Advanced

Explaining the coherency of national stock indices with macroeconomic variables: Time-series correlation and Cross-sectional correlation approaches

Spies, Michael LU (2010) NEKM01 20101
Department of Economics
Abstract (Swedish)
The phenomenon of increasing correlation between asset returns in economic downturns will be investigated with two different approaches and tried to be explained by different macroeconomic variables. The first approach, namely the classic method of measuring correlation with time series is contrasted with an extended method of cross-sectional correlation measurement proposed by Solnik (2000). The method was applied to sub-indices of the German stock market. Adjacent to the sub-index returns several macroeconomic variables were used in OLS regressions as regressors. In order to test for time variability of the variables’ explanatory power subsamples were built. The models were tested with
monthly data starting in January 1991 and ending in... (More)
The phenomenon of increasing correlation between asset returns in economic downturns will be investigated with two different approaches and tried to be explained by different macroeconomic variables. The first approach, namely the classic method of measuring correlation with time series is contrasted with an extended method of cross-sectional correlation measurement proposed by Solnik (2000). The method was applied to sub-indices of the German stock market. Adjacent to the sub-index returns several macroeconomic variables were used in OLS regressions as regressors. In order to test for time variability of the variables’ explanatory power subsamples were built. The models were tested with
monthly data starting in January 1991 and ending in December 2009. Furthermore, several econometric tests were accomplished to evaluate the econometric quality of the different approaches. Several results were found: The classic time series approach outperforms the cross-sectional approach in terms of econometric quality. Moreover, the former backed the
theory of increasing correlations in down-states whereas the latter could not. Nevertheless, the findings of the regressions were very similar: No variable is consistent enough to be used as predictive variable, but in general the amount of credits given to enterprises and the number
of unemployed people help to explain return correlation movements over time. However, all regressors suffer from time variability. Splitting the results to the different sub-indices and its appendent correlations gives further sector specific results. (Less)
Please use this url to cite or link to this publication:
author
Spies, Michael LU
supervisor
organization
course
NEKM01 20101
year
type
H1 - Master's Degree (One Year)
subject
keywords
time-series correlation, cross-sectional correlation, downside risk, DAX
language
English
id
1611760
date added to LUP
2010-06-14 13:38:27
date last changed
2010-06-14 13:38:27
@misc{1611760,
  abstract     = {The phenomenon of increasing correlation between asset returns in economic downturns will be investigated with two different approaches and tried to be explained by different macroeconomic variables. The first approach, namely the classic method of measuring correlation with time series is contrasted with an extended method of cross-sectional correlation measurement proposed by Solnik (2000). The method was applied to sub-indices of the German stock market. Adjacent to the sub-index returns several macroeconomic variables were used in OLS regressions as regressors. In order to test for time variability of the variables’ explanatory power subsamples were built. The models were tested with
monthly data starting in January 1991 and ending in December 2009. Furthermore, several econometric tests were accomplished to evaluate the econometric quality of the different approaches. Several results were found: The classic time series approach outperforms the cross-sectional approach in terms of econometric quality. Moreover, the former backed the
theory of increasing correlations in down-states whereas the latter could not. Nevertheless, the findings of the regressions were very similar: No variable is consistent enough to be used as predictive variable, but in general the amount of credits given to enterprises and the number
of unemployed people help to explain return correlation movements over time. However, all regressors suffer from time variability. Splitting the results to the different sub-indices and its appendent correlations gives further sector specific results.},
  author       = {Spies, Michael},
  keyword      = {time-series correlation,cross-sectional correlation,downside risk,DAX},
  language     = {eng},
  note         = {Student Paper},
  title        = {Explaining the coherency of national stock indices with macroeconomic variables: Time-series correlation and Cross-sectional correlation approaches},
  year         = {2010},
}