Designing Markets for Screening: Optimal Composition and the Geometry of Surplus
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.
Ruth: I'm Ruth.
Warren: And I'm Warren, here with Iris.
Ruth: We're looking at the paper: "Designing Markets for Screening".
Warren: Here's the gist of the paper. The paper characterizes market compositions that generate Pareto efficient outcomes within a screening model featuring type-dependent costs, focusing on maximizing principal or total surplus versus agent-oriented optima.
Ruth: When maximizing principal or total surplus, the market concentrates at the highest type, extracting all surplus.
Warren: However, agent-oriented optima serve all agents without bunching and efficiently serve a positive share of agents with the highest type.
Ruth: Separation is valuable even under agent surplus maximization because it propagates information rents.
Paper summary: Ruth: Let’s dig into what this paper, "Designing Markets for Screening: Optimal Composition and the Geometry of Surplus," is really saying about market design under screening. It starts by looking at how we build markets when there are these type-dependent costs involved in screening agents.
Warren: The main thesis seems to be that maximizing principal or total surplus leads to a very simple outcome where all the activity concentrates at the highest type, and they extract all available surplus. But agent-oriented optima use screening differently; they serve every agent without bunching, and they ensure a positive share of the highest type gets served efficiently.
Iris: What this implies is that even when you're trying to maximize what the principal gets or what the total surplus is, having a market structure that serves everyone efficiently—instead of just bunching at the top—is a distinct way to achieve welfare.
Ruth: Exactly. And they highlight that separation between types remains valuable even under agent surplus maximization because it lets information rents propagate around. This means heterogeneity matters for consumer welfare, even when you’re trying to make things fair for the agents.
Warren: The paper also points out that under mild cost-curvature conditions, every Pareto efficient outcome is implemented by a unique market structure. But if those costs are stronger, the set of possible surpluses can become non-convex.
Iris: That non-convexity is tied to whether there’s a type that has no gains from trade or if the values are private. When those conditions aren't met, you can get a nonconvex set of single-market payoffs because of those type-dependent costs.
Ruth: So the key finding here is that without certain types or without private values, and with these specific cost structures, we can't guarantee every payoff has a single market structure implementing it. That’s a big limitation on the nice structural results.
Warren: They also mention that market segmentation creates outcomes you don't see when you look at the maximum agent surplus at any level of principal surplus, so splitting up the market doesn't help increase consumer welfare in that comparison.
Iris: It’s about showing that while there are many ways to achieve Pareto efficiency, most of them rely on preserving heterogeneity rather than just trying to make everything uniform. That’s a key insight for how institutions should think about screening design.
Conclusion: Ruth: So, looking at the whole paper, "Designing Markets for Screening: Optimal Composition and the Geometry of Surplus," it really boils down to how we structure markets when agents have private information that firms or governments are trying to screen them on.
Warren: The title tells you two things: first, they’re figuring out the best way to design the market composition, and second, they’re mapping out the geometry of all the possible surplus outcomes.
Iris: What this means for us is that if you're designing a system—say, a platform or an insurance menu—you have to think about whether you want to focus purely on maximizing what the principal gets from that structure, or if you want to design it so that everyone in the market is served well.
Ruth: Right. The paper shows that for some conditions, like mild cost-curvature, every single efficient outcome has one specific way the market can be set up. But when those costs get complicated, like type-dependent costs without a payoff-neutral type, things get messy and you end up with a nonconvex set of possible surpluses.
Warren: The big picture is that preserving some level of heterogeneity in the market seems to be crucial for keeping consumer surplus high, even if it means the principal doesn't get as much surplus as they could from bunching everyone at the top.
Iris: It’s a warning about when simplifying a market structure might actually hurt overall efficiency because you lose those information rents that keep agents motivated and well-served.
Ruth: So, in short, it’s about finding the sweet spot between concentrating power to maximize principal gain and designing a structure that allows for efficient service across all types.
European University Institute · Digital Platforms: Pricing, Variety, and Quality Provision (DIPVAR)
econ.TH
Submitted: 2026-04-10
Updated: 2026-11-09
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 92/100
The gist: The paper characterizes market compositions that generate Pareto efficient outcomes within a screening model featuring type-dependent costs…
Key concepts
- Principal or Total Surplus Maximization
- This approach aims to maximize the total value extracted from the market, often by concentrating all surplus at the highest type. This contrasts with agent-oriented optima, which serve all agents without bunching and efficiently serving a positive share of agents with the highest type.
- Agent-Oriented Optima
- This strategy focuses on serving all agents efficiently rather than maximizing total surplus. These markets serve all agents without bunching and are shown to be efficient in serving a positive share of the highest type, even when separating markets are valuable.
- Surplus Set Convexity
- The set of achievable payoffs is convex under certain conditions, such as when values are private or when types without gains from trade can be included. However, type-dependent costs can lead to a nonconvex surplus set if there is no payoff-neutral type.
- Market Segmentation
- Segmented markets generate outcomes that are missing from the full set of Pareto efficient options. Crucially, market segmentation does not increase the maximum agent surplus achievable at any level of principal surplus.
Terminology
Summary
The paper characterizes market compositions that generate Pareto efficient outcomes within a screening model featuring type-dependent costs, focusing on maximizing principal or total surplus versus agent-oriented optima. p. 0 Designing Markets for Screening: Optimal Composition and
the Geometry of Surplus‗
Panagiotis Kyriazis†
October 14, 2026
Link to Latest Version
Abstract
We characterize the structure of market compositions that generate Pareto efficient outcomes
and the geometry of the surplus set in a screening model with type-dependent costs. Maximizing
principal or total surplus concentrates the market at the highest type, with all surplus extracted.Where the paper says it
When maximizing principal or total surplus, the market concentrates at the highest type, extracting all surplus. p. 0 Designing Markets for Screening: Optimal Composition and
the Geometry of Surplus‗
Panagiotis Kyriazis†
October 14, 2026
Link to Latest Version
Abstract
We characterize the structure of market compositions that generate Pareto efficient outcomes
and the geometry of the surplus set in a screening model with type-dependent costs. Maximizing
principal or total surplus concentrates the market at the highest type, with all surplus extracted. Maximizing
principal or total surplus concentrates the market at the highest type, with all surplus extracted. Agent-oriented optima instead use, rather than suppress, screening: all agents in the market
are served, there is no bunching, and a positive share of agents have the highest type and are
served efficiently. Agent-oriented optima instead use, rather than suppress, screening: all agents in the market
are served, there is no bunching, and a positive share of agents have the highest type and are
served efficiently. Separation is valuable even under agent surplus maximization because it
propagates information rents. Separation is valuable even under agent surplus maximization because it
propagates information rents. Under a mild cost-curvature condition, each Pareto efficient
outcome is implemented by a unique market. Under a mild cost-curvature condition, each Pareto efficient
outcome is implemented by a unique market. Under a stronger cost-curvature condition, the
surplus set is convex whenever values are private, or types with no gains from trade can be
included in the market. Without such
types and with type-dependent costs, the surplus set can be nonconvex. Market segmentation
generates the missing outcomes but does not increase the maximum agent surplus at any level
of principal surplus. Market segmentation
generates the missing outcomes but does not increase the maximum agent surplus at any level
of principal surplus. We also derive comparative statics along the Pareto frontier and with
respect to value-cost co-movement, distinguishing changes in market composition from changes
in individual allocations and rents.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Theorem 1 details the structure of optimal markets: if beta at least 1, the unique optimizer is the degenerate market delta thetā. p. 0 Designing Markets for Screening: Optimal Composition and
the Geometry of Surplus‗
Panagiotis Kyriazis†
October 14, 2026
Link to Latest Version
Abstract
We characterize the structure of market compositions that generate Pareto efficient outcomes
and the geometry of the surplus set in a screening model with type-dependent costs. Maximizing
principal or total surplus concentrates the market at the highest type, with all surplus extracted. If the primitive type
domain contains a type 𝜗
0
, weakly below all admissible types, satisfying e𝜈(𝜗
0
) = 𝐶e′
(0), an affine
change of variables maps 𝜗
0
to zero and yields 𝑚(0) = 𝐶(0) = 𝐶
′
(0) = 0, while rescaling agent and
principal surplus by the same factor. Strict convexity makes 𝜗
0
the highest type that does not trade
under the first best. Whether the characterization of this frontier is enough to characterize the entire surplus set
depends on the availability of a payoff-neutral type, meaning a type that would not trade even under
the first best. Diluting any market by adding an atom at a payoff-neutral type reduces consumer
surplus and seller profit in the same proportion.Where the paper says it
Where the paper says it
Where the paper says it
Theorem 2 describes the geometry of the surplus set: under Assumptions 1–2, adding mass to the highest first-best no-trade type scales payoffs. p. 2 Whether the characterization of this frontier is enough to characterize the entire surplus set
depends on the availability of a payoff-neutral type, meaning a type that would not trade even under
the first best. Diluting any market by adding an atom at a payoff-neutral type reduces consumer
surplus and seller profit in the same proportion. As a result, the Pareto frontier and the
lower-efficient boundary generate the whole set by diluting the market with payoff-neutral types. Under our stronger curvature condition, we prove that the surplus set is compact and convex, and
3 Without a payoff-neutral type, this scaling argument is unavailable. Under
private values, a different adjustment can hold seller profit fixed while varying consumer surplus,
so the attainable surplus set remains compact and convex. An example with a quadratic type-dependent cost component and a constant elasticity
quality cost has two minimizing markets with distinct payoffs, but no single market generates the
convex combination of those payoffs. With type-dependent costs, therefore, the surplus set can be
nonconvex when a payoff-neutral type is infeasible.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
The paper concludes that every nondegenerate frontier market combines a strictly separating interior with an efficiently served top atom. p. 2 The minimum-profit endpoint is also
uniquely implemented when a type that would not trade even under the first best is feasible, or
values are private. Therefore, every nondegenerate frontier market has two components: a strictly separating
interior and a positive atom at the top. The same framework identifies when expected consumer gains need not reflect better treatment
of individual types. With sufficiently steep type-dependent costs near the top and a sufficiently high
common minimum type, greater consumer orientation raises expected consumer surplus while
lowering rent at every type above that minimum.Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Market design can be optimized by choosing between a single market and a set of segmented markets, where segmentation is valuable when
no single market generates all possible outcomes
because itconvexifies it by generating the missing outcomes on the minimum consumer surplus (or rentsuppression) side of the set.
-
AI systems can implement screening strategies that balance information rents and efficient trade by maintaining a connected strictly separating interior and an efficiently served top atom in markets, as this structure is characterized as
a canonical way of optimally balancing information rents and efficient trade in one-dimensional screening environments.
-
The system can differentiate between private values and type-dependent costs to understand how
gross willingness to pay and costs rise more closely together,
allowing it to predict which environment will generatemore policyholder surplus
orhigher expected consumer surplus at a given profit.
-
AI systems can perform comparative statics by varying the weight on profit, learning that for a lower relative profit weight,
Consumer surplus is weakly higher under 1/2,
and for an even lower weight, the optimal market structure changes to favor a higher minimum support type. -
The system can use cost attenuation to model hold-up costs where agents invest in privacy or better options before screening, determining whether
the degenerate top market is optimal even when profit weight is less than 1.
Abstract
We characterize the structure of market compositions that generate Pareto efficient outcomes and the geometry of the surplus set in a screening model with type-dependent costs. Under mild cost-curvature conditions, agent-oriented optima use rather than suppress screening to sustain information rents: all agents are served, allocations strictly separate types, and an atom at the highest type receives the efficient allocation. Each Pareto efficient payoff is implemented by a unique market. When costs are type independent or a type with no gains from trade is feasible, we also characterize the entire surplus set and establish its convexity. Otherwise, type-dependent costs can generate nonconvexity. Market segmentation can expand attainable outcomes only on the rent-suppression side of the set, without increasing maximum agent surplus at any level of principal surplus. The characterization and comparative statics link market formation to the distribution of surplus and identify the limits on payoffs attainable under screening.