Can I average out a constant (intercept) in OLS regression?

Asked Dec 12 '18 at 10:54

Active Dec 12 '18 at 10:54

Viewed 61 times

I have a OLS regression in the form:

$$Y_t=\alpha +\beta X_{t-1}+\varepsilon_{t}$$

Can I average out the constant during the OLS estimation/derivation and report,

$$y_t=\beta X_{t-1}+\varepsilon_{t}$$

Could I explain this as "I have averaged out the constant during the OLS procedure?"

So the intercept is there, the estimation procedure I use (code I have) averages it out, is there some correct technical jargon for this?

asked Dec 12 '18 at 10:54

user22485

1

Did you mean centre the data before running the regression? – Huy Pham Dec 12 '18 at 11:59
1

Yes, I did!!!!!!!!!!!!!!! – user22485 Dec 12 '18 at 12:02
1

See the third answer in https://stats.stackexchange.com/questions/201919/utility-of-the-frisch-waugh-theorem – Christoph Hanck Dec 12 '18 at 12:30
1

@ZarinaAkhtar so i mean, unless I'm missing something, that's what it's called. You mean centred the data so the intercept=0. Your residuals should still be ok. – Huy Pham Dec 12 '18 at 13:30

0 Answers0