00

02

00

01

10

00

2Tar

gete

dsh

are

pla

cem

ent

012

00

00

00

25

32

12

Boa

rdof

Dir

ecto

rsis

sues

Shar

ehold

erad

viso

ryco

mm

itte

e0

06

00

00

12

20

16

Maj

ority

ofouts

ider

son

boar

d0

01

06

00

00

22

37

Maj

ority

ofouts

ider

son

com

pens

atio

nco

mm

itte

e0

02

05

00

00

21

47

Sepa

rate

chai

rman

and

CE

O0

02

04

00

00

01

56

Inde

pende

ntno

min

atin

gco

mm

itte

e0

00

04

00

00

00

44

Tota

l22

5447

2811

529

3050

5046

2239

266


Table 2Activist pension fund holdings in target "rms (1986}1994)

This table reports summary statistics on pension fund holdings based on quarterly 13F "lings fromJune 1986 to June 1994. The New York City funds are omitted from this table because they do not"le 13F "lings directly. Instead, their holdings are reported by their external managers. The targetingdate is de"ned as the quarter of the proposal's outcome, which is typically the annual meeting date.Since many "rms are targeted multiple times, the table below speci"es whether the statistics arerelative to the "rst or last targeting. Values in brackets are medians

SWIB CALPERS CALSTRS CREF

Average total domestic equity portfolio value($ millions)

9,400 19,000 12,700 31,800

Average number of total stock holdings inthe portfolio

576 1,112 3,739 1,767

Number of unique targets (1986}1994) 17 35 17 32

Percent of portfolio value in all target "rmsin year before "rst targeting

12.64 6.94 3.48 7.85

Total dollar investment in target "rms inyear before "rst targeting (millions)

939 1,130 422 2,151

Average dollar holding in target "rms inyear before "rst targeting (millions)

55.3 34.2 24.8 67.2[51.4] [25.5] [9.9] [67.5]

Average percentage ownership stake intarget "rms in year before "rst targeting

2.3 1.0 0.4 1.1[1.6] [0.7] [0.4] [1.1]

Average percentage ownership stake in yearafter last targeting

0.4 0.7 0.5 1.0[0] [0.6] [0.4] [1.0]

Average percent of portfolio invested intarget "rms in year before "rst targeting

0.74 0.20 0.22 0.25[0.60] [0.18] [0.08] [0.24]

Average percent of portfolio invested intarget "rms in year after last targeting

0.34 0.17 0.19 0.28[0] [0.10] [0.10] [0.21]

retain e!ective voting control over 98.9% of the stock they hold, compared to66.4% for the average institutional investor. Together, these numbers suggestthat the funds in our sample have the potential to be in#uential in their portfoliocompanies, and thereby have an economic incentive to become active.

3. Variation in pension fund strategies, objectives, and economic incentives

Our approach to examining fund motivation and impact is to identify andincorporate the heterogeneity in our sample fund's activism objectives andinvestment strategies. We seek to examine whether the heterogeneous objectives


and strategies of the funds appear to be consistent with their own measures ofsuccess, and with the normative goal of fund value maximization. Given theaggregate nature of most tests, it is possible that this underlying heterogeneity ispartly responsible for the lack of consensus in the literature on the e!ectivenessof public pension fund activism. In this section, we outline key sources ofheterogeneity among funds. These di!erences could shed light on the observedbehavior of funds and their impact on target "rms.

One important source of heterogeneity is the extent to which a fund ispassively indexed. Funds devoted to indexing do not have the &Wall Street Walk'option of selling when displeased with stock price performance. Thus, relative toactively managed funds, indexed funds have an incentive to promote spillovere!ects that boost the performance of the stock market overall, rather than thatof speci"c stocks. Publicity is one potentially e!ective tool in promoting suchspillover e!ects, since it a!ects not only the direct target of a funds' activism butalso other companies who observe it. The threat of publicity might give fundsleverage with target management, and might also motivate other companies toproactively improve their corporate governance structures without beingexplicitly targeted.

Although spillover e!ects are di$cult to measure, anecdotal evidencesuggests that the kinds of practices that pension fund activists have beenpromoting in individual targets are being adopted more broadly. For example,The Business Roundtable, an association of CEOs of large corporations, re-leased a Statement on Corporate Governance in September 1997 that lists theirrecommendations on the best practices regarding governance issues. Severalrecommended practices in this report, including having a majority of indepen-dent directors on the board and only independent directors on key boardcommittees (audit, compensation, nominating), were common topics of share-holder proposals in the early 1990s. The signi"cance of this is best summed upin a Business Week report: &Indeed, the biggest news in the Roundtable's reportmay be that once radical ideas about corporate governance are now "rmlyin the mainstream } and are accepted by a group made up of 200 CEOs of thenation's largest corporations and headed by the chieftains of Caterpillar,Johnson&Johnson, Chase Manhattan, and General Motors' (Business Week,September 22, 1997, p. 36).

Although we cannot unambiguously attribute these changes to the e!orts ofthe pension fund activists, Hawley et al. (1994) provide additional anecdotalevidence that externality e!ects exist. They report evidence based on interviewswith top CalPERS o$cials that nontargeted "rms pay attention to CalPERS'interactions with target "rms: &Indeed, in the last two years, CalPERS has beensolicited via telephone calls, letters, faxes, and personal visits from numerousCEOs of non-targeted "rms (underperforming and well performing) seeking toopen the lines of communications should the need arise to explain futureproblems in order to stay o! CalPERS' target list'. These observations suggest


that publicity can be an e!ective tool to promote broad, marketwide changes.Monks and Minow (1995) make a similar point: &Perhaps the public pensionfunds' most signi"cant contribution has been to make the world an uncomfort-able place for a director of an underperforming company'.

Another source of heterogeneity is the internal versus external nature of fundmanagement. Funds that delegate investment functions to external managerse!ectively disconnect their activism e!orts from their investment actions, thuspreventing them from pro"tably trading on any private information that resultsfrom their activism. This is important since a shareholder large enough toin#uence corporate management has access to a source of gains not available toan ordinary stock-picker. Maug (1998) and Kahn and Winton (1998) presentformal models of a large shareholder's choice between engaging in activism orselling the position. These models show that a shareholder in#uential enough toe!ect change can generate trading pro"ts by buying shares at the low price thatdoes not yet re#ect the impact of the improvements. In other words, the initialmonitoring actions of the shareholder are private information, and pro"ts canbe generated by buying stock prior to any publicity concerning the activism.Once the market learns of the intention to monitor (perhaps from observinga very large stake), the free-rider problem comes into play since the in#uentialshareholder must buy additional shares at the higher price. Thus, unlike anindex fund, an internally and actively managed fund would have little incentiveto publicize its activism e!orts.

Table 3 provides summary information about the organization, investmentstrategy, and activism goals of each of the funds in our sample, as revealed to usin telephone interviews with key decision makers (full transcripts of the inter-views are available from the authors by request). The funds have similargovernance and oversight systems in that they each have investment andactivism programs developed and implemented by fund sta! and overseen andapproved by boards of trustees. There are, however, substantial di!erencesacross the funds in investment strategy, goals of proposals, de"nitions ofsuccessful proposals, and views on using publicity.

The funds range from mostly indexed (CalPERS, CalSTRS, CREF, and NYC)to almost entirely active stock picking (SWIB). The California and NYC fundsinvest heavily through outside money managers, while SWIB and CREF havevery active in-house investment analysts. Even though CREF is 80% indexed,they tend to take large bets with the remainder of the fund. This is con"rmed ina Wall Street Journal interview of Douglas Dial, CREF's in-house moneymanager, who reports that 16% of CREF's portfolio is devoted to taking bigstakes in 100 to 150 companies (April 4, 1995, p. C1). A more recent story on theDow Jones newswire titled &$100 Billion Fund: Indexing Plus Stock PickingPays O!' also con"rms these numbers (October 20, 1997). Unlike the funds thatinvest through outside managers, CREF retains discretion over their non-indexed stock positions, which would allow for pro"table trading in stocks that


Tab

le3

Sum

mar

yofpen

sion

fund

orga

niza

tion

and

activi

smobje

ctiv

es

This

table

isa

sum

mar

yof

tele

phone

inte

rvie

wsw

ith

seni

orde

cision

mak

ersat

the

funds

,using

ast

anda

rdiz

edse

tof

ques

tions

(con

duc

ted

Feb

ruar

yan

dM

arch

1997

).In

the

case

ofC

AL

PER

Sw

eco

ndu

cted

the

inte

rvie

ww

ith

the

deci

sion

mak

erdur

ing

the

sam

ple

per

iod,r

athe

rth

anth

ecu

rren

tdec

isio

nm

aker

CA

LP

ER

SC

ALSTR

SC

REF

NE

WY

OR

KC

ITY

SWIB

Ric

hard

Koppe

s,fo

rmer

Chi

efC

ouns

elC

EO

,Jan

ice

Hes

ter-

Am

ey,

Cor

por

ate

A!ai

rsA

dvisor

Pet

erC

lapm

an,C

hief

Coun

sel,

Inve

stm

ents

Jon

Lukom

nik,

Dep

uty

Com

ptro

ller

Kurt

Schac

ht,C

hie

fC

ounse

l

Org

aniz

atio

nBO

T"B

oard

ofTru

stee

s

BO

Ta

mix

ofgo

vern

or

appo

inte

es,ben

e"ci

ary

reps,

and

elec

ted

o$

cial

s.C

hief

Cou

nsel

sta!

prop

ose

sac

tivi

smpla

n,

BO

Tdi

scuss

esan

dap

prov

es.

BO

Ta

mix

ofgo

vern

or-

appr

ove

dap

poin

tees

and

elec

ted

o$ci

als.

Gove

r-na

nce

pol

icie

sse

tby

Inve

stm

entC

om

mitte

e,ra

ti"ed

byBO

T,an

dim

plem

ente

dby

sta!

.

100%

bene"

ciar

y-el

ecte

dBO

T.C

hief

Couns

elst

a!re

com

men

ds

activi

smpr

ogra

m,BO

Tre

view

san

dap

pro

ves.

Fiv

epl

ans

with

sepa

rate

BO

Ts,

each

with

am

ixof

polit

ical

appo

inte

esan

del

ecte

do$

cial

s.C

om

ptro

ller

over

sees

activi

smpolic

ies

for

all

plan

s.Sta!

reco

mm

ends

polic

y,B

OT

appro

ves.

BO

Ta

mix

ofgo

vern

or

and

ben

e"ci

ary

appo

inte

es.C

hief

coun

selst

a!re

com

men

ds

activi

smpr

ogra

ms,

BO

Tap

prov

es.

Inve

stm

ent

stra

tegy

:A

lleq

uitie

sex

tern

ally

man

aged

,with

full

disc

retion

tofu

nd

man

ager

s.80

%in

dex

ed,

20%

active

.A

ctiv

ism

isno

tlink

edto

buy/

sell

deci

sions.

All

equitie

sex

tern

ally

man

aged

,with

full

disc

retion

tofu

nd

man

ager

s.L

arge

lyin

dex

ed.A

ctiv

ism

isnot

linke

dto

buy

/sel

lde

cisions.

All

equitie

sm

anag

edin

tern

ally

.80

%in

dexe

dan

d20

%ac

tive

.Pos

t19

95m

aylink

trad

ing

and

gove

rnan

ceissu

es,

pre

not.

Ext

ernal

lym

anag

edan

dhe

avily

index

ed.V

oting

righ

tsre

tain

edby

fund

s.A

ctiv

ism

isno

tlinke

dto

buy/

sell

dec

isio

ns.

Act

ive

inte

rnal

man

agem

entw

ith

valu

ein

vest

ing

focu

sin

80%

of

port

folio

.20

%in

exte

rnal

lym

anag

edin

dex

fund

s.A

ctiv

ism

isno

tlink

edto

buy/

sell

deci

sions.


Obje

ctiv

espro

posa

lac

tivi

ty:

Initia

llyto

get

man

agem

ent's

atte

ntion.

As

CA

LPE

RS

bec

ame

better

know

n,pro

pos

als

used

asle

vera

geto

enco

ura

geac

tion

and

shar

ehol

der

valu

efo

cus.

Mor

ege

ner

ally

to&rai

seth

eoce

anin

order

tolif

tou

rboa

ts'.

To

dra

wat

tention

toth

epoo

rper

form

ance

,get

abe

tter

unde

rsta

ndin

gof

the

com

pan

yan

dits

long-

term

stra

tegy

.&N

ever

inte

rest

edin

puttin

ga

com

pan

yin

pla

y'sinc

eth

eyty

pic

ally

ow

nth

eac

quirin

g"rm

asw

ell.

To

hav

eth

esp

eci"

cre

ques

tad

opte

d.B

elie

veev

entu

ally

gove

rnan

cest

ruct

ure

a!ec

tsper

form

ance

.Pro

duct

ive

dial

ogue

ongo

vern

ance

issu

esan

dbe

tter

rela

tions

with

port

folio

com

pani

esar

eside

ben

e"ts

of

prop

osa

ls.

&To

mak

em

oney'.

Tw

ow

ays:&rai

seal

lbo

ats'

bych

angi

ngco

rpor

ate

gove

rnan

ceen

viro

nm

ent;

chan

gego

vern

ance

stru

cture

and

shar

ehold

erva

lue

focu

sof

spec

i"c

com

pani

es.

To

hav

eth

esp

eci"

cre

ques

tad

opte

d.B

elie

vea

good

gove

rnan

cest

ruct

ure

willa!

ect

per

form

ance

ove

rth

elo

ng

term

.How

ever

,pr

oposa

lsar

ea

seco

ndar

yta

ctic

tosim

ply

selli

ngst

ock

.

Why

prop

osa

ls?&N

oth

ing

rive

tsth

eat

tention

ofth

eco

rpor

ate

min

dlik

ea

pro

posa

l'.C

urre

ntly

use

only

ifin

itia

ldi

scuss

ions

with

man

agem

entunpr

oduc-

tive

.Pro

xyco

ntes

tsar

eto

oex

pensive

.

Chi

e#y

beca

use

ofth

elo

wco

st.A

lrea

dy

ow

nta

rget

stock

sdu

eto

index

ing.

Do

notw

ish

toex

pend

grea

tre

sourc

eson

som

ethin

gth

atdoe

snot

initia

tein

vest

men

tdec

isio

ns.

Subm

itting

pro

pos

als

star

tsdi

alog

ue,ha

sim

pact

beca

use

show

spre

par

edto

take

the

issu

eto

ash

areh

olde

rvo

te.

Pro

xy"gh

tsar

eal

mos

tal

way

sto

oex

pen

sive

.

Pre

fer

priva

tediscu

ssio

nsw

ith

man

agem

ent"rs

t,but&u

sepro

pos

als

ifw

eth

ink

we

nee

dto'.

Pro

xy"gh

tsan

dta

keove

rsar

eto

oco

stly

and

wou

ldra

ise

the

spec

ter

ofa

pol

itic

alba

ckla

sh.

To

get

additio

nal

shar

ehold

erin

puton

the

issu

e,w

hic

hm

ayin#ue

nce

resist

ant

man

agem

entto

mak

ere

ques

ted

chan

ge.U

sed

onl

yif

man

agem

ent

will

not

mak

ech

ange

saf

ter

priva

teco

mm

uni

cation.

Wha

tis

succ

ess?

Ifth

epro

pos

alpas

ses;

ifit

resu

lts

insign

i"ca

ntch

ange

sin

the

com

pan

y(str

ateg

icpl

an,to

pm

anag

emen

t,vi

sible

atte

mpt

sto

incr

ease

shar

ehol

der

valu

e.)&T

heto

pic

itse

lfis

notth

ere

alissu

e'.

Ifre

ceiv

eat

leas

t30

%ofth

evo

te.If

in-d

epth

discu

ssio

nw

ith

man

agem

enton

per

form

ance

and

stra

tegy

resu

lts

inag

reem

enton

futu

resign

sofpro

gres

sor

com

plet

ion.

Whe

nm

anag

emen

tad

opts

the

requ

este

dch

ange

.

Whe

nm

anag

emen

tis

resp

onsive

toim

pro

ving

the

shar

ehol

der

valu

efo

cus

ofth

eco

mpan

y.Eve

ntu

ally

,when

perfor

man

ceim

pro

ves.

Inso

me

case

s,w

hen

the

reques

ted

chan

geis

mad

e.

Whe

nth

esp

eci"

cch

ange

isad

opt

ed,an

dth

ego

vern

ance

stru

cture

isim

pro

ved.

Bel

ieve

tran

slat

esin

tohi

gher

retu

rns

ove

rth

elo

ng

term

.


Tab

le3.

Continue

d.

CA

LP

ER

SC

ALSTR

SC

REF

NE

WY

OR

KC

ITY

SW

IBR

icha

rdK

oppe

s,fo

rmer

Chi

efC

ouns

elC

EO

,Jan

ice

Hes

ter-

Am

ey,

Cor

por

ate

A!ai

rsA

dvisor

Pet

erC

lapm

an,C

hie

fC

ounse

l,In

vest

men

tsJo

nLuko

mnik

,D

eputy

Com

ptro

ller

Kurt

Sch

acht,

Chi

efC

ounse

l

Wha

tac

tion

take

nif

not

succ

essful?

Rep

eat

the

pro

posa

lth

efo

llow

ing

year

.K

eep

com

ing

back

.

Rep

eat

proposa

lsfo

ran

oth

ertw

oye

ars.

Rep

eat

the

pro

posa

lth

efo

llow

ing

year

.Sel

ldec

isio

nunr

elat

edto

pro

posa

l'ssu

cces

s,so

lely

rela

ted

tost

ock

fundam

enta

ls.

Rep

eat

the

pro

posa

lth

efo

llow

ing

year

.C

ons

ider

esca

lating

tact

ics,

such

aspu

blic

ity.

Giv

eup

ifse

ems

count

erpr

oduc

tive

.

Rep

eat

prop

osa

lor

bec

ome

pas

sive

ifse

ems

succ

ess

isve

ryun

likel

y.M

ayse

llst

ock

ifst

ock

fundam

enta

lsdi

ctat

e.

Tar

get

sele

ctio

n:In

itia

lly,ta

rget

ing

was

base

don

the

gove

rnan

ceissu

e.P

rim

arily

per

form

ance

bas

edaf

ter

1989

,ch

oosing

from

the

bot

tom

50pe

rform

ers

inth

eir

por

tfol

io.A

lso

cons

ider

size

ofow

nst

ake

and

spec

i"c

gove

rnan

ceissu

e.

Inea

rly

year

s,pure

lyby

the

gove

rnan

ceissu

e,pic

kin

gpoo

rgo

vern

ance

pra

ctic

es.Per

form

ance

bas

edaf

ter

1989

.Also

consider

gove

rnan

cest

ruct

ure

,in

sider

and

inst

itutional

ow

ners

hip

,an

dsize

ofow

nst

ake.

Bas

edon

the

spec

i"c

issu

e.If

feel

asp

eci"

cissu

eis

import

ant,

look

thro

ugh

thei

rhold

ings

tose

ew

hich

com

panie

shav

eth

atpr

oble

m.A

lso

look

atsize

ofst

ake

and

inst

itutional

ow

ners

hip

.

Origi

nal

lybas

edon

the

spec

i"c

gove

rnan

ceissu

e.Lat

erin

clud

edpe

r-fo

rman

ceco

nsid

erat

ions

.

Cho

ose

targ

ets

from

the

&wor

stof

the

wors

t'in

the

por

tfol

io,t

hen

look

for

gove

rnan

ceissu

eto

targ

et.O

nly

targ

etif

pot

ential

ups

ide

and

stru

ctur

alissu

eto"x.

Vie

won

public

ity:

Publ

icity

can

be

anim

men

sely

pow

erfu

lto

ol

toge

tle

vera

gew

ith

man

agem

ent,

espe

cial

lyif

egre

giou

sgo

vern

ance

pro

ble

ms

exist.

Pre

ssat

tention

togo

vern

ance

has

help

edpro

mote

chan

ge.

Public

ity

isth

egr

eate

stso

licitat

ion

and

educ

atio

nto

olav

aila

ble,

and

islo

wco

st.M

ore

awar

enes

sof

the

fund

iscr

eate

d,b

utfu

nd

o$

cial

sdo

not

bene"tper

sonal

lyfrom

public

ity.

Pre

fer

not

tous

epu

bli-

city

,bu

tw

illbring

the

issu

eto

avo

teif

man

agem

entw

on'

tad

opt

reques

ted

chan

ge.

Use

dpub

licity

bef

ore

gain

edre

puta

tion,

but

now

star

tw

ith

priva

tediscu

ssio

ns

and

only

use

public

ity

when

nec

essa

ry(a

ses

cala

ting

tact

ic).

&Goal

isto

e!ec

tch

ange

,not

toge

tat

tention'.

Pre

fer

not

tous

epub

licity;

unco

mfo

rtab

lew

ith

havi

ng

ahi

gher

pro"le

.P

ubl

icity

isan

oth

erle

vera

geon

man

agem

ent,

but

isof

ten

unn

eces

sary

.O

bje

ctiv

eis

toim

pro

vepe

rfor

man

ce,n

otto

get

atte

ntion

forin

div

idua

ls.


are targets of activism. Given that CREF is largely indexed but also devotesa signi"cant dollar amount to active management, it is unclear whether theymight behave more like the indexed funds or more like SWIB. In order to makethis assessment, we must consider not only the results on activism and publicitystrategies analyzed here, but also CREF's trading strategies analyzed in a latersection.

The funds' stated views on publicity and de"nitions of targeting success aregenerally consistent with our earlier predictions. The heaviest users of indexingand outside managers (CalPERS, CalSTRS, and NYC) feel strongly that public-ity is a useful and e!ective activism tool. Their stated goals include changing thegovernance environment and encouraging a shareholder focus in portfoliocompanies. They feel their campaign is successful if it results in change (ratherbroadly de"ned) at the company. For these funds, the topic of the proposal itselfis not as important as generating a response from management, possibly in theform of a change from previous strategies or company direction. In contrast, thefunds with investment strategies conducive to coordinating activism and tradinge!orts, SWIB and CREF, state that their immediate goal is to have the speci"cmeasure in the proposals adopted. Both state a belief in a link between gover-nance structure and individual stock price performance. Both are very reluctantto use publicity, and do so only as a necessary last resort to make credible theirthreat to take the issue to shareholders. SWIB and CREF's avoidance ofpublicity and their narrow, "rm-speci"c goals for activism seem consistent withthe value-maximizing strategies outlined in the formal models discussed earlier.SWIB also states that they often choose to sell their stake rather than intervene,increasing the uncertainty as to whether they will monitor a given "rm. As themodels predict, this improves their ability to generate trading pro"ts from theirmonitoring.

Di!erences in investment strategies and activism goals will likely carry overinto proposal topic selection as well. Speci"cally, we might expect that fundswith narrow activism goals, such as the internally and actively managed funds,would sponsor proposals that require greater "rm-speci"c knowledge. Con-versely, funds with broad, marketwide goals might tend to sponsor proposalsthat are more generically good for shareholders. Data from Table 1 on proposaltopics broken down by sponsor generally support these predictions. Forexample, 91% of SWIB's proposals are on the topic of poison pills, while 70% ofNYC's proposals request con"dential voting. While most agree that con"dentialvoting is good for shareholders since it mitigates management's in#uence onvoting outcome, the bene"ts of poison pills are more "rm-speci"c. For example,Brickley et al. (1994) "nd that poison pills announced by companies witha majority of independent outsiders on their board generate signi"cant positiveabnormal returns, while those announced by companies without such a major-ity generate signi"cant negative abnormal returns. Consistent with this reason-ing, John Lukomnik of the heavily indexed NYC fund reports that they do not


sponsor poison-pill proposals because they &require too much company-speci"cknowledge'. In contrast, Kurt Schacht at the actively managed SWIB tells usthat &SWIB has always had a focus on antitakeover issues'. He stresses that theirchoice of targets is based on detailed knowledge of the companies, and thatproposals are a secondary option to selling the stock.1

The funds also di!er in whether they tend to choose poorly performing "rmsas targets. Based on the prior "ve-year buy-and-hold stock returns, CalPERS,NYC, and SWIB tend to do performance-based targeting, while CREF andCalSTRS do not. Speci"cally, the median "ve-year stock returns for targets, inexcess of the S&P 500, are !97% for CalPERS, !87% for NYC, !72% forSWIB, 4% for CREF, and 19% for CalSTRS. Thus, whether or not a fund doesperformance-based targetings does not seem to be related to its investmentstrategy. Even though we might expect both types of funds to choose poorperformers, there are plausible reasons why they might focus on other criteria.We might expect actively managed funds to target poor performers since thoseare the "rms with room for improvement and hence higher trading pro"ts. Thefund might also intervene, however, because they know that the "rm is about todo worse. We also might expect funds with marketwide goals to choose poorperformers since those are more likely to attract press attention or have a morereceptive target management. However, they may also target a "rm with goodpast performance but with a governance structure that could lead to futureproblems. For example, despite strong past performance, Disney was recentlycriticized for lacking independent directors that are not beholden to their CEO,Michael Eisner. Consistent with this reasoning, Hermalin and Weisbach (1998)in a formal model of board monitoring show that CEOs with substantialbargaining power, such as those with an exceptional performance record, areless likely to be scrutinized or disciplined by the board or to increase the numberof independent directors.

Overall, activist pension funds show heterogeneity in their objectives forproposal activity, their propensities to use publicity, and their de"nitions oftargeting success, each of which is sensibly related to their investment strategies.For example, the externally managed index funds that bene"t most fromactivism strategies that &raise all boats' state broader goals of corporate change.In contrast, the internally managed funds that have the #exibility to buy and selltarget stocks have the more narrow goal of having the target "rm make thespeci"c change requested in the proposal. Despite their apparent similarities,large non-corporate pension funds are not a homogeneous set of investors.

1One exception to the predictions outlined here is CalSTRS' choice of proposal topics. Given thatthey are heavily indexed (with their active portion externally managed), we might expect them toprimarily choose proposal topics that do not require "rm-speci"c knowledge. However, 79% of theirproposals are on antitakeover issues.


4. Measuring the impact of shareholder proposals on target 5rms

As we turn to the empirical analysis, one issue is how to determine whethera proposal is successful or e!ective. Some possible de"nitions are whether theproposal receives more than 50% of shareholder votes, whether managementtakes the action requested in the proposal, or whether the proposal generatesstatistically signi"cant abnormal returns at the announcement date. Thesede"nitions are similar to those proposed by SWIB and CREF as well as thoseused in the activism literature. For example, Karpo! et al. (1996), Wahal (1996),and Gillan and Starks (1998) examine stock returns around announcementdates of shareholder proposals and generally "nd no signi"cant abnormalreturns. However, as we discuss in Section 5, substantial event date uncertaintyand other problems can erode the ability of an event study to detect the impactof proposals.

An alternative measure of the impact of proposals is whether the managers oftarget "rms make signi"cant changes in corporate policies or whether proposalsappear to mobilize support for further governance activity such as takeovers.This is the explicit objective of CalPERS, NYC, and CalSTRS. Similar issueshave been examined for proxy contests. Dodd and Warner (1983) "nd thatinsurgents win control of the board in only 25% of proxy contests, leadingShleifer and Vishny 1986, p. 472, to conclude that these governance mechanismsare &not e!ective'. However, DeAngelo and DeAngelo (1989) examine changes atcompanies following proxy contests, and "nd extensive turnover, asset sales,mergers, liquidations, and other major changes, even in companies at whichproxy contests are deemed unsuccessful.

4.1. Analysis of changes in corporate policies following shareholder proposals

In this section we provide evidence that pension fund activism has a signi"-cant impact on target company business policies, organization, and governance.To measure the impact of proposal activity, we examine data on changes inmanagement policy and signi"cant events in the life of a corporation andcompare these to data for a control sample with similar performance, size, andindustry characteristics. We present analysis for the sample as a whole, as well asof subsamples of proposals by fund sponsor, topic, and voting outcome. Weshow that the heterogeneous objectives and tactics of the funds are re#ectedin target management's response to a shareholder proposal, and that evenproposals that do not receive a majority of votes are e!ective in promotingchange.

We examine data collected from The Wall Street Journal Index on news ofsigni"cant corporate events, either control-related (e.g., hostile tender o!ers) orthose that might indicate managerial responses or signi"cant corporate change(e.g., restructurings, turnover, changes in payout policy, employee layo!s). We


collect news starting with the "rst announcement of a proposal and continuingfor a total of four years, event years 0 to #3.

We collect news event data over the same time period for a control samplematched to the targets on size, industry, and performance in the "scal year-endprior to the "rst targeting announcement. We match on accounting perfor-mance, de"ned as operating income over assets, because poor performers aremore likely to experience control changes and corporate restructuring relatedactivities. We match on market capitalization and two-digit SIC industry codesbecause our test relies on corporate events being announced in The Wall StreetJournal, and we expect that similar size "rms in the same industries are likely toget similar coverage. The median operating income/assets ratio is 0.152 for bothtarget and control samples; the median market capitalization is $2816.7 millionfor targets and $2257.1 for control "rms. Of the 125 "rms receiving proposals, 80could be matched with control "rms. Most of the target companies for whichmatches are not found are in industries such as retailing and defense that aredominated by a few major companies, almost all of which receive shareholderproposals during the period under study.

We test whether the target sample experiences a greater frequency of an-nounced events than the control sample, employing the nonparametricMann}Whitney U test for signi"cance. (The Mann}Whitney test is equivalent tothe Wilcoxon rank sum test but allows for di!erent sample sizes in the distribu-tions being compared.) Since we are testing for a greater frequency of announcedevents in the target sample, we conduct a one-tailed test. To check whether theresults are driven by those targets without matching control "rms, we conductseparate tests for the sample as a whole and for only those targets with matchesin the control sample.2

Table 4 shows the frequency of several categories of corporate events for (1)our entire target sample, (2) those targets with matches, and (3) the controlsample. Since the sample sizes di!er across these three groups, we report thefrequency of events on a per-"rm basis. For example, there are 0.4 announce-ments per "rm of CEO turnover in the target sample in the four years followingthe "rst announcement of a proposal, or equivalently 0.1 per year. Thus, 10% oftarget "rms per year experience CEO turnover on average, versus 7.5% ofcontrol "rms per year. CEO turnover in the target sample is not statisticallydi!erent from the control sample. Karpo! et al. (1996) and Smith (1996) also "ndCEO turnover to be unrelated to the previous submission of a shareholder

2Additionally, we collect news up through three years from the xnal targeting for those companiestargeted multiple times and repeat our tests using this longer period, with very little change inresults. We also repeat the tests de"ning year 0 as the year of the last proposal. There are fewerdi!erences and lower levels of signi"cance. Thus, "rst-time proposals appear to have the greatestimpact, and the impact of some proposals can take several years to develop.


Table 4Frequency of announced events per "rm in the four years after the "rst targeting

This table contains the post-targeting frequency of events per "rm of several categories of corporateevents announced in the Wall Street Journal. Frequencies per "rm are given for our sample of "rmsthat have received one or more shareholder proposals (target "rms), and for a control samplematched by size, two-digit industry code, and accounting performance. In 45 cases we could not "nda match that "t our stated criteria, so we also include the frequencies for the subsample of target"rms that have matching control "rms.The numbers in parentheses are the test statistics of thenonparametric Mann}Whitney U test. A negative and signi"cant test statistic indicates that thedistribution of news events in the target sample is signi"cantly to the right of the control sample(*,**, and *** indicate statistical signi"cance at the 10%, 5%, and 1% levels for a one-tailed test)

News categories: Frequency of announced events per "rm

All targets(N"125)

Targets withmatches (N"80)

Control "rms(N"80)

Turnover 1.22*** 0.91*** 0.56(!3.09) (!2.36)

CEO turnover 0.40 0.38 0.29(!0.87) (!0.66)

Other turnover 0.82*** 0.54** 0.28(!2.81) (!1.90)

Hostile control attempt 0.22 0.20 0.06(!1.23) (!0.99)

Hostile control change 0.08 0.09 0.01(!0.97) (!0.96)

Any control attempt 0.38 0.38 0.14(!1.24) (!1.16)

Any control change 0.10 0.13 0.08(!0.35) (!0.55)

Governance events 0.52*** 0.48** 0.14(!2.37) (!2.09)

Management response 3.46*** 3.38*** 2.33(!2.90) (!2.74)

Payout Increase 2.26 2.35 2.40(0.68) (0.31)

News category de"nitions:Turnover: CEO retires, CEO resigns unscheduled, director resigns, or other top executive resigns.CEO turnover: CEO retires or CEO resigns unscheduled.Other turnover: Director resigns, or other top executive resigns.Hostile control attempt: a proxy contest attempt or hostile takeover bid.Hostile control change: Win seat(s) in a proxy contest, hostile bid successful, or management buyout.Any control attempt: a proxy contest attempt, hostile takeover bid, or merger talks.Any control change: Win seat(s) in a proxy contest, hostile bid successful, management buyout, or

merger completed.Governance events: Shareholder lawsuit, a non-pension-fund-sponsored shareholder proposal, or

public &no' vote for directors.Management response: asset sales or spino!, restructuring/reorganization, or employee layo!s.Payout increase: increase in dividends, repurchase shares.


Tab

le5

Sum

mar

yof

the

resu

lts

ofth

eM

ann}W

hitn

eyra

nksu

mte

stac

ross

variou

ssu

bsam

ples

ofta

rget"rm

s(e

vents

occ

urring

within

thre

eye

ars

ofth

e"rs

tta

rget

ing)

This

tabl

eco

ntai

nsth

ete

stre

sultsofc

om

par

ing

thepos

t-ta

rget

ing

freq

uency

ofse

vera

lcat

egories

ofc

orpor

ateev

ents

anno

unce

din

the

Wal

lStr

eetJ

ourn

alin

targ

et"rm

sre

lative

toa

contr

ol

sam

ple

mat

ched

bysize

,tw

o-d

igit

indu

stry

code,

and

acco

unting

per

form

ance

.W

eem

plo

yth

enon

par

amet

ric

Man

n}W

hitney;

-tes

t.A

ppen

dix

Aco

ntai

nsdi

sagg

rega

ted

resu

ltsofi

ndiv

idual

corp

ora

teev

entan

nou

nce

men

ts.S

tatist

ical

sign

i"ca

nce

indic

ates

that

the

distr

ibution

ofa

nno

unc

edev

ents

inth

eta

rget

sam

ple

issign

i"ca

ntly

toth

erigh

toft

he

cont

rols

ample

(*,*

*,an

d**

*in

dic

ate

stat

istica

lsig

ni"ca

nce

atth

e10

%,5

%,a

nd1%

leve

lsfo

ra

one-

taile

dte

st).

Pro

posa

lsin

clud

edun

derth

epro

pos

alty

pesu

bsa

mple

s(v

otin

g,an

tita

keove

r,an

dboa

rdissu

es)a

relis

ted

inTab

le1.

An

obs

erva

tion

ispu

tin

them

ultip

le-t

ypeca

tego

ryif

the"rm

rece

ives

multip

lepr

oposa

lsin

di!

eren

tpr

oposa

lcat

egor

ies.

An

obse

rvat

ion

isputin

the19

87to

1989

or19

90to

1993

subsa

mpl

esif

allp

ropos

alsat

that"rm

are

inth

ose

year

s.O

ther

wise,

they

are

putin

ath

ird

dat

eca

tego

ry(n

otre

por

ted).

Outc

om

ere

fers

toth

evo

ting

outc

om

eof

the

pro

pos

al

New

sca

tego

ries

Fund

spons

or

Pro

pos

alty

pe

Dat

eO

utco

me

SW

IBC

REF

CA

LP

ER

SC

ALSTR

SN

YC

Vot

ing

Anti-

takeo

ver

Boar

dissu

esM

ultip

lety

pe

1987

to 1989

1990

to 1993

Pas

sed

With-

dra

wn

Fai

led

Turn

over

***

****

***

****

**

****

CEO

turn

over

***

Oth

ertu

rnov

er**

***

**

***

***

***

**


Host

ileco

ntr

olat

tem

pt*

Host

ileco

ntr

olch

ange

Any

cont

rolat

tem

pt

**

Any

cont

rolch

ange

*G

ove

rnan

ceev

ents

***

**

***

***

**M

anag

emen

tre

spons

e**

**

***

****

**

****

Pay

out

incr

ease

**

New

sca

tego

ryde"ni

tions:

Turn

ove

r:C

EO

retire

s,C

EO

resign

sunsc

hed

ule

d,di

rect

or

resign

s,or

other

top

exec

utive

resign

s.C

EO

turn

ove

r:C

EO

retire

sor

CE

Ore

sign

sun

sched

ule

d.O

ther

turn

over

:D

irec

tor

resign

s,or

other

top

exec

utiv

ere

sign

s.H

ost

ileco

ntro

lat

tem

pt:

apro

xyco

nte

stat

tem

ptor

host

ile

take

over

bid

.H

ost

ileco

ntro

lch

ange

:W

inse

at(s)in

apro

xyco

nte

st,h

ost

ilebi

dsu

cces

sful

,or

man

agem

ent

buyo

ut.

Any

contr

olat

tem

pt:

apr

oxy

cont

est

atte

mpt,

host

ileta

keove

rbid

,or

mer

ger

talk

s.A

ny

contr

olch

ange

:W

inse

at(s)in

apr

oxy

cont

est,

host

ilebi

dsu

cces

sful,

man

agem

entbuy

out,

or

mer

ger

com

plet

ed.

Gove

rnan

ceev

ents

:Sh

areh

old

erla

wsu

it,a

non-

pen

sion-

fund-s

pon

sore

dsh

areh

old

erpro

pos

al,or

public&n

o'vo

tefo

rdirec

tors

.M

anag

emen

tre

sponse

:as

set

sale

sor

spin

o!,r

estr

uct

urin

g/re

orga

niza

tion

,or

empl

oyee

layo!s.

Pay

out

incr

ease

:in

crea

sein

div

iden

ds,re

purc

has

esh

ares

.


proposal in their samples. There are, however, other notable events that occurwith signi"cantly greater frequency in the overall target sample.

The target sample experiences greater turnover in top management (CEO orother senior executive, typically a CFO or president of a corporate unit), morenoncontrol governance events (shareholder suits, non-sample proposals orletters, and &no' votes for directors), and greater management response (assetsales, restructuring, reorganizations, and layo!s) than the control sample, at the1% signi"cance level. For example, the target sample has 0.52 noncontrolgovernance events on average in the four years after the "rst targeting. Incontrast, the performance-, industry-, and size-matched control sample experi-ences only 0.14 such events. The target sample experiences 1.22 senior manage-ment turnovers and 3.46 management response events, compared to 0.56 and2.33 for the control sample.

Table 5 summarizes signi"cant di!erences in the frequency of announcedevents between subsamples of target "rms and their matched comparison "rms.We conduct the Mann}Whitney tests comparing the frequency of news eventsfor subsamples of fund sponsor, proposal type, calendar date, and votingoutcome. Among all the fund sponsors, CalPERS is the one most associatedwith subsequent changes at target "rms. The subsample of CalPERS targets hassigni"cant di!erences from the control sample in "ve of ten news categories,including turnover, control attempts, noncontrol governance events, and man-agement response. CalPERS is the only proposal sponsor associated witha higher frequency of announced external control attempts. Appendix A supple-ments Table 5 by showing a "ner breakdown of news events and showsthat CalPERS targets experience signi"cantly more voluntary internal changes,in addition to events associated with external monitoring or control threats.For example, internal changes with a higher frequency of announcementsinclude senior executive turnover, asset sales, voluntary restructurings, andlayo!s.

As evidence of heightened external monitoring, CalPERS targets havea greater frequency of shareholder lawsuits, block purchases, and hostile take-over defenses. In analysis conducted for robustness checks, we "nd thatCalPERS also has the most immediate e!ect of all the funds, with manysigni"cant di!erences occurring in as few as two years after the "rst news of theproposal (year 0 to year #1). Repeating the tests for the subsamples ofCalPERS targets that do and do not subsequently experience control activityprovides an interesting insight. Only in the noncontrol activity subsample isthere evidence of voluntary management response. For example, this is the onlysubsample to have a signi"cantly greater frequency of voluntary restructuringsand reorganizations, whereas hostile takeover defenses were only signi"cant inthe control activity subsample. This suggests that poorly performing "rms thatdo not respond to shareholders such as CalPERS by making signi"cant volunt-ary changes are more likely to become takeover candidates.


NYC and CalSTRS are the only other sponsors whose targets are associatedwith signi"cantly greater changes, albeit for fewer event categories and withlower statistical signi"cance. The three funds who in their interviews indicatedbroader &company response' goals for their activism appear to be successful ineliciting a response from corporate management. It is not surprising thatCalPERS is the most successful among these three since they have the largestownership stakes, and arguably the highest pro"le. (According to the January23, 1995 issue of Pensions and Investments, CalPERS had $78.5 billion in assets,CalSTRS $48.5 billion, and NYC $35.3 billion at the end of 1994.) Perhaps theclout of the size of their stake, or the credible threat of publicity, makesmanagers more responsive to them.

At the other extreme, SWIB and CREF do not seem to generate signi"cantactivity in their target "rms. These funds' main objective is for the target "rm tomake the speci"c change requested in the proposal. Wahal (1996) reports that43% of SWIB's requests are adopted by corporate management. Carleton et al.(1998) report that &CREF was able to convince the "rms to enact the changes itdesired in 69 of 72 "rms (95.8%) during our 1987}1996 sample period'. Notethat in some cases it takes several years of targeting for CREF's target com-panies to make the requested changes. Although they are generally successful bytheir own measure, they do not seem to generate a broader response fromcorporate management. One possibility is that corporate managers are lessreceptive because they view these funds as less permanent shareholders than thethree externally managed index funds, an idea explored later in this study.

Table 5 also shows results by proposal type. In this analysis, a target is ina proposal category if all proposals received by a "rm are related to that samecategory. If a target, for example, receives both voting and antitakeover propo-sals, they would be in the multiple type category. We expect voting-relatedproposals to have the least impact as the subject matter is relatively benign, andthose relating to the board of directors to have the greatest impact as theyembody more direct intervention by shareholders in governance and manage-ment processes. The results are consistent with these predictions. Speci"cally,"rms targeted with board-related proposals have signi"cantly greater an-nouncements of CEO turnover, non-CEO turnover, management response, andnoncontrol governance events than the control sample, while voting proposalsare only associated with greater non-CEO turnover at the 10% level. Targets ofboard-related proposals are the only subsample to have signi"cantly greaterCEO turnover. Antitakeover proposal targets are the only proposal categoryassociated with greater announced control change attempts.

In a comparison of events in target "rms by voting outcome, Table 5 showsthat even proposals that fail to get a majority of voting support are associatedwith signi"cant changes at target "rms. Speci"cally, failed proposals are asso-ciated with higher turnover and management response, such as layo!s. Target"rms in which a shareholder proposal is withdrawn by the sponsor, usually in


Tab

le6

Furt

her

test

son

annou

nced

even

tsper"rm

This

tabl

eco

ntai

ns

the

freq

uenc

yofev

ents

per"rm

inth

esa

me

cate

gories

ofco

rpora

teev

ents

asin

Tab

les

4an

d5.

The"rs

ttw

oco

lum

ns

repl

icat

eth

ere

sults

inT

able

4fo

rea

seofco

mpar

ison.T

he

next

two

colu

mns

ofre

sults

are

for

a61

-"rm

subs

ampl

eofth

eor

igin

al12

5-"rm

sam

ple.

This

subsa

mpl

ere

mov

esth

e25

%of"rm

sw

ith

cont

rol"rm

sco

nside

red&p

oor'm

atch

eson

pre

viou

s"ve

-yea

rbuy-

and-

hol

dst

ock

retu

rns.

The

Pea

rson

corr

elat

ion

bet

wee

nst

ock

retu

rnsoft

heta

rget

san

dco

ntr

ols

inth

issu

bsam

ple

is0.

78,a

nd

thedistr

ibutions

ofr

etur

nsar

eve

rysim

ilar.

Thela

stfo

urco

lum

nsof

resu

lts

com

pare

the

freq

uenc

yofc

orpo

rate

even

tsin

the

fourye

arsaf

terta

rget

ing

toth

efo

urye

arsbe

fore

targ

etin

gfo

ra

69-"

rmsu

bsam

ple

oft

he

origi

nal

targ

etsa

mple

of12

5"rm

s.F

irm

sar

ese

lect

edfo

rth

issu

bsa

mpl

ean

alys

isif

the"rm

nam

est

arts

with

the

lett

ers

Bth

roug

hM

.The

last

two

colu

mns

ofre

sults

com

pare

thefreq

uen

cyofc

orpora

teev

ents

inth

efo

urye

arsaf

terta

rget

ing

toth

efo

urye

arsbef

ore

targ

etin

gfo

ra

44-"

rmsu

bsa

mpl

eoft

heor

igin

alco

ntr

ol

sam

ple

of8

0"rm

s.Firm

sar

ese

lect

edfo

rth

issu

bsa

mpl

ean

alys

isif

they

are

aco

ntr

ol"

rmfo

ra

targ

et"rm

with

anam

est

arting

with

the

letter

sB

thro

ugh

M.A

sin

Tab

le4,

the

num

bers

inpar

enth

eses

are

the

test

stat

istics

ofth

eno

npar

amet

ric

Man

n}W

hitne

y;

-tes

t.A

nega

tive

and

sign

i"ca

ntte

stst

atistic

indic

ates

that

the

dist

ribu

tion

ofne

wsev

ents

issign

i"ca

ntly

toth

erigh

toft

heco

rres

pond

ing

contr

ols

ample

(*,*

*,an

d**

*in

dica

test

atistica

lsig

ni"

cance

atth

e10

%,5%

,and

1%le

vels

for

aone-

taile

dte

st)

New

sca

tego

ries

:Tab

le4

resu

lts

com

paring

targ

ets

that

hav

em

atch

ing

contr

ols

Clo

sem

atch

eson

pre

viou

s"ve

-yea

rbuy

-an

d-ho

ldst

ock

retu

rn

Tar

get

sam

ple

Con

trolsa

mple

Tar

gets

(N"

80)

Con

trols

(N"

80)

Tar

gets

(N"

61)

Con

trols

(N"

61)

After

targ

etin

g(N

"69

)

Bef

ore

targ

etin

g(N

"69

)

After

targ

etin

g(N

"44

)

Bef

ore

targ

etin

g(N

"44

)

Turn

over

0.91

***

0.56

1.03

***

0.59

1.26

1.12

0.57

0.55

(!2.

36)

(!2.

56)

(0.2

6)(0

.42)

CEO

turn

ove

r0.

380.

290.

460.

300.

350.

300.

340.

27(!

0.66

)(!

1.23

)(!

0.33

)(!

0.55

)

Oth

ertu

rnov

er0.

54**

0.28

0.57

**0.

300.

910.

810.

230.

27(!

1.90

)(!

1.84

)(0

.27)

(0.6

9)

Host

ileco

ntro

lat

tem

pt0.

200.

060.

180.

050.

25*

0.04

0.07

0.07

(!0.

99)

(!0.

80)

(!1.

50)

(!0.

17)


Host

ileco

ntro

lch

ange

0.09

0.01

0.08

0.02

0.10

0.00

0.00

0.00

(!0.

96)

(!0.

62)

(!1.

18)

(0)

Any

contr

olat

tem

pt

0.38

0.14

0.39

0.15

0.38

**0.

100.

140.

16(!

1.16

)(!

0.99

)(!

1.64

)(0

.18)

Any

contr

olch

ange

0.13

0.08

0.13

0.07

0.15

*0.

000.

050.

00(!

0.55

)(!

0.62

)(!

1.47

)(!

0.37

)

Gove

rnan

ceev

ents

0.48

**0.

140.

48*

0.13

0.46

**0.

230.

110.

11(!

2.09

)(!

1.63

)(!

1.72

)(!

0.18

)

Man

agem

ent

resp

onse

3.38

***

2.33

3.56

**2.

483.

35**

2.36

2.05

*1.

30(!

2.74

)(!

2.28

)(!

1.93

)(!

1.47

)

Pay

out

incr

ease

2.35

2.40

2.20

2.30

2.35

2.80

2.66

2.39

(0.3

1)(0

.44)

(1.2

1)(!

0.68

)


exchange for concessions by the "rm's management, also have signi"cantlygreater governance and management response events, the latter primarilyannounced asset sales. However, high shareholder support does seem to berelated to control activity, the most dramatic form of corporate change. Appen-dix A shows that passed proposals are the only outcome category associatedwith greater announcements of control activity such as proxy contests, hostiledefenses, merger talks, and block purchases.

4.2. Further tests on announced xrm activities associated with pension fundproposals

The results in Tables 4 and 5 indicate that there is a signi"cantly higherfrequency of corporate activity in "rms targeted by our sample pension fundsrelative to a control sample matched on two-digit industry codes, marketcapitalization, and accounting performance. In this section, we investigatewhether these results can reasonably be linked to intervention by pension fundsrather than our failure to control for a relevant variable. For example, if pensionfunds choose targets that are likely to engage in the organizational changes wemeasure, we might mistakenly attribute such changes to the targeting activity.We conduct two sets of robustness checks on our results. First, we examinea subsample of target "rms for which the previous "ve-year buy-and-hold stockreturns match closely with their size- and industry-matched control "rm. Thisaddresses the concern that the results are driven by poor stock performers thata priori are more likely to experience events such as control activity andmanagement changes. Second, we compare the frequency of events before andafter targeting for a subsample of targets and matching control "rms. Thisaddresses the concern that our results are driven by di!erences in reporting byThe Wall Street Journal for the target and control samples, or alternatively, thatpension funds simply target &high-activity' "rms.

Table 6 contains the results of the additional tests, using the same format andvariables as Table 4. For ease of comparison, the full sample results of Table 4are repeated here. For the "rst test, we repeat the analysis on the 75% ofmatching control "rms that also match closely on previous "ve-year stockreturns. (The results are also robust in a subsample analyzing the 50% of thesample with the closest match in stock returns.) Within this subsample of 61"rms, the distributions of stock returns for targets and controls are very similar,and the correlation between the returns of targets and controls is 0.78. This testindicates that our results are not likely to be driven by poorer stock performanceat target "rms. The results are quantitatively similar to the full sample, withsimilar levels of statistical signi"cance. Furthermore, the subsample analysis bysponsor is also very similar to the results in Table 5.

For the second test, we compare the frequency of announcements in the fouryears before targeting to the four years after targeting for a subsample of targets


and for control "rms. This is similar to the method used by Huson (1997) toanalyze CalPERS' targets. To select a representative sample from our original125 targets, we conduct this subsample analysis on a total of 69 target "rms withnames beginning with the letters B through M. Of these 69 "rms, 44 havematching control "rms. The last four columns of Table 6 show that the onlysigni"cant increase in activity for the control sample from the period before toafter targeting is in the management response category (asset sales, restructur-ing, reorganizations, and layo!s). In contrast, the target sample experiencessigni"cantly higher frequencies of activities in "ve of the ten categories. Interest-ingly, the only event that is signi"cantly more frequent in the period beforetargeting is the adoption of takeover defenses. This is true for both the targetand control sample, although the frequency is higher for target "rms.

Management turnover is not signi"cantly di!erent in the periods before andafter targeting, but the governance events and management response results arerobust to this additional test. Despite similar levels of corporate control activityin the before period for the target and matching control "rms, we "nd a signi"-cant increase in control activity for the target sample only. Analysis by pensionfund sponsor indicates that the results in Table 5 are also generally robust, withthe exception of those for management turnover. CalPERS continues to be themost e!ective sponsor by this measure, and the only sponsor associated withincreased corporate control activity. SWIB and CREF, the sponsors with thenarrow goal of adoption of the proposal's speci"c request, continue to beassociated with no signi"cant company response in any category of event.

In sum, we "nd that shareholder proposals are associated with signi"cantchanges at target companies in that they are followed by both governance eventsand corporate policy changes. The impact of proposals is heterogeneous, but isconsistent with the objectives of the funds. Speci"cally, the three funds whode"ne success as an ability to promote change at portfolio companies appear tobe successful by this measure. The pattern of events following proposals appearsto be logically related to the proposal topic (e.g., takeover attempts followantitakeover-related proposals), but less related to the amount of shareholdersupport received. The evidence also suggests that a proposal's impact might bethrough its inclusion in a series of reinforcing corporate control events, a resultwe explore further below.

4.3. Further analysis of post-targeting control activity in target xrms

One view of the role of shareholder proposals is that they serve to mobilizesupport for more costly governance activity such as takeovers. This view isexpressed by former SEC commissioner Sommer (1992):

Moreover, a substantial vote in favor of a shareholder proposal opposed bymanagement might be seen as a vote of &no con"dence' in management and


stir the juices of would-be &raiders', either proxy or takeover variety. Indeed,some managements have expressed the notion that a twenty percent favor-able vote on a shareholder proposal is extremely disquieting to management.

In this section we look more closely at the hypothesis that shareholderproposals are precursors to subsequent control activity. A hostile controlchange (tender o!er or proxy contest) is attempted within three years of the "rsttargeting in 16% (20 out of 125) of target "rms and is successful 50% of the time(ten out of 20 attempts). In contrast, only 6.3% ("ve out of 80) of control "rmsexperience a control change attempt. Of these "ve, one is successful. Of the27 target "rms that subsequently experience control activity, 13 (48%) areCalPERS targets, 15 (56%) exclusively receive antitakeover proposals, 20 (74%)receive antitakeover proposals, and "ve (19%) have a passing proposal. Interest-ingly, although only 11 "rms in the entire sample have passing proposals,"ve (46%) of these subsequently experience control activity. Similarly, one-third ofSWIB's and 37% of CalPERS' targets subsequently experience control activity.

We investigate these suggestive univariate relationships more formally usinga probit analysis, reported in Table 7. We classify all events with an attempted orsuccessful change in control as hostile or nonhostile, based on a reading ofarticles in The Wall Street Journal about the control activity. Evidence ofmanagement resistance to the control change leads us to classify events ashostile. For our sample of proposal targets and matching control "rms, weregress the indicator variable for control activity on a number of variablesthought to be related to the probability of takeover. We use the same controlvariables as Comment and Schwert (1995), including stock and accountingmeasures of performance, "rm size, and sales growth. We also include dummyvariables for whether the "rm is in the target or control sample, whether it isa CalPERS target, whether it exclusively received antitakeover proposals, andwhether it received a passing proposal.

We "nd that "rms receiving exclusively antitakeover proposals, those tar-geted by CalPERS, and those with passing proposals are independently asso-ciated with a signi"cantly higher probability of a takeover attempt aftercontrolling for other variables related to takeover probability. These resultssupport the view that strong shareholder voting support sends a message to&would-be raiders', and perhaps suggests that CalPERS is viewed as a perma-nent shareholder that is likely to support a takeover bid.

Successful hostile takeovers are signi"cantly positively related to the antitake-over proposal dummy, and negatively to sales growth. For successful controlchanges (both hostile and nonhostile), "rm size is also signi"cantly negativelyrelated, in addition to sales growth and antitakeover proposals. In a related"nding, Bizjak and Marquette (1999) report that corporate management is morelikely to restructure (rescind or raise the #ip-in trigger) its poison pill if a "rmreceives a poison pill proposal, especially one sponsored by a pension fund.


Although a dummy variable for SWIB's targets is not statistically signi"cantin the regression, 100% of SWIB's proposals are antitakeover proposals. Inter-estingly, by analyzing ownership stakes in target "rms, we "nd signi"cantdi!erences in SWIB's percentage owned at the proposal outcome in "rms thatsubsequently experience control activity. Speci"cally, they own 4.7% on averageof the control activity subsample and 1.5% on average of the rest of their targets,a signi"cant di!erence at the 1% level. In contrast, there are no di!erences inownership detected by separating CalPERS targets that subsequently experi-ence control activity from those that do not.

5. Other measures of the impact of shareholder proposals on target 5rms

In this section we provide evidence on both short-term and long-term eventstudy evidence on the valuation e!ects of shareholder proposals. This allows usto examine the wealth maximization hypothesis and to establish comparabilityto other studies. In addition, we investigate whether the results are consistentwith heterogeneous fund objectives and with the evidence of the previoussection. If pension funds are in#uential but pursue political objectives, we mightexpect to observe a reduction in "rm value and operating performance upontargeting. Alternatively, we might expect to observe no valuation e!ects onaverage if proposals are value-neutral, or as we argue below, if traditionalmethods for measuring valuation e!ects are not powerful enough to capturea proposal's impact.

5.1. Short-term stock price ewects

The "rst three columns in Panel A of Table 8 contain the results of an eventstudy using the market model to estimate normal returns and the Center forResearch in Security Prices (CRSP) value-weighted index as a proxy for themarket.3 Market model parameters are estimated over the interval from 250days through 50 days before the "rst news of a proposal. We report abnormalreturns for three windows: the two-day period around the "rst news of a propo-sal, around the outcome, and in the period that includes both dates. Theannouncement date is de"ned as the earliest of a Wall Street Journal orLexis/Nexis newswire announcement of a proposal, or the date of the proxystatement that contains the proposal. The outcome date is the annual meetingdate, or news date of a withdrawn proposal.

3The market model is a concern in this context since the targeted "rms tend to perform poorlyprior to targeting, so we repeat the event study using a simple market adjustment, with no changes ininferences. Our statistical tests use standard errors that take into account the serial dependence inforecast errors from cumulating abnormal returns using a single set of market model parameters, asdescribed by Mikkelson and Partch (1988).


Tab

le7

Pro

bit

model

regr

essions

ofco

rpora

teco

ntr

olac

tivi

tyon"rm

char

acte

rist

ics

This

table

pre

sents

there

sultsofp

robit

regr

essionsth

atpr

edic

twhic

h"rm

sw

illb

esu

bjec

tto

corp

ora

teco

ntro

lact

ivity

within

thre

eye

arsofa

shar

ehold

erpr

oposa

l.T

he

sam

ple

incl

udes

targ

et"rm

san

dco

ntr

ol"

rmsas

desc

ribed

inT

able

4.T

heta

rget

dum

my

iseq

ualt

oon

eif

the"rm

isin

ourta

rget

sam

ple

and

equal

toze

rofo

rea

chm

atch

edco

ntr

ol"rm

.The

Cal

PER

Sdum

my

iseq

ualto

one

ifC

alPER

Sis

the

spons

oran

dze

rooth

erw

ise.

The

antita

keo

ver

prop

osa

ldum

my

iseq

ual

toone

forta

rget"rm

sth

atex

clusive

lyre

ceiv

edan

tita

keove

rpro

posa

lsan

dze

root

her

wise.

Thepa

ssed

pro

pos

aldum

my

iseq

ual

toon

eif

thepr

oposa

lpas

sed

with

am

ajor

ity

ofv

ote

san

dze

root

her

wise.

The

rem

ainder

ofth

eva

riab

lesar

eas

des

crib

edin

Com

men

tan

dSc

hwer

t(1

995)

.P-v

alue

sar

ein

par

enth

eses

(***

,**,

*in

dic

ates

stat

istica

lsign

i"ca

nce

atth

e1%

,5%

,an

d10

%le

vels)

Inde

pende

ntva

riab

les:

Dep

ende

ntva

riab

les

Dum

my

equa

lto

one

if"rm

expe

rien

ces

aho

stile

takeo

ver

atte

mpt

Dum

my

equa

lto

one

if"rm

expe

rien

ces

asu

cces

s-fu

lhost

ileta

keo

ver

Dum

my

equa

lto

one

if"rm

expe

rien

ces

any

takeo

ver

thre

at

Dum

my

equal

toone

if"rm

expe

rien

ces

any

cont

rolc

han

gein

cludin

gfrie

ndly

mer

ger

Coe$

cien

tC

oe$

cien

tC

oe$

cien

tC

oe$

cien

tIn

terc

ept

1.09

7!

0.30

0!

0.26

42.

111

(0.4

6)(0

.89)

(0.8

3)(0

.19)

Tar

get

dum

my

!0.

400

!0.

009

!0.

610*

!0.

577

(0.3

2)(0

.99)

(0.0

8)(0

.18)

Cal

PER

Sdum

my

0.86

1***

0.72

30.

735*

*0.

467

(0.0

1)(0

.13)

(0.0

3)(0

.28)

Ant

itak

eove

rpr

oposa

ldu

mm

y0.

909*

**1.

160*

*1.

019*

**1.

070*

**(0

.01)

(0.0

3)(0

.00)

(0.0

1)

Pas

sed

pro

posa

ldu

mm

y1.

055*

0.74

21.

145*

*!

0.03

7(0

.07)

(0.3

9)(0

.04)

(0.9

6)


Abn

orm

alre

turn

!75

.725

!29

.437

!55

.824

44.0

91(P

rior

four

-yea

rav

e.)

(0.5

9)(0

.87)

(0.6

5)(0

.74)

Firm

size

!!

0.20

4!

0.16

6!

0.06

5!

0.40

7**

(inprior

year

)(0

.16)

(0.4

5)(0

.60)

(0.0

2)

Sal

esgr

owth

!1.

926

!6.

498*

*.0

18!

4.32

3**

(Prior

four

-yea

rav

e.)

(0.2

2)(0

.04)

(0.9

9)(0

.03)

Liq

uid

ity

1.45

23.

000

1.32

00.

127

(Prior

four

-yea

rav

e.)

(0.2

8)(0

.15)

(0.2

6)(0

.92)

Deb

t/eq

uity

0.00

40.

321

0.24

70.

262

(Prior

four

-yea

rav

e.)

(0.9

9)(0

.46)

(0.4

0)(0

.45)

Mar

ket

/Book

!0.

359

!0.

323

!0.

214

0.11

9(P

rior

four

-yea

rav

e.)

(0.1

2)(0

.38)

(0.2

2)(0

.50)

Price

/ear

nin

gs!

0.02

4!

0.03

4!

0.01

9!

0.02

3(P

rior

four

-yea

rav

e.)

(0.3

0)(0

.25)

(0.3

5)(0

.31)

Num

ber

ofobs

erva

tions

175

175

175

175

Chi

-squa

rest

atistic

25.1

9***

21.0

1**

22.6

2**

20.2

4**

(deg

rees

offree

dom

)(1

1)(1

1)(1

1)(1

1)

!Log

ofto

talas

sets

inye

arprior

to"rs

tta

rger

ing.


Tab

le8

Short

-ter

man

dlo

ng-t

erm

even

tst

udy

resu

lts

atth

ean

nou

ncem

ent

ofa

shar

ehol

der

prop

osa

l

Pan

elA

.A

vera

geab

norm

alre

turn

sfo

rth

een

tire

sam

ple

and

by

spon

sor,

voting

outc

om

e,an

dpro

posa

lto

pic

acro

ssfo

urev

ent

win

dow

s

This

pan

elco

nta

insav

erag

eab

norm

alre

turn

sfo

rth

efu

llsa

mple

and

forsu

bsa

mpl

esby

sponso

r,pro

posa

lou

tcom

e,an

dpr

opos

alto

pic

over

fourev

ent

win

dow

s.P

-val

ues

are

inpar

enth

eses

and

the

per

cent

age

ofpo

sitive

CA

Rs

are

inbra

cket

sunl

ess

other

wise

note

d.*,

**,an

d**

*in

dic

ate

stat

istica

lsign

i"ca

nce

atth

e10

%,5

%,a

nd1%

leve

ls.A

Dis

the

dat

eofth

e"rs

tne

ws

that

ata

rget

com

pan

yw

illre

ceiv

ea

propo

sal.

OD

isth

eea

rlie

rda

teofth

ean

nual

mee

ting

orth

epre

ssan

nounce

men

tof

withdr

awal

orre

solu

tion

(neg

otiat

edout

com

e).F

orex

ampl

e:A

D-2

0to

OD#

20in

clud

esre

turn

sfrom

20da

ysbef

ore

annou

nce

men

tdat

eth

rough

20day

saf

ter

the

outc

om

e.&'

25%'an

d&'

50%'re

fer

tope

rcen

tofvo

tes

cast

infa

vorofa

pro

pos

al.&

Pas

sed'

mea

ns

the

propo

salpa

ssed

acco

rdin

gto

com

pany

byla

ws

(may

require

more

than

50%

ofvo

tes

cast

or

50%

ofal

lsh

ares

,inc

ludi

ngno

n-vo

ted)

.&F

aile

d'

mea

nsth

epr

oposa

ldi

dnotpas

sac

cord

ing

toco

mpan

ybyl

aws.

For

prop

osa

lsw

ith

iden

tica

lev

entda

tes

(e.g

.,both

Cal

PER

San

dSW

IBta

rget

edIB

Mw

ith

di!er

entpro

pos

als

ondi!

eren

tto

pic

s),e

ach

sponso

ran

dto

pic

are

cred

ited

with

the

pro

pos

alin

the

subs

ample

resu

lts.

How

ever

,th

epr

oposa

lis

coun

ted

only

once

inth

efu

llsa

mpl

ere

sults.

For

the

shor

tw

indow

s,w

euse

the

mar

ket

model

toes

tim

ateth

epar

amet

ersto

calc

ulat

enorm

alre

turn

s,an

dou

rst

atistica

ltes

tsus

est

anda

rder

rors

that

take

into

acco

untth

ese

rial

depen

den

cecr

eate

dby

cum

ula

ting

abnor

mal

retu

rnsca

lcula

ted

using

asing

lese

tof

mar

ket

model

par

amet

eres

tim

ates

,as

des

crib

edby

Mik

kel

son

and

Par

tch

(198

8).F

or

the

long

win

dow

,w

eco

mpute

mea

nbuy

-and-

hol

dre

turn

s(B

HA

R)u

sing

them

etho

dofB

arber

and

Lyo

n(1

997)

.Spec

i"ca

lly,B

HA

Ris

the

sim

ple

aver

age

ofc

om

pou

ndta

rget"rm

retu

rnsov

erth

eth

ree

year

saf

ter

the

announce

men

tof

apro

posa

lm

inus

the

sim

ple

aver

age

of

com

poun

dco

ntr

ol"rm

retu

rns

over

this

sam

eper

iod.

Tha

tis,

BH

AR

*T"

<(1#

Rit)!

<(1#

Rct),

wher

eB

HA

RiT

isth

eper

iod¹

buy-

and-h

old

retu

rnfo

rse

curity

i,R

itis

the

mon

thly

raw

retu

rnfo

rse

curity

i,an

dR

ctis

the

month

lyra

wre

turn

for

the

cont

rol"rm

Win

dow

AD

-1to

AD

Tw

o-d

ayan

noun

cem

ent

OD

-1to

OD

Tw

o-d

ayou

tcom

eA

D-2

0to

OD#

20U

pto

thre

eye

ars

afte

r

Full

sam

ple

(198

7}93

)!

0.00

000.

0007

0.02

56!

0.03

52n"

224

(0.8

470)

[48.

7](0

.388

9)[4

9.1]

(0.3

289)

[51.

8](0

.742

6)[n

"73

]

Sponso

r:Tw

o-d

ayan

nounc

emen

tTw

o-d

ayout

com

e(A

D-2

0to

OD#

20U

pto

thre

eye

ars

afte

rSW

IB!

0.00

035

0.00

804*

*0.

0794

8!

0.09

95n"

20(0

.954

5)[5

5.0]

(0.0

241)

[55.

0](0

.137

9)[6

5.0]

(0.7

226)

[n"

9]C

REF

!0.

0015

20.

0001

10.

0098

80.

0294

n"52

(0.5

727)

[38.

5](0

.976

1)[4

8.1]

(0.7

148)

[51.

9](0

.841

1)[n

"17

]C

ALPER

S0.

0038

90.

0085

2**

!0.

0119

9!

0.06

46n"

40(0

.381

2)[5

0.0]

(0.0

268)

[50.

0](0

.915

3)[5

0.0]

(0.8

039)

[n"

20]


CA

LSTR

S0.

0033

10.

0022

2!

0.00

845

!0.

1587

n"28

(0.9

103)

[53.

6](0

.378

3)[4

2.9]

(0.5

286)

[39.

3](0

.572

4)[n

"10

]N

YC

funds

0.00

345

!0.

0020

90.

0694

5*0.

0961

n"10

1(0

.308

6)[5

1.5]

(0.5

121)

[52.

5](0

.089

5)[5

4.5]

(0.5

613)

[n"

31]

Out

com

eFai

led

0.00

142

0.00

128

0.02

233

!0.

0466

n"20

0(0

.464

8)[4

9.0]

(0.1

657)

[50.

0](0

.448

0)[5

1.0]

(0.7

092)

[n"

52]

Pas

sed

0.00

815

0.00

971

0.02

442

0.38

94n"

11(0

.398

9)[4

5.5]

(0.1

250)

[63.

6](0

.874

5)[4

5.5]

(0.3

214)

[n"

6]N

egotiat

ed/w

ithd

raw

n0.

0010

9!

0.00

362

0.05

756

!0.

0674

n"16

(0.7

302)

[50.

0](0

.206

3)[3

1.2]

(0.4

077)

[68.

8](0

.601

5)[n

"16

]'

25%

vote

sin

favo

r0.

0014

20.

0028

0**

0.01

539

!0.

0647

n"17

1(0

.464

8)[4

9.1]

(0.0

425)

[50.

3](0

.414

0)[5

0.3]

(0.6

700)

[n"

49]

'50

%vo

tes

infa

vor

0.00

815

0.01

077*

*0.

0288

10.

6403

n"15

(0.3

989)

[46.

7](0

.013

4)[6

0.0]

(0.6

518)

[40.

0](0

.164

8)[n

"7]

Pro

posa

lto

pic

Vot

ing

issu

es0.

0035

50.

0001

00.

0087

5!

0.14

83n"

111

(0.1

028)

[53.

0](0

.602

5)[5

4.1]

(0.6

392)

[54.

1](0

.535

0)[n

"21

]A

ntitak

eove

rissu

es!

0.00

247

0.00

214

0.01

066

!0.

1945

n"10

2(0

.189

4)[4

4.1]

(0.2

103)

[42.

2](0

.775

4)[4

7.1]

(0.2

066)

[n"

24]

Boar

dissu

es0.

0132

50.

0041

80.

1935

9*0.

4567

n"19

(0.1

770)

[47.

4](0

.693

9)[6

3.2]

(0.0

550)

[57.

9](0

.226

6)[n

"9]


Tab

le8.

Continu

ed.

Pan

elB.A

nno

unc

emen

tdat

eav

erag

ecu

mul

ativ

eab

nor

mal

retu

rns

for

prop

osal

spa

rtitio

ned

byw

heth

erth

ean

nounc

emen

tdat

eis

earlie

rth

anor

the

sam

eas

the

pro

xyst

atem

ent

date

This

pan

elco

nta

insth

ere

sultsof

part

itio

nin

gth

esa

mple

acco

rdin

gto

when

the

mar

ket

lear

nsoft

hesh

areh

old

erpr

oposa

l.A

nan

nounce

men

tdat

eea

rlie

rth

anth

epr

oxy

stat

emen

tda

tein

dic

ates

that

wear

eab

leto"nd

apr

essan

nou

nce

men

tfo

rth

epro

posa

lpriorto

the

proxy

stat

emen

tda

te.I

fwe

are

unab

leto"nd

apr

ess

announ

cem

ent,

the

pro

xyst

atem

ent

date

isth

ean

nounc

emen

tdat

e.p-

valu

esar

ein

pare

nth

eses

(*,**

,an

d**

*in

dic

ate

stat

istica

lsign

i"ca

nce

atth

e10

%,5

%,an

d1%

leve

ls)

Tw

o-d

ayan

noun

cem

ent

dat

ecu

mul

ativ

eab

norm

alre

turn

s

Ann

oun

cem

ent

date(

pro

xydat

eA

nnoun

cem

ent

date"

pro

xydat

e

Full

sam

ple

0.00

4!

0.00

1(0

.158

8)[5

1%po

sitive

]MN

"10

2N(0

.686

4)[4

5%pos

itiv

e]MN

"11

5N

1990

to19

93pro

xyse

asons

only

0.00

8**

0.00

1(0

.045

5)[5

6%po

sitive

]MN

"61

N(0

.769

5)[4

7%pos

itiv

e]MN

"60

N

Ant

itak

eove

ran

dboa

rdpro

posa

lson

ly(e

xclu

des

voting

prop

osa

ls)

0.00

4!

0.00

5*(0

.617

8)[4

6%po

sitive

]MN

"48

N(0

.051

7)[4

1%pos

itiv

e]MN

"64

N

Ant

itak

eove

ran

dboa

rdpro

posa

lsin

the

1990

to19

93pro

xyse

asons

only

0.01

0!

0.01

2*(0

.121

8)[5

3%po

sitive

]MN

"32

N(0

.059

1)[1

9%pos

itiv

e]MN

"16

N


There are three reasons why an event study methodology might not bepowerful enough to capture the true impact of shareholder activism. First, thereis substantial event date uncertainty for proposal announcements because onlya minority are announced in the usual "nancial news sources such as the WallStreet Journal. Second, since the proposals are nonbinding and management'sresponse to the proposal is unknown, uncertainty regarding the valuation e!ectsof proposals continues over the long period between the announcement andoutcome dates. For example, management might adopt the proposal eventhough it does not obtain a strict majority (in 1988 Gillette implemented aproposal relating to its poison pill even though the proposal did not receive amajority of votes), or might not adopt the proposal even though it does wina majority of votes (USAIR, also in 1988). Even if a proposal does not pass, ahigh percentage in favor can signal broader discontent with management anda call to further action. For example, SWIB tells us that in some cases manage-ment is totally unresponsive until their proposal generates signi"cant supportfrom other shareholders, at which point &the board may look at it di!erently'.Thus, one view of proposals is that they are an inexpensive way to mobilizesupport or alert directors to oust incompetent managers.

Finally, we learn from the interviews that often a proposal is publicly "ledonly after private communications with target management have failed to satisfythe proposal sponsor. Since many pension fund proposals are successfullynegotiated and withdrawn prior to the publication of the proxy statement,appearance of a proposal on the proxy can signal the bad news that manage-ment is unresponsive to shareholders. Thus, a negative stock price reaction tothe announcement of a proposal might not necessarily be evidence that publicpension funds are politically motivated. We provide some evidence consistentwith this conjecture below.

Panel A of Table 8 presents the results for the full sample as well as forsubsamples partitioned by sponsor, voting outcome, and proposal topic. Dataon voting outcomes are from the IRRC Corporate Governance Service. Consis-tent with the results of other studies, we "nd little evidence that there aresigni"cant valuation e!ects for broad samples of proposals at the announcementof a shareholder proposal submission. The two-day announcement return forthe entire sample of 224 proposals from 1987}1993 is not signi"cantly di!erentfrom zero. We "nd that none of the subsamples have signi"cant returns in theannouncement window, but CalPERS, SWIB, proposals receiving greater than25% of votes, and proposals receiving greater than 50% of votes have positivesigni"cant abnormal returns in the two-day outcome window. With the excep-tion of the result for SWIB, however, these results are sensitive to one outlier(Avon Corp.) that experienced a takeover bid within the outcome window.Targets of proposals on board-related issues exhibit positive and signi"cantabnormal returns of 19.4% in the longer window encompassing the periodbetween the announcement and outcome dates. This is consistent with the


results of our corporate news analysis, which "nds board-related proposals tohave the broadest impact.

Panel B of Table 8 contains results for the announcement date partitioned bywhen the market learns of the shareholder proposal. An announcement dateearlier than the proxy statement date indicates that we are able to "nd a pressannouncement for the proposal prior to the proxy statement date. If we areunable to "nd a press announcement, the proxy statement date is the announce-ment date. We "nd that negative and signi"cant announcement day CARs areconcentrated in antitakeover and board proposals that have the "rst announce-ment on the proxy statement. Within this subsample, the CARs are even morenegative on average in the 1990s when negotiated settlements became morecommon, which is consistent with the conjecture that the negative reaction isdue to the market learning that negotiations failed. Proposals in the 1990sannounced prior to the proxy statement have signi"cantly positive abnor-mal returns. There is no di!erence in CARs from the proxy date or earlier datesfor voting proposals, however, which we attribute to their relatively benigncontent.

5.2. Long-term performance of targets

As other studies argue, the bene"ts of activism might only be revealed overa long time period. We also investigate the long-run stock price performance oftarget "rms relative to the control sample and relative to broad market indexes.We note that while long-term abnormal stock price performance is a desirablemetric to capture, even the best methods are imprecise and should be interpretedwith care. We calculate monthly buy-and-hold returns following the methodrecommended in Barber and Lyon (1997) for target and control "rms throughthree years after the "rst announcement of a targeting, and test for signi"cantdi!erences between the two groups. The last column of Panel A of Table 8 showsthat for both the full sample and subsamples by sponsor, outcome, and proposaltopic, there is no evidence of abnormal returns.

We also compare the returns of our target sample to the S&P 500, anapproach similar to Nesbitt (1994). We "nd, but do not report, that targetcompanies generally have higher returns than the index, but not signi"cantlyhigher, in the three years after targeting. We "nd the same result for our controlsample of "rms of similar industry, size, and performance, which leads us tosuspect that the stronger performance Nesbitt attributes to CalPERS' targetingis due instead to mean reversion or a tendency to rebound from poor perfor-mance.

We also examine, but do not report, whether shareholder activism has animpact on the operating performance of target companies within three years oftargeting. Using the method used in Karpo! et al. (1996), we investigate whethertarget companies rebound more quickly from poor performance by comparing


the accounting performance measures of our target sample to the performance-,industry-, and size-matched control sample described in Section 4. Speci"cally,we calculate operating income divided by assets (ROA) and operating incomedivided by sales (ROS) as well as sales and asset growth for four years, beginningin the year before the "rst targeting. Consistent with the "ndings of others, target"rm performance does not rebound faster than that of performance-matchedcontrol "rms.

Overall, we "nd few statistically discernible results from our analysis of stockand operating performance of target "rms. Given the stated measurementdi$culties and lack of power, however, this is not surprising. Thus far, we "ndno evidence to support the hypothesis that motivations other than fund valuemaximization are the impetus behind public fund activism. In the next sectionwe provide further evidence on fund motivation.

6. Evidence on the normative goal of fund value maximization

In this section we discuss the alternative hypotheses to fund value maximiza-tion discussed in the literature, and provide some evidence that the funds' targetselection and trading behavior does not appear to be consistent with thesealternatives.

Romano (1993) and Murphy and Van Nuys (1994) maintain that agencyproblems within the funds themselves prevent them from being e!ective corpo-rate monitors. They argue that managers of public funds may use their in#uenceto further their own political or personal goals, instead of maximizing bene"ci-ary wealth. For example, Murphy and Van Nuys predict that public pensionswill tend to &target high-pro"le companies, generating large nonmonetary re-wards for the fund managers in the form of publicity (which in turn may a!ecttheir future employment opportunities)'. Others argue that the absence ofactivist private pension funds, or other types of institutional investors withsuperior incentive structures, is evidence that public funds have ulterior motives.There are, however, reasonable alternative explanations for these observations.As we have argued, the use of publicity might be part of an overall strategy foran index fund to maximize value. Further, other types of institutions, such ascorporate pension funds, bank trusts, and insurance companies, might forgoactivism because they are reluctant to appear openly antagonistic to manage-ment or to jeopardize business relationships with the "rm. Anecdotally, in 1987,CEOs of seven Fortune 500 companies wrote letters to their fellow CEOs urgingthem to instruct their pension managers to vote against shareholder proposals(Institutional Investor, June 1988, p. 162.) Also, Pound (1988) "nds that institu-tions such as banks and insurance companies are more likely to side withmanagement in a proxy contest, possibly in order to ensure future business tieswith the "rm. Similarly, Van Nuys (1993), in a case study of a proxy "ght at


Honeywell Corporation, "nds that bank trusts and insurance companies aremore likely to support management-sponsored antitakeover proposals than arepublic pension funds.

There is a growing literature that "nds indirect evidence that activist fundsare politically motivated. Based on a sample of Fortune 500 "rms, Woidtke(1996) "nds a negative relation between "rm performance (industry-adjustedTobin's Q) and percentage ownership by activist public pension funds. Wagsterand Prevost (1996) "nd that "rms targeted by CalPERS have signi"cantlynegative stock price reactions to the announcement of the 1992 proxy rulechanges. Johnson et al. (1997) "nd that a dummy variable for CalPERS 1992&hit-list' "rms is negatively related to both CEO compensation changes andpay-for-performance sensitivity changes, which they interpret as evidence thatCalPERS acts more like a populist crusader against executive pay levels thana wealth-maximizing shareholder.

6.1. Target selection

We might expect a wealth-maximizing fund to choose targets for which thereis a high probability of having an impact, garnering support from other share-holders, and recouping the return on investing in this activity. Of course, toassess returns from proposals one must consider costs as well as bene"ts. All "vefunds emphasize the cost-e!ectiveness of shareholder proposals. Their ownestimates of the annual cost of their entire activism programs range from$50,000 to $1 million, which is less than half of a basis point for these funds.4Given the substantial ownership stakes documented in Table 2, there appears tobe ample economic incentive to become active. Even CalSTRS, the fund with thesmallest investment in targets at $422 million, has a strong incentive to submitshareholder proposals. If their monitoring activities were to improve stockprices a mere 0.5% at target "rms, they could increase their portfolio value by $2million. There is, of course, an even greater incentive at the larger and moreactive funds.

The selection criteria, as described by the funds and con"rmed by empiricalstudies, are arguably consistent with fund value maximization. During the earlyyears of our sample, funds target "rms based on the speci"c issue at hand, butlater use poor performance as a criterion. The only exception is CREF, who still

4Smith (1996) reports that CalPERS spends approximately $500,000 annually on all activismactivities. This represents only 0.002% of the value of their domestic equity holdings. Carleton et al.(1998) report that CREF spends $1 million annually, or 0.002% of assets on its activism program.SWIB states to us in the interview that the cost of activism is &a tiny fraction of our assets, not evena blip on our screen'. In contrast, Pound (1991) estimates the legal costs of soliciting support fora full-control proxy solicitation at $1.5}3.5 million.


selects targets on the basis of speci"c issues (e.g., if a portfolio companyimplements a poison pill, they receive a proposal regardless of performance orother factors). All the funds state that in selecting targets they also consider thestake they own, since this is directly linked to the payo! from value-improvingactions, as well as to the clout with the company's management. They alsoconsider how much of the target's stock is in institutional hands, becauseinstitutional investors are easier to coordinate with and are considered morelikely to support a proposal.

Recent research is consistent with what the funds have told us about theirselection criteria. For a sample of S&P 500 "rms in the 1991}1992 proxy season,John and Klein (1995) "nd that the likelihood of a "rm receiving a proposalsponsored by a pension fund is signi"cantly negatively related to the previoustwo-year stock return, and positively related to "rm size and institutionalownership. In addition, these two variables are not signi"cant for any other typeof proposal sponsor. Similarly, Karpo! et al. (1996) "nd that the probability ofreceiving a proposal is signi"cantly positively related to "rm size and institu-tional ownership, and negatively related to previous accounting performanceand leverage. Carleton et al. (1998) "nd that the likelihood of being targeted byCREF is positively related to institutional ownership, negatively related toinsider ownership, and unrelated to prior stock performance. Finally, Smith(1996) reports that target selection by CalPERS is positively related to "rm sizeand institutional ownership, but unrelated to performance measures. However,the insigni"cance of performance is possibly due to the fact that by constructionhis comparison "rms are also poor performers.

6.2. Portfolio analysis: adjustments of holdings in target xrm stocks

Examining buying and selling behavior around proposals allows us to verifyempirically what the funds tell us about their targeting and investment strat-egies, thereby providing a more complete picture of their behavior. Relatingtheir targeting decisions to their trading patterns provides further evidence onfund motivation since we can assess the likely impact of their behavior on fundwealth.

All four primarily indexed funds claimed that over our sample period there isno relation between their activism programs and investment decisions. SWIBclaims a minimal relation in that the investment sta! is involved when choosingtargets, but not after that point. Table 2 provides summary statistics on fundownership stakes in target "rms before and after the submission of a shareholderproposal. We do not analyze the ownership patterns for NYC since they do notreport their portfolio holdings in a 13F "ling directly to the SEC, which is thesource of our portfolio data for the other funds.

Consistent with their indexing investment strategies, and with evidence inWahal (1996), there is little evidence of trading around proposals by either


CalPERS, CalSTRS, or CREF. Even though we would expect CREF to exerciseits ability to link trading decisions to their activism programs in the 16% of theirinternally managed portfolio that is not indexed, there is no evidence that theydo. Their average ownership stake in target "rms is 1%, the magnitude onewould expect for an index fund, and the stake does not change much aroundtargeting. In only six out of 32 targets does the stake appear to be signi"cantlylarger than 1%. The ownership stake in each "rm under indexing can beestimated by dividing the dollars CREF devotes to indexing by the total marketvalue of the index. (On 3/30/90 CREF indexed $23.91 billion and the value of theS&P 500 index was $2260.28 billion, which indicates that CREF's percentageownership stake in each "rm in the index should be 1.1%.) In contrast, SWIB'saverage ownership stake drops from 2.3% in the year before the "rst targeting to0.4% in the year after the last targeting, while the median drops from 1.6% tozero.5 In the remainder of this section, we focus on the trading behavior of theactively managed SWIB, since they exercise the greatest #exibility in theirportfolio decisions.

We expect a wealth-maximizing pension fund to submit a shareholder propo-sal only if they expect it to have a positive impact on "rm value, and if thisimpact outweighs the cost. To pro"t from their anticipated positive impact onthe stock price, an actively managed fund might increase their stake in the target"rm prior to submitting a proposal (or at least not reduce their stake). Aftertargeting, actions consistent with fund value maximization will depend on thesuccess of the proposal. It could be that the fund will maintain the same stakeeither because they want to hold on long enough to reap the bene"ts of theirtargeting, or because they have positions too large to unload quickly. However,reducing their stake post-targeting is also consistent with fund value maximiza-tion since they could either be cutting their losses in an unresponsive target "rmor realizing their positive gains from targeting. It makes less sense for the fund toincrease their stakes post-targeting, but our strongest prediction is that wewould expect a wealth-maximizing active fund to trade on their information. Wetest these predictions below.

5Our con#icting results for SWIB are most likely due to Wahal's methodology of studyingchanges in activist funds' stakes in all target "rms (i.e., those targeted by any fund), instead of in onlythose "rms that they themselves target. We analyze a fund's ownership in only their own targets fortwo reasons. First, we are most interested in relating their alterations in portfolio investments totheir decision to target "rms with proposals. Assuming they react to the targeting decisions of otherfunds can mask this relation. Second, in interviews the funds were each very clear that they neitherknow nor care to know the details of what other funds are doing. For example, according to RichardKoppes at CalPERS, &We don't really know how [other funds] put their program together, haven'tspent the time to "nd out. We love and support our fellow public funds, but don't know what they'redoing in advance'. Similarly, according to Kurt Schacht at SWIB, &We don't own a lot of thecompanies that CalPERS owns'.


Panel A of Table 9 contains statistics on changes in the level of funds'ownership stakes in target "rms beginning three years prior to the "rst targetingand ending one year after the last targeting. Ownership at time zero is de"ned asthe holdings reported in the quarter ending after the outcome date, which istypically the annual meeting date. For example, if the outcome date is April 20,1991, then time zero holdings are as of June 30, 1991. Thus, if the pension funddivests a stock immediately after the annual meeting, the time zero holdingcould be zero. They must hold the stock until then since shareholder proposalrules require the sponsor to present the proposal at the annual meeting.

For each fund and event period, we provide a benchmark value that we labelthe expected change in ownership stake. This benchmark is calculated using anestimate of each individual fund's portfolio turnover (excluding target holdings),and can be interpreted as the expected change in ownership stake if targetholdings are turned over with the same frequency as in the rest of the portfolio.This measure is listed in the table as a positive value, even though eithera positive or negative change of this magnitude in ownership stake would beconsidered normal for this pension fund. An advantage of this benchmarkmeasure is that it accounts for di!erences across funds in portfolio turnover andin the size of the stake in target "rms. Thus, if a fund has low turnover becausethey are heavily indexed, the expected change in target holdings would also below. Appendix B describes this benchmark measure in more detail.

The observed patterns appear to be consistent with the funds' stated invest-ment strategies and with fund value maximization. There are few changes intarget holdings for the indexed funds, while SWIB appears to build positions intarget "rms prior to targeting and then divests them within one year of the lasttargeting. The change in ownership stake is 1.1% from three years to one yearprior to the "rst targeting, and !1.8% from one year before to one yearafter the last targeting. Both of these changes are signi"cantly di!erent inabsolute value from the expected changes using a standard t-test. In addition, allchanges in SWIB's target ownership are signi"cantly di!erent from zero at the5% level.

Panel B of Table 9 shows that SWIB completely divests 65% of its targets (11out of 17) within one year of the last targeting, and 29% (5 out of 17) within onequarter. In contrast, none of the other funds divest target "rms, consistent withtheir indexing strategy. These ownership patterns are especially interesting whencombined with the results reported earlier. For example, we "nd that CalPERS'targets have signi"cantly more announced corporate events in the years follow-ing the proposal than a matched control sample, while SWIB's targets do not.Since here we show that the majority of SWIB's targets are divested within thelast year, it is possible that corporate managers are unresponsive to SWIBbecause they know that SWIB will go away, while CalPERS will not. However,we also "nd that SWIB's targetings are associated with signi"cantly positiveabnormal stock returns at the outcome date. Thus, their reduction in target


Tab

le9

Cha

nges

inta

rget

hold

ings

for

four

pens

ion

fund

s

Pan

elA

.Thi

span

elre

port

sav

erag

ean

nual

chan

gesin

pension

fund

hol

din

gsbas

edon

qua

rter

ly13

F"lin

gsfrom

June

1986

toJu

ne

1994

.The

targ

etin

gda

teis

de"ned

asth

equa

rter

oft

hepr

oposa

l'sout

com

e,w

hich

isty

pica

llyth

ean

nual

mee

ting

date

.Sin

cem

any"rm

sar

eta

rget

edm

ultip

letim

es,t

he

tabl

ebe

low

spec

i"es

whe

ther

the

stat

istics

are

rela

tive

toth

e"rs

tor

last

targ

etin

g.F

orex

ampl

e,th

e"rs

tro

wofr

esultsre

port

sch

ange

sin

hold

ings

from

thre

eye

ars

befo

reto

two

year

sbef

ore

the"rs

tta

rget

ing.

The

expec

ted

chan

geben

chm

ark

isca

lcul

ated

using

anes

tim

ate

ofea

chin

divi

dual

fund's

port

folio

turn

ove

ran

dca

nbe

inte

rpre

ted

asth

eex

pect

edch

ange

inow

ners

hip

stak

eif

targ

etho

ldin

gsar

etu

rned

over

with

the

sam

efreq

uency

asth

ere

stofth

epo

rtfo

lio.A

ppen

dix

Bdes

crib

esth

isbe

nchm

ark

mea

sure

inm

ore

det

ail.

Stan

dar

dde

viat

ions

are

inpa

renth

eses

SWIB

CA

LP

ER

SC

ALSTR

SC

REF

Act

ual

chan

geExp

ecte

dch

ange

Act

ual

chan

geExp

ecte

dch

ange

Act

ual

chan

geExp

ecte

dch

ange

Act

ual

chan

geExp

ecte

dch

ange

(!3,!

2)0.

011!

0.00

20.

0012

0.00

090.

0011

!0.

0006

!0.

0006

0.00

12(t"

0is"rs

tta

rget

ing)

(0.0

22)

(0.0

027)

(0.0

012)

(0.0

027)

(!2,!

1)0.

009

0.00

30.

0005

0.00

130.

0005

0.00

07!

0.00

04!

0.00

11(t"

0is"rs

tta

rget

ing)

(0.0

22)

(0.0

05)

(0.0

019)

(0.0

021)

(!1,

0)!

0.01

00.

0046

!0.

0001

0.00

110.

0011

0.00

060.

0008

0.00

12(t"

0is

last

targ

etin

g)(0

.021

)(0

.006

3)(0

.003

5)(0

.003

4)(0

,1)

!0.

007

0.00

24!

0.00

21!

0.00

110.

0001

0.00

08!

0.00

180.

0012

(t"

0is

last

targ

etin

g)(0

.014

)(0

.003

7)(0

.004

1)(0

.004

9)(!

1,1)

!0.

018!

0.00

7!

0.00

220.

0022

0.00

120.

0014

!0.

0007

"0.

0024

(t"

0is

last

targ

etin

g)(0

.026

)(0

.006

6)(0

.007

5)(0

.000

5)


Pan

elB

.This

pan

elre

por

tsth

epe

rcen

tage

ofa

fund's

targ

etsth

atar

eco

mple

tely

div

este

dw

ithi

ntw

oye

ars,

one

year

,and

one

quar

teraf

terth

eou

tcom

edat

eof

the"na

lta

rget

ing.

SWIB

CA

LP

ER

SC

ALSTR

SC

REF

Per

cent

ofta

rget

hold

ings

div

este

dw

ithin

Tw

oye

ars

ofla

stta

rget

ing

713

130

One

year

ofla

stta

rget

ing

653

74

One

quar

ter

ofla

stta

rget

ing

290

03

!Sig

ni"ca

ntly

di!er

ent

from

the

expe

cted

chan

geat

the

10%

leve

l."S

igni"

cant

lydi!

eren

tfrom

the

expec

ted

chan

geat

the

5%le

vel.


holdings after the outcome date is also consistent with them realizing some ofthe gains from their e!ort by selling stock, a possibility we explore below.

6.3. Further analysis of SWIB'S ownership patterns

In measuring SWIB's ownership patterns, we focus on the changes before thexrst targeting and after the last targeting. This could make it appear that SWIBgets in and out of stocks quickly, when actually there are several cases in whichthey target a "rm several years in succession. To investigate potential motiva-tions behind a decision to repeat targeting or to sell out of a position, wecompare the average two-day outcome date CAR in the targets that are sold inthe year after the outcome date versus those that are held. The average CAR inthe held group is 1.1%, versus 0.4% in the sold group. This di!erence isstatistically signi"cant at the 10% level in a two-tailed test, and is con"rmed bya regression with the change in percentage ownership from the year before tothe year after the outcome date as the dependent variable and the outcomedate CAR as the explanatory variable. This evidence is not consistent withSWIB exiting their position to immediately realize the outcome date abnormalreturn.

Another possible explanation for SWIB's exit pattern relates to the targetselection strategy they describe to us in the interview. Speci"cally, they target the&worst of the worst' performing "rms in their portfolio, and only give up on themwhen it appears that the "rm is not going to improve. To examine whether theirbehavior is consistent with this strategy, we compare the average excess returnsin the year following the outcome date for two groups of stocks: SWIB's targetsthat are sold within a year of targeting and those that are held for that year.Using the CRSP value-weighted market index, the average market adjustedreturn is !5.5% for those sold by SWIB and 0.9% for those held. Using theCRSP equally weighted market index, the market adjusted returns are !2.8%and 4.3%. Because this test is based on a small sample, the sold and held groupsare not statistically di!erent from each other. However, 83% of the sold group(ten out of 12 targets) have returns below the CRSP value-weighted index and67% (eight out of 12) have returns below the CRSP equally weighted index. Thisevidence is at least consistent with SWIB maximizing fund value with respect tothe decision regarding whether to target the stock again or to sell out.

Overall, we "nd no evidence to support the hypothesis that public pensionfund managers are politically motivated. In the previous section, we "nd noevidence that proposals sponsored by pension funds tend to decrease "rm valueor operating performance. In addition, our analysis of target selection andtrading patterns around proposals shows that the funds' behavior appears to begenerally consistent with their investment strategies and with fund value maxi-mization. Even though the index funds in our sample cannot recoup theirinvestment in activism through their trading, it is plausible that they bene"t


from spillover e!ects from activism tools such as publicity that boost theperformance of their portfolio as a whole. An exception, however, is CREF.Although its activism strategies and goals appear to be consistent with theinternally and actively managed portion of the fund, we "nd that its tradingstrategies seem more consistent with its indexed portion.

7. Conclusions

We examine the impact and motivation of pension fund activism by studyingthe shareholder proposals submitted by "ve of the largest and most activistfunds. To better understand the activism process, and consequently to formmore precise tests, we interview key decision makers at these funds. We "nd thatthe sample funds di!er in their activism objectives, their use of publicity as anactivism tool, and their impact on target "rms, and that these di!erences aregenerally consistent with the funds investment strategies.

We "nd that shareholder proposals are followed by signi"cant additionalcorporate governance activity and broad corporate change, such as asset salesand restructurings. We also "nd evidence that proposals play a complementaryrole to other governance mechanisms. Speci"cally, "rms targeted by CalPERSor subject to a proposal on antitakeover issues are signi"cantly more likely toreceive a hostile takeover bid, and targets of antitakeover proposals are morelikely to experience a change in control than their matching comparison "rms.We "nd no evidence that this activity has signi"cant e!ects on stock return oraccounting measures of performance in the three years following an initialtargeting, and only sketchy evidence of positive e!ects in the short term. Finally,we "nd that portfolio movements of the pension funds are generally consistentwith fund value maximization and with their stated investment strategies andactivism objectives. We conclude that shareholder proposals are e!ective inpromoting change at target companies, and that pension fund activism is notinconsistent with fund value maximization.

One caveat to our conclusions is that our sample period ends in 1993. Certainpractices by the pension funds have changed since 1993 (e.g., CREF no longerpublicly announces their targets, even after a successful negotiation), and thusthe results of their activism today may di!er from our "ndings. In addition, oursample is restricted to shareholder proposals, and thus represents only a portionof the governance activities of pension funds. For example, Smith (1996) andWahal (1996) include fund letters to management in their samples, Opler andSokobin (1997) study the Council of Institutional Investor's &hit list', Carletonet al. (1998) study CREF's private negotiations with target "rms, and Stricklandet al. (1996) study USA's Target 50 list. Other activism activities not studied inthis paper include lobbying the SEC for changes in the proxy rules and &just voteno' campaigns.


Overall, our evidence indicates that shareholder proposals can be consideredcomplementary elements in an array of governance mechanisms, each of whichcan be utilized as the situation demands. As there are a broad range of potentialcon#icts between shareholders and managers, it seems optimal to also havea broad range of mechanisms to resolve them. With only costly and highlycontentious mechanisms such as tender o!ers and proxy contests, smallercon#icts would perhaps remain unresolved and worsen. Thus, when judging thee!ectiveness of shareholder proposals, it is important to recognize their poten-tial value in a full spectrum of corporate governance tools.

Appendix A. Disaggregated results of Table 5

Table 10 contains the test results of comparing the post-targeting frequency ofindividual types of corporate events announced in the Wall Street Journal intarget "rms relative to a control sample matched by size, two-digit industrycode, and accounting performance. We employ the nonparametric Mann}Whitney U test. Statistical signi"cance indicates that the distribution of an-nounced events in the target sample is signi"cantly to the right of the controlsample (*,**, and *** indicate statistical signi"cance at the 10%, 5%, and 1%levels for a one-tailed test). In addition to the results of tests on the entire sample,we include the subsample results by proposal sponsor, proposal type, timeperiod, and by whether the proposal passed, was withdrawn by the sponsor, orfailed to get a majority of votes.

Appendix B. Calculation of the expected change in ownership stake benchmark

The fund- and period-speci"c benchmark value reported in Table 9 can beinterpreted as the expected change in ownership stake if target holdings areturned over with the same frequency as the rest of the portfolio (non-targetholdings). Speci"cally, we calculate the following for each fund and eventwindow t!1 to t:

Expected change in % ownership staket~1,t

"Average turnover ratio/0/v5!3'%5

*Average % ownership stake in target "rmst~1

.

This measure is listed in the table as a positive value, even though what we arereally saying is that either a positive or negative change of this magnitude inownership stake would be considered normal for this pension fund. For eachfund we calculate the non-target stock turnover ratio for every year from 1986 to1994. The average of these ratios is our estimate of the non-target turnover ratio.


Tab

le10

New

sca

tego

ries

Full

Sam

ple

Fund

spons

or

Pro

pos

alty

peD

ate

Out

com

e

SW

IBC

REF

CA

L-

PE

RS

CA

L-

ST

RS

NY

CV

otin

gA

nti-

take

-ov

er

Boar

dissu

esM

ultip

lety

pe19

87to 19

89

1990

to 1993

Pas

sed

With-

dra

wn

Fai

led

CEO

retire

sC

EO

resign

suns

ched

ule

d*

*D

irec

tor

resign

sO

ther

top

exec

utive

resign

s**

***

***

***

***

**

***

Shar

ehold

erla

wsu

it*

***

**

Non-s

ample

pro

posa

l,et

c.*

***

***

Pro

xyco

ntes

tat

tem

pte

d*

Win

seat

(s)(p

roxy

conte

st)

Host

ileta

keo

ver

bid

Host

ileta

keo

ver

defe

nse

s*

*H

ost

ilebi

dsu

cces

sful

Mer

ger/

acqui

sition

talk

s*

Mer

ger/

acqui

s.co

mple

ted

Ass

etsa

les

orsp

ino!

***

**

****

Maj

or

asse

tpurc

hase

*M

anag

emen

tbu

y-ou

tLar

gein

sider

purc

has

eLar

gein

sider

sale

New

/incr

ease

maj

or

blo

ck*

***

Maj

or

blo

cksa

le*

Res

truc

turing

/reo

rgan

izat

ion

**Lay

o!s

***

***

***

***

**In

crea

sediv

iden

ds**

Dec

reas

edi

vide

nds

**

*R

epurc

has

esh

ares

***

Issu

esh

ares

Ban

kru

ptcy

decl

ared

Acc

oun

ting

irre

gula

rities

&No'vo

tefo

rdi

rect

ors


The following equations de"ne our measure of annual non-target stock turn-over, which is an estimate of the value-weighted turnover in non-target stocks.

<Pt

" Value of the entire portfolio at time t<TAR

t" Value of the target stocks held at time t

<NONt

" Value of the non-target stocks held at time t"<Pt!<TAR

t~1FPt

" Flow in or out of the portfolio at time t"<Pt!<P

t~1(1#RP

t)

FTARt

" Flow in or out of target stocks at time t"<TARt

!<TARt~1

(1#RTARt

)FNONt

" Flow in or out of non-target stocks at time t"FPt!FTAR

t

Turnover ratio/0/~5!3'%5

"

FNONt<NON

t~1

"

(<Pt!<P

t~1(1#RP

t))!(<TAR

t!<TAR

t~1(1#RTAR

t))

<NONt~1

"

(<Pt!<TAR

t)!(<P

t~1!<TAR

t~1)!(<P

t~1RP

t!<TAR

t~1RTAR

t)

<NONt~1

"

(<NONt

!<NONt~1

)!(<Pt~1

RPt!<TAR

t~1RTAR

t)

<NONt~1

.

Because the returns on the portfolio components (RPt, RTAR

t) are unavailable,

we assume they are zero and use the following estimate:

Turnover ratio/0/v5!3'%5

"

(<NONt

!<NONt~1

)

<NONt~1

.

This turnover estimate will be biased upwards as long as <Pt~1

RPt!

<TARt~1

RTARt

'0, which will be true as long as returns are positive. Thus, the test isconservative as long as returns are generally positive since it will be harder toreject the null hypothesis of target holdings turnover being the same as &normal'turnover. The estimates of average annual non-target turnover ratios are 0.2062for SWIB, 0.1103 for CREF, 0.1244 for CalPERS, and 0.1499 for CalSTRS.

References

Barber, B., Lyon, J., 1997. Detecting long-run abnormal stock returns: the empirical power andspeci"cation of test statistics. Journal of Financial Economics 43, 301}339.

Bizjak, J.M., Marquette C.J., 1999. Are shareholders all bark and no bite? Evidence from share-holder resolutions to rescind poison pills. Journal of Financial and Quantitative Analysis,forthcoming.

Black, B., 1997. Shareholder activism and corporate governance in the United States. In: Newman,P. (Ed.), The New Palgrave Dictionary of Economics and the Law, forthcoming.


Brancato, C., 1993. Patterns of institutional investment and control in the USA. Columbia Univer-sity Institutional Investor Project, New York.

Brickley, J., Coles, J., Terry, R., 1994. Outside directors and the adoption of poison pills. Journal ofFinancial Economics 35, 371}390.

Carleton, W., Nelson, J., Weisbach, M., 1998. The in#uence of institutions on corporate gover-nance through private negotiations: evidence from TIAA-CREF. Journal of Finance 53,1335}1362.

Comment, R., Schwert, G.W., 1995. Poison or placebo? Evidence on the deterrence and wealthe!ects of modern anti-takeover measures. Journal of Financial Economics 39, 3}43.

DeAngelo, H., DeAngelo, L., 1989. Proxy contests and the governance of publicly held corporations.Journal of Financial Economics 23, 28}59.

Dodd, P., Warner, J., 1983. On corporate governance: a study of proxy contests. Journal of FinancialEconomics 11, 401}438.

Gillan, S., Starks, L., 1998. Corporate governance proposals and shareholder activism: the role ofinstitutional investors. Unpublished working paper. University of Texas, Austin.

Gordon, L., Pound, J., 1993. Information, ownership structure, and shareholder voting: evidencefrom shareholder-sponsored corporate governance proposals. Journal of Finance 48, 697}718.

Hawley, J.P., Williams, A.T., Miller, J.U., 1994. Getting the herd to run: shareholder activism at theCalifornia Public Employees Retirement System (CalPERS). Business and the ContemporaryWorld 6, 26}48.

Hermalin, B.E., Weisbach, M., 1998. Endogenously chosen boards of directors and their monitoringof the CEO. American Economic Review 88, 96}118.

Huson, M., 1997. Does governance matter? Evidence from CalPERS intervention. Unpublishedworking paper. University of Alberta, Canada.

John, K., Klein, A., 1995. Shareholder proposals and corporate governance. Unpublished workingpaper. New York University.

Johnson, M., Porter, S., Shackell, M., 1997. Stakeholder pressure and the structure of executivecompensation. Unpublished working paper. University of Michigan, Ann Arbor.

Kahn, C., Winton, A., 1998. Ownership structure, speculation, and shareholder intervention. Journalof Finance 53, 99}129.

Karpo!, J., 1998. Does shareholder activism work? A survey of empirical "ndings. Unpublishedworking paper. University of Washington, Seattle.

Karpo!, J., Malatesta, P., Walkling, R., 1996. Corporate governance and shareholder initiatives:empirical evidence. Journal of Financial Economics 42, 365}395.

Maug, E., 1998. Large shareholders as monitors: is there a trade-o! between liquidity and control?Journal of Finance 53, 65}98.

Mikkelson, W., Partch, M.M., 1988. Withdrawn security o!erings. Journal of Financial andQuantitative Analysis 23, 119}133.

Monks, R.A.G., Minow, N., 1995. Corporate Governance. Blackwell, Cambridge.Murphy, K., Van Nuys, K., 1994. State pension funds and shareholder inactivism. Unpublished

working paper. Harvard Business School, Cambridge, MA.Nesbitt, S., 1994. Long-term rewards from shareholder activism: a study of the &CalPERS e!ect'.

Journal of Applied Corporate Finance 6, 75}80.Opler, T., Sokobin, J., 1997. Does coordinated institutional activism work? An analysis of the

activities of the Council of Institutional Investors. Unpublished working paper. The Ohio StateUniversity, Columbus.

Pound, J., 1988. Proxy contests and the e$ciency of shareholder oversight. Journal of FinancialEconomics 20, 237}265.

Pound, J., 1991. Proxy voting and the SEC. Journal of Financial Economics 29, 241}285.Romano, R., 1993. Public pension fund activism in corporate governance reconsidered. Columbia

Law Review 93, 795}853.


Sharara, N., Hoke-Witherspoon, A., 1993. The evolution of the 1992 shareholder communicationproxy rules and their impact on corporate governance. Business Lawyer 49, 327}358.

Shleifer, A., Vishny, R.W., 1986. Large shareholders and corporate control. Journal of PoliticalEconomy 94, 461}488.

Smith, M., 1996. Shareholder activism by institutional investors: evidence from CalPERS. Journal ofFinance 51, 257}262.

Sommer Jr., A.A., 1992. Corporate governance: the search for solutions. University of San FranciscoLaw Review 26, 695}716.

Strickland, D., Wiles, K., Zenner, M., 1996. A requiem for the USA: is small shareholder monitoringe!ective? Journal of Financial Economics 40, 319}338.

Van Nuys, K., 1993. Corporate governance through the proxy process: evidence from the 1989Honeywell proxy solicitation. Journal of Financial Economics 34, 101}132.

Wagster, J., Prevost, A., 1996. Wealth e!ects of the CalPERS hit list to SEC changes in the proxyrules. Unpublished working paper. Wayne State University, Detroit, MI.

Wahal, S., 1996. Pension fund activism and "rm performance. Journal of Financial and QuantitativeAnalysis 31, 1}23.

Woidtke, T., 1996. Pension fund ownership and valuation e!ects. Unpublished working paper.Texas A and M University, College Station.


The motivation and impact of pension fund...

Documents

Transcript of The motivation and impact of pension fund...