Organizational Adaptation on Rugged Fitness Landscapes

Luigi Marengo - University of Trento

That business firms, and economic organizations more in general, are systems of tightly coupled interdependent elements has been widely recognized by business historians (cf., for instance, Chandler (1962)) and organization scientists (e.g. Weick (1979)). But a thorough investigation of the analytical implications for the economic theory of organizations of such interdependencies (or complementarities) has been undertaken only in very recent times (cf. Milgrom and Roberts (1990)). The recognition of the central role played by interdependencies does not only have profound implications for the standard profit maximizing model of the firm, but also and perhaps more interestingly for the dynamics of leaming and adaptation in firms. In fact when applied to tightly coupled systems, selective pressure might be unable to select out harmful characteristics and drive eventually the system to optimality. Intemal constraints are likely to bind the system to suboptimal states, regardless the strength and speed of selection.

Such observations have been put forward very clearly by research in genetic biology and given analytical content by such models as the so-called fitness; landscapes model and, in particular, by Kauffman's N-K model (cf. Kauffman (1993)). They seem to have very important implications for the economic theory of the firm: if sheer fitness-based selection may prove insufficient to effectively drive adaptation and learning of an organization because of internal constraints posed by interdependencies there is room for acting on such constraints, i.e. on the organizational structure, in order to improve the adaptation and learning performance. We can see here a possible point for the market vs. hierarchies debate: organisations can prove useful in order to overcome the limitations encountered by a market selection mechanism in driving the adaptation.

This paper will attempt an investigation of this line of inquiry. In particular, I shall concentrate on the following issues: in interdependent systems mere fitness signals are not sufficient to effectively direct decentralized adaptation processes because these are also constrained by the bounds with interdependent units. A richer set of signals is thus necessary and this can be provided by hierarchically super-imposed units whose function is to complement ambiguous and insufficient market signals. Of course a perfectly knowledgeable planner could, having perfect knowledge of all the interdependencies, implement a perfect set of signals; but, leaving aside this extreme and unlikely situation, we can wonder whether an internal organizational language can adaptively emerge which complements fitness (market) signals. In the simple simulation experiments I will briefly present below the answer is yes.

For the time being I will only concentrate on adaptation of a single organization. Of course this line of inquiry has to be complemented with more population-level arguments (cf. Levinthal (1996)), but this is left for further research.

Scheduled for Session 1.4 Evolutionary Models - I