Adherence to the OECD Model Tax Convention: A Textual Analysis of Member and Non-Member Countries' Double Taxation Agreements

This paper examines the textual design of Double Taxation Agreements (DTAs). Using Natural Language Processing (NLP) techniques we analyze the heterogeneity of a country's set of DTAs and the drivers that lead to (non-)adherence to the OECD Model Convention. We observe that wealthier countries hold a more diverse DTA set compared to poorer countries and more frequently diverge from the OECD Model Convention. Furthermore, we show that negotiation results are less likely to follow the OECD Model Convention when treaty partners are closely related in terms of trade volume or share a colonial history. Institutional factors, such as OECD or EU membership, can counteract this effect and thus lead to a stronger adherence to the OECD Model Convention. As such, our study highlights how socio-economic, institutional, and geopolitical factors influence (non-)adherence to the OECD Model Convention and how this depends on the article under consideration.