Skip to content
FinObservatory

Redistribution / Methodology

Redistribution methodology

How the FinObservatory redistribution layer is built: what each number means, which income concept it uses, where it comes from, and what it deliberately does not claim. This layer measures fiscal redistribution: the effect of cash taxes and cash transfers on measured income inequality. It is one part of how societies redistribute, not the whole of it (see "What this does not measure").

The income concepts

Redistribution can only be read once the income concept is named, so every exhibit names it.

The OECD Income Distribution Database (IDD) reports the Gini coefficient on three concepts of household income, equivalised for household size and ranked over people:

  • Market income: income before taxes and transfers (labour and capital income, private transfers).
  • Gross income: market income plus cash public transfers, before direct taxes.
  • Disposable income: gross income minus direct taxes and social contributions. This is the household's cash "take-home".

Fiscal redistribution here is the difference the tax-and-transfer system makes to the Gini:

redistribution (Gini points) = Gini(market income) - Gini(disposable income)
percentage reduction          = [Gini(market) - Gini(disposable)] / Gini(market) x 100

The Congressional Budget Office (CBO) uses a parallel but not identical set of concepts for the United States:

  • Market income: labour, business, capital income, and other non-transfer income.
  • Income before transfers and taxes: market income plus social insurance benefits (Social Security, Medicare, unemployment insurance, and workers' compensation). CBO treats social insurance as pre-existing income, not as a means-tested transfer.
  • Income after transfers and taxes: income before transfers and taxes, plus means-tested transfers (Medicaid and CHIP, SNAP, Supplemental Security Income, and others) minus federal taxes (individual income, payroll, corporate income, and excise taxes). State and local taxes are not netted out. Dollar figures are in the report's inflation-adjusted dollars; households are ranked after an adjustment for household size, and each income group holds an equal number of people.

Because CBO folds social insurance into "income before transfers and taxes", its US redistribution numbers are not directly comparable to the OECD's market-to-disposable gap. The two are presented as separate exhibits, each labelled with its own concept.

Datasets

OECD Income Distribution Database (IDD). Dataflow OECD.WISE.INE:DSD_WISE_IDD@DF_IDD(1.0), pulled from the OECD SDMX API (sdmx.oecd.org, the API behind data-explorer.oecd.org). Three Gini measures on the 0-1 scale (market, gross, disposable), all reference areas, all age bands, both methodology eras. Cross-sections use the current series only: total population, "income definition since 2012" (METH2012), current definition (D_CUR). See scripts/build_oecd_idd.py.

The 2012 methodology break is real and is respected. The OECD revised its income definition; the database carries the pre-2012 and since-2012 series as separate methodology codes, with overlap-year "previous definition" and "income component" variants at the seam. Splicing the two eras would join incomparable series, so cross-country exhibits never mix them; the parquet preserves every era so the seam is auditable.

CBO, The Distribution of Household Income, 2022. Congressional Budget Office, January 2026, cbo.gov/publication/62300. The researcher archive (61911-additional-data-for-researchers.zip) carries three sets of twelve tables, one set per ranking measure, 1979 to 2022. Four tables are ingested: average household income (Table 3), means-tested transfer rates (Table 8), federal tax rates (Table 9), and household income shares (Table 10). The US exhibits use the "ranked by income before transfers and taxes" set, all households, the same basis as CBO's own supplemental workbook. See scripts/build_cbo_income.py.

OECD Social Expenditure Database (SOCX). Dataflow OECD.ELS.SPD:DSD_SOCX_AGG@DF_SOCX_AGG(1.0), social spending as a % of GDP by expenditure source. This is the policy-input side (what governments spend), not a measured redistribution effect. See scripts/build_oecd_socx.py.

IMF GFS COFOG social protection. Dataflow IMF.STA:GFS_COFOG(11.0.0), general-government (S13) expenditure on social protection (COFOG division GF10) as a % of GDP, via the api.imf.org SDMX route. Broad country coverage, used as the wide-coverage complement to SOCX. See scripts/build_imf_cofog_sp.py.

The gross-versus-net social-spending nuance

Public social spending as a share of GDP (SOCX ES10) understates how much a country actually devotes to social purposes, and does so unevenly across countries. The OECD's net social expenditure accounting adjusts gross public spending for the tax treatment of benefits, for tax breaks with a social purpose (for example the favourable tax treatment of employer-provided health benefits and of household pension saving), and for mandatory and voluntary private social spending.

The United States is the standard illustration: it ranks mid-to-low on gross public social spending, but near the top on net total social spending (ES50) once tax breaks and its large private social spending are netted in. Both series are shown, at the same year, so the difference is visible rather than asserted. The net series (ES40 net public, ES50 net total) require the tax-treatment module and stop earlier than the gross public series; the on-page comparison uses the latest year both are published for.

The IMF COFOG social-protection series is not comparable to SOCX public social spending: COFOG classes health under a separate division (GF07), so COFOG social protection (GF10) excludes health entirely, while SOCX public social spending includes it. COFOG is shown for its country breadth, with the concept difference stated on the page, never spliced onto the SOCX levels.

What this does not measure

The market-to-disposable delta ranks fiscal redistribution at a point in time. It is not the whole of how inequality is reduced:

  • Predistribution is excluded. Wage compression, collective bargaining, minimum wages, and the structure of the labour market shape market incomes before any tax or transfer. A country can achieve a low disposable-income Gini by compressing market incomes rather than by redistributing heavily, and would then show a small delta despite low inequality. Market-income Ginis are shown alongside disposable so this is visible.
  • In-kind services are excluded. Publicly provided health care and education reduce the effective inequality of living standards, but they are not cash and do not enter the disposable-income Gini. Studies that value them find inequality lower still; the disposable-income measure does not capture this.
  • Point-in-time is not lifetime. Much of what public pensions and social insurance do is move income across a person's own life course (contributions while working, benefits in retirement) rather than between rich and poor households. A single year's snapshot reads life-cycle smoothing as redistribution; the true interpersonal component is smaller.

None of this is a criticism of the measure; it is the boundary of what a one-year cash-income Gini can say. The layer reports the fiscal effect precisely and states the boundary plainly.

Units and comparability

  • OECD IDD Ginis are on equivalised household disposable (or market, or gross) income, ranked over individuals. CBO ranks households adjusted for size, with equal numbers of people per group. The two are close in spirit but not identical constructions.
  • The latest available year differs by country in the OECD IDD (2011 for China and India, 2024 or 2025 for most OECD members). Every cross-country exhibit carries each country's own data year and never blends years.
  • Bangladesh is not covered by the OECD IDD or SOCX; it is therefore absent from those panels. Its absence is stated, not silently dropped. It does appear in wider IMF COFOG coverage where reported at the general-government level (it currently reports only budgetary central government, so it is absent from the general-government COFOG panel too, by construction).

Verification anchors (quoted from the publishers, checked at build)

The build scripts abort unless these reproduce exactly:

  • OECD IDD (total population, current definition, since-2012), 2023: United States Gini market 0.506, gross 0.437, disposable 0.394; Finland market 0.509, gross 0.319, disposable 0.269. Reproduced at the OECD Data Explorer's displayed three-decimal rounding.
  • OECD SOCX: public social spending % of GDP in 2024, France 30.6 and United States 19.8; net total social spending in 2021, France 32.0 and United States 33.2. Reproduced at the explorer's one-decimal, half-up display rounding.
  • CBO (supplemental Table 3, all households, 2022): the lowest quintile's income before transfers and taxes of $26,200 and income after transfers and taxes of $44,800, with all 60 Table-3 cells for 2022 matching the researcher CSVs exactly.
  • IMF COFOG: France general-government social protection 23.31% of GDP in 2023 and United States 7.80% in 2024, reproduced at the API's full served precision.

Licenses

  • OECD IDD and SOCX: the OECD terms and conditions (oecd.org/en/about/terms-conditions.html) permit use with attribution. Displayed with attribution; no redistribution surface.
  • IMF GFS COFOG: IMF copyright, served publicly via api.imf.org; displayed with attribution; no redistribution surface.
  • CBO: a legislative-branch federal agency; its published data are US government works in the public domain.

FinObservatory displays these data with attribution and never re-serves files, rows, or bulk data (owner ruling, 2026-07-12). There are no download buttons and no data API on this layer.