International tax policy: domestic expenses that produce foreign income

The latest issue of the National Tax Journal contains a brief "Comment and Reply" section on Jim Hines' 2008 article, "Foreign Income and Domestic Deductions." The comment is by Johannes Becker of the Planck Institute in Munich and Clemens Fuest of the Said Business School at Oxford (coincidentally, two places I am going this summer), and Hines offers a very brief reply.

The exchange is important for people interested in the topic. However, as it's written mainly for economists (with equations playing a prominent role), it may not get the broad attention and readership among other interested individuals that it deserves.

Hines starts from the premise that adopting exemption for the foreign source income of domestic firms is optimal. The question presented is what should then be done with respect to the domestically incurred expenses of producing foreign source income. E.g., suppose a U.S. firm borrows in the U.S., incurring interest expense, in order to invest abroad. Or suppose a portion of its domestic headquarters operations are devoted to producing foreign source income that the U.S., having hypothetically adopted exemption, is not going to tax.

There has been widespread agreement that, at least in principle, one would want to disallow domestically incurred expenses that produce untaxed foreign source income. In effect, since the expenses are an input to determining foreign source income, which one has decided to exempt/ignore, they should be ignored as well. But Hines' article argues that "the only policy consistent with efficiency ... is to permit full domestic deductibility of expenses incurred in the home country." With all due respect to Jim, I think it's fair to say that this has met with widespread skepticism.

I had noticed that Jim's argument seemed too lawyer-like in a bad way (if I may say so as a lawyer). That is, the claim is largely one of logical consistency with exemption, in effect treating the decision to adopt exemption as if it were a precedent.

Becker and Fuest make this a lot clearer. They start by quoting the following statement from Hines: "Exempting foreign income from taxation implies that the government values equally one dollar of after-tax domestic income earned by home-country firms and one dollar of after-foreign tax foreign income, since home-country firms make this tradeoff at the margin."

Let's try a simple numerical example to make this more salient. If the U.S. adopts exemption and has a 35% domestic rate, a U.S. firm will be indifferent between earning $100 in the U.S. ($65 after paying U.S. tax) and earning $65 abroad after paying foreign taxes. Since exemption gives the firm this tradeoff, Hines asserts that this tradeoff is indeed the socially optimal one from a U.S. national standpoint, and is the one that the government, if benevolent, employs in its social welfare function. As Becker and Fuest note, this necessarily means that, in the benevolent government's social welfare function, domestic taxes paid are worth zero (!!!!!!!!!!!!!).

A more plausible social welfare function would hold that a dollar of domestic taxes paid is worth $1, since the government could simply give it to someone.

One way of describing the underlying error (made clear by Becker and Fuest) is that it confuses first-best with second-best optimality. Suppose an exemption system is best for reasons that relate to the government's lack of market power with respect to taxing foreign source income on a corporate residence basis. (The argument is ultimately similar to that against imposing tariffs if the government lacks the market power to make foreigners bear the incidence of the tax.) Even though this might make exemption, like not having a tariff, the best choice, it doesn't imply that domestic taxpayers have optimal incentives in all respects given, for example, that other domestic taxes which are second-best need to be imposed.

Analogously, suppose that imposing a wage tax is best, all things considered, as in a typical optimal tax model where efficiency and distributional objectives are being traded off (and where, as there's just one period, there is no difference between an income tax, consumption tax, or wage tax). The fact that a wage tax is optimal, all things considered, does NOT mean that tax revenues have a social value of zero and that people making work versus leisure decisions (where the return to work, but not the value of leisure, is taxed) have exactly the set of incentives that the government considers optimal.

Becker and Fuest, using a model with a more plausible social welfare function in which domestic tax revenues have value, find that deductions yielding foreign source income should not be allowed. And if expenses producing domestic versus foreign source income cannot be distinguished, partial deductibility of some kind (such as from using an apportionment formula) is optimal. Finally, they find that full deductibility "subsidizes the ownership of foreign affiliates and therefore leads to inefficiently high levels of such ownership."

Hines' reply appears largely to demur. Rather than defending his earlier line of analysis, he argues that foreign investment yields positive externalities domestically. Hence, it should be subsidized, rather than exempt. He does not address, however, what the optimal Pigouvian subsidy for positive-externality-yielding foreign investment should look like. I would be surprised if it took the form of subsidizing such investment that happens to use domestically incurred expenses.