Mystery Math: The U.S.-China Phase-1 Purchase Figures Do Not Add Up

Available Downloads

CSIS Briefs


The U.S.-China Phase-1 commercial agreement brought to a close, at least temporarily, a bitter trade war that witnessed a series of tariff hikes and increasingly hostile rhetoric. A central component of the deal is China’s commitment to expand purchases of American goods and services by over $200 billion by 2021, a massive and unprecedented increase in just two years. Although seemingly a major American victory, an examination of the details reveals some serious weaknesses with this part of the deal. The baseline figures appear to be artificial constructs of the negotiating parties, and the growth targets are not realistic – even before the slowdown resulting from the coronavirus outbreak. The purchase agreement has three other problematic elements: its details are secret, implementation will require extensive government micromanagement, and enforcement very well could be arbitrary. The policy community needs to closely monitor this deal and use it as a benchmark to evaluate the Trump administration’s overall approach toward the global trading system.


With the world appropriately focused on responding to the daunting challenges created by the COVID-19 outbreak, the U.S.-China Phase-1 deal, which seemed so important just weeks ago, has temporarily faded from view. But before long, it will come back into focus, and when it does, we will find implementation, particularly on purchases, to be a tall task. The reason is not simply that the numbers were unrealistic to begin with, even leaving aside the challenges created by the virus, but that they were literally unreal.

Before the pause button was pressed, the policy community and stakeholders were having a constructive debate about whether the deal was worth it. Did China’s concessions— on intellectual property rights, non-tariff barriers for agricultural goods, currency, openings in financial services, the purchases, and a novel enforcement mechanism—justify the tariff war? (My colleague Claire Reade reached a mixed conclusion.) Embedded in that specific question is a broader one about the efficacy of shifting away from an international trading system built on multilateral free trade toward a regime rooted in bilateral managed trade agreements.

Determining whether the Phase-1 deal is worth it, in the short and long term, requires more than judgments about vague developments. It requires knowing the actual value of what is in the deal and being able to track its implementation going forward. Unfortunately, as far as the purchase commitments go, we are in no position to do so. Those details are not only confidential; they also do not make sense.


Many international treaties and agreements have confidential side notes or annexes—language and commitments the parties need to keep from public view for everyone to get to “yes” and to promote the greater good. Yet the Phase-1 deal is distinctive in that its signature element, China’s commitment to increase purchases of American goods and services by $200 billion by the end of 2021, is shrouded in secrecy.

Chapter Six of the agreement, entitled “Expanding Trade,” covers the purchase commitments. It spells out that Chinese imports of American goods and services should grow by $76.7 billion in 2020 over the base year of 2017 and by another $123.3 billion in 2021. It also lists each of the 558 individual items that the deal covers across each of the four basic categories—manufactured goods, agriculture, energy, and services. That said, the text does not specify the 2017 baseline amounts for either overall U.S. exports, the four categories, or any of the hundreds of individual items. This leaves us wondering what those baselines are.

The simplest way to determine those baseline figures is to use U.S. Commerce Department data for each item. This method (see Table 1), first attempted by the Peterson Institute’s Chad Bown (and replicated by my team), indicates that the covered items collectively account for $134 billion of the $186 billion in 2017 U.S. exports to China; if true, total U.S. exports to China should rise by 57 percent in 2020 and 92 percent by 2021. Planned growth for energy exports under this method is particularly massive: 242 percent in 2020 and by a total of 443 percent by the end of 2021.

However, it turns out that this method is inaccurate. When the agreement was released, the U.S. Trade Representative also issued fact sheets on each part of the deal, including one on the purchases. That document provides specific figures on the absolute levels that U.S. exports should reach in 2020 and 2021 for each of the four categories. One can then subtract the amounts each category should grow, provided in Chapter 6, from these targets, and that yields the correct 2017 baselines for the four categories. As Table 2 shows, doing so results in baseline numbers far higher in each of the four categories, reaching $210 billion in total, $75.5 more than via the first method. This is seemingly impossible given that total U.S. exports in 2017 were only $186 billion.
So how is that possible? The agreement’s true baseline figures are not derived only from American export data but also from Chinese import data. Chapter 6 states that, “Official Chinese trade data and official U.S. trade data shall be used to determine whether this Chapter has been implemented.” This does not mean that both sides are free to use their own data to subjectively determine compliance but that the American and Chinese negotiators together created the baseline figures by combining elements from both set of numbers. Chinese figures are typically higher because of the difference in how U.S. and Chinese customs treat Hong Kong. American Commerce Department data treats as an export to Hong Kong the full value of goods that reach the city even if a portion of those goods are re-exported elsewhere, including to mainland China; Chinese import data includes that re-exported component in its own import figures.

According to China, its total 2017 imports of the items in the agreement were not $134 billion but $195 billion (out of a total of $242 billion). The actual baseline figure for each item in the deal was chosen from either the American export data or the Chinese import data (Excel, PDF). As Figure 1 shows, this yields an overall larger baseline number than would have been possible using either side’s data on its own.

The two sides agreed to keep the item-by-item baseline figures secret, and because there are so many items, there is no easy way for outsiders to determine the specific mix of American and Chinese data that yield the true amounts for each item and the overall totals. The Trump administration has not provided the actual baseline figures because in certain instances that would be equivalent to disclosing the specific export information of individual companies, some of whom likely account for a very large share of any individual item. This might in turn result in strange trading patterns of these firms’ stock or other undue influence on their business activities. As a result, the U.S. and Chinese governments have made these figures classified.



There are two other mysteries surrounding the deal’s purchase numbers. The first is why 2017 instead of 2018 or 2019 was chosen as the baseline year. One possible explanation is the tariff war. According to American data, total U.S. exports, which had been growing gradually prior to 2017, fell over $10 billion in 2018 and 2019, with the biggest drops occurring in agriculture and services. (By contrast, Chinese data shows no overall drop off, with only a dip in imports of agricultural goods.) Using a more recent year may have permitted setting lower targets in some categories, but it may have highlighted the embarrassing slowdown caused by the conflict (see Table 3). A second alternative is that 2017 was the last year before higher tariffs, and hence, would mark a more “natural” baseline. One challenge with this rationale is that the American side viewed the overall pattern of trade as artificial from the start and that many of the tariffs from the trade war have been left in place. And still yet another more mundane possibility is that the deal was being negotiated for so long that the two sides did not dare undo already agreed upon targets for fear of scuppering the entire package and needing to start over.
A final puzzle concerns the choice of the actual products, the mix of American and Chinese baseline amounts, and the specific targets for 2020 and 2021. Based on my interviewing, it does not appear that the targets were set based on formal modeling of how exports for individual items would rise if Chinese tariff and non-tariff barriers were lowered; nor were they determined through extensive consultation with companies or industry associations about their hopes if China should constrain industrial policy and further open its markets. Although there likely were some logical inferences made on the assumption of reduced barriers, it appears that the selected items and final targets were based heavily on extensive bargaining between the two sides, bounded by a core American concern that the total headline export growth figure would be $200 billion, an important goal of President Trump.


The mathematical contortions of the trade data may have been necessary to make the numbers “work” and reach a deal, but the ultimate result is a figment of the imagination of both sides. There is a chance that China will radically expand its imports to America’s satisfaction, but it is more likely that the purchase deal will unravel, in no small part due to the manner in which it was reached.

The ambiguity of the numbers also is closely tied to three other problematic features of the purchase agreement.

Secrecy. While confidentiality for sensitive negotiating issues is understandable, an important consequence in this instance is to deprive outsiders of the ability to monitor China’s compliance with this critical part of the agreement. Without knowing the baseline numbers, even when the U.S. and Chinese customs authorities publish new quarterly data and full-year 2020 figures in early 2021, observers will be unable to determine whether Chinese imports fall short, meet, or exceed the targets. Such a consequential part of an agreement between the world’s two largest trading nations should be open to external evaluation. Beyond this principle is a practical one. Yes, there are differences in the two sides’ specific figures, but by indicating which items are included, speculators still have useful information to set their own investment strategies. So, it is not clear that their chosen level of secrecy actually provides the intended protections.

Micromanaged Trade. A second consequence of this deal is the depth of government intervention that will be required to achieve the purchase targets or agreement to modify them. Previous American experiments with managed trade have revolved around a small number of goods, for example, automobiles and rice with Japan. They have never covered the entire trade relationship with another country. Many of the 558 items in the agreement are composites of still more specific products. One can only wish good luck to the Chinese and American bureaucrats responsible for tracking real-time commerce across so many goods.

Reducing extensive Chinese government intervention was the original aim of the Trump administration’s tough approach. Ironically, this agreement, instead, reinforces the legitimacy of Chinese authorities directing economic activity (not to mention legitimating American interventionism as well). The result is even more distortions, not less.

One might counter by suggesting that the biggest distortions existed in the status quo ante, as a result of Chinese restrictions, and that the agreement’s targets, although set by bureaucrats, would be the likely outcome were China to liberalize its import regime. To some extent this makes sense, but it is extremely difficult to know precisely by how much Chinese imports would rise across each of the individual categories and how quickly such growth would occur. In fact, although American exports to China ought to rise over the longer term, as one long- time China expert told me, U.S. exports could very well fall initially because a reduction in subsidies and protection would initially lead to a drop in Chinese consumption of all goods, domestic and foreign.

Arbitrary Enforcement. The purpose of having a bilateral, self-managed enforcement system was to raise the likelihood of Chinese compliance by speeding up the evaluation process and giving the United States the right to quickly mete out penalties without waiting on the WTO. But speed—the agreement sets up a step-by-step appeals process totaling 70 days—does not guarantee consistency or clarity. Given the vast number of purchase items and other commitments in the agreement, decisions about how to treat violations could very well be based on convenience of the moment, business lobbying, or other ambiguous criteria.

And because the appeals process is also confidential, it is possible outsiders may never know when complaints are filed or how they are resolved. If another goal of the agreement was to eliminate moral hazard and reinforce a commitment to the rule of law, it is hard to see how these provisions forward that mission.

The exact same challenges exist with regard to how to potentially provide relief to China as a result of the coronavirus. As China’s economy has slowed dramatically, consumption of all sorts, domestic and foreign, has fallen, making it even more unlikely China could reach the Phase-1 deal’s purchase targets. The agreement has a force majeure provision (Article 7.6-2), but it does not give China the unilateral right to modify or suspend its commitments, allowing only that, “In the event that a natural disaster or other unforeseeable event outside the control of the Parties delays a Party from timely complying with its obligations under this Agreement, the Parties shall consult with each other.” As with every other element, such discussions may occur entirely without the public’s knowledge, and their outcome could be based on a whole range of economic and political factors not rooted in the text of the agreement itself.


American trade policy is being fundamentally remade. While we will be keeping a scorecard on the Phase-1 purchase deal despite its hidden elements, it is equally important to be fully aware of the tectonic shifts in the American approach to trade that this agreement represents and to raise concerns about the efficacy of these changes.

The Trump administration did strike a major deal with Beijing, and markets breathed a sigh of relief—at least until the virus outbreak. It is possible that their risky approach of unilateral pressure will yield greater progress than was achieved through patient multilateralism and result in an unprecedented boon for American business, strengthen the American economy, and provide a stronger foundation for the global trading system. But it is also possible that the opposite will occur and that the promised benefits of the U.S.-China deal will not materialize. The purchase agreement’s mishmash of secret figures does not bode well. Moreover, it is conceivable that the U.S.-China deal will generate a chain reaction of other bilateral deals. If so, the current complex “spaghetti bowl” of regional and plurilateral agreements could be replaced by a more fragmented and problematic “bean bowl” of discriminatory bilateral deals, and they could be as secret as the U.S.-China purchase agreement. The result would be an evisceration of the multilateral trading system’s principle of most-favored nation (MFN).

And so while it is tempting to give the Trump administration the benefit of the doubt, focus on the individual elements of the agreement, and wait a year or more to draw conclusions on the efficacy of this approach, observers need to be more proactive and not succumb to a new kind of unjustified patience built of faith, expediency, apathy, or fear. The policy community needs to discuss and debate not only the details of any individual deal but the administration’s overall vision and strategy and provide a basis for determining its success or failure. If we do not, the United States and the global economy may be much worse off than had Washington maintained the posture adopted by earlier administrations.

Scott Kennedy is senior adviser and Trustee Chair in Chinese Business and Economics at the Center for Strategic and International Studies in Washington, D.C.

The author is indebted to Shining Tan, the CSIS Trustee Chair’s research associate, for the outstanding research assistance she provided for this brief.

This brief is made possible by general support to CSIS. No direct sponsorship contributed to this brief.

CSIS Briefs are produced by the Center for Strategic and International Studies (CSIS), a private, tax-exempt institution focusing on international public policy issues. Its research is nonpartisan and nonproprietary. CSIS does not take specific policy positions. Accordingly, all views, positions, and conclusions expressed in this publication should be understood to be solely those of the author(s).

© 2020 by the Center for Strategic and International Studies. All rights reserved.

Please consult the PDF for references.

Scott Kennedy
Senior Adviser and Trustee Chair in Chinese Business and Economics