This looks like a
feature but is a bug by my standards: missing values are
being forward-filled when lags are taken of variables.
(where "forward-filled" means that the missing value is skipped and
the valid entry before/earlier than that is used)
Hi,
I agree it would be a pretty hefty bug. However, I cannot confirm
this so far:
<hansl>
nulldata 5
setobs 1 1 --time-series
series y = 0
y[3] = NA
print y
series x = y(-1)
print x
</hansl>
This preserves the NA alright. So you would have to tell us exactly
how you got your results.