Skip to content
Codex Populi · Printed from codexpopuli.org · latest release 2026.10.1
Codex Populi
The Record

Every forecast, kept — and scored as the years arrive

13
UN revisions kept
4
independent forecasts
2.1B
spread in their 2100 figures

What each revision projected for 2050 and 2100

World population: the 2024 revision (estimates solid, current projection dashed) with each earlier revision's forward projection from its vintage.

World population, projected 2050 and 2100, by UN WPP revision
Revision 20502100
UN 1998——
UN 2000——
UN 2002——
UN 2004——
UN 2006——
UN 2008——
UN 20109.31B—
UN 20129.55B10.85B
WIC 20139.14B8.95B
UN 20159.73B—
UN 20179.77B—
WIC 20189.38B9.28B
UN 20199.74B10.88B
IHME 20209.55B8.79B
UN 20229.71B10.35B
WIC 20239.59B9.89B
Census 2025——
UN 2024 (current)9.66B10.18B
  • No 2050 figure for these revisions. Their last projected year is 2049: UN 1998 8.88B, UN 2000 9.28B, UN 2002 8.89B, UN 2004 9.05B, UN 2006 9.16B, UN 2008 9.12B.

un-wpp (archival revisions) · release 2026.10.1

World population, 2024 revision and each earlier revision's projection

4 series: UN 2024 (1990–2100, latest 10.18B, projected); IHME 2020 (2017–2100, latest 8.79B, projected); UN 2022 (2021–2100, latest 10.35B, projected); WIC 2023 (2020–2100, latest 9.89B, projected).

UN · IHME · WIC · release 2026.10.1

UN · IHME · WIC · release 2026.10.1 · codexpopuli.org

Licences: UN — CC BY 3.0 IGO · IHME — IHME Free-of-Charge Non-Commercial User Agreement · WIC — Custom (IIASA/WIC, unconfirmed)

IHME 2020: Vollset SE, et al. Fertility, mortality, migration, and population scenarios for 195 countries and territories from 2017 to 2100: a forecasting analysis for the Global Burden of Disease Study. The Lancet. 2020.

WIC 2013: Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 1.0, 2013.

WIC 2018: Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 2.0, 2018.

WIC 2023: Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 3.0 (Beta), 2023.

IHME and WIC data shown with attribution for display only; not licensed for redistribution. IHME: non-commercial use only.

How far off were they?

Bias, by revision and horizon

Total fertility rate. Positive mean bias: the revision’s projected TFR came in above what materialized (over-projected). Negative: under-projected.

Horizon (years) Mean bias MAPE Over-projected
UN 1998
1-5n=177+0.0429.2%58.9%
6-10n=177+0.01912.9%62.2%
11-15n=177-0.11014.8%47.8%
16-20n=177-0.17416.2%62.2%
21-25n=177-0.15719.1%77.8%
26-30n=177-0.10524.5%95.6%
UN 2000
1-5n=178+0.14212.5%49.5%
6-10n=178+0.04216.3%44.0%
11-15n=178-0.01017.3%48.4%
16-20n=178+0.02418.6%60.4%
21-25n=178+0.09422.2%85.7%
UN 2002
1-5n=182+0.09710.3%57.6%
6-10n=182-0.02113.9%48.9%
11-15n=182-0.08615.4%56.5%
16-20n=182-0.07117.2%67.4%
21-25n=182-0.01220.9%90.2%
UN 2004
1-5n=182-0.01812.0%43.5%
6-10n=182-0.06514.3%52.2%
11-15n=182-0.03616.5%68.5%
16-20n=182+0.02820.8%83.7%
UN 2006
1-5n=185-0.0669.8%41.1%
6-10n=185-0.10812.5%53.7%
11-15n=185-0.08015.1%69.5%
16-20n=185-0.01819.1%85.3%
UN 2008
1-5n=185-0.13410.1%54.7%
6-10n=185-0.10613.4%70.5%
11-15n=185-0.04518.0%85.3%
UN 2010
1-5n=186-0.09810.0%68.8%
6-10n=186-0.04313.7%71.9%
11-15n=186+0.03718.9%85.4%
UN 2012
1-5n=191-0.0147.6%69.1%
6-10n=191+0.03311.5%77.3%
11-15n=191+0.11116.7%88.7%
WIC 2013
1-5n=185-0.28312.7%56.8%
6-10n=185-0.19516.2%74.7%
UN 2015
1-5n=191+0.0389.9%73.2%
6-10n=191+0.11115.0%86.6%
UN 2017
1-5n=191+0.0459.1%78.4%
6-10n=191+0.11814.3%83.5%
WIC 2018
1-5n=191-0.11213.1%79.4%
UN 2019
1-5n=191+0.10812.0%86.6%
IHME 2020
1-5n=945-0.00712.7%70.7%
6-10n=189+0.03215.8%78.9%
UN 2022
1-5n=406+0.0516.2%72.9%

n counts the forecasts behind mean bias and MAPE. The over-projected share counts only below-replacement TFR forecasts. Regional and world aggregates are scored with the countries, so n is not a count of countries.

release 2026.10.1

Country probe

Pick a country to compare its actual total fertility rate against what each revision projected.

Total fertility rate, Japan: actual vs. revisions

UN · Census · IHME · WIC · release 2026.10.1

UN · Census · IHME · WIC · release 2026.10.1 · codexpopuli.org

Licences: UN — CC BY 3.0 IGO · Census — Public domain (17 U.S.C. §105) · IHME — IHME Free-of-Charge Non-Commercial User Agreement · WIC — Custom (IIASA/WIC, unconfirmed)

IHME 2020: Vollset SE, et al. Fertility, mortality, migration, and population scenarios for 195 countries and territories from 2017 to 2100: a forecasting analysis for the Global Burden of Disease Study. The Lancet. 2020.

WIC 2013: Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 1.0, 2013.

WIC 2018: Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 2.0, 2018.

WIC 2023: Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 3.0 (Beta), 2023.

IHME and WIC data shown with attribution for display only; not licensed for redistribution. IHME: non-commercial use only.

Error by years ahead

Population forecasts from each scored revision, compared with the current estimates, by how many years ahead they looked. Each line is one revision; each point is the average size of its error, over or under (MAPE), across the countries and aggregates scored, at one horizon.

Population forecast error, by years ahead

12 series: UN 1998 (horizons 1–28 years, MAPE 18.9% at 28 years, n 197 at 1 year, 197 at 2 years, 197 at 3 years, 197 at 4 years, 197 at 5 years, 197 at 6 years, 197 at 7 years, 197 at 8 years, 197 at 9 years, 197 at 10 years, 197 at 11 years, 197 at 12 years, 197 at 13 years, 197 at 14 years, 197 at 15 years, 197 at 16 years, 197 at 17 years, 197 at 18 years, 197 at 19 years, 197 at 20 years, 197 at 21 years, 197 at 22 years, 197 at 23 years, 197 at 24 years, 197 at 25 years, 197 at 26 years, 197 at 27 years, 197 at 28 years); UN 2000 (horizons 1–23 years, MAPE 17.5% at 23 years, n 197 at 1 year, 197 at 2 years, 197 at 3 years, 197 at 4 years, 197 at 5 years, 197 at 6 years, 197 at 7 years, 197 at 8 years, 197 at 9 years, 197 at 10 years, 197 at 11 years, 197 at 12 years, 197 at 13 years, 197 at 14 years, 197 at 15 years, 197 at 16 years, 197 at 17 years, 197 at 18 years, 197 at 19 years, 197 at 20 years, 197 at 21 years, 197 at 22 years, 197 at 23 years); UN 2002 (horizons 1–23 years, MAPE 16.7% at 23 years, n 197 at 1 year, 197 at 2 years, 197 at 3 years, 197 at 4 years, 197 at 5 years, 197 at 6 years, 197 at 7 years, 197 at 8 years, 197 at 9 years, 197 at 10 years, 197 at 11 years, 197 at 12 years, 197 at 13 years, 197 at 14 years, 197 at 15 years, 197 at 16 years, 197 at 17 years, 197 at 18 years, 197 at 19 years, 197 at 20 years, 197 at 21 years, 197 at 22 years, 197 at 23 years); UN 2004 (horizons 1–18 years, MAPE 14.9% at 18 years, n 197 at 1 year, 197 at 2 years, 197 at 3 years, 197 at 4 years, 197 at 5 years, 197 at 6 years, 197 at 7 years, 197 at 8 years, 197 at 9 years, 197 at 10 years, 197 at 11 years, 197 at 12 years, 197 at 13 years, 197 at 14 years, 197 at 15 years, 197 at 16 years, 197 at 17 years, 197 at 18 years); UN 2006 (horizons 1–18 years, MAPE 12.0% at 18 years, n 199 at 1 year, 199 at 2 years, 199 at 3 years, 199 at 4 years, 199 at 5 years, 199 at 6 years, 199 at 7 years, 199 at 8 years, 199 at 9 years, 199 at 10 years, 199 at 11 years, 199 at 12 years, 199 at 13 years, 199 at 14 years, 199 at 15 years, 199 at 16 years, 199 at 17 years, 199 at 18 years); UN 2008 (horizons 1–13 years, MAPE 10.4% at 13 years, n 199 at 1 year, 199 at 2 years, 199 at 3 years, 199 at 4 years, 199 at 5 years, 199 at 6 years, 199 at 7 years, 199 at 8 years, 199 at 9 years, 199 at 10 years, 199 at 11 years, 199 at 12 years, 199 at 13 years); UN 2010 (horizons 1–13 years, MAPE 9.0% at 13 years, n 200 at 1 year, 200 at 2 years, 200 at 3 years, 200 at 4 years, 200 at 5 years, 200 at 6 years, 200 at 7 years, 200 at 8 years, 200 at 9 years, 200 at 10 years, 200 at 11 years, 200 at 12 years, 200 at 13 years); UN 2012 (horizons 1–13 years, MAPE 7.8% at 13 years, n 202 at 1 year, 202 at 2 years, 202 at 3 years, 202 at 4 years, 202 at 5 years, 202 at 6 years, 202 at 7 years, 202 at 8 years, 202 at 9 years, 202 at 10 years, 202 at 11 years, 202 at 12 years, 202 at 13 years); UN 2015 (horizons 1–8 years, MAPE 6.8% at 8 years, n 202 at 1 year, 202 at 2 years, 202 at 3 years, 202 at 4 years, 202 at 5 years, 202 at 6 years, 202 at 7 years, 202 at 8 years); UN 2017 (horizons 1–8 years, MAPE 5.9% at 8 years, n 202 at 1 year, 202 at 2 years, 202 at 3 years, 202 at 4 years, 202 at 5 years, 202 at 6 years, 202 at 7 years, 202 at 8 years); UN 2019 (horizons 1–4 years, MAPE 4.7% at 4 years, n 202 at 1 year, 202 at 2 years, 202 at 3 years, 202 at 4 years); UN 2022 (horizons 1–2 years, MAPE 2.8% at 2 years, n 203 at 1 year, 203 at 2 years).

UN · release 2026.10.1

UN · release 2026.10.1 · codexpopuli.org

Licences: UN — CC BY 3.0 IGO

How many forecasts each line rests on (n), at its first and last point:

UN 1998
n 197 at 1 year ahead, 197 at 28 years ahead
UN 2000
n 197 at 1 year ahead, 197 at 23 years ahead
UN 2002
n 197 at 1 year ahead, 197 at 23 years ahead
UN 2004
n 197 at 1 year ahead, 197 at 18 years ahead
UN 2006
n 199 at 1 year ahead, 199 at 18 years ahead
UN 2008
n 199 at 1 year ahead, 199 at 13 years ahead
UN 2010
n 200 at 1 year ahead, 200 at 13 years ahead
UN 2012
n 202 at 1 year ahead, 202 at 13 years ahead
UN 2015
n 202 at 1 year ahead, 202 at 8 years ahead
UN 2017
n 202 at 1 year ahead, 202 at 8 years ahead
UN 2019
n 202 at 1 year ahead, 202 at 4 years ahead
UN 2022
n 203 at 1 year ahead, 203 at 2 years ahead

Years ahead counts from the last year the revision treated as an estimate. n is the number of forecasts averaged into a point: it is listed above for the first and last point of each line, and shown on hover at any point. Regional and world aggregates are scored with the countries, so n is not a count of countries. The same number of years ahead is a different calendar year for each revision. Longer horizons come from fewer revisions, so the right-hand end of the chart rests on the oldest ones. A line of one or two points says little about how the error changes.

Methodology

How scores are computed

Every projection score compares one revision’s projected value, at a given future year, to the value the scoring release’s current data now reports for that same year and entity.

Method m-81678e8c: 8 scored series across 6 indicators; horizons of 1 to 30 years in 6 buckets for the bucketed views. The label is computed from the scoring configuration in the code now serving this page, and changes when a scored series, a horizon bucket or an exclusion changes. It is not stored with a release: a release’s score rows are written when it is built, and its buckets are computed when the page is requested.

error
projected − actual
relError
error / actual
horizon
year − estimateEndYear

Mean bias is the average of error across scored rows; MAPE is the average of |relError|. MAE is the average of |error|, in the indicator’s own units. Infant mortality, under-5 mortality and net migration rate are scored in their own units: their relError is not stored, their MAPE is null (never 0), and MAE is their headline measure. Life expectancy is scored for all, male and female separately, and a five-year period figure is scored at the year the UN labels as its period’s MidPeriod (2018 for 2015–2020). For the revisions that publish five-year periods, a net migration rate projection is a five-year average scored against a single year’s actual, so part of its error is year-to-year change, not error in the projection. Rows are bucketed by horizon into five-year buckets (1-5, 6-10, 11-15, 16-20, 21-25, and 26-30 years out); cells stop at 30 years. view=raw, and the scores.csv.gz in releases that carry one, hold every scored row of the sources the API serves; display-only sources are left out.

view=horizon gives mean bias and MAPE for every horizon year instead of a bucket, plus the median and the 10th and 90th percentiles of relError (null where MAPE is), and MAE. It returns every horizon present, beyond 30 included.

Rows for UN location codes (LocIDs) with no entity on this site are left out of every UN revision. Those in the 1998 to 2008 revisions include Sudan (Former), Netherlands Antilles, Serbia and Montenegro (1998 to 2004) and State of Palestine (2000 to 2008). Today's Sudan, South Sudan and Taiwan have no rows in the 1998 to 2008 revisions.

Actuals are the scoring release's current best estimates and are themselves revisable.

Forecasters

  • Wittgenstein Centre (WIC) — SSP2 (Medium) scenario

    The Wittgenstein Centre for Demography and Global Human Capital (IIASA) publishes population and total-fertility-rate projections disaggregated by education level, aggregated here to national and world totals; three vintages are scored on the accuracy page (2013, 2018, 2023).

    License: Custom (IIASA/WIC, unconfirmed). Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 1.0, 2013. Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 2.0, 2018. Wittgenstein Centre for Demography and Global Human Capital (WIC), Wittgenstein Centre Data Explorer Version 3.0 (Beta), 2023.

    Not available through this platform’s public API under this source’s license; shown on this page with attribution only.

    Wittgenstein Centre (WIC) raw data ↗

  • IHME — reference scenario

    The Institute for Health Metrics and Evaluation (IHME) publishes a reference-scenario population and fertility forecast for 195 countries and territories through 2100 (Vollset et al., The Lancet, 2020); one vintage (2020) is scored on the accuracy page.

    License: IHME Free-of-Charge Non-Commercial User Agreement. Vollset SE, et al. Fertility, mortality, migration, and population scenarios for 195 countries and territories from 2017 to 2100: a forecasting analysis for the Global Burden of Disease Study. The Lancet. 2020.

    Not available through this platform’s public API under this source’s license; shown on this page with attribution only.

  • U.S. Census Bureau (IDB) — central (single, no named alternatives) scenario

    The U.S. Census Bureau's International Database (IDB) publishes a single central population projection per country, with no named alternative scenarios; one vintage (2025) is scored on the accuracy page.

    License: Public domain (17 U.S.C. §105). U.S. Census Bureau, International Database.

    U.S. Census Bureau (IDB) raw data ↗

Every forecaster above is scored against the same actuals (the current UN WPP release) and the same error/relError/horizon/bucket formulas on this page, so MAPE and mean bias are directly comparable across forecasters as arithmetic -- but each targets a different central scenario (see above), so a comparison also reflects each forecaster's scenario framing, not forecasting skill alone.

World Bank, OECD, and PRB population projections aren’t included here: they largely republish the UN medium variant rather than forecast independently.

Every source’s licence, coverage and API availability: Sources.