game, traders may be disinclined to default if their reputation damage prevents gains from future trades.
Without intermediaries, the dissemination of information on defaults may be limited and the reputation
sanctions correspondingly small. However, if trades are conducted through intermediaries (without contact
between buyer and seller) the intermediaries can act as a clearing house for information. Given frequent
interaction of each party with the broker, the sanction for default can be enhanced thereby lowering the
incentive to cheat. We also use intermediaries as a clearing house for information on performance and allow
the broker to impose a sanction for non performance. The difference lies mostly in objective. Whereas
Kingston is looking generally for an operative mechanism to promote trust when contracts cannot be enforced,
we are focused on the insurance market. Thus, we look in detail at the structure of the partially complete
contracts and market relationships that permit transfers to be made for non verifiable losses. Moreover,
instead of assuming a “nuclear” trigger for sanctions, we look at how information is updated and how brokers
are compensated to see whether threatened sanctions are indeed credible.
3 “Insuring” Non-Verifiable Losses without Brokers
In this section, we consider a simple world with a risk-averse individual and multiple, risk-neutral insurers.
The individual is endowed with initial wealth w0 and exposed to a loss, L, with probability p.Theloss
may be verifiable (and therefore insurable under a conventional policy) with a probability q or non-verifiable
but ex-post insurable with a probability 1 - q. The non-verifiable loss is observed by both the individual
and the insurer, but cannot be contracted upon. For example, this may be a loss that could not have
been anticipated at the time the contract was written but, once it has occurred, it is clearly observed by all
parties.10
Now consider the following infinite period problem. The individual buys a contract from a competitive
insurer to cover an amount c of the verifiable loss (probability pq). However, the premium exceeds the
actuarial value of the verifiable loss, P>pqc. Now suppose that the insurer expects that the policyholder
will renew this coverage indefinitely. With these renewal expectations, the insurer would make rent having
a present value of (P - pqc)/r where r is the discount rate.
The expected rent only will be realized if the policyholder does renew the contract. This “hoped-for”
10 For simplicity, we will assume that all non-verifiable losses are ex-post insurable. The addition of ex-post uninsurable risks
to our model would create a background risk.
10