Sunday, October 11, 2015

The tell

U.S. international tax policy debate is ramping up for a big fight over the OECD BEPS proposals. There is clearly going to be a lot of scholarship with competing bottom lines, and also a lot of straight-out advocacy, which on the corporate side will be very well-funded. The lines between scholarship and advocacy, and between plausible advocacy and hack work, will not always be clear.

With this in mind, I was disappointed by a piece I read this weekend, lead-authored by the eminent economist Gary Hufbauer, whose work I have admired and indeed cite in my international tax book. Hufbauer et al argue that OECD-BEPS is predominantly bad for the U.S. 

Okay, this is one side in the debate, and it is potentially plausible in at least some scenarios, although I lean more to the other side. Just on a personal level, I happen to have friends whom I respect on both sides of the debate.

But then I go to Appendix A at the end of this piece. It makes the claim that criticism of Apple and its affiliates for paying too little tax is factually inaccurate.  To quote:

"Apple and its stakeholders probably paid to the IRS around $25 billion between June 2014 and June 2015. This works out to a robust 45 percent of global profits in that year. Excluding the [shareholder-level] capital gains estimates, Apple and its stakeholders still contributed nearly $16 billion, or about 30 percent of global profits. The accusation that Apple and its stakeholders are shirking their responsibility as US taxpayers is not supported by the facts. "

Wow, all that well-documented tax planning to such little effect. Should Apple perhaps fire its CFO and tax director?

But here are a few details about how Hufbauer et al made these calculations:

--The measure of U.S. taxes paid that they use is almost certainly false as used. It's derived from GAAP reporting that treats deferred U.S. taxes that will in theory be paid by Apple at some point in the future as if they had been currently paid. This would be fine if the expected present value of the deferred taxes was the same as the amounts reported. But this is quite unlikely under real-world circumstances that permit companies to anticipate never paying this tax, at least at the current 35% rate. (Consider future tax holidays, lowering of the U.S. corporate rate, enactment of exemption without a present value-equivalent transition tax, etc.)

Apple would probably have had no problem persuading its accountants to treat the associated earnings as "permanently reinvested abroad" (or "PRE" in standard lingo), permitting it to treat the deferred taxes as zero rather than as equivalent to current cash taxes. It's well-known that Apple doesn't use the PRE designation to nearly the same degree as many other companies, apparently because they want to position themselves in public debate as relative good guys. Given not just this point but Apple's actual repatriation practices, we know with certainty that its current period cash taxes are far lower than its GAAP-reported taxes. Given the issues concerning how, if, and what tax rate future taxable repatriations will occur, there is a strong argument that current period cash taxes are generally as good or better a measure of actual tax burdens, even though what actually matters is true expected present value.

For this reason, work that I respect generally attempts to figure out cash taxes paid, even though they aren't reported. For a study of Apple, given their unusual non-reliance on PRE, it's absolutely necessary to try to do this, or at least to acknowledge the point.

Ignoring (or not knowing) this, as Appendix A does, is not what one would expect of competent professional work.

--The estimated tax on employee compensation is included in the numerator. But the employees' taxable income from that compensation is not included in the denominator.  That fails the test of using a consistent standard that compares apples to apples, etc.

I titled this blog post "the tell." This of course is a poker term, referring to giving yourself away. I had particularly in mind the following two sentences: "To sum up these calculations, Apple and its stakeholders probably paid to the IRS around $25 billion between June 2014 and June 2015. This works out to a robust 45 percent of global profits in that year." 

This is just really bald. From one sentence to the next you have a shift between the inconsistent numerator and denominator, written in such a way as to obscure it. Whose are the global profits, after all? Not those of "Apple and its stakeholders."

BTW, they even include Social Security and Medicare taxes estimated to have been paid by Apple employees. While one could in principle argue for this - leaving aside both incidence questions and the issue of linkage between Social Security taxes and benefits, which could turn these into prepayments for retirement income and services rather than "taxes" - this is a large departure from accepted frameworks of comparison. Once again, it means that the careless reader may end up comparing apples to oranges.

--They also include, in the taxes-paid numerator, shareholder-level capital gains and dividend taxes.  Now, here at least (unlike with employee compensation) there is a reason for keeping the income that gave rise to these tax liabilities out of the denominator. If you have a pure double tax on corporate income, you wouldn't count the same income twice, at both the entity and shareholder levels, in the course of determining the tax burden on the overall enterprise.

But capital gains can relate to past earnings that might not have been taxed at the time, and/or to expected future earnings that might not be taxed in the future.

Hufbauer et al note that the period they measure includes an "exceptional period for Apple [in terms of] its stock market valuation." Especially if this was an atypical period for Apple stock, significant adjustments might be required. Now, Appendix 1 does suggest that they made certain adjustments for both timing and period-related elements of the capital gain. But once I've seen them using the GAAP taxes-paid measure without reference to its possible inaccuracy, and once I've also seen their putting taxes on employee compensation in the numerator but keeping the associated income out of the denominator, it becomes much harder to give them the benefit of any doubt on how they adjusted. There is simply no reason to trust them on this issue, once one has seen how they handled other issues.

No comments: