coverage cannot be achieved as either the verification cost to the insurer or the falsification cost to the
policyholder is embodied in the equilibrium contract. In both approaches, it is assumed that the transfer
schedule is enforceable in front of a court. In this paper, we investigate situations with an even higher degree
of “incompleteness”, namely events that do not allow for any transfer schedules to be enforceable ex-post by
a court. The insurer has no contractual obligation to indemnify the policyholder ex-post. The mechanism
we present in this paper is based on the hold-up power the policyholder derives from future rents that the
insurer is able to extract from an ongoing relationship. As those rents are costly, i.e. the policyholder has
to pay an implicit loading, full coverage is precluded by this mechanism. The broker’s ex-post judgement
and leverage influences the size of hold-up and so affects the transfers from the insurer to the policyholder.
The role of the broker in our model in some ways resembles the role of the courts in Anderlini et al (2003
a,b).9 Both courts and brokers must exercise judgment in deciding whether an unanticipated loss should be
covered under the contract and in both cases they are guided by the efficiency gains this “precedent” implies
for future contracts. In Anderlini et al, the court’s desire for efficiency gain is implied by its adoption of
a social welfare function. In our model, brokers seek future efficiency gain because they can capture rents
directly from value added. However, there are differences. In our case, there is a modeled efficiency gain
in the current period. The parties can choose to contract with one or another of a number of competing
brokers. In making this choice, the degree of hold up, and therefore the terms of ex post bargaining, can
be bounded by the parties. The second difference lies in the nature and expertise of the institutions. The
nature of the ex post judgment (whether a revealed loss is ex post insurable) is technically quite complex.
Courts, by their nature, have no core skills to address such problems and rely on the expertise imported
by the parties in an adversary process. Brokers do have these core skills and derive rents directly from
the creation of value. But perhaps the biggest difference between our model and theirs lies in its purpose.
Whereas they address unanticipated events that frustrate the purpose or outcome of a contract (i.e., they
are an unintended nuisance), we address a situation where the parties know unanticipated events can happen
and we are trying to expand the domain of the contracting relationship to encompass such events (i.e., to
provide insurance). This is why, for us, it is important to address efficiency gains in the current period.
Perhaps, most closely related is the contemporaneous paper of Kingston (2005). He considers an economy
where people can defect on trades. In a single period, this is a prisoner’s dilemma. However, in a repeated
9The papers of Anderlini et al. and ours both build on the incomplete contract literature (see Grossman and Hart, 1986,
Hart and Moore, 1990, Hart, 1995).