09 Aug 2021
In the finance bill for 2020, the French tax and customs authorities have been allowed, on an experimental basis, to use information from social networks and online sales websites to trigger tax and customs inquiries. The implementing rules have now been published.
This article looks at the scope of this new process, how it will work in practice, and what kind of data will be processed.
As in many countries, the French tax authorities (FTA) and the French customs authorities (FCA) have been working for several years on new ways to collect and explore data to better detect fraud.
In 2019, a new step was taken as both the FTA and the FCA have been allowed, initially on an experimental basis, to collect data from social networks and online sales websites to look for tax or customs offenses.
Official guidelines have been released and both the FTA and the FCA may start collecting data from websites.
New Ways to Manage Data
Data Mining and Artificial Intelligence
- How to manage the massive flow of data?
In most states around the world, tax returns are now filled electronically and fed into internal databases. Much data (bank accounts, rulings, country-by-country reports, etc.) is also automatically exchanged by states.
Therefore, all tax and customs authorities are faced with the same difficulty: how to develop tools for exploiting the data they receive?
In France, the FTA and FCA have chosen to increase the use of data mining and artificial intelligence to improve the detection of potential breaches and offenses by relying on predictive analysis based on a taxpayer’s behavior modeling algorithms.
- The means implemented by France
The French government’s goal is to increase the share of tax audits targeted by artificial intelligence to 50% in 2021. To achieve this goal, both the FTA and the FCA have created a department and developed an information processing tool which is fed by tax internal databases.
Need to Find New Sources of Information
- The limits of using internal databases
The FTA have developed tools using their internal databases, which are fed with all the tax returns filed in France and information received from other states through automatic exchanges (bank accounts, etc.)
Even if it represents billions of data that can be useful in the fight against fraud, there are some limits insofar as only the data contained in tax or customs returns may be used. If there is no tax or custom return and no bank account, there is no data.
Indeed, some tax or customs offenses are quite simple and cannot be detected if no return is filed.
- a garage owner carries out repairs to a vehicle. No invoice is issued, and payment is made in cash. Nothing is declared in the mechanic’s tax return
- a watch seller based in Australia makes sales in France directly from his website. The watches are then sent by post and no custom return is filed. No value-added tax (VAT) is then declared and paid in France.
Even if these cases are very simple, they are almost undetectable with an algorithm which would only work with data from tax and custom returns. The most difficult tax or custom fraud to detect for authorities is always the one for which no return has been filed.
- The value of searching for external data
By exploiting data that does not appear in tax returns, tax authorities will be able to enrich the quality of their analysis work considerably. With the development of online social networks and sales websites, much information is available and could be used to develop algorithms to search for tax and custom offences.
Example: a tool manufacturer sells goods in France directly from its website. This information, if collected by software, could easily be cross-checked with tax and customs data.
By Thierry Viu, Bloomberg Tax, 6 August 2021
Read more at Bloomberg Tax
RiskScreen: Eliminating Financial Crime with Smart Technology
Advance your CPD minutes for this content, by signing up and using the CPD WalletFREE CPD Wallet