The n-person war of attrition (Q1119210)

A two-person war of attrition (WA) is a game in which contestants compete for a prize worth V \((V>0)\), and the one who is prepared to wait longer collects the prize; both contestants incur a cost equal to the length of time taken to resolve the contest. Symbolically, if E(x,y) denotes the amount gained by a contestant prepared to wait time x when the opponent is prepared to wait y, we have \[ E(x,y)=V-x\quad if\quad x>y,\quad and\quad E(x,y)=-x\quad if\quad x<y. \] Such a game has precisely one evolutionarily stable strategy (ESS), viz. to wait for a time x drawn at random from the exponential distribution with mean V, i.e. density \(V^{-1}\exp (-x/V)(x>0).\) The authors analyze several different models for the generalization of this WA to n-person conflicts. In Model A, n players compete for one reward, value V, and each player independently selects a time he is prepared to wait, in the hope of outlasting the (n-1) opponents. Once chosen, this time is fixed; as some players drop out, those remaining are not allowed to alter their initial ``bids''. If \(X_ i\) is the waiting time chosen by player i, the reward to player 1 (say) is \[ E(X_ 1;X_ 2,...,X_ n)=V-W_ 1\quad if\quad X_ 1>W_ 1 \] and \[ E(X_ 1;X_ 2,...,X_ n)=-X_ 1\quad if\quad X_ 1<W_ 1 \] where \(W_ 1=Max(X_ 2,...,X_ n)\). Theorem 1 says that this model has a unique ESS, when each player chooses a value \(X_ i\), independently, from the distribution functions \(G(x)=(1-\exp (-x/V))^{1/(n-1)}(x\geq 0).\) Other models, in which the players may alter their bids or in which several prizes are available, are also discussed and the ESS investigated (there may be none).