On Dynamic Contracts without Transfers

Project: Research

View graph of relations



Consider a designer who has a long-term project in which he must decide whether to assign up to one unit of good to one agent in every period. The agent receives a positive payoff from consuming the good in every period. However, the associated social benefit is identically and independently distributed across time and is privately known to the agent. The good is assigned based on the agent’s reports. In each period, the designer  can verify an agent’s private information at a cost and punish the agent by burning a fraction of her surplus. There are no monetary transfers.There are several important economic environments roughly corresponding to this model. For example, every year, a developing aid agency such as the World Bank must decide whether to allocate a grant to a local organization to serve some beneficiary groups. The local organization privately knows how the grant will benefit the intended groups. The development aid agency can evaluate the benefit by reviewing the recipient’s report and limiting the grant size it manages in the future. However, it cannot recover the funds allocated to the organization or disbar it from future grant applications out of humanitarian concerns. A funding agency must decide whether to give a grant to a researcher to support their new project every year or every few years. The researcher privately knows the intellectual merits of their projects. The funding agency can evaluate a project's impact by reviewing the researcher's final report and limiting the grant size they receive in the future. However, it cannot recover past grants or disbar the researcher from future grant applications.In this project, we study the design of optimal mechanisms in the environments described above and the properties of optimal mechanisms. These theoretical exercises help us understand how repeated interactions affect the characteristics of optimal mechanisms and provide guidance for the use of different mechanisms in practice.Several earlier papers have studied this problem in a static model in which the designer and the agent interact only once. In this project, instead of considering a static allocation problem, we consider a dynamic allocation problem, in which the designer and the agent interact repeatedly. As a first step of the analysis, we analyze two special cases: pure limited punishments and pure costly verification. More work is required to characterize the optimal mechanism in the general environment and to study their properties.


Project number9043441
Grant typeGRF
StatusNot started
Effective start/end date1/01/23 → …