1

Market Partitioning Under Varying one-Dimensional Resource Spaces

Market Partitioning Under Varying One-Dimensional Resource Spaces

César E. García-Díaz

Department of International Economics & Business

Faculty of Economics

University of Groningen (The Netherlands)

April 21, 2004

1. Introduction

It has been almostthree decades since Organizational Sociology witnessed the emergence of Organizational Ecology (OE), a research program that focus in the study of populations of organizations and the principal effects on organizational founding and mortality rates (Hannan & Freeman, 1989; Witteloostuijn, 2000). OE tries to give answers to the fundamental question “why are there so many kinds of organizations?” (Hannan & Freeman, 1977: 936) using a Darwinian approach of environmental selection features (Levins, 1968), by which explicitly opposes to the classical rational adaptation theories of organizational theory (Kamps, 2000). One central claim of OE states that, when facing environmental changes, the organizations of the population not able to fit the new conditions are replaced by new ones and a population-level adaptation occurs. This also implies that a selection process takes place at the individual level (Hannan & Freeman, 1977).

In its attempt to understand population-level adaptation through the behavior of founding and mortality rates and the proliferation of organizational forms, Organizational Ecology (Hannan & Freeman, 1977, 1989) has brought alternative views to classical organization theory’s contingency approach regarding optimal strategies in uncertain environments through the formulation of Niche Width Theory (Freeman & Hannan, 1983; Péli, 1997; Bruggeman, 1997; Bruggeman & O’Nualláin, 2000; Baum & Amburgey, 2002; Hannan, Pólos & Carroll, 2003), to organizational change theories through the formulation of Structural Inertia Theory (Hannan & Freeman, 1984; Péli et al., 1994; Péli et al., 2000; Hannan, Pólos & Carroll, 2003) and to economists’ neoclassical theory regarding the understanding of markets composition. Organizational Ecologyhas also introduced its own vision of market structures and, as emphasized by Vermeulen & Bruggeman (2002) and Carroll & Hannan (1995), has proposed opposite perspectives to classical industrial organization theory views on the role and effect of market concentration, as is the case in Resource Partitioning Theory (Carroll, 1985).

Carroll (1985)’s seminal paper in Resource Partitioning Theory (RPT) in Organizational Ecology gives explanation about the coexistence of generalist organizations with specialist organizations in a two-dimensional resource space characterized by scale economies and a center. RPT emphasizes that some necessary (but no sufficient) conditions are needed for such dual market structure: heterogeneity of resources, the presence of a market center and economies of scale/scope.

Although Witteloostuijn & Boone (2003) develop a market structure typology with eight typical cases, there is no theoretical investigation that connects the emergence of such typology with different sets of initial conditions for the n-dimensional resource space in which such market evolves. A theory to explain the emergence of these market configurations is needed. Some attempts like the first-order logic model developed by Vermeulen & Bruggeman (2001) have been made. However, this model states that resource partitioning occurs independently of organizational mass, size-localized competition, diversifying consumer tastes and changes in niche width, missing important elements needed to fully understand the dynamics process that generates market partitioning.

Through computer simulation of different one-dimensional resource spaces, which differ in its level of resource heterogeneity, we want to understand how such different resource spaces generate specific partitioned markets. We want to explore which are the thresholds in the degree of homogeneity at the market center that account for specific levels of generalist concentration (and consequently, it will be our interest to explore if size-localized competition (Baum & Amburgey, 2002) is effectively related to size or if it is a consequence of certain levels of homogeneity of the resource space). In the search of conditions for sufficiency in market partitioning, we also explore which are the heterogeneity thresholds that allow market partitioning to appear.

2. Resource Partitioning Theory

2.1. Fundamental ideas

Carroll (1985)’s shaping paper in the theory of resource partitioning gives explanation about the coexistence of generalist organizations with specialist organizations in a market characterized by scale economies and a center. Generalist and specialists organizations are, in the context of resource partitioning theory, differentiated by the range of resources they take from the resource space: generalists are characterized by taking a broader portion of resources, while specialists take a narrower segment.

Generalists/specialist definition is slightly different from the original inter-temporal definition of OE’s niche theory, in which both specialists and generalists define the width of their niches among range of resources that are not present simultaneously throughout time (Wezel & Witteloostuijn, 2003).

The resource space might be, for instance, the n-dimensional space of product space characteristics, where each n-vector represents a specific combination of consumer preferences (Péli & Nooteboom, 1999).

Resource partitioning assumes that the center of the market is abundant in resources (it means that the resource distribution is unimodal), so that intensive competition might be hold around the center. Provided that the total resource space has a fixed carrying capacity[1], such intensive competition will generate two different and simultaneous outcomes:

a)Some of the firms that fight for taking resources at the market center will not succeed and will have to leave the market. It means that the fewer winning organizations will take over the market center and, as a result, the market concentration will increase. Those successful organizations could grow big (generalists) due to the departure of some of their competitors and the advantages provided by the scale economies, but will be less in number as well. Also, “generalists tend to differentiate themselves by differentiating their product offers, positioning their niches apart from each other” (Péli & Nooteboom, 1999:1135). In other words, “as market concentration rises, the amount of unique resource space covered by the combination of all generalist organizations contracts” (Carroll et al. 2003:14).

b)Since the total resource space owned by the all the generalist organizations is contracted, the peripheries of the resource space will be available for organizations that might take resources from this non-abundant portion of the resource space. Those “small” organizations (specialists) might proliferate as the market concentration rises[2],[3]. It means that the consolidation of generalists creates the conditions for specialist proliferation: “The surviving generalists thus become larger and more general as time passes. However, because of the ever-widening range of the surviving generalist’s target area, it become increasingly difficult to secure the entire free area. This is because doing so involves uncertainty, it might prove very difficult or more costly than it is worth, or it might entail loss of some of the organization’s existing target areas because its identity or capabilities would be undermined” (Boone, Carroll & Witteloostuijn; 2002:272).

So far, there are two explored sources of resource space release for specialists: first, the intensive competition among generalists, which produces that some of them leave the market, while others conquer the freed space near the market center (Carroll, 1985; Carroll et al., 2003). Second, the emergence of new tastes due to specialists’ effort to enter the market and differentiate themselves (Carroll & Hannan, 1995). The latter effect is explained by changes in the dimensionality of the resource space (Peli & Nooteboom, 1999)[4].

2.2. Latest Developments

Hannan, Carroll & Pólos (2004a,b) undertake the task of unifying niche theory and RPT. They faced many differences and tensions: i) Niche theory defined niches under absence of competition (fundamental niches) while RPT emphasizes niche differentiation under competition (realized niches)[5], ii) environmental conditions vary between two environmental states and depending on variability, grain and dissimilarity of the environmental states, while RPT consider one single environmental configuration, or better, several similar resource conditions at the same time iii) in RPT, generalist can perform simultaneously in the range of environmental resources in which they define their niches, while in niche theory they alternate their operations in two non-simultaneous environmental states, iii) niche theory states its predictions without assuming scale economies or nonhomogenous environmental distribution, unlike RPT.

With the aim of integration of these two fragments, Hannan, Carroll & Pólos (2004a,b), redefine the principle of allocation, stating that the level of organizational commitment or engagement is fixed for a population at a given point of time. Hannan, Carroll & Pólos (2004a,b) make a distinction between two kinds of appeal (i.e. intrinsic and actual). Intrinsic appeal refers to the degree to which an organizational offering fits a specific taste due to a socio-cultural affinity. The intrinsic appeal is the abstract “attraction” between the taste and the offering, which becomes materialized by factors like availability of the offering or the way it is presented to customers (i.e. through the capacity of engagement of the organization). So we can say that the level of actual appeal depends on the capacity of organizational engagement in the social positions in which it has intrinsic appeal. Note that it is assumed that every single social position in a social space has a unique taste and that, consequently, tastes vary among social positions.

Given the above framework, the fundamental niche is redefined as the set of positions for which an organization’s offering has nonzero appeal (Hannan, Carroll & Pólos (2004a,b)[6]. In order to reconcile the niche theory’s fundamental niche perspective with the RPT’s realized niche viewpoint, the latter is defined as the set of positions that generate positive returns for the organization. The organization’s return is calculated by multiplying the organization’s “share” in each social position among all the organizations that compete for the same taste (which is defined as organization’s fitness) with the “total expenditure” of each social position (given, for instance, by number of consumers at that position). Considering that an organization gets positive returns from the set of positions with nonzero actual appeal, it can be proved that the organization’s realized niche is a subset of its fundamental niche (Hannan, Carroll & Pólos, 2004a).

Taking a step further in the explanation of the resource partitioning process, Hannan, Carroll & Pólos (2004a) divide the market in three distinguishable segments: the center, the near-center and the periphery, and try to explain why the near-center organizations, as concentration rises, become prone to failure while the organizations in the periphery raises their survival chances. Hannan, Carroll & Pólos (2004a) leave apart the definitions of generalists and specialists, and use the maximum scale advantage as a proxy for concentration, given that scale is positively correlated to size (Carroll & Swaminathan, 2000) and that concentration is positively correlated to the size of the largest organization in the market. For a dynamic perspective, it is necessary to take into account two kinds of effects in this partitioning process: size-localized competition and scale-based competition.

Given size-localized competition effects (Péli & Nooteboom, 1999; Baum & Amburgey, 2002), the near-center organizations tend to disappear because they face strong competition with both center-located and peripheral organizations. As mentioned before, specialists and generalists, which are different in size and structure, don’t compete directly and coexist in the same resource space.

On the other hand, the scale-based competition hypothesis states that “among scale-based (generalist) competitors within an organizational population, the greater the sum of distances of a firm from each of its larger (generalist) competitors, the higher its mortality hazard” (Carroll et, al, 2003:12).

Carroll & Swaminathan (2000) point out that there is no contradiction in the two concepts: while the sized-localized competition applies to every organization in the population, the scale-based competition applies only to the generalists. As mentioned before, the small generalists will face the most fierce competition because they will experience an increase in their mortality rates due to both effects:i) size-localized competition with specialists and,ii) scaled-based competition with other generalists. This situation will eventually generate the partitioning of the market[7].

RPT has concentrated in explaining interaction in a fixed environment.The consideration of changing resource spaces and its effects on market structures will extend further the theory through the consideration of spatial and temporal changes in social positions’ tastes. “Market structure and firm conduct are co-determined by the underlying features of the environmental resource” (Witteloostuijn & Boone, 2003:7).

3. A simple discrete-event simulation model

In this section we present a conceptual framework for a simple discrete-event simulation for market partitioning in one-dimensional resource spaces, which will allows us to study the evolution of market concentration, organizational density and the relative performance of specialist organizations versus generalist organizations[8]. This is, we want to investigate the effects on concentration and density under varying degrees of resource heterogeneity. To keep the model simple, and to use it as a step stone for future works, we explore the effect of varying heterogeneity in one-dimensional resource spaces.

i) Resource space generation: there is a one-dimensional resource space with a market center, which represents the space of social positions, where it is supposed that there is a number n of social positions. The space has a Gaussian-like distribution and will be generated for several degrees of resource heterogeneity (represented by its variance, ²). The space is generated as a discrete rendering derived from a Gaussian distribution. Each social position generated Sj, j=1,…, n, has an associated “budget” bj, j=1,…,n. This budget represents the amount of resource offered by each Sj, j=1,…, n. For convenience, the possibility of considering a range of tastes in each single social position Sj is not taken into account. The resource space generation is represented by two (n x 1) vectors, S and b.

ii) Organizational founding: Simulation horizon is divided equally into T discrete time intervals i, i= 1,…,T, each of them representing a single iteration in the simulation environment. Firm will enter the market following with a Poisson process with an arrival rate (per time interval) of i, which means that P(arrival of any firm at interval i)= ixe-i/x! Although a more precise model will have the rate i dependent on the number of organizations in the market, we won’t do this in our model just to keep it simple. That is, i = , i = 1, 2,,…,T. The dependency of the arrival rate on the number of firms in the market might be allowed to integrate density dependence theory and resource partitioning (at least regarding founding rates), since founding events are somehow related to processes of legitimation, and such founding events are dependent on crowding effects. Noteworthy to say is that such assumption of considering i as a function of density will work out to represent density dependence, but not to prove it. Outcome will be recorded in the (1xT) vector X.

iii) Endowment: Although it might be assumed that firms have some endowment at the beginning of the simulation, this feature will not be considered, since it is our interest to primarily focus on studying the process of resource partitioning. Regarding the way this simulation is built, the inclusion of organizational endowment will delay the emergence of market partitioning since the firms will be allowed to have some amount of negative profits and will remain in the market. We think that this inclusion of endowment will just give us a different cut-off point to select firms out, but won’t modify the behavior of market formation throughout the simulation runs. Obviously, if the simulation model includes some other features (e.g. density dependence), endowment do count for market structure emergence.

In addition, it might be argued that a difference in the final results would be generated if some firms, in spite they have got losses, are able to survive and re-enter the market again. For that reason, we assume that a firm with no social positions (provided that it has positive cumulative profits) will remain in the market and will have the opportunity to fight again for niche positions (i.e. re-entry costs to market are cero).

iv) Niche center / niche width selection: Let us assume that Oi,k is the k-th firm that appears in the market at iteration i. In general terms, each organization generated in each step, Oi,k, i=1,…,T; k=1, 2, 3,… chooses a strategy that has two components: The niche center, pi,k, and the breadth of the niche, wi,k. At iteration i=1 each firm chooses its niche center, pi,k, whose value will remain constant during the whole time horizon. Oi,k’s niche width will have an starting value for all organizations, wo, which will be updated in subsequent interactions, depending on the outcomes of price competition and economies of scale (i.e. t=2,…,T). The niche position pi,k of each Oi,k is chosen randomly according to a uniform distribution. The arguments for such assumptions come next.

We can say that the fact of picking up a given position pi,k by a firm is partly a consequence of identifying resource peak points by the firm actors. Those points of peak intrinsic appeal (Hannan, Pólos & Carroll, 2004a,b) are chosen by entrepreneurial activities and, supposedly, such activities are more likely to be carried out in the most abundant regions of the resource space.

As found in empirical research (Boone, Carroll & Witteloostuijn, 2002), the more concentrated the resources are, the higher the concentration among generalists. It means that the effects of concentration might be partly a consequence of crowding along the process of niche center selection. So it is expected that founded firms, at the beginning, are more likely to locate their niche centers in the most abundant region, rather than in the peripheral regions. However, when the market gets crowded, the risk of failure is higher in the region near the center since competition is expected to be higher in such area respect to the level of competition in non-abundant regions of the resource space. In the absence of more detailed assumptions, thetrade-off between attractiveness and risk that entrepreneurs leads us to consider that the generation of niche centersmight be distributed uniformly.

As mentioned before, we assume that firms are totally inert, so pi,k remains unchanged throughout the time horizon, and no possible niche center shifts are possible. Firms that are not able to maintain themselves in the starting position abandon the market and leave their niches free. The outcome of niche centers at each iteration t is stored in a matrix(T x max(X1, X2, X3,..., XT)) named P.