Linear panel data: imposing MA(6) correlation errors

[Paula Pereda]

Hello Statalist! I'm doing an exercise from Microeconometrics Using Stata, by Cameron and Trivedi, Exercise 3 of Chapter 8 (page 278).

"For the model and data of section 8.4, compare PFGLS estimators under the following assumptions about the error process: independent, exchangeable, AR(2) and MA(6)."

 use "http://fmwww.bc.edu/ec-p/data/mus/mus08psidextract.dta" 
 xtreg lwage exp exp2 wks ed, pa corr(ar 2) vce(robust) nolog

I can do the AR(2) part but I don't find how to impose a MA(6) process to the error.

Any suggestion?

[Phil Bromiley]

Look at the time series documentation - it provides more flexible error structures.