[Trac] #199: the output of option forecast in estimation is incorrect in the Reference Manual
#199: the output of option forecast in estimation is incorrect in the Reference Manual ---------------------------+------------------------------------------------ Reporter: michel | Owner: Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Documentation | Version: Keywords: | ---------------------------+------------------------------------------------ If the forecast takes place after computing the posterior mode, the output is indeed saved in oo_.forecast (as correctly described in the manual) But, if the forecast takes place after Metropolis iteration and is really the posterior distribution of forecasts, the output is saved in oo_.MeanForecast and oo_.PointForecast
#199: the output of option forecast in estimation is incorrect in the Reference Manual ---------------------------+------------------------------------------------ Reporter: michel | Owner: Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Documentation | Version: Keywords: | ---------------------------+------------------------------------------------
Comment(by sebastien):
Michel: Stéphane and I tried to understand what oo_.MeanForecast and oo_.PointForecast contain, and we could not understand. It seems that both contain uncertainty wrt parameters, but only PointForecast contains uncertainty wrt to shocks. But at 1st order uncertainty about shocks averages out. So we must have misunderstood this.
#199: the output of option forecast in estimation is incorrect in the Reference Manual ---------------------------+------------------------------------------------ Reporter: michel | Owner: Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Documentation | Version: Keywords: | ---------------------------+------------------------------------------------
Comment(by sebastien):
See also this related question on the forum:
http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=3542
#199: the output of option forecast in estimation is incorrect in the Reference Manual ---------------------------+------------------------------------------------ Reporter: michel | Owner: sebastien Type: bug | Status: assigned Priority: major | Milestone: 4.3 Component: Documentation | Version: Keywords: | ---------------------------+------------------------------------------------ Changes (by sebastien):
* owner: => sebastien * status: new => assigned
#199: the output of option forecast in estimation is incorrect in the Reference Manual ---------------------------+------------------------------------------------ Reporter: michel | Owner: michel Type: bug | Status: accepted Priority: major | Milestone: 4.3 Component: Documentation | Version: Keywords: | ---------------------------+------------------------------------------------ Changes (by michel):
* owner: sebastien => michel * status: assigned => accepted
Comment:
should be solved at the same time as ticket #102
#199: the output of option forecast in estimation is incorrect in the Reference Manual ---------------------------+------------------------------------------------ Reporter: michel | Owner: michel Type: bug | Status: accepted Priority: major | Milestone: 4.3 Component: Documentation | Version: Keywords: | ---------------------------+------------------------------------------------
Comment(by michel):
check also if it is necessary to keep forcst_unc.m
#199: the output of option forecast in estimation is incorrect in the Reference Manual ---------------------------+------------------------------------------------ Reporter: michel | Owner: michel Type: bug | Status: accepted Priority: major | Milestone: 4.3 Component: Documentation | Version: Keywords: | ---------------------------+------------------------------------------------
Comment(by michel):
The difference between oo_.MeanForecast and oo_.PointForecast concerns the confidence interval not the central forecast that should be identical in theory (but the sampling is different, I believe). oo_.MeanForecast provides a confidence interval taking only into account estimated parameter uncertainty oo_.PointForecast provides a confidence interval taking into account both estimated parameter uncertainty and future shocks
#199: the output of option forecast in estimation is incorrect in the Reference Manual ----------------------------+----------------------------------------------- Reporter: michel | Owner: michel Type: bug | Status: closed Priority: major | Milestone: 4.3 Component: Documentation | Version: Resolution: fixed | Keywords: ----------------------------+----------------------------------------------- Changes (by sebastien):
* status: accepted => closed * resolution: => fixed
Comment:
Fixed in 35a2f38d6a8f551cd142551600c58a1a1bde7078
-
Trac Dynare