Open sbenthall opened 4 years ago
There is a port of the KS REMARK now in master. I think the next step on this is to include a parallel dolo version of KS.
@albop is there somebody on your team who is set up to contribute a dolo version of KrusellSmith? Alternatively, is dolo in shape such that I could try to build the dolo-KS implementation?
Based on the current email thread with @mnwhite , it sounds like there are several versions of KS in HARK, but none of them do it quite right and Matt is working on a revised version now.
If the HARK version of a model is unstable, I'm not sure how we can meaningfully write a matching Dolang version of it, even being very imaginative about the expanded possibilities of a Dolang syntax.
There's only one item labeled KrusellSmith in HARK right now. The problem is that it solves a problem that is not quite identical to the actual model in the original paper by KS, because of a limitation in HARK. The model that is solved is actually like a first cousin rather than an identical twin.
You're right that it would be best to hold off on the KS chimera until Matt and Mridul finish the project of getting the identical twin working. This will require a fairly small change to the markov aggregate shock model, and a bit of careful constructing of matrices, but shouldn't be too hard.
As this model requires a market aggregation step, I assume that this should be implemented with dolark
.
I believe the KS model in HARK has now been fixed, as per @llorracc 's earlier comment?
See: https://github.com/econ-ark/HARK/issues/414
Using KrusellSmith REMARK: https://github.com/econ-ark/REMARK/issues/17
Make a DARKolo chimera.