3SLS with low durban watson- how can that be?
Posted: Wed May 02, 2012 9:16 am
by svogel
Our model uses a three-stage least squares but our durbin watson statistic still indicates a fair deal of autocorrelation is present Why is this the case? Are our estimators still BLUE?
Re: 3SLS with low durban watson- how can that be?
Posted: Wed May 02, 2012 9:42 am
by startz
The Durbin-Watson tests for serial correlation. This has nothing in particular to do with 3SLS. The estimators are not BLUE. (3SLS is technically never BLUE, but can be approximately so in a large sample.)