IRFs - combined effect of two shocks

[kanimbla]

Dear all,

I am fairly new to Dynare so maybe/hopefully the following problems are easy to solve.

1.) I have a New Keynesian model with a couple of AR(1)-shocks. My goal ist to generate IRFs that show the combined effect of different shocks (let's say two shocks) that hit the economy simultaneously. How can I implement this?

2.) I want the economy to be hit by a series of shocks, say, a demand shock that hits the economy for 4 quarters in a row. How can I implement this?

3.) If I delay one of the shocks, it implies that the shock is perfectly anticipated, right?

Thanks a lot for support!

[kyri82]

Hi,

1) I am not sure what you mean here. IRF by definition, is the response of the economy to a single shock. Do you mean to have two IRFs on the same graph?

2) That I do not know, sorry.

3) Yes, if you delay the shock it is anticipated. Suppose the shock is e, if you put it in Dynare as e(-1) it is perfectly anticipated.

Kyriacos

[kanimbla]

kyri82, Thank you for your comments and the clarification regarding my question 1.

What I would like to plot is the response of, for instance, output when two shocks hit the economy simultaneously. Do you know whether it is possible to do this?

I think that just adding up the two sperate IRFs will give a wrong picture if the shocks are correlated...

[kyri82]

Again, IRF is the response of the economy to one shock and one shock only! That is what the IRF is, if you try to somehow plot a combined effect it will be misleading and counter intuitive. Basically, you cannot do it! And you should definitely not add the responses.

[jpfeifer]

As long as you know what you are doing, having two shocks simultaneously is no problem (and does in particular not contradict the logic of an IRF as it can be interpreted as a single new shock that hits two different variables at the same time) and can be done in Dynare. Moreover, it is a recurring theme. For a potential solution, see http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=2515. In case you want to use two different but correlated shocks, Dynare performs a Cholesky decomposition, see http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=2574

[kanimbla]

Thanks a lot for the additional hint!!