NTA Annual Meeting in New Orleans, part 3

As noted in earlier blog entries, I wanted to offer a brief account of the other two panels in which I participated at the NTA Annual Meeting in New Orleans. Herewith the first additional entry, concerning a panel at which I commented on two empirical papers on international taxation.

The first paper on which I commented was Eric Allen & Susan Morse, "Firm Incorporation Outside the U.S.: No Exodus Yet," which is available here. Allen and Morse use hand-collected data (a phrase in general usage that I nonetheless find charmingly antiquarian, since I would assume people mostly do it with computers) to make a very useful rifleshot finding that contributes to our understanding of U.S. incorporation practices.

As I noted in my residence electivity article, if the U.S. pushes worldwide taxation of U.S. companies especially hard, one might expect rising foreign incorporation by U.S. start-ups. The article notes anecdotal evidence, offered to me by leading practitioners, explaining why in their experience home incorporation remains the norm in most business sectors, even for the potential multinationals of the future. Home incorporation turns out to have significant advantages in the start-up phase. Plus, so long as one places one's valuable international property abroad for tax purposes, the U.S. regime generally is not all that onerous anyway. But I noted suggestive evidence from an article by Mihir Desai and Dhammika Dharmapala that tax haven incorporations have risen in recent years. Might this be the start of a trend? Both I and the authors of that article stroked our chins (metaphorically speaking) and said yes, it's just possible that it might be.

Allen and Morse lay this to rest, however, by finding that the recent tax haven incorporation boomlet appears to reflect almost exclusively action involving start-ups from China and Hong Kong. U.S.-headquartered companies are not contributing to it at any significant level.

Good to know this. Now, I noted that this finding doesn't rebut the possibility that rising tax-elasticity of overseas investment by U.S. companies might still be a current trend and a problem, operating along other margins (e.g., new equity issuances, and clientele effects regarding who ends up investing where). Nonetheless, Allen and Morse have made a very nice contribution by making this new finding about tax haven start-ups, and I imagine that they will be cited regularly for this (certainly by me).

The second paper on which I commented was "Taxes and the Clustering of Foreign Subsidiaries," by Scott Dyreng, Brad Lindsey, Kevin Markle, and Douglas Shackelford, which is not yet available on-line. This paper notes that the empirical literature to date on how U.S. (and other) multinationals (MNEs) shift income between affiliates has generally operated under the assumption that all company choices regarding where to place an overseas affiliate are made independently.

For example, if Acme Products U.S. decides to place a controlled subsidiary in the Netherlands, this is assumed to have absolutely no effect on whether it will also put one, say, in Belgium, France, Germany, Ireland, or the Bahamas (to name just a few where we know that in fact there might be interacting causation). The literature adopts this view, not because anyone actually believes that affiliate choices are independent, but because we don't know how they interact with each other.

The Dyreng-Lindsey-Markle-Shackelford paper attempts to figure out relationships by coming up with "expected" correlations between where one has an overseas affiliate in the absence of a tax motivation, and then comparing this to the actual correlations that are found. E.g., if one found lots and lots of MNEs with affiliates in Germany plus the Caymans, this might suggest that the pairing is tax-motivated unless there are other grounds for expecting it.

The difficulties in deriving findings on this very interesting topic (which potentially would help inform policymakers), include the following:

--It's hard to say what correlations would be expected absent tax considerations. For example, should companies be expected commonly to have affiliates in neighboring countries? Leaving tax aside, one could imagine that a Netherlands sub is either a substitute or a complement for one in Belgium. On the one hand, the MNE is active in the area and thus might want to go to more countries that are nearby. On the other hand, perhaps the Netherlands is close enough to Belgium to reduce the need for a specifically Belgian sub if one starts being active there.

--Likewise, tax-induced relationships between affiliates may turn on either substitution or complementarity. E.g., being in one tax haven may either make a second tax haven less needed than otherwise, or else make the second one all the more valuable, as in "Double-Dutch-Sandwich" type schemes that require laundering profits through several successive locations. In addition, by going to one tax haven, a company may provide evidence that it is the type of company that likes to use tax havens. So, if it was also in other tax havens at "unexpected" levels, this might reflect endogeneity rather than complementarity.

--The real action may involve multi-jurisdictional clustering, not just two-country pairings. But that makes testing for empirical relationships even harder.

--Suppose that having a sub in Bermuda is a substitute for having one in the Caymans for companies following Strategy A, but a complement for those that are following Strategy B. Then one might fail to find net correlations that reflect tax planning, but it would still be going on in case after case.

The authors are quite aware of all these problems, and are struggling ingeniously to refine their empirical strategy. I look forward to the end result, as it could be both interesting and useful. But in the interim, virtue (choosing an interesting project even though it is hard) may serve as its own unfair punishment.