Speed and Search: Designing Organizations for Turbulence and Complexity

TLDR

The study investigates how environmental turbulence and complexity influence the optimal formal design of organizations. The authors use an agent‑based simulation of multidepartment firms with varying designs to test responses to controlled turbulence and complexity. The simulation yields two sets of testable hypotheses: one identifies formal designs that excel in turbulent, complex, or combined environments, while the other shows that the effects of design elements on speed and search depend on department‑head powers, revealing counterintuitive outcomes such as bottom‑level processing power slowing firm improvement.

Abstract

We use an innovative technique to examine an enduring but recently neglected question: How do environmental turbulence and complexity affect the appropriate formal design of organizations? We construct an agent-based simulation in which multidepartment firms with different designs face environments whose turbulence and complexity we control. The model’s results produce two sets of testable hypotheses. One set pinpoints formal designs that cope well with three different environments: turbulent settings, in which firms must improve their performance speedily; complex environments, in which firms must search broadly; and settings with both turbulence and complexity, in which firms must balance speed and search. The results shed new light on longstanding notions such as equifinality. The other set of hypotheses argues that the impact of individual design elements on speed and search often depends delicately on specific powers granted to department heads, creating effects that run contrary to conventional wisdom and intuition. Ample processing power at the bottom of a firm, for instance, can slow down the improvement and narrow the search of the firm as a whole. Differences arise between our results and conventional wisdom when conventional thinking fails to account for the powers of department heads—powers to withhold information about departmental options, to control decision-making agendas, to veto firmwide alternatives, and to take unilateral action. Our results suggest how future empirical studies of organizational design might be fruitfully coupled with rigorous agent-based modeling efforts.