EViews.com

Dear tcfoon

Thank you for good explanation VAR Model. I am facing the same type of problem. I have 3 variables and 4 lags are appropriate. Output of VAR model, 4 coefficients for each variable. So, how do I explain it. You mentioned that the contemporaneous effect (t or t-1) only, alternatively we may also use the summation of the coefficients. I do not understand clearly. Does it mean that I need to explain 4 coefficients separately or added four coefficients and explain? Please explain it.

Thanking you
Ali

Hi, I am an undergrad economics student and i am currently working on my thesis.:) I need to conduct this cointegration test but i find it hard to interpret the results since this particular topic was not covered in our undergrad curriculum. I am having a problem regarding the coefficients in my initial regression. At first, the variables of interest have coefficients that are not significant but theoretically correct signs. then in order to test for cointegration, as this is the subject of my study, i first-differenced the said variables because they are non-stationary at level form. The said variables are found to be cointegrated but I am concerned if the initial regression results which show that the variables have insiginificant coefficients affect the validity of the cointegration test. Can you please help me regarding this matter? Big thanks to all. :)

Please help!!!!!!!!!!


1. If Johansen cointegration test confirms that an explanatory variable has a significant long term impact on the dependent variable, should the variable also have a short term significant impact when estimating the vector error correction model?


Stop right there.

Cointegration has nothing to do with saying whether an explanatory variable has an impact on a dependent variable. Cointegration tests whether there is a common nonstationary variable driving two or more observed series.

The cointegrating equations do, as you suggest, give the long-run relation between the variables. But there is no "dependent" and "independent" variable and no causality.

i want to ask 1 question. in my research i am dealing with returns and for returns i took stock prices and apply return formula ln (pt/pt-1) to calculate returns. i took these returns to eviews. now plz guide me either i am going to treat data at level or it is 1st difference? because if i apply ADF on returns, the series is stationary at level.