From perfect to practical: Partial identification methods for causal inference in strategic management research

• 成果类型: Article
• 署名作者: Frake, Justin; Gibbs, Anthony; Goldfarb, Brent; Hiraiwa, Takuya; Starr, Evan; Yamaguchi, Shotaro
• 署名单位: University of Michigan System; University of Michigan; Purdue University System; Purdue University; University System of Maryland; University of Maryland College Park; University of Wisconsin System; University of Wisconsin Madison
• 刊物名称: STRATEGIC MANAGEMENT JOURNAL
• ISSN/ISSBN: 0143-2095
• DOI: 10.1002/smj.3714
• 发表日期: 2025
• 页码: 1894-1929
• 关键词: Causal Inference; partial identification; endogeneity; instrumental variables; difference-in-differences

摘要

Research SummaryStrategy and management scholars have increasingly used difference-in-differences (DD) and instrumental variables (IV) designs to identify causal effects. These methods rely on untestable identifying assumptions to interpret the results as causal. partial identification techniques allow researchers to draw causal inferences from imperfect identification strategies by quantifying how results change with the severity of a violation of the identifying assumption. We explain how these tools work in the context of DD and IV designs, provide practical guidance to apply them, and illustrate their use in an empirical example that investigates how first patents affect inventor mobility. In doing so, we emphasize the role of theory, context, and judgment when deciding how strongly to infer a causal relationship from an empirical result.Managerial SummaryManagers seeking to understand the causal effects of their strategic decisions may struggle to do so when their choices cannot be randomized. In such cases, difference-in-differences (DD) and instrumental variable (IV) approaches may be a viable estimation strategy. However, these methods still rely on untestable identifying assumptions and it may not be clear how to interpret the results if those identifying assumptions do not hold. In this study, we describe how partial identification methods for DD and IV designs allow managers to draw causal inferences even when the identifying assumptions do not hold exactly. We explain how these tools work, provide practical guidance to apply them, and illustrate their use in an empirical example that investigates how first patents affect inventor mobility. In doing so, we emphasize the role of theory, context, and judgment when deciding how strongly to infer a causal relationship from an empirical result.