1988/07/01 by Robert Drago, Geoffrey K. Turnbull
Social Sciences · Decision Sciences · Economics, Econometrics and Finance · #Experimental Behavioral Economics Studies #Decision-Making and Behavioral Economics #Game Theory and Voting Systems
paper · doi:10.2307/1058859
Much recent research concerning work incentives has analyzed tournaments, or contests between workers for rewards based on a comparison of their relative performance [3; 5; 8; 9; 12]. Bull, Schotter and Weigelt [2, 2] argue that tournaments are empirically pervasive, characterizing incentives for many salespeople competing for bonuses, managers vying for high-level positions, academics striving for limited tenured positions, and politicians whose election hinges on relative votes received. Contrarily, Nalebuff and Stiglitz [12, 40] suggest that tournaments are rare. For example, many employees face incentives produced by upward-sloping age-earnings profiles [6], individual piece rates, profit sharing, or group bonus schemes [14], or potential firing [11]. Responding to this lack of agreement over the pervasiveness of tournaments, recent theoretical research has begun to identify circumstances where tournaments fail or succeed. The corpus of such work generalizes the conditions under which tournaments are optimal. For example, tournaments can respond effectively to common environmental shocks to productivity [5; 12], heterogeneous workers [13], employee and employer moral hazard problems [3], monitoring difficulties [9], and inefficiencies associated with firing as an incentive [11]. Less work has addressed circumstances where tournaments may fail, the question which concerns us here. To date, three circumstances where tournaments fail have been identified. First, as Lazear and Rosen noted in their seminal article on tournaments, such incentives fail if individual effort can be measured perfectly or precisely inferred from output. As they put it, a difference of opinion is necessary for a horse race [8, 845]. If the winner is known before the race, only the winner will run, and slowly at that. Second, Lazear [7] analyzes circumstances where agents can engage in sabotage, or increase the probability of winning by reducing their opponent's measured output. Tournaments then fail because the saboteur can reduce his or her own output while