THE PROMISE OF EJO HEZA A Brighter Future for All Rwandans © 2023 International Bank for Reconstruction and Development/The World Bank. 1818 H Street NW, Washington, DC 20433, USA. Telephone: 202–473–1000; Internet: www.worldbank.org. Some rights reserved This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Guven, Melis, and Himanshi Jain. 2023. The Promise of Ejo Heza: A Brighter Future for All Rwandans. ©World Bank. License: CC BY 3.0 IGO. Translations—If you create a translation of this work, please add the following disclaimer along with the at- tribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third‑party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202–522–2625; e-mail: pubrights@ worldbank.org. Cover and text images: World Bank photo collection, https://www.flickr.com/people/worldbank/. Contents n  iii Contents Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii Specific design-related questions of interest to the RSSB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix Key factors for Ejo Heza success and lessons for other schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi World Bank policy recommendations based on analysis in the report. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Section 1: Background of Ejo Heza Long Term Savings Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Section 2: Statistics of Ejo Heza Long Term Savings Scheme and Highlights from the Phone Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain . . . . . . . . . . 18 Assess . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Enroll . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Contribute. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Provide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Manage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Section 4: Ejo Heza Member Preferences to Select Design Changes . . . . . . . . . . . . . . . . . . . . . . . . . 40 Short-term access to funds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Increase in the minimum savings level to be eligible for government incentives. . . . . . . . . . . . . . . . . . . . . . . . . . 42 Scaling up coverage through mandates and/or identifying aggregators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Eligibility restrictions based on age. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Section 5: Viability Assessment of the Ejo Heza scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Section 6: Policy Recommendations 1. Communications: Keep doing what’s working, fix the gaps, and explore new avenues. . . . . . . . . . . . . . . . . . . 53 2. Focus on improving adequacy, persistency, and preparing for withdrawals/payouts. . . . . . . . . . . . . . . . . . . . 54 3. Revisit incentives as ‘experience’ on the scheme develops. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4. Improve interoperability with other government systems and integrate with private sector. . . . . . . . . . . . 56 5. Strengthen learning, monitoring, and evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 6. Build human resource capacity and leverage external expertise as needed. . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 7. Collect better data and mine it for policy making. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Annex 1: Sampling Frame of Ejo Heza Phone Survey. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Annex 2: Indicators Recommended to be Monitored and Included in IT Dashboard. . . . . . . . . . . 61 Annex 3: Design and Messaging Recommendations for a Digital Contract. . . . . . . . . . . . . . . . . . . . 62 Annex 4: Contribution Targets by District for 2021/22. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 iv  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Boxes 1. Cooperatives are to Ejo Heza what employers are to formal sector pension schemes.......................................................................23 2 The Ejo Heza Information Technology (IT) team on challenges facing the system.................................................................................30 3 Key factors for Ejo Heza success & lessons for other schemes........................................................................................................................52 Figures 1. Family members of eligible Ejo Heza members receiving insurance compensation.................................................................................8 2. Ecosystem map of Ejo Heza.................................................................................................................................................................................................9 3. Ejo Heza member experience with existing and potential intermediaries...................................................................................................10 4. Employment type and gender subscribers...................................................................................................................................................................12 5. Household income by employment type....................................................................................................................................................................12 6. What is your plan for financial survival in old age....................................................................................................................................................13 7. How strongly do members feel about saving for old age....................................................................................................................................13 8. Ejo Heza’s subscriber and active saver growth over time.....................................................................................................................................13 9. Active male and female savers by province and overall........................................................................................................................................14 10. Count of subscribers and average saving by Ubudehe category......................................................................................................................15 11. Count of subscribers, active savers, and average savings by province............................................................................................................16 12. Reimagined the social insurance delivery chain.........................................................................................................................................................19 13. Social insurance delivery chain applied to Ejo Heza...............................................................................................................................................20 14. Posters and banners with tailored messages...............................................................................................................................................................21 15. Facebook post on how to access balance through the online portal............................................................................................................22 16. Ejo Heza roadshow (left) and member registration in field (right)...................................................................................................................24 17. Voluntary registration process diagram.........................................................................................................................................................................26 18. What an intermediary registration process might look like................................................................................................................................27 19. Members rating of the fiscal benefits in Ejo Heza................................................................................................................................................... 33 20. Subscribers’ rating of three key activities by level of difficulty........................................................................................................................37 21. Who do members ask if they have a question on Ejo Heza?.............................................................................................................................38 22. How much early access to funds would members like?.......................................................................................................................................42 23. Reported ability to save if minimum savings limit for government incentive was increased.............................................................43 24. Should U3 and U4 save mandatorily in Ejo Heza?....................................................................................................................................................44 25. Preference to mandate coverage for U3 and U4......................................................................................................................................................45 26. Reported willingness to save more if the government automatically contributes some tax revenue directly into your account.........................................................................................................................................................................................................................................45 27. Age distribution of Ejo Heza savers and non-savers...............................................................................................................................................46 28. Equation used to project Ejo Heza scheme revenues in the SVAT..................................................................................................................49 Tables 1: Program rules under Ejo Heza as of 2022 ...................................................................................................................................................................... 7 2: Reported savings in Ejo Heza, by employment and household income ......................................................................................................15 3: Insurance statistics, 2021........................................................................................................................................................................................................17 4: Percentage of members who would like partial access to Ejo Heza savings .............................................................................................42 5: Reported ability to save more compared with reported savings in Ejo Heza ...........................................................................................43 6: Summary of baseline and scenario results from Ejo Heza’s viability assessment carried out using December 2020 data...50 Abbreviations n  v Abbreviations GDP Gross Domestic Product HCI Human Capital Index IBNR Incurred but not reported ID Identity Document LI Life Insurance LODA Local Administrative Entities Development Agency LTSS Long Term Savings Scheme MEIS Monitoring and Evaluation Information System MINECOFIN Ministry of Finance and Economic Planning NAV Net Asset Value NIDA National Identity Agency NPS National Pensions Scheme NST1 National Strategy for Transformation 1 RCA Rwanda Cooperative Agency RNIT Rwanda National Investment Trust Ltd. RSSB Rwanda Social Security Board RwF Rwandan Francs SACCO Saving and Credit Cooperative Society SCD Systematic Country Diagnostic (World Bank) SMS Short Message Service U1 Ubudehe Category 1 U2 Ubudehe Category 2 U3 Ubudehe Category 3 U4 Ubudehe Category 4 UNCDF United Nations Capital Development Fund USSD Unstructured Supplementary Service Data VSLA Village Savings and Loan Association VUP Vision Umurenge Program XML Extensible Markup Language API Application Programming Interface vi  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Acknowledgements Call-center team of Ejo Heza. The report also includes findings from the field visit to Kirehe district by the T he report was written by Melis Guven and task team where we interviewed Mr. Nzirabatinya, Vice Himanshi Jain from World Bank in collabo- Mayor of Kirehe District; the Ejo Heza District Coordi- ration with the Ejo Heza team from the nator of Kirehe District; members of the Koprike Rice Rwanda Social Security Board (RSSB), which Cooperative; and three agents from the CARE VSLA comprised Augustin Gatera and Dan Rwiyamirira. In- group. We express our sincere gratitude to all senior puts to the report were provided by Abhishek Kham- local officials, and partners who have shared remarks in rai and Aditya Jagannath and Dr. Fareeha Armugham these consultations/interviews. The phone survey of assisted with the phone survey instrument design Ejo Heza members was carried out in September 2022 and Sunalika Singh provided specific inputs. Eric Gires by the RSSB call center under the guidance of DDG provided significant inputs to the Scheme Viability Regis Hitimana. Analysis. The team benefitted from excellent comments from The report would not have been possible without the formal peer reviewers Gustavo Demarco, Fiona Eliza- excellent support and coordination of the Ejo Heza beth Stewart, and Suleiman Namara. The team appre- team who conducted written interviews on behalf of ciates support from Iftikhar Malik and Silas Udahemuka the task team with the District Mayor of Burera and in the preparation of this report. We would also like to Gakenke district, Katecogro tea cooperative, Dukunde thank the RSSB management CEO Regis Rugemans- Kawa Cooperative of coffee farmers, and two CARE huro and Chief Benefits Officer Dr. Regis Hitimana for agents from Karongi district. Virtual consultations their support of this task. Essential editorial and design were carried out with Mr. Herbert Asiimwe from Min- support was provided by Deepika Davidar and Michael istry of Finance and Economic Planning (MINECOFIN); Alwan respectively. The report was produced under Germain Gatabazi and Saidi Sibomana from LODA; Mr. the overall guidance of Rolande Simone Pyrce (Rwanda Jean Bosco and Agnes Uwanyiligira from Access to Fi- Country Manager) and Paolo Belli (Practice Manager nance Rwanda; and with the Communication, IT, for Social Protection & Jobs, Africa East). Executive Summary The low pension coverage, notably within the informal sector/self-employed/part- time workers, has motivated countries across Africa, Asia, and Latin America to opera‑ tionalize voluntary schemes. The voluntary schemes globally differ in design, incentives, who they target, and withdrawal rules, but the objective for all is to offer a scheme that meets the needs of those traditionally excluded from formal sector schemes. Among such schemes, Rwanda’s Ejo Heza Long-Term Saving Scheme is emerging as a leader given its rapid coverage growth, notwithstanding the impact that COVID-19 had on its mobilization strategy. The only other voluntary scheme that has achieved success in coverage at the same scale is China’s RURS scheme (at 38 percent of working age population), and it offers a generous subsidy to the poor and vulnerable groups to encourage saving in the scheme (Jain and Palacios 2021). Over 80 percent of Rwanda’s labor force works in the agricultural sector, is self-em‑ ployed, or works informally in the economy as drivers, hawkers, artisans, and so forth, and therefore is not mandated to be covered under the Rwanda Social Security Board (RSSB). This leaves these workers and their dependents exposed to the risks of poverty during old age. Rwanda’s formal sector pension scheme covers only 8 percent of the labor force. The high proportion of workers in the informal sector combined with changing demographics (i.e., falling fertility and rising life expectancy contributing to rapid aging) poses challenges. In recognition of these challenges, Ejo Heza was introduced in 2018 under Law N° 29/2017. It was designed as a voluntary defined contribution scheme, with short- term withdrawals under specific contribution levels are reached , and targeted incentives to cater to the characteristics of informal sector workers. Administered by the RSSB, the Ejo Heza savings scheme has registered a remarkable increase in coverage since its imple- mentation in 2019. The fast growth in Ejo Heza’s coverage has meant that any analysis based on the date must be caveated. This report includes data from 2019, 2020, 2021, and 2022. The latest numbers would likely have changed by the report’s release. The report’s primary aim is not to provide the most recent numbers on Ejo Heza, which is a moving target. In- stead, it aims to comment on the trends and highlight the success factors and recommen- dations to improve the scheme. As of December 2022, the scheme has registered 2.9 million individuals, or 22 percent of the Rwandan population, or 37 percent of the working age population, many of whom are from low-income households. The number of active savers in Ejo Heza was 2.4 million as of December 2022. Data as of 2021 shows that about 49 percent of the subscrib- ers in Ejo Heza are females, while females account for only 32 percent of total contributors in the RSSB formal sector scheme. Informal sector workers comprised 87 percent of savers, while 12 percent were from the formal sector. viii  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Ejo Heza, a local term that translates to ‘a better fu‑ cost-effectively is also an area to focus on. Unless in- ture,’ not only holds the promise of income security dividuals save for a long duration, the benefits they at older ages but can also be a much-needed source can receive will not be adequate. Investment returns for domestic resource mobilization in the country. A can play a significant role in ensuring the growth of the boost to the domestic savings rate would propel individuals’ fund balance, but with a low contribution Rwanda’s growth and allow the country to scale up in- density, there are limits to the growth possible through vestments in the medium to long run. The assets under prudent investments alone. management as of January 2023 stood at USD 42 mil- lion, and as Ejo Heza coverage expands, the pool of While the cooperatives, as an aggregator, have savings will also grow. The scheme can grow to be one proven to be a prominent feature behind Ejo Heza’s of the largest institutional funds in the country and success, the scheme should also focus its efforts to could be a source for domestic resource mobilization. target informal sector workers with regular in‑ This would require robust governance arrangements comes. These include household enterprises, micro and sound investment policies in place so that the and small enterprises and SMES particularly those in fund is well-managed and provides fair returns to the urban areas. In doing so urban aggregators will need to members who save with the scheme. be identified (maybe banks/telcos/MFIs) and innova- tive methods to target MSMEs, and household enter- The ecosystem, delivery arrangements and political prises will be explored. Overall savings in Ejo Heza are commitment are some of the main factors for suc‑ beneficial from the micro and macro perspective. The cess. The ecosystem of Ejo Heza, based on a strong gains made by Ejo Heza despite the COVID-19 pan- central digital platform interoperable with the national demic is striking, and the low numbers in urban areas government ID and the social safety net databases, is could have been the direct impact of COVID on mo- a stable foundation for Ejo Heza and contributes to its bility restrictions in these dense settings. The scheme success. The ease of contribution, flexibility in product is still in early phases of design improvements and rec- design, and short-term incentives make this scheme ommendations as well as international experience can better cater to the needs of the uncovered popula- be helpful for Ejo Heza to achieve its goals. tion. Lastly and most importantly, the political will, clarity in policy direction, and ownership across differ- Coverage expansion in the voluntary schemes of ent levels of government make this a scheme that in- other countries (such as Thailand, India, Colombia, formal sector workers trust. Vietnam, and Kenya) has not been as high as that of Ejo Heza but there are three lessons learned which Despite its success, there are concerns to be ad‑ are considered to be applicable for Ejo Heza. First, dressed and room for improvement. Notable among flexibility in withdrawals is desired (India’s latest the challenges is a need to focus on improving savings scheme, Pradhan Mantri Shram Yogi Maan-dhan, offers adequacy and managing expectations. A back-of-the- much more flexibility than its predecessor scheme, envelope calculation suggests that an annual contribu- APY). Second, bundling can be attractive if it is com- tion of RF 45,000 for at least 25 years will be needed to municated well and if it allows informal sector workers generate a fund balance that can produce pensions to get similar risk coverage as the formal sector enjoys equivalent to the poverty line for 240 months. Current e.g., maternity, sickness, occupational Hazard is cur- average savings range between RWF 9000 - 12,000, in- rently only offered to formal sector in Rwanda. Co- dicating a need for a strong focus on improving ade- lombia and Vietnam in their voluntary schemes are quacy. If saving levels fail to increase, there is a risk of beginning to extend coverages for short term benefits unmet expectations and low pensions when people to the informal sector as well. Third, and most impor- retire. Similar reputational risks have arisen in manda- tantly, finding aggregators that can replicate the ‘ob- tory DC pension schemes, not because of design faults serve and auto-deduct’ role that employers play for but due to low or irregular contributions from the formal sector pension schemes is a gamechanger for members. Ensuring persistence in savings and doing so scaling up coverage. India has attempted to do so with Executive Summary n  ix Banks, some East Asian countries are attempting the assessment. The World Bank customized the Scheme same through platforms like Uber/Gojek, while some Viability Assessment Tool (SVAT) at the behest of are exploring partnerships with Telcos. the RSSB, which wanted to have a model that esti- mates the breakeven point for Ejo Heza under differ- This report is a product of a two-year engagement ent assumptions—customized for the scheme at the between the World Bank team and RSSB. It aims to request of RSSB. Data used for the SVAT assessment overview the Ejo Heza scheme, highlight its key was as of December 2020, when the number of sub- strengths, and provide recommendations to im‑ scribers was a quarter (about 789,900) of current num- prove coverage, adequacy, operational efficiency, bers. The fast evolution of the scheme since the SVAT and long-run viability. For this purpose, a rich set of assessment reduces the validity of modeling results. qualitative and quantitative information collected dur- However, the results highlight the benefits of doing a ing the engagement has been analyzed and compiled. viability assessment, especially in countries with limited The engagement included meetings with Ejo Heza fiscal resources, to support incentives indefinitely. core teams; interviews with stakeholders and partners of Ejo Heza; an assessment of Ejo Heza’s operational The findings from the report are presented in three processes using the social insurance adapted delivery areas to cater to the needs of different audiences. First chain framework; field visits in 2021; use of the World are specific design-related questions that policymak- Banks’s Scheme Viability Assessment Tool (SVAT) to as- ers at RSSB were interested in exploring using the sess Ejo Heza’s viability under multiple scenarios; and phone survey and through any international experi- most recently, a phone survey of 386 subscribers of Ejo ence. Second is the lessons learned from Ejo Heza’s Heza. The target audience for this report includes poli- success which is helpful for global audiences. Third, the cymakers at RSSB looking for scheme design and deliv- World Bank team’s policy recommendations are ery system recommendations to improve the desir- grouped into seven areas. ability and efficiency of Ejo Heza and global audiences who would like to learn lessons from Ejo Heza and understand the potential reasons behind its success. Researchers and practitioners studying voluntary Specific design-related schemes can also benefit from the frameworks, meth- questions of interest to the RSSB ods, and tools used for analysis in this report. Specifically, the report uses the Delivery Chain (1) Allowing short-term access to funds in Ejo Heza. framework, phone survey results, and the Scheme Sixty percent of phone survey respondents said they Viability Assessment Tool to structure the findings. had access to funds in case of emergencies or to meet The phone survey was carried out in 2022 with the Ejo short-term needs, while 33 percent indicated that they Heza communications team. A random sample of 386 had no other saving access. Ninety-two percent of all Ejo Heza subscribers was surveyed, out of whom half respondents said they would like partial access to their were females, half percent were below the age of 40, savings in Ejo Heza. These results point to a significant and half had an active contribution status in the last six proportion of members, irrespective of alternate sav- months while the rest had not contributed (dormant) ings, wanting partial access from Ejo Heza. A promising in the last six months. The findings from the phone finding is that subscribers do not wish to withdraw all survey of Ejo Heza members showcase respondents’ funds. When questioned on how much access to socioeconomic characteristics and preferences. The funds would be desirable, 33 percent of respondents delivery chain framework developed by the World said they would like to access less than 30 percent of Bank (Lindert et al. 2020) allows an assessment of pro- their savings in Ejo Heza; 52 percent said they would grams and schemes from Outreach to the Exit stage. like between 30-50 percent fund access and less than 1 This framework was reimagined for a social insurance percent would like full access. Based on the findings scheme to analyze Ejo Heza and desk research; inter- and in recognition of the liquidity needs of the target views and field visits were used to complete the group, Ejo Heza could consider a side-car account x  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans structure with up to 30 percent or 40 percent in a ID coverage, digital payments, and growing interoper- short-term liquid account and the rest in a long-term ability across different government systems in Rwanda saving account that can only be accessed once savings will, in the near future, make it possible for the govern- reach the target of RF 4 million or when an individual ment to divert collected VAT revenues directly into reaches age 55. individual Ejo Heza accounts. About 60 percent of re- spondents in the survey said that they would most (3) Increasing the minimum savings level to be likely save if the government automatically contrib- eligible for government incentives. uted some tax revenue into their account; 27 percent The high coverage growth rates that Ejo Heza has said they might continue to save, and 4 percent said achieved are only sobered by the finding that despite they might not save. Such a move would be ingenious the incentives, average savings have remained low. Low and bold, but consultations are warranted to test the savings in the contribution phase will translate to feasibility of this proposal. Another possibility for ex- lower pensions on retirement. Individuals can save panding coverage could be to mandate that the more more or more often to improve pension adequacy, or well-off individuals (in categories U3 and U4) contrib- the scheme can earn higher returns. Currently, govern- ute to Ejo Heza. Surprisingly there appears to be sup- ment incentive is offered when individuals meet a port for such a move among phone survey respon- minimum saving threshold. The threshold varies de- dents (who one should note are registered subscribers). pending on the household’s poverty status (Ubudehe About 68 percent of respondents said that the gov- category) to whom the individual belongs. It is hypoth- ernment should mandate Ejo Heza coverage for indi- esized that increasing the threshold for the incentive viduals belonging to Ubudehe U3 and U4 (the more might encourage individuals to save more and, in turn, well-off income categories). improve their adequacy. The phone survey results found that 70 percent reported they would save more (4) Age restrictions in joining Ejo Heza or accessing if the threshold were increased, while 30 percent ex- insurance benefits. pressed an inability to save more. Financial inclusion Ejo Heza is an age-inclusive scheme that allows chil- literature points to optimism in individuals’ reported dren and older adults to save. While this is a policy ability to save, and most current subscribers cannot choice for Rwanda, to not exclude anyone from an op- benefit from the matching. So while some individuals portunity to save, it comes with its challenges. Indi- might be able to meet the higher threshold, it is likely viduals who start saving at older ages (for example, age that those with lower and unpredictable incomes will 52) are allowed to withdraw at 55, but their savings need more generous matching by the government to would not have grown by much. There is a risk that meet those thresholds. these individuals would have unmet expectations from the scheme. If the age distribution of Ejo Heza (2) Scaling up coverage through mandates or iden- gets skewed towards older adults (maybe because tifying aggregators. they are more worried about older age), then the Universal pension coverage remains a bold aspiration scheme will need to adjust its investment policy and for the Ejo Heza scheme. So far, working with coopera- keep cash in hand to be able to make payments. This tives to enroll their members has proved to be a suc- will hurt the scheme’s ability to invest in long-term se- cessful modus operandi. Since cooperatives in Rwanda curities that provide higher and more stable returns. observe their members’ incomes and can auto-deduct Without age restrictions, a private insurer’s premium their earnings, they can take on a similar role as em- for life insurance is high because they need to consider ployers in the formal sector. To scale up coverage to adverse selection in their pricing. By restricting the age include all uncovered Rwandans, the scheme will need for eligibility for the life insurance benefit, RSSB and to look for other aggregators, ideally, one that has MINECOFIN can negotiate globally competitive rates oversight over the entire population. The only such ag- on life insurance under Ejo Heza. The government can gregator is a tax authority, and consumption (VAT) is retain the age-inclusive characteristic of Ejo Heza the only tax that all Rwandans incur, irrespective of whereby an individual of any age can save, but the life employment or income type. The sweeping increase in and funeral benefits are provided to savers below a Executive Summary n  xi certain age. Lump sum or pension payouts can also be unavailable to informal sector workers are easily restricted to those with a minimum of five years of understood and feel like a good ‘bargain’ for lock- membership or after reaching savings of RF 4 million ing away their money. (whichever occurs sooner). 7. Balance flexibility in design: This would meet the needs of informal sector workers without acutely compromising the objective of retirement security. Key factors for Ejo Heza 8. Be adaptable and agile: The informal sector is dy- success and lessons for namic and has evolving needs, so a scheme needs other schemes to be designed to ‘meet the clients where they are found’. The traditional methods of social security organizations that cater to the formal sector are 1. A sound legal basis for the scheme and align‑ unlikely to be successful. ment with GoR’s strategic priorities- A key inter- vention under the Economic Pillar of the NST1 9. Leverage digital infrastructure like ID and pay‑ 2017-2024 was to operationalize a long-term sav- ment to make registration, contribution, and ings scheme and pension for all Rwandans, includ- payments easy and efficient. It also makes the ing those in the informal sector as a basis to sup- individual account portable across jobs, locations, port long-term domestic investments. Following and service providers, which is critical for informal this strategic priority the GoR, through the MINE- sector workers who are mobile and would like to COFIN established EjoHeza Long Term Savings be assured that their modest savings are safe Scheme under Law N° 29/2017 of 29th June 2017. 10. Explore partnerships at the sub-national and civil society level: Ejo Heza’s partnership with 2. Set targets at sub-national level and recognize mayors, CARE VSLA, Rwanda Cooperative Author- those who meet them: This helps with increased ity, and volunteer village-level mobilizers help ownership, higher accountability and leads to in- with mobilization efforts. novative campaigns by district level staff who want to be recognized. The targets themselves 11. Ensure clear, consistent, and intuitive communi‑ need to be well designed and the central office cation: This can be done through leaflets, video, needs to support districts e.g., through a desig- radio messages, call centers, and district coordinators nated District Coordinator in Ejo Heza. who are knowledgeable enough to answer ques- tions and are committed to the scheme’s success 3. Build trust in the institution and political will: (through monetary or non-monetary incentives). The Government of Rwanda and RSSB enjoy their people’s widespread trust, which is a key advan- tage. In countries that might not have that, efforts World Bank policy need to be made at the policy and program level to establish trust and maintain it. recommendations based on 4. Identify aggregators to onboard subscribers: Aggregators who can ‘observe and auto deduct’, analysis in the report like cooperatives, can serve as an effective substi- The Ejo Heza scheme has become a success story tute to employers for those in the informal sector. globally and is a signal to countries that voluntary cov- erage expansion can be a reality for low and emerging 5. Offer fiscal incentives: Especially in the initial economies. The engagement’s findings show that Ejo stages, when the scheme is likely to have opera- Heza must also focus on improving adequacy, scheme tional teething troubles, a monetary benefit is ex- viability, and customer experience while expanding pected to get more people interested in joining. coverage and making operations smoother. Accord- 6. Offer a tangible benefit for the short run: Ben- ingly, the report concludes by providing recommenda- efits such as life and funeral insurance, currently tions grouped under seven thematic areas. xii  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Recommendation 1: Communications: v. Explore the potential for ‘auto deduction’ in part- nership with Mobile Money operators and Banks. Keep doing what’s working, fix the gaps, and explore new avenues. vi. Promote the registration of spouses and chil‑ dren of existing members. i. Reshape the narrative on why people should vii. Improve system readiness to process withdraw‑ save in Ejo Heza—steering away from incentives als, pension payouts and evaluate their cost/in- to highlighting prospects of a pension, and com- vestment implications. petitive and safe returns ii. Recognize and develop different targeting and communication strategies for urban/rural; Recommendation 3: Revisit incentives USSD/smartphone users; women/men, individu- as ‘experience’ on the scheme develops als with/without phones. i. Assess (empirically) whether RF 4 million is a iii. Create short (<1 minute) ‘how to’ videos that hard-to-meet target for most Rwandans. can be shared on WhatsApp/YouTube and show ii. Renegotiate insurance premiums after imposing how to register, contribute, check balance, and age restriction on eligibility get enrollment information. iii. Consider incentives for agents/intermediaries iv. Build a community of mobilizers (paid or vol‑ based on a cost-benefit analysis and experience unteers) and make them feel part of one ‘Ejo from countries. Heza family’ by holding quarterly meetings and awarding tokens of appreciation. iv. Pilot other short-term incentives to members, for example, accident insurance, maternity ben- v. Engage with other aggregators who might have efits to savers in Ejo Heza, or points in a grocery an ‘observe and auto-deduct’ capability espe- store/mobile money top-ups/chance to win a cially in urban areas. lottery if one saves persistently. vi. Adopt a ‘Tech with touch’ motto to avoid ex- cluding those with limited digital access (women, children, low-income members) and in vulnerable Recommendation 4: Improve situations (after death of a member or disabled). interoperability with other vii. Create ‘digital contracts’ using information col‑ government systems and integrate lected at the registration stage that can be ac- with the private sector cessed by members with a #text and can serve as i. Ensure real time updates on the Ubudehe cat‑ proof of membership. egory of members by registering on the ‘messag- ing queue’ of LODA MEIS. Recommendation 2: Focus on ii. Integrate Ejo Heza platform with banks, SAC‑ improving persistency and preparing COs, mobile money operators so that they can for withdrawals/payouts be intermediaries (similar to cooperatives) who register and collect contributions. i. Revisit targets for districts using objective cri‑ iii. Integrate with IREMBO so individuals can submit teria to reflect socio-economic differences by claims/check balance. district. iv. Integrate the system with RCA to allow sharing ii. Enhanced efforts to target urban dwellers of ‘real time’ information on cooperatives. iii. Assess and manage members’ expectations on v. Integrate with the insurance benefits provider expected fund balance in old age, given their sav- (Sonarwa currently) so that Ejo Heza teams can ing levels. see the application status and reason for delay. iv. Highlight the competitive and stable returns in vi. Review and ensure that data privacy and secu‑ the communication material, and in-person visits. rity controls are in place for registration through intermediaries. Executive Summary n  xiii Recommendation 5: Strengthen vi. Hire a PR firm to work with the communica‑ tions team to craft tailored and creative mes‑ learning, monitoring, and evaluation sages targeting children, urban workers, i. Generate quarterly reports on monitoring and women, and youth (age 16–30). evaluation (M&E) indicators for management. vii. Hire a statistical expert to analyze data, design ii. Redo the World Bank SVAT analysis based on research questions, and monitor call center new data and consider withdrawal patterns. staff during a survey iii. Organize study tours with countries that have relevant experience to share. Recommendation 7: Collect better iv. Engage with researchers to carry out Impact data and mine it for policy making Evaluations. i. Build a more extensive profile of members by cross-referencing data from other government Recommendation 6: Build human databases as relevant. resource capacity and leverage ii. Add a question at registration, such as ‘Who external expertise as needed helped you register?’ . i. Evaluate the human resource policy for Ejo iii. Collect data on ‘occupation’ at the time of reg‑ Heza team members with attention to staff istration and use the same categories as Rwan‑ shortage, capacity constraints, and short-term da’s household survey data. nature of contracts. iv. Analyze panel data of participants to uncover ii. Recognize and reward better HQ, district, and trends over time. sector-level performers. v. Propose a module on social insurance in the next iii. Build technical capacity and review staffing of round of ECIV (HHLD) survey by collaborating call center. with the National Institute of Statistics of Rwanda. iv. Engage in regular training of agents (paid and vi. Add monitoring and evaluation indicators in volunteer). the IT dashboard of Ejo Heza . v. Adapt systems and build capacity to generate vii. Add a question on the ‘cause of death’ to build ‘intermediary logins’ at the district level. experience ratings and better price life insurance. Introduction R wanda has registered a threefold increase in per capita income in the last two decades, coupled with major progress in human development and pov‑ erty reduction. While still a low-income country, Rwanda is ahead of more than 20 countries in Sub-Saharan Africa in terms of per capita GDP. Between 2001 and 2017, poverty rates fell from 58.9 percent to 38.2 percent while extreme poverty fell from 40 percent to 16 percent (measured by the national poverty line) (World Bank 2022c). Strong governance has been a driving force behind Rwanda’s rapid growth. With substantial investments in health and education, Rwanda has made excellent progress in im- proving adult survival, reducing maternal mortality, and achieving high rates of primary school enrollment. Despite this progress, low human capital has been identified as one of the key struc‑ tural bottlenecks to achieving Rwanda‘s inclusive growth and poverty reduction agenda. The 2020 Human Capital Index (HCI), which quantifies the contribution of health and education to the productivity of the next generation of workers, estimates that a child born today in Rwanda will grow up to be an adult who is only 38 percent as productive as she would be had she achieved good education and health as defined by the HCI (World Bank 2020).1 In 2020, Rwanda experienced its first recession since 1994 due to COVID-19, which pushed a large proportion of vulnerable people into poverty. The crisis is estimated to have raised poverty in Rwanda by about five percentage points in 2021, representing over half a million additional poor people. The pandemic has affected human capital formation, poverty reduction, and employment, with a marked impact on women and girls. Employ- ment is depressed despite the strong recovery. Female-headed households were poorer and less likely to be employed in the formal sector even before the pandemic. Having disrupted progress on human capital attainment, COVID-19 impacts may continue to be felt in the medium to long term (World Bank 2022c). The Government of Rwanda successfully navigated waves of COVID-19 in 2020-21 and the country is positioned for a strong economic rebound. The Government’s Economic Recovery Plan (ERP) for May 2020 to December 2021 aimed to support vulnerable house- holds and boost employment in the wake of the pandemic. The ERP monetary value was estimated at 4.4 percent of GDP and included fiscal and monetary policy measures, con- tinuous social assistance, and support for firms. It specifically committed to scale up social safety net programs, build key infrastructure for recovery, and support strategic enter- prises to offset the crisis impact. After contracting by 3.5 percent in 2020, the economy is 1. Productivity-linked indicators included in the HCI are (i) probability of survival to age 5, (ii) expected years of schooling, (iii) harmonized test scores as a measure of the quality of learning, (iv) the adult survival rate, and (v) proportion of children not stunted. 1 2  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Social assistance schemes played a critical role in Rwanda’s response to COVID-19 and a resilient recovery would entail efforts to further strengthen social assistance, social insurance, and social protection delivery systems projected to grow by 5.1 percent in 2021, rising to 7.0 programs via schemes administered by the Rwanda and 8.1 percent in 2022 and 2023 respectively. Apart Social Security Board (RSSB). While the main objec- from base effects, the rebound is supported by a re- tive of social assistance is to lift people out of chronic covery in agriculture, industry, and services. Notwith- poverty, social insurance schemes are designed to standing these projections, the employment-to-pop- build resilience by protecting against idiosyncratic risks ulation ratio8 remains nearly 8 percentage points such as health, unemployment, and old age. In 2001, below the pre-pandemic level (World Bank 2022c). health coverage was offered through the RSSB Medical scheme to eligible public and private employees and Social assistance schemes played a critical role in pensioners.2 RSSB also manages the formal sector pen- Rwanda’s response to COVID-19 and a resilient re‑ sion scheme with the aim of improving income secu- covery would entail efforts to further strengthen rity in old age. The coverage of the health and old-age social assistance, social insurance, and social protec‑ pension scheme for the formal sector stood at 3.5 tion delivery systems. Significant pre-COVID-19 in- percent and 7 percent of the working age population vestments in its flagship safety net, the Vision 2020 respectively, in 2021 ( NISR 2020). Given the large and Umurenge program (VUP), allowed Rwanda’s social persistent informal sector, the Community Based protection system to respond immediately to the pan- Health Insurance (CBHI), also known as Mutuelle de demic crisis by scaling up safety nets, including emer- Sante, was introduced to provide health coverage to gency transfers. The social insurance schemes re- the informal sector in 2007 after eight years of pilot- sponded by waiving penalties on employers for ing. In 2015, RSSB took over management of the CBHI delayed or missed contributions. Payments to pen- program. Continuing with the focus on social insurance sioners were digitized to allow continued and safe ac- for the informal sector, the Ejo Heza Long Term Sav- cess to pension benefits. Given the sharp economic ings Scheme (Ejo Heza LTSS), a defined contribution slowdown and increase in poverty, social protection scheme, was launched in 2018 to encourage savings for programs are more important than ever. Besides sup- old age. The scheme was managed by the Ministry of porting the poor, they can help protect Rwanda’s most Finance and Economic Planning (MINECOFIN) during vulnerable citizens from falling deeper into poverty. the incubation period. In 2021, RSSB took over the They can also play a central role in helping households management of the Ejo Heza scheme, although MINE- build resilience and avoid detrimental coping strate- COFIN continues to pay for the fiscal incentives and gies such as selling productive assets or drastically re- operational costs of running the scheme. ducing food consumption ( World Bank 2022b). The health insurance scheme (CBHI) for informal In addition to social assistance programs, the Gov‑ sector workers has achieved coverage rates as high ernment of Rwanda has invested in social insurance as 86 percent of the population and is hailed as a 2. The medical scheme caters to all public employees; private sector employees of companies having a payroll of 7 or more ; as well as pensioner affiliates who previously contributed to the medical insurance scheme. The scheme also extends to the spouses and children of these affiliates. Introduction n  3 The rollout of the Ejo Heza scheme is timely given the rapid aging of the population and the large proportion of workers in the informal economy who lack retirement security success story under the Universal Health Coverage works informally in the economy as drivers, hawkers, agenda. The CBHI is a solidarity health insurance sys- artisans, and so forth, and therefore is not mandated tem where all household members pay contributions to be covered under the RSSB. This leaves these work- to access healthcare with the exception of any mem- ers and their dependents exposed to risks of poverty ber insured under other medical insurance schemes. As during old age. Recognizing these challenges, Ejo Heza of 2020-21, the Government of Rwanda achieved 85.7 was designed to account for the characteristics of the percent coverage of the population (more than 9 mil- large informal sector such as low and irregular earnings. lion beneficiaries), up from 7 percent in 2003, 74 per- Administered by the RSSB, the Ejo Heza savings cent in 2013, and 79.6 percent in 2019 (Kalisa 2015). A key scheme became operational in 2018 and registered re- contributing factor to the success of the CBHI is the markable increase in coverage. Despite the pandemic, generous subsidies from the government—notably which affected people’s incomes, Ejo Heza covers 2.1 paying premia for Ubudehe category 1 households and million individuals (as of April 2022), accounting for 27 waiving the waiting periods between contribution and percent of the population between ages 15-64.5 De- access to various health services and facilities. Other spite these steady increase in coverage, there is still factors include the standardization of premiums for room to increase coverage as well as level of savings. the poor, establishment of health centers across the country, and institutional leadership community- The rollout of the Ejo Heza scheme is timely given the based support frameworks that facilitate higher par- rapid aging of the population and the large propor‑ ticipation. The investments made under CBHI have led tion of workers in the informal economy who lack to improvements in key health indicators such as life retirement security. Rwanda is a demographically expectancy at birth, which increased from 49 years in young country with four in every five Rwandans below 2000 to 65 in 2012 and 69 years in 2020.3 Mortality rates the age of 40. Declining fertility and increases in life (under 5) decreased from 185 per 1,000 live births in expectancy, however, mean that the number of el- 2000 to 41 in 2020.4 derly people is expected to triple in the next three decades. As of 2022, 38 percent of the population is Following the success of CBHI, the Government of below age 15 (5.2 million people). By 2050, the below-15 Rwanda focused its attention on expanding pension age group will grow to 6.7 million people. The number or long-term savings coverage for the informal sec‑ of working people ages 15-59 will double during the tor. Under the Law N° 29/2017 and after a pilot study, same period—from 7.7 million in 2022 to 14.1 million by MINECOFIN established the Ejo Heza Long Term Sav- 2050. The number of people aged 60 and older will ings Scheme (LTSS)—a voluntary defined contribution grow the fastest, from 0.7 million to 2.1 million.6 This scheme designed with the needs of the informal sec- phenomenon of rapid societal aging is common in tor in mind. The majority of Rwanda’s labor force much of the developing world. Young developing works in the agricultural sector, is self-employed, or countries can mitigate risks while realizing the 3. https://data.worldbank.org/indicator/SP.DYN.LE00.IN?end=2020&locations=RW&start=2000 4. https://data.worldbank.org/indicator/SH.DYN.MORT?end=2020&locations=RW&start=2000 5. Working age population (15-64) in 2022 was 8,036,935 (UN Population Prospects, 2019). 6. https://population.un.org/wpp/Download/Standard/Population/. 4  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans This report provides an overview of the Ejo Heza scheme and suggests recommendations to improve coverage, adequacy, operational efficiency, and long-run viability of the scheme opportunities that aging presents, but it will require savings will also grow. Over time, Ejo Heza could grow timely action and a fundamental rethink of work, so- to be one of largest institutional funds in the country cial protection, care, and other policies to build, use, and could be a source for domestic resource mobiliza- and protect their human capital across the life course tion. This would require strong governance arrange- (World Bank 2022). ments and sound investment policies in place so that the fund is well-managed and provides fair return to The changing demographic structure of Rwanda re‑ the members who save with the scheme. It would also quires social protection instruments across the life help the government achieve its domestic savings rate cycle. In addition to enhancing social protection in- target of 23.9 percent by 2024 (Republic of Rwanda struments to address poverty and human capital de- 2019). Ejo Heza’s growth therefore has micro (higher velopment through working age, there is a need to financial resilience of individuals) and macro (domestic prepare for old age financial protection of current and resource mobilization) benefits. future elderly. This would require enhancing links be- tween different social protection instruments, includ- Against this background, this report aims to provide ing social assistance and social insurance, at the policy an overview of the Ejo Heza scheme and provide rec‑ and delivery levels, in addition to continuing efforts on ommendations to further improve coverage, ade‑ individual program level enhancements. To improve quacy, operational efficiency, and long-run viability income security during old age and reduce the poten- of the scheme. This report is a product of a two-year tial fiscal burden of social assistance to protect the engagement between the World Bank team and RSSB growing number of elderly from poverty, it is impor- that has spanned multiple rounds of consultations and tant to further expand coverage under Ejo Heza and interviews with stakeholders and partners of Ejo Heza; increase the level of savings. meetings with Ejo Heza core teams to understand the operational processes; field visits in 2021 once CO- As more Rwandans save with the Ejo Heza scheme, VID-19 restrictions were relaxed; adaptation of the the fund reserves of the scheme will grow and could World Bank’s Scheme Viability Assessment Tool be a source for domestic resource mobilization in (SVAT)7 tool to assess Ejo Heza’s viability; and, most the country. Public investment has been driving trans- recently, a phone survey of 386 subscribers of Ejo formation in Rwanda—growing from 5 percent of Heza. The rich set of qualitative and quantitative infor- gross domestic product (GDP) in the early 2000s to an mation collected during the engagement has been average of 15 percent in recent years (World Bank 2019). analyzed and compiled for this report. The target audi- Domestic savings rate in Rwanda has fluctuated from 6 ences are policy makers at RSSB who are looking for percent to 10 percent in recent years. Given the rela- specific recommendations to improve the scheme tively high investment rate in Rwanda and the need to (section 6) and subscriber preferences into design scale up investments in the medium to long run, a changes like short-term access to funds, mandating boost to domestic savings rate would propel Rwanda’s savings, changing eligibility rules (section 5); scheme growth. As Ejo Heza coverage expands, the pool of administrators at Ejo Heza who would like an overview 7. See Section 5 of this report for details on the tool. Introduction n  5 of the delivery chain (section 4); and global audiences Ejo Heza subscribers. Section 3 is a detailed assess- who would like to learn about the success of Ejo Heza ment of Ejo Heza’s operational processes using the and potential reasons behind it. This report is also writ- adapted (for SI) delivery chain framework discussing ten to help policy makers, practitioners, and research- elements of the scheme from outreach to exit stage. ers supporting voluntary pension schemes in other Section 4 discusses four design changes that in prin- countries gain critical insights into Ejo Heza, and to ciple can either help meet member needs (providing learn about techniques and tools that can be used to short-term access); improve adequacy (increase mini- analyze such schemes. mum saving level for matching); expand coverage (through mandates or new aggregators); or improve Section 1 presents background on Ejo Heza, notably scheme governance (through select age restrictions). the design rules and institutional ecosystem. Section Section 5 presents the SVAT analysis carried out by 2 presents scheme statistics (coverage growth, sav- the World Bank at the request of RSSB to assess the ing levels, density/distribution of savings) and socio- viability of the Ejo Heza scheme. Section 6 concludes economic highlights from the 2022 phone survey of with recommendations under seven thematic areas. SECTION 1 Background of Ejo Heza Long Term Savings Scheme “Ejo Heza is a very good scheme for our members because it will help us in old years. Government contribution and even interest on our contributions—we are very happy with them” – Head, KATECOGRO tea cooperative “Being a government program gives me assurance and safety. We finally have a pension program, and pensions are no longer reserved for formal sector” – Member, Koprike Rice Cooperative “I saw what my parents were going through because of no savings. I want to be assured of their future” – Members, CARE VSLA E jo Heza is a voluntary defined contribution scheme that offers enrolees short- term withdrawals once they meet specific conditions and lump sum and monthly/scheduled payouts on retirement. No minimum contribution is re- quired and individuals can choose how often and how much they wish to con- tribute. Ejo Heza allows short-term withdrawals if individuals have at least RF 4 million (USD 3,701) in savings. On reaching the pension eligibility age of 55, individuals can choose lump sum and/or scheduled withdrawal options for the payout phase. They can also choose to keep their savings in the account beyond the age of 55 and withdraw at a later time. Formal sector workers who make mandatory contributions to the RSSB pension scheme can also enrol in Ejo Heza to improve the adequacy of their retirement income. Rwandan citizens with a national ID or foreigners residing in Rwanda can participate in the Ejo Heza scheme, but the scheme is explicitly designed to extend pension coverage to the informal sector. A unique feature of the scheme is that even children can enroll in it. Children under the age of 16 can enroll using the temporary ID issued to them by the National Identity Agency (NIDA) and their Ejo Heza account is linked to their parents’ na- tional ID number. At age 16, once children are issued their national ID, the Ejo Heza contri- butions will continue in their own names.8 By allowing children to have accounts in Ejo Heza, administrators have increased the probability that even small contributions in the early years will lead to substantial savings both during the contributors’ youth and when the contributors become old. MINECOFIN provides fiscal support to Ejo Heza by financing the operational costs of the scheme and matching contributions and insurance premiums for members who meet the eligibility conditions. Following the success of CBHI in expanding health 8. The last four figures of a child’s temporary ID in Rwanda are the same as the ID of the parent. 6 Section 1: Background of Ejo Heza Long Term Savings Scheme n  7 TABLE 1:  Program rules under Ejo Heza as of 2022 Contribution Rules Ubudehe category Annual saving in Ejo Heza Government Co-Contribution (in year Life insurance Funeral insurance of contribution) (available next year) (available next year) Ubudehe RF 15,000 or more 100% matching amount (max of RF RF 1,000,000 RF 250,000 Cat. 1 & 2 18,000) Ubudehe Cat. 3 RF 18,000 or more 50% matching amount (max of RF RF 1,000,000 RF 250,000 18,000) Ubudehe Cat. 4 RF 72,000 or more Nil RF 1,000,000 RF 250,000 Withdrawal rules If savings >RF 4 million Up to 40% in excess of RF 4 million can be withdrawn for housing or education expenses Full withdrawal Allowed at any age in case of permanent disability, terminal illness, or emigration of a non-Rwandan Payout rules on reaching pension eligibility age of 55 If fund balance @55 < RF 4 million Lump sum allowance or scheduled withdrawal based on frequency chosen by member If fund balance @55 > RF 4 million Up to 25% of fund balance as lump sum + monthly pension calculated as remaining fund balance divided by 240 Source: insurance coverage to the informal sector, MINECOFIN scheduled to continue for 3 years from December 2018 piloted the Ejo Heza scheme. Two years after imple- to December 2021, the process to extend it is ongoing. mentation, the scheme was passed on to RSSB to ad- Members who meet the annual saving minimum for minister along with the other social insurance schemes their respective Ubudehe category are automatically in the country. Ejo Heza is the youngest scheme under eligible for an RF 1,000,000 life insurance cover and an RSSB and its operational expenses (staff costs, admin- RF 250,000 funeral insurance cover in the year follow- istrative costs related to the scheme, and fund ex- ing their minimum annual contribution (Figure 1). This penses) continue to be financed by MINECOFIN. To benefit is also available to those in Ubudehe category attract individuals to save in the scheme the Govern- 4 and the insurance premium on behalf of eligible ment of Rwanda offers targeted9 matching contribu- members is currently paid for by MINECOFIN. tions (known as co-contributions) and life and funeral benefits to those who meet a minimum savings level The partial access to funds allowed in Ejo Heza aims (see Table 1 for rules). These fiscal incentives are de- to balance the objective of retirement security for signed to motivate individuals to save an annual mini- informal sector workers with their need for liquid‑ mum. Individuals who belong to lower-income house- ity. However, the average savings of members suggest holds (based on the community-based targeting that meeting the target of RF 4 million might be diffi- system ‘Ubudehe’ which is used for the social safety cult for most, unless members start saving more. net system10) receive a higher matching contribution. Short-term accessibility to funds is an increasingly Matching contributions are transferred directly by common feature in pension schemes for informal sec- MINECOFIN on a quarterly basis to the Ejo Heza ac- tor workers (Guven et al. 2021) to meet their liquidity counts of eligible members. The matching was initially needs and make a voluntary scheme attractive. Ejo 9. Depending on the Ubudehe category of the household. 10. To identify beneficiaries for social safety nets and other pro-poor social sector programs, Rwanda has been using a commu- nity-based targeting system known as “Ubudehe” since 2002. The Ubudehe system uses a few, mostly subjective, variables for communities to classify households into four categories, numbered 1 to 4. Most pro-poor programs benefit households in Ubudehe category 1 (poorest) with some also reaching households in category 2. 8  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans FIGURE 1.  Family members of eligible Ejo Heza members receiving insurance compensation Source: Press release in Kagilito on 29-10-2021.. Ejo Heza is built on a strong digital platform interoperable with the government national ID and the social safety net databases Heza restricts short-term access for education, and Vietnam Social Security), Ejo Heza stands out as a housing needs. Even in these instances, members can unique case whose ecosystem is based on a central only access up to 40 percent in excess of RF 4 million digital platform but with multi-stakeholder and strong of savings. Members can also use the 40 percent in grassroots engagement (Figure 2). This approach is excess of RF 4 million amount as collateral toward a largely responsible for the resounding success the personal loan. While these are novel features, the aver- scheme has had in covering 27 percent of the working age annual savings of RF 13,000 as of 2021 suggest that age population (age 15-64) in the four years since its it will take individuals at least 37 years of saving to inception. China’s RURS scheme, which covers 38 per- meet the RF 4 million target, assuming the scheme cent of the working age population, is the only other earns a 10 percent nominal return each year.11 scheme that has achieved coverage at a scale similar to that of Ejo Heza.12 On average, voluntary schemes, ad- Ejo Heza is built on a strong digital platform in‑ ministered by a government entity (China, India, Thai- teroperable with the government national ID and land, Vietnam, Colombia), sustain themselves with or the social safety net databases. Its ecosystem in- without innovating. However, schemes led by the pri- cludes a range of private and civil society actors. vate sector (MBAO in Kenya, schemes in Ghana) face Among the schemes for the informal sector that exist financial challenges sooner if the coverage increase is globally (such as China’s Rural and Urban Resident not as rapid as envisioned. Scheme (RURS), India’s Atal Pension Yojana (APY)/Na- tional Pension Scheme (NPS), Colombia’s Solidarity RSSB administrates Ejo Heza while the Rwanda Na‑ Pension Fund, and Vietnam’s voluntary scheme in tional Investment Trust (RNIT) serves as the 11. The annual return credited to the individual accounts of Ejo Heza members in 2021 was 9 percent, hence this assumption. It should be noted though that the investment earnings of the RSSB formal sector pension scheme have been lower than 10 percent, historically speaking. 12. See Jain and Palacios (2021) for scheme details of China’s RURS. Section 1: Background of Ejo Heza Long Term Savings Scheme n  9 FIGURE 2.  Ecosystem map of Ejo Heza Ecosystem Diagram National Bank of Rwanda MINECOFIN Steering Committee (NBR) Rwanda Social Security Board CUSTOMER CHANNELS Registrations Mobile Administrator Operators Existing NIDA (National ID) Central IT CARE Int + VSLA Proposed Direct Channel LODA (Local Custodian Bank Fund Manager (RSSB)+ Agents supervision) KCB Bank Rwanda National Rwanda Investment Trust RSSB – Digital Direct Channels Contributions Payouts/ Transactions and Service Banks Cooperatives PSP/Banks enabling Training, Monitoring (SACCO etc) payments via National Switch (RIPPS) Community/Faith based DEATH AND FUNERAL MFI CREDIT SERVICE CHANNELS BENEFITS Various NBR IREMBO approved Financial Helpline/Call Implementation LIFE INSURER Institutions center partners Source: RSSB Ejo Heza team. RSSB administrates Ejo Heza while RNIT serves as the investment manager and the KCB Bank serves as the custodian bank for the scheme investment manager and the KCB Bank serves as the The scheme relies on the NIDA database to verify custodian bank for the scheme. As scheme adminis- the identity of subscribers and on the database of trator, RSSB is responsible for managing the operations the Local Administrative Development Agency of Ejo Heza—maintaining records, ensuring payment (LODA) to retrieve the Ubudehe category of house‑ of life insurance and funeral benefits, and compliance holds. Interoperability with government databases al- and implementation of Ejo Heza with applicable laws. lows Ejo Heza to provide a seamless customer experi- RNIT is responsible for managing and overseeing the ence for members, ensuring that registration takes investment policy operations of Ejo Heza in accor- only a few minutes. Based on the results of the phone dance with the law and investing the assets in accor- survey, partnerships with other public, private, and dance with the scheme’s investment policy. KCB, as civil society platforms can be explored in the future. the custodian bank, is responsible for safekeeping of For example, a partnership could be possible with the scheme’s assets, receiving and accurately recording IREMBO,13 the primary government-to-person (G2P) contributions within the timeframes set forth in its channel, which could allow people to save in Ejo Heza custodian agreement, and reporting to the trustees if when they visit IREMBO for other purposes. Another any transactions could jeopardize the interests of the channel to explore is partnership with mobile money scheme. operators and banks, as a large majority of members 13. IREMBO is a one stop service portal of where individual can request different government services online. This can be done via IREMBO web page (https://irembo.gov.rw/home/citizen/all_services) or by walking up to the IREMBO agents who are com- monly found on city streets. 10  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans FIGURE 3.  Ejo Heza member experience with existing and potential intermediaries District Coordinators CARE agents IREMBO Mobile money Banks 0 50 100 150 200 Not Good Good Excellent Worse Average Source: Phone survey results of a random sample of 386 Ejo Heza members (2022). report having a good and excellent experience with place. In an interview, Mr. Nzirabatinya, Vice Mayor, these institutions (Figure 3). Kirehe district, noted that the targets for each district are decided in a participatory fashion and while there Partnerships with local officials at the district level is no empirical methodology used for setting targets, and with aggregators (like cooperatives) have al‑ the officials and Ejo Heza team account for district lowed Ejo Heza to have a grassroots presence while characteristics and ability to save when setting targets. still benefitting from the fiscal support and scalabil‑ However, the limited variation in contribution collec- ity that a national scheme offers. As of 2022, 88 per- tion targets across districts (see Annex 4) suggest that cent of cumulative subscribers have enrolled via an there is scope to use objective methods to calculate intermediary. These intermediaries include coopera- these targets. The Ejo Heza communications team re- tives or local government institutions at the district vealed that at the start of the fiscal year they make levels. The Imihigo14 contract which the political lead- presentations to districts where they not only propose ers, Ejo Heza team, and districts agree on each year are targets but also include steps on how the district can an effective way to decentralize targets and increase achieve them (for example, by making a list of coop- ownership all the way to district and cell level. The tar- eratives16 the district can approach). The targets are set gets under the Imihigo contract are simple but the two for each month and once every two weeks the Ejo indicators15 ensure that besides registering new mem- Heza team at Kigali (HQ) meets with the district team bers, strategies to attract new members and encour- to review progress. age them to save more and persistently need to be in 14. Since 2006, Imihigo’s have been a tool for accelerating the development outcomes that have been envisaged in both the long and medium-term development ambitions. Imihigo have been a reliable tool for recalibrating performance at the central and district levels. Source: Imihigo reports | National Institute of Statistics Rwanda 15. The targets are number of new savers and amount of total savings (of existing and new subscribers). See Annex 4 for targets by district in year 2021-22 16. This data is available with the Rwanda Cooperative Agency (RCA). Currently the Ejo Heza team requests data from RCA but going forward interoperability between RCA and Ejo Heza systems can be explored, similar to how Ejo Heza and LODA systems interact. SECTION 2 Statistics of Ejo Heza Long Term Savings Scheme and Highlights from the Phone Survey E jo Heza was implemented in 2019; in less than four years the scheme has cov‑ ered 27 percent of the working age population, many of whom are from low- income households. The administrative data of Ejo Heza showed, at the end of 2021, a total of 1.85 million registered subscribers, 78 percent of whom contrib- uted to the scheme actively17. About 49 percent of the subscribers in Ejo Heza are females while female account for only 32 percent of total contributors in the RSSB formal sector scheme. Informal sector workers comprised of 87 percent of savers while 12 percent were from the formal sector. The numbers as of Dec 2022 have already gone up by over 58 percent, from 1.85 million subscribers to 2.87 million. In 2022 the World Bank team in collaboration with Ejo Heza carried out a phone survey of 386 Ejo Heza subscribers (see Annex 1 for sampling strategy details). A stratified random sampling was done from among the 2.1 million subscribers based on gender (50 percent male/50 percent female), age group (50 percent below 40/50 percent above 40) and con- tribution status (active in last six months/dormant in last six months18). The findings from the phone survey of Ejo Heza members showcase the socioeconomic characteristics and preferences of respondents: • About 34 percent of all respondents are farmers, 20 percent are daily wagers, and 29 percent are paid employees (Figure 4). • Females make up 48.4 percent of all respondents (enforced by sampling design) and are not concentrated in any particular employment category. • The average monthly household income among respondents is reported to be RF 88,000. About 69 percent of members reported a monthly household income of under RF 50,000 (US$50), 14 percent between RF 50,000-100,000, and 9 percent between RF 100,000-200,000. Paid employees on average make more than the median household income (Figure 5). • The average monthly saving of the household is reported to be RF 10,893 and the me- dian savings is RF 2,500. 25 percent of individuals reported no monthly saving. 17. ‘Active saver’, in this report, refers to someone who made at least one contribution in the last cal- endar year (January-December). 18. ‘Active’ means at least one contribution between July-December, 2021; ‘dormant’ means no contri- bution between July-Dec 1st 2021. 11 12  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans FIGURE 4.  Employment type and gender subscribers FIGURE 5.  Household income by employment type 40 Unemployed 30 Self-employed Percent Paid employee 20 Home-based worker 10 Farmer 0 Daily wager ee ed ed er er r ke ag rm oy oy oy or w Fa pl pl pl w ly em em em 0-50,000 Rwf 50,000-100,000 Rwf ed ai as lf- Un D id Se b Pa 100,000-200,000 Rwf 200,000-400,000 Rwf e- om H Female Male 400,000- 600,000 Rwf 800,000- 1,000,000 Rwf Source: Phone survey results, 2022. Source: Phone survey results, 2022. – Among those who save, 46 percent save with (6 percent); and “already save in too many places” (7 neighbors, 11 percent with workmates, and 16 percent). percent with cooperatives. • About 24 percent of members did not have a plan – Aside from Ejo Heza, common channels used for financial survival in old age, 19 percent felt that for savings are cash loaned to family/friends (37 they would need to work in their old age, about 36 percent); mobile money (14 percent); personal percent made some plans such as having their chil- bank account (13 percent), a saving scheme (16 dren support them or using other assets/savings, percent). Only 1 percent keep cash at home while 15 percent knew that they needed to save which highlights that when individuals have sav- more (Figure 6). ings, they are sufficiently financially aware to • Questions related to immediate versus long-term either invest or loan it to family/friends. savings revealed that members of Ejo Heza under- • Approximately 80 percent of members responded stood why saving and planning for long-term needs “No” when asked if they “contribute any amount was important (Figure 7). to Ejo Heza from their savings”.19 When asked an The scheme has witnessed an exponential growth in alternate question –“How much have you contrib- subscribers and a steady improvement in the num‑ uted to Ejo Heza’’—12 percent said “no savings”, ber of active savers over time. The number of sub- and 30 percent said they “don’t remember”. Out of scribers has been growing (Figure 8) despite COVID-19 the 42 percent of members who reported Ejo Heza which significantly affected the lives and livelihoods of savings the average contributions to date were RF those in the informal sector. As of 2021, females made 13,455 and median savings were RF 14,300. These up 49 percent of total subscribers. Of the 1.85 million numbers are close to the average savings reported subscribers in 2021, 76.1 percent or 1.41 million were ac- by the administrative data. tively saving, up from 70.6 percent in 2020, and 39 per- • Reasons for not saving in Ejo Heza included “I don’t cent in 2019. In a 2021 survey caried out by RSSB, the have any more money to save” (about 78 percent); main reasons for not saving were liquidity needs and “don’t see the benefit of saving in Ejo Heza” emergencies. About 22.1 percent of members 19. The sample selection was such that 50 percent of members were those who contributed in July-December 2021 and 50 percent were dormant during that period. A higher percentage of inactive than the stratification strategy would suggest could mean that individuals do not remember contributing. The Ejo Heza team can cross-check against their latest administrative records to test if this is the case. Section 2: Statistics of Ejo Heza Long Term Savings Scheme and Highlights from the Phone Survey n  13 FIGURE 6.  What is your plan for financial survival in FIGURE 7.  How strongly do members feel about old age saving for old age 80 I have no plan 24% 70 I will continue working until I die 19% 60 50 I have other plans/savings that will 18% help me Percent 40 I have to save more now so that i 15% 30 can help myself in my old age I have assets that I will sell if I need 20 12% to (house, cattle, jewelry) 10 My children will support me 6% 0 It is wiser to spend on It is better to prioritize Saving for old age is immediate needs than retirement saving over important to me I have to find more assets witch will to save for future immediate spending 6% help me in financial survival Strongly Agree Agree Neutral Disagree Strongly Disagree Source: Phone survey results, 2022. Source: Phone survey results, 2022. Ejo Heza has witnessed an exponential growth in subscribers and a steady improvement in the number of active savers over time FIGURE 8.  Ejo Heza’s subscriber and active saver growth over time 2,500,000 2,174,408 1,854,366 2,000,000 Number of subscribers 1,500,000 1,127,622 874,197 1,000,000 337,971 500,000 37,287 826,319 0 2018 2019 2020 2021 As of April 2022 Male active savers Female active saver Total subscribers (savers+dormant) Source: Phone survey results, 2022. Note: 2022 data is not for the complete year at of time of writing this report. 14  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans “couldn’t/didn’t know how to save,” 15 percent were minimum required to avail of the government “not interested in saving after opening an Ejo Heza ac- matching or life/funeral insurance benefit. At pres- count”, and 9.3 percent didn’t save because of unem- ent, the average saving in Ejo Heza is equivalent to ployment due to COVID-19 (RSSB 2021). about two percent of the income per capita. In com- parison, formal sector workers are required to contrib- Females save as much as males and are more likely to ute six percent of their income to the RSSB pension be active savers than males (Figure 9). The gender bal- scheme. Although informal sector workers make up 87 ance observed in the subscriber count is also observed percent of subscribers to Ejo Heza, their annual aver- in the saving amounts. Women on average contributed age savings were only RF 10,195 as of 2021, a third of the RF 8,740 in 2021 while men contributed RF 8,696 average savings of those from the formal sector (RF (US$8.4). The savings were higher in 2020 (male average 31,831). According to data received from RSSB, only was RF10,606; female average was RF 9,454) and the 203,677 members received matching contributions in drop in 2021 could be attributed to the impact that the 2020, up from only 22,247 members in 2019. This phe- pandemic had on incomes. In terms of persistency of nomenon of few people qualifying for matching is an saving, women outperform men. In 2022, the ratio of area which needs more scrutiny. The consultations and active females versus total female subscribers was 83 results of phone survey point to high levels of awareness percent, while the same ratio for men was at 78 per- and attractiveness of the matching contributions and life cent. Persistency of savings on average was higher for insurance. A hypothesis is that the COVID-19 pandemic women in Eastern, Northern, Southern, and Western might have caused people to not meet the minimum provinces and in 22 out of 30 districts overall. Surpris- eligibility for matching, even though they knew about ingly, the persistency of savings is extremely low in Ki- it and would like to benefit from it. Results from the gali for both male and female subscribers. phone survey (2022) also confirm that savings of most members are lower than the minimum eligibility require- The administrative data showed that the average ments. Those with a higher household income are likely annual saving per participant was RF 8,718 in 2021 to save more. Paid employees have higher savings and RF 10,030 in 2020, making both lower than the given the relative stability of their incomes (Table 2). Females save as much as males and are more likely to be active savers than males FIGURE 9.  Active male and female savers by province and overall 100 89% 89% 92% 90% 86% 86% 85% 80 79% 77% 80% 58% 60 52% Percent 40 35% 29% 20 0 Diaspora Eastern Kigali Northern Southern Western Overall Active male savers Active female savers Source: Phone survey results, 2022. Section 2: Statistics of Ejo Heza Long Term Savings Scheme and Highlights from the Phone Survey n  15 TABLE 2:  Reported savings in Ejo Heza, by employment and household income Household income/average Home-based Paid Self- savings in Ejo Heza to date (in RF ) Daily Wager Farmer worker employee employed Unemployed RF 0-50,000 RF 10,488 RF 12,940 RF 15,083 RF 19,396 RF 11,958 RF 6,730 RF 50,000-100,00 RF 9,000 RF 9,429 RF 24,968 RF 15,857 RF 100,000-200,000 RF 18,000 RF 25,852 RF 6,000 RF 200,000-400,000 RF 19,600 RF 150,000 RF 400,000- 600,000 RF 84,000 RF 72,000 RF 800,000- 1,000,000 RF 40,000 Average RF 11,503 RF 12,827 RF 15,083 RF 24,189 RF 20,604 RF 6,525 Source: Phone survey results, 2022. According to the RSSB administrative data, 82 per‑ number of uncategorized (see details in Section 3). As cent of subscribers are from households in Ubudehe a result of this work, the percentage of those missing categories 1, 2, and 3. Ubudehe category 1 comprises Ubudehe categories was brought down to 13 percent of the lowest income households and accounts for 21 in June 2022. Despite being poorer on average, those in percent of subscribers, category 2 accounts for 33 per- Ubudehe 1 saved RF 15,321, which is higher or equal to cent, and category 3 accounts for 28 percent. About 17 those in category 2 (RF 9,729) and category 3 (RF 15,354) percent of subscribers did not have an Ubudehe cat- but significantly lower than those in category 4 egory and classified as “uncategorized.” The Ejo Heza (RF 89,740) (Figure 10). team has been working with LODA to reduce the According to the RSSB administrative data, 82 percent of subscribers are from households in Ubudehe categories 1, 2, and 3 FIGURE 10.  Count of subscribers and average saving by Ubudehe category 700,000 100,000 603,014 90,000 600,000 80,000 504,933 Average savings (in Rwf) 500,000 70,000 390,506 60,000 400,000 Count 316,039 50,000 300,000 40,000 200,000 30,000 20,000 100,000 10,000 1,816 0 0 Cat 1 Cat 2 Cat 3 Cat 4 Uncategorized Source: Phone survey results, 2022. 16  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans The number of subscribers across provinces ranges between 400,000–500,000, except for Kigali, which has a lower number of contributors and a much lower number of active savers FIGURE 11.  Count of subscribers, active savers, and average savings by province 600,000 566,493 565,402 20,000 513,053 18,000 495,165 500,000 16,000 407,173 412,396 14,000 Average saving (Rwf) 400,000 364,504 343,026 12,000 309,146 Count 300,000 10,000 8,000 200,000 6,000 101,009 4,000 100,000 2,000 0 0 Eastern Kigali Northern Southern Western Subscribers Active savers Average savings Source: Phone survey results, 2022. If the low income meets the minimum saving re‑ withdrawals, and collateral on loans that are allowed quired to be eligible for the government’s matching once individuals accumulate RF 4 million. contribution, they will reach the RF 4 million target for short-term withdrawals a decade earlier than The number of subscribers across provinces ranges those who fail to meet the minimum. Calculations between 400,000-500,000, except for Kigali, which suggest that an individual who saves RF 13,000 annually has a lower number of contributors and a much will take 38 years to meet the RF 4 million target, as- lower number of active savers. Kigali stands out as a suming a 10 percent return on savings from Ejo Heza. unique case where savings on average are higher, but Alternatively, if an individual from category 1 or 2 saves the count of subscribers is much lower relative to the the minimum RF 15,000 annually, they will receive a 100 population of Kigali (Figure 11). The number of active percent matching contribution from the government savers in Kigali is also unusually low relative to other and, assuming the matching contribution continues, provinces, raising concern about the persistency of they will reach the RF 4 million target in 28 years.20 Low savings. The unexpectedly low coverage rates in Kigali savings is a cause of concern because it not only leaves could be an unintended effect of Ejo Heza’s focus on contributors ill-prepared for old age but also reduces cooperatives as the aggregator. The nature of coop- the ability of contributors to take advantage of the eratives is that they are more likely to be active in rural government matching contribution, short-term or peri-urban areas. If so, this points to a need for 20. This calculation assumes government matching will continue indefinitely, which might not be the case, but the longer indi- viduals can benefit from the matching the more likely they are to reach the RF 4 million target. Section 2: Statistics of Ejo Heza Long Term Savings Scheme and Highlights from the Phone Survey n  17 TABLE 3:  Insurance statistics, 2021 Members covered under Life insurance 210,213 members Total premia paid by MINECOFIN to the insurance company RF 1.3 billion Claims paid as of September 2021 298 members Claims and fees due as of December 2021 884 members Premium paid per eligible member RF 18,000 Source: RSSB 2022. Only 15 percent of Ejo Heza active savers had life insurance coverage as most individuals did not meet the minimum savings required to be eligible for life and funeral insurance identifying aggregators in the urban context (for ex- contributions were about RF 19 billion, while gov‑ ample, banks in the case of India, mobile money opera- ernment matching contributions and investment tors, or transport workers as in Kenya) and a communi- return of member savings were RF 1.9 billion and RF cation strategy that will resonate with urban dwellers. 2 billion respectively. Ejo Heza savings more than doubled from 2020 to 2021 and are heavily invested in Only 15 percent of Ejo Heza active savers had life government bonds. RSSB manages assets worth RF 1.4 insurance coverage as most individuals did not meet trillion; Ejo Heza’s assets make up only 1.65 percent of the minimum savings required to be eligible for life RSSB’s total assets.21 However, Ejo Heza is growing rap- and funeral insurance. The RSSB has partnered with idly with total savings of Ejo Heza participants increas- Sonarwa Life Assurance Limited to provide life and fu- ing from RF 9.1 billion in 2020 to RF 23.2 billion in 2021. neral benefits to eligible participants of Ejo Heza. The The latest numbers, as of 30th June 2022 were at RF life insurance premia are paid by MINECOFIN to the 31,293,470,518 or ~USD 31 million. A significant propor- insurance company. Out of the 1.42 subscribers only tion of the fund balance (~87.7 percent) is invested in 210,213 were eligible for life insurance offered to its government bonds and the remaining 13 percent is members by Ejo Heza (Table 3). Despite this low cover- kept in cash or fixed deposit accounts. The investment age, MINECOFIN pays RF 1.3 billion to Sonarwa for life policy of Ejo Heza aims for stable and risk-free returns. insurance and funeral benefits, which is 6 percent of The returns are also increasingly competitive as the Ejo Heza’s assets of RF 23.2 billion as of 2021. The nominal return on investments of Ejo Heza has grown blended premium per contributing member for life in- from 1 percent in 2020 to 9 percent in 2021. The Ejo Heza surance and funeral expense coverage is RF 18,000 per pooled contribution is unitized and invested such that member, which is higher than the average RF 13,000 the daily net asset value of the portfolio informs mem- savings per member (RSSB 2022). bers of the current value of their savings. It is important to ensure that members know about the investment As of December 2021, total assets of Ejo Heza were return and can see it on their account when they login RF 23.2 billion (US$23 million) out of which member via web, app, USSD,22 or call to talk to an agent. 21. As of 2020, the RSSB investment portfolio including Ejo Heza was managed by a professional team of 47 staff members, which includes 30 support staff and 14 managers, as per data documents from RSSB. 22. USDD = Unstructured Supplementary Service Data, or Smart Code. SECTION 3 Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain D elivery systems constitute the operating environment for implementing social protection benefits and services. In recent years, delivery systems have been improved by the rapid development and deployment of three systems: (i) identification systems, (ii) social registries, and (iii) payment sys- tems. All three work beyond the social protection system, serving broader functions for governments. Each has proven critical for improving the quality of delivery within social protection programs and systems. In many countries, social protection programs are not only the biggest users of these systems but have also contributed to their widespread adoption (World Bank 2022b). How countries deliver social protection has implications for the outcomes of a social protection program. An analysis of social protection delivery systems can provide an- swers to crucial questions such as how various elements of delivery systems come to- gether to implement programs as they were intended to function, how they ensure easy access to programs, how they can be leveraged to promote better coordination and inte- gration, and how social protection programs can provide a better client experience for their intended populations. The Sourcebook on the Foundations of Social Protection Delivery Systems includes a framework for analyzing the delivery chain (Lindert et al. 2020). The delivery chain framework in the Sourcebook was reimagined for a social insur‑ ance/pension scheme for the purposes of analyzing the Ejo Heza scheme in this report. Virtually all social protection programs pass through similar implementation phases along the delivery chain. The Sourcebook is mainly focused on social assistance (non-contribu- tory programs). Using the framework developed in the Sourcebook, this section reimagines the delivery chain for social insurance (contributory programs). The most important adap- tation from social assistance programs is the inclusion of contribution collection in addi- tion to the delivery of benefits. Contributory social protection schemes must manage the contributions of plan members accurately over long periods of time. Throughout the phase of payment of contributions, the social insurance institution needs to keep accurate records and manage contribution collection. In the payout phase, the benefits corre- sponding to the contributions need to be made based on the rules of the scheme and a lump sum at retirement or a pension annuity is provided. The Delivery Chain Mapping for Ejo Heza was carried out through a mix of desk research; interviews with operational staff (notably IT, communication, call center department, and the Ministry of Local Government), agents, district coordinator, and cooperative head; and a field visit to Kirehe district.23 Where relevant, findings from the recently concluded 23. The field visit on July 13, 2022, included consultations with Vice Mayor of Kirehe District, Rice Co- operative Head, seven of its members in Kirehe District, and three Ejo Heza members who joined via CARE VSLA (one of them was a volunteer agent). 18 Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  19 SOCIAL INSURANCE SERVICE DELIVERY SYSTEM - LTSS RWANDA FIGURE 12.  Reimagined the social insurance delivery chain ASSESS ENROLL CONTRIBUTE PROVIDE MANAGE RECURRING CYCLE RECURRING CYCLE Outreach Registration Verification Onboarding Contribution Benefit Eligibility Withdrawal Benefit Payout Monitoring Exit Source: Authors adapted from Sourcebook (2020). The delivery chain framework in the Sourcebook was reimagined for a social insurance/pension scheme to analyze the Ejo Heza scheme in this report phone survey of a random sample of 384 Ejo Heza sub- scribers are also included. The reimagined delivery chain includes five main stages (Assess, Enroll, Contrib- ASSESS ute, Provide, and Manage) and each stage has two sub- steps as shown in Figure 12. Figure 13 includes a visual of the delivery chain recreated for the Ejo Heza scheme Channels Registrant and the subsequent paragraphs talk about the innova- validation for Informal tions under each stage and highlight areas of improve- Associations Self or assisted by agent ments based on the World Bank team’s assessment Resident c On demand borrowing from international best practices. Social media registration Diaspo VSLA and Care Assess International Resident Fo The first stage of the delivery chain, Assess, includes Radio & TV Using two substages—Outreach and Registration. As these Posters & National ID, are the first touchpoints that a customer has with the Banners no & Date o scheme, it is important to focus on behavioral nudges Using Real time data that confront the limited attention of the customer Community (during Outreach), and the limited ability to digest Meetings USSD NIDA complex information (during Registration). Roadshows & Mobilisation Web Portal LODA M Outreach Efforts to raise awareness, build confidence, and propagate an understanding of a pension scheme can help informal sector workers make appropriate Outreach Registration Verifica and informed decisions about participation and en‑ courage persistent savings for old age. Potential 20  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans FIGURE 13.  Social insurance delivery chain applied to Ejo Heza SOCIAL INSURANCE SERVICE DELIVERY SYSTEM - LTSS RWANDA ASSESS ENROLL CONTRIBUTE (RECURRING CYCLE) PROVIDE (RECURRING CYCLE) MANAGE Confirm Contribution channels Based on Based on Support Channels Registrant Ubedehe category Mobile money savings & Savings & Informal validation for age Greivences (UC) UC Associations Self or assisted by agent Online & mobile Resident citizens If savings > RwF 4M Address, personal banking Before age (55) On demand Capped maximum Govt. Saving > RwF 4M 40% savings as loan Compliance Social media & family details End of Withdrawal registration Cash deposits Co Contribution - RwF18K 40% Partial collateral, leaving Diaspora Period Enter withdrawals; savings in atleast RwF 4M UC 1/2 - saving > RwF15k VSLA and Care Card payments excess of Rwf 4M Data Update International Contribution Resident Foreigners 100% Co-Contribution frequency & amount At age (55) Death of Beneficiary To UC 3 - saving > RwF18k Saving < RwF 4M Claim Application Radio & TV Using Death claim If insured Occupation KCB Custodian 50% Co-Contribution Lumpsum or equated National ID, Phone RwF 1,000,000 Posters & Bank Account Installment to total Using no & Date of Birth Insurance Benefits Life Insurance claim Banners Onboarding drawdown Using Notification Invested by RNIT into for citizens (for next year) Rwf 250,000 Call center / Email Real time data from At age (55) Community UC 1/2 - saving > RwF15k Funeral Insurance Saving > RwF 4M Meetings USSD NIDA SMS Govt. bonds UC 3 - saving > RwF18k claim Roadshows & UC 4 - saving > RwF72k Up to 25% partial Online Forms Mobilisation Life + Funeral lumpsum withdrawal + Web Portal LODA MEIS email Term deposits remaining monthly Insurance pension for 240 months Walk ins Premium RwF18k/person Outreach Registration Verification Onboarding Contribution Benefit Eligibility Withdrawal Benefit Payout Monitoring Exit Source: Created by authors. Ejo Heza has adapted its communication and outreach strategy based on emerging experience from the scheme participants in a scheme would want to know the rules and isolated relative to their formal sector peers of the scheme, their various options for participation, whom they often consider privileged in their access to the channels for making contributions, the modalities government services and resources. The outreach and for the payment of the pension benefit, the age of communication approach needs to map out all poten- pension eligibility, and so on. Communicating these tial partners at different levels (national, district, village) details in a clear and consistent way helps build trust in and establish the roles and responsibilities of each. the scheme and scheme administrators. This is particu- larly important in encouraging informal sector workers Ejo Heza has adapted its communication and out‑ to save for retirement in a voluntary pension scheme reach strategy based on emerging experience from as individuals are not mandated to participate and the scheme and has identified three distinct target myopia can affect their desire to save in a long-term groups to receive tailored messages—adults with saving scheme. jobs, youth seeking jobs, and adults with children in school. Since its inception, the scheme has prioritized The success of outreach and communication strat‑ resources for outreach and communication in its bud- egy depends on close collaboration with partners in get. Its threefold strategy involves focusing on specific public, private (mobile operators, banks), and civil groups as targeted potential beneficiaries, raising society sectors, as well as identifying and forging awareness and encouraging them to participate, and new partnerships (for example, with cooperatives educating existing participants on the importance of or Village Savings and Loan Associations (VSLAs). persistent savings. There is no evidence of rigorous The lack of a clear administrative relationship between analytics that guide the targeting strategy, but the government and informal sector workers means that scheme administrators have an educated strategy on in addition to partnerships at the national level to en- identifying informal sector workers that are more or- sure scalability and cost effectiveness, grassroots mo- ganized (for example, cooperatives and associations). bilization is needed to get the buy-in of informal sec- As coverage has grown, the targeting strategy has also tor workers. Informal sector actors feel marginalized evolved. Partnering with the Ministry of Education, the Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  21 communications team focused on children ages 3-16 The main channels used for outreach are (i) radio, TV, and delivered child-friendly messages on why parents and promotional street banners; (ii) social media; (iii) should contribute to their children’s pension. As a re- VSLAs and Care International; (iv) informal sector as- sult, the Ejo Heza team is expecting about RF 5 million sociations; and (v) local community meetings, road- of savings from this age group in 2022. They had in- shows, and mobilization campaigns. tended to pilot in 10 schools but so far 81 schools have joined. Next, going after low-hanging fruit, the com- (i) Radio, TV programs, banners munications team targeted large agricultural coopera- The weekly radio programs every Thursday on Radio tives. The next step is to target non-agricultural coop- Rwanda RBA 100.7 have contributed to increased eratives. There are also ongoing discussions with the awareness for the scheme. Posters and banners with Private Sector Federation. A missing area under Ejo tailored messages (Figure 14) are among the latest de- Heza’s target population are urban dwellers. This group velopments. An RSSB 2019 survey showed that about while more geographically mobile than a cooperative 12 percent of respondents had heard about Ejo Heza is an important target group because of their higher through radio or other advertising. The numbers are likelihood to save, digital and financial savviness. similar in the recent (2022) phone survey of Ejo Heza members with about 15 percent of respondents re- The scheme relies on five channels for communica‑ portedly hearing about the scheme from the radio, 1 tion and outreach, but the interviews and surveys percent from TV, and 0 (zero) percent from social me- reveal that the local government efforts are most dia. Men were more likely (10 percent) than women (5 successful in increasing awareness of the scheme. percent) to have heard about the scheme from the FIGURE 14.  Posters and banners with tailored messages Options under USSD (top left); why children should save in line with focus on enrolling kids (top right); promotion banners on street (below left); Ejo Heza promotion by mobile network operator partners (below right) Sources: RSSB headquarters in Kigali (top left and top right); Facebook images (bottom left and bottom right). 22  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Ejo Heza relies on five channels for communication and outreach, but interviews and surveys reveal that local government efforts are most successful in increasing awareness radio. This finding reinforces the importance of pursu- could be uploaded on YouTube and shared as What- ing a mix of outreach strategies while being mindful of sApp links. The social media handles on the Ejo Heza variations in access, familiarity, and usage by gender. website were not linked to the respective social media site at the time of writing this report. The Ejo Heza IT (ii) Social media team can also consider including analytics and monitor- While Ejo Heza has posted messages on social media, ing software on the website to detect user page visits, notably on Facebook and Twitter (Figure 15), the phone location of visit, what information they are most inter- survey findings suggests that it is not the source that ested in, and track the number of new visitors, returning builds large-scale awareness. The cost-effectiveness visitors, and registrations. and reach of social media and streaming websites like YouTube and Facebook make them powerful tools for (iii) Informal sector cooperatives communicators. However, ensuring that the content is There are estimated to be around 4,000 cooperatives creative and catchy is important given the limited at- nationwide, covering about 9 million people (Box 1). tention span of social media users. Ejo Heza could con- The organized nature of these cooperatives and high sider partnering with social media influencers to dis- levels of peer-to-peer trust within these groups makes seminate information on the scheme, in public interest. them an effective aggregator in the Rwandan context. The Ejo Heza communication team could produce a Ejo Heza’s strategy of identifying aggregators who can ‘how-to’ video series in the local language featuring offer scale while still being an organized group is a bite-sized videos on how to register, contribute, check valuable lesson for countries looking to expand cover- balance, update a profile, and so forth. These videos age in voluntary schemes. A cooperative in Kirehe FIGURE 15.  Facebook post on how to access balance through the online portal Source: Ejo Heza Facebook page. Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  23 BOX 1: Cooperatives are to Ejo Heza what employers are to formal sector pension schemes Over 60 percent of current Ejo Heza members were enrolled through cooperatives. Using cooperatives as an aggregator has been key to the success of Ejo Heza as they assume a role similar to that of employers in the formal sector pension scheme. But unlike a formal sector scheme where pension contribution is mandated, the cooperative contribution is agreed upon by all its members. Cooperatives auto-debit the contribution from the harvest income of members. Members of cooperatives (for example, agriculture, fishing, farming, and services) typically have seasonal incomes so the frequency and amount of payment is decided collectively by each cooperative. The heads of the cooperatives send com- mitment forms to the members, members fill them in, and the heads collect those forms and keep them as records in their finance office. At the time of harvest, the agreed contribution is auto-deducted and deposited in a bank. Ejo Heza then credits the amount into individual accounts of cooperative members and sends a notification (via text) to all members in the cooperative. Members can check their balance any time and they are not charged for doing so. Cooperatives don’t receive any special incentives for enrolling members and have reported that many of their members are members of other savings groups and understand the importance of saving. The World Bank team spoke to the head of the Katecogro tea cooperative in Karongi district and the head of the Koprike rice cooperative in Kirehe district. Katecogro has 1,260 members, out of which 985 members are subscribers in Ejo Heza (35 percent of them are females). Koprike has 3,400 members, all of whom contributed to Ejo Heza as of July 2022 (40 percent of whom are women). Insights were also received from the Dukunde Kawa cooperative of coffee farmers which has 1,193 small-scale farmer members (24.5 percent female). All the members are active savers in Ejo Heza. Insights from Katecogro: The auto-debit feature should in principle ensure 100 percent compliance, but the cooperative head clarified that the only time members do not pay is if they don’t have a harvesting in that period. However, on average, individuals are able to get back to paying contributions in two months once their harvesting cycle is back. The cooperative members are well connected through WhatsApp groups. Their exposure to Ejo Heza is primarily through the (i) Ejo Heza District coordinator, who mobilizes members; (ii) Community Radio of Isangano that discusses Ejo Heza with local leaders; (iii) meetings with sector staff in inteko z’abaturage, umuganda, Umugoroba w’imiryango;a and (iv) and the Rwanda Radio RBA, where Ejo Heza is discussed every Thursday. When asked about potential additional features, members responded that they would like a ‘membership card’ to have a record that they can show to their children. Insights from Koprike: All cooperative members have agreed to an auto-debit of RF 9,000 after each harvest season (twice a year), for a total of RF 18,000 annually. This is the minimum annual contribution in Ejo Heza to be eligible for matching and insurance benefits and is about 10 percent of the harvest income. The cooperative head deposits the total contributions in the bank and Ejo Heza credits savings into the individual accounts of the members. Recommendations from members on improving service delivery included “getting the life insurance claims processed faster” and “resolving delays in crediting of matching contributions due to missing Ubudehe category”. Speaking about challenges, the cooperative head noted that, “Convincing people initially was hard” but, “it is a process and when some people receive money through [life and funeral] insurance then people get motivated”. Insights from Dukunde Kawa: The cooperative is divided into 31 groups that meet monthly and attend general assembly meetings two times a year. The farmers heard about Ejo Heza through the community mobilization campaigns jointly carried out by Ejo Heza agents and district authorities. Life and funeral benefits are most appreciated as most members are over age 65 and ‘pension coverage is yet to be comprehended’ by them. Common questions include “Where is the contract? Shall I cash out my savings? What are the conditions? How are old people benefiting from pension coverage? What is the difference between Ejo Heza and other pension services? How can I check my member contribution without a phone number?” Cooperatives in Rwanda have become an intermediary that has allowed the scheme to rapidly expand coverage and keep costs low by ensuring persistency in saving. However, experience of using intermediaries notably trade unions in Latin Ameri- can contexts, has displayed a need for close monitoring to ensure that the objective here remains on coverage and persis- tency and to prevent any incentives that could lead to a conflict of interest and induce sub-optimal decision for members. a. This refers to Community meetings, Community, Family Night 24  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans district deposited as much as RF 25 million last year savings scheme like Ejo Heza. However, interviews with (US$25,000) into the Ejo Heza scheme on behalf of its CARE agents reveal that VSLA participants are, in fact, members. The district coordinator of Ejo Heza ap- more likely to save in Ejo Heza as they understand the proaches cooperative heads and aims to convince benefits of saving and appreciate the benefits that Ejo them (and through heads, cooperative members) on Heza offers. The benefits of partnering with an institu- the benefits of joining the scheme. tion such as CARE that has grassroots experience, knowledge, and trust is another relevant lesson learned (iv) VSLAs and CARE International from Ejo Heza. CARE has a memorandum of understanding with Ejo Heza. CARE agents support outreach and enrollment (v) Local community meetings, roadshows, efforts for Ejo Heza by sensitizing their large networks and mobilization campaigns (Figure 16) of VSLAs to Ejo Heza. CARE’s flagship VSLA methodol- About 42 percent of respondents of a 2019 LTSS survey ogy was developed in Niger more than 25 years ago said that they were made aware of the scheme through and supports informal savings group members in build- mobilization activities and roadshows (RSSB 2019a). In ing their financial skills and assets. Women’s economic the 2022 phone survey, 63 percent of respondents (31 empowerment is a priority area for CARE’s work, and percent female, 32 percent males) reported that they financial inclusion has been identified as one of the heard about Ejo Heza through their local government. four interrelated and key pathways to achieving this Community meetings are a consistent and common goal. Given the limited incomes of informal sector feature across all districts in Rwanda. The social fabric workers, a concern is whether workers will be able to of the country lends itself to inclusive participation in participate in saving groups (like VSLAs) that meet these meetings where local leaders (mayor, vice mayor, short-term needs and also save long term under a district coordinators of RSSB) meet with members of About 42 percent of respondents of a 2019 LTSS survey said that they were made aware of the scheme through mobilization activities and roadshows FIGURE 16.  Ejo Heza roadshow (left) and member registration in field (right) Source: Twitter account of Ejo Heza. Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  25 Much of Ejo Heza’s current outreach strategy focuses on communicating the benefits of saving in Ejo Heza, incentives being offered, ways to register, and so on the community to discuss, plan, and implement their Much of Ejo Heza’s current outreach strategy fo‑ ideas. The mayor for Gakenke district, which has cuses on communicating the benefits of saving in around 88,000 subscribers and a contribution density Ejo Heza, incentives being offered, ways to register, of 91 percent, notes, “To achieve our targets and and so on. This is a good approach in the initial stages goals, GAKENKE District organizes different commu- when the aim is to get the target group interested in nity meetings within churches, through family meet- the scheme and build financial awareness. However, as ings every Tuesday afternoon, within community coverage grows, scheme administrators need to shift works (UMUGANDA), meetings within cooperatives, focus to persistency and managing expectations of etc.” A senior local government office of Kirehe dis- members. Only 6 out of 386 respondents in the 2022 trict remarked, “Mobilization is key, more important phone survey reported that they knew how much than any incentives”’ When asked about ways to mo- pension they would receive from Ejo Heza if they con- bilize cost effectively, the vice mayor of Kirehe district tinued saving at their present rate. Estimating pension noted that testimonials in community meetings were income in a defined contribution is difficult even for by far most effective. He mentioned how in commu- literate and financially aware individuals but experi- nity meetings he often asked people to check if they ence has shown that unmet or unclear expectations of could see their savings on the app; many “yes” re- pension income can lead to erosion of trust in the sponses were a good sign, and even a few “no” re- scheme when the first cohort starts retiring. sponses would dampen spirits. The vice mayor also suggested that award ceremonies could be popular. Registration Ejo Heza has a large network of paid staff and volun‑ The information provided in the outreach stage teer agents involved in the mobilization efforts at serves as the foundation for the intake and registra‑ different levels. There are three Ejo Heza staff24 in tion stages. It is important to provide a familiar and Kigali RSSB office; 30 District Coordinators (one in each easy-to-use registration channel to make the process district), 416 at Sector Level and 2148 at Cell Level, paid simple for the intended population, while ensuring by districts (who also do mobilization of other district that the systems and local field agents are in place to programs); 14,837 Village Volunteer Agents (from CARE support the registration process. Once people have VSLAs); and 14,837 Village Chiefs. Mobilization for the decided to participate, they need to know how and scheme is a key priority for District Coordinators. In- where to register and how to navigate the registration terviews with agents revealed that they would appre- process. The self-service registration model is for indi- ciate ‘training and financial incentives’ but they would viduals who wish to participate in Ejo Heza at any time also appreciate being recognized for their efforts and at their convenience. It is distinct from group reg- (through tokens like caps and t-shirts that distinguish istration or intake on specific days as part of an admin- them as agents). The large network of mobilizers helps istrator-driven mass strategy. A benefit of self-service with the grassroots outreach but Ejo Heza, at the cen- registration is that interested applicants get familiar tral level, should maintain a database of mobilizers, with the interface which might motivate them to save organize quarterly meetings, and ensure that mobiliz- more often. Conversely, if the registration process is ers feel part of one ‘Ejo Heza family.’ not intuitive or simple, this could be a barrier, 24. One of them focuses on City of Kigali, another on Northern and Western Province, and a third on the Eastern Province and Southern Province. 26  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans especially for the poor or vulnerable who are at a marketing materials and how ‘lack of communication larger risk of exclusion because of behavioral bottle- facilities (phone), lack of equipment (umbrella, boots, necks and limited financial literacy. Ejo Heza depends notebook) and COVID-19 prevention material’ were on both voluntary and group registration (especially in some of the challenges he faced. An incentive struc- case of cooperatives), which is a good mix in the initial ture for agents (cash and/or goods) should be consid- years while the registration process is being made user ered. Private entities such as banks and telcos also friendly and people get familiar with the scheme and have a strong grassroots presence and could be used processes. to assist with Ejo Heza registration, borrowing from India’s experience where banks receive incentives for Most of the current registration happens through every new person they register. the vast network of agents who engage in group and door-to-door registration, with only 25 percent of Figure 17 highlights what the self-service registra‑ subscribers to date using self-service registration. tion process looks like from the systems angle. The Most agents are unpaid or not on salaried contracts, new registrant enters their national ID, date of birth, and effective agents do not receive any specific incen- and phone number as part of the registration process tives. Agents receive initial training (through CARE or on the Ejo Heza platform. A request through the API is Ejo Heza); however, an agent from the Rulindo District sent to NIDA to validate the identity. Once established who has enrolled 364 individuals to date reported as valid, the Ejo Heza system requests information needing retraining on the topics of ‘Declaration, Re- from the LODA database and receives sociodemo- porting, and Ejo Heza scheme’ to assist with the regis- graphic data and Ubudehe categories as part of the tration process. The agent reported not having any backend processes. When promoting self-service Most of the current registration happens through the vast network of agents who engage in group and door-to-door registration diagramProcess Self Registration FIGURE 17.  Voluntary registration process New Registrant Ejo Heza system NIDA LODA Enter “National ID”, date of birth & phone no NID validation request Wrong “National ID” entry PLEASE ENTER CORRECT ID NID response, if NID invalid Enter “National ID”, date of birth& phone no NID validation request Correct “National ID” entry NID response, if NID valid “National ID” of Head of Household Request Response Head of Household details DATA FIELDS AUTO POPULATED Head of Household details First Name, Last Name, NID, Age, Category of Ubudehe First Name, Last Name, NID, Age, Household Members details Category of Ubudehe First Name, Last Name, NID, AGE, Village code/village name, Sector, Household Members details District, Province, Number of Dependents First Name, Last Name, NID, AGE, Village code/village name, Sector, District, Province, Number of Dependents Source: Authors’ depiction based on desk research and interview with LODA and Ejo Heza IT team. Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  27 registration, it is important to consider the interface central IT team at Ejo Heza. Using those credentials, and customer experience for those registering on the they can enroll subscribers. Currently, providing inter- web, smart phone, or USSD devices. Consultations and mediary credentials depends on the manpower and interviews with call center representatives suggest bandwidth at central IT and field offices; if it becomes that while the registration experience on the web and easy to generate intermediary logins, then the registra- smart phone has improved, the registration with USSD tion scale-up can be quicker, seamless, and cost-effec- devices “is confusing, requires a PIN at multiple levels, tive. Figure 18 depicts what an automated intermediary and can be frustrating for the user.” This needs to be registration process might look like for a cooperative. rectified as agriculture workers and those in the lower Verification could be done through the Rwanda Coop- Ubudehe categories are more likely to use phone reg- erative Agency (RCA), which maintains the database of istration for Ejo Heza. all cooperatives. Once verified, the cooperative would be able to enroll members in a matter of minutes. Simi- Registration is currently possible via intermediaries lar links could be explored with other aggregators like like cooperatives and district/field agents (Figure banks and Saving and Credit Cooperative Societies 18). The intermediaries express their interest in being (SACCOs). While such efforts are explored there is a part of the Ejo Heza scheme to the district coordina- need to tighten controls on data privacy and data tor and receive intermediary credentials from the security. Registration is currently possible via intermediaries like cooperatives and district/field agents FIGURE 18.  What an intermediary registration process might look like Intermediary registration 1 2 Initiate registration request on intermediary page Ejoheza IT interfaces with RCA to verify cooperative credentials Request: <> 3 4 Response: <> cooperative 5 Cooperative then proceeds to login and enroll members Rwanda Intermediary (e.g Cooperative) Ejoheza IT Cooperative Association (RCA) Source: Authors’ depiction based on desk research and interview with LODA and Ejo Heza IT team. 28  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Enroll mandating an applicant’s national identification num- ber at the time of registration. By linking the social in- The second stage of the delivery chain, Enroll, includes surance database to the ID database, other personal two substages—Verification and Onboarding. At this information stored in the ID database (name, date of stage, the individual has made a decision to join the birth, or possibly a phone number) can be requested scheme; however, the experience at the verification both for cross-checking and/or limiting data fields re- and onboarding stage can create a favorable or unfa- quested at the registration stage. For example, details vorable impression. A smooth enrollment process that of verified applicants can be pulled from other admin- meets expectations will enhance the customer experi- istrative databases to identify the residency status, ence of the applicant. citizenship status, and socioeconomic level to create a complete social profile. Interoperable data systems can allow for schemes to put together a social profile on the Verification basis of which incentives can be targeted or allow for An applicant’s identification needs to be verified pattern analysis to be conducted. Such data exchanges against other administrative systems and, in some across systems are preferably done over API integra- instances, interoperability is needed to ensure that tions to enable real time data ‘push and pull’ (two-way data can be shared seamlessly, in real time, and traffic). Additionally, sharing data real time can ensure without errors. In countries with near-universal that any changes in user data (such as socioeconomic unique IDs, as is the case of Rwanda, a social insurance classification, address, and phone numbers) can be re- system can leverage this foundational structure by flected automatically in the social insurance system. In Ejo Heza, an applicant’s verification is done through integration with NIDA database. The appli- cant may be a local Rwandan resident or a diaspora ENROLL CONTRIBUTE member (Rwandan abroad). The CYCLE) (RECURRING citizens residing ap- plicant completes the registration form by providing Confirm the mandatory identification information—date of Contribution channels Based on Registrant birth, national identification number, and mobile Ubedehe category Mobile savings money is a diaspora member, & validation for phone number. If the applicant (UC) UC sted by agent they must provide additional information such as their Online & mobile Resident citizens Address, personal demand country, city, andbanking address of residence and upload a & family details Capped maximum Govt. stration copy of their Rwandan passport. If the applicant is a Cash deposits Co Contribution - RwF18K Diaspora child age 16 or younger, he or she will use the phone Enter number and national UC 1/2 - saving > RwF15k identification number of his or Contribution Card payments Resident Foreigners frequency & amount 25 her guardian or parent (RSSB 2019b). The 100% Co-Contribution applicant identification form has data field pattern validation UC 3 - savingto > RwF18k Using To Occupation ensure that the data inputs are formatted correctly, 50% Co-Contribution National ID, Phone and CAPTCHAKCB 26 toCustodian prevent robot attacks. The system no & Date of Birth Bank Account 27 system Onboarding transmits the applicant’s data to the NIDA Insurance Benefits Real time data from Notification over an APIInvested by RNIT for validation into and for citizens (for next year) verification. UC 1/2 - saving > RwF15k USSD NIDA SMS Govt. bonds UC 3 - saving > RwF18k Once the data at the registration stage is validated and UC 4 - saving > RwF72k verified by NIDA, the next step is to retrieve an b Portal email Life + Funeral LODA MEIS Term deposits Insurance 25. This is required because the child may not a na- yet haveRwF18k/person Premium tional identification number and may only possess a birth certificate document. tration Verification Onboarding 26. CAPTCHA = Contribution Completely Automated Public Turing testEligibility Benefit to tell Computers and Humans Apart. 27. NIDA is the government institution which maintains the national population registry. Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  29 In Ejo Heza, an applicant’s verification is done through integration with NIDA database applicant’s socioeconomic information from LODA.28 charges a fee for this service and it is recommended The real time electronic data communication with that RSSB request this integration for Ejo Heza as well. LODA occurs via the web services integration with the LODA Monitoring Evaluation Information System Aside from NIDA and LODA, the Ejo Heza system is (MEIS). This communication involves two steps. First, a currently not sharing or receiving data with any web service request call is made from Ejo Heza to the other public or private sector institution. During an LODA MEIS with the national ID of the applicant as the interview, the call center team of Ejo Heza noted that input parameter. The second step is the response from a common question was on the status of the life insur- LODA MEIS system to the requesting Ejo Heza system. ance/funeral insurance claim. If the call center team If the national ID is found registered in LODA MEIS, could view the status or stage of processing for the then the household head and member details29 are claim during the call, from the life insurance company, sent in an XML format to Ejo Heza. If there is an error it would allow them to provide meaningful informa- in matching the ID, a message is sent to say that the ID tion to the caller. As it now stands, they can offer gen- is not found in the LODA MEIS database. eral information on the time it takes for processing, to the frustration of the caller. In the immediate future, Enhancements to the integration architecture be‑ interoperability with the private life insurance firm tween Ejo Heza and the LODA MEIS systems can ad‑ (SONWARA) could be considered. For the next phase dress a common concern from Ejo Heza members on of expansion, the Ejo Heza team should explore in- the delay/error in receiving government co-contri‑ teroperability with public or private sector systems/ butions owing to missing or incorrect Ubudehe cat‑ databases (such as those of banks, SACCOs, telcos, and egories. In 2019-20, 17 percent of the subscribers in Ejo chain stores) to reach more consumers, better target Heza were missing an Ubudehe category; this number incentives, and/or improve the customer experience. dropped to 13 percent by 2021. The absence of an Ubu- dehe category specifically hurts those who would fall under the U1, U2, and U3 categories, all of which are Onboarding eligible for matching contributions and life and funeral insurance benefits. This issue does not arise in case of Onboarding is the last stage in the enrollment pro‑ CBHI, where eligibility for subsidies is also determined cess where the applicant confirms the profile de‑ based on Ubudehe categories. This is because pro- tails. In most social insurance programs, verification spective CBHI members need to update their Ubu- and onboarding can happen simultaneously. At this dehe classification in LODA MEIS prior to registration point, the applicant can view his verified personal in- in CBHI. This is not a requirement for Ejo Heza. An- formation and can see data imported from other ad- other challenge is that unlike CBHI, the Ejo Heza sys- ministrative systems (if the feature exists). The appli- tem is not connected to LODA MEIS via the LODA cant will be able to update personal information and messaging queue engine. This prevents Ejo Heza from non-mandatory fields like desired contribution automatically syncing with the LODA MEIS system amount, preferred payment, and communication when a member’s Ubudehe category changes. LODA channel at this stage. 28. LODA is the Local Administrative Entities Development Agency. It is a government agency under the supervision of the Ministry of Local Government. LODA is mandated with the implementation of the social protection and local economic de- velopment programs. 29. The member details sent by LODA are the household head’s first and last names, age, Ubudehe category, household code; also sent are details about each household member, including their first and last names, age, village, sector, district, and prov- ince, as well as the number of dependents. 30  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans BOX 2: The Ejo Heza Information Technology (IT) team on challenges facing the system The Ejo Heza IT unit is a team of four people: a Systems tester, Development Head, Data Systems Lead, and a Database Administrator. This team manages IT for only Ejo Heza and is distinct from the RSSB IT team (which supports the other six RSSB schemes). RSSB is working toward an integration of the platform of Ejo Heza with the main RSSB IT platform but there are transition-related challenges. These challenges have arisen because the program language used for the Ejo Heza platform is different from the one used by RSSB. While an integration is feasible and desirable, staff shortages and the short-term nature of contracts of Ejo Heza IT team members (3-6 months) make it hard for them to manage day-to-day operations while also focusing on integration. The IT team also noted four other challenges: • Need to update infrastructure: Ejo Heza achieved fast coverage growth in a short time and since the systems were initially not ready for this growth they would slow down. The IT team would work non-stop and over the weekends to bring the system back to speed. According to the team, “The slowness in the system has largely been resolved with the application-level changes made.” If, however, coverage was to expand rapidly, more resources will be needed to make the system ready for an increase in volume. • Slowness in generating reports and accessing data from the system: The IT team manages a dashboard which allows staff with access to generate reports and evaluate trends. The IT team acknowledged that generating these reports takes longer than desired. The call center team also noted that because of the slow system, members would call to ask, “I am saving but why is my balance not being updated? Where are my government matching contributions?” This is an area the systems developer has been working on and by end of 2022 they expect this to no longer be an issue. The limited human resources were once again noted as the main challenge in being able to address this issue promptly. • Challenges in making Ejo Heza system interoperable with LODA MEIS: This challenge arises because LODA only has data on the head of household and is only recently beginning to map out all members in each household. Since the Ejo Heza scheme is at an individual level whereby all subscribers are not necessarily household heads, there is a higher chance of missing or incorrect Ubudehe categories when integrating with the LODA system, relative to CBHI where benefits are also targeted at household level. • Building a rich dataset: The subscriber is only asked for their ID, name, phone, date of birth, occupation, and frequency of subscription at the time of registration. However, the IT team retrieves a lot more data at the back end from NIDA and LODA database. For example, the Ejo Heza administrative database is a rich data set and includes address, province, district, photo, whether they were registered by agent, agent details, and more. When asked if the additional variables are of good quality and can be used to estimate which registration channel is most effective, the IT team clarified that it would be better to add a question at the time of registration such as ‘Who helped you register in the scheme?’ and include a drop-down menu. The IT team also noted that they can easily add more options (for example, under occupation) if more details were needed. Most systems have a default notification channel discounted rates with telcos in the initial years but re- (like a text message or a paper receipt30 in case of sorted to auto-debit (as in the case of India’s APY in-person registration) to let the applicant know scheme where contributions are deducted directly that the onboarding to the scheme is completed from bank accounts); others are experimenting with successfully. Applicants can also select a preferred sending notifications via their scheme app (Vietnam) communication channel (text, email, WhatsApp, or or WhatsApp (Gift a Pension in India). phone call for example). The preferred or default com- munication channels can be used to send reminders to In the Ejo Heza scheme, applicants can confirm the members based on their desired contribution fre- data prepopulated from LODA MEIS while finishing quency. Since sending reminders can be expensive, the enrollment process. The applicant can access in- some schemes (like Ejo Heza) have negotiated free/ formation about their Ubudehe category, registered 30. This is the case in the voluntary scheme of Vietnam where agents issue a card at time of registration and update the card every time a contribution is made (like a passbook). Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  31 Individual or group (cooperative) subscribers to Ejo Heza can pay contributions through mobile money, bank transfers, cash deposits, or via credit card address (province, district, sector, village), and depen- informal sector workers have. The reach of digital pay- dents. They may confirm their desire to retain the ments is expanding rapidly in this group in Rwanda, as same address. At this stage the applicant is also re- in other African countries. While they should be en- quired to enter their occupation and specify the fre- couraged for their ease and cost effectiveness, care quency and amount of their contribution. The appli- should be taken that groups (for example, women, cant may include their email address as an optional poorer groups without a phone) should not be ex- field. Data as of 2020 indicates that about 80 percent cluded only because they do not have digital access. of members chose the monthly contribution fre- quency, 9 percent quarterly, and 6 percent annual. Individual or group (cooperative) subscribers to Ejo However actual contribution trends suggest little ad- Heza can pay contributions through mobile money, herence to the chosen payment frequency. bank transfers, cash deposits, or via credit card. The scheme allows for a wide variety of ways to Contribute The third stageENROLL of the delivery chain, Contribute, in- CONTRIBUTE (RECURRING CYCLE) PROV cludes two substages—Contribution and Benefit Eligi- bility. At the Contribute stage, the individual has made Confirm Contribution channels Based a decision to make at least the first contribution into Based on Registrant the Ubedehe category Mobile money savings & Saving validation forscheme and would be interested in knowing the (UC) UC age benefits and incentives available. A lack of understand- Online & mobile ing citizens Resident or a lack of appreciation of the Address, incentives available personal Before age (55 banking to an individual can cause & family the details incentive to be ineffec- Capped maximum Govt. Saving > RwF 4 Cash deposits Co Contribution - RwF18K 40% Pa Diaspora tive. It is therefore not only important to include tar- Enter withdrawals; geted incentives in scheme design but equally impor- UC 1/2 - saving > RwF15k Contribution Card payments excess of R Resident tant to ensure they are understood by members and Foreigners 100% Co-Contribution frequency & amount At age (55) are deemed attractive, given the characteristics of the To UC 3 - saving > RwF18k Saving < RwF Using informal sector. Occupation 50% Co-Contribution Lumpsum o National ID, Phone KCB Custodian Bank Account Installment no & Date of Birth Contribution Onboarding Insurance Benefits drawd Real time data from Notification Invested by RNIT into for citizens (for next year) At age (55) The next step after registration is for the member to UC 1/2 - saving > RwF15k Saving > RwF NIDA make a decision to contribute SMS to the scheme. If Govt. bonds UC 3 - saving > RwF18k there isn’t an auto-debit design in place (for example, UC 4 - saving > RwF72k Up to 25% Life + Funeral lumpsum wit the LODA Heza or banks in India’s case of cooperatives in Ejoemail MEIS Term deposits Insurance remaining APY), a voluntary scheme needs to focus on ensuring pension for 2 that individuals know how to contribute and do so Premium RwF18k/person continually. Members contributing for the first time Verification may need handholding from Onboarding the agent network or Contribution Benefit Eligibility Withdr helpline. The method of contribution collection needs to be mindful of usage and access of instruments that 32  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Individual or group (cooperative) subscribers to Ejo Heza can pay contributions through mobile money, bank transfers, cash deposits, or via credit card contribute: (i) mobile money digital payment using Air- (1 percent), cash account (4 percent), and treasury tel money or MTN money; (ii) bank transfers or online bonds (95 percent). The nominal investment return and mobile banking with partner banks—for example, credited to member accounts was 9 percent in fiscal National Bank of Rwanda, the Bank of Kigali, Equity year 2021 (RSSB 2022). Bank, and Cogebanque Bank; (iii) deposits in cash at partner banks; or (iv) credit card payments using Visa or In the first three years of the scheme, Ejo Heza, in MasterCard. When a subscriber contributes for him- the interest of raising public awareness, partnered self/herself, they can use their preferred channel and with a telecom company, and requested that they only need to provide their national identity number as offer free text messages as contribution reminders reference for the transaction. When cooperatives or to members. The scheme no longer offers text re- groups make payments on behalf of their members, minders as they proved to be a costly and recurring the Ejo Heza system allows for uploading a declaration communication expense. Behavioral studies have also with each member’s ID and payment amount. The sys- shown that generic reminders are not as effective as tem in turn creates a unique payment reference ID for tailored reminders. Cost permitting, tailored reminders the bulk payment. Subscribers can access their ac- with the name of the subscriber, a reference to their count balances via USSD and through the Ejo Heza desired contribution frequency, and a motivating mes- online web platform. The individual account balance sage, are much more effective. Interviews with the Ejo of a subscriber is displayed under three sub-headings: Heza team has revealed that the strategy going for- (i) contribution payments, (ii) government matching ward is to look for effective aggregators like the coop- contributions, and (iii) investment returns. The account eratives who, by design, ensure higher persistency with balance at any given point is the sum of these three minimal recurring costs. components. Benefit eligibility The contribution from the subscribers is collected in KCB Bank’s custodian account before it is trans‑ Ejo Heza is a voluntary defined contribution scheme ferred to be invested by the fund manager Rwanda which means that the benefits are largely a function National Investment Trust (RNIT). RNIT invests the of contributions paid by members and returns funds in the custodian account according to the earned by the scheme. The level of contributions and scheme’s investment policy. Ejo Heza’s investment frequency are not mandated in a voluntary scheme so holdings totaled RF 23 billion as of December 31, 2021, ensuring persistent contributions from members is key and were spread out between fixed deposits to retirement adequacy from the scheme. Scheme In the first three years of the scheme, Ejo Heza, in the interest of raising public awareness, partnered with a telecom company, and requested that they offer free text messages as contribution reminders to members Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  33 administrators face a number of trade-offs when de- 15,000 annually for Ubudehe categories 1 and 2 and RF signing pension schemes for informal sector workers. 18,000 annually for Ubudehe category 3. The amount One among them is striking a balance between the of government matching is higher (100 percent match- short-term needs of informal sector workers and the ing for U1 and U2, 50 percent for U3) and the level of long-term goals (for example, ensuring income secu- savings needed to qualify for matching is also lower for rity in old age) of a pension scheme. Spreading aware- Ubudehe categories 1 and 2. This is progressive given ness and propagating a long-term saving culture is cer- that individuals from these households are expected tainly a crucial piece of the policy agenda. However, to be poorer and have lower ability to save. Members international experience has shown that including de- in U4 do not receive government matching contribu- sign features that offer tangible short-term benefits in a tions as they are considered to be well-off. However, voluntary pension scheme can go a long way in attract- to encourage them to save for old age, the govern- ing informal sector workers to save (Guven et al., 2022). ment offers members a tax exemption if they save at least RF 72,000 in Ejo Heza annually. The short-term The Ejo Heza scheme is novel in its design as it offers benefits of life and funeral insurance are among the flexibility, features to attract individuals to save a most motivating and attractive features of Ejo Heza, as minimum level, and tangible short-term benefits to highlighted in interviews by district mayors, agents, members. Members have the flexibility to partially and cooperatives. The 2022 phone survey found that, withdraw from the scheme once they have saved at on average, 89 percent of subscribers found the fiscal least RF 4 million. They can also access the loan feature benefits to be valuable or very valuable. Surprisingly once they have hit this level of minimum savings. The there is not much variation in how respondents value matching contributions offered by the government the three benefits—matching contributions, life, and (see Table 1) incentivize members to save at least RF funeral benefits (Figure 19). The Ejo Heza scheme is novel in its design as it offers flexibility, features to attract individuals to save a minimum level, and tangible short-term benefits to members FIGURE 19.  Members rating of the fiscal benefits in Ejo Heza 60 50 40 Percent 30 20 10 0 Not valuable at all Less Valuable Neutral Valuable Very Valuable Life insurance Funeral insurance Govt matching Source: Phone survey results, 2022. 34  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans The benefits offered under the voluntary scheme are a cost to MINECOFIN but they are explicit and capped and therefore not a contingent liability like the defined benefit scheme of the formal sector pension program The benefits offered under the voluntary scheme members. The Ejo Heza team should investigate why are a cost to MINECOFIN but they are explicit and the take up is low despite the attractiveness of these capped and therefore not a contingent liability like benefits. If more individuals can access these benefits the defined benefit scheme of the formal sector pen‑ while they last it will increase Rwandan’s familiarity sion program. Government matching for U1, U2, and U3 with the scheme and lead to a coverage increase via is capped at RF 18,000 per saver. This makes the fiscal word-of-mouth growth. While focusing on ensuring costs explicit and capped31 unlike in defined benefit more individuals get access to these fiscal benefits, the pension schemes where there is a contingent liability if team also needs to think of a long-term strategy. A the contributions collected are not actuarially fair. The hassle-free contribution process in Ejo Heza where in- life and funeral insurance are not administered by the dividuals get competitive market returns are a good government. Instead, they have been contracted out start as the informal sector workers currently do not to a private sector insurance firm, Sonarwa, following have any other scheme that offers them any income a competitive selection process. The insurance is only security in old age. The phone survey (2022), however, available for those who save the minimum levels (see revealed that only 14.2 percent of members knew the Table 1 for rules). Currently, MINECOFIN finances the investment return that was earned on their Ejo Heza cost of the insurance premium and pays RF 18,000 in account. Among those who knew the investment re- premium to Sonarwa for each eligible member. turn in Ejo Heza, 64 percent were satisfied and 34.5 percent were highly satisfied. The awareness levels did The matching contribution and insurance were not vary by gender, pointing to a need for a concerted unanimously touted as the most attractive features effort by the Ejo Heza team to highlight the inflation- of Ejo Heza during the consultations but surpris‑ competitive and stable returns being offered by the ingly only 25.5 percent of active savers in 2020 saved scheme. Communication material should feature the the minimum levels to qualify for these benefits. historical return, how people can check their return, While the incentives themselves are attractive and a and testimonies of satisfied individuals. The trust signal of policy commitment from the government, Rwandans have in RSSB, and the Government of most individuals are not being able to access these Rwanda is a vital positive factor for ensuring long-run benefits. Careful investigation is needed to understand success of the scheme. As the insurance market deep- what might be happening. Is the minimum saving level ens, coverage grows, and more experience develops, it too high for individuals or are most unaware of benefits may be possible to negotiate more competitive premi- or forget about them after they have registered? ums. In case of Colombia, for example, the social secu- rity institution (ColPensiones) has negotiated the pack- The incentives under Ejo Heza are not time bound age of inclusive insurance (lump sum in case of death, currently but it is imperative to think of ways to sus‑ sickness, disability) with an insurance company and tain momentum if and when fiscal incentives from transfers 1 percent of the individual’s saving as pre- MINECOFIN are no longer available or limited in mium to the insurance company (see annex of Guven scope. The matching contributions as of 2021 formed et al. (2021) for details). In doing so, the fiscal burden is only 8.1 percent of the Ejo Heza fund balance and life no longer on the government. insurance premiums were paid for only 210,213 31. It can also be time bound; for example, India’s APY scheme only offered matching for those who registered in year 1 of the scheme. Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  35 The pension eligibility age in Ejo Heza is 55 years and the options available to withdraw after that age depend on whether the fund balance is less than or more than RF 4 million Provide CONTRIBUTE (RECURRING The fourth stage CYCLE) of the delivery chain, Provide, in- PROVIDE (RECURRING CYCLE) cludes two substages—Withdrawals and Benefit Pay- out. At the Provide stage, the individual needs to inter- Contribution channels act with Based the scheme to evaluate on is eligible for if he/she Based on Support Mobile money early savings withdrawals. The Benefit Payout & stage refers to Savings & UC age Greive the substage at which the individual interacts with the Online & mobile If savings > RwF 4M scheme to receive a benefit (retirement or early with- Before age (55) banking drawal) payout. Capped maximum Govt. Saving > RwF 4M 40% savings as loan Compl Cash deposits Co Contribution - RwF18K 40% Partial collateral, leaving UC 1/2 - saving > RwF15k withdrawals; savings in atleast RwF 4M Card Withdrawals payments excess of Rwf 4M Data U 100% Co-Contribution At age (55) In a pension scheme, the withdrawals, in most in‑ To UC 3 - saving > RwF18k Saving < RwF 4M Death claim If insured Claim App stances, begin once the pension eligibility age, as KCB Custodian 50% Co-Contribution Lumpsum or equated per the scheme rules, is reached. Increasingly, in both RwF 1,000,000 Account and voluntary schemes, earlier withdrawals Bankmandatory Installment to total Using Life Insurance claim Insurance Benefits drawdown by short-term Invested or RNIT into withdrawals for citizens are (for alsonext year) offered to meet Rwf 250,000 At age (55) Call cente UC 1/2 - saving > RwF15k the liquidity needs of individuals and increase the ap- Funeral Insurance Saving > RwF 4M Govt. of saving withUC bonds peal 3 - saving these schemes. > RwF18k On the flip side, claim UC 4 - saving allowing early access impacts adequacy > RwF72k of the scheme Up to 25% partial Online Life + Funeral be limited to lumpsum withdrawal + so it is Term deposits recommended that withdrawals Insurance remaining monthly emergency needs (for example, health expense, death, pension for 240 months Wal Premium disability). Alternatively, design RwF18k/person the scheme to offer a short-term (accessible anytime) and long-term Contribution Benefit account,32 or (as in case Eligibility of Ejo Heza) allow withdrawals Withdrawal Benefit Payout Monit once individuals meet a minimum saving amount. The pension eligibility age in Ejo Heza is 55 years and the options available to withdraw after that age de‑ interest even as the drawdown of savings continues. pend on whether the fund balance is less than or Individuals with fund balance more than RF 4 million at more than RF 4 million. Individuals who have a fund retirement can receive up to 25 percent of the funds balance less than RF 4 million at retirement can with- as lump sum and the remaining as a pension annuity draw their entire account as a lump sum, or request an that is calculated as remaining fund balance divided by ‘equated installment.’ An equated installment is one 240. Members are allowed full access to funds prior to where the subscriber chooses the frequency of with- reaching age 55 under three circumstances: permanent drawals and the amount is calculated based on their disability, terminal illness, and emigration to another fund balance. The fund balance will continue to accrue country by non-nationals. 32. This is also referred to as the side-car account structure (see Stewart, Jain and Sandbrook 2018). 36  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans The pension eligibility age in Ejo Heza is 55 years and options available to withdraw after that age depend on whether the fund balance is less than or more than RF 4 million The scheme also offers withdrawals in the short eligible for this incentive. The next of kin of eligible term (before the pension eligibility age of 55) if fund beneficiaries can apply for a claim by visiting the Ejo balance is more than RF 4 million. A member who Heza website (ejoheza.gov.rw), choosing the nature of meets this criterion can withdraw up to 40 percent of claims in the claim’s module, and submitting relevant their savings, in the excess of RF 4 million, for housing documents. Funeral benefits in the amount of RF and education expenditures. This feature ensures that 250,000 are paid out within 24 hours, while life insur- members retain at least RF 4 million (US$4,000) in their ance death benefits in the amount of RF 1,000,000 are retirement saving while also getting access to their paid out within 7 days. While digitization of the claims funds to engage in productive activities. While there process reduces costs and improves efficiency, care are members who have become eligible for pensions should be taken that members have access to alter- (because they reached age 55), there is no member so nate ways to submit claims (for example, through the far who has saved RF 4 million, given that the scheme district coordinator or agents) especially in what is an has been operational for only four years. As the emotionally difficult time for next of kin. Interviews scheme matures, it is likely that some members will with two agents revealed that they are currently not qualify for withdrawal in the short term. able to help members with claim submission. Aside from monetary benefits, members can also Benefit payout get loan collateral guarantees from a financial insti‑ Benefit payments under the Ejo Heza scheme take tution for an amount equal to 40 percent of savings place either in case of short-term withdrawals, with‑ in excess of RF 4 million. The Ejo Heza scheme is still drawal after reaching pension eligibility age, or if a young and there is no saver who qualifies for the loan life or funeral insurance claim is made. Calculating guarantee under scheme rules. This is an innovative and processing payments once someone reaches pen- feature and could attract individuals to join the scheme sion eligibility age is straightforward as Ejo Heza has if and when some are able to benefit from this feature. date of birth records and detailed contribution history Since the scheme itself is not offering the loans to in- of each individual. Claims under life or funeral insur- dividuals and the collateral comes with a cap (only 40 ance, however, require a proof of death and a claim percent in excess of RF 4 million), Ejo Heza will be able request from the beneficiary’s next of kin. Since the to avoid contingent liabilities and keep administrative insurance benefit is contracted to Sonarwa, a private costs in check. However, scheme administrators need to sector insurance firm, the claim verification and issu- evaluate if RF 4 million is a reasonable level that mem- ance is also their responsibility. Not all members are bers can reach. If most members are unable to reach automatically eligible for the life and funeral insurance that target, then this incentive only remains good on benefit. As noted in Table 1 (rules), only those who paper, and it might be more prudent to reduce the sav- meet the minimum saving (by Ubudehe category) are ings level at which loans can be offered to members. Benefit payments under the Ejo Heza scheme take place either in case of short-term withdrawals, withdrawal after reaching pension eligibility age, or if a life or funeral insurance claim is made Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  37 Manage PROVIDE (RECURRING CYCLE) MANAGE The fifth and last stage of the delivery chain, Manage, includes two substages—Monitoring and Exit. At the Based on Monitoring stage, the scheme aims to collect feed- Support Savings back&from the individual on customer experience and age Greivences grievances and monitors their changing attitudes/ If savings > RwF 4M Before age (55) preferences. In the Exit stage, the intention is to check Saving > RwF 4Mthe scheme does 40% that savings not include as loan any member who is Compliance End of Withdrawal 40% Partial no longer allowed to be collateral, leaving part of the scheme. Period withdrawals; savings in atleast RwF 4M excess of Rwf 4M Data Update At age (55)Monitoring Death of Beneficiary Saving < RwF 4M Death claim If insured Claim Application Current members, prospective registrants, and Lumpsum or equated RwF 1,000,000 those from the target groups might become familiar Installment to total Using with the scheme throughLife Insurance one of theclaim many touch‑ drawdown At age (55)points (agent network, Rwf 250,000 walk-ins at the counter, web Call center / Email platform, Saving > RwF 4M call-center, Funeral email, Insurance local community). Mon- claim itoring refers to the process of ensuring that the vari- Up to 25% partial Online Forms ous touchpoints lumpsum withdrawal + are able to meet the needs of the remaining monthly individual or point them to a resource channel that is pension for 240to able months do so. Some examples of the support chal- Walk ins lenges individuals face might be associated with regis- tration, data management, benefit disbursements, Withdrawal Benefit Payout withdrawals, claims, and other grievance solutions. Re- Monitoring Exit sults of the phone survey (2022) found that, on aver- age, most subscribers find the contribute and provide stages of the delivery chain ‘easy’ to navigate (Figure Ejo Heza relies on four key resource channels to an‑ 20). Among the three key processes, finding out about swer questions or provide support to current and incentives was deemed, on average, to be more difficult prospective members. These include a dedicated than contributing or finding enrollment information. toll-free helpline (4044, 5006) that is available between FIGURE 20.  Subscribers’ rating of three key activities by level of difficulty 90 80 70 60 Percent 50 40 30 20 10 0 Very difficult Difficult Neutral Easy Very Easy Contribute to Ejo Heza Find out enrollment info Find out about incentives Source: Phone survey results, 2022. 38  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Current members, prospective registrants, and those from the target groups might become familiar with the scheme through one of the many touchpoints (agent network, walk-ins at the counter, web platform, call-center, email, local community) the hours of 8 a.m. and 10 p.m.; email at the support schemes. An increase to staffing could be considered as address (info@ejoheza.gov.rw); walk ins to any RSSB calls on Ejo Heza have increased substantially—from regional office; and/or reaching out to the District Co- 5,500 calls a month in 2019 to 16,000 in 2021 and about ordinator of RSSB in their district. When subscribers to 30,000 as of July 2022. The annual call center report Ejo Heza were asked in the phone survey who they (2019) noted that commonly requested inquiries at the reach out to if they have a question on Ejo Heza about call center were about balance checks, how to con- 43 percent said they do not know who to ask (Figure tribute, how to update profile information, how to 21). At least 13 percent of them were comfortable update phone number, pin reset, and how to register reaching out to the call center while 14 percent would (RSSB 2019c)Interviews with the call center team re- ask their cooperative. vealed that the kind of questions have also evolved over time, pointing to the rising awareness. In early 2019, The call center is a centralized team in the RSSB Ki‑ the questions were: “What is Ejo Heza? How can I reg- gali headquarters and can provide real time insights ister? Am I eligible to register? Can I register with the into the challenges individuals face, questions they scheme if I am saving in other schemes? Can my kids have, and how they evolve over time. The call cen- register?” After 6-12 months from the launch, the ques- ter’s responsibilities have been contracted out to a tions were focused on matching contributions: “Which private sector firm. There are currently seven agents category gets how much benefit? How long will I get the dedicated to Ejo Heza and five for all other RSSB benefit? Why are you only giving benefit for 2/3 years?” Currently (as of July 2022), the common questions are: “I am saving but why is my balance not being re- FIGURE 21.  Who do members ask if they flected? Where are my government contributions?” have a question on Ejo Heza? Comparing call center reporting data from 2018–19 and 2020–21 revealed that the incoming call traffic 13% increased by 15 percent only, while subscribers in‑ 30% creased by sixfold during that same period. The ef- ficiency of the call center team appears to have im- proved during this period as the ‘dropped call’ rate33 decreased from 31 percent to 12 percent. The cited reason for the high call abandonments were limited 43% 14% human resources. During this period, the ratio of bal- ance enquiries reportedly tripled, and profile update Call Ejo Heza Call center I don’t know whom to ask 33. Dropped call rate refers to the share of total calls that call My cooperative My friends/family center couldn’t get (got dropped) because customer hung up while they were waiting in queue. Source: Phone survey results, 2022. Section 3: Assessing the Ejo Heza Scheme using the Social Insurance Delivery Chain n  39 The exit stage identifies the criteria and processes of a social insurance scheme to determine when a beneficiary is no longer eligible for benefits in the program queries increased by 14 times. The call center staff Exit received a 5-day refresher training in early 2022. Dur- ing an interview with the World Bank team, the call The exit stage identifies the criteria and processes center team reported that training by Ejo Heza staff of a social insurance scheme to determine when a on ‘key messages’ was most useful. They noted that beneficiary is no longer eligible for benefits in the some callers would make calls only to verify something program. Reasons for program exit under the Ejo Heza they might have heard on the radio, so it was impor- scheme could be retirement itself, early withdrawal tant for the call center team to be abreast of key mes- under exceptional circumstances, or death of the ben- sages being highlighted to ensure that they weren’t eficiary. Typically, programs follow a four-step process providing unclear or conflicting messages to the caller. in exiting beneficiaries: an exit trigger, criterion valida- The efficiency and resource constraints of the call cen- tion, exit decision, and beneficiary notification. Exit ter should also be looked at, in light of the increasing triggers are updates that result in changes to a benefi- call volume. ciary’s basic information. Criterion validation is an ac- countability check to ensure that a member is being The call center serves another critical purpose of exited for the reasons that are being cited and the carrying out phone surveys which allows policy rules for exiting are being met. Exit decision is a yes/no makers in Rwanda to get representative insights depending on whether the criteria are validated or from members across districts, age, gender. The Ejo not. Notification of the decision is then sent to the Heza Net Promoter Score survey report (2021) found beneficiary or the next of kin (in case the exit is due to that 41.05 percent of respondents were satisfied with death) by the preferred communication channel. Ejo Heza and 12.8 percent were dissatisfied. The top three concerns of dissatisfied members were that: Accurate records and a transparent certification (i) contributions were not reflected on the system process are important to ensure that members have dashboard, (ii) verification of account balance was a a smooth experience at the exit stage and there is challenge, and (iii) government co-contribution was no leakage in the scheme. More than other social pro- not credited (NPS RSSB, 2021). Those who were satis- tection schemes, pension systems are more likely to fied with the scheme noted that Ejo Heza offered receive cases of death benefits and need to process them a chance to be able to ‘retire in dignity’ and it was payments promptly and accurately. At the same time, easy for them to retrieve their account balance. The there are examples of schemes with ‘ghost’ pensioners dichotomy in responses on ability to view fund bal- that arise due to leakages, errors, or fraud.34 Ejo Heza ance shows that awareness building has a role to play should ensure that the nominated beneficiary data- in addressing some of the members’ concerns. A sur- base is up to date, the date of birth of both beneficiary vey that asks more details on access/usage of phone, and nominee is accurate, and that the phone number or educational background, financial awareness, and loca- address is the latest. A clear, accountable mechanism tion will allow Ejo Heza teams to assess if dissatisfac- should be in place to certify through official means tion is concentrated, for example, with individuals who (such as a death certificate) that the beneficiary has have feature phones, with those above a certain age, passed away. Ensuring record accuracy and transparent or with a specific group like women. exit processes while the scheme is still young will avoid legacy challenges as the scheme matures. 34. See Omondi (2019) and Singh (2022) for examples. SECTION 4 Ejo Heza Member Preferences to Select Design Changes T he need for solutions to increase pension coverage among informal sector/ self-employed/part time workers has motivated countries across Africa, Asia, and Latin America to operationalize voluntary schemes. The voluntary schemes differ in design, incentives, who they target, and withdrawal rules, but the objective of all is to offer a scheme that meets the needs of those traditionally ex- cluded from formal sector schemes (see annex of Guven et al. 2021 for details on other schemes). Among such schemes, Ejo Heza is emerging as a leader given its rapid coverage growth, notwithstanding the impact that COVID-19 had on its mobilization strategy. The only other voluntary scheme similar to Ejo Heza in terms of coverage is China’s RURS scheme—at 38 percent of working age population—which offers a generous subsidy to the poor and vulnerable groups35 to encourage saving in the scheme (Jain and Palacios 2021). China introduced the rural resident scheme in 2009 and urban resident scheme in 2011 and unified them as one single scheme, Rural and Urban Resident Scheme (RURS) in 2014 to cover informally employed workers. Coverage expansion in the voluntary schemes of other countries (such as Thailand, India, Colombia, Vietnam, and Kenya) has not been as high as that of Ejo Heza but there are three lessons learned which are considered to be applicable for Ejo Heza. First, flex- ibility in withdrawals is desired (India’s latest scheme, Pradhan Mantri Shram Yogi Maan- dhan,  offers much more flexibility than its predecessor scheme, APY). Second, bundling can be attractive if it is communicated well and if it allows informal sector workers to get similar risk coverage as the formal sector enjoys. Third, and most importantly, finding ag- gregators that can replicate the ‘observe and auto-deduct’ role that employers play for formal sector pension schemes is a gamechanger for scaling up coverage. The ‘observe’ feature enables an administrative relationship through which trust comes into play and compliance or nudges can be exercised. The ‘auto-deduct’ feature enables the overcoming of behavioral bottlenecks and makes it both cost effective and efficient for individuals to save. Cooperatives are an excellent example of an aggregator in Ejo Heza as they can ob- serve incomes/behavior of their members and can auto-deduct the committed saving amount at the time of harvest. India’s APY scheme depends on the large network of na- tional and state banks to be an aggregator. The government initiative of Jan-Dhan Yojana opened a total of 200 million bank accounts for the undeserved and these accounts are now all eligible to join APY. Banks are offered monetary incentives for every individual that 35. About 25 percent revenue in the RURS comes from individual contributions and the rest from national and subnational government subsidies. The matching subsidies to individual accounts come from local governments In 2019, around 22 million workers had their full contributions paid by the central government. 40 Section 4: Ejo Heza Member Preferences to Select Design Changes n  41 Countries across Africa, Asia, and Latin America have operationalized voluntary schemes to increase pension coverage among informal sector/self-employed/part time workers joins and continues to save in the scheme. Contribu- • Scaling up coverage through mandates and/or tions to APY are auto-deducted from bank accounts. identifying aggregators In Kenya, the Haba Haba scheme is exploring using • Eligibility restrictions based on age. digital taxi companies, Boda-boda drivers and Safari- com (the largest telco company) as aggregators. Coun- tries that have strong foundational systems and a digi- Short-term access to funds tal payment ecosystem can also explore an aggregator at the scale of tax authorities. While incomes are not Individuals working in the informal sector, self-em‑ observable for many in the informal sector or those ployed or gig workers on average have higher liquid‑ who don’t meet the minimum tax slab, consumption- ity needs than those in the formal sector. Not all based taxes like the value added tax (VAT) are in theory who work in the diverse informal economy are low- traceable in a digital economy. The tax authority can, income individuals but the lack of adequate insurance in principle, leverage both the ‘observe’ and ‘auto-de- (for health, employment, sickness, and so on) and sta- duct’ feature and can divert a percentage of VAT taxes ble income sources mean that many face liquidity into every individual’s pension account. challenges and have to tap into savings to meet con- tingencies. A literature review note (Stewart & Jain et The differences in country context and starting al. 2019) showed that when faced with adverse shocks, points mean that not all lessons from global experi‑ limited access to retirement savings can in fact help ence are relevant for Rwanda. However, scheme de- increase the overall welfare of these individuals but sign changes which are also aligned with the aspiration there are adequacy and relevance issues if unlimited of Ejo Heza i.e. expanding adequate pension coverage access to pension savings is allowed. The need to get to all Rwandans, can be considered for discussion. The this balancing act right is being felt increasingly as cov- phone survey allowed for an opportunity to elicit erage is expanded to the non-salaried/informal work- member preferences on some design changes. These force. The report includes country examples and changes are not recommendations of the authors but points to four methods36 that are commonly allowed where relevant benefits, risks, and feasibility of each for early withdrawal access to pension funds (Stewart change ae highlighted. It should also be noted that et al. 2019). It argues for the need to balance current these design changes individually or together are not interest with future needs of members by facilitating meant to solve the coverage, adequacy, and viability controlled access or offering a dual-account system objectives of the scheme. Section 6 of this report with varying degrees of liquidity. presents thorough policy recommendations. The four design changes discussed in this section were selected In the 2022 phone survey, respondents were asked if based on consultations with the Ejo Heza team and they had access to other savings they could use in international experience. They are: case of emergencies or to meet short-term needs. Sixty percent of them reported having access to such • Short term access to funds funds and 33 percent indicated that they had no access • An increase in the minimum savings level to be eli- (see row totals in Table 4). Another question posed to gible for government incentives the respondents was whether they desired partial 36. These methods are permanent withdrawals; loan and repayment; pension savings as collateral; and the dual account (also known as hybrid savings or side-car account structure). 42  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans TABLE 4:  Percentage of members who would like partial access to Ejo Heza savings Desire to partially withdraw Ejo Heza savings anytime and for any purpose Access to saving for short-term needs/emergency No Yes Total No 2% 31% 33% No, but I can ask my family and friends for help 0% 7% 8% Yes 5% 54% 60% Total 8% 92% 100% Source: Phone survey results, 2022. access to their savings in Ejo Heza. Ninety-two percent the target of RF 4 million or when an individual reaches of all respondents said they would like partial access age 55. It should be noted that allowing short-term and, as seen in Table 4, individuals who reported having access would involve additional administrative ar- access to other savings also desired short-term access. rangements. Systems and administrative capacity will This points to the overwhelming desire by members, need to be developed to handle partial withdrawals. irrespective of alternate savings, to want partial access from Ejo Heza. It is promising that most members don’t desire unfettered access. When asked how much Increase in the minimum access to funds would be desirable, 33 percent of re- savings level to be eligible for government incentives spondents said they would like access to less than 30 percent of Ejo Heza savings. About 52 percent of re- spondents would like access between 30-50 percent The annual savings in terms of GDP per capita that of funds and less than 1 percent would like full access is needed to be eligible for the government match‑ to funds (Figure 22). Based on the survey findings and ing and life insurance benefit is only 1.8 percent of in recognition of liquidity needs of the target group, GDP/capita for those in Ubudehe category 1 and 2; Ejo Heza could consider a side-car account structure 2.1 percent for Ubudehe category 3; and 8.7 percent with up to 30 percent or 40 percent in a short-term for Ubudehe category 4 (see Table 1 for rules).37 If liquid account and the rest in a long-term saving ac- the aim is to prevent individuals for falling into old age count that can only be accessed once savings reach poverty, a stream of pensions equivalent to the pov- erty line will be needed for the 240 months of the FIGURE 22.  How much early access to funds payout phase. A back-of-the-envelope calculation would members like? suggests that an annual contribution of RF 45,000 for 1% at least 25 years38 will be needed to generate a fund balance that can produce pensions equivalent to the 15% poverty line39. The average savings under Ejo Heza is 33% only between RF 9,000–RF 12,000, and therefore much lower than this estimated annual contribution . If the low savings in the scheme continue it could pose a reputational risk in 20-25 years when cohorts start 52% 37. The GDP per capita as of 2021 for Rwanda was RF 824,345. 38. Assuming the scheme earns a 10 percent return each year. We assume 25 years of contribution as experience suggests Less than 30 percent Between 30–50% that it is difficult to ensure persistency over long periods of Between 50–70% Access to all my savings even time especially in voluntary scheme. before old age 39. The national poverty line assumed for this calculation is RF 159,375 (2014) and it has been inflated to 2022 numbers and Source: Phone survey results, 2022. amount to RF 223,361 annually. Section 4: Ejo Heza Member Preferences to Select Design Changes n  43 In the 2022 phone survey, respondents were asked how they would react if the government raised the minimum savings needed to be eligible for matching contributions retiring. If individuals can save more or for a longer pe- FIGURE 23.  Reported ability to save if riod, or if the scheme earns higher returns, a higher minimum savings limit for government pension can be expected.40 incentive was increased 2% In the 2022 phone survey, respondents were asked how they would react if the government raised the 18% minimum savings needed to be eligible for matching 35% contributions. Eighteen percent said they might not be able to save, 10 percent were uncertain, and 2 per- 10% cent were sure they would not be able to save more; 70 percent of respondents said they ‘might’ or are ‘most likely’ to save more (Figure 23). The optimism in being able to save more to meet a higher matching is 35% intriguing especially since administrative data of Ejo Heza suggests that many members do not meet cur- I will not be able I might not be able I am uncertain to save more to save more rent matching.41 The individuals who say they might or I might save more I will most likely save more are most likely to save more also report a higher sav- Source: Phone survey results, 2022. ings in Ejo Heza to date (Table 5). The correlation be- tween current savings in Ejo Heza and reported ability able to save more. If the goal of the scheme is to en- to save more points to the fact that those who are able members to enjoy pensions at least equal to pov- saving lower amounts currently are less likely to be erty line for the 240 months of payout phase, then the TABLE 5:  Reported ability to save more compared with reported savings in Ejo Heza Average savings in Ejo Heza to date Std. Deviation I will not be able to save more RF 5,500 RF 8,544 I am uncertain RF 16,219 RF 12,425 I might not be able to save more RF 16,737 RF 13,156 I might save more RF 16,141 RF 19,335 I will most likely save more RF 17,463 RF 15,986 Grand Total RF 16,594 RF 16,429 Source: Phone survey results, 2022. 40. Since the remaining fund balance after retirement will continue to accrue interest, the scheme might be able to offer infla- tion-indexed pensions as long as the scheme is making good returns. 41. In follow-up analysis of the phone survey the World Bank will aim to link the contribution records of the respondents to look at the administrative data and see if the individuals who report being able to save more are meeting the current minimum requirements for eligibility. 44  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans An agent suggested that Ejo Heza's coverage could be scaled up by making the scheme mandatory for all those in the informal sector scheme needs to motivate people to save more than Mandates themselves can come in different forms. the average saver currently. A higher threshold for gov- They can be selective for e.g., only apply to richer ernment incentives could motivate some members households like U3 and U4 or they can be universal but but for others the higher threshold would need to be financed without imposing an additional cost to indi- accompanied by a more generous matching by the gov- viduals, especially the poor. When Ejo Heza members ernment.42 The high ratio of U1 and U2 savers in Ejo Heza were asked if there should be mandatory coverage for points to the interest they have in the scheme but their individuals belonging to Ubudehe U3 and U4 (the more inability to meet the minimum saving level also shows well-off income categories), 68 percent of respon- that these groups might need more subsidies from the dents answered ‘Yes’ (Figure 24). There is a diversity in government, similar to the subsidies under CBHI. responses across the income spectrum (Figure 25) but on average there is acceptance for a mandate even among the more well-off individuals who are more Scaling up coverage through likely to be affected by such a mandate. It should be mandates and/or identifying noted that despite digital advances, enforcement chal- aggregators lenges with mandates remain. A case in point are man- dates on the formal sector which have been a promi- In an interview, an agent from the CARE VSLA group nent feature of social security but institutions including from Kirehe district suggested that coverage within RSSB continue to need compliance teams to enforce Ejo Heza could be scaled up by making the scheme and monitor. mandatory for all those in the informal sector. This suggestion is in line with decades of pension his- FIGURE 24.  Should U3 and U4 save tory experience where mandates are the single most mandatorily in Ejo Heza? effective tool used by policy makers to provide pen- 2% 0% sions to formal sector workers. However, there are three main challenges that policy makers face in man- dating contributions for the informal sector: (i) in- comes are unobservable and vary, making it adminis- tratively hard to mandate based on income; (ii) ensuring 30% compliance of the mandate is difficult, as there is no registered ‘employer’ through whom administrators can track contributions; and (iii) the low and irregular incomes of this group of workers means that the ad- 68% ministrative cost of collecting and managing their con- tribution can be high. These challenges were insur- mountable in the era of paper-based records, lack of digital systems, and limited digital awareness. However, I don’t know Maybe the sweeping increase in use of digital payments and No Yes digital media globally and particularly in Africa has Source: Phone survey results, 2022. opened a range of possibilities. 42. There will soon be a social registry managed by LODA that will be used for classifying households into distinct income categories. This objective targeting criteria would be better suited to target Ejo Heza subsidies to those who need higher matching to meet the proposed (higher) saving threshold. Section 4: Ejo Heza Member Preferences to Select Design Changes n  45 FIGURE 25.  Preference to mandate coverage for U3 and U4 RF 400,000–600,000 2 2 RF 200,000–400,000 4 12 RF 100,000–200,000 14 19 RF 50,000–100,000 12 41 RF 0–50,000 80 179 0% 50% 100% No Yes Source: Phone survey results, 2022. At the other extreme towards achieving universal diversion of even nominal VAT revenues from the pension coverage, is to identify an aggregator that Treasury coffers into Ejo Heza accounts would mean a has the ‘observe and auto-deduct’ capacity like co‑ slight drop in immediate capital expenditures. But as operatives, but for the whole Rwandan economy. the economy moves from cash to digital the overall The only such aggregator currently is the tax authority VAT revenue is likely to go up significantly, thereby not and consumption taxes (VAT revenues) is a proxy indi- hurting cashflows. The views of the subscribers were cator of individual consumption and paid indirectly by elicited in the phone survey through a question on almost all individuals. If all VAT revenues were to be whether individuals would save (or save more) if the digitized and traced back to individuals via National ID, government were to automatically contribute tax rev- then one could fathom a diversion of a nominal per- enue into individuals’ accounts. About 60 percent of centage of VAT revenues into pension accounts for respondents said they will most likely save and an- individuals. In doing so, the state would effectively other 27 percent saying they might continue to save. ‘mandate’ consumption smoothing for the individual Only 4 percent suggested they might not continue to but without imposing any additional tax on them. A save (Figure 26). FIGURE 26.  Reported willingness to save more if the government automatically contributes some tax revenue directly into your account Yes, I will most likely save 60% I might continue to save 27% No, I will not save 1% I might not continue to save 4% I am uncertain 7% 0 10 20 30 40 50 60 70 Percent Source: Phone survey results, 2022. 46  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Rwanda has already implemented a green policy meet the minimum savings of RF 15,000 for U1 and U2; (National Strategy of Transformation, NST1), aiming RF 18,000 for U3; and RF 72,000 for U4, they are eligible to be a cashless and paperless economy by 2024 for the matching contributions as well as for the life (Deutsche Welle 2020). The digital advances, strategic and funeral insurance benefit. Administrative data vision, success with CBHI, and ingenuity shown in case shows that 52 percent of all savers in 2022 were in the of Ejo Heza havealso shown that Rwanda makes bold age group 26-45 and the rate of dormancy (or inactiv- policy decisions. The phone survey results are a pre- ity in contribution payment) was lowest among older liminary indication that members might not be averse adults (Figure 27). The lack of any age restriction cre- to this idea, but further consultations are warranted to ates two unintended challenges for the scheme. test the feasibility of this proposal. The supply side chal- lenges, notably having the right technology and adher- First, premia charged by the Sonawara, the life insur‑ ence to data privacy rules, are also important consider- ance company, are higher than global estimates of ations that need to be carefully thought through. similar schemes, as there is a risk of adverse selec‑ tion. As of 2021, 22 percent of savers were above the pension eligibility age of 55. To be eligible for the insur- Eligibility restrictions based ance benefits they only need to save the minimum on age and the state pays the premium on their behalf, which is equivalent to the annual minimum savings for Ubu- From the outset, the policy vision of Ejo Heza was to dehe category 3 (RF 18,000). Comparing the life insur- be an inclusive scheme for all Rwandans and for‑ ance benefit offered by Sonarwa in Rwanda to a similar eigners residing in Rwanda. This is why there were no program offered by the Life Insurance Corporation in any age restrictions to join the scheme. If members India, called the PMJJBY (Pradhan Mantri Jeevan Jyoti From the outset, the policy vision of Ejo Heza was to be an inclusive scheme for all Rwandans and foreigners residing in Rwanda FIGURE 27.  Age distribution of Ejo Heza savers and non-savers 350,000 300,000 250,000 200,000 Savings (RF) 150,000 100,000 50,000 0 0-18 19-25 26-35 36-45 46-55 56-65 65+ Age group Dormant members Active savers Source: Phone survey results, 2022. Section 4: Ejo Heza Member Preferences to Select Design Changes n  47 In keeping with the vision of policy makers to keep Ejo Heza age-inclusive, an individual of any age could be allowed to save, but the life and funeral benefit could be provided only to savers below a certain age Bima Yojana):43 PMJJBY has an age limit of 18-50 and the There is a risk that these individuals would have unmet subscriber pays a premium of INR 436 ($5.35) in PMJJBY expectations from the scheme. If the pension fund has for a cover of INR 200,000 ($2,453). This scheme is in- a growing proportion of older adults, it will also need tended for a similar cohort of unorganized workers (in- to adjust its investment policy and keep more cash in cluding rural and agricultural workers). In Ejo Heza, a hand to be able to make lump sum payments on retire- premium of RF 18,000 ($16.79) is charged for a total cover ment. This could hurt its ability to invest in long-term of RF 1,250,00044 ($1,166.3). The rate per thousand for securities that provide higher and stable returns. the Ejo Heza scheme ($1,166/$16.7 = $68.5) is approxi- mately 7 times that for the PMJJBY scheme in India. In keeping with the vision of policy makers of keep‑ Allowing for some age restrictions when accessing the ing Ejo Heza age-inclusive, an individual of any age life insurance benefit can allow MINECOFIN to achieve could be allowed to save, but the life and funeral globally competitive rates as insurance companies. benefit could be provided only to savers below a certain age. There could also be a requirement of pay- Second, individuals who start saving at older ages out being allowed only after a minimum of five years (for example, age 52) will still be eligible to withdraw of membership or after reaching savings of RF 4 million at 55 but the savings would not have grown by much. (whichever occurs sooner). 43. https://financialservices.gov.in/insurance-divisions/Government-Sponsored-Socially-Oriented-Insurance-Schemes/Prad- han-Mantri-Jeevan-Jyoti-Bima-Yojana (PMJJBY). 44. RF 1,000,000 death benefit + RF 250,000 for funeral expenses. SECTION 5 Viability Assessment of the Ejo Heza scheme V iability assessment of a voluntary scheme like Ejo Heza enables policy makers to understand when a scheme would become financially viable— that is, when the revenues into the scheme might begin to pay for the expenses of operating the scheme. The answer to that question depends in large part on how many individuals the scheme is expected to attract, what their expected level of contributions are, what the operational and asset management costs of the scheme, and so forth. For an existing scheme like Ejo Heza, the assumptions can be based on historical trends and projections based on grassroots experience. The viability assessment for Ejo Heza was carried out by the World Bank team using a tool (SVAT)45 customized for the scheme at the request of RSSB. The assessment was carried out using administrative data as of December 2020. The scheme had been imple- mented for approximately two years at the time of this assessment, so it was still a young scheme and had less than half of subscribers (789,900) it has currently. The matching con- tributions by MINECOFIN were due to expire in 2021 and the SVAT analysis was used to estimate the likelihood of the scheme’s viability under four different scenarios, one of which included a continuation of the matching contributions for three more years, until 2024. The viability assessment used for Ejo Heza was designed to assess the financial viabil- ity of the scheme, but its outputs can also be customized to look at adequacy indicators like expected fund balance on retirement. The SVAT model for Ejo Heza made revenue and expenditure projections for the scheme from 2020 to 2060. The revenues of the scheme were projected by making an assumption on the number of contributors who are expected to join the scheme (based on take-up rate distribution), the average annual contributions individuals are likely to pay, and the average government matching they would be eligible for given scheme rules (Fig- ure 28). The detailed administrative data as of 2020 shared by Ejo Heza was used to cali- brate the distributions used for the model and the assumptions of coverage increase over time informed the scale up assumption over time. It is worth noting that the Ejo Heza team targeted 2.5 million active savers by the end of 2025, and that the scheme already has 1.4 million active savers as of December 2021.46 Once contribution revenue is calculated, the investment return net of management fees is applied to calculate the total funds accumu- lated into the scheme, or assets under management (AUM). The model assumes that, in the baseline case, 2 percent of the AUM is set aside to cover the scheme’s expenditures. This is also referred to as the ‘cost recovery’ and is the only revenue source for the scheme. 45. The Scheme Viability Assessment Tool (SVAT) was developed by the World Bank Social Protection and Jobs Team in collaboration with World Bank Treasury Pension Department to assess the viability of long-term voluntary schemes for the informal sector. The tool can be used to assess viability of an existing scheme as well as provide guidance for policy makers where such a scheme is considered. 46. The target as of 2021 for the model was assumed to be 1,070,000. 48 Section 5: Viability Assessment of the Ejo Heza scheme n  49 FIGURE 28.  Equation used to project Ejo Heza scheme revenues in the SVAT Targeted Average Average Gvt Demographic Take-up rate Take-up contribution by matching Projection distribution by rate age and gender contribution by (PROST) age and gender age and gende gender Calibrated on Individual data Source: World Bank staff's schematic depiction of calculations in SVAT. Note: PROST is the World Bank pension tool used for actuarial projections of mandatory pension schemes. The baseline scenario for the viability assessment of Ejo Heza projected the scheme to break even by 2025 under what was then a realistic set of assumptions The scheme is deemed to break even when the cost The baseline scenario for the viability assessment of recovery or revenue of the scheme is sufficient to Ejo Heza projected the scheme to break even by meet the operating costs of the scheme. The scheme 2025 under what was then a realistic set of assump‑ is fully funded by design as it is a defined contribution tions. The World Bank team in conjunction with Ejo scheme. Members are only expected to receive their Heza assumed the baseline scenario to be one where contributions plus investment returns plus govern- government matching continued only until 2021, aver- ment matching if any, at the payout phase. The incen- age savings increased by inflation, the scheme earned tives under the scheme namely the matching contri- a ‘real’ investment return of 5 percent and had a cost butions and the premium to the life insurance company recovery of 2 percent.47 The World Bank team noted are both paid by MINECOFIN. The expenditures to the that since the scheme was growing rapidly it was nec- scheme are therefore limited to operational expenses essary to have real time (at least annual) updates so only. These include salary costs, marketing cost (fixed), that the most recent count, savings, matching contri- acquisition cost per customer (variable), and fixed cost bution, and investment data are input in the SVAT of updating IT hardware once every 10 years. Similar to model at all times. A limitation of the model was that revenues, the expenditures are expected to rise over it did not take into account any endogenous effects time as the number of members grows, with econo- that could affect scheme viability (for example, con- mies of scale for fixed costs. It should be noted that tinuing or discontinuing matching contributions could the results would look very different (and breakeven impact take-up rates and savings). These effects were would occur at much later stage) if the costs currently not included in the model as they were hard to quan- incurred by MINECOFIN were to be included in the tify without a rigorous impact evaluation that mea- scheme’s viability analysis. sured the impact of matching contribution on likeli- hood of saving and level of saving. An extension of the 47. The year of breakeven and amount needed to finance the scheme was found to be most sensitive to the cost recovery assumption. Higher cost recovery however means lower account balance for participants. It was recom- mended that after the scheme reaches breakeven, the cost recovery-percent be reduced to improve adequacy of benefits. 50  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans TABLE 6:  Summary of baseline and scenario results from Ejo Heza’s viability assessment carried out using December 2020 data Breakeven Cumulative net AUM by 2040 Scenario year cost (in million) (as % of GDP) Baseline scenario 2025 4,178 3.0% Scenario 2: Government funded matching 2024 3,784 3.1% contributions continued till 2024 Scenario 3: Cost recovery reduced by 1% 2028 10,084 3.2% Scenario 4: Low coverage growth (only 1.5 million 2026 5,921 1.8% active savers as of 2025) Scenario 5: 20% Higher savings from contributors 2025 3,898 3.5% Source: Calculations by WB team using the SVAT tool. model that was considered was to incorporate with- expected to cost more to the government as it would drawal patterns. In 2020, the scheme was too early for need to finance the operating costs of the scheme. withdrawals and balances were low; but over time they are both expected to rise and could impact the The SVAT model was customized for Ejo Heza based scheme’s viability and finances, and hence they are im- on its features and allowed policy makers to under‑ portant to observe and model. stand the impact that policy decisions and take-up/ saving rates by members can have on Ejo Heza’s fi‑ Four sensitivity scenarios were carried out to illustrate nancial viability. It is recommended that Ejo Heza con- the impact of a change in assumptions on break-even duct at least one expected versus actual analysis every year, deficits, and AUM. The results in baseline de- year to test validity of assumptions and recalibrate the pended on the validity of assumptions used and the model if need be. This is especially important in the scenario analysis carried out was used to help policy initial years of the scheme as it grows, and the profile makers understand the range of possibilities (Table 6). of contributors changes rapidly. The model itself can Scenario 2 where matching contribution was extended be made more robust if panel data is maintained so for 3 more years relative to the baseline, was projected that persistency of contributions and later withdrawals to cost less (cumulatively) as break even was to be can be analyzed. The high inflationary environments achieved sooner. A reduction in cost recovery (Sce- since 2020 onwards also poses a risk on the fund’s ability nario 3) relative to the baseline meant higher returns to generate stable real returns and a scenario analysis for members but delayed break even and was with lower real returns is recommended. SECTION 6 Policy Recommendations E jo Heza, a voluntary long-term saving scheme launched with the intent to of‑ fer retirement protection to the vast majority of informal sector has been an inspiration for policy makers and practitioners globally. In the four years since its launch in November 2018, it has subscribers equal to 27 percent of the working age population. The need for innovative and scalable solutions to provide social insurance coverage to the informal sector has been noted by international institutions since 2000s, made more urgent post COVID-19 (Guven et.al. 2021). Under the universal health coverage (UHC) agenda, more countries have had success in scaling up health insurance in large part because of government subsidies. Rwanda’s health scheme for informal sector workers (CBHI) is yet again a widely cited example. The global south has been experimenting in approaches to expand coverage to the informal sector. The World Bank’s 2019 white paper called for a blurring of lines between social assistance and social insurance and for govern- ment resources to financially protect and prepare those most vulnerable for their old age (Packard et al. 2019). However, many countries face the reality of resource constraints es- pecially as they reel from the fiscal costs of providing subsidy in health insurance. Rwanda, through the Ejo Heza scheme, has demonstrated that political will, smart solutions, and the willingness to experiment via new partnerships can make financial protection in old age a possibility even for those outside the formal sector. Developing countries globally are experimenting with different designs and partnerships to scale up coverage among informal sector workers. As international experience grows, the strategies best suited to register more people, ensure persistent contributions, manage expectations, and meet stated goals, will become more apparent. For now Ejo Heza, for much of the developing world, is an active ground both for lessons learnt and experimentation (see Box 3). The vision of Ejo Heza continues to be bold with a desire to cover all those in the infor- mal sector. This report and the broader World Bank technical support to Ejo Heza is di- rected at supporting this vision via three key activities: (a) a thorough assessment of the scheme and documenting good practices/gaps by using the World Bank Delivery Chain assessment (b) using the phone survey to elicit member needs, preferences, and opinions on design changes to the schemes, and (c) showcasing how models like SVAT can be used to ascertain scheme viability in the long run. Section 4 gave recommendations on four significant design changes, most relevant to Rwandan policy makers. This section lists rec- ommendations for practitioners (grouped under seven themes) that would bring Ejo Heza closer to its vision, irrespective of any design changes. To meet its stated goals of expanding pension coverage to all Rwandans especially those in the informal sector Ejo Heza scheme will need to focus on improving scheme viability, improving customer experience, ensuring persistency in saving, continued coverage expansion, smoother operations and policy making. The recommendations in this section highlight (in green) the areas which each of them can help improve. 51 52  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans BOX 3  Key factors for Ejo Heza success & lessons for other schemes The report calls for rigorous impact evaluation to ascertain what factors might be responsible for Ejo Heza’s success but in the absence of such evaluations, feedback from numerous consultations, trends in administrative data, knowledge of initiatives attempted in other countries, and the phone survey responses of a random sample of 384 Ejo Heza subscribers offer insights. Below are some key factors that stand out: 1. A sound legal basis for the scheme and alignment with GoR’s strategic priorities- A key inter- vention under the Economic Pillar of the NST1 2017-2024 was to operationalize a long-term sav- ings scheme and pension for all Rwandans, including those in the informal sector as a basis to support long-term domestic investments. Following this strategic priority the GoR, through the MINECOFIN established EjoHeza Long Term Savings Scheme under Law N° 29/2017 of 29th June 2017. 2. Set targets at sub-national level and recognize those who meet them: This helps with increased ownership, higher accountability and leads to innovative campaigns by district level staff who want to be recognized. The targets themselves need to well be designed and the central office needs to support districts e.g., through a designated District Coordinator in case of Ejo Heza. 3. Build trust in the institution and political will: The Government of Rwanda and RSSB enjoy the widespread trust of their people which is a key advantage. In countries that might not have that, efforts need to be made, at the policy and program level, to establish trust and maintain it. 4. Identify aggregators to onboard subscribers: Aggregators who can ‘observe and auto-deduct’, like cooperatives, can serve as an effective substitute to employers for those in the informal sector. 5. Offer fiscal incentives: Especially in the initial stages when the scheme is likely to have opera- tional teething troubles, a monetary benefit is likely to get more people interested in joining. 6. Offer a tangible benefit for the short run: Benefits such as the life and funeral insurance lump sum benefit, which is currently unavailable to informal sector workers, are easily understood and feel like a good ‘bargain’ for locking away their money. 7. Balance flexibility in design: This would meet the needs of informal sector workers without acutely compromising the objective of retirement security 8. Be adaptable and agile: The informal sector is dynamic and has evolving needs, so a scheme needs to be designed to ‘meet the clients where they are found’. The traditional methods of social security organizations that cater to the formal sector are unlikely to be successful. 9. Leverage digital infrastructure like ID and payment to make registration, contribution and payments easy and efficient. It also makes the individual account portable across jobs, loca- tions and service providers, which is critical for informal sector workers who are mobile and would like to be assured that their modest savings are safe 10. Explore partnerships at the sub-national and civil society level: Ejo Heza’s partnership with mayors, CARE VSLA, Rwanda Cooperative Authority, and volunteer village level mobilizers helped with mobilization efforts. 11. Ensure clear, consistent and intuitive communication: This can be done through leaflets, video, radio messages, call centers, and district coordinators who are knowledgeable enough to answer questions and are committed to the scheme’s success (through monetary or non- monetary incentives). Section 6: Policy Recommendations n  53 1. Communications: Keep doing what’s working, fix the gaps, and explore new avenues A communication and mobilization strategy is key to the continued success of the scheme. Communication teams in Ejo Heza need to identify what is working (do more of it), hear from members (fix the gaps), and not hesitate to be creative (explore new avenues). The objective should be to work hard behind the scenes to keep it simple for members. What it can help with Improve scheme Customer Persistency Coverage Policy Smoother Recommendation viability experience in saving growth making operations i. Reshape the narrative on why people should save in Ejo Heza— steering away from incentives to highlighting prospects of a pension, and competitive and safe returns ii. Recognize and develop different targeting and communication strategies for urban/rural; USSD/ smartphone users; women/men, individuals with/without phones. iii. Create <1min ‘how to’ videos that can be shared on WhatsApp/ YouTube and show how to register, contribute, check balance, and get enrollment information. iv. Build a community of all mobilizers (paid or volunteers) and make them feel part of one ‘Ejo Heza family’a by holding quarterly meetings and awarding tokens of appreciation. v. Engage with other aggregators who might have an ‘observe and auto- deduct’ capability especially in urban areas. vi. Adopt a ‘Tech with touch’ motto to avoid excluding those with limited digital access (women, children, low- income members) and in vulnerable situations (after death of a member, or disabled). vii. Create ‘digital contracts’ using information at registration stage that can be accessed by members with a #text and can serve as proof of membership (see Annex 2 for suggested format). a. Interviews with volunteers CARE VSLA agents revealed that they are glad to volunteer as they see Ejo Heza as a scheme that will help their com- munity, but they would appreciate if Ejo Heza team gave them caps, t-shirts, umbrellas, and pens, for example, to recognize their efforts and help them stand out in the community. 54  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans 2. Focus on improving adequacy, persistency, and preparing for withdrawals/payouts The focus of the scheme so far has been on coverage expansion (and rightly so). The strategy used for coverage expansion in Ejo Heza by identifying aggregators like cooperatives and schools is a unique feature of the scheme. It allows the scheme to target the ‘low hanging fruit’ as far as coverage expansion is concerned. Further, group characteristics of aggregators guarantees persistency and ef- ficiencies in communication. Such efforts are laudable but given that Ejo Heza is a long-term saving scheme, a strategy to improve adequacy, persistency will be needed to ensure pensions at minimum provide protection against poverty in old age. There is a risk with not diversifying to other target groups notably urban dwellers who are likely to have a higher ability to save than those in co- operatives. Not managing expectations can also lead to reputational risks not because of design faults but due to low or irregular contributions from the members. Lower income individuals in Ejo Heza are more likely to save less and in turn receive low pensions. Lastly, the scheme is currently young so payout and withdrawals are limited but the frequency will go up as the scheme matures. The administrative costs will need to be accounted for and scheme viability and a seamless process of payouts will be crucial for maintain- ing attractiveness of the scheme. What it can help with Improve scheme Customer Persistency Coverage Policy Smoother Recommendation viability experience in saving growth making operations i. Revisit targets for district using objective criteria to reflect socio- economic differences by district. ii. Assess and manage expectations of members on expected fund balance in old age given their saving levels. iii. Enhanced efforts to target urban dwellers iv. Highlight the above inflation, competitive and stable returns in the communication material and in-person visits. v. Explore potential for ‘auto deduction’ in partnership with Mobile Money operators and Banks.a vi. Promote the registration of spouses and children of existing members. vii. Improve system readiness to process withdrawals, pension payouts and evaluate their cost/ investment implications. a. A pilot where individuals are tracked for six months or longer can be carried out to assess individual’s reaction to auto-deduction and if saving patterns are any different. Section 6: Policy Recommendations n  55 3. Revisit incentives as ‘experience’ on the scheme develops The limited successful global experience around voluntary schemes means that the Ejo Heza scheme will need to ‘learn by doing’. This report has attempted to bring in international experience which might be relevant for Ejo Heza (notably from China, India, Co- lombia) but the learnings from Ejo Heza itself will be the most valuable for improving the scheme. Scheme practitioners should use data, empirical analysis, and the vast community network as their guide to inform policy making. What it can help with Improve scheme Customer Persistency Coverage Policy Smoother Recommendation viability experience in saving growth making operations i. Assess (empirically) whether RF 4 million is a hard to meet target for most Rwandans. ii. Renegotiate insurance premiums after the age eligibility restriction, based on scheme experience, comparable international benchmarks. iii. Consider incentives for agents/ intermediaries based on a cost- benefit analysis and experience from countries. iv. Pilot other short-term incentives to members, for example, accident insurance, maternity benefits to savers in Ejo Hezaa or points in a grocery store/mobile money top- ups/chance to win a lottery if one saves persistently. a. Colombia’s experience can be relevant here where 1 percent of the savings is diverted to life insurance company in exchange of lump sum benefits in case of sickness, death (maternity benefit is under consideration to being included) 56  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans 4. Improve interoperability with other government systems and integrate with private sector Making it easy to save and view benefits stands out as critical in ensuring that Ejo Heza meets its vision of extending pension coverage to all those currently excluded. The scheme needs to work seamless across systems. Rwanda has an exemplary network of founda- tional systems and with the policy direction of becoming a cashless economy by 2024, strengthening social registries, and the willing- ness of the private sector to collaborate, Ejo Heza is well suited to scale up if it can create a web of integrated systems. What it can help with Improve scheme Customer Persistency Coverage Policy Smoother Recommendation viability experience in saving growth making operations i. Ensure real time updates on Ubudehe category of members by registering on ‘messaging queue’ of LODA MEIS.a ii. Integrate Ejo Heza platform with banks, SACCOs, mobile money operators so that they can be intermediaries (similar to cooperatives) who register and collect contributions. iii. Integrate with IREMBO so individuals can at minimum submit claims/check balance. iv. Integrate the system with RCA to allow for sharing of ‘real time’ information on cooperatives. v. Integrate with the provider of insurance benefit (Sonarwa currently) so that Ejo Heza teams can see status of application and reason for delay. vi. Review and ensure that data privacy and security controls are in place for registration through intermediaries. a. Interviews with LODA IT team revealed that CBHI is currently registered on the messaging queue so this service can easily be added to Ejo Heza on payment of a fee to LODA. Section 6: Policy Recommendations n  57 5. Strengthen learning, monitoring, and evaluation The Ejo Heza team has demonstrated an agility that other voluntary schemes and even formal sector schemes (including the scheme at RSSB) can learn from. The scheme is, however, now at a scale where strengthening learning, monitoring and evaluation are impor- tant. The field presence of Ejo Heza leadership and constant communication with district coordinators has allowed the team to be in tune with what members want and the challenges they face. As numbers grow and the team expands, data will need to do more of the talking. Consultations will continue to be critical, but the M&E systems will alert scheme administrators of issues at a national level and by sub-group as well as theme (coverage or persistency or claim challenges). The M&E systems should be designed to pre- serve the agility that embodies Ejo Heza—the feedback from the members should allow strategies to speedily adapt to the emerging realities of the field. What it can help with Improve scheme Customer Persistency Coverage Policy Smoother Recommendation viability experience in saving growth making operations i. Generate quarterly reports on M&E indicators for management (see Annex 1 on suggestions). ii. Redo the SVAT analysis based on new data and take into account withdrawal patterns. iii. Organize study tours with countries that have relevant experience to share. iv. Engage with researchers to carry out Impact Evaluations. 58  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans 6. Build human resource capacity and leverage external expertise as needed Ejo Heza was set up almost like a start-up with Access to Finance Rwanda as the sponsor for the first two years, a consulting firm to do the market scoping, and a dedicated team with experts from MINECOFIN that led the development. In hindsight, this could be a major factor behind the agility shown by Ejo Heza (missing from other schemes set up by social security institutions globally) but the rapid scale of the scheme will now also need to be supported by commensurate resources. Strengthening internal capacity, attract- ing/retaining talent, and leveraging external expertise as needed (from partners or consulting firms) is recommended. What it can help with Improve scheme Customer Persistency Coverage Policy Smoother Recommendation viability experience in saving growth making operations i. Evaluate the human resource policy for Ejo Heza team members with attention to staff shortage, capacity constraints, and short-term nature of contracts. ii. Recognize and reward better performers at the HQ, district, and sector level. iii. Build technical capacity and review staffing of call center. iv. Engage in regular training of agents (paid and volunteer). v. Adapt systems and build capacity to allow ‘intermediary logins’ to be generated at the district level. vi. Hire a PR firm to work with the communications team to craft tailored and creative messages targeting children, urban workers, women, and youth (age 16-30). vii. Hire a statistical expert to analyze data, design research questions for survey, monitor call center staff during survey. Section 6: Policy Recommendations n  59 7. Collect better data and mine it for policy making A clear understanding of enrollment and contribution patterns among informal workers remains severely limited by the data environ- ment. Ejo Heza has developed a short registration and enrollment intake form that makes it easy for individuals to open their ac- counts. Adding a few optional questions could provide valuable insights on outreach strategies that are working. Further, collabora- tion with the National Institute of Statistics of Rwanda would allow RSSB administrative data (not only for Ejo Heza but also for other schemes) to be complemented with household/labor force survey data with the intent of getting more policy-relevant insights on the population. Rwanda can leverage its “exemplary identification ecosystem” (World Bank 2016) to produce the type of linked ad- ministrative and survey data sets that have been instrumental in pension reform design in a country like Chile.48 48. Encuesta de Protección Social: https://www.previsionsocial.gob.cl/sps/investigadores/estudios-previsionales/encuesta-de-proteccion-social-2 What it can help with Improve scheme Customer Persistency Coverage Policy Smoother Recommendation viability experience in saving growth making operations i. Building a ‘richer data set’ by pulling variables from other government databases as relevant. ii. Adding a question at time of registration such as ‘Who helped you register’? iii. Ensure that the drop down under ‘Occupation’ at the time of registration includes the same categories as Rwanda’s household survey data to allow for easier mapping. iv. Analyze panel data of participants with policy relevant questions in mind. v. Propose a module on social insurance in the next round of ECIV (HHLD) survey by collaborating with the National Institute of Statistics of Rwanda. vi. Add indicators in the IT dashboard (see Annex 1 for suggestions). vii. Add a question on cause of deatha to build ‘experience ratings’. a. Assuming Sonarwa already collects this data, the Ejo Heza IT team should ensure they get this data from Sonarwa (ideally through interoperable data system) as this information can be used to develop the ‘experience’ of the scheme that actuaries then use it to assess premium fairness. 60  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Annex 1: Sampling Frame of Ejo Heza Phone Survey The Phone survey was designed to test five hypotheses: • The reason why Ejo Heza members do not save consistently in Ejo Heza is the lack of financial literacy aware- ness or a culture of savings. • Ejo Heza members value short-term benefits more than generous longer-term benefits. • Ejo Heza members do not think saving for old age is important. • Ejo Heza members understand that the pension they will receive from Ejo Heza depends on how much they contribute today and how regularly they do. • Ejo Heza members have trouble accessing information related to Ejo Heza. The survey only includes those who are in the Ejo Heza database, but the results can be generalized to the entire population as the sample is a ‘representative sample” of 2 million (30 percent of labor force) and good enough for generalizability and internal validity. SURVEY SAMPLING 1. The phone survey was done with 386 Ejo Heza members who were randomly selected from the pool of total Ejo Heza members. Currently Ejo Heza has details of 2 million members so a good sample size at 95 percent confidence interval is = 386 interviews. Formula used for calculating Sample Size = (2×(1−)/2) / (1+ ((2×(1−)/2N)) where N=2 million, p=0.5, z=1.96, e=5 percent 2. Population: All Ejo Heza members who registered before December 2021 (this includes people who have contributed once, more than once or never contributed and only registered). The reason why the popula- tion is being limited to those who registered before December 2021 is because matching contributions are paid once every four months so it is likely that someone who registered in 2022 would not have received the matching contribution at the time of this survey and the questions on matching would not be as rele- vant for them. 3. Respondent: Out of all those who have been in the Ejo Heza database since December 2021, the sample was stratified based on gender (male/female), age group (below 40/above 40), and contribution status (ac- tive/dormant49). 49. By active we mean at least one contribution between July-December 2021, by dormant we mean no contribution between July-December 1, 2021). Annexes n  61 Annex 2: Indicators Recommended to be Monitored and Included in IT Dashboard Coverage and persistency indicators 1. Number of active contributors—disaggregated by broad age group (e.g. age 15-25, 25-35 etc.) and gender (active refers to those who have paid contributions consistently in the last three months). 2. Number of new contributors—gender disaggregated (‘new’ refers to new joiners to the scheme). 3. Number of dormant contributors—gender disaggregated (‘dormant’ refers to those who are part of the scheme but have not paid any contributions in the last quarter). 4. Number of registered contributors by Ubudehe category and occupation and by gender (‘registered’ refers to all those whose data is in the system. If all data is correct, then at any given point in time number of registered >= number of active. If there are no dormant contributors, then number of registered will be equal to number of active). Savings capacity indicators 1. Total Contributions—split into individual and matching contributions, as of the date when report is gener- ated for e.g. March 31st, June 31st, Sept 31st, Dec 31st if reporting data quarterly. 2. Total reserves of the fund, as of the date when report is generated. 3. Average individual savings by age group, gender, and Ubudehe category (age group can be any broad age classification 0-15, 15-19, 20-24 etc.). 4. Median individual savings by age group, gender, and Ubudehe category. 5. Percentage of members eligible for matching contributions by age, gender, Ubudehe category. Investment indicators50 1. Most recent asset allocation along with expected returns in a pie chart. 2. Nominal returns from Ejo Heza investments (using data from AUM). 3. Nominal returns credited to member accounts (using individual data)—an analysis of individual records showed that the average returns credited to individuals were only 4 percent lower than what one would expect given the assumptions of cost recovery. This could be because there are delays in crediting gains from investment to member accounts, but this should be monitored, and attempts made to quickly transfer realized gains to member accounts). Data and service quality 1. Percentage of individuals in the system who are missing Ubudehe categories by gender. 2. Percentage of Individuals in the system who are eligible for matching but whose accounts have not yet been credited as yet. 3. Number of calls (and type of calls) to call center or some indicator to understand rates of grievances and what they are. 4. Expenses related to ‘communication and awareness initiatives’ in the last quarter. 5. Percentage of members who used mobile money to pay contributions (a pie chart can also be created to show the delivery channel(s) being used by members for e.g. cash payment in office, mobile money, money order etc.). 6. Percentage of members who inquired about their fund balance (using their mobile phone or in person). Withdrawals and other benefits 1. Percentage of individuals by Ubudehe category who claimed additional benefits like insurance. 2. Expenditure on additional benefits. 3. Withdrawal51 count by broad age (e.g., 15-25, 25-35...) and gender. 4. Average Withdrawal amount by broad age (e.g., 15-25, 25-35…) and gender. 50. Pension funds are recommended to develop a detailed investment performance report, at minimum on a monthly basis, with monthly, cumulative and annualized returns by asset class. They are also advised to look at benchmarks (or the target port- folio), some risk metrics and indicators of the actual performance versus the benchmark (excess of return, Sharpe ratio, etc.) 51. Withdrawal could be because the individual reaches age 55 or has accumulated in excess of 4 million RF and is hence eligible to withdraw up to 40 percent of excess. 62  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans Annex 3: Design and Messaging Recommendations for a Digital Contract Box 3: Design and messaging recommendations on scheme contract A • Clear information regarding EjoHeza details in succinct form EjoHeza Scheme Contract • Authenticity via signature/seal to certify the document A JOHN SILAS 14 JUNE 1974 MONTHLY NAME OF SUBSCRIBER DATE OF BIRTH CONTRIBUTION FREQUENCY • Simple depiction of what the subscriber gets and what he/she does not CATEGORY 2 14 MARCH 2018 A33467678 B • Illustration of benefits that can be personalized to the subscriber as per his/her UBUDEHE CATEGORY REGISTRATION DATE CONTRACT NUMBER Ubudehe category 1234567890 EJOHEZA SUBSCRIBER ID • Clear call to action on how the subscribers can receive more information C • Both USSD and Inbound call center numbers referred EXCLUSIONS WHAT DO YOU • Branding of RSSB to leverage trust and authenticity GET B • Government co- • If you save < 15,000 RwF contributions if your there will be No contributions exceed 18,000 RwF government co contributions Design Principles • Life and funeral expenses (based on eligibility • Life and funeral expenses will not be available if you • Messaging can be family oriented to highlight collective benefits of long term savings criteria) contribute < 15,000 RwF • Usage of simple and uncluttered messages, with complex information being • When you are 55, you will • Exclusion 3 communicated in infographic/visual form get back savings in monthly form • Exclusion 4 • Focus themes on safety and stability • With 10 years of • Focus the messages on meeting needs of the family and how the EjoHeza contributions contribution, you will can help receive Rwf 30,000 per month when you are XX • Leverage the communication channels that are available and provide the relevant proofs/documents in SMS formats as well Dial *506# for any queries C • Graphically depict legal information like Eligibility criteria, Terms and Conditions and Call XXXXXXXXXX for more support user consent. Annexes n  63 Annex 4: Contribution Targets by District for 2021/22 64  n   The Promise of Ejo Heza: A Brighter Future for All Rwandans References Rwanda Social Security Board (RSSB). 2021. Report on Impact of Government Incentives. May 2021. RSSB, Kigali. Rwanda Social Security Board (RSSB). 2022. Ejoheza Presenta- tion and Data. RSSB, Kigali. Deutsche Welle. 2020. “Rwanda to Start New Cashless Era Singh, Dalip. 2022. “Haryana Transferred ₹100 cr to Ghost Pen- after COVID-19.” Nature and Environment section, July 3. sioners in 2017–20.” The Hindu Businessline, August 9. https://www.dw.com/en/rwanda-to-start-new-cashless- https://www.thehindubusinessline.com/news/national/ era-after-covid-19/a-54038132. haryana-transferred-100-cr-to-ghost-pensioners- in-2017-20/article65751525.ece. Guven, Melis (2019). Extending pension coverage to the infor- mal sector in Africa. Social Protection & Jobs Discussion Stewart, Fiona Elizabeth, Himanshi Jain, and Will Sandbrook. Paper No. 1933. Washington DC: World Bank. 2019. “Early Access to Pension Savings: International Experience and Lessons Learnt.” World Bank, Washington, Guven, Melis, Himanshi Jain, and Clement Joubert. 2021. Social DC. https://documents.worldbank.org/pt/publication/ Protection for the Informal Economy: Operational Les- documents-reports/documentdetail/7570015517 sons for Developing Countries in Africa and Beyond. 15193169/early-access-to-pension-savings-international- Washington, DC: World Bank. https://openknowledge. experience-and-lessons-learnt. worldbank.org/handle/10986/36584 License: CC BY 3.0 IGO. Republic of Rwanda. 2019. “7 Years Government Programme: National Strategy for Transformation (NST1) 2017–2024.” Jain, Himanshi, and Robert Palacios. 2022. “Assessment of https://www.nirda.gov.rw/uploads/tx_dce/National_ VSS’s Voluntary Pension Scheme and Policy Recommen- Strategy_For_Trsansformation_-NST1-min.pdf dations.” World Bank, Washington, DC. https://docu- ments1.worldbank.org/curated/en/099000105132210101/ United Nations. 2019. World Population Prospects 2019. Unit- pdf/P1742580d88ccc0f1093e10fc126c222f5c.pdf. ed Nations, Department of Economic and Social Affairs, Population Division. New York, NY: United Nations. Kalisa, I., S. Musange, D. Collins, U. Saya, and T. Kunda. 2015. https://population.un.org/wpp/Publications/. “The Development of Community-Based Health Insur- ance in Rwanda—Experiences and Lessons.” University of World Bank. 2016. “ID4D Country Diagnostic: Rwanda.” World Rwanda College of Medicine and Health Sciences— Bank, Washington, DC. License: Creative Commons Attri- School of Public Health, Kigali, Rwanda and Management bution 3.0 IGO (CC BY 3.0 IGO). Sciences for Health, Medford, MA, USA. World Bank. 2019. Rwanda Systematic Country Diagnostic. Lindert, Kathy, Tina George Karippacheril, Inés Rodríguez Cail- World Bank, Washington, DC. © World Bank. https:// lava, and Kenichi Nishikawa Chávez. 2020. Sourcebook on openknowledge.worldbank.org/handle/10986/32113 Li- the Foundations of Social Protection Delivery Systems. cense: CC BY 3.0 IG Washington, DC: World Bank. © World Bank. https:// World Bank. 2020. The Human Capital Index 2020 Update: openknowledge.worldbank.org/handle/10986/34044 Li- Human Capital in the Time of COVID-19. World Bank, cense: CC BY 3.0 IGO. Washington, DC. © World Bank. https://openknowledge. National Institute of Statistics Rwanda (NISR). 2020. Labour worldbank.org/handle/10986/34432 License: CC BY 3.0 Force Survey Annual Report. NISR, Kigali. IGO. Omondi, Michael. 2019. “Treasury Headcount to Catch Ghost World Bank. 2022a. “Adapting Social Protection & Jobs to an Pensioners.” Business Daily, January 23. https://www.busi- Actively Aging World. Social Protection & Jobs.” Brochure. nessdailyafrica.com/bd/economy/treasury-headcount- World Bank, Washington, DC. https://thedocs.world- to-catch-ghost-pensioners-2235850. bank.org/en/doc/9c1462c00ab0d8dc0ed712c58964c 9c9-0200022022/related/SPJ-Aging-Brochure-World- Packard, Truman, Ugo Gentilini, Margaret Grosh, Philip Bank.pdf. O’Keefe, Robert Palacios, David Robalino, and Indhira San- tos. 2019. Protecting All: Risk Sharing for a Diverse and World Bank. 2022b. Charting a Course Toward Universal Diversifying World of Work. Human Development Per- Social Protection: Resilience, Equity, and Opportunity for spectives. Washington, DC: World Bank. doi: 10.1596/978- All. World Bank Group, Washington, DC. © World Bank. 1-4648-1427-3. License: Creative Commons Attribution CC https://openknowledge.worldbank.org/handle/10986/ BY 3.0 IGO. 38031 License: CC BY 3.0 IGO Rwanda Social Security Board (RSSB). 2019a. LTSS Survey Re- World Bank. 2022c. “Project Information Document (PID).” port 2019. RSSB, Kigali. Report No: PIDA33836. Rwanda—Social Protection Transformation Project (SPTP) (P177492). World Bank, Rwanda Social Security Board (RSSB). 2019b. Long Term Saving Washington, DC. https://documents1.worldbank.org/ Scheme (EjoHeza) Registration Process (Web Portal and curated/en/099705007092222871/pdf/P17749202cb8c USSD) User Manual. RSSB, Kigali. f0f095e701db4a222a664.pdf. Rwanda Social Security Board (RSSB). 2019c. Ejo Heza Annual Report Dec2018- Nov2019.