{
  "id": "lorenz-curve",
  "title": "Lorenz Curve",
  "lang": "en",
  "source_lang": "en",
  "summary": "A Lorenz curve plots cumulative population shares against cumulative shares of income, wealth, or another quantity to describe how unequally it is distributed.",
  "category": "economics",
  "subcategory": "macroeconomics",
  "type": "concept",
  "aliases": [
    "Lorenz concentration curve"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/lorenz-curve/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/lorenz-curve/",
      "json": "https://aiwiki.page/zh/lorenz-curve.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The Lorenz curve is a graphical representation of how a quantity is distributed across a population. Most commonly used to study income inequality and wealth inequality, it plots the cumulative share of people, ranked from the lowest to the highest value, against their cumulative share of the total quantity. Under its standard assumptions of nonnegative values and a positive total, the curve lies on or below the diagonal representing perfect equality. The farther it lies below that diagonal, the more concentrated the distribution is, although curves that cross cannot be ranked unambiguously by this criterion alone."
    },
    {
      "heading": "Origin and interpretation",
      "text": "The American statistician Max O. Lorenz introduced the curve in 1905 in an article titled “Methods of Measuring the Concentration of Wealth.” It became a central tool in distributional statistics and economics, particularly through its relationship with the Gini coefficient.\n\nThe horizontal axis records the cumulative proportion of the population, beginning with those who have the least income or wealth. The vertical axis records the cumulative proportion of total income or wealth received or owned by that population.\n\nFor example, a point \\((0.40,0.15)\\) means that the poorest 40% of the population receives 15% of total income. It follows that the richest 60% receives the remaining 85%.\n\nThe diagonal from \\((0,0)\\) to \\((1,1)\\), called the line of equality, represents a distribution in which everyone has the same amount. In that case, the poorest 20% receives 20% of the total, the poorest 50% receives 50%, and so forth.\n\nA Lorenz curve describes relative shares, not absolute living standards. Two populations can have identical curves even if every person in one population receives ten times the income of the corresponding person in the other."
    },
    {
      "heading": "Mathematical definition",
      "text": "Let \\(X\\) be a nonnegative random variable representing income or another measured quantity, with finite positive mean\n\n\\[\\mu=\\mathbb{E}[X]>0.\\]\n\nLet \\(F\\) be its cumulative distribution function, and let \\(Q(u)\\) denote its quantile function:\n\n\\[Q(u)=\\inf\\{x:F(x)\\geq u\\}, \\qquad 0<u<1.\\]\n\nThe Lorenz curve is\n\n\\[L(p)=\\frac{1}{\\mu}\\int_0^p Q(u)\\,du, \\qquad 0\\leq p\\leq 1.\\]\n\nThe integral in the numerator measures the cumulative quantity associated with the lowest fraction \\(p\\) of the population. Dividing by the population mean expresses it as a share of the total.\n\nFor an equally weighted sample of \\(n\\) observations sorted so that\n\n\\[0\\leq x_{(1)}\\leq x_{(2)}\\leq\\cdots\\leq x_{(n)},\\]\n\nthe empirical curve passes through\n\n\\[p_i=\\frac{i}{n}, \\qquad L(p_i)=\\frac{\\sum_{j=1}^{i}x_{(j)}}{\\sum_{j=1}^{n}x_{(j)}},\\]\n\ntogether with \\((0,0)\\). These points are conventionally joined by straight line segments.\n\nSurvey weights require a corresponding adjustment: population shares are calculated from cumulative weights, while income shares are calculated from cumulative weighted incomes.\n\nMain properties\n\nUnder the standard assumptions:\n\n- Endpoints: \\(L(0)=0\\) and \\(L(1)=1\\).\n- Monotonicity: the curve never decreases.\n- Convexity: its slope is nondecreasing because observations are ordered from smallest to largest.\n- Position: \\(0\\leq L(p)\\leq p\\).\n- Scale invariance: multiplying every observation by the same positive constant leaves the curve unchanged.\n- Anonymity: exchanging incomes between named individuals leaves the curve unchanged, because only the ordered distribution matters.\n\nWhere differentiable,\n\n\\[L'(p)=\\frac{Q(p)}{\\mu}.\\]\n\nThus, the slope at a population rank equals the income at that rank relative to mean income.\n\nAdding the same positive amount to every income, unlike multiplying all incomes by a common factor, generally moves the curve toward the equality line."
    },
    {
      "heading": "Numerical example",
      "text": "Suppose five equally weighted individuals have incomes of 10, 20, 30, 40, and 100 units. Total income is 200 units.\n\nCumulative population share | Cumulative income | Cumulative income share\n0% | 0 | 0%\n20% | 10 | 5%\n40% | 30 | 15%\n60% | 60 | 30%\n80% | 100 | 50%\n100% | 200 | 100%\n\nThe poorest 80% receives half of total income, while the richest 20% receives the other half. Plotting the final column against the first produces the empirical Lorenz curve.\n\nThese observations also illustrate why population shares and income shares must be distinguished: each person represents 20% of the population, but not 20% of income."
    },
    {
      "heading": "Relationship to the Gini coefficient",
      "text": "The Gini coefficient summarizes the area between the Lorenz curve and the equality line. Since the area below the equality line is \\(1/2\\),\n\n\\[G=1-2\\int_0^1 L(p)\\,dp.\\]\n\nPerfect equality gives \\(L(p)=p\\) and \\(G=0\\). Under standard nonnegative-income assumptions, greater concentration produces a larger Gini coefficient.\n\nFor empirical data joined by straight lines, the area under the curve is calculated exactly using trapezoids. Equivalently,\n\n\\[G=1-\\sum_{i=1}^{n} \\bigl(L(p_i)+L(p_{i-1})\\bigr)(p_i-p_{i-1}).\\]\n\nThe five-person example has a Gini coefficient of \\(0.40\\).\n\nIn a finite, equally weighted population of \\(n\\) people, the largest uncorrected empirical Gini coefficient is \\(1-1/n\\), attained when one person receives everything. Some reported measures apply finite-sample adjustments, so conventions matter when comparing estimates.\n\nThe Gini coefficient compresses an entire curve into one number. Different Lorenz curves can enclose the same area and therefore have the same Gini coefficient despite differing substantially in the distribution of income among particular population groups."
    },
    {
      "heading": "Lorenz dominance and distributional comparisons",
      "text": "Distribution A Lorenz-dominates distribution B if\n\n\\[L_A(p)\\geq L_B(p) \\quad\\text{for every }p\\in[0,1],\\]\n\nwith strict inequality for at least one interior value of \\(p\\). A is then less unequal according to the Lorenz ordering.\n\nThis ordering is stronger than a comparison of Gini coefficients: Lorenz dominance implies a weakly lower Gini coefficient, but a lower Gini coefficient does not imply Lorenz dominance.\n\nA transfer from a richer person to a poorer person, holding total income constant and not reversing their ranks, moves the distribution toward equality in the Lorenz sense. This is the basis of the Pigou–Dalton transfer principle.\n\nIf two curves cross, neither distribution Lorenz-dominates the other. One may allocate a larger income share to the poorest groups but a smaller share to a broader lower-income group. Different inequality measures may then rank the distributions differently because they place different weight on different parts of the distribution.\n\nGeneralized Lorenz curves\n\nThe generalized Lorenz curve retains information about the level of income by multiplying the ordinary curve by mean income:\n\n\\[GL(p)=\\mu L(p)=\\int_0^p Q(u)\\,du.\\]\n\nIts endpoint is \\(GL(1)=\\mu\\), rather than 1. Unlike an ordinary Lorenz curve, it changes when all incomes are multiplied by a common factor.\n\nGeneralized Lorenz comparisons are used in welfare economics to assess income levels and distribution together. If one generalized curve lies everywhere above another, it is preferred by the class of social welfare criteria based on averaging a common increasing, concave function of income. This interpretation incorporates both a preference for higher income and an aversion to inequality."
    },
    {
      "heading": "Applications and measurement choices",
      "text": "Lorenz curves are used to compare distributions across countries, regions, demographic groups, and periods. They can describe income, expenditure on consumption, landholdings, or other nonnegative quantities.\n\nFor income analysis, the measured concept is crucial. Market income, income after taxes, and disposable income can produce different curves. Comparisons must also distinguish individual income from household income. Household resources are often adjusted using an equivalence scale to account for household size and composition.\n\nA comparison of market-income and disposable-income curves describes how the measured distributions differ after taxes and transfers. By itself, however, it does not establish the causal effect of a policy, because observed income distributions can also reflect behavioral responses and other influences.\n\nOutside income analysis, the same mathematical construction can describe concentration in firm sizes, resource ownership, or scientific output. Interpretation depends on what is being ranked and what quantity is being accumulated."
    },
    {
      "heading": "Limitations",
      "text": "Negative or zero values\n\nZero observations are compatible with the standard definition if total income remains positive. The curve then has an initial horizontal segment.\n\nNegative observations, which can arise in income data and especially in net worth data, complicate interpretation. With a positive overall mean, cumulative shares may initially be negative, and the curve may fall below the horizontal axis. The conventional Gini coefficient can then exceed 1.\n\nIf total income is zero, the normalized curve is undefined; a negative total also undermines the usual interpretation of cumulative shares. Standard geometric properties and inequality interpretations therefore should not be assumed without checking the data.\n\nLoss of information about income levels and people\n\nAn ordinary Lorenz curve does not reveal mean income or directly measure poverty. A poorer population can have a curve closer to equality than a richer population.\n\nThe curve also does not identify who occupies each rank. Similar curves in successive years can coexist with substantial movement of individuals up or down the income distribution. Nor does a national curve reveal inequalities within households or between particular social groups unless those dimensions are examined separately.\n\nData quality and grouped observations\n\nHousehold surveys may underrepresent very high incomes, omit some income sources, or measure wealth imperfectly. Administrative records have different coverage limits and may not capture people outside the reporting system.\n\nPublished quintile or decile shares identify only selected points on a Lorenz curve. Joining those points with straight lines ignores inequality within each group. Under standard nonnegative-data assumptions, the resulting Gini coefficient is a lower bound on the value consistent with the full underlying distribution.\n\nSampling error also affects estimated curves. An apparent crossing or dominance relationship in sample data need not establish the same relationship in the population.\n\nRanking conventions\n\nA Lorenz curve ranks people by the quantity whose cumulative share is plotted. A related concentration curve may rank people by one variable, such as income, while plotting another, such as tax payments or health expenditure. Such a curve need not have the convex shape of a standard Lorenz curve, and its interpretation depends on the ranking variable."
    }
  ],
  "links": [
    "income-inequality",
    "wealth-inequality",
    "statistics",
    "gini-coefficient",
    "random-variable",
    "probability-distribution",
    "integral",
    "welfare-economics",
    "consumption-economics",
    "taxation",
    "disposable-income",
    "net-worth",
    "poverty"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}