The EconomistAugust 8th 2020 Business 492 trant’s share price. Thereafter, one theory
goes, increased common ownership of
companies by large index funds, which
hold stakes in listed rivals, reduces com-
petitive pressure and leads to juicier share-
holder returns all around.
However, as passive ownership of as-
sets in mutual and exchange-traded funds
has swelled—from around 3% of their port-
folios in 1995 to more than 40% at the start
of 2020—the fillip from inclusion in the in-
dex appears to have weakened. It may even
have become a burden.
That, at least, is the conclusion of a new
paper by Benjamin Bennett of Tulane Uni-
versity, René Stulz of Ohio State University
and Zexi Wang of Lancaster University.
They looked at freshly minted s&p 500
firms between 1997 and 2017, five days be-
fore their ascension, then up to a year later.
In the first decade of the sample, index in-
clusion increased “abnormal” stockmarket
returns—ie, those relative to pre-inclusion
expectations—by up to 4.9% in a firm’s first
week in the index. This outperformance
then gradually disappeared over 12
months. In the second decade, by contrast,
the initial pop was negligible. Worse, ab-
normal returns turned negative within
three weeks, and stayed there.
The researchers suggest two possible
reasons. First, unlike active stockpickers,
passive investors are indifferent to the
price of a firm’s shares—if it is in an index,
they must hold them. What they pay thus
conveys less information about their view
of a firm’s strengths and weaknesses than
an active stockpicker would. Distorted
market signals may make managers over-
confident, warp investment decisions and
lead to shabbier long-term earnings—
which share prices in principle reflect. Sec-
ond, an increase in passive holdings as a
proportion of a firm’s shares dilutes the
power of active (and activist) shareholders.
Since they hold managers to account, cor-
porate governance can slip after joining an
index, also hurting long-term prospects.
Tesla, and other s&p500 wannabes, had
better take note. 7
Standard deviation
S&P 500, cumulative abnormal returns*
related to index inclusion, percentagepointsSource:“DoesJoiningtheS&P 500 IndexHurtFirms?”,by
B. Bennett, R. Stulz & Z. Wang, NBER working paper, July 2020*Differencebetweenactualandpredictedreturn6
4
2
0
-2
-4
151050-5 21
Days before/after inclusion2008-171997-200795% confidence
interval
For weeks rumours have swirled that
Nvidia, an American company which
designs and sells computer chips, wants to
buy Arm, a firm which produces the funda-
mental blueprints on which most such
chips are built. Arm, which is based in Brit-
ain, has since 2016 been owned by Soft-
Bank, a Japanese technology conglomer-
ate, which bought it for an extraordinary
$32bn—the highest price ever paid for a
European technology company. On August
3rd news reports said that Nvidia was close
to a deal. It may splurge as much, or more.
That is because Arm is not a normal
company. The firm’s core products are a set
of fundamental designs for computer
chips called instruction-set architectures
(isas). Arm sells access to isas to the likes
of Apple, Qualcomm and Huawei, giving
those firms freedom to design and manu-
facture Arm chips however they want. The
powerful chips in Apple’s iPhones are the
product of this process, as are those in just
about every smartphone in the world. Arm
also creates its own chip designs, which it
calls “cores”, and licenses them to compa-
nies that need a cookie-cutter starting-
point for chips to put in their devices, as
well as cars, connected fridges or anything
else hooked up to the internet. As a result,
Arm is everywhere.
Oddly for the sole supplier of technol-
ogy that undergirds the global semicon-ductor industry, Arm makes relatively little
money. In 2019 it pulled in less than $2bn in
revenue. Nvidia, which faces plenty of
competition, generated $2.8bn in net pro-
fit, mostly by selling powerful specialised
chips used to play video games, or bolted
onto existing machines in data-centres run
by Google, Microsoft and other cloud-com-
puting providers, where they are used to
train artificial intelligence.
Chipmakers that currently rely on un-
fettered access to cutting-edge chip de-
signs worry about losing it once Nvidia gets
hold of Arm. Under SoftBank, which is not
a chip company, none of this was an issue.
But if Nvidia, which is one, owned Arm, it
would exercise a degree of control over po-
tential rivals. Chinese chipmakers in par-
ticular may fret that the American firm,
whose operations are located almost en-
tirely in its domestic market, will give the
federal government power to block Arm’s
designs from getting to China, as long as
America’s tech standoff with its Asian rival
continues. Nvidia declined to comment.
Nvidia remains tight-lipped about what
exactly its intentions are. It has not con-
firmed it is in talks with SoftBank. One pos-
sibility is that it wants to combine its scale
and Arm’s designs to create general-pur-
pose chips that keep data-centres hum-
ming. At the moment this lucrative busi-
ness is dominated by Intel, another
American firm (Arm servers exist, but they
are mostly designed by firms like Tesla and
Amazon for their own purposes).
Nvidia could in principle compete with
Intel by licensing Arm’s designs without
buying the firm. Acquiring Arm would
hand it control over the technology road-
maps of countless high-tech firms. Many
companies would pay an arm and a leg for
that sort of power. 7Nvidia tries to get its hands on a
critical chip designerSemiconductorsWorth an arm and
a leg