An Improved Technique for Obtaining Current Sub-state Income Estimates
This paper compares some specifications for estimating current sub-state (county or MSA) income. A large number of trials are generated, using the data for Tennessee, in a procedure akin to a series of Monte-Carlo experiments. The out-of-sample forecast errors are examined to see which specifications perform best, and then to obtain some insight into the approximate error one would obtain using the best specification.