Hassett’s Evidence on Transfer Pricing and the U.S. Trade Deficit

In my last post, I questioned Kevin Hassett’s claim that transfer pricing manipulation was responsible for half of our trade deficit and asked what was the paper he referenced. We have the text of his speech:

There is another important factor to consider when thinking about how these changes will affect the economy. A recent NBER working paper (Guvenen, Mataloni, Raisser and Ruhl 2017) argues that profit shifting by large multinational firms causes part of their economic activity to be attributed to their foreign affiliates, leading to an understatement of U.S. GDP. Moreover, this profit-shifting activity has increased significantly since the mid-1990s, resulting in an understatement of measured U.S. aggregate productivity growth. The authors correct for this mismeasurement by “reweighting” the amount of consolidated firm profit that should be attributed to the U.S. under a method of formulary apportionment. Under this method, the total worldwide earnings of a multinational firm are attributed to locations based upon apportionment factors that aim to capture the true location of economic activity. The authors use equally weighted labor compensation and sales to unaffiliated parties as proxies for economic activity. Applying the formulary adjustment to all U.S. multinational firms and aggregating to the national level, the authors calculate that in 2012, about $280 billion would be reattributed to the U.S. Given that the trade deficit was equal to about $540 billion, this reattribution would have reduced the trade deficit by over half in 2012.

Formulary apportionment can take on many forms. One form is to allocate taxable profits by sales but this approach would likely lead to a different allocation of income than a true arm’s length approach especially for a nation that imported a lot of sourced produced abroad. Wasn’t this realization central to that debate over the Destination Based Cash Flow Tax idea? This NBER paper, however, does something else as noted by this summary:

To correct for this mismeasurement, we use confidential MNE survey data collected by the Bureau of Economic Analysis to construct apportionment factors that distribute total transactions in income on FDI among the parent and affiliates in the MNE. We consider both labor compensation and sales to unaffiliated parties as apportionment factors, as these variables are most likely to identify the real production taking place in each location.

Using BEA data to infer the extent of transfer pricing manipulation is a Herculean task so I guess we should applaud any effort to do so. And while the addition of labor compensation might improve upon the formulary approach over a pure sales based apportionment, this approach still strikes me as misleading. The value-added under arm’s length pricing can be seen as the sum of labor compensation and profits from the tangible and intangible assets created in a jurisdiction. Unless capital to labor ratios are internationally equal, apportionment based on labor compensation still misses the mark. Apportionment based on sales rewards the distribution function ignoring where the product was produced. Of course measuring the market value of capital is difficult especially if capital includes intangible assets and multinationals often transfer the rights to intangible assets to tax havens at valuations far below fair market value. As such, I would take these estimates of transfer pricing manipulation with a grain of salt. And as has been noted, the ability to manipulation transfer pricing has little to do with the effect of corporate tax rates on how productive activity is sourced.