Advanced Econometrics

27.03.2023:

The course will be taught in English. Lectures and exercise classes will take place in the classroom. For students who are not able to attend in person last year's screencasts of lectures and exercise classes will be available via the GRIPS course site.

Grading during the course (semesterbegleitenden Leistungen): There will be graded presentations of the exercises (up to a maximum of 15 points) and a mid-term exam. More details can bei found on the GRIPS course site.

Course Description

The course Advanced Econometrics builds on the Master's course in Methods of Econometrics and teaches econometric methods that significantly extend and go beyond the applicability of the multiple regression model. Both the underlying models and the properties of the estimation methods presented are analyzed. Graduates of the course should be able to acquire adequate econometric methods for more sophisticated empirical analyses.

The course will focus on nonlinear regression models as they are used for forecasting purposes, for modeling regime transitions and for estimating microeconomic behavioral and technological equations. Panel data models are also discussed. A widespread problem in empirical economic research is the endogeneity of explanatory variables, for example in simultaneous systems, in the case of omitted regressors or measurement errors with regard to explanatory variables. Despite the resulting correlation between regressor and error term, which leads to the inconsistency of conventional least-squares methods, causal effects or structurally interpretable relationships can be identified using instrumental variable estimators under suitable assumptions.

Furthermore, more general econometric estimation principles are discussed, which can be applied to a variety of questions, even beyond regression models. The generalized method of moments (GMM) does not require a complete specification of the data generating process and has become particularly popular for the estimation of individual behavioral equations of (dynamic) economic models. In contrast, the maximum likelihood (ML) principle uses all the information resulting from the specification of the entire distribution of the model variable(s). If such assumptions are appropriate, maximum estimation precision can be expected, which is why ML estimation has the status of a standard tool in the empirical sciences.

An essential part of the course is the application of the methods to economic issues. Examples covered include non-linear microeconomic cost functions in the energy sector, forecasting stock returns with non-linear smooth transition regression models, the demand for cigarettes and related tax effects in a panel data set, the relationship between (endogenous) institutions and economic development using instrumental variables, estimating the New Keynesian Phillips curve with GMM and determining the demand for doctor's visits in count data regression models with ML. The free statistical software R and its available packages are used.

Course Outline

1. Repetition and motivation
2. Nonlinear regression
3. Panel data models (fixed-effects estimators, random-effects estimators)
4. Instrument variable estimation
5. Generalized Method of Moments (GMM)
6.  Maximum likelihood estimation

Literature

Davidson, R. und MacKinnon, J.G. (2004). Econometric Theory and Methods. Oxford University Press. Corrections since publication

Further Literature

Davidson, J.(2000). Econometric Theory. Blackwell Publishers. Corrections since publication

Audience / Qualification

The course Methods of Econometrics is a necessary qualification for this course. 

Grading System

The overall grade for the course is determined by the written examination, a learning objectives test and the exercises presented in the tutorials. To pass the course, students must achieve an overall grade of no worse than 4.0. Details can be found on GRIPS.

Schedule