Structural reconfiguration and informal coordination in administrative organizations

This paper considers the problem of determining the optimal design of public organizations in terms of maximizing their reliability against institutional failures. To capture both the individual and the system-level aspects of organizational decisionmaking, first we present an analytical model that characterizes the optimal decision behavior of a single decision maker (unit, agent, in general: DM) in the context of a binary decision task. In this sense, reliability of a DM against the two possible error types: implementation of the wrong policy (error of comission, Type I error) and failure to act when it is necessary (error of omission, Type II error) are interpreted as the result of a particular decision strategy. Individual expertise is represented in the form of a Relative Operating Characteristic (ROC) curve that, in turn, depicts the necessary trade-off between the two errors when selecting an appropriate decision strategy. Component decisions are then combined along the lines of organizational structure which is described using a graph formalism. We show that the task of finding the best organizational design involves a joint optimization over structure and strategy, and implement the normative model in the context of a detailed example. Our numerical results suggest that when DMs coordinate their decision rules, there is little difference in the performance of various organizational structures.This work was supported by the Office of Naval Research under ONR contract #N0014-93-I-0793.