Page 1 of 1
equation needed for heteroscedasticity test
Posted: Fri Jan 06, 2012 7:46 am
by Mike
any tip is appreciated... I like to test heteroscedasticity using Eview. From another program, I have demand data, forecast data, and forecast errors (residuals). How to formulate my equation in Eviews?? forecast=c(1)+(c2)*demand does work but the residuals do not equal my forecast errors... however, they should as to perform the tests.. any idea for the equation???
Re: equation needed for heteroscedasticity test
Posted: Fri Jan 06, 2012 8:14 am
by startz
If the residuals don't match the forecast errors in EViews, you're missing something.
Having said that, you can run the heteroskedasticity test in EViews after an estimate, and EViews shows you the equation it estimates for the test.
Re: equation needed for heteroscedasticity test
Posted: Fri Jan 06, 2012 9:20 am
by Mike
I see, thanks, I am not an expert ...
So, then I would need to transform my forecast from SPSS to Eviews. In SPSS, let's say I found the following seasonal model appropriate ARIMA (1,2,3)(4,5,6) with 12 periods. How to write the equation in Eview? Something like:
ls(forecast) c ar(1) i(2) ma(3) sar(4) si(5) ma(6)? (where to write 'demand'?)
Re: equation needed for heteroscedasticity test
Posted: Fri Jan 06, 2012 9:56 am
by startz
EViews notation is different from the notation of SPSS. The first variable in the list is the dependent variable (demand?). Search the online help for how to specify AR and seasonal AR terms. If you really have a model that is integrated of order 2 (unusual but certainly possible), then specify the dependent variable as being second differenced, d(y,2) for example.