ARDL Bounds Test - 3of6 (Lag Selection & Unit Root Test)

[@justineborja556]

Is it okay to understate the lag length in ARDL if the understated lag creates the better diagnostic results? Thank you

[@alauddinfahad3084]

Hi Pat Obi: Thanks for your nice explanation. You have determined optimal number of lag through var specification (which is 1 in this example). But this optimal lag is not used anywhere in augmented dicky fuller test. Then why did we need to determine lag if it is not used in any other steps?

[@nguyentu3776]

dear Sir, when should we use Break point Unit root test? how we could determine which is appropriate approach?

[@PatObi]

Hi Nguyen, I'm sorry I'm not quite sure. However such an approach is generally appropriate when your data stretches over a period that you suspect includes significant shifts in the economy such as the 2008 financial crisis or 1997 Asian Contagion..

[@nguyentu3776]

@@PatObi thank you so much

[@rafaeldemoraislima9080]

Thank you for the video Sir. I got one question: My dependent variable are I(0) and my regressors are a mix of I(1) and I(0), can I use ARDL in this case? I haven't found any solid answer for this question. Cheers from Brazil!

[@PatObi]

Thanks, Rafael. Great question! Here are my thoughts. If you want your Y-variable to be I(0), then you should run a long-run OLS model against X-variables that are also I(0). There is the temptation to, instead, use the differenced series of the I(1) independent variable since in its differenced form, it is I(0). But I think that's complicated because differenced series are for short-run dynamics. I encourage you to reach out to others who may know better than I :-)

[@olubukolaadegbe9876]

thanks for the video sir. How can i get the powerpoint please? thanks

[@cgdino7096]

Hi! Do I always need to use ADF for unit root? My data seems to have a structural break and whenever I use ADF, one of my variables turns out to be integrated of order higher than 1. Thank you!

[@TheEnyoy]

Hey sir! I have a question dr. Adeleye selected the lags by specifying the dependent variable in the endogenous box and the other variables in the exogenous variables box in eviews. I get different selected lags. I am using annual data. Your method gave me 4 lags, specifying the variables in the exogenous box of eviews gave 1 lag.

[@desientertainment8239]

VERY INFORMATIVE VIDEO BUT I AM UNABLE TO WRITE THE EQUATIONS IN WORD. CAN YOU SEND THESE EQUATIONS?

[@PatObi]

Go to Insert in Word and then look for EQUATION.

[@patot7]

Hi Sir! You have been quite clear, so I want to really thank you! I am currently working on my final work to get my degree as an economist. I wanted to know how can I get an optimal lag selection for each variable in my ARDL model. I have seen some models which have different lag orders in the independent and dependent variables. Thank you a lot again! Greetings from Argentina.

[@PatObi]

Gracias Patricio for connecting with me. You can specify the variables with their respective lags manually on the ARDL equation. I think EViews will nevertheless give you the optimal specification.

[@rinasafitri1541]

hello sir,, i'm rina... i'm sorry if i disturb you.. but can you tell me about markov switching vector error correction heteroskesdasficity?

[@PatObi]

Hi Rina. No, you're not disturbing me :-). Sorry, I'm unfamiliar with that model at this time.

[@rinasafitri1541]

Pat Obi :-) thank you prof for the answer, but can you explain in detail what co-integration actually coherently and clearly, or maybe there is a book that is recommended?

[@rinasafitri1541]

Pat Obi I don't really understand what cointegration is, why is it called a long-term relationship, and how is the interpretation of the results, please help me professor, thank youu

[@PatObi]

@@rinasafitri1541 If two time series variables, X and Y, are cointegrated, it simply means they have a long run relationship. They move together in the long run either positively or negatively. That's all.

[@PatObi]

@@rinasafitri1541 Interpret the regression coefficients like any other least squares regression result. When you regress Y on X, the slope coefficient (beta) is your measure of the degree and direction of impact of X on Y. If you need a refresher on regression, you can watch my video entitled Regression in Brief.