On the stability of utilitarian aggregation

Leandro Nascimento · 2025

Read On the stability of utilitarian aggregation by Leandro Nascimento (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-17061.txt.

Introduction

Leandro Nascimento asks how close a group choice rule stays to a simple weighted sum when fairness rules hold only roughly, in a paper from two thousand twenty-five. The setting joins personal value scores into one group score. A classic theorem says exact fairness forces the weighted sum form.

Nascimento shows that small breaks from the fairness rules mark rules that sit near the weighted sum form. The gap from the sum is at most half the size of the fairness break. In plain terms, near fairness means near the classic result, with a clear bound. Readers in social choice get proof that the classic theorem is stable, not brittle.

Catalog record

CREATORLeandro Nascimento
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE110 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-17061

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On the stability of utilitarian aggregation Leandro Nascimento Abstract In the context of aggregating von Neumann-Morgenstern utilities, we show that bounded violations of the Pareto conditions characterize aggregation rules that are approximately utilitarian. When a single utility function is intended to represent the preference judgments of a group of individuals and the Pareto principles are nearly satisfied, we prove that its distance from a weighted sum of individual cardinal utilities does not exceed half of the positive parameter that differentiates our weaker versions of the Pareto conditions from their conventional forms. This result suggests the stability of Harsanyi ’s ( 1955 ) aggregation theorem, in that small deviations from the Pareto principles lead to aggregation rules that remain close to utilitarian aggregation. Keywords : aggregation of preferences, Pareto unanimity, utilitarianism. JEL classification : D70, D81. 1 Introduction Harsanyi ’s ( 1955 ) aggregation theorem relates a utility function u_{0} , intended to represent the collective judgment of a group of N individuals, to the utilities u_{i} of these individuals. In a setting with von Neumann-Morgenstern (vN-M) utilities over alternatives x , Harsanyi’s theorem gives conditions that characterize the equality between u_{0} and some utilitarian aggregation function \displaystyle w(x)=\sum_{i=1}^{N}a_{i}u_{i}(x)+b, (1) by showing that u_{0} is a utilitarian aggregation of the individual utilities u_{i} when they satisfy a suitable Pareto unanimity principle. While the standard Pareto principles characterize precise forms of utilitarian aggregation, economic models of behavior are only approximations of reality, and we should not expect their assumptions to hold exactly. In fact, many theoretical results in individual and collective decision-making rely on exact conditions that are false in many significant situations. This raises an important question about the extent to which the conclusions of these results can be preserved when their underlying assumptions are modified or weakened. For example, a decision maker (DM) seeking to provide a social assessment of competing alternatives may not enforce Pareto unanimity in every comparison, instead accepting alternatives that are merely near-optimal. Hence, when the precise conditions required by the Pareto unanimity principles are not met – and so exact utilitarian aggregation does not obtain – we seek conditions under which the strong conclusion of linear aggregation still holds, at least approximately. This paper shows that the conclusions of Harsanyi’s aggregation theorem hold approximately when its assumptions are nearly satisfied. Our main goal is to define and characterize the notion of approximate utilitarian aggregation within an expected utility framework. We study how u_{0} , which we refer to here as the DM’s utility rather than a social or group utility function, relates to individual utilities u_{i} under weaker variants of the standard Pareto conditions. These variants relax the standard Pareto constraints, requiring only that the DM’s choices be near-optimal rather than fully consistent with unanimous preferences. Allowing this form of satisficing behavior introduces flexibility into the aggregation procedure and permits us to quantify departures from utilitarianism. We show that the extent to which the Pareto principles are violated bounds the distance from u_{0} to the set of utilitarian aggregators. Moreover, in many important settings, we obtain an exact quantitative relationship between this distance and the degree of violation of the Pareto principles. More specifically, under our near Pareto unanimity conditions, we prove that, rather than u_{0} being identical to a utilitarian aggregator w as in ( 1 ), a residual term r(x) arises, so that \displaystyle u_{0}(x)=w(x)+r(x)\quad\text{for all }x. (2) The magnitude of r(x) in ( 2 ) is at most half of the positive parameter \epsilon that bounds the extent to which the standard Pareto unanimity conditions are violated. Likewise, we show that our weaker Pareto principles are equivalent to a representation of the form \displaystyle u_{0}(x)=\sum_{i=1}^{N}a_{i}u_{i}(x)+e(x)\quad\text{for all }x, (3) with e denoting a function whose oscillation does not exceed that same \epsilon . This additional structure on the size of the residual term r in ( 2 ) and, equivalently, on the oscillation of e in ( 3 ) yields a precise upper bound on the distance of the DM’s utility function to the set of utilitarian aggregators w . When the alternatives x belong to certain domains that are common in economic applications, and the utility functions are sufficiently well-behaved, we exactly determine the distance of a non-utilitarian u_{0} to the class of utilitarian aggregators w by means of the violations of the Pareto conditions. To illustrate, our version of the Semistrong Pareto principle requires that \displaystyle u_{0}(x)\geq u_{0}(y)-\epsilon\quad\text{ when }\quad u_{i}(x)\geq u_{i}(y)\text{ for all }i=1,\dots,N. (4) For interpretation, if alternative x is an optimal choice from \{x,y\} for each individual i , then the DM considers x satisficing, meaning that x is \epsilon -optimal over the set \{x,y\} in the sense of Radner (1980) . Note that in ( 4 ) we can have a preference for y over x for the DM, but this violation of the standard Pareto unanimity principle is bounded, or constrained, by the parameter \epsilon in our cardinal setting. Hence, we allow some flexibility in the DM’s preference for x over y . Rather than requiring a preference for alternative x , the condition on the DM’s preferences merely requires that x should not be significantly inferior to y . More important, our results establish the equivalence between \epsilon -Pareto conditions as in ( 4 ) and representations of u_{0} as an approximate utilitarian aggregator like in ( 2 ) with \left\lvert r(x)\right\rvert\leq\frac{\epsilon}{2} for every alternative x , and in ( 3 ) with e(x)-e(y)\leq\epsilon for every pair of alternatives x and y . The characterization of the \epsilon -Pareto axiom just described resembles a quantitative version of Harsanyi’s theorem on utilitarian aggregation and, in a sense, demonstrates its stability. Specifically, small violations of the Pareto principle yield an aggregation rule that approaches linear aggregation. For settings with vN-M utilities that are common in many applications, we also establish in this paper a duality formula that connects both the norm of the residual term r in ( 2 ) and the oscillation of the function e appearing in ( 3 ) to violations of the standard Semistrong Pareto principle. In particular, the norm of the function r and the oscillation of e vanish precisely when the right-hand sides of ( 2 ) and ( 3 ) define a utilitarian aggregation function. At the same time, our measure of the violations of Semistrong Pareto equals zero in precisely this case. When approximate utilitarian aggregation is viewed as a quantitative version of Harsanyi’s aggregation theorem, this paper contributes to a significant literature that examines the extent to which a given object deviates from satisfying certain ideal assumptions and, consequently, from taking a specific form. One early example, in the context of the existence of macro production functions, is Fisher (1969) . He investigates the regularity conditions a function must satisfy to be close to a mapping in which two or more arguments are grouped into a single aggregate. 1 1 1 Fisher’s question is addressed in further detail in Mak (1988) , who gives a sharper quantitative result linking the degree of near separability of a function to its deviation from a truly separable mapping. More recently, results of a similar kind have appeared in Mossel ’s (

Read the complete text