rangestat error

[Olena Onishchenko]

Dear Stata users

My data is in panel format, daily frequency (id, date). I need to run a regression for each firm (id) each day (date) and obtain regression coefficient, constant and residuals.

I have tried the following code:

rangestat (reg) BuyInst sent_vol_nonzero_lag, interval (date 0 0) by(id)

Stata returned the following error: no result for all obs: reg BuyInstVolN sent_vol_nonzero_lag

What's wrong with the code? Any help is much appreciated.

Thank you.

[Nick Cox]

rangestat is from SSC, as you are asked to explain.

From what you say you are trying to fit each regression to one data point (each firm and each day). rangestat cannot help you any more than regress could. The difference is that no error messages are given. rangestat isn't surprised that any window contains insufficient data to do the required operations and therefore not surprised even if they all contain insufficient data.

You don't give a data example but yours is on all fours with examples like

. webuse grunfeld

. regress mvalue invest if company == 1 & year == 1940
insufficient observations
r(2001);

[Olena Onishchenko]

Nick

Thank you. I think what I need is a regression by id, month.

This one seem to be working:

rangestat (reg) BuyInst sentiment_volume_1d_n_lag1, interval(month . 0) by(id)

The output gives coefficients, standard errors but no residuals. Is there any chance rangestat can produce it?

[Nick Cox]

It's much more likely that you want

interval(month 0 0)

If you think about what rangestat gives you, it is immediate that

gen residual = BuyInst - b_cons - b_sentiment_volume_1d_n_lag1*sentiment_volume_1d_n_lag1

[Olena Onishchenko]

Nick

Thank you. I am not sure I can calculate residuals in a way you have suggested. When I run regression each month id (based on daily data) I get the following output:



When I apply the above equation to compute residuals I get residuals compared to daily BuyInst...