Structural Vector Autoregressive (SVAR) Modelling in Eviews

This video provides some useful steps on how to estimate a Structural Autoregressive (SVAR) Model in Eviews.
For your requisite knowledge of Traditional VAR and SVAR models, kindly refer to the following materials:
Sims (1980); Sims (1986); Blanchard and Quah (1989); McCoy (1997); Narayan (2013).
#svar #var #identifyingrestrictions #structuralvectorautoregression #structuralvar #vectorautoregression #econometrics #eviews #chekwubemadichie #vidatasolutions #identification #restrictions #justidentified #overidentified #reducedformvar #AMatrix #BDiagonal

[@suleimansalisu1255]

Thank you sir for this clarification.

[@ChekwubeMadichie]

Welcome

[@safeedafaisal3447]

thank you so much sir

[@mengeshamanedo6317]

thank you keep it up

[@factorier]

thnx for video

[@zakeniachoirunnisah8149]

Thanks sir for this video is very useful. Sorry beforehand sir, can you also make a video for the steps to regress the svar method using panel data?

[@ChekwubeMadichie]

I'm already working on a video for Panel svar.

[@zakeniachoirunnisah8149]

@@ChekwubeMadichie thank you very much sir for the response, this really helps me who is processing the data for the thesis

[@Econotitans]

great work sir

[@Econotitans]

hello plz make a video on extracted garch series to be further analysed with svar model

[@ChekwubeMadichie]

Noted

[@suleimansalisu1255]

Thank you sir for this video. My question : why did you log the interest rate variable when it is already in in percentage? .

[@ChekwubeMadichie]

Dear Salisu, the only set of variables that you don't need to log are those whose values range from 0 to 1 to avoid the problem of unnecessarily large numbers. Whether a variable is captured in percentage or not, log transformation can still be applied. You only have to be mindful of your interpretation. For instance, if interest rate was 5% last year and 6% this year, it means that there has been a 1% point increase in interest rate and at the same time, it means that there has been 20% rise in interest rate. Thus, if you do not log it, the former interpretation (1% point increase) is used, and the later goes with logged interpretation. I hope you understand better?

[@fadijasmin1]

Thanks for this explaination , but I havea question about the interpretation of structural model coefficients and precisely in minute 11 ...you said that the first coeeficient represents that effect of CPI in GDP and the second is the effect of exchange rate on Gdp , But i guess if that is the inverse meaning that is the first and the second coefficients is the effect of gdp on CPI and EX respectively , as we estimate a recursive model according to cheolesky decomposition ?

[@tuhinahmed4952]

Yes, I have a same question here.

[@kelvsOse]

Thank you Sir for the video. I’m having problem with estimating structural factorization. When i click on it, I don’t get a dialogue box similar to yours. No restriction preset to choose from or a Matrix A. When I do proceed anyway I get an error message; “Number of SR restrictions must be equal number of endogenous variables in SVAR”. What can I do to fix this please?

[@ChekwubeMadichie]

I guess you're using eviews version earlier than version 10. However, you can manually define your matrices and restrictions in structural factorisation dialogue box based on your model structure.

[@kelvsOse]

@@ChekwubeMadichie I thank you for your response Sir. I’m new to running regressions and I’m currently using Eview9. I really do not know how I can manually impose restrictions. I’m working on a monthly panel data of 10years with 4 variables.

[@ChekwubeMadichie]

@@kelvsOseSend a dm to my WhatsApp [phone]. Let me look at your SVAR model.

[@suleimansalisu1255]

Great video. It appears to me that the coefficients of the estimated SVAR model are short run. How do I obtain the long run coefficients in the SVAR model?

[@ChekwubeMadichie]

The shortrun coefficients in my video follow from the difference of variables in the example I gave. I had to difference all the variables because they were individually non-stationary and no cointegration was found. Hence, there was no need to bother about the longrun coefficients since they were not cointegrated. However, if, in your own case, you found evidence of cointegration among non-stationary variables, you may have to perform the SVAR analysis on the level of variables. Then, in the definition of the SVAR structure, you may set up the longrun.

[@suleimansalisu1255]

Thanks a lot for your guidance on this note. Now, I need further clarification. Suppose I set up and estimated my SVAR at level (on the understanding that there is cointegration between the variables of interest), does it mean that the estimated coefficients are long run? if so, can I still interpret these coefficients in the same manner as in standard or traditional VAR ( by interchanging the sign against each long run coefficients). Furthermore, how do I estimate the short run coefficients/VECM for the SVAR.?

[@Lala_atksr]

Sir , I can't open output var estimate,

[@suleimansalisu1255]

Well done sir. Suppose that after performing the stationary tests, I discovered that my variables of interest are a combination of 1(0) and I(1), now given this circumstance, how do I estimate the structural VAR model. Secondly, is there the need to perform the conventional diagnostic tests on the estimated model; and if yes , how do I perform it, please? Thanks a lot.

[@vickym7133]

After watching this video. I have the same question as Salisu. May you please clarify this sir. Thank you very much

[@GarimaGupta]

Well, I'm facing the same issue in my model. I have some variables as I(0) and some as I(1). Should I still go for SVAR?

[@anjalyb8420]

@@GarimaGupta Yes, you can use SVAR

[@Econotitans]

i have 8 variables and im facing problem in estimating svar. is it ok to estimate svar with 8 variables?? please answer as soon as possible

[@ChekwubeMadichie]

What is the sample size? You should have a very large sample size.

[@Econotitans]

Its 233 observations, monthly data of 19 years

[@eristyaahdya4535]

Sir, i want to ask some question, can svar be used for data forecasting?

[@ChekwubeMadichie]

Yes

[@eristyaahdya4535]

@@ChekwubeMadichie Is the same way to use it as in the video?

[@SamuelSNtege]

Hello Sir, could you have this in STATA? Kindly share

[@ChekwubeMadichie]

Yes

[@ayanda.sikhosana]

@@ChekwubeMadichie Hi. Thanks for the video. I'm also trying to estimate SVAR in STATA. Please share link.