Discussion

Bill Murray of Chubb suggested a closer examination of the role of the broker. In reality brokers, rather than acting disinterestedly, generally act on different motivations. In many cases, they receive compensation from the insurers to whom they present business and with whom they sometimes have profit-sharing arrangements. The status of a broker's compensation may partly influence where the broker chooses to place business. The paper seems to suggest that the insured should always act through a large national broker, who can provide greater clout in the market and have a greater effect on the insurer's reputational interest; in reality, however, regional brokers thrive. Moderate or midsize companies may receive greater attention from regional brokers. Murray was impressed to see a mathematical justification for reputation, something his corporation values but has difficulty quantifying in self-evaluations.

Richard Zeckhauser of Harvard University highlighted the role of the broker as an arbitrator and reputation spreader. Market participants probably want an outcome that they would consider "fair" or "anticipated," rather than the Nash equilibrium solution, which disadvantages the risk averse. He suggested modifying the model to give the insured analytic ability to distinguish by reputation the "fair" insurer from the hard bargainer. The Better Business Bureau and eBay are two examples of the demand for reputation spreading. Mike O'Malley of Chubb agreed, adding that a broker concerned with reputation constitutes a three-party game. In reality brokers do not always act as impartial arbiters. Thanking Zeckhauser for his suggestion, Neil Doherty expressed interest in modifying the model to allow firms to develop reputation by volunteering a settlement beyond the Nash bargaining solution. Developing the model into a three-way bargaining game seems necessary to develop the study further. Although legally it maybe difficult to establish that brokers act as agents for the insurer, receiving compensation from the insurer certainly points to an agency relationship with the insurance corporation.

Richard Darrig of the Automobile Insurers' Bureau suggested that fairly priced future contracts ought to have future profits set to zero. The loss then should not be forgone profits, but sunk costs in capital or expenses. He also suggested that reputation can be thought of as franchise value, which would allow a firm to command higher prices in the market. Finally, rather than setting β_n = 0 in failed bargaining, some negative value representing an unpleasant outcome for one side or the other should be reflected in the equation. Doherty responded that zero profit makes sense in a competitive market without an intertemporal aspect to product pricing. Making relationship-specific investments that are recouped in the future distributes profit over time and provides some leverage in bargaining.

Howard Kunreuther of the Wharton School pointed out that insurance firms rely on deductibles to screen out unverifiable claims that may be the result of moral hazard. The importance of the deductible warrants further investigation, and verifiability must be defined more clearly. The challenge of verification may be in the severity of the loss or in whether the cause of the loss was an event covered by the policy. Also a client may evaluate the decision between firms of differing reputation depending on the client's own opportunity cost of litigation. Doherty clarified that "verifiable" in the context of this paper indicates not that loss actually occurred and is quantifiable, but that the event causing the loss is specified in the contract. An event unspecified in the contract constitutes a non-verifiable claim.

Robert Litan of Brookings expressed his belief in the paper's relevance to large commercial insurance. Although individuals may have recourse to a regulator, they hardly have bargaining power. Knowing what percentage of claims are litigated or arbitrated rather than just unilaterally declared would give a greater sense of the magnitude of the issue. There may be more to the choice of broker than the reputation staked: the degree of choice, for example. Finally, considering that most settlements are the result of negotiation rather than arbitration, Nash modeling probably provides the best real-world approximation.

Responding to comments, Doherty continued. Having an expectation of future bargaining would certainly increase premiums, but beyond that it may provide a means...