10-16-2015 01:55 PM - edited 10-16-2015 02:04 PM
All SAS modeling procedures do this. You need to understand that with classification effects, there is no unique set of coefficients and intercept; there are an infinite number of solutions that give the exact same model fit and model predictions.
So it is relatively meaningless to ask what the actual value of the coefficient for a particular level is.
The more relevant questions are: