EViews: Engle and Granger Approach to Cointegration Test (Estimation and Interpretation)

[@sajdabano1813]

Well explain👌. Thnx

[@joyjahmed]

Sir, I have two dependent and two independent variables, please what Co-integration method will I use and what method of estimation is suitable ? NB: 11 observations All variables are stationary at I(1)

[@joyjahmed]

Please what method of estimation is suitable for this type of cointegration? Thank you for this clear explanation on E-G cointegration test

[@obezipacademy]

@@joyjahmed Single equation Models that accommodates I(1) variables such as FMOLS or DOLS

[@joyjahmed]

⁠@@obezipacademy Okay. Thank you so much for your prompt response sir. Also, after estimation using either FMOLS or DOLS the next step is post estimation tests right?

[@obezipacademy]

@joyjahmed3269 Yes

[@joyjahmed]

@ I am so Grateful sir. Thank you 🙏

[@blissezekiel6485]

Esther Ezekiel Thank u sir

[@femitex]

Are you guys still producing video tutorials?

[@obezipacademy]

Yes

[@jemimabulus2140]

Thank you sir

[@zainmalik4323]

Having 01 dependent and 05 independent variables can we use Engle-Granger test? Or this test is only when having 1 dependent 1 independent variable ??

[@obezipacademy]

One dependent variable with one or more independent variables. It must be a single equation model

[@zainmalik4323]

@@obezipacademy OLS estimation shows 3 out of 5 significant variables and residuals stationary at level, that means only these 03 statistically significant variables are having cointegration with the dependent variable, right??

[@obezipacademy]

@@zainmalik4323 All combined are cointegrated. You don't segregate

[@zainmalik4323]

@@obezipacademy keeping cointegration aside, How can I conclude which variables are significantly related to dependent variable? (All variables are stationary at 1st difference)

[@princessjudii]

The unit root test for variables. Do they need to be stationary at levels or first difference !!

[@obezipacademy]

For cointegration to exist, the error term must be stationary at levels.

[@princessjudii]

@@obezipacademy thank you ✨

[@philemononowoareke2960]

What other test do I conduct if it doesn't pass the cointegration test

[@obezipacademy]

You simply estimate a first differenced ARDL model

[@kealebogaditshabe3272]

I have a question, after test unit root for the et variable, if the variable turns out to be insignificant, does it mean i did something wrong? The variable is negative but not significant which makes me feel like everything is wrong.

[@obezipacademy]

Insignificant means that there is no long run equilibrium relationship between the dependent and independent variables. As such, just run first differenced ARDL MODEL

[@sunnyolamide1142]

After establishing the fact that there is long term relationship/contintegration; what next please?

[@obezipacademy]

Estimate error correction model

[@vishalvarsani669]

shouldnt you be using the diffrenced data when making the regression equation ???

[@obezipacademy]

U mean using differenced data to generate the error term meant for cointegration test?

[@philemononowoareke2960]

Understood Sir. Thanks a lot. However I am working on a project that has just one independent variable but two dependent variables. Do I run the cointegration test separately?

[@obezipacademy]

Yes

[@paulinusozemede5181]

Thank you sir. The lecture is well understood. However, my area of concern is when conducting test of Et on constant and trend, assume only trend is found to be statistically significant, what do we do when conducting the Unit root test? Are we picking on trend and intercept or none since there is no option for trend alone.

[@obezipacademy]

You pick "trend and intercept" since there is no option for only trend

[@paulinusozemede5181]

@@obezipacademy Thank you sir.

[@happydauda5699]

Thank you sir

[@Safehaven_1]

Thank you sir

[@angayejoseph6049]

Thank you sir