Lorenz Curve¶

Read a whole distribution's inequality off a single curve by plotting the cumulative share of a quantity against the cumulative share of its ranked holders, so the sag below the perfect-equality diagonal shows how concentrated it is and whether two distributions can be safely ranked.

Core Idea¶

The Lorenz curve, introduced by Max Lorenz in 1905, is a graphical instrument for measuring distributional inequality: it plots the cumulative share of a positive quantity (income, wealth, land, citations) held by the cumulative share of its holders, with units sorted from smallest to largest. The diagonal line from the origin to the upper-right corner represents perfect equality, where each x% of units holds exactly x% of the quantity. The Lorenz curve sags below the diagonal in proportion to the inequality in the distribution — a perfectly equal distribution produces the diagonal itself, while a perfectly unequal one (all quantity held by a single unit) produces a curve that hugs the bottom and right edges. The Gini coefficient, the standard scalar summary of inequality, is defined as twice the area between the Lorenz curve and the diagonal, ranging from 0 (perfect equality) to 1 (maximum concentration).

The curve provides a partial order over distributions: if one Lorenz curve lies entirely above another (is closer to the diagonal at every point), the first distribution is unambiguously less unequal, and every standard inequality measure that satisfies the Pigou-Dalton principle of transfers will agree on the ranking. When two Lorenz curves cross, the ranking depends on which part of the distribution is weighted more heavily, and summary statistics like the Gini can disagree — the curve makes this ambiguity visible rather than hiding it inside a single number.

Because the instrument depends only on cumulative ranked shares, it applies to any positive quantity distributed across units — income, wealth, or land holdings within the inequality-measurement literature. The epidemiological concentration curve (used to measure health-care utilisation relative to income) is a direct structural sibling that relaxes the requirement that units be sorted by the quantity being measured.

Structural Signature¶

Sig role-phrases:

the ranked units — the holders of a positive quantity, sorted from smallest to largest
the cumulative-share axes — cumulative fraction of units (x) against cumulative fraction of the quantity held (y), the unit-square construction
the equality diagonal — the fixed reference line where each x% of units holds exactly x% of the quantity
the sag below the diagonal — the curve's departure from equality, deeper the more concentrated the distribution
the Gini coefficient — the scalar summary, twice the area between curve and diagonal, running 0 (perfect equality) to 1 (maximal concentration)
the dominance partial-order — the engineered guarantee: when one curve lies wholly above another, every transfer-respecting (Pigou-Dalton) measure agrees it is unambiguously less unequal
the crossing caveat (its limitation) — when two curves cross, no unqualified ranking is warranted and a Gini comparison smuggles in an unstated value judgment about whether the bottom or the top matters; the shape must then be read directly ("unequal where?")
the measure-not-phenomenon boundary — the curve displays and quantifies concentration but is not the generative fact producing it (heavy tails, power-law / cumulative-advantage dynamics), so a high Gini describes rather than explains

What It Is Not¶

Not the cause of concentration — a display of it. The Lorenz curve quantifies how concentrated a distribution is; it does not explain why. The generative fact that a few holders command most of the quantity — heavy tails, power-law processes, cumulative advantage — is a separate matter with its own mechanisms. Reading "high Gini" as an explanation rather than a description confuses the lens for the thing seen.
Not a total order over distributions. Inequality is only a partial order: when two Lorenz curves cross, no transfer-respecting measure is entitled to an unqualified ranking, and which distribution is "more unequal" depends on whether one weights the bottom or the top. A clean ranking holds only under dominance (one curve wholly above the other).
Not equivalent to its Gini coefficient. The Gini is a many-to-one summary that discards the curve's shape: two very different societies — one sagging near the top, one near the bottom — can share an identical Gini. Under crossing curves a Gini comparison smuggles in an unstated value judgment, so the scalar is safe to trust only when the curves are nested.
Not a normative verdict. A curve far from the diagonal describes high concentration; it does not by itself say the distribution is unjust or that equality is the right target. The diagonal is a mathematical reference for perfect equality, not a moral baseline the curve endorses.
Not a heavy-tailed distribution. The Lorenz curve applies to any positive quantity distributed across units, heavy-tailed or not; it is a measurement instrument, not a claim about tail shape. A distribution can be measured by a Lorenz curve without being power-law or heavy-tailed, and the curve is the readout, not the structural property of the tail.

Scope of Application¶

Because the Lorenz curve is a measurement instrument, not a mechanism, it is not bounded to one subject matter: it applies literally wherever its single precondition holds — a positive quantity distributed across units rankable from smallest to largest holder. Its home is inequality measurement in economics, but the fields below apply the identical construction (the same cumulative-share axes, equality diagonal, Gini, dominance test, and crossing caveat), not a metaphor; the boundary to respect is instrument-reach versus over-reading (the Gini is a partial order, not total, and the curve displays concentration without explaining it).

Income and wealth inequality (economics) — the canonical use Max Lorenz introduced in 1905: distributions of income, wealth, and land holdings across a population and over time, with the Gini and dominance ordering carrying across those quantities untranslated.
Bibliometrics — cumulative citations plotted against cumulative papers (or authors), quantifying the concentration of scholarly impact in the "80/20" style.
Ecology — cumulative biomass (or abundance) against cumulative species, where the sag below the diagonal measures community evenness.
Computing and networks — load distributed across servers or traffic across links, read as a concentration curve to expose hotspots and imbalance.
Epidemiology and health economics — the close structural sibling, the concentration curve, ranks units by an external variable (typically income) rather than by the measured quantity, to gauge how health-care utilization or disease burden concentrates across socioeconomic strata.

Clarity¶

The Lorenz curve's central clarifying service is to make explicit that inequality is a partial order, not a total one — and to show exactly where the order runs out. A scalar like the Gini coefficient always returns an answer, ranking any two distributions; the curve reveals that some of those answers are arbitrary. When one Lorenz curve lies wholly above another, the ranking is unambiguous and every transfer-respecting inequality measure agrees, so the analyst can speak of one distribution being more equal without qualification. When two curves cross, no such statement is warranted: the verdict now depends on whether one weights inequality at the bottom or the top, and competing summary statistics can legitimately disagree. The curve puts that ambiguity on the page instead of burying it inside a single number — telling the analyst when a Gini comparison is trustworthy and when it is smuggling in an unstated value judgment about which end of the distribution matters.

It also separates two things a bare inequality figure fuses: the shape of a distribution's concentration and a one-number summary of it. Because the curve depends only on cumulative ranked shares, it reads the same way across any positive quantity — income, wealth, land, citations — so an analyst can ask not just "how unequal?" but "unequal where?": a curve that sags mostly near the high end (a thin top stratum holding much) describes a different society from one that sags near the low end (a large base holding little), even at the same Gini. The diagonal anchors all of this as a fixed reference for perfect equality, turning "how far from equal?" into a literal, visible distance rather than an abstraction.

Manages Complexity¶

A distribution of income, wealth, or land across a whole population is a high-dimensional object — in the limit one number per holder, in practice a full density with its own mean, spread, skew, and tail behaviour — and comparing two such distributions, or one population against itself across a decade, means somehow weighing all of that without committing to a particular functional form. The Lorenz curve collapses that object to a single curve in the unit square: cumulative share of the quantity against cumulative share of holders, sorted smallest to largest, with the diagonal as a fixed perfect-equality reference. The entire distribution's concentration is now one sag below the diagonal, and the standard scalar — the Gini, twice the area between curve and diagonal, running 0 to 1 — compresses even that shape to one number for headline comparison. An analyst tracks a curve (or a single coefficient) rather than a population's worth of holdings, reads "how far from equal" as a literal visible distance, and compares societies or years on a common, distribution-free instrument. Because it depends only on cumulative ranked shares, the same compression serves any positive quantity — income, wealth, land, citations — so one tool replaces a separate apparatus per domain.

What makes this more than lossy summary is that the curve also reports the reliability of its own compression, which is the deeper complexity-management move. Inequality is only a partial order, and the curve makes the boundary of safe comparison explicit. When one Lorenz curve lies wholly above another, the ranking is unambiguous — every transfer-respecting (Pigou-Dalton) measure agrees — and the analyst may collapse to the Gini and speak of "more equal" without qualification. When two curves cross, that collapse is unsafe: the verdict now turns on whether one weights the bottom or the top, and competing summary statistics can legitimately disagree. So the curve hands the analyst a branch test before any single number is trusted — dominance, in which case one scalar suffices, versus crossing, in which case the scalar smuggles in an unstated value judgment and the shape must be read directly. That same residual shape carries information the Gini discards: a curve sagging near the high end (a thin top stratum holding much) and one sagging near the low end (a large base holding little) describe different societies at the same Gini, so the analyst can ask not only "how unequal?" but "unequal where?" The compression is therefore graded — one number when curves are nested, the full shape when they cross — and the instrument itself tells you which regime you are in.

Abstract Reasoning¶

The Lorenz curve licenses reasoning about inequality as a graded operation — sometimes a clean ranking, sometimes only a shape to be read — and its moves all turn on the relationship between two curves, or between a curve and the equality diagonal.

The foundational move is dominance-ordering as unambiguous comparison. To rank two distributions by inequality, the analyst checks whether one Lorenz curve lies entirely above the other (closer to the diagonal at every point); if it does, the reasoning concludes that the first distribution is unambiguously less unequal, because every transfer-respecting (Pigou-Dalton) inequality measure must agree on the ranking. The inference runs from a geometric relation between curves to a verdict that holds across the whole family of inequality measures at once, so the analyst can assert "more equal" without committing to any particular index. This is the move that makes a Lorenz comparison stronger than any single statistic could be.

The decisive companion move is the crossing test as a reliability gate on scalar summaries. When two curves cross, the analyst reasons that no unqualified ranking is warranted: the verdict now depends on whether one weights inequality at the bottom or the top of the distribution, and competing summary statistics (including the Gini) can legitimately disagree. So before trusting any one-number comparison, the analyst runs a branch test — dominance, in which case collapsing to a scalar like the Gini is safe, versus crossing, in which case the scalar is smuggling in an unstated value judgment about which end of the distribution matters and the shape must be read directly. The reasoning is diagnostic about the instrument itself: the curve tells the analyst when a Gini comparison is trustworthy and when it is arbitrary, which is a claim about the limits of measurement rather than about any distribution.

A third move is reading concentration location off the sag — "unequal where?", not just "how unequal?". Because the curve preserves the full shape that the Gini discards, the analyst reasons from where the curve departs from the diagonal to the structure of the inequality: a curve sagging mostly near the high end (a thin top stratum holding much) describes a different society from one sagging near the low end (a large base holding little), even at an identical Gini. The move recovers distributional information from the residual shape, letting the analyst distinguish top-heavy from bottom-heavy concentration that any scalar summary would fuse.

A fourth move is distance-from-the-diagonal as a literal measure. The analyst treats "how far from equal?" as a visible geometric distance — the area between curve and diagonal, doubled, is the Gini, running 0 (the curve is the diagonal, perfect equality) to 1 (the curve hugs the edges, maximal concentration). The reasoning converts an abstract question about a distribution into reading off a fixed reference, with the diagonal anchoring perfect equality and the magnitude of the sag quantifying departure from it.

Finally, the curve supports a substrate-portable application by shared form. Because the instrument depends only on cumulative ranked shares of a positive quantity, the analyst reasons that the same construction and the same vocabulary (Gini, dominance, equality diagonal) apply to income, wealth, land, or citations without modification — so a concentration verdict derived for one quantity transfers in method to any other. The move is to recognize a distribution of any positive quantity across units as a Lorenz object and to import the dominance-and-crossing reasoning wholesale, with the epidemiological concentration curve as the recognized sibling that relaxes the requirement that units be sorted by the measured quantity itself.

Knowledge Transfer¶

The Lorenz curve is a measurement instrument, not a causal mechanism, so it does not transfer by analogy and lose its force off-substrate — it transfers literally, intact, wherever its single mathematical precondition is met. That precondition is minimal: a positive quantity distributed across a set of units, the units rankable from smallest to largest holder. Given only that, the construction (cumulative share of the quantity against cumulative share of holders), the equality diagonal, the Gini as twice the enclosed area, the dominance partial order, and the crossing caveat are all defined and all mean exactly what they mean for income. The interpretive vocabulary travels with the object because it is the object; an analyst who can read an income Lorenz curve can read any other with no relearning.

That is why the device ranges so widely, and the range is a true extension of one instrument rather than a family of metaphors. Its home is inequality measurement in economics — income, wealth, and land holdings, the use Max Lorenz introduced in 1905 — where the diagnostics carry across those quantities untranslated. Beyond economics the same construction is applied verbatim: bibliometrics plots cumulative citations against cumulative papers (the 80/20-style concentration of impact); ecology plots cumulative biomass against cumulative species to measure evenness; computing plots load across servers or traffic across links; epidemiology uses the close structural sibling, the concentration curve, which relaxes only the requirement that units be sorted by the very quantity being measured, ranking instead by an external variable such as income. In each, the Gini, the dominance test, and the "unequal where?" reading off the sag apply unchanged. This is the application of one mathematical object to many datasets — the same status as admitting "the histogram" travels everywhere — not the recurrence of a mechanism, and that is exactly the right way to understand a measure's reach.

For an instrument the boundary worth marking is instrument-reach versus over-reading, and the Lorenz curve invites two over-reads in particular — one of which it is unusually good at flagging itself. The first is reading the Gini, a many-to-one summary, as if it were a total order: when two Lorenz curves cross, no transfer-respecting measure is entitled to an unqualified ranking, and a Gini comparison there is smuggling in an unstated value judgment about whether the bottom or the top matters more. The curve's signal service is to put that ambiguity on the page — collapse to the scalar only under dominance, read the shape directly under crossing — so the over-read is precisely the failure to consult the curve before trusting its summary. The second over-read is mistaking the measure for the phenomenon: a curve sagging far from the diagonal displays and quantifies concentration, but it is not itself the structural fact that produces concentration. That a few holders command most of the quantity is a claim about the underlying distribution — heavy tails, power-law generative processes, cumulative-advantage dynamics — and those are separate patterns with their own primes and their own mechanisms; the Lorenz curve is the lens, not the thing seen. Reading "high Gini" as an explanation rather than a description confuses the readout for the cause. The instrument answers exactly the question it was built for — how concentrated, and how safely can that be ranked — and the discipline is to ask it that question and not another (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

Take a five-person economy with incomes 1, 2, 3, 4, and 10 (total 20), sorted smallest to largest. Cumulative population shares step 20%, 40%, 60%, 80%, 100%; the matching cumulative income shares are 1/20 = 5%, 3/20 = 15%, 6/20 = 30%, 10/20 = 50%, and 100%. Plotting income share against population share gives the points (0.2, 0.05), (0.4, 0.15), (0.6, 0.30), (0.8, 0.50), (1.0, 1.0) — a curve bowing below the 45° line, deepest where the top earner's share pulls it down. Computing the enclosed area by trapezoids, the area under the curve is 0.30, so the area between it and the diagonal is 0.5 − 0.30 = 0.20 and the Gini is twice that, 0.40.

Mapped back: The sorted incomes are the ranked units; income-share-against-population-share are the cumulative-share axes; the 45° line is the equality diagonal; the bow is the sag below the diagonal, deepest near the top holder. Doubling the enclosed 0.20 to get the Gini coefficient of 0.40 is the distance-from-the-diagonal reading made a literal computation.

Applied / In Practice¶

National statistical agencies and the World Bank build income Lorenz curves from household survey microdata to publish Gini coefficients and to compare distributions across countries and over time. The instrument's discipline matters most when comparisons are contested: two countries with similar Gini values can have Lorenz curves that cross — one more equal at the bottom deciles, the other more equal at the top — so an unqualified "country A is more unequal" is not warranted, and analysts fall back to Lorenz-dominance checks or to inequality indices weighted toward the tail they care about. This is why cross-country inequality rankings are reported with caution rather than as a clean ordering.

Mapped back: The survey-built curves invoke the dominance partial-order — when one country's curve lies wholly above another's, every transfer-respecting measure agrees and the ranking is safe. The crossing cases are exactly the crossing caveat: a bare Gini comparison there smuggles in an unstated judgment about whether the bottom or the top matters, so the shape must be read directly. It also respects the measure-not-phenomenon boundary — the Gini describes the concentration, it does not explain the tax, wage, or capital dynamics producing it.

Structural Tensions¶

T1: Partial order versus the scalar that always answers (dominance versus crossing). The Gini coefficient never refuses: hand it any two distributions and it returns a ranking. The Lorenz curve's deeper service is to show that some of those rankings are unearned — only when one curve lies wholly above another (dominance) does every transfer-respecting measure agree, and there the collapse to a scalar is safe. When curves cross, no unqualified ranking is warranted, and a Gini comparison smuggles in an unstated value judgment about whether the bottom or the top matters. The tension is that the very convenience of a one-number summary — it always ranks — is also its danger, because it hides the boundary of safe comparison the curve makes visible. Diagnostic: Do the two curves nest or cross — and if they cross, has the Gini comparison silently chosen which end of the distribution counts?

T2: How unequal versus unequal where (the shape the scalar discards). The Gini is a many-to-one summary: two societies — one sagging near the high end, a thin top stratum holding much; one sagging near the low end, a large base holding little — can share an identical coefficient while describing entirely different structures of concentration. Compressing to a scalar buys a headline comparison at the cost of the location information the curve preserves in its sag. The tension is not that compression is wrong but that it is silent about what it drops: the same number answers "how unequal?" while erasing "unequal where?" A designer who reports only the Gini has a portable common metric; one who reads the curve keeps the distributional structure but forfeits the single clean ranking. Diagnostic: Does the question at hand turn only on the magnitude of concentration, or on where in the distribution the concentration sits?

T3: Description versus explanation (the lens is not the thing seen). A curve sagging far from the diagonal displays and quantifies concentration, but it is not the fact that produces concentration. That a few holders command most of the quantity issues from heavy tails, power-law generative processes, or cumulative-advantage dynamics — separate patterns with their own mechanisms — and the Lorenz curve is the readout, not the cause. The pull is to treat "high Gini" as an account of why inequality obtains, when it is only a measurement of that it obtains and by how much. The tension is that a vivid, quantitative display invites causal over-reading precisely because it looks like an explanation: the sag is so legible that the absence of a mechanism behind it is easy to forget. Diagnostic: Is the curve being asked to report the degree of concentration, or is it being conscripted to explain the process that generated it?

T4: Descriptive instrument versus normative verdict (the diagonal is a reference, not a baseline). The equality diagonal is a fixed mathematical reference where each x% of units holds x% of the quantity — it is not a moral target the curve endorses. A curve far from it describes high concentration; it does not by itself say the distribution is unjust or that equality is the right goal. Yet the geometry all but invites the reading: "distance from equality" sounds like distance from what ought to be, and the instrument's whole visual grammar is departure from a line labelled perfect equality. The tension is that a purely descriptive tool carries a normatively loaded anchor, so its readouts are constantly borrowed as arguments. Diagnostic: Is the sag being reported as a measured fact about concentration, or being wielded as a claim that the distribution ought to be flatter?

T5: Rank by the measured quantity versus rank by an external variable (Lorenz curve versus concentration curve). The Lorenz curve sorts units by the very quantity whose concentration it measures — poorest to richest by income, when income is what is plotted — which guarantees the curve stays below the diagonal and the Gini runs cleanly 0 to 1. Its epidemiological sibling, the concentration curve, relaxes exactly this: it ranks units by an external variable (typically income) while plotting a different quantity (health-care use), so the curve can rise above the diagonal when the ranked-poor consume more. The tension is that loosening the self-ranking requirement extends the instrument to questions Lorenz cannot ask — how does X concentrate across a Y-ordering? — but forfeits the guarantees (bounded sag, single-signed area) that make the Lorenz reading so disciplined. Diagnostic: Are units ranked by the same quantity being measured (Lorenz, with its guarantees) or by an external ordering (concentration curve, where the sign of the departure itself carries meaning)?

T6: Autonomy versus reduction (its own named instrument or an instance of a general construction). "Lorenz curve" is a canonically named device with a 1905 pedigree, a signature Gini summary, and a home in economic inequality measurement. Yet what travels to bibliometrics, ecology, computing, and epidemiology is not proprietary to economics: it is the general cumulative-share concentration construction and the dominance partial-order — a substrate-free mathematical object that applies verbatim wherever a positive quantity is distributed across rankable units. Cumulative citations against papers, biomass against species, load across servers all are Lorenz objects, not metaphors for one. The tension is between a standalone economic instrument that earns its own name and study and the recognition that its cross-domain reach already belongs to the general concentration-and-dominance structure it instantiates. Diagnostic: Resolve toward the general cumulative-share concentration construction and dominance partial-order when asking what carries across income, citations, biomass, and traffic; toward the named Lorenz curve when reading income or wealth inequality in situ.

Structural–Framed Character¶

The Lorenz curve sits at the mixed-structural position on the structural–framed spectrum — the most structural entry in this batch, close in register to how isostasy is characterized, and held off the structural pole not by any framedness of substance but by being a specific named measurement instrument rather than the fully general construction it instantiates. Four of the five criteria read structural, and strongly. Evaluative_weight is nil: the entry is explicit that the curve is "not a normative verdict" and the equality diagonal is "a mathematical reference for perfect equality, not a moral baseline the curve endorses" — a curve far from the diagonal describes concentration and blames nothing. Vocab_travels is high, unusually so: the interpretive vocabulary (cumulative share, equality diagonal, Gini, dominance, crossing) travels because it is the object, applying verbatim to income, wealth, land, citations, biomass, and server load — the entry likens this to "admitting the histogram travels everywhere." Import_vs_recognize is recognition, not analogy: cumulative citations against papers or biomass against species genuinely are Lorenz objects, the same mathematical construction applied to new data, not a resemblance borrowed. Institutional_origin is low in substance: while the name and the 1905 attribution belong to a discipline, the object itself — the cumulative-share plot, the enclosed-area Gini, the dominance partial order — is a mathematical fact about any distribution, not an artifact of a survey or agency.

Two things keep it in the mixed-structural band rather than at the pole or past the domain-specific bar. First, its relationship to human practice is that of a measurement instrument: it does not run observer-free in nature the way a rebounding lithosphere does — it is a lens an analyst constructs and reads — yet what it measures (the concentration structure of a distribution, the dominance ordering between two) is a substrate-independent mathematical property that exists whether or not anyone plots it, so it is far less practice-bound than the significance-testing instruments (look-elsewhere, Lord's paradox) whose very existence depends on p-values and reporting conventions. Second, and decisively for its domain-specific status, is the measure-not-phenomenon boundary the entry draws: the curve displays and quantifies concentration but is not the generative fact producing it (heavy tails, power-law and cumulative-advantage dynamics live elsewhere), so it is a structural instrument, not a structural mechanism.

The portable structural skeleton is the general cumulative-share concentration construction plus dominance partial-order — a substrate-free object that applies wherever a positive quantity is distributed across rankable units, with the epidemiological concentration curve as the recognized sibling that relaxes the self-ranking requirement. That skeleton is genuinely portable, which is exactly why the Lorenz curve reaches so widely. But it does not lift "Lorenz curve" itself to a prime, because that general construction is precisely what the named instrument instantiates as its inequality-measurement specialization, not what makes the specific 1905 device travel: the cross-domain reach belongs to the general concentration-and-dominance object, while the domain-accented specifics — units ranked by the very quantity measured, the Gini as the standard scalar, the income/wealth home and the concentration-curve sibling — are the part that stays keyed to inequality measurement. Its character: a genuinely structural, evaluatively neutral, recognized-not-analogized measurement instrument whose portable core is a substrate-free cumulative-share construction, mixed-structural because it is a named specialization of that general object and a lens on concentration rather than the mechanism generating it.

Structural Core vs. Domain Accent¶

This section settles why the Lorenz curve is a domain-specific abstraction and not a prime — and, because the curve is so nearly structural, the case turns less on framedness than on the difference between a named instrument and the general object it specializes.

What is skeletal (could lift toward a cross-domain prime). Strip inequality economics away and a thin, substrate-free relational structure survives: a positive quantity distributed across units rankable smallest-to-largest, plotted as cumulative share of the quantity against cumulative share of holders, whose departure from an equal-division diagonal quantifies concentration — together with a dominance relation that yields a partial order over such distributions (when one curve lies wholly above another, every transfer-respecting measure agrees). That is the general cumulative-share concentration construction plus dominance partial-order: ranked units, cumulative-share axes, an equality diagonal, a sag whose depth reads concentration, and a nesting-versus-crossing test on whether two distributions can be safely ranked. It is genuinely portable — it recurs in bibliometrics (citations against papers), ecology (biomass against species), and computing (load across servers), and those recurrences are the same mathematical object re-applied, recognition not metaphor. But it is the core the curve shares, not what makes "Lorenz curve" the distinctive named device.

What is domain-bound. The accent that does not lift is everything that keys the instrument to inequality measurement. The Gini coefficient as the standard scalar (twice the enclosed area, 0–1); the 1905 economic pedigree and the income/wealth/land home; the Pigou-Dalton transfer principle that licenses the dominance ordering; the reading of the equality diagonal as perfect equality rather than as a bare reference line; and the self-ranking convention — units sorted by the very quantity measured — that guarantees the bounded, single-signed sag. The decisive test is by contrast with the curve's own sibling: relax the self-ranking requirement and rank units by an external variable, and the guarantees dissolve into the looser epidemiological concentration curve; strip the inequality framing entirely and "how far from equal?" reduces to a bare geometric distance in a unit square. And the measure-not-phenomenon boundary is load-bearing here: the curve displays concentration but does not generate it — heavy tails, power-law processes, and cumulative-advantage dynamics are separate primes with their own mechanisms — so it is a structural instrument, not a structural mechanism that could itself be a prime.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose reach is recognition of the same mechanism. The Lorenz curve is unusual in that its transfer is literal everywhere its one precondition holds — a positive quantity across rankable units — so its reach across income, citations, biomass, and server load is genuine recognition rather than analogy. But that very fact is what keeps it below the bar: what travels intact is the general cumulative-share-and-dominance construction, of which "Lorenz curve" is the inequality-measurement specialization, and the parts that stay home — the Gini, the income substrate, the Pigou-Dalton grounding, the concentration-curve boundary — are exactly the named-instrument accents that do not travel. When the substrate-spanning lesson is actually wanted — how to read concentration off a distribution, and when a scalar ranking is safe versus arbitrary — it is already carried, in more general form, by that construction and its dominance partial_order. Admitting "Lorenz curve" as a prime would enshrine a named specialization where the portable content is the general object it instantiates; the cross-domain reach belongs to the construction, while the 1905 device carries measurement-economics baggage that should stay in inequality analysis.

Relationships to Other Abstractions¶

Current abstraction Lorenz Curve Domain-specific

Parents (3) — more general patterns this builds on

Lorenz Curve is part of Order Prime

Lorenz construction contains order twice: units are ranked by holdings and non-crossing curves induce a concentration-dominance partial order.
Lorenz Curve is a decomposition of Aggregation Prime

Removing inequality framing leaves a deliberate many-to-one collapse from unit holdings to cumulative population and quantity shares.
Lorenz Curve is a decomposition of Measurement Prime

Removing inequality vocabulary leaves an instrument-procedure-scale mapping from a distribution's concentration attribute to a cumulative-share curve.

Children (1) — more specific cases that build on this

Gini Coefficient Domain-specific presupposes Lorenz Curve

Gini presupposes the Lorenz concentration object because its defining scalar is twice the area between that curve and the equality diagonal.

Hierarchy paths (5) — routes to 5 parentless roots

Lorenz Curve → Order → Comparison → Self Checking

Show alternative paths (4)

Not to Be Confused With¶

Gini coefficient. The scalar summary derived from the curve — twice the area between curve and diagonal, running 0 to 1. It is a many-to-one compression that discards the curve's shape: two societies sagging near opposite ends can share a Gini yet differ entirely. The Lorenz curve is the full distributional object; the Gini is one number read off it, trustworthy only when curves nest, not cross. Tell: does the claim rest on the full shape and a dominance/crossing check (Lorenz curve), or on a single concentration number that always returns a ranking (Gini)?
Concentration curve. The close structural sibling that ranks units by an external variable (typically income) while plotting a different quantity (say health-care use), rather than sorting units by the very quantity measured. Relaxing the self-ranking requirement lets it ask "how does X concentrate across a Y-ordering?" but forfeits the Lorenz guarantees (bounded sag, single-signed area) — a concentration curve can rise above the diagonal. Tell: are units ranked by the same quantity being plotted (Lorenz curve), or by a separate ordering variable so the sign of the departure itself carries meaning (concentration curve)?
Pareto / power-law / heavy-tailed distribution. The generative phenomenon that often produces concentration — a functional claim about tail shape and the process behind it (cumulative advantage). The Lorenz curve is a measurement instrument that displays and quantifies concentration for any positive quantity, heavy-tailed or not, without explaining it. Confusing them reads "high Gini" as an explanation rather than a description. Tell: is the object a claim about the distribution's tail and its generating process (power law), or a readout of how concentrated the realized distribution is (Lorenz curve)?
Cumulative distribution function (CDF). The other cumulative curve one meets in statistics — cumulative probability plotted against the variable's value, describing where mass sits along the value axis. The Lorenz curve plots cumulative share of the quantity against cumulative share of ranked holders, both on 0–1 axes, and measures inequality, not the value distribution. Tell: does the y-axis accumulate probability against a raw value (CDF), or accumulate the quantity's share against the population's share (Lorenz curve)?
ROC curve. The classification-performance curve, also drawn in the unit square with a diagonal reference and an area summary (AUC), where AUC relates to a Gini-type index — which invites conflation. But ROC plots true-positive against false-positive rate to gauge a classifier's discrimination, an entirely different question from a quantity's distributional concentration. Tell: are the axes true-positive vs false-positive rate measuring a classifier (ROC), or cumulative quantity-share vs population-share measuring inequality (Lorenz curve)?
The general cumulative-share concentration construction + dominance partial order (umbrella). The substrate-free mathematical object the Lorenz curve instantiates — cumulative-share axes, an equality diagonal, a sag reading concentration, and a nesting-vs-crossing dominance test — which applies verbatim to citations, biomass, and server load. This is the super-type, not a peer; "Lorenz curve" is its inequality-measurement specialization with the Gini and income substrate. Tell: strip the Gini, the income home, and the perfect-equality reading and what remains — bare concentration-and-dominance on any ranked positive quantity — is the general construction (treated more fully in Structural Core vs. Domain Accent).

Neighborhood in Abstraction Space¶

Lorenz Curve sits in a sparse region of the domain-specific corpus (83^rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (309 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12