Abstract
Are the large measured wage differentials for on-the-job computer use a true return to computer skills, or do they just reflect that higher wage work