Identifying VARs with sign restrictions

[TomDoan]

Dear Tom,

I use code for four shocks from the forum.I am not quite sure whether the code is correct. Could you please give me some suggestions and feedbacks whether my code is correct?

Thank you very much for your help and kindness.

pei

First off, do not start with a NOPRINT option on your ESTIMATE. Look at the output and see if there are problems. Once you're sure that the VAR itself is OK, you can put NOPRINT back on. In your case, it's not OK. ST is the difference between FIRON and PIRON, so you can't include all three in a VAR. You also aren't including a CONSTANT in the system, so the residuals appear to be much more correlated that they are. Offhand, it also appears that you're including in levels data series which would ordinarily be modeled in logs.

Your HSTART is also wrong; it needs to be the start of the estimation period (1986:7 with six lags), rather than the start of the data.

Finally, don't start with four shocks. Start with one. See if you can get that to work successfully. As it is, you're not even getting successful draws for the first shock.

[pei]

Dear Tom,
Thank you very much for your suggestions.
I am a beginner, so there will be many problems cannot understand. Thank you for your patience to answer my questions.
I modified the sample，it works partially. But I don’t know what is worry with it, and how can I correct the mistakes.
Please give me some suggestions.


Sincerely,
pei

[pei]

Dear Tom,
The above is my code, I have modified the code for three shocks and four variables. Nevertheless, results do not appear.
I might do something wrong. Could you please give me some suggestions?


pei

[FRegensburg]

Dear Mr Doan,

I need to calculate a statistic, presented by Mountford and Uhlig in their 2009 article, that allows one to arrive at the impact multipliers of the three policy experiments (as in pages 983-985 in the original paper)

the authors calculate, at different horizons, the impact multipliers, according to this formula:

Multiplier for GDP=(GDP response/initial -lag zero- fiscal shock)/(average fiscal variable share of GDP).

The last ratio is from the raw data, so that's not a problem. But what lines of code do I need add to the replication file code to extract the hard numbers to compute the first ratio?

I also need to calculate the maximum and minimum multipliers of the policy experiments, with confidence intervals (16th and 84th quantiles), with the same formula.

I look forward to hear from you, as I don't know how to do this.

[TomDoan]

Those are already being generated at the end of the mu2009x1.rpf and mu2009b1.rpf programs. They are being graphed by @MCGRAPHIRF. If you want the values to put into a table, use @MCPROCESSIRF instead (or in addition).