An Empirical Framework for Testing Theories About Complementarity in Organizational Design

This paper studies alternative empirical strategies for testing for complementarity in organizational design. We study a model where firms choose whether or not to adopt multiple organizational practices, and these choices interact to determine performance according to an “organizational design production function.” Unobserved variation in the returns to individual practices influences both the decision to adopt practices and realized performance. As a result, common cross-sectional empirical approaches used to test for complementarity may be subject to bias. Our analysis (a) identifies specific and plausible conditions under which bias exists, (b) analyzes whether the sign of bias can be determined, and (c) provides interpretations in terms of the underlying economic and managerial environment. In certain cases, several different procedures all result in qualitatively similar biases. We further propose a structural estimation approach that includes simultaneous equations for productivity and the demand for organizational practices. Testing for complementarity requires choice-specific instrumental variables (i.e., observable factors impacting the adoption of individual practices but independent of other practices and measured performance) and an appropriate treatment for selectivity. Our analysis offers implications for data gathering, survey design, and the interpretation of empirical evidence concerning complementarity in organizational design.