Leap of Faith: The Economic Impact Assessment of the Pillar One and Pillar Two Blueprints

Oct. 12, 2020, was a holiday for North Americans. Above the 49th parallel, Canadians gathered for Thanksgiving dinner, celebrating the fall harvest and the end of summer. Below the 49th parallel, Americans gathered to celebrate Columbus Day/Indigenous People’s Day. Oct. 12 was also important for a third reason: the OECD Secretariat released the Pillar One Blueprint and the Pillar Two Blueprint (the Blueprints, for short). Since then, tax professionals around the world have been busy reading and sharing their views on the Blueprints in webinars, virtual panels, blogs, and articles.

The Blueprints are not short. The Pillar One Blueprint is 230 single-spaced pages totaling, by my count, nearly 122,000 words; the Pillar Two Blueprint another 246 pages with nearly 123,000 words. The OECD Secretariat has requested comments by Dec. 14 on a list of assigned questions about the Blueprints, which are available in a third document, Questions for Public Comments (Questions, for short), also issued on Oct. 12 and itself another 5,000 words. In Questions, the OECD Secretariat asked for public comments on 12 areas that it identified as open issues in the Pillar One Blueprint; the 12 areas collectively contain 38 individual questions, some of which have their own sub-questions. For the Pillar Two Blueprint, the Secretariat identified 10 major issues areas and a total of 62 separate questions for comments. And, yes, 38 + 62 = 100; the tax community was assigned 100 questions by the OECD Secretariat to answer in detail in two months!

How busy tax and transfer pricing professionals, already overwhelmed with issues related to Covid-19 and the worst global recession since the 1930s Great Depression—and in the last quarter of the fiscal year—can possibly address most, much less all, of the 100 questions within 63 days is not clear. It’s unfortunate that the Secretariat has set such a short window in the middle of a pandemic and financial crisis as I doubt that multinational enterprises (MNEs) and tax practitioners will have time to do the appropriate due diligence given the limited window in which they have been asked to provide answers.

In the two earlier rounds of comments in March and November 2019, only a few of the largest MNEs became involved in the consultative process for Pillars One and Two (primarily pharmaceutical and digital MNEs since they were likely to be the most severely affected). Despite recent requests for more MNEs to “get involved” in assessing the Blueprints, I suspect that there simply will not be the time available for the private sector to comment in any detail.

Moreover, given the U.S. election results, with former Vice President Joe Biden declared the next President on November 7, it may be months before the new U.S. Treasury Deputy Assistant Secretary for International Tax Affairs (who serves as the U.S. delegate to the OECD Committee on Fiscal Affairs) is in place. President-Elect Biden is also expected to propose substantial changes to U.S. tax policy in 2021, including changes to the U.S. GILTI (global intangible low-taxed income) regulations, which would clearly affect policy choices for Pillar Two.

As a result, I suspect individual MNEs and practitioner groups will target the questions most closely related to their own interests and hope (another leap of faith) that someone else will cover the other questions. The proposed radical changes to the existing international tax system will therefore likely receive far less detailed analysis and commentary than the Blueprints deserve. This is unfortunate because the public comments are supposed to provide guidance to the OECD Secretariat and members of the Inclusive Framework on the next round of revisions to the proposals. In addition, as I complained about on the previous public consultation round, the questions for public comment are narrowly focused on technicalities of implementation rather than on open debate of the policy aspects.

My conclusion is that the Pillars One and Two process will become, this time around, an even greater leap of faith than the earlier rounds, and the OECD/G20 might well be advised to put the whole process on the backburner until at least the summer of 2021. At the very least, the Secretariat could have extended the deadline for responses to the Questions through the first quarter of 2021, providing a longer window for informed feedback.

The Need for Transparency

When the Blueprints and Questions were released, they were accompanied by a fourth document: the Economic Impact Assessment (EIA, for short). The EIA makes an heroic attempt to estimate the economic impacts of the proposals in the Blueprints. The EIA is 283 pages totaling nearly 140,000 words, which brings the total word count for the four documents to 390,000 words to be read, absorbed, and commented on in 63 days—more than 6,000 words per day!

The Acknowledgements to the EIA states that the analytical work in the project was carried out by five OECD economists, with the assistance of and contributions from other individuals and organizations (EIA, p. 5). My heartiest congratulations to these five “unsung heroes” for their herculean efforts to quantify the Blueprints!

However, after all the extraordinarily hard work involved in calculating the economic impacts of the Blueprints, the empirical results for each jurisdiction (country) were unfortunately not publicly released. What the public received were the estimated effects for 222 jurisdictions, grouped by per-capita income categories (with number of countries in brackets), into high- (64), medium- (105) and low- (29) income countries and investment hubs (24). The EIA reports that “there was no consensus over whether or not jurisdiction-specific estimates should be publicly released as part of the economic impact assessment” (EIA, p. 19) so the decision was made to aggregate results such that individual country results could not be determined.

Even more disconcerting was the fact that the OECD Secretariat did not share the full results with any of the national tax authorities that are members of the Inclusive Framework. Tax authorities that requested access to the figures were provided—on a “confidential and bilateral basis” (i.e., OECD Secretariat to tax authority)—with their own jurisdiction’s results, not with all the data or all the empirical work or all the results. They were provided also with “revenue estimation tools” where a tax authority could vary the percentages and “estimate the impact on tax revenues in their jurisdiction” (EIA, p. 121).

Industry groups have also needed to do their own modeling of the effects of the Blueprints and, given the short time window, only the largest and most affected MNEs likely have the time and resources to perform their own analyses. This is unfortunate because the impacts appear to be highly firm specific, depending on factors such as the MNE’s activities (whether they are in-scope for Pillar One), global organizational structure (centralized or decentralized), and geographic footprint (including tax havens and investment hubs).

Given that the OECD/G20 is proposing such major and important changes to the international tax system, to not provide the most affected parties, MNEs and tax authorities, with the detailed analysis done by the Secretariat to predict the effects, forces them to also engage in a leap of faith!

As an academic and scholar, I know that transparency in research is a critically important step to improving the rigor of the scientific process. Sharing data and methods (“opening the book” on what was done) enables other researchers to review, discuss, confirm or challenge and suggest revisions. All major scholarly journals are committed to transparency in the peer review process as a core ethical value, necessary for ensuring public trust in their research findings. That the EIA’s data, methods, and findings, at the individual country level, are not an “open book” for national tax authorities (particularly for jurisdictions that lack the capacity to perform their own analyses) is of great concern to me, and I hope also to other tax scholars, professionals, and policymakers. Without better access to the underlying data, methods, and results at the country level, how can any democratically elected government assure its citizens that its policymakers have done their due diligence on such major changes to the existing international tax rules?

Still, within the limits of what is available, it may be useful to shed some light on the economic impacts of the Blueprints by deconstructing the Economic Impact Assessment. With little time to prepare responses to the Questions, I expect most practitioners will focus their attention on the Blueprints and Questions and at best quickly scan the EIA. Moreover, none of the 100 questions mentions the EIA, and that lack of mention also discourages reading and analyzing the document. That is unfortunate because, in my opinion, the EIA does provide a very helpful road map to understanding (or trying to understand) how the Pillar One and Pillar Two proposals might work in practice and where the problematic “bumps on the road” are likely to occur. It may even be possible to tease out some likely effects at the jurisdiction level from the available data even though the OECD Secretariat has not provided these results.

Below I provide a short guide for tax professionals to the EIA roadmap. The guide may help answer the question I posed to myself, and perhaps others have asked themselves also: How accurate are the EIA estimates of the revenue impacts that tax jurisdictions can expect if the Blueprints are adopted? Can its numbers be trusted or do they also involve an additional leap of faith? With that purpose in mind, I provide an overview of the EIA’s modelling of both Blueprints. In a follow-up article I provide a more detailed analysis of the EIA’s modelling of Pillar One.

The Strengths and Weaknesses of Economic Impact Assessments

The purpose of an economic impact assessment for any policy change is to determine the size and direction of the economic effects on the affected parties, calculating both direct and indirect impacts and over both the short and long runs. The short-run (static) analysis compares the effects against an assumed counterfactual such as the status quo, with “all else” being held “equal.” The long-run (dynamic) effects allow for affected parties to adjust their behaviors in response to the policy change, for example, by changing prices, quantities, or investment. Economists generally expect the long-term effects to be larger than the short-term ones due to the greater elasticities of demand and supply.

Economists are well aware of the strengths and limitations of the approaches and techniques used in economic impact assessments. The study must be based on the best and most recent data years, cover all affected parties, exclude or control for confounding events, and use confidence internals and sensitivity tests to ensure that the results are robust to alternative specifications of the models. Robustness checks and triangulation of data and methods are recommended best practices for empirical work that are well established and understood by social scientists. Economists are also aware that even the best of their econometric models are flawed. The goal of economic modeling is to identify trends and deliver proof of concept, not necessarily to achieve a precise quantification or to design public policies.

The purpose of the EIA was structured along the lines of a typical economic impact assessment; as the EIA itself states, “this report analyses the economic and tax revenue implications of the Pillar One and Pillar Two proposals” (p. 10). The Blueprints were the proposed policy change and the economists were asked to predict the short- and long-run effects at the global and country levels.

Somewhat similar big questions have been addressed by economists before, typically involving preferential trade agreements. For example, in 1988, three economists, for the Commission of the European Communities, assessed the economic impacts of deepening regional integration in Western Europe by moving from a customs union to a common market. This year, seven International Monetary Fund economists issued a report assessing the potential economic impact and challenges of the new African Continental Free Trade Area. Economists have also assessed the impact of tax policy changes on foreign direct investment (FDI) and tax revenues; for example, modeling a 10% cut in EU corporate income tax rates and estimating the revenue losses for developing countries from double tax treaties.

The EIA: A Herculean Task

The economists writing the EIA, however, faced an even more difficult task than in earlier assessments, for two reasons. First, the Blueprints have multiple places where decisions still must be made (e.g., thresholds, percentages, items that are in or out of scope). Admittedly, most of these gaps are likely to be filled by decisions based on political “horse trading” rather than on their economic impacts, but the economists still had to “guesstimate” what the likely final decisions might be and use those parameters in their estimates.

Second, the data simply did not and do not exist for the economists to be able to accurately—or even with reasonable accuracy—answer the research question they were assigned. Having worked with a variety of public, confidential, and proprietary datasets on MNEs and FDI for more than 30 years, I understand the difficulties and hurdles that the EIA authors have had to overcome. The fine-grained data needed to assess the short-run, much less the long-run, economic impacts of Pillars One and Two for all countries simply do not exist. Even the most sophisticated, best resourced tax administrations in terms of tax gap analysis—such as the IRS or HMRC—cannot perform this analysis properly because it is simply impossible to gather and properly connect the necessary data.

To create their dataset, the EIA authors were therefore forced to cobble together data from a variety of existing datasets; collect and clean the data for the countries, firms, and markets where reasonably solid data exist; and come up with various ways to fill in the missing observations. These research challenges are well understood by data scientists and university researchers. To the authors’ great credit, the EIA does discuss the problems, and provides some explanation for and assessment of the choices that were made.

Still, unless you have worked with huge datasets, been forced to make “best efforts” to address the holes and problems, done the empirical work, and written up and published the results, you really cannot understand the leap of faith involved in addressing “big questions with big datasets” that are riddled with holes and inconsistencies.

Data Problems and Solutions

The EIA economists faced data problems and made decisions that readers of the Blueprints who have not also studied the EIA may well have missed. While no one of these issues on its own creates a major data hurdle for the results, their number should provide policymakers with a “Hic Sunt Dracones?” as I worried in an earlier article.

Some of the specific data problems and solutions in the EIA estimates are the following:

• Only one year of data (2016) is used in the calculations; recommended best practice is to use at least three years. The year 2016 is also prior to major tax changes such as the 2017 U.S. Tax Cut and Jobs Act and implementation of the first BEPS round.
• Most variables have data available only for a subset (often small) of jurisdictions so that the missing observations for the remaining jurisdictions must be proxied. See, for example, how consumer-facing business (CFB) data for 16 jurisdictions was used to estimate data for 222 jurisdictions (EIA, pp. 40-43).
• Some jurisdictions are so small they have no MNEs; the EIA assumes that all jurisdictions have at least 50 CFB and 15 ADS multinationals to ensure these jurisdictions have In-scope Destination-based Sales (EIA, n. 19, p. 73).
• Outliers are excluded from some calculations (e.g., Hong Kong, India, and others from some of the destination-based sales calculations (EIA, n. 12, p. 72).
• Amount B, the proposed minimum floor for routine marketing and distribution activities, is left out of the analysis and estimates for Pillar One.
• The Subject to Tax Rule (STTR) is left out of the analysis and estimates for Pillar Two.

In addition, the overall calculations in the EIA are affected by the following decisions:

• The Pillar One Amount A estimates assume all Market jurisdictions receive 100% tax relief on their share of Amount A from Residence and/or Source jurisdictions, and 100% compliance by all jurisdictions with no defections and no U.S. safe harbor. This seems highly unlikely; instead, given the number and variety of countries involved, iterative gaming behaviors with partial tax relief and partial compliance is more likely. Some EIA modeling of the possible gaming scenarios, especially by “tax relieving” jurisdictions, would have been helpful in assessing the longer term effects.
• Implementation of Pillars One and Two will likely cause strategic responses not only by governments but also by MNEs. Strategic responses by MNEs were not built into the EIA estimates, with the exception of modeling changes in MNE investment costs. It would also have been useful, but further afield, to explore how Pillars One and Two would affect the MNE-state political bargaining games familiar to international business scholars.
• Chapter 4 of the EIA estimates the costs of non-consensus on Pillar One but uses a “worst case” scenario where non-consensus leads to proliferation of digital services taxes (DSTs) and an international trade war. Not surprisingly, the costs of non-consensus are predicted to be very high. A more realistic comparison would have been a world where some (but not all) jurisdictions sign on to Pillar One and/or Pillar Two, with partial and varying levels of implementation and compliance, and a continuation of DSTs in some but not all jurisdictions. In other words, the most likely scenario is a much messier one that lies between the two polar cases modelled in the EIA of full adoption and implementation of Pillars One and Two versus an all-out trade war.

Conclusion

The OECD is an international organization focused on economic cooperation and development. International tax policy therefore must be a critical mission for the OECD, given the importance of tax revenues for national economic development and the destructive effects that poorly designed taxes can have on distorting FDI flows. We do not know the true costs of the existing international tax system on either tax jurisdictions or FDI. The costs were likely much higher before governments began implementing the OECD’s first round of BEPS tax policy changes in 2017–2018. We therefore should not idealize the status quo and we should encourage national tax authorities (including the U.S. Treasury) to implement the Multilateral Instrument (MLI). However, the OECD Blueprints and the political processes on which they have been built, in my view, require multiple leaps of faith by the different parties involved in this process. The leaps are too large for comfort, especially given the high stakes involved.

The lure of data is undeniable. Everyone likes “hard numbers.” Economists and other data scientists know, however, that hard numbers often do not reflect reality. The estimates based on hard numbers can be at best somewhat inaccurate and at worst highly misleading, and the uninitiated are likely to place far too much faith in the results. For these reasons, responsible data science requires accountability, data transparency, and “open books” to build public trust. Without full and frank exchange of information, how can we trust our policymakers to perform the due diligence necessary to ensure that the Pillar One and Two proposals are unlikely to backfire?

History is littered with failed policy experiments that have had disastrous consequences, such as China’s “one child” policy and Venezuela’s current political and economic disaster. Government policy mistakes can bankrupt an economy for years, wreaking more devastation than natural disasters. We should be particularly careful when the proposed policy changes affect large numbers of households, businesses, and countries, especially when they have already been weakened by a worldwide pandemic.

The Blueprints are clearly a leap forward in terms of their level of innovativeness and their distance from the existing international tax rules. The Economic Impact Assessment is also a leap forward in terms of its sophisticated economic modeling and use of “big data to address big problems.” However, as Oliver Treidler has argued and I also fear, the proposed leap forward will not be to a better place but rather into the “abyss” of greater tax uncertainty, complexity, and disputes. That brave new tax world is likely to be much worse that the current international regime with its well-understood principles, standards, and dispute settlement mechanisms, especially once the first round of BEPS Action Items are fully implemented and the MLI broadly adopted. There are problems that still need to be addressed, which may best be handled by unbundling the individual proposals in the Blueprints and moving forward only on the ones that strengthen the existing international tax rules. There are better, simpler, and less invasive ways to tax the global profits of digital MNEs in the 21st century.

This column does not necessarily reflect the opinion of The Bureau of National Affairs, Inc. or its owners.

Author Information

Lorraine Eden is Professor Emerita of Management and Research Professor of Law at Texas A&M University, leden@tamu.edu. Helpful comments on earlier drafts were provided by Joan Hortalà Vallvé, Jeffrey Owens, Philippe Paumier, Niraja Srinivasan, Ognian Stoichkov, Oliver Treidler, and Charles Hermann. The opinions expressed herein are of the author, who accepts responsibility for any errors or omissions.