When individual incentives determine a resource-sharing rule

A division problem with joint restrictions

Dividing a fixed total becomes more complicated when feasible shares must satisfy more than a budget constraint. Restrictions might limit the dispersion of allocations or the total received by any group of a specified size. Chen studies constraints that treat agents interchangeably: relabeling the agents must preserve feasibility. This property is called permutation invariance.

The model also requires the feasible set to be nonempty, closed and convex. Closedness includes limits of feasible allocations; convexity means that weighted averages of feasible allocations remain feasible. Every allocation must distribute the same fixed total.

Each agent cares only about her own share and has a continuous, strictly single-peaked preference. Her peak is her uniquely preferred quantity. Below it, more is better; above it, less is better. However, two preferences with the same peak can disagree about which of two quantities on opposite sides of that peak is preferable.

That distinction matters because agents report their complete preferences, not merely their preferred quantities. A mechanism could potentially use those additional comparisons when assigning shares.

What uniform gains does

The uniform-gains rule begins from equal division, which is feasible under the model’s symmetry and convexity assumptions. It considers allocations in which every agent’s share lies between equal division and that agent’s peak.

For each such allocation, the rule measures each agent’s movement from equal division toward her preferred quantity. It then uses the leximin ordering: maximize the smallest movement first, then the next smallest, and continue in that order, subject to feasibility. These gains are changes in quantities, not numerical measures of satisfaction.

The paper studies when specified requirements uniquely select this existing rule, rather than proposing uniform gains as a new mechanism.

Three central requirements concern outcomes and incentives. Efficiency means that no feasible alternative can make everyone weakly better off and someone strictly better off. Symmetry means that relabeling complete preference reports relabels the resulting allocation. Individual strategy-proofness means that an agent cannot obtain a strictly preferred share by misreporting while everyone else’s complete report remains fixed.

Two characterization results

The first theorem covers every finite population of at least two agents and every nonempty closed convex permutation-invariant feasible set within the fixed-total constraint. It adds a fourth requirement: own-coordinate peak-onlyness.

This condition says that changing a report without changing its peak must leave the reporting agent’s own share unchanged. It does not directly require other agents’ shares to remain unchanged. Chen shows that the four requirements together identify the entire uniform-gains allocation uniquely.

The second theorem removes that additional reporting restriction for populations of two through six agents, on every nonempty compact convex permutation-invariant feasible set. In this finite-dimensional setting, compactness means closedness and boundedness. Within that population range and under those feasibility conditions, efficiency, symmetry and individual strategy-proofness alone characterize uniform gains.

Because the identified rule has stronger protection against coordinated manipulation, these characterizations also imply strong group strategy-proofness: no coalition can misreport so that all its members weakly benefit and at least one strictly benefits.

How the proofs recover everyone’s shares

The study is mathematical, not an empirical test. Its central difficulty is that individual strategy-proofness directly constrains only the share of the agent changing her report. The other shares still need to be determined.

With own-coordinate peak-onlyness, Chen derives a median identity that determines an agent’s share from her peak and the allocations associated with two bounding reports. Without that condition, the proof instead examines the shares actually attainable by changing one complete report while holding all others fixed.

Reports at the lowest and highest feasible quantities establish attainable bounds. Symmetry, feasible transfers, the fixed budget and efficiency then identify the remaining shares. An induction introduces interior reports successively, with additional starting configurations needed for five and six agents.

Scope and the unresolved question

A separate corollary characterizes continuous rules using individual rather than coalition incentives on compact feasible sets that are polytopes, or strictly convex with full dimension relative to the fixed-total hyperplane. It does not assert that continuity holds on every feasible set considered.

The work is an arXiv preprint, with no journal publication identified in the supplied record. Its results concern the specified preference domain and symmetric convex constraints. Whether the three economic requirements alone suffice for populations larger than six on arbitrary compact feasible sets in this model remains open.