May 20, 2021
Still plodding on with this. I did a fairly big refactor of how incomes where handled, and that’s simplified things bit. Need to do another one to cut down some of the huge enumerations used for employment and education status and the like, and simplify many of the crazily long names.
I’ve got as far as non-means-tested benefits now. This is more of a struggle than I’d hoped. It throws up all sorts of wierdness in the FRS. A sample:
- receipts of some benefits are imputed, but the results of the imputations seem wrong to me sometimes.
Here’s some workings where it seems people are getting duplicate amounts of some benefits. Also, PIP seems to be imputed on a calendar year basis, rather than financial years.
This is annoying because I’m relying on these amounts as a simple way of modelling some benefits without going in to the mechanics of the assesment procedure. Ideally I’d like to be able to model the actual assessment for disability benefits but information is just not there: there’s a set of ADL - like measures in the FRS, but not the ones used for PIP. I’ve got some rough regressions of pip takeup from a while back that I could maybe adapt later on.
I can’t remember if I knew all this back when I had a proper job.