Goodness of Fit Measures for Forecasting

[mornington]

Hi all,

My question relates to measuring goodness of fit in the GMM model and for the resulting forecast. I'm dealing with data on the federal funds rate, and used GMM to estimate the Taylor Rule prediction for the federal funds rate.

1) Is it okay to use R^2 to compare different estimations, given that I'm using a GMM model? What would be the best goodness of fit measures?

2) I estimated the above data for 1983-2003 and then used the model to forecast predictions for 2003-2006. I now want to check how well this prediction matches with the actual rates in 2003-2006. What measure(s) should I use?

thanks
T