Selection Efficiency in Multiple-Prize Tournaments with Sabotage

Baoting Huang; Shao-Chieh Hsueh; Min Qiang Zhao

doi:10.1515/bejeap-2024-0172

Artikel

Selection Efficiency in Multiple-Prize Tournaments with Sabotage

Baoting Huang , Shao-Chieh Hsueh und Min Qiang Zhao

Veröffentlicht/Copyright: 16. Dezember 2024

Veröffentlicht von

Veröffentlichen auch Sie bei De Gruyter Brill

Manuskript einreichen Informationen für Autor*innen Erkunden Sie dieses Fachgebiet

Aus der Zeitschrift The B.E. Journal of Economic Analysis & Policy Band 25 Heft 1

Abstract

Sabotage in competitive environments, like sales contests and sports, can hinder the effectiveness of rank-order tournaments in selecting the most capable individuals. Traditional winner-take-all tournaments may unintentionally level the playing field, making it difficult to distinguish the best. Despite extensive research on tournament design, the impact of sabotage on selection efficiency remains underexplored. This paper addresses this gap by investigating how the introduction of multiple-prize structures in rank-order tournaments affects selection efficiency in the presence of sabotage. Our analysis reveals that multiple prizes can improve the selection of high-performing contestants by redirecting sabotage toward weaker opponents, resulting in corner equilibria. In contrast, winner-take-all structures often result in interior equilibria, where promotion chances are equalized. By outlining the conditions under which these equilibria arise, we demonstrate that strategic prize design can enhance performance incentives, mitigate the negative impact of sabotage, and ultimately improve the selection efficiency of rank-order tournaments.

Keywords: prize design; tournament; sabotage; selection efficiency; corner equilibria

JEL Classification: C72; D23; D40

Corresponding author: Shao-Chieh Hsueh, Department of International Business, School of Economics and Management, Xiamen University of Technology, Xiamen, China, E-mail: schsuechieh@msn.com

Funding source: Natural Science Foundation of Xiamen, China

Award Identifier / Grant number: 3502Z202373068

Acknowledgments

We thank the editor and two anonymous reviewers for their helpful comments and suggestions. We are also grateful to the participants of the 10th China Meeting on Game Theory and Its Applications (July 2023) for their insightful feedback.

Research funding: Shao-Chieh Hsueh acknowledges financial support from the Natural Science Foundation of Xiamen, China (3502Z202373068), and the Social Science Foundation of Fujian, China (FJ2024B034). Baoting Huang acknowledges financial support from the Doctoral Research Start-up Foundation of Liming Vocational University (LRB202408).

Appendix A

A.1 Derivation of Probability and Expected Benefit Functions

The probability that contestant i achieves the highest rank (r _i = 1) can be computed as follows:

(A.1) P r i = 1 = Prob q i ≥ q j & q i ≥ q k = Prob f u i , θ j i + θ k i − f u j , θ i j + θ k j ≥ ε j − ε i & f u i , θ j i + θ k i − f u k , θ i k + θ j k ≥ ε k − ε i = ∫ − ∞ + ∞ ∫ − ∞ ε i + f u i , θ j i + θ k i − f u j , θ i j + θ k j g ε j d ε j × ∫ − ∞ ε i + f u i , θ j i + θ k i − f u k , θ i k + θ j k g ε k d ε k g ε i d ε i .

From equation (A.1), we can readily derive the following relationship:^[8]

(A.2) ∂ P r i = 1 ∂ u i = ∂ P r i = 1 ∂ θ i j + ∂ P r i = 1 ∂ θ i k .

Equation (A.2) demonstrates that the marginal increase in the probability of winning associated with productive effort is equivalent to the combined marginal increases in the probability of winning from sabotaging each of the other two contestants. Similarly, the probability that contestant i is ranked last (r _i = 3) can be expressed as:

(A.3) P r i = 3 = Prob q i ≤ q j & q i ≤ q k = Prob f u i , θ j i + θ k i − f u j , θ i j + θ k j ≤ ε j − ε i & f u i , θ j i + θ k i − f u k , θ i k + θ j k ≤ ε k − ε i = ∫ − ∞ + ∞ 1 − ∫ − ∞ ε i + f u i , θ j i + θ k i − f u j , θ i j + θ k j g ε j d ε j × 1 − ∫ − ∞ ε i + f u i , θ j i + θ k i − f u k , θ i k + θ j k g ε k d ε k d ε i g ε i d ε i .

The marginal winning probabilities of engaging in productive and sabotage activities can be connected in a similar way to equation (A.2):

(A.4) ∂ P r i = 3 ∂ u i = ∂ P r i = 3 ∂ θ i j + ∂ P r i = 3 ∂ θ i k .

Let p i u i , θ i j , θ i k denote the expected benefit of exerting productive and sabotage efforts in the payoff function (2):

(A.5) p i u i , θ i j , θ i k = W 1 P r i = 1 + W 2 P r i = 2 + W 3 P r i = 3 = W 2 + Δ W 1 ∫ − ∞ + ∞ G ε i − f j i G ε i − f k i g ε i d ε i − Δ W 2 ∫ − ∞ + ∞ 1 − G ε i − f j i 1 − G ε i − f k i g ε i d ε i .

A.2 Proof of Equation (A.2)

Given f _ji, as defined in Section 3.1.1, we can obtain f k i = f u k , θ i k + θ j k − f u i , θ j i + θ k i . Using the definition of P r i = 1 from equation (A.1), we can define the following equations:

(A.6) ∂ P r i = 1 ∂ u i = d f d u i ∫ − ∞ + ∞ g ε i g ε i − f j i ∫ − ∞ ε i − f k i g ε k d ε k d ε i + ∫ − ∞ + ∞ g ε i g ε i − f k i ∫ − ∞ ε i − f j i g ε j d ε j d ε i ,

(A.7) ∂ P r i = 1 ∂ θ i j = − d f d θ i j ∫ − ∞ + ∞ g ε i g ε i − f j i ∫ − ∞ ε i − f k i g ε k d ε k d ε i ,

(A.8) ∂ P r i = 1 ∂ θ i k = − d f d θ i k ∫ − ∞ + ∞ g ε i g ε i − f k i ∫ − ∞ ε i − f j i g ε j d ε j d ε i .

Since d f d u i = 1 , d f d θ i j = − 1 and d f d θ i k = − 1 , it is easy to show that

(A.9) ∂ P r i = 1 ∂ u i = ∂ P r i = 1 ∂ θ i j + ∂ P r i = 1 ∂ θ i k .

A.3 Proof of Proposition 1

The proof assumes that either ΔW ₁ or ΔW ₂ is greater than zero. According to the Kuhn–Tucker conditions, we can obtain

(A.10) Δ W 1 ∂ P r i = 1 ∂ θ i j − Δ W 2 ∂ P r i = 3 ∂ θ i j − ∂ C i ∂ θ i j ≤ 0 ,

(A.11) θ i j ≥ 0 , θ i j ∂ U i ∂ θ i j = 0 ,

where U i = W 1 P r i = 1 + W 2 P r i = 2 + W 3 P r i = 3 − C i u i , θ i j + θ i k . Since ∂ C i u i , θ i j + θ i k ∂ θ i j θ i j + θ i k = 0 = 0 , each contestant will sabotage at least one of the opponents.^[9]

Given W ₁ ≠ W ₂ or W ₂ ≠ W ₃, we can show that under each of the three conditions listed in Proposition 1, if f _j ≤ f _k, then contestant i will not sabotage opponent k. In contrast, suppose that contestant i sabotages opponent k θ i k > 0 .

Let us consider an alternative arrangement such that contestant i decreases θ _ik to 0 while increasing θ _ij by the same amount, which will not change the total cost. The corresponding benefit of such an arrangement is described by p i u i , θ i j + θ i k , 0 below:

(A.12) p i u i , θ i j + θ i k , 0 = W 2 + Δ W 1 ∫ − ∞ + ∞ G ε i − f j i + θ i k G ε i − f k i − θ i k g ε i d ε i − Δ W 2 ∫ − ∞ + ∞ 1 − G ε i − f j i + θ i k 1 − G ε i − f k i − θ i k g ε i d ε i .

Given that G ⋅ is strictly log-concave, g x G x should strictly decrease with x (An 1998). Hence, h ′ θ > 0 . Since h 0 = 0 , h θ should be positive for all θ > 0. Given equations (A.5) and (A.12), we can obtain

(A.13) p i u i , θ i j , θ i k − p i u i , θ i j + θ i k , 0 = Δ W 1 − Δ W 2 ∫ h θ i k g ε i d ε i − Δ W 2 ∫ F θ i k g ε i d ε i .

If p i u i , θ i j , θ i k ≤ p i u i , θ i j + θ i k , 0 , contestant i will not sabotage opponent k.

Condition (a). G ⋅ is convex, and ΔW ₁ − ΔW ₂ ≤ 0.

It is easy to verify condition (a) because G ⋅ is a convex function that ensures that F θ ≥ 0 .

Condition (b). G ⋅ is convex, ΔW ₁ − ΔW ₂ ≥ 0, and Δ W 1 − Δ W 2 ∫ h θ i k g ε i d ε i ≤ Δ W 2 ∫ F θ i k g ε i d ε i .

Given that h θ ≥ 0 , G ε i − f j i G ε i − f k i ≤ 1 , and ∫ g ε i d ε i = 1 , we have 0 ≤ ∫ h θ i k g ε i d ε i ≤ 1 . Next, we rearrange ∫ F θ i k g ε i d ε i as follows:

(A.14) ∫ F θ i k g ε i d ε i = ∫ ∫ ε i − f j i ε i − f j i + θ i k g ε i d ε i − ∫ ε i − f k i − θ i k ε i − f k i g ε i d ε i g ε i d ε i .

Given that F θ ≥ 0 , ∫ ε i − f j i ε i − f j i + θ i k g ε i d ε i ≤ 1 , and ∫ g ε i d ε i = 1 , we can obtain 0 ≤ ∫ F θ i k g ε i d ε i ≤ 1 . If ΔW ₂ is sufficiently large, it is easy to identify a set of multiple-prize allocations that can ensure p i u i , θ i j , θ i k ≤ p i u i , θ i j + θ i k , 0 .

Condition (c). G ⋅ is the increasing failure rate distribution. Let Y θ i k = h θ i k + F θ i k . Then, p i u i , θ i j , θ i k ≤ p i u i , θ i j + θ i k , 0 requires that Δ W 1 ∫ h θ i k g ε i d ε i ≤ Δ W 2 ∫ Y θ i k g ε i d ε i . Given that f _j ≤ f _k, G ⋅ is an increasing failure rate distribution ( g θ 1 − G θ increases with θ), which implies that:

(A.15) ∂ Y θ i k ∂ θ i k = g ε i − f j i + θ i k 1 − G ε i − f k i − θ i k − g ε i − f k i − θ i k 1 − G ε i − f j i + θ i k > 0 .

Since Y 0 = 0 , Y θ i k should be positive. If ΔW ₂ is sufficiently large and ΔW ₁ is sufficiently small, it is easy to identify a set of multiple-prize allocations that can ensure p i u i , θ i j , θ i k ≤ p i u i , θ i j + θ i k , 0 .

A.4 Proof of Lemma 1

Without loss of generality, for contestant i, we assume k is the underdog opponent such that f _ji > f _ki. Let Δ P 1 = ∂ P r i = 1 ∂ θ i j − ∂ P r i = 1 ∂ θ i k and Δ P 3 = ∂ P r i = 3 ∂ θ i j − ∂ P r i = 3 ∂ θ i k . From equation (2), we can obtain:

(A.16) ∂ U ∂ θ i j − ∂ U ∂ θ i k = Δ W 1 Δ P 1 − Δ W 2 Δ P 3 .

To prove Lemma 1, we need Lemmas 5 and 6.

Lemma 5.

If G ⋅ is strictly log-concave, then ΔP ₁ > 0.

Proof.

Let us expand ΔP ₁ as follows:

(A.17) Δ P 1 = ∂ P r i = 1 ∂ θ i j − ∂ P r i = 1 ∂ θ i k = ∫ − ∞ + ∞ g ε i g ε i − f j i ∫ − ∞ ε i − f k i g ε k d ε k d ε i − ∫ − ∞ + ∞ g ε i g ε i − f k i ∫ − ∞ ε i − f j i g ε j d ε j d ε i = ∫ − ∞ + ∞ g ε i − f j i G ε i − f k i − g ε i − f k i G ε i − f j i g ε i d ε i .

Since G ⋅ is strictly log-concave, g ( θ ) G ( θ ) should strictly decrease with θ. Given that f _ji > f _ki, it is easy to show that ∂ P r i = 1 ∂ θ i j > ∂ P r i = 1 ∂ θ i k . Q.E.D.

Lemma 6.

If G ⋅ is an increasing failure rate distribution, then ΔP ₃ > 0.

Proof.

Let us expand ΔP ₃ as follows:

(A.18) Δ P 3 = ∂ P r i = 3 ∂ θ i j − ∂ P r i = 3 ∂ θ i k = − ∫ − ∞ + ∞ g ε i g ε i − f j i 1 − ∫ − ∞ ε i − f k i g ε k d ε k d ε i + ∫ − ∞ + ∞ g ε i g ε i − f k i 1 − ∫ − ∞ ε i − f j i g ε j d ε j d ε i = ∫ − ∞ + ∞ g ε i − f k i 1 − G ε i − f j i − g ε i − f j i 1 − G ε i − f k i g ε i d ε i ,

Given that G ⋅ has an increasing failure rate , g ( θ ) 1 − G ( θ ) should increase with θ. Given that f _ji > f _ki, it is easy to show that ∂ P r i = 3 ∂ θ i j > ∂ P r i = 3 ∂ θ i k . Q.E.D.

G ⋅ is assumed to be strictly log-concave. Given that f _j > f _k, Lemma 5 implies that ΔP ₁ > 0. Let Δ = Δ P 3 − Δ P 1 = ∫ − ∞ + ∞ g ε i − f k i − g ε i − f j i g ε i d ε i . Then, ΔP ₃ = ΔP ₁ + Δ. Equation (A.16) can be rearranged as follows:

(A.19) ∂ U ∂ θ i j − ∂ U ∂ θ i k = Δ W 1 − Δ W 2 Δ P 1 − Δ W 2 Δ .

Case (1): If G ⋅ is convex ( g ′ ⋅ > 0 ), f _ji > f _ki implies that g ε i − f k i − g ε i − f j i > 0 . In other words, Δ = Δ P 3 − Δ P 1 = ∫ − ∞ + ∞ g ε i − f k i − g ε i − f j i g ε i d ε i > 0 . Given that ΔW ₁ − ΔW ₂ ≤ 0 and ΔP ₁ > 0 (from Lemma 5), it is easy to show that equation (A.19) should be negative. That is, ∂ U ∂ θ i j < ∂ U ∂ θ i k . It will be optimal to sabotage the underdog opponent.

Case (2): If G ⋅ is an increasing failure rate distribution, Lemma 6 implies that ΔP ₃ > 0. Then, if W ₁ = W ₂ and W ₂ > W ₃, equation (A.19) can be simplified as ∂ U ∂ θ i j − ∂ U ∂ θ i k = − Δ W 2 Δ P 3 < 0 . It will be optimal to sabotage the underdog opponent k.

Case (3): If W ₂ = W ₃ and W ₁ > W ₂, equation (A.19) can be simplified as ∂ U ∂ θ i j − ∂ U ∂ θ i k = Δ W 1 Δ P 1 . Since ΔW ₁ > 0, ∂ U ∂ θ i j − ∂ U ∂ θ i k > 0 . It will be optimal for contestant i to sabotage the favorite opponent j.

Case (4): If G ⋅ is concave ( g ′ ⋅ < 0 ), f _ji > f _ki implies that g ε i − f k i − g ε i − f j i < 0 . Then, Δ < 0. Given that ΔW ₁ − ΔW ₂ ≥ 0 and ΔP ₁ > 0, it is easy to show that equation (A.19) should be positive. That is ∂ U ∂ θ i j > ∂ U ∂ θ i k . It will be optimal to sabotage the favorite opponent.

A.5 Proof of Lemma 2

Suppose to the contrary that f _j ≤ f _k. If one of the conditions in Proposition 1 holds and f _j ≤ f _k, Appendix A.3 shows that p i u i , θ i j , θ i k ≤ p i u i , θ i j + θ i k , 0 . In other words, θ _ik should be zero, which contradicts with θ _ik > 0.

Furthermore, if one of the conditions in Proposition 1 holds and f _j > f _k, it is easy to show that θ _ij = 0 based on Appendix A.3.

A.6 Proof of Lemma 3

By replacing C i ⋅ with γ i C ⋅ , the objective function defined in (2) can be modified as follows:

(A.20) U i = W 2 + Δ W 1 P r i = 1 − Δ W 2 P r i = 3 − γ i C u i , θ i j + θ i k .

To evaluate the Hessian matrix of the objective function, let us first focus on its second derivatives, illustrated below:

(A.21) U u i u i = A − γ i ∂ 2 C ∂ u i 2 ,

(A.22) U u i θ i j = H − γ i ∂ 2 C ∂ u i ∂ θ i j + θ i k ,

(A.23) U u i θ i k = D − γ i ∂ 2 C ∂ u i ∂ θ i j + θ i k ,

(A.24) U θ i j θ i j = Q − γ i ∂ 2 C ∂ θ i j + θ i k 2 ,

(A.25) U θ i j θ i k = E − γ i ∂ 2 C ∂ θ i j + θ i k 2 ,

(A.26) U θ i k θ i k = F − γ i ∂ 2 C ∂ θ i j + θ i k 2 ,

where

A = ∫ g ′ ε i − f j i G ε i − f k i g ε i d ε i + 2 ∫ g ε i − f j i g ε i − f k i g ε i d ε i + ∫ g ′ ε i − f k i G ε i − f j i g ε i d ε i Δ W 1 − Δ W 2 + ∫ g ′ ε i − f j i g ε i d ε i + ∫ g ′ ε i − f k i g ε i d ε i Δ W 2 ,

H = ∫ g ε i − f j i g ε i − f k i g ε i d ε i + ∫ g ′ ε i − f j i G ε i − f k i g ε i d ε i Δ W 1 − Δ W 2 + ∫ − ∞ + ∞ g ′ ε i − f j i g ε i d ε i Δ W 2 ,

D = ∫ g ε i − f j i g ε i − f k i g ε i d ε i + ∫ g ′ ε i − f k i G ε i − f j i g ε i d ε i Δ W 1 − Δ W 2 + ∫ g ′ ε i − f k i g ε i d ε i Δ W 2 ,

Q = ∫ g ′ ε i − f j i G ε i − f k i g ε i d ε i Δ W 1 − Δ W 2 + ∫ g ′ ε i − f j i g ε i d ε i Δ W 2 ,

E = ∫ g ε i − f j i g ε i − f k i g ε i d ε i Δ W 1 − Δ W 2 ,

F = ∫ g ′ ε i − f k i G ε i − f j i g ε i d ε i Δ W 1 − Δ W 2 + ∫ g ′ ε i − f k i g ε i d ε i Δ W 2 .

Given that C ⋅ is strictly convex, we can show that ∂ 2 C ∂ u i 2 > 0 , ∂ 2 C ∂ u i 2 ∂ 2 C ∂ θ i j + θ i k 2 − ∂ 2 C ∂ u i ∂ θ i j + θ i k 2 > 0 .

To identify the conditions that ensure that the Hessian matrix of the objective function is negative semidefinite, we examine the determinants of all upper-left Hessian matrices of the objective function. First, given that C ⋅ is strictly convex, if γ _i is sufficiently large, it is easy to show that U u i u i = A − γ i ∂ 2 C ∂ u i 2 is negative and U u i u i U u i θ i j U u i θ i j U θ i j θ i j = ∂ 2 C ∂ u i 2 ∂ 2 C ∂ θ i j + θ i k 2 − ∂ 2 C ∂ u i ∂ θ i j + θ i k 2 γ i 2 − A ∂ 2 C ∂ θ i j + θ i k 2 + Q ∂ 2 C ∂ u i 2 − 2 H ∂ 2 C ∂ u i ∂ θ i j + θ i k + A B − H 2 is positive.^[10]

Next, the determinant of the whole Hessian matrix can be derived as follows:

(A.27) U u i u i U u i θ i j U u i θ i k U u i θ i j U θ i j θ i j U θ i j θ i k U u i θ i k U θ i j θ i k U θ i k θ i k = 0 · γ i 3 + I + L Δ W 1 − Δ W 2 + J + K Δ W 2 × ∂ 2 C ∂ u i 2 ∂ 2 C ∂ θ i j + θ i k 2 − ∂ 2 C ∂ u i ∂ θ i j + θ i k ∂ 2 C ∂ u i ∂ θ i j + θ i k γ i 2 + H 2 + D 2 + 2 E A − 2 H D − F A − A Q ∂ 2 C ∂ θ i j + θ i k 2 + 2 F H + 2 Q D − 2 D E − 2 H E ∂ 2 C ∂ u i ∂ θ i j + θ i k + E 2 − F Q ∂ 2 C ∂ u i 2 γ i + 2 D H E + F A Q − D 2 Q − E 2 A − H 2 F ,

where

I = ∫ − ∞ + ∞ g ′ ε i − f j i G ε i − f k i g ε i d ε i − ∫ − ∞ + ∞ g ε i − f k i g ε i − f j i g ε i d ε i

L = ∫ − ∞ + ∞ g ′ ε i − f k i G ε i − f j i g ε i d ε i − ∫ − ∞ + ∞ g ε i − f k i g ε i − f j i g ε i d ε i

J = ∫ g ′ ε i − f k i g ε i d ε i

K = ∫ g ′ ε i − f j i g ε i d ε i

If γ _i is sufficiently large, a negative sign of I + L Δ W 1 − Δ W 2 + J + K Δ W 2 can ensure that the determinant of the whole Hessian matrix is negative.

Condition (a) – G ⋅ is convex, ΔW ₁ − ΔW ₂ ≥ 0.

Given that G ⋅ is convex and ΔW ₁ − ΔW ₂ ≥ 0, it is easy to show that 0 ≤ N < J < M, 0 ≤ N < K < M, g + ∞ = B , and g − ∞ = S .

Using integration by parts, we can modify the expression of I as follows:

(A.28) I = ∫ − ∞ + ∞ G 2 ε i − f k i g ε i d g ε i − f j i G ε i − f k i = g 2 + ∞ − 2 ∫ − ∞ + ∞ g ε i − f k i g ε i − f j i g ε i d ε i − ∫ − ∞ + ∞ G ε i − f k i g ε i − f j i g ′ ε i d ε i .

Based on the upper and lower bounds of g ⋅ , it is easy to show that the second component of (A.28) is bounded as follows:

(A.29) − 2 B 2 < − 2 ∫ − ∞ + ∞ g ε i − f k i g ε i − f j i g ε i d ε i < − 2 S 2 .

The third component of (A.28) can be rearranged as follows:

(A.30) 0 < ∫ − ∞ + ∞ G ε i − f k i g ε i − f j i g ′ ε i d ε i < B ∫ − ∞ + ∞ g ′ ε i d ε i = B B − S .

Given (A.28), (A.29) and (A.30), we can obtain that

(A.31) − B 2 − B B − S < I < B 2 − 2 S 2 .

Similarly, we can also show that L is bounded between − B 2 − B B − S and B ² − 2S ². The following condition can ensure that the sign of I + L Δ W 1 − Δ W 2 + J + K Δ W 2 is negative:

(A.32) Δ W 2 Δ W 1 − Δ W 2 < 2 S 2 − B 2 M .

In other words, given that G ⋅ is convex, ΔW ₁ − ΔW ₂ ≥ 0, and Δ W 2 Δ W 1 − Δ W 2 < 2 S 2 − B 2 M , when γ _i is sufficiently large, the Hessian matrix of the objective function is negative semidefinite.

Condition (b) – G ⋅ is concave,ΔW ₁ − ΔW ₂ ≤ 0.

Given that G is concave and ΔW ₁ − ΔW ₂ ≤ 0, it is easy to show that N < J < M ≤ 0, N < K < M ≤ 0, g + ∞ = S , and g − ∞ = B .

The boundary condition is still applied to the second component of (A.28), but the third component is modified as follows:

(A.33) B S − B < ∫ − ∞ + ∞ G ε i − f k i g ε i − f j i g ′ ε i d ε i < 0 .

Given (A.28), (A.29) and (A.33), we can obtain that

(A.34) S 2 − 2 B 2 < I < − S 2 + B B − S .

Similarly, we can also show that L is bounded between S ² − 2B ² and − S 2 + B B − S . The following condition can ensure that the sign of I + L Δ W 1 − Δ W 2 + J + K Δ W 2 is negative:

(A.35) Δ W 2 Δ W 2 − Δ W 1 > B 2 − S 2 − B S M .

In other words, given that G ⋅ is concave,ΔW ₁ − ΔW ₂ ≤ 0, and Δ W 2 Δ W 2 − Δ W 1 > B 2 − S 2 − B S M , when γ _i is sufficiently large, the Hessian matrix of the objective function is negative semidefinite.

A.7 Proof of Proposition 3

Proposition 3 implies that each contestant sabotages only one opponent with a relatively lower ability level if one of the conditions in Proposition 2 is satisfied. We show that no contestant will deviate from these equilibria. Let us use contestant i as an example. Suppose to the contrary that θ _ik > 0. Then, according to Lemma 2, we have θ _ij = 0 and f _j > f _k. Let us consider an alternative arrangement such that contestant i decreases θ _ik to 0 while increasing θ _ij by the same amount, which will not change the total cost. The difference between the benefit associated with θ _ik > 0 and θ _ij = 0 (denoted by p i u i , θ i j , θ i k , see equation (3)) and the benefit under the alternative arrangement (denoted by p i u i , θ i j + θ i k , 0 , see equation (A.12)) is provided in equation (A.13).

Before we examine the difference between p i u i , θ i j , θ i k and p i u i , θ i j + θ i k , 0 , we need to show how productive and sabotage efforts change with the ability parameter γ, as summarized in Lemma 7 below.

Lemma 7.

If one of the conditions in Proposition 2 holds, both productive and sabotage efforts decrease with the ability parameter γ.

Proof.

Without loss of generality, let us focus on contestant i. We first show that ∂ 2 U i ∂ u i ∂ θ i j ≥ 0 holds if one of the conditions in Proposition 2 is satisfied.

Given (A.20), ∂ 2 U i ∂ u i ∂ θ i j can be expressed as follows:

(A.36) ∂ 2 U i ∂ u i ∂ θ i j = Δ W 1 − Δ W 2 R + Δ W 2 T − γ i ∂ 2 C ∂ u i ∂ θ i j + θ i k ,

where

(A.37) R = ∫ − ∞ + ∞ g ε i − f k i g ε i − f j i g ε i d ε i + ∫ − ∞ + ∞ g ′ ε i − f j i G ε i − f k i g ε i d ε i ,

(A.38) T = ∫ − ∞ + ∞ g ′ ε i − f j i g ε i d ε i .

According to integration by parts,

(A.39) R = ∫ − ∞ + ∞ g ε i − f k i g ε i − f j i g ε i d ε i + ∫ − ∞ + ∞ g ′ ε i − f j i G ε i − f k i g ε i d ε i = ∫ − ∞ + ∞ g ε i d G ε i − f k i g ε i − f j i .

Under Condition (a) of Proposition 2, we can obtain that g′ ≥ 0, g − ∞ = S , and g + ∞ = B . Then, R can be simplified as follows:

(A.40) R = B 2 − ∫ − ∞ + ∞ g ε i − f j i G ε i − f k i g ′ ε i d ε i .

Since S ≤ g ≤ B, 0 ≤ G ≤ 1 and g′ ≥ 0, the second component of (A.40) is bounded as follows:

(A.41) − B B − S = B − ∫ − ∞ + ∞ g ′ ε i d ε i < − ∫ − ∞ + ∞ g ε i − f j i G ε i − f k i g ′ ε i d ε i < 0 ,

Given (A.40) and (A.41), we can obtain that 0 ≤ BS < R < B ². For T, since g′ ≥ N ≥ 0, T = ∫ − ∞ + ∞ g ′ ε i − f j i g ε i d ε i > 0 . Given that ΔW ₁ − ΔW ₂ ≥ 0, R > 0, T > 0, and ∂ 2 C ∂ u i ∂ θ i j + θ i k ≤ 0 , it is easy to show that ∂ 2 U i ∂ u i ∂ θ i j ≥ 0 under Condition (a) of Proposition 2.

Under Condition (b) of Proposition 2, g′ ≤ 0, g − ∞ = B , g + ∞ = S , and equation (A.40) can be modified as follows:

(A.42) R = S 2 − ∫ − ∞ + ∞ g ε i − f j i G ε i − f k i g ′ ε i d ε i .

Similarly, we can show that 0 ≤ S ² < R < S ² + B(B − S). For T, since N ≤ g′ ≤ 0, we obtain that N = N ∫ − ∞ + ∞ g ε i d ε i < T < 0 . Given γ _i is sufficiently large (i.e., γ i ≥ Δ W 2 − Δ W 1 S 2 + B B − S + Δ W 2 − N − ∂ 2 C / ∂ u i ∂ θ i j + θ i k > 0 ) and the boundary conditions for R and T it is easy to show that ∂ 2 U i ∂ u i ∂ θ i j ≥ 0 under Condition (b) of Proposition 2.

Under Proposition 2, contestant i sabotages only one opponent. Without loss of generality, assume contestant j is sabotaged by contestant i. The resulting first-order conditions are presented below:

(A.43) ∂ U i ∂ u i = Δ W 1 ∂ P r i = 1 ∂ u i − Δ W 2 ∂ P r i = 3 ∂ u i − γ i ∂ C ∂ u i = 0 ;

(A.44) ∂ U i ∂ θ i j = Δ W 1 ∂ P r i = 1 ∂ θ i j − Δ W 2 ∂ P r i = 3 ∂ θ i j − γ i ∂ C ∂ θ i j = 0 .

To examine how γ _i affects u _i and θ _ij, we conduct a counterfactual analysis. Given a corer equilibrium, if γ _i increases slightly and all the previously determined optimal conditions are not allowed to change, then both ∂ U i ∂ u i and ∂ U i ∂ θ i j should drop below zero because ∂ C ∂ u i > 0 and ∂ C ∂ θ i j > 0 . Given that U _i is a concave function, we can obtain that ∂ 2 U i ∂ u i 2 ≤ 0 , ∂ 2 U i ∂ θ i j 2 ≤ 0 , and ∂ 2 U i ∂ u i ∂ θ i j 2 ≤ ∂ 2 U i ∂ u i 2 ⋅ ∂ 2 U i ∂ θ i j 2 . It is clear that the new equilibria cannot be obtained if u _i and θ _ij change in opposite directions because ∂ 2 U i ∂ u i 2 ≤ 0 , ∂ 2 U i ∂ θ i j 2 ≤ 0 , and ∂ 2 U i ∂ u i ∂ θ i j ≥ 0 . If both u _i and θ _ij increase, ∂ 2 U i ∂ u i ∂ θ i j 2 ≤ ∂ 2 U i ∂ u i 2 ⋅ ∂ 2 U i ∂ θ i j 2 cannot be satisfied. Therefore, both productive and sabotage efforts decrease with the ability parameter γ. Q.E.D.

Since each contestant differs only by his ability parameter γ, Lemma 7 implies that the contestant with a higher ability level should expend more productive effort u. In other words, γ _i > γ _j > γ _k implies that u _k > u _j > u _i.

Given that f _j > f _k, u _k > u _j > u _i, f _i = u _i − θ _ki − θ _ji, f _j = u _j and f _k = u _k − θ _ik, it is easy to show that ɛ _i − f _ji > ɛ _i − f _ki − θ _ik. Then, G ⋅ , which is strictly log-concave, can ensure that h θ i k > 0 for any positive value of θ _ik. According to Appendix A.3, p i u i , θ i j , θ i k − p i u i , θ i j + θ i k , 0 < 0 if one of the conditions listed in Proposition 2 is satisfied. This contradicts θ _ik > 0. Therefore, contestant i has no incentive to deviate from these equilibria, and f _j < f _k according to Lemma 2.

By the same token, neither contestant j nor contestant k has any incentive to deviate from these equilibria.

A.8 Proof of Corollary 1

If one of the conditions in Proposition 2 holds, contestant i will sabotage only one opponent. Contestant i is indifferent to choosing between j and k because γ _j = γ _k, which gives rise to two cases under Corollary 1.

Next, we show that neither contestant j nor contestant k has any incentive to deviate from those equilibria. Let us first focus on contestant j and Case (1). Suppose to the contrary that θ _jk > 0.

Lemma 2 implies that θ _ji = 0 and f _i > f _k. Next, let us consider an alternative arrangement such that contestant j decreases θ _jk to 0 while increasing θ _ji by the same amount, which will not change the total cost. The difference between the benefit associated with θ _jk > 0 and θ _ji = 0 (denoted by p j u j , θ j i , θ j k ) and the benefit under the alternative arrangement (denoted by p j u j , θ j i + θ j k , 0 ) is provided below:

(A.45) p j u j , θ j i , θ j k − p j u j , θ j i + θ j k , 0 = Δ W 1 − Δ W 2 ∫ h ̃ θ j k g ε j d ε j − Δ W 2 ∫ F ̃ θ j k g ε j d ε j ,

where h ̃ θ j k = G ε j − f i j G ε j − f k j − G ε j − f i j + θ j k G ε j − f k j − θ j k , and F ̃ θ j k = G ε j − f i j + θ j k + G ε j − f k j − θ j k − G ε j − f i j − G ε j − f k j .

Since j and k are identical contestants and act simultaneously, they must expend the same amount of productive effort, which should be greater than that of contestant i, who has lower ability according to Lemma 7. That is, u _k = u _j > u _i. Similarly, contestants j and k should expend more sabotage effort than contestant i. That is θ _ki = θ _jk > θ _ij.

Given u _k = u _j > u _i, f _i = u _i − θ _ki, f _j = u _j − θ _ij and f _k = u _k − θ _jk, we can easily show that ɛ _j − f _ij > ɛ _j − f _kj − θ _jk. Then, G ⋅ , which is strictly log-concave, can ensure that h ̃ θ j k > 0 for any positive value of θ _jk. According to Appendix A.3, under the conditions in Proposition 2, p j u j , θ j i , θ j k − p j u j , θ j i + θ j k , 0 < 0 . This contradicts θ _jk > 0. Therefore, contestant j has no incentive to deviate from this equilibrium, and f _i < f _k according to Lemma 2.

Next, let us focus on contestant j and Case (2). Again, suppose to the contrary that θ _jk > 0. Similarly, equation (A.45) is still applied. Given θ _ki = θ _jk > θ _ik, u _k = u _j > u _i, f _i = u _i − θ _ki, f _j = u _j and f _k = u _k − θ _ik − θ _jk, we can easily show that ɛ _j − f _ij > ɛ _j − f _kj − θ _jk. Then, based on the property of G ⋅ and Appendix A.3, it is easy to show that p j u j , θ j i , θ j k − p j u j , θ j i + θ j k , 0 < 0 . This again contradicts θ _jk > 0. Therefore, contestant j has no incentive to deviate from this equilibrium under case (2).

Since both j and k are identical contestants, neither contestant k will have any incentives to deviate from these equilibria under both cases, and f _i < f _j. In addition, if contestant i sabotages j, it is easy to show that f _j < f _k, and vice versa.

A.9 Proof of Corollary 2

If one of the conditions in Proposition 2 holds, contestant k will sabotage only one opponent. Contestant k is indifferent to choosing between i and j because γ _i = γ _j, which gives rise to two cases under Corollary 2.

Next, we show that neither contestant i nor contestant j has any incentive to deviate from those equilibria. Let us first focus on contestant i and Case (1). Suppose to the contrary that θ _ik > 0.

Lemma 2 implies that θ _ij = 0 and f _j > f _k. Next, let us consider an alternative arrangement such that contestant i decreases θ _ik to 0 while increasing θ _ij by the same amount, which will not change the total cost. The difference between the benefit associated with θ _ik > 0 and θ _ij = 0 (denoted by p i u i , θ i j , θ i k , see equation (3)) and the benefit under the alternative arrangement (denoted by p i u i , θ i j + θ i k , 0 , see equation (A.12)) is provided in equation (A.13).

Since i and j are identical contestants and act simultaneously, they must expend the same amount of productive effort, which should be lower than that of contestant k, who has greater ability according to Lemma 7. That is, u _k > u _j = u _i.

Given u _k > u _j = u _i, f _i = u _i − θ _ki − θ _ji, f _j = u _j and f _k = u _k − θ _ik, it is easy to show that ɛ _i − f _ji > ɛ _i − f _ki − θ _ik. Then, G ⋅ , which is strictly log-concave, can ensure that h θ i k > 0 for any positive value of θ _ik. According to Appendix A.3, p i u i , θ i j , θ i k − p i u i , θ i j + θ i k , 0 < 0 if one of the conditions listed in Proposition 2 is satisfied. This contradicts θ _ik > 0. Therefore, contestant i has no incentive to deviate from the equilibria in Case (1), and f _k > f _j according to Lemma 2.

Next, let us focus on contestant i and Case (2). Again, suppose to the contrary that θ _ik > 0. Similarly, equation (A.13) is still applied. Given u _k > u _j = u _i, f _i = u _i − θ _ji, f _j = u _j − θ _kj and f _k = u _k − θ _ik, it is easy to show that ɛ _i − f _ji > ɛ _i − f _ki − θ _ik. Then, based on the property of G ⋅ and Appendix A.3, it is easy to show that p i u i , θ i j , θ i k − p i u i , θ i j + θ i k , 0 < 0 . This again contradicts θ _ik > 0. Therefore, contestant i has no incentive to deviate from this equilibrium under case (2).

Since both i and j are identical contestants, neither contestant j will have any incentives to deviate from these equilibria under both cases, and f _k > f _i. In addition, if contestant k sabotages i, it is easy to show that f _j > f _i, and vice versa.

A.10 Proof of Lemma 4

To prove the equivalence between the two inequalities, let us first assume that u j − θ i j + θ k j > u k − θ i k + θ j k . That is, f _j > f _k. To simplify our notation, let Z = Δ W 1 ∂ P r i = 1 ∂ θ i j − Δ W 2 ∂ P r i = 3 ∂ θ i j − Δ W 1 ∂ P r i = 1 ∂ θ i k − Δ W 2 ∂ P r i = 3 ∂ θ i k . We can further expand Z as follows:

(A.46) Z = Δ W 1 − Δ W 2 ∫ − ∞ + ∞ g ε i − f j i G ε i − f k i − g ε i − f k i G ε i − f j i g ε i d ε i − Δ W 2 ∫ − ∞ + ∞ g ε i − f k i − g ε i − f j i g ε i d ε i .

Given that G ⋅ is strictly log-concave, g x G x strictly decreases with x. Therefore, f _j > f _k implies that g ε i − f j i G ε i − f k i − g ε i − f k i G ε i − f j i > 0 .

For Case (a), when ΔW ₁ > ΔW ₂ = 0, it is easy to show that Z > 0. For Case (b), given that G x is a concave function, g ′ x < 0 . In other words, g ε i − f k i − g ε i − f j i < 0 . Then, ΔW ₁ − ΔW ₂ ≥ 0 can ensure that Z > 0.

Next, let us assume that Z > 0. For Case (a), when ΔW ₁ > ΔW ₂ = 0, Z > 0 implies that ∫ − ∞ + ∞ g ε i − f j i G ε i − f k i − g ε i − f k i G ε i − f j i g ε i d ε i > 0 . Given that G ⋅ is strictly log-concave, only f _j > f _k can satisfy Z > 0, which means that u j − θ i j + θ k j > u k − θ i k + θ j k .

For Case (b), s is the opposite of f _j ≤ f _k. G x being a concave function implies that g ε i − f k i − g ( ε i − f j i ) ≥ 0 . Given that G ⋅ is strictly log-concave and f _j ≤ f _k, g ( ε i − f j i ) G ε i − f k i − g ε i − f k i G ( ε i − f j i ) ≤ 0 . Then, from (A.46), ΔW ₁ − ΔW ₂ ≥ 0 implies that Z ≤ 0, which contradicts Z > 0. Therefore, f _j > f _k. That is, u j − θ i j + θ k j > u k − θ i k + θ j k .

A.11 Proof of Proposition 4

Under interior equilibria, each contestant will sabotage all rivals. If u j − θ i j + θ k j > u k − θ i k + θ j k , Lemma 4 implies that Δ W 1 ∂ P r i = 1 ∂ θ i j − Δ W 2 ∂ P r i = 3 ∂ θ i j > Δ W 1 ∂ P r i = 1 ∂ θ i k − Δ W 2 ∂ P r i = 3 ∂ θ i k . Then, contestant i can decrease θ _ik and simultaneously increase θ _ij by the same amount, which will not change the total cost but can help increase the expected payoff. Therefore, u j − θ i j + θ k j = u k − θ i k + θ j k is required to reach interior equilibrium. The probability of winning across contestants will be equalized.

A.12 Necessary Conditions for the Existence of Interior Equilibria

To simplify the notation, let us define:

(A.47) W ̃ = Δ W 1 p 1 − Δ W 2 p 3 ,

where p 1 = ∂ P r i = 1 ∂ u i f j i = f k i = 0 = 2 ∫ − ∞ + ∞ g 2 ε G ε d ε ,

p 3 = Δ W 2 ∂ P r i = 3 ∂ u i f j i = f k i = 0 = − 2 ∫ − ∞ + ∞ g 2 ε 1 − G ε d ε .

Under the interior equilibria in which f _i = f _j = f _k, equations (4)–(6) can be modified as follows:

(A.48) u i = W ̃ γ i , u j = W ̃ γ j , u k = W ̃ γ k ,

(A.49) θ i j + θ i k = W ̃ 2 γ i , θ j i + θ j k = W ̃ 2 γ j , θ k i + θ k j = W ̃ 2 γ k

(A.50) u i − θ j i + θ k i = u j − θ i j + θ k j = u k − θ i k + θ j k

Let Θ_i denote the total sabotage effort aimed at contestant i. That is Θ_i = θ _ji + θ _ki. Then, the total amount of sabotage effort can be expressed as

(A.51) Θ i + Θ j + Θ k = W ̃ 2 r i + W ̃ 2 r j + W ̃ 2 r k .

Next, given (A.48) and (A.50), we can obtain

(A.52) W ̃ γ i − Θ i = W ̃ γ j − Θ j = W ̃ γ k − Θ k .

By combining (A.51) and (A.52), we can solve for Θ:

(A.53) Θ i = W ̃ 6 − γ i γ k + 5 γ j γ k − γ i γ j γ i γ j γ k ,

(A.54) Θ j = W ̃ 6 − γ j γ k + 5 γ i γ k − γ j γ i γ i γ j γ k ,

(A.55) Θ k = W ̃ 6 − γ k γ j + 5 γ i γ j − γ k γ i γ i γ j γ k .

Therefore, Θ_i > 0 requires that γ j + γ k < 5 γ j γ k γ i ; Θ_j > 0 requires that γ i + γ k < 5 γ i γ k γ j ; and Θ_k > 0 requires that γ i + γ j < 5 γ i γ j γ k .

References

Amegashie, J. A. 2012. “Productive versus Destructive Efforts in Contests.” European Journal of Political Economy 28 (4): 461–8. https://doi.org/10.1016/j.ejpoleco.2012.05.005.Suche in Google Scholar

An, M. Y. 1998. “Logconcavity versus Logconvexity: A Complete Characterization.” Journal of Economic Theory 80 (2): 350–69, https://doi.org/10.1006/jeth.1998.2400.Suche in Google Scholar

Bagnoli, M., and T. Bergstrom. 2005. “Log-Concave Probability and its Applications.” Economic Theory 26: 445–69. https://doi.org/10.1007/s00199-004-0514-4.Suche in Google Scholar

Balafoutas, L., F. Lindner, and M. Sutter. 2012. “Sabotage in Tournaments: Evidence from a Natural Experiment.” Kyklos 65 (4): 425–41. https://doi.org/10.1111/kykl.12000.Suche in Google Scholar

Balafoutas, L., E. G. Dutcher, F. Lindner, and D. Ryvkin. 2017. “The Optimal Allocation of Prizes in Tournaments of Heterogeneous Agents.” Economic Inquiry 55 (1): 461–78. https://doi.org/10.1111/ecin.12380.Suche in Google Scholar

Becker, G. S. 1968. “Crime and Punishment: An Economic Approach.” Journal of Political Economy 76 (2): 169–217. https://doi.org/10.1007/978-1-349-62853-7_2.Suche in Google Scholar

Bhattacharya, S., and L. Guasch. 1988. “Heterogeneity, Tournaments, and Hierarchies.” Journal of Political Economy 96: 867–81. https://doi.org/10.1086/261567.Suche in Google Scholar

Brown, A., and S. M. Chowdhury. 2017. “The Hidden Perils of Affirmative Action: Sabotage in Handicap Contests.” Journal of Economic Behavior & Organization 133: 273–84. https://doi.org/10.1016/j.jebo.2016.11.009.Suche in Google Scholar

Brown, J., and D. B. Minor. 2014. “Selecting the Best? Spillover and Shadows in Elimination Tournaments.” Management Science 60 (12): 3087–102. https://doi.org/10.1287/mnsc.2014.2014.Suche in Google Scholar

Chen, K. P. 2003. “Sabotage in Promotion Tournaments.” Journal of Law, Economics, and Organization 19 (1): 119–40. https://doi.org/10.1093/jleo/19.1.119.Suche in Google Scholar

Chowdhury, S., and O. Gürtler. 2015. “Sabotage in Contests: A Survey.” Public Choice 164 (1–2): 135–55. https://doi.org/10.1007/s11127-015-0264-9.Suche in Google Scholar

Clark, D., and C. Riis. 2001. “Rank-Order Tournaments and Selection.” Journal of Economics 73 (2): 167–91. https://doi.org/10.1007/BF02340174.Suche in Google Scholar

Curry, P., and S. Mongrain. 2009. “Deterrence in Rank-Order Tournaments.” Review of Law & Economics 5 (1): 723–40. https://doi.org/10.2202/1555-5879.1338.Suche in Google Scholar

Del Corral, J., J. Prieto-Rodriguez, and R. Simmons. 2010. “The Effect of Incentives on Sabotage: The Case of Spanish Football.” Journal of Sports Economics 11 (3): 243–60. https://doi.org/10.1177/1527002509340666.Suche in Google Scholar

Delfgaauw, J., R. Dur, A. Non, and W. Verbeke. 2015. “The Effects of Prize Spread and Noise in Elimination Tournaments: A Natural Field Experiment.” Journal of Labor Economics 33 (3): 521–69. https://doi.org/10.1086/679670.Suche in Google Scholar

Ehrlich, I. 1972. “The Deterrent Effect of Criminal Law Enforcement.” Journal of Legal Studies 1 (2): 259–76. https://doi.org/10.1086/467485.Suche in Google Scholar

Fu, Q., J. Lu, and Z. Wang. 2014. ““Reverse” Nested Lottery Contests.” Journal of Mathematical Economics 50: 128–40. https://doi.org/10.1016/j.jmateco.2013.08.007.Suche in Google Scholar

Fu, Q., X. Wang, and Z. Wu. 2021. “Multiprize Contests with Risk-Averse Players.” Games and Economic Behavior 129: 513–35. https://doi.org/10.1016/j.geb.2021.07.003.Suche in Google Scholar

Ghosh, C., D. Huang, N. H. Nguyen, and H. V. Phan. 2023. “CEO Tournament Incentives and Corporate Debt Contracting.” Journal of Corporate Finance 78: 102320. https://doi.org/10.1016/j.jcorpfin.2022.102320.Suche in Google Scholar

Gürtler, O., and J. Münster. 2010. “Sabotage in Dynamic Tournaments.” Journal of Mathematical Economics 46: 179–90. https://doi.org/10.1016/j.jmateco.2009.11.003.Suche in Google Scholar

Harbring, C., and B. Irlenbusch. 2011. “Sabotage in Tournaments: Evidence from a Laboratory Experiment.” Management Science 57 (4): 611–27. https://doi.org/10.1287/mnsc.1100.1296.Suche in Google Scholar

Hvide, H., and E. Kristian. 2003. “Risk-Taking in Selection Contests.” Games and Economic Behavior 42 (1): 172–9. https://doi.org/10.1016/S0899-8256(02)00538-9.Suche in Google Scholar

Ishida, J. 2012. “Dynamically Sabotage-Proof Tournaments.” Journal of Labor Economics 30 (3): 627–55. https://doi.org/10.1086/664945.Suche in Google Scholar

Kale, J. R., E. Reis, and A. Venkateswaran. 2009. “Rank-Order Tournaments and Incentive Alignment: The Effect on Firm Performance.” The Journal of Finance 64 (3): 1479–512. https://doi.org/10.1111/j.1540-6261.2009.01470.x.Suche in Google Scholar

Kini, O., and R. Williams. 2012. “Tournament Incentives, Firm Risk, and Corporate Policies.” Journal of Financial Economics 103 (2): 350–76. https://doi.org/10.1016/j.jfineco.2011.09.005.Suche in Google Scholar

Klunover, D. 2023. “Punishment for Sabotage in Dynamic Tournaments.” Journal of Mathematical Economics 106: 102841. https://doi.org/10.1016/j.jmateco.2023.10284.Suche in Google Scholar

Lazear, E. P. 1989. “Pay Equality and Industrial Politics.” Journal of Political Economy 97 (3): 561–80. https://doi.org/10.1086/261616.Suche in Google Scholar

March, C., and M. Sahm. 2018. “Contests as Selection Mechanisms: The Impact of Risk Aversion.” Journal of Economic Behavior & Organization 150: 114–31. https://doi.org/10.1016/j.jebo.2018.03.020.Suche in Google Scholar

Minchuk, Y., B. Keren, and Y. Hadad. 2018. “Sabotaging in Contests with Monitoring Efforts.” Managerial and Decision Economics 39 (6): 674–81. https://doi.org/10.1002/mde.2937.Suche in Google Scholar

Moldovanu, B., and A. Sela. 2001. “The Optimal Allocation of Prizes in Contests.” The American Economic Review 91 (3): 542–58. https://doi.org/10.1257/aer.91.3.542.Suche in Google Scholar

Münster, J. 2007. “Selection Tournaments, Sabotage, and Participation.” Journal of Economics and Management Strategy 16 (4): 943–70. https://doi.org/10.1111/j.1530-9134.2007.00163.x.Suche in Google Scholar

O’Keeffe, M., W. K. Viscusi, and R. J. Zeckhauser. 1984. “Economic Contests: Comparative Reward Schemes.” Journal of Labor Economics 2 (1): 27–56. https://doi.org/10.1086/298022.Suche in Google Scholar

Polinsky, A., and S. Shavell. 2000. “The Economic Theory of Public Enforcement of Law.” Journal of Economic Literature 38 (1): 45–76. https://doi.org/10.1257/jel.38.1.45.Suche in Google Scholar

Sisak, D. 2009. “Multiple-Prize Contests – The Optimal Allocation of Prizes.” Journal of Economic Surveys 23 (1): 82–114. https://doi.org/10.1111/j.1467-6419.2008.00557.x.Suche in Google Scholar

Stracke, R., W. Höchtl, R. Kerschbamer, and U. Sunde. 2014. “Optimal Prizes in Dynamic Elimination Contests: Theory and Experimental Evidence.” Journal of Economic Behavior & Organization 102 (1): 43–58. https://doi.org/10.1016/j.jebo.2014.02.018.Suche in Google Scholar

Szymanski, S., and T. M. Valletti. 2005. “Incentive Effects of Second Prizes.” European Journal of Political Economy 21: 467–81. https://doi.org/10.1016/j.ejpoleco.2004.07.002.Suche in Google Scholar

Tsoulouhas, T., C. Knoeber, and A. Agrawal. 2007. “Contests to Become CEO: Incentives, Selection, and Handicaps.” Economic Theory 30 (2): 195–221. https://doi.org/10.1007/s00199-005-0060-8.Suche in Google Scholar

Yumoto, Y. 2003. “Who is Target of Sabotage? The Dark Side of Promotion Tournaments”. Mimeo. https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=9175c1f7620804f7fcd8393d1f5751cdb654a5b8.Suche in Google Scholar

Received: 2024-05-19

Accepted: 2024-11-13

Published Online: 2024-12-16

Sie haben derzeit keinen Zugang zu diesem Inhalt.

Artikel in diesem Heft

https://doi.org/10.1515/bejeap-2024-0172

Schlagwörter für diesen Artikel

prize design; tournament; sabotage; selection efficiency; corner equilibria