I am using the WinRATS procedure @paneldols.src but I don't understand very well the mechanism of this. Can you tell me briefly how this procedure work and what is the main difference between this procedure and an usual panel by OLS ? Moreover, how can I choose the option "Average" in this procedure: average=[simple]/sqrt/precision ?
Best regards
procedure @panelDOLS.src
Re: procedure @panelDOLS.src
It's for panel Dynamic OLS - which estimates a cointegrating vector using Stock and Watson's dynamic OLS technique of adding lags and leads of the differenced RHS endogenous variables to correct for bias. If you're not trying to estimate a cointegrating vector, then you're using the wrong procedure.
It does a separate regression on each cross sectional unit and uses some type of average of those estimates as the overall estimate (so it's assuming a common cointegrating vector). The unweighted average is done using AVERAGE=SIMPLE. The other two are for weighted averages which give higher weight to the estimates from cross sections with more precise estimates.
It does a separate regression on each cross sectional unit and uses some type of average of those estimates as the overall estimate (so it's assuming a common cointegrating vector). The unweighted average is done using AVERAGE=SIMPLE. The other two are for weighted averages which give higher weight to the estimates from cross sections with more precise estimates.