Estimating residuals in an ARIMA(0,1,1) model

[Abel]

In the following post:
http://forums.eviews.com/viewtopic.php?f=7&t=465

there is an explanation of how to estimate the residuals of the following process:

y c x ma(1)

It possible for me to reproduce the residuals of the example in an excel file.

When I want to estimate the following model:

d(y) c d(x) ma(1)

I can not reproduce the residuals of this model in my excel file (exact copy of excel file in that post). What should I change to get the correct residuals in a differenced model?

[EViews Gareth]

Exactly the same way, but use the difference of the variables, rather than the variables.

Here's an EViews program:

Code: Select all

rndseed 1 create u 100 series x=nrnd series y=nrnd equation e1.ls(ARMA=CLS, Z, OPTMETHOD=LEGACY) d(y) c d(x) ma(1) 'Static Forecast smpl 1 2 series e=d(y)-c(1)-c(2)*d(x) smpl 3 100 series e=d(y)-c(1)-c(2)*d(x)-c(3)*e(-1) e1.fit(d) yhats1 'EViews fitted values (static) series yhats2=c(1)+c(2)*d(x)+c(3)*e(-1) show yhats1 yhats2

[Abel]

Thanks for your quick response

I see the difference is in the estimating method, not in how to estimate the residuals.

I used GLS and I had backcasting on. What do these things do and why does it makes a difference?

[EViews Gareth]

Such econometric discussions are beyond my scope.

The EViews documentation provides some details, as well as references to more.