I want to apply long-run and short-run Granger causality test for time series data. The one I find in Eviews is very simple and did not show the direction of causality as well as F-statistics and Wald test.
VECM itself tests for short and long run causality among cointegrated variables.
Mousumi Bhattacharya . Thank you, I know that for which purpose we use this test. My question is there are two types of VECM test the one is very simple which can run through eviews or SPSS. However, many paper used detailed form. I am looking for detailed one.
Please have a look the sample of relevant test.
Dear Shinwari Riazullah,
You already got long-run and short-run results with direction. If you need more info, you can come to HUST. Before coming, call me pls at 13129915142
Testing for non-causality in VECM in by far not straightforward. I think if you really need to study short and long-run non-causality, then Toda Phillips gives the way to proceed. Now if you want to test for causality in a model including a mixture of I(0), I(1) variables possibly cointegrated, then use Bauer and Maynard (JoE). This is an extension of Toda and Yamamoto and it works really fine. No need to search for cointegrating relationhips.
Philippe de Peretti Thank for ur informative sharing. actually I have used ARDL-bound cointegration test. The resutls of this prove the existence of long-run ans short-run cointegration. Meanwhile, variables are I(0) and I(1).
Hi Shinwari. Cointegration should be only among integrated variables. The ECM is a rewriting of the ARDL model. I think I would investigate Toda Yamamoto or Bauer & Maynard tests. They only requiere the maximal number of integration. Bauer and Maynard would be my first best and is really easy to implement.
For finding direction of causality from VECM, you need to run all variables as dependent variable one by one, for example :
A=B+C+D....eq 1
B=A+C+D....eq 2
C=B+A+D....eq 3
D=B+C+A....eq 4
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running