ABSTRACT
This chapter presents the empirical results of models tested to explore the relationship between metro-scale employment churn dynamics and productivity growth. It begins by examining the descriptive statistics on individual dynamic process measures. First, the zero-order correlations among variables are analyzed. Second, the basic model is tested using Ordinary Least Squares (OLS) and Feasible Generalized Least Squares (FGLS) estimation techniques for comparing the estimates. Then, fixed-effect models are employed to control for unmeasured metro-specific contributions. The robustness of models is examined to get a clearer picture of the roles of dynamic processes across multiple conceptions of economic space and time. The chapter also compares the estimates obtained from both OLS and heteroscedasticity-corrected FGLS estimation techniques. The OLS technique is applied both with and without panel-corrected standard errors based on the alternative specification forms for the assumed covariance of the disturbances across the panels.
