A wavelet improved test for Granger causality in the presence of GARCH effects

Author

Månsson, Kristofer

Author_Institution

Dept. of Econ. & Stat., Jonkoping Univ., Jönköping, Sweden

fYear

2010

fDate

18-20 June 2010

Firstpage

121

Lastpage

123

Abstract

The size and power of the F-test and a new wavelet-based approach for testing for Granger causality is evaluated in this paper by means of a Monte Carlo study in which the error term follows a GARCH process. The Monte Carlo simulation includes 4 different data generating processes (DGP) and it shows that the F-test tends to over-reject the true null-hypothesis under the presence of GARCH errors. The study also shows that the new wavelet-based approach improves the size of the Granger causality test for all different DGPs.

Keywords

Monte Carlo methods; autoregressive processes; causality; wavelet transforms; F-test; GARCH effects; Granger causality; Monte Carlo study; data generating process; generalized autoregressive conditional heteroskedasticity consistent process; null-hypothesis; wavelet improved test; Discrete wavelet transforms; Error analysis; Filtering; Global Positioning System; Monte Carlo methods; Noise reduction; Power engineering and energy; Power generation economics; Statistical analysis; Testing; GARCH; Granger causality; Power JEL Classification: C32; Size; Wavelet;

fLanguage

English

Publisher

ieee

Conference_Titel

Financial Theory and Engineering (ICFTE), 2010 International Conference on

Conference_Location

Dubai

Print_ISBN

978-1-4244-7757-9

Electronic_ISBN

978-1-4244-7759-3

Type

conf

DOI

10.1109/ICFTE.2010.5499413

Filename

5499413