Hi

I have got an issue, I estimated a model, I used Prais-Winsten autoregressive model (lag one and constant), I forecasted the value for the next two years, see table below:

For 95% confidence intervals, t(4, 0.025) = 2.776

     Obs Global_Sales    prediction    std. error        95% interval

    2004        125.00       125.47

    2005        139.00       139.01

    2006        151.00       150.70

    2007                         166.82              0.693       164.89 -   168.74

    2008                         181.81              0.709       179.84 -   183.77

My question is; how the std. error is calculated, I could calculate the std. error for 2007 (summarize of the squares of residuals, divided by n-2), but I do not know how it is calculated for the next year (2008 – 0.709)

Does anybody know the formula or how it is calculated? I am interested in any paper or book where I can find it.

Thank you very much

Kind regards

Francisco Sosa

dd: +44 (0) 20 7448 3875

email: francisco@digitab.uk.com

digitab

tel: +44 (0) 20 7588 6669

fax: +44 (0) 20 7256 7540

digitab  4th floor 6-8 bonhill street  london   ec2a 4bx

please consider the environment before printing this document