EBRDo

Harmonize countries in the EBRD style. Available in Julia, R, Python, and Stata.

EBRDo takes a table with a country column (ISO2, ISO3, ISO-numeric, or country name) and adds five columns that classify each country the way the European Bank for Reconstruction and Development (EBRD) does. The same tool is available in all four languages, generated from one source of truth and tested to give identical results.

# R
library(EBRDo)
ebrdo(data.frame(country = c("Kazakhstan", "Germany", "Kosovo", "United States")))
#>         country ebrd_coo     ebrd_name          ebrd_region   transition_group  combined_group
#> 1    Kazakhstan        1    Kazakhstan         Central Asia       Central Asia            <NA>
#> 2       Germany        0       Germany      Advanced Europe    Advanced Europe Advanced Europe
#> 3        Kosovo        1        Kosovo South-eastern Europe    Western Balkans EBRD Emerging…
#> 4 United States        0 United States                 <NA>               <NA>            <NA>

What it adds

Non-COO comparator groups (Advanced Europe, Advanced Economies, the IMF aggregates, EBRD Emerging Europe) can be placed alongside the EBRD groups, and are configurable. Ireland is always excluded from comparators, because multinational profit-shifting distorts its macro data. Selecting overlapping comparator groups raises an informative error.

The four ports

One source of truth, verified parity

All ports derive their reference data from a single curated definition built on countrycode/CountryCodes/country_converter. That definition covers the 41 Countries of Operation, the EBRD/Transition/IMF groupings, the special spellings, and the Kosovo handling. The Stata crosswalk and the Julia/R/Python lookup tables are all generated from it. A cross-port parity test runs the same adversarial inputs (mixed schemes, accents, whitespace, Kosovo aliases, numerics, unresolvable values) through the ports and confirms they return identical classifications, value for value. None of them crash, and none corrupt their input. (The Python port resolves names through country_converter; a few non-COO fallback spellings can differ from countrycode — see python/README.md.)

Installation

Julia

using Pkg
Pkg.add(url = "https://github.com/angrahed/CountryCodes.jl")          # dependency, until registered
Pkg.add(url = "https://github.com/angrahed/EBRDo", subdir = "julia")

R

# install.packages("pak")
pak::pak("angrahed/EBRDo/r")
# or:  remotes::install_github("angrahed/EBRDo", subdir = "r")

Python

uv add "ebrdo @ git+https://github.com/angrahed/EBRDo#subdirectory=python"
# or:  pip install "git+https://github.com/angrahed/EBRDo#subdirectory=python"

Stata

The most reliable route ships the reference data: clone the repo and put stata/ on the adopath.

* git clone https://github.com/angrahed/EBRDo.git
adopath ++ "EBRDo/stata"
help ebrdo

net install works for the program and help, but Stata treats the .dta reference data as ancillary (not copied to the adopath), so the clone route above is recommended:

net install ebrdo, from("https://raw.githubusercontent.com/angrahed/EBRDo/main/stata") replace
* then fetch and place the datasets on your adopath (e.g. PERSONAL):
net get ebrdo, from("https://raw.githubusercontent.com/angrahed/EBRDo/main/stata") replace

Quick start

# Julia
using DataFrames, EBRDo
df = DataFrame(country = ["Kazakhstan", "Germany", "Kosovo"])
ebrdo(df, :country)                      # adds the five columns
is_ebrd_coo("Kazakhstan")                  # 1
ebrd_name("TUR")                           # "Türkiye"
ebrd_coos()                                 # all 41 COO names, in reference order

# R
library(EBRDo)
df <- data.frame(country = c("Kazakhstan", "Germany", "Kosovo"))
ebrdo(df)                                # country column auto-detected
ebrd_region(c("DEU", "POL"), comparators = "Advanced Economies")
ebrd_coos()                              # all 41 COO names, in reference order

# Python
import pandas as pd, ebrdo
df = pd.DataFrame({"country": ["Kazakhstan", "Germany", "Kosovo"]})
ebrdo.ebrdo(df)                          # country column auto-detected
ebrdo.ebrd_name("TUR")                   # "Türkiye"
ebrdo.ebrd_coos()                        # all 41 COO names, in reference order

* Stata
ebrdo country, comparators("Advanced Economies")

Each port supports the same options (comparators, advanced, abbreviate, output naming), documented in its folder's README and in the Julia/R/Stata help. Input is matched case- and accent-insensitively; ISO2, ISO3, ISO-numeric, and name all resolve, and Kosovo (which has no official ISO3) is accepted as KOS, XKX, XK, KV, "Kosovo", and so on.

Repository layout

EBRDo/
├── julia/   EBRDo.jl    Julia package (depends on CountryCodes.jl)
├── r/       EBRDo       R package (CRAN-style; tibble/data.table/duckdb)
├── python/  ebrdo       Python package (pandas; depends on country_converter)
├── stata/   ebrdo       Stata package (ado + help + bundled crosswalk)
├── CITATION.cff
└── LICENSE  (MIT)

Citing

See CITATION.cff, or cite as: Anton Grahed, EBRDo: Harmonize countries in the EBRD style (Julia, R, and Stata), 2026.

Acknowledgements

Country code and name harmonization is powered by Vincent Arel-Bundock's countrycode (R) and its Julia port CountryCodes.jl; the Python port uses country_converter.

License

MIT © Anton Grahed.