The course is built around the issue of identification of production function coefficients and thus is more leaning towards empirical Industrial Organizations. We will start with classical setting of the productivity analysis and go through discussion of two most important sources of endogeniety, simultaneity selection biases, and proposed solutions to them.
Understanding of assumptions of a standard linear regression model and working knowledge of methods of panel data analysis and using instrumental variables approach are important prerequisites of the course.
After successful completion of the course, students will have acquired a scope of applications for productivity analysis, and will master several methods which have recently become standard in the literature. They include such as parametric fixed-effects and instrumental variable estimators, semi-parametric estimators by Olley and Pakes (1996), Levinoshn-Petrin (2003), Ackerbarg-Caves-Frazer (2015), and Wooldridge (2009), as well as dynamic estimators in the spirit of Arellano-Bond or Blundell-Bond (1999). Also, students will be exposed to intuitive treatment of non-parametric estimation and stochastic frontier analysis.
On the practical side, students will learn to embrace complexity of identification due to various data issues, such as attrition, non-balanced panels, pecuniary revenue vs. physical volumes of output, issue of outliers. Students will also understand possible sources of data and various limitations of such data set.