Professional Documents
Culture Documents
Sustainable Competitive Advantage
Sustainable Competitive Advantage
Copyright 2006 John Wiley & Sons, Ltd. Received 20 October 2003
Final revision received 11 July 2005
216 R. Adner and P. Zemsky
Value creation presents a distinct set of plots the price premium as a function of the per-
challenges. Consider for example the market formance premium that a given processor offered
for computer microprocessors. For over 30 years in comparison with the slowest processor in the
microprocessors have followed a steep technology annual product line. Note that while higher pro-
trajectory, with processing speed approximately cessing power always commanded a price pre-
doubling every 2 years in accordance with mium at a point in time, the rightward shift of
Moore’s law. The market has been controlled the curves shows that this premium decreased
by the duopoly of Intel and AMD, with Intel between 1996 and 2000. This suggests that despite
dominating the market based on impressive Intel’s ability to maintain the pace of performance
product and process resources. Despite its strong improvements, its ability to create value for con-
position, from the late 1990s Intel faced problems sumers was hampered by consumers’ decreasing
that were more fundamental than threats of marginal utility from these performance improve-
resource imitation by AMD. ments. This was a major challenge that reshaped
Figure 1 shows the price premium for differ- Intel’s strategy (see Intel Corporation, 1998).3
ent microprocessors in Intel’s product line in 1996 Concurrently, Intel faced increasing consumer het-
(Pentium) and 2000 (Pentium and Celeron).2 It erogeneity and responded by introducing the
3
While Figure 1 is consistent with consumers having decreasing
willingness to pay for faster processors as highlighted by discus-
of valuation by consumers, which underlies the ideas of com- sions of Intel in the business press (e.g., ‘This is not the Intel
petitive positions and substitution, its research agenda too seems we all know,’ Business Week, August 16, 2004: 32), we expect
to have been primarily focused on firm-firm interactions. that some of the shift also reflects increasing capabilities of the
2
We thank Joel Baum for providing these data. rival AMD.
7
Pentium product line 1996 2000 Pentium 3c
Pentium product line 2000 1133 mhz, $990
Price of microprocessor / Price of slowest processor in product line
5
1996 Pentium 133 Mhz, $198 2000 Celeron
2000 Pentium 3c 550 Mhz, $163 766 Mhz, $138
2000 Celeron 366 Mhz, $43
4
1996
Pentium MMX
200 mhz, $550
3
1
1 1.2 1.4 1.6 1.8 2 2.2
Speed of microprocessor / Speed of slowest microprocessor in product line
Figure 1. Price premium for performance premium in Intel’s product lines in 1996 and 2000, where the rightward
shift shows Intel’s decreasing ability to extract a price premium for performance improvements. Adapted from Adner
(2004)
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 217
Celeron product line to target the low-end market and Zemsky, 2003, 2004). In both cases the prior
segment, maintaining the Pentium line to target the work has focused on the competition between firms
high-end market segment. to acquire resources, and has tended to take for
In this paper we explore the ways in which granted how these resources create value for con-
decreasing marginal utility and consumer hetero- sumers, which is our focus here.
geneity across market segments affect the sustain- A final strand seeks to articulate a demand-
ability of competitive advantage through shifts in based perspective on strategy. Adner and Levinthal
consumer willingness to pay.4 We consider the (2001) introduce the construct of decreasing margi-
implications for a variety of threats to sustainabil- nal utility to study patterns in the evolution of
ity: (i) how the threats from substitute technologies technology. Adner (2002) considers how decreas-
change over time and whether these threats will be ing marginal utility interacts with consumer hetero-
permanent or transitory; (ii) how rents from dif- geneity across discrete market segments to study
ferent resources change over time and the extent the emergence of disruptive technologies. Adner
to which imitation and decreasing marginal util- and Zemsky (2005) build on these simulation stud-
ity have similar effects; (iii) how the viability of ies to develop an analytic model that considers the
different competitive strategies changes over time implications of discrete market segments and dis-
and whether a market can support strategic diver- ruptive technologies for a variety of classic IO
sity such that both a cost leader and a differentiator concerns such as industry concentration, social
coexist. welfare, and the effect of mergers on market struc-
ture. These studies, however, focused on questions
of technology strategy. The present paper extends
Formal foundations of strategy
the demand-based perspective to address questions
This paper contributes to three strands in an emerg- at the firm and resource levels of analysis.
ing literature that is developing the formal foun-
dations of strategy. One strand, the added-value
Key elements
approach to business strategy (Brandenburger and
Stuart, 1996), seeks to leverage cooperative game Just as Lippman and Rumelt (1982) use a simple
theory. This literature starts with value creation by model to demonstrate the importance of uncer-
coalitions of different economic actors. In contrast tain imitability for explaining differences in firm
to the non-cooperative approach used in standard performance, our objective is to demonstrate the
IO models, in cooperative games the competition importance of demand-side factors for the analy-
for value capture takes the form of free-form bar- sis of sustainability. We develop a formal model
gaining. Subsequent work in this strand has tended that offers a stylized characterization of firms and
to focus on the conditions that govern the ability competition, which allows us to focus on the impli-
of different parties to capture value (e.g., Lippman cations of the demand-side elements of interest.
and Rumelt, 2003; MacDonald and Ryall, 2004). On the supply side we consider a duopoly set-
We complement this work by shifting attention ting where firms differ either because they use
toward the drivers of value creation. different technologies or because they hold dif-
A second strand seeks to formalize the resource- ferent resources. The performance of firms’ offers
based view. Much of this literature draws on infor- improves over time along technology trajecto-
mation economics to elucidate the workings of ries (e.g., the increasing performance of IT hard-
strategic factor markets in which firms acquire ware; increasing safety of automobiles; increasing
valuable resources (e.g., Makadok, 2001; Makadok breadth and timeliness of financial information).
and Barney, 2001). In addition, it explores resource Our approach to competitive interactions follows
accumulation strategies in the presence of time the added-value methodology (e.g., Brandenburger
compression diseconomies (Pacheco de Almeida and Stuart, 1996), which assumes that a firm’s
ability to capture value is governed by its added
4
We associate the sustainability of competitive advantage with value. In our analysis, we make the notion of
the extent to which firm rents do not erode over time. This is a competitive advantage precise by equating it with
common approach in the literature, but it differs from approaches added value. A firm’s ability to sustain competitive
that focus on the extent that firm strategies remain unique despite
attempts at imitating their benefits (Barney, 1991; Makadok, advantage is thus equivalent to its ability to sustain
1998). added value.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
218 R. Adner and P. Zemsky
By explicitly linking firms’ supply-side activi- in a market’s development there are positive syn-
ties to demand-side value creation we are able to ergies to holding multiple resources that arise from
derive a typology of resources. Process resources advancing the time at which the firm has positive
reduce a firm’s production costs. Product resources value creation. As markets mature, so that even an
increase the performance of a firm’s offer by unresourced firm creates positive value, we find
a fixed amount. Timing resources give a firm a that most resource combinations exhibit neutral or
head start in developing its technology. Finally, even negative interactions.
innovation resources enhance a firm’s technology We extend our base model to allow firms to
trajectory. change their competitive positions over time by
We introduce two new elements to the added- adjusting their resource portfolio. We identify con-
value approach. The first is the extent to which ditions under which firms pursue a Resource Gen-
consumers have decreasing marginal utility (DMU) eralist strategy by investing in product and process
from performance improvements. This key resources simultaneously. We find that such firms
demand-side driver of value creation relates the pioneer markets sooner than Cost Leaders and Dif-
extent to which increases in product performance ferentiators due to the initial synergies in their
are reflected in increases in consumers’ willingness resource portfolio.
to pay. DMU reduces the value of performance The next section introduces the model and dis-
improvements over time. One implication is that cusses the key assumptions. The following section
DMU erodes the rents from product resources, and defines competitive advantage in the context of
does so in a way that parallels the effects of classic the added-value approach. Once these preliminar-
resource imitation. Further, we find that the effects ies are in place, we proceed to the results, with the
of DMU on entry timing and resource strategy are formal proofs given in the Appendix.
contingent on the nature of a firm’s competitive
advantage.
DMU is of interest as an independent variable MODEL
because it varies across markets and market seg-
ments. Generally, in settings where the product is In our model, there are two firms, which we index
more critical to the consumer, whether in direct by i = 1, 2, each of which has a single product
consumption or as an input into a process, perfor- or service offer that is improving over time. Con-
mance improvements are more highly valued and sumers belong to one of two market segments,
DMU is lower. In settings where the product is less which we index by m. There is a high-end mar-
critical, DMU is higher. For example, improve- ket segment (m = H ) with sH > 0 consumers and
ments in display resolution face different levels a low-end market segment (m = L) with sL > 0
of DMU depending on the application (e.g., we consumers. Each consumer buys at most one unit.
would expect lower DMU for display resolution Following Brandenburger and Stuart (1996, 2003),
improvements in medical imaging devices, where we assume that firm i’s profit in segment m is pro-
the stakes are high, than in personal organizers). portional to its added value, which we denote by
The second key element we introduce is con- Aim (t).
sumer heterogeneity. We model such heterogeneity
with discrete market segments that vary in their
Value creation
taste for quality. These differences in taste for qual-
ity across segments determine the relative impor- A key element of an added-value analysis is a pre-
tance of performance-based and cost-based com- cise specification of the value creation of a firm’s
petitive advantages. For example, we show how offer. From Brandenburger and Stuart (1996), an
the extent of consumer heterogeneity determines offer’s value creation for a consumer is the differ-
whether a market can support strategic diversity ence between the consumer’s willingness to pay
such that firms pursuing different resource strate- (WTP) for the offer and the firm’s opportunity cost
gies can profitably coexist. of serving an additional consumer. We denote a
In addition to DMU and consumer heterogene- consumer’s willingness to pay for the offer of firm
ity, an important driver of our results is the dis- i by wim (t). Note that WTP varies over time and
continuities that arise as new firms begin to create across firms and segments. We simplify by assum-
value in a segment. We find, for example, that early ing that there are no capacity constraints so that a
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 219
firm’s opportunity cost is just a constant marginal The exponent β parameterizes the extent of
cost ci > 0. Thus, a firm’s value creation, vim (t), DMU. For example, the utility from increasing the
is maximum speed of an automobile is much greater
when the maximum speed is 40 miles per hour than
vim (t) = wim (t) − ci (1)
when it is 100 miles per hour. We assume that β
is between 0 and 1 so that consumers’ willingness
Willingness to pay to pay increases as performance improves, but at
a decreasing rate.6
We assume that consumers’ WTP is given by In summary, firm i’s value creation for segment
m at time t is
wim (t) = am [xi (t)]β (2)
vim (t) = am [bi (t + hi ) + ri ]β − ci (4)
where am > 0 is consumer taste for quality, xi (t)
is product performance at time t, and β ∈ (0, 1)
parameterizes the extent of consumers’ decreas- Firm heterogeneity
ing marginal utility (DMU) from performance
improvements. We consider two sources of firm heterogeneity in
Consumers’ taste for quality am can be inter- value creation. First, firms might use different tech-
preted in several ways. If consumers are individ- nologies. This is our approach when we analyze
uals, am can be interpreted as an ability to pay the threat posed by a substitute technology. Sec-
which is (usually) increasing in income level, or ond, firms using the same technology might differ
as the intensity of interest in the category (e.g., in their resources. This is our approach when we
audiophiles). If consumers are organizations, am address the sustainability of firm rents and compet-
can be interpreted as the importance of the input itive positions. Equation 4 gives rise to a natural
(e.g., airplane engine performance might be more typology of resources. For firm resources to impact
important for military than for civilian buyers) value creation, they must impact one of the firm-
or as the frequency with which the offer will be specific parameters (i.e., bi , hi , ri , and ci ). We
used. We assume that the high-end segment has a develop this typology in the section on resource
greater taste for quality than the low-end segment, rents.
aH > aL . We characterize consumer heterogeneity
in terms of variation in segments’ taste for quality
(aH − aL ).5 ADDED VALUE
Product performance—the speed of a micropro-
cessor, the accuracy of medical testing, the reli- A critical first step in the analysis is to identify a
ability of logistics services—is given by xi (t). firm’s added value in a market segment at a point
Although many offers have multiple performance in time because this is what drives profitability.
attributes, we simplify the analysis by focusing on Firm i’s added value is the total pay-off to all
a single key attribute. We assume that performance parties when firm i is in the market compared to
is increasing over time according to the total pay-off were firm i to withdraw from the
market. In our context, this is the value a firm cre-
xi (t) = bi (t + hi ) + ri (3) ates relative to a consumer’s next best alternative,
which may either be purchasing a rival’s offer or
where bi > 0 is firm i’s technology trajectory, hi ≥ not making a purchase at all.
0 allows a firm to have a head start in developing
its offer, and ri allows for a fixed increment to the 6
Formally, marginal utility is ∂wim /∂xi = am β/[xi ]1−β . We
performance of firm i’s offer. restrict attention to xi > 1 so that marginal utility is increasing,
and DMU is decreasing, in β. We note that DMU is decreasing
in am as well. For expositional simplicity we focus on the effects
5
Our simplifying assumption of consumer homogeneity within of β on DMU; in the proofs we show that the results hold as well
market segments yields a demand function that decreases in two when it is shifts in am that affect DMU. Finally, note that the
discrete steps. As the number of segments in a market increases, exponential form xi (t)β is conservative in that consumer utility
and as the size of these segments decreases, demand in our model is unbounded and hence limx→∞ wij (t) = 1. Imposing an upper
more closely approximates the downward sloping demand curves bound on WTP would imply that DMU is even more pronounced
found in traditional economics textbooks. than with our functional form.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
220 R. Adner and P. Zemsky
Consider the added value of firm 1 in segment offer. When it is the rival that has positive added
m. When firm 1 creates less value than firm 2 (i.e., value, the firm’s offer is not purchased.7
v1m (t) ≤ v2m (t)), then firm 1’s added value is 0.
This is because the consumer buys at most one
unit and that unit is purchased from firm 2; hence, Added value and firm profits
if firm 1 is not in the game, nothing is lost. When Firm 1’s profits in a segment are given by
firm 1 creates more value than firm 2, three cases A1m (t)sm , recalling that A1m (t) is the added value
arise as illustrated in Figure 2. for a consumer in segment m and sm is the number
First, firm 1 might have negative value creation, of consumers in the segment.8 Firm 1’s total profits
which occurs when its marginal costs exceed the across the high-end and low-end segments are
consumer’s WTP for its offer, in which case the given by
firm has no added value. Second, firm 1 might have
positive value creation while firm 2 has negative 1 (t) = A1H (t)sH + A1L (t)sL − F (6)
value creation. In this case, firm 1’s added value is
the entirety of its value creation, v1m (t), since this where F ≥ 0 allows for the possibility of a fixed
is what it brings to the game. Finally, when both cost to maintaining resources, which we make use
firms have positive value creation, firm 1’s added of in the section on how resource strategy changes
value is the difference in value creation, v1m (t) − over time.9
v2m (t), since firm 2’s value creation would remain
if firm 1 were to exit the game. 7
Note that when the two best options have the same value
creation, added value and hence rents are zero. Although an
Lemma 1. The added value of firm 1 is added-value approach does not specify which offer consumers
purchase in this case, from a rent perspective it does not matter.
A1m (t) = 8
Note that we are assuming that firms capture all of their added
value. One could easily consider the effects of bargaining power
v1m (t) if v1m(t) > 0 ≥ v2m (t) in our model by introducing a parameter αm such that a firm
v1m (t) − v2m (t) v1m (t) > v2m (t) > 0 (5) is able to capture a fraction αm of its added value. Because sm
and αm are both multiplicative terms, our assumption that firms
0 otherwise capture all their added value (i.e., αm = 1) is without loss of
generality. That is, sm can be interpreted as reflecting both a
An analogous expression defines A2m (t). segment’s size and its bargaining power.
Thus, when a rival begins to create value in 9
There are several alternative approaches to linking firm value
a segment, the determinant of the focal firm’s creation to firm profits. Our approach is to assume that profits
are proportional to a firm’s added value. We would get the
added value shifts from its absolute level of value same profit function if we assumed Bertrand price competition.
creation to its relative value creation. In a working paper (Adner and Zemsky, 2002), we consider
What is the connection between added value Cournot competition. The main difference is that under the
added-value approach each segment is winner-take-all, while
and consumer choice? If a firm has positive added with Cournot the firms coexist in the segment if their value
value in a segment, then consumers purchase its creation is sufficiently close. Using a Shapley value approach
Added value of
firm 1, A1m (t)
Value Creation
v1m (t)
v2m (t)
Added value of
firm 1, A1m (t)
v1m (t)
0
v2m (t)
Time
v1m (t)
v2m (t)
Figure 2. The value creation and added value of the firms at three points in time. Firm 1’s added value is calculated
relative to the next best alternative, either firm 2’s offer or not purchasing
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 221
Added value and competitive advantage example, by the phenomenon of disruptive tech-
nologies (Christensen, 1997). Disruptions occur
We now make explicit the relationship between
when existing industry boundaries are redrawn by
competitive advantage and added value. It is useful
the entry of firms using new technologies that
to separate competitive advantage into two parts:
start in a niche segment and, as they improve,
relative costs and relative differentiation. Firm 1’s
displace incumbent technologies from mainstream
cost (dis)advantage is
segments. The recent technology bubble, in which
many promising new technologies turned out not
Ac1 = c2 − c1 (7)
to be disruptive, underscores the need to critically
assess the substitution threat posed by new tech-
and its differentiation (dis)advantage is
nologies.
We consider how change in the relative value
Ad1m (t) = w1m (t) − w2m (t) (8)
creation of technologies shapes the evolution of
Note that a firm’s cost advantage in our model market boundaries. We address whether a new
is the same across segments, while its differenti- technology will displace an incumbent technology
ation advantage varies by segment and over time. from a given segment; if so, when this will occur;
Analogous expressions define Ac2 and Ad2m (t). and finally, whether substitution will be permanent
Firm 1’s net competitive advantage is or transitory. We show how the answers depend on
the extent of DMU, technology trajectories, and
Ad1m (t) + Ac1 = (w1m (t) − c1 ) − (w2m (t) − c2 ) cost positions.
Specifically, in this section we consider the
= v1m (t) − v2m (t) following situation. Firm 1 represents the new
technology, whose performance depends on its
Thus, we have shown the following. technology trajectory
Proposition 1: If the technologies have the same Consistent with this proposition, there are numer-
trajectory, the new technology substitutes for ous cases of new technologies that start with
inferior performance but, by improving along a
the incumbent technology only if it has a cost
superior technology trajectory, eventually substi-
advantage. The greater the extent of DMU the
tute for incumbent technologies. Classic examples
sooner substitution occurs.
are radial tires substituting for bias ply tires and
steamships substituting for sailing ships (Foster,
Proposition 1 highlights the role of DMU in shift- 1986).
ing the emphasis from differentiation to costs as More subtly, Proposition 2 makes predictions
technologies improve. For example, as computer about how the impact of shifts in the demand
hard drive capacities increased, unit costs played environment depends on the relative costs of the
an increasingly important role in determining mar- competing technologies. Consider the effects of
ket outcomes in the desktop segment. Those new the rapid adoption of the World Wide Web in the
hard drive technologies that had lower unit costs late 1990s. The ready availability of rich content
than higher performance incumbent technologies increased consumers’ willingness to pay for com-
(3.5 vs. 5.25-inch; 2.5 vs. 3.5-inch) were able to munication bandwidth, which, consistent with the
displace incumbents from the PC segment; disk proposition, raised the threat posed by higher-cost,
drive technologies that had a cost disadvantage broader-band communication technologies such as
(1.8-inch and 1.3-inch) were not disruptive (Adner, cable modems and ADSL to the incumbent dial-
2002). up technology. Conversely, because the Internet
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 223
allowed users to exploit the power of remote Eventually, the incumbent’s differentiation advan-
servers, the Internet lowered the willingness to pay tage offsets the entrant’s cost advantage and it
for desktop processing power, which increased the re-enters the segment at time tb .
threat posed by lower-cost PC architectures (e.g., Decreases in DMU serve to magnify the dif-
the ‘sub-$1000’ PC). ferentiation advantage of the incumbent, which
both delays the onset of substitution and advances
the time at which the incumbent retakes the
Worse trajectory segment.
Ac1
Ac1
Firm 1’s cost advantage
0
ta tb
Time
Figure 3. The duration of market entry by a substitute technology when b1 < b2 . Firm 1 is in the segment between
time ta and time tb , during which its cost advantage exceeds its differentiation disadvantage. Without a sufficient cost
advantage (i.e., Ac1 < Ac1 ) firm 1 never enters the segment
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
224 R. Adner and P. Zemsky
3 3
Resource rent
Resource rent
with imitation
with DMU
sm∆c
0 0
t1 t2 20 t1 t2 20
Time Time
smv1m (t) Smv1m (t)
Figure 4. The evolution of resource rents when firm 1 has a process resource (for c = 1; c = 3, sm = 1, and b = 1)
under assumptions of DMU (left panel, β = 0.5) and imitation (right panel, z = 0.15). The dashed lines show each
firm’s value creation
on, firm 1’s rents are proportional to its relative segment, its rents first increase over time and
value creation and the size of the segment. then stabilize at sm c .
The evolution of rents from time t2 onward
depends on the type of resource.
Product and timing resources
Both product and timing resources improve the
Process resources
performance of firm 1’s offer by a fixed amount.
Suppose firm 1 has a process resource. In this case, In the case of a product resource, we have x1 (t) =
value creation is increasing at the same rate for bt + r , such that the performance difference
each firm but firm 1’s value creation is always between the firms is x1 (t) − x2 (t) = r . In the
higher by the amount of its cost advantage, Ac1 = case of a timing resource, we have x1 (t) = b(t +
c . Figure 4 (left panel) illustrates this. After time h ), such that the performance difference between
t2 , when both firms have positive value creation the firms is bh . Hence, firm 1 always has a dif-
and firm 1’s rents are proportional to its relative ferentiation advantage, but this advantage erodes
value creation, rents stabilize at sm c , which is over time due to DMU. From time t2 , firm 1’s
constant over time. rents decay as illustrated in Figure 5 (left panel).
Proposition 4: Suppose that firm 1 has a process Proposition 5: Suppose that firm 1 has either
resource. From the time of firm 1’s entry into the a product resource or a timing resource. From
1.5 1.5
Resource rent
Resource rent
with imitation
with DMU
0 0
20 20
Time Time
smv1m(t )
smv1m(t )
smv2m(t )
smv2m(t )
-1.5 -1.5
Figure 5. The evolution of resource rents when firm 1 has a product resource (for r = 4, c = 2, sm = 1, and b = 1)
under assumptions of DMU (left panel, β = 0.4) and imitation (right panel, z = 0.1). The dashed lines show each
firm’s value creation. The left panel also illustrates the case where firm 1 has a timing resource with h = 4
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
226 R. Adner and P. Zemsky
the time of firm 1’s entry into the segment, its Thus we find that the rents from an innovation
rents first increase and then decrease to zero resource are sustainable, but that even in the best
over time. case of a non-imitable innovation resource profit
growth rates are not.
Thus, even a firm possessing an inimitable re-
source, such that it maintains a constant level The effect of imitation on resource rents
of performance superiority, can see its resource
How would these rent profiles differ if resources
rents decay over time as DMU erodes consumers’
could be imitated? In this subsection we relax the
willingness to pay for this performance difference.
assumption that resources are inimitable.10 At the
Consider how this proposition applies to Apple
same time we assume that there is no DMU (i.e.,
Computer’s ability to profit from its head start in
β = 1). This allows us to compare rent profiles
developing easy-to-use personal computers. Over
that result from just imitation with those discussed
time, despite its ability to maintain superior ease
above that arise from just DMU. We continue
of use over rival offers, consumers’ WTP for this
to focus on the case where only one firm has a
advantage seems to have declined.
resource and where there is a single focal segment.
We assume that resources are imitated over
time. Specifically, the degree of resource unique-
Innovation resources
ness at some time t is given by the decreas-
With an innovation resource, firm 1 again has a dif- ing function U (t) = 1/(1 + zt). With this imita-
ferentiation advantage. In this case, x1 (t) = (b + tion process, initially resources are entirely unique
b )t and the difference in performance between (U (0) = 1) and in the long run uniqueness dis-
the firms is x1 (t) − x2 (t) = b t. In contrast to appears (limt→∞ U (t) = 0). The parameter z gives
product and timing resources, this differentiation the ease of imitation in that the larger is z the faster
advantage grows over time. It does so, however, uniqueness erodes.11 Consider the case where firm
at a decreasing rate due to DMU. Figure 6 (left 1 has a process resource. Over time, firm 2 acquires
panel) illustrates this. this resource as well so that firm 1’s cost advan-
Hence, firm 1’s rent continues to increase even tage erodes according to Ac1 = c U (t). Refer to
after firm 2 has positive value creation, although Figure 4 (right panel) for the resulting rent pro-
the rate of profit growth decelerates. file. As in the case without imitation, the resource
asymmetry results in an interval of time during
Proposition 6: Suppose firm 1 has an innovation 10
We thank an anonymous referee for suggesting this section.
resource. From the time of firm 1’s entry into 11
In this subsection, we restrict attention to product, process,
the segment, its rents increase over time, but at and innovation resources. In a single segment context it is not
a decreasing rate. clear what it means to imitate a timing resource.
Resource rent
2 with DMU 2
Resource rent
with imitation
0 0
20 20
smv1m(t ) Time smv1m(t ) Time
smv2m(t ) smv2m(t )
-2 -2
Figure 6. The evolution of resource rents when firm 1 has an innovation resource (for b = 0.3; c = 2, sm = 1, and
b = 1.3) under assumptions of DMU (left panel, β = 0.7) and imitation (right panel, z = 0.15). The dashed lines show
each firm’s value creation
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 227
which only firm 1 creates value for the segment Proposition 9: With imitation and no DMU, from
and in which its rents increase with its level of the time of firm 1’s entry into the segment, the
value creation. After time t2 , when firm 2 starts to rents from innovation resources increase over
create value as well, rents depend on the net com- time, but at a decreasing rate.
petitive advantage c U (t), which is eroding due
to imitation. For both product and innovation resources we
Now consider the case of a product resource, as find that DMU and imitation give rise to qual-
illustrated in Figure 5 (right panel). Over time, firm itative similar rent profiles. Hence, sustainability
2 acquires this resource so that firm 1’s differenti- of advantages rooted in superior product perfor-
ation advantage erodes as the gap in performance, mance is threatened not only by imitation but also
x1 (t) − x2 (t) = r U (t), shrinks due to imitation. by consumers’ decreasing willingness to pay for
The resulting rent profile follows the same pattern a superior product. Thus, imitation and DMU are
as with a process resource. Thus we have: substitute threats such that it is worth expending
resources to neutralize one only to the extent that
Proposition 7: With imitation and no DMU, from the other threat can be avoided as well. Broadly,
the time of firm 1’s entry into the segment, the strategy formulation needs to anticipate not only
rents from both product resources and process competitor imitation but also shifts in the bases of
resources first increase and then decrease to zero consumer value creation. Consequently, it would
over time. be useful to extend the empirical literature on sus-
tainability (e.g., McGahan and Porter, 2003) to
While the rents from product resources are eroded decompose the sources of rent erosion into supply-
by imitation and DMU in qualitatively similar side and demand-side drivers.
ways, the rents from process resources are sus-
tainable under DMU but not under imitation.
With either imitation or DMU we have an initial A MULTI-SEGMENT,
period of increasing rents. Without DMU, how- MULTI-RESOURCE EXAMPLE
ever, value creation for both firms increases faster.
The increase in firm 1’s rate of value creation To this point we have restricted the analysis to
leads its rents to increase at a faster rate. However, a single segment and to the rents from a sin-
because firm 2’s rate of value creation benefits gle resource. These simplifications are useful for
from an equivalent increase, the interval of time developing some basic intuitions about demand-
during which only firm 1 creates value is shorter. side drivers of sustainability. Reality, of course, is
Thus we have: more complex. A market can have multiple seg-
ments, a firm can hold multiple resources, and its
Proposition 8: With imitation and no DMU, the rivals can hold their own unique resource bun-
rate of increase in rents from product and pro- dles. In this section, we begin to extend our model
cess resources is greater than with DMU and to incorporate these complexities to more closely
no imitation, but the period during which rents approximate empirical settings.
increase is shorter. We consider an example, loosely inspired by
the competition between Sony and Matsushita
Finally, consider the case where firm 1 has an in consumer electronics. Firm 1 (Sony) holds
innovation resource, as shown in Figure 6 (right both a product resource (r = 1) and a timing
panel). With imitation, Firm 2’s technology trajec- resource (h = 2), while firm 2 (Matsushita) holds
tory increases over time, converging to the trajec- a process resource (c = 0.7). We return to the
tory of firm 1. What matters for value creation, assumptions that there is DMU (β = 0.5) and that
however, is not the rate of increase at a point in resources are inimitable. The high-end segment has
time, but rather the absolute levels of performance a taste for quality of aH = 2, the low-end segment
that have been achieved. Consequently, conver- has a taste for quality aL = 1.4, b = 1, sH = sL =
gence in trajectories only implies a deceleration in 100, c = 3.5. Figure 7 shows each firm’s rent pro-
the growth of firm 1’s competitive advantage. This file.
results in an increasing rent profile that parallels Due to the combined strength of its timing and
the profile with DMU. product resources, firm 1 is the first to create value
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
228 R. Adner and P. Zemsky
100
Firm 1 s rents
Firm 2 s rents
0
t1 t2 t3 t4 t5 t6 20
Time
Figure 7. The evolution of firm rents in a two segment setting where firm 1 holds performance and timing resources
and firm 2 holds a process resource
in the market. Firm 1 creates value first in the from the low-end segment. At time t6 , its differ-
high-end due to that segment’s higher taste for entiation advantage has eroded so much that firm
quality. From time t1 until time t2 , it is only firm 1 is displaced by firm 2 in the high-end segment
1 that is creating value and therefore its rents as well. After time t6 , firm 2’s rents continue to
increase as its offer improves. From time t2 , firm increase as firm 1’s differentiation advantage fur-
2 joins firm 1 in creating value in the high-end ther erodes, converging to a final profit level of
segment. Firm 1’s rents from that segment, which sm c = 70 in each segment.
are now proportional to its relative value creation, Thus, by assembling different elements of our
decay over time as DMU erodes its differentiation simple model, one can characterize relatively com-
advantage. To this point, the rent profiles are the plex patterns that include shifts in market leader-
same as those illustrated in Figure 5 (left panel). ship across segments and shifts in firms’ absolute
At time t3 , firm 1 returns to a period of profit and relative profits over time.
growth as its offer begins to create sufficient The example raises several questions that are
value to serve the low-end segment. Now, firm worthy of further exploration. First, what is the
1 starts earning rents from the low-end segment interaction among multiple resources held by a
which are proportional to its value creation and single firm and how does it change over time? Sec-
which increase as its offer improves. The increas- ond, what explains firms’ order of entry into dif-
ing rents from the low-end segment more than ferent market segments? Third, what are the condi-
offset the continuing decay in rents from the high- tions that allow firms with heterogeneous resources
end segment. to coexist in a market? We address these questions
At time t4 , firm 2 begins to create value in the in the remainder of the paper.
low-end segment, triggering a permanent decline
in firm 1’s fortunes. Now, firm 1’s rents from both
segments are decaying over time. MULTIPLE RESOURCES
Note that until time t5 firm 2 does not sell to
either of these segments. It is firm 2’s increasing There is a prior literature exploring the value
relative value creation that limits firm 1’s rents.12 to firms of combining multiple resources (e.g.,
At time t5 DMU has eroded firm 1’s differentiation Amit and Schoemaker, 1993; Teece, Pisano, and
advantage sufficiently that firm 2’s cost advantage Shuen, 1997; Makadok, 2001). This literature
gives it superior value creation in the low-end seg- focuses on how internal firm capabilities govern
ment. Consequently, at time t5 firm 1 is displaced the effectiveness with which firms can deploy
their resources. Our focus in this section com-
plements the internal focus of the received lit-
12
Note, however, per the discussion of exogenous technological erature, with an external focus on how different
progress in the section on substitute threats, that firm 2 might
be selling its products to consumers in an (unanalyzed) niche resource combinations affect value creation in the
segment. See also footnote 9. market.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 229
In the above section on resource rents, we char- One significance of superadditivity, or resource
acterized the rent profiles from holding individual synergies, is that resources cannot be valued in
resources on their own. We now consider the rents isolation. Rather, resource valuation depends on
from combining multiple resources within a single interactions within a firm’s portfolio of resources.
firm.13 When is the rent profile of a multi-resourced In contrast to prior work on resource combination,
firm different from the simple sum of the rents the synergies in our model do not arise from
from holding each of the resources on its own? internal capabilities. Our synergies arise from the
Holding two resources together can result in rents interaction between the resource portfolio and the
that are less than, equal to, or greater than the sum demand environment, which ultimately determines
of the rents from holding each on its own. This the threshold for value creation. One implication
leads to the following definitions: is that firms pioneering new market segments have
greater incentives to acquire resources because
• Definition 1. Two resources are subadditive each additional resource speeds the time of entry
when the rent from holding both resources and the onset of rents.
simultaneously is less than the sum of the rents Do these synergies persist as markets mature?
from holding each on its own. As firm value creation increases, it is no longer
• Definition 2. Two resources are additive when the ability to break into the market that matters,
the rent from holding both resources simulta- but rather the level of value creation. Consider
neously is equal to the sum of the rents from a firm holding a timing and a product resource.
holding each on its own. Both of these resources increase the performance
• Definition 3. Two resources are superadditive of the firm’s offer and hence its value creation.
when the rent from holding both resources Due to DMU, the greater the level of performance,
simultaneously is greater than the sum of the the lower the effect of an additional performance
rents from holding each on its own. enhancement on WTP. Hence, the net effect on
WTP (and therefore on value creation) of the two
All types of resources increase firm value creation resources in combination is less than the sum of the
and a firm holding two resources must create effects when the resources are held on their own.
more value than a firm holding either resource Therefore, timing and product resources become
on its own. Hence, a firm with two resources subadditive as the market matures.
starts creating value in a segment earlier, and starts Next consider a firm that holds a process re-
earning rents sooner, than a firm holding either source and some other, performance-enhancing
resource on its own. Consequently, rents must resource type. Since the shift in value creation is
initially be superadditive because neither resource the sum of the cost reduction and the increase in
held on its own would have yielded positive value WTP from the performance enhancement and since
creation. Even at a time when one of the resources DMU does not impact the value of cost reductions,
is enough for positive value creation, the resources there is an additive interaction between process
remain superadditive because adding the second resources and the other resource types.
resource increases rents even though this second Finally, consider a firm holding a timing and an
resource, on its own, would not be have positive innovation resource. Here there is a superadditive
value creation. interaction between the two resources: the greater
Therefore: the technology trajectory, the greater the benefit to
having a head start on development; and the greater
Proposition 10: Consider a firm with two re- the head start, the greater the benefit to increasing
sources facing an unresourced rival in a seg- the firm’s trajectory.
ment. There is an interval, starting at the time
where the firm has positive value creation, dur-
ing which the two resources are superadditive. Proposition 11: Consider a firm with two re-
sources facing a rival with none. After the initial
interval of superadditivity, the resources can
13
Alternatively, one could interpret the discussion below as be additive, subadditive, or they may remain
applying to resources spread across multiple firms that are using
an alliance, joint venture, or other contractual mechanism to superadditive, depending on the identity of the
jointly deploy their resources. resources. (See Table 1 for details.)
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
230 R. Adner and P. Zemsky
Table 1. Resource interactions after the initial interval resources are split among competing firms? In any
of superadditivity, classified as subadditive (−), additive given segment, the combined profits of the two
(+), or continued superadditivity (++)
firms depend on the added value of the firm with
Resource Process Product Timing Innovation the greatest value creation. Splitting the resources
type between the firms necessarily lowers the value
creation of the leading firm. At the same time,
Process + + + + it necessarily increases the value creation of the
Product + − − − lagging firm. The net result is lower added value
Timing + − − ++
Innovation + − ++ − and hence lower industry profits.
the order of entry is contingent. Echoing the logic differentiation advantage are sometimes very suc-
of Proposition 2, if the new firm has a cost advan- cessful (e.g., Kim and Mauborgne, 1997; Besanko,
tage relative to the incumbent, then entry occurs Dranove, and Shanley, 2000). To explore this, we
while the new firm still has a differentiation disad- allow a firm to pursue a Resource Generalist strat-
vantage. The new firm therefore enters the low-end egy, by holding a product and a process resource
segment first since that is where differentiation is simultaneously.
less important. Conversely, if the new firm has a We first explore the sustainability of different
cost disadvantage, then entry only occurs when it resource strategies. We then consider the condi-
has an offsetting differentiation advantage. In this tions under which a market can support strategic
case, the new firm enters the high-end segment first diversity, where firms pursuing different strategies
since that is where differentiation is more highly can profitably coexist.
valued. In every period firms first choose their resource
strategy, which determines their value creation,
Proposition 15: In an industry with a high-end and then compete in the market as before. For-
and a low-end market segment both being served mally, we are studying a biform game as intro-
by an incumbent, a new technology enters the duced by Brandenburger and Stuart (2003). We
low-end segment first when it has a cost advan- assume that there is a fixed per-period cost to main-
tage. It enters the high-end segment first when it taining resources of Fc > 0 for process resources
has a cost disadvantage. and Fr > 0 for product resources. If a firm holds
both resources simultaneously, there is an addi-
Thus, low-cost mini-mill technology started in tional fixed cost associated with organizational
low-end segments like reinforcement bars before complexity of K ≥ 0. Following Porter (1996), the
moving into higher-end segments like structural parameter K captures the extent to which inter-
steel, while jet engine technology was first adopted nal trade-offs make it difficult to pursue cost and
in high-end military market segments before mov- differentiation advantage simultaneously.
ing into long-haul commercial segments. The
proposition can be applied not only at the technol-
Single segment analysis
ogy level, but at a firm level of analysis as well.
For example, the discount broker Schwab began Suppose that there is a single segment. To justify
in low-end segments and has gone on to penetrate the fixed investment in maintaining resources, a
higher-end segments as the quality of its offer has firm must have rents and therefore positive added
improved, while Dolby’s noise reduction technol- value for some customers. Since at most one firm
ogy moved from the professional recording market can have added value in a segment, with a single
to the mass market. segment at most one firm will maintain resources.
How will the firm’s resource strategy change over
time? The three main drivers of resource strat-
RESOURCE STRATEGY OVER TIME egy in a single segment are time, the costs of
complexity, and the stand-alone profitability of
To this point, we have considered how the value each resource. From Proposition 4, the profitabil-
of resources changes over time, taking resource ity of maintaining the process resource on its own
endowments as given. We now take a first step reaches its maximum at the time when the unre-
toward endogenizing a firm’s resource strategy. sourced firm starts creating value, and is a constant
We restrict attention to three generic strategies thereafter. Let c be the maximum profitability of
for holding product and process resources. As in maintaining the process resource on its own. From
Porter (1980), a firm can be a Differentiator, by Proposition 5, the profitability of maintaining the
holding a product resource, or a Cost Leader, by product resource is highest at the point where the
holding a process resource. Porter argues that firms unresourced firm starts creating value, and declines
face a choice between positioning with a cost or a thereafter. Let r be the maximum profitability of
performance focus, and that those firms that do not maintaining the product resource on its own. We
choose one of these positions risk being stuck in consider the case where r > c > 0.
the middle. Others, however, have observed that Figure 8 shows how the firm’s resource strategy
firms that simultaneously pursue both cost and varies over time and with the cost of complexity.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
232 R. Adner and P. Zemsky
40
Cost of Complexity (K )
Differentiator
Cost
no Leader
Πc entry
Resource
Generalist
0
20 70 120
Time
Figure 8. The effect of the cost of organizational complexity on a firm’s resource strategy over time in a single
segment (for c = 10, c = 1.4, Fc = 100, r = 90, Fr = 60, β = 0.4, am = 1, sm = 100)
At first the firm is out of the market and (i) If the cost of complexity is sufficiently small
holds no resources because no resource combina- (K < c ), then the firm enters the market
tion results in consumer willingness to pay that segment with a Resource Generalist strat-
is greater than firm costs. What happens next egy.
depends on the cost of complexity. Suppose that (ii) For higher costs of complexity (K > c ),
the cost of complexity is high, specifically K > the firm enters the market segment with a
c . Then firms will only maintain one resource at Differentiator strategy.
a time. The firm starts with a Differentiator strat- (iii) For any value of K, in the long run the firm
egy because, initially, the profits from maintaining pursues a Cost Leader strategy.
the product resource are greater (an implication of
r > c ). Over time, however, DMU erodes the The proposition shows how one can place bound-
rents from the product resource so that at some ary conditions on the pursuit of different generic
point the process resource offers higher returns. strategies.14 While one may see a firm pursuing
At this point, the firm shifts to a Cost Leader the classic generic strategies of Cost Leadership or
strategy. Differentiation, we identify conditions under which
Now suppose that the costs of complexity are a firm does better, for a limited interval of time,
low, specifically K < c . Then the firm enters the with a Resource Generalist strategy. The shift over
time away from product resources and towards
market with a Resource Generalist strategy. Recall
process resources echoes the technology life cycle
from Proposition 10 that when pioneering a market
literature, where a central empirical regularity is
any two resources are superadditive. As a result,
the shift from product to process innovation as
the entry time for a Resource Generalist is earlier
markets mature (Utterback, 1994). Unlike classic
than that for a Differentiator. Resource superad-
explanations based on the emergence of a domi-
ditivity is partly offset by the cost of complexity nant design (Utterback and Abernathy, 1975), in
and hence the lower is K the earlier the entry. Over our theory the transition is driven by DMU (see
time the rents from the product resource erode and also Adner and Levinthal, 2001).
the firm eventually drops the product resource from Consider the effects of DMU on resource strat-
its portfolio, shifting to a Cost Leader strategy, as egy. Not only does it accelerate the erosion of the
in the case with a high K.
14
For simplicity, we restricted the analysis to the case of r >
c > 0. Consider the case where c > r > 0. For low K
the firm still enters as a Resource Generalist and subsequently
Proposition 16: Suppose r > c > 0 and switches to be a Cost Leader, while for high K the firm now
there is only a single segment. enters directly as a Cost Leader.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 233
returns from the product resource, it also delays the greater relative value creation for the Differentia-
time at which the firm is able to enter the market. tor. Thus, the Differentiator must have added value
Thus there is a clear negative relationship between in the high-end segment, which implies that the
the extent of DMU in the market and the time dur- Cost Leader must have added value in the low-end
ing which the firm maintains the product resource. segment to be viable.
In contrast, the effect on the holding of a process Both firms must have sufficient added value in
resource is contingent: for low costs of complex- their respective segments to cover their resource
ity, the two resources are deployed simultaneously costs. For the Differentiator, this means that the
and hence higher DMU delays the deployment of taste for quality in the high-end segment cannot
the process resource. On the other hand, for high be too low (aH ≥ a H ). For the Cost Leader, the
costs of complexity, the firm is choosing between taste for quality in the low-end segment cannot be
the two resources and higher DMU speeds the shift too high (aL ≤ āL ), as otherwise the Differentia-
from the product to the process resource. tor would create too much value there. However,
neither can the low-end segment’s taste for qual-
Proposition 17: Suppose r > c > 0 and ity be too low (aL ≥ a L ) because decreases in aL
there is only a single segment. Increases in DMU increase the Cost Leader’s added value only until
reduce the time during which the firm maintains the point where the Differentiator no longer creates
the product resource, but have an ambiguous value for the segment. After that, further decreases
effect on the time during which the firm main- in the segment’s taste for quality reduce the Cost
tains the process resource. Leader’s added value, eventually driving it to zero.
Our basic model and the associated intuitions on Proposition 18: Strategic diversity, where one
the evolution of resource rents can be extended, firm has a process resource and the other has
in a straightforward way, to look at the choice of a product resource, requires a sufficiently high
resource strategy. This extension is a step toward cost of complexity, a sufficiently high taste for
addressing a weakness in the received literature on quality in the high-end market segment, and an
positioning, that ‘our understanding of the dynamic intermediate level of taste for quality in the low-
processes by which firms perceive and ultimately end market segment.
attain superior market positions is far less devel-
oped [than our understanding of advantage at a Thus, beyond internal trade-offs in the form of
point in time]’ (Porter, 1991: 95; see also Rumelt, costs of complexity we find that sufficient con-
1987). We think that explicitly linking resource sumer heterogeneity is required to support strategic
choice to the evolution of value creation is a diversity. For example, in the airline sector the
promising avenue for deepening our understanding high level of consumer heterogeneity (e.g., busi-
of the dynamics of positioning. ness vs. leisure travelers) helps support strategic
diversity (e.g., full service vs. no-frills airlines).
How do changes in market size and in the fixed
Strategic diversity
costs of maintaining resources affect the level of
Under what conditions would one expect to consumer heterogeneity (a H − āL ) required to sup-
observe strategic diversity, where one firm pur- port strategic diversity? For example, in 2001–02
sues a Cost Leader strategy and the other firm a the airline sector experienced reductions in market
Differentiator strategy? From the single segment size, especially at the high end, as well as higher
analysis, we need a sufficient cost of complexity fixed costs due to increased security requirements.
to preclude the industry from being dominated by a The greater is the fixed cost of maintaining the
Resource Generalist. Moreover, because each firm product resource and the smaller the size of the
must have added value in some segment, there high-end segment, the higher must be the Differ-
must be at least two different market segments. entiator’s added value for each high-end customer
How different must they be? As before, consider and hence the lower bound on the high-end seg-
a market with a high-end segment and a low-end ment’s taste for quality, a H , must increase. Con-
segment that vary in the taste for quality, aH > aL . versely, the greater is the fixed cost of maintaining
A higher taste for quality places greater empha- the process resource relative to the size of the low-
sis on product performance and hence results in end segment, the higher must be the Cost Leader’s
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
234 R. Adner and P. Zemsky
added value for each low-end customer and hence those settings where DMU and consumer hetero-
the lower must be āL and the higher must be a L . geneity are important, to develop intuitions regard-
ing their effects on sustainability.
Incorporating a demand-based perspective into
Proposition 19: The extent of consumer hetero- empirical research will enrich the hypotheses that
geneity (a H − āL ) required to support strategic can be tested in both cross-sectional and longitu-
diversity is increasing in the level of fixed costs dinal studies. Although new tools will be needed
required to maintain the resources and decreas- to characterize the demand environment, many
ing in the size of the segments. have already been developed for the purpose of
constructing quality-adjusted price indices (e.g.,
Griliches, 1961; Trajtenberg, 1990) and in the new
CONCLUSION empirical industrial organization literature (e.g.,
Berry, Levinsohn, and Pakes, 1995). These tools
can provide the inputs that would be used to study
Our approach to sustainability starts on the sup- strategy questions.
ply side with product market competition and One limitation of our approach is that firms are
improving technologies. The novelty is that we assumed to engage in intense (i.e., Bertrand) price
introduce an explicit treatment of how technology competition and thus we cannot address changes in
improvements affect consumer choice among com- the intensity of rivalry. This is a substantive lim-
peting offers. This leads us to focus on demand- itation because technological progress and DMU
side drivers: marginal utility from performance might affect the intensity of rivalry in some set-
improvements, consumer taste for quality, and the tings. Similarly, firm choices regarding market
extent of consumer heterogeneity. We combine entry and resource portfolios could also impact
these elements in a simple model that allows us to the extent of rivalry. While these interactions are
address a wide range of issues related to the sus- beyond the scope of the current paper, we highlight
tainability of competitive advantage. By explicitly them as potential avenues for future research.
linking resources and utility, we have attempted There are two additional elements in this paper
to complement the traditional focus in strategy on that we think are a promising basis for future theo-
competition and value capture with a focus on con- retical work. First, the formalization of competitive
sumers and value creation. advantage as superior value creation provides a
At the level of firm resources, we show how simple and yet compelling foundation for strategy
competitive advantage can erode not only because theorizing. A natural way to build on this founda-
imitation undermines the uniqueness of resources, tion is to incorporate firm actions that shape value
but also because consumer valuation of firm dif- creation. Our analysis of resource strategy is one
ferences declines due to the effects of decreasing step in this direction. Second, a focus on the inter-
marginal utility. At the level of firm positions, we action between consumer heterogeneity and firm
show that strategic heterogeneity is rooted not only strategy offers a promising avenue for building
in differences between firms’ internal resources, a richer theory of firm–environment fit. In this
but also in the extent of consumer heterogeneity paper, we focus on the link between consumer
in the firms’ demand environment. heterogeneity and strategic diversity. Future work
While DMU and consumer heterogeneity are along this trajectory might fruitfully explore how
characteristic of many (though perhaps not all) set- firms should respond to consumer heterogeneity by
tings, they have been largely ignored in strategy their market segmentation choices, and relatedly,
studies. We note that the assumptions of DMU and their market diversification decisions.
consumer heterogeneity are critical to our results The interactions between firm strategy and the
and without them most of our observed patterns demand context suggest new dimensions along
would disappear. This, of course, is precisely the which to consider firm strategy. In particular,
point: to highlight the role of demand-side ele- important features of demand are subject to influ-
ments in determining sustainability, and to assure ence by firms. For example, Intel responded to
that these demand-side threats are not overlooked. decreasing marginal utility for processing power
We thus hope, first, to encourage the field to be by investing billions in venture capital directed
more sensitive to their existence; and second in at suppliers of complements that would increase
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 235
demand for processing power. In addition to invest- Amit R, Schoemaker PJH. 1993. Strategic assets and
ing in complements, firms can influence thresholds organizational rent. Strategic Management Journal
for acceptable performance through advertising, 14(1): 33–46.
Barney J. 1986. Strategic factor markets: expectations,
standard setting, and regulation. This raises the luck and business strategy. Management Science 31:
question of how firms’ resource allocation pro- 1231–1241.
cesses should incorporate the possibility for such Barney J. 1991. Firm resources and sustained competitive
demand-side strategies (see Adner, 2004, for fur- advantage. Journal of Management 17(1): 99–120.
ther discussion). Barney J. 2001. Is the resource-based ‘view’ a useful
perspective for strategic management research?
The fit between a firm and its environment has Academy of Management Review 26(1): 41–56.
been a fundamental concern in strategy since the Barney J, Peteraf M. 2003. Unraveling the resource-
inception of the field. Over time, a tendency has based tangle. Managerial and Decision Economics 24:
developed to equate environmental analysis with 309–323.
competitor analysis. We hope to have demonstrated Berry S, Levinsohn J, Pakes A. 1995. Automobile prices
the value of focusing attention not just on firm in market equilibrium. Econometrica 63(4): 841–890.
Besanko D, Dranove D, Shanley M. 2000. Economics of
resources and competition, but on the demand Strategy (2nd edn). Wiley: New York.
environment as well. Brandenburger A, Stuart HW. 1996. Value-based busi-
ness strategy. Journal of Economics and Management
Strategy 5: 5–24.
ACKNOWLEDGEMENTS Brandenburger A, Stuart HW. 2003. Biform games.
mimeo, New York University.
Christensen CM. 1997. The Innovator’s Dilemma.
We would like to thank seminar participants at Harvard Business School Press: Boston, MA.
the Atlanta Competitive Advantage Conference, Dierickx I, Cool K. 1989. Asset stock accumulation and
BYU Winter Conference, Duke, Michigan, NYU, sustainability of competitive advantage. Management
Ohio State, Strategy Research Forum, Tel Aviv Science 35(12): 1504–1511.
University, Toronto, Washington University, and Foster R. 1986. Innovation, the Attacker’s Advantage.
Simon & Schuster: New York.
Wharton, as well as Jay Barney, Adam Branden- Ghemawat P. 1991. Commitment: The Dynamic of
burger, Ramon Casadesus-Masanell, Pankaj Ghe- Strategy. Free Press: New York.
mawat, Javier Gimeno, Constance Helfat, Anne Grant RM. 1998. Contemporary Strategy Analysis: Con-
Marie Knott, Bruce Kogut, Michael Leiblein, Dan cepts, Techniques, Applications (3rd edn). Blackwell
Levinthal, Rich Makadok, Anita McGahan, Patrick Business Press: Cambridge MA.
Morton, Mary O’Sullivan, Subi Rangan, Miles Griliches Z. 1961. Hedonic price indexes for automobiles:
an econometric analysis of quality change. In Price
Shaver, and Louis Thomas for their helpful com- Statistics of the Federal Government . National Bureau
ments. of Economic Research: New York; 173–196.
Intel Corporation. 1998. Letter to our stockholders.
Annual Report.
REFERENCES Kim C, Mauborgne R. 1997. Value innovation: the
strategic logic of high growth. Harvard Business
Adner R. 2002. When are technologies disruptive: a Review 75(1): 102–112.
demand-based view of the emergence of competition. Lippman SA, Rumelt RP. 1982. Uncertain imitability:
Strategic Management Journal 23(8): 667–688. an analysis of inter-firm differences in efficiency
Adner R. 2004. A demand-based perspective on technol- under competition. Bell Journal of Economics 13(3):
ogy lifecycles. In Business Strategy Over the Industry 418–438.
Lifecycle: Advances in Strategic Management , Vol. 21, Lippman SA, Rumelt RP. 2003. A bargaining perspective
Baum J, McGahan A (eds). Elsevier: Oxford; 25–43. on resource advantage. Strategic Management Journal
Adner R, Levinthal D. 2001. Demand heterogeneity 24(11): 1069–1086.
and technology evolution: implication for product MacDonald G, Ryall M. 2004. How do value creation
and process innovation. Management Science 47(5): and competition determine whether a firm appropriates
611–628. value? Management Science 50(10): 1319–1333.
Adner R, Zemsky P. 2002. Strategy dynamics through Makadok R. 1998. Can first-mover and early-mover
a demand-based lens: the evolution of market advantages be sustained in an industry with low
boundaries, resource rents, and competitive positions. barriers to entry/imitation? Strategic Management
INSEAD working paper series 2003/01/SM. Journal 19(7): 683–696.
Adner R, Zemsky P. 2005. Disruptive technology and the Makadok R. 2001. Toward a synthesis of the resource-
emergence of competition. Rand Journal of Economics based and dynamic-capability views of rent creation.
36(2): 229–254. Strategic Management Journal 22(5): 387–401.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
236 R. Adner and P. Zemsky
Makadok R, Barney J. 2001. Strategic factor market does not occur. Conversely, if Ac1 > 0, then there
intelligence: an application of information economics exists a tE such that Ad1m (t) + Ac1 > 0 iff t > tE
to strategy formulation and competitor intelligence.
where tE satisfies am (b1 tE )β − am (b2 (tE + h2 ))β +
Management Science 47(12): 1621–1638.
McGahan A, Porter ME. 2003. The emergence and sus- Ac1 = 0. Using this equality and the implicit func-
tainability of abnormal profits. Strategic Organization tion theorem,
1: 109–115.
Pacheco de Almeida G, Zemsky P. 2003. The effect ∂tE (b1 tE )β − (b2 (tE + h2 ))β Ac1 /am
of time-to-build on strategic investment under =− =
∂am ∂Ad1m /∂t ∂Ad1m /∂t
uncertainty. Rand Journal of Economics 34(1): (12)
166–182.
Pacheco de Almeida G, Zemsky P. 2004. Time compres- which is positive since Ac1 > 0. Similarly, we have
sion diseconomies and sustainable competitive advan-
tage. Mimeo, INSEAD. am (b1 tE )β ln(b1 tE ) − (am (b1 tE )β
Peteraf MA. 1993. The cornerstones of competitive ∂tE +Ac1 ) ln(b2 (tE + h2 ))
advantage: a resource-based view. Strategic Manage- =− (13)
ment Journal 14(3): 179–191. ∂β ∂Ad1m /∂t
Porter ME. 1980. Competitive Strategy. Free Press: New
York. which is positive since Ac1 > 0 and ln(b2 (tE +
Porter ME. 1991. Towards a dynamic theory of strategy. h2 )) > ln(b1 tE ) > 0 as we are assuming that
Strategic Management Journal , Winter Special Issue b1 tE > 1 so that DMU is increasing in β. Q.E.D.
12: 95–118.
Porter ME. 1996. What is strategy? Harvard Business
Review 74(6): 61–79. Proof of Proposition 2
Priem RL, Butler JE. 2001. Is the resource-based ‘view’ a
useful perspective for strategic management research? Suppose b1 > b2 . Since Ad1m (0) + Ac1 < 0, ∂Ad1m /
Academy of Management Review 26(1): 22–40. ∂t > 0 and limt→∞ Ad1m (t) = ∞, there exists a
Rumelt RP. 1984. Towards a strategic theory of the firm. tE > 0 such that Ad1m (t) + Ac1 > 0 iff t > tE where
In Competitive Strategic Management, Lamb R (ed). tE satisfies am (b1 tE )β − am (b2 (tE + h2 ))β + Ac1 =
Prentice-Hall: Englewood Cliffs, NJ; 556–570.
Rumelt RP. 1987. Theory, strategy and entrepreneurship.
0. Thus, ∂tE /∂am is given by Equation 12, which
in The Competitive Challenge: Strategies for Industrial is positive iff Ac1 > 0. Finally, ∂tE /∂β is given
Innovation and Renewal , Teece D (ed). Ballinger: by Equation 13, which is positive if Ac1 > 0 since
Cambridge, MA; 137–158. in this case b1 tE < b2 (tE + h2 ), while conversely
Rumelt RP. 2003. What in the world is competitive ∂tE /∂β < 0 if Ac1 < 0 since in this case b1 tE >
advantage? UCLA, policy working paper 2003-105.
b2 (tE + h2 ). Q.E.D.
Teece DJ, Pisano G, Shuen A. 1997. Dynamic capabili-
ties and strategic management. Strategic Management
Journal 18(7): 509–534. Proof of Proposition 3
Trajtenberg M. 1990. Economic Analysis of Product
Innovation: The Case of CT Scanners. Harvard Suppose b1 < b2 . Then Ad1m (t) < 0 for all t ≥ 0
University Press: Cambridge, MA. and there exists a t ∗ such that ∂Ad1m (t)/∂t > 0 iff
Utterback JM. 1994. Mastering the Dynamics of Inno- t < t ∗ . If Ac1 < −Ad1m (t ∗ ), firm 1 never enters the
vation: How Companies Can Seize Opportunities in
the Face of Technological Change. Harvard Business segment. If Ac ≥ −Ad1m (t ∗ ) then firm 1 enters the
School Press: Boston, MA. segment at some ta ∈ (0, t ∗ ]. As limt→∞ Ad1m (t) =
Utterback JM, Abernathy W. 1975. A dynamic model −∞, entry is reversed at some time tb > t ∗ .
of process and product innovation. Omega 3(6): We have that ∂tE /∂am and ∂tE /∂β are given by
639–656. Equations 12 and 13 with tE replaced by tj for j =
Wernerfelt B. 1984. A resource-based view of the firm.
Strategic Management Journal 5(2): 171–180. a, b. The numerators of both equalities are posi-
tive since Ac1 > 0. Since ∂Ad1m (ta )/∂t > 0, we have
∂ta /∂am > 0 and ∂ta /∂β > 0. Since ∂Ad1m (tb )/∂t >
APPENDIX 0, we have ∂tb /∂am < 0 and ∂tb /∂β < 0. Hence,
the interval of entry [ta , tb ] is falling in am and β.
Proof of Proposition 1 Q.E.D.
and is never displaced. We have that ti is defined the period during which rents increase is shorter
by vim (ti ) = 0. From Lemma 1, we have that firm with imitation and no DMU than with DMU and
1’s rent from its resource is 0 if t ≤ t1 , sm v1m (t) no imitation.
if t ∈ (t1 , t2 ], and sm (v1m (t) − v2m (t)) if t > t2 . Suppose firm 1 has a process resource. With
For all resource types, v1m (t) > 0 as performance imitation and no DMU, we have that v1m (t) = b
increases in t and hence rents are always increas- and with DMU and no imitation we have that
ing for t ∈ (t1 , t2 ). For a process resource, v1m (t) −
v1m (t) = bβ/(bt)1−β , which is less than b when
v2m (t) = Ac1 such that rents are constant for t > t2 . t > t1 . With DMU and no imitation, we have that
For all other resources, v1m (t) − v2m (t) = Ad1m (t), t1 = c11/β /b and t2 = c21/β /b and hence t2 − t1 =
which varies with time. For performance and c21/β /b − c11/β /b, which is decreasing in β. Hence
timing resources, ∂Ad1m /∂t < 0 and rents decline no DMU (β = 1) and imitation results in a smaller
for t > t2 with limt→∞ Ad1m (t) = 0. For innovation t2 − t1 than DMU and no imitation. This completes
resources, ∂Ad1m /∂t > 0 and rents increase over the proof of Proposition 8. Q.E.D.
time but at a decreasing rate since ∂ 2 Ad1m /∂t 2 < 0.
Q.E.D.
Proof of Propositions 10 and 11
Proof of Propositions 7–9 Suppose that firm 1 has two resources which
we index by ρ = A and B. As usual, let v1 (t)
For all four resource types, Ad1m (t) + Ac1 > 0 for
be firm 1’s value creation. Define v1A (t) and
all t > 0 and hence firm 1 enters the segment
v1B (t) as firm 1’s value creation when it only has
first and is never displaced. We have that ti is
resource A and B respectively. Define the follow-
defined by vim (ti ) = 0. From Lemma 1, we have
ing critical times: v1 (t1 ) = 0, v2 (t2 ) = 0, v1A (t1A ) =
that firm 1’s rent from its resource is 0 if t ≤ t1 ,
0, v1B (t1B ) = 0 and then define t1M = max{t1A , t1B }
sm v1m (t) if t ∈ (t1 , t2 ], and sm (v1m (t) − v2m (t)) if
and t1m = min{t1A , t1B }. Note that t1 < t1m ≤ t1M <
t > t2 . For all resource types, v1m (t) > 0 as perfor-
t2 . We now consider resource interactions for var-
mance increases in t and hence rents are always
ious values of t.
increasing for t ∈ (t1 , t2 ). For a process resource,
For t < t1 even the firm with both resources has
v1m (t) − v2m (t) = c U (t) such that rents decrease
negative value creation and there are no rents; the
to zero for t > t2 . Similarly, for a product resource
resources are additive. For t ∈ (t1 , t1m ), we have
we have that v1m (t) − v2m (t) = r U (t) and rents
that possession of both resources yields positive
decrease to zero for t > t2 . This completes the
value creation and rents, while possession of either
proof of Proposition 7.
individually results in negative value creation and
For innovation resources, the willingness t to pay
no rents; hence any two resources are superaddi-
for firm 2’s offer at time t is given by 0 (b + (1 −
tive. For t ∈ [t1m , t1M ], there is one resource that
U (t))b )dt = (b + b )t − zb ln(1 + zt). We yields rents with or without the other. However,
then have v1 (t) − v2 (t) = zb ln(1 + zt), which is value creation and rents are greater with both
increasing in t but at a decreasing rate because of resources and the second resources has negative
the concavity of the ln function. This completes value creation and no rents on its own; hence
the proof of Proposition 9. any two resources are superadditive. What hap-
Suppose firm 1 has a process resource. With pens for t > t1M depends on the identity of the
imitation and no DMU, we have that v1m (t) = b two resources.
and with DMU and no imitation we have that Suppose that the firm has a product and a timing
v1m (t) = bβ/(bt + r )1−β . For t > t1 we have that resource. When t ≥ t2 , then rent to holding both
bt + r > 1 and hence bβ/(bt + r )1−β < b and resources is sm (v1 (t) − v2 (t)) = sm am (b(t + h ) +
the rate of increase in rents is greater with imitation r )β − (bt)β ), while the sum of the rents from
and no DMU than with DMU and no imitation. holding both individually is sm am ((bt + r )β −
With DMU and no imitation, we have that t1 = (bt)β ) + sm am (b(t + h ))β − (bt)β ). For r = 0,
(c1/β − r )/b and t2 = c1/β /b, and hence t2 − t1 = these expressions are equal and the second expres-
r /b. This difference is independent of β and sion increases faster in r ; hence this resource
hence it holds when there is no DMU and no combination is subadditive for t ≥ t2 . For t ∈
imitation. As imitation reduces t2 , we have that [t1M , t2 ], the difference in rents between holding
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
238 R. Adner and P. Zemsky
the resources together and holding each individu- Proof of Propositions 16 and 17
ally is sm (am (b(t + h ) + r )β − c) − sm (am (bt +
r )β + am (b(t + h ))β − 2c), which is falling in Suppose that sL = 0 and let aH = a and sH = s.
t. Thus there exists a tˆ ∈ (t1M , t2 ) such that for With only a single segment, at most one firm
t1 < t < tˆ the resources are superadditive and for will pay the fixed costs to maintain resources.
t > tˆ the resources are subadditive. Let vc (t) = a(bt)β − c + c be the value created
Similar arguments establish the existence of a by a Cost Leader; let vr (t) = a(bt + r )β − c be
critical tˆ for a portfolio with a product and inno- the value creation by a Differentiator; and let
vation resource as well as for a portfolio with two vrc (t) = a(bt + r )β − c + c be the value cre-
timing, two product, or two innovation resources. ation by a Resource Generalist. Firm 2 starts
Suppose that resource A is a process resource. creating value at time t2 = c1/β /b. From previ-
For t ∈ [t1M , t2 ), the rents from holding both re- ous analysis, the rents from the Cost Leadership
sources are sm v1 (t) = sm (v1B (t) + c ) while the strategy are maximized for t ≥ t2 and are ¯c =
rents from holding both individually are (recall- sc − Fc , and the rents from the Differentiator
ing that only one resource gives positive value strategy are maximized at t = t2 and are equal
creation on its own) sm max{v1B (t), v2 (t) + c } < to ¯ r = s(a(c1/β + r )β − c) − Fr . Suppose that
sm (v1B (t) + c ); hence the resources are super- ¯
r > ¯ c > 0.
additive in this range. For t ≥ t2 , the rent from Suppose that t ≤ t2 . The firm chooses the strat-
holding both resources is sm (v1B (t) + c − v2 (t)), egy at each point in time that maximizes its profits
the same as the sum of the rents from holding both where a strategy without any resources yields prof-
individually, and the resources are additive. Hence, its of 0. Being a Cost Leader yields svc (t) − Fc ,
for t1 < t < t2 the resources are superadditive and being a Differentiator yields svr (t) − Fr , and being
for t ≥ t2 the resources are additive. a Resource Generalist yields svrc (t) − Fc − Fr −
The final case to consider is a timing and K. Since vc (t) > vr (t) and ¯r > ¯ c , we have that
an innovation resource. For t ≥ t2 , the differ- a strategy of Differentiation dominates Cost Lead-
ence in rents between holding both and hold- ership for t ≤ t2 . Subtracting the profits from being
ing each separately is sm am ((b + b )(t + h ))β − a Resource Generalist from those of being a Differ-
(b(t + h ))β − ((b + b )t)β + (bt)β ), which is entiator yields ¯ c − K and hence for K < ¯ c the
equal to zero for b = 0 and increasing in b . firm enters the market segment as a Resource Gen-
Hence, the difference is positive and the resources eralist and for K > ¯ c the firm enters the market
are superadditive. It is straightforward to show that segment as a Cost Leader.
the difference in rents is monotonic in t for t ∈ Suppose t > t2 . Being a Cost Leader yields
[t1M , t2 ]. Hence, timing and innovation resources ¯ c , being a Differentiator yields sa((bt + r )β −
are superadditive for all t > t1 . Q.E.D. (bt)β ) − Fr , and being a Resource Generalist
The proofs of Propositions 12, 13 and 14 are yields sa((bt + r )β − (bt)β + c ) − Fc − Fr
straightforward and are omitted. − K. The latter two are both falling in t and con-
verge to a value of less than ¯ c and hence in
the long run the firm pursues a strategy of Cost
Proof of Proposition 15
Leadership. This completes the proof of Proposi-
Suppose that firm 2 (the incumbent) is in both tion 16.
segment H and L at some point in time but For K > ¯ c , the firm starts maintaining the
that firm 1 displaces it from these segments. The product resource when svr (t) − Fr = 0 and the
proofs of Propositions 1, 2, and 3 characterize start time is decreasing in β and a. The firm stops
the effect of am on the time at which firm 1 maintaining the product resource when sa((bt +
displaces firm 2 from segment m and show that r )β − (bt)β ) − Fr = ¯ c and the stop time is
the entry time is increasing when firm 1 has a cost increasing in β and a. For K ≤ ¯ c , the firm starts
advantage and decreasing when firm 1 has a cost maintaining the product resource when svrc (t) −
disadvantage. Since aH > aL , firm 1 first displaces Fr − Fc − K = 0 and the start time is decreasing
the incumbent from the high-end segment if it has in β and a. The firm stops maintaining the prod-
a cost disadvantage and from the low-end segment uct resource when sa((bt + r )β − (bt)β ) − Fr −
if it has a cost advantage. Q.E.D. K = 0 and the stop time is increasing in β and a.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)
Demand-Based Perspective 239
For K > ¯ c , the firm starts maintaining the pro- does not create value in the low end, which occurs
cess resource when it stops maintaining the product for aL (bt + r )β − c < 0. Then the profits of the
resource and hence the start time for the product Cost Leader are sL (aL (bt)β − c + c ) − Fc , which
resource is decreasing in β and a. For K < ¯ c, are non-negative for
the firm starts maintaining the process and product
resource at the same time and hence the start time Fc c − c
aL ≥ a L = β
+ (16)
for the product resource is increasing in β and a. sL (bt) (bt)β
Q.E.D.
A Differentiator does not increase its prof-
its from shifting to a Resource Generalist strat-
Proof of Propositions 18 and 19 egy if sH (aH D − c ) − Fr ≥ sH aH D + sL aL D −
Formally, we consider the one-shot game where Fr − Fc − K or K ≥ sH c + sL aL D − Fc . Sup-
firms simultaneously choose their resources. For pose that the Differentiator creates value in the
it to be a Nash equilibrium that one firm is a low end. A Cost Leader does not increase its prof-
Differentiator and the other is a Cost Leader, it its from shifting to a Resource Generalist strat-
must be that each firm has non-negative prof- egy if sL (c − aL D) − Fc ≥ sL c + sH c − Fc −
its and that neither can increase its profits by Fr − K or K ≥ sH c + sL aL D − Fr . Suppose that
pursuing a Resource Generalist strategy. As the the Differentiator does not create value in the low
Differentiator has more added value in the high end. A Cost Leader does not increase its profits
end than the low end, we can restrict attention to from shifting to a Resource Generalist strategy if
parameters for which the Differentiator has pos- sL (aL (bt)β − c + c ) − Fc ≥ sL (aL (bt + r )β
itive added value in the high end and the Cost − c + c ) + sH c − Fc − Fr − K or K ≥ sL aL D
Leader has positive added value in the low end. + sH c − Fr . Hence, a necessary condition for
Note that if the Cost Leader has added value in strategic diversity to be a Nash equilibrium is
the low end then it has positive value creation in that K ≥ K̄ = max{0, sH c + sL aL D − Fc , sH c
both segments, while it is not the case that the + sL aL D − Fr }. Note that it is possible that K̄ = 0.
Differentiator necessarily has positive value cre- So far we have been concerned with the con-
ation in the low end. Define D = (bt + r )β − ditions for the existence of an equilibrium with
(bt)β . a Cost Leader and Differentiator. These condi-
The profits of the Differentiator are sH (aH D − tions do not assure uniqueness. For uniqueness one
c ) − Fr and these are non-negative for needs to rule out the equilibrium where one firm
is a Resource Generalist and the other does not
c Fr invest in resources. The pay-off to the resourced
aH ≥ a H = + (14) firm in this case is sH (aH D + c ) + sL (aL D +
D sH D
c ) − Fc − Fr − K. Note that for K sufficiently
which is increasing in Fr and decreasing in sH . small, the conditions aH ≥ a H and aL ≥ a L assure
If the Differentiator has positive value creation in that the Resource Generalist equilibrium exists
the low-end segment, then the profits of the Cost as well. Hence, for Cost Leadership and Dif-
Leader are sL (c − aL D) − Fc and these are non- ferentiation to be the unique Nash equilibrium
negative for we require that K > sH (aH D + c ) + sL (aL D +
c ) − Fc − Fr , which is greater than zero when-
c Fc ever the Cost Leader–Differentiator equilibrium
aL ≥ āL = − (15)
D sL D exists. Q.E.D.
Copyright 2006 John Wiley & Sons, Ltd. Strat. Mgmt. J., 27: 215–239 (2006)