Unique Strategies and Average Treatment Effects

Abstract

What explains the increasing disconnect between empirical strategy research and real-world strategy? Although empirical methods have evolved, notably in the use of techniques that support causal identification, we argue that empirical strategy’s focus on estimating well-identified average treatment effects from a population-level model of performance is fundamentally misaligned with the strategist’s need for firm-specific insights about how to build firm-specific models that guide strategic decision making. This misalignment poses a critical problem because strategic decisions are typically one-shot, nondiversifiable, and deeply interconnected with other firm-specific choices. Thus, strategists require localized, firm-specific models and estimates rather than population-level average treatment effects. Through an extended empirical example, this paper demonstrates the limits of strategy’s current approach of ever-more-precise identification and offers guidance about methodological approaches that can help bridge the relevance gap.