Interdependence, Complementarity, and Ruggedness of Performance Landscapes

Abstract

Interdependences among organizational choices are widely seen as central to the mapping between those choices and performance outcomes (i.e., performance landscape). However, interdependency lacks specificity, and there is little agreement on how the definition and features of interdependencies shape performance landscapes. For example, prior formalizations split between unimodal production functions and rugged landscapes with many local peaks, and scholars disagree on how complementarity among choices, an empirically salient property, impacts the ruggedness of landscapes. I introduce a new formalization of performance landscape that builds on the empirically familiar notion of interaction effects among organizational choices. Studying the link between features of interdependency and the shape of performance landscape, I find that (1) the number of local peaks may grow much more slowly with the number of choices (N) than previously formalized; (2) complementarity significantly reduces the number of peaks—in fact, under maximum complementarity (i.e., supermodularity), one rarely observes more than two local peaks, and that number declines with N; and (3) modeling the evolution of landscapes in light of innovation offers a bridge between the divergent views of performance landscapes. Over time, performance-enhancing (innovative) mutations increase complementarity and reduce the number of local peaks, potentially to a single one. Results clarify the link between complementarity and the shape of landscapes, highlight the relevance of task structures with a few local peaks where management’s role is more salient than either extreme, and provide an explanation for the emergence of complementarity among organizational choices, as well as dominant designs in industry life cycles.