Can algorithms expose tax fraud?
Ignacio González, Agencia Estatal de Administración Tributaria; Alfonso Mateos, Universidad Politécnica de Madrid; Adapted by: Eli Vivas and Carina Bellver (StoryData)
For administrations to be able to optimise the design of the two components of any tax policy – revenues and expenditures – they need to know the distribution of wealth, and also to identify the natural and legal persons who possess it, in order to prevent tax evasion. They use algorithms, new big data techniques and artificial intelligence to detect, with a precision that would have been unimaginable only a few years ago, both the wealth that some endeavour to conceal under a network of companies and the various tax fraud mechanisms.
1Algorithms detect hidden wealth, misuse of aggressive fiscal engineering, money laundering and fraud.
2The use of algorithms has enabled the Spanish Tax Administration Agency ('Agencia Estatal de Administración Tributaria' or AEAT) to identify more than 170 million undeclared family relationships on top of the 87.6 million family relationships declared in personal income tax and inheritance tax. In this way, the total number of family relationships that can be used to detect how companies are controlled is at present nearly 258 million.
3On the basis of all tax returns, the total wealth of all Spain’s taxpayers had been estimated at 3.6 trillion euros. The use of algorithms has made it possible to identify the real owners of more than half a trillion euros more, concealed behind networks of enterprises.
4Algorithms have revealed that the richest 6% of the Spanish population possess part of their wealth indirectly, through a network of enterprises that are not listed on the stock market. This group comprises 2,532,964 citizens whose wealth is much greater than would appear from their holdings in these companies.