83145 KIÚTPROGRAM Social Microcredit Program Case Study June 11, 2013 The World Bank 1 ACKNOWLEDGEMENTS This report was prepared, in alphabetical order, by Robin Audy (Consultant, World Bank), Sandor Karacsony (Consultant, World Bank), Barbara Kits (Consultant, World Bank), Joost de Laat (Senior Economist), György Molnár, Senior Research Fellow, Institute of Economics, Hungarian Academy of Sciences, and Lea Tan (Consultant, World Bank). Sujani Eli (ECSHD Program Assistant) provided team support. Detailed comments by Roberta Gatti, Sector Manager and Lead Economist, ECSH4, and peer reviewers Ms. Aglaia Zafeirakou, Senior Education Specialist (HDGPE) and Ms. Sophia V. Georgieva, Social Development Specialist (ECSSO) are gratefully acknowledged. World Bank ECA Vice President : Phillippe H. Le Houerou World Bank ECCU5 Country Director : Mamta Murthi World Bank HD Sector Director : Ana Revenga World Bank HDE Sector Manager : Roberta Gatti World Bank Task Team Leader : Joost de Laat This report has been prepared for the European Commission; however, it reflects the views only of the authors, and the Commission cannot be held responsible for any use which may be made of the information contained therein. 2 TABLE OF CONTENTS ACKNOWLEDGEMENTS .................................................................................................................................. 2 TABLE OF CONTENTS ..................................................................................................................................... 3 1 Introduction ........................................................................................................................................... 4 2 Kiútprogram – Program Description ..................................................................................................... 6 2.1 Mission and Rationale.................................................................................................................. 6 2.2 Group Lending Concept ............................................................................................................... 7 2.3 Institutional Framework ............................................................................................................... 8 2.4 Daily Fieldwork Activities ........................................................................................................... 9 3 Kiútprogram Client Characteristics ..................................................................................................... 13 4 Program Operating Performance ......................................................................................................... 17 4.1 Loan Portfolio Performance ....................................................................................................... 17 4.2 Repayment and Client Characteristics ....................................................................................... 19 4.3 Program Costs Per Loan Issued ................................................................................................. 22 4.4 Client Feedback ......................................................................................................................... 23 4.5 Stakeholder Feedback ................................................................................................................ 26 4.6 Post-Pilot Contiunation .............................................................................................................. 26 5 Discussion ........................................................................................................................................... 28 3 1 INTRODUCTION This report is a case study of the Kiútprogram (‘Way-out’) financial services pilot project by the Polgár Foundation for Opportunities. Kiútprogram was a European pilot project 1 focusing on the provision of microcredit for Roma inclusion that operated in the most disadvantaged microregions of Hungary from June 2010 until December 2012. The objective of this case study is to learn from the practical experiences by the Kiútprogram pilot project in providing microcredit services and thus provide practical insights for policy makers and program managers designing financial services promoting Roma inclusion. Was Kiútprogram able to reach poor Roma households and, if so, with which specific lending approach and at what cost? How important were other, complementary, financial services provided by Kiútprogram? And, is there evidence that the access to microcredit provided a way out of poverty and into inclusion? Answers to these questions will provide guidance to the design of financial services programs aimed at promoting Roma inclusion. The insights – mainly operational – provided by this case study are based on a combination of qualitative and quantitative information. Qualitative information comes from interviews conducted with management, staff and its sponsoring financial institution, the policy lessons document prepared by Kiútprogram (2012) itself, as well as the analysis of client feedback data from Kiútprogram management information system (MIS). This MIS was developed in partnership with the World Bank and implemented by Kiútprogram field staff. It included a detailed survey with each potential beneficiary before the formal loan application (the ‘intake questionnaire’), which collected information on the socio-economic background characteristics, and a (proposed) business questionnaire. The MIS also included a second survey toward the end of the loan cycle with the actual clients, which focused on client feedback. Finally, the MIS collected loan repayment information and information on the time allocation across different tasks by field workers. As such, this case study does not aim to provide a rigorous assessment of the project impact on the beneficiaries. This was not feasible nor was the MIS designed with this objective in mind. It was impossible to design a rigorous prospective evaluation for several reasons. First, as described in greater detail below, it was a real challenge for the project to find qualified applicants, making a randomized evaluation design based on oversubscription impossible. Randomization at the community level might have been possible, but – mainly in the first phase of the project – with very few individuals per community ultimately interested and qualifying, such a design would have risked being underpowered. 1 For more information about the European pilot projects for Roma inclusion, visit http://ec.europa.eu/regional_policy/activity/roma/pilot_en.cfm 4 And, as a pilot project, the design evolved over the course of the project duration. Instead, the information used in this case study provides the basis for important insights – mainly operational - for the provision of microcredit and related financial services targeting poor Roma. This report follows the companion report “Reducing Vulnerability and Promoting the Self- Employment of Roma in Eastern Europe through Financial Inclusion” (World Bank, 2012), which provided an assessment of two complementary questions: first, whether Roma in Eastern Europe lack access to a wide range of basic financial services; and second, whether improving access to microcredit can be a means to substantially raise self-employment. Key findings from the companion report were based on the UNDP/World Bank/EC regional Roma survey (2012) survey data. The companion report finds that self-employment levels are relatively low, but there is considerable interest among Roma in becoming self-employed. Among working Roma, the share that is self- employed is low in the Czech Republic, Hungary, and Slovakia, while noticeably higher in Bulgaria and Romania. The vast majority of Roma express a desire for stable jobs, similar to the responses of non- Roma neighbors, and many express an interest in starting a business and becoming self-employed. However, the companion report finds that most Roma interested in becoming self-employed face significant obstacles that limit access to credit and which are not easily addressed by regular MFIs. Not many suppliers of microfinance target start-up businesses in general, and even fewer are reaching (starting) Roma entrepreneurs. A key reason is that many potential Roma entrepreneurs do not qualify for credit in light of lack of savings and indebtedness, very low levels of education, even when compared to the segment of the general population that is being refused credit, and little employment experience, especially lacking experience starting and operating a business in the formal economy. In light of this, and global experiences around the provision of financial services to the poor, the companion report concludes that a comprehensive approach to financial inclusion is needed that goes beyond strictly providing microcredit. This approach addresses first basic barriers, such as the limited outreach of financial service providers in Roma communities and the (related) lack of access to savings technology, but also addressing low financial literacy and business skills. The study documents these barriers and highlights various examples from Europe and elsewhere to achieve this. These include examples of programs providing families with incentives for targeted savings; e.g. saving for housing improvements, saving for secondary education, savings for preschool etc. Similarly, introducing non- stigmatizing and accessible electronic government-to-citizen payment systems can promote financial 5 inclusion. Finally, a ‘Graduation approach’ (CGAP-Ford Foundation) promotes business skills and grants/loans for self-employment, but after financial literacy and savings promotion. The Kiútprogram was defined by its founders as ‘social microcredit’, and aimed to enable the unemployed living in deep poverty – primarily, but not exclusively Roma – become self-employed. To achieve this, the Kiútprogram provided participants with not only financial services (importantly, unsecured microloans), but also related social services from counselling to physically joining applicants to local municipalities for the registration of their businesses. Results of the Kiútprogram indicate that facilitating the creation of sustainable enterprises among participants required intensive social work by well-trained field workers familiar with social work, lending, and business. Key features of Kiútprogram’s inclusive social microlending model were formulated as follows: i. a focus on participants’ employment (including self-employment) outcomes rather than profitability for the lender; ii. the potential to significantly increase inclusion and economic empowerment among participants by demonstrating the participants’ commitment to break out of poverty; and iii. the ability of the borrower to operate a business as a profitable investment in the formal economy. The remainder of this case study introduces the operational experience of Kiútprogram and attempts to draw broader policy lessons on the provision of microcredit that aims to reduce poverty and promote inclusion, especially among extremely vulnerable and frequently discriminated populations such as many Roma. 2 KIÚTPROGRAM – PROGRAM DESCRIPTION 2.1 MISSION AND RATIONALE The primary mission of the Kiútprogram was to facilitate Roma inclusion by assisting clients to establish sustainable formal businesses that connect with the economic mainstream. Disadvantaged areas of Hungary - the target area - are characterised by low levels of business activity, underdevelopment of infrastructure, low levels of investments and low share of the service industry. Previous attempts to generate employment opportunities in such areas have largely failed. Kiútprogram’s rationale was based on the assumption that the development of micro- and small businesses could be a viable step forward for 6 many, as there is no real alternative to entrepreneurship other than participation in public work programs. In the words of one client: I wanted to make my dreams come true and open my own pastry shop called 'Sweet Dreams'. The project started in 2010 in our region: that's when a field worker came to see us. The reason why I joined is because I was unable to create funds to start the business. I sell pizza, pastry and cake, and a lot of young people come to me to play pool and table soccer. There is no discrimination here; the customers are both Gypsies and Hungarians. This project can truly be an opportunity for the poor if they take it. I also have further plans, I want to expand the shop and build an outside sitting area and a bakery right next to the shop. 2.2 GROUP LENDING CONCEPT The concept of the Kiútprogram was inspired by Muhammad Yunus’ Grameen model operating in Bangladesh.2 Muhammad Yunus launched the model in 1976, providing unsecured loans of relatively small amounts to people living in deep poverty in order to enable income generation. He established the Grameen Bank in 1983 after processing and assessing the results of early initiatives. The Grameen model aims to provide financial services, loans and information to those living in deep poverty in order to enable them to utilise their own resources to rise from their disadvantaged living conditions. The primary prerequisite of success is mutual trust, in addition to the loan. The core of the lending activity is a voluntary, self-nominated group of five loan recipients where the traditional safeguards required by banks are replaced by joint liability. Lending in the Grameen scheme is sequential: group members receive their loans one after the other. If the first loan is repaid as agreed, the loan recipient may be granted another loan, which is larger in amount. The idea of adapting the Grameen model to Hungary was proposed by the Polgár Foundation for Opportunities in 2008. Kiútprogram was developed with an institutional structure adapted to the specific conditions of Roma inclusion. Loans were formally issued by Raiffeisen Bank; Kiútprogram administered lending operations as the bank’s agent. Extensive client selection and client support – described in more detail below – was a very time intensive process, going well beyond the Grameen model. Clients were recruited from a pool of potential entrepreneurs with incomes of less than 60% of Hungary’s median household income. Loans were administered to groups of 5-6 members through the use of sequential lending, i.e. a loan would first be provided to two members, then a second to two other members, and 2 See Yunus, M. (1999): Banker to the Poor. Microlending and the battle against world poverty. Public Affairs, New York. 7 finally the group leader would receive the loan. While loans were provided in group settings, they were not, however, jointly liable. According to Kiútprogram staff, the group-based approach increased participants’ self-confidence, and helped them learn more easily the skills necessary for building a social or business network. Indeed, as shown in the client feedback below, many clients (70%) valued this element. The loans had a 1-year tenure and a weekly repayment schedule, and 5% of the loan amount was paid to a common group fund. The maximum loan amount was set at HUF 1 million (approximately EUR 3,400 at the time). 2.3 INSTITUTIONAL FRAMEWORK Kiútprogram Microcredit Initiative Non-profit Ltd. was established by private shareholders as a not-for-profit shareholder group. Kiútprogram was responsible for program implementation, management and operations, as well as promotion of the microcredit product as an agent. Kiútprogram also trained and managed field workers; recruited and trained clients, as well as supported business development through mentoring; assisting the development of business plans; preparing, evaluating and submitting loan contracts; serving as a focal point with employment services, central and local government agencies and the financial institution; monitoring and evaluation; and program administration. Figure 2. Kiútprogram’s institutional framework EU DG Region Institutional structure Roma integration pilot project Largest donor POLGAR (1.43 M EUR) Foundation Supervision EU connections RAIFFEISEN Main shareholder BANK Civil connections Analysing social impact Loans and savings Loan guarantee World IT system KIÚTPROGRAM Bank/UNDP MFI Impact evaluation Monitoring support Advocacy Field work Accepting business plans EXPERTS, Client and program administration Government Loan agency SHAREHOLDERS Monitoring & evaluation Advocacy Quality control Feedback Networking Financing welfare bridge and other related costs Regulation, laws 8 Kiútprogram’s partners played pivotal roles. Kiútprogram’s partner financial institution, Raiffeisen Bank, took on expenditures and losses of lending up to HUF 20 million within the framework of the bank’s corporate social responsibility program. Raiffeisen Bank did not charge Kiútprogram clients commission or other banking fees for its services, and provided training to staff in regionally designated local branches. It also offered savings accounts and provided IT support. Polgár Foundation for Equal Opportunities, the main shareholder of Kiútprogram Non-profit Ltd., was the primary partner to the EU agreement. Polgár carried out project management and evaluation tasks. In addition, it was responsible for the dissemination of program results. Financing for Kiútprogram – both its operational costs and loan portfolio – amounted to approximately EUR 1.4 mln and were primarily provided by the EU project titled Pan-European Coordination of Roma Integration Methods – Roma Inclusion: Self-employment and microcredit, granted by an agreement signed in June 2010. The loan portfolio was financed exclusively from private sources: the Raiffeisen Bank and individuals. Support was also granted by the Hungarian government (which has been discontinued). In addition, another indirect source contributed to the program: in Hungary, new entrepreneurs can apply for state assistance to counterbalance the requirement for paying social security contributions starting in the same month as they get their licence (even if there is no business income) and personal income tax (if there is income). Assistance can be received for 6 months, with a maximum monthly amount equal to the minimal wage. The Kiútprogram duration was initially set at 2 years, and was extended by 3 months at Kiútprogram’s request, as 2 years was considered to be too short for a program issuing loans with one-year duration. 2.4 DAILY FIELDWORK ACTIVITIES The intense involvement of field workers providing a variety of business and social support services was a key feature of the program. Field workers provided a multitude of services, including providing support developing the business plan, business registration support, and accounting support. Further, during operations the scope of tasks of the field workers was larger than originally anticipated. Field workers had to personally meet the clients at least once a week and provide continuous technical assistance support to the operation of the business. Specifically, field workers were actively involved in:3 3 This description is adopted from “Social microcredit, self-employment and Roma inclusion. Lessons of Kiútprogram, a social microcredit pilot", 2012. Available at: 9 1. Community selection. Field workers came from the region in which they worked, but did not work in their actual place of residence – except for Budapest. Communities were selected jointly with Kiútprogram management, often following consultations with other organisations such as the Roma Minority Self-Government and other civil organisations. Community selection involved personal visits, getting familiar with local residents and finding one or more contact persons; people well familiar with the community. 2. Organising a community meeting. The field worker first organised a community meeting, often connecting it to another community occasion to reach more people, to provide information about Kiútprogram and generate interest to start the personal interactions. Field workers conducted recruitment activities and organized forums usually in areas with higher density of Roma, however, becoming a client was not subject to ethnicity. Only income above the given threshold caused ineligibility. Nonetheless, the fact that primarily Roma clients were targeted reportedly inhibited the interest of some non-Roma. The proportion of the Roma is 80% among both the applicants and the actual clients. 3. Family visits. Field workers would next meet with interested candidates, often at the home of the family to get a real picture of the environment and the financial situation and to explore whether the candidate has a viable business idea. During this visit, field worker completed the intake questionnaire. Based on objective and subjective factors, the field worker decided if there was scope for follow-up with the candidate. At least 8 to 9 suitable candidates within a given location had to be identified to create at least one group, considering dropout rates. 4. Group meetings. After identifying 8 to 9 candidates, initial group meetings were organised to learn about group dynamics. Before receiving the loan, people tend to present their favourable side. During this initial meeting the business questionnaire was filled to explore earlier business experience – virtually all informal - and assets, and to discuss the new business ideas. During this meeting, the candidates also selected the group leader, usually the most active, most motivated member. It was the field worker’s task to prepare and lead the group meetings. Beside the pre-arranged programs (see later) the group meetings gave a good occasion to discuss the problems of the local community, and the personal/family problems of the group members. The organized debates served also as a kind of communication training. After gaining their trust, the field workers became a general mentor of the group members. http://www.kiutprogram.hu/index.php/en/documents-f?download=92:social-microcredit-self-employment-and-roma- inclusion-lessons-of-kiutprogram-a-social-microcredit-pilot-and-policy-recommendations-for-the-new-structural- funds-period 10 5. Filtering. To officially become group members, the candidates had to meet certain criteria:  Were there any unpaid taxes or utility debts? Tax arrears can prevent being issued a government business license. For serious candidates, field workers would help with the settlement of tax backlog, public utilities debt and unpaid loans by rescheduling them, applying for partial remittal or paying in instalments.  Were there any unpaid loans? Debt per se was not a reason for exclusion, but too much debt was. Raiffeisen Bank assisted in providing credit reports. In the first year of the pilot, the field workers ran into a few situations where certain candidates were not really intending to start a business, but rather wanted the loan to pay back informal debts owed to loan sharks (“usurers”). With experience, field workers were increasingly able to identify these situations early.  Prepare a simple but realistic business draft (on a piece of paper; so-called “checked paper”) containing rough income and cost figures and list potential customers and suppliers.  Complete and evaluate a business scoring card. This was a management evaluation tool summarising the information collected from the base questionnaire, the business draft and personal meetings. Subjective factors were also included such as the candidate’s motivation, communication skills, existing web of relationships, risk-taking and decision-making skills, family background, business experience, etc. The field worker would send the business draft and the scoring card with his or her recommendation to the Kiútprogram management team where the Credit Committee (CC) would review these materials. The members of the CC included the professional manager, the financial manager at the bank (an employee of the Kiútprogram) and a dedicated expert member of the management board. 6. Finalizing business plans. After filtering, a decision was made about the group and, if given the green light, the remaining group members would create their finalised business plans. By this time, the ideal number of group members was 6. Field worker would meet the group at least once a week and also work with the clients individually. Business plans followed a template, including a cash flow plan for 13 months, and detailed descriptions of the intended loan use, market connections of the planned business, potential partners, risk analysis, analysis of future outlook, and finally the assessment by the field worker. The business plan was recorded by the field worker based on the client’s detailed inputs. The group members discussed and approved each plan before submission. Finalised business plans were approved by the CC. Finalizing the business plans took approximately half of the field worker’s time dedicated to the groups. 7. Formalization of the group. After the finalised business plans were approved, as well as the loan sequence (with the leader last), the group was officially formed during a ceremony with members of 11 the Kiútprogram management team. Each group members received a certificate, signed the group foundation document (including lifestyle rules defined by the members) and the certification of group membership. The group also received a Group Diary to record the group meetings, repayments, and taking and returning cash from/to the group fund. 8. Preparing and issuing the loans. Following the loan sequence, the first two recipients started their business. Business accounts were opened for the cash component of the loan, as well as security deposit accounts and group fund accounts. The security deposit worked as a savings account with a 15% interest at sight. Additionally, a personal account for the client - not connected with the loan – was opened. Furthermore, Kiútprogram prepared the application for the support to become a formal entrepreneur or primary producer and clients were required to contact a bookkeeper affiliated with the program. These were identified in each region and paid for by the program. At this point, clients also received a so-called “loan recipient’s booklet” to track repayments, income, and expenses on a daily level. 9. Creating and maintaining the business. The field worker assisted in the purchase of one-time investments and inventory, accompanying the client if needed or providing necessary transportation. Field worker support was also useful in cases of prejudices against the Roma and the poor. During acquisitions, field workers continued to provide practical training about budgeting. This was important as clients had rarely seen this much money before, and they might have been prone to – for example – purchase goods that were unnecessarily expensive. According to Kiút, such support was a delicate balance, since the clients also had to become independent. Field worker support with obtaining official licenses was also common. Most businesses were categorised as either retail or agricultural ventures. A small group chose forestry; the others are scattered, unique activities. Most businesses required one or more licenses: and operating license, a license to use public premises, an animal health control license, a food safety license, a fire safety license, a legacy protection license, etc. According to Kiút, prejudices by officials providing licences played a role in about half of the cases. 10. Tasks connected with the repayment of loans. Finally, field workers spent considerable time tracking repayments, and working with clients and Kiútprogram management to resolve situations of payment arrears. In case of default due to ill faith (maliciousness), the group or Kiútprogram management decided to remove the member from the group, and the bank put a halt on lending and initiated debt collection. However, this was more of a psychological measure since the loans were unsecured, leaving usually nothing to collect. The client was automatically added to the banking “black list”. 12 According to Kiútprogram staff, the role of the fieldworker as simultaneously being social worker and loan agent providing various services became critical. Had these two areas been treated separately, field workers operating as social workers would have slid into the role of advocates for the clients, regardless of the level of the client’s actual business potential; on the other hand, field workers acting exclusively as loan agents would only have motivated clients to repay loans as opposed to encouraging the creation of feasible businesses. Kiútprogram was concerned that such separation would have diminished the performance of the program, or field workers would have sought to contract clients of higher socio- economic status. In some instances, field workers were also reported to be essential in counteracting discrimination faced when new entrepreneurs registered their businesses, for example. There was, however, considerable turnover among field staff. Over the two-year period, Kiútprogram recruited field workers five times. Selected applicants underwent a training program and were required to pass an exam. The training sessions took 4-5 weeks, and included both a theoretical component and a field component, with the latter in one of the settlements. Altogether 57 people started the training, 48 people completed it, and 37 passed the exam. The training was modified each time based on the feedback and recommendations of the trainees and the program experiences. The high number of trainees reflects in part relatively high turnover. Some field workers simply quit, finding the work burden (primarily psychological) too much. Others were let go. During full operations, approximately 19 field workers were in place. 3 KIÚTPROGRAM CLIENT CHARACTERISTICS The program put in considerable efforts to identify clients. During the course of the program, 202 settlements were visited and lending groups were formed in 38 settlements. More than 900 interested candidates were contacted personally, and 44 actual groups were formed out of 60 pre-groups. As of October 2012, loans with an average value of EUR 1,800 had been issued to 138 clients. 75 clients had become individual entrepreneurs (mostly in retail), and 63 clients had become primary producers operating animal husbandry or crop cultivation businesses. TABLE 1: CLIENT SELECTION INDICATORS Settlements screened 202 Persons screened (personal connection with field workers) 900 Number of intake questionnaires 447 Pre-groups 60 13 Groups formed 44 Settlements with formed groups 38 Group-members 192 Number of loan recipients (clients) 138 Entrepreneurs 75 Mobile retailers 33 Shopkeepers 14 Forestry, timbering 10 Other 18 Licensed primary producers 63 Animal husbandry 32 Plant cultivation 31 Number of loan disbursements a 153 Value of total portfolio EUR 252,000 b Average loan per person EUR 1,825 a One person can get more than one loan b Calculated at 280 HUF/EUR exchange rate Applicants were overwhelmingly confident that they could improve their lives. According to the intake questionnaire administered by the field workers, 44% of applicants were dissatisfied with their lives in general, and 64% with their financial position. The applicant pool median total monthly household revenue stood at HUF 68,250, or approximately EUR 240. Average household size was 4.4 family members, while average applicant age was 36 years. Applications were relatively evenly split between women and men, with women comprising 40% of all applicants. The level of education among applicants was generally low, with only 13% having secondary education or higher. Most applicants lived in a rural environment, and access to sewerage inside the home was significantly lower than access to electricity (64% and 99% respectively). Only 9% of applicants lived in settlements predominantly populated by Roma – 78% of applicants instead lived in mixed settlements. The questionnaire allowed for double ethnic identity, because it is typical among the Hungarian Roma population. 36% of the applicants declared themselves as firstly Roma and secondly Hungarian, 32% firstly Hungarian and secondly Roma, 12% only Roma, and 19% only Hungarian. The ethnic distribution of the loan recipients was almost the same. 14 TABLE 2: BACKGROUND CHARACTERISTICS OF THE LOAN APPLICANTS Demographic / Income characteristics Median total monthly household revenue HUF 68,250.00 Share of female applicants 40% Average age of applicants 36 years Average household size 4.4 people Share of applicants with secondary education or higher 13% Community characteristics Share of applicants... ... living in a rural environment 87% ... living in a settlement predominantly populated by Roma 9% Median distance to nearest bank branch 10 minutes Median distance to nearest employment office 29 minutes Personal / Welfare characteristics Share of applicants... ... that is dissatisfied with their life in general (as opposed to 44% ’satisfied’ or ’very satisfied’) ... that believes they can do something to improve their life 93% ... characterizing him- / herself as a ’risk taker’ 44% ... with access to sewerage inside the home 64% 86% of applicants were either unemployed or outside the labor force. The majority of applicants were unemployed (67%), followed by those out of the labour force (19%), those employed as employees (11%), and finally the self-employed. Of those who were employed (either self employed or as employees), 14% worked in the informal sector. The formally employed had, on average, 10.7 years of work experience. The informally employed had an average of 2.3 years of work experience. 15 FIGURE 1: EMPLOYMENT STATUS OF APPLICANTS Personality characteristics of the applicants appeared mixed. Though 44% of applicants described themselves as ’risk takers’, 85% indicated that they plan tasks carefully. With regards to group dynamics, 57% of applicants mentioned that they find satisfaction in influencing others, and 45% of applicants would prefer to direct group activities rather than follow instruction. In terms of business characteristics, as shown below, slightly less than half the businesses to which loans were granted were in ’agriculture, forestry and fishing’, and slightly more than half in trade. Only very few businesses remained unregistered by the end of the program. Regarding business owners’ experiences with problems in starting and operating the business, a lack of finance stands out. For operating the business, a number of other challenges, including a lack of customers were also reported. Only 10% of all business owners reported that they had no major difficulty at all in operating their business. TABLE 3: KEY BUSINESS CHARACTERISTICS OF SUCCESSFUL LOAN APPLICANTS Key Business characteristics Industry: agriculture, forestry and fishing 45% Industry: trade 55% The business is registered 99% Lack of finance was the greatest problem encountered in setting 68% up the business Loan from bank was the main source of money for setting up 95% 16 the business What is the greatest problem you encounter in operating your business? No major difficulty 10% Lack of finances/credit 21% Lack of customers 22% Government licensing and regulations 10% Taxation 18% Average level of sales in an ' average sales' month HUF 68,312.5 Percentage of businesses which have had either paid or unpaid 51% workers in the last 12 months When in operation during the past 12 months, at least one 14% person besides the owner usually worked in the business 4 PROGRAM OPERATING PERFORMANCE 4.1 LOAN PORTFOLIO PERFORMANCE Drawing on the early operating experience, Kiútprogram had to adjust its operating model in early 2011. The recruitment of clients during the Fall of 2010 took off much slower than expected, and the program soon also presented several signs of crisis, notably the increase of late- or non-payment on loan instalments. In response, in Spring of 2011 Kiútprogram field officers started to place bigger emphasis on personality traits, capabilities, motivation, credit history, the supporting role of the family and any earlier experiences. Furthermore, (i) Kiútprogram made the loan product and group formation rules more flexible, (ii) hired new field workers and improved their training; (iii) added new regions to the target areas and (iv) attempted to take on economic network building activities to connect its clientele with the economic mainstream. Loan repayment performance significantly improved after model correction. Clients whose groups were formed before the correction belong to the so-called 1st batch (classification is based on the time of selection rather than the time they received the loan: due to sequential lending, a number of clients’ groups were formed before the modification of the model, but received the loans following the modifications). Kiútprogram implemented further operational changes in 2012, shifting toward integrated businesses solutions. In particular, Kiútprogram brokered a tripartite arrangement among an agricultural cucumber 17 buyer, the Kiútprogram itself, and its 23 clients (the 3rd batch) in a village in Northeast Hungary leading to individual contracts in Summer 2012, which effectively guaranteed a market for the business products (cucumbers). Average loan. During the pilot the Kiútprogram staff concluded that in the case of relatively larger loans the realistic time to see return on investment for most established businesses is longer than one year. Given the time span of the project, they found diminishing the loan size to be a feasible solution. However, the smaller loan size of the 3rd batch reflects also the nature of the business activity for which the loans were provided. TABLE 4: SUMMARY LOAN INDICATORS BY BATCH 1st batch 2nd batch 3rd batch Total/average Groups formed 12 26 6 44 Loan recipients 41 74 23 138 Operating as entrepreneur in 13 59 23 95 December 2012 Loan disbursements 49 81 23 153 Average loan per person EUR 2,528 EUR 1,832 EUR 550 EUR 1,825 Average duration 52 weeks 43 weeks 26 weeks Lending results. It is estimated that the average amount of arrears will be approximately 30% for clients of the 2nd batch, who received their loans after March 2012. In case of the cucumber-growing project (3rd batch), Kiútprogram expects that that (nearly) all clients will repay the loans on time. Approximately one- third of businesses of the 1st batch are still operating, while this figure is two-third in the 2nd batch. Kiútprogram estimates that about 60% of the older, and 70% of the newer businesses of the 2 nd batch will remain sustainable. 18 TABLE 5: REPAYMENT INDICATORS OF CLIENTS RECEIVING LOANS PRIOR TO MARCH 2012 Arrears per Arrears Arrears per Currently payments due Loan per payments due at Persons operating as th at the 180th day repaid payments the 120 day after entrepreneur after due, % borrowing, % borrowing, % 1st batch 41 13 0 62 28 33 2nd batch 46 31 7 39 12 20 Total 87 44 (95)a 7 52 20 27 a All clients of the 2nd batch not listed here and clients of the 3rd batch still operate their businesses; if they were included, the full number of active entrepreneurs would be 95. The project itself was continued in a smaller extent from private sources after the expiration of the pilot phase. Kiútprogram staff followed old clients and expanded the cucumber-growing project. We present the final repayment indicators. TABLE 6: REPAYMENT INDICATORS, STATUS ON 15 MAY 2013 Still operates as Loan totally Arrears per Payment per Persons entrepreneur repaid payments due (%) credit ratio (%) 1st batch 41 4 0 62 44 2nd batch 74 37a 14 45 62 3rd batch 23 22 2 22 81 Total 138 63 16 51 55 a 1 of them went back to the informal economy. According to the knowledge of Kiútprogram staff the others remained in the formal sector. 4.2 REPAYMENT AND CLIENT CHARACTERISTICS Most background characteristics are similar when comparing unsuccessful applicants4, defaulters and borrowers repaying on time. As shown below, the share of women was largest among borrowers repaying on time. Surprisingly, the share of borrowers repaying on time that had secondary education was 4 Unsuccessful applicants contain both the refused applicants and those who after all decided not to apply for a loan beside the given conditions. 19 smaller than the same share among defaulters and unsuccessful applicants. Conversely, the share of employed applicants was the lowest among borrowers repaying on time. Satisfaction with life in general was higher among borrowers repaying on time as compared to the other two groups. The share of those who took part before the project at least weekly in discussions and decisions of some kind of local community was higher among borrowers repaying on time as compared to the other two groups. This finding suggests that the intensity of local connections helps the success of the new enterprise. Ethnic identity did not influence repayment. Lastly, the share of applicants characterizing themselves as ’risk takers’ was slightly higher among defaulters than among borrowers repaying on time. TABLE 7: UNSUCCESSFUL APPLICANTS VS. DEFAULTERS VS. BORROWERS REPAYING ON TIME Unsuccessful Defaulters Borrowers Applicants Repaying on Time Number of applicants belonging to 318 71 67 each group Share of applicants... ... living in a rural environment 86% 89% 96% ... living in a settlement 9% 6% 8% predominantly populated by Roma … of female applicants 38% 41% 51% Average age of applicants 37 35 34 ... with secondary education or 15% 14% 5% higher ... that is employed 13% 28% 5% ... that is unemployed 66% 58% 79% ... that is inactive 20% 14% 16% ... that is dissatisfied with their life in general (as opposed to 43% 40% 54% ’satisfied’ or ’very satisfied’) ... that believes they can do 94% 95% 85% something to improve their life ... characterizing him- / herself as a 43% 48% 42% ’risk taker’ Median total monthly household HUF 66,500 HUF 80,000 HUF 63,000 revenue 20 There was no big difference between successful and unsuccessful applicants in confidence levels regarding their ability to obtain other loans. Among the successful applicants (including both defaulters and those who paid on time), 23% believed that if they had to obtain a loan of 100,.000 HUF (apr. 360 Euros) within the next month, they could obtain such a loan. When the same question was asked regarding a loan of 500, 000 HUF (apr. 1800 Euros), this shrank to 17%. Among the unsuccessful applicants, these figures amounted to 28% and 18%, respectively. Defaulters generally appeared slightly wealthier in terms of land and livestock ownership than both unsuccessful applicants and borrowers repaying on time. Land and livestock ownership among defaulters was higher than among borrowers repaying on time by 9 percentage points and 5 percentage points respectively. The percentage of unsuccessful applicants owning land was the lowest of all three groups. The share of clients owning a car was 5 percentage points lower for defaulters than for both borrowers repaying on time and unsuccessful applicants. Mobile phone ownership was highest among defaulters, dropping slightly for borrowers repaying on time and unsuccessful applicants. TABLE 8: ASSET OWNERSHIP BY BENEFICIARY TYPE Share of applicants... Unsuccessful Defaulters Borrowers Repaying Applicants on Time ...owning land 4% 18% 9% ... owning livestock 36% 41% 36% ... owning a car 38% 33% 38% ... owning a mobile phone 93% 94% 90% Savings were generally very low among all groups of applicants. Conversely, a sizable share of applicants in all three groups had made use of credit in the recent past, and an even larger share had public or public utility debts. A marginally lower share of defaulters possessed any savings than borrowers repaying on time and unsuccessful applicants. The share of applicants who had borrowed money in the last 12 months or had public or public utility debts was highest among unsuccessful applicants. 21 TABLE 9: SAVINGS AND DEBT BY BENEFICIARY TYPE Share of applicants... Unsuccessful Defaulters Borrowers Repaying Applicants on Time ... that has any savings 7% 5% 7% Among those with savings: average number of months the applicant’s household 1 month 1 month 1 month could live on these savings as the sole source of income ... that has borrowed money from a family member or friend, or a bank, savings 38% 33% 31% cooperative, money lender, shopkeeper or any other source in the past 12 months ... that has public and public 51% 45% 43% utility debt Previous business experiences. 20% of the successful loan applicants formalized existing informal activity with the help of the program. In this respect there was no significant difference between borrowers repaying on time and defaulters. Among the unsuccessful applicants only 15 % had an informal business within the previous 12 months. Finally, both successful and unsuccessful applicants displayed relatively similar self-reported character traits. 91% of successful applicants and 84% of unsuccessful applicants described themselves as planning tasks carefully, while 85% of successful applicants and 72% of unsuccessful applicants indicated that they made decisions quickly. Successful applicants were more likely to enjoy multi-tasking than unsuccessful applicants (81% and 70% respectively). Both groups were equally likely to prefer to direct group activities (45%), though a higher percentage of unsuccessful applicants than successful applicants gained satisfaction from influencing others (25% and 20% respectively). 4.3 PROGRAM COSTS PER LOAN ISSUED Operating costs were high in the two-year pilot period: the cost of identifying, issuing, and subsequently supporting a loan of EUR 1,800 was almost EUR 6,000. The largest costs were incurred by the field workers’ wages and material expenditures. For each 20 field workers, 5 staff members were required in the Kiútprogram program headquarters, including highly qualified lending professionals. Continuous training, legal support, external communications and bookkeeping for the clients required 22 further contractors. Serving 300 clients per month cost a gross amount of EUR 62,0005. EUR 22,000 was paid in taxes and contributions accumulated by the microcredit institution only (excluding clients’ contributions). Kiútprogram estimates that continuous operations at full capacity cost approximately EUR 210 per month per client (gross), or EUR 130 per month per client (net, i.e., without taxes and contributions). Estimated lending losses totalled approximately 30% under regular operating conditions. Given the average loan amount of EUR 1800 for the 2nd batch of clients (i.e., following the adjustment of Kiútprogram’s operating model ), the lending losses were EUR 540 per client per year, or EUR 45 per month. The operating costs of lending activities were amply covered by interest margin, and the full net cost of regular operation was EUR 175 per client. As Kiútprogram shifted its operating model toward implementing integrated projects (such as to the cucumber-growing venture) on a larger scale, lending losses were be further reduced. Comparison with the cost of public works suggests that the Kiútprogram net operating costs are lower. The net wage paid to public work participants with only primary education in Hungary was EUR 168 during the program period (77% of the net amount of the minimal wage). Considering the costs of public work programs, this amount must be supplemented by the costs of employing managers, the administrative costs and the costs of equipment. Altogether, the annual client costs of Kiútprogram may be comparable to the cost per person per year in the public work program. While public work does not necessarily increase the chances of getting employed, within the framework of Kiútprogram a majority of clients established a self-sustaining business paying taxes and contributions in the long run. 4.4 CLIENT FEEDBACK Field workers collected client feedback at the end of the project, providing useful information on project design. However, the findings below may overstate positive responses for two reasons. First, not all those who defaulted in batch 1 were willing to be re-interviewed, and second respondents may have been more positive toward the Kiútprogram fieldworkers than they really felt. The interest payment scheme of the Kiútprogram model attracted client interest, while other features, including group lending, were more neutral but not important impediments for clients. For two of the other three indicators measured, namely the Kiútprogram group lending model and the size of the loans, the majority of successful applicants was indifferent. For both of these program characteristics, 23 about one third of successful applicants reported that these elements increased their interest to join, and smaller shares reported that it had the opposite effect. The length of the loan period did not affect interest for about half of all loan recipients: the other half was almost equally divided between a decrease in interest to join and an increase in interest to join. TABLE 10: KIÚTPROGRAM PROGRAM CHARACTERISTICS AND BENEFICIARY INTEREST TO JOIN Interest to join: share of applicants that believes the [program element] reduced, increased, or did not affect their interest to join Kiút. Increased No difference Reduced Kiútprogram group lending 34% 53% 14% model... Loan sizes... 35% 55% 9% Lenght of loan period... 22% 47% 31% Interest payment... 52% 36% 12% The majority of clients interviewed indicated that the loan contributed to improving their businesses, while there was more variation in the perceived impact on the family’s financial situation. 90% of participants responded that the loan very much improved their businesses. In 48% of cases, participants noted that the loan either very much improved their family’s financial situation, while 7% of participants experienced a negative impact on their family’s financial situation as a resul t of the loan. 45% of participants also mentioned that the loan increased their fulfillment in life. TABLE 11: BENEFICIARY OPIONS ON LOAN IMPACT Very much Somewhat Same as Negative before impact Did the loan contribute to improving your 90% 8% . 1% business? Did the loan contribute to improving your 48% 32% 14% 7% family's financial situation? Did the loan contribute to improving your 45% 27% 22% 5% fulfillment in life? 24 A large proportion of loan recipients found the participation in the program a useful experience. 70% of participants found the functioning of the group to be beneficial, while only 7% found it counterproductive. In terms of support provided, only 3% of participants found the quality of information provided inadequate, while the percentage participants that found support in running of the business and in book-keeping inadequate was 6% and 1% respectively. The close neighborhood of participants was generally supportive. By contrast, the wider neighborhood was only reported to be supportive by 38% of all loan recipients. TABLE 12:KIÚTPROGRAM PROGRAM CHARACTERISTICS AND LOAN PERFORMANCE: BENEFICIARY FEEDBACK Helped my Neutral Counterworked activity Functioning of the group 70% 23% 7% Very adequate Adequate Inadequate Quality of the information provided 59% 38% 3% Support in running my business 58% 37% 6% Support in book-keeping 60% 39% 1% Helpful / Neutral Hostile Sympathetic How did your close neighborhood 77% 8% 14% relate to your participation? How did your wider neighborhood 38% 35% 27% relate to your participation The general opinions of successful applicants on the importance of other forms of support when starting a business were positive, shown in the table below. The majority of successful applicants felt that financial literacy training, financial savings support, and microcredit for business were either somewhat or very important. Across the three services, savings support was considered to be very important by the greatest share of clients, followed by financial literacy training, and then microcredit. 25 TABLE 13: IMPORTANCE OF DIFFERENT FINANCIAL SERVICES: BENEFICIARY FEEDBAC Very Somewhat Not important important important Offering financial literacy training improves your 53% 41% 5% life Offering financial savings support improves your 69% 18% 13% life Offering microcredit for business improves your 45% 37% 18% life 4.5 STAKEHOLDER FEEDBACK The Kiútprogram helped address negative stereotypes at the local level. According to Kiútprogram clients and staff members, the feedback received from local office managers and administrators of employment centres, banks and other offices was that some were surprised to see that the Roma clients could handle the administrative hurdles of the enterprises and repay the loans. Officials and administrators would not have thought that these people were willing take on such a huge financial risk if there is a slight chance ofto improving their own lives and their families’ future. Similar opinions arrived from some of the business partners. Furthermore, the mayor of the village where the cucumber project was accomplished stated that Kiútprogram was more successful in promoting economic integration than a local public work program in which the unemployed receive allowances in return for agricultural work. Given the cucumber project success, several mayors from neighboring villages invited Kiútprogram field workers to introduce the program also in their settlements. 4.6 POST-PILOT CONTIUNATION In September 2012 private sponsors decided to continue the Kiútprogram on a small scale after the EU pilot program terminated and after the Hungarian government support had been discontinued. Kiútprogram staff followed contracted clients up to the end of their loan period, but – with the exception of one county – new loans were not delivered. From March 2013 only the so called “cucumber-growing” project has been continued, with a professional leader, two field workers and an assistant. The 2012 cucumber results attracted considerable interest in the surrounding villages of the area. Beside the 23 original clients, the program has 90 new 26 clients in 7 settlements. With more field workers even larger clientele could have been reached according to Kiútprogram. The average loan size is 500 Euro, while the planned average operational cost is 1100 Euro/client. The enlargement necessitated the involvement of a new buyer, who – according to his statement – previously never had a connection with Roma producers of Roma origin. The program found also a new partner bank delivering the loan, namely a local saving cooperative6. The encouraging results from this project point to considerable financial inclusion of Roma living in these communities. One of the lessons for Kiútprogram from the cucumber project experience is the need to have prospective borrowers use the loan for activities for which there is a demonstrated demand in the market; i.e. a clear market chain in which the funded activities fit. In particular, Kiútprogram helped broker an agreement between an agricultural cucumber buyer and Kiútprogram borrowers. 6 Hodász-Porcsalma Saving Cooperative 27 5 DISCUSSION The Kiútprogram was successful in targeting poor and mostly unemployed people, the majority Roma. Finding suitable clients was initially difficult, however. Over the two year project period, 138 loan contracts were signed. This low number reflects in part the initial difficulty of finding suitable clients, and the subsequent decision to increase client screening and support before approving loan requests. 86% of applicants were either unemployed or outside the laborforce. However, applicants were overwhelmingly confident that they could improve their lives. The program required intensive follow-up by well-trained field workers familiar with social work, lending, and business. The operating costs were high in this pilot period: the cost of issuing a loan of EUR 1,000 was almost EUR 6,000. Estimated lending losses totalled approximately 30% under regular operating conditions. Assessing the cost effectiveness of the Kiútprogram pilot program is not straightforward, but the program may be considered cost effective compared to common active labor market policies. Is EUR 6,000 ‘expensive’ for a EUR 1,000 loan? First, it depends of course on the program impact. In the absence of a plausible counterfactual, this is impossible to isolate rigorously. Several indicators, however, suggest that the project improved the lives of clients: for example, approximately two-thirds of clients continue to operate a business beyond the loan cycle, and more than two-thirds indicate that the program improved their financial situation and improved their fulfilment in life. Second, it also depends on the nature of the comparison. The Kiútprogram argues that social microcredit should not be benchmarked against regular microcredit, but against active labor market programs. For example, the annual client costs of Kiútprogram may be within the range of costs per person per year in public works programs, or even lower when considering the taxes and social contributions paid by new entrepreneurs. Public works, however, have not proven to sustainably increase employability. Furthermore, Kiútprogram fieldworkers point to the programs role in changing perceptions among local officials and the general public regarding Roma entrepreneurship. For example, people often did not expect that poor Roma would be willing to take on financial risk and invest considerable energies to start a new enterprise that might lift them out of poverty. And, successful experiences such as the cucumber project led to increased demand among officials from neighboring communities requesting the program’s support there as well. The Kiútprogram made several adjustments to improve the performance of the program. Given the comparable circumstances of Roma in neighboring countries, these adjustments likely have wider application. They include: 28  Providing a broad range of financial and social services aimed to enable the unemployed living in deep poverty – primarily, but not exclusively Roma – become self-employed. The range of services started in many instances with social support followed by a range of businesses support services from business plan development to financial literacy training, including savings support, and ultimately to (unsecured) microloans provided in group settings. Experience demonstrated that the comprehensive approach was necessary and client feedback at the end of the project suggests that this wide range of services were valued. These findings are also consistent with the companion World Bank (2012) report emphasizing a comprehensive approach toward the financial inclusion of Roma, focusing on access to bank accounts, (targeted) savings, financial literacy, and business skills training.  Careful selection of people that seemed to have good – but difficult to measure - entrepreneurial skills such as trustworthiness, motivation, ability to articulate a business plan, etc., improved repayment outcomes. Fieldworkers recommended the use of real skill-based evaluations of applicants that can measure such character traits.  Flexibility of the loan product. During the pilot, staff and clients reported that the realistic time to see return on investment for most established businesses is much longer that one year. In most cases, field workers reported that entrepreneurs struggling with repayment difficulties were not lacking the willingness but the capability to repay. For this reason, Kiútprogram signed agreements toward the end of the pilot enabling some clients whose business was operational to pay on deferred terms.  Actively support potential clients identify business opportunities for which there is demonstrated demand. One of the lessons from the Kiútprogram cucumber project experience is the need to have prospective borrowers use the loan for activities for which there is a demonstrated demand in the market; i.e. a clear market chain in which the funded activities fit. Kiútprogram helped broker an agreement between a large cucumber buyer and program borrowers, effectively securing a market for their produce. Following the completion of the pilot phase, Kiútprogram has continued on a smaller scale with private resources, applying these lessons. The so called “cucumber-growing” project has had good initial results, which attracted considerable interest in the surrounding villages and has brought in 90 new clients in 7 settlements. 29 The careful documentation of the Kiútprogram experience demonstrates valuable operational lessons for financial services targeted at the poor. There are several useful global resources supporting greater evidence based policy making more broadly around financial inclusion, including (a) the World Bank’s financial inclusion program – www.worldbank.org/financialinclusion; and, (b) the financial inclusion (FINDEX) database - http://datatopics.worldbank.org/financialinclusion. For example, FINDEX measures how people save, borrow, make payments, and manage risk. The Financial Capability Survey (CPFC) survey captures the way people manage their money, their ability to budget and to stick to the budget, to plan ahead, to make provisions for their retirement age as well as for expected and unexpected expenses and by identifying how people chose financial products and to which financial products and services people have access. Another resource is the (c) the Global Financial Inclusion initiative at Innovations for Poverty Action - http://www.poverty-action.org/financialinclusion - , which ‘manages two research funds seeking to test, evaluate and replicate interventions to improve products, delivery channels, and tools to help users make the most of their financial portfolios.’ This includes rigorous randomized counterfactual evaluations, or social policy experiments. Finally, while Kiútprogram was successful in continuing the project on small scale beyond the pilot phase with the support of private funding, there is currently no clear instrument supporting this kind of ‘social’ microcredit in Europe. Kiútprogram it experienced difficulty in identifying sources of financial support that fit the broader ‘social microcredit’ approach to self-employment. A main obstacle for this type of program is that ‘sustainability’ of microcredit operations is currently a prerequisite for existing microcredit instruments. These instruments’ target groups include existing enterprises, or th ose who are very close to being “bankable”: such sustainability requires a clientele with higher social status than the target group of social microcredit. 30