The dynamics of competitive intensity

Administrative Science Quarterly, March, 1997 by William P. Barnett

The evolutionary significance of organizational size results from this separation. To see this, compare how the baseline evolutionary process affects each of two hypothetical organizational populations, one a population of small, unitary organizations and the other a population of large, multiunit structures. Several assumptions help in this comparison. First, assume that the same amount of total organized activity occurs at the outset in both populations: There are n unitary organizations, and likewise there are n units randomly assigned across the large, complex organizations. Assume also that in each population, the n units are randomly assigned some initial fitness level, drawing in each case from a distribution with the same mean and variance. Finally, assume that both populations face the same selection threshold, set al the mean of the fitness distribution for units. By these assumptions, the two populations look almost identical at the outset, and each faces the same selection environment. The only difference is that one population has its units organized hierarchically into super-ordinate structures.

Now let environmental selection operate on each population as it moves from time 1 to time 2. For the small, unitary organizations, this is just the baseline evolutionary model: above-average units are retained and below-average units are de-selected or learn at some rate, leaving the population more fit at time 2. This process continues if we move through additional time periods, with selection or development improving the fitness of the population over each period so that older organizations generate stronger competition according to [w.sub.j] = [Sigma] [Gamma.sub.t] [Tau.sub.j], as in the baseline model.

In contrast, for the large organizations the process is different. Fitness varies at the level of the n units, but environmental selection operates at the organization level. So we must establish an aggregation rule to determine the fitness of each organization from that of its units. I assume "additive" fitness, in which the fitness of each of the large organizations is just the average fitness of its component units. This is a plausible aggregation rule, in that it equally rewards organizations for their strengths and penalizes them for their weaknesses. This aggregation of units into large organizations changes the fitness distribution. Averaging attenuates the variance in fitness at the organization level compared with the variance at the unit level, so that the fitnesses of larger organizations tend to cluster around the mean of the fitness distribution.(5) The fitness distributions of large and small organizations both have the same mean, but the distribution for large organizations has a lower variance. Along the lower tail of the distribution, this clustering results in improved fitnesses among large organizations, compared with small organizations, whereas on the upper tail of the distribution, clustering amounts to a reduction in the fitness levels of large organizations compared with small organizations.