structural VAR and restrictions

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

kingomar
Posts: 5
Joined: Sat Aug 08, 2020 2:36 am

structural VAR and restrictions

Postby kingomar » Sat Aug 08, 2020 10:45 pm

i'm currently analyzing the economic impact of a group of country being pegged to another my major interest is impulse response function, my variables are LN_GDP_EURO , LN_GDP_AFRICA, CPI_EURO , CPI_AFRICA, ln_REER_AFRICA, and i have one dummy which captures when Europe adopted Euro as a currency. My study period is annual from 1968 to 2015. I have tried a long run cointegration among my I(1) variables (GDP_EURO , GDP_AFRICA, ln_REER_AFRICA but I did not have a correct ect( positive and non-significant), i turned all my I(1) in first difference so i have D_LN_GDP_EURO , D_LN_GDP_AFRICA , D_ln_REER_AFRICA , CPI_EURO , CPI_AFRICA. I ran a structural VAR with both short-run and long-run restrictions but every time I'm applying my restrictions i have the same message "convergence achieved before restrictions on S/F were satisfied"
Im using reviews 11 lite.
you can find attached my data and the restrictions i have applied, I'm struggling any help will be welcomed
Attachments
restrictions.docx
(107.28 KiB) Downloaded 367 times
data.xlsx
the data i used my VAR order is D_GDP_EURO, D_GDP_AFRICA , EURO_CPI., AFRICA_CPI, D_LN_REER
(12.83 KiB) Downloaded 332 times

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: structural VAR and restrictions

Postby EViews Matt » Mon Aug 10, 2020 2:36 pm

Hello,

That error ideally only occurs when the restrictions you've chosen don't agree with the underlying data, i.e., EViews cannot find results that simultaneously satisfy your restrictions and match the VAR's residual covariance matrix. Given that you've over-constrained the SVAR, that result is not unexpected.

Also, the restrictions on S are redundant and can be removed. Since you've restricted A to be unit lower triangular and B to be diagonal, S will naturally be lower triangular, thus there's no need to specify any zeros in the upper triangle of S.

kingomar
Posts: 5
Joined: Sat Aug 08, 2020 2:36 am

Re: structural VAR and restrictions

Postby kingomar » Sat Aug 15, 2020 11:06 pm

Thanks a lot for your reply, I put my restrictions on A matrix and everything seems to go smoothly. I have another question, my IRF(impulse response function) is not significant, I read on another thread that if you are interested in IRF and not in coefficient, you should run your SVAR at level with your I(1) variables, i had to turn my I(1) to I(0) to run my VAR because they were no cointegration among my I(1) variables, do you think i can run my SVAR with my I(1) at level to have a better IRF results

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: structural VAR and restrictions

Postby EViews Matt » Wed Aug 19, 2020 9:21 am

Hello,

You can certainly try it. Just keep in mind that since the SVAR is derived from an underlying VAR, if the VAR model of your data at levels is poor then I doubt you can place much confidence in the SVAR IRFs.

kingomar
Posts: 5
Joined: Sat Aug 08, 2020 2:36 am

Re: structural VAR and restrictions

Postby kingomar » Sun Aug 23, 2020 6:30 am

thanks a lot

kingomar
Posts: 5
Joined: Sat Aug 08, 2020 2:36 am

Re: structural VAR and restrictions

Postby kingomar » Mon Aug 24, 2020 10:26 pm

I have another question concerning structural Var. After plotting my variables, i did again unit root test, this time I did a unit root structural break test and I realized that 2 of my variables have a structural break with different dates, from the literature I read that I have to include dummies, I need some clarification, does it mean I have to include the dummies in my structural var as endogenous variables with the other variables when I will be doing my structural VAR analysis?

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: structural VAR and restrictions

Postby EViews Matt » Tue Aug 25, 2020 9:27 am

Hello,

The VAR would not determine where the structural breaks occur (the values of the dummy series), so the dummies would be exogenous variables.

kingomar
Posts: 5
Joined: Sat Aug 08, 2020 2:36 am

Re: structural VAR and restrictions

Postby kingomar » Tue Aug 25, 2020 9:35 pm

Thanks again, im very happy for this quick reply, you helping me a lot :lol: I have another question, do I have to include the structural break dummies(0 before and 1 after) in my unit root test or I can directly go to run a VAR and include it as exogenous dummy. thanks again

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: structural VAR and restrictions

Postby EViews Matt » Wed Aug 26, 2020 4:14 pm

Performing the unit root test with the dummies would help confirm that you've accounted for observed non-stationarity, but you can still run the VAR to see what happens regardless.

Muhammad Javid
Posts: 3
Joined: Sat Feb 27, 2021 9:16 am

structural VAR and restrictions

Postby Muhammad Javid » Sat Feb 27, 2021 9:20 am

I'am trying to estimate the impact of fiscal policy shocks on economic growth. To this end, I estimated a Structural VAR (SVAR) model by using EViews 11 software. In Standard SVAR procedure, Impulse response of a variable is based on one standard deviation shock. My question is how one can generate impulse response to one percent shock and two percent shock and the difference of one percent and two percent shocks

EViews Matt
EViews Developer
Posts: 560
Joined: Thu Apr 25, 2013 7:48 pm

Re: structural VAR and restrictions

Postby EViews Matt » Mon Mar 01, 2021 2:40 pm

Hello,

A number of other threads have broached this subject, such as One SE Bands on IRF and Impulse response to Cholesky 1% shock. The short answer is that you will need to create custom shocks for the impulse response procedure to use, which the sirf addin can help with.


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 26 guests