W&SWiiJ
HD28
.MA 14
ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
DIFFUSION OF INFORMATION TECHNOLOGY
OUTSOURCING: INFLUENCE SOURCES AND THE KODAK
EFFECT
Lawrence Loh
and
N. Venkatraman
Working
'King Paper No. 3476-92BPS
c
October 1992
345
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
DIFFUSION OF INFORMATION TECHNOLOGY
OUTSOURCING: INFLUENCE SOURCES AND THE KODAK
EFFECT
Lawrence Loh
and
N. Venkatraman
Working Paper No. 3476-92BPS
October 1992
NI.I.T.
NOV
LIBRAh
I 1992
Diffvsion of Information Technology Outsourcing:
Influence Sources and the Kodak Effect
Lawrence Loh
Sloan School of Management
Massachusetts Institute of Technology
50 Memorial Drive E52-509
Cambridge
02139
(617) 253-3857
MA
N. Venkatraman
Sloan School of Management
Massachusetts Institute of Technology
50 Memorial Drive E52-537
Cambridge
02139
MA
(617) 253-5044
Version: September
15,
1992
Forthcoming
in Information Systems Research
Submitted: January 9, 1991; Accepted: October 3, 1992
Acknowledgements. This research was supported by the consortium on Managing
Information Technology in the Next Era (MITNE) of the Center for Information Systems Research at
the Massachusetts Institute of Technology and IBM's Advanced Business Institute.
thank two
associate editors and three journal reviewers for comments on earlier versions of the manuscript.
We
Diffusion of Information Technology Outsourcing
Diffusion of Information Technology Outsourcing:
Influence Sources and the Kodak Effect
Abstract
The governance
infrastructure
is
of an organizational information technology (IT)
steadily shifting
away from pure
hierarchical
mechanisms toward hybrid and partnership modes
In particular, IT outsourcing has recently
emerged
and market
that involve external vendors.
as a significant administrative
innovation in an organization's IT strategy. This paper seeks to explore the sources
of influence in the adoption of this innovation. For this purpose,
comprehensive sample of outsourcing contracts
in the U.S.
we
generated a
using an electronic
bibliometric search process. Using diffusion modeling, our empirical analysis
that the adoption of IT outsourcing
imitative behavior) than
Subsequently,
we
is
motivated more by internal influence (or
by external influence amongst the user organizations.
considered the widely-publicized Eastman Kodak's outsourcing
decision as a critical event to assess whether this internal influence
pronounced
show
shows
in the post-Koda^ r
that internal influence
is
gime than
dominant
in the
in the
is
more
pre-Kodak regime. Our results
post-Kodak regime but not in the
pre- Kodak regime. Implications and directions for future research are discussed.
Keywords: Diffusion models; information technology governance; information technology
outsourcing; information technology strategy.
Diffusion of Information Technology Outsourcing
Introduction
In recent years, the
has shifted beyond
its
governance of information technology
traditional hierarchical locus
towards
a
(IT) infrastructure
hybrid
mode
involving external organizations. The newer approaches include joint ownership of
IT networks
(e.g., in
an industry-wide electronic data interchange), co-development
of applications software
(e.g., in a
multi-company consortium) as well
as several
other types of partnerships between users and vendors. Within such a spectrum,
focus on one governance
outsourcing of
all
mode
that has
become prominent
we
in recent years:
or significant components of the user's IT infrastructure. In simple
terms, this involves the transfer of property or decision rights in varying degrees
over the IT infrastructure by a user organization to an external organization such as
a
technology vendor or a systems integrator.
The trend toward IT outsourcing
Dearden
(1987),
who argued
is
proffered in a prognostic
that the IT organization
is
"withering away" as end-
users gain greater control of their computing environments and
more importantly,
as external software specialists increasingly take charge of corporate
development. Indeed,
this shift in IT
governance
multi-billion dollar outsourcing market.
Group estimated
billion,
A
to reach
all
in IT
sectors.
vendor. 3
is
On
Inc.
a similar vein,
two other
separately forecasted that the
rapidly evolving.
market encompasses organizations
Many
types of outsourcing totaled $33
outsourcing would be about 17% for 5 years after 1990. 2
The IT outsourcing business
side, the
currently characterized by a
$49 billion in 1994. ! In
consulting organizations: G2 Research and Input,
annual growth rate
systems
leading IT consulting firm, The Yankee
that outsourcing revenues for
and are expected
is
commentary by
On
the
in the for-profit
demand
and
(i.e.,
user)
not-for-profit
users in these different sectors are simultaneously served by the
the supply
(i.e.,
vendor) side, the industry
is
same
dominated by the top
five
vendors: Anderson Consulting, Computer Sciences Corp. (CSC), Digital Equipment Corp.
(DEC), Electronic Data Systems Corp. (EDS), and International Business Machine Corp.
Diffusion of Information Technology Outsourcing
(IBM). All these IT suppliers provide a
full
range of outsourcing services across
multiple industries and sectors. 4 '5
With the increasing significance
this
innovative administrative practice
of IT outsourcing as a viable strategic option,
is
being viewed with considerable attention
within the professional community. Against this backdrop and the high potential
profits in the marketplace, 6 there has
manage
been an aggressive push by the vendors
to
the IT infrastructure of the users. For instance, Martin Marietta Information
Systems Group ran an advertisement with the caption: "You don't
plant for your electricity
...
Why own
a data center for
own
a
power
your information systems?" 7
and Information Week — two leading industry publications
Further, Computerworld
have devoted extensive coverage of the outsourcing phenomenon. As a Merrill
Lynch outsourcing analyst put
rolling downhill. This
current year
among
it,
this
industry
is
"a huge, burgeoning snowball
not a market that's burning out
is
the leading outsourcers will be
intensified scramble for differentiation, jockeying in
values." 8
a curious
In this light,
anomaly
The
centers,
hopes
will build
it
IBM
agreement,
and 300 Kodak workers
to cut costs as
impact of
..hr
fierce competition,
an
ranks and staggering deal
has been contended that "outsourcing suddenly went from
interest in IT outsourcing escalated
to this
marked by
exploding... the
to a veritable craze." 9
agreement under which
According
it
-- it's
much
as
this large contract
when IBM announced "an unusual
and operate
a data center for
work done by four Kodak
"will take over the
will
become IBM employees."
50% by turning
Eastman Kodak." 10
In addition,
the operation over to IBM."
"Kodak
...
The
involving two prominent companies had been
profound: IT outsourcing suddenly became a serious strategic choice for firms, and
the "consideration of outsourcing"
emerged
as
one of the top ten issues
for success
(or survival) in the 1990s. 11
Within the research community, IT governance
as
an important component of IT strategy
is
in conceptual
beginning to be recognized
frameworks (Henderson
Diffusion of Information Technology Outsourcing
and Venkatraman, 1992) and analytical models (Richmond, Seidmann and
Whinston, 1992; Whang, 1992). Empirical studies on IT governance adopting
sectional
b).
approach are beginning
a cross-
appear recently (Loh and Venkatraman, 1992a,
to
This paper adds to the empirical stream by adopting a longitudinal approach to
explore the sources of influence underlying the diffusion process of IT outsourcing
as an administrative innovation.
used by Mahajan, Sharma, and
We
apply the analytical models of diffusion as
Bettis (1988) in the case of
M-form
organizational structure but with a significant methodological
(multidivisional)
improvement
through our more powerful nonlinear estimation and specification
Specifically,
we:
April 1988 to
(a)
sample of 60 outsourcing decisions
identify a
August 1990 using an
tests.
in the U.S.
from
electronic bibliometric search; (b) adopt
nonlinear estimation methods and specification tests to examine the sources of
influence
—
internal, external,
of outsourcing;
and
differential
that could explain the diffusion pattern
the prominent Kodak outsourcing decision as a 'critical-
(c) treat
event' in delineating
and mixed —
two regimes — "pre-Kodak" and "post-Kodak
'
to assess the
impacts of the influence sources within each of the regimes.
TJieoretical Perspectives
IT Outsourcing as an Administrative Innovation
In general terms, outsourcing falls within a class of
decisions in an organization.
manufacturing of parts
It
in the
"make-versus-buy"
has been studied in several settings such
automobile industry
(e.g.,
Walker and Weber,
the sales function in the electronic industry (Anderson, 1985);
of equipment, components,
and Weitz,
We
and supplies across
a
broad
as:
the
1984);
and the distribution
set of industrial firms (John
1988).
define IT outsourcing as the significant contribution by external vendors of
the physical
and/or human resources associated with the
components of the IT infrastructure
entire or specific
in the user organization. 12
This definition
consistent with recent discussions of IT outsourcing by researchers
and
is
practitioners.
Diffusion of Information Technology Outsourcing
For instance, Elam (1988) suggested the possible use of three forms of cooperative
arrangement
(i.e.,
full-equity ownership, partial ownership,
and no ownership)
involving external vendors that can be used for the joint development of corporate
information systems. Further, a leading FT periodical described outsourcing as the
"turning over, or sharing, responsibility for
all
or part of an organization's
information technology function with a third party." 13
We
how
argue that FT outsourcing represents
firms govern or
manage
a
fundamental point of departure
their FT infrastructure,
an administrative innovation (Teece, 1980; Mahajan
and hence could be viewed
et. al.,
and Koh (1992) defined an administrative innovation
1988).
in
as
Venkatraman, Loh,
as "involving significant
changes in the routines (or behavioral repertoires) used by the organization to deal
with
(a)
its
tasks of internal arrangements
and
external alignments." This definition reflects
the critical notion of first-time adoption
Zaltman, Duncan, and Holbeck, 1984);
(b)
by an organization (Rogers,
the changes in the routines
1983;
and
procedures of organization and management involving substantial set-up costs and
organizational disruptions (Nelson and Winter, 1982); and
(c)
a comprehensive
view of administrative tasks as organization-environment coalignment
(Thompson, 1967; Snow and Miles,
Table
1
1983).
summarizes our arguments supporting the treatment of FT
outsourcing as an administrative innovation.
the
mode
of governance
--
First,
it
represents a significant
shift in
from the traditional locus of control and coordination
within the hierarchy (combined with relatively standardized market transactions
with vendors) towards newer modes that could be characterized as hybrids
(Williamson, 1991) or partnerships (Kanter, 1989; Powell, 1990). As
Markus
(1984: pp.
125-127) articulated: 'In the early days of computing, in-house development
was
the
only alternative available to those organizations that wished to explore the
commercial potential of computers.... most companies with sizeable internal
development
efforts usually
run most of
their applications
on
internally
owned and
Diffusion of Information Technology Outsourcing
Table
1
IT Outsourcing as an Administrative Innovation
Key Component
of the Definition of
Administrative Innovation
Diffusion of Information Technology Outsourcing
operated computers."
A
good example
of a recent outsourcing contract involving a
fundamental reorganization of the IT function
outsourced a major part of
its
is
General Dynamics Corp. that has
IT infrastructure (with 2600 employees and a data
center of over 1000 MIPS) to the Computer Sciences Corp.
mode
of governance
is
its
In fact, such a shift in
associated with profound transformations in the strategic
and operational mechanisms
within
-14
that are necessary for
current mission or scope.
It is
an organization
....
view
consistent with Evan's (1966: p. 51)
administrative innovation as "the implementation of an idea for a
pertaining to
to position itself
new
of
policy
the allocation of resources, structuring of tasks, authority, of
rewards."
Second, IT outsourcing represents significant changes
the user organizations.
Kanter (1989:
p. 141)
in the internal processes of
argued that multi-organizational
partnerships alter "internal roles, relationships, and power dynamics for the
organizations entering into them." This perspective applies well to IT outsourcing
in
many
left
cases.
behind
to
As
a leading IT periodical observed, "Information
mind
may suddenly
the shop after parts of their operations have been outsourced
out of water: They're forced to learn almost an entirely
feel like fish
different set of job
systems executives
skills....
A
large bulk of the responsibility IS
overseeing outsourcing contracts
is
managers take on
in
monitoring the outsourcing vendor's work." 15
This clearly calls for a principal adjustment in the nature and degree of internal
routines. For example, a large oil
with just five managers within
have the responsibility
for
and mining company
its
for
changes
mechanisms
--
ensuring that their department develops an internal IT
by virtue of
in internal processes
for joint
and operations
left
IT organization. 16 These managers, however,
role that supports the firm's business strategy.
that IT outsourcing
was
that has outsourced
its
Beyond
this specific case,
radical shift in the locus of
we
argue
governance
involving redefinitions of liaison roles,
--
calls
new
decision-making in areas such as development, maintenance
of the IT infrastructure, allocation of critical resources given possible
Diffusion of Information Technology Outsourcing
divergence
in goals,
and redistribution of authority as well
as criteria for
performance appraisal.
Third, IT outsourcing constitutes a significant change
in the organizational
away from an
routines used to deal with the external environment. Specifically, a shift
arms-length approach for dealing with the IT marketplace (with multiple competing
bids and price-dominated short-term contracts) towards a "quasi-firm"
mode
(with
longer-term alliances involving a few selected partners) requires increased mutual
understanding, enhanced goal compatibility and recognition of complementary
skills
and requirements. For instance, National Car Rental System,
the operational
National and
management
EDS planned
its
IT infrastructure to EDS. 17
to jointly
initially installed at National
Inc.
As part
has outsourced
of this alliance,
develop applications and systems that would be
but could subsequently be marketed to other car rental
companies. This implies that the routines for dealing with the external partner are
different
from the
traditional user-vendor relationship as
development and changes
(beyond merely renting
outsourcing
may
involves joint-
business scope of National Car Rental System
in the
cars).
it
As Kirkpatrick
new form
point toward a
(1991: p. 112) noted, "Ultimately,
of corporate interdependence with
implications that extend beyond the computer room."
A
Diffusion-of-Innovation Perspective
The
diffusion of an innovation
communicated through
system" (Rogers, 1983:
the process by
certain channels over time
p. 5).
this
(any idea, object, or practice that
of individuals
which the innovation
among
the
members
"is
of a social
Extant research in this area (see for instance, Mahajan
and Peterson, 1985) describes
community
is
process using four
is
critical
perceived to be new); the
and/or organizations
innovation), the channels of communication (the
elements: the innovation
social
system (the
that are potential adopters of the
means by which information
is
transmitted to or within the social system); and time (the rate or speed by which
members
of the social system adopt the innovation). This perspective has been
Diffusion of Information Technology Outsourcing
adopted
to
10
understand the pattern of diffusion of
a
wide range of innovations
technological and administrative innovations (see Mahajan, Muller, and Bass, 1990
for a recent
review pertaining
to the case of
marketing). In the context of IT
innovations, the diffusion approach has been used in studies such
for
modern software
practices,
Brancheau and Wetherbe (1990)
for
as:
Zmud
(1983)
spreadsheet
software, Gurbaxani (1990) for BITNET, and Nilakanta and Scamell (1990) for
database design tools and techniques
This paper
is
where the relevant
concerned with IT outsourcing as an administrative innovation
social
system
is
the set of organizations leveraging IT
infrastructures to achieve their mission. Specifically, our referent
encompasses the computer-using organizations
that
might consider the adoption of
IT outsourcing. Channels of communication in IT outsourcing take
horizontal channel (namely:
another
--
for
how
the
members
community
two forms:
of a social system interact with
example, through direct interpersonal contacts or indirect
observations within the IT user community) and vertical channel (namely:
members
one
of a social system interact with the outside agents
--
for
how
the
example, through
promotional efforts by the IT vendors, results of research studies or summaries of
state-of-the-art practices
from other countries).
Rogers and Agarwala-Rogers (1976:
p. 7) related
organizational decision making: "an organization
is
communication
an elaborate
to
set of
interconnected communication channels designed to import, sort, and analyze
information from the environment and export processed messages back to the
environment. Communication provides a means for making and executing
decisions, obtaining feedback,
and correcting organizational objectives and
procedures as the situation demands" Communication can also be facilitated within
the context of a professional network comprising corporate executives across
different organizations (Blair, Roberts,
communication
is
and McKechnie,
1985; Mueller, 1986).
Thus
an integral aspect of the diffusion model. The inter-temporal
Diffusion of Information Technology Outsourcing
11
pattern of outsourcing contract announcements represents the time element of the
diffusion of IT outsourcing.
Influence Sources in the Diffusion of IT Outsourcing
two
In diffusion models, there are
external
—
—
an innovation by members of
that drive the adoption of
In addition, these
basic types of influence
two can be aggregated
to constitute
what
is
internal
and
a social system.
termed as mixed
influence.
Internal influence. This perspective purports that diffusion takes place solely
through channels of communication within a specific community or social system.
The diffusion
rate can
be represented
as:
^^- = qN(t)[m-N(t)]
dt
where N(t)
is
(1)
the cumulative
number
of adoptions at time period
and
coefficient of internal influence (q>0),
the social system.
A
plot of N(t) with
The internal-influence model
model (Mahajan and Peterson,
increase in the
the
number
t
is
of adoption
is
the potential
results in a standard
q
number
is
the
of adopters in
S-shaped curve.
structurally equivalent to the imitation
1985: p. 17).
As evident
in the equation, the rate of
depends on the extent
community have already adopted
for diffusion is
m
t,
in
which the members of
the innovation. In other words, the impetus
an emulation of prior adopters by potential adopters. In our study,
imitation in the adoption of IT outsourcing
have hitherto maintained
may
take place
their IT function inhouse,
make
when
organizations, that
their sourcing choice
based on other organizations that have already outsourced.
The
internal-influence
model has received empirical support from
well-cited
studies such as Griliches (1957) and Mansfield (1961). In the former study, Griliches
analyzed the diffusion of hybrid seed corn amongst farmers in crop-reporting areas
within the U.S. In the
latter study,
Mansfield examined the diffusion of several
industrial innovations such as pallet loaders, diesel locomotives,
and continuous
Diffusion of Information Technology Outsourcing
mining machines among
more
firms. In a
12
recent work,
investigated the comparative diffusion pattern of
mechanisms
(specifically, joint ventures
that the internal-influence
The authors argued
the innovation
model
that this
which
is
is
and M-form
structures)
verified
conceptually due to the relative ease of imitability of
and
communication are instrumental
factors:
in the diffusion of
(i.e.,
unique
Within
social complexity.
Brancheau and Wetherbe (1990) proposed that internal
amongst
was
-- it
in the case of joint ventures.
brought about by the absence of three
is
et al. (1992)
two corporate governance
more appropriate
historical conditions, causal ambiguity,
of
Venkatraman
IS research,
interpersonal) channels
spreadsheet software
later adopters.
External influence. This perspective specifies that diffusion
is
driven only by
information from a communication source external to the social system. Thus, the
rate of diffusion at time
t is
dependent only on the potential number of adopters
present in the social system at time
t
without attributing any diffusion
between ^rior adopters and potential adopters. Formally,
this
to interaction
model can be written
as:
_=
dN(t)
m/*m
r
p[ m-N(t)]
(2)
where p
is
the coefficient of external influence (p>0). Plotting N(t) with
curve that increases
at a
decreasing
rate. In the specific
t
results in a
context of our study, external
influence on the potential adopters of IT outsourcing can emanate from efforts by
vendors
(e.g.,
EDS and
or trade periodicals
IBM), consulting firms
(e.g.,
(e.g.,
The Yankee Group and Input,
Inc.)
Computerworld and Information Week).
The external-influence model has been empirically supported by Coleman,
Katz, and Menzel (1966)
where the diffusion
of a
new drug amongst
being studied in four midwestern communities. The model
when members
innovation
is
of a social system are isolated or
when
is
physicians
is
especially appropriate
information on the
not obtained through interpersonal (or interorganizational)
communication but has
its
origins from outside the system (Mahajan
and Peterson,
Diffusion of Information Technology Outsourcing
1985).
Venkatraman
et al. (1992)
inherent in the diffus on of
;
13
established that external-influence
M-form
structures, arguably
due
external information channels can facilitate the diffusion of
practices (or
new system development
methodologies) and
is
to the imperfect
Zmud
imitability of the innovation. Within the IS literature,
model
(1983) verified that
modern software
this is
contingent on
organizational factors such as size, professionalism, task complexity, and context.
Brancheau and Wetherbe (1990) argued that early adopters are more
external channels of communication
(i.e.,
mass media)
in the
likely to rely
on
adoption of
spreadsheet software. Nilakanta and Scamell (1990) found limited support for the
influence of external information sources
diffusion of database design tools
Nidumolu, Ramiller, and Ward,
and communication channels
and techniques
(see also
1991). In a recent review,
Swanson,
in the
Fuller,
Gurbaxani, King,
Kraemer, McFarlan, Raman, and Yap (1990) highlighted that several external
institutions (e.g.,
governments, international agencies, and educational institutions)
are critical in shaping the international diffusion of IT.
Mixed
influence. This perspective
influence models
^p
internal-
and external-
and can be represented by the following equation:
= [p+qN(t)][m-N(t)]
The cumulative
distribution of the mixed-influence
generalized logistic curve
the
subsumes both the
model gives
whose S-shape depends on
the coefficients p
most general form and widely used because the assumptions
external- or the internal-influence
rise to a
and
q. It is
of either the
models are seldom met unequivocally. The
mixed-influence model has been successfully applied in the work of Bass (1969)
where the
sales of several electric appliances are being forecasted. In the IT arena,
Gurbaxani (1990) discovered
more
that the diffusion of
specifically, a logistic curve.
of coverage
The impetus
BITNET followed
for
its
diffusion
by the media and the early adoption by highly
universities.
was
a S-curve, or
a combination
visible research
Diffusion of Information Technology Outsourcing
Based on the above discussion,
we
14
seek to assess the influence type that
captures the pattern of diffusion of IT outsourcing.
Research Question
1: Wliat
Our
first
research question
is:
source of influence best characterizes the diffusion of IT
outsourcing?
Kodak's Outsourcing as a Critical Event
In July 1989,
its
Kodak announced a unique contract under which
data center operations to
three
IBM.™
it
outsourced
This involved a transfer of four data centers and
hundred IT personnel within the IT function
of
Kodak
to
IBM. 19
We
will
argue
that this specific case constitutes a critical event in the overall pattern of diffusion of
outsourcing. In particular,
differ before
The
and
we
seek to test whether the types of underlying influence
after this critical event.
possibility that
one single adoption can drive the diffusion of
innovations has been documented in the
management
literature.
Rogers and
Kincaid (1981) argued cogently that adoption by an opinion leader can have a
profound
effect
on subsequent adoptions. This
is
because opinion leaders have
connections to multiple and diverse sources of relevant information; in addition,
they usually occupy boundary-spanning and central positions within the network of
potential adopters. In general, such personnel serves to reduce the incidence of
uncertainty inherent in the process of adoption by providing information access,
sources of reliability, and symbols of legitimacy to potential adopters. In the IT
context,
MIS opinion
training of users
leaders can facilitate the promotion of
(McLeod and
Fuerst, 1982) as well as foster
new
applications and
MIS
infusion (Kwon,
1990).
The most appropriate example
case of
BITNET
(Gurbaxani, 1990).
the City University of
New
of a critical adoption driving diffusion
BITNET was
is
the
established through a link between
York and Yale University. Subscription to the network was
based on the operating philosophy that a
new
adopter has to provide a line to the
nearest network node. In the formative stages,
BITNET was confined
to institutions
^
Diffusion of Information Technology Outsourcing
in the U.S. east coast.
As Gurbaxani
BITXET was
evolution of
"A
(1990: p. 74) noted,
15
significant event in the
the adoption by the University of California at Berkeley.
who
This greatly lowered the access costs to other western universities
subsequently
joined the network."
We
build on Gurbaxani (1990) by considering a critical event and develop two
regimes to see the differential types of influence on the diffusion pattern.
Specifically,
we
consider Kodak's decision as a
critical
event for the following
reasons: One, this represented a higher level of visibility for the administrative
due
practice than prior user-vendor relationships
to the
companies and the magnitude of the contract ($500
have provided
a
major impetus for managers
evaluate outsourcing as a governance option.
prominence of the two
Two,
million).
this
to
in other organizations to seriously
As Kirkpatrick
(1991: p. 103) noted,
No company
'The move startled computer managers across corporate America.
Kodak's size and prominence had ever turned over
Indeed, this decision sparked a greater level
seems
.>f
its
computers
interest in this
an outsider..."
to
mode
of
of IT
governance: "For the majority of organizations, however, the idea became thinkable
only
when
large
and
visible corporations
such as Eastman Kodak
Co....
began
to
publicize their plans to divest and openly discuss the cost benefits of outsourcing." 20
A
leading outsourcing consultant attributed the current interest in outsourcing to
the Kodak decision:
that
Kodak
is
doing
"The Kodak contract was
this
sends a message that
other single decision seems to have as
really the
it is
OK
watershed event. The
to
go
much impact during
fact
to outsourcing." 21
No
the time period that
we
considered in this paper.
We
critical
use the terms "Kodak effect" to signify the importance of the Kodak
event in driving the diffusion pattern of IT outsourcing. Specifically,
examine whether the post-Kodak regime exhibits
than the pre-Kodak regime. For this purpose,
1988 to July 1989-- as the
first
we
diffusion regime,
we
a different pattern of influence
consider the time period
and the time period
--
--
April
August 1989
to
Diffusion of Information Technology Outsourcing
August 1990—
as the
16
second diffusion regime. Our second research question
is:
2: What source of influence best characterizes the diffusion of IT
outsourcing before and after the Kodak-IBM contract?
Research Question
Methods
Sample
We
generated a sample of outsourcing contracts by an electronic search of two
CD-ROM databases - Newspaper Abstracts Ondisc and
produced by University Microfilms
Business Dateline Ondisc
International (UMI).
The
first
seven major daily newspapers, while the second provides the
--
contains indexes of
full text of articles of
nearly 180 regional magazines, daily newspapers, and wire services.
We
performed
comprehensive search of these information sources starting from January 1985
a
(as
constrained by the databases) and identified a total of 60 outsourcing contracts
spanning the time period April 1988
to
management outsourcing and systems
1990. 22
August
We
included both
integration outsourcing,
facilities
which are two
major modes of IT outsourcing practiced currently and are consistent with our
Appendix
A
provides a detailed breakdown of the contracts. The
average size of the contracts
is
$191.1 million (with a standard deviation of $589.9
earlier definition.
million), while the average duration of the contracts
deviation of 5.0 years). Figure
1
shows
a time-plot of
is
5.6 years (with a
'ihe
standard
outsourcing adoption.
Analytical Framework
Mahajan
et
al.
(1988) developed
the imitation hypothesis in the case of
and used
M-form
a linear
analogue approach to
structure. This
subject to econometric limitations such as multicollinearity
standard errors for the crucial parameters
we adopt
inferential
(p, q,
approach
is,
test
however,
and non-availability
and m) of equations
(1) to (3).
nonlinear least squares (NLS) estimation of the three models. The
procedures are based on a special class of model specification
tests
developed by Davidson and MacKinnon (1981) for the case of a non-nested
of
Hence,
I
I
I
Diffusion of Information Technology Outsourcing
17
Figure 1
Diffusion Pattern of IT Outsourcing
A
60
t
50
-
40
•-
30
-
20
-
10
-
d
o
P
t
i
o
n
Number
i
>—r;~I^T-> -^ <*":-N --
<
1988
1988
1988
1989
1989
1989
1989
Apr
Jul
Oct
Jan
Apr
Jul
Oct
1
—
1
1990
Jan
—
1
I
1—
I
—I—
1990
1990
Apr
Jul
Diffusion of Information Technology Outsourcing
18
nonlinear alternative hypothesis. For testing the linear null hypothesis and a
we apply
nonlinear alternative hypothesis,
and alternative hypotheses, we apply the
as the conventional F-test
P-test.
inappropriate
is
the
J-test;
and
for testing nonlinear null
These specialized
when
tests are
the alternative hypothesis
nested with respect to the null hypothesis. Under such circumstances,
simply constrain a portion of the hypothetical functional form
a schematic representation of the analytical
A
The null hypothesis.
follows a white-noise or
that the difference
necessary
to
we
is
non-
cannot
be zero. Figure 2
is
framework.
stringent null hypothesis
is
that the diffusion pattern
random walk process (Mahajan
between the numbers of adopters
at
t
et al, 1988).
and
(t-1) is
This specifies
random,
implying that the rate of diffusion will be driven by the error term only. The
mathematical form of the null hypothesis posits that the
first
differences in the
noncumulative adoption time-series are random:
where
mean
x(t) is
(e(t) is
x(t)
=x(t-l)
the
number
+
e(t)
(4 )
of adopters at time
uncorrected with
e(t-k) for all
t,
and the residuals
nonzero
k).
e(t)
Equation
have
a zerc
(4) specifies that
the adoption time-series will proceed by a sequence of unconnected steps, starting
each time from the value
previous period.
in the
The internal-influence model. Solving equation
for estimation,
we
^
1+
where mo represents
-m^ ex P ^ mt)
(
the
number
(5)
of adopters in the initial period.
we
obtain the following functional form:
x(t)
=m
+
(m-mo)
mQ
,
.
exp (-qmt)
(m-m
1 +
mQ
,
The external-influence model. From equation
m
[l-exp(-pt)]
Using a discrete
1
1
1
N(t) =
obtain the functional form
have:
N(t) =
formulation,
(1) to
)
1V
exp (-qm(t-l))
(2) earlier,
.
.
.
(6)
we
get:
(7)
Diffusion of Information Technology Outsourcing
19
Figure 2
Analytical Framework
Specification of
c
Start
White-Noise Model
as Null Hypothesis
I
Specification of
Individual Influence Models
as Alternative Hypotheses
i
Nonlinear Estimation of
Individual Influence Models
i
Testing of Influence Models
against White-Noise Model
uisng J-Test
I
Are
Two or More
Influence Models Significant?
i
No
Yes
Finer Specification of
Influence Models as
Null and Alternative Hypotheses
I
Comparison of
Influence Models
using P-Test
Stop
"^
Diffusion of Information Technology Outsourcing
20
This results in the following functional form for estimation:
m [exp(-p(t-D)
=
x(t)
exp(-pt)]
-
An NLS
The mixed-influence model.
influence
this
(8)
estimation
method
model has been used by Srinivasan and Mason
(1986)
nonlinear estimation procedure performed better than
(OLS) and
maximum
likelihood estimation
for the
its
mixed-
who showed
that
ordinary least squares
(MLE) counterparts. Accordingly, we
use the following functional form:
l-exp(-(p+q)
l-exp(-(p+q)t)
m
=
x(t)
1
+^ exp
(-(p+q)t)
1
-3 exp
(t-1))
(-(p+q)(t-l))
(9)
Test of alternative models against the null white-noise model. In the null
hypothesis,
we have
As reasoned
outsourcing.
model
is
linear,
Accordingly,
x(t)
=
model
specified a white-noise
usual F-test
earlier, the
we employ
inapplicable. Since our null
the J-test as derived by Davidson
we
estimate the regression:
(1-
a)
f(t)
is
for the diffusion of IT
+ a g<t)+
and MacKinnon
u(t)
(
where
f(t)=x(t-l)+e(t)
is
the null white-noise model, g(t)
an appropriate alternative model based on
some
constant,
and
u(t) is a
distributed with zero
the estimation
influence
random
mean and
error
a
which
is
10)
the predicted value under
likelihood estimation,
(10)
enable us to
test the alternative
the conventional asymptotic one-tailed t-test for the
Testing the influence models against one another. Here, both the null
alternative hypotheses nonlinear.
1981). This
is
The
P-test
is
=
similar to the J-test, except that
(1-oc) f(t)
+
a
g(t)
+ F B +
and the
thus used (Davidson and
we now
estimate the
following function:
x(t)
is
normally and independently
null hypothesis that a=0.
MacKinnon,
a
constant variance. The econometric properties of
and inference using equation
model by applying
maximum
is
(1981).
u(t)
ai)
Diffusion of Information Technology Outsourcing
where
F
is
a
and
f(t)
row
g(t) are the
21
appropriate null and alternative models respectively and
vector containing the derivatives of
f
with respect to B (the parameters of
0, evaluated at B.
Evaluating the two research questions.
using the analytical framework (Figure
If
first
research question
2) for the entire diffusion
estimation of the three influence models
assess the relative
The
is
performed, and the
adequacy of an influence model
two or more influence models turn out
directly tested
is
regime. Nonlinear
J-test is
used
to
vis-a-vis the white-noise
we perform
significant,
model.
the P-test that
allow us to directly compare the individual models. The second research question
is
appraised using the analytical framework in the two regimes (pre-Kodak and postKodak).
The
specification tests
do not permit
a direct verification of the
spanning across two different regimes. Our approach
is
based on the standard
econometric technique of testing structural change by breaking up a
into separate regimes using
1984).
The research question
and/or
of the
P-test results within
dominant type
some conceptually conceived
is
models
time series
full
cutoff points (Johnston,
then evaluated by comparing the separate
J-test
each of the regimes and interpreting the results in terms
of influence in the
two regimes.
Results
Research Question
Table 2
three models.
(a)
1:
Influence Sources in the Diffusion of IT Outsourcing
summarizes the parameter estimates and the model
The estimated values
for
m
fit
for the
range from 149 to 206, which are
in line
with predicted numbers of outsourcing adoption by a leading consulting firm. 23
We
note that amongst the three influence models, the internal-influence model has the
best absolute
fit.
Table 2
(b)
presents the results of comparisons of the alternative
specifications with the white-noise
the results
the J-test,
(p <.01).
comparing
all
model using the
J-test,
while Table 2
(c)
shows
different alternative specifications using the P-tests. Based
three specifications are significantly superior to the white-noise
However, based on the
P-test, the internal-influence
model emerges
on
model
as
Diffusion of Information Technology Outsourcing
providing the better
(p <.01).
It is
models.
model
We
to the data
when
tested against the external-influence
model
interesting to note that the internal-influence
able to reject the
not able to reject either the internal- or the external-influence
is
believe this
specification that gives
other hand, being
is
model
model but not the mixed-influence model. Further, the mixed-
external-influence
influence
fit
22
the
it
more
due
is
to the
power
parsimony of internal-influence model
to reject the external-influence
fully specified, the
mixed-influence model
model; on the
is
difficult to
be
rejected.
To ensure robustness
we
of the results,
carried out the
same analyses with
data aggregated at a quarterly level. These findings are summarized in Appendix B
where we can observe generally consistent
data.
Most
interestingly,
the null white-noise
influence
we
see that the external-influence
model under
model under the
models of influence: internal and
influences.
Given
is
the J-test, but can
now
model cannot even
reject
be rejected by the mixed-
P-test.
The mixed-influence model
our primary concern
monthly
results as those obtained using
is
an additive combination of the two basic
external.
While
this is
included for completeness,
with the distinction between the two basic models of
that the mixed-influence
model
surprising that the internal-influence model per se
is
is
more
fully specified,
unable
to reject the
it is
not
mixed-
influence model outright in our results. Nevertheless, between internal and
external influences, our results strongly support the
dominance of
internal
influence in the diffusion of IT outsourcing.
Research Question
2:
Kodak's Outsourcing as a Critical Event
Tables 3 and 4 summarize the results for the pre- and post-Kodak regimes
respectively.
The patterns
in the
parameter estimation and model
with those obtained earlier under the
models are
statistically significant
full
when
fit
correspond
regime. In the pre-Kodak regime,
tested against the white-noise
all
three
model using
Diffusion of Information Technology Outsourcing
Table 2
Diffusion Results for IT Outsourcing in the Full Regime
2(a)
Parameter
23
Diffusion of Information Technology Outsourcing
the J-test (p <.05).
However, using the
24
P-test,
no one model emerges
superior to another. In the post-Kodak regime,
statistically
statistically significant
internal-influence
under the
model
J-test (p <.05).
three
all
However, using the
rejects the external-influence
model
as being
models are
P-test, the
(p <.05).
The
results
pertaining to the mixed-influence model are similar to those discussed earlier under
the full regime.
Our
critical
event
is
management outsourcing
a facilities
we
Kodak and IBM. Thus, to ensure robustness of the results,
question using only the subsample composed of
for the pre-
and post-Kodak regimes. The
can see that the patterns of
with those based on the
Our
statistical
facilities
results are
management agreements
summarized
inference under the
J-
is
that while
all
regime that
is
is
is
and
in
Appendix
C.
We
P-tests are consistent
the models provide adequate
data in the pre-Kodak regime, the external-influence model
The implication
evaluate this research
full dataset.
overall conclusion
model wlwm compared
contract between
to the internal-influence
that the
Kodak event seems
model
to the
clearly an inferior
is
in the
to distinguish
fit
post-Kodak regime.
between an antecedent
not dominated by any single influence and a consequent regime that
characterized by internal influence (vis-a-vis external influence). Thus,
contend that there
is
a Kodak effect underlying this
we
unique observed pattern of IT
outsourcing diffusion.
Discussion
Diffusion of IT Outsourcing: Imitation as an Explanation
Our
model
is
analyses and results of this study suggest that the internal-influence
a better explanatory
mechanism
of the diffusion pattern of IT outsourcing
than the external-influence model. Further decomposition of the time period into
pre-
and post-Kodak regimes indicates
that the critical
Kodak event significantly
differentiates the pattern of influences in the diffusion of IT outsourcing
influence dominates in the post-Kodak regime, but not before.
-
internal
Diffusion of Information Technology Outsourcing
Table 3
Results
Diffusion
for IT Outsourcing
3(a)
Parameter
in the
Pre-Kodak Regime
Diffusion of Information Technology Outsourcing
Table 4
Diffusion Results for IT Outsourcing in Post-Kodak Regime
4(a)
Parameter
26
Diffusion of Information Technology Outsourcing
Mahajan and Peterson
perspective
adopting
it
is
27
(1985: p. 18) noted that the internal-influence
"most appropriate when an innovation
places social system
disadvantage in business)"
....
members
Our
....
socially visible,
at a 'disadvantage' (e.g., a
(emphasis added). Mahajan
imitative behavior in the adoption of
innovation perspective.
is
M-form
structure within the diffusion-of-
analyses are consistent with those of Mahajan et
Our
imitative behavior explanation
is
more pronounced
in the post-Kodak
isomorphism presented by DiMaggio and Powell
isomorphic phenomenon arises from
"the major factors that organizations
organizations" (Aldrich, 1979:
is
directly
p. 265).
a basic organization-theoretic
must take
argument:
into account are other
The tendency
for organizations to
model
through explicit interaction or an external change agent. The compelling
legitimization.
is
best
examined
As DiMaggio and Powell
in the context of organizational
'1983: pp. 151-152) articulated:
organizational technologies are poorly understood, [and]
when
(1983). Essentially,
often a response to uncertainty. Such organizational models can diffuse
logic of imitative behavior
the environment creates symbolic uncertainty,
model themselves
more
regime than
the theory of imitative behavior, our results are in line with the notion
of institutional
or
visibility since the
pre-Kodak regime.
From
others
al.
results strongly
support Mahajan and Peterson's discussion of the role of social
this
competitive
et al. (1988) tested the
(1988) but with significant methodological enhancements.
in the
and not
when
...
"When
goals are ambiguous,
organizations tend to
after similar organizations in their field that they perceive to
be
legitimate or successful."
Nelson and Winter
(1982: p. 123) discussed imitation as
an organizational
routine and noted: "envious firms, then attempt to duplicate [the] imperfectly
observed success."
It is
important to recognize that although the performance effects
of specific administrative innovations are never perfectly
known, imitation
nevertheless could occur (see Burt, 1987 and Scott, 1987 for elaborations of this
Diffusion of Information Technology Outsourcing
28
argument), possibly as an insurance against being locked-out of access to
new
resources or
when
others
sources of competitive advantage in the marketplace.
the above set of arguments contends that organizations
summary,
In
new
mimic
the underlying administrative processes are complex, especially under
conditions of environmental uncertainty. The
management
of IT infrastructure has
been made particularly complex given the uncertainties associated with the
proliferation of
competing standards, fluctuations
introduction of
new
obsolescence
(e.g.,
cost-performance trends,
in
systemic functionalities as well as risk of technological
Keen, 1988, 1990; Strassmann, 1991). Beyond uncertainties within
the technological environment, the role of IT in the business operations has
changed from an administrative support
strategic
Such
advantage
a shift
in the
marketplace
role
(e.g.,
towards providing
new
sources of
McFarlan, 1984; Scott Morton, 1991).
could compel organizations to mimic seemingly successful
administrative practices adopted by others within the social system.
While imitative behavior
adoption
may appear
to
as an underlying determinant for innovation
be non-rational, our position
broad sense, isomorphic behavior
termed
is
linked to what
is
that
Accordingly, rationality
is
need not be
DiMaggio and Powell
as "collective rationality." Further, Fulk, Schmitz,
developed the concept of rationality
it
in the context of
subjective, retrospective,
^.
and
(1983)
Steinfield (1990)
social influence
and attributable
model.
to influence
through information provided by others. However, the rational argument
advanced
Farrell
in the
economics
and Saloner,
It is
choice that
literature
on the adoption
so. In a
is
best
of standards (see for instance,
1985).
then possible that a firm adopts FT outsourcing as a deliberate strategic
is
based on a sound and rationally-conceived economic analysis.
A
recent
event study of stock market reactions to IT outsourcing established a positive impact
on the market value of the user firms adopting
and Venkatraman, 1992b). The implication
is
this
governance mechanism (Loh
that corporate
performance
is
Diffusion of Information Technology Outsourcing
enhanced through such
29
efficient choice of IT
governance mechanisms which
is
consistent with organizational economics (Williamson, 1991).
Our paper has added
to the
administrative innovation (such
proposing
a set of
arguments
growing
as:
literature
M-form
on the diffusion of
structure and joint venture) by
outsourcing as an
for the consideration of IT
administrative innovation. Further, from a methodological point of view,
contributed by developing and using
more powerful
we have
analytical estimation
techniques than those traditionally used. In terms of IT research, this paper
constitutes an attempt to empirically assess the diffusion pattern of a significant
current administrative innovation within the IT
domain and complements
and
the
study conducted by Gurbaxani (1990) in the area of diffusion of an important
technical innovation.
Limitations and Extensions
Contract
Our
size.
results are
based on outsourcing contracts that are
sufficiently large to merit reporting in professional trade periodicals
biased to the extent that smaller outsourcing decisions might be
professional
in
community
of IS
may have
Time-period.
is
we
we do
was
Our
due
to their
lower
visibility,
underlying imitative behavior in the
not think that the exclusion of the smaller
bibliometric search
located before April 1988.
the trade periodicals
may have changed
of IT outsourcing contracts. Further,
had outsourced
to the
adversely affected our results and interpretations.
the starting date of the databases) and
definition
believe that
critical catalysts
diffusion of IT outsourcing. Thus,
contracts
made known
managers through information sources not covered
our bibliometric search. However,
smaller contracts are not
and may be
it
was performed from January 1985 (which
no coverage
Our
of IT outsourcing that
met our
results are limited to the extent that
their criteria
has been
their IT operations historically
and emphasis
known
in the reporting
that several organizations
from the 1960s. As Markus (1984)
uncovered, such usage of external vendors was confined
to
time-sharing or
Diffusion of Information Technology Outsourcing
30
processing services involving service bureaus, limited use of integrated computer
systems and application software involving systems houses, and some
management involving
facilities
professional services firms (see also Phister, 1979 for
additional perspectives on the history of computing and information systems).
These forms of vendor participation
facilities
management had been
are,
however not widespread
largely confined within hospitals
institutions. Further, these contracts
were not usually reported
--
for instance,
and educational
in the
media
as they
are normally operational agreements related to the routinal conduct of the business
(e.g.,
custom programming and item processing).
The contemporary form
coverage, constitutes a
more fundamental
infrastructure strategy in the organization.
use of outsourcing
is
in line
from the extant media
of outsourcing, as evident
reconstitution of the overall IT
To some
extent,
we
believe this current
with the evolutionary importance attached to the
organizational information infrastructure. Thus,
we
are comfortable in the time
frame selected for our analysis. More importantly, even
if
we
include the few
organizations that have outsourced before our starting point, the actual diffusion
curve would be more S-shaped;
make
the internal-influence
Had we uncovered
pattern ex post, a
the
in other
words, the direction of the bias would
model more pronounced than we may have found.
dominance
of an external-influence
more rigorous examination
of this bias
model
in the diffusion
would be an important
extension of this paper.
Specification
diffusion
to the
models
and estimation of models.
that
phenomenon
is
We
specified
and estimated a
class of
consistent with our conceptual level of analysis pertaining
as well as the current state-of-the-art in estimation procedures.
Within the substantive domain of information technology, our model specification
represents an improvement over prior
work
(e.g.,
Gurbaxani, 1990). As the
analytical progress using the diffusion perspective gains
momentum
within IT
research, investigators should assess whether the theoretical arguments could be
Diffusion of Information Technology Outsourcing
31
framed within other forms of diffusion models such
(e.g.,
Chow,
1967;
Dodson and
Muller, 1978) where
as
we
dynamic diffusion models
can relax the assumption of a
constant adoption population.
Expanding
be
to
the scope of the model.
A
useful direction for future research
expand the scope of the constructs considered. For
assumed
a
models
population with homogeneous mixing, where the likelihood of
communication between any pair
analytical point of view,
it is
of prior
and potential adopters
(Strang, 1991).
support. For instance,
However,
it
may be
IT outsourcing within the
is
equal.
From an
possible to relax this assumption and incorporate the
effects of social structure into the diffusion
method
instance, our
would
this
modeling using an event-history
should be based on corresponding theoretical
possible to develop arguments that the diffusion of
US economy
is
channeled by structural characteristics of
the social relations that link the adopters. Then,
it
may be
appropriate to use the
degree of interlocking directorates between the user organizations and the
technology vendors /systems integrators as well as membership in
common
professional bodies and associations. Such an approach could build from the
Burt (1987)
who
work
considered the effects of cohesion and structural equivalence in the
well-cited diffusion of medical innovation discussed in
Complementing
a cross-sectional
Coleman
model of IT outsourcing.
et al. (1966).
Our study
constitutes a
longitudinal approach in explaining the incidence of IT outsourcing. This
complements
a cross-sectional
approach used by Loh and Venkatraman (1992a)
where the authors empirically examined the
level
level of IT outsourcing using firm-
determinants derived from two contexts: business cost structure, business
performance, financial leverage, IT cost structure, and FT performance. Our study
suggests that the institutional context at the social system level
delineating a
of
more comprehensive model
may add
value in
that can be empirically verified.
Diffusion of Information Technology Outsourcing
32
Conclusion
It is
the
clear that IT outsourcing
US economy
today.
is
an innovative governance mechanism within
Our paper sought
to
the pattern of diffusion of this governance
support for the internal-influence model;
the much-publicized
with an enthusiastic
sectional
understand the sources of influence
mode.
this
We
provide strong empirical
model was more pronounced
Kodak-IBM outsourcing contract than before.
call for
further
We
conclude
work on IT outsourcing — both along
and longitudinal perspectives.
after
cross-
in
Diffusion of Information Technology Outsourcing
33
References
Aldrich, H., Organizations and Environments
,
Englewood
Prentice-Hall,
E., "The Salesperson as Outside Agent or Employee:
Analysis," Marketing Science, 4 (1985), 234-253.
Anderson,
Bass, F.M, "A New Product
(1969), 215-227.
Growth Model
for
Consumer
A
Cliffs,
NJ, 1979.
Transaction Cost
Durables.," Management Science, 15
Blair, R., K.H. Roberts, and P. McKechnie, "Vertical and Network Communication in
Organizations: The Present and the Future," in R.D. McPhee and P.K. Tompkins (Eds.),
Organizational Communication: Traditional Themes and New Directions, Sage Publications,
Beverly Hills, 1985.
Brancheau, J.C. and J.C. Wetherbe, 'The Adoption of Spreadsheet Software: Testing
Innovation Diffusion Theory in the Context of End-User Computing," Information
Systems Research, 1 (1990), 115-143.
Burt, R.S., "Social Contagion and Innovation: Cohesion versus Structural Equivalence,"
American Journal of Sociology, 92 (1987), 1287-1334.
Chow,
G.C., "Technological
Review, 57(1967), 1117-1130.
Change and
the
Demand
for
J.S, E. Katz, and H. Menzel, Medical Innovation:
Indianapolis, 1966.
Coleman,
Computers," American Economic
A
Diffusion Study, Bobbs-Merrill,
Davidson, R. and J.G. MacKinnon, "Several Tests for Model Specification
Alternative Hypothesis," Econometrica, 49 (1981), 781-793.
Dearden,
J.,
"The Withering
Awav
of the IS Organization," Sloan
in the
Presence of
Management Review,
(1987),
87-91.
DiMaggio, P. and W. Powell, 'The Iron Cage Revisited: Institutional Isomorphism and
Collective Rationality in Organizational Fields," American Sociological Review, 48 (1983), 147160.
Dodson, J. A. and E. Muller, "Models of New Product Diffusion Through Advertising and
Word-of-Mouth," Management Science, 24(1978), 1568-1578.
Elam, J. J., "Establishing Cooperative External Relationships," in J.J. Elam, M.J. Ginzberg,
P.G.W. Keen, and R.W. Zmud (Eds.), Transforming the IS Organization, ICFT Press,
Washington, DC, 1988.
Evan, W.M., "Organizational Lag,"
Human
Organizations, 25 (1966), 51-53.
Farrell, J. and G. Saloner, "Standardization, Compatibility, and Innovation," Rand Journal of
Economics, 16 (1985), 70-83.
Fulk,
}., J.
"A Social Influence Model of Technology Use," in J.
Steinfield (Eds.), Organizations and Communication Technology, Sage Publications,
Schmitz, and C.W. Steinfield,
Fulk and
C
Newbury
Park, 1990.
Griliches, Z.,
"Hybrid Corn:
An
Econometrica, 25 (1957), 501-522.
Exploration in the Economics of Technological Change,"
Diffusion of Information Technology Outsourcing
"Diffusion in Computing Networks: The Case of BITNET," Communications
33 (December 1990), 65-75.
Gurbaxani,
of the
34
V.,
ACM,
King, K.L. Kraemer, F.W. McFarlan, K.S. Raman, and C.S. Yap,
of Information Technology," in
Proceedings of the Eleventh International Conference on Information Systems,
Gurbaxani,
V., J.L.
and the International Diffusion
"Institutions
Copenhagen, Denmark,
1990.
Henderson, J. and N. Venkatraman, "Strategic Alignment: A Model for Organizational
Transformation via Information Technology," in T. Kochan and M. Useem (Eds.),
Transforming Organizations, Oxford University Press, New York, 1992.
John, G. and B.A. Weitz, "Forward Integration into Distribution: An Empirical Test of
Transaction Cost Analysis," Journal of Law, Economics, and Organization, 4 (1988), 337-355.
Johnston,
Econometric Methods, McGraw-Hill,
J.,
Kanter, R.M.,
and Careers
York, 1984.
When Giants Learn to Dance: Mastering the Challenge
Simon and Schuster, New York, 1989.
of Strategy,
Management,
in the 1990s,
Keen, P.G.W., Competing
Ballinger,
New
Cambridge,
in
MA,
Time: Using Telecommunications for Competitive Advantage,
1988.
Keen, P.G.W., Shaping the Future: Business Design through Information Technology Harvard
Business School Press, Boston, MA, 1991.
,
Kirkpatrick, D.,
"Why Not Farm Out Your Computing,"
Fortune, September 23, 1991, 103-
112.
Kwon,
"A Diffusion
T.H.,
of Innovation
Approach
to
MIS
Infusion: Conceptualization,
Methodology, and Management Strategies," in Proceedings of the Eleventh
International Conference on Information Systems, Copenhagen, Denmark, 1990.
Loh, L. and N. Venkatraman, "Determinants of Information Technology Outsourcing:
Cross-Sectional Analysis," Journal of Management Information Systems, 8 (1992a), 7-24.
A
Loh, L. and N. Venkatraman, "Effects of Information Technology Outsourcing on Firm
Values: Evidence from the Stock Market," Sloan School of Management, Massachusetts
Institute of Technology, Working Paper, 1992b.
V and
Mahajan,
Hills,
CA,
Mahajan,
R. .Peterson,
Models for Innovation Diffusion, Sage Publications, Beverly
1985.
V., E.
Muller, and F.M. Bass, "New Product Diffusion Models in Marketing:
for Research," Journal of Marketing, 54 (1990), 1-26.
A
Review and Directions
Mahajan,
Sharma, and R. Bettis, 'The Adoption of the M-Form Organizational
Test of Imitation Hypothesis," Management Science, 34 (1988), 1188-1201.
V., S.
Structure:
A
Mansfield,
E.,
'Technical
Markus, M.L., Systems
in
Change and
Organizations
the Rate of Imitation," Econometrica 29 (1961), 741-766.
,
,
Pitman Publishing
McFarlan, F.W., "Information Technology Changes the
Business Review, bl (1984), 3, 98-103.
McLeod,
R.
and W.L.
Fuerst, "Marketing the
Inc.,
Marshfield,
Way You
MIS during Times
MA,
1984.
Compete," Harvard
of
Resource Scarcity,"
Diffusion of Information Technology Outsourcing
MIS
35
Quarterly, 6 (September 1982), 45-54.
Mueller, R.K., Corporate Networking: Building Channels for Information and Influence, The Free
New York, 1986.
Press,
Nelson, R.R and S.G. Winter, An Evolutionary Theory
Cambridge, MA, 1982.
Economic Change, Belknap Press,
of
Nilakanta, S. and R.W. Scamell, 'The Effect of Information Sources and
Communication Channels on the Diffusion of Innovation in a Data Base Development
Environment," Management Science, 36 (1990), 24-40.
Phister, M., Data Processing Technology
and Economics, Digital Press, Bedford,
MA,
1979.
Powell, W.W., "Neither Market nor Hierarchy: Network Forms of Organization," in B.
Staw and L. Cummings (Eds.), Research in Organizational Behavior, JAI Press, Greenwich, CT,
1990.
Richmond, W.B., A. Seidmann, and A.B. Whinston, "Incomplete Contracting Issues
Information Systems Development," Decision Support Systems, 1992, in press.
Rogers, E.M., Diffusion of Innovations, The Free Press,
New
Rogers, E.M. and R. Agarwala-Rogers, Communication
York, 1976.
in
in
York, 1983.
Organizations,
Rogers, E.M. and D.L. Kincaid, Communication Networks: Toward a
The Free Press, New York, 1981.
New
The Free
Press,
New
Paradigm for
Research,
Scott Morton, M.S. (Eds.), The Corporation of the 1990s,
1991.
Scott, W.R.,
"The Adolescence of
Oxford University Press,
New
York,
Institutional Theory," Administrative Science Quarterly, 32
(1987), 493-511.
Snow, C.C. and
R.E. Miles,
Lamb
Organizations," in R.
"The Role of Strategy
(Eds.),
Advances
in the
in Strategic
Development of a General Theory of
Management, JAI Press, Greenwich,
CT, 1983.
Srinivasan, V. and C.H. Mason, "Nonlinear Least Squares Estimation of
Diffusion Models," Marketing Science, 5 (1986), 169-178.
Strang, D.,
"Adding
Sociological
Methods
Social Structure to Diffusion Models:
Research, 19 (1991), 324-353.
An
New
Product
Event History Framework,"
&
Strassmann, P. A., The Business Value of Computers, The Information Economics Press,
Canaan, Connecticut, 1990.
New
Swanson, E.B., M.K. Fuller, S. Nidumolu, N. Ramiller, and S.G. Ward, "Illusive Effects
on the Diffusion of an Innovation: A Comment," Management Science, 37(1991), 15001506.
Teece, D., "The Diffusion of an Administrative Innovation," Management Science, 26 (1980),
464-470.
Thompson,
J.D.,
Organizations in Action, McGraw-Hill,
New
York, 1967.
Diffusion of Information Technology Outsourcing
36
Venkatraman, N., L. Loh, and J. Koh, 'The Adoption of Corporate Governance
Mechanisms: A Test of Competing Diffusion Models," Sloan School of Management,
Massachusetts Institute of Technology, Working Paper, 1992.
Walker, G. -nd D. Weber, "A Transaction Cost Approach to
Administrative Science Quarterly, 29 (1984), 373-391.
Whang,
S.,
Make-or-Buy Decisions,"
"Contracting for Software Development," Management Science, 38 (1992), 307-324.
Williamson, O.E., "Comparative Economic Organization: The Analysis of Discrete
Structural Alternatives," Administrative Science Quarterly, 36 (1991),
Zaltman, G., R. Duncan, and J. Holbeck, Innovations
Publishing, Malabar, FL, 1984.
Zmud,
&
Organizations Robert E. Krieger
,
R.W., 'The Effectiveness of External Information Channels in Facilitating
Innovation within Software Development Groups," MIS Quarterly ,7 (June 1983), 43-58.
i
D
fiu sion o*
Information Technology Outsourcing
—
E
3
- - - "5 re
c c c c c
—
O
(*.
C
—
C
—
re
c
r-,
>C f.
X
CO x
- JS
c c
—
re
re
re
c c c c
o
r*
lc
<vi
^
c
o
r~,
if,
f,
—
£2
—
—
l/".
iri
£
r*l
.— Or
t-.
f
re
re
c c
tl
c
x
r"
;
^
-Z
f. " *£
r. tN
;
ir,
c-^
i/-.
re
re
~^
c c
2
£
<
Cm
22_22222222_22_
______________
______ 2_222
— _ _ _
_
LT.
c
U
c
U
</.
E
if.
cj
E
O
IT.
~.
E
O
U
x.=c
5
Z
5
o
5-
=
W>* K
B
3
c E
a
P-o. E
e c
"re
E
t
re
o
O
-
o
UU
'E
—
c
c
v.
Ci
c C
ss
s
to
>>
sj
C.
qQ
6C
c
o
re
a
re
3
.£
°-5
c
C
3
3
5
CI
a:
U
_=
c c 3
c,
c 3
"
c
u
CCO
=
—
3 3
>
-> >
a.
ovJ
o _
u 2C:
<
o
Hi
t:
u
a..o
5
E2 *2
Ey
E E
£
U
3
c
—
6£
E
E
Ci
E
d-u*.
_
r
ci
c
eui
'X
rcS2o---:E
t/5
Is
3
5
E
o
c ?
Cl
U
u-,
•tf
P
c
o
tr.
c
>->
a.
E
c
c
N
'E
6C
5
<
I/)
>
f-
5
i
Q
U
c
_^
Si
<o
n
>,
-'
*3
g
c-
M
2 S
*
I*
Ct •{-
C C '
.= tu
3
3
in
Cl.
IB
ZC
%9 .E
—
u
Si
-.
—
QC
C
U
c
u.
CI
_o
n
o
£
— i)—
K
re
2
il
S
A u.C U" 35
U
x- U
o 8'F
5
<
m.
—
<
3
§
^^ l
u
c
CI
P
3DZ^ uj2ZUcu2uh~ujc2lu=>
tntn
13
cl
r B
- E £ t
c c
<< -I-
—
« «
ci
O
it
<
o
I
Z
u
1
2
'
ffjl
£Sxx£555x£55xxxxxxxSxx£xxx£xxx
&i
Ci
b
...Is xE ~E
is. all
<
iiil< < <
Cf,
6
ci
_c
E
Cl
o
> > C
£ o
Cl
k,
>> >^
c
t.
i.
u.
%%££
—
1 B B tl tl
— — — _ x
E E E E E
C
^ U
2ZCQ^2222 5;ZCOCC22Z^2
*r
l-«
i_i
t_*
(_*
Diffusion of Information Technology Outsourcing
ir,
in
SS
—
c
—
re ir.
c
—
c c
rvi
3^
Cinn^-irv^-^^iAir;
—
^_
SilBs
>*
in
4.
38
c
2
re
c
i
r^ rj
oo*?o
— —— -
re
re
cr,
i/i
o-
to
tn
22
22222222_2222s222_222222_22
— —— —
— — s —
— s
s — —
"
j-
U. U. U. U.
-
WZ
E
fc
Z
60
c
C
c
«
s-
o
C
y
=
o
n
3
<£
E
S
I
5
-
(A
c
%
I
U
a:
c
&
IT
U E
c
01
o
"C -if.
l£
DH
2-
C
- _
- o
c- _
<
<.<£
-J
c
A
'E
6C
^ —
— a
— <
>-,—
IT.
-</,
re
sc
5 c
^ *^ Q ^ ^ ^
=
\r.\r.u. y.
-
-
Diffusion of Information Technology Outsourcing
39
Appendix B
Diffusion Tests in the Full Regime using Quarterly Data
We
further
aggregated
at
examine our Research Question
1
using the diffusion data
the quarterly level. Table Bl presents the inferential results of testing
an influence model against
(a)
the white-noise
another influence model using the
P-test.
The
model using
the
J-test;
and
(b)
J-test indicates that the internal-
and
mixed-influence models reject the white-noise model (p <.05 and p<.01
respectively), while the external-influence
the data than the white-noise model.
model
is
unable to provide a better
However, based on the
model
is
influence
model
(p<.05). This set of results establishes the robustness of those
by the internal-influence model (p<.10) and the mixed-
obtained using monthly data as discussed earlier in the
text.
Table Bl: Inference Results for the Full Regime using Quarterly Data
Alternative
Hypothesis
to
P-test, the external-
influence
rejected
fit
Diffusion of Information Technology Outsourcing
40
Appendix C
Diffusion of Facilities Management Outsourcing
in the Pre- and Post-Kodak Regimes
We
assess the robustness of our results pertaining to Research Question 2
analyzing the diffusion pattern using only
contracts. This
a facilities
is
to
account for the
fact that
management agreement, and
facilities
our
by
management outsourcing
critical
event
(i.e.,
Kodak contract)
the influences in the social system
is
may
only operate via this specific type of outsourcing. Our results (see Tables CI and C2)
supporting the dominance of the internal-influence model over the externalinfluence
model
replicate those based
on the
full
sample discussed
Table CI: Inference Results for Facilities
in the
Alternative
Hypothesis
Management
Pre-Kodak Regime
in the text.
Diffusion
Diffusion of Information Technology Outsourcing
41
NOTES
1
2
Network World, February 17, 1992: pp. 1, 31-36
See Computerworld, December 25, 1989/January
1,
1990: p. 8
and Information Week, October
7,
1991:
p. 44.
3
The breakdown
follows:
of industry outsourcing revenue for the year 1990
35% commercial; 38%
federal;
20%
international;
7%
amongst the various sectors
government
is
as
state/local
(Computerworld, May 20, 1991: p. 141).
^Network World, February 17, 1992: pp. 1, 31-36.
5 Kirkpatrick (1991:
6
p. 104) reported that the 1990 revenue and market share for the five biggest
outsourcing vendors are as follows: (1) EDS ($6110m, 20.7%); (2) IBM ($3900m, 13.2%); (3)
Anderson Consulting ($1880m, 6.4%); (4) DEC ($1800m, 6.1%); (5) CSC ($1500m, 5.1%).
The gross margin in the business has been estimated to be as high as 25-30% (Kirkpatrick , 1991:
p. 104).
7
Computerworld, June 4, 1990: p. 80
° Computerworld, January 13, 1992:
' Computerworld, June 11, 1990: p. 1
p. 8
10 Wa// Street Journal,
July 26,.1989: p. 3
11
Computerworld, December
25, 1989/January 1, 1990: pp. 14-15
not include the case where the IT infrastructure is transferred to a whollysubsidiary as the mode of governance is still "hierarchical" as opposed to "market."
12 In our definition,
owned
we do
^Computerworld, April 16, 1990: p. 77
^Computerworld, September 30, 1991: p. 1
^Computerworld September 16, 1991: p. 113
1
,
16
Computerworld, November 11, 1991; p. 70
17 Computerworld, January
14, 1991: p. 10
^Wall Street Journal, July 26, 1989: p. 3. At the same time, Kodak also outsourced as microcomputer
systems operations to Businessland.
19 See Harvard Business School cases, "Eastman Kodak Co.: Managing Information Systems Through
Strategic Alliances," 1-192-030 and "Digital Equipment Corporation: The Kodak Outsourcing
Agreement," N9-191-039 for background details.
10 Computerworld,
January 8, 1990: p. 67
21 Interview
with Howard Anderson, the managing director of The Yankee Croup reported in
Computerworld, January 8, 1990: p. 76.
^First, we conducted an exploratory search to discern the patterns of reporting for outsourcing
contracts in the databases. Next, we generated our sample using a generalized search procedure
which avoided any bias to any vendor or user. We used the following search commands to draw
out all the contracts in the databases: term(contract) AND (computer OR system OR
information) AND xxx, where we respectively used outsourcing, facilities management, systems
integration, time-sharing, network management, systems management, information management,
and data processing management for xxx. Throughout this procedure, we are confident that the
sample is adequately comprehensive for the results to be generalizable.
^Network World, July 2, 1990: pp. 1, 47
2799
5
Date Due
05 199a
*tf
MAY 2 31P5
•
ste*.
»g
2o
MAR.
1
tiff
1M6
MAY 11
^fc
nz
1
1JJ95
4 }9%
1
-
Lib-26-67
MIT
3
TOflO
LIBRARIES
00711414
2