I'm currently working on a Panel BVAR using the toolbox and, with a number of units for the panel greater than 4, the gibb sampling is taking forever. I have 6 variables for each unit and a time frame of 20 years, so this shouldn't be a heavy model. The prior is the Random Hierarchical one, I don't think that the issue is related to any particular kind of identification (even if I'm using sign restrictions), because the issue is strictly related to Gibbs sampling (the bar basically never starts the computation of the sampling). I know for a fact that there are people which are able to run specifications with a double number of units that I am allowed to run, but I can't figure out what is the issue because we are working on the same release of the toolbox. Of course dropping variables it works for an higher number of units, but that's detrimental for my specification and shock identification. I suppose that the issue is related to the .m file "panel4gibbs.m", which generate the sampling for the specific type of panel BVAR that I am running. Any suggestions? thanks in advance
Hi,
I'm currently working on a Panel BVAR using the toolbox and, with a number of units for the panel greater than 4, the gibb sampling is taking forever. I have 6 variables for each unit and a time frame of 20 years, so this shouldn't be a heavy model. The prior is the Random Hierarchical one, I don't think that the issue is related to any particular kind of identification (even if I'm using sign restrictions), because the issue is strictly related to Gibbs sampling (the bar basically never starts the computation of the sampling). I know for a fact that there are people which are able to run specifications with a double number of units that I am allowed to run, but I can't figure out what is the issue because we are working on the same release of the toolbox. Of course dropping variables it works for an higher number of units, but that's detrimental for my specification and shock identification. I suppose that the issue is related to the .m file "panel4gibbs.m", which generate the sampling for the specific type of panel BVAR that I am running. Any suggestions? thanks in advance