ArticlePDF AvailableLiterature Review


Reduction of new product development cycle time and improvements in product performance have become strategic objectives for many technology-driven firms. These goals may conflict, however, and firms must explicitly consider the tradeoff between them. In this paper we introduce a multistage model of new product development process which captures this tradeoff explicitly. We show that if product improvements are additive (over stages), it is optimal to allocate maximal time to the most productive development stage. We then indicate how optimal time-to-market and its implied product performance targets vary with exogenous factors such as the size of the potential market, the presence of existing and new products, profit margins, the length of the window of opportunity, the firm's speed of product improvement, and competitor product performance. We show that some new product development metrics employed in practice, such as minimizing break-even time, can be sub-optimal if firms are striving to maximize profits. We also determine the minimal speed of product improvement required for profitably undertaking new product development, and discuss the implications of product replacement which can occur whenever firms introduce successive generations of new products. Finally, we show that an improvement in the speed of product development does not necessarily lead to an earlier time-to-market, but always leads to enhanced products.
is collaborating with JSTOR to digitize, preserve and extend access to
Management Science.
New Product Development: The Performance and Time-to-Market Tradeoff
Author(s): Morris A. Cohen, Jehoshua Eliashberg and Teck-Hua Ho
Management Science,
Vol. 42, No. 2 (Feb., 1996), pp. 173-186
Published by: INFORMS
Stable URL:
Accessed: 10-03-2015 15:35 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
Morris A. Cohen * Jehoshua Eliashberg * Teck-Hua Ho
The Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6366
Graduate School of Management,
University of California at Los
Angeles, Los Angeles, California 90024
eduction of new product development cycle time and improvements in product perfor-
mance have become strategic objectives for many technology-driven firms. These goals may
conflict, however, and firms must explicitly consider the tradeoff between them. In this paper
we introduce a multistage model of new product development process which captures this trade-
off explicitly. We show that if product improvements are additive
(over stages), it is optimal to
allocate maximal time to the most productive development stage. We then indicate how optimal
time-to-market and its implied product performance targets vary with exogenous factors such
as the size of the potential market, the presence of existing and new products, profit margins,
the length of the window of opportunity, the firm's speed of product improvement, and com-
petitor product performance. We show that some new product development metrics employed
in practice, such as minimizing break-even time, can be sub-optimal if firms are striving to
maximize profits. We also determine the minimal speed of product improvement required for
profitably undertaking new product development, and discuss the implications of product re-
placement which can occur whenever firms introduce successive generations of new products.
Finally, we show that an improvement in the speed of product development does not necessarily
lead to an earlier time-to-market, but always leads to enhanced products.
(New Product
Development; Time-to-market;
New Product
1. Introduction
Many technology-driven firms compete on new product
development cycle time. Stalk (1988) coined the term
competition to highlight the importance of
quick time-to-market in today's intensive competitive
environment. Clark (1989) estimates that for a $10,000
car, each day of delay in introducing a new model rep-
resents a $1 million loss in profit. A recent McKinsey
study reports that, on average, companies lose 33% of
after-tax profit when they ship products six months late,
as compared with losses of 3.5% when they overspend
50% on product development. In their book Developing
Products in Half the Time, Smith and Reinertsen (1991)
argue that it is necessary to adopt an incremental ap-
proach to product innovation in order to reduce time to
market. This is because incremental product innovation
reduces the amount of effort and learning that must be
done and, consequently, the amount of time needed to
invest in the new product prior to its launch. Such a
perspective has led some companies (e.g., General Elec-
tric, Hewlett Packard) to adopt time-to-market as their
principal product development metric.
There exists an alternative school of thought that em-
phasizes product performance. Several empirical stud-
ies have shown that a new product's success depends
critically on its performance and its value to customers.
Zirger and Maidique (1990), for example, examined 330
new products in the electronics industry and showed
that these factors significantly affected product profit-
ability. Cooper and Kleinschmidt (1987) demonstrated
that product superiority in terms of unique features, in-
novativeness, and performance is a key factor that
Copyright X 1996, Institute for Operations Research
and the Management Sciences MANAGEMENT SCIENCE/VOl. 42, NO. 2, February 1996 173
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product
Figure 1 The Return
Map for HP Pocket Calculator (House and Price
I1 (DaR) It4W
O /Imction I Deyopaelt I , l I
4 6 12 16 18 24 30 32 36 42
MR Thn (-tb)
differentiates new product winners from losers. This
perspective, for instance, has led Boeing to specify per-
formance as the key metric for its new 777 aircraft. The
highly successful Excel 3.0 software program is another
case in point. It has 100 more new features than its pre-
decessor and is considered to be a much friendlier and
"smarter" system (Dyson 1991). New product perfor-
mance is often the decisive factor in the purchase of
technologically advanced products like software pack-
ages. Indeed, most consumer product guides give a
heavy weight to the performance of a software package
(Foster 1990). These observations provide support for a
strategy of making significant improvements in new
product performance over existing products. Unfortu-
nately, such improvements often take more time to de-
velop and can significantly delay the product launch
(see Griffin 1992, and Yoon and Lilien 1985 for empirical
Clearly, there can be a tradeoff between the objectives
of minimizing time-to-market and maximizing perfor-
mance of the new product. Significant improvements in
product performance have the potential to capture a
larger market share from competing (or substitute)
products, but they may take too long to accomplish,
and, consequently, the company will miss the window
of opportunity. An example of this is the Apple's Lisa-
Macintosh development effort in the early 80s. The de-
velopment project was extremely ambitious and aimed
to make major leaps in both product performance
(hardware and software) and manufacturing process
development. The delay, by several quarters, of the
product's introduction drove Apple's earnings down
dramatically and caused the stock of the company to
fall to less than half its early 1983 value (Hayes et al.
1988). Less ambitious improvements in product perfor-
mance can be achieved quickly, but they may not attract
too many customers. In fact, rushing to the market can
be disastrous. General Electric's introduction of a new
with a rotary compressor which failed in the
field has been retrospectively explained as a case where
a product was launched too early. Over one million re-
frigerators had to be recalled and fixed (The Wall Street
Journal 1990). Therefore, there are benefits as well costs
involved in invoking each of these metrics. This sug-
gests that employing integrative new product devel-
opment metrics, which simultaneously capture time-to-
market as well as product performance criteria, might
be more advantageous.
This observation motivated Hewlett Packard's "BET/
2" metric, which is directed toward reducing break-
even time (BET) by one-half for its new products
(House and Price 1991, Young 1991).
Figure 1 depicts the return map employed by Hewlett
Packard (House and Price 1991) for managing a new
pocket calculator development process. As shown, the
break-even time (32 months) is the point at which total
cumulative investment in the development project is
equal to total cumulative net revenue. Reducing break-
even time can motivate the product development team
to address the crucial balance between a high product
performance target and a short time-to-market. A sig-
nificant improvement in the product performance target
is likely to increase the slope of the sales (revenues)
curve, at a cost of delaying the new product launch.
Incremental product improvements, on the other hand,
are likely to generate sale curves that are less steep, but
which bring revenues to the firm earlier.
In this paper we develop a modeling framework that
allows explicit consideration and examination of this
tradeoff for those product markets characterized by (1)
a short and fixed window of opportunity, (2) a high rate
of product obsolescence, and (3) customers who under-
stand and respond to product performance improve-
ments. Industries that exhibit these characteristics in-
clude packaged software, computer hardware and pe-
ripherials, and consumer electronics. Using Dolan's
SCIENCE/Vol. 42, No. 2, February 1996
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product
(1993) terminology, such new products bring "low-
medium" newness to the market. In addition, they have
a "medium-high" opportunity cost and "medium" de-
velopment costs.
The contribution of our modeling framework is based
on the following three aspects. First, it recognizes the
multi-stage nature of product performance improve-
ment processes. This perspective allows us to study
how development resources and time should be allo-
cated across development stages. Second, it considers
both the productivity and the return of development
projects in producing product performance improve-
ments. Prior research focused on either productivity or
return, but not both together (see ?2 for further details).
Third, we adopt an integrative perspective over the new
product development time horizon. This embeds de-
velopment as well as marketing/production cycles.
Both the cumulative costs and the revenues of the new
product, over its entire life cycle, are considered. Our
model yields the following policy insights:
(1) If product improvements are additive (over
stages), it is optimal to allocate maximal time to the
most productive development stage.
(2) Faster is not necessarily better if the new product
market potential is large and if the existing product (to
be replaced) has a high margin. In addition, it is better
to take time to develop a superior product when the
firm is faced with an intermediate level of rivalry.
(3) Minimizing break-even time may lead to prema-
ture new product introduction.
(4) The development capability hurdle needed to un-
dertake profitably a new product development project
increases with the total existing product performance
(that of the developing firm as well as its competitors)
in the market and decreases with the product category
demand rate, the new product margin, competitors'
market share, and the time window of opportunity.
(5) An improvement in the new product develop-
ment capability does not necessarily lead to an earlier
time to market, but it always leads to enhanced prod-
These results are of interest since not all of them are
intuitive. Moreover, the sensitivity of these conclusions
to parametric/environment changes and the explicit
representation of the tradeoffs in the model can further
stimulate empirical and analytical research.
The paper is organized as follows. In the next section,
we review the related literature. In ?3, we provide a
model formulation that captures explicitly the tradeoff
between time-to-market and new product performance.
The structure of the policies implied by the model is
characterized in ?4. Various insights are provided and
stated in terms of testable propositions. Section 5 pro-
vides conclusions and suggestions for future research
directions. The proofs of the formal results can be found
in Cohen et al. (1995).
2. Literature Review
In ?1 we discussed the relationship and tradeoff be-
tween time-to-market and new product performance.
Time-to-market and product performance can also be
affected by the overall level of development resources
assigned to the project. Indeed, the economics/R&D
race literature has often assumed a fixed target of prod-
uct performance level and focused on the tradeoff be-
tween time-to-market and total development resources.
This literature consists of two streams of research: the
decision theoretic approach (for review, see Kamien
and Schwartz 1982) and the game theoretic approach
(for review, see Reinganum 1989). A standard assump-
tion made here is that more severe compressions of de-
velopment cycle ("crashing" the project) are achieved
at increasingly high levels of total development cost;
that is, the relationship between development cycle and
total cost has been taken as strictly convex (see Scherer
1984 and Mansfield et al. 1977, for empirical evidence
for this premise). Another assumption often made in
this literature is that the firm that is first to the market
wins the whole pie, the so-called "winner-takes-all" hy-
pothesis. The winner-takes-all hypothesis and the fixed
performance target assumption are reasonable under
scenarios where firms compete on a patentable break-
through technology. However, many firms spend a sig-
nificant amount of their development resources com-
peting against incumbents in terms of product improve-
ments (Dolan 1993). More often than not, product
development is assumed to be completed, and its de-
velopment cost is not explicitly considered.
As noted earlier, our modeling framework considers
both the productivity and the return of new product
development over time. In this respect, our model
SCIENCE/Vol. 42, No. 2, February 1996 175
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product
framework attempts to integrate the operations and
marketing literatures. We focus on studying the tradeoff
between time-to-market and product performance,
given a specified level of total resource inputs, for three
reasons. First, little or no attention has been devoted to
studying this tradeoff analytically. To the best of our
knowledge, this is the first model-based study of the
issue. Second, the McKinsey study appears to suggest
that the tradeoff between time-to-market and product
performance is more critical than the tradeoff between
time-to-market and level of development resources in
those product markets that we are interested in mod-
eling. Third, industry leaders are beginning to realize
that new product development teams should be kept
small, constant, and manageable. Large development
teams involve expensive administrative coordination
and communication and can delay the decision making
process. For example, the size of the development team
responsible for the successful IBM Laptop that was in-
troduced in 1991 was only nineteen. This is about a
tenth the normal size at IBM (The Wall Street Journal
1991). Hence, it is reasonable to assume that firms will
fix the size of the product development team such that
there is no opportunity to "crash" development pro-
grams. Consequently, incremental product innovations
are necessarily accompanied by a short time-to-market,
and significant improvements in product performance
require a long time to market.
3. Model Formulation
Figure 2 shows the firm's product performance in the
marketplace over time for the situation we wish to cap-
ture here. It is assumed that there is a fixed window of
opportunity T, beyond which the new product has no
value. T can be interpreted as the demand window for
the new product (Clark and Fujimoto 1990 and Dolan
1993). Such a demand window often exists for high-
technology product markets where there is a high rate
of product obsolescence. House and Price (1991) indi-
cate that many of HP's products (e.g., calculators) ex-
hibit such demand characteristics. Other industries that
have such demand windows include packaged soft-
ware, computer hardware and peripherials, and con-
sumer electronics. Indeed, Krubasik (1988) suggests
that a major risk in these product markets is one of miss-
Figure 2 The Perfornance of the Firn's Product
in the Marketplace
over Time
Perfonnunce Ql
New Product
TP T Tume
ing the fast moving demand window. The firm of inter-
est here has an existing product with performance Qo.
At time Tp, a new product with performance Q, is
launched. We assume that the introduction of the new
product makes the old product completely obsolete
(e.g., the latest version of a software program often
makes its predecessor completely obsolete). The stra-
tegic marketing decision is therefore to determine when
to introduce the new product (i.e., replace it with the
existing product) and what the target performance level
should be for the new product. The strategic develop-
ment decision is to determine the allocation of the de-
velopment time and effort across development stages
(to be discussed later). The objective of the firm is to
maximize profits over the time window T.
The development of the new product occurs in mul-
tiple steps. In particular, they include:'
(1) Concept Generation
(2) Product Design
(3) Engineering Analysis
(4) Process Analysis and Design
(5) Prototype Production and Testing
For the purpose of this paper we group these steps
into two more aggregate stages of activities, i.e., Design
and Process.
The Design stage includes steps 1, 2, and 3
above. The Process stage includes steps 4 and 5. After
the Process stage, the new product is launched in the
market (Market stage). We note that there are many
ways in which the activities embodied in these stages
l Both the marketing literature (e.g., Urban and Hauser 1980) and the
production literature (Hayes et al. 1988) have acknowledged the se-
quential nature of the new product development process.
SCIENCE/Vol. 42, No. 2, February 1996
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product Development
can be organized. In particular, the recent movement
toward Simultaneous Engineering (see Nevins and
Whitney 1989) suggests that many of the activities in-
volved in new product development should be carried
out in a concurrent (as opposed to a sequential) manner.
The impact of simultaneity is to reduce total develop-
ment costs and time-to-market as well as to improve the
manufacturability of the product. For the purpose of
this paper, however, we will treat the two macro stages,
defined above, in a sequential manner. Our interest is
in understanding how the new product performance is
affected by the time duration of each stage. We are es-
pecially concerned with how the new product perfor-
mance is "transferred" between stages, i.e., the output
of the new product performance at the end of the Design
stage becomes the input at the beginning of the Process
Figure 3 illustrates the overall structure of our
model. As in previous papers (Clark and Fujimoto
1991, Adler et al. 1992), we focus on the engineering
labor resource inputs. Mathematically, LD and LP
note the sizes of the development team (labor inputs
measured in man-hour per unit time) for the Design
and the Process stage, respectively. A more micro ver-
sion of this model could consider more detailed
classes of inputs (e.g., designers, technicians, drafts-
men).2 Under the model, a day spent in the Design or
Process stage means a day lost in sales (i.e., a day lost
in the Market stage). This premise allows us to cap-
ture explicitly the pressure to compress the develop-
ment process of the new product in the product mar-
kets described above.
In Figure 3, TD and TP are completion times (cal-
endar dates) for the Design and the Process stage, re-
spectively. The cumulative product performance level
at the end of a stage is determined by the following
2The composition of the development team can be an important factor
in the development of the new product. Moreover, it may be different
across development stages. Our model captures this factor, albeit in-
directly and aggregatively, via two exogenous parameters: aD and a,p
(see below). These labor productivity parameters are determined to a
large extent by the composition of the team. A more explicit way to
capture the composition of the team is to have detailed classes of labor
resource inputs categorized by their expertise. We plan to pursue this
as future research.
3 A Multistage
Product Development
and Time
1D Lr
I <~~~~~~I
Dsp Qi(T~D) Prces__(p Market
TD n T
variables: input performance level from the prior
stage, the duration of the stage, and the size of the
development team employed during that stage. En-
hancements in performance are parameterized in
terms of use of raw materials, innovative technology,
manufacturing processes, ergonomics, etc. The firm is
assumed to have numerous performance improve-
ment opportunities so that the increase in the perfor-
mance can be represented as a continuum, i.e., per-
formance is captured by a real-valued index Q(t).
The key factors in our model are the speeds
of improve-
ment for the Design and the Process stages. Specifically,
we define the speed at which performance is being im-
proved during each of the two development stages to
QD = KDL5D, 0 C t - TD, (3.1)
Qp = KpLpP, TD C t ? Tp, where (3.2)
Cj = time derivative representing the speed of im-
provement during stage j [units of performance/time],
Lj = size of the development team for stage j [man-
aj = labor productivity parameter for stage j [units of
Kj = capital productivity parameter for stage j [units
of performance/time],3 and
j = D (Design) or P (Process).
3 Qj, Lj,
and Kj are measured in logarithmic scales (see, for example,
Walters 1963).
MANAGEMENT SCIENCE/Vol. 42, No. 2, February 1996 177
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product Development
Figure 4 shows graphically the evolution of the prod-
uct performance over the total development time. It fol-
lows from (3.1)-(3.2) that the enhancements (i.e., the
increments in product performance) during the Design
and Process stages are QDTD and Qp(Tp
- TD), respec-
tively. The multistage development process links the
performance improvements in an additive manner. Im-
proving product performance is like climbing up a per-
formance ladder (Grossman and Helpman 1991). Thus,
the performance level of the new product from the time
it is launched until the end of T is
Qi(TD, TP)
= QO
+ Qp(Tp -TD), TP
- t - T.
The total man-hours spent in the development project
E(TD, Tp) = LDTD + Lp(Tp
- TD). (3.4)
The speed of improvement (Equations 3.1 and 3.2) is
of the Cobb-Douglas form, and it is taken as analogous
to a production function. The Cobb-Douglas
forms are
conceptually appealing. It is also worthwhile, however,
to examine their empirical validity. Some empirical sup-
port for a Cobb-Douglas type of speed of improvement
can be found in Kamien and Schwartz (1982, chapter 3)
for hardware equipment and chemicals, and in Bohem
(1982) for software. Additional evidence is discussed
Ho (1993) collected primary data from a major food
processor and analyzed secondary data from the auto-
mobile industry to support the assertion that the Cobb-
Douglas is a reasonable functional form. The primary
data from a major food processor consisted of 51 new
food development projects that were undertaken by the
company from 1991 to 1993. For each food development
project, more than 20 resource input variables and two
performance measures of the new product were col-
lected. Three resource inputs that could significantly
predict the performance measures of the new product
were total engineering hours spent in the project, aver-
age experience level of the design team, and the sample
size of the focus groups used during product testing.
Ho (1993) tried several functional forms in regressing
the resource inputs against the performance measures
and found that the Cobb
Douglas form.
provided the best
fit for both performance measures.
4 Improvements
of Product Performance over Time
- - - - - - - - - - - - - - --
QiOD) -
Pxis O _ I
Design lb PrOCess TP Mai e T Tim
The secondary data involving the automobile indus-
try were derived from Clark and Fujimoto's (1991)
study. Clark and Fujimoto conducted a benchmarking
study of new product developments by different firms
for four strategic-regional groups (Japan, United States,
Europe (high-end), Europe (Volume)). They measured
the outcomes of the development process in terms of
lead time (months), total product quality (units of per-
formance, an index ranges from 1 to 100), and total en-
gineering-hours spent in development for 31 different
new car projects. By adjusting the above three measures
relative to a reference point, which represented a stan-
dard car development project, the authors compared the
projects' development efficiency. Ho (1993) divided the
sample into three efficiency groups based on the speed
of performance improvement (high productivity group,
medium productivity group, and low productivity
group) and found that in all three productivity groups,
the Cobb
Douglas form provided the best fit.
Returning to the model formulation, we assume
that at each stage engineering resources are invested.
Hence, the total development costs of the new prod-
uct is
TC(TD, Tp) = WDLDTD + WpLp(TP- TD). (3.5)
This development cost function assumes that engineer-
ing labor costs at each stage j (j = D, P) are charged at
wage rate Wj measured as dollars per man-hour. For
expository purposes, we ignore discounting.
Revenues from the new product can be realized
only during the Market stage, [Tp, TI. A reasonable
178 MANAGEMENT SCIENCE/Vol. 42, No. 2, February 1996
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product Development
market share function, frequently used in the mar-
keting literature, is the logit model that was devel-
oped in discrete choice theory (McFadden 1980). The
sales (demand) rate at time t for the firm that develops
and introduces the new product is the product of the
product category demand rate and the firm's market
M eu(Qo) + eU(Q,) 0 t < Tp,
D(Q(t)) = eU(Q1(TD,TP))
+ eu(QC), T t ,
D(Q(t)) = sales rate at t for the firm that develops the
new product [units sold/time],
M = product category demand rate [units sold!
= performance level of the existing product [units
of performance],
Qc = competitive product performance level [units of
The logit model has received extensive empirical sup-
port. It has been employed widely in the marketing lit-
erature (Green and Krieger 1988, Lilien et al. 1992). It
basically assumes that the customer's utility is the sum
of two components-a deterministic component ob-
servable by the firm, and a random unobservable com-
ponent. The deterministic part is a monotonic function of
product performance and is represented by U(Q(*)) in
the expression above. The random part is assumed to
have a double exponential probability distribution func-
tion. The probability that a randomly chosen consumer
buys from the firm is simply the probability that the
firm's product gives the highest utility to the customer
(Luce and Suppes 1965). In this paper, we use a log
utility function for performance. That is, U(Q(*))
4We have assumed that M is constant. The model structure can be
extended to incorporate nonstationary demand for the category, i.e.,
5We assume a stationary competitive environment. Competitive ac-
tions and reactions can be studied using the above framework by al-
lowing one or more competitors deciding Qc.
We plan to pursue this
as future research.
= ln(Q( )).6 Since eln(x)
= x, the sales rate for the firm
which develops and introduces the new product is
given by7
QO O t < T,
D Q (0) Qo + QC
D(Q(t)) = Qli(TD, TP) T <t<T
Ql (TD, TP) + Qc
The firm's cumulative profit is given by
Tp) = TR(TD, Tp) - TC(TD, Tp), (3.8)
where TR(TD,
and TC(TD,
are total net revenues
and costs (see Equation (3.5)), respectively. The total net
revenues function is given by
= MrO +
O Tp
+ Mr, Ql(TD,TP) + (T - Tp), (3.9)
where ro = margin of the existing product, r, = margin
of the new product.
We are now in a position to define the firm's profit as
a function of the complete set of the new product de-
velopment decisions. The decision set i\, is defined as
A = (TD, TP}. (3.10)
Note that decisions concerning TD
and TP
define the
length of the Process stage (TP - TD). Combining
Equations (3.1) through (3.10) it is straightforward to
generate an explicit representation of the firm's cu-
mulative profit as function of the decision variables.
This substitution yields the following optimization
6We assume a logarithmic utility function because it seems plausible
to have utility as a log function of product performance (like utility as
a log function of money payoff so commonly used in microeconomics
analysis) (Kreps 1988). We tested the robustness of our results to the
functional form of U(Q) through numerical simulation. We experi-
mented with the quadratic (Q - aQ2)
and power (Qa) forms and found
that most qualitative results remain unchanged.
7See Schmalensce (1978) for a related model which links advertising
effort and quality to market share.
SCIENCE/Vol. 42, No. 2, February 1996 179
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product
[P1] TH*(6*) = max
[Mro QO
+ Q TP
+ Mr1
f Qo
+ KpLpP(Tp-T)
- (
IQo + KDLYDTD + KpLpP(Tp-TV) + Q- J
+ WpLp(Tp
- TD))1. (3.11)
Technological constraints may also be included for min-
imum time spent in the Design stage (TD) and Process
stage (Tp), i.e.,
0 < TD ' TD, (3.12)
0 < TP '?TP-TD. (3.13)
This model structure is quite rich and can be em-
ployed to answer the following managerial questions:
(1) What should the launching timing and perfor-
mance targets be?
(2) How should total developmental time be allo-
cated across the two developmental stages?
(3) What are implications for break-even time (BET)
(4) Under what conditions should a new product de-
velopment project be undertaken?
(5) What is the impact of technological and organi-
zational improvements for the new product develop-
ment process which will act to increase the speed of
performance enhancement? How should improvement
capabilities be exploited?
4. Analysis of the Optimal Policies
4.1. Optimal Design and Process Times
Our first proposition is concerned with the structure of
the optimal solution of [P1] in terms of the relative time
allocation between the Design and Process stages. It
specifies circumstances under which it is optimal to con-
centrate on either the Design or the Process stage.
then either TD = TD
or TP-TD = Tp
PROOF. See [8]. L
Proposition 1 implies that the firm should identify its
new product development strengths and concentrate its
resource base on performance enhancements which are
most productive. At the individual project level, this
policy implies that life-cycle profits can be increased via
a focused resource allocation strategy. This focused re-
source allocation strategy appears to have been prac-
ticed by Japanese automakers in the 1970s and 1980s
when they channelled their development resources to
improving product reliability (i.e., Process stage).
Tandy computer adopted a similar strategy by subcon-
tracting out its Process activities. Mansfield (1988) stud-
ied several industries in Japan and the United States and
reported that Japanese companies allocated their devel-
opment resources unevenly across the new product de-
velopment stages. In particular, an unusually large pro-
portion of resources was allocated to Process activities.
American companies, on the other hand, spread their re-
sources more evenly across the developmental stages. At
the firm level, this result suggests that an increased spe-
should be considered. Firms
that focus and cap-
italize on their design strengths tend to hire outside sup-
pliers for their
less efficient activities. These subcontracting
opportunities will make specialized design services viable
and flourishing. It is interesting to note that the popular
notion of core competence (Prahald
and Hamel 1990) also
appears to be consistent with above result. It is operation-
alized here as the firm's productivity
in delivering product
per unit time.
4.2. Optimal Time-to-market and Product Performance
Our next result expresses the optimal time-to-market as
a function of the model's parameters. In particular, we
wish to study the optimal time to market TP, given that
we know from Proposition l that mathematically, the
optimal solution is a corner solution (i.e., focusing on
only one of the two new product developmental stages).
Without loss of generality, we assume that the Process
stage is more productive than the Design stage (i.e.,
D= TD). In the next proposition, we provide a closed-
form solution for TP
and investigate its properties.
2. Let QO=Qo + (KDLDP
- KpLpP)TD.
utility function
is logarithmic,
then the
time to market
Tp is
| Mr1QC(Q0
+ KpL-PT
+ Qc) _
N AMri
+ WpLp -MroQo/( Qo
+ Qc)
P ~~~~KpLPP
PROOF. See [8]. L
180 MANAGEMENT SCIENCE/Vol. 42, No. 2, February 1996
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product
Proposition 2 has several implications that can be ob-
tained via standard comparative statistical analytical
procedures (presented in the appendix). First, it implies
that the compression of the Design and Process stages
(Tp) is less sensitive than the time window compression
(T). This is true because Tp is a square-root function of
T. We illustrate it with a numerical example. If, for in-
stance, Mr, = 1.0, WpLp
= 0.01, Qo = 0.0, Qc = 1.0, TD
= = 0, KpLpP
= 0.2, and T = 10, then T* = 3.7. If T is
compressed by 100% to 5, then T* = 2.1. While the time
window of opportunity drops by 100%, the optimal
time-to-market of the new product only drops by 76%
(3.7 -+ 2.1). Second, it is better to develop a superior new
product rather than to move fast to the market when the
margins of the new and existing products (r1
and ro)
high and the product category demand rate (M) is large.
Thus, the conventional wisdom that "faster is better"
may not hold under these scenarios. Third, the firm
should introduce a greater leap in product performance
when it faces an intermediate level of rivalry (Q,).
Again, it is suboptimal to rush a product too quickly to
the market. On the other hand, developing an ambitious
new product and thus delaying the time-to-market too
long is suboptimal if the competitive performance level
is very low or very high.8
The optimal level of the new product performance
during its Market stage is Q1(T,, T*) = Qo + KpLp Ip.
It can be readily shown that the optimal product per-
formance level increases with Kp and ap. Thus, with
a higher values for the parameters characterizing the
speed of performance improvement, the firm should
8Consider if Q, is allowed to prevail at some time t,, (t, > 0) during the
time window of opportunity (presently
we have t, = 0). We checked via
numerical simulation how T* might be affected if this was indeed the
case. It was found that if t, is less than the original T*, the optimal
time-to-market remains unchanged and is greater than t, (this is the
case where the firm is the follower). If t, is marginally greater than the
original T*, the revised T* is greater than the original time-to-market
and is identical to t, (this is the case where firms launch their products
simultaneously). If t, is significantly higher than the original T*, the
revised time-to-market is greater than the original time to market but
is less than t, (this is the case where the firm is the leader). Only if t,
is very large would the revised time-to-market be smaller than the
original time-to-market. Thus, the overall impact of an Q, which pre-
vails at some time t,. > 0 is to make the firm more aggressive by de-
laying and launching a higher performance product.
strive to increase its performance level target. That is,
better performance always pays. We shall show, how-
ever, that it is not necessarily optimal to reduce the
time-to-market with higher values of Kp
and ap.
4.3. Break-even Time Reduction
It is interesting to compare TP (given in Equation 4. 1)
with T**, the market release time that minimizes the
break-even time (discussed in ?1 and illustrated in
Figure 1). Break-even time has been employed as a
practical guideline for new product launching (see,
for example, House and Price 1991). The break-even
time as a function of TD and TP can be obtained by
replacing T with TBET in Equation (3.11) and setting
TH equal to zero. In doing so, we obtain
+ WpLp(Tp
- TD) - Mr, Q TP
+ Q+Q (4.2)
+ KpLpp(Tp
+ KpLaP(Tp-TD) + Q-
Note that the break-even time is simply the sum of the
time-to-market (i.e., Tp) and the elapsed time taken to
recoup the cumulative net investment (i.e., cumulative
development cost minus cumulative net revenue from
the existing product). The latter time is simply the ratio
of the cumulative investment and the net revenue rate
from the new product (i.e., the second term in the right-
hand side of (4.2)).
3. Minimizing BET leads to premature
In particular,
Tp* < Tp.
PROOF. See [8]. L
The above proposition suggests that using BET
alone as a metric for new product performance leads
to suboptimized profits in new product launching
under the scenarios captured by our model. In par-
ticular, new products launched under minimizing
the BET metric will tend to be incremental types
9Our model does not include a fixed cost, such as overhead and sale
costs, associated with the launching of the new product. Including a
fixed cost will not affect Tp as long as it remains profitable to undertake
new product development. Including a fixed cost will delay the break-
even related release time, T**.
MANAGEMENT SCIENCE/Vol. 42, No. 2, February 1996 181
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product Development
rather than quantum leaps because the firm does not
spend enough time to fully "bake" them. This is akin
to the old saying "no wine before its time." The gap
between Tp* and Tp is smaller, however, when T is
shorter. Since the BET metric is simple and helps
functional coordination, it seems plausible that the
metric be used along with other metrics, especially
those that capture explicitly the life-cycle profits of
the new product. It has been our observation that
firms adopt time-based metrics (minimizing Tp or
BET) without considering life-cycle profits because
of their ease of measurement.
4.4. Product Replacement
Product replacement is an important factor in new
product strategy whenever firms introduce successive
generations of new products that completely replace
existing versions via improvements and enhance-
ments. To analyze its effect on the optimal time to
market, we need to obtain the optimal time-to-market
under no replacement. That is, we wish to analyze the
difference in time-to-market of a successive genera-
tion (Tp) vis-a-vis first generation of new products
This can be done in our model by setting ro = 0
or Qo = 0 in (3.11) and solving for the new optimal
This yields
+ Qc) A
=* l' Mr, + WpLp -Q CQo
PRoposmoN 4. Product replacement always increases
the time to market
(i.e., Tp**
< Tp). The amount of delay,
ATp = Tp - Tp**,
with T and Qo and decreases
with KpLpP.
PROOF. See [8]. 0
This proposition shows that a firm should delay
launching the successive generation of a new product if
it already has a superior existing product. The superior
existing product (high QO)
allows the firm to earn sig-
nificant revenues during the development of the new
product and thus reduces the "pressure" to launch a
new product quickly. The proposition also suggests that
if the firm has a superior new product development ca-
pability (i.e., high value of KpL"P),
then the amount of
delay due to product replacement can be reduced. This
is so because a significantly better new product can be
developed within a shorter time frame with a superior
development capability. If the time window of oppor-
tunity is short, the delay due to product replacement
will become less significant because the pressure to
catch the window becomes the firm's dominant con-
4.5. The Minimally Required Speed of
The model also allows us to investigate the minimal
speed of performance improvement which is required
for a profitable undertaking of the new product devel-
opment project. This minimal speed is useful because it
indicates to the firm whether it has the development
capability needed to undertake a new product devel-
opment for a given market situation. The firm will un-
dertake a new product development only if the optimal
cumulative profit associated with new product devel-
opment is greater than the cumulative profit when there
is no new product development. That is,
, Tp) > Mro QO
0 T. (4.4)
Note that the right-hand side of inequality (4.4) is not
zero. This is because when there is no new product de-
velopment, the firm still enjoys some profits from the
existing product. To find the minimal speed of improve-
ment required for undertaking the new product devel-
opment challenge, we need only to identify the condi-
tions which guarantee the validity of (4.4). We find that
it is necessary that the speed of improvement (KpLpP)
greater than some lower bound. For a fixed develop-
ment team size, this lower bound on the speed of de-
velopment imposes minimal values for the parameters
a and K that the firm must possess in order to undertake
profitably the new product development project. This
insight is summarized in the following proposition.
5. If TD = Tp = 0, then the speed of im-
has to satisfy the following condition
for under-
taking profitably
new product development:
SCIENCE/Vol. 42, No. 2, February 1996
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product
Op = KpLpP Q2 TQ
W(LpQ o + (r, - ro)Qo)] (4.5)
x (Mr,Qc)/(Qo + QC) rjQC ]
In particular,
if r, = ro, then
Qp=KLaP~QC +QO =WQpL(.6
= KpLcp'P
2 T (Mr,Qc)/(Qo + Q)
PROOF. See [8]. 0
The form of the minimal speed of improvement re-
quired for a profitable undertaking of a new product
development shown in (4.6) is interesting. It increases
with total existing product performance in the market
(Qo + Qc)
(while keeping Qc/(Qc + Qo) fixed). Thus,
a product market that is flooded with many superior
products is difficult to enter. The minimal required
speed is also higher when the time window is shorter.
These implications appear to support the notion that
many Japanese firms employ their fast development
capability as a competitive weapon to raise entry bar-
riers. Firms that have slow development capability
will not be able to catch up with the fast and effective
new product developers. Hence, the rapid developers
can use the RHS of (4.6) as a strategic barrier. The
minimal required speed of improvement decreases,
however, with increases in the product category de-
mand rate (M), the new product profit margin (rl),
and the competitor's market share (Qc/(Qo + Qc))
(while keeping Qc
+ Qo
fixed). The last point suggests
that a firm that already has a higher market share has
a bigger challenge and thus a higher hurdle speed of
improvement than a firm with a lower existing market
It can be readily shown that Qp is convex in Qc
and that for Qc > (<)Qo, Qp increases (decreases)
with Qc. If we think of the firm as a follower given
that some pioneer has already introduced a product
of performance level Qc > Qo, then the inequality
(4.6) can also be used by the pioneer to determine
the preemptive
product performance Qc,
above which
it is not profitable for the follower to introduce
a new product since to do so would require a mini-
mum new product development speed of improve-
ment capability-which is either unattainable or pro-
hibitively expensive.
4.6. Exploitation of Improved Speed of Performance
Enhancement Capability
Our next proposition shows that it is not necessarily
optimal to reduce the time-to-market, even with better
speed of performance enhancement capability.
PRoPosITIoN 6. Whenever
the speed of improvement is
within certain bounds, it is optimal to increase
the time to
market with a more effective speed of performance
ment. Specifically,
is between
and QP2.
then OTp*/(KpLUP)
> 0, where
= QP (4.7)
QP2[Q Op+ (Qo + Qc)(ri
- ro)Qo
+_ p], and (4.8)
Mr,Qc + 2M(ri - ro)Qo
(Qo + Qc)WpLp
M2r Qc(r1 - r0)Q0
+ M2(r - 2Q
+ Qc)WpLp12
PROOF. See [8]. 0
It is worth noting that, mathematically, Qp1
< QP2
Equations 4.7 and 4.8). Proposition 6 shows that im-
provements in the firm's speed of performance capabil-
ity may lead to a longer time-to-market rather than a
shorter time-to-market. Lilien and Yoon (1990) have
studied timing of entry and have shown empirically
that if the performance of a follower's new product can
be readily improved relative to that of the existing prod-
ucts, then delaying the market entry timing may lead to
better market performance (Proposition 10 in their pa-
per). If we assume that some pioneer has already intro-
duced a new product of performance QC,
and our firm
is the follower, then our results become consistent with
Lilien and Yoon's empirical evidence if the firm has a
speed of improvement capability bounded from above
and below, between (Pi and QP2.
Thus, an improve-
ment in the speed of product improvement does not
necessarily lead to an earlier time-to-market, but always
SCIENCE/Vol. 42, No. 2, February 1996 183
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product Development
leads to enhanced products (see also the discussion on
Proposition 2).
5. Conclusion and Further Research
In this paper, we have focused on the tradeoff between
target performance and the time-to-market a new prod-
uct. Under an additive multistage model of the perfor-
mance "improvement" process, we have shown that it
is optimal to concentrate efforts on the most productive
stage. In addition, it is possible to determine the optimal
time-to-market and the product performance target,
both of which are functions of the parameters relating
to the firm's cost structure and to the market character-
istics. We have also derived the minimal speed of im-
provement capability required for undertaking profit-
ably new product development projects and shown that
this lower bound is a fairly complex function of the
firm's rate of development labor expense, current per-
formance in the market, product category demand rate,
the new product profit margin, competitor's market
share, and time window of opportunity. Finally, we
have shown that replacing
existing products always de-
lays the time-to-market and product performance target
for the new product vis-a-vis introducing the first gen-
eration of products. Moreover, product replacement
should be delayed further when the existing product
has a high performance. It should, on the other hand,
be introduced faster when the time window is short or
when the firm has a fast development capability. Sen-
sitivity analyses of the optimal time-to-market indicate
that an incremental improvement from the minimal
speed of improvement may lead to delayed
rather than
quicker times-to-market; i.e., the optimal strategy is to
use the faster speed of improvement to develop a better
product rather than to develop a product faster. These
results, in general, contradict some conventional wis-
dom concerning the dominance of incremental over sig-
nificant improvements in product enhancements.
Like any analytical model, our modeling framework
relies on certain assumptions. In particular, we assume
that product performance is additive
over the new prod-
uct developmental stages. The additive assumption is
reasonable if the new product can be structured into
modules and if teams are well coordinated. Proposition
1 is driven mainly by this assumption. Invoking this
assumption facilitates studying the relative allocation of
development time across stages. Since Proposition 1 ap-
pears to have received some empirical support, we con-
jecture that the additivity assumption is a good approx-
imation in certain industries under certain situations.
Proposition 1 is quite robust structurally, however. It is
not affected by several model extensions. For example,
the proposition remains true even if the labor inputs LD
and LP
are taken to be time-varying decision variables.
Other model assumptions, such as stationary product
category demand rate (M), competitive product perfor-
mance (QC),
and fixed size of development teams (LD
and LP)
can be relaxed easily. Relaxing these assump-
tions would allow us to pursue other managerial issues.
For example, allowing product category demand rate to
be a function of price would enable the analysis of pric-
ing the new product. If the market for the product varies
over time, then it is possible to study the revised optimal
timing by allowing M to be a function of time. Com-
petition among firms can also be a,nalyzed if a game
theoretic approach is adopted and QC is expressed as a
function of the competitor's time-to-market. By letting
and LP
be (possibly dynamic) decision variables, we
can study how the firm may choose to compress the
time to market by employing over time more resources
in the new product development process, i.e., by "crash-
ing" the project.
Another possible extension of our modeling frame-
work might be to allow the firm to enhance its product
performance in the Marketing stage via advertising.
Our current model assumes that there is little oppor-
tunity for the firm to do that. Such an assumption is
reasonable in industrial products or products whose
performance can be verified easily by the consumer. In
experience goods, where product performance is not
easily verified, firms can influence the consumers' per-
ception of the product performance by investing in ad-
vertising. Our modeling framework can be easily ex-
tended to incorporate this phenomenon (see Ho 1993).
Propositions 2-6 generate several interesting propo-
sitions which may be subject to empirical scrutiny. For
example, Proposition 2 suggests that optimal time-to-
market is a square-root function of time window T. A
cross-sectional study can be conducted to test whether
this is true. Specifically, it is possible to collect data on
new product development times and their time win-
SCIENCE/Vol. 42, No. 2, February 1996
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product
dows of opportunities across relevant industries and
test the proposition. Another proposition which may be
examined empirically in a fairly straightforward man-
ner is Proposition 5. Factors that determine the firm's
decision to undertake a new product development pro-
ject can be collected and analyzed. Proposition 5 pre-
dicts that these factors include the total existing product
performance, the competitor's market share, the length
of the time window, the product category demand rate,
and the margin of the new product.
Our modeling framework can also be used to evalu-
ate various industry practices such as the target timing
approach, target performance, and target costing. Each
of those practices can be constructed as a restricted case
of a globally optimal procedure based on our modeling
framework. Comparisons of each practice against the
global optimal procedure with respect to the size of the
development team, time-to-market, new product per-
formance level target, and unit cost of the product can
then be made (see Cohen et al. 1993 and forthcoming).
We have also used our modeling framework as the basis
for a real-world implementation and development of a
support system (see Cohen et al. 1994).
Adler, P., A. Mandelbaum, V. Nguyen, and E. Schwerer, "From Project
to Process Management in Engineering: Strategies for Improving
Development Cycle Time," Working Paper Graduate School of
Business, Stanford University, Stanford, CA, 1992.
Bohem, B. W., Software Engineering Economics,
Prentice-Hall, Engle-
wood Cliffs, NJ, 1982.
Clark, K., "Project
Scope and Project Performance: The Effect of Parts
Strategy and Supplier Involvement on Product Development,"
Sci., 35, 5 (1989), 1247-1263.
and T. Fujimoto, Product Development Performance:
Strategy, Or-
ganization, and Management
in the World Auto Industry, Harvard
University Press, Cambridge, MA, 1991.
Cohen, M. and T. Ho, "Ingersoll-Rand: New Product Development
Process and Strategy," University of Pennsylvania Decision Sci-
ences MBA Teaching Case, Philadelphia, PA, 1993.
, J. Eliashberg, and T. Ho, "New Product Design Strategy Anal-
ysis: Evaluating Industry Practices," Working Paper, University
of Pennsylvania Decision Sciences, Philadelphia, PA, 1993.
, and , "New Product Design Strategy Analysis: A Mod-
eing Framework," in S. Dasu and C. Eastman (Eds.), Management
of Design: Engineering
and Management
Kluwer Aca-
demic Publishers, 1994, Norwell, MA, 45-59.
, and , "New Product Development: The Perfor-
mance and Time-to-market Tradeoff," Working Paper, Uni-
versity of Pennsylvania Operations and Information Manage-
ment, Philadelphia, PA, 1995.
Cooper, R. and E. Kleinschmidt, "New Products: What Separates Win-
ners from Losers?," J. Product Innovation
4 (1987),
Dolan, R., Managing the New Product Development Process: Cases and
Notes, Harvard University Press, Cambridge, MA, 1993.
Dyson, E., "Microsoft's Spreadsheet, on Its Third Try, Excel," Forbes,
147, 7 (April 1, 1991), 118.
Foster, J. (Ed.), InfoWorld:
Consumer Product
Guide, Volume II, Brady,
NY, 1991.
Grossman, G. and E. Helpman, "Quality Ladders in the Theory of
Growth," Rev. of Econ. Studies, 58 (1991), 43-61.
Graves, S., "The Time-cost Tradeoff in Research and Development: A
Review," Engineering
Costs and Production
Econ., 16 (1989), 1-9.
Green, P. and A. M. Krieger, "Recent Contribution to Optimal Product
Positioning and Buyer Segmentation," European J. Oper. Res., 41
(1989), 127-141.
Griffin, A., "Metrics for Measuring New Product Development Cy-
cle Times," University of Chicago Working Paper, Chicago, IL,
Ha, A. and E. Porteus, "Optimal Review Frequency in a Concurrent
Engineering Environment," Stanford University Working Paper,
Stanford, CA, 1992.
Hayes, R., S. Wheelwright, and K. Clark, Dynamic Manufacturing,
Free Press, NY, 1988.
Ho, T.-H., "Product Design Strategy Analysis: The Marketing-
manufacturing Interface," Unpublished Ph.D. Dissertation, Uni-
versity of Pennsylvania, Philadelphia, PA, 1993.
House, C. H. and R. L. Price, "The Return Map: Tracking Product
Teams," Harvard Business
Rev., Jan-Feb (1991), 92-100.
Kamien, M. and N. Schwartz, Market
Structure and Innovation,
bridge University Press, Cambridge, MA, 1982.
Kotler, P., Marketing Management,
Seventh Edition, Englewood Cliffs,
NJ, 1991.
Kreps, D., Notes on the Theory of Choice,
Westview Press, Boulder, CO,
Krishnan, V., S. Eppinger, and D. Whitney, "Ordering Cross-
functional Decision Making in Product Development," MIT
Working Paper, Cambridge, MA, 1992.
Krubasik, E. G., "Customize Your Product Development," Harvard
Rev., Nov.-Dec. (1988), 46-53.
Lilien, G. and E. Yoon, "The Timing of Competitive Market Entry:
Exploratory Study of New Industrial Products," Management
36, 5 (1990), 568-585.
Luce, D. and P. Suppes, "Preference, Utility, and Subjective Prob-
ability," in D. Luce, R. Bush, and E. Galanter (Eds.), Handbook
of Mathematical Psychology, John Wiley and Sons, New York,
McFadden, D., "Econometric Models for Probabilistic Choice Among
Products," J. Business, 53, 3 (July 1980), 513-530.
Mansfield, E., J. Rapoport, E. Romeo, E. Villani, S. Wagner, and F.
Husic, Production and Application of New Industrial Technology,
W. W. Norton, New York, 1977.
MANAGEMENT SCIENCE/Vol. 42, No. 2, February 1996 185
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
New Product Development
Mansfield, E., "The Speed and Cost of Industrial Innovation in Japan
and the United States: External vs. Internal Technology," Man-
agement Sci., 34, 10 (1988), 1158-1170.
Neveins, J.
and D. Whitney, Concurrent Design of Products
and Processes,
McGraw-Hill, New York, 1989.
Norton, J. and F. Bass, "A Diffusion Theory Model of Adoption for
Successive Generations of High-technology Products," Manage-
ment Sci., 33, 9 (1987).
Nowery, D. and N. Rosenberg, Technology
and the Pursuit of Economic
Growth, Cambridge University Press, Cambridge, MA, 1989.
Prahald, C. K. and G. Hamel, "The Core Competence of the Corpo-
ration," Harvard Business Rev. (May-June 1990), 79-91.
Reinganum, J. F., "The Timing of Innovation: Research, Develop-
ment, and Diffusion," in R. Schmalensee, and R. D. Willig
(Eds.), Handbook of Industrial Organization, Chapter 14, 849-
908, 1989.
Scherer, F., Innovation and Growth. Schumpeterian
Perspectives, MIT
Press, Cambridge, MA, 1984.
Smith, P. G. and D. G. Reinertsen, "Developing New Products in Half
the Time," Van Nostrand Reinhold Books, New York, NY, 1991.
Stalk, G., "Time-The Next Source of Competitive Advantage," Har-
Business Rev., July-August (1988), 41-51.
Therrien, L., "Kraft is Looking for Fat Growth from Fat-free Foods,"
Business Week, 3151 (Mar. 26, 1990), 100-101.
Urban, G. and J. Hauser, Design and Marketing of New Products,
Prentice-Hall, Englewood Cliffs, NJ, 1980.
Wall Street- Journal, "GE Refrigerator Woes Illustrates the Hazards in
Changing a Product," May 7, 1990.
, "IBM Bends Its Rules to Make a Laptop; New Process Slashes
Development Time," April 16, 1991.
Wilson, L. and J. Norton, "Optimal Entry Timing for a Product Line
Extension," Marketing Sci., 8, 1 (Winter 1989), 1-17.
Young, J., "Investing in Innovation,"
Excellence, 8, 8 (1991), 20.
Zeithmal, V., "Consumer Perceptions of Price, Quality, and Value: A
Means-end Model and Synthesis of Evidence," J. Marketing, 52
(1988), 2-22.
Zirger, B. and M. Maidique, "A Model of New Product Development:
An Empirical Test," Management
Sci., 36, 7 (1990), 867-883.
Walters, A., "Production and Cost Functions: An Econometric Sur-
vey," Econometrica,
31 (1963), 1-66.
Accepted by Abraham
received December 1994. This paper
has been with the authors 1 month
for 1 revision.
SCIENCE/Vol. 42, No. 2, February 1996
This content downloaded from on Tue, 10 Mar 2015 15:35:07 UTC
All use subject to JSTOR Terms and Conditions
... A large number of International Business (IB) studies have found that gradual internationalization can help MNEs manage the liability of foreignness because of timecompression diseconomies (Dierickx and Cool, 1989) and learning from prior experiences (Barkema and Bell et al., 1996;Johanson and Wiedersheim-Paul, 1975). In contrast, some other studies suggest that rapid OFDI expansion may affect firm performance positively due to time-based competition and first-mover advantages (Cohen and Eliashberg et al., 1996;Salomon and Martin, 2008), especially in today's rapidly changing business environment. Given these conflicting views on the impact of rapid expansion, Chang, and Rhee (2011) finds the conditions under which rapid OFDI expansion is a feasible strategy, that is, rapid OFDI expansion can enhance performance of firms in industries where globalization pressures are high or when it is adopted by firms with superior internal resources and capabilities. ...
... Timecompression diseconomy is another reason for poor performance caused by rapid international expansion (Dierickx and Cool, 1989;Barkema and Drogendijk, 2007;Vermeulen and Barkema, 2002). On the other hand, some researches find that rapid international expansion may lead to higher firm performance according to time-based competition (Cohen and Eliashberg et al., 1996;Salomon and Martin, 2008). For instance, Salomon and Martin (2008) find that speed of investment increases firms' performance in the semiconductor and oil industry respectively. ...
... Universities must keep on coming up with new products since with time the programs are adopted by other universities therefore profitability cannot be sustained for a longer period of time (Ayuya, 2015). A number of studies relates innovation and product development and their positive effects of profitability and overall organizational performance (Bonner et al., 2002;Cohen et al., 1996;B. Mbithi et al., 2016;Zhang et al., 2009). ...
... McPhee and Eberlein [51] introduced practices adapted for reducing time-to-market. Cohen et al. looked at development performance and time-to-market trade-off [52]. None of these studies focus on startups per se, but show that there is current knowledge that could be useful for startups, or at least can function as a starting point for performing research into solutions for startups. ...
Full-text available
Software startups are newly created companies with no operating history and oriented towards producing cutting-edge products. However, despite the increasing importance of startups in the economy, few scientific studies attempt to address software engineering issues, especially for early-stage startups. If anything, startups need engineering practices of the same level or better than those of larger companies, as their time and resources are more scarce, and one failed project can put them out of business. In this study we aim to improve understanding of the software development strategies employed by startups. We performed this state-of-practice investigation using a grounded theory approach. We packaged the results in the Greenfield Startup Model (GSM), which explains the priority of startups to release the product as quickly as possible. This strategy allows startups to verify product and market fit, and to adjust the product trajectory according to early collected user feedback. The need to shorten time-to-market, by speeding up the development through low-precision engineering activities, is counterbalanced by the need to restructure the product before targeting further growth. The resulting implications of the GSM outline challenges and gaps, pointing out opportunities for future research to develop and validate engineering practices in the startup context.
... Padmanabhan and Bass [19] worked on the entry timing of the generational product and presented the optimal pricing policy. In the work by Purohit [20], the et al. [10] assumed that products can only be sold during a fixed time window. Sohn and Ahn [22] demonstrated a cost−benefit analysis based on the Norton and Bass model [18] for computing the introduction time of a newer generation. ...
Despite the growing interest in the role of chief executive officers' (CEOs') environmental awareness in enhancing performance, its impact on new product development (NPD) performance, as well as the underlying mechanisms and contextual bounds of the impact, remains unexplored. Based on upper echelons theory, we suggest that CEO environmental awareness promotes NPD performance through corporate environmental practice. We further theorize that the role of CEO environmental awareness is contingent on environmental institutional pressure. To validate the model, we collected survey data from 534 senior executives of 178 manufacturing firms in China. The main results show that corporate environmental practice completely mediates the relationship between CEO environmental awareness and NPD performance. The relationship is positively moderated by environmental normative pressure, whereas environmental regulatory pressure negatively moderates this process. Theoretical and practical implications of how to leverage CEO environmental awareness and environmental institutional pressure to boost NPD performance are discussed.
Should a firm begin its production even before its new product is approved? In a competitive market with time‐sensitive consumers, a firm may choose to adopt the “concurrent process” by conducting the approval process and the production process in parallel so that the product will become available for sale once approved. However, to avoid incurring upfront (production related) investments that are nonrecoverable should the product fail to receive approval, a firm may opt for the “sequential process” so that the production process will only begin after approval. But such a sequential process can delay product launch, making the firm less competitive. These trade‐offs between the concurrent and sequential development processes and the recent Covid‐19 vaccine development motivate us to examine the process choice in the presence of three salient factors: (a) uncertain product approval, (b) time‐sensitive consumers, and (c) asymmetric firm competition —firms have different ex ante approval probabilities. Our equilibrium analysis reveals that it is possible for the laggard firm (the firm with lower ex ante approval probability) to aggressively adopt the concurrent process whereas the leading firm (the firm with higher ex ante approval probability) adopts the sequential process. First, as the approval requirement tightens, both firms have a lower chance of receiving approval, and the laggard firm is more likely to adopt the sequential process than the leading firm. Second, as consumers become more time sensitive, the leading firm is more eager to adopt the concurrent process than the laggard firm. Finally, when the firm asymmetry is large, competition does not always soften and both firms may compete directly by adopting the sequential process. We also examine the case when consumers are “forward‐looking” instead of “myopic.” We find that forward‐looking behavior increases competition and motivates both firms to adopt the concurrent process. Finally, we consider the case when the government offers subsidies to defray the nonrecoverable investments to increase the chance of having an approved product available sooner. Interestingly, we find such a subsidy can backfire resulting in lower consumer welfare and lower profit for the firms especially when consumers are time insensitive.
Recently, an increasing number of firms sell product bundles with after‐sale services. It is challenging for firms to select the best product bundle of the product and services since the product and the services are intercorrelated. In this paper, we study a product service system where a manufacturer provides products to a retailer, who sells them bundled with additional after‐sales services to customers under a service‐level commitment. The market demand is uncertain and depends on the service‐level commitment and product quality. We derive the optimal product (production quantity and product quality) and service (service‐level commitment and service capacity) decisions for this supply system under demand uncertainty. Additionally, this paper designs a coordination mechanism with the service‐level commitment and cost sharing. The coordination mechanism can achieve a Pareto improvement (i.e., both manufacturer and retailer can be better off) and render a higher service‐level commitment and product quality, resulting in a win‒win situation for both the firms and customers. The analytical results demonstrate that service‐level commitment and product quality are substitutes for the firm's profit. Namely, the impact of the commitment of the service level on the firm's profitability increases as product quality decreases. Conversely, service capacity and product quality are complements to the firm's profit. Namely, the better the product quality is, the higher the marginal effect of larger service capacity on profitability.
The objective of this chapter is to discuss managerial approaches to the management of the product development process. At first, the chapter presents the main phases of product development and the defining features of this process, such as the lever effect, the pervasive presence of iterations and the need to balance tradeoffs associated to project cost, time and outcomes. After presenting classical approaches to the management of product development (i.e., sequential, design reviews, concurrent engineering) and the support that can be derived from information technology systems, the discussion explains the role of flexibility and the emerging approaches that is enabled by flexibility (i.e., Agile and Lean).
Full-text available
For most firms producing fast-moving consumer packaged goods, line extension is central to their new product development (NPD) strategy. The authors present a decision-support system for managing the NPD process in this industry, which explicitly evaluates the financial prospects of new line extension concepts. The system developed is based on an in-depth analysis of 51 new product projects launched over a three-year period at a major food manufacturer. It embodies historical knowledge about the productivity of the firm's NPD process and captures some key research and development resource inputs that can affect this productivity. It also provides shipment forecasts at various stages of the NPD process and thus can be used at new product project review gates to evaluate line extension concepts systematically. Finally, the system also can be used to improve the practice of the NPD process by enabling its users to take a product line perspective, using incremental sales evaluation, and by facilitating cross-functional and inter-project learning. Although the system has been developed specifically for a packaged food company environment, its underlying design principles are generic and applicable to a wide range of industries.
This dissertation develops a model framework for the strategic management of tradeoffs among product performance, time to market, unit product cost, and development resources. A new product development process is conceptualized as a product performance improvement process, where improving the mean product performance is like climbing a performance class ladder, and reducing the variance of the product performance is similar to enhancing the reliability of product performance. The model framework integrates three streams of literatures: (1) operations, (2) marketing, and (3) economics R&D, and captures three types of innovations: (1) product, (2) process, and (3) software within a single framework. The novel feature of the model framework is the joint determination of the level of development resources and product launching timing decisions.^ We characterize optimal new product development policies under different product development strategies and competitive scenarios. These characterizations provide insights for the management of strategic tradeoffs and for the setting of project goals. We apply the framework to benchmark several industry practices and highlight the potential pitfalls of each practice. For instance, we show that minimizing project break-even time leads to premature product introduction and that a design approach that ignores the impact of unit product cost can lead to significant loss in life-cycle profits in product innovations.^ We also present empirical evidence to validate the model framework. We examine the model structure and some model assumptions. We test the hypothesis that the productivity of a new product development process can be measured by a product performance improvement function. Overall, the model framework receives empirical support.
Equilibrium conditions for a firm marking decisions with respect to price, quality, and advertising expenditure. Theorems on optimal advertising and optimal quality with fixed prices.
The extension of PERT and Critical Path planning models for analysis of cost-time trade-offs and resource allocation is discussed. Various scheduling algorithms with and without resource constraints are noted and compared as reported in the literature. Key assumptions and some practical limitations are analyzed. Recommendations for further research are offered.