Estimating a MA(1) ARIMA Model in Stata

[Adam Guerrero]

Dear Statalisters,

I am trying to reproduce an MA(1) model using a dataset out of "Forecasting, Time Series, and Regression" by Bowerman et al. (2005) (the dataset is attached).
The authors provide a value for the MA coefficient equal to -.3534. When I run the same model in Stata, I get a value for the MA coefficient equal to .3518. The code is as follows

import excel sales.xlsx, firstrow clear
tsset time
arima sales, arima(0 1 1)

It is my understanding that the authors used least squares regression to estimate their MA(1) model, whereas Stata uses Maximum Likelihood. According to their text, these estimators should provide nearly identical results.

My questions are: 1) Did I estimate this model correctly using the code provided? 2) How can I reproduce the authors' result, or at least something close to their result, using least squares in Stata?

Thanks in advance,
Adam A. Guerrero

[Steve Samuels]

This forum is for trying out the forum software (practicing MathJax or LaTex, for example) not for asking substantive questions about statistics or Stata code. Re-ask on the General Forum http://www.statalist.org/forums/foru...ussion/general

[Adam Guerrero]

Yes, I know. I am just practicing. I probably should have chosen a different title - e.g., "practice post, estimating a MA(1) model... " Thanks!

[Adam Guerrero]

Trying to upload a Stata dataset (testing).