OPTIMAL PRICING WITH POSITIVE NETWORK EFFECTS: THE BIG BENEFITS OF JUST A LITTLE DISCRIMINATION

摘要: We study the revenue-optimal pricing strategies of a monopolist selling divisible good (service) that exhibits positive externalities to consumers embedded in social network. The our model mean consumer's usage level depends directly on his neighbors Optimal may therefore involve offering different prices based consumers' position. first consider setting where can offer individualized and derive an explicit characterization optimal price for each consumer as function network show such policy amounts solution quadratic program it is seller charge proportional measure importance called Bonacich centrality. next constrained whereby choose no more than k distinct prices. While problem tractable uniform pricing, i.e. k=1, we generally NP-complete there polynomial-time approximation scheme solution. Next, relaxation k-price solvable provide lower bounds revenue fully discriminatory policy. latter gap naturally decreasing centrality variance network’s degree distribution. results suggest knowledge important network's density its distribution increases. Put another way, seller's capability discriminate between settings which relative influence other (or, equally, susceptibility influence) from varies greatly. Significantly, also allowing even modest number provides Finally, seeding special case k-pricing given subset free. Seeding always represents large improvement over unconstrained policy, benefits are greatest high-variance networks. seeds substantively depending whether unilateral or bilateral, extracted remains largely unaffected by reciprocity influence.