November employment situation

Ok, very quickly, then I'm headed on the road. I just ran two vector autoregressions: one with initial unemployment claims and payroll employment and another with continuing claims and the unemployment rate. I used the VARs to forecast November's unemployment and payroll employment numbers. Then I took data for unemployment claims in the second week of November (big drop in both numbers - pretty remarkable, and good news for the labor market), used the historical correlation between surprise changes in those numbers and surprise changes in unemployment and payroll employment, and I get the following predictions:

November unemployment rate: 10.0%.
November change in payroll employment: -10,795.

I don't know whether to believe those numbers or not. The standard error of those point estimates is very large, and the models may be misspecified to begin with. But let's let those predictions stand and see what is reported a week from Friday. If the numbers look that good, there are going to be a lot of people scrambling to figure out how they could have been so wrong about predictions for a jobless recovery!