What works to increase financial inclusion and women’s financial autonomy? Intentional designs showing promise

ABSTRACT Despite gains in financial inclusion and evidence that increased access to services improves women’s ability to realise their goals, a gender access gap to financial services of 9 percentage points prevails in lower-income countries. This stubborn reality calls for new approaches to reaching and supporting women to access financial services, including greater consideration of the gender norms and institutions that inhibit women’s full use of financial services. This article shares two experiences where CARE worked with partner banks and local non-profits to deliver financial services coupled with intentional interventions designed to address gender norms.


Introduction
While there have been myriad programmes and interventions designed to increase women's financial autonomy, many have focused on the financial system, structures, and the delivery of mass-market products. While such interventions are critical to address many of the barriers women face regarding access to and use of financial services, they often fail to consider informal structural domains and other barriers that inhibit women's ability to take advantage of otherwise "accessible" financial services, realise their financial goals and participate equally in household financial management. These barriers range from informal gender norms that limit women's autonomy to open a bank account, to structural barriers that limit women's numeracy and financial literacy levels (Groupe Speciale Mobile Associates 2018). Yet, equitable financial access and use are not impossible. If gender barriers are intentionally addressed, women's access to and use of formal financial services can be increased, the gender access gap reduced, and women's economic empowerment advanced. This fundamental hypothesis guided a series of CARE initiatives designed to close the gender gap and increase women's financial autonomy. These initiatives built on over 25 years of CARE's experience increasing women's access to financial servicesformal or informalthrough the promotion of community-based savings groups.
CARE began implementing projects to increase women's financial autonomy and inclusion nearly 30 years ago. To improve women's resilience and economic well-being, CARE harnessed the power of community-based savings groups, codifying the practices that led to the highest calibre, most sustainable groups and then scaling those practices globally; to date, CARE has supported over 7 million women and men to join savings groups, 81% of whom are women. Community-based savings groups of 15-30 members meet weekly or bi-weekly, and have proven to be a reliable route to improving women's access to and control over resources (Beaman, Karlan, and Thuysbaert 2014;Ksoll et al. 2016). Eighty-nine per cent of groups continue to operate five years after formation (VSLA Associates 2019) and groups replicate organically at rates between 2:1 and 3:1 (Anyango et al. 2007;Financial Sector Deepening Kenya 2012;Mine et al. 2013). As savings groups mature, they mobilise larger sums of money and look for more savings and credit options as individuals and collectively, driving demand from many to access formal financial services (FFS).
Since 2009, CARE has worked with leading financial service providers (FSPs) to design products and services to meet this demand. As a result, over 1 million savings group members have opened accounts with partner FSPsthe first formal bank account for most. CARE has worked on myriad digital solutions unlocking opportunities for many group members who were previously too rural or remote to be served by traditional bricks and mortar operations. However, many customers continue to prefer to operate their accounts via branches and ATMs.
The business case and evidence base for supporting savings groups to open bank accounts were promising (Datassist 2017); however, to more deeply advance women's financial autonomy required additional interventions. Women faced barriers including requirements for a co-signer on application forms, expectations to consult a woman's husband before opening an account, lack of numeracy or understanding of banking operations and expectations of customers. To understand the incentives that would motivate FSPs and other stakeholders to invest in addressing these barriers required further operational research. In response, CARE designed programming to explore three key questions: (1) Could community-based savings groups be provided financial services on a commercial basis at scale? (2) If so, what impacts would this have on savings groups and their members? (3) How could interventions address formal structural barriers (e.g. lack of available infrastructure or member identifications and group registrations), as well as informal structural barriers (e.g. social and gender norms that limit women's exercise full choice in opening and using an account)?
This article shares implementation experience and learning from two distinct but related CARE projects in East Africa: the LINK Up project, implemented in Kenya and Tanzania, where CARE tested financial products and business models designed to link community-based savings groups with FSPs in the formal sector; and the Digital Sub-Wallets project, implemented in Uganda, where CARE has completed a pilot and is currently implementing the full-scale project which includes a digital financial platform tailored to meet women's unique financial needs and addresses gender norms through intra-household dialogue sessions.

LINK Up
The LINK Up project operated from 2014 to 2017 with support from the Bill and Melinda Gates Foundation. LINK Up was designed to test two core hypotheses: (1) Providing access to formal financial products for community-based savings groups would deepen the positive impacts groups had on the lives and livelihoods of their members. (2) Delivering these services at scale would provide a compelling business opportunity for FSPs. LINK Up leveraged "alternative channels" (e.g. mobile phones, mobile bank agents, and mobile money) in Kenya and Tanzania to support savings groups' and individual members' access to FFS and improve group security as cash would no longer be stored in physical boxes. The project sought to support 8,500 savings groups representing 195,500 members to access and utilise FFS while supporting 52,000 savings group members to establish and utilise personal accounts. While implementation was more geographically spread out in Tanzania, implementation was more concentrated in the Northern and Western provinces in Kenya. Further descriptive statistics of a representative sample of the LINK Up participants are in Table 1. To develop and deploy products tailored to the needs of savings groups and their members, CARE cultivated partnerships with leading FSPs: Equity Bank and Kenya Commercial Bank (KCB) in Kenya, and National Microfinance Bank (NMB) and Access Bank in Tanzania. The LINK Up intervention package constituted: (1) The development of tailored financial products for savings groups including group and individual savings accounts and group loans that could be fully accessed through alternative channels (detailed in Table 2) 1 (2) Capacity-building initiatives targeting savings group members to enable them to better assess and decide whether to pursue, register for, and use the developed products (3) Facilitated onboarding support to savings groups who opted to open accounts to ensure groups were confident users of alternative channels and were able to fully utilise their accounts.

Methodology
CARE partnered with Datassist and Ipsos to evaluate LINK Up. The evaluation design constituted a mixed-methods approach combining a statistically representative quantitative survey with indepth qualitative research. The evaluation was designed to determine the degree to which the project targets had been reached, the underlying social, economic and behavioural changes and challenges that influenced these outcomes, as well as the lived experiences of project participants. LINK Up used pre-post data from a quantitative, panel survey collected through a rolling baseline and single point end line to isolate results based on the duration of interaction project participants had with LINK Up interventions. The survey sampling method looked at households as a unit, not solely women participating in LINK Up; in total, 389 households including both male and female-headed households (MHH and FHH) were surveyed. Quantitative survey findings were then used to inform the design of the qualitative approach, including a series of 32 focus group discussions (FGDs) conducted at project end with 362 individuals representing 160 savings groups. LINK Up account usage data was then compared and benchmarked against third-party data from the World Bank Findex, national Demographic Health Survey, and Savings Group Information Exchange (SAVIX) data to contrast LINK Up participant experience and outcomes with those of the broader population within project implementation areas.
Where possible, analyses of the data included disaggregation by sex as well as by MHH versus FHH status.

Results
By project end, 13,165 savings groups opened accounts with LINK UP partner FSPs, serving an estimated 322,000 members. Critically, these figures include both groups reached directly by CARE and those reached by bank partners through their traditional sales channels once the products were shown to be promising commercially. CARE's evaluation found that groups with accounts realised higher returns, and group cohesion remained the same or slightly improved. Women who opened group and individual accounts reported higher degrees of control over resources and increased ability to achieve financial goals (Vandergaag, Kwilasa, and Krause 2017). Offering group accounts resulted in increasing women's individual account ownership. Among groups directly supported by CARE (n = 1827 savings groups/46,512 members) 38% of women members participating in LINK Up in Tanzania had bank accounts at end line compared to 11% of women members at baseline. By contrast, women's bank account ownership among the general population in Tanzania during the same period increased by only four percentage points from 11% to 15%. Similar trends are seen in Kenya, where LINK Up participants saw an increase in women's access to FFS from 16% to 45% while the change was 15% to 30% nationally. Across both Kenya and Tanzania, LINK Up exceeded its original targets with 7,719 savings groups reached directly by CARE accessing project-developed financial services and products. The real illustration of success was 5,446 additional savings groupsunaffiliated with CARE but attracted by FSP marketing campaignsopening group accounts at partner banks. The reach was enabled in part by CARE's partner selection, which focused largely on large banks with branches nationwide in each country. By project end, CARE had directly reached 177,912 individuals (126,318 women); additionally, over 81,000 individuals (80% women) had opened individual accounts. For most members this was their first-ever bank account (Vandergaag, Kwilasa, and Krause 2017).
Financially, these linkages proved valuable for groups and members. Linked groups experienced a higher average return on savings (33%) than non-linked groups (19%), translating to gains of US$40-55 per member per cycle. Linked group members invested more money in their savings group and took and repaid more loans than unlinked groups with a similar maturity. This change was particularly intense for groups with the highest proportions of women members, suggesting that women in these groups were generating, gaining access to, and exercising control over additional resources. Groups that opened accounts also appear to have equally strong if not stronger degrees of group cohesion. The evaluation used group attendance and drop-out ratescomparing groups with and without formal accounts; findings suggest that while the average annual attendance for LINK Up groups supported by CARE was 83%, the attendance rate for other INGO-supported groups without any linkage training ranged from 73% to 91% (Vandergaag, Kwilasa, and Krause 2017). Qualitative findings reinforced this conclusion, with members citing an increased sense of security and trust in their group as a result of opening group accounts with formal institutions; some women shared they had re-joined groups they had previously left due to their trust in the structure and mechanisms by which funds were now managed.
Women in LINK Up households reported an increased ability to personally decide how to spend their savings. Meanwhile, within both MHH and FHH, women's decision-making agency in regard to accessing a loan also improved. Within both types of households, women experienced an increase in decision-making agency, but the change among women in MHH was more significant. Within FHH, 2 women were 38% more likely to be able decide to take out a loan in comparison to baseline, while women in MHH were 80% more likely to be able to decide to take out a loan in comparison to baseline 3 ; findings were statistically significant. This metric had a higher baseline value for FHH than for MHH, and the change observed at end line was smaller (Datassist 2017). Evaluation results also suggest that women experienced a change in how overall financial decisions were made in their households, shifting from male-dominated decision-making to more collaborative decision-making. Within FHH, there was a 26% increase in women exercising decision-making power over their savings; in MHH, there was a 45% increase. Statistical modelling to explain the variation attributed 68% of the increase to linkages facilitated by LINK Up and 32% of the increase to normal dynamics among savings group member households, whether the group was linked or not. By comparison, women from the broader communities in Tanzania and Kenya reported an 8% increase in decision-making power over the same period. This analysis suggests that LINK Up interventions had a notable impact beyond the changes happening throughout society (Datassist 2017). Specifically, savings group members were more likely to see improvements in intra-household dynamics than non-members, and members who were also in linked groups saw a further improvement in these dynamics over non-linked savings groups.
LINK Up demonstrated that having a group account was an effective vehicle for increasing women's overall trust in and confidence navigating FFS. CARE observed that this change was largely related to the effect that group solidarity had on individual agency; group members worked together to build individual women's confidence in the group and overcome their apprehensions about accessing FFS. Analysis of FGD data suggests that participation in the project had a notable impact on the way both men and women felt about financial institutions, particularly banks.

Challenges
Despite these promising results, the pace of account openings fell short of initial expectations. After an in-depth look into protocols, our implementation approach and specifically the linkage protocol were adjusted. It was determined that the branch staff needed more help understanding product details and how best to explain them to customers. Additionally, CARE adjusted compensation for frontline trainers to align with the additional time investment required to support savings groups; and, most critically, CARE adjusted the outreach approach to more efficiently assess initial group interest. This process was refined to five steps: (1) demand creation event; (2) planning VSLA accounts; (3) account opening and activation; (4) account opening and usage training; and (5) follow-up and follow-through. Community-based trainers engaged with groups at every stage, providing on-demand support to groups based on individual questions. Rather than contacting or visiting all groups in a region, CARE and FSP partners began convening large cluster meetings, inviting groups that were interested in the product or opportunity to send a delegation to learn more. For interested parties, CARE provided goal-setting support to help participants un-pack and distinguish between individual and group financial goals and explained how financial services could help to achieve these goals. This included one-on-one help making their first deposits and transfers, building confidence through action plans to help them actualise their financial goals over time, and suggestions on how to leverage their accounts to support savings behaviours. While CARE's experience suggests that making a plan for an account supports increased usage, the project team also added content to the onboarding training to ensure that group members understood how to maintain active accounts and thus avoid previous experiences with account dormancy. The revamped delivery approach substantially reduced the number of groups that were trained but subsequently did not open accounts and increased the level of account usage among those who opened accounts.
Beyond accounts, the project also illustrated the limitations of an approach focused on improving financial products and supporting women to access them. Qualitative insights illustrate how, even after opening accounts, women had varying degrees of trust and transparency within their households (Vandergaag, Kwilasa, and Krause 2017).
. "Depending on the kind of husband, whether he is supportive or not, will determine whether you tell him about your savings". (Female savings group member, Kibaha, Tanzania) . "If my husband shows me what he is making/saving, I will show him what I am saving" (Female savings group member, Morogoro, Tanzania) . "It all depends on the type of relations you have at the household level. With a good relationship, you will share this information willingly". (Female savings group member, Morogoro, Tanzania)

Lessons learnt
. Interventions were not sufficient to support gender norm change: Providing linkages to financial products and institutions proved beneficial, to a point. For many women, access to a personal bank account provided the privacy necessary to maintain control over earned and saved income; however, it left unchanged women's concern that were their spouses to learn of the funds, women would lose control over them. Based on assessment using CARE's Gender Marker, the intervention therefore may reflect more of a gender "sensitive" approach than one that is "transformative" (CARE International 2016). 4 The unequal gender norms and power relations within households that prompted women to keep their income secret were accommodated through the provision of a financial product that supported this secrecy, instead of directly addressing the gender norms within households such that finances were shared transparently, and household financial management, more collaborative. With a gender "transformative" approach, the project would have sought to transform the gender norms that underpinned women's inability to openly share their finances and work to support equal relations between household members. . Digital channels proved challenging: Perceived risks surrounding lack of phone ownership and financial literacy among members motivated CARE to focus on group-based interventions and FSP linkages; however, overall group adoption of the digital channels was lower than anticipated. CARE found that despite awareness of mobile banking and the potential for time savings, many savings groups continued to use bricks and mortar outlets. Myriad drivers contributed to this: fear of making mistakes; perceived high fee rates in the mobile system; a lack of confidence with mobile banking; frequent network failures or challenges with mobile agents; and preference for paper receipts for group records.

Digital Sub-Wallets project
Few, if any, mobile platforms designed around women's financial habits and needs exist. To incentivise use, savings technologies may use soft or strong commitment mechanisms, both of which have found success in encouraging individuals to save. Soft commitment saving mechanisms rely on the flexibility of "mental labelling"the allocation of funds for a specific purpose, but not restricting them for that purpose. Strong commitment saving mechanisms (i.e. those which dictate what savings can be used for) focus on overcoming time-inconsistency (i.e. situations in which user preferences or priorities change over time) as a constraint to saving. A study in Kenya demonstrated that mental labelling is preferred to strong commitments, in this case group savings allocated for healthcare (Dupas and Robinson 2013). As CARE's experience through LINK Up underpinned, financial inclusion of women cannot be fully achieved by focusing on financial products, gender norms must also be addressed. For example, providing women access to ATM cards in Kenya reduced account usage by women with low bargaining power in the household; without addressing inequitable gender norms within the household that underpinned women's limited control over their funds, the ATM cards simply made women's accounts and their funds more vulnerable to appropriation by their husbands or other members of the household (Schaner 2016). Building on these evidence and learnings from LINK Up, and CARE's Mind the Gap study in Rwanda (CARE International 2012), CARE recognised the need to design a savings group approach that would simultaneously address women's unique needs in terms of formal structures (e.g. FFS) as well as individual agency, relations, and informal structures (e.g. social and gender norms) which underpin the barriers women face in accessing financial services and exercising control over earned income. In the Rubirizi and Bushenyi districts of rural western Uganda, these factors are amplified by women's restricted mobility and extensive distances to FFS further limiting their access. The design of the Digital Sub-Wallets project sought to address these factors based on a set of hypotheses (see Figure 1): (1) By addressing inequitable gender norms and relations at the household level the unintended consequences of targeting women with financial innovations (e.g. digital financial products) could be mitigated and women could realise their financial autonomy and participate more equally in household financial management.
(2) By facilitating space for dialogue, households would be able to recognise the diversity of their goals and reach a joint vision ultimately elevating the status of women's financial needs and goals. (3) Through these combined interventions, along with financial and mobile money literacy training, women would realise increased agency, participation in household financial management, and financial autonomy.
Based on these hypotheses, the Digital Sub-Wallets project used a two-pronged design: a digital financial product tailored to meet the unique needs of women largely excluded from financial systems, and a dialogue intervention to transform inequitable gender norms and relations within households.

The pilot
CARE conducted a six-month pilot targeting 586 savings group members (450 of which were women) in the Bushenyi and Rubirizi districts of western Uganda before beginning full-scale implementation of the project. Participants were selected at random and were not excluded from participation based on their literacy or numeracy skills (or lack thereof). Demographic information about pilot participants is in Table 3. Pilot participants were split into two groups: The Digital Sub-Wallet intervention offered to 287 savings group members in Bushenyi district; and the household dialogue sessions were offered to 299 savings group members in Rubirizi district. The pilot objectives were to: gain implementation experience with two new intervention designs; inform key research design elements; and build a relationship with CARE's private sector partner, PostBank.

The digital sub-wallet
Before savings group members could be offered the digital sub-wallet, CARE and PostBank, had to design it. Building on the concept of mental labelling, CARE worked with research partners DoubleX Economy and Innovations for Poverty Action (IPA), to identify savings categories that women prioritised for longer-term savings goals; these ranged from seasonal savings needs (e.g. school fees), to emergencies (e.g. health/medical), to longer-term goals (e.g. land). With the identified savings categories CARE worked with PostBank to design a user-friendly mobile platform called "Women in Progress" which users could access using their simple mobile phone running USSD. Instead of seeing their savings in one lump sum in their account, Women in Progress allows users to allocate savings towards specific medium to long-term savings goals. Once developed, the project was able to begin activities to roll-out the application.
The project was designed to work with savings groups that had been operational for two years or more; while many participants had been part of a savings group for at least two years, few had a bank account of their own. However, instead of starting immediately with bank account sign-up, CARE and local partners Community Volunteer Initiatives for Development (COVOID) and Ankole Private-Sector Promotion Centre Limited (APROCEL), recognised the need to take a graduated approach. Activities with the savings group members offered the digital sub-wallet started with financial literacy training focused on core concepts of financial management: budget management, financial planning and goal-setting, and banking literacy. Following the financial literacy training, savings groups participating in the sub-wallet arm of the pilot were offered the opportunity to sign up for PostBank bank accounts and trained by APROCEL on how to access and use Women in Progress.

The household dialogue
Before the pilot began, CARE drew on principles of gender norm transformation and intergenerational conflict resolution to develop a household dialogue curriculum; the curriculum includes a series of modules designed to foster equity among all household members and specifically engages all household members who influence decision-making (CARE Uganda 2017). Once completed, CARE and partners trained female and male community-based facilitators to conduct the household dialogue sessions in the pilot. The modules of the curriculum were covered during seven weekly sessions covering seven domains: communication and decision-making, participatory visioning, growing collaborative power, household discussion; aligning visions; developing financial goals; and financial planning.
Savings group members were first introduced to the household dialogue intervention through their group; members who expressed interest in participating in dialogues then had to attain agreement from all other adults influencing decision-making within their respective households to participant. Given the nature of households with multiple adults living at home within Bushenyi and Rubirizi, due to practices of polygamy and extended family living with their siblings or children, gaining support from all household members was particularly challenging for some savings group members. Once all household members had agreed to participate, the community-based facilitators began the first session. Community-based facilitators for the household dialogues received an intensive week-long training, simulated in-house training prior to starting household dialogue sessions; and continued support from COVOID thereafter. Facilitators focused first on fostering an open and safe space for discussion allowing household members to share their respective financial goals, discover the diversity inherent in them, and come to alignment. This process sought to foster balance and equity within the household as well as elevate the status of women's needs and goals and increase their financial autonomy and role in household financial management.

Methodology
CARE partnered with DoubleX Economy, LLC and Innovation Poverty Action (IPA) to evaluate the project and conduct the research for the pilot as well as the randomised control trial. 5 The pilot research design examined the effect of three arms: (1) savings group members in Bushenyi (n = 287) were offered the digital sub-wallet product, on-boarding and financial literacy training; (2) savings group members in Rubirizi (n = 299) who participated the household dialogue sessions; and (3) a control sample of savings group members from a comparable area (based on socio-cultural factors and demographics) within which no project activities were implemented. 6

Pilot results and scale-up
The pilot provided key insights into the adoption and acceptability of the interventions and results showed strong uptake of both interventions. As highlighted in Figure 2, of the 287 members offered the digital sub-wallet intervention, 57.8% of women (and 60.7% of men) submitted paperwork to open an account; these findings reflect the realities faced in the pilot that fewer women than men were able to produce the necessary documentation to open accounts and that the pilot period was not sufficient for participants to complete the process of opening their accounts. This uptake was high in comparison to previous studies of similar products in Uganda (Intermedia 2018). In addition, 31.8% of women (and 44.6% of men) completed the process to open bank accounts; this was especially notable given that a mere 5% of women sampled for the pilot had bank accounts at baseline. Of particular note, however, was that a mere 16.1% of women activated the digital sub-wallet mobile access to their accounts in comparison to 35.7% of men (Scott, Custers, and Dolan 2017). Pilot findings confirmed demand for the household dialogues was high; of the 299 savings group members offered the household dialogue (228 women; 71 men), 89% enrolled. 7 In addition, 80% of Figure 2. Bank account uptake, digital sub-wallets pilot participants. enrolled households completed all seven sessions over seven-week period. Finally, field reports as well as the final pilot evaluation suggested increases in women's participation in financial planning and decision-making at the household level among women in the household dialogue arm (Scott, Custers, and Dolan 2017).
Following the pilot, CARE began full project implementation in February 2018, scaling interventions to reach 1,320 more women. In contrast to the pilot, in one arm of full-scale implementation, project participants are participating in both the household dialogue as well as the digital subwallet interventions while the second arm is participating in the sub-wallet and financial literacy training only. Current findings from ongoing monitoring and evaluation suggest that those participating in the household dialogue sessions and financial literacy training for the digital sub-wallet are experiencing and driving a lot of change, such as improved communication between spouses and expression of positive feelings. During field visits, participants engaged in the household dialogue sessions in particular are sharing experiences of significant change in their relationships, decisionmaking processes, and financial health. In discussions with project field staff during monitoring visits, participants mostly attribute these changes to project activities.
There continue to be myriad types of households (e.g. polygamous families, couples living with extended family, etc.) and CARE has identified that they each have unique needs and additional monitoring and support visits to these households are helpful. It has also been noted that increased interaction with CARE and partner staff may alter the experience of these households with the project and thus the research results. As such, CARE and partner staff have been careful to document the frequency of visits and have avoided providing additional counselling to these households while remaining vigilant in noting cases in which additional services may be needed and have followed up with COVOID and APROCEL to link these households with further services (e.g. social services, HIV counselling and testing, etc.).

Challenges
CARE and our partners faced a number of unique challenges during project implementation: . Household dialogue sessions: Community-based facilitators found that men dominated the discussions and left little space for women in the dialogue, reflecting a gap in training prior to the start of the pilot. To address this, an intensive week-long gender training was designed and conducted for all partner staff as well as community-based facilitators prior to beginning full-scale implementation. The training covered concepts of gender, norms, power dynamics and how to navigate them, and how to address them. The training also covered how to use the gender-based violence referral pathwaydeveloped specifically for the projectand provide referrals to survivors. . Diverse family structures: Practices such as polygamy were higher in the implementation area than anticipated; during the pilot, community-based facilitators struggled to adapt the onboarding approach to households with multiple wives, especially in cases where one of multiple wives was enrolled in the savings group, or in cases where only one of multiple wives was interested in the household dialogue session. In response, the CARE team developed a set of supplementary guidance for the household dialogue curriculum to support community-based facilitators through the process of onboarding polygamous households as well as other households who fell outside of heteronormative family archetypes for which the curriculum had originally been designed. To accommodate FHH, which are common in Eastern Uganda for myriad reasons, it was determined that while women representing single-headed households could not participate in the household dialogues given the nature of the intervention, that they could have access to the financial literacy component offered through the digital sub-wallet activities through their savings group. . Changes in data requirements: During the pilot, mobile network operators, the National Identification Registration Agency and banks were capturing data from clients in different ways that were not linked with banking "know your customer" requirements which demand uniform personal data; this presented challenges which have been reduced in full-scale implementation and uptake of harmonised guidance for personal data collection. The new guidance and use of National Identification have made it possible to remove the requirements for third-party signoff and a recommendation from the village local council which were previously requested to confirm the location of both men and women participants seeking to open accounts.

Lessons learnt
. Pilot findings demonstrated higher levels of digital illiteracydefined as the ability to use technology (in this case a phone) to access a serviceamong targeted savings group members than anticipated; in response, interventions for full-scale implementation were designed to include intensive digital literacy training for women, including one-on-one support on the use of the Women in Progress application. . Feedback received during the pilot suggested that the sub-wallet labels did not reflect women's most common savings goals. CARE and PostBank updated the interface and labels for each of the sub-wallets to better align women in savings groups had identified during a rapid assessment following the pilot; the notable changes were additions for "wedding" and "personal needs" as reflected in Table 4.

Recommendations
. Linking savings groups to FFS and bank accounts has proven promising; however, in order to be successful, development actors and FSPs must partner together and explore different partnership models to create an enabling environment and develop the necessary infrastructure to overcome embedded social and gendered barriers such that women's access to and use of financial products increases. This partnership must begin at project start-up and be maintained throughout implementation. Further, design decisions must be made based on the desired outcomes of the project. Starting with paper-based savings groups for a few years may be necessary to reach the maturity necessary to link with FFS; however, if enabling transactions between group members is the end goal, then starting with a digital platform from the outset may be possible. . Resources should be allocated to enable sufficient training prior to the start of implementation for project staff and throughout implementation for participants. Training was a key factor in the success of both projects. For staff, adequate training prior to the start of implementation on the digital product as well as gender was essential. For participants, especially women participants, training sessions designed to improve financial literacy and build their confidence using a mobile device for banking proved successful. . Addressing both the financial product needs of savings groups and members as well as the gender dynamics that influence women's access to and control over resources is critical to the success of initiatives designed to increase women's financial autonomy and economic empowerment. LINK Up focused on increased access to financial services and delivered training and product access to individual women and savings groups. The results were promising; however, the combined effect of product development and training with household dialogues to address gender norms within households as seen within Digital Sub-Wallets appears more promising in facilitating women's economic empowerment. . Related, women must be engaged in every step of the process from identifying major barriers to informing product design and implementation approaches. CARE's projects worked with existing savings groups to understand the biggest pain points they faced in pursuing their economic goals. Partners were brought in to collectively internalise this feedback and develop potential interventions. Through a process of iteration alongside women in the projects, the solutions and roles of all partners were defined and proven before implementation was scaled up. . Further operational research is required to determine how best to deliver interventions that address the gender and social norms and structures that inhibit women's control over financial decision-making at scale. Until gender norms change, financial inclusion efforts risk providing marginal gains but ultimately falling short of the ambitions laid out in Sustainable Development Goals (SDG) 1 (ending all poverty); SDG 5 (gender equality), and SDG 8 (creation of decent employment for all and promotion of inclusive growth).

Conclusion
It is necessary to consider and integrate gendered approaches in order to overcome the informal structural barriers that women face in accessing financial services and realising their financial goals. The Digital Sub-Wallets project shows that a dual-pronged approach focused on women's financial autonomy as well as intrahousehold relationships can succeed. While the development of tailored financial products designed to meet the unique needs of women can address formal structural barriers and increase women's access to financial services, implementation experience suggests that not only are household dialogue sessions acceptable and desirable to savings group members and women in particular, but they can also create space for familial units to discuss their financial goals, ameliorate communication and negotiation skills, and ultimately increase shared planning and decision-making around household financial management. In both LINK Up and Digital Sub-Wallets, the existence of high-quality savings groups with high rates of women's participation and a history of working together to identify and meet economic and non-economic goals offered the projectsand partner banks and NGOsa low-cost entry point and effective vehicle for introducing and advancing the intervention. While these approaches and interventions have shown promise in addressing gender norms and increasing women's financial autonomy and savings, results are preliminary and it will take more time to see and measure sustainable change; the project length in both projects was simply insufficient to glean insights surrounding broader and longer-term sustainable changes in economic and other development outcomes (e.g. health).
Notes 1. CARE's theory of change for supporting VSLAs and their members to access FFS focuses on supporting savings-led linkage to allow first-time account holders to build confidence and experience, and establish a track record with their FSP. Credit and other products are introduced later. 2. Even in FHH, women often have men from their extended family who exercise power over their decision-making.