NARDL Bounds Test - 1of5 (Introduction)

This video is the first of 5 videos in my NARDL cointegration series. It explains the motivation for this asymmetric modeling approach and describes the NARDL Bounds Testing framework.
Please note that the "intuitive" example shown in minute 7:11 is NOT an accurate depiction of the NARDL data structure. Apologies. Correct method: X+ is obtained by calculating the cumulative partial sum (i.e. running totals) of positive changes in X (DX+). Likewise, values for X- are the running totals of negative changes in X (DX-). Please refer to my published study at www.mdpi.com/1....

[@Azam_Pakistan]

The best explanation on the web. Your intuitive explanation with data was too good which most teachers avoid.

[@PatObi]

Thanks for your kind comment :-)

[@adityarazpokhrel7626]

No one explains NARDL better than you sir. Salutations from Nepal, South Asia.

[@PatObi]

Thanks Aditya!

[@fidelisobaniyi9103]

The best explanation so far.

[@yeweixia8210]

Thank you so much for the excellent explaination of NARDL. But I am just curious about the example you did in the video at 7:52, cuz isn't the X+ or X- should be the 'sum' of the changes in X, instead of the simple changes in X. I am not sure about the answer, and please forgive me if there is any offensive expression I might not notice. ;)

[@PatObi]

Thanks for your note. Please refer to my video description below the video for the correct calculation method.

[@Maria-tn4cn]

excellent explanation thanks a lot

[@PatObi]

You're welcome.

[@Maria-tn4cn]

@@PatObi please make vedio on adaptive lasso for variable selection

[@alauddinfahad3084]

AWESOME!

[@meenakshigautam9457]

is NPARDL is done in same way kindly reply

[@naomymuindi8432]

Wow,I love this. Could you please help run Panel NARDL model.

[@PatObi]

Please check out my new videos under the playlist "Panel VAR"

[@beb12745]

Hi. Can you explain how we can use unknown multiple threshold in nardl model in eviews? as in some case we cannot simply apply "0" as threshold value for decomposition. Thanks

[@PatObi]

I'm not sure. I'll post a response once I figure it out.

[@najeebkhan4246]

thanks for nice explanation... can i use the NARDL model for my model where my dependent variable includes mostly negative and few positive values while independent variables values are positive . After checking stationarity both dependent and independent variables are stationary at 1st difference.. either i should go for simple ARDL model... please answer

[@PatObi]

The sign of your values has nothing to do with the estimation model you choose. Since both X and Y are I(1), you can run ECM or ARDL. But if you wish to determine whether asymmetries exist in the relationship between X and Y, you should run NARDL.

[@KnowledgePromo]

Could you please share your slides?

[@ЛюдмилаКшнясева-я5ш]

a0=mt ?!

[@angeleren8393]

Sir, my result for wald test is symmetry. Is my NARDL still okay?

[@PatObi]

Yes, however no evidence of asymmetric relationship.

[@a.nelson7394]

A