Efficiency in hierarchies: implementing the first-best solution by sequential actions

A model of a firm with one principal and two agents is considered. The actions of the agents are not observed by the principal; hence there is moral hazard. It is shown that when the agents act sequentially-that is, agent 2 takes his action conditional upon agent I 's choicethe first-best actions can be implemented by a suitable choice of the incentive schemes given to the two agents. The result relies both on the sequential nature of the game and on the particular specification of the effect that the agents' actions have on output. Several extensions and applications of the framework are suggested. In particular, it is shown that the analysis generalizes to the case of more than two agents.