When Fairness and Truthfulness Point to the Same Allocation Rule

An indivisible object creates an immediate fairness problem: if two people both want it, giving it to either person leaves the other envious. Rather than demanding an impossible outcome, allocation researchers ask which weaker fairness guarantees can coexist with efficiency and honest reporting.

In the November 3, 2026 version of their research preprint, Florian Brandl, Warut Suksompong and Nicholas Teh study this question under a restrictive but simple preference model. Each person assigns every object either zero or one unit of value, and a bundle’s value is the sum of its items’ values. People therefore report approvals, not different degrees of enthusiasm.

The authors’ contribution is a mathematical characterization: particular combinations of requirements force an allocation rule to maximize Nash welfare. This is not an empirical test of how people divide possessions.

Why the objective needs explaining

Maximum Nash welfare is commonly described as maximizing the product of people’s utilities. But a plain product becomes uninformative when someone must receive zero. The paper first maximizes the number of people receiving positive value, then maximizes the product of those positive values.

Under binary additive preferences, this objective coincides with leximin, which maximizes the lowest utility, then the second-lowest, and so on. The paper also notes its equivalence to additive welfare objectives using strictly concave functions. The results therefore do not establish a victory for Nash welfare over leximin: in this setting, they describe the same allocation principle.

Nor do fairness and efficiency alone identify that principle. The paper illustrates this with two people and eight goods. One approves the first four; the other approves the last six, with two overlapping approvals. Giving each person four approved goods maximizes Nash welfare. Yet giving the first person only the two exclusively approved goods and the second person the remaining six is also envy-free and Pareto-optimal. Neither person prefers the other’s bundle, and nobody can be helped without hurting someone else.

The characterization needs requirements that connect decisions across different preference reports and instances.

Five requirements that force Nash welfare

For any fixed number of people of at least two, the authors prove that a rule satisfying five properties must always return a maximum Nash welfare allocation:

• Envy-freeness up to one good: any envy disappears if a suitable single item is removed from the envied person’s bundle. This is a comparison, not a requirement to remove the item in practice.
• Strategyproofness: nobody can obtain strictly greater true value by misreporting their own approvals.
• Neutrality: renaming objects does not change each person’s received value.
• Minimal completeness: every object approved by someone is allocated, while objects approved by nobody remain unallocated.
• Invariance under disapproving unassigned goods: if someone withdraws approval for an object they did not receive, the entire allocation remains unchanged.

The last condition concerns an object unassigned to that particular person, not necessarily an object left outside everyone’s bundles.

Together, minimal completeness and this invariance condition imply Pareto-optimality. They prevent a valued object from being assigned to somebody who does not value it.

The authors also construct two-person counterexamples showing that dropping any one of these five requirements permits a rule that satisfies the other four but does not maximize Nash welfare.

Two people, another route—and a tie-breaking caveat

For two people, the paper offers an alternative characterization. It replaces the invariance condition with non-redundancy—every allocated object goes to someone who approves it—and resource-monotonicity: adding an object must not reduce anyone’s utility.

Tie-breaking remains important. The results force the welfare objective, not one uniquely specified assignment of individual objects. With two people, the first characterization additionally requires a consistent favored person across relevant ties with the same total number of goods and the same number of valued goods. The alternative characterization requires consistency for a given number of valued goods, regardless of how many unwanted goods are present.

The independence claim for the alternative result has a qualification: removing resource-monotonicity admits a counterexample that still maximizes Nash welfare but violates this stronger tie-breaking requirement. The paper does not show that removing this condition alone permits a different welfare objective.

The boundary of the findings

The binary additive assumption is central. It excludes differing strengths of preference and interactions between objects. The paper notes that, with general additive values, Nash welfare can lose strategyproofness and resource-monotonicity, and its computation becomes difficult.

Extending the alternative characterization beyond two people remains open. Within the binary model, however, the results clarify exactly how fairness, reporting incentives and consistency requirements can constrain allocation rules—not merely individual outcomes.