On the Choice of Insurance Distribution Systems

Abstract

INTRODUCTION The insurance industry employs a variety of distribution systems: insurance contracts are sold through direct writers, exclusive agents, independent agents, and brokers. In the direct writer system, the sales agent is an employee of the insurance firm. An exclusive agent also represents a single insurer, yet is not technically the firm's employee. An independent agent represents more than one insurance company. Finally, a broker represents the customer and negotiates with multiple insurers.(1) In this article, we examine the insurer's choice of distribution system. We identify complimentarities among corporate policies that arise from particular problems within contracting relationships among the insurer, the agent, and the insured. Much of the literature on distribution systems in the insurance industry focuses just on contracting problems between the insurer and the selling agent (for examples, see Marvel, 1982; Grossman and Hart, 1986; and Sass and Gisser, 1989). While including these contracting problems, our analysis also incorporates the contracting relationships involving policyholders. Milgrom and Roberts (1995) examine complementarities among inputs to explain corporate choices of organizational structure, technology, and strategy. The standard definition of complementarity in economics states that two inputs to a production process are complements if a decrease in the price of one causes an increase in the use of the other. But Milgrom and Roberts use this term not just in its traditional sense of a relation between pairs of inputs, but also in a broader sense as a relation among groups of activities. They introduce a broader definition: several activities are strategic complements if doing more of one activity increases the marginal profitability of each of the other activities. If the activities can be expressed as differentiable functions, this corresponds to positive mixed partial derivatives of the payoff function - the marginal returns to one activity are increasing in the levels of other activities. Yet their analysis emphasizes that continuity, differentiability, and convexity of the payoff functions are not necessary - only an ability to order the various activities is required. This framework is particularly useful here, where we want to examine various ownership structures, like stocks and mutuals, as well as different distribution systems, like independent and exclusive agents. The key idea in the Milgrom and Roberts analysis is that, if choosing a common stock ownership structure (rather than a mutual) changes the payoffs so that the returns from using independent agents rise, then ownership structure and distribution systems are strategic complements. We argue that the use of independent agents better bonds the insurer's promise to provide services to the policyholder and helps control potential expropriative behavior by the insurer; thus, the independent agency system is more valuable for ownership structures where these incentive problems are more severe. Under those circumstances, we should observe a correlation between the choices of ownership structure and distribution systems across firms in the insurance industry. Our tests employ data from a large sample of property-liability insurance firms.(2) We examine differences between independent agency and exclusive agency insurers. Our evidence documents strong associations between the choice of distribution system and both the insurance company's ownership structure and its lines of business. Our analysis also provides support for several additional hypotheses including the Grossman and Hart (1986) and Marvel (1982) hypotheses that link advertising policy and the choice of distribution system. Finally, we document a richer interaction between distribution systems and product mix than suggested by Marvel (1982). In the next section, we review theories on the choice of distribution system and identify testable implications. …