Page 1 of 2

Diebold-Yilmaz index

Posted: Wed Apr 25, 2018 4:32 am
by dakila
This thread is about the dyindex add-in that estimates the Diebold-Yilmaz index of spillover using forecast error variance decomposition method of a VAR model.

Re: Diebold-Yilmaz index

Posted: Mon Apr 30, 2018 1:29 am
by paolo.zanghieri
Great. Very useful add in. Is there a way to recover the time series of the pairwise interconnection measures?
Thanks

Re: Diebold-Yilmaz index

Posted: Tue May 01, 2018 7:03 pm
by dakila
The add-in is updated to include the option to save the pairwise spillovers.

Re: Diebold-Yilmaz index

Posted: Tue May 08, 2018 1:58 am
by paolo.zanghieri
thanks a lot. Just a final clarification. the series pc_i_J must be interpreded as share of unit i's variability due to j? Moreover each column in the fromvar (tovar) matric is the chare of variance due to (contributing to) other units, right? thanks again

Re: Diebold-Yilmaz index

Posted: Tue May 08, 2018 11:00 am
by dakila
yes

Re: Diebold-Yilmaz index

Posted: Sat Oct 06, 2018 4:53 am
by dlodlod
Thank you for the add-in. I need help please. When I try to run the add-in I get an error message which says "Sizes do not match in matrix function". I cannot figure out how to fix this. I do see that the matrices have different orders but my challenge is how do I get around this.
Background to my research: I want to calculate a spillover index for monetary policy variables which are measured on a quarterly basis. I have attached the file so you can see the VAR I have run (named var01). When I try to run the add-in from the VAR, that's when I get the error message. Any assistance/advice will be greatly appreciated.

Re: Diebold-Yilmaz index

Posted: Mon Oct 08, 2018 2:31 pm
by dakila
Did you read the instruction document ? (dyindex.pdf file located in add-in folder)
You cant run the add-in from the VAR. You should open it first from Add-ins menu. Or try command line.

Re: Diebold-Yilmaz index

Posted: Thu Oct 18, 2018 6:57 am
by bibi.dostmohamed2
Hello ,am having "illegal lag selection" when using the dyindex.Can you help me solving this please ?

Re: Diebold-Yilmaz index

Posted: Fri Oct 19, 2018 10:31 pm
by dakila
Could describe the data? Or upload data file?

Re: Diebold-Yilmaz index

Posted: Sat Oct 20, 2018 7:05 am
by bibi.dostmohamed2
Thank you for responding .I need to calculate the return spillover with those returns.Tests about the lag etc were done and was 1

Re: Diebold-Yilmaz index

Posted: Wed Oct 24, 2018 6:45 pm
by dakila
The sample size of time series is too short. So you cannot choose lags greater than 1.

Re: Diebold-Yilmaz index

Posted: Wed Oct 24, 2018 9:03 pm
by bibi.dostmohamed2
Thank you for your reply.But when am inputting the number of lags 1, i am having illegal lag specification.

Re: Diebold-Yilmaz index

Posted: Thu Oct 25, 2018 7:21 am
by dakila
Change the rolling windows parameter. for example, the following command worked for me

Code: Select all

dyindex(window=60) 2 @ usd_aud usd_cny usd_jpy usd_rub

window=60 - rolling window parameter

Re: Diebold-Yilmaz index

Posted: Thu Oct 25, 2018 8:33 am
by bibi.dostmohamed2
Thank you loads.It worked.

Re: Diebold-Yilmaz index

Posted: Tue Nov 13, 2018 5:03 pm
by abdhut123
hi, I am getting an error "sizes do not match in matrix function" when I'm doing generalized rather than Cholesky. When I'm doing Cholesky I'm able to get the result'.....but I want to do generalized...please help me