Hi Vincent,
They are treated by literally removing those TRs from the
regression analysis.
It is possible to imagine cases where it might be better to
censor them all, but it seems to generally make sense not to
throw away samples of some betas because you lost a few of
some others.
If the sample number gets too small for some betas, then they
become unreliable, which should be reflected in the partial
t-stats.
- rick