Incomplete contracts can be optimal

When you write a contract, you want to lay out what happens in all circumstances. The intuition that this is best comes from the idea that complete markets are ex-ante optimal for everyone. If some circumstances have been left out, then the contract has to be renegotiated, which may lead to issues for example if there is a hold-up problem or worse if lawyer and courts need to get involved. The uncertainty may also lead to insufficient effort by contracting parties. The accepted wisdom is thus that incomplete contracts are bad.

Well, not always. Maija Halonen-Akatwijuka and Oliver Hart argue that one may not want too complete contracts. While there is no denying that it may be costly to draw a complete contract, their argument is that any contingency that is specified may acts as a reference point that may make negotiating for unspecified contingencies more difficult. If I understand this right, complete contracts may still be optimal, but if you have to have an incomplete one, you better not make it too complete. This non-monotonicity stems from the fact that contracted contingencies act as reference points. If there are too many contingencies and a state of the world is realized that is not accounted for, the parties will disagree which contingency should be taken as reference (each party will take the one that suits its interests best). Renegotiation becomes then costly. Have fewer contracted contingencies, and this is less likely to happen.