Greater Bay Area
Market Design
Workshop 2026

Ride-hailing platforms increasingly use rank-based dispatch: peak-period compliance earns points that translate into lean-period priority. We study two consequences: an involution mechanism (rankings promote peak-time acceptance, expand hours and mileage, suppress unit pay, increase fatigue-related safety risk) and a collusion mechanism (rank becomes reputation capital that discourages multihoming, partitions the driver pool, and softens wage competition). We develop a dynamic model of rank-based dispatch and formally derive the involution and collusion mechanisms. Using driver surveys, municipal monitoring reports, and a staggered quasi-experiment around a cab platform's ranking intensification, we document significant empirical support. Findings motivate targeted regulation of rank-based incentives.

Inspired by major transition programs, this paper studies the design of exchange mechanisms within a framework that incorporates both distributional constraints and dual-priority-respecting constraints. We propose two mechanisms that match students to majors in an efficient way, while respecting each major's distributional constraints and dual-priority-respecting constraints. The efficient mechanisms are based on eligibility maximization (EM) and two processes for identifying improvement cycles — TiE cycles and ToE cycles.

Motivated by the pervasive need for privacy, we investigate the confidentiality of dynamic mechanisms that implement a given social choice function. We introduce two privacy criteria, intrusiveness and exposedness. A mechanism is confidential if and only if neither of the two proposed transformations (subtracting duplicate actions and obfuscating superfluous information) can be applied to the associated gradual mechanisms. We provide confidential mechanisms for the second-price auction rule and the hierarchical exchange rule.

We study the conditional acceptance mechanism in the standard school choice model. Equilibrium sets are nested in the cap and coincide with the stable set for kappa ≤ 2, but unstable equilibrium outcomes can arise for any kappa ≥ 3. Our main results compare conditional acceptance with deferred acceptance: Ergin-acyclicity is sufficient but not necessary for conditional acceptance to implement the set of stable matchings.

In 55 countries, volunteer, non-remunerated donations meet less than 22% of blood demand, necessitating replacement donor programs. Current first-come, first-served practices are inefficient. We introduce feasible schedule menus — a novel concept that enables flexible exchange rates — and propose weighted utilitarian mechanisms. Under suitable menus, these mechanisms are incentive compatible for donor revelation; priority mechanisms also ensure incentive compatibility for utility function revelation.

We experimentally examine to what extent non-truthtelling in strategy-proof matching mechanisms can be attributed to a failure of contingent thinking (FCT). When we remove the need to think contingently, we find no increase in truthtelling. If instead we inform applicants of contingency probabilities, removing the need to think contingently increases truthtelling. We conjecture non-truthtelling is instead caused by non-standard preferences or beliefs.

We study stability in matching markets with two-sided incomplete information. We first introduce a notion of stability based only on iterative reasoning about rational blocking behavior in the absence of communication. We then develop a refinement that allows for self-enforcing blocking agreements. We discuss the relationship between these two notions and their implications for matching under incomplete information.

We investigate a T-period version of the classic static one-to-one marriage problem proposed by Gale and Shapley (1962), allowing agents’ preferences to change over time based on their sequence of past partners. Agents contemplate blocking with the worst ‘credible’ outcome in mind after the block. Credible means that no continuation of a blocking can itself be blocked by the same criterion from the next period onwards. We also impose additional novel restrictions based on the timing of blocking, which determines how agents outside the blocking pair can respond in the period of blocking. We propose dynamic notions of a matching being stable, considering the worst-case scenario over current responses compatible with the timing of blocking and credible continuation matchings. We study conditions under which such a stable matching exists. Furthermore, we link these stable matchings to equilibrium outcomes of suitably defined non-cooperative games.

Allocating public budgets to incentivize performance is difficult when absolute performance levels cannot be compared across heterogeneous units. This paper develops an axiomatic framework for budget allocation based on performance trajectories. We characterize families of rules that satisfy time consistency together with invariance conditions, and show how they govern treatment of historical performance and status-quo allocations.

India's affirmative action system — the world's largest — has been repeatedly reviewed by the Supreme Court of India. We formalize the legal principles established in Indra Sawhney (1992) and Ashoka Kumar Thakur (2008) as axioms and derive the corresponding allocation procedures. Our assignment rule eta* implements the Justice Bhandari remedy by reallocating unfilled OBC positions according to merit while holding the cutoff fixed, and merit-dominates every assignment rule consistent with the governing legal principles.

We study strategic application behavior in centralized university admissions. Using administrative data from a Chinese province, we document that students strategically submit portfolios of choices. Exploiting two sequential institutional reforms — an expansion in the number of choices and a shift from pre-exam to post-score preference submission — we find that the expansion shifts female students towards less selective first-choice universities, while post-score submission substantially improves targeting precision.

Many markets are regulated for paternalistic reasons. We develop a mechanism-design framework that derives the optimal policy mix among price regulation, quantity controls, and prohibition. A planner allocates trade between privately informed buyers and sellers while disagreeing with their valuations. When the planner’s assessment preserves the ranking of agents, price instruments (e.g., sin taxes) are optimal. When willingness to trade instead signals bias, optimal regulation can restrict quantities or even prohibit transactions despite gains from trade. Behavioral interventions (e.g., nudges) offer two benefits: they improve selection into markets, and raise demand or supply for goods that are insufficiently traded.

We study social choice on network-constrained dichotomous domains, in which agents' approved alternatives are restricted to form a connected subset of an exogenously given network over alternatives. This restriction nests approval voting as the special case of a complete network, while capturing settings in which an agent's favored alternatives must be structurally clustered - adjacent public projects, compatible products in a retail network, or related roles within an organization. Our main result characterizes exactly which networks admit a social choice function that is onto and pairwise strategy-proof - a weakening of group strategy-proofness to coalitions of size at most two: such a function exists if and only if the network is a path graph. On path-graph networks, equivalently the single-plateaued dichotomous domain, we obtain a full characterization: every onto, pairwise strategy-proof social choice function is either a Left-Biased Rule or a Right-Biased Rule, two families that always select a commonly approved alternative when one exists and otherwise resolve conflict by favoring, respectively, agents positioned toward one end of the underlying order.

The Central Educational Institutions (Reservation in Teachers' Cadre) Act, 2019 provides for reserving teaching vacancies in India's central educational institutions for beneficiaries of its affirmative action policy. Reservation of teaching vacancies had been a contentious issue, and the act was introduced to resolve it after the Supreme Court's solution was met with protests from the Teachers' Union. Our paper demonstrates an impossibility result in the Supreme Court's solution and the act, which are flawed in reserving seats simultaneously at both the university and within its departments. To overcome this impossibility, we propose an alternative solution based on approximate implementation of fractional assignments, offering a promising middle-ground between the two disputed solutions practiced in India. This novel application demonstrates the practical relevance of the approximate implementation approach (Akbarpourand Nikzad(2020)) beyond the constraint structures examined in the literature.