Home

Awesome

💸 {Digital,Cloud,Electronic,Online} Services VAT Rate Database

Some countries requires businesses to apply, for any sales of online/electronic/digital/cloud services to consumers (B2C), a value-added tax (VAT) on all purchases made by their citizen. This impose all foreign companies to track the residency of all their customers, to apply the right tax.

This trend started January 1st 2015 with all European country, and is about to be implemented by others too.

As the rate depends on the locality of the customer, this project aims to centralize, in a machine-readable format (currently a plain CSV file), the list of applicable rates for each country of residence, and all their territorial exceptions.

This is a painful job, worth sharing with a community, so please help me keep this database up to date! :)

Testimonials

I'm impressed with your BFPO detail. -- Tim Whitlock

I'm glad to see independently researched data confirming mine! -- Will Bond

VAT Application Rules

All B2C customers matching the locality in that file are subject to the corresponding tax.

Your B2B customers are exempted of VAT, as long as they provide a registered VAT number. You can check their validity on the VAT Information Exchange System (VIES). I recommend using a third-party library to automate the process, like pyvat for Python. A B2B customer without VAT number is considered as a simple B2C customer, so local rate applies.

Note that starting January 1st, 2015, these rules applies to all non-European SaaS businesses with European customers.

Status

This matrix expose the current completeness of the database:

Administrative familyEU member statesSpecial territories, states, countries, collectivities, islands, departments, towns, …
Number:white_check_mark: 28 / 28:white_check_mark: 57 / (?)
Standard rates:white_check_mark: All:white_check_mark: All
Reduced rates:x: None:x: None
Increased rates:x: None:x: None
Parking rates:x: None:x: None
Currency codes:white_check_mark: All:white_check_mark: All
Historical standard rates:white_check_mark: All:x: None
Historical reduced rates:x: None:x: None
Historical increased rates:x: None:x: None
Historical parking rates:x: None:x: None
Historical currency codes:warning: Wrongly aligned to current one:warning: Wrongly aligned to current one

Schema

start_date is an inclusive ISO 8601 calendar date from which the rate starts to apply.

stop_date is an inclusive ISO 8601 calendar date from which the rate is no longer valid.

territory_codes is a list of (eventually mixed):

currency_code is the de jure ISO 4217 currency code (a.k.a. legal tender), not de facto's one.

rate is the decimal rate.

rate_type is the kind of rate. Either:

description human-readable description of the territory the rate applies to, and eventual rationale behind the application.

Rows are sorted by territory_codes, then start_date.

Interpretation

Starting from this database, your next step is to interpret the data.

By looking at the dates, you can compute if a rate is either current, historical or future. Beware, some rates changes in the middle of a month. That means on theory, your billing system should support pro-rata application of several rates on a monthly invoice.

To choose the right rate, you then need to guess the location of your customer. I advise you to derive this data from the billing address, as it's the most common element with the necessary administrative granularity. An address that is properly normalized is precise enough, down to the postal code, to select the right VAT rule, including territorial exceptions. To solve the territory complex, I wrote a Python module to parse and normalize postal addresses.

Sources

The process of building up this database is somewhat fuzzy.

This database is unequivocally founded on the latest official VAT Rates document from the EC portal. It provides all member states' rates and their historical values. You'll also find there a description of regions and territories where special or no VAT rates applies.

Still, the hardest part of establishing this database lies in the characterization of locality. Member states and some regions are easy: they have a dedicated country code. For these we rely on ISO 3166-1 alpha-2, with an extra compatibility layer for European Commission country codes (i.e. the GB/UK and GR/EL pairs).

When this is not enough, we go down to a lower administrative level and leverage subdivision codes from ISO 3166-2.

Things get messy once VAT rules only applies to areas as small as a town. In which case I guesstimated the geographic zone with postal codes fetched from individual Wikipedia pages.

Finally, for completeness, I compiled the catalog of member's states special territories and restarted the locality characterization process for these. I was able to add the missing entries based on the list of included and excluded zones of the EU VAT area.

Other resources

History

I decided to create this database because all the current VAT libs were quite naive about the territory definition. Most of the time it's only based on the country, while the territory a tax applies to, in a fiscal context, is a much more insidious concept carrying administrative, political and historical weight.

To match the place the supply takes place against the VAT database, I created a Python module to normalize and parse postal addressed of my customers.

License

The content of this repository is licensed under a BSD 2-Clause License.