C4: Nov. 2000 exam, #12

[New at pd]

Hi everyone,
I tried this problem on my own w/o looking at the answer sheet. I got the correct answer, but not the way the answer key does it.
The Cochrane-Orcutt procedure says to get the residuals, then perform a regression on the residuals to get the estimated value of rho, which in Actex is defined as the lag one correlation coefficient.
The answer key does some sort of sum that doesn't use the values that I did.
Any enlightenment would be greatly appreciated.

Thanks.

[Higher Authority]

The actex solution is exactly the same as the solution from the Cochrane-Orcutt proceedure. (it is refined, though) The actex solution is just a result of doing a least squares fit for the model Y(i+1)=p*Y(i) + error. (for i = 1 to 4) and Y(i) is the difference between the fited data and the origional data.

[New at pd]

Thank you so much, Needo. That totally clears it up for me.

[Guerilla poster]

Test