Basement
:,ji>-^^<^^us^
V
HD28
The International Center for Research on the
Management of Technology
First
in Financial
Mover Advantages
Service Innovations
Luis E. Lopez
Edward
B.
Roberts
Sloan
I
We
2
WP
October 1997
2
^
WP#
#
167-97
3991
INCAE
MIT Sloan School of Management
appreciate funding assistance provided by
Research on
the MIT International Center for
and the
(ICRMOT),
of Technology
INCAE
the Management
MIT Center for Innovation
Product Development
EEC-9529140,
(National Science Foundation Grant #
October 1, 1996).
in
1997 Massachusetts Institute of Technology-
Sloan School of Management
Massachusetts Institute of Technology
38 Memorial Drive, E56-390
Cambridge, MA 02139-4307
FEB osigt
Abstract
A
well-established stream of research indicates that innovative firms must careliilly time
their entry into
new
markets, and that there are advantages to early market entry.
Such
empirical regularity has not been extensively tested in financial services, despite the
importance that the timing order of market entry might have for financial innovations.
Such importance derives
This
innovators.
fi^om the
characteristic
approach toward appropriation.
debit cards,
in financial
other
and pension
industries,
traditional
flinds),
services innovations.
particularly
weak
leaves
appropriability regimes faced by financial service
innovators
Analyzing three
we
lines
find important
These
with pioneering
of
financial
as
an
important
products (credit cards,
market share advantages to early entry
results are in line with other research
consumer products.
We
find,
model of order of market entry doesn't convey
all
performed
in
however, that using a
the information that
is
revealed through a detailed qualitative analysis of longitudinal data.
Moreover, traditional
models of order of market entry do not distinguish simultaneous entries fi"om entries
separated by long periods of time. A model using elapsed time since first entry seems
more appropriate and renders stronger
We
results
and better interpretations.
thank Professors Donald Lessard and Scott Stern for their
and suggestions.
Please direct questions or
comments
Luis E. Lopez
Apartado 960-4050
Alajuela, Costa Rica
e-mail: lopezl@mail.incae.ac.cr
to
many
thoughtfijl
comments
-]
in financial
advantages
se.,ces^
examine
To answer ..s ,ues„cn we
,he relations.,
products.
for a set of financial
long term n.arke. share
and
en.^
market
between order of
literature on pioneenng
provides a review of the
paper
the
of
pan
firs,
Accordingly, the
fo^ulate hypotheses and
second pan we
,„ ,he
.Kem
This
first entr^:
is
for test.ng
prov.tie a research design
and conclus.ons
followed by data analysts
,n
var.ahle pernnts ^uali^.ng
as an .ndependent
Our
findtngs tndicate a negat.ve
more
detati
such relat.onsh.p
Literature
InUoductioimidRe.vi'-w "' "i"
The
.Hare for
brand
and long term market
order of nnarke. en,,,
between
relauonsh.p
en,p,r,ca,
Urban ,1,92, analyzed ,8
new products
,s
well-established
product
entrants in S different
,n,o
,„at later entrants
i.,ance, as shown
in
new markets
Table
.,
m
Kalyanaram and
packaged goods and found
categones of consumer
suffer long-term
market share d.sadvantages,
For
the ptoneer
brands in the market,
the presence of 7
occur only through advantages
whether such advantages
,n
producon
costs, advents.ng,
for
to find after correcting
the product hne, only
of
breadth
and
pnce ,ual.ty, distnbutton,
an
imponant, and to d.scover
infiuences are not
such
that
vanables.
,He effects of those
for late entl^.
apparently inherent penalty
Table
1.
Order of entry
vs.
market share (Taken from Kalvanaram and Urban. 1992)
advantages
A .nmm^rv of some selected emEiri^^' studies on pioneering
T.hiP o
Characteristics of the study and findings
Study, year
Bond and Lean,
1
977
Reviewed
first
1
1
new
entrants maintained market share leadership
years in market
Found
producb. in two prescnption drug markets
after
more
that
10
tlian
Later entrants needed to otTer substantial additional
benellts to surpass the first-mover.
held over
Studied 7 cigarette markets and found that pioneenng brands
Wluttcn, 1979
50%
of the market several years aller
llieir
introduction.
These cigarette
and widely
brands were also those which were actively promoted
dislnbuted.
Kohiiison and I-onicll, 1985
Anahyed
Irbun cl al
1
,
average,
first-
market share of
movers had a market share of 2(1%. early followers had a
17%, and
I
On
mature consumer goods businesses.
.''71
late entrains licld
1
3%
of the market
hnportant penalties
Aiialwed 129 enlnes across 34 product categones
for product
controlling
Table
(see
)
entrants
were reported for late
9S6
1
positioning and marketing
Robinson,
.ambkin,
1
activity.
Analwed 1209 mature induslnal goods manufactunng
9X8
businesses. Tlie
Earlv Ibllowers and
market pioneer holds 29% of the market on average.
respectively
market
the
15%
of
and
late entrants retain 21%
on29
Considered 2 subsamples
I'JHK.
start-up businesses (data for their
first
1X7 adolescent businesses (data for their
fi.ur years of operation) and
an important market
second lour years of operation). Pioneers attained
vs. 9.7% for
pioneers
96"/n
for
(23.
entrants
advantage over later
share
m
late entrants
start-up
subsample and 32.56%
vs.
12.95%
m
the
adolescent subsample).
Lihcn and Yium,
l')9().
The likelihood of success
for first
and second entrants was lower than
and fourth entrants in a sample ol
the likelihood of success for the third
112
Colder and
Tellis. 1993.
1
994
Hutr and Robinson, 1994
industrial products
from 52 French firms.
Using an historical approach found
m
Brou-n and Lattin,
new
success.
They
Modeled the
report a
efiect of
47%
time
pioneenng
may
Re-anaK-zed Urban
el al "s
product categones
Found
efforts
market on pioneering advantage
exert a beneficial efiect
that lime-in-market
market share reward
in
that not all
ended
failure rate for pioneers.
Found
upon the pioneer
(1994) data on 95 surviving brands in 34
that longer lead-time increases the pioneer's
appear .o be qu,.e consistent
acoss many produc,
lines,
as can
be seen
.n
Table
2.
generalizations that applied
(1995) summarized some
Urban
and
Robinson,
Kalyanaram.
,0 prescriptton ami-ulce.
reproduced
in
Table
goods
dmgs and consumer packaged
These generalizations are
3.
OrdeorfjMlVfl^MIX^ncLl^^
Tables
^i^escri^tionmTtiriilcer drut;L
-p^;7;::;:^;q77;ri^ ^h.re Relative"^ the
-}
Pioneerin£B7^nd__
"
'
Prescription Anti-
UlcerDrm^s
ConsumerPackaMed^oodi
Kalyanaram and
Rerndlelal(1994)
Urban (1992)
L„.le
work
has deal, with services
existence of
nnancal mnova.ions, the
Tnfano (1989) found,
,He
life
,<,S9:
23.)
He showed
remam consistent
such advamage seems .o
of the product.
The same
innova.ions
imi.a.ions .ban wi.h
firm
sample of 58
pa.icularly
son,e first-mover advantages,
and marke.tng
.rad.ng, underwntmg,
form of lower costs of
„„f,„„
for a
,s
shown
in
,n
.ha. p.oneers
subsequent years of
less share
.0 cap.ure subs.an.ially
This can be observed
in
Table
4,
the
w„h
pnces for underwriting
p.oneers did not charge higher
Tufano's data showed that
market
prices, generated larger
charging s.m.lar or lower
than later entrants, and, by
from lower costs of trading and
advantages were reaped primanly
shares. Thus, pioneer
accrued through higher pncing.
not from mcreased revenues
from secondary trades and
prov.des banks w.th sources of
suggests that innovation
Tufano's (1989) study
search
to reduce information
.nformafon that allow them
difficulties
involved
in further pioneering.
others (as
be more innovative than
is
thus
in turn, that certain
The
also apparent in Table 4).
with additional
have not been confirmed
T.hle4
This suggests,
costs,
reducing the
banks tend to
findings,
however,
studies,
Percema<4ec£the.ddiM,Yalu^^
ji^e^sitein^Shlhe^^^
Markets
in
wluch bank pioneers the
Number
an
of
Average
share as
Numhcr
A>erage share as
of
products
follower
pioneer
(%)
imitated
(%)
innov ated
Boston
Morgan
Goldman Sachs
Slicarson Lehman
Stanley
Memll Lynch
Smith Barney
ofTers
products
Salomon Brothers
First
which bank
in
imitative product
product
Investment hank
Markets
Hams
Blyth Eastman Dillon
47.8
16
7
6
54.4
12
187
168
18 3
45,3
12
5
35.5
10
12.4
4
47.3
15
15 8
4
3
37.4
12
23.0
54.7
2
2
2
61.6
Drcxcl Burnham
1
Bear Stearns
Shcarson Locb Rliodes
1
1.4
42 9
1
15.1
69.2
7
3
4.5
11.7
62.2
0.6
1
1
154
47.9
Mean
\,i7a^,
luiaiiu (1989)
om Tufano
Taken from
Smdying innovation
should be enlightening
in
of
services from .he perspective
tlte
pioneering literature
se.ice oriemed,
As economies become more
the
R&D
fttnct.on
within service firms emerges.
Practitioners should benefit fi"om guidelines that can help to
determine expected returns fi'om products entering markets
at different
stages of market
evolution
Moreover, the need to explore these issues
particularly interesting in services
is
because there are possibly important differences between services and consumer or
industrial products.
These differences occur along many dimensions.
products are bundles of physical
objects
and intangible attributes
however, have a high degree of intangibility. This makes
it
For example,
Many
difficult to assess
all
services,
the needs of
the customer before creating a product and to measure objectively the satisfaction derived
from purchasing a service (Bitran and Lojo, 1993)
This thereby adds complexity to the
design and testing of services, thus increasing the risk associated with
1987)
development (Anderson,
moment of
are generated, they are
Moreover, some services must be delivered
accommodation, health
consumed
many
at
once
care,
The
and the
like.
When
such services
o^ producing and consuming the
act
cases, simultaneously,
and such services cannot be
product occurs, therefore,
in
inventoried for later use
Although the assets necessary to generate a
ser\'ice can,
course, be kept ready to deliver, the product per se cannot be delivered except
(thus,
yesterday's
the
at
For instance, customers must be present to receive services
their inception
like transportation, hotel
new product
unused hotel room or
airline
seat
is
lost
forever and
in real
of
time
cannot be
inventoried for later usage).
It
is
customer as
to
humans
this reason,
also salient that
their
many
services have to be delivered by humans, have the
primary input, or both
Patients are the
each delivery
is,
Health care, for example,
is
delivered by
humans
"raw material" of very complex production processes
by definition,
different.
For
As Bitran and Hoetch (1990: 89)
some
note, in
specifications-,
sei^ices
"i.
is
to
quality simply as •ccnfomiance
no. sufficien. .0 define
cannot be completely specified."
because the human encounter
Finally,
impossible').
(although, of course, no.
protect
difficult to
products or processes are very
newly created service
attribute this lack of
Bitran and Lojo (1993)
nature of services.
appropriability to the imangible
necessarily imply a difference
These differences do no,
of innovation
are
in all
The phenomenon of
innovation.
In financial services
dynamics of innovation:
highly relevant to the
a vety
envtronmen. and, most .mponan.ly,
(1992), for ms.ance,
creation
of many new
the
prices
advances
ulobalization
in financial
augmemed
changes
creatton of
new
argue that
m
m
Marshall and
weak appropnabUi.y regime
contributed to the
environmental factors have
environmental
volattli.y are:
^
^ong
technological advances,
of markets, regulatory changes,
in the speed, frequency,
schemes
regard, Tufano (.989)
,s
Pctruz,.,.
2
an incensing
to
For instance, the
variations requires the
and magnitude of price
manage
reporis, for a
in.crcs, inTlic poss,bi,i,i«s
the associated risk
This results
monopoly allowed
sample of 58 innovations
orpa.»n„6 nnancial
Del Vallc. and '"<"»"«''««'.
compiled a paniai
Finned) (1988). for instance,
seventies nnd 1988
the factors
frequem and abrupt changes
in
iisi
„,„„
oi mti
,„„
in a
more
to innovators.
the secunties
year after the ptoneer in
the market less than one
industry, that nvals enter
iThcrc
on
two parameters are
particular,
and Bausal, 1992,
theo^^ and others (Marshall
financial
that take
to the
increased volatility with respect
reduced duration of the transient
rapid dhftsion and a
In this
some parameters
the last 20 years
financal tnst^men.s durtng
responsible for such mcrease
,„
motivations and related mechanisms,
nature that have
mdus.ries, events of s.milar
1992)
different values (Geoffron,
Bausal
,ts
fundamental nature
in the
60%
innovations. See for
of the
e.a.ple
,„„„„ems created bcnvecn
,h= early
cases.
The mean number of
deals the pioneer
is
able to close before rivals enter
is in all
cases less than two.
The feebleness of
First, little
protection
in
from several causes.
the form of copyright or trade secret laws.
example, very few patents have been awarded for
Lynch was awarded a patent
Merrill
company was
allowed either
is
financial services, for
When
appropriabiiity regimes in services stems
able to obtain
for
Patents are awarded to tangible things
products.
Cash Management Account, the
its
on the grounds that
it
new
In
it
was
a novel
In financial services patents
computer program.
may
only be given to
-i
systems (for example, mechanical or electronic systems) that make the product
different or that serve to operate
In these
it
weak
appropriabiiity regimes,
somehow
new
services
are copied and improved after very short periods of time (Bitran and Lojo, 1993).
case of financial services
(SEC)
many
in
cases
the United States, the Securities and
in
forces
"inventors"
products they create (Tufano, 1989)
created and marketed everv' year.
is
In spite
As
have occurred
in
changes
cards), process innovations
EFTs
(ATMs
this,
numerous
..the
word revolution
in financial institutions
and instruments
financial
-automated
"
teller
new
machines,
CMA
-electronic fijnd transferring) and financial strategies
-cash
which implies the commitment of
development process,
Anderson and
is
a great deal
-leveraged
In this regard,
1989)
This
is
Clearly
of resources into the
undertaken as a profit-maximizing activity (Tufano,
Harris, 1986).
credit
management
(LBOs
financial products requires large financial outlays (Tufano,
financial innovation,
securities (zero
instruments (debit cards,
buyouts, swaps, corporate restructuring) (Finnerty, 1988, Miller, 1986)
developing
the
financial instruments are
These innovations include
the past twenty years."
coupon bonds, junk bonds), consumer-type
accounts,
of
Exchange Commission
information about
detailed
Miller (1986:459) indicates:
entirely appropriate for describing the
that
disclose
to
In the
1989,
carried out despite imitators needing to invest only a
and to
replicate a financial product
original investment to
fracion' of the innovator's
in the ntarket,
product
,„e
some
As a
law.
about infringing any paten,
wooing
cycles ,n
many
in
|<,89),
life
not
as
types of financial services,
This
is
especally
of secudties (Tufano,
for the case
we saw
many sendees
result, in
some, ephemeral.
industries are shc«, and in
sell .t
development cycles must be
almost instantaneously and
where products are copied
(Quinn, 1992).
reduced to the order of months
Levin
(1987)
al
et
competi.,ve advantages of
mechanisms firms use
argue
new
that
allernafve
use
firms
means
and products
or .mproved processes
,0 appropriate the returns
of mnova.ive
They
activities:
the
protect
to
1)
identify six
patents to
ume, 5)
tncome, 3) secrecy, 4) lead
patents to secure royalty
prevent duplication, 2)
moving quickly down
sales or service efforts
the learning curve, and 6)
effecve as a means of
Patents are not very
few industries
As Von Hippel(19S8 47)
: c lud ng
The patem
L tato"
grant
is
pro.ecng innovations except
indicates:
and/o! capturtng royalty
mcome The ans-r
no, useft.1 for either purpose
m
industnes
instance,
pha^aceuttcal industries but
in
patents
are
pari.culariy
ineffective or not
more
un.formly found
most mdustnes
patems do appear ,0 offer
Thou..h uenerally ineffect.ve,
For
very
in
some
effective
effect.ve than
,n
protection in ceriain
the
some
chemical
and
other alternat.ves
telecommunication industries.
the electronics and the
to offer
Similarly, secrecy appears
little
pro.econ
know-how trading among
repons the extstence of informal
325%
10
50%. Tufano 1990
to innovators
Von
Hippel
specialists ,n a given discipline
These
specialists
might find
economically reasonable to trade information that could be
it
considered proprietary, even with
Given the
(Von
rivals
of some forms of appropriation
relative inefficiency
firms probably utilize a combination of
That
effective.
on
depending
is,
appropriation methods
likely to
is
Hippel, 1988).
means of appropriation
the
industry,
methods.
ineff'ective.
Moreover, products, once released, are
(1993) report
financial
services,
In the financial ser\'ices industry,
In order to
inefficiency
of some
and
secrecy
are
nonexistent
relatively easy to imitate.
or
Bitran et
al.
ready by the time the
first
plausible alternative
one
way
is
institution introduces a
to assess the strengths and
it
is
in
type of
weaknesses of the
the service industry in terms of
new
important to have the next service almost
introduced.
to protect and derive rents
to introduce products that are hard to
new
very quickly, but often introduce an improved
be a leader
product introduction, therefore,
et al
patents
whenever an
it
some time
service since they have
original service
Bitran
be cost-
likely to
that:
account, others not only copy
A
relative
is
contexts,
be balanced with a greater emphasis placed on other
alternative
In
the
that
in certain
copy
This
is
difficult in
(1993) indicate, because products are not
from innovative
some
intrinsically
activity
is
financial services, as
complex
The process
of creation may be complex, involving high doses of creativity and the participation of
specialists
from
different areas, but replication
replicating a financial product
Even products
may
render them a
bit
is
is
easy.
on the order of 50% to
Tufano estimates
75%
Morone and Berg (1993)
difficult to replicate are
indicate,
"...
of
less than the innovator.
that require intricate technological or institutional
more
that the cost
networking that
not necessarily difficult to copy.
As
the relative benefits of pioneering versus fast
following are particularly open to question since information technology can often be
quickly copied, and later entrants often enjoy the benefits of lower-cost hardware."
Hence, adding complexity to products through a more intensive use of information
technology or other means
competitors'
will
movement down
Given
this,
we would
not necessarily cause delays in product difHision or in
the learning curve.
expect
that, at least in financial services,
be an important form of appropriation
if
not the dominant
protection and secrecy are ineffective and, for
down
If patent-related
and moving
practical purposes, absent,
the learning curve occurs almost instantaneously because of the relative simplicity of
the products, then lead time (being
new products
No
in financial
services
that,
related to order
in
at
consumer markets.
general, market share
is
For these types of products
a)
The dynamics of order of
entry
new
disregarding the characteristics of firms that launch the
The
are
a
has been
literature
sometimes
national distribution" (e.g..
by national advertising
Urban
(e.g.,
is
products.
et al.,
in
evaluated
Thus, whether
in
the
firm's
seldom studied.
identifies as first entrant a
product that
first
"achieves
1986) or requires that the product be supported
Whitten, 1979)
advertising or national distribution
of entry,
number of
sometimes
because they lack strategic relation with other products
portfolio, or for other reasons not related to order
b)
in
in\ersely (though not necessarily lineariy)
on pioneering advantages does, however, contain
literature
fail
it
saw,
of market entry
problematic aspects
products
we
on pioneering advantages has primarily concentrated, as
packaged goods aimed
The
to market) and a quick response in introducing
study, however, aside from Tufano's, has looked into this problem in depth
literature
determined
first
might be the most effective methods of appropriating the returns of
innovation efforts
The
all
one.
pioneering should
"Hence,
a national market,
11
if
it
a firm does not have national
does not qualify as an entrant.
(Robinson, Kalyanaram and Urban, 1994:2). This might be problematic because indirectly
it
means
that the product that qualifies as an entrant has
an important
set
of complementary assets
thus precluding
distribute),
methodology one runs
(i.e.,
been launched by a firm that has
being able to nationally advertise or
many products launched by
smaller firms.
By
the risk of confounding "entry" with financial muscle.
using this
The method
also censors entrants that have failed to stay in the market, thus overstating the effect of
entry on the survivors.
c) Units
of analysis tend to vary.
while others analyze organizations
Some
Making
studies center their attention
on products
generalizations from the literature as a whole
is difficult.
d) Censoring:
Many
studies cannot or
do not consider the products'
entire historv',
thus overestimating or underestimating pioneering.
e)
A
rapidly changing market
subsequent entrants
in
a
ver\'
may
defy order of
entr>'
dominance, especially
if
environment are able to incorporate into their
volatile
products characteristics that take such changes into account
f)
Finally,
premature market conditions
obscure the effect of order of entry
fijture
Pioneers
consumers or creating network
others to invade
As
may end up
have died
in
may
also
bearing the weight of educating
externalities yet unconsciously level the field for
though empirical
Some of
this
results
controversy
appear strong, empirical generalization
show
for a variety
their youth.
of products
Colder and
specified scale of commercialization and
is
discussed by Colder and Tellis (1993) who,
is
using an historical approach, argue that moving
contrary, they
pioneering "too early")
it.
a result,
controversial.
(i.e.,
first
that almost
Tellis
achieves no advantage.
all
of those
12
the
who have moved
first
to reach
any
do not require the pioneer
many of those pioneers do
On
not have a meaningfijl
share of the market
in
the long run.
information from articles
some
In their
as far back as 1842 to prove their point.
approach solves the problem of reverse causality
historical
being identified quasi-automatically as pioneers),
market
methodology, Colder and
that are not well
gather
Though
this
high market share firms
can also overlook other entries into the
documented.
Such aspects should be taken
study the phenomenon
it
(i.e.,
Tellis
when choosing an appropriate methodology
into account
to
in financial services.
Hypothesis
On
we
account of the aforementioned
hypothesize
first
mover advantages
for
financial services innovations:
H
There
of
is
a negative relationship
financial
services
between order of market entry and market share
innovations,
controlling
for
firm
size
and
product
interrelatedness.
This proposed relationship
been discovered
in
many consumer
products,
consumer packaged goods (Robinson,
Colder and
with the same empirical regularity that has
falls in line
et a!.,
in particular
the pharmaceutical industry and
1994), but runs contrary to the findings of
Tellis (1993).
13
Methods
The study
is
based primarily on historical analyses.
financial products are
Product histories for several
assembled and reconstructed using archival records.
These records
are information available in published sources of information (which include primarily
newspapers,
regulatory
institutions.
informants
in
and research papers) and from other sources such as
trade journals,
The
historic
information
is
then
launched
in
Costa Rica within the past 15 years.
lines (credit card, debit card,
sample
we
in
which product
sample of Costa Rican
we
and pension
fijnd)
A
is
is
a set
of
used
whose
in
Thus, through
three product
this
choice
This study
is
oi'
part
of commercial banks were reconstructed for
For the purposes of
financial institutions.
included only those products
done with the
line histories
financial services
sample of 36 entries
are trying to minimize the problem of lack of fine-grain data.
of a larger one
a
from
data
pertinent institutions and by others familiar with the industry.
For convenience of data gathering, the sample chosen
•"t
with
validated
histories
this paper,
were reconstructed
however,
This was
in detail.
intention of a) correctly identifying pioneers, b) avoiding censoring, and c)
choosing products which had been launched
potentially different market effects
in
roughly similar time-frames (to control for
when agglomerating
the sample).
The sample of products was circumscribed purposefully
to a small market
under
progressive deregulation, and, thus, historical information could be obtained to permit an
accurate determination of pioneering and subsequent entries.
which these products have flourished provided an
related innovation.
No more
Moreover, the
interesting case for studying service-
than 20 years ago, the commercial banking sector
Rica was a heavily protected and regulated domain.
domains of
activity
were allowed
14
to
in
Costa
Economic deregulation dramatically
changed the structure of the industry since 1980 and companies
specific
industr>' in
offer
new
that
services
were confined to
in
areas
that
had
traditionally
belonged to banks
offer a wider range
of products
AJI financial institutions
in
have been forced to renew and
order to remain competitive
and the development of new products
in
Over
15 years, innovation
the financial services industry have
become
increasingly important.
Our choice of sample
includes products that have been created
in
the near past.
This allows for a relatively simple process of information gathering from public records.
also
permits us to gather first-hand information from people
launching of the products
Another important advantage
dynamics of the dependent
product-lines
whose
variable,
is
that
we were
Thus,
within organizations
who may
not be entirely accurate.
able to assess the
Moreover,
we chose
comparison and contrast of
trying to avoid overreliance on single informants
overstate pioneering status, or
To avoid
participated in the
we were
market share, over time
history could be reconstructed through
several data sources.
who
It
whose
recollections might
these problems, multiple measures were collected both
from organizational informants and from secondary sources, allowing us to cross validate
and triangulate data from different origins (scholars, experts, journalists, etc
sources
who were
An
fact
not interested
in
be censoring the dependent variable, since
case the long term
Although
marketing
field,
may
sur\'ival
and
component
is little
has been widely used
Utterback
in their
is
plausible that not
that
we may
in
enough time has
In this
not yet be long enough.
particular interest are the studies
1992),
it
is
and performance of the different market entrants.
historical analysis
it
particularly
overstating the pioneering status of a particular firm.
important disadvantage with the chosen sample of products
elapsed for assessing the
),
(1994)
used
in
in
studies of pioneering carried out in the
studies of technological innovations.
Of
done by Cooper and Schendel (1976), David (1986,
AJI
these
work.
15
authors
incorporate
a
strong
historical
The dependent
variable in this study will be total market share
of the nth entrant
at
the time of the study, measured as percentage of total customers using a particular
financial
instrument
committed
and
reports
archival
A
periodicals.
financial,
Data sources for
to the financial instrument.
Archival
data.
manual search of
all
of
percentage
as
alternatively,
and,
total
of dollars
portfolio
this variable included informants'
were obtained primarily from
data
the issues of the four
most important
financial
local business,
and economic journals since 1985 (and a selective search of some older issues)
was coupled
to both electronic and
newspapers.
We
libraries, the
manual searches of the countries' most important
also searched several libraries in the country (for
Central
Bank
Librar>', the library
example
universities'
of the Stock Exchange Commission, and
several others) for sources of information which included books, unpublished theses and
papers, and
monographs
In
all
we were
able to assemble over 100 articles or printed
pieces of information which referenced the product lines under scrutiny
was
This information
validated with data provided by organizational level informants and three industry
For
experts.
pension
funds
Commission which required
portfolio
size,
information
we
all
we
contacted
pension
fijnd
number of customers, and
an
incipient
operators
the
like
in
Pension
Fund
Regulatory'
the country to report data
When we
on
found contradictory
either sought a third piece of confirmatory evidence or asked an industry
expert for clarification.
Because the products were recent, confirmatory evidence was
generally found with only a moderate degree of difficulty.
operationalized
as
the
Pioneers are
entry will be used as an independent variable.
The order of market
first
entrant.
Time of
entry
informants.
Most organizations had personnel who were
of entry of
their product.
was obtained using
able accurately to recall the time
This was confirmed with archival data and,
16
primarily
in
the case of
pension funds, with the date reported to the Pension Fund Regulatory Commission as the
first
year of operation
An
additional independent variable will be the elapsed time delay since the
The
entry until the nth entry.
rationale here
is
that the impact
vary depending on the time elapsed since the
we had
determined once
The
of being the nth entrant
entry.
first
first
will
Elapsed time was easily
dates of product introduction per institution.
institution's size, in terms
of
total
Number of employees
study, will be used as a control variable.
These
correlated with total assets
size
number of employees
measures are
leverage existing resources to support a product
reputation, and distribution channels
in
a
is
at
the time of the
highly and positively
proxy for the bank's
ability to
terms of advertising expenditures,
This information
was obtained through informants
within the organizations
rest
of the institution's products can also have a moderating
variable
People may be inclined to switch to certain institution's
Relatedness with the
effect
upon the outcome
products because of the added convenience of carrying on
organization
For
this
businesses with
purpose a measure of strategic focusing was used
of the point cloud on the client-product matrix of each
of product interrelatedness
may be
all
Institutions with larger/ampler arrays
better positioned to leverage
which quantified the change
institution
in the
new product
offerings'*
The dispersion
was used
as a
measure
of products and
An
one
clients
alternative measure,
dispersion of each bank's product cloud,
was
also used
as a measure of interrelatedness (small values indicating small departures fi"om current
product offerings).
Finally, a
dummy
hypothesized relationship.
variable
This
was used
for assessing the regulatory effect
dummy was
•Please sec Appendi.v 4 for details.
17
upon
the
assessed with the help of three industry
experts
who were
independently asked which of the product lines was influenced by
regulation (when operation required permission from the regulator, or
only permitted certain institutions to operate
The model
to be tested can be
a2
^nc
'
as:
a3 ^
^nc
nc
the regulator
the time of entry).
summarized
ai
^
^nc
•fe
at
when
a^
ctA
'^nc
'-•tie
!
where,
Snc = market share of the nth entrant
in
Enc" ^ Order of entr>' of nth product
Tnc "=Elapsed time
category
in
c, in
category
percent
c.
delay between nth entry and the
first
=Re!ative size of institution that released the nth product
Anc
as a ratio of total employees with respect to largest entrant
Rnc
Gpc
entry in category
"
in
for
^Regulatory variable (dummy
variable,
18
(measured
in
category
c.
dichotomous).
Following a section of exploratory data analysis,
regression models.
category' c
months.
category).
bank releasing product n
=Index of product relatedness
in
c, in
we
will
fit
a
taxonomy of
Results
Qualitative Analysis
In general,
ail
products examined here present different patterns of entry,
seemingly different relationships between order of entry and performance.
exit,
and
For some
products, particularly those which only add features or are exclusively driven by changes
the effect of pioneering
in regulation,
upon performance seems
strategic products that enter the market in the absence
expected
negative
relationship
between
order
to be small.
However,
of regulation reveal an apparent
of market
entry
and
measures
of
performance.
Pension Funds.
Pension funds
in
this
market had traditionally been administered by the
Economically active people contributed a fixed percentage of
fund.
In such environments,
their salaries to a national
private pension fijnds tend to be successful
operated pension system gives small pensions
at
high retirement ages.
customers would have an incentive to transfer to
Problems with the state-administered
fijnd
fijnds
that
opened an opportunity
offer
In
if
the state-
such a case,
more
until July
of 1995.
benefits.
for other institutions to
enter the market and offer complementary pensions This industry operated with
no supervision
state.
little
In fact, before that time, private pension funds
or
could
not operate openly as such, but only as providers of additional coverage to that given by
the state.
Even
of 1988.
In
so, the first private
pension
fiind
operator entered the market
in
August
1995 the funds were authorized to operate as such and also received an
19
additional impulse
in
the form of a tax break'.
between 1991 and September of 1996. Table
number of clients, and
Table 5
-i
5
Ten more operators entered the market
summarizes the order of market entry, the
the size of the portfolio for the last three years.
% of market
When
share
is
measured
in
terms of portfolio size (not shown on graph), the
pioneer also appears to retain an important portion of the market.
%
of
market
(Sept 96 as
total
%
of
customers)
2.5
Order
Figure
2:
Order of market
measured as
% of total
7.5
5.0
entrN' vs.
of
10.0
Market Entry
market share for pension funds
1996
customers as of Sept
Market share
Figure 3 shows a longitudinal history of total market shares (measured as
total
customers) for
in this
all
market for a
participants
total
The pioneer
of 23 months
in this
According
first
pension funds entrant gained a
maximum market
50%
of the pioneer's
share.
%
of
market was able to operate alone
Firms' market shares start leveling off after
approximately 4 years since the
entry.
is
.
to these
graphs and data, an early
share over time that
In this case, not considering the
is
approximately
one apparently atypical data
point (PF6), firms that entered the market within an elapsed time delay between 27 and 57
21
months eventually averaged 57% of the
pioneer's share.
Very
stabilized with market shares that represent approximately
(8%
share
appear to have
late entrants
5%
to
10% of
the pioneer's
in this case).
Order of entry
vs. marltet
share for pension funds
Entrant
10CW.
•-
90%
-
80%
-
70%
-
—V
60%
-
I
50%
—
—
-•
ia
-o
2nd
-—
-—
m\
3fd
-
-'
V
5»i
—
-a
30%
•
OT*
•
10%
-
-•
-
0% o-
—
em
7th
8th
S=^
92
93
Year
Figure
3
Order of market entry vs market share
for pension funds.
Other factors, however, particularly organizational
effect
upon market share
size,
seem
Figure 4 shows a scatterplot of order of
to have an important
entr>' vs.
percentage of
the pioneer's market share (measured as the relative size of its portfolio vs that of the total
market).
same
1996.
The squares
for the year
Symbol
represent this percentage for the year of 1994, the triangles say the
of 1995, and the
size
is
circles represent this percentage at the
an indication of organizational size (the larger the symbol, the larger
the size of the institution measured
in
total
scatter plot reveals an apparent influence
market shares.
end of August of
Notice that entrant number
of
number of employees)
size
five,
22
Inspection of this
upon the dynamics of an incumbent's
which
is
also the largest as
measured by
symbol
size, increases its
combining the pension
offer
market share noticeably
fiind business
also banks, but
8,
three years, signaling the possibility of
with a larger platform of products, thus being able to
more complete product packages
entrants 7 and
in
it
is
exclusively operates this pension fijnd) or
A
to customers.
similar effect
not observed in entrant
is
number
observed
three (which
number four (an insurance company). Size may
also signal a greater ability to leverage sales efforts and achieve scale effects.
relationship
to
between SIZE and market share
is
positive and significant for the years
The
1994
S-r=0 70, p<0 05, for 1995 S-r=0.63, p<0.1, for 1994 S-r=0 61,
1996 (for 1996:
p<0
1).
Entry vs
%of
pioneer's share 1996
1.2
Variables
"o ol pioneer'*:
PFl
share
O
9
PV(,
0.9
Entry vs.
^
Entry vs.
E
Entry vs.
%of
%of
%of
pioneer's share
1
pioneer's share
1
pioneer's share
1
PF2
PK5
6
0.5_
in
"P
J
-
PF7
0.2PF4
O
PF8
e
PF9
10
0.0
Order of entry
Figure 4 Order of market entry vs pioneer's share for three years
23
P ension funds
996
995
994
Table 6 shows a matrix of Spearman-rank correlation coefficients for order of
market entry and market share measured as a fraction of the pioneer's share
total
customers) for the years ended
in
(in
terms of
December of 994, December of 995, and August
1
1
1996.
Tabic
6:
ENTRY and the
Spcarman-rank correlations between the variable
\anablcs SIZE and market share for \anous \ears (Market share measured
as
% of the pioneer's share m terms of total customers)
Pension
Spearman-rank correlation
Variable
SIZE
-034
MARKET SHARE:
%PION%
-0.72*
%PION95
%PION94
-0
81**
-0
80**
fiinds.
coefficient.
**p<{).()l.*p<().()5
As expected,
a negative relationship
between order of entry and market
strong and statistically significant throughout
(only last three years shown).
these years the relationship conserves the expected direction, but
decrease (from 0.8 to
upon the outcome
relationship
72),
presumably because other variables
between order of market entry and SIZE.
ones) are not, according to these data, the
pioneers
fijnds
But these advantages
associated with
some
first
seem
are,
ones or the
fairly
Throughout
coefficient starts to
start exerting influence
Table 6 also suggests that there
variable market share
These data on pension
its
is
is
no
significant
Larger institutions (or smaller
last
ones to enter the market.
to indicate that there are
some rewards
for the
however, quickly overcome by other factors
organizational characteristics of subsequent entrants.
24
Later entrants
into this financial
market are not delayed by any
sort
of proprietary considerations, hence
the pioneer enjoys only a relatively short period to operate as a monopolist.
Debit Cards
Debit cards were recently introduced into the financial system under study.
payment instrument requires the cardholder
to
have a checking account with the issuing
institution, with the debit card ser\'ing to automatically
from the checking account
businesses.
The product
and electronically deduct payments
requires a fairly extensive network of affiliated
Moreover, debit cards become more
useflil
if
they can also be utilized to
obtain cash from automated teller machines or from bank offices
would expect
Hence, a
priori,
large banks, or banks with extensive geographical coverage to be
successfijl in this market.
Table 7
Raw
data for debit card industry
Tot. assets.
Operator
DCl
This
Dale of
Order of
Customers as
12/95
entr>
entr\
of 10-96
(mu)'
# employees
one
more
The
debit card entered the market being studied in
first
additional operators had entered the market by 1996 and another
enter soon.
The above data
commercial banks.
between the pioneer and the next two entrants
caused by regulatory changes that allowed
made
it
Six
two were planning
Table 7 summarizes some characteristics of these operators.
institutions are
This change
August of 1987.
to
All debit card
reveal a fairly short elapsed time
A wave
of new entrants occurs
in
1996,
banks to provide regular checking accounts.
all
unnecessary for banks to circumvent
this regulatory constraint,
which
they often did, through alternative and more creative means.
Figure 5
total
is
a scatterpiot that
market share measured as
industry' at
%
the time of the study
shows order of market
of customers for
all
entr>'
versus percentage of
participants in the debit card
Inspection of this scatterpiot doesn't suggest the
presence of a negative relationship between order of market entry and long term market
share
The pioneer
doesn't have any significant advantage.
able to capture market share
one
sixth
The
It
is
the third entrant
pioneer's market share, in this case,
of the third entrant's market share
A
is
who
is
approximately
very late entrant (with a total elapsed time
of 106 months after the pioneer's entry) appears to capture approximately 50 percent of
the pioneer's share in a relatively short period of time.
26
60
4
2
80
Order of market entry
Figures
datapoinis
No
Order of market entry
is
vs.
market share For debit cards
longitudinal
data could be obtained for this product
notwithstanding, an examination, even a cursory one,
factors
may be
inferred
that additional factors
of the
institution,
data pomt
The
size
of the
an indication of the relative size of the institution.
It
might be that
this
is in
much
determined by number of employees,
minimum
technological platform
is
In Figure 5 the size
represented by the size of the
is
Obviously, the very large institutions entered
suggesting that a
shorter life-cycle, and
of pioneering.
effect
this
market
in isolation
As
must be kept by the customer with the issuing operator.
drive customers to adopt one or the other debit card,
overcome the
effect
of eariy entry
number of ATMs each incumbent
first,
perhaps
necessary for firms to enter into this
market because convenience to the customer drives the adoption of
Moreover, the product cannot be adopted
This
particular.
order so that the effect of other
product has a
have already overcome the
in
a
minimum,
a
this
product.
checking account
Convenience, however,
and
this
factor
may
may
quickly
Figure 6 shows, through the size of the data point, the
has.
Entrant number three holds the largest
27
number of
followed by entrant number five with
those by
far,
nearly as
many
as entrant
5.
Though
60%
as many.
this relationship evidently
the cross-sectional information at hand, there
is
Entrants
1
and 2 have
cannot be established with
apparently a positive correlation between
market share growth rates and # of ATMs.
Market
sluirc (as
%
of pioneers).
4
2
6
Order of market entry
Figure 6
Order of market entry and market share
The
debit cards.
size
of the data point
is
as
% of pioneer's market share for
an indication of the number of ATMs each
participant has
The
entry.
entries
effect
of
entPy'
may
also be influenced
The absolute order of market
entry could hide nearly simultaneous entries or
which are separated by long periods of
market share
at
by the time elapsed since the pioneer's
time.
In Figure 7
we
have plotted
the time of the study against elapsed time since the pioneer's entry.
groups can be cleariy recognized
in
the graph.
One group of
Two
eariy entrants introduced
debit cards into the market within a total elapsed time of 17 months.
28
total
The next entry
occurred 58 months
and more than eight years
later,
after the pioneer's entry
we
can
observe a second group of institutions entering the market.
0,8
Market
sliarc.
-,
1996
•1.
85
50
15.0
120
Elapsed months since pioneer entered
Figure
7:
Elapsed time since pioneer's entry vs
Two
observations are
inconveniences of utilizing
in
order
First,
total
the data on debit cards indicate the potential
order of market
strict
might be preferable to moderate
this strict
market share for debit cards
entry' as
Alternatively, time elapsed since
independent variable.
Second,
different institutions entering a
strategic while the later
wave of
first
entry could be used as an
can see that different motivations
market
In this case, early entrants
entrants
It
order by coding entries as early entries, early
followers, and so on
we
an independent variable.
may
chan^e.
29
just
may
may
stand behind
well be termed as
be taking advantage of a regulatory
Despite
this,
we do
significant relationship
8
observe for the case of debit card introduction a
between order of market entry and long term market share
shows a matrix of Spearman-rank
correlation
coefficients
interest.
We
indicative
of total number of customers), the two variables
total
%PION_SHR
number of employees),
total
and
%
Notice that market share (as a
that
the
time elapsed since
for
is
measure
SIZE! and -0.96 with p<0 001
between market share and number of
for
The observed
SIZE2).
ATMs
is
come
ATMs)
positive
relationship
p<O.OI.
Though
between order of market entry and
may
models
give an incorrect picture of the
though
in this
case no firm conclusion can be drawn, that
into play that eventually dilute the pioneering effect.
case of debit cards
network of
-0 86, p<0.01
observation.
Finally, the data suggest,
other factors
negatively
also important to note that examination of statistical
alone, without further expIorator>' data analysis,
phenomenon under
=
S-r
statistically significant at
the table generally supports a negative relationship
is
and
are using ranks, with
negatively correlated with both measures of entrant size:
it
is
terms of
ELAPSED
is
of
In this case the large established players entered first (time
first entr\')
long term market share,
(which
size (in
of the pioneer's share)
we
Table
variables
CLIENTS
variables
correlated with order of market entr>' (and similarly, given that
elapsed
for several
have included for reference purposes the variable
and
assets
statistically
we
observe that entrants with larger coverage
are apparently able to
data also suggest that, for this product, a
grow much more
much
30
In this particular
(in
the form of a
rapidly than pioneers.
shorter transient
monopoly
The
existed than
for pension funds.
Three firms entered the market within a period of 17 months.
second wave of entry, presumably driven by changes
months
after.
are less than
Table 8
in
regulatory conditions, occurred 58
Nonetheless, very late entrants appear to stabilize with market shares which
8%
of total.
Spearman-rank correlation coefficients for variables of interest.
Debit cards
A
all
banks
in
1979
in
the United States offered a credit card plan (Federal Reserve Bulletin, 1973),
71%
(Mandell,
of all banks had
(BankAmericard
shown
in
By 1976
1990).
Figure
(later
8.
and
credit card plans;
there were
VISA) and Master
in
1985 the figure had grown to
two dominant organizations
Card).
Notice that explosive growth
The growth of
in
90%
the business
their total
accounts
is
starts after 1975.
r^f^.r^r-^r-^rv.'^*''*-'''^ CO GO ^
CO
co
aa
^ <c as o^ oi o>
O^O^a^C^o^o^<T'O^C^tcT>0^0^<x>o^CTlcno^O^O^O^O^CTlCTl
oo
fv.
Figure
8:
Groulh of Credit Card accounts
in
the
US (VISA and Master
Card)
Taken from
Krumme. 1987
The
first
credit card plan in the financial system
by a commercial bank
(this is a relatively early
second entrant appeared
advantages
in
may have been
the market after 108
obtained
in
under study was created
entrance even by U.
months
In this
S.
in
1976
standards).
The
environment pioneering
terms of network externalities
Card operators
derive income from customer's interest payments and also from the commisions paid by
vendors, thereby making
it
crucial to develop a large
network of
affiliated businesses.
Figure 9 displays a scatterplot of total credit card market share for three years (1993,
1994, and 1995) versus total elapsed time since the
32
first entry.
As shown,
the very early
first
entry doesn't sustain an above average market share over time.
(108 months
after) captures
In this case
is
it
approximately
45%
of the market and retains
is
having any important
appears that large institutions (as indicated by symbol size) enter
In
relationship
it
entrant
consistently.
not apparent that size of the institution or the plausible combination of this
product with a more ample product platform
entry
The second
effect.
in a
Moreover,
it
second wave of
however, the observed scatterplot seems to support a negative
general,
between order of market entry and market
05
share.
Variables
•
A
ElDSdtotvs mi<t%95
ElDSOtotvs mkt%94
ElpsStotvs mkl%93
0.4
02
1
-0
V-
-
1
25
-50
100
175
250
Elapsed months since pioneer's cntp.
Figure 9
months since the pioneer's entry versus market share
of total customers.
Market share measured as
Elapsed
credit cards
total
time
in
for
%
Figure 9 reports total market shares for the credit card market and Figure 10
market shares measured as a percentage of the pioneer's share
percentage of total customers).
(in
both cases as
Notwithstanding that the pioneer had a period of
approximately seven years operating as a monopolist,
33
it
fails
to reap long
term market
share advantages.
In this case the
second entrant became (and stayed) the market leader
After this period another three competitors entered the market during a three-year period.
Credit card marl^et shares
Figure
Market shares
10;
Notice also
increases
This
is
may
its
in
for credit cards
Figure
1
1,
that the pioneer retains the
number of customers somewhat
reflect different strategic intentions
growing
overall.
of the
from other incumbents but rather from the
any type
and
in
the early entrants' total
third entrants into this
customers they
This
may
intrinsic
grow without
taking market share
growth of the market, perhaps
explain in part the absence of erosion of
1
market do exhibit consistent growth
34
Moreover, the market
different firms.
number of customers. Figure
attract.
(in fact
consistently) despite losing market share.
This means that later entrants can
attacking different market segments.
number of customers
1
shows
in
that the
the total
second
number of
^}
Figure
first
1
number of customers
shown
Total
1
four entrants
for participants in Credit
Card business
Only
Table 9 displays Spearman-rank correlation coefficients for several variables of
interest.
Notice, as observed
elapsed time since
first
in
the scatterplots, the strong negative relationship
entry and market share
in
the credit card business
negative relationship between order of entry and market share
statistically
significant throughout, the coefficient
coefficients
of -0.94, -0.90, and -0 69, p<0.001).
negative relationship between size (measured by
0.69,
p<0
001).
As observed when
later in this case.
Finally,
market share for
diflferent
Though
becomes smaller
Notice also the
There
between
is
also a
this relationship is
as time passes (S-r
statistically significant
number of employees) and entry
(S-r
=
-
exploring the scatterplots, large firms seem to enter
one can observe
a positive relationship
between
all
measures of
years but, again, the sizes of the coefficients diminish.
could signal that the early entrants' market shares are starting to erode.
35
This
Tabic
9:
related to
Spcarman-rank correlation coefficients for \anables
Our
goal
in this part
of the paper
for the dependent performance variable
of the
institution's total
is
to build a sensible and eflficient baseline
SHARE,
exist
controlling for institution size, the effect
product offerings, and the effects of regulation.
construct a model to include question predictors
model
ENTRY
and
We
will
ELAPSED. Should
then
a link
between the question predictors and product performance, controlling for the
aforementioned
accordingly or,
effects, practitioners
at least, qualify their
could then design their product launching initiatives
perceptions about possible performance outcomes of
financial products.
First,
are
shown
Table 10
in
the
raw data
Table
set
and the distributions of
all
variables
were examined
These
10.
Descriptive statistics for selected variables
Normaiit> of
distribution C^)
Variable
SHARE
Mean
s
d
Min
Max.
this to the large
average elapsed time since
entry (125.4 months),
first
we
can infer a
possible negative relationship between elapsed time and share.
The average
size
of the
institutions
Given
institution in that product category.
influence on the
outcome
variable
is
SHARE.
approximately
this,
it
is
27%
the size of the largest
plausible that size will exert
Such presumption
is
some
reinforced with non-
normalities observed for this variable.
In general, the non-normalities
comparison
of
coefficients
for
Pearson
the
observed signal the presence of curvilinearity.
correlation
variables
at
issue
relationships which included the variables
announcing the need for transformations.
Tabic
1
with
coefficients
rendered
large
SIZE and
Spearman-rank
differences
SHARE
(as
for
shown
A
correlation
all
in
bivariate
Table
11),
positive relationship between order of market entry and elapsed time.
This relationship
not a straight hne because of the different product categories present
Second,
we
sample (though
it
is
Third, there appears to be a clear negative and curvilinear relationship
ENTRY
INNOV
appears to indicate a strong negative relationship between the two
at
various markets,
i
e.,
more innovative
in this
regard) tend to enter
This graph, however, does contain a few seemingly atypical data points
the scatterplot of order of entry
real
between
ELAPSED
time.
(ENTRY)
ELAPSED
the graph of
consider
when
our
This could indicate that those firms which have wider product platforms (more
products aimed
earlier,
in
between
Fourth, inspection of the scatterplot of
order of market entry and market share.
variables
not clear whether such entries are prototypical
This clearly indicates the need to control for size
for certain product categories)
versus
the sample.
can see that a group of notably larger institutions appear to have entered
relatively early in the
analyses
in
is
A
and
vs
similar thing
INNOV
vs.
SIZE suggests
SIZE suggests
happens when
that
we
Fifth,
first.
although
that large firms tend to enter
this
look
must be qualified
at
if
we
the relationship extant
This suggests that different results could be obtained
using either variable as predictor, thus pointing to the need for differentiating any
implications
in
terms of order of entry and timeliness of entry.
39
Figure 12: Scatterplots for selected bivariate relationships
.
ENTRY vs ELAPSED
ENTRY
250
vs SIZE
2
1
• •'
SIZE
175
••
•
.••'
09:
•••
1000
•• ©••
.••
••
"it*.
§•
50 CL
00
00.
18
12
6
12
6
ENTRY
ENTRY
SHARE
ENTRY
SHARE
VS
INNOV
OS
ENTRY
VS
INNOV
80
<
06
• •
••A*
••••si
•
••«
02
0_?
fiS«e«**««««A«*»4
6
12
18
0„
25
ELAPSED
SIZE
1
25
12
J
ENTRY
ENTRY
EUPSED
INNOV
VS SIZE
2.
VS
INNOV
80
•
A
*
f
* •
•
40
03
•- •
20
>.•
Hi.
1000
175
250
25
100
ELAPSED
ELAPSED
40
175
250
As suggested by
transformations
on
the
results
variables
the
of the exploratory data
SHARE
transformations and the Rule of the Bulge,
We
perform best for these cases
Appendix
1).
and
we presumed
nevertheless tried
In fact, the natural log
Tukey's
ladder
that natural logarithms
several
possibilities
showed
that both sariables
Figure 13
Normal Probability
Normal
Plot of
will
analysis utilizing the
be designated here as
probability plots (coupled with
SHARE*
had attained normality, as indicated
probability plots for
SHARE*
SHARE*
in
Figure 13.
and SIZE
Normal Probability
15
15
Normal Distnbution
00
Normal Distnbution
41
^
Plot of
-80
00
were
Kolmogorov-Smirnov
00.
15
in
Various analyses were performed to assess the convenience
Normal
of such transformations
of
would
(shown
Bi-variate scatterplots confirmed that the selected transformations
adequate (see Appendix 2)
SHARE-
we performed
of both variables seems successfully to restore
transformed versions of the original variables (which
tests)
Using
Then we proceeded with our data
linearity in the relationship.
and SIZE*)
SIZE.
analysis,
SIZE*
A
comparison of Spearman-rank correlation coefficients and Pearson correlation
coefficients
Appendix
showed
Once comfortable with
3).
had been minimized (see
that the previously observed differences
the variables to analyze,
we developed
bivariate correlations, estimated through Pearson correlation coefficients
summarized
in
Table
12,
revealed that the dependent variable
negatively correlated to both
(Pearson p = -0 69, p<0.001).
ENTRY
Moreover, the variable SIZE*
SHARE*
Pearson correlations
ELAPSED
SHARE* was
strongly
EL.APSED
negatively correlated to
(Pearson p = -0 67, p<0.001) and
(Pearson p = 0.64, p<0.001).
(pair-wise deletion)
is
of
This analysis,
(Pearson p = -0.78, p<0.001) and
(Pearson p = -0.68, p<0.001) and
positively correlated to
Table 12
ENTRY
a table
suggesting that institutions with a larger array of products are not necessarily the ones that
move
first
in
the case of regulation-driven entries.
This interrelation
class
of question predictor.
defined as alterations
SHARE*
in total
predictors.
which
We
established
The
entries are driven
their size
and
REGUL
was
by regulatory changes
in
problems
This
INNOV
The
variables
ENTRY
as control
of the
and
level to
ELAPSED
the analysis.
among
will, evidently,
the highly correlated and redundant variables,
then, to build a
SIZE* and
classified as a control predictor
noted that the relatively large correlations
result in multicollinearity
one
network with an ample
ability to
classified the variables
were established as question predictors
We
classes of control predictors and
market share from entrants which are able to push average
Thus we
variable
two
estimated that "networking effects" could take place,
up or down on account of
platform of products
-)
we
of course not
dummy REGUL
conclusive given the dichotomous characteristics of the
Following these analyses,
is
practically
become
ELAPSED
and
all
variables could
a problem in the case
ENTRY
It
of
was decided,
sequence of nested models paying attention to the plausible occurrence of
collinearity.
Three regression models were subsequently
of creating an appropriate baseline model
Second,
SHARE* was
regressed on
predictors in the regression analysis
The
and, finally,
results are
43
and analyzed with the intention
SHARE* was
First,
ELAPSED,
fitted
regressed on
we used
summarized
in
ENTRY.
both variables as
Table
13.
Table
Regression models
13:
in
which market
market entry and elapsed time since
first
SHARE*, is predicted by order of
ENTRY and ELAPSED, singly and
share,
entry.
jointly.
Model
ENTRY
Intercepts
1
-1.79***
2
-1.82***
3
-183
EL.AJSED
-0.24***
R2 (%)
df Error
61.59
34
-0.015***
47.47
34
-0.0018
6165
33
-025***
***p<().0()l. **p<() Ol. *p<0.05. ~p<().l
For model
residuals
did
not
probability plot
distributed.
(in
1
Table
reveal
evidence
of heteroscedasticity
the
were obtained
residuals
for
of models
models 2 and
1
overestimates share for intermediate values of
and
ELAPSED
Table
as predictors
13, the introduction
seems to correct
Variance Inflation
statistic
accepted cut point of
5,
and
2
entr>'
this
Inspection
of the normal
A
Figure
3, as illustrated in
suggests
that
model
14.
1
slightly
The introduction of both
ENTRY
discrepancy
of both predictors renders
caused by the already noted multicollinearity
of this
of scatterplots of raw and studentized
and histograms of residual distributions reveal them to be normally
Similar results
Comparison of
13), the inspection
ELAPSED
However,
as
in
This
is
= 0.20) and
a
non-significant
tolerance statistic (Tol
shown
(VIF) of 4 89 do indicate values approaching the usually
thus indicating multicollinearity problems (particularly
size).
44
in
a dataset
Figure
Model
14.
Histograms of residuals for three
fitted
models
Model 2
1
Model
HttSaQTtm
Re*.duais of
Two
SHARE*
Residii*u of
additional factors
model
baseline
First,
and
has a
coefficients
tor
maximum
somewhat
tests
more
undergo a
1
full
ancillary,
we have
Entry has a
used
as
units
maximum
in
value of 21 while
very small Beta
SHARE,
of
predictor
have different
that
This results
thus
Second and more important, a difference
and 2 with respect to model
of the R-squared
significant increment in the value
with the reduced model number
the
deciding upon the adoption of a sensible
value of 246 (months).
difficult.
performed for models
in
differently.
ELAPSED when
interpretation a bit
SHARE*
were considered
dimensions and must be interpreted
ELAPSED
3
Hiatot^m
1
(Fobserved = 5.16 >
3
statistic
F^pitical
(Fobserved ^ 12.20
with the reduced model number 2
shows
=
>
in
rendering
R-squared
that both
Comparing
J 29),
models
the
fijll
and comparing
F^ritjcal
-
3.29),
we
notice in both cases, but particularly in the second case, the real and important change in
the
magnitude of the R-squared
statistic.
However, the observed
multicollinearity
problems, and the potential interpretation difficulties urged us not to choose our baseline
model as the most
sensible one, but rather analyze
models
1
and 2 separately.
After reaching a decision about the choice of question predictors, a taxonomy of
multiple regression models
was
fitted.
This taxonomy included ihe "networking effect"
45
dummy
control variables and the
summarizes the
An
results
REGUL,
variable
Table
entered sequentially.
14
of these analyses.
examination of
this table
shows
that the control predictors
SHARE*,
important influence upon the dependent variable
either
do not
singly
exert any
or jointly.
Further inclusion of these variables into the model does not change the intercept and the
coefficient of
(p<0.00]
in
ENTRY. ENTRY
We
most cases)
ENTRY
INNOV
and
REGUL
seems
remains negative throughout with a value close to -0.25
observe that interactions of SIZE with the variables
have no significant
to have an important effect
effect
either.
The
upon the outcome
interaction
variable (Beta coefficient
=0.49, p<0.05) which permits to reject the null hypothesis that the slope
population)
Likewise, the interaction between
ENTRY
an important influence upon the outcome variable.
and
When
the baseline model, both the intercept and the slope of
REGUL
is
zero
in
the
does seem to exert
this variable is
ENTRY
of SIZE and
introduced into
remain stable, but
its
own
beta coefficient acquires a value of -0.24 with p<0 05 (which permits rejecting the null
hypothesis that this slope
is
zero
in
the population)
In both cases differences in
R-
squared tests support the inclusion of these variables into the baseline model (for model
10,
Fobserved
Fcritical=3.29).
=
515 2
upon the dependent
present.
Fcritical=3 29
These interactions seem
regulation, size, and entry
effects
>
do
not,
to
and
make
for
model
sense.
It
12,
is
Fobserved=520.5
>
perfectly plausible that
through their additive terms, completely specify their
variable
SHARE*
because synergies or other effects
may be
These interactions taken together would take the place of a good approximation
of the combined
effect
of regulation, firm size and entry upon the dependent variable.
46
-3
00
t^
r^,
r^
r^i
r-.
I- c:
3
r^
r^^
r-\
r^.
r-i
r^
o
9 o
00
a:
P
Z
o
o
'J
^•t
-3
'/i
-3
O
O
a.
a
E
\0 00
o
>
E
—
r^i
c
o
U
O
o
O
c/:
3
(.-
11
o
>
E
D.
ooooocoocpoo
o
c
o
o
«y^
c
o
oV
C
00 w^ r^
iAi
-rr
'/^
o
V
c
CO
^
»Ai
o
r^ 00 c^
However, when both
statistical significance.
ENTRY_REG
correlated
The Beta of SIZE_REG goes from 0.49
fluctuates
(but
"pointing"
information, a composite
Components Analysis
combined
standardized variables
We
variance
from -0.24 to
was created
SIZE_REG
Based on
order
not
to
lose
valuable
agglomerate the interaction using Principal
to
To
ENTRY_REG
to denote the
create the composite
These two
we
first
of variances
units
total
we
fitted a final
The model
model
in
is
model adding the composite
summarized
in
Table
to the
15.
which the dependent variable
SHARE* is predicted
REGUL.
by
and a composite that captures the interactions of ENTRY, SIZE, and
Final
Intercept
ENTRY
-1.65***
25***
-0
PCI
R-sq (%)
0.44*
67.52
dfE
33_
*p<().05
***p<()()()l
difference in R-squared test for this model again
important change
to
In
composite that ended up containing 98.7% of the
this,
Final fitted regression
Model
A
directions).
then calculated scores for the composite and used the scores to construct
baseline model already chosen
ENTRY
and
of
This denotes that both variables are
of these three variables.
into the single
the variable at issue
Table 15
different
in
0.2.
to -0.15 and the Beta
The composite thus created has been termed PCI
interactive effect
became subsumed
variables are included jointly they change signs and lose
524 9 which
is
in
studentized
that there
the magnitude of the R-sq. statistic (Fobserved
greater than the critical F value of 3.29).
appears to capture
shows
many of the
residuals
for
variables of interest.
this
model
permits
48
An
is,
The model
is
is
in this
a real and
case, equal
parsimonious and
inspection of the scatterplot of
validating
the
assumptions
about
independence of errors (see Figure 15)
was
The assumption about normality of the
validated through inspection of the normal probability plot (Figure 15).
residuals
Furthermore,
no evidence of multicollinearity was found when the model was tested through the
tolerance statistic (Tol=0 95).
Figure
-•(
15.
Plot of residuals and normal probability plot for final
model
Figure
16.
Cook's
Plot of influence statistics. Hat
H vs.
Cook's
D
.
Hat vs Cooks
D
0.5
116
0.4
0.3
0.1
0.0
00
0.2
0.1
0.3
0.4
Hat
The
analysis revealed the presence of
having a considerable impact on the
impact of
this
observation on the
fit
fit
have an unusual impact on the model.
SHARE
sensitivity analysis
showed
of the chosen model to climb from 67
dependent variable
A
one single data point (case 16) which was
52%
that
to
its
performed to investigate the
removal caused the R-squared value
68.46%.
Case 15 was also
identified to
These cases show unusually low values of the
for their relative sizes, thus causing
high values on the created composite PCI, as
shown on Table
50
16.
them
to
show unusually
Table 16
Impact of atypical data points on
fitted
model
Inspection of plots of raw and studentized residuals permitted
histogram of the .esiduals.
validation of the assumptions about independence of errors as
Figure
17.
Normal Probability
Residual vs Predicted
Plot of
.
10|
•
•
.
Residuals of
20
•
-1
,0-
Figure
in
./•
•
y
•20l_
55
-4
00
-2 5
Normal
Predicted
The
final
Table 17
17.
Plots of residuals and normal probability plot for final regression model
Residuals
20
shown
model obtained
is
shown
Fitted final model:
in
Table
17''.
Distribution
1
5
SHARE*
.
parsimonious
coefficient
fitted
model regresses
SHARE*
on
ELAPSED
(intercept
=
-1.82, p<0.001;
of ELAPSED = -0.00154, p<0.001).
Summary and Conclusion
We
wanted
controlling for
enti^'
to assess the effect of order of market
some "networking
initiatives in the
The goal was
effects"
development of new
upon market
to determine whether innovation
semces may render
financial
share,
better performance if
launched during certain time frames.
We found
market entry
that a baseline
(ENTRY)
model
in
which
70%
explained almost
SHARE* was
of the
variability in
size and the amplitude of
control variables such as the firm's
have
a sizable effect
Our
upon our baseline model
most of the prior research work
that has
regressed against order of
its
performance
Including
product offerings did not
analysis turned out to be in line with
been performed
in
other industries, particularly
consumer products.
Our
analysis
shows
timing the release of the>r
for different orders
results
that financial industry practitioners
new products
of market
entry.
The
Figure 18
is
may
benefit
a visual display of our
visual display has
53
model
results
been "umransformed" and
the pioneer's share to
have been computed as a percentage of
interpretation.
from correctly
facilitate
Figure
18.
Results of Final model of order of market entry versus market share for a
prototypical case.
4
6
5
Order of Market Kntr>
Our
findings indicate a negative relationship
long term market share
A
second entrant
pioneer's eventual market share
results
appear to be
such a comparison.
in line
It
entrants are penalized
A
is
between order of market entry and
likely to
tenth entrant will only capture
with those performed
in
heavily.
54
1
]% on
78%
of the
average.
The
other industries. Table 18 establishes
appears that our model slopes
more
capture approximately
downward more
rapidly and later
Table
18:
Order of market entry and market share for consumer packaged goods.
prescription anti-ulcer drugs, and financial services innovations.
Forecasted Market Share Relative to the Pioneering Brand
Consumer Packaged Goods
Entry order
")
Prescription
Financial
Anti-Ulcer Drugs
Products
We can
also observe that the use
of order of market entry alone tends to average
out elapsed time between entries. With our data such average comes out to approximately
15
months between
entries.
This means that the results for the second,
third, fourth,
and
so forth entrants are roughly equivalent to entries occurring after 15, 30, and 45 months
respectively in the second model.
Use of elapsed time
these models.
the pioneer
which
is
We
may
since
first
entry also adds to a managerial interpretation of
see that, for our sample of products, entrants which follow shortly after
obtain substantial market shares relative to the pioneering brand, a result
not obvious by looking at order of market entry alone.
Figure 19
Results of final model of elapsed time since
first entry'
a_prototypical case
21
41
61
Elapsed time
56
81
in
101
months
versus market share for
In
summary,
several important results
First, there are
innovations.
important advantages to early entry
this research.
in financial
services
This result extends established generalizations to other industries and settings
in a field that " relies heavily
either the
emerge from
Urban
et
al.
on North American manufactured goods that are included
Assessor data or the
PIMS
data" (Kalyanaram et
al.,
in
1995: 9).
Second, several outcomes of this research contribute more generally to the
-)
literature
on order of market entry
matters.
If
we
look
at individual
operationalize pioneering as the
retain an appreciable
lines
in
explored
This
On
the one hand
observe that level of analysis
products, establish their longitudinal histories, and
first
entrant,
we
find that the pioneers
market share advantage. This
is
we
is
do not necessarily
two of the three product
true in
consistent with the exposition of Golder and Tellis (1993) although
our case the product histories are more concise because of the youth of the product
lines studied
On
early entry effect
the other hand,
is
observed.
when our product sample
We get
results
which are
is
in line
aggregated, an important
much
with
on pioneering advantages. Evidently by aggregating the data we can only
advantages of eariy entry and not those of pioneering
strictu sensu.
other literature
assert to the
History reveals a
more
detailed picture that cannot be completely captured in cross sectional studies.
Third,
it
appears to be
usefial
and important to reframe the original research
question here posited and to consider elapsed time since
first
entry
when
evaluating these
models. These effects have been, to the best of our knowledge, explored only
57
in
two
The study of Huff and Robinson (1994) determined
studies.
competitive rivalry
(i.e.,
that increasing the years
of
reducing the pioneer's lead time) tended to reduce the pioneer's
market share advantage. Hurwitz and Caves (1988) examined 56 pharmaceutical products
and
performance
their
after patent expiration.
They report
that longer patent life
longer pioneer lead time) renders appreciable market share advantages
of entry
in
terms of elapsed time since
literature, particularly the
work of Sutton
mover advantages may have important
time, a first-mover
first
may spend more on
(i.e.
Evaluating order
entry also ties to the industrial organization
(1995).
effects
Sutton posits that the presence of first-
upon industry
structure.
Given enough lead
advertising, thus preempting or relegating
subsequent entrants to weak positions. Sutton's (1995) empirical examples do indicate
that such
preeminence might be
subsequent entries
Because
tightly linked to the time elapsed
first
mover advantages
affect
between
first
and
market structure, one could
plausibly argue for a negative relationship between elapsed time and market concentration
Such elapsed time might obey
indicates).
exogenous
Moreover,
it
is
to historical occurrences (history matters, as Sutton (1995)
interesting to speculate that in the absence
influences, historical or otherwise, the
of all sorts of
phenomenon of innovation could render
sequential entries with characteristic periods per industry Variance in such "characteristic
frequencies" could point to
more or
less difficulties in imitating products, establishing the
necessary network externalities for products to diffuse rapidly and dominate the market, or
both; thus establishing a firmer
tie
innovations and dominant designs.
with the literature about product and process
Moreover,
58
it
appears that such frequencies should be
related to characteristic
should be evaluated
in
life
cycles of revenues
whereupon entry
at
any point
new product
into the firms current product offerings (using our client-product matrix
similar ones) clearly deserves fiarther evaluation over longer periods
larger samples of detailed data.
it
in fact a
seems compelling
firm
pioneering
is
may
Although
to use such
in
our case
we
not be advantageous
measures of hedging are not
in
left
current strategy.
away from
its
core,
case of banks, particularly where alternative
available, such occurrences could point to
These ideas are
methodology or
did not find any significant
measures to proxy deviations
In the
(or brand)
of time and with
venturing to pioneering a product that appears too far
institution's risk profile
time
terms of its concomitance with the characteristics of such cycles.
Likewise, the idea of using the variation exerted by the
effect,
in
changes
in
the
here as suggestions for further research
59
If
References
New Product Development
Anderson. R. (1987).
m the Service Industry.
Thesis. Massachusetts Institute of Technolog)', Sloan School of
Unpublished Master
Management.
Anderson. R. and C. Harris (1986). "A model of innovation with applications to new financial
products." Oxford
Bcmdt.
E. R.. L. T. Bui.
and
G R
M
and
R
L.
Urban (1994). "The
roles
of marketing, product
S. anti-ulcer
drug
"The humanization of service: Respect
at the
moment of
Lojo (1993) "A framework for anahzing service operations." European
Mancii^emcnt Journal. Vol
Bond.
G
growth and composition of the U.
19-94. Massachusetts Institute of Technolog>'.
Hoech (1990)
J
in the
38, 203-218.
Sloan Management Review. Winter. 89-96.
truth."
Bitran.
H. Reilh and
Working Paper
"
industr>
G
D
and price competition
qualitx.
Bitran.
Economics Papers.
S.
1
I.
and D. F Lean (1977)
No.
3.
271-282.
"Sales promotion and product differentiation in
two
prescription drug markets."' Staff Report to the Federal Trade Commission. Februan..
Brown. C. L and
M.
J.
Lattin (1994). "Investigating the relationship between time in market
pioneering advantage."
Cooper.
.\
and
D
Management
Schcndcl
Horizons. Vol.
19.
No
(
1.
"Strategic responses to technological threats."
1976)
and
Science. Vo\. 40. No. 10, October, 1361-1369
Business
Fcbruar>. 61-69
Cooper. R. G. (1995). "De\'eloping new products on time,
in
time
"
Research-Technology
Manav,emenl. September-October. 49-57.
QWERTY'; The necessity of history
Economic History and the Modern Economist. W. N. Parker, ed. New York: Basil
David. P
In
A
(1986)
"Understanding the economics of
Blackwell
David,
P.
A. (1992). "Heroes, herds and hysteresis
Corporate Change, Vo\.
1,
No.
Federal Reserve Bulletin (1973).
1,
in
technological historv." Industrial
and
129-180.
"Credit-card and check-credit plans at commercial banks."
September.
Finnertv.
J.
(1988)
Management.
"Financial engineenng in corporate finance;
17. 14-33.
60
An
over\'iew." Financial
"
in international
T (1983) "Market share, technology leadership, and competition
Flahcm"
Innovation. Management, and
In Research on Technological
semiconductor markets
JAI Press, 69-102.
Pohcy. R. Rosenbloom, cd. Greenwich, CT:
M
GeoffVon.
Economies
et singularites
"Analogies
(1992).
P.
et Societes.
No
des processus d'lnnovations financiere
et reelle."
Sept-Oct. 73-96.
9,
significance of the credit card
(1967). "A studv of the strategic and operational
Gibson D
Reserve Bank of Boston. Research Report, Boston:
for commercial banks." Federal
Administration.
Harvard Univcrsit>'. Graduate School of Busmess
M
advantage: Marketing logic or marketing legend?"
Golder. P and G. Tellis (1993). "Pioneer
158-170.
Journal of Marketing Research. Vol. 30. May.
"f
andW T
Huff L C
Robinson (1994)
"Note:
The impact of leadtime and years of
advantages."
competitive nvalr>' on pioneer market share
10. October.
Management
Science. Vol. 40. No.
1370-1377
Promotion and the
\ and R E. Caves (1988). -Persuasion or information''
pharmaceuticals " Journal of Law and Economics.
shares of brand name and generic
Hur^vltz
M
31.
October. 299-320.
the order of entry on market share.
and G. Urban (1992). "Dynamic effects of
purchases for frequently purchased consumer goods."
trial penetration, and repeat
235-250
Marketing .Science. Vol.1 1. No. 3. Summer.
Kalvanaram
G
Kalvanaram
G.
W
Robinson, and
G
Urban (1995)
"Order of market entry
Estabhshcd
:
empirical generalizations, and ftiture research.
empirical generalizations, emerging
Working
Technology
Paper. Massachusetts Institute of
Klevonck
A K R
C. Levin.
significance of intenndustrv'
the sources and
R. Nelson and S. G. Winter (1995). "On
Research Policy.
differences in technological opportumties."
R
24. 185-205.
Kmmme.
D.
(1987)
Amencan Bankers
Banking and the plastic card. Washington. D.C.:
Association.
performance
Lambkin. M. (1988). "Order of entry and
Management Journal, 9 (Summer). 127-140.
in
new markets."
Strategic
"Appropnating the
Klevonck, R. R. Nelson, and S. G. Winter (1987).
Activity,
development." Brookings Papers on Economic
returns from industrial research and
Levin
3.
R C A K
783-820.
61
G. and E.Yoon (1990). "The timing of competitive market entn:
Lilicn.
of new
industrial products."
Management Science.
J.
F and V.
K
Financial Innovation.
Meyer, M. H. and
J.
IEEE
success."
explorators study
Vol. 36. (May). No. 5. 568-585.
Mandcl. L. (1990). The credit card industry: A history
Marshall.
An
Boston: T\va\Tie Publishers.
Bansal (1992). Financial Engineering, a Complete Guide to
New York: New York
M. Utterback
(1995).
Institute
of Finance.
"Product development cycle time and commercial
Transactions on Engineering Management, Vol 42. No.
4.
November,
297-304.
W.. B. Gartner and R. Wilson (1989). "Entn,' order, market share, and competitive
advantage: A study of their relationships in new corporate ventures." Journal of Business
Miller. A..
Venturing. 4 (M2i\). 197-209.
M
Miller.
and Quantitative
Financial
Moronc.
"Financial innovation:
(1986).
and
J
D
last
twent\ \ears and the
."
ne,\t."
Journal of
Analysis. Vol. 21. No. 4. 459-471.
Berg (1993). "Management of technology
the banking industrv
1.
The
in the service sector:
Practices in
The Journal of High Technology' Management Research. Vol.
4,
No.
123-137.
Pctruzzi.
C
.
M. Del Valle and
S.
Judlowc (1988). "Patent and copyright protection for
innovations in finance." Financial Management. (Winter), 66-71.
Quinn.
B. (1992).
J
Robinson.
industries
Robinson.
W.
New York The
Free Press.
"Sources of market pioneer advantages: The case of industrial goods
Journal of Marketing Research. Vol. 25, Februar>-, 87-94
W. T
"
Intelligent Enterprise
(1988)
T. and C. Fomell (1985)
"Sources of market pioneer advantages
in
consumer
goods industries." Journal of Marketing Research. 22 (August), 305-317.
new
markets:
A
G
Urban (1994). "First-mover advantages from pioneering
survey of empirical evidence." Review of Industrial Organization. 9, 1-23.
Robinson. W.. G. Kalyanaram, and
Robinson. W.. C. Fomell and M. Sullivan (1992) "Are market pioneers intrinsically stronger
than later entrants':*" Strategic Management Journal. Vol.13, 609-624.
Sunon,
Tufano,
J
(1995)
P. (1989).
Sunk costs and
industrv' structure.
Cambridge:
MIT
Press.
"Financial innovation and first-mo\er advantages." Journal
Economics. 25 ,213-240.
62
of Financial
Urban. G.. T. Carter.
brands;
No.
6.
pioneering
Gaskin and Z. Mucha (1986). "Market share rewards to
S.
An empincal
analysis and strategic implications."
Management
Science, Vol. 32,
June. 645-659
Utterback.
J.
M. (1994). Mastering
the
dynamics of innovation, Boston: Harvard Business
School Press.
Von
Hippel, E, (1988).
Whitten,
cntn,'.
I.
T. (1979).
1913-74
"
The sources of innovation.
"Brand performance
New York;
Oxford Umversit>' Press.
in the cigarette industn'
and the advantage
Staff Report to the Federal Trade Commission. June,
63
to early
Appendix
1
Scatter-plot matrix report for various transformations of
1
SHARE and
SIZE
Appendix 2
Selected scatterplots after transformations.
ENTRY
o
.
o
< ^
ENTRY
SHARE-
o
^o'^^„'^
a.
vs
.
ai
1^°o
vs SIZE*
Co
1
qo
-
o
O
o
O
r
qOO
*-
"•f
125
125
J
ELAPSED
ELj^PSEDvs SHARE*
o c
C.Co
VS SIZE*
<b
o
9 -o
o
< »
25
188
ENTRY
ENTRY
o
'y
Oo
a.
6
o
-I
'=t.
(%
'rc'
<o
—
-1
•50
25
100
ELAPSED
INNOVvs SHARE*
'tC'
o
100
ELAPSED
INNOV
vs SIZE*
175
250
Appendix
3.
Table Ap3-1: Difyerence between Pearson and Spearman correlations
( pair-wise
deletion)
Variables
ENTRY
ENTRY
0.000
ELAPSED INNOV
REGUL
SHARE*
Appendix 4
Measure of Strategic Focus
To
obtain a measure of strategic focus,
product matnx. Though normally used to
we used
test
we used
the chi-square distribution of each bank's client
hypothesized relationships
among
categories of data points,
chi-square in this case as a measure to establish degree of strategic focusing.
institutions' clients
and products and de\eloped "client-product"
matrices.
We
first classified
Therupon, two measures were
developed.
A)
^i
In the first
measure, a comparison was made between the chi-square obtained for particular client-product
matrices with respect
one with
a large
to a client-product inalri.x
number of products
each
in
cell
of a totally unfocused institution (for practical purposes
of the client-product matri.x). The procedure used was as
follows:
Let
Ncp = number observed
ccp= number expected
large
in (c.p)th cell
in the (c.p)th cell
and equal number of products
forc=
1.2..V
andp=
of the client product matn.\. and
in
under the assumption of
each
cell.
1.2.3.
Let
Np = '^Ncp
Nc =
Then we
Y,N.-j,
calculated a chi square as follows:
67
.and
a totally
unfocused institution with a
7
J
58
Date Due
NuV 1 3 1998
MAY
0%
t
Lib-26-67
|[ii|!iiiiiiimiMi™S
3 9080 01444 5057