Different computers - different results!

[jason_ll]

Hello,
When I run an equation on one computer at work, I get certain results, but then running the exact same equation from the exact same workfile on a different computer is giving me significantly different results.

I've found this baffling. I will omit some work-sensitive information, but I'll post everything else:

First computer results:
Method: Least Squares
Date: 10/13/11 Time: 16:59
Sample (adjusted): 1987Q2 2011Q1
Included observations: 96 after adjustments
Convergence achieved after 19 iterations
MA Backcast: 1986Q2 1987Q1
R-squared 0.874374 Mean dependent var 0.019300
Adjusted R-squared 0.867395 S.D. dependent var 0.061226
S.E. of regression 0.022295 Akaike info criterion -4.708419
Sum squared resid 0.044737 Schwarz criterion -4.548147
Log likelihood 232.0041 Hannan-Quinn criter. -4.643634
F-statistic 125.2828 Durbin-Watson stat 1.473017
Prob(F-statistic) 0.000000

Inverted AR Roots .91
Inverted MA Roots .99 -.00+.99i -.00-.99i -.99





Second computer results:



Method: Least Squares
Date: 10/14/11 Time: 01:31
Sample (adjusted): 1987Q2 2011Q1
Included observations: 96 after adjustments
Failure to improve SSR after 3 iterations
MA Backcast: 1986Q2 1987Q1

R-squared 0.760000 Mean dependent var 0.019300
Adjusted R-squared 0.746667 S.D. dependent var 0.061226
S.E. of regression 0.030816 Akaike info criterion -4.061087
Sum squared resid 0.085468 Schwarz criterion -3.900816
Log likelihood 200.9322 Hannan-Quinn criter. -3.996303
F-statistic 57.00003 Durbin-Watson stat 1.784136
Prob(F-statistic) 0.000000

Inverted AR Roots .88
Inverted MA Roots .82 .00-.82i .00+.82i -.82


I did notice that in the second estimation, it says "Failure to improve SSR after 3 iterations", while in the first it said "Convergence achieved after 19 iterations". However, I tried to fool around with the estimation settings and I couldn't fix the error.

Please help. Thank you!

[EViews Gareth]

Either a different version of EViews (check the build date under Help->About EViews), or different starting values.