Published in: Economica 74 (2). (2007). 259-297. (with Randolph Sloof and Joep Sonnemans)
Gibbons (1998) identifies a tradeoff between up-or-stay and up-or-out promotion rules. Up-or-stay never wastes skills of those not promoted but may provide insufficient incentives torninvest in skills. Up-or-out on the other hand can always induce investment in skill acquisition but may waste the skills of those not promoted. This paper reports an experiment designed to study this tradeoff. Under the up-or-out rule parties behave (almost) just as theory predicts them to do. But under up-or-stay (and stay-or-stay) rules results differ markedly from theoretical predictions. Workers invest rather frequently although the subgame perfect prediction is that they should not do so. Deviations from theoretical predictions can be explained by reference to different reciprocity mechanisms.