A Dynamic Model of the Transmission of Price Information in Auto Insurance Markets Abstract A model of the demand structure of auto insurance markets is constructed in which consumers learn about firms through friends and neighbors. Consumers firms when the price differential exceeds a certain threshold, the switch cost. Data from New York and Alberta auto insurance markets are used to measure costs and the effective speed of transmission of information in these markets. Introduction A dominant theme in the literature exploring the implications of imperfect knowledge on the part of market participants has been that the way in which information is transmitted among and processed by agents is a critical determinant of market dynamics and outcomes [Rothschild (1973), Holmstrom (1984), Rogerson (1983), Shapiro (1982)]. In discussing the consumer's problem of searching for the lowest price in markets with price dispersion, Rothschild (1974) noted that, although demand functions are determined by the rules which searchers follow, very little empirical work had been done to determine how consumers actually gather information in these markets. In this article, a model of the demand structure of auto insurance markets is constructed which is consistent with the observed process by which consumers gather price information in those markets. The model is static on the firm side with no entry or exit. At period t = 0 a given group of firms sets prices which do not vary over the remaining periods. Consumers gradually learn about these prices. Proceeding in this manner allows modelling in detail the dynamics of the demand structure, generating the market shares of all firms in each time period. The focus of the literature on price dispersion in markets has been on equilibrium outcome [Carlson and McAfee (1983), Pratt, Wise and Zeckhauser (1979)] which in real-world markets may be reached only after many years, or not at all, given the random events and structural changes in the institutional environment. Constructing a dynamic model of the demand structure allows measurement of the effective speed of transmission of information in the markets for auto insurance in Alberta from 1976 through 1980 and in New York State from 1977 through 1983. The result is a model in which markets slowly tend to an equilibrium as agents learn about the structure of the marketplace, rather than one in which markets are always in equilibrium, instantly adjusting to exogenous shocks. The focus is on only the demand side of the market. Equilibrium search models which illustrate the behavior of firms rely on simplifying assumptions regarding the demand structure. In this model the complexity of the demand structure precludes the use of analytical techniques on the firm side. Dahlby and West (1986) have recently used an extensive set of auto insurance data from Alberta to test Carlson and McAfee's equilibrium price dispersion model which makes the following assumptions: (1) consumers know the distribution of prices, (2) all firms are equally likely to be sampled in the search process, and (3) the market is in equilibrium. In the model presented in this article, on the other hand, consumers are not assumed to know the distribution of prices. More importantly, a key feature which is not present in other search models is that a firm with larger market share is more likely to be sampled in the search process because it will have more consumers disseminating information about the firm, and will generally have more marketing outlets facilitating contact with prospective consumers. The model is fitted to data from New York State and the Dahlby and West study, and the results are compared. The empirical evidence on consumer information acquisition patterns and the existence of price dispersion in auto insurance markets are first discussed. Then the model of consumer demand is developed and empirically tested, followed by a summary. …