¹ OECD, ‘Leveraging Digital Trade to Fight the Consequences of COVID-19’, p. 5.

² UNCTAD, ‘COVID 19 and E-Commerce’, 2021, p. 124.

³ The Geneva Ministerial Declaration on Global Electronic Commerce, Ministerial Conference: Second Session Geneva, 18 and 20 May 1998.

⁴ The New Zealand–Singapore FTA signed in 2000 is known as the first trade agreement which stipulates a digital trade related provision in Article 12 (Paperless Trading) under Section 4 (Customs Procedure).

⁵ The Australia–Singapore FTA in 2003 had the separate chapter on electronic commerce. The US–Japan Digital Trade agreement is recorded as the first stand-alone type of digital treaty.

⁶ The digital trade chapter of the USMCA is the most representative example of the recent digital trade agreement. Paragraph 1 of its Article 19.2 states the following: ‘the Parties recognize the economic growth and opportunities provided by digital trade and the importance of frameworks that promote consumer confidence in digital trade and of avoiding unnecessary barriers to its use and development’. This is not the case only for developed countries. Digital trade rules by developing countries, such as the ASEAN Agreement on Electronic Commerce, also stress their function against trade barriers. Also, the preamble of the ASEAN Agreement on Electronic Commerce focuses on its contribution in lowering barriers to entry and operating costs for businesses.

⁷ Bossche, P. Van Den and Zdouc, W. (2021) The Law and Policy of the World Trade Organization, 4th edn. Cambridge University Press, p. 38CrossRefGoogle Scholar.

⁸ For instance, Article 19.4 stipulates the non-discrimination principle, paragraph 4 of Article 19.2 indirectly prescribes the principle on market access, and Article 32.1 contains the principle on the conflict between free trade and social value.

⁹ ‘The principles’ encompass non-discrimination principles of the WTO as well as RTAs (including recent digital trade agreements). This study particularly focuses on non-discrimination principles under RTAs.

¹⁰ WTO, ‘Work Program on Electronic Commerce Decision of 2 December 2009’, WT/L/782, 11 December 2009.

¹¹ More examples and descriptions are found in the United States’ ‘National Trade Estimate (NTE)’ report (USTR, 2022) and M. Fink (2020) ‘Legal Analysis of International Trade Law and Digital Trade’, Think Tank European Parliament. For example, USTR annually publishes a long list of digital trade barriers by trading partner countries of the US in the NTE (National Trade Estimate) Report. Under the section on ‘Digital trade and electronic commerce’, it points out the barriers to cross-border data flows, including data localization requirements, discriminatory practices affecting trade in digital products, and restrictions on the provision of internet-enabled services, which are often de facto discriminatory to foreign suppliers.

¹² Ministerial Text for Trade Pillar of the Indo-Pacific Economic Framework for Prosperity, Pillar I (trade).

¹³ González, J. López and Ferencz, J. (2018), ‘Digital Trade and Market Openness’, OECD Trade Policy Paper No. 217, OECD Publishing, Paris, p. 34Google Scholar.

¹⁴ A representative example must be Mitchell, A.D. and Mishra, N. (2018) ‘Data at the Docks: Modernizing International Trade Law for the Digital Economy’, Vanderbilt Journal of Entertainment & Technology Law 20(4)Google Scholar.

¹⁵ Okanga, O.O. (2021) ‘Testing for Consistency: Certain Digital Tax Measures and WTO Non-Discrimination’, Journal of World Trade 55(1), 101–126CrossRefGoogle Scholar.

G. Rajgopalan (2018) ‘Taxing Digital Economy − Applicability of Non-Discrimination Rules in International Agreements’, Background Paper for New Delhi Conference of the International Fiscal Association, April 2018, 12 Pages, posted 4 June 2018. This study also points out that the EU digital tax regime is open for questioning violation of the principles of GATS.

¹⁶ Elisabet, M. and Komalasari, Y. (2021) ‘Digital Services Tax Regulations and WTO Non-Discrimination Principle: Is the Deck Stacked?’, Indonesian Journal of International Law 19(1), art. 2Google Scholar. The imposition of PTE by Indonesia and the DST by the EU does not violate the WTO's non-discriminatory principles, namely the National Treatment and Most-Favored Nation principles in Article II and Article XVII of the GATS.

¹⁷ India's Digital Service Tax is called the Equalization Levy, or Equilibrium Levy under Section 166A.

¹⁸ S. Lester (2021) ‘Digital Trade Agreements and Domestic Policy’, Free Trade Bulletin, CATO Institute, 14 April 2021, Number 79.

¹⁹ Wunsch-Vincent, S. (2009) ‘Trade Rules for the Digital Age’, GATS and the Regulation of International Trade in Services. Cambridge University PressGoogle Scholar..

²⁰ WTO (1998) ‘The Declaration on Global Electronic Commerce’, adopted by Ministers at the second session of the Ministerial Conference, WT/L/27430 September 1998.

²¹ IMF, ‘Towards a Handbook on Measuring Digital Trade: Status Update’, BOPCOM−18/07, p.6.

IMF defines digital trade in an extensive and comprehensive approach. According to IMF, the three dimensions of digital trade are referred to as follows: (1) Digitally ordered trade transactions include all the international trade transactions that are classified as the sale or purchase of a good or service,conducted over computer networks by methods specifically designed for receiving or placing orders (following the existing definition of ecommerce, but including additional clarification that e.g. in-app purchases should also be considered in scope). (2) Digital intermediary platform-enabled trade transactions include all the international trade transactions via digital intermediary platforms, which in turn are defined as online interfaces that facilitate, for a fee, the direct interaction between multiple buyers and multiple sellers, without the platform taking into account the economic ownership of goods or providing the services that are being sold. (3) Digitally delivered trade transactions include all the trade transactions involving products (predominantly services) that are delivered remotely over ICT networks, following the definitions developed by the TG Service Task Group and the TFITS for ICT-enabled services.

²² Moreover, the OECD further defines digitally ordered trade as an international sale or purchase of a good or service, conducted over computer networks by methods specifically designed for the purpose of receiving or placing orders.

²³ USICT (2017) ‘Global Digital Trade 1: Market Opportunities and Key Foreign Trade Restrictions’, p. 14. USICT defines digital trade as ‘the delivery of products and services over either fixed-line or wireless digital networks. It also describes the use of digital products and services by a wide range of industries, including online sales of goods and services over e-commerce platforms.’

²⁴ EU Commission website, defines digital trade as ‘commerce enabled by electronic means’, https://ec.europa.eu/trade/policy/accessing-markets/goods-and-services/digital-trade/ as EU.

²⁵ According to RCEP by China, it stresses trade facilitation by allocating a separate Section B (Trade Facilitation), while avoiding digital contents and services. See Chapter 12 (Electronic Commerce) of RCEP.

²⁶ Many FTAs allow for the indirect figuring out of the concept and scope of digital trade by providing the definitions of ‘electronic means’ and ‘using electronic means’ instead of defining ‘digital trade’. For example, FTA between EAEU and Vietnam includes a separate chapter on ‘Electronic Technologies in Trade’. Under this FTA, ‘electronic commerce’ is defined as ‘trade with the use of electronic technologies’. Center for WTO and Innternational Trade -VCCI, https://wtocenter.vn/upload/files/fta/174-ftas-concluded/188-vietnam---eurasian-/241-full-text/FTA%20VN%20-%20EAEU%20-%20Full%20text.pdf.

²⁷ Wunsch-Vincent, supra n. 20, p. 5, According to this study, it is not visible at the moment for the WTO participating members to agree on the classification of digital products whether they are governed by GATT, GATS or some unique category deserving its own set of trade rules.

²⁸ OECD, ‘The Impact of Digitalization on Trade’, www.oecd.org/trade/topics/digital-trade/.

²⁹ Under Article 23.1 of the Korea–US FTA, ‘Article XIV of GATS is incorporated into and made part of this agreement, mutatis mutandis for the purpose of the e-commerce chapter’, https://ustr.gov/trade-agreements/free-trade-agreements/korus-fta/final-text.

³⁰ Although many issues have been resolved in recent RTAs, it seems that many countries are still reserving a clear attitude on how to classify digital trade. For instance, even a cutting edged type of digital trade agreement, DEPA, cites both GATT and GATS exceptions without distinguishing goods and services in digital trade.

³¹ OECD, ‘Digital Trade and Market Openness’, OECD Trade Policy Papers No. 217, p. 38.

³² Ibid.

³³ WTO, ‘Report of the Working Party (97–109) Border Tax Adjustments’, adopted on 2 December 1970.

³⁴ Ibid.

³⁵ Zdouc, W. (1999) ‘WTO Dispute Settlement Practice Relating to the GATS’, Journal of International Economic Law 2(2), 333Google Scholar.

³⁶ Elisabet and Komalasari, supra n. 16, 47. According to this study, the CPC defines only the activity of the service provider and not the service provider.

³⁷ For example, the Annex 7-A-1 of the EU–Korea FTA states that ‘In identifying individual sectors and sub-sectors:  …  (b) CPC ver. 1.0 means the Central Products Classification as set out in Statistical Office of the United Nations, Statistical Papers, Series M, N° 77, CPC ver 1.0, 1998’.

³⁸ Handley and Limão (2015) show that trade agreements can bring a positive trade effect by credibly reducing trade policy uncertainty (TPU) (K. Handley and N. Limão (2015) ‘Trade and Investment Under Policy Uncertainty: Theory and Firm Evidence’, American Economic Journal: Economic Policy 7(4), 189–222).

³⁹ According to the 2022 NTE report, data localization requirements are imposed in many countries such as Brazil, Canada, China, Ecuador, El Salvador, EU, India, Indonesia, Kenya, Korea, Malaysia, Nigeria, Norway, Pakistan, Philippine, Russia, Saudi Arabia, Swiss, Thailand, Turkey, UAE, UK, and Vietnam, which can raise costs for the construction or use of unnecessary and redundant local data storage. As even more extreme data localization measures, Russia and China plan to introduce measures to centralize, control, and restrict Internet infrastructure services, driving Internet fragmentation at all levels (refer to Russia's ‘Sovereign Internet’ law and Chinese ‘Draft Data Security Management Measures of 2020’, article 29) (Internet Society, ‘Internet Way of Networking Use Case: Data Localization’, 30/Sep/2020).

⁴⁰ The 2022 NTE report presents that local presence requirements are imposed in many countries, such as Jordan, Korea, Pakistan, Russia, Thailand, Turkey, and Vietnam, which can limit the ability of firms to provide services on a cross-border basis. As an example, Russia implemented the ‘Landing Law’ in 2001, which requires local presence for a website or application with more than 500,000 daily users (2022 NTE report, p. 437). 2022 NTE report website, https://ustr.gov/about-us/policy-offices/press-office/press-releases/2022/march/ustr-releases-2022-national-trade-estimate-report-foreign-trade-barriers.

⁴¹ China, Egypt, Kenya, Pakistan, Russia, Singapore, Thailand, UAE, and Vietnam restrict or block access to certain websites that are deemed politically or culturally inappropriate. Specially, China currently blocks legitimate websites and imposes significant costs on both suppliers and users of web-based services and products, and it has been estimated that more than 10,000 sites are blocked, affecting billions of dollars in business, including communications, networking, app stores, news, and other sites (2022 NTE report, p. 107).

⁴² Horn, H. (2006) ‘National Treatment in the GATT’, American Economic Review 96(1), 394–404CrossRefGoogle Scholar.

⁴³ R.W. Staiger (2021) ‘Does Digital Trade Change the Purpose of a Trade Agreement?’, mimeo, www.nber.org/papers/w29578.

⁴⁴ Goldfarb, A. and Tucker, C. (2019) ‘Digital Economics’, Journal of Economic Literature 57(1), 3–43Google Scholar.

⁴⁵ On 1 July 2019, Bangladesh imposed VAT on sales of B2C and B2B transactions ranging from 15% (a standard rate) to 5% (a reduced rate), to non-resident providers of digital services to domestic consumers (R. Asquith, ‘Bangladesh VAT on Foreign Digital Services’, 15 December 2021).

⁴⁶ On 1 August 2019, Turkey's Radio and Television Supreme Council (RTUK) published the Regulation to require providers of Internet streaming services to establish a commercial presence in Turkey and to obtain a broadcasting license. On 21 February 2022, RTUK warned three international online media outlets. Deutsche Welle, Voice of America, and Euronews, to obtain broadcast licenses. Failure to obtain the licenses could result in RTUK blocking access to or removing content on the sites and imposing criminal sanctions. The broadcast license, which covers 10 years, costs $7,382 (100,000 Turkish liras) for digital streaming platforms and online TV broadcasting (E. Sahinkaya (2022) ‘Turkey's Media Regulator Forces VOA and Others to Obtain License’, 10 February 2022).

⁴⁷ M. Mann and D. Cheung (2019), ‘Measuring Trade in Services by Modes of Supply’, Eurostat Statistical Working Papers, https://ec.europa.eu/eurostat/documents/3888793/10282481/KSTC-19-007-EN-N.pdf/730bfc0b-8c13-db03-a903-1dbb0c69013f.

⁴⁸ For both BaTiS and TISMOS, see www.wto.org/english/res_e/statis_e/trade_datasets_e.htm.

⁴⁹ Suh, J. and Roh, J. (2023) ‘The Effects of Digital Trade Policies on Digital Trade’, The World Economy 46(8), 2383–2407CrossRefGoogle Scholar.

⁵⁰ We use the version of 11 June 2021. For details on TAPED, see Burri, M. and Polanco, R. (2020) ‘Digital Trade Provisions in Preferential Trade Agreements: Introducing a New Dataset’, Journal of International Economic Law 23(1), 187–220Google Scholar.

⁵¹ Some trade agreements (e.g. Singapore–US FTA, China–Korea FTA, CPTPP, etc.) include the principle for a sector-specific article such as the financial service sector in the service chapter under the title ‘Market Access for Financial Institutions’.

⁵² For example, the Singapore–US FTA specifies the principles along with the imposition of no customs duty in the article of ‘Digital Products’ in the digital trade chapter and does the principles in the articles of ‘local presence’ and ‘national treatment’ in the service chapter. It also includes a sector-specific clause mentioning the principles under the title ‘Market Access for Financial Institutions’ in the service trade chapter.

⁵³ Scoring involves transforming qualitative information into quantitative data. Weights balance the relative importance of measures. Aggregation is the final step to calculate the cumulative index as a weighted average of the scores. See J. Ferencz (2019) ‘The OECD Digital Services Trade Restrictiveness Index’, OECD Trade Policy Papers 221, OECD Publishing, for a detailed explanation on the categorized framework of OECD Digital STRI.

⁵⁴ The Digital STRI framework is categorized as following. ‘Class 1 – Infrastructure and connectivity’ is a measure affecting cross-border data flows. ‘Class 2 – Electronic transactions’ is a measure affecting electronic transactions. ‘Class 3 – Payment systems’ is a measure affecting the payments made through electronic means. ‘Class 4 – Intellectual property rights’ is a measure of domestic policies related to the protection and enforcement of trademarks, copyright and related rights. ‘Class 5 – Other barriers’ is a measure of barriers to trading in digitally enabled services that do not fall under the previous policy areas (e.g. performance requirements, limitations on downloading and streaming, or restrictions on online advertising).

⁵⁵ USTR NTE 2022, https://ustr.gov/about-us/policy-offices/press-office/press-releases/2022/march/ustr-releases-2022-national-trade-estimate-report-foreign-trade-barriers.

⁵⁶ Anderson, J. and Van Wincoop, E. (2003) ‘Gravity with Gravitas: A Solution to the Border Puzzle’, American Economic Review 93(1), 170–192Google Scholar.

⁵⁷ Silva, J. Santos and Tenreyro, S. (2006) ‘The Log of Gravity’, The Review of Economics and Statistics 88(4), 641–658Google Scholar.

⁵⁸ Given that our digital trade agreement related variables were made based on the effective date, it is understood that these trade promotion effects take into account the effects on digital trade after the entry into force of the agreements.

⁵⁹ One may wonder whether this is because countries with relatively liberal approaches and more data flows find it easier to agree to ambitious digital trade commitments, including non-discrimination. However, such causal issue of whether an agreement containing provisions related to digital trade itself contributes to digital trade flows has been empirically tested and validated in another study by Suh and Roh, supra n. 49, which showed that there is no reverse causality as the aforementioned argument.

⁶⁰ This result makes sense in that the policy variables constituting DTRI appropriately cover digital trade specific elements that are not in STRI (e.g. electronic transactions, payment system, intellectual property rights, and other barriers affecting trade in digitally enabled services).

⁶¹ It is worth noting that these results are based on the ‘final values’ (the result of the estimations and adjustment procedures used to ensure complete consistency of the dataset) from BaTiS rather than the ‘reported values’ (the trade as reported by the relevant statistical authorities), so there is a risk of recovering the gravity type coefficients that are used to estimate the missing values by default. To address this point, we conduct the same analysis using the ‘reported values’ and find similar results, only quantitatively different. That is, we can find that the effect of NDs on digital trade flows is positive and statistically significant even under the dataset with reported values and a moderating effect for DSTRI. Also, other gravity variables show similar statistical significance and signs of coefficients. One caveat is that the coefficient on DSTRI itself is negative but statistically insignificant, which is likely because the standard deviation of DSTRI for countries that report trade flows in services is about twice as low as that for countries that do not.

⁶² We re-construct the variables of ‘data flow' and ‘no data localization' based on the TAPED. According to the TAPED codebook, ‘data flow' (TAPED variable # 1.28.1) is coded based on whether the e-commerce chapter include provisions on data flows, and ‘no data localization' (TAPED variable # 1.28.4) on whether a trade agreement has provisions to prohibit the use of data localization requirements. It is worthwhile to note that there is no case for ‘imposing data localization requirement’ (TAPED variable # 1.28.3) since our study covers up to 2019. TAPED Codebook, chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.unilu.ch/fileadmin/fakultaeten/rf/burri/TAPED/2020_06_08_TAPED_Burri_Polanco_Codebook.pdf.

⁶³ Compared to the ‘ND’ (‘no ND’ obs. 24 and ‘ND’ obs. 73), 21 out of 97 agreements include the provisions on ‘data flow’, 7 agreements of which also include the provision on ‘no data localization’. There are no agreements that include only ‘no data localization’ without ‘data flow’. In addition to 7 agreements including both ‘data flow’ and ‘no data localization’, additional 13 agreements including only ‘data flow’ also have ‘ND’.

⁶⁴ One may argue that examining a small number of provisions may not be sufficient to ascertain the effectiveness of the principles. And it makes sense especially considering the correlated structure of clauses in trade agreements. It is often that when one clause is included in a trade agreement, other related clauses tend to be included at the same time. And such correlation structure of clauses makes it difficult to distinguish the effect of a particular clause from others. To address this concern, we conduct the principal component analysis in the Appendix. Over every relevant variable in TAPED, we constructed principal components (PCs), and found that PC containing ND provision is one of the major components and the PC actually shows a positive sign on digital trade flow.

⁶⁵ According to the Wall Street Journal 31 August 2021, the law amends South Korea's Telecommunications Business Act to prevent large app-market operators from requiring the use of their in-app purchasing systems. It also bans operators from unreasonably delaying the approval of apps or deleting them from the marketplace—provisions meant to head off retaliation against app makers.

⁶⁶ The National Trade Estimate Report on Foreign Trade Barriers (NTE) 2021, p. 333.

⁶⁷ Article 15.3 (Digital Products).

⁶⁸ In a report of the inspection on the government office by the Korean National Assembly 2021, according to Korean news article, the market share of Google and Apple app stores was revealed during the parliamentary inspection of the administration (2021.09.15).

⁶⁹ According to this news article, USTR raised the possibility of the violation on the national treatment obligations under WTO GATS and the Korea–US FTA.

⁷⁰ For this purpose, we exclude #1.3, #1.4, #1.8.1, #1.8.3, #1.28.1, #1.28.3, and #1.28.4 in terms of TAPED variable number. The first four variables are used for generating the ND variable, and the last three variables are sub-clauses of data flow.

⁷¹ TAPED Codebook, chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.unilu.ch/fileadmin/fakultaeten/rf/burri/TAPED/2020_06_08_TAPED_Burri_Polanco_Codebook.pdf.