The Digital Trade Protocol (‘the Protocol’) of the African Continental Free Trade Area (AfCFTA) is a document intended to define the environment for digital trade between members of the African Union (AU). This refers to continental member states as well as island nations such as Mauritius and Madagascar. ‘Digital trade’ primarily refers to ‘digitally-delivered services’ (DDS) such as video streaming, software as a service (SAAS), cloud services and increasingly, AI services. However, some aspects of the definition refer to digitally-originated trade, whether for physical goods (‘e-commerce’) or for non-digitally delivered services, for example platform services such as ride-hailing or cleaning services.

The draft protocol as it stands – and it needs to be pointed out that although it is a ‘mature draft’, a final text adopted by the Assembly of the African Union is not yet available – is a progressive document that aims to encourage digital inclusion of citizens, digital skills development and promote digital economy integration by MSMEs. However there are several important exceptions to this:

1. The Protocol refers to an Annex on rules of origin (ROO) for digital services, which seems to imply that some member states (MS) are expected to implement customs duties on DDS originating from third countries (Article 7). Although this does not encourage duties on DDS, a more progressive and pro-development approach would have been to clearly discourage trade barriers against DDS, which would have been consistent with the current WTO e-commerce moratorium^[1].

2. The Protocol allows exceptions to digital trade best practice such as in the case of cross-border data transfers (Article 21). This article is less progressive than that specified in say, the EU Digital Trade Title^[2], in the following ways:

   • It specifies that cross-border data flows shall not be restricted “provided the activity is for the conduct of digital trade”. However, a large volume of cross-border data is part of social media content, which is not strictly defined as ‘digital trade’ but could rather be seen as ‘communications’, ‘media’ or even ‘entertainment’. In other words, there is a large component of cross-border data that is non-commercial from the perspective of the users, even though the platform itself is commercialised in the sense that it may host commercial advertising. However, the average user is able to use social media platforms ‘free of charge’ and not as part of a transaction with the platform owner.

   • It appears to permit localisation of data because, even when read in conjunction with Article 23 (location of computing facilities), the only pertinent requirement on MS is to not require a person or other MS to locate computing facilities in that MS territory as a prerequisite for doing business. A country can permit cross-border data flows while at the same time requiring certain categories of data to be stored only locally. An example would be an encrypted transmission that passes across a border but is transient for the purposes of a discrete transaction (such as a payment) and where the original financial data is required to be stored locally.

   • It appears to permit a MS to prohibit another MS from storing and processing the first MS’s data in the second MS’s territory.

These considerations are in regard to conventional cross-border data flows, i.e., data flow concessions in the Protocol all have the standard waiver for ‘legitimate public policy objective or to protect essential security interests’ (articles 21 and 23). The data flows referred to in the bullet points above would not fall within the waiver.

Taken together, these exceptions to digital trade policy best practice constitute potential barriers to digital trade with regard to the rest of the world (bullet 1) and with regard to other member states (bullet 2).

Nevertheless, a key characteristic of the internet and one of its most misunderstood is the fact that it is essentially borderless. The internet, even websites that are local to a country or city, do not necessarily ‘exist’ on that locality. In fact, they are more than likely to exist in multiple places and multiple countries, even if their web host is a local business. This means that any attempts to limit cross-border data flows, to localise data or to force the location of data centres will encounter important practical challenges and certainly lead to efficiency losses/cost increases.

The internet’s borderless nature is a result of several core aspects of the internet’s architecture and operation:

• Decentralised Network Structure: The internet is a network of networks, where data packets travel across diverse paths to reach their destination, irrespective of geographical or political boundaries. This decentralised approach ensures that information can flow freely across the network, without a central controlling authority. The design, based on the pioneering work of Vint Cerf and Bob Kahn on the TCP/IP protocols, was intended to ensure robustness and resilience, but it also inherently disregards geographical borders.

• Content Distribution Networks (CDNs): CDNs are a tangible manifestation of the internet’s borderless nature. They are networks of servers located around the world, designed to deliver content efficiently to users regardless of their location. By caching content at strategically distributed points, CDNs optimise delivery speeds and reliability, underscoring the internet’s global reach and its ability to serve content across borders with minimal delay.

• Cloud Computing: The advent of cloud computing further emphasises the borderless nature of the internet. Cloud services allow data and applications to be stored and accessed over the internet from anywhere in the world. This means that businesses and users can leverage computing resources located in different countries, often without knowing the exact location of these resources.

The opposition to African data flowing freely to the tech firms located in the developed world appears to be based on the notion that that data has a commercial value, which value is appropriated by firms in the rest of the world. It is important to note however, that the platforms used to capture the data are not based in Africa, therefore the transmission of African data across the border happens at the initial stage. In order to prevent African data flowing cross border to the ROW, it would be necessary to not use the platform in the first place. Finally, the owners of these platforms pay for infrastructure, staff and bandwidth in order to facilitate platform operation. The users’ data has minimal value to them against the significant costs of maintaining these operations.

An additional misconception relates to the access to source code and technology transfer. Certain commentators have alleged that the Protocol as it stands will limit technology and skills transfer from developed countries to African countries. This springs from a reading of Article 25, which forbids MS from requiring access to source code except for legitimate, non-commercial reasons. This provision is consistent with global norms and standards, for example the earlier-cited EU Digital Trade Title^[3]. All African WTO members are likewise subject to the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), which came into effect on January 1, 1995. Although this agreement does not specifically cover digital information/designs, the related World Intellectual Property Organisation (WIPO) Copyright Treaty (WCT)^[4] of 1996 does. The WCT has been ratified or acceded to by numerous countries worldwide, including a significant number of African nations. These countries have committed to incorporating the treaty’s provisions into their national laws, thereby enhancing the protection of copyright holders in the digital era.

Even so, would relaxing adherence to international IP agreements necessarily assist in technology and skills transfer to African countries? New research indicates that technology transfer from developed to developing countries happens spontaneously in IP-protected environments, to the benefit of developing countries^[5]. The findings of this research highlight a significant increase in cross-border patent flows, particularly from wealthy to poorer countries, growing over 300% between 1995 and 2018. This trend, facilitated by policy and globalisation, benefits both sets of countries but disproportionately aids poorer nations.

A final area of clarification around the Protocol relates to its scope. The preceding claim that technology transfer would be hampered by the protection of source code required by Article 25 is actually erroneous not just by content but by scope as well, even though it has been made in a published article. The AfCFTA is intra-African; it defines the expected interactions between African member states, not those between African states and third countries/the developed world. Although third countries are mentioned in places, this is only as regards to the rights of MS to form policies with regard to third countries in the particular area of an Article (such Articles 21 and 8).

The Digital Trade Protocol of the AfCFTA aims to set the stage for intra-African digital trade even as the world experiences a fundamental shift in digital services trade, that relating to AI technology. It is important that we in Africa position ourselves to take full advantage of this groundbreaking technology, first as users but eventually as innovators too. The Protocol needs to be understood as defining the intra-African digital trade environment but also needs to reflect the needs of businesses, investors and consumers in order to maximise the benefits of the technology.

---

^[1] There is consensus among experts that imposing trade barriers against DDS would harm domestic businesses, especially MSMEs. For example, this paper, which addresses the issue from the perspective of Indonesia:
https://infisum.com/index.php/2023/12/08/the-value-of-cross-border-digital-transmissions-to-msmes-in-indonesia-implications-for-participation-in-the-wto-e-commerce-moratorium-2/

^[2] European Union. 2019. EU Proposal for a digital trade title: explanatory note. Brussels: European Union

^[3] Ibid.

^[4] https://www.wipo.int/treaties/en/ip/wct/

^[5] LaBelle, J., Martinez-Zarzoso, I., Santacreu, A. and Y. Yotov. 2024. Determinants and impact of cross-border patents: Evidence from a new dataset. VoxEU/CEPR
https://cepr.org/voxeu/columns/determinants-and-impact-cross-border-patents-evidence-new-dataset