First, this paper specifies a structural model of the firm, the standard principal-agent model augmented with an investment decision, and then uses that model to conduct empirical work on the connection between performance and ownership. We calibrate the model exactly to data on managerial ownership and the level of investment in productive assets from Execucomp and Compustat. For each firm-year observation, this generates estimates of structural productivity parameters for both investment and managerial input. Based on variation in these exogenous parameters, we find that Tobin’s Q and managerial ownership exhibit the patterns documented in McConnell and Servaes (1990). Thus, our augmented principal-agent model can explain the hump-shaped empirical relation between performance and managerial ownership. No additional factors, such as managerial entrenchment overtaking incentive alignment at high ownership levels, are required. Second, the calibration creates a data panel for which we know the underlying structural model and appropriate empirical specification. This allows us to quantify the statistical and economic importance of specification error and endogeneity in empirical work. Including firm fixed effects or controls for firm size (investment or sales) adds explanatory power, but the spurious relation between Q and managerial ownership typically remains. In this setting, standard approaches to the

This paper specifies a structural model of the firm, the standard principal-agent model augmented with an investment decision, and then uses that model to conduct empirical work on the connection between performance and ownership. We calibrate the model exactly to data on managerial ownership and initial investment in productive assets from Execucomp and Compustat. In addition, we find that Tobin’s Q and managerial ownership exhibit the patterns documented in McConnell and Servaes (1990). Thus, our augmented principal-agent model can explain the empirical relation between performance and managerial ownership. No additional factors, such as managerial entrenchment swamping incentive alignment at high ownership levels, are required. The calibration creates a data panel for which we know the underlying structural model and appropriate empirical specification. This allows us to examine the importance of specification error and endogeneity in empirical work. Including firm fixed effects or controls for firm size (investment or sales) adds explanatory power, but the spurious relation between Q and managerial ownership remains. In this setting, standard approaches to the endogeneity problem fail to provide a solution. The endogeneity problem is substantial and it is difficult to correct using control variables and fixed effects. Nevertheless, our procedure illustrates how a structural model of the firm can isolate the