Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal...

35
Hedge Fund Activism and Internal Capital Markets * Sehoon Kim Fisher College of Business The Ohio State University August, 2016 Abstract This paper studies the effects of hedge fund activism on the activity and efficiency of target companies’ internal capital markets. I find that firms targeted by activist hedge funds significantly increase investment cross-subsidies between divisions, pre- dominantly by enhancing the efficiency of their internal resource allocations. Following Schedule 13D filings by activist hedge funds, segment investments of targeted compa- nies become more sensitive to cash flow generated elsewhere in the firm, and this increase in cross-subsidization is primarily driven by the redirection of firm cash flows toward segments with high Tobin’s Q. The increases in the activity and efficiency of internal capital markets due to hedge fund activism are unlikely to be driven by mea- surement errors in Tobin’s Q or changes in unobserved correlations across segments. * I am grateful to Ren´ e Stulz (Chair), Kewei Hou, and Berk Sensoy for their invaluable advice and guidance. I would also like to thank Alon Brav for providing data on Schedule 13D filings. Finance PhD Candidate. 810 Fisher Hall, 2100 Neil Avenue, Columbus OH 43210. E-mail: kim.4295@fisher.osu.edu Preliminary. Please do not circulate or quote without permission from the author. 1

Transcript of Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal...

Page 1: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Hedge Fund Activism and Internal Capital Markets∗

Sehoon Kim†

Fisher College of Business

The Ohio State University

August, 2016‡

Abstract

This paper studies the effects of hedge fund activism on the activity and efficiency

of target companies’ internal capital markets. I find that firms targeted by activist

hedge funds significantly increase investment cross-subsidies between divisions, pre-

dominantly by enhancing the efficiency of their internal resource allocations. Following

Schedule 13D filings by activist hedge funds, segment investments of targeted compa-

nies become more sensitive to cash flow generated elsewhere in the firm, and this

increase in cross-subsidization is primarily driven by the redirection of firm cash flows

toward segments with high Tobin’s Q. The increases in the activity and efficiency of

internal capital markets due to hedge fund activism are unlikely to be driven by mea-

surement errors in Tobin’s Q or changes in unobserved correlations across segments.

∗I am grateful to Rene Stulz (Chair), Kewei Hou, and Berk Sensoy for their invaluable advice andguidance. I would also like to thank Alon Brav for providing data on Schedule 13D filings.†Finance PhD Candidate. 810 Fisher Hall, 2100 Neil Avenue, Columbus OH 43210. E-mail:

[email protected]‡Preliminary. Please do not circulate or quote without permission from the author.

1

Page 2: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

1 Introduction

Since the early 2000s, hedge funds have joined the fray of shareholder activism, transforming

both the means and ends of exacting influence as owners of corporations. With the decline

of activity in the market for corporate control in the 1980s, shareholder initiated proxy pro-

posals and interventions, particularly those carried out by large institutional investors, have

become an important aspect of shareholder monitoring. Among such institutional share-

holders, hedge funds, despite having received much public spotlight and suspicion for their

confrontational tactics, have stood out as successful at imposing lasting change on the value

and operating performance of target companies (see Brav, Jiang, Partnoy, and Thomas

(2008), Bebchuk, Brav, and Jiang (2015), and Brav, Jiang, and Kim (2015)). How do they

bring about such change? An important yet relatively unanswered question is how and to

what extent hedge funds influence firms’ inner operations in the long-run.1 This paper aims

to narrow this gap by exploring how interventions from activist hedge funds impact the ef-

ficiency of the firm’s internal capital allocation decisions.

A widely debated proposition in finance is that diversified firms are valued less than un-

diversified firms, and that this discount is due to the inefficient investment allocation across

divisions (see Lang and Stulz (1994), Berger and Ofek (1995), and Shin and Stulz (1998)).

If this is the case, hedge fund activism may affect the internal capital markets of target

companies for several reasons. A direct reason is that hedge funds care about the firm’s

investment policies as part of their goal to improve the operational efficiency of the firm,

and therefore intervene so that internal capital is directed toward projects with the greatest

investment opportunities. Brav et al. (2008) report that among 1,059 activist events be-

tween 2001 and 2006, improving operational efficiency is stated as the main goal in 131 of

the cases. For example, Brav et al. (2015) show that firms targeted by activist hedge funds

1Only recently has research started to shed light on this aspect. See, notably, Brav et al. (2015) forevidence on the positive impact of hedge fund activism on the production efficiency of target firms.

2

Page 3: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

invest in technologies that increase the productivity of their plants. Furthermore, hedge

funds are concerned with alleviating agency problems and divisional power struggles within

the firm, which are potentially at the roots of inefficient capital allocation across divisions

(see Rajan, Servaes, and Zingales (2000) and Scharfstein and Stein (2000)). Hedge funds

may indirectly affect the efficiency of internal capital markets by addressing such political

and social frictions within the organization, particularly when they do not have the expertise

to intervene in the firm’s technologies, operations, and assets. Consistent with this notion,

Klein and Zur (2009) show that hedge funds focus on addressing agency problems associated

with cash flows, in which sense hedge funds have similar governance oriented objectives as

other activist shareholders (see Hartzell and Starks (2003) and Ertimur, Ferri, and Muslu

(2011)). Whichever the underlying rationale may be, it is reasonable to expect that a more

efficient internal capital market may arise as a result of activist campaigns by hedge funds,

if they indeed intend to increase the value of the firm as they purport.

Alternative views in the conglomerates literature challenge the assertion that internal cap-

ital markets are inefficient, by arguing that the empirically observed diversification discount

and internal capital allocations are consistent with optimal firm behavior (see Maksimovic

and Phillips (2002, 2013) and Gomes and Livdan (2004)); that the diversification discount

is related to the reduction of risk and uncertainty about the firm’s prospects rather than

operational inefficiencies (see Mansi and Reeb (2002), and Hund, Monk, and Tice (2010));

or that the conclusions about internal capital market efficiency and the value of diversified

firms are products of selection biases or measurement errors (see Whited (2001), Campa and

Kedia (2002), Graham, Lemmon, and Wolf (2002), Villalonga (2004), and Custodio (2014)).

If there really is no inefficiency in the way resources are allocated within conglomerates,

interventions by activist hedge funds should have little impact on how diversified firms use

their internal capital markets. At the least, significant shifts in resource allocations follow-

ing activist campaigns would imply that hedge funds take the view that the internal capital

3

Page 4: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

markets of their target companies are inefficient.

A good example illustrating the intent and consequences of hedge fund activism is the

case of Pershing Square Capital Management, an activist hedge fund led by Bill Ackman,

and Canadian Pacific Railway, a major railroad company in which the hedge fund acquired

a 14% stake at the end of 2011. In early 2012, Mr. Ackman launched an activist campaign,

pushing for changes in the company. His complaints were focused on the lack of operating

efficiency, low asset utilization, substantial underinvestment, failed acquisitions, and sluggish

market share growth. Mr. Ackman argued that the incumbent CEO and the CEO-friendly

board were inept at addressing such problems, and pushed for change of management with

the full support of the shareholders of the company. After the new CEO, industry legend

Hunter Harrison, and board members chosen by Mr. Ackman were brought in, significant

operational improvements took place, including substantially more investment allocations to

areas with high growth opportunities such as intermodal services and the firm’s rail network

covering North Dakota’s Bakken shale region. From September 2011 to December 2014,

Canadian Pacific’s stock price rose from around $49 to $220.

Overall, theory and evidence provide tension for the hypothesis that the internal capital

markets of firms targeted by activist hedge funds may become more active and efficient. To

investigate this hypothesis, I implement a difference-in-differences framework to analyze the

differential impact of hedge fund activism on segment investment with respect to the cash

flows and investment opportunities of business segments. As in previous studies of hedge

fund activism, I rely on Schedule 13D filings submitted by hedge funds to the SEC to iden-

tify targeted companies. Many of these papers examine short-term stock price movements

surrounding the 13D filing date, or study the actions of hedge funds and their impact on the

performance of target companies over a short period of time mostly before they exit their

investments (see Brav et al. (2008), Klein and Zur (2009), Brav et al. (2015), and Gantchev,

Gredil, and Jotikasthira (2015)). In contrast, I abstract from the investment horizon of hedge

4

Page 5: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

funds and analyze the long-term impact that they have on the internal capital markets of

targeted companies.

Consistent with the hypothesis that activist hedge funds push firms to facilitate and opti-

mize internal capital markets, I find that investments made by the segments of targeted firms

become significantly more sensitive to cash flows generated in other parts of the company

after it first becomes an investment target in a Schedule 13D filing, and that most of this

increase in cross-subsidization is driven by reallocating the firm’s cash flows toward segments

with high Tobin’s Q. The sensitivity of a typical segment’s investment to cash flow generated

by other segments of the firm increases by 0.046, much of which is contained among seg-

ments with the highest Tobin’s Q in the firm: It increases by 0.056 more for the highest Q

segments than for segments that have lower Q. I interpret the results as evidence that hedge

funds push firms to rectify their internal capital markets such that cash flows generated in

various segments of the firm are redirected toward the segment with the greatest investment

opportunities.

To mitigate the concern that segments with the highest Tobin’s Q do not correctly iden-

tify those with the greatest investment opportunities due to measurement errors in Tobin’s

Q (see Whited (2001)), I show that the results are robust to categorizing segments with

respect to their Tobin’s Q in several ways. Separately estimating the effect of hedge fund

activism on cross-subsidization using dummy variables for (1) highest Q segments, (2) above

median Q segments, (3) above average Q segments, and (4) lowest Q segments within firms

all corroborate the conclusion that hedge funds influence firms to redirect cash flows to seg-

ments with the greatest investment opportunities. The results also hold for an industry-size

matched sample and a propensity score matched sample, making it less likely that inferences

about the effects of hedge fund activism are confounded by systematic differences between

targeted and non-targeted firms. Finally, I alleviate the concern that estimated changes in

cross-subsidization may in fact be a manifestation of changes in unobservable correlations

5

Page 6: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

across the investment opportunities of segments (see Chevalier (2004)), by showing that

there is little evidence that targeted firms refocus their businesses by reducing the number

of their business segments or industries they serve.

This paper contributes to the literature on active monitoring by outside shareholders,

and the nascent field of hedge fund activism in particular. Earlier studies cast doubt on the

effectiveness of activist campaigns in the 1980-1990s carried out by institutional shareholders

such as mutual funds and pension funds. Karpoff, Malatesta, and Walkling (1996), Smith

(1996), Wahal (1996), and Gillan and Starks (2000), for instance, suggest that the impact

of interventions by such activist institutions on firm value and operating performance are

modest or non-existent. This has recently changed with the advent of activist hedge funds.

Brav et al. (2008), Bebchuk et al. (2015), and Brav et al. (2015) collectively show that

interventions by hedge funds lead to increased operating performance, CEO turnover, and

production efficiency in target firms, which are accompanied by large abnormal returns that

are not reversed in the near future. Moreover, Gantchev et al. (2015) find that hedge fund

activism has spillover effects on non-targeted firms as well. This paper sheds more light on

the real effects of their interventions.

The results of the paper also provide new insight into the workings of internal capital

markets, namely that outside shareholders can influence the efficiency at which they oper-

ate. Much focus in the internal capital markets literature has been placed on their general

lack of efficiency (see Shin and Stulz (1998)), and ample theory and evidence have been

put forth explaining the role of agency conflicts and intra-firm politics therein (see Rajan

et al. (2000), Scharfstein and Stein (2000), Gertner, Powers, Scharfstein (2002), Ozbas and

Scharfstein (2010), Duchin and Sosyura (2013), and Glaser, Lopez-de-Silanes, and Sautner

(2013)). My findings suggest that the fates of internal capital markets can be altered by

force of actively monitoring shareholders. While more research can be done on this front,

the results of this paper are consistent with activist hedge funds removing social and political

6

Page 7: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

barriers within the firm that prevent capital from being allocated efficiently.

In Section 2, I discuss the empirical strategy of the paper. Key results are presented in

Section 3. Finally, I conclude in Section 4.

2 Empirical Strategy

2.1 Data

To investigate the effects of activist interventions from hedge funds on the efficiency of

internal capital markets, I examine whether cross-divisional investment subsidies within the

firm become more aligned with division-level investment opportunities after a firm is targeted

by an activist hedge fund.

Hedge fund activism is identified using Schedule 13D filings, or “beneficial ownership

reports”, submitted to the SEC. Section 13(d) of the 1934 Securities Exchange Act stipulates

that investors who (1) own more than 5% of a voting class of a company’s equity securities

and (2) intend to influence control of the issuer must disclose the amount and intent of

ownership within 10 days of acquiring such a stake. The investor can file a shorter 13G filing

in lieu of a 13D in the absence of the intent to control, which implies that a Schedule 13D

filing meaningfully indicates an active intervention to follow. 13D filers are narrowed down

to hedge fund managers based on the identity descriptions of the reporting entities. I then

take a firm to be targeted by hedge fund activism, with a dummy variable set equal to 1

(HFA= 1) and 0 otherwise, if a Schedule 13D had initially been filed with the firm as the

investment target at least a year earlier.

The sample of my study covers the Compustat universe of firms that report segment

level information over the period 1996 to 2012.2 Segment level financial data are obtained

2I start the sample period in 1996 to ensure a minimum gap of two years with 1994, the year of the firstavailable 13D filing date, so that biases arising from unobserved activist events before 1994 are balancedwith a long enough sample period.

7

Page 8: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

from Compustat business segment files, and firm level information are supplemented from

the Compustat fundamentals annual database. Segment level investment and cash flows are

both normalized by total assets of the firm, since a dollar of cash flow should have the same

impact on segment investment after controlling for investment opportunities, regardless of

where the cash flow originates within the firm.3 Detailed variable descriptions and screening

procedures to address various reporting biases are elaborated in the data appendix.

Table 1 presents a snapshot of the data. There are 875 (2,413) firms (segments) of which

95 (280) are targeted by activist hedge funds.4 Targeted firms have an average asset size

of $ 2,065 million, while non-targeted firms have $ 5,412 million. This sharp difference in

firm size is consistent with the notion that hedge funds are less likely to target bigger firms

because the large amount of capital it takes to acquire a 5% stake in a large company could

introduce significant portfolio risk. Firms that become targets tend to hold smaller cash

balances, be more indebted, be more profitable, have greater cash flow, have lower Tobin’s

Q, have lower sales growth, invest less, and be more diverse in terms of how dispersed their

segment sales are.5 Targeted firms also tend to have fewer segments, consistent with them

being smaller firms. Furthermore, there are 115 distinct hedge funds that invest in these

diversified firms, and the Herfindahl-Hirschman index of funds based on either target size or

target numbers is very small (0.10 and 0.06, respectively), indicating that it is unlikely the

case that a small number of major funds might drive the results of the analysis.

3To the extent that firms can tap into their internal capital markets when they are credit-constrained,the cash flow of a segment should affect the investment of the firm only through its impact on firm cash flow(see Shin and Stulz (1998)).

4Due to the segment and industry-year fixed effects included in the baseline specification, I require thatthere are more than one observation for each segment, and more than one segment within each industry-year.

5Lang and Stulz (1994) and Comment and Jarrell (1995), for instance, use the Herfindahl index computedfrom segment sales as a measure of firm diversification.

8

Page 9: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Table 1. Sample Statistics

This table presents summary statistics for the final sample over the period 1996 to 2012. The number offirms and segments are shown for the full sample, the subsample of firms targeted by hedge fund activism,non-targeted firms, and non-targeted control samples matched on industry (exact match on 2-digit SICcode) and firm size or propensity scores (closest match). Propensity scores are predicted values from alogistic regression of a hedge fund activism dummy (HFA) on lagged firm level Tobin’s Q, cash flow, assetsize, cash holdings, and long-term debt. Average firm characteristics such as asset size, cash holdings overassets, long-term debt over assets, return on assets, cash flow, Tobin’s Q, sales growth, capital expenditureover assets, within-firm Herfindahl-Hirschman indices of segment level sales, and number of segments arepresented. t-statistics on the difference of means between the targeted vs non-targeted sample and targetedvs matched control samples are shown as well. The number of activist hedge funds in the sample and theirHerfindahl-Hirschman indices in terms of target asset size and number of targets are shown at the bottom.

Full Non- Size P -ScoreSample Treated Treated (t-stat) Matched (t-stat) Matched (t-stat)

Firms 875 95 780 149 254Segments 2,413 280 2,133 437 753

Firm CharacteristicsSize 4,908 2,065 5,412 (-4.23) 1,525 (1.60) 2,168 (-0.32)

Cash 0.07 0.07 0.07 (-0.72) 0.08 (-1.04) 0.08 (-1.96)

Debt 0.24 0.26 0.23 (3.01) 0.26 (-0.07) 0.22 (3.33)

ROA 0.01 0.01 0.01 (0.84) 0.02 (-0.70) 0.02 (-0.58)

Cash Flow 0.12 0.12 0.12 (0.57) 0.13 (-1.32) 0.12 (0.09)

Tobin’s Q 1.47 1.40 1.49 (-1.99) 1.43 (-0.40) 1.41 (-0.13)

Sales Growth 0.23 0.13 0.24 (-0.78) 0.15 (-0.42) 0.14 (-0.34)

Capital Expenditure 0.06 0.06 0.07 (-2.07) 0.06 (-0.46) 0.06 (-0.37)

HHI of Segment Sales 0.44 0.43 0.44 (-0.85) 0.44 (-0.44) 0.45 (-1.00)

Number of Segments 2.87 2.78 2.89 (-2.05) 2.69 (1.36) 2.79 (-0.09)

Funds 115HHI (Target Size) 0.10HHI (No. of Targets) 0.06

2.2 Endogeneity

While observable firm characteristics are explicitly controlled for in my analysis, the signif-

icant differences between the firm characteristics of targeted and non-targeted firms raise

9

Page 10: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

the concern that the distribution of these characteristics may be different between the two

groups. One may then suspect that they may also be inherently different in unobservable

ways. If that is the case, the observed effects of hedge fund activism on internal capital mar-

kets may potentially include confounding influences from those unobservable dimensions that

are correlated with hedge fund activism. The primary reason this endogeneity is a concern

is that it then becomes unclear whether it is the action of hedge funds that causes internal

capital markets to become more active and efficient, or if fund managers simply choose to

invest in firms where such changes are imminent even without any activist intervention.

Prior research on the effects of hedge fund activism largely refutes the possibility that the

latter might be the case on average. Notably, hedge fund activist campaigns are associated

with changes in target companies that are unlikely to take place in the absence of pressure

from activists, such as a sharp increase in the firm’s CEO turnover rate, from less than

6% to over 12% over the 2001 to 2006 period (see Brav et al. (2008)). Activist campaigns

also entail significant costs for the activist investor: Campaigns that end in proxy fights are

estimated to cost on average near $11 million (see Gantchev (2013)). Furthermore, studies

that compare announcement returns of 13D filings with 13G filings document higher returns

for activist stake disclosures than for passive ones (see Clifford (2008) and Klein and Zur

(2009)). These findings cannot be explained if the average activist hedge fund engages in

passive stock picking rather than active engagement. In essence, the evidence to date sug-

gests it is unlikely that the endogeneity problem will seriously confound the results of this

study.

Econometrically, it is a difficult if not infeasible task to rule out the stock picking story by

estimating the average treatment effect of hedge fund activism in a randomized experiment,

because hedge funds are likely to choose their battles in firms where there is a clear problem

they want to solve and where they can readily effect change to their liking. In the absence

of an instrument for hedge fund activism or an exogenous shock thereof, a partial remedy

10

Page 11: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

is to use a matched sample where targeted and non-targeted firms are reasonably similar so

that treatment by hedge fund activism is the only source of observable variation across the

two groups. The hope is that once the two groups are made to be similarly distributed in

observable ways, other systematic variations will be minimized, except for whether a firm is

targeted by hedge fund activism. For this purpose, I construct two matched samples that

employ only non-targeted firms that are similar to target firms as the control group. Each

year, targeted firms are matched to non-targeted companies in the same two-digit SIC in-

dustry that are closest according to some metric. In the first matching method, firms are

matched by asset size, where the difference (computed as|ATtargeted−ATcontrol|

(ATtargeted+ATcontrol)/2) is required to

be smaller than 50%. In the second method, firms are matched by their propensity scores,

which are computed as the predicted values from a pooled logistic regression of a hedge fund

activism dummy (HFA) on a set of lagged firm level variables (i.e. Tobin’s Q, cash flow, asset

size, cash holdings, and long-term debt). These industry-size or propensity score matched

non-targeted firms are then used as the control group.

The last four columns of Table 1 show the characteristics of these matched firms. There

are 149 (254) and 437 (753) industry-size (propensity score) matched control firms and seg-

ments, respectively. What is notable is that across the various firm characteristics, there is no

longer a significant difference between matched control firms and targeted firms. For exam-

ple, the average asset size of non-targeted firms drops from $ 5,412 million to $ 1,525 million

after the industry-size matching, and the difference with targeted firms, which is significant

with a sizable t-statistic of -4.23 before matching, becomes insignificant (t-statistic=1.60).

Characteristics other than asset size are also well matched as a result of the industry-size

matching scheme. The difference in long-term debt (t-statistic goes from 3.01 to -0.07), To-

bin’s Q (-1.99 to -0.40), capital expenditure (-2.07 to -0.46), and number of segments (-2.05

to 1.36) all become insignificant after matching on industry and size. This suggests that the

industry-size match does a good job at making targeted and non-targeted firms comparable.

11

Page 12: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

The propensity score matching scheme also does a reasonable job at minimizing the differ-

ences between targeted and non-targeted firms for most characteristics, with the exception

of long-term debt. I conduct my analysis using the full sample as well as the industry-size

and propensity score matched samples, and report results for each case.

2.3 Methodology

I employ these samples to test the hypothesis that hedge fund activism makes the sensitivity

of a division’s investment to cash flow generated elsewhere in the firm more responsive to

whether the division has greater opportunities than other segments of the firm. To this end,

I implement a difference-in-differences regression of segment investments (INVS ) on an indi-

cator variable for hedge fund activism treatment (HFA), a set of explanatory variables meant

to capture the extent and efficiency of investment cross-subsidies across divisions, and the

interaction terms between those variables and the HFA dummy variable. These explanatory

variables include (1) the segment’s own cash flow, (2) cash flows from all the other segments

within the firm, and (3) their interactions with an indicator variable for whether the segment

has the greatest investment opportunities among all segments of the firm.

An important hurdle in this approach is the measurement of segment level investment

opportunities. While Tobin’s Q is conventionally used as a measure of investment opportu-

nities for firms, it is not explicitly calculable for segments because segment market values are

unavailable. As widely done in the conglomerates literature to circumvent this challenge, I

use the median Tobin’s Q of “pure-play” single-segment firms in the same industry as the

segment, defined at the two-digit SIC level, as a proxy for the segment’s Tobin’s Q (see Shin

and Stulz (1998), Rajan et al. (2000), and Ozbas and Scharfstein (2010) for notable papers

that employ this methodology).

This method by itself does not resolve the potential bias due to the measurement er-

ror in Tobin’s Q. As discussed by Whited (2001) and many others, observable measures of

12

Page 13: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Tobin’s Q can diverge substantially from the actual unobservable measure of investment

opportunities implied by standard intertemporal models of investment: marginal Q. In the

context of the difference-in-differences framework used in this study, this can be a problem

if the measurement error in Tobin’s Q distorts the ordering of investment opportunities of

segments within firms, or is somehow correlated with whether a firm is targeted by hedge

fund activism. While it is difficult to imagine why measurement errors would correlate with

hedge fund activism, it is plausible that such errors could result in falsely categorizing a

segment as having the highest investment opportunities when in fact it does not.

One way to address this issue is to define whether a segment has high or low Tobin’s Q

compared to other segments of the firm in different ways, and check if the results from those

alternative Q categorizations convey a consistent message. I alternatively identify segments

that have Tobin’s Q greater than the firm’s median and average and show similar results

with identifying those with the highest Tobin’s Q. Furthermore, identifying segments with

the lowest Tobin’s Q yields sharply opposite results. These mitigate the concern that a

segment might be falsely identified as having the greatest investment opportunities in the

firm because of measurement errors. Also, I use lagged segment sales growth as an addi-

tional control measure of segment investment opportunities that is based on data from the

segment.

A related issue that often makes it difficult to estimate the extent of internal capital

market transfers is that apparent cross-subsidization may in truth be a manifestation of

unobserved correlations of investment opportunities across segments (see Chevalier (2004)).

Using Tobin’s Q and sales growth to explicitly control for segment investment opportunities

partially alleviates this problem, only to the extent that they accurately measure invest-

ment opportunities. The difference-in-differences framework overcomes this limitation and

further remedies the problem, insofar as the relatedness between segments do not change

substantially after firms are targeted by hedge fund activism. However, it is a legitimate

13

Page 14: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

concern that hedge funds may indeed influence how close segments are to each other by

making the firm more focused, rendering the difference-in-differences specification invalid for

the purpose of testing internal capital market activity. After all, it is well known from pre-

vious research that activist shareholders often push for the divestiture of underperforming

assets (see Bethel, Liebeskind, and Opler (1998), Brav et al. (2008), Klein and Zur (2009),

Bebchuk et al. (2015), and Brav et al. (2015)). In the next section, however, I show that

activist interventions do not appear to be associated with significant changes in the number

of business segments operated by firms or industries served by them. It appears that while

hedge funds do push firms to divest unprofitable assets, they seldom go so far in eventuality

as to shut down or sell off entire industry segments.

The difference-in-differences specification can be written as:

INV Si,j,t = α + β ·HFAi,t

+ δ1 · CFi,−j,t + δ2 · CFi,j,t

+ ϕ1 · CFi,−j,t ×HFAi,t + ϕ2 · CFi,j,t ×HFAi,t

+ γ1 · (CFi,−j,t ·HighQi,j,t−1) + γ2 · (CFi,j,t ·HighQi,j,t−1)

+ λ1 · (CFi,−j,t ·HighQi,j,t−1)×HFAi,t + λ2 · (CFi,j,t ·HighQi,j,t−1)×HFAi,t

+ µ′ ·Xi,j,t−1 + τ ′ ·Yi,t−1 + ai,j + bt + εi,j,t

where INV Si,j,t denotes gross investment of the jth segment of firm i during year t, scaled

by firm i’s total assets as of year t− 1. HFAi,t indicates whether it has been at least 1 year

as of year t since firm i had first become a target of hedge fund activism. CFi,j,t, or CF,

denotes the cash flow of segment j of firm i during year t, scaled by t− 1 total assets as of

the firm. CFi,−j,t, or Other CF, denotes the cash flows of all segments of firm i other than

segment j. HighQi,j,t−1 indicates whether the segment has the highest Tobin’s Q among all

segments of the firm. Xi,j,t−1 is a vector of lagged segment level control variables such as

14

Page 15: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

segment Tobin’s Q and sales growth, and Yi,t−1 includes firm level controls such as asset

size, profitability, cash holdings, and leverage. Segment and industry-year fixed effects are

controlled for as well.

The baseline specification captures the activity and efficiency of internal capital markets,

and more importantly, the impact of hedge fund activism on both of these aspects. For

example, the coefficient δ1 quantifies the increase in segment level investment with respect

to an incremental increase in cash flows generated elsewhere in the firm, hence the level of

activity of internal capital markets. γ1 measures how greater the investment cross-subsidy

is when the segment has the greatest investment opportunities in the firm. This coefficient

assesses how efficient the firm’s internal capital market is by estimating how much more

capital is allocated toward the greatest opportunities.

Most critically, ϕ1 and λ1 capture the impact of hedge fund activism on the activity

and efficiency of internal capital markets, respectively. ϕ1 evaluates the effect of hedge fund

activism on the sensitivity of segment level investment to cash flow from other segments in

the firm. λ1 allows for a different coefficient on the interaction of other segments’ cash flow

and HFA for segments with the highest Q in the firm. The idea is that if activist hedge

funds have a real effect on the capital allocation between divisions of target companies, one

would expect either of the coefficients ϕ1 or λ1 to be economically large and statistically

significant.

In the next section, I explain the key results from implementing this methodology on the

full sample as well as on the industry-size matched sample and the propensity score matched

sample.

15

Page 16: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Table 2. Hedge Fund Activism and Internal Capital Markets

This table presents results from difference-in-differences regressions of segment investments (INVS ) on atreatment indicator variable for whether the firm is a target of hedge fund activism (HFA), a set of explana-tory variables: other segments’ cash flow (Other CF ); own segment cash flow (CF ); their interactions witha dummy variable indicating whether the segment has the highest Tobin’s Q among all segments of the firm(High Q), and the interaction terms between these explanatory variables and HFA. All specifications includesegment and industry-year fixed effects, where industry is defined as the segment’s 4-digit SIC code. SegmentQ and sales growth, and firm asset size (log of assets), profitability (ROA), cash holdings and long-term debt(fraction of assets) are included as control variables. Standard errors adjusted for clustering at the firm-yearlevel are reported in parentheses. (*** p<0.01, ** p<0.05, * p<0.1)

Dependent Variable: INVS

Other CF × High Q × HFA 0.056**(0.025)

CF × High Q × HFA -0.007(0.046)

Other CF × HFA 0.046** 0.028(0.023) (0.024)

CF × HFA -0.044 -0.038(0.032) (0.035)

HFA 0.004 0.003(0.004) (0.004)

Other CF × High Q -0.001 -0.005(0.009) (0.009)

CF × High Q 0.003 0.005(0.018) (0.019)

Other CF 0.027*** 0.022** 0.017* 0.023** 0.019*(0.010) (0.010) (0.010) (0.010) (0.010)

CF 0.085*** 0.080*** 0.084*** 0.078*** 0.082***(0.013) (0.013) (0.014) (0.015) (0.016)

No. Obs. 7,893 7,893 7,893 7,893 7,893Segment Controls Yes Yes Yes Yes YesFirm Controls No Yes Yes Yes YesSegment / Industry-Year FE Yes Yes Yes Yes YesAdj R2 0.699 0.703 0.704 0.702 0.704

16

Page 17: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

3 Key Results

Table 2 shows the full sample results from the difference-in-differences framework described

in the previous section. In the first and second columns, I exclude the treatment indicator

for hedge fund activism (HFA) and analyze the effects of a segment’s cash flow and cash flow

from other segments on the segment’s investment in an OLS framework, controlling for the

segment’s investment opportunities measured by Tobin’s Q and sales growth. Consistent

with Shin and Stulz (1998), I find that increases in both the segment’s own cash flow and

other segments’ cash flow lead to significantly more segment investment, but that the effect

of the segment’s own cash flow is much larger in magnitude. The coefficient on the segment’s

own cash flow (0.080) is 3.6 times larger than that on other segments’ cash flow (0.022). The

central question is whether this relative inactivity of cash flow transfers across segments, or

internal capital market failure, is resolved by the intervention of activist hedge funds, and if

so, whether the resolution is attained by increasing the efficiency of internal capital markets.

The difference-in-differences result reported in the third column of Table 2 indicates that

hedge fund activism has an effect of facilitating internal capital markets. The coefficient

on the interaction term between other segments’ cash flow (Other CF ) and the hedge fund

activism dummy variable (HFA) implies a positive and significant impact of hedge fund

activism on the sensitivity of segment investment to other segments’ cash flow: INVS -to-

OtherCF sensitivity increases on average by 0.046 (with a firm-year clustered standard error

of 0.023). On the other hand, hedge fund activism does not appear to affect the sensitivity of

the segment’s investment to its own cash flow. Taking into account the fact that a segment’s

investment is much more sensitive to its own cash flow than to other segments’ cash flow to

begin with, the evidence corroborates the idea that hedge funds come in to remove forces

that prevent resources from flowing from one segment to another. Moreover, the interaction

term between other segments’ cash flow and hedge fund activism partially subsumes the

17

Page 18: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

effect of other segments’ cash flow alone, which means that internal capital markets are even

less active without such activist interventions.

Why are internal capital markets inactive in the first place, and how do hedge funds help

facilitate them? The canonical answer to the first question provided by the literature is that

internal capital markets fail because they do not actively direct internal resources of the firm

to their best use (see Shin and Stulz (1998), Rajan et al. (2000), Ozbas and Scharfstein

(2010), Duchin and Sosyura (2013), and Glaser et al. (2013)). The fourth column of Table

2 shows results consistent with this understanding. Here, the segment’s own cash flow and

other segments’ cash flow are interacted with an indicator variable for whether the segment’s

Tobin’s Q is the highest in the firm (High Q). The coefficient indicates there is no evidence

that even the segment with the greatest investment opportunities receives more internal re-

sources than any other segment does. The crux is, do hedge funds make firms rectify this?

To examine how firms targeted by activist hedge funds activate internal capital markets,

the last column of Table 2 interacts the HFA dummy with the segment’s own cash flow,

other segments’ cash flow, and those for the highest Q segments separately. Confirming

that hedge funds induce firms to redirect cash flows toward the greatest investment oppor-

tunities within the firm, the coefficient for the interaction term between Other CF, High Q,

and HFA is significantly positive. Following hedge fund activism, the sensitivity of segment

investment to other segments’ cash flow for the highest Q segments increases on average

by 0.056 (with a firm-year clustered standard error of 0.025) more than for segments that

do not have the highest Tobin’s Q. More importantly, it subsumes the coefficient on the

interaction term between Other CF and HFA, providing evidence that hedge funds facilitate

the internal capital markets of firms predominantly by making them more efficient. They

push the firm’s internal capital market to redirect cash flows generated in various segments

toward the segment with the highest Tobin’s Q.

As mentioned earlier in the paper, a potential problem with Tobin’s Q-based measures

18

Page 19: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Table 3. Internal Capital Market Efficiency: Alternative High Q Definitions

This table presents results from diff-in-diff regressions of segment investments (INVS ) on a hedge fundactivism dummy (HFA), a set of explanatory variables: other segments’ cash flow (Other CF ); own segmentcash flow (CF ); their interactions with a High Q dummy, and the interaction terms between these explanatoryvariables and HFA. Results are presented for three alternative definitions of High Q : whether the segment’sTobin’s Q is (1) above the median segment’s Tobin’s Q ; (2) above average; (3) lowest among all segments ofthe firm. All specifications include segment and industry-year fixed effects. Segment Q and sales growth, andfirm asset size, profitability, cash holdings, and long-term debt are included as control variables. Standarderrors are adjusted for clustering at the firm-year level. (*** p<0.01, ** p<0.05, * p<0.1)

Alternative High Q Definitions

Dependent Variable: INVS Above Median Above Average Lowest Q

Other CF × High Q × HFA 0.068** 0.065*** -0.078***(0.027) (0.025) (0.026)

CF × High Q × HFA 0.011 0.008 0.001(0.045) (0.044) (0.044)

Other CF × HFA 0.017 0.024 0.072***(0.025) (0.024) (0.027)

CF × HFA -0.049 -0.045 -0.046(0.037) (0.036) (0.038)

HFA 0.003 0.003 0.004(0.004) (0.004) (0.004)

Other CF × High Q -0.011 -0.006 0.006(0.009) (0.008) (0.009)

CF × High Q -0.013 -0.002 0.018(0.019) (0.018) (0.020)

Other CF 0.023** 0.020* 0.014(0.011) (0.010) (0.010)

CF 0.091*** 0.085*** 0.074***(0.017) (0.016) (0.017)

No. Obs. 7,893 7,893 7,893Control Variables Yes Yes YesSegment / Industry-Year FE Yes Yes YesAdj R2 0.704 0.704 0.704

of investment opportunities is that they may contain measurement error. Consequently, the

High Q indicator variable might falsely assign a value of 1 to a segment that is in fact not

19

Page 20: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

the one with the highest marginal Q. If Tobin’s Q happens to diverge greatly from marginal

Q, this will be a problem. I address this issue by using alternative definitions of High Q : The

segment’s Tobin’s Q is (1) above the median Tobin’s Q of the firm’s segments; (2) above

the average Tobin’s Q of firm segments; (3) lowest among all segments of the firm. The

first two alternative definitions allow more room for error in estimated Tobin’s Q, because

there are less likely to be errors in judging whether a segment has greater investment op-

portunities than most segments of the firm, compared to judging whether the segment has

greater opportunities than all the other segments. The third alternative definition, that the

segment has the lowest, not highest, Tobin’s Q, corroborates the baseline results by provid-

ing evidence going in the opposite direction. The combined results alleviate concerns about

measurement errors, since it is unlikely that high Tobin’s Q coincides with low marginal Q

and at the same time low Tobin’s Q corresponds to high marginal Q, on average.

Table 3 shows the results from this analysis. In the first and second columns, I define

High Q to equal 1 if the segment has Tobin’s Q higher than the median and average of

the firm’s segments, respectively, and 0 otherwise. The results are similar to that shown in

the last column of Table 2. In both cases, the coefficient on the interaction term between

other segments’ cash flow, High Q, and HFA is larger in magnitude: 0.068 (significant at the

5% level) and 0.065 (significant at the 1% level) for the above median and above average

High Q definitions, respectively. As in Table 2, hedge fund activism impacts the sensitivity

of segment investment to other segments’ cash flow only for High Q segments, but not for

below median and below average Q segments. In the third column, I assign segments with

the lowest Tobin’s Q in the firm with a value of 1 and assign 0 for all other firms. Not

surprisingly, hedge fund activism has a much lower impact on the sensitivity of segment

investment to other segments’ cash flow for the lowest Q segments, compared to the impact

it has on other segments of the firm with better prospects. The coefficient on the interaction

term between other segments’ cash flow and HFA is 0.072 which is highly significant at the

20

Page 21: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

1% level, and the coefficient differs by -0.078 (also significant at 1%) for the lowest Q seg-

ments. In essence, the intervention of hedge funds ensure that resources are directed toward

the most promising divisions and not to places where internal capital may be wasted.

Another complication discussed earlier is that targeted and non-targeted firms may be

systematically different from each other. This confounds inference about the causal effect

of hedge funds, because hedge funds may be passively picking companies whose internal

capital markets are likely to become more efficient in the future rather than actively effect-

ing change. In the econometric sense, it is difficult to disentangle this problem. However,

because hedge fund activism is usually associated with high campaign costs to the activist

investor and changes in the company that are likely involuntary (e.g. higher CEO turnover),

the endogeneity concern is a benign issue in the context of this study. Notwithstanding, a

feasible and partial econometric remedy is to construct a matched sample so that targeted

and non-targeted companies are made to be similar at least across observable dimensions

with the hope that systematic differences between the two groups will also be minimized. As

elaborated in the previous section, I construct an industry-size matched sample as well as

a propensity score matched sample. Each year, targeted companies are matched with non-

targeted companies in the same two-digit SIC industry with closest asset size or propensity

scores, which are used as the control sample. As shown in Table 1, the matching procedures

successfully minimize differences between the targeted and non-targeted groups across a va-

riety of observable variables. I then implement the difference-in-differences framework on

each of these matched samples.

Table 4 shows results from this matched sample analysis. The bottom-line is that the

results from the full sample hold up well in both of the matched samples. One notable

difference is that the segment’s own cash flow and other segments’ cash flow alone no longer

account for the segment’s investment policy. The coefficients on CF and Other CF range

from -0.022 to 0.048 and from -0.069 to -0.036, respectively, none of which are statistically

21

Page 22: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Table

4.

Matc

hed

Sam

ple

Diff

ere

nce

-in-D

iffere

nce

sR

egre

ssio

ns

Th

ista

ble

pre

sents

mat

ched

sam

ple

resu

lts

from

diff

-in

-diff

regre

ssio

ns

of

segm

ent

inve

stm

ents

(INVS

)on

ah

edge

fun

dact

ivis

md

um

my

(HFA

),a

set

ofex

pla

nat

ory

vari

able

s:ot

her

segm

ents

’ca

shfl

ow(O

ther

CF

);ow

nse

gm

ent

cash

flow

(CF

);th

eir

inte

ract

ion

sw

ith

aHighQ

du

mm

y,an

dth

ein

tera

ctio

nte

rms

bet

wee

nth

ese

exp

lan

atory

vari

ab

les

an

dHFA

.E

ach

year,

targ

eted

firm

sare

matc

hed

wit

hn

on

-targ

eted

firm

sin

the

sam

etw

o-d

igit

SIC

ind

ust

ryw

ith

clos

est

asse

tsi

ze(P

an

elA

)or

pro

pen

sity

score

s(P

an

elB

).P

rop

ensi

tysc

ore

sare

pre

dic

ted

valu

esfr

om

alo

gist

icre

gres

sion

ofa

hed

gefu

nd

acti

vis

md

um

my

(HFA

)on

lagged

firm

leve

lT

obin

’sQ

,ca

shfl

ow,

ass

etsi

ze,

cash

hold

ings,

an

dlo

ng-t

erm

deb

t.R

esu

lts

are

pre

sente

du

sin

gal

tern

ativ

ed

efin

itio

ns

ofHighQ

:w

het

her

the

segm

ent’

sT

ob

in’s

Qis

(1)

the

hig

hes

t;(2

)ab

ove

med

ian

;(3

)ab

ove

aver

age;

(4)

low

est

inth

efi

rm.

All

spec

ifica

tion

sin

clu

de

segm

ent

an

din

du

stry

-yea

rfi

xed

effec

ts,

as

wel

las

firm

an

dse

gm

ent

leve

lco

ntr

olva

riab

les.

Sta

nd

ard

erro

rsar

ead

just

edfo

rcl

ust

erin

gat

the

firm

-yea

rle

vel.

(***

p<

0.0

1,

**

p<

0.0

5,

*p<

0.1

)

Pan

elA.Indu

stry-SizeMatching

Pan

elB.Propensity

Score

Matching

Alt

ernat

ive

Hig

hQ

Defi

nit

ions

Alt

ernat

ive

Hig

hQ

Defi

nit

ions

Ab

ove

Ab

ove

Ab

ove

Ab

ove

Dep

enden

tV

aria

ble

:IN

VS

Hig

hes

tQ

Med

ian

Ave

rage

Low

estQ

Hig

hes

tQ

Med

ian

Ave

rage

Low

estQ

Other

CF×

HighQ×

HFA

0.11

4**

0.10

9*0.

118*

*-0

.111

**0.

092*

0.09

6*0.

106*

*-0

.098

*(0

.051

)(0

.057

)(0

.052

)(0

.056

)(0

.049

)(0

.054

)(0

.051

)(0

.059

)

CF×

HighQ×

HFA

-0.0

36-0

.014

-0.0

51-0

.015

-0.0

64-0

.048

-0.0

720.

014

(0.0

76)

(0.0

92)

(0.0

88)

(0.0

87)

(0.0

75)

(0.0

83)

(0.0

83)

(0.0

88)

Other

CF×

HFA

0.09

1*0.

044

0.04

00.

045

0.14

8**

0.08

0*0.

044

0.03

20.

038

0.13

5**

(0.0

47)

(0.0

46)

(0.0

52)

(0.0

47)

(0.0

60)

(0.0

42)

(0.0

44)

(0.0

46)

(0.0

44)

(0.0

59)

CF×

HFA

-0.0

11-0

.002

-0.0

030.

010

-0.0

07-0

.064

-0.0

27-0

.034

-0.0

25-0

.063

(0.0

57)

(0.0

60)

(0.0

74)

(0.0

67)

(0.0

66)

(0.0

52)

(0.0

55)

(0.0

65)

(0.0

61)

(0.0

64)

HFA

-0.0

10-0

.006

-0.0

09-0

.008

-0.0

09-0

.001

-0.0

01-0

.002

-0.0

00-0

.002

(0.0

08)

(0.0

08)

(0.0

08)

(0.0

08)

(0.0

08)

(0.0

07)

(0.0

07)

(0.0

07)

(0.0

07)

(0.0

07)

Other

CF×

HighQ

-0.0

13-0

.018

-0.0

070.

030

-0.0

10-0

.004

-0.0

040.

019

(0.0

31)

(0.0

37)

(0.0

32)

(0.0

28)

(0.0

31)

(0.0

37)

(0.0

32)

(0.0

35)

CF×

HighQ

0.12

5**

0.06

50.

124*

-0.0

280.

052

0.02

50.

055

0.02

3(0

.054

)(0

.065

)(0

.067

)(0

.053

)(0

.063

)(0

.070

)(0

.069

)(0

.059

)

Other

CF

-0.0

51-0

.051

-0.0

50-0

.054

-0.0

69-0

.036

-0.0

40-0

.042

-0.0

44-0

.052

(0.0

45)

(0.0

47)

(0.0

51)

(0.0

47)

(0.0

44)

(0.0

48)

(0.0

52)

(0.0

53)

(0.0

52)

(0.0

50)

CF

0.02

9-0

.019

-0.0

10-0

.022

0.04

80.

032

-0.0

000.

008

-0.0

010.

022

(0.0

66)

(0.0

70)

(0.0

78)

(0.0

71)

(0.0

74)

(0.0

44)

(0.0

43)

(0.0

48)

(0.0

44)

(0.0

63)

No.

Obs.

879

879

879

879

879

1,01

71,

017

1,01

71,

017

1,01

7C

ontr

olV

aria

ble

sY

esY

esY

esY

esY

esY

esY

esY

esY

esY

esSeg

men

t/

Indust

ry-Y

ear

FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

AdjR

20.

749

0.75

50.

750

0.75

50.

749

0.74

40.

745

0.74

50.

746

0.74

4

22

Page 23: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

significant. This is consistent with the story that hedge funds tend to target firms where

there are more impediments to active resource reallocation, and since the control samples

are made to have similar characteristics to the targeted sample, the overall level of internal

capital market activity is low for firms in the matched samples. The lack of segment invest-

ment sensitivity to the firm’s cash flow can also be explained by the fact that targeted firms

tend to invest less than non-targeted firms, potentially underinvesting as often accused by

activist hedge funds, and therefore the matched samples which are constructed to be similar

to targeted firms exhibit lower investment to cash flow sensitivity.

The first columns in Panel A (industry-size match) and Panel B (propensity score match)

of Table 4 show that hedge fund activism has a sizable impact on how sensitive a segment’s

investment is to cash flows generated elsewhere within the firm. While the statistical signifi-

cance is somewhat weaker than in the full sample results, potentially due to the substantially

smaller sample size, the economic magnitude is large: a 0.091 (0.080) increase in the seg-

ment investment sensitivity to other segments’ cash flow for the industry-size (propensity

score) matched sample. As shown in the remaining four columns in each panel of Table 4,

the efficiency implications are profound in both of the matched samples. Again, the seg-

ment investment to other segments’ cash flow sensitivity gains are statistically significant

and economically large for segments that have the highest Tobin’s Q in the firm, but not

for other segments. For example, the coefficient on the interaction term between other seg-

ments’ cash flow, High Q, and HFA is a striking 0.114 (with a firm-year clustered standard

error of 0.051) in the industry-size matched sample, and the interaction term between other

segments’ cash flow and HFA is insignificant. This result holds up when using alternative

definitions of High Q. The coefficient on OtherCF × HFA is greater by 0.109 (0.118) for firms

with above median (average) Tobin’s Q compared to firms that have lower Tobin’s Q. In the

last column of each panel, I confirm that segments with the lowest Tobin’s Q in their firms

receive substantially less cash flows from other segments compared to segments that have

23

Page 24: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

higher Tobin’s Q, after being targeted by hedge fund activism. For instance, the sensitivity

of segment investment to other segments’ cash flow in the industry-size matched sample

increases dramatically by 0.148 after being targeted, but lowest Q segments are excluded

from enjoying that cross-subsidization. The difference in the sensitivity increase between

lowest Q segments and other segments is both economically large (-0.111) and statistically

significant (at 5%).

Finally, I provide ancillary evidence that it is not the case that firms targeted by hedge

funds become more focused on average. This suggests that while hedge funds may push

for the divestiture of unprofitable assets, they rarely close entire industry segments in ac-

tuality.6 This is an important point for this study, because if firms become more focused

due to hedge fund activism, it may well be the case that the apparent increase in cross-

subsidization following hedge fund intervention is in fact a product of increased unobserved

correlation between segment investment opportunities.7 In Table 5, I report results from

logistic and probit regressions analyzing whether firms change the number of segments or

segment industries after being targeted by hedge fund activism. I analyze four possibilities.

The dependent variable is a dummy variable set to 1 if the firm (1) reduced the number of

its business segments; (2) increased the number of its business segments; (3) reduced the

number of segment industries; (4) increased the number of segment industries during the

year, and 0 otherwise. Industry is defined at the 2-digit SIC level. The independent vari-

able is an indicator for whether the firm has been the target of activist hedge funds for at

least a year (HFA). Firm Tobin’s Q, cash flow, sales growth, asset size, profitability, cash

holdings, long-term debt, and year dummies are included as control variables. Results are

shown separately for the full sample (Panel A), industry-size matched sample (Panel B), and

6See Bethel et al. (1998), Brav et al. (2008), Klein and Zur (2009), Bebchuk et al. (2015), and Brav etal. (2015) for previous research that suggest that hedge funds aim to divest underperforming parts of targetcompanies.

7The concern that cross-subsidization measured from segment cash flows may in fact be due to correlatedinvestment opportunities across segments was raised by Chevalier (2004).

24

Page 25: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Table 5. Hedge Fund Activism and Firm diversification

This table presents results from logit and probit regressions of dummy variables for whether a firm reduced orincreased the number of reported segments or number of segment industries in a given year, on a treatmentindicator variable for whether the firm had been a target of hedge fund activism (HFA). Industry is definedat the 2-digit industry level. Firm Q, cash flow, sales growth, asset size, profitability, cash holdings, and long-term debt are included as control variables. Year dummies are controlled for in all specifications. Marginaleffects are presented as results. Results are reported for the full sample and industry-size matched sampleseparately. Standard errors are adjusted for clustering at the firm level. (*** p<0.01, ** p<0.05, * p<0.1)

Logistic Regressions Probit Regressions

∆Seg. ∆Seg. ∆SIC ∆SIC ∆Seg. ∆Seg. ∆SIC ∆SIC< 0 > 0 < 0 > 0 < 0 > 0 < 0 > 0

Panel A. Full Sample

HFA -0.009 -0.014* -0.012 -0.011* -0.009 -0.015** -0.012 -0.010*(0.008) (0.008) (0.008) (0.006) (0.008) (0.007) (0.007) (0.006)

No. Obs. 3,878 3,670 3,130 3,700 3,878 3,670 3,130 3,700Pseudo R2 0.055 0.107 0.040 0.068 0.054 0.108 0.039 0.067Prob > χ2 0.004 0.000 0.060 0.000 0.004 0.000 0.090 0.000

Panel B. Industry-Size Matched Sample

HFA 0.018 0.001 0.004 0.003 0.020* 0.001 0.006 0.004(0.012) (0.020) (0.016) (0.015) (0.012) (0.018) (0.015) (0.013)

No. Obs. 460 368 303 379 460 368 303 379Pseudo R2 0.272 0.170 0.234 0.081 0.272 0.179 0.234 0.087Prob > χ2 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Panel C. Propensity Score Matched Sample

HFA 0.010 0.001 0.001 -0.004 0.011 0.000 0.001 -0.004(0.014) (0.017) (0.013) (0.010) (0.012) (0.015) (0.011) (0.009)

No. Obs. 550 475 428 554 550 475 428 554Pseudo R2 0.264 0.141 0.243 0.081 0.262 0.145 0.236 0.084Prob > χ2 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Control Variables Yes Yes Yes Yes Yes Yes Yes YesYear Dummy Yes Yes Yes Yes Yes Yes Yes Yes

propensity score matched sample (Panel C). The coefficients are reported as marginal effects

and standard errors are adjusted for clustering at the firm level.

Contrary to the concern that hedge funds may push the firm to become more focused and

therefore confound inferences about their impact on cross-subsidization, the marginal effects

25

Page 26: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

of hedge fund activism on the probability of reducing the number of business segments or the

number of segment industries are both economically and statistically negligible across the

board. Out of the eight specifications testing whether the number of segments or industries

decrease following hedge fund activism, the only indication that hedge funds make firms

more focused is given by an industry-size matched sample probit regression of whether firms

reduce the number of segments. The marginal effect is a 2% increase in probability which is

significant at the 10% level. On the other hand, there is slightly more evidence that targeted

firms become less likely to further diversify their businesses. Four specifications using the

full sample yield significant results, though the magnitude of the marginal effects are mod-

est. For example, logit (probit) regressions show that targeted firms are 1.4% (1.5%) and

1.1% (1.0%) less likely to increase the number of their business segments and the number

of industries in which they operate, respectively. Overall, Table 5 shows that hedge fund

activism does not appear to be associated with large changes in the configuration of business

segments of target companies, mitigating the possibility that the estimated effects of hedge

fund activism on internal capital markets are due to changes in unobservable cross-segment

correlations.

In the next section, I summarize the paper and provide concluding remarks.

4 Conclusion

This paper sheds new light on the impact of hedge fund activism on the inner workings

of firms, namely that the intervention of hedge funds push firms to facilitate their internal

capital markets and improve the efficiency at which they allocate resources across business

segments. This is an important finding given the large literature documenting the limits of

internal capital markets and the social barriers that prevent them from working properly. It

also provides insight into the widely debated role hedge funds play in the companies they

26

Page 27: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

launch activist campaigns against.

I find that following initial Schedule 13D filings, investments made by the segments of

targeted firms become substantially more sensitive to cash flows generated in other parts of

the company, and that most of the increase in cross-subsidization comes from the reallocation

of firm cash flows toward segments with high Tobin’s Q. These findings are robust to catego-

rizing segments with respect to their Tobin’s Q in a variety of ways, mitigating the concern

that measurement errors in Tobin’s Q might drive the results. Allowing for separate estima-

tion of the effect of hedge fund activism on cross-subsidization for (1) highest Q segments,

(2) above median Q segments, (3) above average Q segments, and (4) lowest Q segments all

deliver a consistent message that hedge funds redirect the firm’s cash flow to segments with

the greatest investment opportunities. The results also hold in an industry-size matched

sample as well as a propensity score matched sample, rendering it less likely that systematic

differences between targeted and non-targeted firms confound inferences about the effects of

hedge fund activism. Finally, there is no evidence that targeted firms refocus their businesses

by reducing the number of their business segments or industries they serve, alleviating the

concern that apparent changes in cross-subsidization may in fact be a symptom of changes

in unobservable correlations across segments.

It should be noted that the findings of this paper are silent about the precise rationale

behind the facilitation of internal capital markets that hedge funds achieve. It could be the

case that hedge funds are concerned about productivity, and therefore directly impact the

effectiveness of the firm’s use of capital and assets. On the other hand, hedge funds may

correct inefficiencies indirectly by preventing political frictions between CEOs and divisional

managers from distorting resource allocations. This can particularly be true when hedge

funds, as outside investors, are not experts about the firm’s detailed operations. The truth

is likely somewhere in between.

Activist campaigns carried out by hedge funds are often at the center of public contention,

27

Page 28: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

where debates frequently deviate from the facts. With corporate control and shareholder

wealth at stake, it is important for academics to understand the real implications of hedge

fund activism regarding the operations of firms. While our knowledge has advanced in re-

cent years, there is still much room for research to understand how activist hedge funds

affect the utilization of labor and capital within the firm, and how they influence the social

connections and internal power struggles of divisional managers. Such endeavors may also

help understand how hedge funds differ from other types of shareholders who actively seek

to influence corporate policies. I look forward to fruitful investigations in the future.

28

Page 29: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

A Data Appendix

This section describes the construction of variables used in this study. To mitigate the

influence of outliers, all continuous variables are winsorized at the 1st and 99th percentiles,

except for segment Tobin’s Q which is used to construct categorical variables.

A.1 Variables

• HFA: Dummy variable that equals 1 if the date of the first Schedule 13D filing with

the firm as the investment target was at least 1 year ago, and 0 otherwise

• INVS : Segment level gross capital expenditure (item CAPXS) divided by lagged firm

total assets (item AT)

• CF : Segment level cash flow, computed as operating income before depreciation (item

OIBDPS) or operating profit (item OPS) plus depreciation (item DPS), as available,

divided by lagged firm total assets (item AT)

• Other CF : Sum of cash flows (item OIBDPS or items OPS plus DPS as available) of

all other segments in same firm, divided by lagged firm total assets (item AT)

• Segment Q: Segment level Tobin’s Q is the median Tobin’s Q of single-segment firms

in the same two-digit SIC code industry as the segment. Tobin’s Q of a single-segment

firm is computed as the ratio of firm value (defined as the market value of equity (item

CSHO multiplied by item PRCC F) plus the book value of total assets (item AT)

minus the book value of equity (item CEQ plus item TXDB)) to the book value of

total assets (item AT).

• High Q: Dummy variable that equals 1 if the segment’s Tobin’s Q is

– highest among all segments in the same firm

29

Page 30: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

– above the median Tobin’s Q of segments in the same firm

– above the average Tobin’s Q of segments in the same firm

– lowest among all segments in the same firm

• Control Variables: Segment level Tobin’s Q, segment level sales (item SALES)

growth rate, firm level asset size (log of item AT), firm level profitabiltiy (ROA, defined

as income before extraordinary items (item IB) divided by lagged total assets (item

AT)), firm level cash holdings (item CHE divided by item AT), firm level long-term

leverage (item DLTT divded by item AT), all lagged by 1 year

A.2 Data Screening

FASB No. 14 and SEC Regulation S-K require firms to report audited footnote information

for business segments whose sales, assets, or profits comprise more than 10% of the firm’s

consolidated totals. In June of 1997, FASB No. 14 was superseded by FASB No. 131,

under which firms are required to report such segment data insofar as “it is used internally

for evaluating segment performance and deciding how to allocate resources to segments”.

The Compustat segment database reports segment information based on this requirement.

To ensure that the reporting requirement change does not affect the results of the paper,

I redo the analysis using the sample period beginning with the fiscal year 1998 so that all

variables, including the lagged ones, use data strictly after the change occurred. The results

are virtually unchanged.

To construct the variables described above, I begin by following Shin and Stulz (1998)

and require segments to contain complete information on net sales (item SALES), identifiable

total assets (item IAS), capital expenditures (item CAPXS), operating profit (loss) (item

OPS), depreciation (item DPS), and SIC code. I exclude financial segments (SIC codes

between 6000 and 6999), since applying Tobin’s Q as a measure of investment opportunities

30

Page 31: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

may be problematic in these industries.

There are a number of widely recognized issues with the Compustat segment database.

For example, Compustat reports only up to ten segments, meaning smaller segments may be

neglected. Moreover, firms may choose to allocate their financial reporting across segments

with some discretion. As a result, firms may not fully allocate accounting items across the

reported segments. To address this problem, I follow Berger and Ofek (1995), Billett and

Mauer (2003), and Seru (2014) and require the sum of segment sales (assets) to be within

1% (25%) of firm totals, after which I apply a multiple to explicitly allocate unallocated

sales, assets, capital expenditure, and cash flow. Another important problem is that firms

may reorganize their segments over time and this may distort the identification of particular

segments. To minimize any bias arising from this issue, I take cue from Shin and Stulz

(1998) and require the following ratios to be less than one: segment capital expenditure to

segment assets, other segment capital expenditure to other segment assts, segment capital

expenditure to firm total assets, other segment capital expenditure to firm total assets,

segment cash flow to firm total assets, and other segment cash flow to firm total assets.

To ensure that the sample of firms are truly diversified, I follow Shin and Stulz (1998) and

Billett and Mauer (2003) and require firms to have at least two segments serving different

two-digit SIC industries and further exclude firms in which the smallest and largest segments

are in the same industry.

31

Page 32: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

References

Bebchuk, Lucian A., Alon Brav, and Wei Jiang, 2015. The long-term effects of hedge fundactivism. Columbia Law Review, Vol. 115, No. 5, pp. 1085-1156.

Berger, Philip G. and Eli Ofek, 1995. Diversification’s effect on firm value. Journal ofFinancial Economics, Vol. 37, pp. 39-65.

Bethel, Jennifer E., Julia Porter Liebeskind, and Tim Opler, 1998. Block share purchasesand corporate performance. The Journal of Finance, Vol. 53, No. 2, pp. 605-634.

Billett, Matthew T. and David C. Mauer, 2003. Cross-subsidies, external financing con-straints, and the contribution of the internal capital market to firm value. Review of Finan-cial Studies, Vol. 16, No. 4, pp. 1167-1201.

Brav, Alon, Wei Jiang, and Hyunseob Kim, 2015. The real effects of hedge fund activism:Productivity, asset allocation, and labor outcomes. Review of Financial Studies, Vol. 28,No. 10, pp. 2723-2769.

Brav, Alon, Wei Jiang, Frank Partnoy, and Randall Thomas, 2008. Hedge fund activism,corporate governance, and firm performance. The Journal of Finance, Vol. 63, No. 4, pp.1729-1775.

Campa, Jose Manuel and Simi Kedia, 2002. Explaining the diversification discount. TheJournal of Finance, Vol. 57, No. 4, pp. 1731-1762.

Chevalier, Judith A., 2004. What do we know about cross-subsidization? Evidence frommerging firms. The B.E. Journal of Economic Analysis and Policy, Vol. 4, No. 1, pp. 1-29.

Clifford, Christopher P., 2008. Value creation or destruction? Hedge funds as shareholderactivists. Journal of Corporate Finance, Vol. 14, pp. 323-336.

Comment, Robert and Gregg A. Jarrell, 1995. Corporate focus and stock returns. Journalof Financial Economics, Vol. 37, pp. 67-87.

Custodio, Claudia, 2014. Mergers and acquisitions accounting and the diversification dis-count. The Journal of Finance, Vol. 69, No. 1, pp. 219-240.

Duchin, Ran and Denis Sosyura, 2013. Divisional managers and internal capital markets.The Journal of Finance, Vol. 68, No. 2, pp. 387-429.

32

Page 33: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Ertimur, Yonca, Fabrizio Ferri, and Volkan Muslu, 2011. Shareholder activism and CEOpay. Review of Financial Studies, Vol. 24, No. 2, pp. 535-592.

Gantchev, Nickolay, 2013. The costs of shareholder activism: Evidence from a sequentialdecision model. Journal of Financial Economics, Vol. 107, pp. 610-631.

Gantchev, Nickolay, Oleg Gredil, and Chotibhak Jotikasthira, 2015. Governance under thegun: Spillover effects of hedge fund activism. Working Paper.

Gaspar, Jose-Miguel and Massimo Massa, 2011. The role of commonality between CEOand divisional managers in internal capital markets. Journal of Financial and QuantitativeAnalysis, Vol. 46, No. 3, pp. 841-869.

Gertner, Robert H., Eric Powers, and David S. Scharfstein, 2002. Learning about internalcapital markets from corporate spin-offs. The Journal of Finance, Vol. 57, No. 6, pp. 2479-2506.

Gillan, Stuart L. and Laura T. Starks, 2000. Corporate governance proposals and shareholderactivism: The role of institutional investors. Journal of Financial Economics, Vol. 57, pp.275-305.

Glaser, Markus, Florencio Lopez-de-Silanes, and Zacharias Sautner, 2013. Opening the blackbox: Internal capital markets and managerial power. The Journal of Finance, Vol. 68, No.4, pp. 1577-1631.

Gomes, Joao and Dmitry Livdan, 2004. Optimal diversification: Reconciling theory andevidence. The Journal of Finance, Vol. 59, No. 2, pp. 507-535.

Graham, John R., Michael L. Lemmon, and Jack G. Wolf, 2002. Does corporate diversifica-tion destroy value? The Journal of Finance, Vol. 57, No. 2, pp. 695-720.

Hartzell, Jay C. and Laura T. Starks, 2003. Institutional investors and executive compen-sation. The Journal of Finance, Vol. 58, No. 6, pp. 2351-2374.

Hund, John, Donald Monk, and Sheri Tice, 2010. Uncertainty about average profitabilityand the diversification discount. Journal of Financial Economics, Vol. 96, pp. 463-484.

Karpoff, Jonathan M., Paul H. Malatesta, and Ralph A. Walkling, 1996. Corporate gover-nance and shareholder initiatives: Empirical evidence. Journal of Financial Economics, Vol.

33

Page 34: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

42, pp. 365-395.

Klein, April and Emanuel Zur, 2009. Entrepreneurial shareholder activism: Hedge fundsand other private investors. The Journal of Finance, Vol. 64, No. 1, pp. 187-229.

Lang, Larry H.P. and Rene M. Stulz, 1994. Tobin’s q, corporate diversification, and firmperformance. Journal of Political Economy, Vol. 102, No. 6, pp. 1248-1280.

Maksimovic, Vojislav and Gordon Phillips, 2002. Do conglomerate firms allocate resourcesinefficiently across industries? Theory and evidence. The Journal of Finance, Vol. 57, No.2, pp. 721-767.

Maksimovic, Vojislav and Gordon Phillips, 2013. Conglomerate firms, internal capital mar-kets, and the theory of the firm. Annual Review of Financial Economics, Vol. 5, pp. 225-244.

Mansi, Sattar A. and David M. Reeb, 2002. Corporate diversification: What gets discounted?The Journal of Finance, Vol. 57, No. 5, pp. 2167-2183.

Ozbas, Oguzhan and David S. Scharfstein, 2010. Evidence on the dark side of internal capitalmarkets. Review of Financial Studies, Vol. 23, No. 2, pp. 581-599.

Rajan, Raghuram, Henri Servaes, and Luigi Zingales, 2000. The cost of diversity: Thediversification discount and inefficient investment. The Journal of Finance, Vol. 55, No. 1,pp. 35-80.

Scharfstein, David S. and Jeremy C. Stein, 2000. The dark side of internal capital markets:Divisional rent-seeking and inefficient investment. The Journal of Finance, Vol. 55, No. 6,pp. 2537-2564.

Seru, Amit, 2014. Firm boundaries matter: Evidence from conglomerates and R&D activity.Journal of Financial Economics, Vol. 111, pp. 381-405.

Shin, Hyun-Han and Rene M. Stulz, 1998. Are internal capital markets efficient? TheQuarterly Journal of Economics, Vol. 113, No. 2, pp. 531-552.

Smith, Michael P., 1996. Shareholder activism by institutional investors: Evidence fromCalPERS. The Journal of Finance, Vol. 51, No. 1, pp. 227-252.

Villalonga, Belen, 2004. Diversification discount or premium? New evidence from the Busi-ness Information Tracking Series. The Journal of Finance, Vol. 59, No. 2, pp. 479-506.

34

Page 35: Hedge Fund Activism and Internal Capital Markets · 2019-07-29 · Hedge Fund Activism and Internal Capital Markets Sehoon Kimy Fisher College of Business The Ohio State University

Wahal, Sunil, 1996. Pension fund activism and firm performance. Journal of Financial andQuantitative Analysis, Vol. 31, No. 1, pp. 1-23.

Whited, Toni M., 2001. Is it inefficient investment that causes the diversification discount?The Journal of Finance, Vol. 56, No. 5, pp. 1667-1691.

35