and Technology Studies
KITeS - Università Bocconi
Via Röntgen 1- 20136 Milano
tel. +39 025836.3395/7
fax +39 025836.3399
Franco Malerba and Luigi Orsenigo
innovation and the evolution of
market structures under alternative
User-producer relations, innovation and the evolution of
market structures under alternative contractual regimes
Franco Malerba* and Luigi Orsenigo**
* Department of Economics and KITeS, Bocconi University, Milan, Italy
** DIMI, University of Brescia and KITeS, Bocconi University, Milan, Italy
We thank Luca Bonacina for excellent research assistance and the comments of two
referees. Support for this paper comes from the Italian FIRB Program.
In this paper we examine the effects that user-producer interactions have on innovation and the
dynamics of market structure of two vertically related industries under alternative contractual
regimes. User-producer relationships have been a key theme in the economics of innovation ever
since the seminal contribution by Lundvall (1992): user-producer relationships may generate
interactive learning, improvements in products and processes and innovation. Yet, despite the
widely recognized importance of user-producer interactions as an important source of innovation,
very little attention has been paid to the analysis of how the existence of technological advantages
arising from tight interactions among suppliers and producers influences the co-evolution of
innovation and market structure. The empirical literature has focused mainly on the attempt at
identifying how and at what conditions user-producer relations may foster innovation at the level of
individual firm, region and country, without considering in any detail the consequences on the
dynamics of concentration: if anything, the discussion has been mainly framed in terms of the
relative advantages of establishing tight relations among suppliers and customers as opposed to
vertical integration or other forms of governance of the innovative process. In a more theoretical
perspective, while a few models (Ciarli et al., 2008, Windrum et al., 2009) have studied the effects
of vertical specialization and outsourcing on firms performance and innovation, no formal model
has yet been produced so far which explicitly focuses on the effects of close interactions between
users and producers and the patterns of industry evolution.
In a previous paper (Malerba and Orsenigo, 2008), we began to explore how users-producers
interactions affect innovation and the dynamics of market structure in industry evolution. In this
paper we extend the analysis by examining how the benefits of user-producers interactions influence
the rates of innovation and the evolution of market structure in the two related industries under
alternative contractual arrangements. We do not address the issue of why specific contractual forms are
used: more generally, we do not enter the realm of contract theory. Nor do we consider here the
possibility that firms may vertically integrate. More modestly, as an initial step, we simply compare
the dynamic effects of different contractual forms regulating market transactions, namely the length
and the exclusivity of the contracts.
The choice for a highly simplified analysis is supported by the consideration that even in this
extremely stylized setting, the model generates a rather complex behaviour. In fact, the existence of
advantages stemming from users-producers relationships introduces a dynamic “matching” problem
between firms characterized by heterogeneous capabilities and imperfect information who act in a
continuously changing environment but are however able to improve their products also through
interactive and interdependent learning processes. Our results highlight the subtle trade-offs and
dynamic interdependencies that arise in these contexts. In particular, we show that:
a. a trade-off is present between the exploitation of past experience and the exploration
of new suppliers;
b. externalities are present, even if the advantages arising from interactions do not spill
over to other firms;
c. imperfect information and agents heterogeneity are crucial factors in determining the
consequences of alternative contractual arrangements on industry dynamics;
d. vertical interdependencies propagate the effects of specific firms’ decisions across
industries and over time, so that the resulting dynamics is characterized by
interacting path-dependent processes.
It has to be emphasized that the model presented here is not “history-friendly”, i.e. it does not aim at
qualitatively examining the main causal factors and processes that could explain the evolution of a
particular industry (Malerba et al., 1999, 2008; Malerba and Orsenigo 2002). This exercise
originates indeed from a “history-friendly” model (Malerba et al., 2008), but it has not the ambition
to reproduce the main stylized features of any particular industry. As much as we are convinced that
history friendly models are an extremely useful and methodologically sound research tool, they are
not necessarily the only acceptable modeling strategy. We believe that they are complementary with
other more traditional styles. Here we develop a history friendly model in a more abstract way for
two reasons. First, we are interested in investigating and probing a more general phenomenon that
might apply to a variety of industries, very much along the lines of Malerba et al. (2007). Second, in
a complementary perspective, we feel the need of beginning to elucidate some more abstract
theoretical issues which could then guide us towards the construction of models of industries where
user-producer relations play a particularly important role.
The paper is organized as follows. In the next section we discuss the empirical and theoretical
background of the paper. In Section 3 we begin to examine the complex issues that the
consideration of user-producer relations introduces in the dynamics of market structure. In Section 4
we present the spirit and the logic of the model, the structure of which is outlined in Section 5. Then
in Section 6 we run the basic simulation exercises. Finally in Section 7 we draw the main
2. The background
The notion that user-producer relationships might be a fundamental source of technological
advances and innovation was first introduced by Lundvall (1992) and since then it has had a
profound impact on the economics of innovation. Lundvall (1992) emphasized that users are quite
knowledgeable about products uses and applications, and that their interactions with suppliers
provide the latter with information, ideas and feedbacks. The establishment of close relations
between producers and users can facilitate the transfer of tacit knowledge, the development of trust,
the ability to customize products, etc. Thus, the development of close and stable relations between
suppliers and users might be beneficial to both parties and therefore such interactions should be
properly used and organized.
More generally, the recognition that the development of new technologies in many cases requires
close communication and interaction between users and producers has deeply influenced our
understanding and our representations of the innovation process. The relevance of user–producer
interactions appears in a vast range of different streams of research: the literature is so large and
pervasive that cannot be reviewed here. Suffice it to mention, as examples, that von Hippel (1986,
1988, 2005) pointed to lead users as sources of ideas for the innovation process of producers and to
the role of users as innovators themselves. Bresnahan and Greenstein (2001) have emphasized that
users and producers may co-invent new process technologies and introduce organizational
innovations. Producers knowledge and innovations are modified by users when they are adapted to
their specific organization and production processes. Often this takes place with the active
participation of both producers and users. The whole literature on networks of firms has argued that
vertical R&D joint ventures and cooperative agreements between suppliers and users are a major
source of innovation and new product development (Powell and Grodal, 2005).The idea of the
importance of user- producer interactions is a key element in the development of the various notions
of national, regional and sectoral systems of innovation, whereby it is often argued that it is
precisely these relations that may significantly contribute to explain the development of innovative
regional clusters or the export performance of countries in specific industries (Fagerberg 1993,
Porter, 1990). However, it has also been argued that relations between users and producers may not
always be beneficial to innovation: in some cases they may hinder it, especially if new (disruptive)
technologies appear which suppliers are unable to master (Christensen, 1997).
The analysis of user–producer relationships has been typically framed within the broader discussion
of the “boundaries of firms” and of the relative merits and demerits of alternative forms of
organization of innovative activities, ranging from vertical integration at one extreme to pure spot,
arm-length market transactions to the other. Thus, in many instances, user-producer relations are
considered as an intermediate organizational form which requires the development of long-run,
stable (formal or informal) contracts and which can be more efficient than others as a function of
the conventional variables considered in this stream of literature (transaction costs, asset specificity,
information asymmetries, etc..). In this context, for instance, the establishment of tight, long run
relations with suppliers was considered as one of the main advantages of the so-called J-form of
organization as compared to the A-form and in general of the Japanese mode of production (Aoki,
1986). More recently, and in a somewhat different vein, the literature on markets for technology
(Arora et al., 2001) has stressed the advantages of division of labour in innovative activities
between agents who specialize respectively in the production and development of innovations.
User-producers interactions are also at the center of the whole debate on outsourcing and
modularity in products and organizations (Sanchez and Mahoney, 1996; Brusoni, Prencipe and
Pavitt, 2001). It is important to notice, in this respect, that recent contributions have increasingly
relied on a “cognitive” and “capability-based” (Langlois, 2002, Marengo and Dosi, 2005, Jacobides
and Winter, 2005) approach to the analysis of the processes of vertical specialization – as distinct
from more orthodox accounts based on contract theory –. In this context doubts have been raised
about the beneficial effects of outsourcing strategic parts of innovative activities. Indeed, it is
argued that little ( and sometimes) contrary empirical evidence exists in this regard: outsourcing
would appear to have a detrimental effect on firm’s productivity growth in the longer run, especially
when an activity is almost completely outsourced (total outsourcing) (see Windrum et al., 2009 and
the literature thoroughly reviewed there; Gianelle and Tattara, 2007; Cusmano et al., 2009). In these
situations, a firm might gain (short run) advantages in terms of modularization and specialization,
but it might run the risk of losing the ability to develop new product and organizational
It is also worth noting that in this stream of literature only few formal models have been developed
to support empirical analysis. A recent class of models deals with the problems that firms face in
controlling and coordinating innovative activities which involve multiple and interacting modules
ways of decomposing problems, activities and routines (Marengo and Dosi, 2005; Ciarli et al.,
2008; Windrum et al. 2009). However, less has been done as it concerns the effects of these
decisions on the co-evolution of innovation and market structure.
This paper deals precisely with this last issue. While we drastically simplify the cognitive and
organizational aspects of the innovative process within firms, we focus instead on the analysis of
how user-producer interactions affect the (long run) evolution of technological change and market
structure in both the upstream and downstream industries: to our knowledge, no previous formal
model that examines this problem has been produced so far in the evolutionary tradition.
3. User-producer interactions and the dynamics of market structure: an overview
The consideration of the existence of technological benefits arising from interactions between
suppliers and clients introduces several interesting issues in the evolution of market structure which
have to be analyzed simultaneously. In a previous paper (Malerba and Orsenigo, 2008), we began
our investigation by starting and developing an earlier model of the evolution of two vertically
related industries (Malerba et al., 2008) and introducing the possibility that interactions between
producers and users create technological advantages in the design of the products that upstream
firms sell to their customers. Results showed that in general the existence of gains from interaction
tend to increase the rate of technical change and industrial concentration. Indeed, by assumption,
interactions between suppliers and buyers improve technological capabilities and are a source of
innovation. To the extent that innovation is associated to the establishment of some monopolistic
power – essentially through some form of success-breeds- success process - faster rates of technical
change result also in higher degrees of industry concentration, both in the upstream and in the
downstream industry. The stronger are the benefits, the higher is concentration. However, the size
and the very existence of these effect depend on many other factors, which may even reverse the
tendency towards increasing concentration and faster technical change. Not surprisingly, important
differences emerged in alternative scenarios regarding the precise nature of the advantages from
First of all, the degree to which the advantages stemming from interaction can spill over to other
users or are entirely specific to a relationship between two firms is crucial. In the first case, (the
advantages may spill over to other users) the experience and the ideas developed during a relation
between a user and a producer can be (at least partly) transferred to an interaction with a different
firm. Users-producers relationships provide here an externality for all firms. Thus, a supplier
becomes increasingly efficient the higher the number of her/his clients. This in turn increases the
chances of acquiring new contracts. Increasing returns to interaction and a form of network
externalities – interaction externalities - are in place. Under these circumstances, path-dependence is
an inherent feature of industry dynamics. A producer who – perhaps by sheer luck – manages to
establish relations with more than one user gains high probabilities to become dominant. Vertical
interdepencies propagate these effects from one sector to another. For example, a monopoly by a
supplier is associated with more competition among users, to the extent that all of them are using
the same components: the dominance upstream produces an equalizing effect downstream.
In the second case (the advantages from interaction are entirely specific), the gains from interaction
expire if the relationship is interrupted. Since mutual learning does not occur automatically and
immediately, establishing a new relationship implies starting from scratch. Thus, for example, a
user may want to select a new, potentially better, supplier but at the cost of giving up the advantages
accumulated so far: the learning process has to start anew and not necessarily the superior
efficiency of the new supplier is such to compensate for the loss of accumulated experience,
especially if the “quality” of suppliers and the potential for mutual learning cannot be immediately
and correctly evaluated. Thus a trade-off is present between the exploitation of past experience and
the exploration of new suppliers. In such a context, technological progress might be paradoxically
slower (and concentration lower), to the extent that the exploitation of the gains from interaction
hinders the search for better suppliers and their growth.
Even in this case, without spillovers, externalities and path-dependence are still present. By selling
to a user, a supplier make profits, grows, invests in R&D and further improves the quality of the
intermediate product, even not considering the additional specific advantages which arise from
interaction. Competitors in the downstream industry can switch to this supplier and benefit from the
higher quality of the components and increase their own competitiveness. Thus, the selection of a
more efficient supplier by any one downstream user entails a sort of positive externality for the
downstream competitors. In an extreme case, the choice by the current leader in the downstream
sector of the best supplier might actually imply the loss of market leadership in the future1. Success-
breeds-success processes further make the fate of individual firms and the dynamics of market
structure path-dependent: events occurring in the early stages of the process (e.g. a firm acquiring
even a small advantage in the upstream and/or downstream sector) critically affect subsequent
evolution in both industries.
1 The case of IBM in personal computers might be taken as an example: because of the size of IBM, suppliers like Microsoft
and Intel became leaders in their own markets, supplying IBM’s competitors and eventually eroding IBM’s advantages.
A second important factor which modulates the impact of users-producers relations on industrial
evolution relates to the contractual forms governing users—producers relationships. This is the
specific object of this paper. We consider here two types of contractual: length and exclusivity. We
limit our attention to the case of entirely relation-specific advantages from interaction2.
First, contracts can be either signed for a short period of time and then a different supplier might be
chosen, if deemed appropriate; or they can be fixed for longer periods, tying users and producers for
a long time. Intuitively, long-term relations may be necessary for fully capturing the benefits from
interaction. The downside might be that – absent perfect information - users may end up being tied
with inefficient suppliers. Short term contracts, conversely, may have precisely the advantage that
users can link quickly with the best available supplier. It must be noted that in a dynamic context
with heterogeneous firms having imperfect information, the problem of selecting the best supplier is
quite complex. The existence of advantages stemming from users-producers relationships
introduces a dynamic “matching” problem. To the extent that firms have heterogeneous innovative
capabilities, the matching between users and producers with different capabilities plays an
important role. Putting it in an extreme way, if a competent producer links with a non innovative
user, or vice-versa, a leveling effect can counterbalance the growth of (initially) more efficient
firms: the user benefits from the relation with the competent supplier, gaining higher market shares
than it would otherwise had, at the expense of the “best” user coupled with a lower quality supplier.
Similarly, in the upstream industry, the competent producer’s growth prospects, are lower. In
principle market selection will gradually push toward the elimination of the less efficient suppliers:
users select better suppliers as time goes by. But in a dynamic environment populated by
heterogeneous firms which may benefit from interaction, the competencies of users and producers
co-evolve and learning by interacting may complicate this process. Imperfect information on the
efficiency of the suppliers can make it more difficult to identify the “best” suppliers, especially if
the gains in technological capabilities stemming from interaction are entirely relation-specific and
therefore they are not observable by third parties. At the same time less efficient suppliers become
more innovative if they link with highly competent users (and vice versa). The issue gets even more
complex as soon as it is recognized that the advantages from interaction take time to be developed
and fully reaped.
Equally interesting is the case where users and producers establish exclusive contracts. Exclusive
contracts have the advantage for users of preventing competitors to get access to (supposedly) high
2 Contract length and exclusivity do not matter much when the benefits from user-producer relations are generic: a
monopolist quickly emerges in the component sector, irrespective of the characteristics of contracts.
quality supplies, industrial secrets or tacit knowledge, spillovers arising from interaction:
exclusivity can therefore be imposed to suppliers as a pre-condition for a contract. For producers,
however, exclusive contracts constitute a limit to growth. Therefore exclusivity should be
compensated by higher prices. In any event, exclusive contracts eliminate potential spillovers and
tend to constrain the expansion of the best suppliers. As a consequence, technological progress and
concentration in the upstream sector may be lower.
As mentioned in the Introduction, our analysis is consciously kept simple. First, we compare the effects
of alternative contractual regimes, without asking why those contractual forms are used. So, we
abstract from any consideration related to asymmetric information, hold-up problems and the like
and concentrate on the effects that alternative contractual forms have on the ability of firms to learn
from interactions and hence to gain market shares. In a final series of runs, we let firms who have
chosen to rely on contracts with different length to compete among each other. However we do not
enter the realm of contract theory.
Similarly, in this paper we do not consider the possibility of vertical integration. This issue was
investigated in Malerba et al. (2008) and – with specific reference to user-producer relations - in
Malerba and Orsenigo (2008). There, we showed that very strong advantages from user-producer
interactions may trigger vertical integration. This result emerges irrespectively of contract length or
exclusivity. A full discussion of the relative merits of (alternative) contractual forms governing
user-producer interactions vis-a-vis vertical integration, however, would introduce a whole set of
different considerations and it would require another paper.
4. The logic of the model
The model we use is a modified version of a history friendly model developed for examining the
co-evolution of the final product and component industries (Malerba et al., 2008).
The model refers to a component and a final product industry. Components and final products are
characterized by a level of quality, or merit of design (Mod). Other things being equal, the share of
market sales gained by a particular product is a function of their Mod. In the case of the final
product, the Mod is affected also by the quality of the components that go into it. Both of these
Mods tend to improve over time as a result of R&D.
Firms that sell more are more profitable than firms that sell less. High profits translate into growth.
As profitable firms expand their sales, they increase also their R&D spending and hence -
probabilistically – their Mod. Thus, more efficient firms tend over time to gain market shares.
Producers and users are linked through contracts. When seeking for components, a final product
firm scans the market for potential suppliers. The downstream firm signs contracts with a
component producer according to a probability function that reflects the relative merit of the design
of the components offered by different suppliers: the higher the merit of a component, the higher
the probability that its producer signs a contract with a final product firm. A component firm that
signs a contract sells a number of components reflecting its customer’s final product sales and the
number of components required per final product. After signing the contract, the final product firm
is tied to the component supplier for a certain number of periods. When this period expires, a new
supplier might be selected, using the same procedure.
Contracts can have different length. In addition contracts can be exclusive or non exclusive.
Exclusive contracts imply that suppliers are prevented from selling to other customers during the
length of the contract. Non-exclusive contracts allow suppliers to have more than one customer.
In some simulations we assume that all contracts have the same nature for all firms. For example, in
each set of simulations all firms are tied by short run contracts and exclusive contracts. This
assumption partially allows us also to dispense from attributing a compensation (a price) to
suppliers signing an exclusive contract. Then in other simulations, we introduce the presence of
groups of firms with different contract lengths.
Against this basic background, we introduce the possibility that user-producer interactions generate
advantages in terms of higher Mods. In our model, these advantages (called bonus) accrue directly
only to suppliers and only indirectly – in the form of higher quality components - to users3. The
advantages of a user-producer relationship (the bonus) arise only from the direct interaction of user
and a producer advantages and cannot be transferred to other firms. However, when contracts are
non exclusive, there is also an indirect effect. To the extent that the Mod of the component increases
as a consequence of the advantages from interaction, the user becomes more efficient and other
thing being equal, it will grow. The growth of the user implies larger sales for the supplier and
3 We make the strong assumption that firms do not differ in their potential for achieving advantages from interactions.
Basically, this implies that it is common knowledge that users-producers relations bring benefits and that it is indifferent to
interact with one supplier or another, if they offer a product of the same quality.
therefore larger R&D and a higher baseline Mod. Hence, independently from the bonus, the
supplier will attract new customers, who in turn benefit from the better components they get.
In the basic initial runs, it is assumed that producers and users are tied by short-term and non
exclusive contracts. Contracts have a limited duration and allow users to change frequently (if they
wish) their suppliers. A component supplier is also allowed to serve multiple customers. In the next
series of runs we introduce exclusive contracts. Then we move to long term contracts, again with
exclusivity or non exclusivity clauses.
In a final set of runs we allow for the possibility that contracts with different length coexist. In the
simulations, different groups of firm use either short or long time contracts or very long contracts.
Individual firms cannot change the length of their contracts for the all history of the simulation. Yet,
firms characterized by different contract duration compete with each other by exploiting differently
the advantages arising from interaction.
5. The Model
The model is based on Malerba et al. (2008), from which this section draws.
5.1 Final products
At the beginning of the simulation, firms (12 in the current parameter setting) start producing and
selling final products which are characterized by two characteristics: their “performance” and their
cheapness (the inverse of price). As a consequence of firms’ R&D investment, and the advance of
component technology, the characteristics of final products of a given type improve over time.
Final products are produced by combining two main inputs, which we call systems (s) and
components (c)4. The “quality” of the product, i.e. its merit of design, (Mod), M, is given by a CES
t it i
4 This formulation derives from the model developed in Malerba et al. (2008), which had the explicit aim of analysing the
computer industry. However, we believe that this representation is flexible enough to be applied to a large variety of
products and industries
with A = 1, 0 < < 1 and > -1. The elasticity of substitution is: δ = 1/ (1 + ). In the CES
functions the weight attributed to the Mod of components (τ) is higher than the weight on the Mod
5.2 Demand for final products
The demand side of the model is composed by several groups of individual buyers or submarkets.
They are identical in terms of the preferences that lead them to one type of product, but they differ
in specific behavior because they have different histories, which are partly randomly determined in
the model. Buyers respond to the products offered by different firms not only according to the
relative merit of products, but also according to other considerations, including their specific buying
history. Customers of final products are characterized by their preferences about the merit of design
of final products. Markets are represented as being characterized by frictions of various sorts,
including imperfect information and sheer inertia in behaviour, brand-loyalty effects as well as
sensitivity to firms' marketing policies. These factors are captured in a compact form by the share of
final product brands in the overall market for that type time t-1: the larger the share of the market
that a product already holds, the greater the likelihood that a customer will consider that product.
Finally, there is a stochastic element in buyers’ choices between different final products.
We represent the market process by characterizing the probability distribution of buyer’s choices
among the different final products of the desired type. This probability is a re-normalized
counterpart of a purchase propensity that depends on the merit and market share of a particular final
product. Formally, the “propensity”, Lit , for final product i to be sold to a buyer at time t is given by
) 1 (
1, 1 > 0 (2)
is the market share, 1 is the exponent indicating the bandwagon effect on final product
market, and. The probability Pri,t of the final product i being sold to a buyer at time t is given by:
In short, the demand for a final product depends positively on its Mod and on its market share in
the previous period. The probability that a final product from a specific firm is purchased by a buyer
is proportional to the “propensity,” normalized to sum to one. Then, for a final product firm the total
final products sold are equal to Mit times the number of buyers.
5.3 The market for components
At the beginning of the simulation a cohort of component producers (12 in this version of the
model) enters the market.
When a final product firm seeks component supplies, it scans the market for potential vendors.
Competition among component producers is modelled in a fashion that parallels the competition of
final product producers. First of all, here again imperfect information characterizes the market for
components: downstream firms may not be perfectly able to assess the quality (Mod) of the
components offered by different suppliers at any point in time, especially when the differences in
the Mods of the available components are small. Even more so, they are unable to predict accurately
the future quality of the components produced by the various suppliers. Hence, buyers’ choices
between different components include a stochastic element.
Specifically, the process of selection of a supplier by a downstream firm is modelled as follows.
A downstream producer rates the component producers as a function of their Mods. Like the
demand for final products, the rating of component firms is influenced by bandwagon effects
captured by the previous market share. This rating then yields a ranking of suppliers, on the basis of
which downstream producers select their suppliers. Yet, imperfect information makes this selection
process probabilistic. Thus, the “best” component producers have only higher probabilities to be
selected and to sign a contract with downstream companies. Especially when the differences among
suppliers are small, mistakes – i.e. selection of a less efficient supplier – may occur5.
) 1 (
2, 2 > 0
5 Technically, the rating mechanism ranks component firms on a segment of length one as a function of their
probability of selling. In practice, the sum of the Mods of the available components is normalized to one and individual
Mods are rescaled accordingly. Firms are placed on this segment, occupying a length equal to their probability of
selling. Then, a random number between 0 and 1 is extracted from a uniform distribution. A downstream firm selects as
a supplier the first firm whose normalized Mod is higher or equal to the extracted random number. Thus, a higher Mod
(rating) increases the likelihood to be selected as a supplier because the specific proportion of the segment is higher.
14 Download full-text
M is the merit of design of the component, LitC is the propensity of component producer i
to be selected,
is the market share of firm i in the previous period and
is the probability of
a supplier to be selected.
A component firm that signs a contract sells a number of components reflecting its customer’s final
product sales and the number of components required per final product, which in the current
simulations, for sake of simplicity, is set to one6. This assumption implies a very flat (one tier)
supply chain and simplifies drastically the analysis. After signing the contract, the final product firm
is tied to the component supplier for a certain number of periods (a parameter of the model). In the
model, short term contracts refer to 8 periods. Because each period is to be interpreted as a calendar
quarter, a short term contract lasts 2 years. Long term contracts refer to 16 periods – 4 years. In
some simulations we introduce the possibility of very long contracts: 32 periods – 8 years. When
the contract period expires, a new supplier might be selected, using the same procedure.
Component producers sell also to users other than final product firms, i.e. what we define an
external market. The size of this external market is exogenous and parametrically determined. The
external market is conceptualized in the same way as the final product market, i.e., it comprises a
number of heterogeneous buyer groups or submarkets to which component firms may sell.
However, the submarkets of the external component market are not modeled explicitly and firms
gain probabilistically a fraction of it as a function of their merit of design and of their previous
market share, as in the final product market. External demand plays an important role in the model,
since it allows component producers to grow faster than their customers: absent these external
markets, component producers would only face the derived demand from upstream firms. In
particular, the size of external demand plays a crucial role in determining the dynamics of vertical
integration by downstream firms. However, since in this paper we do not examine vertical
6 The simplest way to interpret this formulation is an assumption of a perfectly modular technology product. In this respect,
the main simplifying assumption is that – as we discuss below – the number of components (modules) required to produce a
unit of the final product is set to one and it remains constant over time. In this respect, all suppliers can be considered to be
perfect substitutes, except, of course, for their Mod. Thus, strictly speaking, we are not considering the consequences of
architectural innovations. In Malerba et al. (2008) and Malerba and Orsenigo (2008), however, we explicitly investigate the
consequences of discontinuities in component technology.