EViews.com

Hello,
I am using Bayesian VAR to nowcast GDP and financial sector output, however I have some issues. I will try to describe it step-by-step. First, I estimate the model with data from 1996q1 to 2015q4, then I make model and obtain forecasts for 2016q1 using Dynamic-Deterministic simulation. Until this point, everything is fine. However, as I have 2016q1 data for some of the variables included in the model, I create alternative scenario and override the values for 2016q1 with the data I have. However, when I compare the forecasted values of GDP from this scenario with baseline forecasts there is no change. What am I doing wrong? I will greatly appreciate your comments!

Need the workfile probably

Here is the workfile. Thank you Gareth very much for looking into these matters!

I think none of the remaining endogenous variables depend upon the overriden/excluded variables in the same time period - they all depend on past values. So the forecasts will only change for a forecast of 2016q2 onwards.

Maybe you have some advice how I can make the remaining endogenous variables depend on the overriden variables in the same period?

Change the specification of the underlying relationships

Seems that I found the solution - variables for which data are available are specified as exogenous in VAR dialog!

You do not need to make some variables exogenous forcefully. It is not good idea. You can use the confcast add-in.

Dear Dakila, thank you very much for answer. However, where can I download the confcast add-in? It isn't listed there http://www.eviews.com/Addins/addins.shtml

Sorry. You need to wait. It is new add-in. I hope Eviews will post it soon on the web.

Hi Andrew,

How did you find it?