The OECD/Inclusive Framework's proposal benchmarked the U.S. Global Intangible Low-Taxed Income (GILTI) regime, which imposes a similar top-up tax. The difference between the OECD proposal and the U.S. GILTI is that the former defines the tax rate that triggers the top-up tax based on a global average tax rate, whereas the latter tests the tax rate based on a country-by-country (CbC) basis. Since the United States has signed on to the global tax deal, the Biden administration has proposed several provisions in the Build Back Better (BBB) Act, which raises the U.S. GILTI rate from the current 10.5% to 15% and applies the GILTI rule on a CbC basis.
The current consensus is that a CbC method would 1) collect more tax revenue from MNEs, and 2) prevent harmful competition amongst taxing jurisdictions than global average tax rates. Chris Sanchirico (Pennsylvania) challenges this consensus in his new paper, A Game-theoretic Analysis of Global Minimum Tax Design, by using a sequential theoretical game model. Sanchirico finds that implementing a global averaging method would better accomplish both goals. Readers might find Sanchirico's findings somewhat counterintuitive. It is important to note, however, that Sanchirico's paper constructs a game-theoretic model to predict MNEs' behavioral response—whether MNEs will shift their income from high tax jurisdictions to low tax jurisdictions—under either the CbC method or global averaging method and the subsequent tax policy response of high- and low- tax jurisdictions.
Under the global averaging method, MNEs may shift some profits from high tax jurisdictions to low tax jurisdictions, but a more important observation is that high tax jurisdictions do not have to lower their tax rate to the level of the global minimum tax rate. Suppose that a high tax country has the optimal high tax rate for generating the maximum amount of tax revenue from MNEs. If the high tax country lowers its tax rate, it still collects from such MNEs an amount of tax revenue equal to the imposed global minimum rate times pre-profit-shifting income because the decrease of tax revenue due to lowered tax rate and the decrease of the volume of profit shifting to low tax countries precisely offset. In short, high tax countries have no incentive to lower their rate. Unfortunately, low tax countries would still continue the harmful tax competition, lowering their tax rates to the bottom.
On the other hand, under the CbC method, the MNE's choice is pretty simple. The MNE will shift all their income into the jurisdiction(s) with the lowest effective tax rate. The low tax jurisdictions may continue the tax rate competition, but not beyond the level of the global minimum rate. Then what would be the high tax countries' response? If they maintain the high tax rate, they will collect no revenue from MNEs because MNEs will shift all income offshore. So, to collect the same amount of tax revenue, a high-tax jurisdiction must lower its tax rate to the level of the global minimum tax rate even if the global rate is substantially lower than its domestically optimal tax rate.
Therefore, Sanchirico explains that 1) high tax jurisdictions were able to achieve two goals under the global averaging method—choosing a domestically optimal (high) tax rate and effectively collecting the maximum tax revenue, but 2) high tax jurisdictions cannot achieve both under CbC method and must lower the rate if they do not want to lose the tax revenue. Thus, high tax jurisdictions would prefer the global averaging method, whereas low tax jurisdictions prefer the CbC method. MNEs are indifferent between the two methods. However, Sanchirico also notes that in terms of overall costs, firms end up paying slightly less under the global averaging method than under the CbC method because of the de minimis cost of shifting.
Overall, I believe that Sanchirico's analysis and comparisons are accurate given its various assumptions. However, I wonder whether his dual conclusion—that is, a global averaging method will result in 1) more tax revenue collected from MNEs and 2) less harmful competition—might be more relevant to high-tax jurisdictions. First, let us compare tax revenue. As Sanchirico explains, MNEs are indifferent in terms of their global tax liability under either model. Then global tax revenue collected from either model should be the same. Some countries would be relatively better off in one method over the other: high tax jurisdictions in the global averaging method and low tax jurisdictions in the CbC method. But this paper does not seem to offer an analysis that the total global tax revenue amount is bigger in the global averaging method than in the CbC method. It would be great if Sanchirico would supplement this point in his model. Nonetheless, if we limit the implication to high tax jurisdictions, Sanchirico's conclusion is correct. Second, harmful competition. Consider that the history of harmful tax competition refers to the phenomenon of a race to the bottom, particularly in developing countries. The CbC method is more effective in deterring the race to the bottom among low tax jurisdictions. Therefore, I would suggest understanding the second conclusion as relevant to the behavioral response of high tax jurisdictions. Indeed, global averaging is more effective for high tax jurisdictions not to lower their optimal (high) tax rate to the level of the global minimum tax rate. However, I would be hesitant to describe the rate reduction to the level of global minimum tax as "harmful."
Finally, this paper is a relatively complex economic paper. Nonetheless, this paper walks through an excellent analysis based on a game-theoretic model that provides further insights into how MNEs and tax jurisdictions might react under a global average model and a CbC model and promotes further research on this topic. I anticipate that the referenced separate paper regarding the “motivations, limitations, and potential extensions” of this model will prove valuable in expanding the theoretical understanding of this topic and likely clear up many of the uncertainties discussed in this review. For example, I have thought about testing the proposed 15% global minimum tax rate based on this model. If the global tax deal offers the CbC model as the forthcoming rule, we should expect the behavioral response of high tax jurisdictions as Sanchirico’s paper presents. They would have to lower their domestic corporate tax rate to the global minimum tax rate, which was not necessary under the global averaging model. It will have a significant implication for the United States, which already has a GILTI regime with the global averaging model. One way to mitigate high tax jurisdictions' incentive for lowering the tax rate is to jack up the global minimum tax rate from 15% to, for example, 18% or 21%, as suggested during the negotiation process of the global tax deal. The Tax Justice Network and Oxfam criticized the global minimum tax rate as 'too low' to curb the race to the bottom among developing countries. Sanchirico's model can offer additional support for increasing a global minimum tax rate from a novel perspective—that is, developed countries that are likely to have high tax rates would also be better off with 18% or 21% of the global minimum tax rate because those rates are closer to their domestic corporate tax rates (in the United States, 21%).