This article was downloaded by: [188.90.135.

223] On: 06 October 2024, At: 03:17

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA

Management Science
Publication details, including instructions for authors and subscription information:
http://pubsonline.informs.org

Short-Term Investors, Long-Term Investments,

and Firm Value: Evidence from Russell 2000 Index
Inclusions
Martijn Cremers, Ankur Pareek, Zacharias Sautner

To cite this article:

Martijn Cremers, Ankur Pareek, Zacharias Sautner (2020) Short-Term Investors, Long-Term Investments, and Firm
Value: Evidence from Russell 2000 Index Inclusions. Management Science 66(10):4535-4551. https://doi.org/10.1287/
mnsc.2019.3361

Full terms and conditions of use: https://pubsonline.informs.org/Publications/Librarians-Portal/PubsOnLine-

Terms-and-Conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use or
systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher approval,
unless otherwise noted. For more information, contact [email protected].

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or support
of claims made of that product, publication, or service.

Copyright © 2020, INFORMS

Please scroll down for article—it is on subsequent pages

With 12,500 members from nearly 90 countries, INFORMS is the largest international association of operations
research (O.R.) and analytics professionals and students. INFORMS provides unique networking and learning
opportunities for individual professionals, and organizations of all types and sizes, to better understand and use
O.R. and analytics tools and methods to transform strategic visions and achieve better outcomes.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
 MANAGEMENT SCIENCE
Vol. 66, No. 10, October 2020, pp. 4535–4551
http://pubsonline.informs.org/journal/mnsc ISSN 0025-1909 (print), ISSN 1526-5501 (online)

Short-Term Investors, Long-Term Investments, and Firm Value:

Evidence from Russell 2000 Index Inclusions
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

Martijn Cremers,a Ankur Pareek,b Zacharias Sautnerc

a
University of Notre Dame, Notre Dame, Indiana 46556; b University of Nevada, Las Vegas, Las Vegas, Nevada 89154; c Frankfurt School of
Finance & Management, 60322 Frankfurt am Main, Germany
Contact: [email protected] (MC); [email protected] (AP); [email protected], http://orcid.org/0000-0002-0780-6571 (ZS)

Received: June 27, 2018 Abstract. We document that an increase in short-horizon investors is associated with cuts
Revised: November 23, 2018; March 6, 2019 to long-term investment and increased short-term earnings. This leads to temporary boosts
Accepted: March 26, 2019 in equity valuations that reverse over time. To estimate these effects, we use difference-in-
Published Online in Articles in Advance: differences regressions around firms’ additions to the Russell 2000, comparing firms with
January 31, 2020 large and small increases in short-term ownership. We proxy for the presence of short-term
https://doi.org/10.1287/mnsc.2019.3361 investors using ownership by transient institutions. Our results suggest that short-term
pressures by investors can lead to myopic firm behavior.
Copyright: © 2020 INFORMS
History: Accepted by Shiva Rajgopal, accounting.
Funding: This work was supported by the IRRC Institute.
Supplemental Material: The internet appendix is available at https://doi.org/10.1287/mnsc.2019.3361.

Keywords: short-term investors • long-term investments • Russell 2000 inclusions

1. Introduction Stock prices are then driven up to the valuations of

Survey evidence by Graham et al. (2005) suggests that the most optimistic investors, because short-sale con-
many executives are willing to take short-term actions straints limit rational investors from eliminating any
that are detrimental to long-term firm value, such as overvaluations. CEOs also benefit from temporary
cutting long-term investment, possibly in response to overvaluations because their short-term compensa-
short-term pressures by investors. We show that an tion is tied to short-term stock prices. Equity over-
increase in short-horizon investors is associated with valuations reverse only gradually because it takes
cuts to long-term investment and increased short-term time for investors to understand that the higher earn-
earnings. This leads to temporary boosts in equity val- ings were due to R&D cuts that are detrimental to
uations that reverse over time. We estimate these effects long-term firm value.
through difference-in-differences regressions around Other theories of short-termism also model CEOs’
firms’ additions to the Russell 2000, comparing firms incentives to cut R&D. In Stein (1989) or Von Thadden
with large and small increases in short-term ownership. (1995), reductions in R&D arise because CEOs want
Our tests are motivated by the theory in Bolton et al. to avoid a decline in earnings, which may lead to
(2006), which predicts that short-horizon investors lower stock prices if investor trading is sensitive to
pressure chief executive officers (CEOs) to cut in- earnings news. CEOs are concerned about lower eq-
vestment to increase earnings, which subsequently uity valuations because their incentives may be tied
leads to temporary boosts in stock prices. The model to short-term stock prices or because a decline in
argues that CEOs, incentivized by short-horizon in- stock prices may trigger a takeover. Short-term insti-
vestors through short-term pay, take actions that in- tutional investors are particularly likely to exit over
crease the short-term speculative component in stock disappointing earnings because they hold stakes in
prices, at the expense of long-term firm value. Bolton numerous firms and trade frequently in and out of
et al. (2006) use the case of research and develop- stocks (see Bushee 1998). Although these models also
ment (R&D) cuts that boost short-term earnings as a explain myopic investment cuts, stock prices are un-
specific example of actions that temporarily inflate affected by R&D cuts because markets see through
stock prices. Stock prices overreact to positive earn- CEOs’ incentives to boost earnings.
ings news because of investor overconfidence, which in We estimate the effects of short-term investors within
their model generates disagreements regarding fun- a sample of stocks added to the Russell 2000 index from
damental firm values, as in Scheinkman and Xiong below. Whereas the Russell 1000 includes the 1,000
(2003). Short-term investors benefit from this by selling largest U.S. stocks according to market capitaliza-
stocks to other investors with more optimistic beliefs. tions, the Russell 2000 includes the next 2,000, thereby

4535
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4536 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

representing the U.S. small cap segment. 1 A firm is ownership in the years before index inclusion.2 We do
added to the Russell 2000 from below when its mar- not look at index exclusions because these are often
ket capitalization has sufficiently increased over the due to fundamental shocks, such as bankruptcies or
past year so that it ranks between 1,000 and 3,000. mergers and acquisitions (M&A), that may directly
Our sample of small cap firms has the benefit that affect our outcome variables (Harris and Gurel 1986,
short-sale constraints are more binding for such stocks Shleifer 1986). We focus on index inclusions from
(Reed 2013), making it easier for temporary over- below, rather than from above, because such recon-
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

valuations to arise. stitutions are more than three times as frequent and
A firm’s addition to the Russell 2000 might increase trigger significantly larger increases in short-term
short-term ownership for three nonmutually exclu- ownership.
sive reasons. First, index inclusion might lead to sig- A firm’s addition to the Russell 2000 can also in-
nificant buying by active short-term institutions that crease passive ownership and analyst coverage, and
benchmark their performance to the Russell 2000 these changes may affect our outcome variables. For
(Boone and White 2015, Chang et al. 2015). Second, example, ownership by passive (long-term) inves-
short-term institutions might increase their holdings if tors may allow firms to make long-term investments
they expect more firm transparency after index inclu- without the short-term performance pressures that
sion, which can increase trading opportunities and re- come from active traders (Beyer et al. 2014). Wider
duce the price impact of trading (Bushee and Noe analyst coverage may pressure managers to reduce
2000). Third, membership in the Russell 2000 might R&D to meet or beat earnings expectations (Fuller and
increase firm visibility, for example because analyst Jensen 2010).3 To address the potentially confounding
coverage increases after index inclusion. Increased effects of these variables, we directly control for changes
firm visibility in turn may lead to higher short-term in both variables in our regressions.
institutional ownership. Our analysis proceeds in three steps. In a first step,
Our proxy for the presence of short-horizon in- we examine whether firms that experience a large
vestors is ownership by transient institutions. Intro- (above-median) increase in transient ownership spend
duced by Bushee (1998, 2001), transient institutions less on R&D after index inclusion, relative to firms
are characterized by having high portfolio turnover that only experience a small increase. We find that
and diversified portfolios. We document a large and firms that experience a large transient-ownership in-
persistent increase in ownership by such institutions crease reduce R&D rates by 1.3 percentage points after
after a firm is added to the Russell 2000 from below: index inclusion, relative to firms that do not experi-
transient ownership is 1.9 percentage points higher ence the same increase. This effect equals 11% of the
after index inclusion, an increase by 22% relative to R&D variable’s standard deviation, and it is unaf-
the preinclusion average of 8.5%. The increase in short- fected when we control for changes in passive own-
term ownership does not reverse over subsequent years. ership or analyst coverage. Results are also unaffected
Transient ownership has several benefits over more if we control for changes in short-term trading (rather
trading-based proxies, such as share turnover. First, than short-term ownership) or the general increase in
much of the recent variation in share turnover comes institutional ownership around index inclusion. The
from high-frequency traders that are unlikely to impact decline in R&D is persistent and not reversed in the
longer-term corporate policies. Second, although share years after index inclusion.
turnover also increases around index inclusions be- In the theory of Bolton et al. (2006), short-term in-
cause of portfolio rebalancing needs by passive in- vestors pressure CEOs to cut R&D in order to report
stitutions, this effect is only temporary (Chang et al. higher earnings. Survey evidence (Graham et al. 2005)
2015). Rebalancing-motivated trading is therefore un- indicates not only that CEOs can cut investments on
likely to have lasting effects on corporate policies. short notice but also that this can boost earnings even
We identify the effects of short-term investors by within the same quarter. If temporarily inflated earn-
estimating difference-in-differences regressions around ings are misinterpreted by some investors, this can
a firm’s inclusion in the Russell 2000, comparing firms temporarily boost stock prices. Thus, linking the pres-
with large and small increases in transient ownership. ence of short-term investors to current earnings (and
Although a large increase in transient ownership is un- later to equity valuations) is an important element in
likely to be random, our analysis makes the (relatively Bolton et al. (2006). We find that the earnings (earn-
plausible) assumption that changes in transient own- ings before interest and taxes (EBIT) over assets) of
ership around index inclusion are not determined by firms that experience a large increase in transient own-
the outcome variables that we study. Importantly, all ership rise by 4.7 percentage points after index in-
of our outcome variables exhibit the same trends ac- clusion, relative to those of firms that do not experi-
ross firms with large and small increases in transient ence the surge. This effect equals 22% of the earnings
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4537

variable’s standard deviation. The higher earnings overvaluation once a firm has been included in the
also do not reverse over time. Russell 2000 for a few years, for example as short-sale
In a second step, we explore short-term and long- constraints gradually fade away once institutional
term equity valuations around index inclusions, con- ownership has sufficiently risen (e.g., Asquith et al.
ditional on changes in R&D and transient ownership. 2005).4 High institutional ownership after index in-
If short-term investors pressure managers to behave clusion may also lead to a more efficient processing
myopically, then Bolton et al. (2006) predict that the of information about the negative long-term effects of
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

equity valuation of firms that reduce R&D around R&D cuts (e.g., Piotroski and Roulstone 2004). There
index inclusion should rise temporarily relative to may be similar effects due the generally improved
those of other firms. This effect should be strongest information environment once firms have been in the
among firms with a large, recent increase in short- Russell 2000 for a few years (Hong et al. 2000, Boone
term ownership, because the decline in R&D at such and White 2015). For example, more analysts or an-
firms should be most strongly attributable to the alysts with higher reputation may start to specu-
ownership change. Temporary boost in equity valu- late on the effects of the reduced R&D. It is also pos-
ations would suggest that markets (initially) mis- sible that firms gradually start missing analysts’
interpreted the higher earnings as a positive signal earnings targets in the years after index inclusion
about firm fundamentals. To the contrary, equity valu- (though earnings continue to grow), leading to de-
ations should not rise if markets anticipated man- clines in valuations.
agers’ response to the inflow of short-term investors In a last step, we explore whether firms with more
(Stein 1989). short-term investors grant more short-term incen-
In the year of index inclusion, we find that equity tives to their CEOs. The channel in Bolton et al. (2006)
market-to-book ratios of firms that cut R&D are 1.5 holds that short-term investors incentivize CEOs to
units higher than those of other firm; this valuation act myopically by providing them with short-term
difference corresponds to 39% of the equity valua- pay. This allows CEOs to personally benefit from tem-
tion’s standard deviation. Importantly, the effect of porary increases in equity valuation. Ideally, we would
R&D cuts on equity valuations is strongly concen- test this prediction among Russell 2000 inclusion
trated among firms that experienced a large increase firms, but comprehensive pay data on such firms are
in transient ownership. not available. Instead, we turn to ExecuComp firms
Although this result is consistent with myopia, an for which we can link short-term ownership to short-
alternative explanation is that short-term investors term incentive pay. We find that firms with more
pressure CEOs to optimally reduce investment that transient investors grant their CEOs more options and
had previously been too high (Laux 2012). In that case, restricted stock.
equity valuations would also increase in the short run Our paper contributes to a growing empirical lit-
at firms that undertake such cuts. We distinguish be- erature that links corporate policies to investor ho-
tween these competing explanations by examining the rizons. The first paper in this literature is by Bushee
effect of R&D cuts on long-term valuations. If R&D (1998), who shows that more transient ownership
cuts are indeed myopic, then equity valuations should leads to less R&D spending, especially among firms
decline over time as the costs of the reduced investment where R&D cuts can reverse an earnings decline
are gradually revealed. Conversely, if R&D cuts are (“small-decline firms”). We complement Bushee’s work
efficient, then valuations should not decline in the with more recent data, in a different setting, and with
long run. additional results for earnings and equity valuations,
We find that R&D cuts have significantly negative thereby testing predictions derived from Bolton et al.
effects on future valuation: market-to-book ratios of (2006). We also show that the effects of R&D cuts extend
firms that cut R&D start to decline in the first year beyond small-decline firms, though we find stronger
after index inclusion, and this effect accelerates in effects for such firms.
subsequent years. Again, the decline in valuation is Our paper is further related to Derrien et al. (2013),
strongest among R&D-cutting firms that also experi- who find that greater long-term ownership is asso-
enced a large increase in transient ownership around ciated with increased investments when firms have
index inclusion. Overall, this reversal pattern is con- lower value than predicted by their fundamentals.
sistent with temporary price distortions caused by the Relatedly, Harford et al. (2018) find that firms with
effects of short-term investors on R&D. more long-term shareholders exhibit less fraud and
Our findings raise the question of which mecha- empire building. There is also evidence that firms with
nisms may cause the price corrections to occur in more short-term investors do worse in takeovers as tar-
the years after index inclusion. One possibility is gets or acquirers (Gaspar et al. 2005, Chen et al. 2007).
that it becomes easier to speculate against a possible Two papers relate managerial horizons and investment.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4538 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

Ladika and Sautner (2019) show that a decrease in increased CEO incentive pay (Mullins 2014) and higher
managerial horizon caused by accelerated option vest- dividends (Crane et al. 2016). Aghion et al. (2013) ex-
ing leads to investment cuts. Similarly, Edmans et al. ploit S&P 500 additions to show that greater institu-
(2017) document that imminent vesting of equity in- tional ownership improves innovation.
centives is associated with lower investment spending. Two papers focus on passive ownership. Schmidt
The next section summarizes studies that look at and Fahlenbrach (2017) find that higher ownership
the effects of index inclusions on corporate policies. by index funds or exchange-traded funds (ETFs) leads to
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

The section shows that most prior research focuses more CEO power, fewer independent directors, and
on Russell 2000 inclusion from above, rather than worse M&A decisions. In contrast, Appel et al. (2016)
from below, and on the role of institutional and passive find that an increase in ownership by passive mutual
ownership, rather than of short-term ownership. funds leads to more independent directors, the re-
moval of antitakeover defenses, and more equal vot-
2. Related Literature on the Effects of ing rights. Schmidt and Fahlenbrach (2017) recon-
Index Inclusions cile these findings by arguing that passive ownership
2.1. Firm Policies and Index Inclusions affects governance positively when it comes to low-
Research on index inclusions can broadly be classified cost governance activities such as voting, and nega-
into studies that examine the effects of ownership tively for high-cost governance such as monitoring of
changes around index inclusions, and studies that M&A decisions.
consider how stock prices change around the time
when stocks are added to an index. Within the first 2.2. Stock Market Reactions and Index Inclusions
group, most papers exploit ownership variation around Several studies analyze the short-term market re-
the cutoff used to calculate membership in the Russell action when firms get added to important indexes,
1000/2000. Because portfolio weights within these initially to answer whether demand curves for stocks
two indexes are value-weighted, the smallest stocks slope down or are horizontal. Shleifer (1986) examines
in the Russell 1000 have small index weights, whereas price reactions when firms get added to the S&P 500,
the largest stocks in the Russell 2000 are given large which does not provide new information about fun-
weights (larger by a factor of approximately 10). As a damentals to the market. Nevertheless, he documents
result, for every dollar invested in the two indexes, very an index addition effect of approximately 3% in the
little is invested in stocks at the bottom of the Russell days around the inclusion, which he explains with
1000, but a lot is invested in stocks at the top of the downward-sloping demand curves caused by index
Russell 2000. tracking.5 Further evidence for a positive S&P 500 in-
Several studies exploit the fact that these differ- clusion effect is provided by Harris and Gurel (1986)
ences in index weights lead to higher (lower) insti- and Lynch and Mendenhall (1997). Petajisto (2011)
tutional and passive ownership at firms at the top of and Biktimirov et al. (2004) provide similar evidence
the Russell 2000 (bottom of the Russell 1000), using for the Russell 2000, and Greenwood (2008) for the
either a regression-discontinuity design (RDD) or in- Nikkei 225.
strumental variables. The RDD framework compares Chang et al. (2015) argue that it is difficult to sep-
firms just above (i.e., at the bottom of the Russell 1000) arate index-tracking effects from confounding fac-
and below the threshold (i.e., at the top of the Russell tors, such as the investor recognition associated with
2000), assuming that firms around the threshold are index membership. To isolate indexing effects, they
identical except for the assignment to a different index use an RDD framework and compare firms around
(Boone and White 2015). As an instrument for own- the Russell 1000/2000 threshold (rather than com-
ership, some studies use an indicator capturing whether paring firms added to an index with those that are
firms switch between the two indexes (Fich et al. not). As explained, this design exploits that firms
2015, Schmidt and Fahlenbrach 2017), and others use added to the Russell 2000 from above (i.e., from the
an indicator of whether firms are assigned to the Russell Russell 1000) have disproportionally higher own-
2000 (Appel et al. 2016, Crane et al. 2016). ership by indexing institutions. Using this frame-
In terms of results, Boone and White (2015) show that work, they find a 5% stock price increase for firms
higher institutional ownership due to inclusion in the added to the Russell 2000 from above; this effect is
Russell 2000 leads to greater management disclo- smaller (approximately 3%) and noisier for firms
sure, analyst following, and liquidity. Fich et al. (2015) added from below. Index addition is further asso-
document that higher institutional ownership, espe- ciated with more trading, as indexing investors need
cially by institutions whose holding value is in the top to rebalance their portfolios.
10% of their portfolio, leads to better M&A decisions. A few papers study whether the initial rise in stock prices
Higher institutional ownership is also associated with around index inclusions is persistent. Whereas Lynch
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4539

and Mendenhall (1997) find that large parts of the Bushee identifies “transient” investors as institutions
S&P 500 price effect are permanent, Petajisto (2011) with high portfolio turnover and diversified port-
documents for Russell 2000 stocks that the effect folios. He differentiates these investors from “quasi-
largely reverses within two months, though standard indexer” institutions with low turnover and diversified
errors are large. Greenwood (2008) finds for Nikkei portfolios, and from “dedicated” institutions with low
225 stocks that most of the initial price increase reverses turnover and more concentrated portfolios. Table 1,
over time. panel A, reports that ownership by transient investors
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

averages approximately 13% across our sample. Table 1,

3. Data and Methodology panel B, shows how key variables are correlated with
3.1. Sample and Measures transient ownership.
Our sample consists of stocks added to the Russell
2000 from below between 1990 and 2016. We analyze 3.2. Empirical Model
the two-sided four-year window around each firm’s To capture the effects of short-term ownership, we
addition to the index, covering the four years before estimate the following difference-in-differences model
and the four years after a firm’s inclusion year. This for the two-sided four-year window around a firm’s
event window allows us to study whether trends exist inclusion in the Russell 2000:
for our outcome variables in the preinclusion years
and whether any reversals arise in the post-inclusion Firm Outcomeft
π₁Post Inclusionft
years. Firms are newly added to the Russell 2000 from × Large Increase Transient IOft
below when they were previously outside of the Russell + π2 Post Inclusionft
1000 and 2000 but their market capitalization ranks
have increased over the past year so that they are now + π3 Large Increase Transient IOft
ranked between 1,000 and 3,000. Firms added to the + π4 X ft + λf + µft + εft , (1)
index from below tend to be smaller than the other
Russell 2000 firms. where Firm Outcome is R&D/Assets or EBIT/Assets for
Unlike for passive investors, such as index funds firm f in year t, depending on the hypothesis being
or ETFs, the addition to the Russell 2000 does not tested. Post Inclusion equals one for firm-year ob-
mechanically force active short-term investors to buy servations after a firm gets added to the Russell 2000,
newly added shares. However, Russell 2000 additions and zero otherwise. Some of our tests estimate mod-
are accompanied with significant buying by active ifications of this baseline model in which we replace
short-term institutions that benchmark their per- Post Inclusion with year dummies that measure time
formance to the Russell 2000 (Boone and White 2015, relative to the inclusion year (e.g., Year +1).
Chang et al. 2015) or expect greater firm transpar- Large Increase Transient IO captures a large increase
ency (Bushee and Noe 2000).6 Short-term ownership in transient ownership around index inclusion. We
also rises if index inclusion increases firm visibility, identify such events with an indicator that equals
which in turn may lead to buying by short-term active one in all firm years for firms that experience an in-
institutions. Indeed, we document a large increase in crease in Transient IO that is above the sample median,
ownership by short-term (transient) investors after a measured over the two-year period around a firm’s
stock is added to the Russell 2000 from below. inclusion in the Russell 2000, and zero otherwise.8
We restrict our tests to only the first inclusion of a Our main coefficient of interest is π1. A negative
firm to the Russell 2000 during our sample period, to (positive) π1-coefficient for R&D (earnings) would
avoid problems with overlapping event windows.7 indicate that firms that experience a large increase in
This restriction leaves us with 57% of all firms that transient ownership reduce (increase) R&D spending
were added to the Russell 2000 during the sample (earnings) after index inclusion, relative to firms that
period. We show that results are unaffected if we use only experience a small increase.
all inclusion events instead of only the first inclusions. The vector Xft contains firm-level control variables
The number of inclusion events in our regressions is that may affect changes in short-term ownership and our
further reduced as we require data on short-term own- outcome variables. Similar to Schmidt and Fahlenbrach
ership and corporate policies for the years around (2017), we control for a firm’s market capitalization,
index inclusion. Nevertheless, our final sample has the change in market capitalization rank, and the past
sufficient power to identify the effects of short- stock return. Next to these variables, we account for
horizon investors because it contains 841 unique in- the overall changes in institutional ownership around
clusion events. a firm’s index inclusion. Following Bushee and Noe
Our measure for the presence of short-term institutional (2000), we also control for dividends, leverage, idi-
investors is a firm’s ownership by transient investors osyncratic risk and a firm’s equity beta. We account
(Transient IO), introduced by Bushee (1998, 2001). for these variables as they have been shown to affect
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4540 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

Table 1. Summary Statistics

Panel A. Summary statistics

Variable Mean Median Std. dev. 25th 75th Obs.

Transient IO 13.1 9.5 12.1 3.4 19.3 5,425

Large Increase Transient IO 0.51 5,524
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

Quasi-Indexer IO 23.3 19.5 17.2 10.1 32.7 5,492

Large Increase Quasi-Indexer IO 0.52 — — — — 5,469
IO 45.6 44.0 26.2 23.8 65.3 5,499
Large Increase IO 0.53 5,524
Analyst Coverage 2.90 2.00 3.18 0.00 4.00 5,524
Δ Analyst Coverage 2.03 1.00 3.20 0.00 4.00 5,524
Share Turnover 1.75 1.15 1.75 0.57 2.34 5,523
R&D/Assets 0.05 0.00 0.12 0.00 0.05 5,524
R&D Missing 0.48 — — — — 5,524
Decrease R&D 0.26 — — — — 5,478
EBIT/Assets 0.02 0.06 0.21 0.01 0.11 5,481
Income/Assets −0.03 0.02 0.23 −0.02 0.07 5,524
Market-to-Book Ratio 3.82 2.36 3.82 1.36 4.60 5,370
Log(Market Cap) 5.20 5.19 0.99 4.54 5.80 5,524
Δ Market Cap Rank −131 −101 593 −425 172 5,524
Stock Return 0.27 0.11 0.76 −0.20 0.52 5,520
Dividends/Net Income 0.11 0.00 0.39 0.00 0.10 5,524
Debt/Assets 0.19 0.12 0.21 0.01 0.29 5,524
Idiosyncratic Risk 0.03 0.03 0.02 0.02 0.04 5,524
Equity Beta 1.12 1.03 0.76 0.54 1.57 5,524

Panel B. Correlations

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

Transient IO (1) 1.00 — — — — — — — — — — — —

Large Increase Transient IO (2) 0.31 1.00 — — — — — — — — — — —
Quasi-Indexer IO (3) 0.26 0.11 1.00 — — — — — — — — — —
Large Increase Quasi-Indexer IO (4) 0.18 0.15 0.20 1.00 — — — — — — — — —
IO (5) 0.67 0.23 0.78 0.20 1.00 — — — — — — — —
Large Increase IO (6) 0.30 0.45 0.17 0.57 0.28 1.00 — — — — — — —
Analyst Coverage (7) 0.37 0.16 0.34 0.16 0.44 0.20 1.00 — — — — — —
Δ Analyst Coverage (8) 0.22 0.24 0.12 0.30 0.19 0.35 0.44 1.00 — — — — —
Share Turnover (9) 0.41 0.15 0.20 0.13 0.34 0.20 0.31 0.19 1.00 — — — —
R&D/Assets (10) −0.01 0.00 −0.14 0.00 −0.10 0.00 0.03 −0.01 0.14 1.00 — — —
R&D Missing (11) −0.12 −0.08 0.02 −0.01 −0.05 −0.09 −0.05 −0.06 −0.22 −0.45 1.00 — —
Decrease R&D (12) 0.06 0.09 0.00 0.08 0.03 0.10 0.03 0.07 0.14 0.26 −0.52 1 —
EBIT/Assets (13) 0.13 0.09 0.20 0.05 0.23 0.08 0.07 0.05 0.00 −0.68 0.18 −0.10 1.00
Market-to-Book Ratio (14) 0.20 0.09 −0.05 0.11 0.05 0.12 0.11 0.12 0.33 0.29 −0.22 0.17 −0.11

Notes. Panel A reports summary statistics for the variables used in the analysis, and panel B correlations for key variables. The sample
consists of firms added to the Russell 2000 from below. We report firm-year observations for the two-sided four-year window around
a firm’s inclusion in the Russell 2000 from below. The sample period is 1990 to 2016. Variables are defined in the Appendix. The table
contains observations included in the regression sample in Table 3, column (1). Std. Dev., standard deviation; Obs., number of
observations.

transient ownership and they may also affect R&D 3.3. Russell 2000 Inclusions and the Presence of
and earnings. As Russell 2000 membership tempo- Short-Term Investors
rarily leads to more trading volume, we control for Table 2, columns (1) and (2), examine the effect of in-
Share Turnover for robustness. In R&D regressions we dex inclusion on short-term ownership, measured us-
set missing values of R&D to zero and add a control ing Transient IO. Column (1) identifies the effect of index
variable that equals one for observations with missing inclusion by comparing average short-term ownership
R&D data. λf are firm fixed effects that account for in the four years before and after a firm’s addition to
any systematic differences in our outcome variables the Russell 2000. We find a highly statistically sig-
across firms, and µft are industry-by-year fixed effects nificant and economically large increase in short-term
that control for economic shocks that affect all firms ownership after a firm is assigned to the index. The
within an industry in a given year. estimates imply that transient ownership is 1.9
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4541

Table 2. Short-Term Ownership, Analyst Coverage, and Passive Ownership Around

Russell 2000 Inclusions

Dependent variable Transient IO Transient IO Quasi-Indexer IO Analyst Coverage

(1) (2) (3) (4)

Post Inclusion 1.908*** — 2.154*** 0.224

(3.71) — (3.47) (1.52)
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

Year 0 — 1.499*** — —
— (3.45) — —
Year + 1 — 1.786*** — —
— (3.73) — —
Year + 2 — 1.456*** — —
— (3.01) — —
Year + 3 — 1.401*** — —
— (2.70) — —
Year + 4 — 1.560*** — —
— (2.81) — —
Log(Market Cap) 2.609*** 2.658*** 7.214*** 1.402***
(8.68) (9.93) (15.67) (11.36)
Δ Market Cap Rank −0.144*** −0.157*** 0.276*** 0.034***
(−5.24) (−5.28) (9.10) (4.41)
Stock Return (t − 1) −0.548*** −0.468** −1.429*** −0.242***
(−2.88) (−2.30) (−6.87) (−4.51)
Dividends/Net Income −0.121 −1.777*** −1.059 −0.358***
(−0.37) (−4.84) (−1.45) (−2.78)
Debt/Assets −2.090 1.843* 1.938 0.066
(−1.45) (1.68) (1.22) (0.19)
Idiosyncratic Risk −0.504*** −0.848*** −1.625*** −0.123***
(−4.91) (−8.08) (−10.67) (−3.95)
Equity Beta −0.673** 0.064 −1.417*** 0.058
(−2.08) (0.20) (−3.21) (0.64)
Share Turnover 1.609*** 2.064*** 1.040*** 0.372***
(9.93) (13.22) (4.89) (6.41)
Industry-by-year fixed effects Yes Yes Yes Yes
Firm fixed effects No No No No
No. of observations 5,424 5,424 5,491 5,523
Adjusted R2 0.651 0.431 0.472 0.281

Notes. We report regressions using firm-year observations for the two-sided four-year window around a
firm’s inclusion in the Russell 2000 (in year 0). The sample period is 1990 to 2016. The regressions contain
observations included in the regression sample in Table 3, column (1). Transient IO is the percentage
ownership of transient institutional investors. Quasi-Indexer IO is the percentage ownership of quasi-
indexers. Analyst Coverage is the number of analysts covering a firm. Post Inclusion equals one for firm-
year observations after a firm gets added to the Russell 2000 from below, and zero otherwise. Year 0
equals one in the firm year in which a firm gets added to the Russell 2000 from below, and zero
otherwise. The coefficient on Δ Market Cap Rank is multiplied by 100, and the coefficient on Idiosyncratic
Risk is divided by 100. Variables are defined in the Appendix. t-statistics are based on robust standard
errors clustered at the firm level.
*p < 0.10; **p < 0.05; ***p < 0.01, based on a two-tailed test.

percentage points higher once a firm gets added to We find that short-term ownership remains high over
the Russell 2000, which reflects an increase by 22% the four-year period after index inclusion, indicating
relative to the preinclusion average of 8.5%. The that there is no reversal in transient ownership.
magnitude of this effect increases slightly when we In terms of control variables, Table 2, columns (1)
account for firm fixed effects (not reported). Column (2) and (2), show that short-term ownership is larger
explores whether the increase in Transient IO is persistent. among bigger firms, among firms that moved up more
We replace the post-inclusion dummy with indicators the market capitalization rank (for them Δ Market Cap
that measure short-term ownership in each year after Rank is negative), and among firms with higher past
index inclusion, relative to the preinclusion average. returns. Consistent with Bushee and Noe (2000), firms
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4542 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

with higher transient ownership pay lower dividends, outcome variables by directly controlling for changes
have higher leverage, have lower idiosyncratic risk, in these variables.
and exhibit more share turnover.
As explained above, we restrict our tests to the first 4. Main Results
time a firm is added to the Russell 2000. Internet Ap- 4.1. Short-Term Investors and R&D Expenditures
pendix Table 1 relaxes this restriction and uses all Bolton et al. (2006) predict that short-term investors
inclusions from below. Within this larger sample we pressure CEOs to reduce R&D to surprise markets
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

continue to find that index inclusion significantly in- with higher earnings. We first examine whether firms
creases short-term ownership. spend less on R&D when short-term ownership in-
creases owing to index inclusion. As in Bushee (1998),
3.4. Confounding Effects: Passive Ownership and we focus on R&D to capture long-term investments
Analysts Coverage because these are discretionary expenditures that con-
Index inclusion has a strong positive effect on short- temporaneously depress reported earnings. At the
term ownership. This change is unlikely to occur in same time, R&D projects can take years to complete,
isolation, and a firm’s Russell 2000 assignment may and the benefits from R&D may occur only years into
affect other variables that influence firm outcomes. the future. R&D is particularly susceptible to myopia
We consider two variables in particular, namely changes because managers generally have broad leeway to
in passive ownership and analyst coverage. reduce or postpone R&D in order to increase current
An increase in passive ownership could be bene- earnings. Cutting R&D can boost a firm’s stock price
ficial for long-term investment, because passive in- in the short term if differences of opinion exist and if
vestors tend to be long-term owners interested in investors misinterpret the higher earnings that re-
long-term performance. Passive investors are also less sult from R&D cuts (e.g., if some investors naı̈vely use
likely to use voice (too expensive given the low-cost earnings-based multiples to derive their estimate of
business model) or the threat of exit (not available equity values). These assumptions are plausible as
given their explicit index tracking) to pressure man- stock markets seem unable to properly value R&D
agers to act myopically. We capture the effects of pas- (Cohen et al. 2013), implying that investors may not
sive holdings using firm ownership by quasi-indexers fully incorporate the consequences of R&D cuts. Valu-
(Quasi-Indexer IO) (Bushee 1998). Quasi-indexers in- ing R&D is likely to be more challenging for small
clude purely passive index funds that track the Russell firms, such as those for which we identify the effects
2000 and more actively managed funds that closely of short-horizon investors.
mimic the index. In Table 3, we examine in columns (1) through (4)
Increases in analyst coverage may have the oppo- whether firms that experience a large increase in tran-
site effect, because wider following by analysts may sient ownership reduce R&D spending after index
pressure CEOs to act myopically, so they can meet or inclusion, relative to firms that only experience a small
beat analyst expectations (Fuller and Jensen 2010). increase. The estimates in column (1) show that the R&D
Additionally, wider analyst following may have feed- rates of firms that experience a large rise in transient
back effects if increased firm visibility due to larger ownership decrease by 1.3 percentage points after
analyst coverage triggers some short-term investors index inclusion, relative to those of firms that do not
into buying newly added stocks. We measure analyst experience the same short-term ownership increase.9
coverage using the number of analysts issuing earn- This effect is large economically because the decline
ings forecasts for a firm (Analyst Coverage). in R&D equals 11% of the variable’s sample stan-
Table 2, column (3), shows that it is potentially im- dard deviation.10
portant to account for the role of quasi-indexers, Next, column (2) in Table 3 additionally controls for
because the ownership by such investors significantly the increase in ownership by quasi-indexers around a
increases when a firm gets added to the Russell 2000. firm’s index inclusion. Although a large increase in
The point estimate in the column implies that Quasi- transient ownership continues to have a negative effect
Indexer IO increases by 2.2 percentage points after on R&D, we do not find that a large rise in ownership
index inclusion. This equals 15% of the variable’s pre- by quasi-indexers is associated with changes in R&D
inclusion average of 15%. With regard to changes in spending. Within our sample, we therefore have little
analyst coverage around index inclusion, Table 2, evidence that higher passive ownership better allows
column (4), indicates that analyst following also in- firms to make long-term investments. Column (3) ad-
creases once a firm gets added to the Russell 2000, ditionally controls for the potentially confounding
though this effect is marginally insignificant (t-statistic role of the rise in analyst coverage around index in-
of 1.52). We will account for the potential effects clusion. Although the effect of transient ownership
of passive ownership and analyst coverage on our is unaffected when we control for changes in analyst
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4543

Table 3. Effect of Short-Term Ownership on R&D Expenditures Around Russell 2000 Inclusions

Dependent variable R&D/Assets

(1) (2) (3) (4) (5)

Post Inclusion × Large Increase Transient IO −0.013** −0.014** −0.013** −0.012** —

(−2.29) (−2.40) (−2.33) (−2.18) —
Year 0 × Large Increase Transient IO — — — — −0.010*
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

— — — — (−1.93)
Year + 1 × Large Increase Transient IO — — — — −0.013**
— — — — (−2.17)
Year + 2 × Large Increase Transient IO — — — — −0.017**
— — — — (−2.45)
Year + 3 × Large Increase Transient IO — — — — −0.014*
— — — — (−1.80)
Year + 4 × Large Increase Transient IO — — — — −0.015**
— — — — (−2.00)
Post Inclusion × Large Increase Quasi-Indexer IO — 0.000 0.001 0.001 0.000
— (0.01) (0.16) (0.20) (0.02)
Post Inclusion × Δ Analyst Coverage — — −0.001 −0.001 —
— — (−0.97) (−0.89) —
Share Turnover — — — −0.004*** —
— — — (−2.84) —
Post Inclusion 0.007 0.006 0.007 0.006 —
(1.50) (1.35) (1.38) (1.33) —
Year 0 — — — — 0.002
— — — — (0.39)
Year + 1 — — — — 0.010*
— — — — (1.86)
Year + 2 — — — — 0.011*
— — — — (1.81)
Year + 3 — — — — 0.007
— — — — (0.94)
Year + 4 — — — — 0.009
— — — — (1.07)
Post Inclusion × Large Increase IO 0.002 0.002 0.004 0.004 0.002
(0.36) (0.33) (0.50) (0.57) (0.33)
Log(Market Cap) −0.009*** −0.009*** −0.008*** −0.005*** −0.009***
(−3.77) (−3.68) (−3.75) (−2.62) (−3.48)
Δ Market Cap Rank 0.000 0.000 0.000 0.000 −0.000
(0.31) (0.21) (0.34) (0.03) (−0.60)
Stock Return (t − 1) −0.002 −0.002 −0.002* −0.001 −0.002*
(−1.61) (−1.64) (−1.65) (−0.93) (−1.80)
Dividends/Net Income −0.000 −0.000 −0.000 −0.000 −0.000
(−0.04) (−0.09) (−0.13) (−0.09) (−0.16)
Debt/Assets −0.017 −0.014 −0.014 −0.013 −0.014
(−1.09) (−0.91) (−0.92) (−0.88) (−0.91)
Idiosyncratic Risk 0.158* 0.156* 0.157* 0.226** 0.156*
(1.71) (1.68) (1.70) (2.40) (1.68)
Equity Beta 0.001 0.000 0.001 0.001 0.000
(0.40) (0.12) (0.22) (0.33) (0.14)
R&D Missing −0.027** −0.026** −0.026** −0.025** −0.026**
(−2.40) (−2.31) (−2.29) (−2.25) (−2.30)
Industry-by-year fixed effects Yes Yes Yes Yes Yes
Firm fixed effects Yes Yes Yes Yes Yes
No. of observations 5,524 5,469 5,469 5,468 5,469
Adjusted R2 0.833 0.834 0.834 0.835 0.834

Notes. We report regressions using firm-year observations for the two-sided four-year window around a firm’s inclusion in the Russell 2000
from below (in year 0). The sample period is 1990 to 2016. R&D/Assets is R&D expenditures over assets. Large Increase Transient IO equals one
in all firm years for firms that experience an increase in Transient IO that is above the sample median, measured over the two-year period around
a firm’s inclusion in the Russell 2000, and zero otherwise. Post Inclusion equals one for firm-year observations after a firm gets added to the
Russell 2000 from below, and zero otherwise. Year 0 equals one in the firm year in which a firm gets added to the Russell 2000 from below, and
zero otherwise. The coefficient on Δ Market Cap Rank is multiplied by 100. t-statistics are based on robust standard errors clustered at the firm level.
*p < 0.10; **p < 0.05; ***p < 0.01, based on a two-tailed test.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4544 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

coverage, we find that the increase in analyst cov- main earnings variable in columns (1)–(5) is EBIT,
erage itself has a negative effect on R&D. However, whereas results are similar if we use net income in
this effect is statistically insignificant. Finally, we column (6) instead. We scale both earnings variables
also control in column (4) for share turnover to ac- by total assets.
count for changes in short-term trading. In this re- Column (1) in Table 4 shows that firms that expe-
gression, we continue to find that an increase in rience a large increase in transient ownership report
transient ownership leads to less R&D spending. higher earnings after index inclusion, relative to firms
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

Table 2 documented that the increase in Transient that only experience a small increase. The estimates
IO after index assignment is persistent. This raises the imply that EBIT/Assets of firms that experience a large
question of whether the decline in R&D is also lasting, transient-ownership increase rise by 4.7 percentage
or whether R&D reverses in subsequent years. For points after index inclusion, relative to firms that do
example, some CEOs may reduce R&D in the year not experience such a surge. The effect is large eco-
of index inclusion but then increase it again in sub- nomically, equal to 22% of the variable’s standard de-
sequent years. To explore this possibility, we replace viation. Because the coefficient estimate on Post In-
in Table 3, column (5) Post Inclusion with indicators clusion × Large Increase Transient IO is larger than the
for each year after index inclusion, and interact these corresponding estimate for R&D in Table 3, reduc-
indicators with Large Increase Transient IO. The re- tions in other expenses likely also contribute to the
sulting interaction terms allow us to estimate the ef- rise in earnings.12
fects of the increase in transient ownership for each Columns (2) and (3) in Table 4 show that transient
post-inclusion year separately. We find that the esti- ownership continues to have a positive effect on earn-
mates for all interaction terms are negative, of similar ings after controlling for the change in ownership by
magnitude, and statistically different from zero. Hence, quasi-indexers and in analyst coverage. The same holds
there is no evidence that the decline in R&D reverses in column (4), where we account for changes in short-
at firms that experience a large increase in transient term trading. Column (6) illustrates that results are
ownership. unaffected if we use net income instead of EBIT.
Bushee (1998) documents in his sample that the We showed that the effects of changes in transient
negative effect of transient ownership on R&D is con- ownership on R&D around index inclusions are per-
centrated among firms whose earnings before R&D sistent. Column (5) in Table 4 documents the same
and taxes (EBTRD) declined relative to the prior year, pattern for earnings, which also continue to be per-
but by an amount that can be reversed by cutting sistently higher after index inclusions, thereby mir-
R&D. This raises the question of whether our results roring the lasting effect of the decline in R&D.
are also concentrated among such firms, or whether
R&D cuts are more widespread. To explore this ques- 4.3. Short-Term Investors and Equity Valuations
tion, we create a sample of firms for which R&D cuts If short-term investors pressure managers to behave
can reverse an earnings decline from year t − 1 to myopically, then Bolton et al. (2006) predict that the
year t. We consider the index-inclusion year as year t equity valuation of firms that reduce R&D should rise
and call the resulting set of firms the “small-decline temporarily relative to the valuation of other firms.
sample,” following Bushee (1998).11 A limitation of Such temporary price effects would suggest that mar-
this analysis is that only 187 firms meet this require- kets (initially) misinterpreted the higher earnings as
ment, which reduces the power of our tests. Internet a positive signal about firm fundamentals. The al-
Appendix Table 2 replicates our analysis for the small- ternative (null) hypothesis is that markets correctly
decline sample. Consistent with Bushee (1998), the anticipated managers’ response to the inflow of short-
estimate of Post Inclusion × Large Increase Transient IO term investors, in which case equity valuations should
is much larger than in the corresponding Table 3, not rise (Stein 1989).
column (1), roughly by a factor ten. Table 5, panel A, considers whether R&D reduc-
tions are associated with temporary distortions to the
4.2. Short-Term Investors and Earnings valuation of a firm’s equity, which we capture using
Short-term investors in Bolton et al. (2006) pres- the equity market-to-book ratio in the year of index
sure managers to reduce R&D with the objective to inclusion. Specifically, we examine whether short-
report higher earnings. If temporarily inflated earn- term equity valuations are higher among firms that
ings are misinterpreted by some investors, this can cut R&D, and how this effect varies with short-term
lead to temporary boosts in the stock price. To test ownership. As explained in Section 2, firms added to
for the effects of short-term investors on earnings, the Russell 2000 generally experience an increase
Table 4 provides regressions similar to those in Table 3, in equity valuations upon index inclusion relative
but using measures of earnings instead of R&D. Our to prior years. Our regressions therefore attempt to
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4545

Table 4. Effect of Short-Term Ownership on Earnings Around Russell 2000 Inclusions

Net Income/
Dependent variable EBIT/Assets Assets

(1) (2) (3) (4) (5) (6)

Post Inclusion × Large Increase Transient IO 0.047*** 0.048*** 0.046*** 0.044*** — 0.040***
(4.04) (4.06) (3.96) (3.75) — (3.02)
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

Year 0 × Large Increase Transient IO — — — — 0.033*** —

— — — — (2.80) —
Year +1 × Large Increase Transient IO — — — — 0.055*** —
— — — — (4.11) —
Year +2 × Large Increase Transient IO — — — — 0.059*** —
— — — — (4.16) —
Year +3 × Large Increase Transient IO — — — — 0.051*** —
— — — — (3.35) —
Year +4 × Large Increase Transient IO — — — — 0.068*** —
— — — — (4.35) —
Post Inclusion × Large Increase Quasi-Indexer IO — 0.010 0.008 0.007 0.008 0.006
— (0.83) (0.65) (0.57) (0.68) (0.38)
Post Inclusion × Δ Analyst Coverage — — 0.003 0.003 — —
— — (1.50) (1.34) — —
Share Turnover — — — 0.013*** — —
— — — (4.73) — —
Post Inclusion −0.028** −0.029** −0.030** −0.029** — −0.027**
(−2.29) (−2.26) (−2.31) (−2.25) — (−2.15)
Year 0 — — — — −0.040*** —
— — — — (−2.88) —
Year +1 — — — — −0.076*** —
— — — — (−3.95) —
Year +2 — — — — −0.105*** —
— — — — (−4.25) —
Year +3 — — — — −0.125*** —
— — — — (−4.40) —
Year +4 — — — — −0.155*** —
— — — — (−4.60) —
Post Inclusion × Large Increase IO −0.023* −0.029** −0.032** −0.034** −0.028** −0.023
(−1.96) (−2.19) (−2.32) (−2.42) (−2.12) (−1.37)
Log(Market Cap) 0.031*** 0.030*** 0.028*** 0.020*** 0.030*** 0.029***
(5.49) (5.36) (5.10) (3.44) (5.00) (4.26)
Δ Market Cap Rank −0.002*** −0.002*** −0.002*** −0.002*** −0.001*** −0.002***
(−4.36) (−4.45) (−4.52) (−4.15) (−2.80) (−3.99)
Stock Return (t − 1) 0.015*** 0.015*** 0.015*** 0.012*** 0.013*** 0.014***
(5.35) (5.23) (5.25) (4.27) (4.44) (3.98)
Dividends/Net Income 0.012*** 0.013*** 0.013*** 0.013*** 0.013*** 0.014***
(2.65) (2.80) (2.80) (2.78) (2.91) (2.89)
Debt/Assets −0.143*** −0.155*** −0.154*** −0.157*** −0.144*** −0.267***
(−3.86) (−4.17) (−4.14) (−4.29) (−3.95) (−6.68)
Idiosyncratic Risk −1.121*** −1.101*** −1.104*** −1.313*** −1.003*** −1.703***
(−5.61) (−5.45) (−5.48) (−6.33) (−5.26) (−5.94)
Equity Beta −0.005 −0.003 −0.004 −0.005 −0.002 −0.002
(−0.69) (−0.44) (−0.54) (−0.66) (−0.24) (−0.22)
Industry-by-year fixed effects Yes Yes Yes Yes Yes Yes
Firm fixed effects Yes Yes Yes Yes Yes Yes
No. of observations 5,481 5,426 5,426 5,425 5,426 5,469
Adjusted R2 0.740 0.741 0.741 0.745 0.746 0.633

Notes. We report regressions using firm-year observations for the two-sided four-year window around a firm’s inclusion in the Russell 2000
from below (in year 0). The sample period is 1990 to 2016. EBIT/Assets is EBIT over assets. Net Income/Assets is net income over assets. Large
Increase Transient IO equals one in all firm years for firms that experience an increase in Transient IO that is above the sample median, measured
over the two-year period around a firm’s inclusion in the Russell 2000 from below, and zero otherwise. Post Inclusion equals one for firm-year
observations after a firm gets added to the Russell 2000 from below, and zero otherwise. Year 0 equals one in the firm year in which a firm gets
added to the Russell 2000 from below, and zero otherwise. The coefficient on Δ Market Cap Rank is multiplied by 100. t-statistics are based on
robust standard errors clustered at the firm level.
*p < 0.10; **p < 0.05; ***p < 0.01, based on a two-tailed test.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4546 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

Table 5. Short-Term and Long-Term Valuations Around Russell 2000 Inclusions

Panel A. Short-term valuations Panel B. Long-term valuations

Dependent variable Market-to-Book Ratio Market-to-Book Ratio

Large increase No large increase Large increase No large increase

All Transient IO Transient IO All Transient IO Transient IO
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

(1) (2) (3) (4) (5) (6)

Decrease R&D 1.507*** 2.043** 0.846 — — —

(3.54) (2.77) (1.23) — — —
Year +1 × Decrease R&D — — — −0.609 −1.195** −0.005
— — — (−1.56) (−1.99) (−0.01)
Year +2 × Decrease R&D — — — −1.118*** −1.578** −0.380
— — — (−2.79) (−2.52) (−0.64)
Year +3 × Decrease R&D — — — −1.653*** −2.199*** −1.122**
— — — (−3.71) (−3.14) (−2.06)
Year +4 × Decrease R&D — — — −1.771*** −2.520*** −1.013
— — — (−3.70) (−3.30) (−1.65)
Year +1 — — — 1.329** −2.470** 1.407***
— — — (2.05) (−2.44) (3.60)
Year +2 — — — 1.991 −5.927*** 2.301***
— — — (1.59) (−3.24) (3.59)
Year +3 — — — 2.762 −9.669*** 3.830***
— — — (1.47) (−3.58) (4.24)
Year +4 — — — 3.384 −13.233*** 4.743***
— — — (1.36) (−3.69) (4.05)
Log(Market Cap) 0.545 0.815 0.956** 0.009 0.084 −0.184
(1.67) (1.30) (2.64) (0.06) (0.33) (−0.64)
Δ Market Cap Rank −0.232*** −0.275*** −0.176* −0.164*** −0.196*** −0.142***
(−4.96) (−3.16) (−1.85) (−12.73) (−8.93) (−8.04)
Stock Return (t − 1) −0.036 −0.406 0.114 0.118 0.066 0.035
(−0.18) (−0.87) (0.24) (1.21) (0.52) (0.24)
Dividends/Net Income −0.660 −2.047 −0.534 −0.066 −0.167 0.007
(−1.67) (−1.66) (−0.86) (−0.44) (−0.80) (0.03)
Debt/Assets −2.259** −2.313 −3.670* 0.357 0.444 0.022
(−2.35) (−1.65) (−1.88) (0.43) (0.34) (0.02)
Idiosyncratic Risk 0.101 0.111 0.084 −0.130** −0.092 −0.246***
(0.73) (0.51) (0.23) (−2.34) (−0.84) (−3.35)
Equity Beta 0.400 0.463 −0.198 −0.395** −0.496* −0.327
(1.55) (0.80) (−0.47) (−2.43) (−1.72) (−1.37)
Share Turnover 0.213** 0.102 0.249 0.279*** 0.164* 0.471***
(2.45) (0.53) (1.34) (4.02) (1.77) (4.10)
Industry-by-year fixed effects Yes Yes Yes Yes Yes Yes
Firm fixed effects No No No Yes Yes Yes
No. of observations 826 363 372 4,063 1,789 1,750
Adjusted R2 0.454 0.449 0.399 0.667 0.642 0.696

Notes. Panel A reports regressions for the year (year 0) in which a firm gets added to the Russell 2000 from below, whereas panel B reports
regressions for the [0; +4] year window around a firm’s inclusion in the Russell 2000 from below. The sample period is 1990 to 2016. Columns (2) and (3)
as well as (5) and (6) separate the sample according to whether a firm experiences a large increase in Transient IO around the index inclusion. Market-to-
Book Ratio is the market value of equity over the book value of equity. Decrease R&D equals one in all firm years for firms that reduced R&D/Assets from
two years before to two years after the inclusion in the Russell 2000 from below, and zero otherwise. Large Increase Transient IO equals one in all firm
years for firms that experience an increase in Transient IO that is above the sample median, measured over the two-year period around a firm’s
inclusion in the Russell 2000 from below, and zero otherwise. Year +1 equals one in the first firm year after a firm gets added to the Russell 2000 from
below, and zero otherwise. The coefficient on Δ Market Cap Rank is multiplied by 100 and the coefficient on Idiosyncratic Risk is divided by 100.
t-statistics, calculated in panel A (panel B) based on robust standard errors clustered at the year (firm) level, are reported in parentheses.
*p < 0.10; **p < 0.05; ***p < 0.01, based on a two-tailed test.

examine whether valuations are affected differen- columns (2) and (3) separate the sample according to
tially depending on whether firms cut investment. Col- whether firms experienced a large increase in Transient
umn (1) estimates the effect for all firms, whereas IO around index inclusion.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4547

Column (1) in Table 5 shows that inclusion-year same trend across firms with large and small increases in
equity valuations are substantially higher among firms transient ownership before index inclusion. Columns (4)
that cut R&D. Such firms trade at equity valuations through (6) further show that both sets of firms have
that are 1.5 units higher than those of other firms, statistically indistinguishable R&D rates, earnings, and
a difference that equals 39% of the equity multiple’s equity valuations before index inclusion.
standard deviation. A comparison of columns (2) and (3)
shows that the effects of R&D cuts are strongly con-
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

4.5. Short-Term Investors and CEO Incentives

centrated among firms that also experienced a surge The channel underlying Bolton et al. (2006) is that
in transient ownership. short-term investors incentivize managers to act myopi-
Although these results are consistent with mana- cally by granting short-term incentives. This allows
gerial myopia, an alternative explanation holds that managers to personally benefit from any temporary
short-term owners pressure CEOs to optimally re- increases in equity valuation. Ideally, we would test
duce investment that has previously been too high this prediction among our Russell 2000 inclusion
(Laux 2012), in which case equity valuations would firms. Unfortunately, comprehensive data on CEO pay
also increase at firms that undertake R&D cuts. We for such firms is not available in standardized data
distinguish between these competing views by ex- sets. To still provide some suggestive evidence, we
amining the effect of R&D cuts on long-term equity use data on CEOs in ExecuComp to link proxies for
valuations. If investment cuts are myopic, then equity short-term incentives to short-term ownership:
valuations should decline as the costs of the reduced
investment are gradually revealed. Conversely, if Short-term Incentivesft
π₁Transient IOft
investment cuts are efficient, then equity valuations + π2 X ft + λf + µft + εft , (2)
should not drop in the long run.
Table 5, panel B, reports market-to-book ratios for where Short-term Incentives measures new grants of
each year over the four-year period after a firm’s in- Option Pay or Equity Pay. We either include these vari-
dex inclusion, comparing effects between firms with ables directly or scale them by total pay. We use these
and without R&D cuts. We estimate all effects rela- measures to capture short-term incentives, because more
tive to equity valuations in year 0, the year of index in- pay in the form of options or restricted stock provides
clusion. The estimates reveal that R&D cuts have sig- CEOs with greater incentives to increase the stock price
nificantly negative effects on future equity valuation. by boosting short-term earnings (Cheng and Warfield
Among all firms in column (4), valuations start to decline 2005). CEOs can profit from such behavior by sub-
in first year after inclusion (−0.6), with the decline ac- sequentlyexercisingoptionsand sellingequity.Asbefore,
celerating in the second (−1.1), and the valuations bot- Transient IO measures ownership by transient in-
toming in the third (1.7) and fourth (−1.8) year. Im- vestors, Xft is a vector of firm-level controls, λf are firm
portantly, columns (5) and (6) show that the valuation fixed effects, and µft are industry-by-year fixed effects.
decline is strongest among R&D-cutting firms that Internet Appendix Table 3 documents a positive cor-
experienced a large increase in transient ownership; relation between short-term ownership and short-term
these are precisely the firms that also saw the biggest incentives: firms with more transient investors grant their
surge in short-term valuations at index inclusion. The CEOs more options and restricted stock. We interpret
long-term decline in valuations after index inclusion is these results as associative rather than evidence of any
similar to (or even slightly exceeds) the short-term surge causal effects. The finding that transient ownership is
in the inclusion year (see Table 5, panel A). positively related to short-term incentives is consistent
with Dikolli et al. (2009). It is also related to Cadman
4.4. Parallel-Trends Assumption and Sunder (2014), who document this relation in the
A key identifying assumption for our tests is the context of venture capital investments.
parallel-trends condition. This condition requires that
our outcome variables follow the same trends across 5. Conclusion
firms with large and small increases in transient We provide evidence that the presence of short-term
ownership before index inclusion. In other words, our investors is associated with cuts to long-term invest-
strategy assumes that the evolution of the outcomes ment in order to generate higher earnings, leading to
of firms with a large transient-ownership increase temporary boosts in the stock price. Our findings extend
after index inclusion represents the counterfactual for the work by Bushee (1998) and are consistent with the
firms with a small increase. This condition cannot be model in Bolton et al. (2006). We estimate the effects of
tested directly, but we provide evidence in Table 6 short-term ownership within a sample of stocks added
that R&D, earnings, and equity valuations are not di- to the Russell 2000 from below. Such Russell 2000 ad-
verging in the years before index inclusion: columns (1) ditions are accompanied with a large and persistent
through (3) show that these variables exhibit the increase in firm ownership by short-term institutions.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4548 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

In a first step, we showed that firms that experience a in transient ownership. At the same time, such firms
large increase in transient ownership around index in- experience a long-term decline in equity valuations once
clusion reduce R&D spending and increase earnings, the costs of the reduced investment are gradually
relative to firms that do not experience the same in- revealed. This reversal pattern is consistent with tem-
crease. In a second step, we documented that short- porary price distortions caused by the effects of short-
term equity valuations are higher in the year of index term investors on R&D. In a last step, we showed that
inclusion at firms that cut R&D, an effect that is strongest short-term investors incentivize CEOs to act myopi-
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

among those firms that also experienced a large increase cally by granting short-term incentives.

Table 6. Evaluation of Parallel-Trends Assumptions

Market-to-Book Market-to-Book
Dependent variable R&D/Assets EBIT/Assets Ratio R&D/Assets EBIT/Assets Ratio

(1) (2) (3) (4) (5) (6)

Year −1 × Large Increase Transient IO 0.002 0.015 0.810 — — —

(0.22) (0.74) (0.59) — — —
Year −2 × Large Increase Transient IO −0.005 0.011 −0.013 — — —
(−0.62) (0.54) (−0.01) — — —
Year −3 × Large Increase Transient IO −0.001 0.012 −0.268 — — —
(−0.07) (0.67) (−0.32) — — —
Year −1 −0.016 0.148 −1.223 — — —
(−0.75) (1.35) (−0.45) — — —
Year −2 −0.001 0.080 −1.512 — — —
(−0.07) (1.08) (−0.82) — — —
Year −3 0.000 0.032 −0.845 — — —
(0.01) (0.81) (−0.78) — — —
Large Increase Transient IO — — — 0.005 −0.005 0.128
— — — (0.54) (−0.30) (0.20)
Large Increase IO — — — −0.008 0.036** −0.339
— — — (−0.89) (2.23) (−0.53)
Log(Market Cap) −0.001 −0.014 −0.249 −0.001 0.013* −0.029
(−0.14) (−1.06) (−0.38) (−0.31) (1.82) (−0.09)
Δ Market Cap Rank 0.000 −0.002* −0.109** −0.000 −0.001 −0.196***
(0.18) (−1.88) (−2.11) (−0.42) (−1.18) (−4.28)
Stock Return (t − 1) −0.001 0.013** 0.946* −0.005 0.025*** 1.520***
(−0.38) (2.03) (1.76) (−1.33) (3.30) (3.23)
Dividends/Net Income 0.002 0.017 −0.015 −0.003 0.016* 0.019
(0.61) (1.15) (−0.04) (−0.68) (1.72) (0.05)
Debt/Assets 0.040 −0.203*** −4.061 −0.078*** 0.050 −3.268**
(1.60) (−2.65) (−1.57) (−4.11) (1.14) (−1.97)
Idiosyncratic Risk −0.001 −0.000 0.005 0.011*** −0.035*** 0.663***
(−0.58) (−0.04) (0.03) (5.25) (−6.61) (3.78)
Equity Beta 0.002 0.013 0.867 0.019*** −0.033*** 0.472
(0.26) (0.91) (1.05) (3.34) (−3.12) (1.30)
R&D Missing −0.009 — — −0.086*** — —
(−0.80) — — (−9.47) — —
Share Turnover — — 1.376*** — — 0.681***
— — (3.80) — — (2.70)
Industry-by-year fixed effects Yes Yes Yes Yes Yes Yes
Firm fixed effects Yes Yes Yes No No No
No. of observations 1855 1838 1795 1855 1838 1795
Adjusted R2 0.898 0.795 0.602 0.347 0.236 0.222

Notes. We report regressions using firm-year observations for the [−4; −1] year window before a firm’s inclusion in the Russell 2000 (in year 0). The
sample period is 1990 to 2016. R&D/Assets is R&D expenditures over assets. EBIT/Assets is EBIT over assets. Market-to-Book Ratio is the market value of
equity over the book value of equity. Large Increase Transient IO equals one in all firm years for firms that experience an increase in Transient IO that is
above the sample median, measured over the two-year period around a firm’s inclusion in the Russell 2000, and zero otherwise. Year -1 equals one in
the firm year before a firm gets added to the Russell 2000, and zero otherwise. The coefficient on Δ Market Cap Rank is multiplied by 100 and the
coefficient on Idiosyncratic Risk is divided by 100. t-statistics based on robust standard errors clustered at the firm level are in parentheses.
*p < 0.10; **p < 0.05; ***p < 0.01, based on a two-tailed test.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4549

Acknowledgments Brian Bushee, Brian Cadman, Rüdiger Fahlenbrach, Florian

The authors thank Shiva Rajgopal (the editor), an anonymous Peters as well as participants at the American Finance As-
associate editor, two referees, Pedro Matos (AFA discussant), sociation (AFA) Meetings 2018 and at the European Finance
Zexi Wang (EFA discussant), Tobias Berg, Patrick Bolton, Association (EFA) Meetings 2016.

Appendix
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

Variable Definition Data Source

Post Inclusion Indictor that equals one for firm-year observations after a firm gets FTSE Russell
added to the Russell 2000 from below, and zero otherwise.
Year 0 Indictor that equals one in the firm year in which a firm gets added to FTSE Russell
the Russell 2000 from below, and zero otherwise. Year −4 to Year +4
are indicators that are defined accordingly and reflect the firm year
relative to the firm year of index inclusion.
Transient IO Percentage ownership of transient institutional investors. The Brian Bushee’s web page
measure was introduced by Bushee (1998, 2001). Transient
investors have high portfolio turnover and diversified portfolios.
We obtain the institutional investor classification data from Brian
Bushee’s website and calculate the percentage of a firm’s ownership
by transient institutional investors. Winsorized at 1%.
Large Increase Transient IO Indictor that equals one in all firm years for firms that experience an Brian Bushee’s web page
increase in Transient IO that is above the sample median, measured
over the two-year period around a firm’s inclusion in the Russell
2000, and zero otherwise.
Quasi-Indexer IO Percentage ownership of quasi-indexers. Quasi-indexers have low Brian Bushee’s web page
turnover and diversified portfolio holdings. The measure was
introduced by Bushee (1998, 2001). We obtain the institutional
investor classification data from Brian Bushee’s website and
calculate the percentage of a firm’s ownership by quasi-indexer
institutional investors. Winsorized at 1%.
Large Increase Quasi-Indexer IO Indictor that equals one in all firm years for firms that experience an Brian Bushee’s web page
increase in Quasi-Indexer IO that is above the sample median,
measured over the two-year period around a firm’s inclusion in the
Russell 2000, and zero otherwise.
IO Percentage ownership of institutional investors. Thomson Financial
Large Increase IO Indictor that equals one in all firm years for firms that experience an Thomson Financial
increase in IO that is above the sample median, measured over the
two-year period around a firm’s inclusion in the Russell 2000, and
zero otherwise.
Analyst Coverage Number of analysts making an earnings forecast for a stock (IBES data IBES
item NUMEST).
Δ Analyst Coverage Change in Analyst Coverage over the two-year period around a firm’s IBES
inclusion in the Russell 2000.
Share Turnover Number of a firm’s shares traded throughout the year, divided by the CRSP
number of shares outstanding. Winsorized at 1%.
R&D/Assets R&D expenditures (Compustat data item XRD) over total assets Compustat
(Compustat data item AT). Missing values are set to zero.
Winsorized at 1%.
R&D Missing Indictor that equals one if data on R&D expenditures (Compustat Compustat
item XRD) are missing, and zero otherwise
Decrease R&D Indictor that equals one in all firm years for firms that reduce R&D/ Compustat
Assets from two years before to two years after inclusion in the
Russell 2000 from below, and zero otherwise.
EBIT/Assets Earnings before interest and taxes (Compustat data item EBIT) over Compustat
total assets (Compustat data item AT). Winsorized at 1%.
Income/Assets Net income (Compustat data item NI) over total assets (Compustat Compustat
data item AT). Winsorized at 1%.
Log(Market Cap) Logarithm of the market capitalization of a firm’s equity in million $. Compustat/CRSP
Winsorized at 1%.
Δ Market Cap Rank Year-on-year change in a firm’s market capitalization rank. Compustat/CRSP
Market-to-Book Ratio Market value of equity (Market Cap) over the book value of equity Compustat/CRSP
(Compustat data item CEQ). Winsorized at 5%.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
4550 Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS

Appendix. (Continued)

Variable Definition Data Source

Stock Return Annual stock return. Winsorized at 1%. CRSP

Dividends/Net Income Dividends (Compustat data items DVC + DVP) over net income Compustat
(Compustat data item NI). Winsorized at 1%.
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

Debt/Assets Debt (Compustat data items DLTT + DLC) over total assets Compustat
(Compustat data item AT). Winsorized at 1%.
Idiosyncratic Risk Standard deviation of residuals calculated by regressing excess stock CRSP, Ken French’s web page
returns on the Fama and French 3-factors using the last 12 months
of daily return data before the fiscal end month. Winsorized at 1%.
Equity Beta Equity Capital Asset Pricing Model (CAPM) beta calculated by CRSP
regressing stock returns on the excess return on the market using
the last 60 months of monthly return data before the fiscal end
month (with a minimum of 24 nonmissing observations).
Winsorized at 1%.

Endnotes References
1
Membership in the two indexes is based market capitalization Aghion P, Van Reenen J, Zingales L (2013) Innovation and in-
ranks at the last trading day in May. Each stock’s index weight stitutional ownership. Amer. Econom. Rev. 103:277–304.
is then determined according to end-of-June float-adjusted mar- Appel IR, Gormley TA, Keim DB (2016) Passive investors, not passive
ket capitalizations. owners. J. Financial Econom. 121:111–141.
2 Asquith P, Pathak PA, Ritter JR (2005) Short interest, institutional
We also explore the effects of the persistent increase in transient own-
ership on our outcome variables three and four years into the future. ownership, and stock returns. J. Financial Econom. 78:243–276.
These tests are useful because future changes in R&D or earnings are Beyer A, Larcker DF, Tayan B (2014) Study on how investment
less likely to be affected by simultaneity or reverse causality resulting horizon and expectations of shareholder base impact corporate
from contemporaneous changes in transient ownership. decision-making. Working paper, National Investor Relations
3
The increase in firm visibility due to wider analyst coverage after index Institute and The Rock Center for Corporate Governance, Stan-
inclusion may also directly trigger the buying by short-term investors. ford, CA.
4 Biktimirov EN, Cowan AR, Jordan BD (2004) Do demand curves for
Diamond and Verrecchia (1987) show that short-sale constraints can
small stocks slope down? J. Financial Res. 27:161–178.
impede the adjustment of prices to bad news.
Bolton P, Scheinkman J, Xiong W (2006) Executive compensation and
5
As firms are added to the index, index tracking funds buy newly short-term behavior in speculative markets. Rev. Econom. Stud.
added shares (implying an outward shift in the demand curve), 73:557–610.
which reduces the stocks available for non-indexing investors and Boone AL, White JT (2015) The effect of institutional ownership on firm
increases the price. If demand curves were horizontal, one should transparency and information production. J. Financial Econom.
not expect such a price increase around the index inclusion. 117:508–533.
6
The amounts of money benchmarked to the Russell 2000 are of Bushee BJ (1998) The influence of institutional investors on myopic
similar magnitude as those benchmarked to the Russell 1000 (Chang R&D investment behavior. Accounting Rev. 73:305–333.
et al. 2015). Bushee BJ (2001) Do institutional investors prefer near-term earnings
7
For example, if a firms is added twice to the index over a short over long-run value? Contemporary Accounting Res. 18:207–246.
horizon, then changes in our outcomes variables may be confounded Bushee BJ, Noe CF (2000) Corporate disclosure practices, institu-
by the changes caused by the second inclusion. We verify that in- tional investors, and stock return volatility. J. Accounting Res. 38:
clusion firms in the early years of the sample were not added to the 171–202.
Russell 2000 in the five years before the sample period starts. Cadman B, Sunder J (2014) Investor horizon and CEO horizon in-
8
Because transient ownership is stable in both time-directions away centives. Accounting Rev. 89:1299–1328.
from the year of index inclusion, results are unaffected if we define Chang YC, Hong H, Liskovich I (2015) Regression discontinuity and
the rise in short-term ownership over the [−3, +3] or [−4, +4] year the price effects of stock market indexing. Rev. Financial Stud. 28:
periods. However, this refinement reduces the number of observa- 212–246.
tions, because more data on Transient IO is available in years closer to Chen X, Harford J, Li K (2007) Monitoring: Which institutions matter?
index inclusion. J. Financial Econom. 86:279–305.
9 Cheng Q, Warfield TD (2005) Equity incentives and earnings man-
The individual effects of the transient ownership variable are ab-
agement. Accounting Rev. 80:441–476.
sorbed by the firm fixed effect.
Cohen L, Diether K, Malloy C (2013) Misvaluing innovation. Rev.
10
For the full sample of firms added to the Russell 2000, we continue Financial Stud. 26:635–666.
to find that a large increase in transient ownership is associated with
Crane A, Michenaud S, Weston J (2016) The effect of institutional
less R&D (Internet Appendix Table 1).
ownership on payout policy: Evidence from index thresholds.
11
Formally, the small-decline sample consists of all inclusion firms Rev. Financial Stud. 29:1377–1408.
for which −R&Dt-1 < (EBTRDt − EBTRDt-1) < 0. Derrien F, Kecskes A, Thesmar D (2013) Investor horizons and
12
The surge in earnings is not the result of higher sales: we find no corporate policies. J. Financial Quant. Anal. 48:1755–1780.
evidence that firms that experience a large increase in transient own- Diamond DW, Verrecchia RE (1987) Constraints on short-selling and
ership have lower Sales/Assets after index inclusion, relative to firms asset price adjustment to private information. J. Financial Econom.
that only experience a small increase (not reported). 18:277–311.
 Cremers, Pareek, and Sautner: Short-Term Investors, Long-Term Investments, and Value
Management Science, 2020, vol. 66, no. 10, pp. 4535–4551, © 2020 INFORMS 4551

Dikolli SS, Kulp SL, Sedatole KL (2009) Transient institutional Frankfurt School of Finance & Management, Frankfurt am Main,
ownership and CEO contracting. Accounting Rev. 84:737–770. Germany.
Edmans A, Fang VW, Lewellen KA (2017) Equity vesting and man- Laux V (2012) Stock option vesting conditions, CEO turnover, and
agerial myopia. Rev. Financial Stud. 30:2229–2271. myopic investment. J. Financial Econom. 106:513–526.
Fich EM, Harford J, Tran AL (2015) Motivated monitors: The im- Lynch A, Mendenhall R (1997) New evidence on stock price effects
portance of institutional investors’ portfolio weights. J. Financial associated with changes in the S&P 500 Index. J. Bus. 70:351–83.
Econom. 118:21–48. Mullins W (2014) The governance impact of index funds: Evidence
Fuller J, Jensen M (2010) Just say no to Wall Street: Putting a stop to from regression discontinuity. Working paper, Rady School of
Downloaded from informs.org by [188.90.135.223] on 06 October 2024, at 03:17 . For personal use only, all rights reserved.

the earnings game. J. Appl. Corporate Finance 22:59–63. Management, University of California San Diego, San Diego.
Gaspar JM, Massa M, Matos P (2005) Shareholder investment hori- Petajisto A (2011) The index premium and its hidden cost for index
zons and the market for corporate control. J. Financial Econom. 76: funds. J. Empirical Finance 18:271–288.
135–165. Piotroski JD, Roulstone DT (2004) The influence of analysts, in-
Graham JR, Harvey CR, Rajgopal S (2005) The economic implications stitutional investors, and insiders on the incorporation of market,
of corporate financial reporting. J. Accounting Econom. 40:3–73. industry, and firm-specific information into stock prices. Ac-
Greenwood R (2008) Excess comovement of stock returns: Evidence counting Rev. 79:1119–1151.
from cross-sectional variation in Nikkei 225 weights. Rev. Fi- Reed A (2013) Short selling. Annual Rev. Financial Econom. 5:245–258.
nancial Stud. 21:1153–1186. Scheinkman JA, Xiong W (2003) Overconfidence and speculative
Harford J, Kecskes A, Mansi S (2018) Do long-term investors improve bubbles. J. Political Econom. 111:1183–1220.
corporate decision making? J. Corporate Finance 50:424–452. Schmidt C, Fahlenbrach R (2017) Do exogenous changes in passive
Harris L, Gurel E (1986) Price and volume effects associated with institutional ownership affect corporate governance and firm
changes in the S&P 500 list: New evidence for the existence of value. J. Financial Econom. 124:285–306.
price pressures. J. Finance 41:815–829. Shleifer A (1986) Do demand curves for stocks slope down? J. Finance
Hong H, Lim T, Stein JC (2000) Bad news travels slowly: Size, analyst 41:579–590.
coverage, and the profitability of momentum strategies. J. Finance Stein JC (1989) Efficient capital markets, inefficient firms: A model of
55:265–295. myopic corporate behavior. Quart. J. Econom. 104:655–669.
Ladika T, Sautner Z (2019) Managerial short-termism and invest- Von Thadden EL (1995) Long-term contracts, short-term investment,
ment: Evidence from accelerated option vesting. Working paper, and monitoring. Rev. Econom. Stud. 62:557–575.