Abstract
A principal wishes to transact business with a multidimensional distribution of agents whose preferences are known only in the aggregate. Assuming a twist (= generalized Spence-Mirrlees single-crossing) hypothesis and that agents can choose only pure strategies, we identify a structural condition on the preference $b(x,y)$ of agent type $x$ for product type $y$ -- and on the principal's costs $c(y)$ -- which is necessary and sufficient for reducing the profit maximization problem faced by the principal to a convex program. This is a key step toward making the principal's problem theoretically and computationally tractable; in particular, it allows us to derive uniqueness and stability of the principal's optimum strategy -- and similarly of the strategy maximizing the expected welfare of the agents when the principal's profitability is constrained. We call this condition non-negative cross-curvature: it is also (i) necessary and sufficient to guarantee convexity of the set of $b$-convex functions, (ii) invariant under reparametrization of agent and/or product types by diffeomorphisms, and (iii) a strengthening of Ma, Trudinger and Wang's necessary and sufficient condition (A3w) for continuity of the correspondence between an exogenously prescribed distribution of agents and of products. We derive the persistence of economic effects such as the desirability for a monopoly to establish prices so high they effectively exclude a positive fraction of its potential customers, in nearly the full range of non-negatively cross-curved models.
Introduction
In microeconomics, the "principal-agent" problem describes a scenario where a seller (the principal) attempts to maximize profits from a population of buyers (agents) whose individual preferences are private. While this problem is well-understood in one-dimensional contexts—such as pricing a single good based on a buyer's income—it becomes "dauntingly difficult" when both buyers and products are defined by multiple characteristics. For instance, a car manufacturer must balance fuel efficiency, safety ratings, and luxury features for a diverse population of drivers.
The primary obstacle in these multidimensional settings is that the principal’s optimization landscape is typically non-convex. Without convexity, finding a global optimum is computationally intensive, and the resulting solutions may be unstable or non-unique. This paper identifies the "missing link": a structural condition called non-negative cross-curvature that transforms this complex problem into a tractable convex program.
The Geometry of Preferences
To address the problem, the authors leverage tools from the mathematical theory of optimal transportation. They reformulate the principal's challenge as a minimization over the space of "$b$-convex" functions. These functions are a generalization of standard convex functions, specifically tailored to a preference function $b(x,y)$, where $x$ represents the agent type and $y$ represents the product type.
The core innovation is the application of non-negative cross-curvature. This condition involves fourth-order derivatives of the preference function and acts as a strengthening of the Ma-Trudinger-Wang (MTW) condition. Originally developed to study the smoothness of optimal transport maps, the MTW condition (A3w) is here shown to be the key to ensuring that the set of feasible utility functions is convex.
Key Insights and Economic Implications
The research yields several profound insights into market dynamics:
- Convexity and Tractability: By proving that non-negative cross-curvature is the necessary and sufficient condition for the set of $b$-convex functions to be convex (Theorem 3.2), the authors provide a roadmap for numerical stability. This allows researchers to use standard algorithms to find optimal price menus and proves that the principal's optimal strategy is unique and changes continuously with small market perturbations.
- The Desirability of Exclusion: A recurring theme in monopoly theory is "exclusion" — the practice of setting prices so high that the lowest-end customers are priced out of the market. This paper provides a rigorous, general proof (Theorem 4.7) that exclusion is a robust economic phenomenon in nearly all multidimensional models satisfying the cross-curvature condition. It is not merely an artifact of simplified 1D models but a fundamental strategy for profit maximization.
- Robustness of Bunching: The study suggests that "bunching" — where a principal offers the same product bundle to a variety of different buyer types — is a persistent feature of optimal multidimensional screening.
Theoretical Framework and Results
The paper moves beyond simple bilinear preferences (like the $b(x,y) = x \cdot y$ model) to explore more complex interactions. It establishes that while the "twist" condition (a multidimensional Spence-Mirrlees property) is necessary for defining the problem, it is the cross-curvature that ensures its solvability.
The authors derive conditions under which the principal’s optimal strategy is unique (Theorem 4.5) and stable (Corollary 4.6). These results are invariant under reparametrization, meaning the underlying economic truths remain consistent regardless of how agent or product types are mathematically labeled.
Conclusion and Future Directions
By identifying non-negative cross-curvature, this work bridges the gap between abstract differential geometry and applied microeconomics. It provides the theoretical foundation needed to make multidimensional screening problems computationally tractable.
However, some limitations remain. The current model assumes that the dimensions of the agent and product spaces are equal and focuses on "pure strategies" where agents buy a single specific product. Future research may explore cases where these dimensions differ or where randomized mechanisms (lotteries) are permitted. Nonetheless, this paper marks a significant step toward a unified, tractable theory of multidimensional markets.
Summary updated