Promises in contract design

- We propose a new channel of contract design to boost efficiency,
- which we test experimentally.
- We vary who proposes the contract and whether it includes a
- specification of the worker's effort.
- Highest efficiency when the worker proposes a contract, including the worker's promised effort.

In this paper, we propose a new channel of contract design to boost efficiency. If deviating from one's own words induces a self-imposed moral burden, the optimal contracting procedure with regard to cheap talk shall assign the responsibility for installing the nonbinding promise in the contract to whoever has the residual right to break such promise, in the spirit of Grossman and Hart (1986) and Hart and Moore (1990). To study whether a worker's own promise of effort level governs his real choice of effort in a gift exchange game, we implement four treatments in our experiments by varying two factors: (1) who (the firm or the worker) takes the position of the proposer to propose the contract and (2) whether the proposed contract includes a nonbinding specification of the worker's effort level. Our key finding is that when it is the worker who proposes the contract and the contract includes the worker's promised effort level, both the worker's actual effort choice and the aggregate profits are significantly higher than in each of the other three treatments (and there is little difference in worker effort otherwise).