- 23 Sep 2013
- Working Paper
Applying Random Coefficient Models to Strategy Research: Testing for Firm Heterogeneity, Predicting Firm-Specific Coefficients, and Estimating Strategy Trade-Offs
Executive Summary — Textbooks generally define firm strategy as a set of decisions focused on managing organizational trade-offs in order to achieve long-term competitive advantage. Although strategy models theorize why the same actions by different firms lead to different effects on firm performance, empirical work typically estimates the average effect of an action across firms. The authors discuss how Random Coefficient Models (RCMs) can close the gap between theoretical and empirical research in strategy. Among other advantages to using RCMs, researchers can make a critical distinction between firm actions (or explanatory variables) that are statistically significant and those that are strategically significant. Key concepts include:
- The next-generation methodology of Random Coefficient Models (RCMs) has already been used extensively in education, biostatistics, political science, and other fields to identify, model, and leverage unobserved differences at the individual level.
- In non-technical terms, RCMs can be described as generalized versions of standard methods such as OLS, probit, logit, multinomial logit, and so on.
- The application of RCMs in strategy as a discipline would significantly advance the field by allowing scholars to directly test the many strategy theories based on firm-level differences.
Although Strategy research aims to understand how firm actions have differential effects on performance, most empirical research estimates the average effects of these actions across firms. This paper promotes Random Coefficients Models (RCMs) as an ideal empirical methodology to study firm heterogeneity in Strategy research. Specifically, we highlight and illustrate three main benefits that RCMs offer to Strategy researchers--testing firm heterogeneity, predicting firm-specific effects, and estimating trade-offs in strategy--using both synthetic and actual datasets. These examples showcase the potential uses of RCMs to test and build theory in Strategy, as well as to perform exploratory and definitive analyses of firm heterogeneity.