My various posts on testing for Granger non-causality seem to have been quite popular with readers of this blog.
For example, see here, here, here, here, and also see the Word-Count block in the right side-bar of this page.
The literature involving applications of non-causality testing continues to grow, even though it is now 43 years since Granger's seminal paper on the subject. Regrettably, some of these applications are lacking in various respects, but there are many that are really excellent examples of applied econometric analysis.
Contributions to the various theoretical and methodological issues surrounding the Granger causality literature also continue to emerge. Here are just a few such papers that have emerged in recent months.
- Matthieu Droumaguet & Tomasz Wozniak, 2012, "Bayesian testing of Granger causality in Markov-switchng VARs
- Christophe Hurlin & Elena Dumitrescu, 2012, "Testing for Granger non-causality in heterogeneous panels"
- Yushi Li, 2012, "Estimating long memory causality relationships by a wavelet method
- Joachim Wilde, 2012, "Effects of simultaneity on testing Granger-causality - A cautionary note about statistical problems and economic misinterpretations"
- Tomasz Wozniak, 2012, Granger-causal analysis of VARMA-GARCH models"
These papers cover a lot of important ground, and they're well worth taking a look at if you have an interest in testing for Granger non-causality.
© 2012, David E. Giles
Prof. Giles,
ReplyDeleteI appreciate the wealth of information on those recent papers on G-Causality. However, I seem to have difficulty in finding papers that deals on whether G-Causality is useful for policy analysis. Like for example the presence of causal relationship between food self-sufficiency and poverty or between stable food prices and poverty.
In a commentary in 2003, Granger seems to admit that it is better to have an economic theory first or prior knowledge before testing for causality. In Granger (1988), he suggested that G-Causality is useful for policy evaluation but results should be interpreted with care. Any thoughts? Thanks!
Thanks for the comment. First, I agree that an economic theory base should precede any testing for (non-)causality. When it comes to policy analysis, I think the same degree of caution is appropriate as with any econometric modelling. It's a useful input into the process; if done well it has the advantage of forcing you to expose the assumptions you are making; but it's rarely "definitive".
DeleteDG