I am running a regression of the form:

Code:
reg Y ibn.dow#c.l(12).pc1  ibn.dow , base nocons
where pc1 is the dependant variable that I wish to lag. (and dow is a day of the week dummy variable). My suspicion is that there is not an exact lag of pc1 that gives the best fit. Including more than one lag causes problems with multicolinearity.

My question is: Is there a way of estimating as part of the regression and optimal weighting combination of the lags that gives the best fit?

Thanks
Rob