Research


Publications

AGV mechanism balances the budget with minimum total transfer variance (Journal of Mathematical Economics, 2025)

Two mechanisms are equivalent if they share the same allocation rules and interim transfer rules, carrying incentive properties from one mechanism to the other. I prove the generalized d’Aspremont-Gerard-Varet (AGV) mechanism achieves minimal liquidity risk. It uniquely minimizes total transfer variance among all ex post budget-balanced mechanisms equivalent to an initial mechanism that balances the budget in expectation. Motivated by the contrast between ex ante and ex post fairness, I characterize the optimal mechanism minimizing payoff volatility. Finally, I show that in single-item auctions, a first-price auction with the optimal reserve price jointly maximizes expected revenue and minimizes revenue variance.

Robust lower bounds on monopoly profit with α-concave demand (Economics Letters, 2024)

I extend Condorelli’s lower bound on monopoly profit from log-concave demand to a broader class of α-concave demand, with α = 0 corresponding to log-concavity and α = 1 to concavity. The monopoly profit is at least 1 /(1+α)^(1/α) of the area under the demand curve. I further derive upper bounds for consumer surplus and deadweight loss relative to monopoly profit and show all three bounds are sharp.

Other publications

Geometry of Gains: How Competitive Markets Keep Everyone on Board (Mathematical Intelligencer, 2025)

Under revision

A demand-and-supply approach to markets, core stability, and mechanism design (here, conditionally accepted at Journal of Economic Education)

This paper uses demand-and-supply diagrams to study a production economy with a numeraire, providing elementary proofs to fundamental economic results. We demonstrate that the competitive outcome lies within the core of the economy’s market game and illustrate core convergence to the competitive equilibrium when the economy expands. We further explain the impossibility of efficient trade under private preferences and incentive constraints. Our unified graphical approach makes economic theory more intuitive and easier to teach.

Working Papers

Stability by design (here)

Platforms thrive only if on-platform trading is stable and leaves no gains to off-platform breakouts. I study mechanism design for two-sided platforms with many buyers and sellers who possess private information, act strategically under the platform’s rules, and can collectively defect to contract off-platform. A \emph{core-stable mechanism} is one whose equilibrium is immune to such breakouts: no coalition can exit to form a profitable trading protocol. Private information, interim incentive constraints, and budget balance make the market game nontransferable utility. I show this game is cardinally balanced, meaning coalitions tend to coalesce to create value rather than fragment. Constructively, the \emph{equally weighted second-best mechanism} lies in the inner core and uniquely survives replication; it remains in the inner core of every replica market. As market thickens, its equilibrium payoffs converge to the competitive payoffs of the corresponding large market.

Dynamic “Stackelberg” mechanisms

We refine static “Stackelberg’” mechanisms for efficient allocation with transfers and then extend them to dynamic environments. We propose two static designs that achieve ex post efficiency and ex post budget balance while easing informational demands and strengthening incentives. The first mechanism gives dominant strategies for n-1 agents and makes the remaining agent near-dominant (a near‑VCG property), yet it requires only the marginal independent type distribution of a well-identified player rather than the full prior. The construction extends to fully correlated types with conditional independence structure. The second, under mutual independence, is partially collusion-proof: no coalition that includes the Bayesian-incentive compatible agent can profitably deviate. Both mechanisms generalize to dynamic settings, yielding simple, practical designs when one participant is well understood or when collusion forms around a pivotal agent.

(In)complete information bargaining (here)

Job market paper -This paper develops a two-stage bilateral trade model unifying complete and incomplete information approaches to bargaining. At the ex ante stage, the agents’ selection of trading protocol is modeled as a cooperative game over all incentive compatible, individually rational and budget balanced mechanisms. The interaction between the agents, once private information is realized, is modeled as a game of incomplete information resulting from the mechanism selected. I characterize the set of equilibrium payoffs both directly and indirectly via linear inequalities, then establish general comparative statics in relation to changes in type distributions. Sanctions, k-double auctions bargaining, and mechanism refinement with stricter behavioral conditions are also studied.

The simple economics of the impossibility of efficient trade (here)

The Vickrey-Clarke-Groves (VCG) and price mechanisms both maximize surplus from trades, the former using dominant strategy equilibrium and the latter relying on competitive equilibrium. This paper compares these equilibria in a twosided market with multiple goods, continuous quantities, and quasilinear payoffs. Under VCG, buyers pay less and sellers earn more than in competitive payments, with the difference captured by Harberger triangles. The impossibility of efficient trade follows as the VCG deficit equals the sum of Harberger triangles. Also, VCG payments can be formulated as aggregates of constrained competitive prices, or marginal externalities. The link between Shapley’s inequality and Harberger triangles suggests horizontal mergers worsen the deficit. As the market thickens, VCG payments approach competitive payments while the deficit converges.

Works in Progress

Efficient estimation of income inequality using coarse data (based on ANU 2018 Honours Thesis, here)

In this paper, I developed a new econometrics framework to estimate the size distribution of income and obtain valid standard errors using grouped data. Data acquired are often available in grouped formats: proportions, quantiles, interquantile means. These three types of income data are common and more accessible, while microdata are difficult to obtain due to high costs or confidential reasons. I introduce the GMM-based method of matching to solve the estimation problems involving three data types. I prove consistency and asymptotic normality properties of the estimator and establish asymptotic efficiency results relative to the Cramer-Rao lower bound. Finally, applications to Australian income data using new estimation techniques are considered.