یک مطالعه تجربی از پراکندگی قیمت در مدل جستجوی بهینه با تبلیغات

This paper reports a laboratory experiment to study pricing and advertising behavior with costly buyer search. Sellers simultaneously post prices and may incur a cost to advertise this price. In the unique symmetric equilibrium, sellers either charge a high unadvertised price or randomize in an interval of lower advertised prices. Increases in search or advertising costs raise equilibrium prices, and equilibrium advertising intensity decreases with lower search costs and higher advertising costs. Our experimental results support these comparative static predictions, and sellers also post high unadvertised prices as predicted. In all treatments, however, sellers advertise more intensely than in equilibrium.

Costly information acquisition has remained an important topic of economic research since Stigler (1961). One principal goal of this literature has been to explain nontrivial wage and price dispersion as a stable equilibrium outcome when one or both sides of the market have imperfect information. Economists have developed numerous versions of equilibrium search models citing the role of search frictions (Reinganum, 1979), information asymmetries (Varian, 1980), differing production costs (Salop, 1973), differing consumer search costs (Stahl, 1989 and Stahl, 1996) and ex post consumer heterogeneity (Burdett and Judd, 1983) and other reasons for the failure of the “law of one price”. Although theorists established the relationship between search, incomplete information and price distribution many years ago, these models are gaining renewed significance as search costs and technologies are being transformed by emerging Internet-based markets. Search models can provide an equilibrium explanation for the persistent price dispersion observed in essentially homogenous product markets (Baye and Morgan, 2001, Brynjolfsson and Smith, 2000 and García-Gallego et al., 2004).1 Search models can also provide guidance to help answer important public policy questions, both for formulating the basic principles of the policy and for allocating resources efficiently to implement them. For instance, various consumer protection regulations stipulated by the Federal Trade Commission are aimed towards curbing the harms of imperfect information.

In this study we analyze the effect of two types of informational costs on pricing and advertising behavior in posted offer markets. The experiment is based on a variant of Robert and Stahl's (1993) influential model of seller advertising with costly buyer search. The laboratory data provide almost uniformly positive support for the theory. The data support all of the model's comparative static predictions for prices, profits, and advertising rates, and the only major systematic deviation from the quantitative prediction is that observed advertising rates exceed the predicted rates. Prices are also a bit lower than predicted. The distinguishing feature of our environment is that the information can both be disseminated by the sellers and acquired by the buyers. This is more realistic than the more simplified posted offer market often studied in the laboratory, and it allows us to draw important parallels between our results and behavior observed in the field. For instance, the aggressive advertising strategies adopted by sellers in our experiment are reminiscent of many ‘real world’ price advertising campaigns adopted by firms in markets in which price information could also be gathered by buyers. Our study suggests that this type of aggressive advertising may in fact be deeply rooted in strategic profit considerations, and the overly aggressive advertising does not cause the dispersed prices to differ significantly from the model's predictions.