Page **6** of **7**

### Re: Threshold Structural VAR

Posted: **Sun Jul 16, 2017 6:15 am**

by **nasa**

Hi,

I am using eviews 10 demo license and I have installed thsvar add in but once I specify the thsvar model using the box options and press ok using the box options it is telling me that Eviews 10 has stopped working.

What is the problem,please?

best

### Re: Threshold Structural VAR

Posted: **Sun Jul 16, 2017 2:27 pm**

by **dakila**

I think problem is that you are using the demo version.

### Re: Threshold Structural VAR

Posted: **Mon Jul 17, 2017 2:20 am**

by **nasa**

Ok, but there was a time using the demo version of Eviews 9,I was using threshold svar add in without any problem.

Thanks.

Best

### Re: Threshold Structural VAR

Posted: **Sun Aug 06, 2017 7:24 pm**

by **yizhuorui**

Hello, i want to know the code of THSVAR,when i select File/open/Programs in add-in folder... and open the THSVAR folder and click thsvar.prg,why the dialog display "Encrypted program". How should i do if i want to understand the thsvar.prg(in order to modify THSVAR code to adapt my model )? Thanks.

### Re: Threshold Structural VAR

Posted: **Mon Aug 07, 2017 8:59 am**

by **EViews Matt**

Hello,

Addin authors have the option to encrypt and conceal their programs if they wish. This ability isn't unique to addins, any EViews program can be encrypted. If you're interested in how a program works and/or obtaining the unencrypting source for the program, you can contact the addin author through the appropriate forum or via any contact information provided with the addin's documentation. Just keep in mind that the author chose to encrypt the program for a reason, so there's no guarantee he or she will share the source with you.

### Re: Threshold Structural VAR

Posted: **Sat Aug 12, 2017 7:24 am**

by **nasa**

Hi,

In the balke paper,it says the threshold value is the one that maximizes the log determinant of the structural residuals.But here ,in the add in ,I read the result as specification that minimizes the log determinant.So,is there any difference between this two points? could you give me a general intuition about this,please?

And the second I think we are also using the Tsay test that is avg- wald and exp-wald besides to the hansen test that is sup-wald or LR-sup.Hence,can you give me also some ideas about the Hansen and Tsay tests,please? What exactly are we doing with regard to these two tests?

Best

### Re: Threshold Structural VAR

Posted: **Sun Aug 13, 2017 4:18 pm**

by **EdgarM**

Hello, thanks for the useful add-in.

Is there a way to calculate a 1 percent shock instead of a 1 standard deviation shock?

The problem is that my threshold variable is inflation growth rate, and it's standard deviation is 0.59, so doing a 59% shock is not an useful exercise

Is the response measured in standard deviations too?

Thanks in advance,

Best regards

### Re: Threshold Structural VAR

Posted: **Mon Aug 14, 2017 5:44 am**

by **dakila**

In the balke paper,it says the threshold value is the one that maximizes the log determinant of the structural residuals.But here ,in the add in ,I read the result as specification that minimizes the log determinant.So,is there any difference between this two points? could you give me a general intuition about this,please?

No. there is no difference between these two points.

And the second I think we are also using the Tsay test that is avg- wald and exp-wald besides to the hansen test that is sup-wald or LR-sup.Hence,can you give me also some ideas about the Hansen and Tsay tests,please? What exactly are we doing with regard to these two tests?

I don't know Tsay test.

### Re: Threshold Structural VAR

Posted: **Mon Aug 14, 2017 5:51 am**

by **dakila**

Is there a way to calculate a 1 percent shock instead of a 1 standard deviation shock?

I think there is no way to do so. Because Balke's Threshold VAR model is nonlinear. So it is not scalable.

The problem is that my threshold variable is inflation growth rate, and it's standard deviation is 0.59, so doing a 59% shock is not an useful exercise

Is the response measured in standard deviations too?

If your variable is in log then response is in percentage change. I think 0.59 is 0.59 % shock.

### Re: Threshold Structural VAR

Posted: **Wed Aug 16, 2017 12:52 am**

by **nasa**

Hi,

when i want to run the Add in in eviews 10 demo, it tells me that syntax error.What is that,please?

Best

### Re: Threshold Structural VAR

Posted: **Wed Aug 16, 2017 2:29 pm**

by **EdgarM**

Thank you Dakila for your answer.

Is there a way to obtain the VAR estimation for each regime instead of different outputs for each dependent variable?

It will be useful to get the variance/covariance matrix for each regime, also the R-squared of the VAR estimation.

Best regards,

Edgar

### Re: Threshold Structural VAR

Posted: **Thu Aug 17, 2017 4:34 pm**

by **dakila**

when i want to run the Add in in eviews 10 demo, it tells me that syntax error.What is that,please?

It is a bug for Eviews 10. It is fixed. Just update the thsvar add-in.

### Re: Threshold Structural VAR

Posted: **Thu Aug 17, 2017 4:40 pm**

by **dakila**

Is there a way to obtain the VAR estimation for each regime instead of different outputs for each dependent variable?

It is indeed the VAR estimation. If read any textbook on VAR, it will tell that VAR estimation equal to OLS estimation of each dependent variable equation.

### Re: Threshold Structural VAR

Posted: **Sat Aug 19, 2017 7:02 am**

by **nasa**

Hi,

What is the degree of freedom adjustment? The default value in the dialog box is 0.Can I simply estimate my threshold SVAR model as in the default value with out modification to the degree of freedom,please? What would be the implication of this for my thsvar model?

Best ,

### Re: Threshold Structural VAR

Posted: **Sat Aug 19, 2017 2:58 pm**

by **dakila**

The degrees of freedom adjustment is used to compute the threshold value which is excluded at the beginning and the end of sample.

The formulae: b = ob*(end-start+1)+nvar-dfa

- ob = fraction of buffer period

- start = start of sample

- end = end of sample

- nvar = number of variable

- dfa = degrees of freedom