An Empirical Framework for Testing Theories About Complementarity in Organizational Design

This paper analyzes alternative empirical strategies for studying the effect of organizational design on performance, as well as the factors which determine organizational design. We begin by developing a general economic model of a firm where multiple organizational design practices are endogenously determined. The model includes exogenous variation in the costs and returns to each of the individual practices, which is the source of the heterogeneity among firms. We then argue that in many empirical applications, some of the exogenous costs and benefits of each organizational design practice will be unobserved by the econometrician. We use this model to evaluate how different econometric strategies perform under alternative assumptions, focusing on how the presence of unobserved heterogeneity impacts the interpretation of different empirical procedures. For example, we show that some of the main testing strategies which have been used in the empirical literature on organizational design cannot distinguish between two hypotheses: (H1) a set of choice variables are mutually complementary, and (H2) the choice variables do not interact in determining profits, but the unobserved returns to individual practices are positively correlated. We propose a structural approach to the problem which involves the estimation of a system of equations, including both a productivity equation as well as equations which describe the firm's optimal choices. These equations explicitly incorporate the presence of unobserved heterogeneity between firms. We analyze this approach theoretically and using simulated data, comparing it to other approaches which make different assumptions. The analysis highlights the tradeoffs between approaches in terms of computational burden, data requirements, precision, and sensitivity of the results to the assumptions required