Received: 18 December 2019 Revised: 3 September 2021 Accepted: 10 September 2021

DOI: 10.1002/smj.3339

SPECIAL ISSUE ARTICLE

A formal framework for the RBV: Resource

dynamics as a Markov process

Phebo D. Wibbens

INSEAD, Strategy Area, Boulevard de

Constance, Fontainebleau Cedex, France Abstract
Research Summary: This article introduces a formal
Correspondence framework for long-term resource dynamics under
Phebo D. Wibbens, INSEAD, Strategy
uncertainty. Firms' resource positions are identified
Area, Boulevard de Constance,
Fontainebleau Cedex 77305, France. with states in a stochastic Markov process. Two stylized
Email: [email protected] examples implementing the framework illustrate how
it can yield new strategic insights. One such insight is
the attractiveness paradox. Shorter-lived resources
might seem less attractive for the simple reason that
they will yield returns for a shorter time period. How-
ever, the model in this article shows that when
accounting for the full competitive dynamics such
seemingly less attractive resources can sometimes yield
a more attractive investment opportunity. The ultimate
goal of introducing the Markov framework is to provide
new directions for resource-based view research based
on formal and simulation models, while retaining the
verbal theory's richness and intuitive appeal.
Managerial Summary: A firm's resources (the assets
and capabilities it has access to) vis-à-vis its competi-
tors are the equivalent of the positions of pieces on a
chessboard. They represent the “state” of the game,
and each player tries to optimize this state in its favor.
What sets apart the game of business from the game of
chess is uncertainty: while in chess making a move
leads to a certain outcome on the board, making an
investment usually does not carry such guarantees,
especially if it is in an intangible resource such as tech-
nology. This article introduces the mathematical tools
to study strategic investments in the game of business,

1562 © 2021 John Wiley & Sons Ltd. wileyonlinelibrary.com/journal/smj Strat Mgmt J. 2023;44:1562–1586.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1563

inherently incorporating uncertainty and long-term

competitive dynamics.

KEYWORDS
formal foundations of strategy, Markov models, resource-based
view, stochastic processes, value appropriation

1 | INTRODUCTION

Imagine you are a manager and need to make a decision to invest in either of two markets. The
key difference between them is the speed at which the resources needed to compete—technolo-
gies, brands, plants—are expected to become obsolete: 3 years on average in one market and 10
in the other. After this time, you will need to make new investments in the next generation of
resources to serve the customers in that market. Both markets are similar on all other dimen-
sions, including consumer demand, cost curves, and expected level of competition. Which of
these two markets is the most attractive to invest in?
This might appear to be a simple decision. In absence of competition it would be: longer
lasting resources with the same initial cost and same annual pay-offs are more valuable than
shorter lasting ones for the simple reason that more years of pay-offs means more money in the
bank. However, resources with more valuable pay-offs will generally attract more investment.
This can increase the cost and decrease the pay-offs. Could these competitive dynamics reverse
the attractiveness of the resources, making the market with shorter-lived resources actually the
more attractive one, even in the absence of new entrants?
Although these kinds of attractiveness questions are at the heart of the resource-based view
(RBV), extant literature provides surprisingly few tools to answer them quantitatively. On the
one hand, the seminal papers have used fundamental principles about competitive dynamics
under uncertainty to derive broad managerial guidance, such as the importance of valuable,
rare, inimitable, non-substitutable (VRIN) resources for sustained competitive advantage
(Barney, 1991). These verbal papers are not very suitable though to answer quantitative ques-
tions such as those posed above. On the other hand, formal models have become increasingly
important in the RBV to clear up specific mechanisms such as the role of investment in infor-
mation (Makadok & Barney, 2001), demand-side factors (Adner & Zemsky, 2006), the timing of
resource investments (Pacheco-de-Almeida & Zemsky, 2007), and strategic factor markets
(Chatain, 2014). They provide a solid logical foundation and can be used to derive quantitative
results. However, current models do not integrate the broad principles of the seminal papers.
For instance, many formal RBV models are based on value-based strategy (VBS;
Brandenburger & Stuart, 1996; Lippman & Rumelt, 2003a; Gans & Ryall, 2017). Though VBS
provides a powerful framework for modeling competitive interactions, by itself it does not natu-
rally incorporate long-term investment dynamics and uncertainty, two core tenets of the RBV
that play a critical role in assessing long-term value and thus market attractiveness.
In this article, I introduce a formal framework for resource dynamics with the aspiration of
providing a logically coherent, quantitative foundation for the RBV while preserving the intui-
tive appeal from the seminal works. My central premise is that resource positions of competing
firms in an industry can be identified with states in a Markov model. This premise captures the
idea that a firm's resource base consists of anything that can affect its future evolution in an
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1564 WIBBENS

uncertain world, formalizing broad conceptions of resources such as those in Wernerfelt (1984)
and Helfat et al. (2007). Markov models have been widely used in other disciplines to describe
stochastic processes, but so far have garnered relatively little attention in strategic management
despite the increasing importance of stochastic methods (Denrell, Fang, & Liu, 2015). I adapt
the Markov framework to the RBV context and provide two stylized examples implementing it.
One of the examples provides an answer to the opening question of this article. Counter to
what one might naïvely expect, the market with the shorter-lived resource can be more attrac-
tive due to the competitive dynamics of the sunk cost investments. This is an instance of what
I call the attractiveness paradox: resource investments that appear attractive based on direct
pay-offs might turn out to be less attractive after taking the full competitive dynamics into
account.
The Markov framework allows a deeper investigation into the subtle mechanisms behind
the attractiveness paradox. In this case, there are three effects due to the competitive dynamics
that jointly more than offset the a priori higher expected value of pay-offs for the longer-lived
resource. First, the rationally optimal investment level in the longer-lived resource is higher,
induced by the longer pay-offs. Second, the value of gaining the resource is smaller than might
be expected because there is a higher chance of the resource at some point losing its value due
to the follower obtaining it too. Third, there is a strong loss of value if the competitor gains the
resource first because it will likely take a longer time before the focal firm can successfully chal-
lenge its competitor. The model shows that these three effects jointly can more than offset the
higher value of the longer-lived resources pay-offs. Section 3.5 provides further analysis and dis-
cussion of these mechanisms.
At a deeper level, long-term competitive (re-)investment dynamics are responsible for the
attractiveness paradox. A well-known cousin of this paradox occurs for investments with
unrestricted entry: the more potential profits a market has to offer, the more new entrants that
will invest, the less attractive the market becomes. What sets apart the paradox in this article is
that it occurs even with restricted entry. The reason that seemingly more profitable markets can
turn out to be less attractive even with a fixed number of competitors is that investment is not a
simple yes/no decision decided at market entry. Instead, investments can be set at different
levels and can be adjusted based on changing competitive conditions, for instance when a com-
petitor obtains a resource. Additionally, resources can depreciate over time, thus requiring re-
investments leading to another round of competition to build a favorable resource position. The
Markov framework presented here is uniquely suited to study this continuous dynamic of
changing competitive positions and (re-)investments, exposing the subtle mechanisms leading
to the attractiveness paradox.
Resource depreciation is just one example of how the Markov framework can be used to
study the impact of competitive dynamics on resource attractiveness. Other resource character-
istics that can be incorporated include rareness, imitability, substitutability (Barney, 1991), asset
mass efficiencies, interconnectedness of asset stocks (Dierickx & Cool, 1989), and dynamic
capabilities (Teece, Pisano, & Shuen, 1997). Unlike earlier models, the Markov framework
allows studying the effects of these characteristics on long-term competitive dynamics under
uncertainty.
Finally, the Markov framework would be well suited to incorporate bounded rationality
(Simon, 1955), which current formal RBV models typically do not since they are based on eco-
nomic models assuming rational actors. The example models in this article still assume fully
rational actors with perfect information about the competitive environment. Doing so helps
keep the model parsimonious since no further assumptions need to be made about their
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1565

behavior other than that they optimize long-term value. Moreover, the simple formal model
could by itself be seen as a behavioral, “small-world” representation of a complex competitive
environment (Levinthal, 2011). Nevertheless, firms' expectations about future value will often
deviate from the truth and it would be interesting to understand the impact of such limits to
perfect information on competitive dynamics. Because the Markov framework does not incor-
porate any a priori assumptions about rationality, it is better suited to incorporate more behav-
ioral approaches than typical economic modeling frameworks. In Section 3.6, I will briefly
discuss several of the framework's possible extensions, including how it can incorporate more
behavioral assumptions.
Ultimately, I see the primary contribution of this article as the specific implementation and
mathematical language of the Markov framework as described in Section 3.1. Many earlier
models in strategy have used Markov processes, often implicitly. For instance, a one-period
model can be trivially described in terms of a Markov process, as is the case for most multi-
period games. Unlike earlier models though, the framework in this article emphasizes that the
game of business usually has an infinite horizon (it cannot be described using a fixed number
of periods) and is fundamentally uncertain. My goal for this article has been to incorporate
those two fundamental aspects in an elegant mathematical framework that can be used to study
practical competitive situations. Perhaps the most useful result in the framework is a version of
the Bellman equation (Equation [9]) that links profit, investment, and value in a competitive
game with infinite horizon, as it defines a rigorous and unambiguous notion of “value” within
the RBV.

2 | P R I O R L I T ER A T U R E

2.1 | Foundations of the RBV

While Penrose (1959) introduced many of the RBV's fundamental ideas, a set of highly influen-
tial papers in the 1980s and early 90s have cemented it as a foundational pillar for the strategy
field. These works include Rumelt (1984); Wernerfelt (1984); Barney (1986); Dierickx and
Cool (1989); Barney (1991) and Peteraf (1993). Though each contribution emphasizes different
aspects of the RBV, three fundamental themes run as a thread across them:

1. Heterogeneity. The RBV's raison d'être is performance heterogeneity under competition. Even
firms within the same industry with similar competitive positions can exhibit sustained per-
formance differences (Mueller, 1977; Rumelt, 1991), contrasting with the traditional view in
neo-classical economics that performance differences among direct competitors should be
rapidly competed away. For such differences to exist, the RBV posits that firms must exhibit
heterogeneity in their access to resources: assets and capabilities that allow some firms to
consistently outperform others.
2. Dynamics. The RBV is fundamentally a dynamic theory. To explain why performance
differences can persist over time, resources must be immobile due to isolating
mechanisms (Rumelt, 1984), or alternatively a combination of rarity, inimitability and
non-substitutability (Barney, 1991). Additionally, the RBV needs to account for
how firms can develop heterogeneous resource positions in the first place. The two pri-
mary mechanisms described in the literature are through the workings of strategic
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1566 WIBBENS

factor markets (Barney, 1986) and the resource accumulation process (Dierickx &
Cool, 1989).
3. Uncertainty. Finally, uncertainty plays a fundamental role in resource dynamics, in at least
three ways. First, uncertainty allows firms having heterogeneous expectations about the
value of resources, leading to strategic factor market imperfections: because a firm can have
superior information about the value of a resource, it might obtain a resource below its
“true” value, leading to above-normal returns (Barney, 1986). Second, in addition to uncer-
tainty about external resources, even internally a firm's own resource position might be
unobservable; such causal ambiguity can by itself already lead to sustained performance het-
erogeneity (Lippman & Rumelt, 1982). Third, resource positions can change due to exoge-
nous shocks, or “luck” (Barney, 1986).

These tenets form three foundational pillars of the RBV and it is hard to imagine a logically
coherent description of the relation between resources and persistent performance differences
that does not incorporate these three elements.

2.2 | Critiques and responses

Though the RBV gained significant traction in the scholarly community during the 1990s, at
the same time, several scholars started raising fundamental critiques on the RBV's logical struc-
ture. For instance, Priem and Butler (2001, p. 22) contended that “imprecise definitions hinder
prescription and static approaches relegate causality to a ‘black box’.” Similarly, Foss and
Knudsen (2003, p. 292) stated “that the logical structure of the RBV appears to be unclear.”
Other scholars have echoed similar critiques (Bromiley & Fleming, 2002; Hoopes, Madsen, &
Walker, 2003). Though several responses and clarifications have ensued (Barney, 2001;
Peteraf & Barney, 2003), some scholars remain unconvinced that all critiques have been fully
resolved (Kraaijenbrink, Spender, & Groen, 2010).
In particular, the meaning of “value” has remained troublesome to some scholars, because
Barney (1991) uses it not only as resource characteristic needed to achieve competitive advan-
tage but also to define the resulting competitive advantage itself. Peteraf and Barney (2003)
attempt to circumvent this issue by defining competitive advantage explicitly as “the difference
between the perceived benefits gained by the purchasers of the good and the economic cost to
the enterprise” (p. 314), a definition in the spirit of VBS (Brandenburger & Stuart, 1996; Gans &
Ryall, 2017). However, using this definition it is fairly easy to come up with examples of
resources that can yield sustained higher profits but that would not constitute a “competitive
advantage,” which seems to defeat the purpose of such a definition (Lieberman, 2021;
Siggelkow & Wibbens, 2015).
Formal models are ideally suited to address critiques such as the above on logical struc-
ture, as they provide precise assumptions and reasoning, in unambiguous mathematical lan-
guage (Adner, Polos, Ryall, & Sorenson, 2009; Knudsen, Levinthal, & Puranam, 2019; Oxley
et al., 2010). Indeed, several more formal approaches to structure the foundations of the RBV
have ensued. A number of scholars have proposed VBS as a formal framework for studying
the RBV (e.g., Gans & Ryall, 2017; Lippman & Rumelt, 2003a). Introduced to strategy by
Brandenburger and Stuart (1996), VBS describes competitive interactions in terms of
coalitional economic games and has become one of the most widely used formal frameworks
in the field. Peteraf and Barney (2003) as well as Schmidt and Keil (2013) also build RBV
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1567

arguments on VBS concepts, defining the economic value of resources in terms of customer
benefits (or willingness-to-pay) and economic cost, clearly separating it from the resulting
rent creation.
Biform games extend product market competition in VBS with a first-stage non-coalitional
game representing “moves to try to shape the competitive environment in a favorable way”
(Brandenburger & Stuart, 2007, p. 537). From a resource-based perspective, this first stage
would represent sunk cost investments in resource positions. For instance, in an innovation
game (Example 2.2 in Brandenburger & Stuart, 2007) firms can invest in a technological
resource that increases consumers' willingness-to-pay.
However, these VBS formalizations do not capture the long-term dynamics and uncertainty
of competition, two of the three RBV's core tenets in Section 2.1. This is also true for biform
games, which merely represent two stages instead of a long sequence of path-dependent invest-
ments with uncertainty.
Real option theory (Kogut & Kulatilaka, 2001; McGrath, 1997), another important formal
framework in strategy, does not capture all core tenets of the RBV either. Though real options
do incorporate long-term dynamics and uncertainty, they are typically applied to investments
by a single firm. Therefore, this framework does not allow studying the emergence of heteroge-
neity under competition.

3 | FORMAL AN ALY S I S

3.1 | General framework

To incorporate long-term dynamics and uncertainty into models of competition such as VBS,
one could envision a long sequence of biform games with uncertainty. Sunk cost investments
shape the probability distribution of the resulting competitive outcomes, which in turn shape
the future investment opportunities. This idea can be formalized using Markov models. In a
Markov process describing industry dynamics, a stochastic variable X(t) captures the resource
positions of all relevant firms in an industry at each time t. The industry state X(t) determines
both the competitive outcomes and the investment opportunities of firms, and thus the proba-
bility distribution of industry states X(t + Δt) at a later time.
At its essence, a Markov process is a stochastic process in which causality moves forward in
time. Formally, for any sequence of consecutive time points t1 < t2 < …tr < t, the Markov prop-
erty in terms of the probability distribution of X(t) conditional on the prior time points is:

X ðt Þ j X ðt1 Þ, …, X ðtr Þ=X ðt Þ j X ðt r Þ ð1Þ

In other words, when knowing the latest state X(tr), no additional information about the
future can be gained from knowledge about prior states X(t1), …, X(tr−1). The Markov property
might seem restrictive, but all that it requires is that relevant information about the future is
encoded in the state variable X(t) and not in the equations that describe the probability evolu-
tion of X(t). Section 3.6 provides an overview of how relevant strategic variables can be incorpo-
rated in a process X(t) such that it obeys the Markov property. Markov models are widely used
across scientific areas, including economics (Ericson & Pakes, 1995), computer science
(Rabiner & Juang, 1986), biology (Eddy, 2004), marketing (Netzer, Lattin, & Srinivasan, 2008)
and operations management (Schweidel, Bradlow, & Fader, 2011).
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1568 WIBBENS

In a strategy context, X(t) can represent an industry state consisting of a matrix encoding
the relevant resource states for all incumbents and potential entrants:
0 1
Firm 3
2 1 Firm n
2 3
B Resource 1 X 11 X 1n C
B C
B 6 7 6 7C
B Resource 2 6 X 21 7 6 X 2n 7 C
B
X =B 6 7 6 7C ð2Þ
6 7 6 7C
B .. 6 .. 7   6 .. 7 C
B . 6 7 6 . 7C
@ 4 . 5 4 5A
Resource J X J1 X Jn

Each row of this matrix encodes the extent to which a firm has access to a specific type of
resource. Each column represents a specific firm's resource bundle. The set of all possible values
for X is called the state space. This view of resources is closely related to the idea of viewing
firms and their assets as states in evolutionary systems (Winter, 1987).
I will make three additional assumptions about the structure of the process X(t). These
assumptions simplify the analysis without putting major restrictions on the types of strategic sit-
uations that can be analyzed.
First, I assume that the conditional probabilities in Equation (1) are symmetric across firms,
meaning that interchanging any two firms does not alter any of the dynamic equations. This
assumption is not very restrictive, since it simply requires that any firm heterogeneity should be
encoded by differences in resources states in Equation (2).
Second, I assume that the conditional probabilities in Equation (1) are homogeneous in
time, meaning that the dynamic equations depend only on the value of X(t) and not explic-
itly on t itself. This implies for instance that the probability distribution of X(t + Δt) condi-
tional on X(t) is the same as that of X(Δt) conditional on X(0). Again, this simply requires
that time-dependent processes are codified through the state space and not in the dynamic
equations.
Third, I assume that the state space is finite, meaning that each resource position can only
take on a finite number of values. In analogy to NK models, one could for instance assume that
each firm's resource base can be described by a long string of 0/1-values (Levinthal, 1997).
Values beyond 0/1 are possible too, as long as there are only finitely many of them. A continu-
ous state space could always be discretized to arbitrary precision to yield a finite one.1 A finite
state space implies that each state X can be labeled by an integer i = 1, …, N.
With a finite state space, the probability distribution at any time t can be described by a vec-
tor P(t), with elements:

Pi ðt Þ= PðX ðtÞ =iÞ ð3Þ

The vector P(t) describes the probability of finding the industry in state X = i at time t.
Going forward I will use bold face variable names to denote vectors on the state space in
this way.

1
The time variable t could be discretized too, but the analysis is often actually simpler if it is kept as a continuous
variable (Doraszelski & Judd, 2012). One advantage of a continuous time variable is that the probability that two events
happen simultaneously is zero. Hence simultaneous events do not need to be accounted for in the model.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1569

A matrix Q called the infinitesimal generator completely characterizes the evolution of the
stochastic process X(t). This follows from the Markov and time-homogeneity assumptions
(Bhat & Miller, 2002, p. 195). The transpose of Q describes the transition rate from one state to
another over an infinitesimally short period:

dP
=QT P ð4Þ
dt

The matrix Q is constant over time. The elements Qij ≥ 0 describe the transition rate of going
from state i to j ≠ i. In other words, for a short period of time Δt the probability of going from i to
j is QijΔt. The matrix elements Qii ≤ 0 describe the rate of moving out of state i, with corresponding
probability jQiiΔtj. These relations are derived in Supporting Information S1 Appendix A.
If the generator can be diagonalized such that QT = MDM−1 with D the diagonal matrix of
eigenvalues, then the evolution of P can be directly calculated using:

P ðt Þ=MeDt M − 1 Pð0Þ ð5Þ

If diagonalization is not feasible,2 the evolution of P can be estimated using matrix multipli-
cation over small time steps Δt:
 n
P ðnΔt Þ ≈ I +QT Δt Pð0Þ ð6Þ

Because of the symmetry across firms, Firm 1 can be designated as the focal firm without
loss of generality. Therefore, all variables of interest such as profits π only need to be defined
for Firm 1. The profits for any other firm follow immediately from exchanging it with Firm 1 in
Equation (2). Since by assumption the state X encodes any relevant information about the
industry, profits can be expressed as a vector π on the state. The same is true for any sunk cost
investments y.
The value of Firm 1 is the discounted sum of all future cash flows over time. Assuming a
discount rate ρ and cash flows π−y over time, the value at time t is:
ð∞
V ðt Þ = ðπ ðsÞ −yðsÞÞe − ρðs − tÞ ds ð7Þ
t

The expected value of V can also be a expressed as a vector V, with elements:

V i =EðV ðt Þ j X ðt Þ =iÞ ð8Þ

Because of time homogeneity, V is independent of t—the above equation holds for any time
t ≥ 0. Given V, the expected value over time can be calculated using the probability evolution:

T
2
Equation (4) has P ðt Þ=eQ t P ð0Þ as formal solution. Direct calculation of the matrix exponential is usually only
practical if diagonalization is feasible. Since the computational time required to do so scales with N3 this will only be
possible for moderately small state spaces. Otherwise the small-step approximation in Equation (6) needs to be used,
which can be faster especially if the generator Q is a sparse matrix. In that equation, I represent the N by N identity
matrix.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1570 WIBBENS

P
EV ðt Þ=V T Pðt Þ = V i Pi ðt Þ. Intuitively, this equation states that the expected value of Firm 1 at
i
time t is the sum of the probability of being in state i times the value of that state. The same is
true for any other dynamic variable; for instance Eπ(t) = πTP(t).
The Bellman equation provides a central relation between profit, investment, and value.
Using the above definitions, it takes the form:

π −y +QV
V= ð9Þ
ρ

This equation states that the value V of a resource position equals the present value of its
cash flows π−y plus the flow from the change in expected value QV resulting from the resource
dynamics. The Bellman equation can be used to analyze a wide range of strategic situations
under uncertainty. Supporting Information S1 Appendix B provides a derivation.
If Q is given, Equation (9) can be used directly to calculate the expected values using matrix
algebra:

V = ðρI −QÞ − 1 ðπ −yÞ ð10Þ

Instead of specifying Q directly, often it is more interesting from a strategic perspective to

specify an endogenous investment model. Given the resource state X, a firm can decide its invest-
ment level y, which influences the transition rate of going to a different state—higher investments
increase the probability of obtaining a certain resource. The generator Q, the matrix of all transi-
tion rates, becomes a function of the investment policy y, the vector specifying Firm 1's invest-
ment level for each state X.3 Figure 1 schematically depicts this situation. At any moment t, the
resource state X(t) determines the profit and investment; for X = i the profit is π i and the invest-
ment yi. The investment policy y determines the generator Q(y), which in turn describes the tran-
sition of the probability distribution from X(t) to X(t + Δt), a short time later.
Rational actors will choose an investment policy y that optimizes the value V in each state.
This situation is called a dynamic game and the solution is a Markov perfect equilibrium
(MPE). It has been extensively studied in the industrial organization (IO) literature following
the landmark paper by Ericson and Pakes (1995). Defining the matrix of partial derivatives
∂Q
Q0ij ðyÞ= ∂yij , the optimal investment policy follows from:
i

Q0 ðyÞV =1 ð11Þ

In this Equation (1) denotes a vector of ones. The equation follows from setting the partial
derivatives in the Bellman equation equal to zero: for the optimal investment policy, any small
change in investment yi should not increase the value. Supporting Information S1 Appendix B
provides a derivation.
The combination of Equations (9) and (11) can then be used to solve simultaneously for
the optimal investment policy y and the expected values V, either analytically or using an iter-
ative procedure. Both approaches will be used in the two examples below, which illustrate
applications of the Markov framework to competitive dynamics. The examples show how

3
Recall that each other firm's investment policy follows directly from Firm 1's policy, because of the assumed symmetry
between firms.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1571

F I G U R E 1 Schematic depiction of resource dynamics as a Markov process. X(t) represents the industry state
at time t, consisting of all relevant resource positions for incumbent firms and potential entrants. Profit π and
investment y for the focal firm are functions of the industry state. They can be represented as vectors if the state
space is finite (i.e., X can take only finitely many values, which can be labeled 1, …, N). The infinitesimal
generator Q is a matrix that determines the evolution of the probability distribution of states over a short period
Δt. In an endogenous investment model, Q is defined as a function of the investment policy y

Markov processes can incorporate many familiar strategic concepts, as well as how the combi-
nation of long-term dynamics and uncertainty can lead to perhaps unexpected competitive
outcomes.

3.2 | Example 1: Competing for a single resource

The first example provides a parsimonious representation of competition for a resource that
leads to sustained competitive advantage. The resource can for instance represent a technology
or a brand name. Though a similar setup has been extensively studied as a patent race in the IO
literature (Fudenberg, Gilbert, Stiglitz, & Tirole, 1983; Loury, 1979), it is useful as an example of
Markov dynamics because of its parsimony and existence of an analytical solution. It also pro-
vides an insightful illustration of the meaning of “value” within the RBV and the potential
caveats in defining it. The second example builds on the first, adding resource depreciation.
This model with merely four industry states leads to remarkably rich competitive dynamics
with several interesting strategic insights.

Example 1. Consider two potential entrants that both can create an annual value of
10 with the customers in a new market segment. A firm needs to obtain a resource to
serve these customers, and then can appropriate 50% of value created, for a profit
π = 5 per year into perpetuity. If both firms obtain this resource, they have zero added
value leading to π = 0. Each firm can invest in the resource at an annual rate y ≥ 0;
this is a sunk cost. Investing in the resource increases the probability rate of obtaining
it, according to:

gy
f ðyÞ = ð12Þ
g +y

Firms rationally optimize their investment y at each moment to maximize the

expected value based on complete knowledge of all parameters and resource states.
Assume g = 0.2/year and a discount rate ρ = 0.1/year.

The function f encapsulates the time-compression diseconomies of resource investments

(Dierickx & Cool, 1989). The functional form of f as specified in Equation (12) is very common
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1572 WIBBENS

F I G U R E 2 Graph of the resource investment function f, showing the probability f(y)Δt of gaining a resource
in a short period Δt as a function of the annual investment rate y, as specified in Equation (12). The shape of
f encapsulates time-compression diseconomies: Twice the investment rate yields less than twice the probability
of obtaining the resource (Dierickx & Cool, 1989). This probability always remains below gΔt, an asymptote of
the investment function

in the IO literature (e.g., Besanko & Doraszelski, 2004; Ericson & Pakes, 1995). A more general
form would be f ðyÞ = cgg+y y, with a constant c that denotes the minimum total expected cost of
the resource (Wibbens, 2021). If the profit π is variable one can set c = $1 without loss of gener-
ality, as is done here, essentially fixing the value of a dollar.
Figure 2 shows a graph of the resulting probability of gaining a resource over a short time
period Δt. Without investment (y = 0) the probability of obtaining a resource is zero. Increasing
the annual investment always leads to an increase in the probability of gaining the resource,
but with diminishing returns: twice the investments yield less than twice the probability of
obtaining the resource. In the limit y ! ∞, the asymptote gΔt represents the maximum proba-
bility that a firm can gain a resource in a short period of time Δt. For low values of g the time-
compression diseconomies are very pronounced (the speed of growth is limited no matter how
much a firm invests), while for high values of g they are less so.
An optimal choice of the investment yi for each state i represents a MPE solution of the
dynamic game (Ericson & Pakes, 1995). There are three potential resource states in this
game, denoted by an ordered pair whether Firm 1 and Firm 2 have access to the resource.
In state (0, 0) no firm has access to resource, in (1, 0) only Firm 1 and in (0, 1) only Firm
2. The state (1, 1) will never be attained because as soon as one firm has obtained the
resource the optimal policy for the other firm is to stop investing (since the value of
obtaining the Ð ∞ resource has diminished to zero). It follows that y(1,0) = y(0,1) = 0, V(0,1) = 0
and V ð1,0Þ = 0 πe − ρt dt =π=ρ = 50.
The equilibrium investment y(0,0) and value V(0,0) are nontrivial and need to be calculated
using Equations (9) and (11). Supporting Information S1 Appendix C presents an analytical
solution yielding y: = y(0,0) = 0.95 and V(0,0) = 17.0 as the single symmetric equilibrium.
Supporting Information S2 contains the R code needed to generate all numerical calculations in
this article, including for the table and figures. The resulting generator is (using the ordering of
states as indicated in front of the first matrix):
0 1 0 1
ð0, 0Þ − 2f ðyÞ f ðyÞ f ðyÞ −0:330 0:165 0:165
B C B C
Q = ð1, 0Þ B
@ 0 0 0 C B
A=@ 0 0 0 C
A ð13Þ
ð0, 1Þ 0 0 0 0 0 0
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1573

F I G U R E 3 Competitive dynamics in Example 1

The zeros at the bottom two rows of Q indicate that there are no transitions anymore once state
(1, 0) or (0, 1) is reached. The transition rate from (0, 0) to either (1, 0) or (0, 1) is f(y) = f
(0.95) = 0.165. The upper left cell is minus the sum of these two rates and represents the decay rate
of state (0, 0). It is the parameter of an exponential distribution for the time τ spent in state (0, 0)
until one of the firms obtains the resource. The expected value of this time is Eτ = 1/
0.330 = 3.0 years.
Figure 3 provides an overview of the resulting competitive dynamics and annual value
appropriation. Both firms invest 0.95 per year during τ years until one of the firms obtains the
resource. Then both firms stop investing, with the firm having obtained the resource (depicted
as Firm 1) appropriating π = 5 per year and the other firm appropriating zero. The customers
appropriate a value of five per year after time τ.

3.3 | Different notions of “value”

It is interesting to analyze the various perspectives on resource value that the

competitive outcomes in this example allow. As before, I will focus on the value for Firm
1; the value for other stakeholders (in casu Firm 2 and the customers) can be analyzed
analogously.
The simplest view on the “value” of the resource is the sum of discounted cash flows of the firm
that has obtained the resource. This is the expectation of V conditional on having the resource:

π
EðV j X = ð1, 0ÞÞ=V ð1,0Þ = =50 ð14Þ
ρ

The above value, though, does not take into account the sunk cost that was needed to obtain
the resource. To take this investment y into account, one can calculate the net present value
(NPV) at time zero of the resource conditional on obtaining it at time τ:
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1574 WIBBENS

π y
EðV j X ðτÞ = ð1, 0Þ, τÞ = e − ρτ − ð1 −e − ρτ Þ
ρ ρ ð15Þ
=59:5e − 0:1τ −9:5

The NPV of the resource depends strongly on how long it takes to obtain the resource and
can vary between 50 (for τ = 0) and −9.5 (for τ ! ∞).
The value in Equation (15) is still conditional on obtaining the resource, which has an ex
ante probability of a half. Taking that additional uncertainty into account yields:

π − ρτ y
EðV j τÞ= e − ð1 −e − ρτ Þ
2ρ ρ ð16Þ
=34:5e − 0:1τ −9:5

Finally, one can view the resource value as the expected value at t = 0, when the state is X
(0) = (0, 0):

EðV j X = ð0, 0ÞÞ =V ð0,0Þ =17:0 ð17Þ

These equations illustrate that there are many different ways of defining value; I will discuss
the implications for the RBV in Section 4.2.

3.4 | Example 2: Adding depreciation

The next example adds resource depreciation to the competitive dynamics. Depreciation of a
resource can represent for instance a technology that becomes obsolete or a brand name that
has waned. Adding depreciation leads to much richer competitive dynamics. In Example 1 the
situation always evolved toward a single state, with either firm having obtained the resource.
With depreciation added, resource positions remain in constant flux.

Example 2. Assume the same competitive situation as in Example 1. In addition, the

resource now depreciates at a rate δ = 0.05 per year; i.e., in a short period Δt a firm
that has the resource faces a probability of δΔt of losing it.

In addition to the three possible states X = (0, 0), (1, 0), (0, 1) previously, in this example a
fourth state X = (1, 1) is possible in which both firms have obtained the resource. While in
Example 1 the expected value V(1,1) was zero, now it is positive, because it carries some option
value due to the possibility that the competitor loses the resource—even though the cash flows
in this state are zero for both firms.
Because V(1,1) > 0 it can be optimal for Firm 1 to continue investing in the resource in
industry state X = (0, 1), albeit at a lower rate than in state X = (0, 0). The full MPE solution
to this example is presented in Supporting Information S1 Appendix D, employing an itera-
tive procedure. The optimal investments are y(0,0) = 0.60 and y(0,1) = 0.16. The focal firm's
investments if it has a resource remain zero, y(1,0) = y(1,1) = 0, since by assumption it cannot
obtain a second resource.
This optimal investment policy leads to the following generator of the resulting Markov process:
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1575

0 1 0 1
ð0, 0Þ − 2f 0 f0 f0 0 −0:30 0:15 0:15 0
B C B C
ð1, 0Þ B
B δ − ðδ +f 2 Þ 0 f2 C B
C B 0:05 − 0:14 0 0:09 C
C
Q= B C=B C ð18Þ
ð0, 1Þ B
@ δ 0 − ðδ+ f 2 Þ f 2 C B
A @ 0:05 0 −0:14 0:09 C
A
ð1, 1Þ 0 δ δ − 2δ 0 0:05 0:05 −0:10

Here, the shorthand notation f0: = f(y(0,0)) = f(0.60) = 0.15 and f2: = f(y(0,1)) = f(0.16) = 0.09
is used for the transition rates of the focal firm gaining a resource.
The competitive dynamics generated by Equation (18) are markedly different from those by
(13). In this case, all diagonal elements are non-zero, indicating that each state will eventually
move to a different one. As previously, their absolute values express the decay rates in an expo-
nential distribution. The expected time spent in state (0, 0) is 1/0.30 = 3.3 years; in states (1, 0)
and (0, 1) it is 1/0.14 = 7.1 years; and in state (1, 1) it is 1/0.1 = 10 years. The other entries in each
row represent the transition rate to the respective states. Note that the decay rate of each state is
equal by definition to the sum of the transition rates in the non-diagonal entries in each row.
Indeed, each row sums to zero.
Figure 4 shows the evolution of the probability of each state when both firms start without
the resource—P(0) = (1, 0, 0, 0). This evolution is calculated using Equation (5). Clearly, after a
long time period, the probabilities converge to some stationary distribution Ps, which satisfies:

QT P s = 0 ð19Þ

The solution4 to this equation is Ps = (0.08, 0.24, 0.24, 0.44).

Importantly, the emergence of a stationary distribution does not mean that the competitive
situation does not change. Competitive positions still regularly shift; only the probabilities to
find the firms in a specific industry state converge. Interestingly, in the long run the probability
of both firms having the resource at any given time (and thus not making any profit) is substan-
tial in this example—close to a half.
Figure 5 shows how the stationary distribution varies for different depreciation rates δ. The
relationship is non-trivial. For low δ, the situation is very similar to Example 1, in which the
depreciation rate was zero: either firm ends up with the resource and keeps it for a very long
time. This situation changes abruptly around δ ≈ 0.004. Then, all of a sudden, it becomes attrac-
tive for the laggard firm to start investing in the resource too, essentially taking a real option
that its competitor's resource will depreciate first. In the regime around δ ≈ 0.01, there is a more
than 60% chance at any given time that both firms have obtained the resource, not generating
any cash flows As δ increases further, it becomes less attractive to invest in the resource, since
the benefits of obtaining it are so short-lived that it is not worth significant investment any-
more. Above δ ≈ 0.3, the probability that neither firm has the resource is more than half.
To understand the mechanisms behind the abrupt change around δ ≈ 0.004, it is insightful
to derive analytical expressions for the stationary probabilities. Substituting Q from

4
In this case, the stationary distribution is unique given the constraint that the probabilities sum to one, jjPsjj = 1.
Sometimes multiple stationary distributions exist. For instance, in Example 1, any linear combination of (0, 1, 0) and
(0, 0, 1) is stationary. Mathematically, the stationary states represent the Eigen space of the Eigen value 0 of QT. The
matrix Q is defined such that it always has at least one Eigen value 0, while all other eigenvalues are negative. Note that
Q and QT have the same eigenvalues.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1576 WIBBENS

F I G U R E 4 Evolution P(t) for δ = 0.05 in Example 2

F I G U R E 5 Stationary distribution Ps for varying δ in Example 2

Equation (18) into Equation (19) leads to two conditions for the stationary probability: P(1,0)/
P(0,0) = f0/δ and P(1,1)/P(1,0) = f2/δ.5 The abrupt increase in P(1,1) follows from the latter of these:
because of the low δ, even a small investment y(0,1) and resulting transition probability f2 leads
to a large stationary probability P(1,1). Specifically, while f2 = P(1,1) = 0 for δ = 0.003, a small
increase to δ = 0.004 leads to an annual investment with transition rate f2 = 0.0088 and a
resulting probability ratio of P(1,1)/P(1,0) = f2/δ = 0.0088/0.004 = 2.2: the probability of both firms
having the resource is more than twice as large as that of only the focal firm having it. This
intuitively makes sense, because even though the chance in any given year of getting into state
(1, 1) is small, due to the low depreciation rate the chance of moving out of this state is even
smaller. That is the mechanism behind the abrupt increase of the stationary probability for state
(1, 1) around δ ≈ 0.004 illustrated in Figure 5.
Figure 6 shows the conditional expectations of the resource value E(V jX) for the different
states X for different values of the parameters δ and g. The black lines show the expected value
of the profit π from the current state. This is only nonzero if the focal firm has obtained the
resource, in which case it is equal to:

5
These two equations completely determine Ps, along with the trivial relations P(1,0) = P(0,1) and
P(0,0) + P(1,0) + P(0,1) + P(1,1) = 1.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1577

X (0,0) X (0,1)
50

40

30

20

10
Expected value

0 Profit S
X (1,0) X (1,1) g 0.2
50 g 0.8
40

30

20

10

0
0.001 0.01 0.1 1 0.001 0.01 0.1 1
Depreciation G (log scale)

F I G U R E 6 Conditional expectations E(V jX) of resource value in Example 2. The black lines show the
expected value of the profit π from the current state. The gray lines show the expected value when also taking
into account the flow from future changes in value QV net of the sunk investments y. The dark gray line is based
on the original assumption g = 0.2, while the light gray line is based on g = 0.8, which means that time-
compression diseconomies are less pronounced, allowing firms to grow faster

ð∞
π
EðV π j ð1, 0ÞÞ = πe − δt e − ρt dt = ð20Þ
0 δ+ρ

The gray lines show the expected value when also taking into account the flow from future
changes in value QV net of the sunk investments y. The dark gray line is based on the original
assumption g = 0.2, while the light gray line is based on g = 0.8, which means that firms can
grow faster and time-compression diseconomies are less pronounced.
The three regimes for different regions of δ in the stationary distribution are also apparent
in the conditional values. For δ below 0.004, the conditional expectations are very similar to
Example 1. The value of the focal firm having the resource (state X = (1, 0)) is very close to the
value of the cash profits; the value of the competitor having the resource (state X = (0, 1)) is
almost zero; and the value of not having the resource is based on the expected value of
obtaining the resource net of sunk-cost investment, as in Equation (17) (which is 17 for the orig-
inal parameters and a bit above 20 for g = 0.8).
For δ in intermediate ranges, the situation becomes very different. In this case the expected
value of having the resource E(Vj(1, 0)) becomes significantly lower than the cash profits, espe-
cially if time compression diseconomies are less pronounced (g = 0.8). The reason is that it
becomes attractive for the laggard firm to also obtain the resource, as illustrated by the increas-
ing value of the state (1, 1).
When δ increases further, the value of state (1, 1) diminishes again, for the simple reason
that the direct cash value Vπ decreases due to the rapid depreciation of the resource. In this
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1578 WIBBENS

regime the total value E(Vj(1, 0)) actually exceeds the direct cash value, since the probability of
the laggard catching up becomes very small, while there is still value from being able to regain
the resource once it has depreciated.

3.5 | Mechanisms behind the attractiveness paradox

Perhaps the most surprising outcome in Example 2 is the behavior of E(Vj(0, 0)), when g = 0.8:
for higher depreciation the expected value from the resource can be higher than for intermedi-
ate rates, even though the value of the direct pay-offs from the resource clearly will be lower
after having obtained it (the black line for X = (1, 0)). This is an instance of the attractiveness
paradox alluded to in the Introduction.
The Bellman Equation (9) allows a further investigation of the mechanisms that drive this
perhaps surprising finding. Re-arranging terms, the initial value can be written as6:

f0 f y
V0 = ðV 1 −V 0 Þ + 0 ðV 2 −V 0 Þ− 0 ð21Þ
ρ ρ ρ

This equation uses V0 = V(0,0) for the value of the initial state, V1 = V(1,0) if the focal firm has
obtained the resource and V2 = V(0,1) if its competitor has obtained the resource. In words, it
means that the value of the initial state is the value of the expected return from the focal firm
gaining the resource (V1−V0) plus the expected value from its competitor gaining a resource
(V2−V0, which will be negative) minus the NPV of the annual investments y0 in obtaining the
resource. Recall that f0 denotes the transition rate of obtaining the resource associated with the
investment y0—which is optimal since we are considering a MPE.
Table 1 shows the equilibrium values for each of the three terms in Equation (9) for long and
short-lived resources, with expected lifetime 1δ of 10 and 3 years, respectively. The other parameter
values are as in the light-gray line in Figure 6, that is, g = 0.8 and otherwise as in Example 1.
The first term represents the expected value from gaining the resource. This is of course
larger for the long-lived resource, because it will yield returns for much longer. However, it is
only 42.7/28.6 = 1.5x as large, while the longer-lived resource will yield returns for more than
three times as long. One reason for this discrepancy is that over the 10-year expected lifespan of
the longer-lived resource, there is a probability of 72% that the competitor will also obtain the
resource, after which no firm will receive any pay-offs until one firm's resource depreciates.
This probability is only 45% during the 3-year expected lifespan of the shorter-lived resource.7
The larger expected value for the longer-lived resource is offset by a larger loss in value of
the competitor gaining the resource and the larger investments induced by the longer pay-offs
of the resource. Interestingly, the numbers in Table 1 indicate that especially the former of
these two effects diminishes the market's initial attractiveness V0. This is also apparent from the
top-right chart in Figure 6, which shows an increased value V2 = V(0,1) for larger values of δ

6
Substituting Equation (18) into Equation (9) for state X = (0, 0)
!
P
yields V 0 = ρ1 π 0 −y0 + Q0i V i = ρ1 ð −y0 −2f 0 V 0 +f 0 V 1 +f 0 V 2 Þ= ρ1 ½ f 0 ðV 1 −V 0 Þ +f 0 ðV 2 −V 0 Þ −y0 .
i

7
The probability of the competitor gaining a resource while the focal firm has it already is f f+2 δ, because there are only
2
two ways to move out of state X = (1, 0): moving to state (1, 1) with transition rate f2 and to state (0, 0) with δ. The
probability of either event occurring is proportional to its transition rate and the two probabilities must sum to one.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1579

T A B L E 1 Mechanisms behind the attractiveness paradox

Long-lived (1δ =10) Short-lived (1δ = 3)

f0
ρ ðV 1 − V 0 Þ
Expected value from gaining resource 42.7 28.6
f0 −18.8 −7.7
ρ ðV 2 − V 0 Þ
Expected value from competitor gaining

NPV of annual investment − yρ0 −14.9 −11.6

Initial value (= sum of three above items) V0 9.0 9.3

Note: Illustration of the attractiveness paradox for long versus short-lived resources in Example 2. Depreciation δ as indicated
and the other parameter values are as in the light-gray line in Figure 6, that is, g = 0.8 and otherwise as in Example 1. The
Bellman Equation (9) implies that the sum of the three top rows is equal to the initial value on the bottom row:
V 0 = fρ0 ðV 1 − V 0 Þ+ fρ0 ðV 2 − V 0 Þ− yρ0 . Here, V0 = V(0,0) denotes the value in which both firms have no resource, V1 = V(1,0) the
value in which the focal firm has the resource and V2 = V(0,1) in which its competitor has the resource.

(shorter-lived resources), up to δ ≈ 0.3. This makes sense: in a market with the longer-lived
resources, it is worse to be a follower, because of the longer expected wait to become a leader.
This example illustrates how the subtle dynamics of resource competition can lead to unex-
pected outcomes such as the attractiveness paradox. Moreover, the framework allows a detailed
investigation into the mechanisms driving the outcomes.

3.6 | Extensions

The previous section provided two examples of how Markov processes can be used to analyze
the competitive dynamics of resource competition. The examples incorporated several key con-
cepts in the RBV such as asset erosion (depreciation) and time-compression diseconomies
(Dierickx & Cool, 1989). The general framework presented in Section 3.1 allows many exten-
sions of these examples. Below, I provide an overview of several other RBV concepts that can
be analyzed in this framework.
Resource imitation or substitution (Barney, 1991) can be added by letting the investment
function f(y) depend on the state. For instance, one can use the following specification:

gy
f ðy,X Þ = ð22Þ
gcðX Þ+ y

The function c(X) describes how the minimum expected resource cost varies across states.
An imitable or substitutable resource will be cheaper to acquire once the competitor has
obtained it: c(X = (0, 1)) < c(X = (0, 0)).
Asset mass efficiencies (Dierickx & Cool, 1989) can be added by allowing more resource
states. If each of the two firms can obtain n resources, then there are n2 possible industry states
to consider. Asset mass efficiencies can operate in two different ways. First, in the product mar-
ket if a firm's profit π grows more than linearly with its number of resources X1. Second, in the
resource market if the cost or time-compression diseconomies go down as the number of
resources go up. This can be modeled by making c or g a function of the focal firm's resource
position X1.
Scalability is the extent to which resources incur an opportunity cost in their use. Resources
with low scalability are capacity constrained, while those with high scalability are “scale-free”
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1580 WIBBENS

(Levinthal & Wu, 2010). Wibbens (2021) incorporates scalability in a Markov model through its
effect on the marginal cost functions and the resulting profit π as a function of two firms'
resource positions.
Rareness (Barney, 1991) was implicitly assumed in Examples 1 and 2 because only two firms
could potentially obtain the resource. This could be modeled more explicitly by increasing the
number of firms in the model and constraining the number of resources available and/or the
total potential investment in them. For instance, the time-compression diseconomies in f(y)
could be made dependent on the total investment of all firms jointly, capturing the idea that if
one firms invests for instance in a technology or brand, it might become more expensive for
other firms to do so, since there is only a finite supply of talent or advertising opportunities.
Interconnectedness across resources (Dierickx & Cool, 1989) can be incorporated if multiple
different resource types are included in the state space, as in Equation (2). Each resource will
have its own investment policy y1, y2, and so forth. The interconnectedness can be expressed by
making the investment function f(y) dependent on the level of other resources. For instance,
Dynamic capabilities (Teece et al., 1997) or higher-order resources (Wibbens, 2019), which affect
value not directly through profits in a product market but through their effects on future
resource evolution, can be incorporated in this way. The causal relationships between these dif-
ferent types of resources could be represented using a directed acyclic graph8 or Bayesian net-
work (Ryall & Bramson, 2013). Note that a Markov process itself can be viewed as a simple
Bayesian network consisting of a sequence of states with causality directed forward in time.
The general framework also allows inclusion of bounded rationality (Simon, 1955) or causal
ambiguity (Lippman & Rumelt, 1982; Rumelt, 1984). For instance, instead of a fully rational
optimization of y based on the expected values consistent with the Bellman equation, the invest-
ment could be optimized based on a more myopic view of value. This makes the analysis sim-
pler in fact, because this definition of value does not depend on the investment policy y, and
therefore, the generator matrix can be calculated directly to apply Equation (10), which is con-
siderably simpler to do than finding the MPE.
Moreover, differences in information positions (Barney, 1986) could be included as an addi-
tional state variable. For instance, based on prior experience firms might be able to gain more
information about the value of a resource, which is reflected in a state variable that affects the
investment y, similarly to how other types of higher-order resources are incorporated in the
model.
The framework as presented in Section 3.1 implicitly focuses on investors as the primary
residual claimant of interest. This could be modified to include other stakeholders that have
claims on certain resource bundles, such as employees, suppliers and customers (Barney, 2018;
Lippman & Rumelt, 2003b). The only key requirement from a modeling perspective is that the
state space contains the relevant variables such that payments to other claimants can be defined
as a function of the industry state X(t). This requires for instance that the columns with
resource bundles in Equation (2) not only include the rivals and entrants in an industry, but
also other stakeholders of interest. Additionally, there is no particular necessity for value to be
of the form in Equation (7). The definition could for instance include a stakeholder-specific cost
of capital or other opportunity cost, which even could vary over time—as long as it is formu-
lated as a function of the state X(t).
Also, the model of product market competition could be modified. In Examples 1 and 2, it is
modeled based on VBS, specifying how much value different groups of actors can create and how

8
Thanks to SMJ Associate Editor Olivier Chatain for suggesting the relationship with directed acyclic graphs.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1581

they split value if the resulting coalitional game has an indeterminate core. This is similar to the
specification of a second stage in a biform game (Brandenburger & Stuart, 2007). Noncoalitional
game theory as used in neo-classical economics can be used too, for instance specifying product
market competition using Cournot or Bertrand models. In the end, all that matters for the model
is that profit π can be defined as a function of the state X. This could also be achieved without
using a game-theoretic approach, instead assuming a functional form between X and π with spec-
ified characteristics. For instance, Chatain (2014) employs a two-firm model with π = π(X1, X2)
that has a positive derivative in X1 and a negative one in X2.9 Resource fungibility, the extent to
which resources can be leveraged in different settings (Levinthal & Wu, 2010; Montgomery &
Wernerfelt, 1988), can be incorporated in models with multiple product markets.
Finally, the framework could be extended into a hidden Markov model (HMM), for instance
for empirical research. In an HMM, the state Xt is not directly observable, but inferences about
it can be made based on some observable variables Yt. Profit π could be one such observable
variable. In an HMM, it would not be directly specified as a function of the state π(X), but
rather through a conditional probability distribution πjX. Such a model can be used for empiri-
cal studies, by fitting actual profit data to make inferences about the dynamics of the
unobservable state variables (Wibbens, 2019).

4 | DISCUSSION

The attractiveness paradox discussed in the Introduction and at the end of Section 3.5 illustrates
how the Markov framework can yield strategic insights that would be hard to derive without
such a formal model. Below, I discuss three other insights with implications for future research.

4.1 | Discontinuity of competitive outcomes

In the management literature, most scholars implicitly assume continuous relationships between
constructs. This is apparent from the ubiquity of regression analyses to test our theories. A simple
linear regression yields a relationship of the form y = βx. For a given regression coefficient β, the
smaller the change in x, the smaller the resulting change in y. The same is true for other forms of
regressions, including non-linear ones: one can always achieve an arbitrarily small change in the
dependent variable y by choosing an appropriately small change in an explanatory variable x.10
However, the examples here illustrate that competitive dynamics sometimes lead to abrupt
changes. Figure 5 shows this most clearly: the market outcomes change suddenly around
δ ≈ 0.004. Over a very small range of values for δ the stationary outcome jumps from very likely
having one player in the market to most likely having two players. This finding extends the dis-
continuities found in earlier game theoretic models that did not include long-term investment
dynamics (e.g., Adner, Ruiz-Aliseda, & Zemsky, 2016; Chatain & Zemsky, 2007;
Makadok, 2003).
9
The notation of the original paper has been replaced with this one's, denoting the resource base of firm i with Xi.
Additionally, Chatain assumes various characteristics for the second derivative.
10
Note that this informal notion of continuity corresponds with its formal definition: a function f(x) is continuous if for
each x0 and ϵ > 0 one can always find a δ > 0 such that if x satisfies jx − x0j < δ then jf(x) − f(x0)j < ϵ. Technically, the
change in P(1,1) illustrated in Figure 6b around δ ≈ 0.004 is not discontinuous, but for all practical purposes the change
is very abrupt.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1582 WIBBENS

In future theoretical and empirical work, the potential for discontinuities in competitive
outcomes deserves more attention. In their theorizing, researchers will not only need to derive
the assumed direction of the effect (positive, negative, [inverted] U), but also whether it is con-
tinuous over the entire range. In empirical work, too, the discontinuities should be considered
and potentially tested.

4.2 | Meaning of value in the RBV

Another contribution of this article is that it can alleviate some scholars' persistent critiques on
the RBV's logical structure, particularly pertaining the definition of “value” (see Section 2.2).
The formal approach used in this article forces the use of precise definitions and theoretical rea-
soning. The analysis highlights three key elements of resource value that could easily lead to
confusion if they are not precisely defined.
The first potential source of confusion—and probably the one of which scholars are
most aware—is value creation versus appropriation. Example 1 and Figure 3 illustrate the
distinction: although a firm can create a value of 10 after obtaining the resource, it will
only appropriate five or zero, depending on whether it is a monopolist. Peteraf and Bar-
ney (2003) already highlighted the need to distinguish the two concepts in a resource-
based context.
Second, the meaning of value, rent, or (economic) profit hinges critically on a precise defini-
tion of (opportunity) cost, which is far from trivial. Lippman and Rumelt (2003b) have provided
several persuasive arguments about how tricky it is to arrive at a useful definition of opportu-
nity cost. The potential for confusion is clear in Example 1. One could reasonably define the
value of the resource in several different ways, such as the discounted cash returns
(Equation [14]), the returns net of actual sunk costs (Equation [15]) or net of the expected sunk
costs (Equation [17]). Many other definitions of opportunity cost would be possible if there are
outside options for the use of the resource. Not explicitly specifying which of these is used can
easily lead to confusion or inconsistency.
Third and finally, any notion of value that includes expected cash flows hinges on assump-
tions and information about the future, which could even differ by actor. In a stochastic model,
these are codified in two ways: first in the assumptions made about the competitive dynamics,
such as in the specifications of Examples 1 and 2; and second using expectations conditional on
the value of specific variables. For instance, in Equation (15) the value of the resource is speci-
fied conditional on the focal firm obtaining the resource at time τ, formally expressed as E(V jX
(τ) = (1, 0), τ).
Summarizing, whenever using a concept of value, scholars should explicitly state whether
they mean value creation or appropriation; which (opportunity) cost is taken into account; and
what assumptions and information are used, including whether these are actor-specific or com-
mon knowledge.

4.3 | The origins of value

Finally, the Markov framework provides deeper insights into the origins of value. When consid-
ering value appropriation net of sunk cost and using a Markovian definition of resources, value
can originate ultimately from only two sources: initial endowments (the state X[t = 0]) or
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1583

exogenous shocks (changing states from X(t) to X(t + Δt)).11 This might seem puzzling at first,
and even an artifact of the framework. However, when taking a broad view of what can consti-
tute a resource, such as “anything which could be thought of as a strength or weakness of a
given firm” (Wernerfelt, 1984, p. 172) or “something that the organization can draw upon to
accomplish its aims” (Helfat et al., 2007, p. 4), any superior performance can ultimately be
traced back to an initial endowment or an exogenous shock.
For instance consider the success of Apple—the most value-creating listed company over
the past two decades (Wibbens & Siggelkow, 2020). To explain Apple's exceptionally high level
of value appropriation, one could point to the effectively inimitable resources it has access to,
such as control of the iOS ecosystem including its app store; the iPhone's user-friendly design;
and a large customer base loyal to the Apple brand. But, why was Apple able to get these supe-
rior resources? One could argue due to its higher-order resources (Wibbens, 2019) when enter-
ing the smart phone industry, such as Jobs' strategic foresight of redeploying its capabilities
from the PC market into mobile devices. In the end, Apple's superior value could be traced back
all the way to a combination of endowment (e.g., Steve Jobs as founder) and exogenous shocks
or “luck” (e.g., the right resources available at the right moment to win in mobile).
Some authors consider other origins of value beyond these two. For instance, Barney (1986)
argues for heterogeneous expectations or superior information as an original source of value.
This of course raises the question where such superior information then comes from? Arguably,
superior information can be viewed as a resource in its own right (it passes the VRIN character-
istics from Barney (1991) as well as the definition of resources in Helfat et al. (2007)), and ulti-
mately that too has to originate either from an initial endowment or an exogenous shock. Thus,
whether superior information is a third source of value depends on the definition what consti-
tutes a resource; if it is included in the definition of resources, it ceases to be a fundamental
source of value.

5 | C ON C L U S I ON

Over the past 30 years, the RBV has become one of the most influential scholarly theories in
strategic management, both in research and teaching. With the Markov framework introduced
in this article, I hope to contribute to this very successful stream of research by providing a
starting point for new directions of formal, simulation, and empirical research. The framework
allows adding long-term dynamics and uncertainty to existing models of competition. Hope-
fully, scholars will find it a helpful addition to their toolkit in shaping the next 30 years of RBV
research.

ACK NO WLE DGE MEN TS

The author wishes to thank Michaël Bikard, Laurence Capron, Jerker Denrell, Javier
Gimeno, Ekin Ilseven, Henning Piezunka, Hart Posen, and Phanish Puranam as well as par-
ticipants of the 2021 TOM virtual meeting for helpful feedback. Special thanks to SMJ
11
More formally, future economic profit has an expected value of zero for any given probability distribution of resource
states P(t): The expected value of economic profit over a short period Δt is the sum of expected values of cash flows (E
(π − y)Δt = (π − y)TP(t)Δt) minus the opportunity cost of capital (ρVTP(t)Δt) plus the expected change in resource value
(E(V (t + Δt) − V (t)) = VT(I + QTΔt)P(t) − VTP(t) = VTQTP(t)Δt). Adding up these terms yields (π − y − ρV + QV)TP(t),
which is zero because of Equation (9). Hence, any future economic profit (i.e., above-normal returns) must follow from
exogenous shocks.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1584 WIBBENS

associate editor Olivier Chatain and two anonymous reviewers for very helpful and con-
structive comments throughout the review process. The author takes full responsibility for
any remaining errors.

DATA AVAILABILITY STATEMENT

This study only relies on data generated from numerical calculation and simulation. All results
can be replicated using the R code provided in Supporting Information S2.

ORCID
Phebo D. Wibbens https://orcid.org/0000-0002-1517-3858

R EF E RE N C E S
Adner, R., Polos, L., Ryall, M., & Sorenson, O. (2009). The case for formal theory. Academy of Management
Review, 34(2), 201–208.
Adner, R., Ruiz-Aliseda, F., & Zemsky, P. (2016). Specialist versus generalist positioning: Demand heterogeneity,
technology scalability and endogenous market segmentation. Strategy Science, 1(3), 184–206.
Adner, R., & Zemsky, P. (2006). A demand-based perspective on sustainable competitive advantage.
Strategic Management Journal, 27(3), 215–239.
Barney, J. B. (1986). Strategic factor markets: Expectations, luck, and business strategy. Management Science,
32(10), 1231–1241.
Barney, J. B. (1991). Firm resources and sustained competitive advantage. Journal of Management, 17(1), 99–120.
Barney, J. B. (2001). Is the resource-based “view” a useful perspective for strategic management research? Yes.
Academy of Management Review, 26(1), 41–56.
Barney, J. B. (2018). Why resource-based theory's model of profit appropriation must incorporate a stakeholder
perspective. Strategic Management Journal, 39(13), 3305–3325.
Besanko, D., & Doraszelski, U. (2004). Capacity dynamics and endogenous asymmetries in firm size. RAND
Journal of Economics, 35(1), 23–49.
Bhat, U. N., & Miller, G. K. (2002). Elements of applied stochastic processes (3rd ed.). Hoboken, NJ: Wiley-
Interscience.
Brandenburger, A., & Stuart, H. (2007). Biform games. Management Science, 53(4), 537–549.
Brandenburger, A. M., & Stuart, H. W. (1996). Value-based business strategy. Journal of Economics & Manage-
ment Strategy, 5(1), 5–24.
Bromiley, P., & Fleming, L. (2002). The resource-based view of strategy: A behaviorist critique. In M. Augier &
J. G. March (Eds.), The economics of choice, change and organization: Essays in memory of Richard M. Cyert,
Chapter 15 (pp. 319–336). Herndon, VA: Edward Elgar Publishing.
Chatain, O. (2014). How do strategic factor markets respond to rivalry in the product market? Strategic Manage-
ment Journal, 35(13), 1952–1971.
Chatain, O., & Zemsky, P. (2007). The horizontal scope of the firm: Organizational tradeoffs vs. buyer-supplier
relationships. Management Science, 53(4), 550–565.
Denrell, J., Fang, C., & Liu, C. (2015). Perspective–chance explanations in the management sciences. Organiza-
tion Science, 26(3), 923–940.
Dierickx, I., & Cool, K. (1989). Asset stock accumulation and sustainability of competitive advantage. Manage-
ment Science, 35(12), 1504–1511.
Doraszelski, U., & Judd, K. L. (2012). Avoiding the curse of dimensionality in dynamic stochastic games. Quanti-
tative Economics, 3(1), 53–93.
Eddy, S. R. (2004). What is a hidden Markov model? Nature Biotechnology, 22(10), 1315–1316.
Ericson, R., & Pakes, A. (1995). Markov-perfect industry dynamics: A framework for empirical work. Review of
Economic Studies, 62(1), 53–82.
Foss, N. J., & Knudsen, T. (2003). The resource-based tangle: Towards a sustainable explanation of competitive
advantage. Managerial and Decision Economics, 24(4), 291–307.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WIBBENS 1585

Fudenberg, D., Gilbert, R., Stiglitz, J., & Tirole, J. (1983). Preemption, leapfrogging and competition in patent
races. European Economic Review, 22(1), 3–31.
Gans, J., & Ryall, M. D. (2017). Value capture theory: A strategic management review. Strategic Management
Journal, 38(1), 17–41.
Helfat, C. E., Finkelstein, S., Mitchell, W., Peteraf, M. A., Singh, H., Teece, D. J., & Winter, S. G. (2007). Dynamic
capabilities: Understanding strategic change in organizations. Malden, MA: Blackwell Publishing.
Hoopes, D. G., Madsen, T. L., & Walker, G. (2003). Guest editors' introduction to the special issue: Why is there
a resource-based view? Toward a theory of competitive heterogeneity. Strategic Management Journal, 24(10),
889–902.
Knudsen, T., Levinthal, D. A., & Puranam, P. (2019). A model is a model. Strategy Science, 4(1), 1–3.
Kogut, B., & Kulatilaka, N. (2001). Capabilities as real options. Organization Science, 12(6), 744–758.
Kraaijenbrink, J., Spender, J.-C., & Groen, A. J. (2010). The resource-based view: A review and assessment of its
critiques. Journal of Management, 36(1), 349–372.
Levinthal, D. A. (1997). Adaptation on rugged landscapes. Management Science, 43(7), 934–950.
Levinthal, D. A. (2011). A behavioral approach to strategy–what's the alternative? Strategic Management Journal,
32(13), 1517–1523.
Levinthal, D. A., & Wu, B. (2010). Opportunity costs and non-scale free capabilities: Profit maximization, corpo-
rate scope, and profit margins. Strategic Management Journal, 31(7), 780–801.
Lieberman, M. (2021). Is competitive advantage intellectually sustainable? Strategic Management Review, 2(1),
29–46.
Lippman, S. A., & Rumelt, R. P. (1982). Uncertain imitability: An analysis of interfirm differences in efficiency
under competition. The Bell Journal of Economics, 13(2), 418–438.
Lippman, S. A., & Rumelt, R. P. (2003a). A bargaining perspective on resource advantage. Strategic Management
Journal, 24(11), 1069–1086.
Lippman, S. A., & Rumelt, R. P. (2003b). The payments perspective: Micro-foundations of resource analysis. Stra-
tegic Management Journal, 24(10), 903–927.
Loury, G. C. (1979). Market structure and innovation. The Quarterly Journal of Economics, 93(3), 395–410.
Makadok, R. (2003). Doing the right thing and knowing the right thing to do: Why the whole is greater than the
sum of the parts. Strategic Management Journal, 24(10), 1043–1055.
Makadok, R., & Barney, J. B. (2001). Strategic factor market intelligence: An application of information econom-
ics to strategy formulation and competitor intelligence. Management Science, 47(12), 1621–1638.
McGrath, R. G. (1997). A real options logic for initiating technology positioning investments. Academy of
Management Review, 22(4), 974–996.
Montgomery, C. A., & Wernerfelt, B. (1988). Diversification, Ricardian rents, and Tobin's q. The Rand Journal of
Economics, 19(4), 623–632.
Mueller, D. C. (1977). The persistence of profits above the norm. Economica, 44(176), 369–380.
Netzer, O., Lattin, J. M., & Srinivasan, V. (2008). A hidden markov model of customer relationship dynamics.
Marketing Science, 27(2), 185–204.
Oxley, J. E., Rivkin, J. W., Ryall, M. D., & Strategy Research Initiative. (2010). The strategy research initiative:
Recognizing and encouraging high-quality research in strategy. Strategic Organization, 8(4), 377–386.
Pacheco-de-Almeida, G., & Zemsky, P. (2007). The timing of resource development and sustainable competitive
advantage. Management Science, 53(4), 651–666.
Penrose, E. T. (1959). The theory of the growth of the firm. London: Basil Blackwell.
Peteraf, M. A. (1993). The cornerstones of competitive advantage: A resource-based view. Strategic Management
Journal, 14(3), 179–191.
Peteraf, M. A., & Barney, J. B. (2003). Unraveling the resource-based tangle. Managerial and Decision Economics,
24(4), 309–323.
Priem, R. L., & Butler, J. E. (2001). Is the resource-based “view” a useful perspective for strategic management
research? Academy of Management Review, 26(1), 22–40.
Rabiner, L. R., & Juang, B.-H. (1986). An introduction to hidden Markov models. ASSP Magazine, IEEE, 3(1),
4–16.
Rumelt, R. P. (1984). Toward a strategic theory of the firm. In R. B. Lamb (Ed.), Competitive strategic manage-
ment (pp. 556–570). Englewood Cliffs, NJ: Prentice-Hall.
 10970266, 2023, 6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smj.3339 by Tsinghua University Library, Wiley Online Library on [08/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1586 WIBBENS

Rumelt, R. P. (1991). How much does industry matter? Strategic Management Journal, 12(3), 167–185.
Ryall, M. D., & Bramson, A. L. (2013). Inference and intervention: Causal models for business analysis. New York:
Routledge.
Schmidt, J., & Keil, T. (2013). What makes a resource valuable? Identifying the drivers of firm-idiosyncratic
resource value. Academy of Management Review, 38(2), 206–228.
Schweidel, D. A., Bradlow, E. T., & Fader, P. S. (2011). Portfolio dynamics for customers of a multiservice
provider. Management Science, 57(3), 471–486.
Siggelkow, N., & Wibbens, P. D. (2015). A ladder of competitive advantage. Academy of Management Proceedings,
2015(1).
Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of Economics 69(1), 99–118.
Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Manage-
ment Journal, 18(7), 509–533.
Wernerfelt, B. (1984). A resource-based view of the firm. Strategic Management Journal, 5(2), 171–180.
Wibbens, P. D. (2019). Performance persistence in the presence of higher-order resources. Strategic Management
Journal, 40(2), 181–202.
Wibbens, P. D. (2021). The role of competitive amplification in explaining sustained performance heterogeneity.
Strategic Management Journal, 42(10), 1769–1792.
Wibbens, P. D., & Siggelkow, N. (2020). Introducing LIVA to measure long-term firm performance. Strategic
Management Journal, 41(5), 867–890.
Winter, S. G. (1987). Knowledge and competence as strategic assets. In D. J. Teece (Ed.), The competitive chal-
lange: Strategies for industrial innovation and renewal, Chapter 8, pp. 159–184. Cambridge MA: Ballinger.

S UP PO RT ING IN FOR MAT ION

Additional supporting information may be found in the online version of the article at the pub-
lisher's website.

How to cite this article: Wibbens, P. D. (2023). A formal framework for the RBV:
Resource dynamics as a Markov process. Strategic Management Journal, 44(6),
1562–1586. https://doi.org/10.1002/smj.3339