Is an Investing Institution One Shareholder or a Collection of Separate Funds?

• DOI: http://doi.org/10.1111/corg.12159
• Author: Lynn Hodgkinson,Seth Armitage,Alistair Haig
• Date: 01 January 2017
• Published date: 01 January 2017

Is an Investing Institution One Shareholder or a

Collection of Separate Funds?

Seth Armitage*, Alistair Haig and Lynn Hodgkinson

ABSTRACT

Manuscript Type: Empirical

Research Question/Issue: The paper examineswhether the fund managers in a given investing institution behave in a coordi-

nated manner, in terms of their tradingaround the announcementof a major takeoverby a company in which the institution has

two or more separate holdings.

Research Findings/Insights: Ourdata show that many institutional holdingsconsist of subholdings managedby separate fund

managers. We f‌ind that tradingaround takeover announcements is coordinated in a majority of cases,but there is material dis-

agreement within institutions in a substantial minority of cases, depending on how disagreement is measured. This suggests

that blocks,at the level of the institution,do not always exist in the sense of beingcontrolled by a single agent.Institutional own-

ership is less concentrated than it might appear to be from lists of shareholders in annual reports and databases.

Theoretical/Academic Implications: Research in corporategovernance tends to assume implicitlythat an institutional holding

is a single block.Our f‌indings indicate that it is not safe to make thisassumption, especially in the caseof larger blocks which are

more likely to consistof several subholdings. Some typesof research would benef‌it fromusing data at the level of the managed

fund.

Practitioner/Pol icy Implications: There is much discussion about the merits of coordinated shareholder action between

investing institutions, for example in the Kay Review.Thef‌indings imply that there is scope for greater coordination within

institutions as well as across institutions.

Keywords: Corporate Governance, Institutional Investors,Blockholdings, OwnershipConcentration, ShareholderCoordi-

nation, Measures of Agreement

INTRODUCTION

Principle 5 of the UK Financial Reporting Council’s2012

Stewardship Code states that ‘institutional investors

should be willing to act collectively with other investors

where appropriate’(p. 8). The Kay Review (2012) of the invest-

ment industryhighlights the potential for lackof coordination

between fund managers, and advocates morecollective action

toward companies by the institutionsthat have shareholdings

in a given company.1 Boththe Stewardship Code and the Kay

Review have in mind collaboration between asset manage-

ment groups. They assume, implicitly, that fund managers

within a group act collectively. Our results indicate that this

assumption is questionable. This suggests that there is scope

for greater collective action within groups as well as across

groups.

Asset management groups as a whole are by far the most

important category of shareholder in UK listed companies.

However, individual groups rarely own morethan 20 percent

of a given company, and stakes of less than 10 percent are the

norm. They therefore do not seek to achieve outright control

or even a controlling stake. The fragmentation of institutional

holdings is due, in part, to the advantages of holding a diver-

sif‌ied portfolio,but it is seen as a matter for concern by the Kay

Review and others. For example, Tomasic and Akinbami

(2013) discussthe ability of shareholders to constrainmanage-

rial behavior. They note that “any individual fund manager

will have spreadtheir funds around a number off‌irms so that

it will by itself not have suff‌icient shareholdings in any one

f‌irm to have much inf‌luence if it decides to place pressure

upon the company”(p. 152). The problem is worse when a

group’s investments are divided across a number of separate

funds, as is normally the case.

In the corporate governance literature, shareholders with

large blocks are recognized as having an important role in

monitoring and challenging management. The size of a block

potentially matters to company behavior because it affects

voting power and the incentive and ability of the blockholder

to have a “voice,”by havinga say in such matters as strategic

initiatives,the appointment of directorsand managerial remu-

neration. Blockholders can also inf‌luence company managers

*Address forcorrespondence: Seth Armitage,University of Edinburgh BusinessSchool,

29 BuccleuchPlace, Edinburgh EH8 9JS, UK. E-mail:seth.armitage@ed.ac.uk

© 2016 JohnWiley & Sons Ltd

doi:10.1111/corg.12159

20

Corporate Governance: An International Review, 2017, 25(1): 20–40

through their trading decisions, or “voting with their feet,”

especially “exit”by selling the company’sshares.

Identif‌ication of blockholdings at the level of assetmanage-

ment groups is standard in the literature and has gone

unquestioned until recently. No doubt this is partly because

it is groups that appear in lists of major shareholdings in

annual reports and databases. To identify a block in this

way, when studying corporate governance, involves an as-

sumption that the whole of the group’s holding is controlled

by a single individual or team, with a single point of view.

In fact, group holdings often consist of two or more smaller

blocks controlledby different fund managers.A question then

arises about the extent to which the fund managers act collec-

tively. If the group’s holding consists of severalsmaller blocks

controlled by people who act independently of each other,

then the larger block is incorrectly identif‌ied for the purposes

of research. The block does not really exist, although it might

appear to exist from a list of major shareholdings.

The question we ask is whether or not the fund managersin

a group act in a collective manner with respect to a given

investee company. Specif‌ically, the paper examines the extent

to which fund managers collectively agree on whether to

increase, retain, or reduce theirholdings in a company follow-

ing a value-relevant news release about the company. For

example, if all the funds in the group increase their holdings,

we infer that the fund managers in the group act collectively

or at least make the same investment decision to buy. But if

some funds increase their holdings, while others decrease or

sell out completely, we infer that the group does not act in a

unif‌ied manner. Trading of the company’s shares is recog-

nized as one of the two mechanisms by which a shareholder

with a non-controlling stake can exert inf‌luence over man-

agers. We use the announcement of a large takeover as a case

of value-relevant news. To ensure the takeover is of impor-

tance, we only include takeovers where the market value of

the target company is at least 10 percent that of the acquiring

company.

Existing evidence is limited on the coordination, or other-

wise, of the decisions of funds within groups. Jenkinson and

Jones (2004) mention, but do not answer, this question in their

study of the allocation of shares to bidders in initial public offer-

ings (IPOs).2 Morgan, Poulsen, Wolf,and Yang (2011) study the

voting behavior of US mutual funds on proposals made by

shareholders. There are nearly 20 times more individual funds

in their sample than the 94 fund families (asset management

groups). They f‌ind that 49 fund families coordinate the voting

decisions of their funds, and 45 do not. Our paper on the coor-

dination of trading decisions within groups complements the

evidence of Morgan et al. (2011) on coordination of voting

within groups. Between them, the papers provide evidence

on group behavior bearing on both of the mechanisms by

which non-controlling shareholders exert inf‌luence over

managers.

We f‌ind high levels of agreement within groups, particu-

larly for the sub-sample where groupshave only two or three

funds with holdings in a company. Sixty-nine percent of

groups in this sub-sample have at least two funds making

the same trading decision. The level of agreement decreases,

however, as the number of funds with holdings in a given

acquiringcompany increases. Groupswith four or more funds

display coordination in between 42 and 68 percent of cases,

depending on the criterionadopted for what counts as a coor-

dinated tradingresponse. Weexamine whether thereare some

groups whichconsistently display coordinated trading behav-

ior. However, none of the major groups appears to apply a

strict policy of coordination among its fund managers. The

most important of the robustness tests is to re-def‌ine small

trades of less than 10 percent of a fund’s holding as “no

change.”This increases measured agreement somewhat.

Having found thatlevels of agreement differ acrossgroups,

the paper explores why the fund managers within a given

group might make different trading decisions. Possible rea-

sons include differences in the investment styles of the funds

within a group, the net cash f‌lows into or out of each fund,

and the size of the holding in relation to the size of the fund.

The decisions do not appear to be linked to fund style, with

the exception of index funds and venture capital trusts. Index

funds trade more frequently than other styles, but in small

amounts; venture capital trusts trade less frequently than

other styles. Net cash f‌lows have an impact on trading in the

manner expected, but the relative size of the holding makes

no difference to whether a fund trades. We also carry out

regressionsto explain differences in agreementacross groups,

and f‌ind that agreementis negatively related to the number of

funds in a group, and negatively related to the marketcapital-

ization of the acquiring company.

Our f‌indings suggest that governance-related trading is coor-

dinatedwithingroupsinthemajority of cases. But it is not safe

to assume that there is coordination, especially in the case of

larger blocks at group level, which are more likely to consist of

several subholdings. Institutional ownership is less concen-

trated than it might appear to be from lists of shareholders in an-

nual reports and databases. The f‌indings of Morgan et al. (2011)

on voting behavior within groups point to the same conclusion.

Several papers study the rewarding and disciplining of

managers by shareholders through their trading decisions.

Edmans (2009) presents a model in which trading decisions

by informed blockholders affect management by making the

share price ref‌lect information aboutthe company more accu-

rately. The modelshows that if less-informedinvestors tend to

be short-termist, blockholder inf‌luence on the price will pro-

mote value-adding long-term investment by the company.

Thesizeoftheblockmattersinthemodelbecausealarger

block is assumed to provide more incentive for the blockholder

to become informed. Most empirical papers on disciplining

management examine trading at the level of the institution,

and assume implicitly that an institution is a single share-

holder. Helwege, Intintoli, and Zhang (2012) and Parrino, Sias,

and Starks (2003) document that sales by US institutions in a

f‌irm’s shares increase the probability of subsequent involuntary

resignation of the CEO. A sale is measured as a reduction in the

proportion of shares owned by investing institutions in total, or

as a reduction in the number of holdings at institution level. Dis-

content by some fund managers who sell a company’sshares

within an institution may be masked if others within that insti-

tution do not share that discontent and retain their current hold-

ings or even increase them. Examining trading decisions at the

fund manager level would provide a richer measure of views

toward the company.

Gallagher, Gardner, and Swan (2013)examine the potential

for prof‌it from private information by identifying “swing

trade”sequences by large Australian equity funds. This is a

21ONE SHAREHOLDEROR A COLLECTION OF SEPARATE FUNDS?

© 2016 JohnWiley & Sons Ltd Volume 25 Number 1 January 2017