Posted by: admin 1 year, 12 months ago

(Comments)

Finally, I found a time to create a video in how to, JMulti video. 

So, JMulti is a holy grail in SVAR research, most of you who doesn't know must already use other software such as Eviews, Matlab, STATA, and many other. However, this JMulti is inspired a lot by the work of Lutkepohl (2006), in the new multivariate autoregression. 

So the critical step to reproduce such a step is. 

1. Make sure you have prepared data for this.

2. You also already fully understand the theoretical perspective of what you have to put your variable in the order.

3. Go and select the variable itself.

4. Choose VAR analysis 

5. Specify the lags, first can be 4, second can 10-12.

6. Click model check -> residual analysis -> choose portman -> 20, LM 1 (if the value of P bigger then 0.05 there is no autocorrelation in residual.

7. Click SVAR -> choose IRA -> specify model B, change based cholesky model, the lower part of the matric

8. Choose still SVAR - IRA - choose 20 quarter 

9. Leave for the first option 0,95, bootstrap 1000 repetition, chose any number you like, for example 1,2,3,4 for bootstrap confidence interval

10. Click on display, on the picture, click view option -> convert background white - 

Step by step in getting FEV

Forecast horizon

chose the symbol + or the target 

click on result VAR -> IRA -> SVARA

- the first 4 column is the first variable, response to the 4 shock 

so output repose to outputshock, credit shock, real estate shock, and LTV shock. 

Also happen to the bootstrap with the upper bound and lower bound

later copty to excel or create the graphic using matlab. 

Alright that s all! 

• Tags:
• phdjourney

• ← What is Cholesky decomposition
• I am so happy that finally the work I have with my professor is published →