Searching for 27 million patterns in 6,000 tax treaties
PhD candidate Manon Wintgens is using an algorithm to trawl through thousands of international tax treaties. She hopes to detect a system in the dizzying interplay between countries, businesses and documents. It is a unique research project.
The Data Science Research Programme at Leiden University combines data science with PhD projects in a wide range of disciplines. The programme has been running for over two years, and is producing the first astonishing results. We discuss some of them in this series of articles.
It is an invisible world that has a big influence on the ledgers of countries and their citizens: the immense tangle of international tax treaties. These treaties determine how much tax multinationals have to pay countries for transnational activities, for instance. Experts in tax law research matters such as how to make these agreements as ‘fair’ as possible.
In her data science project, however, Manon Wintgens is taking a different look at this tangle. She wants to see whether it can be understood as a single system, in which we can understand and even predict the behaviour of countries, treaties and businesses. She is doing this with the aid of pattern recognition algorithms, which she is using to compare treaties, among other things. A typical international tax treaty consists of 32 articles. The algorithm recognizes what is different within these 32 articles when treaties are compared with one another. This method is enabling her to conduct research that could never be performed by a person. In one fell swoop, she can compare 6,000 international tax treaties with one another and search for 27 million patterns within them.
What are the results of this comparison?
‘It means that you can recognise, and perhaps even predict, the behaviour of the actors – countries, treaties, businesses. One of the latest findings is that former colonies generally take on tax treaties that were entered into by the former coloniser. We don’t yet know why they do this, but the pattern recognition is an important step in itself. My research also shows that many more treaties are signed in certain regions, such as West Europe, than in a region such as Africa. This was known already, but this analysis has confirmed it.’
What is the advantage of charting the ‘behaviour’ of countries, treaties and businesses?
‘By approaching the field of international taxation as a system, we can gain new insight into how the various actors (jurisdictions, countries and companies) work together in this system. We may also gain new insight into the way in which the international tax system develops over a longer period.’
How does the Data Science Research Programme benefit your research?
‘It means I can make use of the expertise in Leiden in two fields: tax law and data science. I have a background in data science, and I’ve now found an interesting field in which to apply this knowledge. This is fairly new for experts in tax law: comparing 6,000 treaties in one go is something you rarely see.’
The Data Science Research Programme is a University-wide programme that aims to advance data science research and accelerate the use of data science methods at all faculties of Leiden University. The programme is associated with the Leiden Centre of Data Science.