Applications of modern regression techniques in empirical economics

The development of models that go beyond traditional linear regression has been a topic of great interest in statistical research over the last years. As a consequence, a powerful toolbox has emerged, allowing for a realistic modelling of a variety of real data problems. This thesis presents several applications of modern regression techniques and addresses various issues that are currently being discussed in the economics literature. In the first analysis, we semi-parametrically model conditional quantiles of farmland rental rates using Bayesian geoadditive quantile regression. The second analysis investigates the multifaceted dimension of upward social mobility in the United States by modelling all parameters of a multivariate response distribution as a function of covariates employing Bayesian distributional regression. The third analysis is concerned with the optimality of investment decisions and explores the influencing factors on the timing of investment decisions applying Generalised Additive Mixed Models. The results of our analyses are of potential interest for academics and policymakers since the use of advanced regression models allows for revealing additional information in the data that will remain undetected if more traditional models are used