Chair of Econometrics

What is econometrics?

Econometrics helps to create data-based forecasts and assess their reliability, even if only small samples are available as a database. In addition, econometric methods make it possible to identify and quantify (average) causal relationships between economic and other variables on the basis of data. The latter also includes quantifying the reliability of these results and integrating economics expertise into the empirical analysis. Finally, econometric methods can be used to statistically evaluate hypotheses. For all of this, methods of statistical inference are of central importance and therefore form a core of econometric theory and thus of the Chair's training programme. At the heart of this is the multiple linear regression model as well as estimation methods that allow for meaningful estimation of such models under different strict requirements. In times of machine learning and AI, the understanding of statistical inference and econometric methods is more important than ever, because prominent machine learning methods such as neural networks are complex non-linear extensions of the multiple linear regression model, which must be estimated using penalisation methods, e.g. with Lasso - a method originating from statistics. All of this requires solid knowledge of probability theory and statistics as well as modern programming languages such as R or Python for the empirical realisation of the estimates. The availability of large language models in particular, which are gigantic forecasting models, requires their users to have the knowledge to be able to check their results for correctness. Teaching all of this knowledge at all levels is a central concern of the Chair of Econometrics.

An important aspect of machine learning methods is the dimensionality reduction of data. At the Chair of Econometrics, one area of our research focusses on the dimensional reduction of high-dimensional, possibly trended time series using linear factor models.