Savita Bailur, Hélène Smertnik
Digital financial services (DFS), which include mobile money transactions, lending, savings, insurance, and other products, are often touted as the path to financial inclusion for women in emerging markets . The gender gap in access to digital financial inclusion remains high, at 33 percent across low- and middle-income countries. But it varies widely at a country and regional level, narrowing in some areas such as Sub-Saharan Africa (20 percent), East Asia and the Pacific (21 percent), and Latin America and the Caribbean (25 percent). At the same time, the gap remains very high in other regions, such as South Asia (75 percent) . The gender dimension to mobile money has become even more relevant during the Covid pandemic, when the use of mobile money has greatly increased . However, what remains underresearched is whether certain DFS features impact women more than men and whether more investment in R&D is needed for female customers—and if so, why? A critical element of this is design and, therefore, human-computer interaction (HCI).
This article summarizes research we undertook in Kenya and Côte d'Ivoire, two mature mobile money markets in the region, for the Bill and Melinda Gates Foundation on the gender impact of the Level One Project Principles (https://www.leveloneproject.org/). The Gates Foundation launched the Level One Project Principles in 2017 after consultation with hundreds of stakeholders on a minimum set of principles to make DFS systems inclusive, accessible, and affordable. A prototype of this is Mojaloop, open source software that faciliates the interoperability of mobile payments. However, until now, an in-depth gender analysis had not been undertaken on how some of the principles may affect men and women differently.
We deployed 24 focus groups, conducted 28 in-depth interviews (all resulting in a combined 160 end users, both male and female), and spoke to 20 DFS country experts in total. Our overall aim was to understand whether certain DFS features impacted women's access and usage more than men. This led us to realize that the key underlying principle was creating DFS systems that were considered trustworthy by women. We acknowledge that women are by no means a homogeneous group, and the intersectionality of elements such as race, income, class, location, and age are critical (indeed, we talk more about this below). Having said this, we introduce three HCI elements that impact women's perceptions of DFS from the research we conducted .
One clear finding from focus groups and interviews was the confidence afforded by confirmation notifications for women. For example, in Kenya, the dominant mobile money service, M-Pesa, has a multistep process to confirm a transaction, as well as a final 25-second delay built in before "send" is hit. In our focus groups and interviews, more women than men mentioned a delay mechanism as a source of comfort, as they felt it would prevent them from incorrectly sending money to someone else. This was especially important in the context of Kenya, because if an incorrect recipient has Fuliza (an overdraft payment system), the money is automatically put toward repaying the overdraft and is harder to retrieve.
Similarly, in both Kenya and Côte d'Ivoire, women valued real-time notifications because they knew exactly when the money had been transferred. The lack of notifications (often exacerbated by network drops) creates confusion and wastes time. As one interviewee in Côte d'Ivoire stated:
I paid a bill and have not received a message for two days. My brother-in-law and I went to the agency to file a complaint. It was found that indeed I had paid the bill by mobile money and that I had not received a message.
The absence of notifications sows doubts, which leads many women to start withdrawing money from the mobile money account and physically transacting—the antithesis of what DFS are supposed to be about. It is time consuming for women, and also prevents them from making any further transactions if they are uncertain as to where the money is. (It should also be noted that women mentioned the need to have control over how notifications appeared, so they did not infringe on privacy.)
|Helen in her shoe shop, Gikomba market, Kenya.|
|Focus-group discussion with a group of women in Yopougon, outskirts of Abidjan, Côte d'Ivoire.|
|Information posters on costs and fees at a woman's mobile money shop.|
Finally, in both Kenya and Côte d'Ivoire, where most women in our sample were transacting using USSD shortcodes rather than through USSD embedded menus or apps, female customers said they often had to memorize the syntax for commands such as sending money and checking their balance. Eventually, the familiarity with strings of text enabled women to be more confident in their usage of the services, and was particularly valuable for those women who may not have advanced levels of literacy or/and digital literacy. On the other hand, remembering USSD shortcodes did lead women to stick to one provider (likely Orange in Côte d'Ivoire rather than MTN or Moov, as it was the one of the first and users had already memorized their short codes), because they couldn't necessarily remember all the providers' different shortcodes.
Fraud awareness among female customers was high in Kenya and Côte d'Ivoire, to the extent that many women were extremely careful about making transactions, often asking someone younger to do it or going to an agent. The result of this was that, as one expert interviewee stated, "so much emphasis has been placed on fraud awareness, which is good, but it means that cost transparency has been overlooked." In both Kenya and Côte d'Ivoire, women in our sample stated more concerns around DFS transaction costs compared with men—not only that they were high but also that they were opaque and that they could not anticipate charges.
Women valued real-time notifications because they knew exactly when the money had been transferred.
In addition, women hadn't developed the same workaround mechanisms to avoid fees compared with men, who usually were more comfortable using DFS. Some men we spoke with would send multiple lower-value transactions below the threshold of being charged. Very few women, except one young female student in Nairobi, knew about this tactic and were intimidated by the idea of having to spend a longer time doing several transactions, which could potentially lead to mistakes. Instead, female respondents complained of hidden costs and losing money in DFS.
A user-focused, HCI emphasis here would not only clearly display costs before and after each transaction (which systems do) but also have frequent messaging on cost as well as design, which instills trust. Instead, female customers feared being conned by "dark design patterns" . According to an expert interviewee, an example of this was an M-Pesa feature that asks the user if they want to top up their account immediately after sending funds, but resembles the process of sending. This may mistakenly lead the user to buy more credit—which they may not want to do.
Revocability (reversing a transaction after it has been made) was a particular fear for merchants, some of whom shared stories of how a customer would reverse a transaction after buying something, thereby getting an item for free. As Helen, a shoe seller at Gikomba market in Kenya, explains:
When you have finished with the customer and you find they have reversed the money. On calling customer care, they ask a lot of questions and you have to go to the agent and then realize the customer withdrew the money, but it is too late. Now I ask my customers to pay me cash—less trouble.
Irrevocability was therefore valuable for merchants, especially female merchants, who fear being cheated. Our parallel quantitative analysis also suggested that female merchants were likely to be less successful at retrieving their money, compared with men.
Some male respondents had mitigated the risk of a customer revoking a transaction by setting up a bill-pay or invoice system, but doing so requires a separate merchant account, which means a few more steps. Given the costs and procedures, this would be worthwhile only for larger businesses.
The above three elements were some of the overarching DFS design and product features mentioned by women. What also emerged in the gender analysis was that, as always, the human is critical in the system. Purely digital products were not fully trusted. For example, one respondent said, "These Tala [lending] apps, they don't even have agents so do they really exist? How can they help you if there is no one to go to?" Agents and other intermediaries are key here, especially when DFS features are unclear or intimidating. In our sample, female interviewees didn't express a preference for male or female agents; rather, they just wanted a professional who could provide advice and clarity on costs. They also wanted more standardized support and assured grievance-redressal processes rather than varied levels of help, though they acknowledged personal relationships mattered here. Men didn't express this need for agents, often preferring to conduct transactions themselves or ask friends for help.
Though the human touch is a critical part of a functional DFS, this also increases women's dependence on agents. And from an agent perspective, this also seems to add some pressure. One agent in Abidjan, Côte d'Ivoire, resented the consumer education he had to provide:
What also emerged in the gender analysis was that, as always, the human is critical in the system.
It's not my job to raise awareness. When I did my training, I was shown how to do transactions, how to welcome clients, and how to avoid risks and mistakes. I do advise clients how to reduce costs though. I earn very little. It exhausts me and brings me nothing.
This raises the question of whether, alongside the technical aspect of clearer mobile money design features, agents need to be explicitly trained in support as well as compensated for it.
Another issue relating to the human as opposed to the computer element was in "know your customer" (KYC), or ID, requirements. One of the benefits of mobile money is that it often has lower KYC requirements than banking. However, as one uses more advanced features (e.g., lending), more KYC requirements are put into place, which again, some women are cautious of for privacy reasons, or simply don't have. Female customers in Côte d'Ivoire appreciated lower KYC for the simplest aspect of mobile money: cash in/cash out transactions. As one respondent explained, "In banks, they ask you for evidence of your salary, bills—things like that. We don't have all that."
Women typically face greater challenges in obtaining identification credentials than men . This led some female customers to use workarounds, such as a SIM registered in someone else's name. One respondent in Côte d'Ivoire spoke of how a friend had registered a SIM under her name, but when the phone was stolen, she could not travel back to help her friend. Equally, agents in Côte d'Ivoire stated that they tried not to be too stringent on ID requirements because they knew a customer could go to another agent who wasn't as demanding around ID. This can be problematic because it creates unfair competition among agents, which goes unchecked by the mobile operators themselves. All these human elements of DFS, as well as varying standards, create confusion around the requirements and undermine the efficiency and safety that DFS should provide customers.
DFS appear to be rapidly gaining traction, especially during and most likely post-Covid. However, design features can sometimes make them unappealing for women, which led to a number of consequences in our research. First, in our pre-Covid research, female entrepreneurs were beginning to consider cash instead of mobile money, because of a risk of fraud and revocation. Second, hidden fees remain intimidating (though some DFS providers removed fees during Covid). The delay in receiving confirmation messages can also deter women in using services, as it adds stress. Sometimes there is also a lack of interoperability between mobile money networks (e.g., if the sender is on Orange and the recipient on MTN), as well as between mobile money providers and banks. Finally, a lack of standardization in ID and KYC can be problematic for women. The issue of exclusion is even stronger in the context of Covid-19, when payments are increasingly made digitally, negatively impacting women who do not have their own DFS accounts.
What our research found, therefore, was that digital financial services need to a) pay greater attention to the design features that build in trust for women, and b) still think about the human in the human-computer (mobile) interaction. As an aside, the issue is not always about gender, of course—our parallel quantitative analysis in Kenya and Côte d'Ivoire cut by demographic found that income mattered more than gender when it came to transaction amount and volume. Having said that, mobile operators and DFS providers must increasingly be looking to women as new markets, especially the rising, informal social-commerce or digital entrepreneurs; these are the demographics who need to be able to trust DFS with clear user interfaces and messaging.
We would like to thank our research partners on the ground. In Kenya, the AFROES research team, led by Gathoni Mwai and Sylvia Oloo. In Côte d'Ivoire, the Empow'Her research team, in particular Soazig Barthelemy, Chloe Roncajolo, and Serge Couadio. And we would like to thank the Gates Foundation Financial Services for the Poor team for funding this research as well as the Global Centre for Gender Equality at Stanford University for feedback, in particular Nicole Figot, Margaret Greene, and Angela Hartley.
2. 2017 Global Findex Database, ages 15+; https://globalfindex.worldbank.org/
3. The covid-19 crisis is boosting mobile money. The Economist. May 30, 2020; https://www.economist.com/middle-east-and-africa/2020/05/28/the-covid-19-crisis-is-boosting-mobile-money
4. See also Bailur, S. Which DFS features matter more to women than men? Experiences from Côte d'Ivoire and Kenya. Caribou Digital. Mar. 6, 2020; https://medium.com/caribou-digital/which-dfs-features-matter-more-to-women-than-men-3a21765b2236
5. Jaiswal, A. Dark patterns in UX: How designers should be responsible for their actions. UX Collective. Apr. 15, 2018; https://uxdesign.cc/dark-patterns-in-ux-design-7009a83b233c
6. Bailur, S., Srivastava, D., and Smertnik, H. Women and ID in a digital age: Five fundamental barriers and new design questions. Caribou Digital. Sep. 9, 2019; https://medium.com/caribou-digital/women-and-id-in-a-digital-age-five-fundamental-barriers-and-new-design-questions-79caa2a4acb8
Savita Bailur is a research director at Caribou Digital and an adjunct associate professor at SIPA, Columbia University, where she teaches a master's course in digital development. She has a Ph.D. and an M.Sc. in information systems from the London School of Economics. email@example.com
Hélène Smertnik is a senior researcher at Caribou Digital who previously worked on the GSMA Mobile for Development team. She has an M.A. from the University of Sorbonne and a B.A. from the University of Toronto. firstname.lastname@example.org
©2021 ACM 1072-5520/21/03 $15.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2021 ACM, Inc.