Compare databases#
Compare GDP, value added and GHG emissions across MRIO databases. Choose databases, versions, countries, years and sectors, and download results.
Found an error, or know a database we should add? Tell us on GitHub Issues.
Loading the data browser…
Available years and classifications. Missing years, extensions, population or exchange rates remain unavailable. No other year is substituted. World sums every native database region once, including RoW; it is unavailable where the table lacks a complete global total. World per-capita values use the published World Bank WLD population, not a sum of country ratios. Selecting several years shows separate observations for each year; the map is available for one version and year. Sector aggregates follow the source classification and its revision. Combined native categories remain combined. The Excel correspondence lists the native labels, section and weight; every mapped category enters once. Sector aggregates apply to value added and GHG emissions; GDP is whole-economy only. Comparisons without a verified common classification use whole-economy totals.
Value added and GDP. Value added sums the explicitly selected factor rows, preserving signed subsidies and excluding imports, balancing adjustments and product taxes. GDP adds net product taxes on intermediate and final uses. Some parsers omit final-demand taxes, so the bridge is recovered from the source. GDP remains unavailable where this bridge has not been verified. EXIOBASE Hybrid publishes net product taxes only as a row on activities and has no final-use tax row, so its GDP adds that row alone. EXIOBASE readings with zero product-tax blocks are labelled as estimates. WIOD uses the published VA and TXSP rows; international SUT final-use taxes are retained in VY. Its ROW imports are external factors, so international SUTs do not provide a complete World total. FIGARO reconstructs the income total from D1, D29X39 and B2A3G and adds D21X31 on intermediate and final uses, excluding residence-purchase adjustments. Its compilation already uses national-accounts aggregates and World Bank world GDP, so the WDI comparison is not an independent validation. GDP is available only for the whole economy.
World Bank. Current-price monetary values use current US dollars and same-year official exchange rates. WIOD PYP retains previous-year prices and previous-year US dollars; its values are not compared with current-price WDI GDP, and the chart reference is omitted whenever PYP is selected. From 1999, EUR values use the rate reported for Germany. For 1995–1998, the published Eurostat EUR/ECU annual series is used; German DEM/USD values are never interpreted as EUR/USD. Per-capita MRIO totals use the country and year's population; World Bank GDP per capita uses its published series. Sector and aggregate region per-capita values are unavailable. Differences are (database GDP / World Bank GDP − 1) × 100. Source vintages, balancing, projections and accounting boundaries affect the comparison; it is not an accuracy score. Current-dollar time series also reflect prices and exchange rates. Monetary IOT inputs are in million EUR; monetary SUT 3.8.2 income components are in EUR. Each reading retains its own scale.
GHG scope. Industry or activity emissions use E or Ea. Whole-economy totals also include directly emitted final-demand accounts from EY where supplied. EY is not allocated to producer sectors: with sector aggregates it appears separately as Households, so all sectors plus Households equal the whole-economy total. Unavailable EY is not estimated or treated as an observed zero. Each reading has an explicit gas selection. EXIOBASE monetary includes available combustion, process, agriculture and waste rows. Mass-reported gases use EDGAR-compatible AR5 GWP100: CO₂ 1, CH₄ 28, N₂O 265, SF₆ 23,500 and, where reported, NF₃ 16,100. Biogenic CO₂ is excluded; HFC/PFC already expressed in CO₂-equivalent retain the source characterisation, whose GWP basis is unverified. EXIOBASE Hybrid shows an AR5-characterised three-gas subtotal (CO₂, CH₄, N₂O), including separate unregistered-waste source accounts; it omits F-gases and their overlap is not independently verified. EMERGING uses native va as a GDP proxy, without a verified market-price bridge; its seven fossil CO₂ fuel accounts are a CO₂-only proxy under GHG, excluding CH₄, N₂O and F-gases. These caveats are visible beneath the graph and retained in CSV exports. EMERGING-E uses the same-year EMERGING 2.2 CO₂ companion projected onto its native sectors, with electricity split by fuel; these are modelled emissions rather than a separate observed extension. ADB retains its source AR5 GWP100 totals, converted from Gg to tonnes once. No common IPCC basis is claimed: EXIOBASE retains the source characterisation of pre-characterised HFC/PFC mixtures, whose original basis is unverified. Country total GHG units are Gg CO₂eq and World totals use Mt CO₂eq; per-capita MRIO units are tonnes CO₂eq per person.
GLORIA. The v060 source series covers 1990–2029; 2024–2029 are forecast tables. The 1990–2029 readings currently available here expose value added from six native factor rows and the source EDGAR-consistent GHG total, including directly emitted final-demand accounts. Native kilotonnes CO₂eq are converted to tonnes once. Detailed gas rows and the separate OECD-consistent inventory are not added again. Checks against the native gas rows recover AR5 GWP100 factors for CO₂, CH₄, N₂O, SF₆ and NF₃; individual F-gas factors remain unverified, so the source total is retained. Economic tables are dated January 2026 and satellites April 2024. GDP adds signed product taxes and subsidies on intermediate and final commodity uses; independent world income/expenditure checks retain the source residuals. Value-added and GHG sectors use the README’s ISIC Rev. 4 concordance; Other services remains combined S+T.
BONSAI. The native b2 HIOT has reference year 2016, with frozen inputs updated on 21 October 2026. GDP and value added are unavailable: MARIO's factor placeholders are not source observations. GHG shows characterised producer emissions from the native elementary flows with EDGAR-compatible AR5 GWP100 (CO2 1, both methane origins 28, N2O 265). Supplied iLUC contributions to these gases are retained. Accelerated CO2 is excluded because no AR5 factor has been verified; F-gases are not characterised. This is a three-gas subtotal, with producer and land-use boundaries that differ from EDGAR. Separate direct final-demand emissions are unavailable. Whole-economy values are available; the Classifications page provides the qualified NACE/CPA concordances. The native Global block contains global processes, not a World total; the World aggregate sums every native region once.
EDGAR. References appear automatically where available, as separate dash markers. The country reference uses unchanged cells from EDGAR 2026's TOTALS BY COUNTRY worksheet, in Gg CO₂eq with AR5 GWP100. It includes households and excludes LULUCF. International aviation/shipping are reported separately from national totals; published World totals include the reference’s original geographic boundary. Some source territories are grouped, and those values are not split. World totals and country/World per-capita references use unchanged published cells from the 2026 report workbook, in Mt CO₂eq and t CO₂eq per person. MRIO world totals use Mt to match that published reference. EDGAR is not recalculated by sector, summed across countries or interpolated. Its territorial boundary and GWP basis differ from MRIO production accounts; no percentage accuracy score is calculated.
© European Union 2026, European Commission JRC, EDGAR Community GHG database, comprising IEA–EDGAR CO₂, EDGAR CH₄, EDGAR N₂O and EDGAR F-gases. EU-owned material is generally CC BY 4.0; IEA–EDGAR CO₂ is CC BY-NC-ND 4.0. Retain attribution and component-specific conditions; use here is non-commercial. Crippa et al. (2026), GHG emissions of all world countries – 2026 Report, doi:10.2760/7717504. Data are supplied as-is, without warranties from the EU or IEA.
Downloads. The CSV contains the current chart or table observations and source attribution. The MARIO-only Python example parses one local reading and calculates the result; comments explain the remaining aggregation, filtering and conversion. It is a starting point for working with MARIO. Source matrices and coefficients are not distributed here.
Database availability and licences. The indicator browser currently includes EXIOBASE, ADB, OECD ICIO, EMERGING, WIOD, FIGARO, GLORIA and BONSAI. The Environmentally extended filter selects readings with loaded emission accounts; years and versions without them are excluded. Current GTAP releases require a licence; the licence conditions apply to use and redistribution. Older releases may have different access terms. Eora1, Eora2 and Open CEDA provide scope metadata only; Eora1 retains its MARIO parsing guide. GLORIA v060 is included for non-commercial research under CC BY-NC 3.0 Australia, with attribution retained in downloads. Each published reading links to its own source licence.
Additional sources: EXIOBASE classification, IPCC AR5 GWP100 factors, World Bank GDP, GDP per capita, population and exchange rate.