You are on page 1of 19

Information Systems Research

Vol. 15, No. 3, September 2004, pp. 268286


issn1047-7047 eissn1526-5536 04 1503 0268
informs

doi 10.1287/isre.1040.0029
2004 INFORMS
An Economic Model of Product
Quality and IT Value
Matt E. Thatcher
Department of Management Information Systems, Eller College of Management, University of Arizona,
430 McClelland Hall, Tucson, Arizona 85721, thatcher@eller.arizona.edu
David E. Pingry
Department of Management Information Systems and Department of Economics, Eller College of Management,
University of Arizona, 430 McClelland Hall, Tucson, Arizona 85721, pingry@eller.arizona.edu
W
e use an economic model to formalize the complex relationships among IT investments, intermediate per-
formance measures (e.g., product quality and output levels), and economic performance (e.g., productivity,
prots, and consumer surplus). We demonstrate that a prot-maximizing monopolist invests in IT (modeled as
changes in parametric characteristics of the rm) to design a better-quality product and charge a higher price.
While this prot-maximizing adjustment generates more consumer surplus, it also increases production costs
in a way that adversely affects productivity. In contrast, a simple model extension shows that when a rm is
unwilling or unable to improve product quality, then IT investments result in suboptimal improvements in prof-
its, an increase in consumer surplus, and an increase in productivity. Together, these models highlight the way in
which product quality moderates the relationship between IT investments and economic performance. We also
demonstrate that these relationships are robust to the socially optimal case in which a social planner chooses
price and quality to maximize social welfare. In addition, we demonstrate that the results of the monopoly
model hold when considering the design and development of products offered free of charge (e.g., free online
content), but that provide indirect benets to the rm (e.g., more advertising revenues).
Key words: economic value; IT value; IT investments; consumer welfare; productivity; quality; economic
modeling
History: M. S. Krishnan, Associate Editor. This paper was received on June 10, 2002, and was with the authors
1.3 months for 2 revisions.
1. Introduction
For over a decade, empirical studies in the informa-
tion technology (IT) value literature have attempted
to quantify the economic benets realized by rms
and industries from IT investments. Most studies
have focused on productivity as the measure of IT
value (Anderson et al. 2003; Baily 1986; Barua and
Lee 1997; Lee and Barua 1999; Hackett 1990; Lehr and
Lichtenberg 1998, 1999; Loveman 1991; Mandel 1998;
McCrune 1998; McGee and Wilder 2000; Metcalfe
1992; Metheny 1994; Panko 1991; Roach 1991; Strass-
mann 1990, 2001). Some studies have used prot or
consumer surplus as the measure of IT value (Bharad-
waj 2000, Brynjolfsson 1996, Dos Santos et al. 1993,
Hitt and Brynjolfsson 1996, Shin 2001, Strassmann
1997). Still other studies have focused on the impact
of IT on intermediate performance measures such
as product quality and output levels (Banker and
Kauffman 1991, David et al. 1996, Devaraj and Kohli
2000, Mukhopadhyay et al. 1997, Rai 1997, Rai et al.
1996, Weill 1990). The results of these studies have
been mixed.
In an effort to bring some order to the literature,
Hitt and Brynjolfsson reminded us that productiv-
ity, prots, and consumer surplus are different mea-
sures of economic performance, and that while they
are related, they are ultimately separate questions
(1996, p. 121). In this paper we formalize these rela-
tionships using a closed-form, analytical model based
in economic theory. Specically, we model a prot-
maximizing rm that offers a product (or provides
a service) characterized by two attributes, price and
quality. We examine the impact of investments in
IT infrastructure (that improve the efciency with
268
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 269
which a rm designs, develops, and manufactures
a product), on intermediate performance measures
(e.g., product quality and output levels), and on
economic performance (e.g., productivity, prots, and
consumer surplus). The basic model shows that a
monopolist maximizes prots by using IT invest-
ments to design a better-quality product and charge a
higher price. While this prot-maximizing adjustment
increases consumer surplus, it also increases produc-
tion costs in a way that adversely affects productiv-
ity. In contrast, a simple model extension shows that
when a rm is unwilling or unable to improve prod-
uct quality, then IT investments result in suboptimal
improvements in prots, an increase in consumer sur-
plus, and an increase in productivity. Together, these
models highlight the way in which product quality
moderates the relationship between IT investments
and economic value. We also demonstrate that these
relationships are robust to the socially optimal set-
ting in which a social planner chooses product price
and quality to maximize social welfare. In addition,
we show that the results of the monopoly model
hold when considering the design and development
of products offered free of charge (e.g., free online
content), but that provide indirect benets to the rm
(e.g., more advertising revenues). Finally, we interpret
the current body of empirical work on IT value in the
context of our model.
Investments in IT infrastructure that enable a rm
to design, develop, and manufacture a product of
given quality faster and cheaper increase production
efciency (or reduce the marginal cost of improving
product quality). In other words, IT is an input that
enables the rm to seek higher quality of its out-
put. For example, Marsden and Pingry (1993) point
out that when a rm purchases a Decision Support
System (DSS), it does not necessarily lower the cost
of solving decision problems for the rm. It lowers
the marginal cost of searching for problem structure,
which in turn allows the rm the freedom to search
for structures that improve the quality of the solution
to the problem. Other examples of IT investments that
lower the marginal cost of quality include:
CAD/CAE tools and simulation and visualization
tools by manufacturers of automobiles and con-
sumer electronics that reduce prototype iterations and
improve communications with consumers and exter-
nal suppliers, leading to better-quality automobiles,
faxes, printers, PDAs, etc.;
Collaborative technologies by service providers,
such as nancial advisors and travel agents, that
enable individuals and groups to organize data and
communicate ideas more effectively, leading to better
decision making and recommendations;
Prototyping tools, usability lab technologies, and
user-centered design processes by developers of software
applications and consumer electronics that directly
involve end users in the software/product design
process early and often, leading to more functional,
highly usable applications that better meet user
needs;
Website design tools by e-commerce rms that
enable the development of more dynamic, interactive,
online storefronts that provide faster download times,
simpler navigation, and higher usage and conversion
rates.
In the following section we present a brief review
of the empirical work that examines the contribution
of IT investments to economic performance.
2. Literature Review
Early empirical studies of IT value examined the con-
tribution of aggregate IT spending to productivity at
the economy and industry levels of analysis. These
studies typically found little or no improvement in
productivity despite massive investments in IT since
the early 1970s (Baily 1986, Hackett 1990, Panko 1991,
Roach 1991, Strassmann 1990, and see Brynjolfsson
1994 for a review of the empirical work). The pri-
mary explanation for the so-called IT productivity
paradox was that the collection of data aggregated
at the economy and industry levels had led to the
mismeasurement of inputs and outputs in the produc-
tivity measures and, therefore, the underestimation of
productivity gains from IT investments (Brynjolfsson
and Hitt 1996; Brynjolfsson 1993, 1994; McCrune 1998;
Metcalfe 1992; Metheny 1994).
Later studies attempted to address the mismeasure-
ment problem by examining the business value of IT
using rm-level (or disaggregated) data. The major-
ity of studies have continued to focus on measur-
ing the contribution of IT to productivity, with many
nding signicant contributions of IT spending to
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
270 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
rm productivity (Barua and Lee 1997; Brynjolfsson
and Hitt 1996; Brynjolfsson 1993; Lee and Barua
1999; Hitt and Brynjolfsson 1996; Jorgenson and Stiroh
1995; Lehr and Lichtenberg 1998, 1999; Lichtenberg
1995; and see Dedrick et al. 2003 for an extensive
review). Other studies have focused on the relation-
ship between IT investments and rm protability
(see Dedrick et al. 2003 for a review). Strassmann
(1997) and Hitt and Brynjolfsson (1996) have argued
that there is no relationship between IT investments
and measures of rm protability. Alternatively,
Bharadwaj (2000) found a positive and signicant
relationship between a rms IT capability and a
variety of prot performance measures. Still other
studies have identied specic factors that affect the
impact of IT investments on protability. For example,
Dos Santos et al. (1993) found that innovative IT
investments increase rm value, while noninnovative
(or incremental, follow-up) investments do not. In
addition, Shin (2001) found that IT investments do not
improve rm protability unless they are properly
aligned with the rms business strategies. A smaller
set of studies have focused on measuring the ben-
ets passed on to consumers from IT investments.
For example, Brynjolfsson (1996) found that for the
year 1987, IT investments generated approximately
three times their costs in value for consumers. Sim-
ilarly, Hitt and Brynjolfsson (1996), using data from
370 rms from 19881992, found that IT had created
substantial value for consumers.
Other studies further disaggregated IT investments
in terms of specic IT applications and activities
within a rm and examined the contribution of these
specic investments on intermediate performance
measures, such as product quality and rm output
levels, which may, in turn, affect economic perfor-
mance measures. Rai (1997) disaggregated rms IT
budgets into key elements of IT infrastructure and
found that IT budget allocations to IS staff, telecom-
munications, and hardware were positively correlated
to rm output and labor productivity. However, he
found a negative correlation between software invest-
ments and labor productivity. Banker and Kauffman
(1991) found that while ATM network investments
had a positive impact on a banks local deposit market
share, they contributed little to a banks nancial per-
formance. In particular, they found that the presence
of an ATM was associated with a decrease in labor
productivity of branch tellers. Weill (1990) found a
signicant relationship between investments in data-
processing systems and productivity, but was unable
to identify gains associated with strategic systems
or informational investments. He noted that CASE
tools have been particularly disappointing in terms of
their contributions to productivity. David et al. (1996)
found that investments by hotels in back-ofce com-
puter modules led to productivity gains. However,
guest-operated technologies such as in-room infor-
mation, vending, and entertainment and automated
reservation systems led to a decrease in productiv-
ity, but an increase in customer-service levels and
the number of services offered. Mukhopadhyay et al.
(1997) examined the benets of the optical character
recognition and bar code sorting technologies used to
sort mail at the U.S. Postal Service, and found that
such investments led to signicant increases in mail-
sorting output and improvements in quality. Finally,
Devaraj and Kohli (2000) found that investments by
hospitals in DSS that were used to help evaluate con-
tracts led to better-quality products and services and
higher revenues.
3. Basic Model
We present an integrated model based on Thatcher
and Oliver (2000a, b), who examine the impact on
productivity of IT investments by a single-product
monopolist that chooses product price and quality.
Our work extends this model in several ways. First,
we extend the model to examine the impact of IT on
alternative measures of productivity (which we clar-
ify later in this section) and on consumer surplus.
Second, we extend the model to consider the impact
of IT when the rm is either unable or unwilling to
adjust product quality in response to IT investments.
Together, these models enable the interpretation of
the ndings from empirical studies reviewed in 2.
Third, we examine the robustness of the monopoly
result to the case in which a benevolent social plan-
ner chooses product price and quality to maximize
social welfare. Finally, we extend the model to con-
sider products designed and developed by the rm
that are offered free of charge, but that still provide
indirect benets to the rm. Together, these exten-
sions provide a robust and integrated understanding
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 271
of the complex relationships among IT investments,
intermediate performance measures, and economic
performance measures.
Formally, we consider a single-product monopolist
that designs and produces a single product or service
characterized by two attributes, quality (Q) and unit
price (P), which are both determined by the rm.
1
We
model product quality as a numeric level that may be
thought of as a composite measure of some vector of
attributes (e.g., features, performance, reliability, con-
venience, and responsiveness) valued by consumers
(Garvin 1984). In this model, the rm chooses a price-
quality combination (P
m
, Q
m
) that maximizes prots.
The quantity demanded by consumers for this
product is (Lilien et al. 1992)
D=u |P +cQ (1)
where u, |, c > 0 are parameters that can be deter-
mined empirically for each industry. The rms pro-
duction costs are
C =]Q
2
+cQD (2)
where ] , c > 0 are technology parameters that char-
acterize the rms production efciency. The rst cost
term, ]Q
2
, is referred to as the xed cost of quality
and represents the R&D cost (e.g., product analysis,
prototyping, and evaluation) of designing and devel-
oping a product of given quality. This cost term is
assumed to be quadratic because improving the over-
all quality of a product becomes increasingly difcult
(and increasingly costly) as product quality increases.
The quadratic form of this cost term has intuitive
appeal, is widely used in the industrial organization
literature (Barua et al. 1991, Cremer and Thisee 1994,
Gal-Or 1987, Lambertini 1996, Lambertini and Orsini
2002, Moorthy 1988, Moorthy and Png 1992, Schmitt
1
Monopoly or near-monopoly power may be acquired naturally
in the market through standards creation, network externalities,
economies of scale, location, ownership of unique resource, etc., or
may be granted by the government using patents or other legal
means. Therefore, Microsoft and the pharmaceutical industry are
excellent examples of application areas for this model. However,
we note that this model does not apply to monopolies in which
the government retains regulatory control of the product quality
and price, as has traditionally been the case in electric and other
utilities.
2002, Spence 1975), and has been used to analyze a
diverse range of products including software products
and information goods, medical services, and airline
services.
The second cost term, eQD, is the variable production
cost and represents the cost of manufacturing and dis-
tributing products of given quality for consumption.
This cost term is also assumed to increase with prod-
uct quality. This relationship may be due to higher
levels of maintenance activities required to support
higher-quality products. For example, increasing the
quality of travel services (e.g., by offering more travel
options) may require more processing per travel
customer.
The realization of parameters ] and c in Equation (2)
characterizes the efciency of the rms cost structure.
Specically, a reduction in the design cost parameter,
] , implies a more efcient R&D capability that allows
the rm to design a product of given quality at lower
cost. A reduction in the variable production cost
parameter, c, implies a more efcient production pro-
cess that allows the rm to manufacture, distribute,
and maintain a product of given quality at lower unit
cost. The realizations of these parameters depend on,
among other factors, the IT infrastructure (e.g., physi-
cal IT assets and methodologies) implemented by the
rm. For example, investments in CAD/CAE tools,
prototyping tools, and decision support systems tend
to lower ] while investments in electronic distribu-
tion technologies, customer relationship management
(CRM) applications, and enterprise resource planning
(ERP) systems tend to lower c. In the analysis that fol-
lows, we consider changes to each parameter (realized
through investments in IT infrastructure) separately
to develop some initial insights as to their individ-
ual effects on the rms product design choices and
economic performance measures. Below we present
examples of specic IT investments that improve pro-
duction efciency (or lower c and ] ).
The strategy of large pharmaceutical companies is
to invest heavily in the design and development of
new drugs that will qualify for patent protection.
Since the early 1990s, companies such as Bristol-Myers
Squibb Co. and Pzer Inc. have invested millions of
dollars in combinatorial-chemistry technologies to help
automate the drug discovery process (lower ] ). These
technologies can . . . create thousands of chemicals
almost overnight by mixing and matching common
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
272 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
building blocks . . . (Landers 2004, p. A1). Although
these technologies have been slow in generating FDA-
approved drugs, they have . . . helped improve [the
quality of] some drugs that were found by more tra-
ditional means . . . (Landers 2004, p. A1).
For $19.95 a month Major League Baseballs
MLB.com offers a package of 10 different MLB broad-
band products, including live radio and video broad-
casts, video highlights, and archives of the greatest
moments in baseball (Grant 2003). MLB.com recently
rewrote many of its internal applications and made
signicant IT investmentsfor example, two Net
Sealer 9800 content switches (which lower both ]
and c)to redesign the site architecture and eliminate
chronic slowdowns and outages experienced in the
2001 and 2002 baseball seasons (Higgins 2003). These
investments have enabled MLB.com to offer better-
quality content (e.g., reliable, real-time game updates
and scores) and features (e.g., online polling for the
All-Star Game) and to do so 15% faster than it did
before using 30% less bandwidth.
2
Given Equations (1) and (2), revenues, prots, and
consumer surplus, respectively, are
=P(u |P +cQ), (3)
r =(P cQ)(u |P +cQ) ]Q
2
, and (4)
w=
(u |P +cQ)
2
2|
. (5)
Productivity is dened as the ratio of output
value to its related input value. When quantity is
2
Other examples drawn from more competitive industries, but
which may be applied to a monopoly rm, include the following.
AutomotivemanufacturerssuchasBuickhaverecentlyinvested
heavily in CAD/CAE tools (which lower ] ) that allow them to use
simulation and visualization tools to optimize new car designs on
computer (Lyon 2001, Port 2003, Waurzyniak 2001, Vasilash 2001).
United States Plastic Corp. (USPC) invested in VisNetic Mail-
Flow2.0, a Web-basede-mail management program. This CRMappli-
cation (which lowers c) helps USPC to automatically direct customer
e-mails andtracktheprogress of thecustomer requests (Bannan2003).
This has led to faster response times and better customer support
quality.
Canonuseda tool (OneSpace Designer Modeling by Co-Create)
to design the Canon LBP-1210 personal laser beam printer. The tool
allowed the team to reduce prototype iterations, share design data
across teams and departments, and revise the design earlier in the
process, leading to a better-quality product in terms of functionality,
print quality, paper-feeding speed, and quietness.
assumed to be a good proxy for value, productiv-
ity is dened as quantity of output (e.g., number of
checks processed) per quantity of related input (e.g.,
labor hours, number of employees). Measures of pro-
ductivity in early studies on IT value were based
on counts and numbers (see Brynjolfsson 1994 for
a more detailed discussion). However, quantity, in
many casesespecially in the service sectoris not
considered a good proxy for the value created by IT
(Klassen et al. 1998). Therefore, later studies measured
output value based on rm revenues in productiv-
ity measures (e.g., sales dollars divided by the cost
to the producer) to account for some of the intan-
gible value (e.g., better product quality and conve-
nience) associated with IT. Finally, some of the value
created through IT investments are passed on to con-
sumers through competition and will not be captured
in measures of rm revenues. Therefore, others have
attempted to measure the output value of IT invest-
ments based not only on rm revenues, but also con-
sumer surplus.
We consider three measures of productivity, which
we term output productivity, rm productivity, and
social productivity. Output productivity is dened as
the ratio of the number of products produced to its
related input value. Because we are interested in mea-
suring the productivity of the production process,
output productivity is
j
output
=D,C. (6)
Firm productivity is dened as the ratio of output
value to the rm to its related input value,
j
rm
=,C. (7)
Finally, social productivity is dened as the ratio of
social output value (i.e., the sum of rm revenues and
consumer surplus) to its related input value,
j
social
=
+w
C
. (8)
At this point, we note that Thatcher and Oliver
(2000a, b) incorporate the indirect xed costs ( ) of
doing business (i.e., building leases, equipment depre-
ciation and maintenance, and administrative costs)
in the denominators of their productivity measures.
However, including in the input value implies that
two rms, making the same IT investments and mak-
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 273
ing the same optimal adjustments to price and quality
(since optimal choices are not affected by ), will real-
ize different levels of productivity only because one
rm has more equipment depreciation than the other,
or because one rm has more administrative costs
than the other. Therefore, using direct production
costs as the input value for the productivity measures
has more intuitive appeal and is more appropriate in
this model.
Finally, we assume that IT investments (i.e., param-
etric reductions in the cost function) are acquired at
no cost. This assumption provides the best-case scenario
for observing improvements in productivity from IT
investments because incorporating such investment
costs in the input value can only depress the pro-
ductivity measures presented in Equations (6)(8).
Although managers must consider investment costs
when making the investment decision, the assump-
tion that investment costs are zero does not change
the relationships presented in the paper. This model
does not tell rms whether or not to invest in IT. That,
of course, will depend on the difference between the
additional prots the investment is expected to gener-
ate and the cost of the investment. Instead, this model
examines the impact of the IT on intermediate per-
formance measures and economic performance mea-
sures. In fact, we will show that improvements to IT
infrastructure, even when acquired at no cost to the
rm, will lead to a decrease in productivity.
4. Basic Model Results
Given Equations (1)(4), the equilibrium choices of
quality and price are (see tables for derivations)
Q
m
=
u(c |c)
4|] (c |c)
2
and (9)
P
m
=
u(2] +c(c |c))
4|] (c |c)
2
. (10)
Substituting Equations (9) and (10) into Equations
(1)(8) yields the equilibrium values presented in
Table 1. To guarantee a positive and unique equilib-
rium in quality and in price, the following conditions
must hold:
] >
(c |c)
2
4|
and (C1)
c -
c
|
. (C2)
The denominator of each equilibrium value in
Equations (9) and (10) is the determinant of the
2 2 Hessian matrix. Condition (C1) ensures that the
determinant is positive, which guarantees the con-
cavity of the prot function in price and in quality
and guarantees a unique maximum. Condition (C2)
ensures that the rm chooses a positive level of
quality and charges a positive price in equilibrium.
Together, these conditions characterize the relation-
ships between the cost parameters (c and ] ) and the
demand parameters (| and c) that must be satised.
As in all theoretical work, the question of whether the
model conditions are consistent with the real world is
an empirical question.
Condition (C2) |c - c,|] implies that the variable
cost parameter (c) may take on a range of values
in this model. Obviously, condition (C2) will hold
for rms with an infrastructure that enables them
to manufacture and distribute a product at low unit
cost (e.g., software providers and information goods
providers). For example, the unit cost to Microsoft
to reproduce another copy of Windows or Ofce
for distribution, or the marginal cost to MLB.com
to reproduce another audio or video replay of a
baseball broadcast, is essentially zero. In addition,
condition (C2) will hold for rms that manufacture
and distribute a product at a more substantial unit
cost, as long as demand is sufciently sensitive to
product quality relative to price. For example, the
unit cost to manufacture and distribute an automo-
bile may be substantial. However, as long as demand
is sufciently sensitive to car quality (e.g., brand
name, in-vehicle features, safety features, aesthetics,
gas mileage, etc.), condition (C2) will be satised.
We provide some intuition for condition (C1)
|] > (c |c)
2
,4|] by considering two extreme cases:
(1) c =0 and (2) the boundary case c =c,|. In Case 1
(c = 0), condition (C1) becomes ] > c
2
,4|, imply-
ing that the rm incurs substantial marginal costs in
its product design and development processes. For
example, Microsoft, which recently spent over two
billion dollars and two years redesigning its Ofce
Suite (Green 2003), incurs signicant R&D costs (e.g.,
iterative analysis, prototyping, and evaluation pro-
cesses) to design and develop its software appli-
cations. Similarly, MLB.com incurs signicant R&D
costs (e.g., program preparation, script development,
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
274 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
Table 1 Equilibrium Values and Derivatives (Basic Model)
Equilibrium value
c(Value)
cl
c(Value)
ca
0
r
=
a(c ta)
4tl (c ta)
2
c0
r
cl
=
4at(c ta)
(4tl (c ta)
2
)
2
-0
c0
r
ca
=
at(4tl +(c ta)
2
)
(4tl (c ta)
2
)
2
-0
F
r
=
a(2l +a(c ta))
4tl (c ta)
2
cF
r
cl
=
2a(c ta)(c +ta)
(4tl (c ta)
2
)
2
-0
cF
r
ca
=
a(4t
2
al +c(c ta)
2
)
(4tl (c ta)
2
)
2
-0
0
r
=
2atl
4tl (c ta)
2
c0
r
cl
=
2at(c ta)
2
(4tl (c ta)
2
)
2
-0
c0
r
ca
=
4at
2
l (c ta)
(4tl (c ta)
2
)
2
-0
l
r
=
2a
2
tl (2l +a(c ta))
(4tl (c ta)
2
)
2
cl
r
cl
=
2a
2
t(4cl (c ta) +a(c ta)
3
)
(4tl (c ta)
2
)
3
-0
cl
r
ca
=
2a
2
tl (4tcl +(c +2ta)(c ta)
2
)
(4tl (c ta)
2
)
3
-0
0
r
=
a
2
l (c ta)(c +ta)
(4tl (c ta)
2
)
2
c0
r
cl
=
a
2
(c ta)(4tl (c +ta) +(c ta)
2
(c +ta))
(4tl (c ta)
2
)
3
-0
c0
r
ca
=
2a
2
tl (4t
2
al +(2c +ta)(c ta)
2
)
(4tl (c ta)
2
)
3
-0
t
r
=
a
2
l
4tl (c ta)
2
Equation (11) Equation (12)
o
r
=
2a
2
tl
2
(4tl (c ta)
2
)
2
Equation (13) Equation (14)
p
r
output
=
2t(4tl (c ta)
2
)
a(c ta)(c +ta)
Equation (15) Equation (16)
p
r
rm
=
2t(2l +a(c ta))
(c ta)(c +ta)
Equation (16) Equation (17)
p
r
social
=
2t(3l +a(c ta))
(c ta)(c +ta)
Equation (17) Equation (18)
and content collection, verication, and ltering pro-
cesses) to produce and develop its baseball broad-
casts and highlight programs. In Case 2, (c = c,|),
condition (C1) becomes ] > 0. In other words, as
we consider rms with more substantial marginal
costs to manufacture and distribute a given product,
condition (C1) becomes less restrictive. Firms with
higher values of c are likely to meet condition (C2). In
summary, conditions (C1) and (C2) are consistent with
the structure underlying the relationships between the
cost and demand parameters in a range of contexts.
4.1. Impact of IT Investments on Firm Prots and
Consumer Surplus
The impacts on rm prots of a change in technology
parameters ] and c, respectively, are
or
m
o]
=
u
2
(c |c)
2
|4|] (c |c)
2
]
2
-0 and (11)
or
m
oc
=
2u
2
|] (c |c)
|4|] (c |c)
2
]
2
-0. (12)
In addition, the impacts on consumer surplus are
ow
m
o]
=
4u
2
|] (c |c)
2
|4|] (c |c)
2
]
3
-0 and (13)
ow
m
oc
=
8u
2
|
2
]
2
(c |c)
|4|] (c |c)
2
]
3
-0. (14)
The partial derivatives are negative, implying that
improvements (i.e., decreases) in either technology
parameter will lead to an increase in both rm prots
and consumer surplus. The mechanism driving these
results is as follows (see Table 1 for derivations). An
IT investment that reduces ] or c affects the qual-
ity and pricing decisions made by the rm in simi-
lar ways. Specically, such investments will encour-
age the prot-maximizing rm to improve product
quality (Table 1, Row 1). Because demand is sensi-
tive to quality, the rm charges a higher price (Table
1, Row 2) for the improved product and still real-
izes an increase in demand (Table 1, Row 3). Because
more products are sold at a higher price, revenue
for the rm increases (Table 1, Row 4). In addition,
the improvement in product quality and the result-
ing increase in demand together lead to an increase
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 275
in production costs (Table 1, Row 5) despite the
improvement in production efciency realized from
the IT investment. The rm realizes an increase in
prots because the increase in revenue is greater
than the corresponding increase in costs. Finally,
consumers realize an increase in consumer surplus
because the additional value passed on to consumers
from the improvement in product quality outweighs
the higher price charged by the rm.
Proposition 1. Investments in IT infrastructure will
increase rm prots and consumer surplus.
4.2. Impact of IT Investments on Productivity
The impact of a change in technology parameter ] on
productivity is
oj
m
output
o]
=
8|
2
u(c |c)(c +|c)
>0, (15)
oj
m
rm
o]
=
4|
(c |c)(c +|c)
>0, and (16)
oj
m
social
o]
=
6|
(c |c)(c +|c)
>0. (17)
In addition, the impact of a change in technology
parameter c on productivity is
oj
m
output
oc
=
4|
2
(4|
2
c] +c(c |c)
2
)
u(c |c)
2
(c +|c)
2
>0, (18)
oj
m
rm
oc
=
2|(4|
2
c] +c(c |c)
2
)
(c |c)
2
(c +|c)
2
>0, and (19)
oj
m
social
oc
=
2|(c(c |c)
2
+6|
2
c] )
(c +|c)
2
(c |c)
2
>0. (20)
The partial derivatives are positive, implying that
an improvement in either technology parameter will
lead to a decrease in all three productivity measures.
As we saw earlier, a prot-maximizing rm will
respond to an improvement in IT infrastructure by
improving product quality and increasing price in
such a way as to increase output value (as measured
by product demand, rm revenues, and consumer
surplus) and input value (as measured by produc-
tion costs). However, production costs increase by a
larger percentage than demand, revenues, or consumer
surplus, resulting in a decrease in all three productivity
measures.
Proposition 2. Investments in IT infrastructure will
decrease productivity.
In summary, we have shown that IT investments
that reduce the marginal cost of quality (or improve
production efciency) result in higher production
costs and lower productivity, even when account-
ing for the output value passed on to consumers
(Proposition 2). However, prot-maximizing man-
agers should not be concerned because these same
investments improve prots and consumer surplus
(Proposition 1).
5. Fixed-Quality Model Results
In this section we consider a variation of the basic
monopoly model to identify conditions under which
IT investments may lead to productivity gains. Specif-
ically, we consider a single-product monopolist where
Equations (1)(8) continue to hold. However, we now
assume that the product quality offered by the rm
is xed. That is, the monopolist chooses the price
(P
m
q
) that maximizes its prots, holding product qual-
ity constant (i.e., Q = q). As we will see, this model
extension helps to clarify the way product quality
moderates the seemingly counterintuitive relation-
ships between IT and productivity presented in the
basic model.
The assumption that the rm maintains a xed
product quality following investments in IT infra-
structure is neither uncommon nor unreasonable.
Consider technology investments that automate a
rms manufacturing processes. In this case, a rm
may choose to automate existing production prac-
tices without redesigning the products manufactured;
this decision may be due to stafng constraints, for
example. Alternatively, a rm may choose to signif-
icantly redesign the internal structure of a product
while keeping the external structure and functionality
the same. For example, a division of Hewlett-Packard
redesigned a signal generatorone of their electronic
test equipment productsin an effort to lower manu-
facturing costs. There may be valid reasons for such a
strategyfor example, to minimize complications in
the product line.
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
276 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
Table 2 Equilibrium Values and Derivatives (Fixed-Quality Model)
Equilibrium value
c(Value)
cl
c(Value)
ca
F
r

=
a +(c +ta)
2t
cF
r

cl
=0
cF
r

ca
=

2
>0
0
r

=
a +(c ta)
2
c0
r

cl
=0
c0
r

ca
=
t
2
-0
l
r

=
(a +(c ta))(a +(c +ta))
4t
cl
r

cl
=0
cl
r

ca
=
ta
2
2
-0
0
r

=l
2
+
a(a +(c ta))
2
c0
r

cl
=
2
>0
c0
r

ca
=
(a +(c 2ta))
2
t
r

=
(a +(c ta))
2
4t
l
2
Equation (22) Equation (24)
o
r

=
(a +(c ta))
2
8t
Equation (23) Equation (25)
p
r
output,
=
a +(c ta)
(2l +a(a +(c ta)))
Equation (26) Equation (29)
p
r
rm,
=
(a +(c +ta))(a +(c ta))
2t(2l +a(a +(c ta)))
Equation (27) Equation (30)
p
r
social,
=
(a +(c ta))(3a +(3c +ta))
4t(2l +a(a +(c ta)))
Equation (28) Equation (31)
Given Equations (1)(4), the rms equilibrium
price given xed product quality is (see tables for
derivations)
P
m
q
=
u +q(c +|c)
2|
. (21)
Substituting Equation (21) into Equations (1)(8)
yields the equilibrium values presented in Table 2. To
guarantee positive demand in equilibrium, the follow-
ing condition must hold:
c -
c
|
+
u
|q
. (C3)
5.1. Impact of IT Investments on Firm Prots
and Consumer Surplus
The impacts on prots and consumer surplus of a
change in technology parameter ] are
or
m
q
o]
=q
2
-0 and (22)
ow
m
q
o]
=0. (23)
These partial derivatives imply that an improve-
ment in technology parameter ] leads to higher rm
prots without affecting consumer welfare. The mech-
anism driving these results is as follows (see Table 2
for derivations). Because product quality is xed, the
term (fq
2
) in the cost function is a constant. There-
fore, the rm continues to charge the same price for
the product despite the improvement in production
efciency, leaving demand, revenue, and consumer
surplus unchanged (Table 2, Rows 13). However,
because an improvement in ] leads to lower produc-
tion costs (Table 2, Row 4), the rm realizes higher
prots. Of course, the improvement in prots is sub-
optimal when compared to the basic model, in which
the rm is able to adjust both price and quality in
response to IT investments.
The impacts on prots and consumer surplus of a
change in technology parameter c are
or
m
q
oc
=
q(u +q(c |c))
2
-0 and (24)
ow
m
q
oc
=
q(u +q(c |c))
4
-0. (25)
Both partial derivatives are negative, implying that
an improvement in technology parameter c leads to
higher rm prots and higher consumer surplus.
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 277
Because product quality is xed, the rms prot-
maximizing response is to lower its price (Table 2,
Row 1). This adjustment will increase demand
(Table 2, Row 2) in such a way as to increase rm
revenue (Table 2, Row 3) despite the lower price.
Furthermore, because more customers buy the prod-
uct at a lower price, consumer surplus increases.
The impact of this increase in demand on produc-
tion costs depends upon the parameterization of the
model (Table 2, Row 4). However, prots improve no
matter what the impact on production costs.
Proposition 3. Given xed product quality, invest-
ments in IT infrastructure will increase rm prots.
Proposition 4. Given xed product quality, invest-
ments in IT infrastructure that lower the xed cost of
quality parameter (] ) will not affect consumer surplus;
however, investments that lower the variable cost param-
eter (c) will increase consumer surplus.
5.2. Impact of IT Investments on Productivity
Given xed quality, the impact on productivity of a
change in the technology parameter ] is
oj
m
output,q
oj
=
2(u+q(c|c))
(2]q+c(u+q(c|c)))
2
-0, (26)
oj
m
rm,q
o]
=
(u+q(c|c))(u+q(c+|c))
|(2]q+c(u+q(c|c)))
2
-0, and (27)
oj
m
social,q
o]
=
(u+q(c|c))(3u+q(3c+|c))
2|(2]q+c(u+q(c|c)))
2
-0. (28)
The partial derivatives are negative, implying that
an improvement in technology parameter ] will
improve all three productivity measures. As we saw
earlier, given constant product quality, a decrease in
] leads to lower production costs, but does not affect
demand, revenue, or consumer surplus; this results in
an improvement in all three productivity measures.
Given xed quality, the impact on productivity of a
change in technology parameter c is
oj
m
output,q
oc
=
(2|]q
2
+(u+q(c|c))
2
)
q(2]q+c(u+q(c|c)))
2
-0, (29)
oj
m
rm,q
oc
=
|4|
2
c]q
3
+(u+cq)(u+q(c|c))
2
]
2|q|2]q+c(u+q(c|c))]
2
- 0, and (30)
oj
m
social,q
oc
=
4|]q
2
(u+q(c+|c))+3(u+cq)(u+q(c|c))
2
4|q(2]q+c(u+q(c|c)))
2
-0. (31)
Again, the partial derivatives are negative, imply-
ing that an improvement in technology parameter c
will also improve all three productivity measures. As
we saw earlier, a decrease in c leads to an increase
in demand, rm revenue, and consumer surplus. Its
impact on direct production costs depends on the
model parameterization. However, even if production
costs increase, they increase by a smaller percentage
than each output value, leading to an increase in all
three productivity measures.
Proposition 5. Given xed product quality, invest-
ments in IT infrastructure will increase productivity.
In summary, we have shown that IT investments
lead to an increase in prots and consumer surplus
even when holding product quality constant. How-
ever, unlike our ndings in the basic monopoly
model, these investments lead unambiguously to
an increase in productivity. Together, the basic
model and the xed-quality model formalize the
way in which product quality moderates the complex
relationship between IT investments and economic
performance.
6. Welfare Maximization
Model Results
In this section we consider a benevolent social planner
that chooses product price and quality to maximize
social welfare (W), dened as the sum of consumer
surplus and rm prots. As we will see, this exten-
sion demonstrates that the directional impacts of IT
investments on economic performance found in the
basic model are not limited to the monopoly setting,
but are robust to a socially optimal setting.
Given Equations (1)(5), the social welfare maxi-
mization problem becomes
Max
P, Q
W =
(u |cQ+cQ)
2
2|
+(P lQ)(u |cQ+cQ) ]Q
2
. (32)
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
278 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
Table 3 Equilibrium Values and Derivatives (Welfare Maximization Model)
Equilibrium value
c(Value)
cl
c(Value)
ca
0
l
=
a(c ta)
2tl (c ta)
2
c0
l
cl
=
2at(c ta)
(2tl (c ta)
2
)
2
-0
c0
l
ca
=
at(2tl +(c ta)
2
)
(2tl (c ta)
2
)
2
-0
F
l
=
aa(c ta)
2tl (c ta)
2
cF
l
cl
=
2ata(c ta)
(2tl (c ta)
2
)
2
-0
cF
l
ca
=
a(c(c ta)
2
2tl (c 2ta))
(2tl (c ta)
2
)
2
0
l
=
2atl
2tl (c ta)
2
c0
l
cl
=
2at(c ta)
2
(2tl (c ta)
2
)
2
-0
c0
l
ca
=
4at
2
l (c ta)
(2tl (c ta)
2
)
2
-0
l
l
=
2a
2
tal (c ta)
(2tl (c ta)
2
)
2
cl
l
cl
=
2a
2
ta(c ta)(2tl +(c ta)
2
)
(2tl (c ta)
2
)
3
-0
cl
l
ca
=
2a
2
tl ((c +2ta)(c ta)
2
+2tl (2ta c))
(2tl (c ta)
2
)
3
0
l
=
a
2
l (c ta)(c +ta)
(2tl (c ta)
2
)
2
c0
l
cl
=
a
2
(c ta)((c ta)
2
(c +ta) +2tl (c +ta))
(2tl (c ta)
2
)
3
-0
c0
l
ca
=
2a
2
tl (2t
2
al +(2c +ta)(c ta)
2
)
(2tl (c ta)
2
)
3
-0
o
l
=
2a
2
tl
2
(2tl (c ta)
2
)
2
Equation (40) Equation (41)
p
l
output
=
2t(2tl (c ta)
2
)
a(c ta)(c +ta)
Equation (42) Equation (45)
p
l
rm
=
2ta
(c +ta)
Equation (43) Equation (46)
p
l
social
=
2t(l +a(c ta))
(c ta)(c +ta)
Equation (44) Equation (47)
The equilibrium choices of quality and price are
(see tables for derivations)
Q
|
=
u(c |c)
2|] (c |c)
2
and (33)
P
|
=
uc(c |c)
2|] (c |c)
2
. (34)
Note that to maximize social welfare, the planner sets
price equal to marginal costs. Substituting Equations
(33) and (34) yields the equilibrium values presented
in Table 3. To guarantee a unique maximum and a
positive level of quality in equilibrium, the following
conditions must hold:
] >
(c |c)
2
2|
and (C4)
c -
c
|
. (C5)
Condition (C4) ensures that the determinant of the
Hessian is positive, which guarantees the concavity of
the welfare function in price and in quality and guar-
antees a unique maximum. Condition (C5) ensures
that the rm chooses a positive level of quality and
charges a positive price in equilibrium.
Not surprisingly, we observe that the quality level
set by the social planner is higher than that set by the
prot-maximizing monopolist,
Q
|
=
u(c |c)
2|] (c |c)
2
>
u(c |c)
4|] (c |c)
2
=Q
A
. (35)
In addition, the consumer surplus realized under
the social planner is higher than that realized under
the monopolist,
w
|
=
2u
2
|]
2
(2|] (c |c)
2
)
2
>
2u
2
|]
2
(4|] (c |c)
2
)
2
=w
A
. (36)
However, it is interesting to note that the produc-
tivity levels realized in equilibrium under the social
planner are lower than those realized under the
monopolist,
j
|
output
=
2|(2|] (c |c)
2
)
u(c |c)(c +|c)
-
2|(4|] (c |c)
2
)
u(c |c)(c +|c)
= j
A
output
, (37)
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 279
j
|
rm
=
2|c
(c +|c)
-
2|(2] +c(c |c))
(c |c)(c +|c)
= j
A
rm
, and (38)
j
|
social
=
2|(] +c(c |c))
(c |c)(c +|c)
-
2|(3] +c(c |c))
(c |c)(c +|c)
= j
A
social
. (39)
Proposition 6. The levels of product quality and con-
sumer surplus realized under the social planner are
higher than those realized under the prot-maximizing
monopolist.
Proposition 7. The level of productivity realized
under the social planner is lower than that realized under
the monopolist.
Together, Propositions 6 and 7 further demonstrate
the adverse impact of product quality on productivity
shown in the basic model.
6.1. Impact of IT Investments on
Consumer Surplus
The impacts on consumer surplus of a change in tech-
nology parameters ] and c, respectively, are
ow
|
o]
=
4u
2
|] (c |c)
2
|2|] (c |c)
2
]
3
-0 and (40)
ow
|
oc
=
8u
2
|
2
]
2
(c |c)
|2|] (c |c)
2
]
3
-0. (41)
The partial derivatives are positive, implying that,
consistent with results from the monopoly model,
improvements in IT infrastructure will lead to an
increase in consumer surplus.
Proposition 8. Under a social planner, investments in
IT infrastructure will increase consumer surplus.
6.2. Impact of IT Investments on Productivity
The impacts on productivity of a change in technol-
ogy parameter ] are
oj
|
output
o]
=
4|
2
u(c |c)(c +|c)
>0, (42)
oj
|
rm
o]
=0, and (43)
oj
|
social
o]
=
2|
(c |c)(c +|c)
>0. (44)
Equations (42) and (44) are positive, implying that,
consistent with results from the basic model, improve-
ments in ] will lead to a decrease in output produc-
tivity and social productivity. However, we note that
in this setting an improvement in ] does not affect
rm productivity. That is, the optimal improvement
in product quality (Table 3, Row 1) and increase in
price (Table 3, Row 2) in response to an improvement
in ] together lead to proportional increases in rev-
enues and costs (Table 3, Rows 45, 8), meaning that
the ratio of the two values does not change.
In addition, the impacts on productivity of a change
in technology parameter c are
oj
|
output
oc
=
4|
2
(2|
2
c] +c(c |c)
2
)
u(c |c)
2
(c +|c)
2
>0, (45)
oj
|
rm
oc
=
2|c
(c +|c)
>0, and (46)
oj
|
social
oc
=
2|(2|
2
c] +c(c |c)
2
)
(c |c)
2
(c +|c)
2
>0. (47)
The partial derivatives are positive, implying that,
consistent with results from the basic model, improve-
ments in c will lead to a decrease in productivity.
Proposition 9. Under a social planner, investments
in IT infrastructure that lower the variable cost param-
eter (c) will decrease all three productivity measures.
Investments in IT infrastructure that lower the xed cost
of quality parameter (] ) will decrease output produc-
tivity and social productivity but will not affect rm
productivity.
In summary, these results show that the mecha-
nisms underlying the relationship between product
quality and productivity in a monopoly market hold
when considering a social planner that chooses price
and quality to maximize social welfare.
7. Free Product/Service Model Results
As a nal model extension it may be useful to link our
work to that of Barua et al. (1991), who analyze the
strategic impact of IT investments where the services
offered by a rm are not priced initially (e.g., in the
nancial services sector) and where the benets may
come indirectly (e.g., in the form of interest earned on
consumer deposits). Barua et al. presented ATM ser-
vices (banking) and claims settlement services (insur-
ance) as services that t this product category. We
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
280 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
present examples more related to our problem context
below.
In telecommunications, content services (e.g., music
videos, radio programs, movie trailers, video-game
previews, stock quotes) are provided free by many
online providers. Walt Disneys ESPN.com site pro-
vides free content such as sports news, highlights and
interviews, hundreds of hours of commentary from
ESPNRadio.com, and highlights of ESPNs X Games.
While ESPN.com does not charge for access to its con-
tent, it does charge fees to advertisers for commercial
space on the website (Grant 2003). These advertis-
ing fees are indirect benets of the free content. As
another example, Movieix.com offers 1,600 movies
and programs to visitors for free. This free con-
tent lures visitors to the site, allowing the company
to advertise Movieix Plus, a $5.95/month ser-
vice that gives subscribers access to 1,400 other titles
(Grant 2003). The subscription fees earned as some
customers move from free-service to pay-service
are indirect benets to Movieix.com. In hotels,
value-added amenities and services such as high-
speed Internet access, in-room PCs, cordless in-room
phones, and check-in kiosks in the lobby increasingly
are offered to guests free of charge. These free-to-
guest services often provide indirect benets such as
higher occupancy rates and increased guest loyalty.
For example, Wingate Inns found that their free-to-
guest services led to an increase in incremental room
nights, new guests, and repeat guests (Whitford 2000).
Based on the model setup from Barua et al. (1991),
we develop a model to examine the value of IT invest-
ments that support the design and development of
free products/services that provide indirect benets
to the rm. More formally, we consider a single-
product monopolist where Equations (2), (3), and
(6)(8) continue to hold. Because in this context prod-
ucts are free, demand is
D
P=0
=u +cQ. (48)
In addition, because the rm only earns indirect
benets from its product offering, revenue is

P=0
=rD, (49)
where r = the benet (or revenue earned) per cus-
tomer. That is, r represents the indirect benet of
capturing a bigger demand, leading to higher rev-
enues for the rm. In the telecommunications exam-
ples, r may be the average advertising revenue or
subscription revenue per visitor. Because the product
or service is offered free of charge, r is treated as an
exogenous parameter in this model.
Finally, because products are free, consumer sur-
plus is
w
P=0
=QD. (50)
In this model, the monopolist chooses the qual-
ity (Q
m
P=0
) that maximizes its prots. Given Equa-
tions (2)(3) and (48)(50), the rms equilibrium
choice of quality is (see tables for derivation)
Q
m
P=0
=
(cr uc)
2(] +cc)
. (51)
To ensure positive quality in equilibrium the follow-
ing condition must hold:
c -
cr
u
. (C6)
Substituting Equation (51) yields the equilibrium val-
ues presented in Table 4.
7.1. Impact of IT Investments on Firm Prots
and Consumer Surplus
The impacts of changes in technology parameter ]
and c on prots and consumer surplus are
or
m
P=0
o]
=
(cr uc)
2
4(] +cc)
2
-0 and (52)
ow
m
P=0
o]
=
(cr uc)(u] +c
2
r)
2(] +cc)
3
-0, (53)
or
m
P=0
oc
=
(2u] (cr uc)+c(cr uc)(cr +uc))
4(] +cc)
2
- 0 and (54)
ow
m
P=0
oc
=
(u
2
]
2
+2uc
2
]p+c
4
p
2
)
2(] +cc)
3
-0. (55)
The partial derivatives are negative, implying that
improvements in either technology parameter lead to
higher rm prots and consumer surplus. The mech-
anism driving this result is as follows (see Table 4 for
derivations). IT investments that reduce ] or c encour-
age the rm to improve product quality (Table 4,
Row 1), leading to higher demand and revenues
(Table 4, Rows 2 and 3). The improvement in product
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 281
Table 4 Equilibrium Values and Derivatives (Free Products/Services Model)
Equilibrium value
c(Value)
cl
c(Value)
ca
0
r
F =0
=
(cr aa)
2(l +ca)
c0
r
F =0
cl
=
(cr aa)
2(l +ca)
2
-0
c0
r
F =0
ca
=
(c
2
r +al )
2(l +ca)
2
-0
0
r
F =0
=
2al +c(cr +aa)
2(l +ca)
c0
r
F =0
cl
=
c(cr aa)
2(l +ca)
2
-0
c0
r
F =0
ca
=
c(c
2
r +al )
2(l +ca)
2
-0
l
r
F =0
=
r (2al +c(cr +aa))
2(l +ca)
cl
r
F =0
cl
=
cr (cr aa)
2(l +ca)
2
-0
cl
r
F =0
ca
=
cr (c
2
r +al )
2(l +ca)
2
-0
0
r
F =0
=
(cr aa)(cr +aa)
l +ca
c0
r
F =0
cl
=
(cr aa)(cr +aa)
4(l +ca)
2
c0
r
F =0
ca
=
(2a
2
al +c(c
2
r
2
+a
2
a
2
))
4(l +ca)
2
-0
t
r
F =0
=
4alr +(cr +aa)
2
4(l +ca)
Equation (52) Equation (54)
o
r
F =0
=
(cr aa)(2al +c(cr +aa))
4(l +ca)
2
Equation (53) Equation (55)
p
r
output, F =0
=
2(2al +c(cr +aa))
(cr +aa)(cr aa)
Equation (56) Equation (57)
p
r
rm, F =0
=
2r (2al +c(cr +aa))
(cr +aa)(cr aa)
Equation (58) Equation (59)
p
r
social, F =0
=
(2al +c(cr +aa))(2r (l +ca) +(cr aa))
(cr aa)(l +ca)(cr +aa)
Equation (60) Equation (61)
quality and the resulting increase in demand together
lead to an increase in production costs (Table 4, Row
4). The rm realizes an increase in prots because the
increase in revenue (via more ad revenue or more sub-
scriptions to related pay-services) generated by the
growth in demand for the free product is greater than
the corresponding increase in costs. Finally, consumer
surplus is higher because consumers are receiving
better-quality products for free.
Proposition 10. Given a category of products that are
offered free of charge but that provide indirect benets,
investments in IT infrastructure will increase rm prots
and consumer surplus.
7.2. Impact of IT Investments on Productivity
The impacts of a change in ] and c on output pro-
ductivity and rm productivity are
oj
m
output, P=0
o]
=
4u
(cr uc)(cr +uc)
>0 and (56)
oj
m
output, P=0
oc
=
2u(4u
2
c] +c(cr +uc)
2
)
(cr uc)
2
(cr +uc)
2
>0, (57)
oj
m
rm, P=0
o]
=
4ur
(cr uc)(cr +uc)
>0 and (58)
oj
m
rm, P=0
oc
=
2ur(4u
2
c] +c(cr +uc)
2
)
(cr uc)
2
(cr +uc)
2
>0. (59)
The partial derivatives are positive, implying that,
consistent with results from the basic model, improve-
ments in IT infrastructure will lead to a decrease in out-
put productivity and rm productivity. However, the
relationships between the technology parameters and
social productivity are more complex. Specically, the
impacts of a change in ] and c on social productivity
are
oj
m
social,P=0
o]
=
oj
rm,P=0
o]
+
o(w
P=0
,C
P=0
)
o]
=
4ur
(cr uc)(cr +uc)
+
c(cr uc)
(] +cc)
2
(cr +uc)
and (60)
oj
m
social,P=0
oc
=
oj
rm,P=0
oc
+
o(w
P=0
,C
P=0
)
oc
=
2ur(4u
2
c] +c(cr +uc)
2
)
(cr uc)
2
(cr +uc)
2
+
(c
2
(cr +uc)
2
+2u] (u] +c(cr +2uc)))
(] +cc)
2
(cr +uc)
2
.
(61)
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
282 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
While we are unable to sign either partial deriva-
tive, we are able to derive specic insights into the
relationships between the IT parameters and social
productivity.
We rst consider Equation (60). The derivative of
social productivity with respect to ] is positive (i.e.,
oj
m
social, P=0
,o] >0) when the following condition holds:
] >

ucr(cr uc)
2
2uccr
2ur
.
(C7)
That is, if ] is sufciently large, then an improve-
ment in ] will lead to a decrease in social produc-
tivity. This condition is akin to condition (C1) in the
basic monopoly model (4) and to condition (C4) in
the welfare maximization model (6), both of which
restrict ] in similar ways to guarantee a unique max-
imum in those models.
Finally, we consider Equation (61). Because of the
complexity of the relationships among the param-
eters, it is difcult to tease out a simple mathematical
condition in this special case. However, we note that
the sign of Equation (61) also depends on the size
of ] . To provide some intuition, we consider the limit
of Equation (61) as ] becomes large,
lim
]
oj
m
social, P=0
o]
= lim
]
2ur(4u
2
c] +c(cr +uc)
2
)
(cr uc)
2
(cr +uc)
2
+ lim
]
(c
2
(cr +uc)
2
+2u] (u] +c(cr +2uc)))
(] +cc)
2
(cr +uc)
2
,
(62)
which gives
lim
]
oj
m
social, P=0
o]
=+
2u
2
(cr uc)
2
>0. (63)
Therefore, if ] is sufciently large, the derivative
of social productivity with respect to c will be posi-
tive, implying that an improvement in c will lead to
a decrease in social productivity.
Proposition 11. Given a category of products that are
offered free of charge, but that provide indirect benets,
investments in IT infrastructure will:
(a) decrease output productivity and rm productivity
and
(b) decrease social productivity if the xed cost of
quality parameter ] is sufciently large.
In summary, when considering free products and
services that provide indirect benets to the rm, the
model shows that IT investments increase prots and
consumer surplus, but decrease output productivity
and rm productivity. In addition, these same invest-
ments decrease social productivity if the xed cost of
quality parameter ] is sufciently large. These results
are consistent with those found in the basic monopoly
model.
8. Model Summary and Interpretation
of Past Empirical Studies
The basic model shows that a monopolist maxi-
mizes prots by using IT investments to design a
better-quality product and charge a higher price.
While this prot-maximizing adjustment increases
consumer surplus, it also increases production costs
in a way that adversely affects productivity. The xed-
quality model shows that when a rm is unwilling
to improve quality, the rm will realize suboptimal
improvements in prots, an increase in consumer
surplus, and an increase in productivity. Together,
these models highlight the way in which product
quality moderates the relationship between IT invest-
ment and economic value. The welfare maximization
model shows that these results are robust to the case
in which a social planner chooses price and qual-
ity to maximize social welfare. Finally, the free prod-
uct/service model shows that these results hold when
considering a monopolist that designs products that
are offered free of charge, but that provide indirect
benets to the rm. Below, we use the basic model
and the xed-quality model to show that the mixed
empirical ndings presented in 2 are consistent with
economic theory as embodied in our model.
8.1. Basic Model
Many ndings from IT application-level studies are
consistent with the basic model. Several IT applica-
tions have been associated with productivity losses
for example, software infrastructure (Rai 1997), ATMs
(Banker and Kauffman 1991), strategic systems, infor-
mational systems, and CASE tools (Weill 1990), guest-
operated hotel services (David et al. 1996), and DSS
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 283
in hospitals (Devaraj and Kohli 2000). Given our
model, we expect investments in these IT applications
to reduce the marginal cost of quality and lead to
improvements in product quality. In turn, our model
shows that such improvements are expected to reduce
measures of productivity. In fact, consistent with our
model, these studies have linked the IT applications
to improvements in intermediate performance mea-
sures such as product quality and output levels.
At the rm level, empirical results related to the
impact of IT on prots appear consistent with the
model. For example, Dos Santos et al. (1993) found
that innovative IT investments increased rm value,
while noninnovative (incremental, follow-up) invest-
ments did not. In terms of our model, we may expect
that an innovative IT would lower the marginal cost
of improving quality substantially more than the
noninnovative IT, enabling larger improvements in
product quality. Therefore, given the basic model,
we would expect more substantial prot gains from
investments in innovative IT that from investments in
noninnovative IT. We further note that investments in
noninnovative IT are more consistent with the xed-
quality model. Shin (2001) found that IT investments
improve rm prots when IT is properly aligned with
rm strategy. In our model, proper alignment with
rm strategy is represented by the optimal adjust-
ments to price and quality in the basic model. Alter-
natively, suboptimal alignment may be represented by
the xed-quality model. Therefore, Shins ndings are
consistent with our economic model.
8.2. Fixed-Quality Model
Other ndings at the IT application level are consis-
tent with the xed-quality model. For example, some
studies have found signicant productivity gains
from investments in back-ofce data-processing sys-
tems (David et al. 1996, Rai 1997). These tools are
generally used by rms to reduce costs rather than
to improve quality. Therefore, given the xed-quality
model, we expect productivity gains from such IT
investments.
Many rm-level studies have found a positive rela-
tionship between IT investments and productivity
(Barua and Lee 1997; Brynjolfsson and Hitt 1996;
Brynjolfsson 1993; Lee and Barua 1999; Jorgenson
and Stiroh 1995; Lehr and Lichtenberg 1998, 1999;
Lichtenberg 1995). This result may be explained in
terms of the xed-quality model. First, these studies
generally consider the impact of aggregate IT spend-
ing, which may include investments that rms use
for cost reduction (as opposed to quality improve-
ment). In addition, when considering a large num-
ber of rms, only a small subset of rms will be
able to adjust their internal resources and IT capa-
bility to design and develop a substantially better
product quality. For reasons related to implementa-
tion problems or short-run production constraints, we
might expect that many rms will be unable to see
or take advantage of opportunities to improve prod-
uct quality (and may instead focus on cost reduc-
tions). Therefore, it may be reasonable to expect that,
on average, the rms included in these studies will
underinvest in product quality following IT invest-
ments. As shown in the xed-quality model, we may
expect to see signicant productivity gains (combined
with suboptimal prots) from IT investments in these
studies. Finally, Hitt and Brynjolfsson (1996), using
data from over 370 rms, found that IT investments
led to improvements in productivity and consumer
surplus, but did not lead to supranormal prof-
its. These ndings appear to be consistent with the
discussion present above. That is, when considering
such a large sample of rms, we may expect IT invest-
ments to increase both productivity and consumer
surplus, but not necessarily result in supranormal
prots.
9. Limitations and Conclusions
We have presented an integrative economic model
that helps to formalize the complex relationships
among IT investments, intermediate performance
measures, and economic performance measures. The
key nding is that while rms may maximize prots
and improve consumer value by using investments
in IT infrastructure to improve product quality, they
may do so at the expense of productivity.
The economic model presented in this paper has
provided several important insights. However, similar
to the empirical work, our model has several limita-
tions. First, the model conditions (e.g., (C1) and (C2)),
which place restrictions on the technology parame-
ters (] and c), characterize the types of rms and
markets to which the different models may apply.
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
284 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
Going forward, we must empirically determine if
these conditions are consistent with the real world.
Second, we note that the cost function used in our
model is only one of many reasonable possibilities.
Future work may examine the robustness of these
results across alternative cost structures. Third, in our
model we assume a single-product, prot-maximizing
monopolist. While some recent work has begun to
examine IT value in a symmetric duopoly setting
(Quan et al. 2003, Thatcher and Pingry 2004), future
work should examine IT value in an asymmetric
duopoly setting where rms differ in their technol-
ogy infrastructure, product characteristics, etc. This
analysis will increase the applicability of the model
to real-world settings and provide IT managers with
more specic insights into the economic impact of IT
investments. Fourthly, future research should focus
on empirically testing and validating the propositions
derived from the economic model presented in this
paper.
The critical insight of the model is that invest-
ments in IT infrastructure reduce the marginal cost of
improving product quality. In other words, IT is an
input that enables the rm to seek higher quality of its
output. Understanding this is critical to interpreting
the empirical literature. In addition, and perhaps
more importantly, it is critical that IT managers not
view IT as simply a way to reduce total product costs,
because IT may increase costs and reduce productiv-
ity when it is properly aligned with business strategy
to improve product quality.
Acknowledgments
The authors would like to thank the editors and two anony-
mous reviewers for their many constructive suggestions.
Helpful comments on earlier drafts of this work were
received from James Marsden and Jim Oliver.
Appendix
Basic Model
Given Equations (1)(4), the rst-order condition (F.O.C.) of
the prot function with respect to price is
or
oP
=u 2|p +cQ+|cQ. (A1)
Setting Equation (A1) equal to zero and solving for Q yields
Q=
u +2|P
c +|c
. (A2)
Solving for P yields
P =
u +cQ+|cQ
2|
. (A3)
The F.O.C. of the prot function with respect to quality is
or
oQ
=cP 2]Quc +|cP 2ccQ. (A4)
Setting Equation (A4) equal to zero and solving for Q yields
Q=
cP uc +|cP
2(cc +] )
. (A5)
Solving for P yields
P =
2ccQ+2]Q+uc
c +|c
. (A6)
To verify the concavity of the prot function in quality and
in price we consider the second-order conditions using the
Hessian matrix. The determinant (H
d
) of the 2 2 Hessian
matrix is 4|] (c |c)
2
. To guarantee the concavity of the
prot function in quality and in price and to guarantee a
unique maximum it must be the case that H
d
> 0, leading
to condition (C1). Setting Equations (A3) and (A6) equal
and solving for equilibrium quality, Q
m
, yields Equation (9).
Similarly, setting Equations (A2) and (A5) equal and solv-
ing for equilibrium price, P
m
, yields Equation (10). Table 1
presents the equilibrium values and derivatives with respect
to the technology parameters.
Fixed-Quality Model
The derivation of Equation (21) is presented above (see
Equation (A3)). Table 2 presents the equilibrium values and
derivatives with respect to the technology parameters.
Welfare Maximization Model
Given Equations (1)(5), the F.O.C. of the social welfare
function with respect to price is
oW
oP
=|P +|cQ. (A7)
Setting Equation (A7) equal to zero and solving for Q yields
Q=
P
c
. (A8)
Solving for P yields
P =lQ. (A9)
The F.O.C. of the social welfare function with respect to
quality is
oW
oQ
=
u(c |c) Q(2|] (c |c)
2
)
|
. (A10)
Setting Equation (A10) equal to zero and solving for Q
yields
Q=
u(c |c) +|
2
cP
2|] c(c 2|c)
. (A11)
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
Information Systems Research 15(3), pp. 268286, 2004 INFORMS 285
Solving for P yields
P =
u(|c c) +Q(2|] c
2
+2|cc)
|
2
l
. (A12)
To verify the concavity of the social welfare function in qual-
ity and in price, we consider the second-order conditions
using the Hessian matrix. The determinant (H
d
) of the 2 2
Hessian matrix is 2|] (c |c)
2
. To guarantee the concavity
of the social welfare function in quality and in price and
to guarantee a unique maximum, it must be the case that
H
d
>0, leading to condition (C4). Setting Equations (A9)
and (A12) equal and solving for equilibrium quality, Q
|
,
yields Equation (33). Similarly, setting Equations (A8) and
(A11) equal and solving for equilibrium price, P
|
, yields
Equation (34). Table 3 presents the equilibrium values and
derivatives with respect to the technology parameters.
Free Product/Service Model
Given Equations (23) and (4850), the F.O.C. of the prot
function with respect to quality is
or
oQ
=(cr uc) 2Q(] +cc). (A13)
To guarantee the concavity of the prot function in quality
and a unique maximum, we verify that the second-order
condition is negative,
o
2
r
oQ
2
=2(] +cc) -0. (A14)
Setting Equation (A9) equal to zero and solving for equi-
librium quality, Q
m
P=0
, yields Equation (51). Table 4 presents
the equilibrium values and derivatives with respect to the
technology parameters.
References
Anderson, M., R. Banker, S. Ravindran. 2003. The new productivity
paradox. Association Comput. Machinery, Comm. ACM 46(3) 91
94.
Baily, M. 1986. What has happened to productivity growth? Science
234 443451.
Banker, R., R. Kauffman. 1991. Case study of electronic banking at
Meridian Bancorp. Inform. Software Tech. 33(3) 200204.
Bannan, K. 2003. Plastics rm learns CRM makes it more exible.
B to B 88(10) 13.
Barua, A., B. Lee. 1997. The information technology productivity
paradox revisited: A theoretical and empirical investigation in
the manufacturing sector. Internat. J. Flexible Manufacturing Sys-
tems 9(2) 145166.
Barua, A., C. Kriebel, T. Mukhopadhyay. 1991. An economic analy-
sis of strategic information technology investments. MIS Quart.
15(3) 312330.
Bharadwaj, A. 2000. A resource-based perspective on informa-
tion technology capability and rm performance: An empirical
investigation. MIS Quart. 24(1) 169195.
Brynjolfsson, E. 1993. The productivity paradox of information
technology. Association Comput. Machinery, Comm. ACM 36(12)
6777.
Brynjolfsson, E. 1994. Technologys true payoff. InformationWeek
(October 10) 3436.
Brynjolfsson, E. 1996. The contribution of information technology
to consumer welfare. Inform. Systems Res. 7(3) 281300.
Brynjolfsson, E., L. Hitt. 1996. Paradox lost? Firm-level evidence on
the returns to information systems spending. Management Sci.
42(4) 541558.
Cremer, H., J.-F. Thisee. 1994. Commodity taxation in a differenti-
ated oligopoly. Internat. Econom. Rev. 35(3) 613633.
David, J., S. Grabski, M. Kasavana. 1996. The productivity paradox
of hotel-industry technology. Cornell Hotel Restaurant Adminis-
tration Quart. 37(2) 6470.
Dedrick, J., V. Gurbaxani, K. Kraemer. 2003. Information technology
and economic performance: A critical review of the empirical
evidence. ACM Comput. Surveys 35(1) 128.
Devaraj, S., R. Kohli. 2000. Information technology payoff in
the health-care industry: A longitudinal study. J. Management
Inform. Systems 16(4) 4168.
Dos Santos, B., K. Peffers, D. Mauer. 1993. The impact of infor-
mation technology investment announcements on the market
value of the rm. Inform. Systems Res. 4(1) 123.
Gal-Or, E. 1987. Strategic and non-strategic differentiation. Canadian
J. Econom. 20(2) 340356.
Garvin, D. 1984. What does product quality really mean? Sloan
Management Rev. 26(1) 2543.
Grant, P. 2003. Telecommunications (a special report); You have
broadband; now what? Theres plenty of free content available;
and if you want to pay, theres even better stuff. Wall Street J.
(October 13) R8.
Green, J. 2003. A rehab for Microsofts Ofce. BusinessWeek 31(2)
106108.
Hackett, G. 1990. Investments in technologyThe service sector
sinkhole? Sloan Management Rev. 31(2) 97103.
Higgins, K. 2003. If you rebuild it, they will come. Network Comput.
14(16) 67.
Hitt, L., E. Brynjolfsson. 1996. Productivity, business protability,
and consumer surplus: Three different measures of informa-
tion technology value. MIS Quart. 20(2) 121142.
Jorgenson, D., K. Stiroh. 1995. Computers and growth. Econom.
Innovation New Tech. 3 295316.
Klassen, K., R. Russell, J. Chrisman. 1998. Efciency and productiv-
ity measures for high contact services. Services Indust. J. 18(4)
118.
Lambertini, L. 1996. Choosing roles in a duopoly for endoge-
nously differentiated products. Australian Econom. Papers 35(67)
205224.
Lambertini, L., R. Orsini. 2002. Vertically differentiated model with
a positional good. Australian Econom. Papers 41(2) 151163.
Landers, P. 2004. Drug industrys big push into technology falls
short. Wall Street J. (February 24) A1.
Lee, B., A. Barua. 1999. An integrated assessment of productivity
and efciency impacts of information technology investments:
Old data, new analysis and evidence. J. Productivity Anal. 12(1)
2143.
Thatcher and Pingry: An Economic Model of Product Quality and IT Value
286 Information Systems Research 15(3), pp. 268286, 2004 INFORMS
Lehr, B., F. Lichtenberg. 1999. Information technology and its
impact on productivity: Firm-level evidence from government
and private data sources, 19771993. Canadian J. Econom. 32(2)
335362.
Lehr, W., F. Lichtenberg. 1998. Computer use and productivity
growth in U.S. federal government agencies, 198792. J. Indust.
Econom. 46(2) 257279.
Lichtenberg, F. 1995. The output contributions of computer equip-
ment and personnel: A rm level analysis. Econom. Innovation
New Tech. 3(4) 201217.
Lilien, G., P. Kotler, K. Moorthy. 1992. Marketing Models. Prentice
Hall, Englewood Cliffs, NJ.
Loveman, G. 1991. Cash drain, no gain. Computerworld 25(47) 6971.
Lyon, D. 2001. Making it good at Buick: A talk with designer David
Lyon. Automotive Design Production 133(12) 3638.
Mandel, M. 1998. Financial services: The silent engine. BusinessWeek
(Indust./Tech. Edition) 3609 76.
Marsden, J., D. Pingry. 1993. Theory of decision supports systems
portfolio evaluation. Decision Support Systems 9 183199.
McCrune, J. 1998. The productivity paradox. Management Rev. 87(3)
3840.
McGee, M., C. Wilder. 2000. Its ofcial: IT adds up. InformationWeek
782 4251.
Metcalfe, B. 1992. Economists and the productivity paradox.
InfoWorld 14(46) 72.
Metheny, B. 1994. Relying solely on productivity to measure the
impact of info. J. Systems Management 45(3) 24.
Moorthy, K. 1988. Product and price competition in a duopoly
model. Marketing Sci. 7 141168.
Moorthy, K., I. Png. 1992. Market segmentation, cannibalization,
and the timing of product introductions. Management Sci. 38(3)
345359.
Mukhopadhyay, T., S. Rajiv, K. Srinivasan. 1997. Information tech-
nology impact on process output and quality. Management Sci.
43(12) 16451659.
Panko, R. 1991. Is ofce productivity stagnant? MIS Quart. 15(2)
191203.
Port, O. 2003. Design tools move into the fast lane: New software
turns ideas into reality in record time. BusinessWeek (June 2)
84B.
Quan, J., Q. Hu, P. Hart. 2003. IT investments and rms perfor-
manceA duopoly perspective. J. Management Inform. Systems
20(3) 121158.
Rai, A. 1997. Technology investment and business perfor-
mance. Association Comput. Machinery, Comm. ACM 40(7)
8997.
Rai, A., R. Patnayakuni, N. Patnayakuni. 1996. Refocusing where
and how IT value is realized: An empirical investigation.
Omega 24(4) 399407.
Roach, S. 1991. Services under siegeThe restructuring imperative.
Harvard Bus. Rev. 69(5) 8291.
Schmitt, P. 2002. The impact of a marginal cost increase on price
and quality: Theory and evidence from airline market strikes.
Australian Econom. Papers 41(3) 282304.
Shin, N. 2001. The impact of information technology on nancial
performance: The importance of strategic choice. Eur. J. Inform.
Systems 10(4) 227.
Spence, A. 1975. Monoploy, quality and regulation. Bell J. Econom.
6(2) 417429.
Strassmann, P. 1990. The Business Value of Computers. Information
Economics Press, New Canaan, CT.
Strassmann, P. 1997. The Squandered Computer. Information Eco-
nomics Press, New Haven, CT.
Strassmann, P. 2001. Fighting McKinsey. ComputerWorld 35(49) 27.
Thatcher, M., J. Oliver. 2001a. The impact of information technology
on quality improvement, productivity, and prots: An analyt-
ical model of a monopolist. Proc. 34th Hawaii Internat. Conf.
System Sci., (CD/ROM), January 36, 2001, Computer Society
Press, Maui, HI.
Thatcher, M., J. Oliver. 2001b. The impact of technology investments
on a rms production efciency, product quality, and produc-
tivity. J. Management Inform. Systems 18(2) 1743.
Thatcher, M., D. Pingry. 2004. Understanding the business value
of information technology investments: Theoretical evidence
from alternative market and cost structures. J. Management
Inform. Systems 21(2) 6185.
Vasilash, G. 2001. Designing with digits. Automotive Manufacturing
Production 113(4) 5051.
Waurzyniak, P. 1999. Automotive manufacturings technology evo-
lution. Manufacturing Engrg. 123(2) 5460.
Weill, P. 1990. Do Computers Pay Off? ICIT Press, Washington, D.C.
Whitford, M. 2000. Driving demand. Hotel Motel Management 215(5)
4849.

You might also like