How Information Technologies are changing Forensic Finance in Africa

Jean Langlois-Berthelot

read more

Benedikt Barthelmess

read more

Money Laundering and Terrorist Financing remain a serious challenge for governments on the African continent. Rapid technological change has in some instances accentuated the threats emanating from these categories of financial crime and financial authorities, just like elsewhere in the world, will have to invest considerable resources in Anti-Money Laundering and Counter Terrorist Finance measures. Besides technologies that may find direct applications in financial sector, financial institutions and their regulators could also benefit from broader governance innovation.

1. The Role of Technologies in Reducing Money Laundering (ML) and Terrorist Financing (TF) in Africa

In the lesser developed parts of the African continent, contemporary consumers have greatly benefitted from technological progress in finance, notably in the payments sphere. Arguably the greatest achievement of new payments solution has been the sharp increase in people’s access to basic financial services. However, new technologies have also been an enabler of financial crime, the anonymity granted by virtual currencies and the widespread use of prepaid cards being just two examples.

While tighter regulation and supervision has been the way forward to tackling financial crime, technology equally has to play a role if we are “to detect and disrupt financial crime, and ultimately the criminals who seek to exploit the system”, as Megan Butler, Executive Director of Supervision at the British banking supervisor Financial Conduct Authority (FCA), noted in a speech on the 23 October 2019.2

While Butler was mainly referring to the British financial sector, one could argue that this is even more so the case in Africa’s less developed countries where banks and non-banking institutions find it much harder to bear to the costs that have come along with a mix of machine and manual processes employed to comply with new Anti-Money Laundering and Counter Terrorist Finance (AML/CTF) measures. Greater vigilance within the industry has moreover translated into larger numbers of reports filed with national banking supervisory authorities (e.g. Suspicious Activity Reports, “SARs”3). As a consequence, authorities have been struggling to cope with the increased burden of carrying out investigations, which, again, require significant resources. A recent report4 by the AML Working Group of the RegTech for Regulators Initiative provides a particularly alarming assessment of authorities’ readiness for recent technological change:

In general, the information technology (IT) systems currently employed by financial authorities to capture, store, and render data from financial service provides (FSPs) and other sources were not designed for the latest generation of digital products, platforms, and provides that rely on Big Data. As these continue to proliferate across and within financial sectors, the capacity of existing data architectures to fully absorb and digest the data that digital financial services (DFS) generate is steadily diminishing.

Cheaper, technology-based solutions are thus an important way forward in African countries – most notably solutions that tap into Artificial Intelligence (AI), robotics, natural language processing and machine learning.

This article offers an outline of different technological solutions and seeks to explore the extent to which these technologies have already found applications in this part of the world. The remainder of this article is structured as follows: Section 2 focuses on technologies used by financial institutions. This is then followed by an exploration of technologies from the regulatory perspective in Section 3. Section 4 offers some concluding remarks.

2. Financial Institutions’ Use of Technological Solutions
   1. Profiling Technologies

   Financial institutions tend to rely heavily on profiling technologies that attempt to model and simulate ML behaviour in order to capture suspicious transactions. Automated profiling solutions have been in use for many years now; however, these commonplace automated AML systems rely heavily on structured databases that lack the ability to effectively and efficiently identify hidden and complex ML activities, as Han et al.5 note. They further note that this holds particularly true for ML activities with dynamic and time-varying features, i.e., activities that do not follow any straightforward, linear pattern. As a result, systems tend to produce a high percentage of false positives, which then need to be investigated by humans in what are often lengthy and human resource-intensive processes.

   2. Deep Learning Technologies

   The shortcomings of such conventional profiling technologies underline the potential benefits of more advanced technologies, even though they have to date only found very limited applications in the AML/CTF context. Han et al.6 introduce, for instance, a framework that employs deep learning-driven natural language processing (NLP) technologies—a subset of machine learning—to augment AML monitoring and investigation.  What is in its essence a distributed framework, uses NLP technology to analyse news and tweet sentiments, entity recognition, relation extraction, entity linking and link analysis on data sources.

3. Reg- and Suptech: Regulators and Supervisors’ Engagement with Technologies
   Financial regulators and supervisors have now for some time been looking into advanced data collection and analytics tools that are collectively referred to as “regtech” and “suptech”. While there have been considerable advancements in such tools, there has also been a growing focus on creating special regulatory environments—often referred to as regulatory sandboxes—allowing for the testing of new technologies and regulatory measures.
   1. Regulatory Sandboxes

   Regulatory sandboxes are a ringfenced innovation environment promoted by several regulators to allow the FinTech industry to test their innovation and understand the impact of regulatory measures. The first such sandbox was launched by the British FCA in 2015,7 but the idea quickly gained traction and now more than 50 authorities worldwide are toying with such sandboxes. A report by the United Nations Secretary-General’s Special Advocate for Inclusive Finance for Development (UNSGSA) FinTech Working Group8 distinguishes between product testing sandboxes and policy making sandboxes, although the lines between the two types are fluid. Product testing sandboxes allow firms to assess consumer uptake and commercial viability. Meanwhile policy testing sandboxes are more focused on evaluating regulations or policies in view of their potential impact on new technologies and business models. Notable advantages include the reduction of uncertainty and hence the cost of innovation for FinTech developers, while the authorities find themselves in a better position to engage in evidence-based policy making. The FinTech Working Group report also emphasises regulatory sandboxes’ potential to advance financial inclusion.9

   There has, moreover, been some experimentation with multi-jurisdictional sandboxes.  Such sandboxes seem particularly promising when it comes to ML/TF prevention, given the international dimension of ML and TF. An added benefit is that it potentially allows for economies of scale, bringing down individual African regulators’ cost. Yet the FinTech Working Group report also cautions that the initial cost of setting up such a sandbox may be high in light of the challenges the development of a sandbox framework across multiple jurisdictions may pose.

   Regulators in African countries have only limited experience with regulatory sandboxes to date. South Africa’s South African Reserve Bank and the Financial Services Conduct Authority have reportedly announced the creation of an innovation hub along with a regulatory sandbox.10 Other African countries that have been experimenting with regulatory sandboxes are Mauritius, Sierra Leone and Mozambique.11

   2. Data Analytics Tools

A large range of data analytics tools with applications in financial sector regulation have emerged in recent years and include machine learning, cloud computing, natural language processing, text mining, application programming interfaces (APIs), artificial intelligence and machine reading. As Coelho, De Simoni and Prenio12 in a Bank for International Settlement paper note, such technologies can be used for the detection of networks of related transactions, identification of unusual behaviours and, more generally, the transformation of large amounts of structured and unstructured data for analytical purposes. Examples of more specific applications of such technologies with respect to AML/CTF supervision are explored in the following paragraphs.

Supervisory authorities’ mandate to supervise and monitor financial institution entails assessments of governance, risk management, internal controls and processes and systems to prevent financial crime. Coelho, De Simoni and Prenio13 find that analytical tools have been developed to enhance the offsite assessment of the individual institutions’ risk profiles.  Such offsite assessment tools can be complemented by tools to assess the overall risk of supervised entities.  Such analytics tools typically assign a rating to each supervised institution as of their likelihood of non-compliance with AML/CFT requirements.14

When it comes to the evaluation of data submitted to regulators and the identification of patterns and trends, data analytics platforms using data science, machine learning and artificial intelligence have significant potential in helping regulators to cope with growing numbers of structured and unstructured data points.

African countries to date have made limited headway in this area; however, the example of Nigeria’s “Data Stack” data warehouse demonstrates some of the possibilities. As outlined by a R2A whitepaper15, the Central Bank of Nigeria (CBN), alongside the Nigeria Inter-Bank Settlement System Plc and BFA consulting firm, has been working on the redesign of its data infrastructure. The “Data Stack” consists of a transactional data warehouse and dashboards that the CBN and other stakeholders can access to analyse payments data. It is populated via Application Programming Interfaces (APIs) with real-time transactional data from the inter-bank settlements system and CBN compliance data which can be tailored to the analytic requirements of different stakeholders.

Conclusion

ML and TF remain a serious challenge for governments on the African continent. Rapid technological change has in some instances accentuated the threats emanating from these categories of financial crime and financial authorities, just like elsewhere in the world, will have to invest considerable resources in AML/CTF measures. None of this has gone unnoticed in African countries and, to some extent, action has already been taken. However, measures remain incomplete and incoherent, especially when it comes to the use of advanced technologies: there is a difference in employing new tool and turning it into an effective AML/CTF weapon. Among the most promising initiatives are probably, Nigeria’s Data Stack, which has created tremendous opportunities for a much more integrated, data-driven approach to fighting financial crime.

Besides technologies that may find direct applications in financial sector, financial institutions and their regulators could also benefit from broader governance innovation.  Huw van Steenis, Chair of Sustainable Finance at UBS, for instance, argued16 that governments should champion digital forms of identification, following the example of India’s Aadhaar programme, which simplifies the process through which networks can know their customers. The FCA’s sandbox’s fifth cohort included several companies that looked at decentralised and federalised digital identity platforms machine learning identity verification.17 Such new identification mechanisms reduce the need for manual ID checks and help organisations avoid having to do duplicate checks.

*Dr. Jean Langlois-Berthelot, Phd in Applied Mathematics (EHESS-Paris) is currently working on research programs of the French Ministry of Armed Forces.

Mr. Benedikt Barthelmess, MSc (London), MPhil (Oxon), is a former Non-Resident Fellow at the Zambia Institute for Policy Analysis and Research.  

• 1. M. Butler, “Turning Technology Against Financial Crime”, speech delivered at the Royal Services Institute, London, UK, 2019.
• 2. Ibid.
• 3. See for example the guidance provided by the South African Financial Intelligence Centre: Financial Intelligence Centre, Financial Intelligence Centre Guidance Note 4 on Suspicious Transaction Reporting, undated.
• 4. R2A and BFA Global Reports, Financial Authorities in the Era of Data Abundance: RegTech for Regulators and SupTech Solutions, 2018, p. 3
• 5. J. Han et al., “NextGen AML: Distributed Deep Learning based Language Technologies to Augment Anti Money Laundering Investigation”, Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics-System Demonstrations, Melbourne, Australia, Association for Computational Linguistics, 15-20 July 2018, pp. 37–42.
• 6. Ibid.
• 7. FCA (2015). “Regulatory Sandbox”, May 11, 2015, at https://www.fca.org.uk/firms/regulatory-sandbox, (accessed on 1 December 2019)
• 8. UNSGSA FinTech Working Group and CCAF, “Early Lessons on Regulatory Innovations to Enable Inclusive FinTech: Innovation Offices, Regulatory Sandboxes, and RegTech”, Office of the UNSGSA and CCAF: New York, NY and Cambridge, UK, 2019, p. 28
• 9. Ibid. p. 28.
• 10. Gwen Ngwenya, “SA shouldn’t miss opportunity to move the sandbox beyond fintech”, June 21, 2019, Tech Central at  https://techcentral.co.za/sa-shouldnt-miss-opportunity-to-move-the-sandbox-beyond-fintech/90450/, (accessed on 21 November 2019)
• 11. M., L. Wechsler L. Perlman and N. Gurung, “The State of Regulatory Sandboxes in Developing Countries”, SSRN, 2018.
• 12. R. Coelho, M. De Simoni and J. Prenio, “Suptech Applications for Anti-Money Laundering”, FSI Insights on Policy Implementation, no 18, Bank for International Settlements, August 2019, p. 7.
• 13. Ibid. p. 7.
• 14. Ibid. p. 8.
• 15. R2A and FDA, no. 5., 2018, pp. 36-37.
• 16. H. van Steenis, “The Digital Money Revolution”, Project Syndicate, November 13, 2019, at https://www.project-syndicate.org/commentary/digital-money-payments-revolution-by-huw-van-steenis-2019-11, (accessed on 1 December 2019)
• 17. FCA, “Regulatory Sandbox – Cohort 5”, April 29, 2019, at https://www.fca.org.uk/firms/regulatory-sandbox/cohort-5, (accessed on 1 December 2019)