A Tale of Two Monopolies
Yi-Chun Chen, Zhengqing Gui
TL;DR
This paper develops a margin-based approach to optimal selling mechanisms for multiproduct monopolies by perturbing prices over bundles and using the marginal-revenue (MR) calculus. Grounded in the taxation principle, it recasts incentive-compatibility constraints into self-selection pricing and derives first- and second-order conditions (FOCs and SOC) that pin down the structure of revenue-maximizing mechanisms. For symmetric two-dimensional (and, more generally, convex type/allocation spaces) under mild regularity, the framework yields complete characterizations across independence, substitutability, and complementarity, linking MR to the geometry of demand sets via a threshold function $\zeta$ and a one-dimensional indirect utility $\bar u$. The results explain when pure bundling, separate selling, or probabilistic/mixed bundling is optimal, and uncover general features such as pooling bundles, active bundles, and the possible existence of an exclusion set. Overall, the MR approach provides a unified, tractable route to tractable optimal mechanisms in multidimensional screening with broad applicability and clear comparative statics with respect to substitution/complementarity.
Abstract
We apply marginal analysis à la Bulow and Roberts (1989) to characterize revenue-maximizing selling mechanisms for a multiproduct monopoly. We derive marginal revenue from price perturbations over arbitrary sets of bundles and show that optimal mechanisms admit no revenue-increasing perturbation for bundles with positive demand, nor revenue-decreasing perturbations for zero-demand bundles. For any symmetric two-dimensional type distribution under mild regularity, this analysis fully characterizes the optimal mechanism across independence, substitutability, and complementarity. For general type distributions and allocation spaces, our approach identifies bundles that must carry positive demand and provides conditions under which pure bundling or separate selling is suboptimal.
