📊 Add income distribution S3 export for bespoke viz

[marcelgerber]

Our other bespoke viz also upload data to the owid-public S3 bucket (see e.g. #6067 and https://github.com/owid/etl/blob/6e3a93b6d1afcbc99771e00556162131eb3f9c59/etl/steps/export/s3/ihme_gbd/latest/gbd_treemap_json.py), so let's do that for the income distribution plot, too. We also transform the data into a JSON format that's easily consumable in JS.

@paarriagadap If you want, you could now get rid of the data://external/poverty_inequality/latest/thousand_bins_distribution step. I believe that specifying the latest version in this step's DAG should be enough.

Summary

• Add an S3 export step for the 2026 World Bank PIP thousand-bin income distribution data.
• Register export://s3/poverty_inequality/latest/income-distribution in the poverty and inequality DAG.
• Emit compact country-keyed JSON at data/poverty-inequality/income-distribution.2026.json.

Validation

• DRY_RUN=1 .venv/bin/etlr export://s3/poverty_inequality/latest/income-distribution --export --private

[owidbot]

Quick links (staging server):

Site Dev	Site Preview	Admin	Wizard	Docs

Login: ssh owid@staging-site-data-add-income-distribution

chart-diff: ✅

No charts for review.

data-diff: ✅ No differences found

Automatically updated datasets matching excess_mortality|covid|fluid|flunet|country_profile|garden/ihme_gbd/2019/gbd_risk are not included

Edited: 2026-05-15 12:27:40 UTC
Execution time: 7.33 seconds

[paarriagadap]

Looks good!

My only comment is about hardcoding the year, which is not so future-proof. I would prefer to assign the latest year instead.

It's true that the previous step looks redundant now, but I am keeping it for now, because I might use it for static charts.

[paarriagadap]

I would prefer this to be the latest year and not hardcode to 2026

[paarriagadap]

Same here, hardcoding the year can be a problem for future versions of the data

[marcelgerber]

I've now changed it so that it always creates a data file for each year. Then we don't need to hardcode anything here, and can just flexibly use whichever data file we need.
It's not like it costs us much in terms of time or space, anyhow.