Rethinking the Magnitude of Profit Shifting

But if you ask an accounting specialist, such as Tuck professor Leslie Robinson, she might come up with one you’ve never considered: the prevalence of equity income reported on income statements of the various foreign subsidiaries that comprise a multinational enterprise (MNE). Equity income is income reported by a parent company in one country but that is earned, and also reported, by a subsidiary in another country. In 1990, at the dawn of globalization as we know it, equity income represented 27 percent of aggregate foreign profits of U.S. MNEs. By 2016, that percentage had gone up to 67.

What this means, in practical terms, is that U.S. MNEs have become much more complex creatures, with multi-tiered subsidiary structures used to conduct their foreign operations. In research co-authored with Katharina Lewellen, a finance professor at Tuck, Robinson highlighted that these complex structures appear to exist largely so that U.S. MNEs can reduce their tax liability on foreign profits.

But even as this practice grew in popularity, experts disagreed on the fiscal effects of MNE strategies that shift profits to low-tax jurisdictions. Global estimates ranged widely, from annual tax revenue losses of $80 billion to $280 billion. Robinson has always been skeptical of those numbers, and now she and co-author Jennifer Blouin of Wharton can prove their gut feeling was correct. In a new working paper—“Double counting accounting: How much profit of multinational enterprises is really in tax havens?”—they show that errors in previous analyses of profit shifting have resulted in vast overstatements of the problem.

Robinson and Blouin first noticed the error during a conference two years ago. The presenter, who is considered an authority in this area, was explaining some graphs showing the amount of income U.S. firms were reporting they held in tax havens. Robinson and Blouin were very familiar with the Bureau of Economic Analysis (BEA) data on which the graphs were based—they had used it many times before in their own papers—and realized fairly quickly, after looking at it more closely, that the presenter had counted equity income multiple times in her analysis, thus creating a highly inflated final figure. Robinson and Blouin pulled up some aggregate data posted on the BEA website directly following the presentation and saw the error straight away. Other researchers have made the same error. 

The problem arises from a mistaken understanding of the way MNEs are asked to report financial data of foreign operations for purposes of U.S. international economic accounts. Each member of the MNE group prepares a separate company income statement. When one foreign subsidiary owns another foreign subsidiary, as they often do, the owner’s income will include both its own income and that earned in the countries where its subsidiaries are located. When that income from other countries is included in an analysis, it’s essentially counted twice or more. “It’s a consequence of tiering your ownership structure and having your subsidiaries owned indirectly by other subsidiaries,” Robinson says. “So if there are 15 layers, you’d count the same income 15 times. Double counting is an easy slogan, but it’s actually a lot worse than that. A lot of multinationals have more than two tiers.”

Counting equity income multiple times has resulted in numerous misleading estimates of profit shifting, Robinson says. One researcher has asserted that the U.S. in 2012 lost between $77 billion and $111 billion in tax revenue to tax havens. After adjusting the BEA data correctly, and changing nothing else about the methodology, Robinson and Blouin estimate the actual loss to be between $10 billion and $32 billion. Another researcher has argued that in 2013 the U.S. lost $130 billion in tax revenue to profit shifting. The corrected figure from Robinson and Blouin is $80 billion. “The point is that different methodologies will produce very different estimates, but so will misuse of the data,” Robinson says. “I am not sure why these estimates should be controlling the narrative surrounding the magnitude of profit shifting. But they do.”

And the problem is not limited to U.S. data. “We want to highlight that the double counting issue that we address is endemic to all countries’ national statistics,” they write. Robinson and Blouin are working on a second project that will require access to a number of different countries’ national statistics experts. “It’s going to be a long road, but we have to get to bottom of this; we think the issue is even prevalent in other U.S. data sources,” Robinson says.

There are a few reasons why it’s important to have an accurate picture of profit shifting. For one, the BEA just opened up its research program to more academics, so the data is going to be used more frequently in studies. “There will be a lot of people using the data for different things, and we want them to understand how the data works,” Robinson says. More importantly, countries, blocs, and organizations such as the IMF rely on this data to set tax policies and laws. The U.S., for example, reduced the corporate income tax rate from 35 to 21 percent in 2017 in part to incentivize companies to stop using tax havens. If policy makers knew the actual magnitude of profit shifting, they may not have reduced the rate so drastically.

Finally, there is an ongoing global tax policy debate centered on digital business models that is trying to determine the appropriate allocation of taxing rights for various market countries of MNEs. However, Robinson notes, “if we do not have a solid grasp on the actual distribution of profits of MNEs across countries, how can we effectively debate about how the current distribution of profits should be reallocated?”