Error message: Forecast initialization required prior to estimation sample
Posted: Mon Jun 12, 2017 6:23 pm
Hello,
i have tried different approaches to my problem of recursively reestimated state space models.
When i try the following:
1. estimation of state space model for the sample
smpl 7/1/2009 9/22/2014+!h-1
2. then forecasting for the sample
smpl 7/1/2009+!forecasthorizon 9/22/2014+!forecasthorizon+!h-1
' sets the forecast horizon
{%modelname}{!forecasthorizon}.forecast(m=n,n=!forecasthorizon) @state * @signal {%modelname}{!forecasthorizon}_lcdsprognose_d @signalse {%modelname}{!forecasthorizon}_lcdsprognosese_d
' generates the forecasts
for the forecast horizon n=1 the code works, but for the forecast horizon n=2 the code breaks down with the above error message: "Forecast initialization required prior to estimation sample, change forecast period or select eviews or user-specified initial values".
What does this error measure mean in my context. I dont understand, because i has already shifted the forecast sample with "!forecasthorizon" to adjust for the initialization of the kalman filter.
Many thanks and kind regards,
Yohan
i have tried different approaches to my problem of recursively reestimated state space models.
When i try the following:
1. estimation of state space model for the sample
smpl 7/1/2009 9/22/2014+!h-1
2. then forecasting for the sample
smpl 7/1/2009+!forecasthorizon 9/22/2014+!forecasthorizon+!h-1
' sets the forecast horizon
{%modelname}{!forecasthorizon}.forecast(m=n,n=!forecasthorizon) @state * @signal {%modelname}{!forecasthorizon}_lcdsprognose_d @signalse {%modelname}{!forecasthorizon}_lcdsprognosese_d
' generates the forecasts
for the forecast horizon n=1 the code works, but for the forecast horizon n=2 the code breaks down with the above error message: "Forecast initialization required prior to estimation sample, change forecast period or select eviews or user-specified initial values".
What does this error measure mean in my context. I dont understand, because i has already shifted the forecast sample with "!forecasthorizon" to adjust for the initialization of the kalman filter.
Many thanks and kind regards,
Yohan