This is the first of two courses in econometrics in the PhD program in economics. The goal of the module is to familiarize students with modern econometric techniques and equip them with a deeper understanding of identifying assumptions used by econometric models. The core of the course revolves around identification and estimation theory. Those foundations are later used to introduce more advanced topics with the ultimate goal of equipping future PhDs with a wide range of quantitative tools needed to independently conduct empirical research in different subfields of economics.
Core topics: we start by discussing the concept of econometric identification and how it can be achieved under different statistical and economic assumptions. We emphasise the difference between parametric and nonparametric identification, structural and reduced form approaches. We then proceed with a rigorous treatment of a commonly used class of extremum estimators, including least squares, maximum likelihood and (generalized) method of moments. Students will be expected to analyse basic properties (consistency, unbiasedness) and derive asymptotic distributions of popular estimators used to recover parameters of linear and nonlinear econometric models. We will also study how different sampling assumptions, especially time dependency in panel and time series settings, affect the set of assumptions necessary for valid statistical inference.
Additional topics: discrete choice and limited dependent variables, quantile regression, semi- and nonparametric methods (partially linear models, maximum score, kernels, local polynomials, sieves), resampling, measurement error and missing data, machine learning in econometrics (variable selection, causal inference vs. forecasting).
While major parts of the course are theory-oriented, we also emphasise the importance of practical exercises and proficiency with computer software. These exercises will include replicating results from research papers, critically assessing existing studies and implementing estimation routines on real-world data provided in-class.