公平与效率外文文献翻译中英文.docx

1、公平与效率外文文献翻译中英文公平和效率中英文2019英文Fair share and social efficiency: A mechanism in which peers decide on the payoff divisionRod Falvey，Shravan LuckrazAbstractWe propose and experimentally test a mechanism for a class of principal-agent problems in which agents can observe each others efforts. In this mech

2、anism each player costlessly assigns a share of the pie to each of the other players, after observing their contributions, and the final distribution is determined by these assignments. We show that efficiency can be achieved under this simple mechanism and, in a controlled laboratory experiment, we

3、 find that players reward others based on relative contributions in most cases and that the players contributions improve substantially and almost immediately with 80 percent of players contributing fully.Keywords：Experimental economics，Fairness，Mechanism designIt can be difficult for a principal to

4、 observe individual agents effort levels, particularly when agents work in teams. The extensive monitoring that would be required may not be feasible or cost-effective. Profit sharing has been suggested as a response (Weitzman and Kruse, 1990), since giving each of the agents a stake in an enterpris

5、es profits does provide a link between agent contribution and agent reward that is missing from a fixed wage or salary structure. But under an equal sharing allocation, which is the natural allocation for a principal to impose when she cannot observe individual agents behaviour, afree-rider problema

6、rises since each agent bears the full cost of their contribution but only reaps1Nth of the benefit in anN-agent team. Unless the costs of contribution are low or the interdependencies between agent productivities in team production are high (Heywood and Jirjahn, 2009), agents will not contribute the

7、ir social optimal contribution under an equal-sharing regime. If rewards are not related to contribution, an agent who feels under-compensated may end up reducing her contribution.Although the principal may be unable to observe agents contributions, there will be occasions where the agents themselve

8、s are in a position to observe each others actions.1The challenge then for the principal is to design a mechanism that elicits and uses this information to induce the appropriate levels of contribution from the agents. In this context, we consider a simple mechanism in which agents are not only able

9、 to monitor each other, but also in positions to determine each others payoffs. The mechanism we propose takes the form of a two-stage game. In the first stage, each player chooses some contribution level and in the second stage, after having observed each others contributions, each player proposes

10、a fraction of the total surplus to be received by each of the remaining players. A players final share depends on the other players allocation toward her.We label our mechanism the “Galbraith Mechanism” (GM hereafter) as the idea is inspired by John Kenneth Galbraith who, in an aside inThe Great Cra

11、sh 1929, described a bonus sharing scheme used by the National City Bank (now Citibank) in the U.S. in the 1920s. Under this scheme each officer would sign a ballot giving an estimated share of the bonus pool towards each of the other eligible officers, himself/herself excluded. The average of these

12、 shares would then guide the final allocation of the bonus to each of the officers (Galbraith, 1963, p. 171). This sharing mechanism can be applied to many economic problems including games with positive externalities and principal-agent problems in which the principal needs to distribute some commo

13、n resource amongst the agents. The crucial feature of the GM is that how a player allocates shares in the second stage does not affect her own payoff. Therefore, players are able to reward or punish their peers based on the first stage observed actions. A number of studies have demonstrated that pla

14、yers exhibitsocial preferencesto “punish” those who free-ride on the group production (Fehr and Gchter, 2000) and to “reward” those who contribute more than the group average (Sefton et al., 2007; Nosenzo and Sefton,2012). While such social preferences move the outcome towards social efficiency,self

15、-interesttends to restrict their application and the social costs that these punishments and rewards impose on all parties involved tend to limit their ultimate success (seeChaudhuri (2010)for a review).3The GM is based on an endogenous payoff allocation in which players can freely decide on some fr

16、action of the co-players payoffs. Players are free to punish, to reward, to allocate equally or even to allocate randomly to the remaining players, while no costs are incurred by any players in the allocation exercise.The downside of allowing players the freedom to reward and punish in this way is a

17、 potential for multiplicity ofNash Equilibriaat the second stage of the GM game. Without more structure, every allocation is a Nash Equilibrium in the second stage. One method of removing the resulting arbitrariness is by explicitly incorporating a behavioural component into the payoff function, whi

18、ch could be seen as reflecting the players subjective notion of a “fair” allocation.4But rather than imposing a solution in this way, we leave the question of how the players actually allocate to be uncovered in the experiments that follow. That said, our investigation of equilibria in this contribu

19、tion game does reveal a link between efficiency and “fairness”. Much of the theoretical literature on fairness focuses on equality and equal share (e.g.,Fehr and Schmidt, 1999) regardless of contributions. But a growing empirical literature appeals to other fairness criteria to justify unequal alloc

20、ations, e.g.,Adams (1965);Konow(1996;2000;2009) ;Gchter and Riedl (2006);Cappelen et al. (2007);Shaw (2013);Cappelen et al. (2013).5Prominent here is the notion of distributive justice first explored by sociologists (Homans, 1958;Adams, 1965) and later adopted by behavioural economists (Selten, 1978

21、).Distributive justice is often defined by the principle that a players entitlement towards some group outcome should be proportional to her contribution to that outcome.6In the next section, we establish that such “fair” allocation behaviour can support efficiency (full contributions) as part of a

22、SPNE of this contribution game, for all positive returns to total contributions. It is a pivotal case in that other more pro-contribution biased allocations (i.e. allocations that give disproportionately larger allocation shares to those with higher contribution shares) also support the efficient eq

23、uilibrium under the same conditions, but anti-contribution biased allocations (i.e. allocations that give disproportionately larger allocation shares to those with lower contribution shares), such as equal shares, only support the efficient equilibrium at higher returns to total contributions. Propo

24、rtional allocations feature prominently in our experimental results.The GM is also “simpler” than other endogenous mechanisms proposed to solve social dilemma problems. For example,Andreoni and Varian (1999)studied a mechanism where players can agree on a pre-play contract before playing theprisoner

25、s dilemmagame. However, their mechanism does not perform well when tested in laboratory settings (Hamaguchi et al., 2003;Bracht et al., 2008) . While there are other mechanisms that perform better in the laboratory, for example,Falkinger et al. (2000);Masuda et al. (2014)andStoddard et al. (2014), t

26、hey either add an enforcement institution, or “impose” an informed third party to allocate the shares. In the context of an uninformed principal and informed agents, the GM is a method of determining an informed allocation for the principal to make which requires only that the principal collate the

27、allocation shares proposed by the players and distribute accordingly.7Provided the players are inclined to reward contribution in the second stage, and they anticipate this happening at the first stage, the GM should yield outcomes closer to social efficiency than an equal shares mechanism.8Perhaps

28、the model closest to ours isBaranski (2016)who considers a class of voluntary contribution mechanism in which the players shares of the group fund are determined using aBaron and Ferejohn (1989)multilateral bargaining procedure in which each player can be randomly chosen to be a proposer at a certai

29、n period. In each period, after a division is proposed, the remaining players can vote to agree or disagree with the proposal. The bargaining process ends if a majority agree with the proposer and the fund is divided as per the proposal. The most important distinction between this mechanism and the

30、GM lies in the allocation stage which takes the form of a one-shot game in the GM, whereas in Baranskis mechanism, it is given by a multi-stage bargaining process. Furthermore, the GM does not require a randomly chosen proposer as each player simply proposes a share for his peers. In terms of experi

31、ment results, both mechanisms enjoy a substantial increase in the contribution levels once they are introduced.9In summary, the main contribution of this paper is to propose a simple mechanism and to test it experimentally. As noted, this mechanism: (i) allows costless reward and punishment at the a

32、llocation stage; (ii) removes the bias arising when a player proposes an allocation to himself; and (iii) avoids the necessity of imposing an informed allocator or randomly selecting one as part of a bargaining process. Only low returns to scale, where equal shares would not automatically generate f

33、ull contributions, are considered. Provided players reward others based on their contributions at the allocation stage, and anticipate such rewards at the contribution stage, we expect that social efficiency can be achieved under this mechanism. This allocative behaviour does seem to be prevalent in