What are the financial determinants of extreme positive returns of US equity mutual funds? A Generalized Pareto regression approach

Abstract

This master thesis tries to investigate the dependence of the distribution of extreme positive returns of US equity mutual funds. The aim is to identify the determinants of extreme positive returns. We investigated a sample of 11,373 US equity mutual funds and covered a time interval from January 2001 to December 2019. Our set of explanatory variables includes financial, macroeconomic and fund – specific variables. We modeled the extreme positive returns by using the EVT and the POT approach. We assume that the distribution of extreme positive returns follows a Generalized Pareto distribution where the scale and shape parameters depend on the covariates. The selection of the optimal and relevant set of explanatory variables is performed by a regularization procedure based on a penalized – likelihood estimation. This regularization procedure is proposed by Hambuckers, Groll, and Kneib (2018). Lastly, we assessed the robustness of our model by comparing our results to models using a time - varying threshold to define the extreme positive returns or a different time interval.