Optimal Bundling of Technological Products with Network Externality

TLDR

Network externality makes consumer utility depend on user base size, and firms with asymmetric products may consider strategies beyond pure components, pure bundling, and mixed bundling, including a mixed bundling‑1 approach where one product is sold only in a bundle. The study compares how asymmetry or symmetry in network externality and cost affect bundling strategy choice. The authors model a monopolist with two product categories facing heterogeneous consumers. Pure bundling yields higher profits when both products have low marginal costs or high network externality, whereas pure components or mixed bundling‑1 are preferable when costs and externalities diverge, and traditional mixed bundling is optimal in other cases.

Abstract

For many high-tech and Internet-related products, utility to consumers depends in part on the size of the user base, a phenomenon called network externality. A firm with a portfolio of these and other products—that are often asymmetric in their degree of network externality or marginal cost—may have to look beyond the traditional strategies of pure components, pure bundling, and mixed bundling. One such strategic alternative in a two-product case would be a so-called mixed bundling-1 under which the bundle and product 1 are offered, but the other product can only be purchased in a bundled form. The purpose of this study is to compare and contrast the impact of such asymmetry or symmetry in (direct) network externality and cost on the choice of bundling strategies. We model a monopolist firm that has a product in each of two categories and faces heterogeneous consumers. Results suggest that pure bundling is more profitable when both products have low marginal costs or high network externality whereas pure components or mixed bundling-1 is more profitable when the products diverge in their costs and network externality (e.g., only one product has network externality). Traditional mixed bundling is optimal in other instances.