Manipulation through bribes

We consider allocation rules that choose outcomes and transfers, based on a gents' reported valuations of the outcomes. A bribing situation exists when one agent could pay another to misreport his valuations, resulting in a ne t gain to both. A bribe-proof rule eliminates such opportunities. We show t hat under a bribe-proof rule, each agent's payoff is a continuous function of other agents' reported valuations. Furthermore, on connected domains, if the set of outcomes is finite or the domain is smoothly connected, each ag ent's payoff is a constant function of other agents' reports. Finally, unde r a domain-richness condition, a bribe-proof rule must be constant. The res ults apply to a broad class of economies. Journal of Economic Literature Cl assification Numbers: C70, D70. (C) 2000 Academic Press.