Promotion Tournament (Slide for UST Organization Study Group) Kong-Pin Chen 1 1 Introduction Promising young ‘star’ slump in mid-career. It is one who was obscure (or not highly regarded) in early career that eventually wins. Examples ─ Zhao Ziyang, Hu Yaobang (politics) ─ Mario Cuomo, Clinton (politics) ─ Levine vs Nicholas (business) ─ Negative campaign (politics) Why? More than a matter of luck 2 Fact: Promotion is all-or-nothing, based on relative, rather than absolute, performance. As a result, negative effort (sabotage) is a valuable instrument to compete. In order to compete for promotion, the contestants engage in both productive and unproductive (sabotage) activities. When the number of contestants is large enough, the ablest person might become the ‘focal point’ of attack so that his performance might fall behind less able people. Relation to tradition tournament model or all-pay-auction: In these models, the ablest contestant (or the one with highest valuation of the good) always has the greatest chance to win. This is no longer true when they can 3 engage in negative activities. Relation to the literature on ‘influence activity’ (Milgrom, 1998): The contestants not only engage in influence activity to increase promotion chance, but also actively destroy opponents; performance. This phenomenon is not only common in promotion tournament, but also between competing firms. Many firms file perverse antitrust litigation against IBM, Kodak, Microsoft (Brenner, 1987), just as a way of competition. 2 Model n contestants, one promotion. e i : level of productive activity by i . 4 a ij : level of attack by i against j . e i and a ij have different nature: The former only helps person i . The latter not only helps i but also k i, j . As a result, a ij has a flavor of being a ‘public good’. t i : ability of i in productive activity. si : ability of i in sabotage. ri t i / si . The comparative ability of i in productive activity. If ri r j , then we say i has greater comparative ability in productive activity than j. Performance (or output) of i t i ei g ( j i s j a ji ) i ; where j i s j a ji is the total attack person i receives. 5 g (0) 0, g ' 0, g " 0. i and j are IID. ~ f () ; singled-peaked, symmetric around 0: Luck is ‘fair’. Abilities observable among contestants, but not to superior. Wi t i ei g ( j i s j a ji ) , the expected performance of i . The utility of i , ui (e, a) Pr(i is promoted )u v(ei j i aij ) ; where u is the utility of promotion, and v the disutility of effort. The utility of not being promoted is v' 0, v" 0. 6 0. ui (e, a) Pr (Wi i W j j , j )u v(ei aij ) j i i W ji [ f ( i )( j i f ( j )d j )d i ]u v(ei aij ). where W ji W j Wi . Contestant i chooses the values of e i and aij ( j i ) to maximize u i (e, a) . We are interested in the Nash equilibrium of the game. Theorem 1 If ri r j , then k i s k a ki k j s k a kj . Contestant with higher comparative ability in productive activity receives more total attack. 7 Intuition: Other things equal, if i has high value of t i , then opponents need to attack him more at the margin. However, attacking him will force him to revenge, which is not worthwhile the higher is the value of s i . Special case: If si s j , i, j , then abler contestant is subject to more total attack. (l k sl alk ) / si 0 . A person with more talent in negative activity receives less total attack. (l k sl alk ) / si 0 (i k ) . If a person becomes abler in sabotage, then everyone else will be subject to more total attack. A person ablest in productive activity probably does not have greatest promotion chance. 8 For example: a person very talented in negative activity ( s i large), but bad in productive activity (low t i ) will beat all others. (l k sl alk ) / t k 0 . A contestant receives more total attack when he becomes abler in productive activity. (l k sl alk ) / ti 0 (i k ) . A person diverts from sabotage as he becomes more productive. If there is a new member added into the organization (call him person n 1 ), then n 1 l 1,l k sl alk () n s a l 1,l k l lk if rn 1 is small (large): Adding a dove (hawk) into the organization decrease (increases) the total attack inflicted on everybody. (Also Lazear, 1989). 9 3 Two-Person Case If t i t j , then ei aij e j a ji ; that is, i exerts more total effort. This, however, does not mean that i has more chance to win. If the two persons have the same ability in sabotage, then the one with higher value of t is more likely to win. Reason: He can mimic the effort level of opponent and guarantees a greater chance to win. Some interesting comparative statics results: ei / u 0, (i 1, 2) . One works harder in productive activity when prize increases. eij / u 0 . Negative effort level does not depend on prize. ei / t j does not have definite sign: Income and substitution effects. 10 4 Three-Person Case In 2-person case, a person abler in productive activity are more likely to win, when they are equally talented in sabotage. This is no longer true in 3-person case. To contrast the difference. Assume that s i 1 for all i . The ablest person cannot have the lowest chance to win, that is, he is at least No2 in ranking. But he probably does not have the greatest chance to be promoted. Example: r1 5, r2 4, r3 1. v(0) 0, v(1) 1, v(2) 3 , and v (3) 8 . g (0) 0, g (1) 2.5 , and g ( 2) 4 . 1 or 1 , each with probability 1 / 2 . Assume that the contestants have equal chance of being promoted if they have the same performance. 11 Tables 2, 3 and 4 show that the ablest person ends up having the second highest chance of being promoted. Note that this is impossible in 2-person case even for discrete specification. 5 Identical Contestants This case is of interests if contestants’ ability are unknown. The contestants not only exert less productive effort levels (than when they can only engage in productive activities), but also that some of the outputs are destroyed by opponents: Double inefficiency. 12 An increase in number of contestants will decrease level of negative effort: Negative effort is directed to a specific opponent. As they # of contestants increases, the return of negative activity decreases. 6 Alleviating Impact of Negative Activities Pay equality: reducing the value of u reduces the return of negative activity, and thus its level. However, it also reduces the return (and level) of productive effort. (Lazear, 1989. Industrial Politics). Seniority promotion system: In our model, it adds a term related to seniority in Wi , so that the influence of a ij on Wi decreases. Example: Japanese firms. Similarly, this also decreases value (and level) of productive effort. 13 Group compensation: If the promotion chance (or pay) of a member depends also on the performance of the group he belongs to, then the level of negative activity decreases. Example: NBA players. Problem: Free-riding incentive. Early designation of successor: Announcing the winner well in advance, so that negative effort is worthless. Example: Throne succession in Imperial China. Problem: 1. Commitment needed (Kangxi story). 2. Hurt the productive incentives. External recruiting: Since an inside contestant cannot sabotage outsiders, the return of negative effort decreases when there is outside recruitment. Example: Picking presidents in AT&T from presidents in subsidiaries rather than from headquarter. That means, contrary to literature, external recruitment also has an incentive role to pay (Chen, hopefully before Y2K). Problem: Same old problem. 14 If it is difficult to decompose the performance of a contestant into how much is form his own effort, and how much is destroyed by others, then it seems difficult to completely eliminate negative activity in an organization. Whether one wants to use a particular system depends on relative benefit and cost of it. 7 Extension Dynamic model: The contestants’ abilities are unknown, but can be learned from past performance. In this case there will be double jeopardy to efficiency: Since a person performing well earlier in career will be inferred as having higher ability and subject to more attack later, every contestant will hide their ability by exerting even less productive effort than the static model. 15 In reality, people do not compete as single individuals. They form factions (coalition). This is common especially in politics. Reason for forming factions: 1. Forming factions essentially breaks the all-or-nothing nature of promotion: Even if one loses in promotion, as long as his alliance wins, he can still share some of the gains from promotion. Risk sharing. 2. There is synergy in forming factions. Anyone who refuses to join a faction has no chance. Question: What is the structure of coalition in promotion tournament, and what is its consequence? 16 e1 a12 a13 W21 W31 v(e1 a1 ) u1 (a1 , e1 , a*1 , e*1 ) 1 1 0 0.5 -1 3 3 u 3 16 1 1 1 0.5 -3.5 8 1 u 8 4 1 0 1 3 -3.5 3 -3 1 0 0 3 -1 1 -1 0 1 0 5.5 4 1 -1 0 1 1 5.5 1.5 3 -3 0 0 1 8 1.5 1 -1 0 0 0 8 4 0 0 e2 a21 a23 W12 W32 v(e2 a2 ) u2 (a2 , e2 , a* 2 , e* 2 ) 1 1 0 0.5 -1 3 9 u 3 16 1 1 1 0.5 -3.5 8 3 u 8 4 1 0 1 3 -3.5 3 1 u 3 4 1 0 0 3 -1 1 3 u 1 16 0 1 0 5.5 4 1 -1 0 1 1 5.5 1.5 3 -3 0 0 1 8 1.5 1 -1 0 0 0 8 4 0 0 17 e3 a31 a32 W13 W23 v(e3 a3 ) u 3 ( a3 , e3 , a * 3 , e* 3 ) 0 1 0 1 1.5 1 1 u 1 16 0 1 1 1 0 3 1 u 3 8 0 0 1 2.5 0 1 -1 0 0 0 2.5 1.5 0 0 1 1 0 0 0.5 3 1 u 3 8 1 1 1 0 -1 8 3 u 8 8 1 0 1 1.5 -1 3 3 u 3 16 1 0 0 1.5 0.5 1 1 u 1 16 18
© Copyright 2026 Paperzz