Recursive VAR forecasting in STATA

Vote count:

0

I am doing VAR modeling in STATA. I have 192 months of data, two variables with one lag. That means:

y1=c1+b1*y1(t-1)+b2*y2(t-1) y2=c1+b3*y2(t-1)+b4*y1(t-1)

With the "rolling" function and the option recursive I can do recursive estimation meaning I get a vector of coefficients b1 to b4 over time, as more and more data flows into the VAR. Now I know how to use "fcast compute" after the last observation. What my question is, is how to do forecasts based on all coefficients individually. That is at time 24 (I select window(24) in "rolling") I get my first estimate of b1-b4. With that I can make a forecast about t=25. At t=25 I get my second estimate of b1-b4 and then I want to make forecast about t=26. And so forth until forecasting t=193. How can I combine "rolling" "var" and "fcast compute"?

In advance, many thanks for your support!

asked 1 min ago

user2964146

15