📈 economics

On the construction and representation of social welfare orders satisfying consequentialist equity axioms

This paper investigates the constructive nature of social welfare orders on infinite utility streams that satisfy equity axioms such as Strong Equity, Hammond Equity, or the Pigou-Dalton transfer principle, demonstrating that while explicit lexicographic descriptions exist when the utility set is well-ordered, the existence of such orders on general domains necessitates nonconstructive set-theoretic assumptions like the existence of a non-Ramsey collection.

Original authors: Ram Sewak Dubey

Published 2026-06-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ram Sewak Dubey

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 cosmic referee tasked with ranking the happiness of an infinite line of future generations. You have a list of utility streams (like a scorecard for every generation, stretching forever into the future). Your job is to decide which list is "better" than another, but you must follow strict rules of fairness.

This paper by Ram Sewak Dubey is a deep dive into the question: Can we build a fair ranking system for infinite futures using clear, step-by-step instructions (constructive methods), or do we have to rely on magical, invisible tools (non-constructive methods) that we can't actually write down?

Here is the breakdown of the paper's findings using simple analogies.

The Three Rules of Fairness

The paper focuses on three specific "fairness" rules that a good ranking system should follow:

Strong Equity: If you have two scenarios where everyone is the same except for two generations, and in one scenario the "poor" generation gets a boost while the "rich" generation takes a hit (but stays richer than the poor one), the system must prefer the scenario where the poor got the boost.
Hammond Equity: A slightly weaker version of the above.
Pigou-Dalton Transfer: If you take a tiny bit of happiness from a rich person and give it to a poor person (without making the poor person richer than the rich one), the system should prefer this new arrangement.

The Big Conflict: "Building" vs. "Proving It Exists"

The paper distinguishes between two ways of finding a solution:

Construction (The Blueprint): You can write down a specific algorithm or recipe that anyone can follow to compare any two lists and get a result. This is like building a house with a clear set of blueprints.
Representation (The Magic Proof): You can prove that a ranking must exist using abstract math (often relying on the "Axiom of Choice," which is like a magic wand that says "a solution exists somewhere, even if we can't find it"). This is like saying, "There is a perfect house somewhere in the universe," without being able to tell anyone where it is or how to build it.

The Good News: When We Can Build It

The paper finds that if the possible levels of happiness (the "domain") are well-ordered (meaning they are arranged like a ladder where you can always point to the "next" rung, like the natural numbers 1, 2, 3...), we can build a fair ranking system.

The Analogy: Imagine the happiness levels are like steps on a staircase. The author creates a special "lexicographic" (dictionary-style) system. Instead of just looking at the first generation, then the second, this system looks at the thresholds of happiness. It converts every infinite stream of happiness into a long string of binary codes (0s and 1s) based on whether a generation's happiness is above or below certain "threshold" rungs.
The Result: By comparing these binary codes like words in a dictionary, the author creates a perfect, explicit ranking system that satisfies all the fairness rules. No magic wands needed.

The Bad News: When We Need Magic (Non-Constructive)

The paper also finds that if the happiness levels are arranged in a descending order (like negative integers: -1, -2, -3... going down forever) or have a specific "integer-like" structure, we hit a wall.

The Analogy: Imagine trying to rank infinite streams where the happiness levels are like a staircase going down into an endless basement. The paper proves that if you try to build a fair ranking system here, you are forced to create a "Non-Ramsey Set."
What is a Non-Ramsey Set? Think of it as a "chaos collection." It's a group of infinite subsets of numbers that is so jumbled and complex that it cannot be constructed by any logical, step-by-step rule. It's a mathematical object that exists only if you use the "Axiom of Choice" (the magic wand).
The Conclusion: If your happiness domain looks like a descending staircase, you cannot write down a recipe for a fair ranking. You can only prove one exists using non-constructive magic. This means a real-world policy maker could never actually use such a system to make decisions because they couldn't calculate the result.

The Representation Trap (Real Numbers vs. Rankings)

The paper also tackles a common misconception: "If we can rank things, can we just give them a score (a real number)?"

The Finding: For some domains (like the interval [0, 1]), it is impossible to assign a single real number (a score) to every infinite stream while satisfying the fairness rules.
The Twist: However, even if you can't give them a score, you might still be able to rank them explicitly (as shown in the "Good News" section above).
The Takeaway: Just because you can't write a formula that outputs a number (like 85.4), doesn't mean you can't have a clear rule for deciding which is better. The "score" and the "ranking rule" are different things.

Summary of the Paper's Claims

Constructive Success: If your universe of happiness is "well-ordered" (like a ladder going up), we can explicitly build a fair ranking system using a clever dictionary-style code.
Constructive Failure: If your universe of happiness has a "descending" structure (like a ladder going down forever), building a fair ranking system is impossible without relying on non-constructive mathematical magic (Non-Ramsey sets).
Representation Limits: You cannot always turn these fair rankings into simple real-number scores, especially on complex domains like [0, 1].
The Gap: The set of worlds where we can build a ranking system is different from the set of worlds where we can assign a score to every outcome. They don't always match up.

In short, the paper maps out exactly where we can build a fair, step-by-step guide for judging the future, and where we are forced to admit that such a guide is mathematically impossible to write down.

Technical Summary: On the Construction and Representation of Social Welfare Orders Satisfying Consequentialist Equity Axioms

Problem Statement
The paper addresses the fundamental tension between the existence and the explicit construction of Social Welfare Orders (SWOs) on infinite utility streams ( $X = Y^{\mathbb{N}}$ ) that satisfy consequentialist equity axioms. Specifically, it investigates whether SWOs satisfying Strong Equity (SE), Hammond Equity (HE), or the Pigou–Dalton (PD) transfer principle can be explicitly constructed without relying on non-constructive set-theoretic devices, such as the Axiom of Choice (AC) or the existence of non-Ramsey sets.

While previous literature (e.g., Bossert et al., 2007) established the existence of such SWOs using non-constructive methods (Szpilrajn's Lemma), and recent work (Lauwers, 2010; Zame, 2007) confirmed that non-constructive devices are necessary for existence in certain contexts, the precise boundary between domains admitting explicit constructions and those requiring non-constructive objects remains to be fully delineated. Furthermore, the paper seeks to clarify the relationship between the existence of a real-valued Social Welfare Function (SWF) representation and the existence of an explicitly defined SWO.

Methodology
The author employs a dual approach combining constructive order theory and set-theoretic impossibility proofs:

Constructive Approach: For specific domains, the paper constructs explicit SWOs using lexicographic preference relations. Crucially, these constructions do not apply lexicographic ordering directly to generations (which would violate equity) but rather map utility streams into binary threshold-utility codes. These codes are then ordered lexicographically over a well-ordered index set derived from the utility domain $Y$ .
Non-Constructive/Impossibility Approach: To establish the necessity of non-constructive methods, the paper utilizes the concept of non-Ramsey collections. By constructing specific utility streams based on subsets of natural numbers and applying the equity axioms, the author demonstrates that the existence of an SWO satisfying these axioms on certain "rich" domains (specifically those containing subsets of order type $\omega^*$ or $\mu$ ) implies the existence of a non-Ramsey set. Since non-Ramsey sets are non-constructive objects (their existence relies on AC), this proves that no explicit SWO can exist on such domains.
Representation Analysis: The paper distinguishes between the existence of an SWF (a real-valued map) and an SWO (a binary relation). It employs separability arguments (e.g., constructing uncountable disjoint intervals) to prove non-representability results on specific domains like $[0, 1]$ .

Key Contributions and Results

1. Positive Construction Results (Well-Ordered Domains)

Theorem 1: If the utility domain $Y$ is well-ordered by the usual order, there exists an explicitly defined SWO on $X = Y^{\mathbb{N}}$ satisfying Strong Pareto (SP), Strong Equity (SE), Hammond Equity (HE), and the Pigou–Dalton (PD) principle.
Mechanism: The construction maps streams to binary codes based on thresholds in $\bar{Y} = \{t \in Y : \exists s \in Y, t < s\}$ . The ordering is lexicographic over the product space $\bar{Y} \times \mathbb{N}$ , prioritizing lower thresholds and then generations. This method preserves equity-sensitive comparisons while maintaining completeness and transitivity.
Extension: The paper also provides explicit SWO constructions for SE alone under weaker conditions: if the set of potential "pre-transfer poor" utility levels $L(Y)$ is well-ordered, or if the set of "pre-transfer rich" levels $U(Y)$ is reverse well-ordered.

2. Non-Constructive Implications (Rich Descending Domains)

Theorem 2: The existence of an SWO satisfying SE and Monotonicity (M) on a domain $Y$ containing a subset of order type $\omega^*$ (e.g., negative integers) entails the existence of a non-Ramsey collection.
Theorem 2(b): Similarly, for the PD principle combined with Monotonicity, if $Y$ contains sequences with a constant gap descending structure, the existence of an SWO implies a non-Ramsey collection.
Proposition 4: For domains of order type $\mu$ (containing an integer-like subset), the existence of an SWO satisfying SE alone (without efficiency axioms) also entails the existence of a non-Ramsey collection. This identifies a clear class of domains where SE rankings are inherently non-constructive.

3. Representation vs. Construction

Proposition 1: If $Y$ is countable, there exists a real-valued SWF satisfying SE. However, the proof relies on the Szpilrajn Lemma (extension of partial orders), making it non-constructive in general.
Proposition 2: If $Y = [0, 1]$ , no real-valued SWF satisfying SE exists. This impossibility arises from SE alone, independent of efficiency axioms, due to the inability to embed the strict ordering of SE into the real line (via a separability argument).
Distinction: The paper highlights that the domain admitting an explicit SWO construction (e.g., well-ordered $Y$ ) is not necessarily the same as the domain admitting a real-valued SWF representation. Conversely, the existence of an SWF on a countable domain does not guarantee an explicit construction of the underlying SWO.

Significance and Claims
The paper claims to refine the understanding of the "correspondence" between representation and construction in intergenerational equity.

Reformulation of Correspondence: It argues that earlier literature suggesting a coincidence between domains admitting real-valued SWFs and those admitting constructible SWOs is incorrect for consequentialist equity axioms. The domains are axiom-specific and depend on the order-theoretic structure of $Y$ .
Explicit vs. Non-Explicit: The results demonstrate that explicit SWO constructions are possible on well-ordered domains even where real-valued representations fail (e.g., on $Y=\mathbb{N}$ with SE+SP). Conversely, on integer-like domains, the existence of any SWO (constructible or not) is tied to non-constructive set theory.
Policy Relevance: The paper notes that if an SWO is shown to exist only via non-constructive methods (AC), it remains an open question whether a social planner can explicitly rank pairs of utility streams. The constructive results provide a framework for explicit ranking on well-ordered domains, while the non-constructive results delineate the limits of such explicit policy tools.

In summary, the paper establishes that while equity axioms like SE, HE, and PD can be satisfied by explicitly constructed SWOs on well-ordered domains, their satisfaction on domains with descending integer-like structures necessitates non-constructive mathematical objects, thereby limiting the practical implementability of such social welfare orders in those specific contexts.