مطالعه آزمایشگاهی تبلیغات و رقابت قیمت

We use a laboratory experiment to study advertising and pricing behavior in a market where consumers differ in price sensitivity. Equilibrium in this market entails variation in the number of firms advertising and price dispersion in advertised prices. We vary the cost to advertise as well as varying the number of competing firms. Theory predicts that advertising costs act as a facilitating device: higher costs increase firm profits at the expense of consumers. We find that higher advertising costs decrease demand for advertising and raise advertised prices, as predicted. Further, this comes at the expense of consumers. However, advertising strategies are more aggressive than theory predicts with the result that firm profits do not increase.

In contrast to the classical theory of perfectly competitive markets, many markets are characterized by imperfect price information. Consumers often do not know (or perhaps do not care solely about) the prices charged by all sellers in a market, and buyers and sellers must often incur costs to discover or transmit this information. A now well-established literature analyzes how various search frictions can generate imperfect information, and hence affect market performance, and this literature shows that many of the properties of perfectly competitive markets do not carry over to markets with these characteristics. For example, in a market where some consumers are better informed than others about what prices are available, the “Law of One Price” may not hold. That is, in equilibrium, different sellers may charge different prices for a homogeneous product. Moreover, changes in the underlying structure of such markets often have implications that strikingly differ from the perfectly competitive case. For example, Varian's “Model of Sales” (Varian, 1980) analyzes price competition among identical sellers supplying a homogeneous product. Demand for the product comes from two types of consumers. Informed consumers know the prices charged by different sellers and buy at the lowest price (as long as this does not exceed their reservation price). Uninformed consumers do not know what prices are available, and simply choose a seller at random and buy from this seller (again, supposing that the seller's price does not exceed their reservation price). In this model, the unique symmetric equilibrium involves price dispersion as sellers use mixed strategies to generate prices. More broadly, one can view Varian's model as one in which consumers differ in their price sensitivity. Under this view, when consumers are heterogeneous in their price sensitivities, price dispersion is predicted to be the inevitable outcome.

In markets where sellers have to incur costs in order to inform buyers of prices, these costs can have important consequences for sellers’ market power. Our experiment is based on a stylized model of price advertising in a homogeneous product market and, although the model is simple, it nevertheless generates a complicated equilibrium in which sellers randomize in both advertising and pricing decisions. Our experiment provides mixed support for the performance of mixed strategy predictions in this setting. On the one hand, as in previous experiments with (simpler) games involving mixed strategy equilibria, subjects’ decisions do not conform exactly to those predicted by equilibrium. Moreover, we observe some systematic departures from equilibrium play. Most notably, subjects advertise much too frequently in three of our four treatments. As a result, in these three treatments, consumers pay lower prices than predicted, and profits are lower than predicted. On the other hand, the theory is very successful at predicting how subjects’ decisions change as the underlying market environment changes. As advertising fees and numbers of competitors are manipulated, subjects’ decisions change systematically in the direction predicted by comparative static analyses based on the mixed strategy equilibrium. Likewise, the interaction effects of changes in pricing and advertising generate the predicted impact on consumers—they are hurt by higher advertising fees. In contrast, the theoretical prediction that higher advertising fees lead to higher profits is not supported by our data. As fees increase advertising expenditures go up more than predicted, because sellers do not reduce their advertising propensities enough. Although there is an associated change in pricing behavior that allows sellers to extract more revenue from consumers, this does not outweigh the higher advertising expenditures. Thus sellers fail to exploit this facilitating device.