Equilibrium learning in simple contests

The paper studies a repeated contest when contestants are uncertain about their true relative abilities. When ability and effort are complements, a favorable belief about one's own ability stimulates effort and increases the likelihood of success. Success, in turn, reinforces favorable beliefs. We show that this implies that with positive probability players fail to learn their true relative abilities in equilibrium, and one player wins the contest with high probability forever. In this case, the prevailing player may be the actually worse player, and persistent inequality arises. We discuss some features of the model when the complementarity assumption is dropped.