Kevin Sheppard
Kevin Sheppard
It isn't because I need sparse matrices as much as possible for absorbed effects. It should be possible to make a hybrid interface with a formula for the non-absorbed though.
It is also possible that Formulaic which is not the formula engine supports sparse arrays, which I haven't checked.
> See [my comment](https://github.com/scipy/scipy/issues/16803#issuecomment-1207650855) on the corresponding `scipy` issue. I think it's a bug with the `conda` packaging, not upstream. If someone wanted to check, they could (using a name...
@vikjam This seems about right. statsmodels also supports OLS with 2-way clustering.
It is, but needs fixes like this to be made into PRs. Would like to get the readers back in shape then have a release.
Thanks, and sorry for taking so long.
You might ask him if he would put them in github somewhere.
Rolling residual variance estimators should be easy to compute along side other rolling quantities. Since these would be a 1d array and well-defined, this doesn't create the issue that rolling...
Do you variables that do not change in each cross section, for example, time dummy variables?
Is risk premium a linear function of the other variables? Or is risk premium only a function of the year?