Effect of search cost in the presence of search deterring informative advertising

A recent article published in The Wall Street Journal reports that hospitals prevent search for product and price related information. As a result, consumers have to engage into multiple clicks to get information. In response, the Biden administration is proposing to sharply increase penalties in an attempt to clamp down on such practices (https://​www.​wsj.​com/​articles/​u-s-proposes-upping-penalty-for-hospitals-that-dont-publish-prices-11626725949).

Similar informative advertising content is used by car insurance providers such as Progressive, whose campaigns provide specific information comparing their coverage with that of their competitors.

In line with the larger search literature, firms do not know the exact \(\nu _{i}\). In other words, we do not consider the possibility of perfect consumer addressability, and firms cannot target individual consumers with targeted pricing or advertising. The assumption is also consistent with the motivating examples of Miller Lite and Burger King described in the Introduction, where firms send advertising content that is informative but not individually addressed.

Akin to Janssen and Non (2008), advertising and consumer search are substitutes in this paper. In other words, advertising is informative because it fully and truthfully reveals the match-related information, obviating the need for a consumer to engage in the search. However, note that this formulation is qualitatively similar to a model where advertising and search and imperfect substitutes. To see this, consider the model where advertising is truthful with probability \(\beta \) and is noise with probability \(1-\beta \) (Wang, 2017). Given the advertising intensity of firm i, \(\alpha _{i}\), the probability that a consumer is not aware of the true match value and therefore has to engage in a search (to find her fit with product i even after being exposed to advertising from firm i) is \(\alpha _{i}(1-\beta )+(1-\alpha _{i})=1-\alpha _{i}\beta \). In other words, assuming that advertising could be pure noise with probability \(\beta \) changes the ex-ante probability of search to \(1-\hat{\alpha _{i}}\), where \(\hat{\alpha _{i}}\equiv \alpha _{i}\beta \), from our current formulation where the ex-ante probability of engaging in search is \(1-\alpha _{i}\). Thus, the altered model offers qualitative results similar to the model considered.

Precisely for the same reason, firm 1 expects zero demand from subsegment LH, i.e. from consumers with \(\nu _1\le \hat{\nu }-\Delta \) and \(\nu _2>\hat{\nu }\). Consumers in this subsegment, irrespective of which firm they sample first, will buy from firm 2.

We omit the trivial result that advertising effort decreases with advertising cost, which follows directly from the fact that the cost of advertising is convex.

Note that, since a and c are the only two parameters in our model, the equilibrium price dynamics are completely characterized in general by Fig. 3A. Similarly, Figs. 3B and 4 completely characterize the profit and consumer surplus results, respectively.

In Grossman and Shapiro (1984), an increase in the cost of informative advertising reduces the margin, and this margin-reducing effect tends to dominate the positive impact of a reduction in advertising competition derived from less head-to-head competition between the firms.

Note that the profit response to c is in contrast to Janssen and Non (2008), who find (in the context of uncertain product availability) that in the equilibrium where firms advertise, an increase in search cost either monotonically decreases profit or has no impact on profit.