Regularity of pure strategy equilibrium points in a class of bargaining games (Q2494015)

The paper is concerned with \(n\)-player, discounted sequential bargaining games with multidimensional agreement spaces, in particular with determinacy and continuity of their equilibrium correspondences in the parameter space. Lower hemicontinuity is as usual the harder to obtain, and the author, under a strict-concavity or a private-values technical assumption, arrives at the main result for the class of games whose voting rule (determining acceptance/rejection of the stage-proposal advanced by a randomly chosen player) satisfies the oligarchy property that the intersection of all acceptance-determining -- `winning' -- coalitions is itself winning. The class includes the unanimity rule, but not the simple-majority rule. The parameter space is taken to be that of the players' discount factors, \((0,1)^n\); equilibria studied are the Pure Stationary Subgame Perfect ones; and the results of finiteness and lower hemicontinuity of the equilibrium map holds generically in the discount-vectors space. Analytically, the main step consists of devising a map, rooted in the agents' optimization problems, which when single-valued (and to achieve this come in the technical assumptions) allows one to express an equilibrium as a solution to a finite set of equalities and inequalities. Next step is to get rid of inequalites to use the degree theory developed by Christine Shannon; and for this the oligarchy property above is used. Then the author proves that the resulting system of equations satisfies the desired properties exploiting homotopy invariance, under regularity assumptions. To complete the argument it is shown that the aforementioned technical assumption is sufficient to ensure that the regularity of the equation systems involved is in fact generic. By the same analysis the author shows finiteness and local uniqueness for games with an unidimensional agreement space.