Page 1 of 1

forecasting

PostPosted: Thu May 25, 2017 5:39 pm
by algebraic123
Hi,

I have a small question regarding forecasting. I have 176 observations, and I am trying to obtain the 1-step-ahead forecasts for the last 76 observations. I am using nobs=[101:176], forecast=1 in the estimation block. This yields a vector of length 76 for each observable in oo_.RecursiveForecast.mean. Are these the correct 1-step out-of-sample forecasts? Or do I need to use nobs=[100:175] instead?

Thanks in advance.

Re: forecasting

PostPosted: Thu May 25, 2017 5:49 pm
by jpfeifer
I am not entirely sure I correctly understand your question. But for the last observation, i.e. 176, Dynare will provide out-of-sample forecasts starting with the value at time 177. Thus, if you want a forecast for time 176, made at time 175, you need nobs to end at 175.