High dimensional general unrestricted models (GUMs) may include important individual determinants, many small relevant effects, and irrelevant variables. Automatic model selection procedures can handle more candidate variables than observations, allowing substantial dimension reduction from GUMs...