Threshold Structural VAR

[jwojnilower]

No worries.

A couple other quick questions though...

1) Does the moving average include the current period or only previous periods?

2) How should we cite your program?

[dakila]

1) The moving average of threshold variable includes current period, but the delay parameter controls its lag.
2) You must cite the Balke seminal paper (2000).
3) It's up to you to cite the thsvar add-in.

[jiangziyaok]

dear sir
i donot know the means about the results signs of responses function such as the c1 c2 c3 c4,which one is the +2SD,+1SD,-2SD,-1SD? thank you.

[dakila]

That is Eviews bug. You should update your Eviews.

[jiangziyaok]

however,my eviews had updated to eviews9.i want to know how to update the newer version?The symbols c1,c2,c3,c4 wether have a fixed means to the "+1/+2SD"? Thank you very much.

[dakila]

Yes, they represents +-1|+-2 SD.

[jwojnilower]

I have a question regarding how the threshold variable is calculated...

As an example, I have monthly data on credit spreads from 1919m1 through 1941m12. For the threshold variable I'd like to use the moving average of the credit spread with a length of 6. If I set the sample from 1919m6 through 1941m12, will the initial credit spread moving average reflect the 6 months from 1919m1 to 1919m6 or only 1919m6?

Thanks in advance for the clarification.

[dakila]

the 6 months from 1919m1 to 1919m6

[nasa]

Dear
I have a couple of questions regarding the threshold SVAR add in.
1) How can I add exogenous variables: constant and time trends to the ThSVAR. You can address this question from the command line as I cannot see it in the dialogue box option for this.
2) How can I get the generalized impulse response functions in TABLE form rather than in GRAPH form as this will help me to calculate multipliers.How is it possible also to get GIRF's confidence bands ? It can be also addressed from the command line and ,if possible,from the dialogue box.
3) thsvar(girf=1,sample="1961q1 2014q2") 2 dlnrgdp dlnrgdp @ dlnrspending dlnrtax dlnrgdp
The above command line draws GIRFs only for one regime,what about for the other regime? If we assume two regimes only.It is Ok in the dialogue box but in the command line not clear.
4) How do we determine the length of MA average of the threshold variable?

Thanks for your quick response.
B.Regards

[dakila]

Did you read the instruction which is located in the thsvar add-in folder?
1. You can’t add exogenous variable such as time trend. Constant is included in the model. You do not need to include it manually.
2. You can’t get GIRF in table form. It is not allowed. I will try to include this option in the next version. It is not possible to get the confidence band.
3. Use option regime. For example: tvsvar(girf=1, regime=0) …. or regime =1 for the upper regime.
4. Use option length. For example: tvsvar(length =3) 4 cpbill1 d1gdp @ d1gdp d1pgdp fyff cpbill1

[nasa]

Hello,

When it will be released the next version that includes the option in TABle form?

Thanks.

[dakila]

I am not sure when it will released. By the way you can manually calculate the multiplier. If you roll the mouse on the line of IRF, it will display the numbers you need.

[nasa]

HELLO,

One last question is the endogenous variables that I am using are non-stationary in levels and hence I do first difference to get stationary series.However,when I try Johannesn co-integration test in levels,I found that they are co-integrated. My question is the following: is it appropriate to run Threshold SVAR for these variables as long as I am interested in studying the short run effect? or do I need to look for Threshold SVEC model (if you have hint about this model, please suggest me where to get and how it works in eviews).

Thanks.

[alvarezcc]

Hi! I have a question regarding the selection of the moving average order and the delay parameter. What is the best way to select this options? For example, if using a moving average order of 2 I obtain a lower value for the logdet(sigma) when using ma=3, does this means that I should select the ma order of 2? This same logic applies for the delay parameter?

Thanks in advance

[dakila]

HELLO,

One last question is the endogenous variables that I am using are non-stationary in levels and hence I do first difference to get stationary series.However,when I try Johannesn co-integration test in levels,I found that they are co-integrated. My question is the following: is it appropriate to run Threshold SVAR for these variables as long as I am interested in studying the short run effect? or do I need to look for Threshold SVEC model (if you have hint about this model, please suggest me where to get and how it works in eviews).

Thanks.

Since you are interested in the short-run effect I think it is ok to run the thsvar add-in. If you are interested in the long-run effect try the tarcoint add-in. I have no idea about Threshold SVEC model.