Sensitivity of the causality in variance tests to GARCH (1,1) parameters
2011 (English)Other (Other academic)Text
We discuss the sensitivity to the GARCH(1; 1) parameters in the causality of variance tests. Themotivation behind the study is to observe the impact of dierent volatile data sets on volatility spillovertests. We investigate a data generating process AR(1)-GARCH(1,1) with an extensive set of MonteCarlo experiments for dierent GARCH(1; 1) processes. It is found that the causation pattern, dueto causality between two series, is in uenced by the intensity of volatility clustering. Dierent testingprocedures are applied for testing the Causality in variance. We observe a severe size and powerdistortion when the clustering parameter is high and when the process is near integration. Thesendings are noticed when the asymptotic distribution of the statistics is used to dene a critical region.So instead of relying on the asymptotic distribution, we calculate the percentiles of true underlyingprocess with no-spillover eect and use them as a critical rigion for both size and power. We observea meaningful improvement in the results.
Place, publisher, year, edition, pages
SSRN , 2011. , 19 p.
Causality, GARCH, Spillover, Volatility
Probability Theory and Statistics
Research subject Statistics/Econometrics
IdentifiersURN: urn:nbn:se:lnu:diva-50612OAI: oai:DiVA.org:lnu-50612DiVA: diva2:911233