Toda Yamamoto Causality Test in Eviews

[@charifahaouraji7501]

Dear sir, thanks for all your videos.. in my research, i found that my 4 variables are cointegrated ( by using ARDL model) so I run the Toda Yamamoto test to verify the causality... i found that GDP variable not cause any variable and vis versa... this result is normal or cointegration means that all variables should have at least one causality

[@ChekwubeMadichie]

Even though cointegration entails that causality may run in at least one direction, it does not necessarily imply causation. There are occasions where cointegration does not guarantee causality among variables.

[@yusauaudu9044]

Good day, please, what is the best model to adopt when you have a time series data of order 1 and 2 integration?

[@ShivaYadav-wh5jg]

I am getting confused one thing is that to run Toda Yamamoto model, whether variables in level data is to be used or after making variable stationary?

[@philippetrape9295]

Thank you very much for a very didactic and well-explained video - as always. I have just one question regarding the conclusion about the sense of causality: what is the null hypothesis for the Chi-Square statistic? I see that in both cases, GDP and Top, the p-value is 0, meaning the null is rejected. In that case, in the first iteration, the null would generally be that there is no causality from TOP to GDP, GDP being the dependent variable. If the null is rejected, this means that Top is causing GDP. Is that correct? Regards.

[@birkonsevbirkonsev9009]

Very helpful thank you.

[@ChekwubeMadichie]

You're welcome sir. Keep sharing my videos to your friends and colleagues.

[@ubanikatchy5861]

Good day sir, does the Toda Yamamoto test on Eviews work for only two variables ? Like one has to run the test with the dependent vs the independent variable ?

[@manuomprakashsharma5303]

Sir would u be kind enough to make a vedio on asymmetric causality! Any good way to learn that how that can be done in eviews

[@owendeboer5525]

Dear Sir, thank you for the illustrative model. Hypothetically, if I am sure that both of your variables are I(1), would this mean that for the final step, both exogenous variables would be attributed a minus lag length of (-12) rather than (-13)? Additionally, if one variable would be I(0), would this mean that the respective exogenous variable would have a negative lag length of (-11)? Thanks you very much. I look forward to your response

[@ChekwubeMadichie]

What determines what lag comes into the exogenous section are (i) the optimal lag and (ii) the highest order of integration among the variables. For instance, if the optimal lag length is 5, and the highest order of integration is 2 (ie I(2)), then you would put (-7) into the exogenous section. But if the highest order of integration is 1 (ie I(1)), then you would put (-6) in the exogenous section.

[@smsm314]

Hello my Professor, Thank you for this research. Sorry Sir, To make prediction or modeling; we have to take the Var(p) model found in the procedure of TY (model in the levels of the data and no extra lags), Or it is necessary to restimate another model, and in this case the, which model? Cordially.

[@nurehh2271]

Thank you!

[@smsm314]

Good morning Professor, Thank you for this research. Sorry Sir, Your variables GDP and TOP are I(2) or I(1)?

[@OmarAhmed-cm3mv]

Both are l(1) as shown by ADF test

[@JMRG2992]

@@OmarAhmed-cm3mv So why he conducted with 13 lags ? since it was 11 ideal lags, + maximum order of integration which is 1, therefore the t&y test should have been conducted with 12 lags !!!!!

[@macjelly1259]

Why didn't you do Granger causality with stationary data?

[@ChekwubeMadichie]

Thanks George but the main aim of this video is to show the steps on how to perform the Toda-Yamamoto causality test.

[@spinebuster9490]

Granger causality is for I(1) variables only. But for Toda-Yamamoto it doesnt matter if the variables are I(1), I(0) or I(2) as the lags will be distributed accordingly.