Countervailing incentives in multi-agent environments

We consider a principal who signs a centralized grand-contract with two risk-neutral and limitedly liable agents behaving non-cooperatively. Tech-nological features are privately known to each agent and give rise to coun-tervailing incentives to misrepresent costs (the types). We \u85rst assess that, with uncorrelated information, the principal can induce the full information outcome, as both Bayesian and dominant strategy equilibria, thanks to the presence of countervailing incentives. We then show that, with correlated information, yardstick competition helps implement the full information out-come more often in the Bayesian setting, while it does not under dominant strategies. Yet, in either case, it allows to contain distortions and rents when-ever the full information outcome is beyond reach. However, the bene\u85ts from benchmarking tail o ¤ as agentspockets become less deep and some of them eventually disappear