Mitchell Kane's "A Defense of Source Rules in International Taxation"

My colleague Mitchell Kane has recently published a quite interesting article on source rules in international taxation, taking a view that differs more from mine on the surface than I think it does in underlying substance.

A commonly quoted line about determining the source of income, from Hugh Ault's and David Bradford's piece on the subject more than 25 years ago, says that the notion of "source" lacks coherent economic content. I recall Bradford frequently noting that, while there is an intellectually coherent Haig-Simons income concept, there is no such benchmark for source. I've frequently quoted this line, as it's seemed both (a) clearly right and (b) related to the difficulties that source-based taxation presents in practice.  But I've also been aware that (a) it's easy to determine source in some cases, and (b) in other cases it depends on how you define it - e.g., origin basis vs. destination basis (more on this shortly).

Kane agrees at least arguendo that there may be no coherent economic definition of source, but then says: Why would the definition have to be an economic one?  "Household" or "family," for example, can't be satisfyingly defined for tax or other transfer system purposes unless one informs it with ideas taken from somewhere else that express one's underlying purposes.  For source, he sees the purposes as relating to how countries try to divvy up income tax bases between themselves.  He approaches this as a multilateral cooperative process, whereas - just as a matter of taste or interest; either approach can be fine - I tend to think about it more in terms of unilateral processes that may be conducted in the shadow of particular strategic interactions.

Then comes an important point that I've been thinking of writing about, although at the moment I'm engaged in my literature book - origin-based vs. destination-based income concepts. Say I sit at my desk in New York and write a book in Bengali that I will sell for large profits to people on the Indian subcontinent. Under the origin concept, the income is U.S.-source because that's where I did the work.  Under the destination concept, the income is sourced in India and Bangladesh because that's where the sales occurred.

In the context of, say, a retail sales tax or value-added tax, we often hear about the point that they can use either the destination basis (which is the universal norm) or the origin basis (as under some progressive consumption tax models that would use a VAT as part of their structure).  It's a familiar point that, in the consumption tax environment, one can use either, and in the long run it doesn't matter which one uses, leaving aside some extremely important issues pertaining to transition and administrability.

Income is commonly called an origin concept, and I've written about the difficulty of trying to run an income tax off the destination basis.  Absent some very fancy footwork, the equivalence from the consumption tax context is undermined by the fact that how long one saves before consuming (e.g., the time between exports to earn $$ and imports to spend it) affects income tax liability, whereas it hypothetically doesn't affect the present value of consumption tax liability using constant rates across time.

But in fact there are plenty of destination-based source concepts in existing income taxes.  For example, the U.S. income tax sources labor income based on where the work occurs, but it sources royalties based on where the property is used. So what if you use labor to create royalties? Then formalism determines the source of income.

The mix between origin and destination concepts in the income tax creates various tax planning opportunities, but that's not to say, at least right off, that a given country isn't better off using both in different places rather than just one.

Now let's consider source in the taxation of multinational enterprises.  Using transfer pricing between affiliated entities is apparently an origin based concept, but doesn't work too well.  Shifting to one-factor formulary apportionment that was based solely on sales would make it a destination concept.  But as Jerry Seinfeld and George Constanza would say, not that there's (necessarily) anything wrong with that.  When Reuven Avi-Yonah, Kim Clausing, and Michael Durst propose replacing transfer pricing with formulary apportionment, they are surely not "wrong' by reason of urging the use of a destination concept in a mainly origin-based system. After all, suppose the switch has predominantly good effects, whether adopted just by the U.S. or more generally. And the assessment of that depends on all sorts of wholly separate things (e.g., how manipulable would the sales factor be) that are quite distinct from the theoretical choice between income and consumption taxation.

Anyway, back to Kane's article.  The fact that one can use either origin or destination concepts to define the source of income has figured in my thinking as evidence for the prosecution in calling the source concept incoherent. Kane instead views it as evidence for the defense, showing that there are two ways one can actually do the thing, and it is simply a question of deciding which is better.  He mainly comes out pro-origin, because of the points that make it a better fit with the income concept.

Is the choice "really" evidence for the prosecution, as I have thought, or for the defense, as he argues? Once again this is really a matter of perspective - it's not a case where reasoning logically from required premises leads to one conclusion or the other.

Kane has one other main point in defending the coherence (whether it's economic or not) of the source concept.  He notes the problem of, say, deciding where interest should be deducted, when a multinational has both interest expense and gross income that presumably was produced by using the borrowed funds.  (But of course we don't really know what income this "really" is, given that the fungibility of money makes it quite meaningless where the particular loan proceeds were directly sent.)  But he notes that this is simply a broader difficulty of applying the income concept which applies even in the context of one-country taxation where there is no source issue.

Let me broaden that a bit.  Consider the "synergy" problem in transfer pricing. Standard example: U.S. and French company, if separately owned, would have earned $1M each. But they're co-owned by a multinational enterprise, leading to synergies that increase their combined net income to $2.5M. Where did the extra $500,000 of synergy income actually arise? I see it as in principle a bilateral monopoly bargaining problem - one could imagine the U.S. and French entities negotiating over it, if we posit that neither could realize it separately.  But since we can't really say anything about bilateral monopoly bargaining outcomes in the abstract, there's seemingly no good answer here.

So source might be viewed as an economically (and otherwise?) incoherent concept with respect to the $500,000 - though not as to the underlying $1 million that "should" be the minimum reported in each jurisdiction.

Once again, however, Kane's argument, to the effect that this is not a "source" problem as such, can be made here.  Suppose you have related entities in the U.S. that are taxed at different rates - e.g., because we have a special tax rate for the finance industry as compared to the electrical generating industry, and a given conglomerate entity or set of entities does both.  Now it matters which entity has how much income, but there is comparably no coherent answer to the bilateral monopoly bargaining issue that's raised by the synergy income.  So the transfer pricing problem isn't so much a source problem as a related-parties problem.

How much does all this matter in the end?  The issues of ultimate importance really turn on the consequences of alternative rule designs, as judged based on some underlying set of metrics.  So I'm not substantially more (or for that matter less) sanguine on the question of how well or poorly one can use source concepts in practice than I was before reading the Kane article.  But it provides very useful clarification regarding a set of conceptual issues that interest me, and on which I may still write a bit (in light of this article) at some point down the road.