|
on Economic Design |
| By: | Koessler, Frédéric; Skreta, Vasiliki |
| Abstract: | We consider a privately informed sender selecting a mediated communication device to influence players’ actions. In contrast to standard information design, there is no ex-ante commitment, and the communication device relies on information elicited from the informed parties. We define an informed communication equilibrium (ICE) as a perfect Bayesian equilibrium outcome of the informed mediated communication game. We show that the set of ICE is the subset of communication equilibria (CE) that yield sender payoff vectors bounded below by an equilibrium payoff vector of the silent game, under some consistent interim beliefs. The principal’s ex-ante optimal CE may not be an ICE. In sender-receiver games, the set of CE payoff vectors lies in the convex hull of the set of payoff vectors above silence. Hence, if the latter set is convex, as is the case when the sender has transparent motives, the sets of CE and ICE coincide. |
| Date: | 2025–07 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20455 |
| By: | Carnehl, Christoph; Sobolev, Anton; Stahl, Konrad; Stenzel, André |
| Abstract: | We study information design in a vertically differentiated market. Two firms offer products of ex-ante unknown qualities. A third party designs a system to publicly disclose information. More precise information guides consumers toward their preferred product but increases expected product differentiation, allowing firms to raise prices. Full disclosure of the product ranking alone suffices to maximize industry profits. Consumer surplus is maximized, however, whenever no information about the product ranking is disclosed, as the benefit of competitive pricing always dominates the loss from suboptimal choices. The provision of public information on product quality becomes questionable. |
| Keywords: | Information design; Vertical product differentiation; Quality rankings; Competition |
| JEL: | D43 D82 L13 L15 |
| Date: | 2025–08 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20537 |
| By: | Attar, Andrea; Campioni, Eloisa; Mariotti, Thomas; Pavan, Alessandro |
| Abstract: | We study the design of market information in competing-mechanism games. We identify a new dimension, private disclosures, whereby the principals asymmetrically inform the agents of how their mechanisms operate. We show that private disclosures have two important effects. First, they can raise a principal's payoff guarantee against her competitors' threats. Second, they can support equilibrium outcomes and payoffs that cannot be supported with standard mechanisms. These results call for a novel approach to competing mechanisms, which we develop to identify a canonical game and a canonical class of equilibria, thereby establishing a new revelation principle for this class of environments. |
| JEL: | D82 |
| Date: | 2025–07 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20450 |
| By: | Bonatti, Alessandro; Dahleh, Munther; Horel, Thibaut |
| Abstract: | We study linear–quadratic games of incomplete information with Gaussian uncertainty, where each player's payoff depends on a privately observed type and a common state. The designer observes the state, elicits types, and sells action recommendations. We characterize all implementable mechanisms with Gaussian joint distributions of actions and fundamentals, and identify the players-optimal, consumer-optimal, and revenue-maximizing designs. In games of strategic complements (substitutes), these optimal mechanisms maximally correlate (anticorrelate) players’ actions. When type uncertainty is large, recommendations become deterministic linear functions of the state and reports, but remain only partially revealing. |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20794 |
| By: | Hoffmann, Florian; Vladimirov, Vladimir |
| Abstract: | We investigate a seller's strategic choice between optimally-structured negotiations with fewer bidders and an auction with more competing bidders when payments can have a contingent component, as is common in mergers and acquisitions, patent licensing, and employee compensation. The key factor favoring negotiations is that it allows the seller to set her preferred payment structure ---i.e., the revenue-maximizing mix of cash and contingent pay; reserve prices are of secondary importance. Negotiations are more likely to dominate if synergies increase in bidders' productivity types (as with acquirer-target complementarities in M&A). Higher dispersion and magnitude of bidders' private valuations also favor negotiations. |
| Keywords: | Negotiations |
| JEL: | G32 M52 J54 J33 |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20703 |
| By: | Li Guo; Penghuan Yan |
| Abstract: | We study a model in which a sender allocates limited treatment to agents with heterogeneous quality and later recommends selected agents to a receiver, seeking to maximize the number of agents accepted by the receiver. All agents value treatment, which improves agents' quality, but treatment must be allocated before the sender observes agents' initial quality; recommendation occurs only after quality is learned. A natural benchmark is to design the two instruments separately: allocate treatment randomly first, and then recommend agents from the top down afterward. Our main result shows that the sender can do strictly better by coordinating treatment allocation with recommendations. In the optimal joint mechanism, treatment is non-monotone in quality: an intermediate group has a lower treatment probability than both higher- and lower-quality agents, but is compensated with a guaranteed recommendation when treatment is realized. We provide an implementation through contracts that induce self-selection and discuss applications to education, industrial policy, and startup incubation. The takeaway is simple: coordinate treatment allocation and recommendation. |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2606.21120 |
| By: | Madhuparna Karmokar; Ujjwal Kumar; Soumyarup Sadhukhan |
| Abstract: | We study unanimous and strategy-proof probabilistic social choice correspondences (PSCCs), where the selected set of alternatives is interpreted as an interim outcome, and agents evaluate sets using conditional expected utility. We analyze two preference domains introduced by Barbera et al. (2001): the conditionally expected utility consistent (CEUC) domain and the conditionally expected utility consistent with equal probabilities (CEUCEP) domain. Our results characterize all unanimous and strategy-proof PSCCs on these domains and identify cases when randomization enlarges the class of admissible rules. On the CEUC domain, every unanimous and strategy-proof PSCC is a random dictatorship, showing that randomization over sets yields no additional flexibility. In contrast, the CEUCEP domain admits a richer family of unanimous and strategy-proof PSCCs. For at most three agents, these rules are precisely the random bi-dictatorial rules, which are convex combinations of bi-dictatorial rules introduced in Feldman (1980). For four or more agents, the characterization depends on the number of alternatives. When there are exactly three alternatives, the class expands to the larger family of coalition-weighted rules. Thus, randomization enlarges the class of strategy-proof correspondences in the three-alternative case, producing rules that are not convex combinations of deterministic strategy-proof correspondences. However, for four or more alternatives, the class of unanimous and strategy-proof probabilistic correspondences again collapse to random bi-dictatorships. |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.03955 |
| By: | Choné, Philippe; Linnemer, Laurent |
| Abstract: | Sellers face a critical choice: run competitive auctions or strike exclusive deals with preferred buyers. Contrary to conventional wisdom that sellers should rely on open competition, we show that a powerful seller optimally commits to a sequential `flexclusivity' arrangement --a strategic mix of exclusivity and competitive bidding. Under broad conditions, the seller chooses with positive probability to disregard alternative buyers entirely. We demonstrate, in a parsimonious model, that simple option contracts implement flexclusivity efficiently, increasing the expected joint profit of the contracting parties. When a preferred buyer declines the option, this credibly signals his weakness, allowing the seller to extract more rent from stronger buyers in subsequent auctions. The joint gain from such arrangements can represent as much as 75% of what vertical integration would achieve, without requiring commitment beyond the initial contracting stage. |
| JEL: | D44 D82 D86 L22 |
| Date: | 2025–09 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20620 |
| By: | Jingyu Liu; Bolin Zhang; Lin William Cong; Siguang Li; Xuechao Wang |
| Abstract: | Ethereum's Glamsterdam upgrade introduces enshrined proposer-builder separation (ePBS), replacing relay-centric PBS with direct builder bids to proposers. We study how this shift changes the block-building microstructure through a general imperfect-information two-stage auction with verifiable messages, where an early bid serves as both a price offer and a signal. PBS and ePBS are modeled as restrictions of the same block-building game: PBS fixes stopping and disclosure exogenously, while ePBS lets the proposer choose stopping and disclosure ex post. Latency heterogeneity is captured by asymmetric information updates: fast builders observe disclosed early information before rebidding, while slow builders do not. We combine exact perfect Bayesian equilibrium characterizations in tractable cases with calibrated no-regret learning in finite games. For PBS, we show that separating equilibria preserve the standard first-price-auction payoff benchmark and provide conditions for their existence. For ePBS, we demonstrate a ratchet effect: because the proposer can defer block proposal and use early bid information in the second stage, builders anticipate ex-post extraction and shade or pool early bids, generating allocation inefficiency and revenue-efficiency valleys. We interpret this ratchet distortion as a commitment failure. Under full commitment, the optimal policy collapses to the static Myerson auction and removes the ratchet channel. To realize part of this commitment advantage in a feasible mechanism, we propose a Trusted Execution Environment (TEE) sidecar that enforces limited commitment. We formulate the revenue-maximizing TEE mechanism as a bilinear optimization problem. In conservative finite benchmarks, the TEE design increases the proposer revenue relative to the first-price benchmark by approximately \(25\%\). |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.11240 |
| By: | Alex Chan |
| Abstract: | I study insurance markets in which scalable prediction, like AI, designs residual risk rather than merely classifies fixed risk. A complete-contracting benchmark shows that if prevention is observable, contractible, competitively supplied, and fully priced, it does not matter whether consumers, insurers, or vendors supply it. Adverse selection breaks such irrelevance. When high-risk consumers are more "AI-treatable, " efficient prevention makes low-risk contracts attractive to them. A contract intended for low-risk consumers faces a risk-design trilemma: separate, prevent efficiently, or avoid cross-subsidy, but not all three. The result extends Rothschild-Stiglitz from distorted coverage to distorted risk-control technology and offers market design insights of AI in insurance markets. |
| JEL: | D4 D47 D81 D82 D86 G22 G52 I13 O33 |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:35444 |
| By: | Furkan Sezer |
| Abstract: | This paper studies information design when the designer lacks precise knowledge of agents' payoff coefficients. The Calibrated Bayes Correlated Equilibrium (Cal-BCE) is introduced as a solution concept that augments the Bayes correlated equilibrium with a corrector policy preserving incentive compatibility under the designer's structural uncertainty, adapting its revelation principle to this setting. The design problem is nonconvex in general, but under a linear-quadratic-Gaussian structure it admits convex second-order cone and semidefinite reformulations under two-sided probabilistic and conditional value-at-risk (CVaR) constraints, with feasibility guaranteed by a Hadamard invertibility condition. A joint decentralization theorem shows that both designs cap cross-agent action covariances, the CVaR design more tightly at a common tolerance; but because the formulations operate at design-specific feasibility thresholds, the realized ordering is calibration-dependent. Experiments on fifteen sector ETFs confirm the trade-off: the probabilistic design attains higher mean welfare and the CVaR design better tail protection, with neither dominating outright. |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2606.22157 |
| By: | Yi-You Yang |
| Abstract: | We study bilaterally unitary many-to-many doctor--hospital matching with contracts, taking choice functions as primitives. Doctor choices are substitutable and satisfy irrelevance of rejected contracts, while hospital choices are unilaterally substitutable and satisfy the same condition. Every trajectory of the doctor-proposing cumulative offer process terminates at the greatest stable allocation under the doctor Blair order. We also introduce weakly hospital-quasi-stable allocations and show that they form a finite lattice whose greatest element is stable. Hence, the cumulative-offer outcome, the greatest weakly hospital-quasi-stable allocation, and the doctor-optimal stable allocation coincide. The common allocation is hospital-pessimal in the revealed-choice sense. Under the law of aggregate demand, every agent signs the same number of contracts at all stable allocations. |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.10756 |
| By: | Kazumasa Ikeda |
| Abstract: | We introduce a new non-manipulability axiom called worst-case strategy-proofness (WCSP). This axiom is weaker than strategy-proofness and stronger than non-obvious manipulability-worst (NOM-worst) by Troyan and Morrill (2020). WCSP focuses on non-manipulability in a worst-case scenario. We examine the implications of WCSP in a voting model. Although many voting rules, such as the plurality rule, the Borda rule, and the Dowdall rule, satisfy NOM-worst, they violate WCSP. We obtain a necessary and sufficient condition for the anti-plurality rule with fixed-order tie-breaking to satisfy WCSP in terms of the numbers of agents and alternatives. |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2606.22021 |
| By: | Xiaoming Cai; Pieter Gautier; Ronald Wolthoff; Pieter A. Gautier |
| Abstract: | We study a monopoly platform that sets meeting rates between buyers and two seller types: niche sellers, whose higher-quality good appeals to only some buyers, and mass-market sellers, whose good appeals to all. Sellers compete by posting prices à la Burdett and Judd (1983), so buyer surplus requires competition, while platform revenue requires seller rents. This tension creates a systematic distortion: as search capacity grows, the platform keeps niche exposure just past the saturation point---where extra attention erodes rents---and diverts the rest to mass-market sellers. Applied to Amazon product search and Google passage-ranking data, the model indicates buyer-surplus losses of 63 and 44 percent of the planner's benchmark, respectively. Letting buyer participation respond to the platform's recommendations disciplines it and shrinks this loss. |
| Keywords: | attention allocation, Recommendation systems, search frictions, two-sided markets, enshittification of internet |
| JEL: | D62 D83 L12 L40 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_12760 |
| By: | Janssen, Maarten; Roy, Santanu |
| Abstract: | A buyer and a seller can privately learn the quality of an asset - initially unknown to both - by incurring a fixed cost before trading. Asset quality determines their valuations and the seller makes a take-it-or-leave-it price offer. Under a weak "lemons-like" condition, asymmetric information arises endogenously when learning costs are small; as these costs vanish, the seller learns for sure but the buyer remains uninformed with probability bounded away from zero. Nevertheless, efficient limiting equilibria always exist; the buyer earns strictly positive surplus in such an equilibrium if, and only if, she can learn after knowing the price offer. |
| JEL: | L13 L15 D82 D43 |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20786 |
| By: | Yu-Ting Ho |
| Abstract: | We study competition among multiple firms that offer differentiated varieties of the same good to a unit-demand agent. The agent has heterogeneous valuations for goods from different firms. Firms do not observe the agent's exact valuations, but they know their distribution. Firms simultaneously post menus of contracts, after which the agent chooses a firm and one of its contracts to maximize her utility. This defines a game in which firms aim to maximize expected revenue. We introduce a sufficient condition, density-regularity, under which each firm's best response to any arbitrary menu profile posted by its opponents is equivalent to posting a menu that contains only a posted-price contract. Our result is not a direct extension of the canonical Myersonian model with a single seller. The standard argument in the literature breaks down once heterogeneous preferences and competition are introduced. We therefore adopt an optimal-control approach, in which the density-regularity condition is essential for establishing the optimality of posted prices. When this condition fails, posted prices may fail to be a best response. |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2606.22720 |
| By: | Koessler, Frédéric; Renault, Régis |
| Abstract: | A buyer can learn about a product through search or seller-disclosed information. We examine how lower search costs or better seller disclosure affects this interaction. Whereas a drop in search costs improves consumer surplus and decreases profit when the seller can resort to an optimal disclosure strategy, its impact is ambiguous if the seller is unable to provide information. When it is unlikely that the buyer's valuation is below marginal cost, the buyer does not benefit from optimal information disclosure if search costs are high. With such high search costs and no disclosure both parties can be better off than with lower search costs and optimal information disclosure. The seller then adopts a mass market strategy where she posts a low enough price so the buyer always purchases the product without search. By contrast, if it is sufficiently likely that the buyer's valuation is below marginal cost, then the buyer can benefit from sophisticated information disclosure for relatively low search costs. The corresponding outcome is better for both parties than an environment with higher search costs and no information disclosure. The optimal seller strategy targets a niche of high-valuation buyers and prevents wasteful search by buyers with low valuations. |
| Keywords: | Information design; Information acquisition; Advertising; Consumer search |
| JEL: | D42 D82 D83 |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20694 |
| By: | Fleurbaey, Marc; Kornek, Ulrike; Edenhofer, Ottmar |
| Abstract: | An international carbon pricing regime offers significant efficiency gains by avoiding climate change and reducing emissions at least cost. We clarify the role that country-specific prices play with respect to efficiency and burden sharing in an unequal world. Country-specific carbon prices are efficient and serve society's equity objectives if second-best constraints exclude optimal transfers to deal with pre-existing inequalities. This also holds true within the framework of carbon markets, where trade at a common price is just one allocation at the (constrained) Pareto-efficient frontier if the initial permit allocation is fixed. A common carbon price aligns with social objectives if transfers or initial permit allocations can be freely adjusted, and otherwise differentiated prices may be preferable. But the social welfare gains from differentiated carbon prices may vary depending on empirical facts. We present a theoretical model which connects the level of national carbon prices to the choice of social welfare functions that capture various equity principles. With a calibrated integrated assessment model we quantify country-specific carbon prices for a large set of countries and show that they can significantly promote social welfare. |
| JEL: | D62 D63 Q54 Q58 |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20723 |
| By: | Flip Klijn |
| Abstract: | We study the school-choice model of Abdulkadiroğlu and Sönmez (2003) with capped rank-order lists and introduce deferred acceptance with clocked skips (sDA). The mechanism combines DA-style tentative holding with a public round clock: a student who is displaced after being held resumes only at the current clock position and skips list positions that passed while she was held. We first show that every stable matching can be supported as a Nash-equilibrium outcome of the game induced by sDA. With cap one, sDA coincides with immediate acceptance, and its Nash-equilibrium outcomes are exactly the stable matchings. This exact implementation result continues to hold at cap two for every priority structure—a guarantee DA provides only under Ergin-acyclicity. With caps of three or more, however, unstable Nash-equilibrium outcomes can arise even in one-to-one markets; moreover, every Ergin-cycle can be used to support such an outcome for some preference profile. Surprisingly, equilibrium outcomes are not necessarily nested in the cap. These findings identify a niche for sDA in transparent short-list environments while cautioning that the cap is not a monotone design lever: longer lists can both create and destroy unstable equilibrium outcomes. |
| Keywords: | capped rank-order lists, deferred acceptance, deferred acceptance with clocked skips, Ergin-acyclicity, school choice, stability |
| JEL: | C78 D47 C72 |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:bge:wpaper:1583 |
| By: | Debora Daniela Escobar; Wing Fung Chong |
| Abstract: | This paper studies centralized risk sharing with endogenous prices. Multiple policyholders transfer risks to a central insurer through indemnity decisions, while prices are determined by pricing functionals applied to ceded risks. The resulting problem is multiobjective, with Pareto optimality as the natural efficiency criterion. We show that classical Pareto optimality may fail to reveal whether all agents are represented in a balanced decision process that scalarized objectives may assign zero weight to some agents, and group aggregates may obscure individual risk positions. Motivated by bilateral Pareto characterizations through sequential optimization, we introduce inclusive and fair Pareto optimality, a representation-based refinement requiring every agent to appear exactly once, either individually or as part of a group, in a finite ordered sequence of optimizations. Our main result proves equivalence between this concept and balanced sequential optimization, placing it between Geoffrion-proper Pareto optimality and classical Pareto optimality. An illustrative example demonstrates the framework using the Expected Shortfall. |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2606.22956 |
| By: | Antonio Cabrales; Wenhao Cheng |
| Abstract: | This paper studies how organizations should jointly design evaluation rules and assign workers when performance depends on both effort and non-discretionary advantage. Agents choose effort in positions linked by a competition network, while their effective advantage depends on own type and spillovers through a second network. The planner chooses both the assignment and the effort weight in evaluation. Equilibrium effort rises with a position's Katz-Bonacich centrality and falls with effective advantage. The optimal evaluation rule generally differs from true output. When effort is more important in production, the planner lowers the effort weight and uses negative assortative assignment to strengthen incentives. When advantage is more important, the planner raises the effort weight and uses positive assortative assignment to exploit spillovers. We also study a constraint requiring assignments to be pairwise stable, which creates an output loss depending on the intensity of competition. |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.07280 |
| By: | Xinxiang Guo; Yingkai Li; Yifen Mu |
| Abstract: | Decision-makers often rely on multiple probabilistic forecasts that are individually calibrated but need not be fully informative. We develop a framework for aggregating such forecasts when the decision-maker knows only that experts satisfy calibration. We show that the joint distribution of calibrated forecasts can contain decision-relevant information that is unavailable from any single expert, so the standard optimal-in-hindsight (OIH) benchmark may substantially understate attainable performance. To formalize this idea, we introduce a robust max-min benchmark: the best payoff a decision-maker can guarantee against all profile-wise conditional-mean mappings compatible with calibration. This benchmark is tractable, admits a linear-programming formulation, and dominates the OIH benchmark up to calibration error. It can nevertheless be strictly below the Bayesian benchmark, clarifying the value of knowing experts' information structures. Finally, we provide online algorithms that attain the robust benchmark under forecast-only feedback and stronger contextual benchmarks under state feedback. |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2606.31020 |
| By: | Flip Klijn |
| Abstract: | We provide a systematic analysis of the conditional acceptance mechanism in the standard school-choice model. Students play a strategic game in which they may submit preference lists of length at most κ (the cap). Equilibrium sets are nested in the cap (Theorem 1) and coincide with the set of stable matchings for κ ≤ 2 (Propositions 1 and 2), whereas unstable equilibrium outcomes can arise for every κ ≥ 3 (Example 2 and Corollary 1). Our main results compare conditional acceptance with deferred acceptance. Ergin-acyclicity is sufficient (Theorem 2), but not necessary (Example 3), for conditional acceptance to implement the set of stable matchings, so conditional acceptance implements the set of stable matchings whenever deferred acceptance does (Corollary 2). Strikingly, under acceptable-only reports, every unstable conditional-acceptance equilibrium outcome is also a deferred-acceptance equilibrium outcome (Theorem 3). Overall, conditional acceptance may outperform deferred acceptance in producing stable equilibrium outcomes, but cumulative removal makes strategic mistakes more costly. |
| Keywords: | capped rank-order lists, conditional acceptance, Ergin-acyclicity, school choice, stability |
| JEL: | C78 D47 C72 |
| Date: | 2026–06 |
| URL: | https://d.repec.org/n?u=RePEc:bge:wpaper:1582 |
| By: | Yi-You Yang |
| Abstract: | We ask when an incumbent economy with indivisible goods can accommodate an arbitrary new participant while retaining an efficient allocation supported by anonymous item prices. We call this property universal entry robustness. An economy is universally entry-robust if and only if its aggregate welfare valuation is additive. Although the requirement quantifies over all entrant valuations, it can be tested using one canonical entrant whose value for a bundle equals the loss in maximal incumbent welfare caused by removing that bundle. When the characterization holds, one uniquely determined price vector decentralizes the incumbent optimum and every one-agent extension. The proof is direct and uses only demand optimality and welfare comparisons, making transparent why arbitrary-entry robustness eliminates all bundle interactions in aggregate welfare. We finally relate this argument to the robust-integrality interpretation obtained from the linear-programming characterization of Bikhchandani and Mamer (1997). Individual incumbent valuations may be nonadditive and need not satisfy gross substitutes. |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.11658 |
| By: | Taksch Dube |
| Abstract: | Advertisers delegate bidding to autobidders; users delegate tasks to language-model agents. A person describes what they want to an automated proxy that acts in a mechanism on their behalf. This is the revelation principle in production, and it forces a question classical theory assumes away: when is it optimal to describe yourself honestly to your own proxy? We show the answer turns on one quantity, the proxy's within-range regret. The most a principal can gain by misreporting equals the regret of the proxy's honest-report action against those the principal could have steered it to take. Honest self-description is optimal exactly when the proxy already plays the best action it can reach, that is, when it is loyal (Theorem 1). The identity unifies auction-specific autobidding results and pins down when the faithful-communication assumption behind language-model elicitation proxies (Huang et al.) holds. The identity constrains guardrails placed on proxies, from bid caps to a model's alignment layer. No guardrail can be at once binding (it displaces the truthful action from the proxy's best reachable outcome), truthful (honest reporting stays optimal), and capability-preserving (that outcome stays reachable through some report); any two preclude the third (Theorem 2). A safety constraint that alters what a model does while leaving its best output reachable makes honest description of intent suboptimal, so a sharper report can gain. This is the incentive behind prompt-engineering and jailbreaking. Because within-range regret is #P-hard to compute exactly, we estimate it from samples and maintain it as a model is updated, at a cost set by how far the model drifts, not how often it changes. Running it on production language models from five providers under an alignment-style cap, we find honest reporting leaves surplus unclaimed on every model, recovered by inflating the report. |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.14357 |
| By: | Ertner, Sebastian; Janssen, Maarten |
| Abstract: | Many e-commerce retailers adopt strategies that induce consumers to order multiple products at once, inspect their fit at home, and then decide which products to return. These policies introduce a trade-off as they result in consumers acquiring products that better fit their taste, at the expense of the private and social costs associated with products being returned. We determine the conditions under which retailers find it optimal to induce consumers to inspect products simultaneously or sequentially. We also analyze the efficiency properties of market outcomes and state conditions under which inducing simultaneous inspection (surprisingly) leads to fewer returns. An important part of the analysis characterizes the optimal alternative pricing policy that induces consumers to sequentially inspect products after ordering and finds that partial refunds facilitate the extraction of surplus from consumers. |
| JEL: | D40 D83 L10 |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20711 |
| By: | Mingyang Fu; Ming Hu |
| Abstract: | Agentic AI is shifting online shopping from search toward delegated purchasing, where autonomous buying agents monitor markets and decide when to buy on a consumer's behalf. We study the design of such strategic buying agents, which must decide when to purchase within a finite shopping window, translating price observations, the remaining time horizon, and beliefs about future price changes into a purchase policy. We formulate this problem across three information regimes: stationary, Bayesian, and robust, and treat the resulting optimal policies as a policy menu for implementation. In the stationary regime, price adjustments follow a Poisson arrival process with a known post-adjustment price distribution; the optimal policy is a dynamic purchase-threshold rule, with the threshold governed by an ordinary differential equation. In the Bayesian regime, the adjustment intensity is known, but the price-adjustment distribution is uncertain; the optimal rule remains threshold-based, now depending on posterior beliefs, and we bound the value of knowing the true distribution. In the robust regime, the agent has only price bounds and seeks worst-case protection; randomized threshold policies achieve optimal competitive-ratio and minimax-regret guarantees. We evaluate the proposed policies on Amazon price histories from Keepa (367 items, 48, 933 timestamped observations) and examine their integration into language-model buying agents. The stationary and Bayesian policies perform competitively on mean normalized consumer surplus despite their stylized assumptions, while the robust policy performs best at the distribution's 10th percentile. Results suggest language models are better suited to selecting among regimes and calibration samples than to making buy-or-wait decisions directly. |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.04708 |
| By: | Alfredo Di Tillio; Marco Ottaviani; Peter N. S{\o}rensen |
| Abstract: | We compare the informativeness of order statistics in a sample of conditionally independent draws from a distribution \(F(x\mid\theta)\) as the sample size n increases. The k-th highest of n+1 draws is more accurate than the k-th highest of n if and only if the cumulative reverse hazard -\log F(x\mid\theta) is log-supermodular. Symmetrically, the k-th lowest is more accurate if and only if the cumulative hazard -\log(1-F(x\mid\theta)) is log-supermodular. Reversals are exceptional, occurring only for experiments that are, up to increasing transformations, exponential location experiments. In large samples, middle order statistics are asymptotically fully informative, while bounded lower and upper ranks require unbounded informativeness tail conditions. When full learning fails, bounded ranks converge to location experiments, and more central ranks are Blackwell more informative. Extending the analysis from scalar order statistics to blocks of selected data, we obtain multidimensional comparisons under log-supermodularity of hazard rates. The results unify and extend information-aggregation results in auctions and provide a new order-statistic approach to strategic voting. |
| Date: | 2026–07 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2607.10435 |
| By: | Graeber, Thomas; Noy, Shakked; Roth, Christopher |
| Abstract: | Information often shapes behavior regardless of its quality: unreliable claims wield influence, while reliable ones are neglected. We propose that this occurs in part because word-of-mouth transmission tends to preserve claims while dropping information about their reliability. We conduct controlled online experiments where participants listen to economic forecasts and pass them on through voice messages. Other participants listen either to original or transmitted audio recordings and report incentivized beliefs. Across various transmitter incentive schemes, a claim’s reliability is lost in transmission much more than the claim itself. Reliable and unreliable information, once filtered through transmission, impact listener beliefs similarly. Mechanism experiments show that reliability is lost not because it is perceived as less relevant or harder to transmit, but because it is less likely to come to mind during transmission. A simple associative-memory framework suggests that reliability information may be less likely to come to mind either because it is less likely to be cued by transmission requests or because attempts to retrieve it face greater interference. Evidence from our experiments, a large corpus of everyday conversations, and economic TV news supports both of these mechanisms. |
| Keywords: | Memory |
| JEL: | D83 |
| Date: | 2025–09 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20619 |
| By: | Riedel, Frank (Center for Mathematical Economics, Bielefeld University); Spengemann, Marco (Center for Mathematical Economics, Bielefeld University) |
| Abstract: | We study optimal insurance design under linear transaction costs for a policyholder with smooth ambiguity preferences when the underlying loss distribution is identifiable ex post. Under expected utility, Arrow’s classical theorem implies that the optimal contract is a straight deductible. Under ambiguity, this result generally fails. We show that ex post identifiability restores the straight-deductible structure: the optimal contract consists of model-contingent straight deductibles. We also derive comparative statics with respect to transaction costs and characterize the ordering of deductibles under stochastic ordering of loss distributions. |
| Keywords: | Insurance design, risk sharing, smooth ambiguity aversion, identifiability, straight deductible, comparative statics |
| Date: | 2026–07–22 |
| URL: | https://d.repec.org/n?u=RePEc:bie:wpaper:767 |
| By: | Bar-Isaac, Heski; Levy, Raphaël |
| Abstract: | We study training provision in environments where employers privately observe employee performance and thus derive information rents. Training impacts not only overall productivity, but also the extent of adverse selection, and, thereby, the way surplus is shared: firms or workers may strategically over-invest in training to increase their share of the pie. Going beyond the traditional distinction between general and specific training, we highlight how various kinds of training generate different incentives: workers aim at curbing adverse selection and therefore invest in remedial training that primarily benefits the least able or guarantees minimal skills; conversely, firms focus on excellence-oriented programs that boost the productivity of the most able and promote organizational practices that increase match-specific productivity. More generally, the paper highlights how the design and composition of training programs—and their interaction with other HR practices—shape turnover, wages, and firm performance through the adverse-selection channel. |
| Keywords: | Training; Adverse selection; General human capital; Firm-specific human capital |
| JEL: | D82 J24 J31 M53 |
| Date: | 2025–08 |
| URL: | https://d.repec.org/n?u=RePEc:cpr:ceprdp:20530 |