Needs to be added to the GitHub Action

Filing because I've noticed it a few times: openfisca-uk-data frs generate 2018 and FRS.generate(2018) producing different results (the Python interface producing correctly lower income tax and self-employed NI aggregates, and the CLI producing income tax of around £190bn, which is more accurate to the external figures, but it's not right because of the under-reporting of income in the FRS).

For wealth taxes

Hopefully we don't need to do this, but in case, here's an option:

• https://www.irs.gov/pub/irs-soi/14rpmicrodataanalysissystem.pdf
• https://github.com/PSLmodels/Tax-Brain/blob/master/cs-config/cs_config/helpers.py#L89

Include region of the UK in the random forest inputs when imputing data from the Wealth and Assets Survey.

We're currently fusing the 2016-2018 WAS to the 2019-2020 FRS. We should uprate wealth and its components (e.g. land) accordingly, perhaps by nominal GDP growth.

We have some documentation on the imputation of carbon emissions to the FRS at https://policyengine.github.io/openfisca-uk-data/imputation.html#imputing-carbon-emissions-to-the-frs

Let's add some summary statistics:

• Totals for each imputed consumption category
• Contribution to carbon emissions
• Carbon emission totals

The imputation occurs in https://github.com/PolicyEngine/openfisca-uk-data/blob/master/openfisca_uk_data/datasets/frs/frs_enhanced/lcf_imputation.py

Just noticed that private transfer income and maintenance income have some negative values.

Seems like all the relevant code is migrated to openfisca-uk, so we just need to transfer issues over before we archive, is that right @nikhilwoodruff?

Following work in UBICenter/uk-carbon-dividend#6

To be more general than the current WAS-imputed FRS

@nikhilwoodruff has already been working on this, but to formalize: SPI microdata matches administrative totals, and has income of about 100bn pounds more than the FRS, especially at the high end and especially in non-wage income like dividends. We can adjust the FRS data using some quantile-related techniques put forth by the government and IFS, and possibly with random-forests-based synthimpute for people with none of the missing income type (like dividends) in the FRS.

By using the cname arg in https://github.com/peaceiris/actions-gh-pages

Is it possible to download datasets like the FRS from the package directly, as openfisca-us-data does (assuming the user is authenticated with UKDS)?

Placeholder until the data comes out at https://beta.ukdataservice.ac.uk/datacatalogue/series/series?id=200017#!/access-data

Following https://github.com/UBICenter/uk-lvt/blob/main/impute_land.py

Following https://github.com/UBICenter/uk-lvt/blob/main/process_was.py

We currently have land value (x * property wealth + y * corporate wealth) and carbon (x_1 * food spending + x_2 * transport spending, etc.). It'd be better to impute these inputs and apply the final step in OpenFisca-UK.

We need to impute wealth types from the WAS:

• Property wealth
• Corporate wealth

And spending types from the LCFS:

• Food and non-alcoholic beverages
• Alcohol and tobacco
• Clothing and footwear
• Housing, water and electricity
• Household furnishings
• Health
• Transport
• Communication
• Recreation
• Education
• Restaurants
• Miscellaneous