LALLPI -- Latin American Left-Leaning Populism Index

The Latin American Left-Leaning Populism Index (LALLPI) measures active left-leaning populism in ~21 Latin American countries from 2000 onward, by combining populist rhetoric with populist policy outcomes -- rather than rhetoric alone.

POP = POP_R * (EP + IP)

• POP_R -- V-Party's populism (rhetoric) score, [0, 1], president's party.
• EP (Economic Populism) -- average of 4 sub-indices, IEF + EFW variables, [0, 100].
• IP (Institutional Populism) -- average of 6 sub-indices, V-Dem + IEF variables, [0, 100].
• PEP = POP_R * EP, PIP = POP_R * IP, POP = (PEP + PIP) / 2.

Full methodology: see the Documentation page on the project website.

Authors: Nicolas Cachanosky (The University of Texas at El Paso), J. P. Bastos (University of Austin), and Alexandre Padilla (Metropolitan State University of Denver).

Repository layout

Each index vintage is self-contained under matching code/<year>/ and data/<year>/ folders. To start a new vintage, copy code/2025/ to code/2026/, bump INDEX_YEAR in its config.py, and point it at a fresh data/2026/.

code/2025/
  config.py              # all paths + constants; INDEX_YEAR = "2025"
  requirements.txt        # pinned dependencies for this vintage
  01_build_skeleton.py
  02_vparty_prepare.py
  03_vparty_merge.py
  04_load_external.py
  05_build_index.py
  06_export_outputs.py
data/2025/
  countries.csv           # tracked -- country metadata (ISO codes, region, LDC/LLDC/SIDS)
  vparty_overrides.csv    # tracked -- curated governing-party decisions (see below)
  raw/                    # gitignored -- place raw source files here (see below)
  interim/                # gitignored -- intermediate outputs, regenerated by each script
  output/                 # tracked -- final published csv/xlsx/dta + missing-data report

Running the pipeline

cd code/2025
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

python3 01_build_skeleton.py
python3 02_vparty_prepare.py
python3 03_vparty_merge.py
python3 04_load_external.py
python3 05_build_index.py
python3 06_export_outputs.py

Each script reads from and writes to data/2025/interim/, so you can inspect the output of any step directly, or rerun from the middle if only a later step's logic changed.

Raw data sources (not tracked in git)

Place these in data/2025/raw/ before running the pipeline. They are intentionally excluded from version control.

File	Source	Notes
V-Party-2.dta	V-Party Dataset, V-Dem Institute	Populism (rhetoric) score, v2xpa_popul. This repo's copy is trimmed to ~0.13 MB (6 columns, LatAm countries, years >= 1990) -- see note 9 below if you need the full original.
V-Dem-13.dta	V-Dem Dataset v13, V-Dem Institute	Institutional variables
heritage.xlsx	Index of Economic Freedom, Heritage Foundation	Economic freedom sub-indices
efw.xlsx	Economic Freedom of the World, Fraser Institute	"EFW Ratings 1970-2021" sheet

Re-download the same release/version used for the current data/2025/output/ files if you need to reproduce them exactly.

Hand-maintained reference files (tracked in git)

countries.csv -- the ~52 countries/territories this index covers, with ISO codes, region, and LDC/LLDC/SIDS flags. Edit directly to add/adjust coverage.

vparty_overrides.csv -- for each country-year, which party (a V-Party party ID) was actually in government. This is a real human research task: V-Party scores every active party in a country's system each year, not just the governing one, so identifying the correct one requires external knowledge of who held the presidency. Extend it by adding rows directly, with the format ISO3, YEAR, PARTY_GOV, PARTY_CODE, PARTY_NAME.

Website (Quarto site, deployed via Netlify)

The site is a Quarto website project, built from the same published data this README documents above -- every table and chart on the live site reads data/2025/output/index_2025.csv directly at build time rather than embedding hand-copied numbers, specifically to avoid the site quietly going stale relative to the actual pipeline output.

Structure

_quarto.yml          # navbar, footer, social metadata, search
_brand.yml           # UTEP color/typography brand definition
styles.css           # consolidated site-wide CSS
_site_helpers.py      # shared functions used by multiple pages -- see below
index.qmd             # homepage
404.qmd               # custom 404 page (must live at repo root, not pages/)
pages/
  documentation.qmd
  data.qmd
  team.qmd
  countries.qmd        # directory of all country pages
  countries/            # one generated page per country
    venezuela.qmd, argentina.qmd, ...
generate_country_pages.py   # generates pages/countries/*.qmd from a template
requirements-site.txt        # Python deps for RENDERING the site (jupyter,
                              # pandas, matplotlib) -- distinct from
                              # code/2025/requirements.txt, which is for
                              # RUNNING the pipeline
netlify.toml           # installs Quarto CLI + Python deps, then quarto render

_site_helpers.py

Central module imported by every page's Python chunks (import sys; sys.path.insert(0, "."); import _site_helpers). Holds:

• build_ranking_table_html() -- the country ranking table used on both the homepage and Documentation page (also links each country name to its page).
• apply_chart_style() -- shared matplotlib styling (open axes, legend below the plot, UTEP colors) used by every chart on the site, so the charts can't visually drift apart from each other one at a time.
• COUNTRY_SLUGS -- the single source of truth for which countries have a generated page and what URL slug each uses. Both the ranking table's links and generate_country_pages.py read from this same dict, so they can't fall out of sync with each other. 3 countries are deliberately absent: Cuba, Guyana, and Suriname each have some data (V-Party rhetoric, or V-Dem institutional variables) but never a complete POP score for any year, so there's no "latest year" to build a page around. This is also why the site says 22 countries, not the 25 that appear somewhere in the published CSV -- 22 is the count with a complete score; 25 counts anyone with any data at all.
• country_summary(), build_country_trend_chart(), build_subindex_breakdown_chart(), build_regional_comparison_chart(), build_party_history_table_html() -- the building blocks each country page's chunks call.
• build_countries_directory_html() -- the table on pages/countries.qmd.

Adding or updating country pages

Country pages are generated, not hand-written. To add a country, add it to COUNTRY_SLUGS in _site_helpers.py, then run:

python3 generate_country_pages.py

Don't hand-edit files under pages/countries/ directly -- they get overwritten the next time this script runs. If you need to change what's on every country page, edit the TEMPLATE string in generate_country_pages.py and re-run the generator, not the individual generated files.

Known Quarto quirks worth knowing before debugging from scratch

• Raw HTML from a Python chunk must use IPython.display.HTML + display(), not print(). print() inside a #| output: asis chunk can render as literal escaped tag text instead of real HTML -- this actually happened on this site and took a few rounds to track down.
• Only call display() once per chunk. Multiple separate display(HTML(...)) calls in the same chunk only correctly render the last one -- also discovered the hard way. If a chunk needs to output several HTML pieces, concatenate them into one string and call display() once.
• Use root-relative paths (/images/..., /pages/...), never relative ones (../images/...). A page's depth in the folder structure (pages/countries/venezuela.qmd vs index.qmd) changes what a relative path resolves to; a root-relative path is the same from anywhere.
• quarto preview's local dev server and Netlify's production hosting handle 404s differently. Locally, the preview server appears to look for a 404.html scoped to the specific directory of the broken link (e.g. /pages/404.html) rather than falling through to the root 404.html, so the custom 404 page can appear blank in local preview even when it's correctly configured. Test 404 behavior against the actual deployed Netlify site, not just local preview.
• _extensions/ (if you install any Quarto extension) must be committed to git. It's not a build artifact Quarto regenerates -- if a page's YAML references a filters: entry and the corresponding _extensions/<name>/ folder isn't in the repo, the entire site fails to build, not just that one feature.

Deployment

netlify.toml installs a pinned Quarto CLI version and the packages in requirements-site.txt, then runs quarto render, on every push -- see that file's comments for why (Quarto's official Netlify plugin can't execute the Python this site's live tables/charts depend on).

---