This paper develops a multi-period model, in which workers are matched with jobs according to imperfect educational signals and in which their subsequent productivities depend on both their inherent ability and on the quality of the job match. It outlines a sequential process, in which underpaid employees reveal their true productivities and overpaid employees are detected by the firm until every match is perfect. The model produces a time path of the returns to educational signals that is concave, a feature that earlier studies used to dismiss educational signaling. Using a synthetic panel data set from the Current Population Survey the theoretical result is then substantiated empirically. The paper contributes to the literature by establishing the possibility of increasing returns to education over part of a workers life within the signaling framework theoretically and empirically.