I like the way you explain your tutorial videos, they are really beneficial, thanks. My data are stationary at fist difference, I want to do impulse response in VAR, should I use first differenced data or level data?
@juliebasconcillo
Жыл бұрын
Hi! Thank you for the detailed video on diagnostics. It's quite clear. I have a question on residual analysis though. When you tested for serial correlation at lag=3, there is still an autocorrelation at lag=1. Based on previous readings, the lag should be further increased, i.e. more than 3 because there should be no autocorrelation at all lag levels/ periods. Could you please clarify? Thank you.
@subhajitroy4302
7 ай бұрын
Hello Sir, thank you so much for your efforts to make such videos. These are really helpful. I have a question. What is normality test results yield that residuals are not normally distributed and also if the residuals are heteroscedastic, what should be the next step?
@DhavalSaifaleeAaryash
7 ай бұрын
Did u check stationarity of variable first
@subhajitroy4302
7 ай бұрын
@@DhavalSaifaleeAaryash yes, the variables are stationary at first level. I tried both ways, using level data as well as 1st difference. For both am getting the same results
@sagnikmaity6734
Жыл бұрын
Hello Sir, I am getting an error 'Log of non positive number'. How to deal with it?
@SukhwinderSingh-mm5tq
2 жыл бұрын
While performing VAR test, we have included 2 lags in "lag length criteria". My question is - if this 2 is constant in all the cases or we can also increase it to 3 or 4? Because, when I entered 3 lags, its showed better results (lower value). Please guide on this. Thanks
@DhavalSaifaleeAaryash
2 жыл бұрын
If you consider r2 here, increasing lags will definetly increase but it will give rise to multicollinearity. Reasonably dont go above 2 lags as number of terms in final model will go on increasing
@rohtashbhall2671
2 жыл бұрын
Two variables should be stationary according to granger but which order. At level or first difference or same order
@gamana19
2 күн бұрын
At any order, it may be at level or after the first difference, but both the variables should be stationary in the same order
@imaduddinfarih4835
Жыл бұрын
hello Dr. Dhaval.. thanks for very clear explanation. it's very helpfull.. i have question for the wald test to determine significant model. based your explanation the null hypothesis is all the coefficients are equal to zero. so, is that mean the intercept coefficients are included ? or only the variable coefficient ? thank you
@DhavalSaifaleeAaryash
Жыл бұрын
Only coefficient of variable are included
@imaduddinfarih4835
Жыл бұрын
@@DhavalSaifaleeAaryash thanks
@jejedc4965
Жыл бұрын
sir, my variables are stationary after first difference. should i use my data in levels in forecasting or should i difference my variables first before building the var model?
@DhavalSaifaleeAaryash
Жыл бұрын
We hv to difference it
@yasminehandel9642
Жыл бұрын
hello can you answer my question please can i use var test when series are stationary at level
@DhavalSaifaleeAaryash
Жыл бұрын
No
@santatrarakotoarimino5316
8 ай бұрын
Hi sir, if my variable are stationnary at first difference should i estimate the first difference of the variable with var or just the level variable ?
@DhavalSaifaleeAaryash
8 ай бұрын
run normal ols with first difference variable
@santatrarakotoarimino5316
8 ай бұрын
@@DhavalSaifaleeAaryash thanks sir
@imaduddinfarih4835
Жыл бұрын
hello Dr. Dhaval.. I've been carried out heteroscedasticity test for VAR model, but i got the error message "Positive or non-negative argument to function expected". could you help to explain what does it means ? i use data with 30 obeservations and 2 variable. Thank you
Пікірлер: 24