Parts | Days | Selection | Search | Updates | Downloads | Help
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 11: Poster Session
SOE 11.9: Poster
Tuesday, March 15, 2011, 18:05–18:45, P2
Regional clustering of automotive stocks revealed by linear and nonlinear multivariate analysis — Richard Neuberg1 and •Reik V. Donner2,3 — 1Faculty of Economics, Dresden University of Technology, Germany — 2Institute for Transport and Economics, Dresden University of Technology, Germany — 3Potsdam Institute for Climate Impact Research, Potsdam, Germany
Stock markets are extremely dynamic systems with fluctuations on a broad range of different time scales. In this work, we study mutual interrelationships between the 10-day returns of stocks from a set of globally acting car manufacturers recorded for about the last decade. By applying linear as well as nonlinear dimensionality reduction techniques, we find a non-trivial regional clustering in the respective leading-order components, which allows splitting the mean sectoral market evolution into well-defined regional components. The robustness of the respective results obtained with the different multivariate analysis methods is systematically assessed. Finally, we study the temporal changes in the correlation structure between the different stocks and discuss our results in the light of varying global economic conditions.