Regressing into future periods

[Raul Athwall]

Is there a way to regress my dependent variable up to 5 periods in the future? I am regressing FTSE 100 on my x variable, and I want to see how this relationship changes when I am accounting for returns on day t+1, t+2 up tp t+5, keeping my x variable the same.

[Daniel Feenberg]

I am not sure what you are asking for - but does https://www.stata.com/manuals/rpredict.pdf help at all?

[Raul Athwall]

Thanks, Daniel, but I can't seem to find what I am looking for in that manual. I essentially want to regress y on x, then y(t+1) on x, y(t+2) on x and so on up to y(t+5), if that makes sense?

[Daniel Feenberg]

A single observation does not make for a regression. Do you mean to rerun the regression with one more observation, then two more, etc? Maybe you are looking for https://www.stata.com/manuals/tsrolling.pdf (check out the recursive option) but I wonder why you don't start with just an -if- clause on the -reg- command for each additional regression. Maybe I still don't understand the question.

reg y x if t<100
reg y x if t<101
reg y x if t<102
...

[Raul Athwall]

This is from a paper in which they do what I am attempting to replicate. Hopefully, this helps so if it were a regression it would be y(t)= x +u(t), y(t+1) = x + u(t+1) etc

[Joro Kolev]

See the lead (F.x) and lag (L.x) operators in Stata.

[Raul Athwall]

If I want to have 5 forward lead variables for my dependent variable in my regression, does the code look something like this?

. regress FTSE100 F.FTSE100#1-F.FTSE100#5 UKIS

[Bruce Weaver]

For info on the F.x and L.x operators mentioned in #6:

help tsvarlist