Dear Tom,
I came across non-linear causality tests by Kyrtsou and Labys(2006) in the following paper:
https://www.sciencedirect.com/science/a ... 0405000881
Is it possible to conduct these tests in RATS?
Please reply!
Thanks and Regards.
Kyrtsou and Labys(2006)
Re: Kyrtsou and Labys(2006)
They say that you just do OLS to estimate the model and do a standard F.
However, they never really indicate why the Mackey-Glass model is likely to be a reasonable model of the dynamics. (Note that Mackey-Glass was developed for physical processes, not economic ones). If the dynamics are wrong, the test will be wrong. The whole literature on non-linear causality tests is dubious. This seems to be way too specific on the form of non-linearity, while others such as Hiemstra-Jones are too vague (and consequently was shown to be simply wrong).
However, they never really indicate why the Mackey-Glass model is likely to be a reasonable model of the dynamics. (Note that Mackey-Glass was developed for physical processes, not economic ones). If the dynamics are wrong, the test will be wrong. The whole literature on non-linear causality tests is dubious. This seems to be way too specific on the form of non-linearity, while others such as Hiemstra-Jones are too vague (and consequently was shown to be simply wrong).
Re: Kyrtsou and Labys(2006)
Okay. Thanks for your reply Tom!
Regards.
Regards.