Forecasting VECM with restrictions
Posted: Wed Apr 10, 2013 1:22 pm
Hi,
I am currently estimating a VECM with cointegrating restrictions for four different type of exchange rate models and for this i do out of sample point forecasting.The full sample is data from 1981M1-2002M6.The estimation period I focus on is 1981m1-2002m6, and then I conduct an out-of-sample forecasting exercise from 2002m6 for all horizons from h=1,3,6,12,24,36,48 until i exhaust my sample.
On eviews after estimating the VECM I make a model and then solve it for the period with which i wish to forecast e.g. for h=1 i solve the model from the period 2002m6 2000m7 and then use the same method for all other horizons and then forecast the VECM using the VARforecast add in pane. My question is if this forecasting exercise is was conducted properly?
Many thanks.
I am currently estimating a VECM with cointegrating restrictions for four different type of exchange rate models and for this i do out of sample point forecasting.The full sample is data from 1981M1-2002M6.The estimation period I focus on is 1981m1-2002m6, and then I conduct an out-of-sample forecasting exercise from 2002m6 for all horizons from h=1,3,6,12,24,36,48 until i exhaust my sample.
On eviews after estimating the VECM I make a model and then solve it for the period with which i wish to forecast e.g. for h=1 i solve the model from the period 2002m6 2000m7 and then use the same method for all other horizons and then forecast the VECM using the VARforecast add in pane. My question is if this forecasting exercise is was conducted properly?
Many thanks.