Direct Forecasting

[Harvard]

Hello colleagues,

i want to forecast a sample not by iteration, but direct. This means for example, that my AR model looks like this: y(5) c y(-1) y(-2), if the forecast horizon is 5. Because Eviews is not able to forecast it by the commands "fit" or "forecast", I have to program it. Has anybody an idea, how it could look like?

Thanks and greetings

[EViews Gareth]

I don't follow what you mean by this.

You want to estimate Y(5) without estimating Y(4), Y(3), Y(2), Y(1)? If so, then what value do you wish to use for Y(-1) Y(-2)? If you want to use the true values, then a "fit" will do it.

[startz]

Hello colleagues,

i want to forecast a sample not by iteration, but direct. This means for example, that my AR model looks like this: y(5) c y(-1) y(-2), if the forecast horizon is 5. Because Eviews is not able to forecast it by the commands "fit" or "forecast", I have to program it. Has anybody an idea, how it could look like?

Thanks and greetings

Forecast doesn't work with leads and lags. But you could do

Code: Select all

series x=y(5)
ls x c x(-6) x(-7)

[Harvard]

Hello,

the idea "series x=y(5), ls x c x(-6) x(-7)" sounds great. I will try it and post if it works.

Thanks very much

[Harvard]

Dear colleagues,

it works.

Thanks again for the great idea.
Greetings from Germany