EViews.com

Hello,
Please see my small forecasting loop. What I do is to estimate an ARIMA equation in dlogs and forecast in dlogs but I want to estimate the ARIMA in dlogs but forcast in logs. Is there a way of doing it?
thanks

for !j=1 to 36
smpl 1990Q1 1999Q4+!j-1

'estimate the ARIMA model
equation eq22_!j.ls dlcopp c ar(1) ar(2) ma(1) ma(2) dum2 dum3 dum4

smpl 1999Q4+!j 1999Q4+!j+7
'forecast 8 quarters ahead
forecast copp_22_!j

'calculating mean squared errors
series rmse22_!jcopp = (copp_22_!j - dlcopp)^2

next

Rather than using dlcopp as your dependent variable, use dlog(copp) as your dependent variable (assuming you have a variable called copp).

yes, copp is the variable. I changed it the way you suggested, it worked. thanks

Hello,

I have a problem: I have done something similar because I also want to estimate the model in dlogs but to forecast in logs and what I obtain by using as dependent variable dlog(y) is the forecasting in levels of the variable instead of logs.

I would be very grateful if you could help me.
Thanks,
Raquel

Assuming your dependent variable is Y, then create a new series equal to logy = log(y), then use d(logy) as the dependent variable.

Thank you,
Raquel