DAC and CRS code lists

The DAC Secretariat of the OECD maintains various code lists which are used by donors to report on their aid flows to the DAC databases. In addition, these codes are used to classify information in the DAC databases. This repository republishes the lists as CSV files, in a machine-readable format.

Column definitions and types are in datapackage.json.

Data

All files are in data/, as CSV with a header row. Every value is quoted. Most files have English (name_en) and French (name_fr) names.

The columns are:

Known limitations

• Codes are text. They can contain letters (A01 in aid_types.csv), and many lists have blank cells. Read the files as text, so that codes are not turned into numbers and blank cells stay blank:

  import pandas as pd
  df = pd.read_csv("data/sectors.csv", dtype=str, keep_default_na=False)
  

• code is not a unique key in every file. In agencies.csv the donor code repeats for each agency; the key is code with agency_code. In sectors.csv the purpose code repeats for each voluntary_code under it; the key is code with voluntary_code.

• Some cells are blank. For example acronym in agencies.csv, the descriptions in finance_types.csv and sectors.csv, and some French names in channel_codes.csv and finance_types.csv. geography is blank for a few recipients.

• geography mixes broad and narrow regions: some recipients are placed in Africa or Asia, others in a narrower region such as Eastern Africa.

• A description cell can contain a line break inside the quotes. CSV readers handle this; tools that work line by line do not.

• recipients.csv keeps only the DAC income group. The spreadsheet has more columns (for example the World Bank income group and flags such as landlocked), which are not kept.

Data sources

• The DAC and CRS code lists, an Excel file published by the OECD DAC Secretariat and described on the OECD page of reporting resources. Used under the OECD Terms and Conditions, which allow commercial use, adaptation and sharing with credit to the OECD. The exact name of the licence for code lists is unclear (see License). The OECD site refuses automated requests, so those two links may not open from a script.

Changes made to the source data: scraper.py downloads the Excel file (and the OECD's XML version of the lists, which is stored but not used for the CSV files) into source/ unchanged. parsers/excel.py then reads the sheets and columns named in excel_crs_mappings.json, converts every cell to text, removes surrounding spaces, skips header and empty rows, fills down category values where the sheet leaves them blank, removes duplicate rows, joins the English and French sector lists, and writes one CSV file per list. A few cells are replaced by fixed values listed in excel_crs_mappings.json. Columns of the sheets that are not in the mapping are not kept.

Not endorsed by, or affiliated with, the OECD.

Preparation

You need Python 3 (runtime.txt and the workflow file say which versions are used) and access to webfs.oecd.org:

pip install -r requirements.txt
python scraper.py

scraper.py downloads the latest OECD files into source/ and rebuilds all the CSV files in data/. The sheet and columns of each list are set in excel_crs_mappings.json. If the OECD changes the layout of its spreadsheet, the mappings must be changed to match.

A GitHub Action (.github/workflows/scrape.yml) runs scraper.py every day at 02:00 UTC, and on every push. On main, when the result differs, it opens or updates a pull request called "Automated update". Nothing reaches main until that pull request is merged.

License

The code lists are published by the OECD. The OECD Terms and Conditions (section 3, "Data") say:

and that anyone who shares or licenses work made with the Data must "include the same acknowledgment requirement in any sub-licenses that you grant". The same terms add that Data "may be subject to restrictions beyond the scope of these Terms and Conditions, either because specific terms apply to those Data or because third parties may have ownership interests".

What was checked, and what could not be established:

• The OECD terms page refuses automated requests. The quotes above were read on an archived copy dated 31 December 2025, which says it was last updated in July 2024. The live page may differ.
• The OECD names the Creative Commons Attribution 4.0 licence (CC BY 4.0) only for its "written content". The terms do not say whether a code list counts as "Data" or as written content, so the exact licence is unclear. Both readings allow commercial use, adaptation and sharing, need credit to the OECD, and do not require share-alike. For adapted written content the OECD also asks for a disclaimer, which is included in ATTRIBUTION.txt.
• The Excel file and the XML file carry no licence notice of their own, in the sheets and elements we looked at.
• We found no third-party notice in the file, but did not read every cell.

The maintainers place only their own work (the scraper, the column mappings and the CSV layout) under the Public Domain Dedication and License (PDDL). This does not cover the OECD's code lists, in data/ or in source/. See ATTRIBUTION.txt for the credit to give when you use the data.