1. Introduction
The evolution of the data-driven economy is perhaps the most significant trend of the past decade. Digital trade has proven to really matter for growth and innovation in individual economies and the global economy as a whole.Footnote 1 Digital trade regulation has accordingly advanced rapidly in the pursuit of new framework conditions that move away from brick-and-mortar trade law and adequately reflect the practice of data-dependent trade – what we call ‘Trade Law 4.0’.Footnote 2 Since the adoption of the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) in 2018, we have witnessed a keen interest shared by governments around the world to regulate digital trade with specifically designed treaty templates. It should be pointed out that discussions and negotiations of digital trade rules have been taking place in various fora. These venues include the Work Programme on Electronic Commerce of the World Trade Organization (WTO) (WTO E-commerce Work Programme)Footnote 3 and the ongoing E-Commerce Joint Initiative (JI) plurilateral negotiationsFootnote 4 at the WTO and the current advanced e-commerce/digital tradeFootnote 5 negotiations that unfold in preferential trade agreements (PTAs) or stand-alone digital economy agreements (DEAs). This indicates that the impetus propelling the drive to negotiate rules on digital trade is in full throttle. The Trade Agreement Provisions on Electronic-commerce and Data (TAPED) dataset, administered by the University of Lucerne, has traced rulemaking developments in this vibrant field of international economic law since 2000.Footnote 6
TAPED was initially launched in 2017 as part of the research project ‘Governance of Big Data in Trade Agreements’ and received funding from the Swiss National Science Foundation Research Programme (NRP)75: Big Data (2017–2021).Footnote 7 This project was co-led by Mira Burri of the University of Lucerne and Manfred Elsig of the University of Bern. The scope and substance of the dataset were introduced to the research and policy community in 2020.Footnote 8 Since 2021, as part of the European Research Council-funded Consolidator Grant project ‘TRADE LAW 4.0: Trade Law for the Data-Driven Economy’ (2021–2026) led by Mira Burri, TAPED has been regularly updated, expanded, and modified. TAPED provides insights into the significant growth of digital trade provisions in trade agreements globally, as well as into emerging issues of relevance to data governance.
This article seeks to showcase the latest developments in digital trade rulemaking from January 2020 until the latest update of TAPED in November 2023. During this period, an increase in the negotiation of digital trade commitments in PTAs, as well as the emergence of DEAs, was observed. Since January 2020, 49 agreements covering digital trade commitments have been concluded or signed, and more are being negotiated.Footnote 9 While, at the beginning, such commitments were contained in a few provisions scattered throughout the preferential trade treaties, they have now become standard standalone substantive chapters, separate from the services and investment chapters.Footnote 10 Moreover, the treaty commitments remain shallow, reflecting a divergence of domestic regulatory approaches.Footnote 11 However, increasingly clear patterns of broader governance convergence have slowly started to emerge, closing the gap on the erstwhile deep fragmentation that characterized rules in this area.
To keep pace with these important developments, and considering that TAPED has been incrementally used in academic discourse and to inform policymaking,Footnote 12 this article updates and complements the emerging trends in digital trade rulemaking discussed with the first release of TAPED in October 2019.Footnote 13 In addition, this article showcases the evolving nature of the TAPED dataset. In comparison to the 90 different coded items that the dataset initially comprised, the evolution of the provisions in PTAs, as well as the expansion of the issues being covered, warranted the addition of new coded items (in total 34) and the recategorization of old ones into five different areas – i.e., (1) e-commerce; (2) data-dedicated provisions; (3) new data economy issues; (4) cross-cutting issues; and (5) intellectual property (IP). The codebook accompanying the dataset provides more information on each coded item and keywords used to identify the relevant provisions, as well as examples of actual treaty language. In addition to these updates, the methodology guiding the coding of the different items has also been modified. Following the typology developed by Abbott and Snidal for assessing the levels of legal legalization,Footnote 14 the coding of provisions is now according to hard and soft commitments, blending the previous categorization of ‘mixed legalization’ into soft obligations.
This paper is divided into five sections. Following this brief introduction, Section 2 offers an overview of the current landscape of global digital trade governance. Against this backdrop, Section 3 discusses the new developments unfolding in PTAs, while Section 4 addresses the emergence and significance of DEAs. In the latter two sections, we highlight the new dynamics of digital trade rulemaking regarding the actors and the substantive issues being negotiated. Thus, in Section 5, we provide an overview of the new data economy issues observed in PTAs and DEAs and conclude in Section 6, by also providing an outlook on future developments in the broadening domain of digital trade governance.
2. The Current Landscape of Digital Trade Regulation
The developments observed in PTAs and DEAs do not occur in a vacuum. They influence and are influenced by the negotiations on digital trade that are taking place in multilateral and regional fora. Within the multilateral context, in November 2023, 90 members are currently negotiating the Joint Initiative on Electronic CommerceFootnote 15 (E-commerce JI) with the hope of concluding a plurilateral agreement on electronic commerce/digital trade under the umbrella of the WTO by the end of 2023.Footnote 16 The issues that comprise the current E-commerce JI negotiations have been categorized into the following bundles of topics:
• Enabling e-commerce (e-transactions and digital trade facilitation and logistics);
• Openness and e-commerce (custom duties on e-transmission, access to internet and data);
• Trust and e-commerce (consumer protection, privacy, business trust, cybersecurity);
• Cross-cutting issues (flow of information; transparency, domestic regulation and cooperation; capacity building; special and differential treatment);
• Telecommunications (updating the WTO Reference Paper on Telecommunications Services);
• Annex (diverse set of provisions ranging from logistic services, temporary entry of e-commerce related business persons, to goods and services market access)
• Scope and General Provisions (including provisions regarding relationship to other agreements, exceptions, indigenous peoples, taxation and dispute settlement).Footnote 17
Notably, of the 90 JI participants, the majority of countries are high-income economies while only five are least-developed countries (LDCs). The relatively low rate of participation of low-income countries and nearly all LDCs (particularly from Africa) in the E-commerce JI negotiations has raised concerns over how less-resourced countries can effectively participate in crafting rules that will regulate one of the most important areas of trade policy.Footnote 18 This phenomenon could also be attributed to the critical stance they have taken against these negotiations in favour of the potentially less far-reaching WTO E-commerce Work Programme discussions.Footnote 19 Moreover, their skepticism in negotiating cutting-edge rules has been justified by their need to first understand the phenomenon and implications of digital trade before committing to rigid digital trade rules that are broad in scope.
Regional initiatives are also increasingly aiming to set policy frameworks to govern digital trade. Significant examples since January 2020 include the 2021 Mercosur E-Commerce Agreement and the electronic commerce chapter of the Regional Comprehensive Economic Partnership (RCEP). Moreover, the Digital Trade Protocol of the Agreement Establishing the African Continental Free Trade Area (AfCFTA),Footnote 20 and the forthcoming ASEAN negotiations towards a Digital Economy Framework Agreement (DEFA)Footnote 21 underline that no region wants to be left behind in the domain of digital trade regulation and that digital trade policies play an increasingly important role in regional integration.
Against this background, it is interesting to note that the divergent policy priorities of the traditional digital trade legal demandeurs (for instance, the United States (US) has shifted its attention to the Indo-Pacific, adjusted its position as to the negotiated digital trade topics,Footnote 22 and appears less concerned with trade liberalization)Footnote 23 have led to non-traditional actors stepping into the gap to craft rules that best fit their policy priorities. The DEAs are a case in point. They are predominantly being developed by dynamic countries seeking to capitalize on the economic potential of the digital economy.Footnote 24 Countries such as Singapore and Australia, which have not been conspicuous rulemakers, have become leading actors in designing new rules for digital trade. Uncoincidentally, Singapore and Australia, along with Japan, are also the co-convenors of the E-commerce JI.Footnote 25 Moreover, Japan, which currently acts as the G7 president, is an avid promoter of the concept ‘data free flow with trust’ (DFFT) and has recently endeavoured to operationalize this concept by establishing an international data governance body.Footnote 26 This complements previous efforts by the G7 to set principles on digital trade.Footnote 27 These principles tend to be liberal and include the promotion of DFFT; open digital markets; safeguards for workers, consumers, and businesses; digital trading systems; and fair and inclusive digital trade global governance.Footnote 28
As global digital trade rulemaking continues to evolve, evidence-based research can help policymakers to identify the opportunities between the most advanced international treaty frameworks and their own efforts to adopt digital trade rules and accordingly shape domestic regimes. Moreover, an overview of the evolution of international digital trade rulemaking can illuminate the outlook of norm convergence or further fragmentation.Footnote 29 TAPED serves as a source of unbiased, comprehensive, and accessible data, and seeks to inform policymaking in this regard.Footnote 30 The next section discusses the advances in digital trade rulemaking in PTAs.
3. Trends in Digital Trade Provisions in PTAs
The number of digital trade provisions in PTAs has steadily increased over time. Of the 432 PTAs currently comprising TAPED,Footnote 31 which were concluded or signed between January 2000 and November 2023, 214 contain provisions relevant for e-commerce and digital trade, and 122 have dedicated e-commerce or digital trade chapters.Footnote 32 The significant jump in these commitments came about in the past few years. From January 2020 to November 2023, 49 PTAs were concluded or signed. Of these, 44 (or almost 90%) contain provisions on digital trade or e-commerce, and 26 (or 53%) have separate chapters on the subject, indicating that the majority of PTAs being concluded contain provisions on digital trade and most are found in a dedicated chapter. In contrast with the trends observed from January 2016, the year in which the text of the Trans-Pacific Partnership Agreement (TPP) was concluded and from which the US withdrew at the start of the Trump administration, up to December 2019, 68 PTAs were concluded or signed. Of these, 55 (or 80%) contained provisions on digital trade and 33 (just above 48%) had dedicated chapters on the subject. The relatively lower number of PTAs concluded or signed during 2020–2023 could be attributed to the Covid-19 pandemic restrictions. Nevertheless, we see a steady increase in the negotiation of PTAs that include provisions or chapters on digital trade.
During the reference period of January 2020 and November 2023, there has also been increased diversity in the parties negotiating these agreements. This is most apparent in RCEP, whose members include developed and developing countries, as well as LDCs.Footnote 33 Other less visible PTAs also reflect this tendency. An example is the Comprehensive Economic Partnership Agreement (CEPA) between India and the United Arab Emirates (UAE).
In other important developments, we observe a certain repositioning of traditional actors. As noted elsewhere,Footnote 34 the European Union (EU) has repositioned itself on cross-border data flows over time starting with clauses in which the Parties commit to ‘reassess’, within three years of the entry into force of the agreement, the need to include provisions on the free flow of data in the treaty – which has been the case in the EU's deals with Japan and Mexico. The EU's much clearer position on cross-border data flows was first asserted by the post-Brexit Trade and Cooperation Agreement (TCA) with the United Kingdom (UK) and the follow-up free trade agreements (FTAs) with New Zealand and Chile.Footnote 35 The EU's newly found commitments are however linked to the higher standards of personal data protection under its General Data Protection Regulation, essentially creating a conditional data flow regime.Footnote 36 A similar repositioning is observed in China's PTAs. So far, China has negotiated an e-commerce chapter (but not digital trade chapters)Footnote 37 in at least nine of its PTAs.Footnote 38 China's willingness to participate in or influence e-commerce rulemaking is evidenced more clearly in RCEP. This agreement creates, among other things, a conditional framework for cross-border data flows with some obligationsFootnote 39 but also important exceptions and carve-outs, so that the RCEP parties reserve their policy space, particularly in the area of national security.Footnote 40 China's application to join the CPTPPFootnote 41 on 16 September 2021 can be a turning point, as the latter contains less flexible provisions on cross-border data flows and a clear ban on data localization measures, or measures requiring the location of data infrastructure or processing in a certain jurisdiction.Footnote 42 If China's application to accede to the CPTPP is successful, it will mark a radical change in China's PTA policies on data governance and potentially have implications for its involvement in the WTO negotiations. However, it is unlikely that China will be welcomed into the CPTPP because of political frictions with key membersFootnote 43 and the large (and growing) distance between CPTPP's commitments (including beyond digital trade) and China's applied regime and commitments in its current PTAs.Footnote 44
Notwithstanding advances in negotiating rules on e-commerce and digital trade, or even substantive chapters in this area, there remain several contentious issues. We highlight three of them – (1) cross-border data flows; (2) banning or limiting data localization requirements; (3) disclosure or transfer of the source code – and afterwards discuss other observable trends in digital trade rulemaking.
With regard to provisions on cross-border data flows, of all the PTAs coded in TAPED, only 49 (or just above 11% of all PTAs) contain provisions on the free movement of data. Of these, 19 are not legally binding, while 30 are legally binding. This cautious approach to including obligations on cross-border data flows persists in the most recent agreements, although we observe significant improvements. Out of the 49 PTAs concluded or signed since January 2020, 19 (or ca. 38% of the reference period PTAs) contain a provision on the subject of cross-border data flows. Of these, 16 are binding.Footnote 45 These findings are an improvement from the trends observed during the period January 2000 to December 2019. Out of a total of 383 agreements, 30 PTAs contained the relevant provisions and only 14 are binding (for a visualization of this evolution, see Figure 1). Other relevant provisions include review clauses in which the parties agree to review data flow provisions after a certain amount of time. The EU case is the most prominent example in this regard and such an update has recently been concluded with Japan.Footnote 46
Concerning the banning or limiting of data localization requirements, only 32 PTAs out of a total of 432 contain a provision on the subject. This represents 7% of the total number of PTAs. Of these, 30 are binding. Data from January 2020 until November 2023 indicate that out of the 49 PTAs concluded or signed, only 16 PTAsFootnote 49 contain a provision banning or limiting data localization requirements (for a visualization of this evolution, see Figure 2). All of these are binding, indicating a consistency to agree on prohibitions of data localization and a more differential approach to data flows, which is interesting to observe, as there is an interlinkage between these provisions and the commitments made under them.
Another important provision in the context of digital governance relates to the commitments on the disclosure or transfer of the source code. So far, only 23 PTAs of the 432 PTAs contain a provision on the subject. Of these, 22 are binding. From January 2000 to December 2019, 11 PTAs prohibit requirements on the transfer of, or access to, the source code of software owned by a person, as a condition for the import, distribution, sale or use of such software. All of these are binding. Moreover, since January 2020, a further 12 PTAs contain a provision on the subject and there has been some convergence between the US-led and the EU model in this regard, although the EU still inserts several exceptions.Footnote 51 Of these, 11 are binding obligationsFootnote 52 (for a visualization of this evolution, see Figure 3). On a much newer issue, at least six PTAs in the dataset make a separate reference to the transfer of, or access to, an algorithm, which expands the scope of the ban on forced technological transfer of source code.Footnote 53
Less contentious is the inclusion of provisions on data protection, where we can observe an upward trend. Since January 2020, 32 out of the 49 coded PTAs have included a relevant provision. Yet only 16 of these provisions are binding obligations.Footnote 54 This contrasts with previous observations for the period from January 2000 to December 2019, where out of the 113 PTAs containing provisions on the subject, 25 were binding (for a visualization of this evolution, see Figure 4). Notably, there is some convergence in the way provisions on data protection are being included in the latest PTAs, particularly in DEAs. A common reference is to ‘international standards, principles, and guidelines’. In the reference period, 18 PTAs mention these aspects. Of the 13 agreements referring to international principles, sixFootnote 56 include binding commitments, and of those referring to international standards, all but two (in other words, 11)Footnote 57 have binding provisions. In addition, we observe increased reference to ‘compatibility’ and ‘interoperability’ of data protection laws. These aspects are of increased importance in global initiatives, such as the G7 work on DFFT.
Provisions on the facilitation of digital trade are also a rising trend, reflecting the parallel developments occurring in the context of the WTO JI E-commerce negotiations, where the participating members reached some agreement on certain issues, such as online consumer protection; electronic signatures and authentication; unsolicited commercial electronic messages; electronic contracts; transparency; paperless trading; and open internet access.Footnote 58 Nonetheless, while PTAs also include these aspects, the nature of these provisions remains, largely, non-binding. For instance, out of all coded PTAs, 107 PTAs include a provision on paperless trading, but only 19 of these are binding.Footnote 59 Similar metrics are reflected in provisions on customs procedures automatization or custom data exchange systems. In these cases, 106 PTAs contain provisions on these issues, but only 26 of them are binding.Footnote 60 In these and other areas, there is however a concerted effort to achieve interoperability and to comply with existing standards or to participate in international standard-setting activities.Footnote 61
Regarding the relationships between the digital trade chapters with other agreements or chapters of the same agreement, some PTAs regulate this relationship in the event of inconsistency. We observe that from January 2020 to November 2023, only six PTAs contain relevant rules.Footnote 62 In the particular case of the relationship between digital trade chapters with IP rules, only three PTAs contain a relevant provision. This is not trivial, as it conditions the normative value of the agreed-upon digital trade provisions; it is also an issue that deserves further scholarly attention, in particular in combination with the carve-outs and exception clauses formulated in the digital trade chapters and other parts of the treaties.Footnote 63
Concerning the justiciability of these provisions, less than half of the e-commerce and digital trade chapters in PTAs during January 2020 and November 2023 are subject to dispute resolution under the respective treaty. Of the 49 PTAs that have been concluded or signed in this time frame, 28 of these agreements have a dispute settlement chapter that applies to the digital trade chapter.Footnote 64 These observations complement previous studies. Based on data of 275 Regional Trade Agreements notified to the WTO until May 2017, Monteiro and Teh found that out of 76 agreements, 58 of the e-commerce provisions were covered by dispute settlement.Footnote 65 Nine expressly excluded the provisions from dispute settlement, while the other nine excluded some provisions from dispute settlement.Footnote 66 Other studies have also noted that in South–South PTAs, the digital trade provisions tend to fall outside the scope of the dispute settlement mechanism.Footnote 67 Nevertheless, there are some nuances in these observations. For example, the RCEP e-commerce chapter is currently not subject to dispute settlement. However, the dispute settlement chapter will apply to the parties that have agreed to its application after the five-yearly general review.Footnote 68
Another relevant issue that is becoming increasingly clear as a feature of digital trade rulemaking in PTAs is the inclusion of exceptions and carve-outs. These provisions are relevant as countries seek to retain policy space and define the bounds of their digital sovereignty.Footnote 69 These conditionalities are also a vehicle for making the digital trade commitments politically feasible for the treaty parties.Footnote 70 The critical importance of these exception clauses has merited the re-structuring of the items being coded in the TAPED database. Exceptions typically occur where a prohibition is made – for instance, the parties may not impose requirements on the location of computing facilities (data localization). The general tendency is for parties to add a provision stating that nothing prevents countries from requiring the localization of data to comply with national laws or specific national interest considerations immediately after the provisions. In some PTAs, general exception clauses, such as those found in WTO law under Article XX of the GATT 1994 of Article XIV of the GATS apply mutatis mutandis or they are reproduced in the relevant section of the treaty.Footnote 71 Sometimes this reproduction is abridged in that elements of the WTO general exception clauses – in particular, the exhaustive listing of the permissible legitimate objectives – are missing.Footnote 72 These are general public policy exceptions, including privacy protection. Countries also add national security exceptions, which at times refer to Article XXI of the GATT 1994 or Article XIVbis of the GATS, applying mutatis mutandis.Footnote 73 Furthermore, we observe an emerging pattern in the issues expressly excluded from digital trade chapters. The following are frequently excluded to preserve policy space: government procurement,Footnote 74 information held or processed by or on behalf of a Party, or measures related to such information, including its collection,Footnote 75 internal taxation,Footnote 76 financial services,Footnote 77 and audio-visual services.Footnote 78 The EU tends to include in addition a broadly defined ‘right to regulate’, which can potentially cover all policy areas linked to digital trade and undermine the treaty commitments made. There is in this sense a layering of different exceptions, the ultimate effect of which demands, as earlier noted, a careful analysis of each PTA in order to understand the impact of the treaty and the overall policy space that remains available under PTAs.
Among the other interesting trends to observe and that will merit future updates of TAPED is the inclusion of clauses on Special and Differential Treatment (SDT). Some PTAs, such as the CPTPP and RCEP, include clauses on SDT. This is an aspect worth keeping an eye on, as these provisions could enable LDCs, which have not engaged meaningfully in digital trade rulemaking so far, to undertake reduced commitments at the outset and increase them over time or to have longer periods of implementation or support in the implementation process. Previous experiences in negotiating and implementing such types of clauses in the multilateral context could be informative in this regard, showcasing once again the mutually reinforcing relationship between PTAs and multilateral/plurilateral negotiations.
A final aspect to note concerns the scope of the PTAs in the sense of the expansion of the issues being negotiated, which has warranted the update of TAPED. In addition to those provisions on facilitation of digital trade, data protection, and data flows, as well as provisions with any sort of impact on the conditions for digital trade, such as provisions on IP, non-discrimination, general and specific exceptions, PTAs often include, what we call, ‘new data economy provisions’, discussed later in this article. The next section delves first into exploring the DEAs, considered the most innovative agreements on digital trade.
4. Spotlight on the Digital Economy Agreements
As previously mentioned, a significant development in digital trade regulation is the emergence of the so-called DEAs. Since 2019, the following five DEAs have been signed: (1) US–Japan Digital Trade Agreement (DTA), 2019; (2) Singapore–Australia Digital Economy Agreement, 2020 (SADEA);Footnote 79 (3) Digital Economy Partnership Agreement (DEPA) among Chile, New Zealand and Singapore, 2020; (4) UK–Singapore DEA, 2022;Footnote 80 and (5) Korea–Singapore DEA, 2022.Footnote 81 Except for the US–Japan DTA and the DEPA, the other three agreements are part of, and upgrade, an existing FTA.Footnote 82 In other words, the US–Japan DTA and the DEPA are the only true stand-alone DEAs. The conclusion of DEAs raises questions about their compatibility with the multilateral legal framework.Footnote 83
Under WTO law, advanced free trade agreements (FTAs) covering goods are required to comply with Article XXIV of the GATT 1994 and FTAs covering services must comply with Article V of the GATS. As they do not form part of an FTA, within the meaning of WTO law, the legal status of the US–Japan DTA and the DEPA under WTO law remains unclear. Specifically, there is some doubt whether they would qualify as valid agreements contemplated under Article XXIV of the GATT 1994 or even Article V of the GATS.Footnote 84
WTO law has specific requirements that FTAs must fulfil to be exempt from the most-favoured-nation obligation (in other words, to be considered a valid trade liberalization agreement under WTO law). Article XXIV:8(b) of the GATT 1994 requires ‘duties and other restrictive regulations of commerce’ to be eliminated on ‘substantially all the trade between the constituent territories in products originating in such territories’. In Turkey–Textiles, the Appellate Body noted that WTO Members have never reached an agreement on the meaning of the term ‘substantially all the trade’.Footnote 85 However, the Appellate Body considered that it is clear that ‘substantially all the trade’ is not ‘all the trade’ but ‘is something considerably more than merely some of the trade’.Footnote 86 The preamble of the Understanding on the Interpretation of Article XXIV of the [GATT 1994] (Understanding) suggests that omitting any major sector of trade would not fulfil the requirement of ‘substantially all the trade’.Footnote 87 Moreover, during negotiations on the meaning of this term, some WTO Members suggested that an FTA that covers 95% of tariff lines at the six-digit level would fulfil the condition under Article XXIV:8 of the GATT 1994.Footnote 88 Digital trade is a large part of trade nowadays but it has not yet reached its full potential. As an illustration, in the US e-commerce sales currently only constitute around 14.5% of all retail sales.Footnote 89 In 2020, the corresponding figure in Australia was 9.4%, 25.9% in the Republic of Korea, 11.7% in Singapore, and 23.3% in the UK.Footnote 90 It is thus questionable whether a stand-alone DEA would fulfil the requirements of being a valid FTA under Article XXIV:8 of the GATT 1994, as it does not encompass ‘substantially all the trade’.
Similarly, an FTA that is considered to be a valid economic integration agreement under Article V:1(a) of the GATS must have substantial sectoral coverage. Footnote 1 explains that ‘[t]his condition is understood in terms of number of sectors, volume of trade affected and modes of supply. In order to meet this condition, agreements should not provide for the a priori exclusion of any mode of supply.’ Trade volumes aside, there is no consensus on whether services provided via digital trade entail both Mode 1 (cross-border supply) and Mode 2 (consumption abroad) or only Mode 1.Footnote 91 Nevertheless, even if, hypothetically, digital trade involved both those modes of supply, it necessarily excludes, at least, two modes (Modes 3 and 4, i.e. commercial presence and movement of natural persons). Therefore, an agreement based on (maximum) two modes of supply would ostensibly not fulfil the conditions for a valid services FTA under the GATS. In sum, it appears doubtful that the DEPA and the US–Japan DTA would fulfil the WTO requirements to constitute a valid PTA. Interestingly, neither agreement has been notified to the WTO.Footnote 92 Concerns about the legal status and WTO legal compatibility of previous negotiations have also been made, such as the Trade in Services Agreement TiSA.Footnote 93 In that case, the means of its incorporation into the WTO legal framework was uncertain.Footnote 94 This issue remains unresolved because the negotiations were discontinued.Footnote 95
That aside, a few trends have emerged in the parties and typical features of DEAs. Thus far, countries that have signed these agreements are Australia, Chile, Japan, New Zealand, Singapore, Korea, the UK, and the US. Therefore, except for Chile, these agreements are typically concluded between or among high-income countries.Footnote 96 Moreover, Singapore has emerged as the global leader in concluding DEAs. It is involved in each of these agreements except the US–Japan DTA. This indicates Singapore's strong interest in digital trade and its legal entrepreneurship in this area.
Common trends have emerged in all the DEAs and specifically in the ones in which Singapore is involved. We first provide an overview of the common features of the main substantive provisionsFootnote 97 in all DEAs thus far. We then elaborate on the common features in Singapore's DEAs, which are generally cutting-edge topics, including some of the ‘new data economy issues’ addressed in the next section.
The following provisions are common to all DEAs:
• They substantially exclude government procurement and government data.Footnote 98
• They have codified the WTO moratorium on customs duties on electronic transmissions, made it permanent, and provide a carve-out for the imposition of internal taxes.Footnote 99
• The Parties have undertaken to adopt the UNCITRAL Model Law on Electronic Commerce 1996 in the provisions on domestic electronic transaction frameworks.Footnote 100
• The Parties undertake not to prohibit cross-border data flows. However, the provisions include exceptions for legitimate public policy reasons and adopt the necessity test under Article XX of the GATT 1994 or Article XIV of the GATS.Footnote 101
• They all prohibit data localization requirements as a condition for market access or for conducting business in the territory of the other party/parties.Footnote 102
• They all require the adoption of laws protecting personal information.Footnote 103
• They all have provisions requiring the adoption of consumer protection laws.Footnote 104
• They all require the adoption of measures to control Unsolicited Commercial Electronic Messages (spam).Footnote 105
• They all prohibit the disclosure of proprietary information relating to cryptography for ICT as a condition for market access or doing business in the territory of the other party/parties.Footnote 106
• They all have cooperation undertakings to address cybersecurity threats.Footnote 107
• They all have provisions on open government data, generally requiring that they are anonymized, have descriptive metadata, are machine readable, and are in an open format that allows them to be searched, retrieved, used, reused, and redistributed.Footnote 108
While the innovations in the DEAs are exciting, it is also interesting to note the two features of the US–Japan DTA that differ from the agreements involving Singapore. First, the former agreement does not have a dispute settlement provision. This is atypical of previous US agreements. For instance, the US–China Economic and Trade Agreement (the Phase I Agreement), which was signed later than the US–Japan DTA and the USMCA, include a dispute settlement chapter.Footnote 109 Second, the US–Japan DTA has a provision excluding platform liability for tech companies, which necessarily demands the application of Section 230 Communications Decency Act (CDA).Footnote 110 Section 230 CDA insulates platforms from liability for the third party but has been recently under attack, even in the US,Footnote 111 and has become constrained through regulation in other jurisdictions.Footnote 112
It can also be observed that while the US had the first-mover advantage in that it has concluded agreements in digital trade since 2000,Footnote 113 Singapore has certainly taken the baton and ran with it. As previously mentioned, all DEAs but one have Singapore as a party. Of course, the DEAs are not the only agreements that have digital trade chapters. However, the level of ambition and new issues in other contemporaneous agreements do not reach the levels incorporated in the DEAs. For example, during the same period, the UK concluded its post-Brexit PTAs;Footnote 114 RCEP and the India–UAE CEPA were signed; and the negotiations of the EU–New Zealand FTA were concluded. Even with ostensibly like-minded advanced economies,Footnote 115 the number of new data economy issues covered in the digital trade chapters of PTAs that are not DEAs is comparatively low. The next section provides an overview of the ‘new data economy issues’ included in FTAs and DEAs.
5. Charting Unknown Waters: The New Data Economy Issues
As digital trade rulemaking in PTAs continues to evolve, so does TAPED. The addition of new data economy issues is an example in this regard. Since January 2020, we observe, more and more, that PTAs, and in particular DEAs and selected UK PTAs, possibly inspired by Singapore, contain provisions on aspects that transcend traditional digital trade rulemaking. These new provisions are for instance on competition policy related to the digital economy; digital identities; digital inclusion; fintech and lawtech cooperation; artificial intelligence; standardization, interoperability, or mutual recognition regarding electronic means. Most of these issues are found in DEAsFootnote 116 but some of them have also permeated into PTAs’ digital trade chapters. For example, the India–UAE CEPA of 2022 includes provisions on digital identities.Footnote 117 The New Zealand–UK FTA, in addition to the provisions on digital identities, also includes language on artificial intelligence.Footnote 118 Some of these provisions reflect parallel developments occurring elsewhere. For example, policies to facilitate the integration of micro-, small-, and medium-sized enterprises (MSMEs) in electronic commerce are key to harnessing the benefits of the digital economy and figure prominently in the current plurilateral negotiations at the WTO.Footnote 119 Relevant provisions in this regard are also increasingly found in most recent PTAs and DEAs.Footnote 120 Moreover, the Korea–Singapore DEA specifically mentions start-ups as companies that should be promoted and supported in the digital economy.Footnote 121
Interesting is also the incorporation of provisions on digital inclusion, which explicitly appear in five agreements – DEPA, Chile–Paraguay FTA, India–UAE CEPA, Singapore–UK DEA, and the UK–New Zealand FTA. All but the UK–New Zealand FTA have at least one developing-country party. Digital inclusion provisions initially involved providing economic opportunities for MSMEs. However, they have subsequently expanded to encompass women, rural populations, low socio-economic groups, disabled people, and Indigenous Peoples.Footnote 122 The UK–Singapore DEA is unique in that it specifically targets fair labour conditions, worker protection, and improving digital skills.Footnote 123 Moreover, its Parties also recognize the digital divide between countries and undertake to promote the participation of other countries in digital trade.Footnote 124
We also observe that the UK has progressively included some of the recent innovations found in DEAs and added new issues to its negotiating agenda. The inclusion of lawtech in the UK–Singapore FTA makes it the DEA with the most new data economy topics covered so far. Of all the PTAs with provisions on artificial intelligence, the UK–Singapore DEA is the most comprehensive, seeking cooperation on issues and developments relating to artificial intelligence, including among others, ‘ethical use, human diversity and unintended biases, industry-led technical standards and algorithmic transparency’, ‘joint deployment and test-bedding opportunities’, and ‘opportunities for investment in and commercialisation of AI and emerging technologies’.Footnote 125 Similarly, while all the PTAs with provisions on artificial intelligence seek to promote collaboration for the development and adoption of frameworks that support the trusted, safe and responsible use of these technologies, the New Zealand–UK FTA explicitly makes reference to the ‘principles and guidelines of relevant international bodies, such as the OECD and the Global Partnership on Artificial Intelligence’.Footnote 126 The progressive nature of some of the UK's recent agreements goes in tandem with the domestic reforms that the UK has undertaken in recent years to position itself as a tech ‘superpower’, which seems to be increasingly reflected in its digital trade policy.Footnote 127 It also signifies a distancing from the EU model, which can be problematic having in mind the bindingness of the TCA and therewith linked obligations for the UK to provide an essentially equivalent level of personal data protection.Footnote 128
A final remark on artificial intelligence provisions in PTAs is the explicit link that these provisions make to the digital economy, and in some cases, to the promotion of trade and investment flows.Footnote 129 So far, none of the provisions on artificial intelligence in PTAs refers to the human rights implications of these technologies, including possible privacy concerns and issues which have prominently featured in recent discussions on generative artificial intelligence.Footnote 130
6. Concluding Remarks
The recent developments in digital trade rulemaking are exciting and digital trade law has become one of the most active areas of international economic law. We increasingly observe the routine incorporation of digital trade commitments in PTA negotiations, and the inclusion of aspects not observed before – even in ground-breaking agreements, such as the 2018 CPTPP that created a digital trade template that has diffused in a great number of follow-up agreements.Footnote 131 We thus see a clear expansion of global digital trade law. The TAPED database has sought to accordingly adapt and cover these new developments.
While, admittedly, the large majority of the most recent provisions encourage cooperation and are non-binding, there is a tendency to be innovative and include new issues that otherwise have not appeared in negotiating agendas. This seems to indicate that while binding rules are difficult to negotiate (or are not desired in some areas), there is a willingness to include new issues as a way to limit regulatory divergence and create a platform for cooperation. Moreover, we are witnessing the legal entrepreneurship of non-traditional actors, such as Singapore, which have become innovators in the digital trade law domain. Finally, as more and more issues might be included in PTAs’ and DEAs’ negotiating agendas, these types of agreements may increasingly become rule frameworks for all data economy issues, going beyond purely digital trade commitments. The TAPED database is well positioned to keep up-to-date with these developments and to continue providing current and unbiased data to policymakers and other stakeholders to facilitate well-informed decision-making, which may be particularly important for actors with low regulatory capacity, such as developing and least developed countries.