Yesterday at the colloquium, we discussed Omri Marian's The Making of International Tax Law: Empirical Evidence from Natural Langugage Processing (coauthored by Elliott Ash). The paper offers an empirical study of bilateral income tax treaty language, using statistical natural language processing to test for convergence in treaty language over time. The main findings are that (1) there has been convergence in treaty language over recent decades - the treaties now use the very same terms et al to a greater extent than they did in the past - and (2) the OECD Model Tax Treaty appears to have the greatest linguistic influence, outweighing that of the U.N. and U.S. Model Tax Treaties. (Not a surprise when one thinks of how active and well-staffed the OECD is!)
This is interesting and worth knowing, but a key question is what one learns from and should make of it. For example, it's not immediately clear to what extent common or even converging treaty language relates to underlying legal content. One can use common terms, especially vague and general ones such as those used, say, to imply a need for transfer pricing, to mean very different things. Nor is it immediately clear how treaty law relates to domestic international tax law or to unwritten customary practice in the field.
The paper's current draft argues that treaty language convergence is relevant to debate in the international tax policy field regarding Reuven Avi-Yonah's claim that a sufficiently well-developed and binding regime of transnational tax law has arisen to suggest that nation-states "are not free to adopt any international tax rules they please, but rather [must] operate within the context of the regime." In particular, Avi-Yonah argues in favor of a "single tax principle" under which all income should be taxed once, not zero times or twice. Although the Marian paper does not discuss the single tax principle as such - and the treaties generally focus on parceling out non-overlapping rights to tax, whether or not exercised by someone in each case - it states: "While we stop short of concluding that a binding international tax regime exists, we believe our findings lend support to such an argument."
I have historically been somewhat skeptical regarding the "binding regime" claim's centrality with respect to policy debates on which, as it happens, Avi-Yonah and I often find ourselves in significant substantive agreement regarding the bottom line. The binding regime debate raises a rich set of issues all on its own, but the extent to which tax treaty linguistic convergence sheds light on it is probably best evaluated separately from this paper's very interesting empirical findings.