2.1. Key barriers to women’s financial inclusion

The key barriers to women’s financial inclusion are demand-related, supply related, technical and infrastructure related, and financial literacy and legal and policy-related barriers (Central Bank of Nigeria [CBN], 2020). Demand barriers stem from low income, lack of trust in formal financial services, and social norms (CBN, Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020). It implies that women may prefer cash due to lower income and financial literacy levels as well as distance when bank branches are not available or proximate to their locations (CBN, Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020). These barriers persist for digital financial services (DFS). Fewer women have completed fully digital payments and have digital IDs like the National Identity Numbers (NINs; CBN, Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020). The NIN is a new requirement for meeting identification requirements including account opening in Nigeria. Not having the NIN therefore reduces the likelihood of owning an account or accessing the broader range of financial and public services. Supply barriers arise when existing services do not adequately meet women’s needs. Indicators include the absence of suitable financial products, unclear business cases for financial service providers, high costs in reaching women, and the lack of critical mass for digital payments (CBN, Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020). Technical and infrastructure barriers encompass physical, network, and data infrastructure. For instance, women lag men in mobile phone ownership and internet access. A report indicates that in Nigeria, 56 percent of men use mobile internet, whereas only 34 percent of women do so—a notable gender gap of 38 percent (GSMA, 2022). In the context of financial inclusion, gender-disaggregated data are also a crucial (data) infrastructure which has been historically lacking (EIGE, nd).

The Financial literacy barrier is widely acknowledged as limiting women’s financial inclusion. The Organisation for Economic Co-operation and Development (OECD) defines financial literacy as “a combination of financial awareness, knowledge, skills, attitude and behaviours necessary to make sound financial decisions and ultimately achieve individual financial well-being” (OECD, 2012). Financial illiteracy among women is exacerbated by functional illiteracy (higher among women who traditionally lag men in access to formal education in Nigeria), low mobile phone ownership among women (which limits their social, digital, and financial inclusion), limited Internet access, and failure or inability to obtain official ID like the NIN. Legal and policy barriers manifest in a lack of, or ineffective policies (discussed below) on women’s financial inclusion. Other challenges, such as the gender pay gap and cultural practices can hinder women’s economic participation and exacerbate demand barriers, thus requiring legal intervention. A potential intervention could mandate the use of gender-disaggregated data in financial decision-making processes. Although policy already requires the collection of this data, research indicates that it is not yet utilized for strategic decisions aimed at enhancing financial inclusion (WFID, 2022).

2.2. Policy responses

The statistics on financial inclusion in Tables 1 and 2 are an indictment of past and extant policies to accelerate gender financial inclusion and gender equality broadly. The first National Financial Strategy developed in 2010, was revised in 2018 partly because it failed to meet the target of increasing financial inclusion across all indicated groups including women (CBN, 2018). Perhaps, recognising this shortcoming, the CBN established a Framework for Advancing Women’s Financial Inclusion in Nigeria in Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020. The framework contains eight strategic imperatives to spur financial inclusion among women (CBN, Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020). The critical strategic imperatives include the expansion of the financial and digital literacy capabilities and skills programs for low-income women in the context of specific financial behaviours and products, and the expansion of delivery channels to serve women customers close to home (CBN, Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020). Also, in July 2022, the CBN released the Exposure Draft on Digital Financial Services Awareness (CBN, 2022a). The objectives of the Guidelines include addressing gaps in consumer knowledge and financial products development, setting digital literacy standards, and integrating financial literacy into existing financial education programmes (CBN, 2022a, Clauses 1.1 and 1.2). In January 2021, the CBN rolled out a Framework for Regulatory Sandbox Operations. The Sandbox is a formal process for firms to conduct live tests of new, innovative products, services, delivery channels, or business models in a controlled environment, with regulatory oversight, subject to appropriate conditions and safeguards (CBN, 2021a). Its objectives include increasing the potential for innovative business models that advance financial inclusion and ensuring appropriate consumer protection safeguards in innovative products (CBN, 2021a, Clause 1.1). In November 2022, a broader policy approach in the form of a National Fintech Strategy (NFS) was published. The purpose of the NFS is to set an agenda for inclusive fintech that enhances and deepens access to the use of fintech by 2024 (CBN, 2022b). In addition to the policy frameworks, the CBN has disparate and overlapping guidelines for DFS in Nigeria (CBN, 2023, 2021b).

The specific causes of policy inefficiency are not well-documented as agencies like the CBN rarely release progress reports that could elucidate such challenges. This calls for a deeper examination of AI, its potential, and its risks, prior to formulating policies for its deployment in finance or other sectors.

2.3. AI capabilities and risks

Fintech already contributes to the growth of mobile network operators that offer basic financial services in local communities, customer education, and ease of transactions (Kola-Oyeneyin et al., Reference Kola-Oyeneyin, Kuyoro and Olanrewaju2020). Nevertheless, some barriers to women’s financial inclusion persist. Relative to traditional banking, fintech requires advanced financial literacy skills. Points of sale services, often touted as having the capacity to bridge the rural/urban divide in financial inclusion, also require users to be physically present at service kiosks, recreating problems associated with distance from bank branches or financial service points. Service and transaction charges such as ATM transaction fees, or minimum account balances reinvent the challenges of cost and affordability for poor people at the grassroot and rural communities (CBN, 2018). Also, many DFS services require documentation and identification processes as part of Know Your Customer (KYC) and Anti-Money Laundering (AML) regulations, (CBN, 2018) which perpetuate entry barriers for those without required identification like NIN. Furthermore, women are underrepresented in the fintech user base in Nigeria. Statistics indicate a gender gap of 9.6% already exists in DFS, with only 13% of the female population completing a fully digital payment compared to 22.6% of the male population CBN, Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020). AI can be used to minimise these problems. The broad range of AI-powered applications, products and services facilitate transactions, customer services, assets and risk management, fraud detection, financial advice and credit scoring and decision-making. For excluded demographics like women, AI can provide insights into consumer needs and allow providers to customise and personalise financial services. For example, credit scoring algorithms can facilitate access to credit and eliminate credit scoring biases by generating alternative data streams, such as digital footprints for financial decision-making (WEF, 2022). These can unlock insights into data and enhance the creditworthiness of women who have limited credit history. Using voice and face recognition and natural language processing, AI can aid identification and verification of users without official identification. Promising development in voice and speech recognition algorithms, including text-to-image and text-to-video applications, can aid the development of applications that allow people with low literacy to participate in the digital economy (WEF, 2022; Feingold, Reference Feingold2022). Overall, AI can lower the entry threshold (demand related harriers) that has traditionally hindered women’s access to financial services and enhance the quality and diversity of services offered to them, thus addressing supply related barriers.

This is not to suggest that AI is the silver bullet for financial inclusion. Indeed, the application of AI in finance poses significant risks such as bias and disparate impact of credit outcomes, which may stem from inaccurate, non-representative and low-quality data (OECD, 2021). Specifically, machine learning (ML) models in decision-making raise concerns around transparency, accountability, and equity. Nigerian women, who are under-represented in financial datasets due to historical financial exclusion, may find these ML models to be biased, as they often reflect and perpetuate the prejudices present within the data. Consequently, these systems may favour historically privileged groups like men (Women’s World Banking, 2021).

Furthermore, service providers do not appear to be innovating to the needs of financially excluded women (EFInA, 2020). One study, perhaps the only one so far in Africa, found that less than one percent of fintech providers use AI and only for business-to-business tech support purposes that increase operational efficiencies (Ahmed, Reference Ahmed2021). Such use cases include conversational AI and robotic process automation systems for repetitive business processes associated with AML, countering terrorist financing (AML/CTF) and KYC operations. This suggests that the primary use of AI in these contexts is to streamline business operations, not to meet the financial service needs of women who are excluded from the financial system.

The data and analysis in the next section builds on this premise. Since AI application in finance is relatively new, we assume that existing research and explanations may not fully capture the complexities of the AI ecosystem and the behaviour of different players. The research uses four indicators to assess whether and how AI might improve women’s financial inclusion. The indicators are types of licensed services, location and geographical spread of service providers, percentage of service providers using AI in Nigeria and the respective use cases. The objective is to draw inferences from the data about the present use and challenges of designing and deploying AI for women’s financial inclusion.

4.1. Licensed fintech and use cases (indicators A, C, and D)

Table 3 shows that among licensed operators, only mobile money operators (MMO), payment solution service providers (PSSD) and super-agent authorisation appear to offer basic or consumer-facing services like payments and savings, which financially excluded women may need (WFID, 2022). Eighteen out of 169 licensed operators hold these licenses, others in the table appear focused on optimising existing financial products and services and offering sophisticated value-added services, such as trading platforms, and international money transfers, arguably for the already financially included. A reasonable inference is that products that are not relevant to women’s needs will aggravate demand related barriers, while limited licenses to providers of products likely to address such needs widen supply related barriers.

The AI use cases show the same pattern. Although about 45 providers claim to use AI (Table 5), only 13 provide information on their use cases (Table 6). For those without information, it is not clear whether they actively use AI, are in the process of developing their AI tools, or merely include AI to hype their services. Out of the 13 that do use AI, 2 are MMOs and 2 are PSSPs (Table 7). Again, both are the only providers likely to offer consumer-facing services which may be used by financially excluded women, particularly in rural communities. The others are focused on value-added services for theirs or other service providers’ existing products. This corresponds to already identified AI use cases by Fintech in Nigeria. Most fintech services are focused on guaranteeing the optimal performance of theirs or (others) financial products and services, differentiated customer experience, mitigating fraud and fraud detection, and enhancing financial advisory services, price regulation, marketing and managing insurance policies and portfolio and wealth management (Kola-Oyeneyin et al., Reference Kola-Oyeneyin, Kuyoro and Olanrewaju2020; Agidi, Reference Agidi2019). The 17 unlicensed and/or unregulated providers in Table 8 are unlikely to make a difference. They mostly help traditional financial service providers and other fintech providers serve their customers through, for example, providing identity verification services and chatbot services, fraud detection and digital market platforms. Crucially, no service provider seems to offer AI-powered products or services specifically (designed) for women.

There is currently no empirical evidence detailing the motivations behind financial product design, specifically concerning women’s diverse financial needs. However, one can theorise that woman, who are not a monolithic group, might avoid certain financial products due to the lack of “gendered” options that address their varied needs. Research indicates that excluded women may prefer financial services that support savings and micro-credits, as evidenced by the 6.3 million unbanked women in Nigeria who rely on informal savings collectors and merchants (WFID, 2022). Inclusive design, which should prioritise the end-user’s perspective, is crucial for catering to women’s unique financial needs. This concept aligns with Newell and Gregor’s approach to user-centred design for universal usability, advocating for designs that consider the specific characteristics of users, including those with disabilities (Stumpf et al., Reference Stumpf, Peters, Bardzell, Burnett, Busse, Cauchar and Churchill2020; Newell and Gregor, Reference Newell, Gregor and John2000).

The inference can be drawn that complex financial products, such as investment advisory services and trading platforms, may fail to attract certain demographics, like unbanked women, due to their lack of user-centric design and failure to meet the users’ specific needs (Brett et al., Reference Brett and Richard2021, Newell and Gregor, Reference Newell, Gregor and John2000). This gap in design and relevance leads to a diminished interest in learning or trusting the technologies. In Nigeria, a survey showed that while 74% of traders were open to adopting new technologies, over 30% resorted to informal financing due to denial of credit by financial institutions. Notably, most required funds below N30,000 (less than $40), leading to a decline in confidence in financial institutions, a preference for alternative solutions, and reduced willingness to engage with new financial technologies (Omoruyi, Reference Omoruyi2022, Prove, 2022).

4.2. Geography—the urban versus rural divide (indicator B)

Tables 4 and 5 show that most fintech service providers are in or operated from big urban cities like Lagos and Ibadan in Southwest Nigeria. Ordinarily, the location of a fintech might not be significant. Unlike banks, microfinance companies and other traditional financial service providers which rely on proximity to consumers and extensive branch networks to penetrate rural communities, fintech services are mostly digitalised and can be offered from any location. Thus, the models of Fintech companies allow them to bridge this proximity gap and provide services digitally to customers with no geographical boundaries (Frost, Reference Frost2022). M-Pesa, a Kenyan FinTech company, is one of the largest mobile money services in Africa. It operates in multiple countries across East Africa, North Africa, and South Asia, and has about 32 million users in 2018 (Njugunah, Reference Njugunah2022).

Nevertheless, proximity between a financial service provider and customers may be important regardless of the nature and type of service or mode of delivery (Saiedi et al., Reference Saiedi, Mohammadi, Broström and Shafi2022). Lack of awareness is a known cause of financial exclusion (Popescu, Reference Popescu2019), and the location of a fintech provider and its operations may be key to creating awareness in areas with high financial exclusion rates. Locating operational offices close to target users or consumers can create familiarity and make sensitization about products and services easier (Morgan and Trinh, Reference Morgan and Trinh2019). Awareness is also connected to digital and financial literacy. A case can be made that the closer providers are to the communities they serve, the easier they can understand the digital and financial literacy needs and address them (Kaffenberger et al., Reference Kaffenberger, Edoardo and Soursourian2018). Consumers are also more likely to trust providers that are close to them or have physical presence or representatives. This is particularly correct as concerns about privacy and data protection have increased with the proliferation of digital technologies.

For providers using or intending to use AI, trust will be critical to research and data collection, particularly where historical datasets are unavailable. Financial data, being particularly sensitive, are influenced by privacy and data protection concerns which may vary by gender. For example, women may be more hesitant to share personal information (Allen, Reference Allen2000; Park, Reference Park2015). This gender-based reluctance can affect AI training datasets despite assurances of beneficial use if the providers are lesser-known. Conversely, women may be more willing to trust established and proximate micro-finance institutions for services like credit eligibility assessments. According to Fu and Mrinal Mishra (Reference Fu and Mishra2021), the elements of trust in the fintech space include the number of years since a company was established as a measure of stability; whether the provider is headquartered in the same country/locality (as opposed to foreign entrants) as measures of geographic/cultural proximity; number of registered trademarks; and number of media articles as measures of reliability and familiarity respectively.

Perhaps, banks and fintech are developing more collaborations for this reason. Kuda, a digital bank, partnered with GT Bank and Zenith Bank to allow Kuda customers to make over-the-counter deposits at its branches to fund their Kuda accounts. Cowrywise similarly partnered with Sterling and Wema Banks to hold the funds of customers saved on Cowrywise. Piggyvest entered a deal with Providus Bank to create direct deposit account numbers for FinTech’s users (Prove, 2022). However, these partnerships are not always helpful as they tend to lock in existing customers and can perpetuate financial exclusion. This is explained in the next section.

4.3. Understanding the gender-blind approach to AI in finance

The preceding sections reveal fintech’s inadequacies in addressing women’s financial exclusion, regardless of AI integration. Beyond statistics, hindrances to AI’s role in women’s financial inclusion encompass factors like AI technology’s inherent nature, design and policy choices, and profit-centric business models.