📈 economics

Intensity-Efficient and Intensity-Positional Allocations

This paper introduces a framework for ordinal intensity comparisons to define and analyze "intensity-efficient" and "intensity-positional" allocations, distinguishing them from traditional utilitarian or Borda-based approaches in environments where agents can rank the strength of their preferences without quantifying them.

Original authors: Georgios Gerasimou

Published 2026-03-03

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Georgios Gerasimou

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a teacher trying to hand out three different prizes (a gold medal, a silver medal, and a participation ribbon) to three students.

The Old Way (Standard Economics):
Usually, we just ask the students: "What do you want?"
Student A says: "Gold > Silver > Ribbon."
Student B says: "Gold > Silver > Ribbon."
Student C says: "Gold > Silver > Ribbon."

If we just look at these rankings, there are many "fair" ways to give out the prizes where no one can be made happier without making someone else unhappy. This is called Pareto Efficiency. But it's a bit of a mess. Maybe Student A really hates the ribbon and would cry if they got it, while Student B is just "okay" with it. Standard economics can't see that difference; it only sees that they both prefer Gold.

The Problem:
How do we decide who gets the Gold if we can't ask, "How much do you want it?" (because people are bad at giving numbers like "I want it 8.5 out of 10"). We can only ask, "Do you want the Gold more than you want the Silver?"

The New Idea (This Paper):
This paper introduces a new way to look at the problem. It says: "Let's not just look at what you prefer, but how much you prefer it relative to other things."

Think of it like a Taste Test.

Student A says: "I love Gold way more than Silver, but I barely care about the difference between Silver and Ribbon."
Student B says: "I love Gold more than Silver, but I really hate the Ribbon compared to Silver."

Even though we don't have numbers, we can compare the intensity of their feelings by looking at the "rank" of their preferences.

The Two New Rules

The author proposes two new ways to solve the puzzle:

1. The "Intensity-Efficient" Rule (The Fair Swap)

Imagine you have two students, Alice and Bob, and two items, a Pizza and a Burger.

Alice gets the Burger, Bob gets the Pizza.
They both prefer Pizza over Burger.
The Question: Who really wants the Pizza more?

If we look at their "intensity rankings," we might see that Alice's love for Pizza over Burger is ranked as her top intensity difference, while Bob's love for Pizza over Burger is only his third biggest intensity difference.

The Rule: If Alice cares much more about the Pizza than Bob does, Alice should get the Pizza. If we give the Pizza to Bob, we are "intensity-dominated" (we made a mistake). We should only swap items if the person who cares more gets the item.

The Catch: Sometimes, these rules get stuck in a loop. Imagine Alice wants the Pizza more than Bob, Bob wants the Burger more than Charlie, and Charlie wants the Pizza more than Alice. It's like a game of Rock-Paper-Scissors. In these cases, a perfect "Intensity-Efficient" solution might not exist.

2. The "Intensity-Positional" Rule (The Scoreboard)

Since the first rule can get stuck in loops, the author suggests a second, more robust method: The Scoreboard.

Think of this like a video game where you get points not just for winning, but for how hard you fought.

Standard Borda Count (Old Way): If you rank an item 1st, you get 2 points. If you rank it 2nd, you get 1 point. It assumes the gap between 1st and 2nd is the same as 2nd and 3rd.
Intensity-Positional (New Way): We look at your "Intensity Score."
- If you rank Pizza 1st and Burger 2nd, but your intensity ranking shows you care way more about that gap than anyone else, you get a massive bonus.
- We add up everyone's "Intensity Scores" and give the items to the team with the highest total score.

This method always finds a solution, even when the "Fair Swap" rule gets confused. It's like a referee who looks at the whole game's intensity rather than just individual plays.

Why Does This Matter?

The "Harsanyi" vs. "Arrow" Debate:
The paper starts with two famous economists arguing.

Harsanyi said: "We should add up everyone's happiness to find the best outcome." (Utilitarianism).
Arrow said: "Wait, how do you measure happiness? You can't just add up feelings like money!"

This paper finds a middle ground. It says: "We don't need to measure happiness with a ruler (numbers). We just need to know who cares more than whom, based on how they rank their feelings."

The Real-World Analogy:
Imagine you have a ticket to a Mozart concert. You can't go. You have two friends:

Friend A: Loves Mozart, but also loves Jazz. They are happy with either.
Friend B: Hates Jazz and only listens to Mozart. To them, this ticket is a life-changing event.

Standard economics might flip a coin because both "prefer" the ticket to nothing.
This paper says: Give it to Friend B. Why? Because Friend B's "intensity rank" for the ticket is at the very top of their list, while Friend A's is just somewhere in the middle. Friend B's "pain" of missing it is much greater.

The "Mechanism" (How to actually do it)

The paper also figures out how to ask these questions without people lying.

The Trick: You ask simple questions like, "Do you prefer Pizza over Burger more than you prefer Burger over Salad?"
The Result: You can figure out everyone's intensity map in a reasonable amount of time (quadratic time, which is fast for computers).
The Warning: Just like in school elections, if you ask people to report how much they want something, they might lie to get what they want. The paper shows that while we can't stop people from lying about their intensity, we can still build a system that is better than the old "random lottery" systems.

Summary

This paper is about fairness with feelings.
It argues that to be truly fair, we shouldn't just look at what people want, but how much they want it. By using a clever system of ranking "intensity" (how much one preference beats another), we can distribute goods in a way that feels more just, giving the prize to the person who would be most devastated to lose it, even if we can't put a dollar value on their feelings.

1. Problem Statement

The paper addresses the assignment problem (allocating $n$ indivisible goods to $n$ agents) in an environment where agents possess not only ordinal preferences (rankings of items) but also ordinal preference intensities.

The Gap: Standard Pareto-efficiency relies solely on ordinal rankings, often leaving a large set of efficient allocations that may violate intuitive distributive justice.
The Limitation of Existing Solutions: Classical utilitarianism and cardinal-welfarist approaches require cardinal utility (quantifiable intensity differences) and often assume interpersonally comparable utility differences. However, deriving cardinal utility from lottery choices confounds intensity with risk attitudes, and cognitive psychology suggests humans can compare intensity differences (e.g., "I prefer A to B more than C to D") without assigning precise numerical values.
The Objective: To develop allocation criteria that refine Pareto-efficiency using ordinal intensity information (quaternary relations) without requiring cardinal utility or monetary transfers, while establishing the conditions for their existence and operationality.

2. Methodology and Theoretical Framework

A. Measurement-Theoretic Foundations

The author models intensity using quaternary relations ( $\dot{\succsim}_l$ ) on the set of alternatives $A \times A$ .

Representation: An agent's intensity relation is representable by a preference intensity function $s_l: A \times A \to \mathbb{R}$ , unique up to strictly increasing transformation.
Axioms: The framework assumes:
1. Completeness & Reversal: Standard ordering properties.
2. Translocation Consistency: Ensures the intensity relation induces a consistent weak order on preferences.
3. Strictness: Distinct pairs of alternatives have distinct intensity rankings (no non-trivial intensity ties).
Canonical Normalization: The paper introduces a canonical intensity function $s_l$ that maps intensity pairs to a discrete set of integers $\{-k, \dots, k\}$ , where $k = \binom{n}{2}$ . This normalization allows for interpersonal comparisons based on rank positions rather than absolute magnitudes.

B. Interpersonal Comparability

The core innovation is a rank-based criterion for interpersonal comparisons (Assumption 1):

If Agent $i$ ranks the intensity difference between $(a, b)$ higher in their intensity ordering than Agent $j$ ranks the difference between $(c, d)$ , then $i$ is interpreted as having a stronger preference for that specific comparison.
This is analogous to the Borda count (normalizing ordinal utilities) or Relative Utilitarianism (normalizing cardinal utilities), but applied to intensity rankings.

3. Key Concepts and Definitions

A. Intensity-Dominance and Intensity-Efficiency

Intensity-Dominance: Allocation $x$ intensity-dominates $y$ if, for every pair of agents $(i, j)$ where the allocations are "flipped" (i.e., $x$ gives $i$ item $a$ and $j$ item $b$ , while $y$ gives $i$ item $b$ and $j$ item $a$ ), the agent who receives their preferred item in $x$ has a higher intensity rank for that preference than the corresponding agent in $y$ .
Intensity-Efficiency: An allocation is intensity-efficient if it is Pareto-efficient and not intensity-dominated by any other Pareto-efficient allocation.

B. Intensity-Positional Allocations

Intensity-Rank Score ( $r_l(a)$ ): Defined as the maximum rank in an agent's intensity ordering where item $a$ is preferred to some other item. This captures both the number of items ranked below $a$ and the relative intensity of those preferences.
Intensity-Positional Allocation: An allocation that maximizes the aggregate intensity-rank score ( $\sum r_l(y_l)$ ).
Relation to Borda: This concept generalizes the Borda count. While Borda assumes linear intensity gaps (every step down in rank is equal), intensity-positional allocations account for non-linear intensity gaps. They coincide with Borda allocations only when intensity profiles are linear.

4. Key Results

A. Existence of Intensity-Efficient Allocations

$n=3$ : An intensity-efficient allocation always exists. The intensity-dominance relation is acyclic for three agents.
$n \geq 4$ : Existence is not guaranteed. The paper provides a counterexample (Table 2) where the intensity-dominance relation forms a cycle (similar to Condorcet cycles), preventing the existence of a maximal element.
Sufficient Condition (Homeo-Monotonicity): The paper introduces effectively homeo-monotonic profiles. If agents can be partitioned into groups with identical top-preferences, and their intensity rankings over these items are ordered consistently (monotonicity) and preserved across non-consecutive items (homeo-monotonicity), then an intensity-efficient allocation exists.
Necessity: For strictly monotonic top-preference-coincident sets, homeo-monotonicity is a necessary condition for existence.

B. Comparison with Utilitarianism

Distinctness: Intensity-efficient allocations generally diverge from classical utilitarian allocations (which maximize $\sum u_l$ ).
Alignment Condition: They coincide only if the utility profile is balanced, meaning the interpersonal ranking of utility differences perfectly aligns with the interpersonal ranking of intensity differences.

C. Comparison with Borda and Intensity-Positional

Refinement: Intensity-positional allocations refine (are a subset of) Borda allocations.
Existence: Unlike intensity-efficient allocations, intensity-positional allocations always exist because they are defined by a maximization problem over a finite set.
Trade-offs: Intensity-positional allocations may not be intensity-efficient (and vice versa) because they aggregate total scores, potentially ignoring specific pairwise intensity dominance that intensity-efficiency prioritizes.

D. Mechanism Design and Elicitation

Elicitation Complexity: Ordinal intensities can be elicited in $O(n^2)$ time by asking agents to compare pairs of items, leveraging the "lateral consistency" between preferences and intensities.
Algorithm: The author proposes an algorithm extending the Random Priority (RP) mechanism:
1. Run RP to get a Pareto-efficient allocation.
2. Iteratively check for "flipped" pairs where a swap would increase the intensity-dominance score.
3. Perform swaps if they improve intensity efficiency.
Incentive Compatibility: While RP is strategy-proof for preferences, the proposed extension is not strategy-proof for intensities. Agents may misreport intensities to manipulate the outcome in their favor (specifically, to increase the chance of receiving their top choice).

5. Significance and Contributions

New Analytical Domain: The paper establishes a rigorous framework for ordinal intensity in social choice, bridging the gap between purely ordinal (Arrowian) and cardinal (Utilitarian) paradigms.
Operational Refinement: It offers the first operational refinement of Pareto-efficiency that incorporates intensity information without requiring cardinal utility or monetary transfers.
Theoretical Insight: It demonstrates that intensity-based dominance can create cycles (non-existence) in general settings, a phenomenon distinct from standard preference aggregation, and identifies specific structural conditions (homeo-monotonicity) that restore existence.
Generalization of Borda: It introduces the intensity-positional criterion, showing how standard scoring rules can be generalized to handle non-linear intensity differences, providing a more nuanced tool for distributive justice.
Practical Implications: The work highlights the tension between efficiency (intensity-dominance) and incentive compatibility in mechanism design, suggesting that while intensity information improves welfare outcomes, it introduces new strategic vulnerabilities.

In summary, Gerasimou provides a comprehensive theoretical foundation for using ordinal intensity data to improve allocation fairness, defining new efficiency concepts, proving existence theorems under specific structural constraints, and analyzing the trade-offs involved in implementing these concepts via mechanisms.