I regularly learn that policymakers and economists apparently have completely different ideas about what is “obvious” when it comes to the field. For instance, it is both empirically and theoretically easy to find examples of competition raising price when competition is among differentiated goods with negatively correlated valuations. Assume constant marginal cost. If there is one product, the demand curve is relatively flat, and the monopolist prices relatively low in order to let MR=MC. With two products, where half the consumers prefer good 1 and half prefer good 2, the residual demand curves are steeper than in the monopoly case. Therefore, each firm sets price higher than in the monopoly case. Yongmin Chen has a 2006 RAND that derives the exact conditions for price increasing in duopoly vis-a-vis monopoly.
In this paper, Chen and Savage apply this logic to the effect of DSL internet entry in a market that previously has cable. High-value consumers prefer cable because of its speed. The theory suggests that high levels of preference diversity lead to higher likelihood of price increases when moving from monopoly to duopoly. Using panel data from fourteen states, proxying for preference diversity with the standard deviation of education level, they find this to be true, and that at high levels of preference diversity, prices do increase.
http://ideas.repec.org/p/net/wpaper/0713.html (Final WP; published version in July 2010 ReStat)