The Limit of Targeting in Networks

This paper studies the value of network-based targeting within a class of network games withstrategic complements, where the designer can target a group of seed players as first movers. We define an effectiveness index, called the relative network synergy equivalent (RNSE), to quantify the effect of a network-based targeting intervention relative to across-the-board peereffect enhancement. We show that, regardless of the targeting policies and network structures, a universal and tight upper bound for this index is √2 ≈ 1.414. This upper bound is robustto considerations of revenue maximization, costly seeding, random seeding as the alternative benchmark, and multiple stages of moves. Compared with network-based targeting, peer effect enhancing policy has the advantage of being agnostic to the network structure. In the case of small synergy, we provide comparative statics of the RNSE index concerning the network structures: fixing the targeting policy, increasing network links within the seeded group or the unseeded group will decrease the index; meanwhile, adding links across these two groups will increase the index. Our analysis sheds light on policy choices between network-based targetingand peer effect enhancing policies.