Lags and Leads in BRMS formulas

Interfaces brms
1

How does one specify a lag/lead in a formula with time series cross section data?

I effectively want something like
bf(Y ~ lag(X, time=‘t’, group=‘country’, period=1))

But I have been unable to find anything like that in the documentation. There’s an allusion to being able to use indexes in the lengthy (formula help document).

For fixed covariates, I could precompute lags and include them as columns in the dataframe. In my specific case though, I’m imputing missing values and I want to be able to refer to that latent variable column in terms of lags in other formulas.

E.g.
bf(Y~ lag(mi(X), time=‘t’, group=‘country’, period=1) + lag(mi(X), time=‘t’, group=‘country’, period=2))
bf(Z~ lag(mi(X), time=‘t’, group=‘country’, period=3) )
bf(X | mi() ~ some_covariates )

The only relevant post on the forum I can find is “Prediction for a model using a lag as a predictor”.

2

There is no way right now to do this in brms unfortunately. In one of our own projects, we manually edited the stan code of the mi() model to include lags.