pith. sign in

arxiv: 2602.15699 · v3 · submitted 2026-02-17 · 💰 econ.GN · q-fin.EC

Understanding Classical Decomposability of Inequality Measures: A Graphical Analysis

Pith reviewed 2026-05-15 22:07 UTC · model grok-4.3

classification 💰 econ.GN q-fin.EC
keywords inequality measuresdecomposabilitygraphical analysisincome-share simplexbetween-group residualTheil indexcoefficient of variationmean log deviation
0
0 comments X

The pith

Inequality decompositions fail in distinct ways across measures, mainly via negative between-group residuals that render results uninterpretable.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper maps classical decomposability conditions for inequality measures onto the geometry of the three-person income-share simplex. This turns abstract weighting rules into visible restrictions on within-group and between-group residual terms. The resulting diagrams show that decomposability is not an all-or-nothing trait; each measure breaks in its own pattern, yet the between-group residual is the most common point of collapse. Negative values appear for the coefficient of variation and Theil index under population-share weighting, and for the mean log deviation under income-share weighting. Stylised numerical cases illustrate how these negatives produce misleading statements about the contribution of group differences to total inequality.

Core claim

Decomposability is not a binary property as measures fail in qualitatively distinct ways, and the between-group residual is consistently the primary locus of failure. Negative between-group residuals render the decomposition uninterpretable and arise for the coefficient of variation and the Theil index under population-share weighting, and for the Mean Log Deviation under income-share weighting.

What carries the argument

A geometric diagnostic framework that represents three-person income distributions as points on the two-dimensional income-share simplex and translates population-share and income-share decomposability into concrete restrictions on within- and between-group residuals.

If this is right

  • The coefficient of variation and Theil index produce negative between-group residuals under population-share weighting.
  • The mean log deviation produces negative between-group residuals under income-share weighting.
  • Negative between-group residuals make standard statements about group contributions to inequality invalid.
  • Applied researchers can use the simplex diagrams to check whether a chosen measure and weighting scheme will yield interpretable results for a given grouping.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Researchers who routinely decompose inequality by groups may need to add a residual-sign check before reporting between-group shares.
  • The geometric approach could be extended to more than three persons by projecting higher-dimensional simplices onto two dimensions for visual inspection.
  • If the between-group residual negativity proves robust, it would suggest preferring measures that avoid this sign problem, such as the Gini coefficient in the cases examined.

Load-bearing premise

That the violations observed in the three-person simplex case are representative of the failures that occur in larger populations and real empirical applications.

What would settle it

A concrete counter-example in a four-or-more-person income distribution where the between-group residual stays non-negative for the coefficient of variation under population-share weighting would show that the simplex failures do not generalise.

read the original abstract

This paper develops a geometric diagnostic framework for classical inequality decomposability. Representing the simplest nontrivial setting of three-person income distributions as points on the two-dimensional income-share simplex, we translate population-share-weighted and income-share-weighted decomposability into concrete geometric restrictions on within- and between-group residuals, making it possible to localise and characterise violations across measures. Applied to the Mean Log Deviation, the Gini coefficient, the coefficient of variation, and the Theil index, the analysis shows that decomposability is not a binary property as measures fail in qualitatively distinct ways, and the between-group residual is consistently the primary locus of failure. Negative between-group residuals render the decomposition uninterpretable and arise for the coefficient of variation and the Theil index under population-share weighting, and for the Mean Log Deviation under income-share weighting. Stylised numerical examples quantify the resulting misinterpretation scenarios for applied researchers.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper develops a geometric diagnostic framework for classical inequality decomposability by representing three-person income distributions as points on the two-dimensional income-share simplex. It translates population-share-weighted and income-share-weighted decomposability definitions into concrete geometric restrictions on within- and between-group residuals, then applies the framework to the Mean Log Deviation, Gini coefficient, coefficient of variation, and Theil index. The analysis concludes that decomposability is not a binary property, that measures fail in qualitatively distinct ways, and that the between-group residual is the primary locus of failure, producing negative values (and thus uninterpretable decompositions) for the coefficient of variation and Theil index under population-share weighting and for the Mean Log Deviation under income-share weighting. Stylised numerical examples quantify the resulting misinterpretation risks for applied work.

Significance. If the geometric restrictions and observed failure patterns are robust, the framework supplies applied researchers with a visual diagnostic for when standard subgroup decompositions of inequality measures become uninterpretable due to negative between-group residuals. This could improve the reliability of empirical analyses that rely on these classical decompositions.

major comments (2)
  1. [§3 (geometric framework) and §4 (applications to measures)] The central claim that negative between-group residuals arise consistently for CV and Theil under population-share weighting (and MLD under income-share weighting) is derived entirely within the three-person simplex. No analytic extension or numerical verification is supplied showing that the same sign patterns survive when the income vector is embedded in higher-dimensional simplices (n=4 or larger), where the between-group term is defined via weighted averages of subgroup means whose sign and magnitude can change with population size even if within-group residuals remain non-negative.
  2. [§4 and stylised examples in §5] The assertion that the between-group residual is 'consistently the primary locus of failure' rests on the n=3 case; the manuscript provides neither a general proof nor counter-examples confirming that this locus remains dominant outside the simplest nontrivial setting, which is load-bearing for the claim that decomposability fails in qualitatively distinct but predictable ways.
minor comments (2)
  1. [§2 and §3] The notation distinguishing population-share versus income-share weighting in the residual definitions could be made more explicit in the equations to prevent reader confusion when moving between the geometric restrictions and the numerical illustrations.
  2. [Figures 2–5] Figure captions for the simplex diagrams would benefit from explicit labels indicating which regions correspond to negative between-group residuals for each measure and weighting scheme.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. We address each major comment below, clarifying the scope of our n=3 geometric analysis while committing to enhancements in the revision.

read point-by-point responses
  1. Referee: [§3 (geometric framework) and §4 (applications to measures)] The central claim that negative between-group residuals arise consistently for CV and Theil under population-share weighting (and MLD under income-share weighting) is derived entirely within the three-person simplex. No analytic extension or numerical verification is supplied showing that the same sign patterns survive when the income vector is embedded in higher-dimensional simplices (n=4 or larger), where the between-group term is defined via weighted averages of subgroup means whose sign and magnitude can change with population size even if within-group residuals remain non-negative.

    Authors: We acknowledge that our derivations and visualizations are specific to the three-person simplex, which provides the lowest-dimensional nontrivial setting for a complete geometric treatment. The sign patterns for negative between-group residuals stem from the functional forms of the inequality measures and the relative ordering of subgroup means, which we expect to hold more generally. To address this, we will include numerical verifications for n=4 and n=5 in the revised version, demonstrating that the negative residuals persist in analogous configurations. This addition will strengthen the applicability of the diagnostic. revision: yes

  2. Referee: [§4 and stylised examples in §5] The assertion that the between-group residual is 'consistently the primary locus of failure' rests on the n=3 case; the manuscript provides neither a general proof nor counter-examples confirming that this locus remains dominant outside the simplest nontrivial setting, which is load-bearing for the claim that decomposability fails in qualitatively distinct but predictable ways.

    Authors: The characterization of the between-group residual as the primary locus is based on our exhaustive mapping within the simplex, where violations occur predominantly there for the measures considered. A full general proof for arbitrary group sizes is not provided, as the geometric method is tailored to n=3. However, we will add a discussion and numerical counter-examples in higher dimensions to illustrate that the qualitative distinctions remain. We believe this supports the claim without overgeneralizing from n=3 alone. revision: partial

Circularity Check

0 steps flagged

No circularity: direct geometric translation of standard decomposability definitions

full rationale

The paper translates classical population-share and income-share weighted decomposability definitions into geometric restrictions on the 2-simplex for three-person distributions. This mapping is performed by direct substitution of the standard formulas for within- and between-group components into the simplex coordinates, without fitted parameters, self-referential equations, or load-bearing self-citations. The identification of negative between-group residuals for CV/Theil (population weighting) and MLD (income weighting) follows immediately from evaluating those formulas at simplex vertices and edges. No uniqueness theorems, ansatzes, or prior results from the same authors are invoked to force the conclusions; the geometric diagnostics are self-contained consequences of the input definitions. Generalization to n>3 is a scope limitation rather than a circularity in the derivation chain.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on the standard geometric representation of income shares and the classical definitions of decomposability; no free parameters, new entities, or ad-hoc axioms are introduced in the abstract.

axioms (1)
  • domain assumption Three-person income distributions can be represented as points on the two-dimensional income-share simplex
    Standard device in inequality measurement for visualizing population and income shares.

pith-pipeline@v0.9.0 · 5446 in / 1171 out tokens · 44676 ms · 2026-05-15T22:07:08.830675+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.