Data analytics
Prof. Dr. habil. Rouven E. Haschka
Our research deals with the methodological and empirical analysis of causal relationships in economic models under the challenge of endogenous regressors. Methodologically, we develop procedures that do not require external instruments and thus address classical identification problems such as those caused by simultaneity or omitted variables. Central contributions lie in the further development of copula approaches for linear, non-linear, panel and generalized models as well as in the integration of Bayesian inference methods. In addition, we propose ICA-based approaches to correct for omitted-variable bias and use structural vector autoregressions (SVAR) to causally identify bidirectional relationships (such as price-sales). Another focus is on the analysis of asymmetric skewness in stochastic frontier models, which we interpret both theoretically as a signal for market characteristics (e.g. competitive intensity) and methodologically via closed skew-normal approximations.
Overall, our research aims to refine econometric identification strategies to provide robust causal evidence for policy, management and economic research.