Digital Trade Agreements and Domestic Policy

But these policies are not just domestic in nature. The regulations that emerge will have an international dimension due to their impact on companies and individuals in other countries. In an ideal word, domestic regulation would not lead to trade conflict, but in reality it often does. To address the conflicts that might arise from regulatory effects abroad, there have been efforts at the bilateral, regional, and multilateral level to develop international rules on digital governance. The scope of digital trade agreements is still evolving, but, to take one example, the recent U.S.-Japan Digital Trade Agreement defines “digital product” as “a computer program, text, video, image, sound recording, or other product that is digitally encoded, produced for commercial sale or distribution, and that can be transmitted electronically.”¹ With this and other provisions that establish broad coverage, these agreements will have an impact on virtually all commercial activity on the internet and on smartphone apps.

The international rules for digital trade address specific regulatory issues such as consumer protection and spam, but they also include the broad principle of nondiscrimination, which offers more general international oversight. Two key aspects of that principle are the nondiscrimination obligation, which in the international trade context requires that domestic regulations treat comparable foreign and domestic products alike and treat comparable foreign products from different countries alike; and exceptions for various public policies, which ensure that the international trade regime does not interfere with policymaking on nontrade issues. Under the Trump administration, the United States put forward specific language during negotiations on these digital agreements that results in an overly broad approach to the nondiscrimination obligation and a narrow approach to the exceptions. This paper argues that the Biden administration should reconsider current U.S. policy in this area and adjust the proposals it has made in the ongoing World Trade Organization (WTO) e‑commerce negotiations.

As a core principle, international trade agreements generally include an obligation that parties to the agreement must not discriminate on the basis of nationality, with related exceptions that allow for laws and regulations that pursue various public policy objectives even if they violate one of the agreement’s obligations. The precise contours of the obligation and the exceptions have been the subject of intense policy debate in the traditional trade in goods and services context, centered around how to evaluate whether the law or regulation was intended to discriminate and whether it had the effect of discriminating. These debates from the trade in goods and services context should inform the emerging discussion of this principle when crafting rules for digital trade.

With regard to nondiscrimination, one key aspect is how to determine whether there has been a discriminatory effect on foreign goods or services. There are cases in the WTO system (including the General Agreement on Tariffs and Trade (GATT) that preceded it) that address the issue and can help guide the drafters of similar rules in the digital trade context. Things became complicated in the GATT/WTO when the law or regulation did not discriminate against imports explicitly but rather established criteria that applied in a way that led to a disparate impact on imports. In that situation, such an impact can constitute evidence of a discriminatory effect. For example, a measure might tax alcoholic beverages based on their alcohol content, with higher proof products subject to a higher tax. On its face, such a measure would not seem to have a discriminatory effect on foreign products. But if it happens to be, as was the situation in one case, that most foreign products have a higher alcohol content while most domestic products have a lower alcohol content, then a discriminatory effect will be found to exist.²

WTO jurisprudence in this area has evolved over the years and it is now generally accepted that the determination must be made based on the overall impact of a law or regulation on the group of imports as compared to the group of comparable domestic products. For a time, starting with a panel’s ruling in a GATT dispute in 1989, there was a view that if any individual foreign product experienced bad treatment under the regulatory distinction in the law or regulation, and any individual domestic product experienced more favorable treatment, then a discriminatory effect existed.³ That flawed approach could thus find violations even when the overall impact is neutral or favors foreign products. Over the years, though, trade law scholarship and WTO jurisprudence emerged that have largely eliminated the individual product approach from WTO disputes, although governments still argue for it on occasion when it corresponds with their interests in a given case.⁴

Trade agreements contain various obligations and have both general exceptions and exceptions that apply only to specific obligations. In the GATT/WTO system, there are general exceptions under which governments may claim that the law or regulation is tied to policies such as public health or safety, the environment, or public morals, and therefore should be permitted. For example, a government might justify regulation of online pornography, which could violate an obligation to allow the free flow of data, under a public morals exception. These exceptions include conditions to make sure they are not abused, such as a requirement that the measures not be applied in a manner that constitutes “a means of arbitrary or unjustifiable discrimination between countries where the same conditions prevail.”⁵ They also require that the law or regulation be sufficiently connected to the public policy goal. For example, the text might indicate that the measure must be “related to” that goal or “necessary to” achieve the goal.

While the use of “necessary” is a common approach to the wording of exceptions in trade agreements, it is clear that it involves a higher degree of scrutiny of domestic policies than alternative standards, such as “related to.” As interpreted and applied in WTO jurisprudence, roughly speaking, an evaluation of necessity usually involves a consideration of four factors: the importance of the policies; the contribution that the law or regulation makes to its objectives; the trade‐​restrictiveness of the law or regulation; and whether alternative measures could accomplish the same goals in a less trade‐​restrictive way.⁶ In practice, that standard is more exacting and harder to satisfy than one based on whether the measures are “related to” the objective, which involves a more straightforward look at whether there is a close and genuine relationship between the “means” used under the measure and the “ends” that the measure was designed to achieve.⁷

In the agreements it has negotiated to date, and in the proposals it has made in ongoing negotiations at the WTO, the United States has pushed a nondiscrimination obligation that is based on a consideration of the treatment of individual foreign and domestic products and an approach to the exceptions that relies on the necessity standard.

With regard to nondiscrimination, digital trade agreements usually include a basic obligation setting out this principle. For example, under the Trans Pacific Partnership (TPP), in the version negotiated while the United States was involved, the obligation reads as follows:

This continues to be the approach used in the Comprehensive and Progressive Agreement for Trans‐​Pacific Partnership (CPTPP), which was the version of the agreement signed by the other 11 TPP parties after the Trump administration withdrew, and it is also the approach in the Digital Economic Partnership Agreement (DEPA), an effort at creating a broad international agreement in this area led by Chile, New Zealand, and Singapore.⁹

However, during the negotiations on the United States‐​Mexico‐​Canada Agreement (USMCA), the United States shifted its approach. In the initial text released in September of 2018, the language was the same as that of the TPP.¹⁰ But then after a “legal scrub,” a new version of the text was released in November that had changed the language to read as follows:

The switch from “digital products” to “a digital product” could be significant for the interpretation of this provision. This change seems to reflect the GATT/WTO debate over whether to compare the entire group of foreign and domestic products, or to compare individual foreign and domestic products, and it is likely to steer the outcome in the direction of the flawed individual‐​product comparison.

To illustrate this point, imagine a hypothetical world where there are 10 search engines, with five Canadian ones and five American ones. Canada then passes a law which adversely affects—in a de facto way, without targeting nationality explicitly—one of the American search engines and one of the Canadian search engines. Common sense would tell you that this law does not have a discriminatory effect on the basis of nationality, as the number of adversely affected products is equal between the two countries. For each country, four products are not adversely affected, while one is adversely affected. However, under the strain of thinking that says the adverse treatment of any individual foreign product under a measure is enough to count as a discriminatory effect, a violation could be found. If the one adversely affected American company fares worse under the measure than any one of the Canadian companies, there will be a violation, even if overall the American and Canadian companies come out the same (for each country, 20 percent of the companies get worse treatment).

Beyond the individual product versus group comparison, there is also an argument that the language put forward by the United States does not require nationality‐​based discrimination at all because the text only requires that a comparison be between “like digital products” rather than digital products from one country compared to the “like products” of another country.¹² To illustrate this with an example, imagine that a Mexican law established criteria under which Google Search and Microsoft Bing were treated differently. Arguably, the United States could bring a complaint on the basis that one digital product from the United States was treated less favorably than another comparable digital product from the United States.

At the multilateral level, these same issues are being negotiated right now at the WTO. In a leaked version of the WTO e‑commerce text, it appears that the United States continues to push its approach to nondiscrimination because its proposal uses the singular “a digital product.”¹³

Turning to the exceptions, current U.S. policy favors the stricter necessity test, whereas many of the suggestions from other governments use a looser version of this test or do not rely on necessity at all. By supporting the more‐​exacting scrutiny involved with a necessity standard, the U.S. approach may interfere more with domestic policymaking than would a more flexible approach that simply looks at the “means‐​ends” relationship between the measure and its objective.

Exceptions can be established for the agreement as a whole or for a specific subset of obligations. For digital governance rules, a commonly used exception is one that applies in the context of the obligations on cross‐​border trade flows. After setting out a broad obligation that data should flow freely across borders, digital governance agreements then qualify this obligation with an exception. In the current U.S. agreements, a necessity test is always incorporated in this exception in some way.¹⁴

In the WTO e‑commerce negotiations, the difference in approaches is apparent, as the leaked text is at an early stage of discussion and simply notes the language put forward by different parties. The proposed language from the United States and several other countries contains a necessity standard:

But the proposals from several other countries do not. One alternative states:

Another alternative uses “necessary,” but qualified by the term “considers,” which offers the government imposing the measure a great deal of discretion:

Thus, in the WTO context as well, the United States continues its push for a narrower version of the exception.

The Biden administration has inherited U.S. policy on these issues from the Trump administration and now has a choice to make going forward. In order to craft the right balance between international and domestic governance in this area, the new administration should push for a more restrained nondiscrimination obligation and more flexible exceptions.

Amending existing international agreements such as the USMCA or the U.S.-Japan Digital Trade Agreement can be a challenge (although the United States often has the economic leverage to push for such changes), and the USMCA will be particularly difficult because it has been implemented into U.S. law by a statute. However, the administration has wide leeway in formulating its international digital governance policy going forward. In the ongoing negotiations at the WTO, and in future bilateral and regional trade deals, it should consider changing course on U.S. policy regarding nondiscrimination and exceptions in digital trade agreements.

The ongoing domestic policy debates on digital regulation are likely to be contentious, both in the United States and around the world. International rules may be helpful here, but they need to get the balance right to create a system of international cooperation that functions effectively and endures.