Current Projects
The informative value and generalizability of panel surveys depends crucially on whether the panel participants represent the population well. Studies show that the response rates for surveys are falling. This is a major problem for panel surveys in particular. The aim of this project is to contribute to the explanation of non-response in panel surveys and to develop methods to determine the nonresponse bias of estimators.
How well does competition work on fuel markets? By analyzing high-frequency traffic flow data collected by the German Federal Highway Research Institute (BASt, Bundesanstalt für Straßenwesen), this project aims to better explain the causes of price cycles at filling stations. The answer to the question of causes is a crucial part of any political discussion on retail fuel markets, e.g. on taxation, regulation or market transparency. The project is funded by the DFG.
Price algorithms are becoming increasingly important. However, a major problem with price algorithms is that they can prevent competition between companies without explicit agreements and thus lead to excessive consumer prices. In contrast to explicit price agreements between people, self-learning reinforcement algorithms (such as Q Learning) leave no evidence that can be used by the competition authorities. This project analyzes price algorithms in "hybrid" interactions between human actors and algorithms.
Current research findings make this clear: Non-cognitive skills have strong predictive power for key life outcomes such as educational attainment, career success or health. Despite their enormous importance, how non-cognitive skills develop in childhood and adolescence remains largely unexplored. This project aims to close this research gap. The project is funded by the DFG.