ESTIMATING RETURNS TO SCHOOLING - WHEN DOES THE CAREER BEGIN

Because the life cycle is not neatly divided into a period of full-tim e schooling followed by a period of full-time employment, it is unclea r where analysts should ''start the clock'' on the career for purposes of estimating the returns to schooling. This study uses data from the National Longitudinal Survey of Youth to determine how estimated retu rns to schooling are influenced by the choice of career starting date. Schooling and experience measures are defined for four alternative st arting dates and then used to estimate a standard wage model for sampl es of white and non-white men. Estimated returns to schooling increase dramatically as an increasingly later starting date is used because i ncreasingly large amounts of unmeasured, ''pre-career'' work experienc e bias the schooling effects. A specification that controls more accur ately for the accumulation of schooling and work experience is suggest ed as an alternative to the orthodox model. (C) 1998 Elsevier Science Ltd. All rights reserved.