Panel Data SVAR

[Keith Vijaya]

Hi all,

I have some experience with running a SVAR with time series data in STATA. Now, I would like to do the same with panel data.

However, I am not entirely sure whether this is even possible. Does anyone have any recommendations for running an PS-var in Stata?

Many thanks in advance !


With kind regards,

Keith Vijaya

[Justin Niakamal]

I think the Stata Journal article, "Estimation of panel vector autoregression in Stata" would be a good start.

https://journals.sagepub.com/doi/pdf...867X1601600314

[Keith Vijaya]

HI Justin,

Thank you for your reply! My apologies for this late reaction, I did not expect to have such a quick reply back!

I have read the paper by Love and I have re-created her process. However, it is still not clear to me how I can impose structural restrictions to the residuals?

I have heard that there might be a possibility to do this, first I have to run a PVAR (using the command of Love), then get the variance & covariance matrixes.
Do a Cholesky decomposition to get a lower triangular matrix and then graph the IRF`s.

Is there any chance you've heard anything about this?

Stay safe and kind regards,

Kiethan

[Justin Niakamal]

Sorry, Keith Vijaya I am not familiar enough with pvar , but it looks like you need to use the porder option for getting the lower triangular matrix - just be cognizant of the ordering used because it can impact your results. If you're interested, watch the first half of this video for an explanation/example.

[Keith Vijaya]

Hi justin,

Thanks for taking the effort to reply!

I will definitely have a look at the video.

To comment on my original post, I have found out that is indeed not possible to impose structural restrictions on the residuals using PVAR.
One has to find a work-around by doing this manually using the variance-covariance matrix post estimating the PVAR.


Kind regards,

Kiethan