A New Approach to Causality Testing
A new causality test based on Higher Order Cumulants (HOC) is proposed in this paper. The test can be applied on non Gaussian time series. The methodological novelty is the usage of a two- step method based on digital whitening, which is performed by ARMA-HOC filter. To substantiate the method further, an empirical analysis of the relationship between the interest rate spread and real gross domestic product (GDP) growth is presented for the period 1982:q1 -2010:q1. The spread is measured as a difference between 10-year bond yields and three-month Treasury bill rates in the US. The fist step applies ARMA-HOC models to obtain white residuals from a quarterly term spread (TS) and GDP growth. The second step tests the dynamical correlation of TS and GDP growth residuals. The results show that the proposed test can capture the information about non Gaussian properties of the random variables being tested. The test is compared with the Granger-Sims causality test. The paper questions the reliability of the Granger test.
Volume (Year): 44 (2011)
Issue (Month): 1-2 ()
|Contact details of provider:|| Postal: |
Phone: +381 11 2622 357, 2623-055
Fax: +381 11 2181 471
Web page: http://www.ien.bg.ac.rs
More information through EDIRC
When requesting a correction, please mention this item's handle: RePEc:ibg:eajour:v:44:y:2011:i:1-2:p:5-14. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (Zorica Bozic)
If references are entirely missing, you can add them using this form.