Data

Overview

The 2026_06 quarterly update expands coverage with 39 new aggregator and country-level historical sources, improves our continuous automated error-monitoring system, resolves dozens of data-quality bugs, completes a comprehensive audit of the Mitchell historical statistics, and improves the Stata, R, and Python packages.

New Sources

This release adds 39 new sources, bringing the database to 160 sources in total. The additions are dominated by long-run country-level historical sources, complemented by several new cross-country aggregators. Highlights include:

• SECMCA historical statistics: macroeconomic statistics from the Central American Monetary Council for Costa Rica, the Dominican Republic, Guatemala, Honduras, Nicaragua, and El Salvador (1960-2017).
• German historical sources: Ritschl (2002) and Ritschl & Spoerer (1997), adding interwar and German Reich macroeconomic, fiscal, monetary, trade, price, population, and national-accounts series.
• International Historical Database (IHD): additional series with improved source splits and price mappings.
• Archivo de Historia Económica de México: long-run Mexican national accounts, prices, money, and public finances (1876-2025).
• Ghana: historical monetary, inflation, and output series from Ibrahim Abdulai's work on Ghana (1980-2017).
• Australia: Foster's Australian Economic Statistics (1949/50-1994/95).
• Netherlands: the Herengracht house-price index (1628-1973), one of the longest continuous real-estate price series available.
• Korea: Open Fiscal Data from the Republic of Korea's Ministry of Economy and Finance.
• China: Chinese national income estimates, 1661-1933.
• United Kingdom: Feinstein's national income, expenditure, and output series (1855-1965), plus the Sefton-Weale balanced estimates of UK national income (1920-1990).
• Further country-level additions: new or expanded sources for India, New Zealand, Spain, Sweden, Turkey, Japan, Russia, Israel, Bulgaria, Venezuela, and Italy.
• CEPII: the CEPII trade-and-macro source has been refreshed and renamed from CEPII_TRADHIST to CEPII.

Automated Error Monitoring

This release expands our continuous error-monitoring system that scans the database for ratio discrepancies, unit breaks, and other anomalies, then routes each flagged case through an auto-triage workflow for review. The system has already identified and resolved dozens of data-quality issues and will keep running against future updates.

Data Quality and Bug Fixes

Driven by the new monitoring system and reports from GMD users, this release resolves dozens of bugs. The most important corrections fall into two groups:

• Units, scaling, and redenominations: corrected unit-scale and currency-vintage issues across IMF IFS, IMF GFS, Andersson, Bordo monetary aggregates, Flora, the African Development Bank source, HFS, and UN data.
• Splicing and ratio consistency: improved handling of gaps between sources, improved overlap handling, and corrected the US long-term interest-rate series.

Mitchell Historical Audit

We completed a comprehensive audit of the Mitchell International Historical Statistics, fixing issues and correcting the euro-cutoff handling for European government tax series.

Package Updates

The Stata, R, and Python packages have all been improved. The three distributions remain at full feature parity, so users have the same access to the combined database, underlying source data, and documentation regardless of language.

Easy Access to Underlying Raw Data

We have improved access to the underlying raw data, making it easier for users to download and analyze the original sources. Previously, the data was only available by downloading the Excel version of the database or using our Stata, R, or Python packages. From now on, users can simply indicate whether they want access to the underlying raw data on the download page, even where they download the csv or dta versions.

Acknowledgements

Thanks to everyone who reported issues and suggested improvements. Many fixes in this release came directly from user reports.

Overview

This release adds eleven new data sources, introduces two new variables, improves the methodology for splicing government finance ratios, harmonizes all ratio variables, and updates the Python and R packages to full feature parity with the Stata package.

New Sources

We added eleven new sources to the database:

• COMECON: The wiiw COMECON Dataset, covering economic time series for the command economies of Eastern Europe (1944–1994)
• CogneauDupraz: Colonial fiscal, GDP, trade, and population data for French colonies (1833–1962)
• MAFHOLA: The Monetary and Fiscal History of Latin America project, covering eleven Latin American countries (1960–2017)
• Andersson: Central government revenue data from Per F. Andersson, covering multiple countries from the 1800s onwards
• CS1_BOL: Historical real GDP for Bolivia from Herranz-Loncan & Peres-Cajias (2016), extending coverage back to the mid-nineteenth century
• CS1_PER: Historical macroeconomic data from the Banco Central de Reserva del Perú (1922–2021)
• CS1_COL: Historical series from Colombia's central bank, covering government finances, trade, current account, and monetary aggregates
• CS1_HKG: Historical data for Hong Kong including monetary aggregates, government finance, trade, GDP, exchange rates, and prices (1843–2002)

New Variables

We introduced two new consumption variables: household consumption (hcons) and government consumption (gcons). These complement the existing total consumption (cons) variable and provide a finer decomposition of the expenditure side of GDP.

Improved Government Finance Ratio Splicing

We introduced a new methodology for combining government finance ratios (revenue, expenditure, tax, debt, and deficit as % of GDP). We now splice the ratios directly, which avoids compounding errors that arise when the numerator and denominator are spliced separately with different chainlinking adjustments.

Ratio Harmonization

All ratio variables (e.g., govdebt_GDP, exports_GDP, CA_GDP) are now consistently expressed in percent, so that a value of 50 means 50% of GDP.

Package Updates

The Python and R packages have been updated to match the full functionality of the Stata package, including access to underlying source data and documentation features.

Data Quality

We incorporated feedback from GMD users and improved data quality across multiple variables and sources.

Overview

This release introduces significant enhancements to data accuracy and infrastructure. Key updates include a comprehensive revision of the real GDP series and the deployment of a fully automated, cloud-based data processing pipeline to ensure timely future updates.

Real GDP Improvement

We have conducted a major review of the real GDP series. The data has been rigorously corrected and is now consistently rebased to the year 2015, ensuring greater comparability and accuracy across the dataset.

Automated Pipeline

We have implemented a new automated pipeline. This system autonomously handles downloading, processing, and merging data from all sources in the cloud, ensuring the maintenance process and allowing for more frequent and reliable database updates.

Stata Package & Documentation

We have launched a dedicated repository for the official Stata package, now available on GitHub.

Additionally, we have released a comprehensive companion paper, which serves as a detailed guide to using the package effectively.

Overview

This update introduces improved government finance statistics, an extended technical appendix, a major overhaul to the Stata package, bug fixes, and the inclusion of World Bank income groups.

Improved Government Finance Statistics

• We further improved the construction of combined government finance statistics
• The GMD now more frequently relies on a country-specific ordering of sources, which has shown to be more reliable

Extended Technical Appendix

• The documentation is now clearer and more readable
• Going forward, each release will come with its own technical appendix

Major Update to Stata Package

• We rewrote the package from scratch for speed and additional functionalities
• You can now easily pull cleaned data from all sources underlying the GMD with one line of code, plus plot a BibTeX code to cite them
• Within the next few days, you can update using "ssc install gmd, replace"
• A new companion paper introduces the package in detail, including new options

Bug Fixes

• Thanks to many contributors, we identified and fixed several bugs
• Real GDP per capita in Venezuela is a noteworthy example

Overview

This quarterly update introduces improved government finance statistics, streamlined source handling, a new outlier detection process, and numerous fixes and small improvements.

Improved Government Finance Statistics

• Distinguishes between central and general government data
• Included in the GMD as separate series and consolidated aggregates

Improved Download Infrastructure

• Downloads now pull directly from IMF, Eurostat, OECD, and UN rather than dbnomics
• IMF downloads now use the newly released API (3.0)

Pipeline Improvements

• The GMD pipeline was overhauled
• Runtime improved by approximately 10x

New and Improved Sources

• Various IMF and OECD datasets are treated as a single "source" where appropriate
• Added historical monetary statistics for France and unemployment series from Eurostat

Automated Error Checking

• Automated checks now cover multiple error types across the dataset
• Suspicious values are manually reviewed and confirmed

Bug Fixes

• Thanks to many contributors, various small bugs were identified and fixed
• Corrected systematic mistakes identified in the World Bank's WDI and IMF's FPP data

Technical Documentation Updates

• Improved technical documentation for better user experience
• Enhanced clarity of data source descriptions and methodology

Long Term Interest Rate Data Updates

• Updated long term interest rate data with latest available information
• Improved data quality and consistency for interest rate variables

Inflation Data Patch

• Revised priority list to integrate new sources
• Fixed issues in inflation data caused by breaks in the CPI

New Sources and Coverage

• Expanded coverage with six new data sources: RR_infl, Clio, UN_trade, BEL_1, CAN_2, and KOR_2.
• Updated source priority ordering for nGDP, cons, inv, finv, imports, and exports

Government Finance Methodology

• Now construct government finance variables based on their ratio to GDP
• Derived absolute-level series from ratios and applied chainlinking methodology
• Applies to: govdebt, govdef, govexp, govrev, and govtax

Bug Fixes and Improvements

• Improved robustness of the data cleaning pipeline
• Resolved minor inconsistencies across historical sources

Overview

This release includes updated annual data, expanded historical coverage for key countries, resolved inconsistencies in past series, and introduces new ways to access the database through our Python, R, and Stata packages.

Data Updates

All datasets have been updated to include the most recently published annual values as of the release date.

Expanded Historical Coverage

New long-run series have been added for Argentina, Ireland, and Taiwan, enriching the database's historical scope.

IMF WEO Treatment Refined

We now treat the IMF World Economic Outlook (WEO) as two distinct sources: one for historical values and one for forecasts.

World Bank Data Prioritization

We have adjusted our source hierarchy to prioritize data from the World Development Indicators (WDI) over both editions of the IMF WEO.

Exchange Rate Coverage

Monaco has been added to the EUR_fx irrevocable exchange rate list with a fixed rate of 6.55.

Bug Fixes

Corrected inaccuracies in Australia’s long-run historical real GDP in USD figures.

New Access Tools

• Python and R packages: Now available for easier data access.
• Official Stata package: Now available through the SSC Archive.

Initial Release

• Includes 46 core macro-financial variables
• Covers 243 countries and territories
• Annual time series spanning from 1086 to 2024 with projections up to 2030

Data Sources

• Integrated 32 major contemporary sources (e.g., IMF, World Bank, OECD)
• Incorporated 78 historical datasets to extend long-run coverage

Structure and Design

• Unified data model across all countries, variables, and time periods
• Standardized indicator abbreviations and ISO3 country codes
• Laid foundation for subsequent versioned updates and cross-platform access