MICROFINANCE AND HOUSEHOLD WELFARE Cambodia Policy Note © 2019 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 18 17 16 15 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. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. 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: Pimhidzai, Obert; Tong, Kimsun; Anantavrasilpa, Ratchada; Popovic, Andrej; Mel, Sokim; Sanchez Martin, Miguel Eduardo, 2019. Microfinance and Household Welfare. Cambodia policy note. Washington, D.C.: World Bank Group. Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: 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 the Publishing and Knowledge Division, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover design: Florencia Micheltorena, Buenos Aires, Argentina. Interior photo: Chor Sokunthea / The World Bank. MICROFINANCE AND HOUSEHOLD WELFARE Cambodia Policy Note February 2019 Cambodia Policy Note | 4 Foreword Thanks to rapid and sustained growth, Cambodia has become one of the world’s leaders in poverty reduction and shared prosperity. Cambodia’s success so far has ridden on openness to trade and investment, preferential trade treatment, and an abundance of low-skilled, low-cost labor. This facilitated the establishment of an export-oriented and foreign-owned garment sector, which—together with tourism, agriculture, and construction—has been driving growth over the past two decades. Several factors suggest that Cambodia will not be able to rely on its current drivers of growth going forward. With the country becoming more prosperous, it is receiving less generous donor financing, and preferential trade treatment is expected to phase out eventually. At the same time, rising wages are making it increasingly difficult for Cambodia to keep exporting unprocessed rice and low-end garments. Meanwhile, around the globe, automation is displacing jobs, and digital technologies are transforming certain sectors. Coinciding with the possibility of diminished economic prospects are the rising expectations of Cambodian citizens, fueled by the rapid spread of information through mobile phones, the internet, and social media. In light of these factors, Cambodia’s current drivers of growth need to be diversified. While the country’s economic outlook remains positive, Cambodia could begin to explore new drivers of growth that will create jobs and boost prosperity over the next 20 years. The country’s next economic transformation will rest on its ability to empower domestic entrepreneurs and citizens with capabilities and tools for the modern economy of tomorrow. Drawing from the knowledge gaps identified in the Cambodia Systematic Country Diagnostics, the Cambodia Policy Notes analyze four relatively under-analyzed policy areas: Entrepreneurship and Innovation, Digital Economy, Microfinance and Household Welfare, and Social Assistance. These topics are interrelated and will be crucial for catalyzing or complementing reforms in traditional sectors. For example, digital platforms are expected to play a key role in enabling local entrepreneurs to participate in global value chains more effectively. Financial technology (fintech) can provide expanded access to finance for both enterprises and individuals as well as facilitate the growth of e-commerce. Social assistance could help mitigate shocks for households and contribute to building a stronger human capital base to take advantage of new opportunities and safeguard the impressive economic gains of the last two decades. Ellen Goldstein Country Director for Cambodia, Lao PDR, and Myanmar World Bank Group Microfinance and Household Welfare | 5 Acknowledgements This Policy Note has been prepared by a World Bank Group team led by Obert Pimhidzai and Kimsun Tong, with contributions from Ratchada Anantavrasilpa, Sokim Mel, Andrej Popovic, and Miguel Eduardo Sánchez Martín. The authors are grateful to Swapnil Kant Neeraj, and Thuy Thu Bui for their feedback. The note benefits from guidance provided by Ellen A. Goldstein, Inguna Dobraja, Irina Astrakhan, Salman Zaidi, Deepak Mishra, Shabih Ali Mohib, Mark Austin, and Lars Sondergaard. Minna Hahn Tong and Maria Dumpert helped edit the note. The team is grateful to Cambodian government institutions for the support provided during preparation of this Policy Note. In particular, the team would like to thank H.E. Chea Serey (Assistant Governor and Director General of the NBC), H.E. Mey Vann (Director General, General Department of Financial Industry, MEF), H.E. Tep Phyorin (Director General, General Department of Economic and Public Finance, MEF), Mr. Em Kamnan (Director of Legal Department, NBC), Mr. Teng Tithanou (Deputy Director of Financial Market, MEF), Mr. Kith Sovannarith (Deputy Director General and Director On-Site Supervision Department, NBC), Mr. Lun Sam Ol (Deputy Director of Off-Site Supervision Department, NBC), Mr. Men Pheakdey (Deputy Director of On-Site Supervision Department, NBC), Mr. Som Kossom (Division Chief of Off-Site Sup Department, NBC), and Ms. Mak Reaksmy (Division Chief of Legal Department, NBC). The note benefited from the kind cooperation of the Credit Bureau of Cambodia, which provided access to some of the data used in the analysis. Further contributions were provided by private sector representatives, including Mr. Sovanna Yun (General Secretary, Cambodia Microfinance Association), who provided useful insights and feedback for the Note. The team is also grateful to Da Lin for her support during team missions and stakeholder consultations. Cambodia Policy Note | 6 Microfinance Key Messages • Growth in microcredit is having positive financial and welfare impacts for households in Cambodia. The share of households who borrowed from formal sources of credit quadrupled during 2004-2016. This shift increased access to finance for segments of the population who previously relied on unregulated money lenders and provided households with longer loan durations and lower effective interest rates relative to informal lending. • Nonetheless, the cost of providing credit remains high, driven by both operating costs and cost of borrowing. At 5.2 percent in 2016 and 1.2 percent in 2017, the return on equity for Cambodian microfinance institutions (MFIs) is lower than the world average, partly because Cambodian MFIs are less operationally efficient. • Risks are increasing for MFIs and for the Cambodian economy in general, partly reflecting looser lending practices. Over the past five years, the average loan size increased more than tenfold, as did the share of loans for consumption needs and the portfolio-at-risk. These trends are due to a combination of the low penetration of financial instruments, deteriorating lending practices, and low financial literacy. • The introduction of an interest rate cap in April 2017 has resulted in a decline in average borrowing cost, partly offset with a substantial increase in fees charged to borrowers. In most institutions, the decline in interest rates has been partly offset by a substantial increase in fees. Evidence also indicates that some MFIs that were lending at rates below the cap have converged to the 18 percent interest rate, which may effectively increase the cost for some borrowers, and which is consistent with experiences in other countries that resorted to a similar policy. Microfinance and Household Welfare | 7 • At the same time, there has been a reduction in small loans in favor of larger loans with a longer time span, without necessarily slowing overall microcredit growth. Due to the introduction of the interest rate cap, the number of loans of USD 500 or less declined by 48 percent. Notably, the proportion of households in the first quintile (poorest) who borrowed from an informal source increased by 5 percentage points in 2017. Meanwhile, microfinance sector lending grew by 53 percent in the last 3 quarters of 2017 compared to a similar period in 2016, as loan size increased by an average of USD 1,200 and USD 240 for microfinance deposit-taking institutions (MDIs) and MFIs, respectively. • To increase access to finance while minimizing risks, key policy priorities include: (i) introducing regulatory policies that enable reform of the operation and delivery models for MFIs to increase their operational efficiency (including through digital distribution), (ii) addressing funding constraints in the microfinance sector, (iii) broadening financial instruments to meet household consumption borrowing needs, (iv) strengthening consumer protection and empowerment to enhance welfare impacts, and (v) improving transparency, lending practices, and supervision to reduce risks. Introduction institution, and just 50 percent withdrew money from an account during the past year—one of the lowest percentages in the world.1 Thus, Cambodia remains Cambodia’s openness to trade and financial flows a predominantly cash economy, with people enjoying fueled one of the fastest credit growth episodes in easy access to borrowing but with the majority of the Asia, although Cambodia remains a predominantly population still being unable to save. cash-based economy. Capital account openness and dollarization led to rapid growth in foreign currency A remarkable expansion in formal microfinance deposits, averaging close to 25 percent growth year-on- lenders contributed to increased access to credit. year since 2010. This fueled one of the fastest capital Since the promulgation of the Policy and Strategies in the deepening episodes in the region, with Cambodia’s Microfinance Sector (2007), Cambodia’s microfinance credit to private sector as a percentage of gross sector has expanded rapidly, with both assets and credit domestic product (GDP) jumping from 2 percent in 1993 growing at annualized rates of over 40 percent.2 The to 69.7 percent in 2016, already above the average for number of MFIs in Cambodia increased from 16 in 2006 lower middle-income economies. A significant share to 69 in 2017. In addition, Cambodia has 7 microfinance of lending went to households, placing Cambodia deposit-taking institutions (MDIs) and 170 registered among the top developing countries in terms of the microfinance operators.3 Between 2009 and 2017, the percentage of people who borrowed money from a number of borrowers in the microfinance sector4 more financial institution in 2017. However, only 5 percent of than doubled to 1.8 million, three times the number of the population aged 15 and above saved at a financial borrowers of commercial banks. Outstanding loans 1  Global Financial Inclusion Database. 2  Following the enactment of the Law on Banking and Financial Institutions in November 1999, Prakas No B7.00-06 outlined the process of MFI registration and licensing in Cambodia. The National Strategy for Microfinance, promulgated by the Prime Minister in 2007, endorsed the role of microfinance in promoting businesses, increasing agricultural productivity, and ultimately improving the living standards of rural households and reducing poverty. 3  Prakas No.B7.02.49. Microfinance is defined as “the delivery of financial service such as loans and deposits to the poor and low-income households, and to micro-enterprises.” MDIs provide loans and mobilize saving from general public, while MFIs provide only loans. All non-government organizations, associations, and other entities engaged in microfinance are required to register or obtain a license if they meet certain thresholds detailed in the Prakas. 4  In this Policy Note, when referring to the microfinance sector both MFIs and MDIs are considered. Cambodia Policy Note | 8 grew from USD 640 million in 2011 to USD 4.2 billion in 2017. Notably, the seven MDIs accounted for more than Context and Main 85 percent of total customers and loan volume in the Challenges microfinance sector in 2017. The expansion of the microfinance While access to credit helps ease financial sector has provided access to constraints for households, one key concern is how formal credit for a large segment the cost of credit and increased exposure to risk of borrowers who previously and debt levels might affect household welfare. At depended on informal sources the household level, low financial literacy could result in poor borrowing decisions and heighten risks. Thus While the share of households accessing credit far, empirical studies on the impact of micro-credit on has not changed significantly, expansion of the household welfare in Cambodia have been inconclusive. 5 microfinance sector has made it possible for The ongoing policy debate has focused on determining borrowers to shift from informal to formal sources whether households are overindebted, whether the of credit, especially among the poor. The share of market is saturated, what the motivations for borrowing 6 households borrowing from formal sources increased are, and how to reinforce the positive aspects of micro- from 8 percent in 2004 to 30 percent in 2017, while credit while reducing the cost of credit, improving the the share borrowing from informal sources decreased quality of loans, and managing risks. In April 2017, from 32 percent to less than 6 percent (Figure 1, left Cambodian authorities introduced an interest rate cap panel).8 Formal credit has expanded for a segment of aimed at lowering the cost of borrowing. the population who previously relied on unregulated private money lenders, although the overall share This policy note assesses the impact of access of households accessing credit has not changed to credit on household welfare in Cambodia and significantly, hovering around 37 percent over the past provides evidence on the drivers of the cost decade. As access to formal credit has expanded, the of credit in the microfinance sector, as well as population in the poorest quintile has become as likely preliminary estimates of the impact of the interest to borrow from a formal source of credit as those who rate cap. To fill the existing knowledge gaps, this policy are better off (Figure 1, right panel). It is nonetheless note uses the latest available data from official sources worth noting that the proportion of households in the to provide evidence on (i) the impact of microcredit on first quintile (poorest) who borrowed from an informal household welfare, (ii) profitability and cost of credit in source increased by 5 percentage points in 2017, the microfinance sector, and (iii) the effects of the interest which may suggest that some low-end clients turned rate cap in the sector. This note ultimately presents a 7 back to informal sources after the interest rate cap was series of policy options aimed at facilitating affordability introduced in April 2017. and reducing the cost of credit, while maintaining sector profitability and minimizing risks for both households However, some pockets of financial exclusion and the financial system. Policy options have been remain. An analysis of household survey data prepared in consultation with stakeholders. conducted for this policy note found that borrowing from any source is highest in rural areas, where about 5  Phim, 2014; Roth, 2016; Seng, 2017. 6  Mimosa, 2016. 7  This analysis used the Cambodia Social-Economic Survey 2004-2016 and Propensity Score Matching to assess the welfare impact of microcredit, National Bank of Cambodia published data in Annual Supervision Reports 2011-2017 and the application of Data Envelopment Analysis to measure the efficiency of the microfinance sector, and Credit Bureau Cambodia data from January 2015 to March 2018 and application of Regression Discontinuity to analyze the effects of the interest rate cap. 8  These trends are broadly in line with those in the Global Findex Database, where the share of adults who borrowed from a financial institution (formal source) increased from 19 percent in 2011 to 27 percent in 2017. Microfinance and Household Welfare | 9 39 percent of households have a loan, followed by other Access to credit has provided urban areas (31 percent). It is lowest in Phnom Penh (13 some welfare benefits to the poor percent). Households are equally likely to borrow from formal credit sources in rural and other urban areas, but Access to formal sources of credit has improved access to formal credit differs according to household affordability through reduced interest rates and characteristics. Larger households and those with older longer repayment periods. In 2017, reported monthly household heads tend to borrow more from formal interest rates on loans from formal lenders were 1.7 sources. Unmarried household heads, those without percentage points lower than those from informal credit complete primary education, and households belonging sources (Figure 2, left panel). This difference translates to an ethnic minority are significantly more likely to into an annual effective interest rate for formal loans that borrow from informal sources, which highlights some is 22.4 percentage points lower than that of informal exclusion challenges (Table 3 in the Annex). loans. In the formal sector, average loan durations Figure 1. While borrowers have shifted from informal to formal sources of credit, Cambodia’s share of indebted households has not increased significantly over the last decade Percentage of indebted households by source Source of loan by welfare quintile 45 Poorest % of indebted households 40 2 2017 3 35 4 Richest 30 Poorest 25 2 2016 3 20 4 Richest 15 Poorest 10 2 2009 3 5 4 0 Richest 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 04 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 Informal Formal Total Informal Formal Source: Authors’ calculation using CSES 2004, 2007-2017. Figure 2. Interest rates are lower for formal sources Monthly interest rates (%) Loan duration (months) 10 35 30 8 25 6 20 4 15 10 2 5 0 0 04 07 08 09 10 11 12 13 14 15 16 17 04 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Informal Formal Total Informal Formal Total Source: Authors’ calculation using CSES 2004, 2007-2017. Cambodia Policy Note | 10 Box 1: International evidence on the welfare impact of microfinance In Bangladesh, where comprehensive evaluations of microfinance have been conducted, findings on the welfare impact of microfinance have been mixed. The most comprehensive evaluation of microfinance, undertaken over 20 years by the World Bank, examined microfinance institutes in Bangladesh. The initial evaluation found that microcredit helped promote household welfare and that the impacts of credit are higher for women than men.1 While two follow-up studies have confirmed these benefits,2 they are not fully supported by others. Mixed results have also been found in other countries, such as Bolivia, Ghana, India, Indonesia, Morocco, Peru, Thailand, and Sri Lanka. Studies on the impact of microcredit on households’ socio-economic well-being found positive effects on the expansion of existing business enterprises, self-employment activities, and business profits; children’s schooling; household welfare; and management of health and income shocks.3 However, there is also evidence that microcredit had little or no impact on consumption, health, women’s empowerment, education, and business activities.4 A few other studies showed that microcredit reduced men’s health expenditure, increased child labor, and worsened the borrowers’ welfare.5 1  Khandker, 1998; Pitt and Khandker, 1996, 1998; Pitt et al., 1999; Pitt et al., 2006. 2  Khandker, 2005; Khandker and Samad, 2014. 3  Banerjee et al., 2014. Crepon et al., 2014; de Mel et al., 2008; McKernan, 2002. Islam and Choe, 2013. Imai et al., 2010; Imai and Azam, 2012; Akotey and Adjasi, 2016. Gertler et al., 2008; Islam and Maitra, 2012. 4  Duflo et al., 2013; Crepon et al., 2014; Karlan and Valdivia, 2011. 5  Coleman, 1999. Maldonado & González-Vega, 2008. Ganle et al., 2015. increased from around 16 months in 2012, similar to studies in Cambodia have provided inconclusive the loan duration for informal credit at the time, to an evidence on the welfare impact of microfinance average of 30 months in 2017. The extended loan credit.9 Such contradictory findings are also evident tenure has allowed formal credit providers to offer larger in the international literature (Box 1). For this policy loans, with a median of USD 2,000 in 2017, compared note, national household survey data was analyzed to to USD 500 from informal sources. Another benefit of compare various welfare outcomes for borrowing and formal lending is a decreased chance of abusive interest non-borrowing households using improved statistical collection practices. Assessed on these dimensions, the techniques.10 This analysis found that accessing a credit conditions faced by borrowers arguably improved loan increased the chance of a household engaging with the switch to formal sources of lending. in a household enterprise by 4 percent relative to similar households who did not have a loan in 2016. Some evidence indicates that microcredit has Similarly, households with a loan spent 5 percent more improved household welfare in Cambodia. Past on agricultural inputs than those without a loan. The 9  Using panel data (2011-2014) for about 1,000 households in 11 villages in Cambodia and applying the difference-in-difference approach, Roth et al. (2016) concluded that microcredit increased paddy quantity and income, expenditure on inputs of paddy production, and non-land durable assets. Similarly, Phim (2014), also employing panel data (2001, 2004, and 2008) for 827 households and using a combination of propensity score matching and the difference-in-difference method, concluded that microcredit has a positive impact on poverty reduction. However, Seng (2017) used the nationally representative Cambodia Socio-Economic Survey in 2014 to examine the effects of microcredit on household welfare and found that both formal and informal microcredit reduced household expenditure. 10  Specifically, propensity score matching techniques were applied to nationally representative household survey data over several years to compare various welfare outcomes for borrowing and non-borrowing households that had similar financial capacities prior to the borrowing household taking a loan. Microfinance and Household Welfare | 11 analysis also found evidence of positive impacts in terms right panel), the debt-to-consumption ratio is higher of health expenditure (for households with ill members), in urban areas and among the top three quintiles. non-food consumption,11 and poverty reduction (Table 4 The quickly increasing debt-to-consumption ratio has in the Annex). The findings suggest that microcredit has raised concerns about the debt repayment capacity of played an important role in developing entrepreneurial a significant number of borrowers. These trends may activities, commercializing agriculture, and improving be the result of intense competition in the microfinance the overall living standards of the poor. sector, as well as the fact that in most cases, incentives are provided to credit officers based on volume of However, risks may be increasing lending, without considering quality or risks. as the level of indebtedness rises and more funding is directed Moreover, borrowing is being directed increasingly toward consumption toward consumption needs that could have been met with different financial instruments. Loans Household debt levels have increased significantly devoted to agricultural and non-agricultural activities in recent years, as the average loan size has grown decreased from 47 percent of the total in 2004 to 29 faster than household income. Official data from the percent in 2017 (Figure 4). Loans used for non-income- Credit Bureau Cambodia shows that between 2015 and generating activities increased over the past decade 2017, the average loan size in the microfinance sector largely due to a rise in housing improvement and grew by 80 percent, continuing a trend observed since durable goods acquisition loans, reaching more than the beginning of the decade. The ratio of outstanding two-thirds of the total in 2017. There are alternative debt to consumption (extrapolated over the loan financial products better suited to the purposes for duration) for borrowers in the bottom 40 percent (first which most of the non-income-generating loans in two income quintiles) increased from 11 percent in 2010 Cambodia are used. Loans used to cover health- and to 33 percent as of 2017. The increase in the ratio was injury-related expenses (4 percent of total) and funeral more pronounced among the top 60 percent, from 9 expenses could be better financed in insurance markets; percent to 43 percent over the same period (Figure 3, durable goods acquisitions could be funded using retail left panel). While the share of households that borrow credit and installment arrangements; and housing is higher among the bottom 40 percent (Figure 3, improvement, weddings, and other lumpy expenditures Figure 3. Average indebtedness relative to consumption has increased significantly in recent years Borrower debt-to-consumption ratio % indebted households by quintile 60 60 Share of indebted households Debt to consumption ratio 50 50 40 40 30 30 20 20 10 10 0 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2009 2010 2011 2012 2013 2014 2015 2016 2017 Bottom 40 percent Top 60 percent Rural areas Urban areas Poorest Q Q2 Q3 Q4 Richest Q Source: Authors’ calculation using CSES 2004, 2007-2017. 11  No significant impact on food consumption was observed. Cambodia Policy Note | 12 could be financed by innovative saving schemes. While Cambodia, which might result in repayment issues some of these instruments already exist in Cambodia, being underreported.14 To strengthen monitoring and their effective delivery remains an issue, since potential mitigate potential risks, the National Bank of Cambodia customers in many cases do not have the financial passed a Prakas on Credit Risk Grading and Impairment education needed to understand the advantages of Provisioning in December 2017 which includes a more insurance, and MFIs are not allowed to bundle these comprehensive classification of portfolio at risk and products together with microcredit. Low penetration mandates the provisioning of losses when a loan is past or uptake of alternative financial instruments may lead due over 180 days. to an avoidable increase in household debt burden and may also limit the role of microcredit in developing Although the average loan size by MFIs is not entrepreneurial activities. A companion Policy Note on substantially lower than loans by MDIs, MFIs in Entrepreneurship and Innovation discusses challenges Cambodia make riskier loans. The loan portfolio of in access to finance for micro, small, and medium MFIs comprises a higher share of personal loans (58 enterprises in more detail. percent, compared to 21 percent for MDIs), which are riskier on average (Figure 5, right panel). Even for A rise in the share of overdue loans points to rising 12 the same type of loans, delinquencies for MFI loans risks in the microfinance sector. As of end 2017, the are significantly higher. This is consistent with the share of reported NPLs had increased to 2 percent, still higher share of unsecured loans for MFIs (16 percent) low by international standards.13 However, the share compared to MDIs (1.4 percent). MFIs also have most of loans with repayments more than 90 days overdue of their loans secured by other types of titles (i.e. not (portfolio at risk) has increased significantly for MFIs, land or immovable property titles). This pattern suggests reaching nearly 6 percent by mid-2017 before declining that the quality of collateral for MFI loans is poorer, MFIs to around 4.5 percent in early 2018 (Figure 5, left panel). have weaker repayment collection capacity, or risk Another key issue from the asset quality perspective assessment is limited, resulting in a higher portfolio at is the high percentage of refinanced portfolio in risk. Figure 4. Loans used toward non-income-generating activities have increased 2012 2017 1 1 3 Agricultural activities 4 7 5 Non-agricultural activities 17 4 14 31 Household consumption needs 6 Illness/injury/accident 12 Rituals (marriage ceremony, funeral, etc.) Purchase/improvement of dwelling 13 Purchase of consumer durables 30 16 Servicing and existing debts 4 2 30 Other Source: Authors’ calculation using CSES 2004, 2007-2017. 12  Portfolio at risk (PAR), the most accepted measurement of portfolio quality, is the outstanding amount of all loans that have one or more installments of principal past due by a certain number of days (e.g., 30 days or 90 days). 13  The NPL ratio is defined as the amount of NPLs (loans for which payments are more than 30 days overdue) divided by gross loan portfolio. The information is presented in the National Bank of Cambodia Annual Report for 2017. 14  Early refinancing with much larger loans and longer tenor, while conducting just a limited assessment on the repayment capacity of borrowers, has increased significantly. This causes potential repayment risk and deterioration of the portfolio. Microfinance and Household Welfare | 13 Figure 5. The portfolio at risk is significantly larger in MFIs than in MDIs Overdue loan and write off (value) Portfolio comparison 7% 9,5 Overdue business loans (30+ days) 3,2 6% 17,7 Overdue personal loans (30+ days) 1,0 5% 4% 51,2 Secured with other titles 40,0 3% Unsecured loans 16,1 1,4 2% 29,6 1% Busniness loans 48,7 57,8 Personal loans 21,0 0% Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 57,4 2015 2015 2017 2018 Single loans 49,8 0 10 20 30 40 50 60 MDI: of amount with overdue payment (>90 days) MDI: write off MFI: of amount with overdue payment (>90 days) MFI: write off MFIs MDIs Source: Credit Bureau Cambodia. Low financial literacy compounds the current institutions, which mitigates foreign exchange risks but challenges. Most poor borrowers are not informed is also more costly; if the ultimate borrowers are not able enough to choose the right financing instruments. to service their loans, the problem can easily spill over According to the 2014 S&P Global FinLit Survey, only into the broader banking sector. An obsolete insolvency 18 percent of Cambodians could correctly answer framework compounds these risks. The feasibility three out of four questions related to basic financial of potentially introducing deposit insurance to partly literacy (risk diversification, inflation, numeracy, and mitigate these risks needs to be assessed. compound interest). Acknowledging this challenge, the National Bank of Cambodia (NBC) is already making Interest rates in microfinance efforts to improve financial consumers’ awareness of remain high due to operational their rights and responsibilities through the “Let’s Talk inefficiencies Money” campaign launched in 2015, which represents an important step in improving consumers’ capacity for High operational expenses have kept the cost of good financial decision-making. providing credit in the microfinance sector high, while high interest expenses and loan losses The increasing debt-to-consumption ratios and place additional cost pressures on MDIs and deteriorating portfolio quality in the microfinance MFIs, respectively.15 The average cost of credit in sector heighten Cambodia’s macroeconomic risks. Cambodia—consisting of interest rate expenses, An important source of risk is that MFIs (mostly MDIs) operating expenses, and loan losses—varies significantly are borrowing in dollars from abroad, which results in across different types of lenders. At the lower end are currency mismatches (e.g., for activities that generate commercial banks, with an estimated average cost of riel) and maturity mismatches (different loan length) and credit of 9.2 percent in 2016, followed by specialized adds to external private sector debt. At the same time, banks at 14.3 and MDIs at 15.7 percent. MFIs have the smaller MFIs tend to borrow from domestic financial highest cost of credit at 22.5 percent in 2016, since their 15  Cambodia’s lowest microfinance lending rates have been 25 percent and 18 percent per annum in local and foreign currency, respectively— lower than those in Thailand (36 percent) but higher than in Vietnam (13.5 percent). The Thai government has capped the microfinance lending rate at 36 percent since 2014 (The Bangkok Post, 23 October 2013), and the Vietnamese government has also imposed a microfinance lending rate cap of 13.5 percent since 2012 (Vietnam Microfinance Working Group, 2013). In the Philippines, the interest rates have been between 6-20 percent per year (Philippine Daily Inquirer, 31 December 2015). Cambodia Policy Note | 14 Figure 6. The cost of credit structure differs between MDIs and MFIs (a) Cost structure: 2011 – 2017, (b) Correlation between interest % of loan volume yield and cost of fund: 2012 - 2016 30 100 Income - loan ratio (%) 25 80 20 60 15 40 10 5 20 0 0 2011 2013 2015 2017 2011 2013 2015 2017 0 20 40 60 80 100 MDI MFI Interest expense Operating expense NPLs Profit Cost of providing credit - loan ratio (%) Source: Authors’ calculation using National Bank of Cambodia data. Note: Outlier MFIs with losses below 20 percent of their assets are excluded from the weighted average. business model implies catering to rural and potentially improved, while that of MFIs decreased (Figure 9 in the riskier clients. MDIs are more leveraged, relying mostly Annex). There is scope for improving outcomes with the on foreign capital, and interest expenses are around 40 same amount of resources, especially among MFIs. percent of their cost of credit. MFIs are less leveraged but have much higher operating expenses, which Econometric analysis of determinants of efficiency indicates operational inefficiencies (Figure 6, left panel).16 in the microfinance sector suggests that low staff Profits accounted for around 4 percentage points of the productivity and administrative expenses are the interest rate yield in recent years, which means that most important determinants (Table 6 in the Annex). the cost of providing credit is the primary driver of high With a higher asset base and broader geographical interest rates. A high correlation between the cost of coverage, larger MFIs and MDIs tend to be more providing credit and the interest rates charged has been efficient. To a limited extent, higher equity investments, found (Figure 6, right panel). which can spur lenders into efficiency-seeking reforms as shareholders demand higher returns, are also Inefficiencies due to misallocation of resources are associated with higher efficiency. The impact of interest the main driver of high operating costs in Cambodia. expenses on efficiency is not robust, while foreign While the median operating expense ratio (OER)17 in ownership is not a major determinant of microfinance Cambodia is lower than the median values for the world lending efficiency in Cambodia. These findings suggest and East Asia Pacific region, the cost per borrower is that redundant paperwork and processes and excessive significantly higher than in any other region (Table 5 in use of labor are key drivers of the relatively high cost the Annex). An efficiency benchmarking of microfinance of providing credit in Cambodia, especially among lenders in Cambodia using data envelope analysis MFIs. Both redundant paperwork and processes can yields a technical efficiency score of 83 percent in 2016 be largely attributed to underutilization of technology and 2017 compared to 88 percent in 2012. Between to provide credit at a lower cost. Interviews with MFIs 2014 and 2017, it appears that the efficiency of MDIs suggest that some of them have streamlined paperwork 16  The sum of the median for the different components does not add up to the median for the total cost of credit. 17  The operating expense ratio (OER) is a commonly used indicator for measuring the cost efficiency of an MFI, defined as operating expense divided by average gross loan portfolio. The following benchmark was used to assess efficiency of the sample MFIs: Highly efficient = OER ≤ 25%; Efficient = OER > 25% to 35%; Inefficient > 35% (ADB, 2012). Microfinance and Household Welfare | 15 and processes since the introduction of the interest already experiencing a fall in profitability. As of rate cap, which may be related to some reduction in December 2016, before the introduction of the interest operating costs observed in 2017. However, this is not rate cap in April 2017, microfinance lenders charged confirmed by the efficiency frontier analysis, since most interest rates ranging from 25-43 percent (for loans in changes were adopted in the second half of the year. KHR) and 18-41 percent (for loans in USD).18 Yet the return on equity for Cambodian MFIs has been lower The high cost of funds adds to the cost of providing than the regional and world averages (Table 1), and microfinance loans in Cambodia, especially for their return on assets has declined significantly since MDIs. The scaling up of macroprudential requirements 2015. In contrast, Cambodian MDIs—which are highly introduced by the National Bank of Cambodia since leveraged—have an astronomically higher return on 2016 has been a welcome step toward safeguarding equity than the world average and the averages for financial sector stability. At the same time, the increase South Asia and East Asia and the Pacific, although it in deposit reserve requirements and liquidity buffers dropped significantly following the introduction of the have raised the cost of funds for MDIs, which have interest rate cap. significant deposit amounts. The withholding tax on interest from loans acquired from abroad also adds to The interest rate cap has resulted in the reduction the cost of credit. A reduction in the tax from 14 to 10 of small loans in favor of larger loans but has not led percent, introduced by the Ministry of Economy and to the expected slowdown in overall credit growth. Finance in November 2017, has partly compensated for Econometric estimates accounting for seasonal the increase in the cost of credit due to other macro- fluctuations 19 show that the number of loans of USD prudential measures. 500 or less declined by 48 percent after the introduction of the interest rate cap (Figure 7). Nonetheless, total A structural challenge for most MFIs is the limited microfinance lending increased by 8 percent largely access to both domestic and foreign sources of due to a jump in the number of loans between USD funding. While domestic savings increased from an 500-1000 and to some extent those above USD 1000 estimated 11.6 percent of GDP in 2011 to 16.7 percent (Table 7 in the Annex). These estimates are backed by in 2017, they remain low by international standards. the NBC’s annual supervision reports, which show an In addition, the domestic financial market remains increase in total lending volume. Microfinance lenders fragmented and concentrated in retail markets with thus responded to the interest rate cap mostly by limited business-to-business interactions. As a result, increasing loan sizes and extending the repayment MFIs, which do not often enjoy access to international period. As mentioned above, the proportion of sources of funding, also face high cost of credit in the households in the first quintile (poorest) who borrowed domestic market. Those microfinance lenders who from a formal source declined by 5 percentage points borrow from international markets face high interest rates in 2017. due to the high institutional and country risk premium. The interest rate reduction introduced by the cap has The interest rate cap reduced been partly offset by increasing fees. The average issuance of small loans interest yield (interest income to loan ratio) decreased from 22.9 percent for MDIs and 21.7 percent for MFIs in The recent introduction of the 18 percent interest 2016 to around 18 percent in 2017 after the introduction rate cap has put pressure on MFIs, which were of the interest rate cap. To partly compensate for this, both 18  National Bank of Cambodia. 19  The introduction of the nominal interest rate cap on April 1, 2017, a structural break in the series, could be exploited to analyze the impact of the policy by comparing loan issuances immediately before and after the policy came into effect using regression discontinuity analysis. Given the high seasonality pattern of loan issuance in Cambodia, characterized by a dip in lending in the second quarter of the year, the analysis was done using seasonally adjusted figures. Cambodia Policy Note | 16 Table 1. Profitability in the microfinance sector varies significantly Profitability (median) Return on Equity Return on Assets 2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017 World 8.4 8.7 9.6 7.8 8.1 NA 1.8 2.0 2.0 1.9 1.8 NA South Asia 7.8 12.2 14.3 15.1 13.6 NA 1.5 2.2 2.5 2.5 2.5 NA East Asia and the 9.0 16.0 12.2 9.9 8.2 NA 1.9 3.5 3.0 2.6 1.6 NA Pacific Cambodia MDIs 20.0 21.5 23.7 29.4 25.4 13.8 3.4 3.3 2.9 3.3 3.1 2.2 Cambodia MFIs 6.0 5.1 5.6 6.2 5.1 1.2 2.3 2.2 2.3 2.8 1.3 0.4 Note: NA=Not available Source: Authors’ calculation using MIX Market (World, South Asia, East Asia and the Pacific) and National Bank of Cambodia data. Figure 7. The interest rate cap reduced the smallest loans but increased total lending (c) Average number of new micro-loans per commune (d) Average amount of new loans per commune 15 60000 Amount of New Loan (USD) Number of loans <$500 10 50000 40000 5 30000 0 2014w43 2015w41 2016w9 2017w37 2018w35 2014w43 2015w41 2016w9 2017w37 2018w35 Sample average within bin Polynomial fit of order 2 Sample average within bin Polynomial fit of order 2 Source: Authors’ calculation using Credit Bureau Cambodia Data. MDIs and MFIs tripled their non-interest income to loan Thus, average profitability in 2017 declined for MFIs, (mainly through fee increases) to 2.6 percent on average while it was sustained for MDIs. The mean net profit for MFIs. Notably, most microfinance lenders previously among MFIs dropped from USD 0.36 million in 2016 lending at less than 18 percent (mostly MDIs) appear to to USD 0.26 million in 2017. Meanwhile, average net have increased their interest rate upward, while those profit for MDIs increased from USD 16.9 million in 2016 previously lending at above 18 percent (mostly MFIs) to USD 17.8 million in 2017, compared to just USD 3.4 reduced their lending rates (Figure 8). As highlighted million in 2011. The percentage of MFIs making losses by the international experience, interest rate caps increased from 26 percent in 2016 to 35 percent in usually result in unexpected market distortions (Box 2). 2017, while no MDIs are making losses. Microfinance and Household Welfare | 17 Figure 8. While interest rates decreased in 2017, this was partly offset by non-interest income 54 45 36 27 Percentage ponts 18 9 0 -9 -18 -27 -36 Interest yield in 2016 Change in interest yield in 2017 Change in fees Interest cap Source: Authors’ calculation using National Bank of Cambodia, Annual Bank Supervision Reports: 2006- 2017. Box 2: The unintended effects of interest rate caps More than 75 countries, representing 80 percent of global GDP, have introduced interest rate caps with the aim of making credit cheaper and more accessible as well as protecting consumers from usury and exorbitant interest rates. These restrictions have taken many different forms, with some only affecting extreme pricing and others set at “binding levels” below market rates. In either case, the cap can be absolute (affecting all types of credit) or relative (varying based on the level of a benchmark rate). Rather than using a single limit, countries can also choose to set different caps based on the size or type of loan, type of financial institution, socio-economic characteristics of the borrower, or the industry. For example, Australia and Canada regulate payday loans, Nigeria has ceilings on mortgage rates, China has applied caps on MFI loans, and Vietnam has special rules for agricultural lending. However, interest rate caps often have substantial unintended effects, such as increases in non-interest fees, lower credit supply and loan approval rates for small and risky borrowers, and fewer institutions. For instance, when rates in Kenya were capped at 14 percent in 2016 (below the average of 18 percent), the average annual percentage rate (APR) rose to 18.5 percent, increasing the effective cost of loans. Therefore, some countries also regulate non-interest costs, such as Thailand’s 28 percent ceiling that includes both interest and fees. Low caps, however, may also result in reduced quantities of credit, given the high elasticity of credit supply to price changes, particularly for high-risk borrowers. For example, in Bolivia, the number of borrowers decreased by about 35,000 within two years after the introduction of interest rate caps in August 2013. Moreover, the credit market growth rate slowed to 16 percent per year, and importantly, the credit growth of small and medium enterprises became negative. Cambodia Policy Note | 18 Box 2: continued As a result, many countries have recently repealed or reformed their restrictions on interest rates. In 2013, Zambia introduced tiered caps for development MFIs, regular MFIs, and other financial institutions—all at levels below prevailing market rates—but repealed them in 2015 due to the negative effects they had on credit availability and fees. During this time, MFI credit growth dropped from 63 percent to 38 percent, while the ratio of fee income to interest income increased from 5.8 percent to 18 percent. Other countries that have repealed their restrictions include Argentina, which announced the elimination of all caps for credit operations in 2016, and China, which removed all restrictions on commercial lending rates in 2013 and scraped controls on deposit rates in 2015. Given these negative side effects, alternatives and complementary measures for reducing the cost of credit should be considered. These measures, which should be based on the initial source of the distortion, may include: (i) fostering competition, if incumbent financial institutions have significant market power that allows them to profit by setting elevated rates; (ii) strengthening regulatory and supervisory capacities, if limited competition is due to legal and institutional weaknesses rather than collusive behavior; (iii) promoting the creation of credit bureaus, if the reason for high interest rates is information asymmetry; (iv) improving the efficiency of loan foreclosure procedures that allow banks to limit the losses from borrower default; and (v) promoting consumer protection and financial literacy, if the aim is to protect consumers from usury rates. In some cases, a holistic macroeconomic solution may be required, which includes effective fiscal and debt management frameworks as well as capital market development. Source: Based on Ferrari et al., 2018. International Experience cheaper than an in-branch transaction.20 MFIs can also leverage mobile banking platforms for savings mobilization, loan disbursement, and repayments with International experiences point to potential reasonable investments in middleware to automate solutions for tackling challenges such as operational uploading and reconciling of repayment data. This inefficiencies and lack of alternative financial section focuses on the experiences of the Philippines instruments. Cambodia can learn from international and Kenya in supporting digital financing to innovate experiences on fostering product innovation and better on new business models—for example, using mobile delivery of microfinance to lower costs, improving banking for loan origination, application, and due regulations to maintain lending standards and reduce diligence.21 risks, and building the financial literacy of the population to make informed borrowing choices. For example, The Philippines – Policy and agent-based banking can reduce costs by using Regulatory Environment for a network of agents to handle loan disbursement, Microfinance repayments, and acceptance of deposits. A case study of FINCA in the Democratic Republic of Congo shows The policy and regulatory environment for that a single transaction at an agent was 50 percent microfinance in the Philippines is considered 20  Cazacu et al., 2016. 21  For further details, refer to Kumar et al., 2010; Metre, 2011; and Hanouch and Rotman, 2013. Microfinance and Household Welfare | 19 among the best in the world. For example, the savings account to be set up with Equity Bank through Consultative Group to Assist the Poor (CGAP) declared Safaricom’s mobile network. These savings accounts the Philippines’ microfinance industry as “the best in do not have account opening fees, minimum balance implementing programs to reduce poverty” during the requirements, or monthly charges. Only three years after International Year of Microcredit, 2005. More recently, the launch, 9.5 million customers had registered for the Philippines was ranked as one of the top five in the M-PESA, representing more than 45 percent of Kenya’s world in terms of policy and regulatory framework.22 The adult population and twice the number of Kenyans with basic policy framework for the microfinance industry bank accounts. in the Philippines has been the National Strategy for Microfinance, which moved from direct credit and Cross-country studies have found that digital subsidized funding to a market-based orientation, finance can benefit both borrowers and lenders. using government funds for capacity building purposes A 2013 study on mobile banking shows that MFI and emphasizing savings mobilization. The Philippines customers valued the time and cost savings of mobile has also encouraged MFIs to take full advantage of banking services for loan repayment.23 This was mainly technology (i.e. internet and mobile phone) and ensured because agents were more accessible than the MFI that regulations on electronic banking are up to date. branches. For MFIs, mobile banking could lower the The authorities have also considered setting up an cost of funding via deposits mobilized from customers, appropriate performance standard for microfinance. and these costs would be passed back to customers These initiatives offer useful examples for strengthening in the form of lower interest rates. Mobile banking also Cambodia’s microfinance sector to support the use creates opportunities for an innovative microfinance of technology, alternative delivery mechanisms, and business model to process loan applications by using improved standards. transaction history (i.e. initial savings, loan repayment, and saving patterns) as the key tool. This lending model Kenya – Mobile Banking Services does not require loan officers to visit the client’s house for credit decisions, which would reduce the burden The most successful mobile banking service is on loan officers and operating costs. It can therefore M-PESA in Kenya, which allows customers to increase staff productivity, which is a primary driver of make cash withdrawals, deposits, and transfers operational costs in Cambodia. using mobile phones. The main objectives of this service are to reduce low-value transaction costs, extend financial services to remote areas, and increase customer convenience by relying on agents instead of bank branches. The initial concept began as a pilot Policy Options project in 2003 to test a mobile phone-based solution A series of policy measures could help continue the for processing financial transactions in remote areas of expansion of access to credit to the most vulnerable Kenya. Having successfully completed the pilot project, groups, while lowering costs and improving lending a mobile phone operator (Safaricom) and a commercial standards in Cambodia. These measures could include bank (Equity Bank) launched M-PESA in 2007, with (i) introducing regulatory policies that enable reform of the main purpose of providing remittance services. the operation and delivery models for MFIs to increase However, one-fifth of M-PESA customers also used the their operational efficiency and reduce the cost of service as the main tool for saving. For this reason, the providing credit, (ii) addressing funding constraints to the Central Bank of Kenya allowed an interest-collecting microfinance sector, (iii) broadening financial instruments 22  Economist Intelligent Unit, 2012-2016. 23  Hanouch and Rotman, 2013. Cambodia Policy Note | 20 to meet household consumption borrowing needs, (iv) Government policies to promote transparency, strengthening consumer protection and empowerment foster competition on service quality, expose to enhance welfare impacts, and (v) improving inefficiencies, and increase investor demand for transparency, lending practices, and supervision to higher returns could spur efficiency-enhancing reduce risks. These areas resonate with the key strategic reforms. Such policies must be complemented areas laid out in the Financial Sector Development Plan by infrastructure investments and regulations that 2016-2025 regarding strengthening law enforcement incentivize and are supportive of innovative efficiency- and supervision of financial activities, enhancing financial enhancing delivery mechanisms and reduce operating literacy and consumer protection, seeking low funding costs for lenders. sources, identifying the factors that contribute to high costs of funds, and finding ways to reduce interest rates In the short term, MFIs can reduce operating with market mechanisms. Policy options in these areas costs by optimizing processes in four areas. First, are discussed in greater detail below. processes can be streamlined to reduce low-value activities and paper movement, forms can be simplified, Most of the policy options are cost neutral from and redundant functions should be eliminated, a budget perspective, since they require mostly especially in the credit assessment and approval regulatory changes but not significant investments. processes. Second, centralization of processing The exceptions are the medium-term items on functions can reduce efficiency losses. This would establishing e-Government and financial sector reduce costs significantly and improve quality compared systems, in areas (i) and (ii). to a decentralized delivery method in which branches autonomously undertake all credit processing functions Improve the efficiency of MFIs by (including credit approval), resulting in complex branch reducing operating costs structures with high numbers of staff. Third, the credit scoring process can be standardized and automated MFIs in Cambodia can improve operational to reduce time spent on credit assessment and the efficiency by adopting measures that increase number of staff involved, while shifting to risk-based the productivity of staff and reduce the costs of assessment. decentralized delivery to expand outreach. Such measures would include supporting the optimization Meanwhile, authorities should introduce a framework of business processes (including through leveraging and clarify regulations for the use of electronic technology) and adopting innovative delivery models to banking, mobile banking, and agent-based banking reduce the costs of distribution and increase outreach. in the microfinance sector. Mobile financial services Evidence across the world shows that MFIs sometimes are being adopted around the world as a cost-effective fail to improve efficiency because inefficiencies might mechanism for banks and other financial institutions remain hidden in a high-growth environment—as is the to reach new customers rapidly without needing new case in Cambodia—as lenders focus on portfolio growth physical branches. Cambodia does not have defined and expand to underserved markets.24 Furthermore, guidelines or requirements for the adoption of mobile microfinance clients at times struggle to determine and banking in the microfinance sector, which is assessed compare prices, thus reducing competition on product on a case-by-case basis. Similarly, agent-based quality and keeping prices high. Increased competition delivery models can help reduce transaction costs while on product service quality and earnings pressure driven expanding reach to remote areas. While some actors in by higher-equity investments in MFIs can drive lenders the microfinance sector are using Wing or TrueMoney to undertake efficiency-enhancing reforms. to collect debt service payments (Box 3), the adoption 24  Cazacu et al., 2016. Microfinance and Household Welfare | 21 Box 3: Can digital finance be the answer for Cambodia? Recognizing the significant flows of business transactions and remittances (largely dominated by physical transfers) between Cambodia’s rural and urban population, Wing has introduced digital banking services. Overall, 80 percent of Cambodia’s population lives in rural areas, and only around 20 percent of adults have access to financial services (World Bank Global Financial Inclusion: Global Findex 2014). Taking this opportunity, Wing introduced digital banking services—initially offering local money transfers using mobile phones and point-of-sale (POS) devices in 2009, then more recently diversifying to phone top-up, bill payment, online payment, payroll, and international money transfers. In 2014, Wing’s transaction amounts reached USD 4.5 billion with an estimated 1.5 million customers, of which 67 percent are in rural areas and 37 percent are women. Compared to the 1.9 million borrowers and 2.3 million depositors of MFIs, Wing has become a leading player in financial inclusion in Cambodia. Other commercial banks or microfinance institutes such as ACLEDA Bank, AMK, AMRET, and LY HOUR are also increasingly using technology and agents to serve their customers. Differing from other players, AMRET attempts to target different customers—mainly smallholder families—by developing digital goal-based savings (i.e. saving for a specific purpose such as the purchase of cows, child’s education, or child’s wedding), which allow family members to link their individual accounts so they can transfer money without transaction costs or can save jointly. of agents is being approved on a case-by-case basis, also be crucial to unleash the potential of electronic with no clear rules guiding the process. Authorities financial services. should provide a guiding framework and, if needed, adopt secondary regulations that facilitate the adoption Address MFI funding constraints by of these innovative delivery models while protecting minimizing regulatory impact and customers. diversifying sources Investments to support e-Government systems, Another way to reduce the cost of credit is by including biometric IDs, would help enhance credit addressing funding constraints to the microfinance assessment and financial transactions. Investing in sector, which may require optimizing the impact integrated digital information management systems of existing regulations. MDIs can leverage domestic with backend functions integration would help optimize deposits as a source of capital, but the existing processes by staff and facilitate the centralization of liquidity coverage requirements add to the cost. When core activities such as credit approvals. Government accessing funds from abroad, both MDIs and MFIs investments in areas such as an updated digital land end up in practice facing the cost of a withholding registry, a functional civil registry, and biometric IDs tax that would in principle be charged to the foreign would facilitate the linking of information for individuals, investor or lender. Authorities could conduct an land, and credit history, which would help improve assessment aimed at identifying measures to optimize credit scoring to enhance risk assessment as well as liquidity requirements and taxation on foreign capital. speed up transactions and reduce costs. Adopting the In addition, authorities could assess the feasibility of laws on Cybercrime and Electronic Commerce would establishing a microfinance apex institution25 that could 25  As discussed in Forster et al. (2012), “an apex is a second-tier (or “wholesale”) fund that channels resources to multiple retail financial providers— typically lenders—in a single country. Apexes provide mainly local currency loans, but may also offer loan guarantees, equity investment, grants for operational cost support, and technical assistance.” Cambodia Policy Note | 22 provide long-term wholesale funds in local currency to bodies for both markets could discuss and agree on how MFIs as well as, potentially, technical assistance (see to best establish procedures and complaint resolution Forster et al., 2012). mechanisms that are streamlined and satisfactory for financial and insurance products offered in a package. In the medium to long term, developing domestic In addition, authorities could facilitate the development capital markets and further building the financial of new products aimed at channelling remittances from system can help diversify the sources of financing migrants and targeting other savings for education and and reduce funding costs. Currently, capital markets lifetime events such as weddings and retirement. in Cambodia are underdeveloped, which limits the saving and investment options available to economic Strengthen consumer protection agents. Developing the legal and institutional framework and empowerment to reap the for a domestic debt market and starting to issue benefits of microfinance while sovereign debt would help provide a benchmark for reducing household vulnerability private sector corporate debt issuances, which could be another source of financing for the microfinance To maximize the benefits of microfinance while sector. Cambodia would also benefit from implementing reducing household vulnerability, borrowers need a National Payment System, which would allow for to be protected as well as empowered to optimize reliable and real-time payments and settlements and their financial decisions. The rise of new delivery could be used by different actors, ultimately supporting mechanisms raises new challenges for consumer the financial sector as well as broader economic protection, while predatory lending practices leave development. Concurrently, financial infrastructure households worse off and raise systemic risks. such as a good accounting system, reliable reporting Consumer capacity should be enhanced so consumers system, sufficient disclosure, strong governance system can better choose among lending options—from (both public and private), court systems, and reliable discerning the most attractive terms across lenders to insolvency regime should be built. The Financial Sector selecting the appropriate financial products for their Development Strategy 2016-2025 is a good starting needs, recognizing the risks and repayment limitations. point for this long journey of financial sector development The National Bank of Cambodia has taken some and must be implemented with the full engagement of important steps toward improving financial consumer all stakeholders to achieve success. protection (FCP) in recent years, including the creation of an FCP unit, the establishment of an in-house consumer Broaden financial instruments to hotline, and the adoption of regulations on resolution of meet households’ borrowing needs complaints about financial institutions. The launch of the “Let’s talk money” campaign and the integration of Going forward, the focus of the microfinance sector financial education in the formal education curriculum in Cambodia could shift from credit-led growth are significant steps toward strengthening financial to microinsurance, and new products could help literacy. Some additional policy options are proposed facilitate an increase in deposits and remittances. below. To provide appropriate products to meet households’ diverse borrowing needs, authorities need to develop Given its immediate impact, strengthening consumer secondary regulations that facilitate financial innovation protection should be a short-term policy priority. In while protecting customers. While health and funeral the short term, a financial protection law or consumer insurances already exist in Cambodia, distribution of protection law with explicit reference to financial these types of products remains limited. Currently, the services should be adopted to help shield borrowers bundling of financial and insurance products is not from predatory lending practices. It would also be allowed, which reduces their uptake. The regulating beneficial to expand access to enable more people to Microfinance and Household Welfare | 23 use deposits, in conjunction with the development of and Impairment Provisioning. Supervision would an action plan for establishing an autonomous deposit focus on systematic monitoring of business practices protection institution to provide guarantees for deposit in relation to payment, credit, and savings products, as accounts in regulated financial institutions, thereby well as on assessing compliance with macroprudential enhancing public trust in the banking system. Additional measures. Tighter supervision should be complemented medium-term measures include implementation of by rigorous application of pre-announced and graduated equal treatment requirements; development and sanctions on non-compliant institutions. The NBC can implementation of consistent, mandatory, binding, make regular use of off-site data analysis as part of its and enforceable external dispute resolution (EDR) monitoring and supervision activities. schemes; and establishment of specialized complaint handling units. In the age of big data and digital finance, Second, regulatory requirements and licensing strengthening data protection would also be critical. requirements are needed for the variety of actors providing microcredit. Currently, some micro-lenders These efforts can be complemented with measures such as NGOs and pawn shops are licensed under the aimed at strengthening financial literacy. Authorities Ministry of Interior and Ministry of Economy and Finance, can build on the current financial literacy campaign respectively, and fall outside the regulatory supervision and update the curricula to cover financial planning, of the NBC. Licensing and supervision regulations exposure to new products, and borrower rights. Potential should be revised to strengthen the NBC’s mandate borrowers should be walked through the customer for financial consumer protection across microfinance information sheet to help them understand the different providers. This would help curb unfair practices and interest rates and fees that add to the cost of credit. ensure a level playing field for all actors. This should be complemented with dissemination of financial planning toolkits, such as easy-to-use financial Finally, for mitigating risks, adopting lending calculators, that prospective borrowers could use to guidelines for the microfinance sector can be a good compare the financial implications of different loans or complement to macroprudential measures. A short- products. term priority would be for sector actors to agree on a code of conduct and common lending guidelines that Improve transparency, lending can help prevent predatory and risky lending practices. practices, and supervision to A concrete measure that could be introduced is reduce risks common incentives for credit officers that are linked not only to lending volume (which has led to the explosion Managing risks and containing NPLs are a key part of in loan size) but also to the number of customers and reducing the cost of credit in Cambodia, especially portfolio quality (NPLs). In addition, limits on loan size among MFIs which have a riskier portfolio. Excessive to income and refinancing could be established. Banks risk lending behaviors that can result in deteriorating who operate in the microfinance sector should be loan portfolio quality should be discouraged. Increasing subject to the same lending practices. In the medium to accountability and reporting in the sector to expose long term, the Cambodia Microfinance Association, in hidden inefficiencies and systemic weaknesses is also which all sector actors must be registered by law, could important. Tighter supervision, stricter enforcement be empowered to develop and enforce self-regulation, of regulations, and harmonization of regulatory and which would require greater resources and capacity licensing requirements are therefore required. building. This would help alleviate the NBC’s oversight burden and would ultimately render some restrictions First, sector supervision should be enhanced, such as the interest rate cap unnecessary. including by fully implementing and closely monitoring the new Prakas on Credit Risk Grading Cambodia Policy Note | 24 Table 2. Summary of policy options Objectives Short-term options (1-2 years) Medium- and long-term options (3+ years) i. Improve the efficiency of • Introduce a framework and clarify • Invest in supporting microfinance institutions regulations on the use of electronic e-Government systems, including by reducing operating banking, mobile banking, and agent- digital ID to enhance credit costs based banking in the microfinance sector assessment • Optimize processes to reduce operating costs ii. Addressing funding • Evaluate the impact of liquidity • Set up systems and regulations constraints to requirements and taxation on the cost of to develop domestic capital microfinance institutions funds for financial institutions markets by minimizing regulatory • Consider the feasibility of introducing a • Implement the national payment impact microfinance apex institution that could system strategy provide long-term funding iii. Broaden financial • Promote savings through innovative instruments to meet products aimed at capturing remittances households’ borrowing and targeting lifetime events needs iv. Strengthen consumer • Enhance financial literacy outreach by • Establish a deposit protection protection and focusing on financial planning, exposure to institution empowerment to new products, and borrower rights • Implement consistent, mandatory, reap the benefits of • Deploy easy-to-use and accessible binding, and enforceable external microfinance while financial planning and compliance dispute resolution schemes reducing household verification toolkits vulnerability • Adopt a Consumer Protection Law, with explicit reference to financial services v. Improve transparency, • Implement the Prakas on Credit Risk • Empower financial sector lending practices, and Grading and Impairment Provisioning associations to develop and supervision to reduce • Harmonize regulatory requirements and enforce self-regulation risks licensing requirements for micro-credit actors (including pawn shops and rural credit providers) • Adopt a code of conduct and lending guidelines for the microfinance sector Microfinance and Household Welfare | 25 References Duflo, Esther, Abhijit Banerjee, Rachel Glennerster, and Cynthia G. Kinnan. The Miracle of Microfinance? Evidence from a Randomized Evaluation, Working Asian Development Bank. Microfinance Development Paper. National Bureau of Economic Research, 2013. Strategy 2000: Sector Performance and Client Welfare. Manila: Asian Development Bank, 2012. Economist Intelligence Unit. “Global Microscope on the Microfinance Business Environment.” Accessed March Akotey, Joseph Oscare, and Charless K.D. Adjasi. Does 6, 2018. www.eiu.com Microcredit Increase Household Welfare in the Absence of Microinsurance. Vol. 77, World Development, 380- Ferrari, Aurora, Oliver Masetti, and Jiemin Ren. Interest 394. Amsterdam: Elsevier, 2016. Rate Caps: The Theory and The Practice. Policy Research Working Paper, No. 8398. Washington, DC: Banerjee, Abhijit, Esther Duflo, Rachel Glennerster, and World Bank. 2018 Cynthia Kinnan. The Miracle of Microcredit? Evidence from a Randomized Evaluation, Working Paper, Boston: Forster, Sarah, Eric Duflos and Richard Rosenberg. MIT, Department of Economics, 2014. A New Look at Microfinance Apexes. Focus Note, Consultative Group to Assist the Poor (CGAP), 2012. Cazacu Diana, Tatyana Dolgaya, Till Bruckner, Michael Kortenbusch. Potential for Cost-Reducing and Efficiency Ganle, John Kuumuori, Kwadwo Afriyie, and Alexander – Increasing Measures in Financial Institutions. Frankfurt: Yao Segbefia. Microcredit: Empowerment and KfW Development Bank, 2016. Disempowerment of Rural Women in Ghana. Vol.66, World Development, 335-345. Elsevier, 2015. Chantanusornsiri, Wichit. “Maximum rate for microfinance at 36%,” The Bangkok Post, October Gertler, Paul, David I. Levine, and Enrico Moretti. 23, 2013. http://www.bangkokpost.com/business/ Do Microfinance Programs Help Families Insure news/375909/maximum-rate-for-microfinance-at-36 Consumption Against Illness? Vol. 18, Health Economics, 257-273. Boston: Institute for Economic Colman, Brette E. The Impact of Group Lending Development, Boston University, 2008. in Northeast Thailand. Vol.60, Issue 1, Journal of Development Studies, 105-141. Amsterdam: Hanouch, Michel, and Sarah Rotman. Microfinance and Elsevier, 1999. Mobile Banking: Blurring the Lines? Washington, DC: Consultative Group to Assist the Poor (CGAP), 2013. Crepon, Bruno, Florencia Devoto, Esther Duflo, and William Pariente. Estimating the Impact of Microcredit Imai, Katsushi S., Thankom Arun, and Samuel Kobina on Those Who Take it up: Evidence from a Randomized Annim. Microfinance and Household Poverty Reduction: Experiment in Morocco. Boston: Department of New Evidence from India. Vol. 38, Issue 12, World Economics, MIT, 2014. Development, 1760-1774. Elsevier, 2010. de Mel, Suresh, David McKenzie, and Christopher Imai, Katsushi S., and Md. Shafiul Azam. Does Woodruff. Returns to Capital in Microenterprises: Microfinance Reduce Poverty in Bangladesh? New Evidence from a Field Experiment, Policy Research Evidence from Household Panel Data. Vol 48, Issue 5, Working Paper. Washington, DC: World Bank, 2008. Journal of Development Studies, 633-653. 2012. Cambodia Policy Note | 26 Islam, Asadul, and Chongwoo Choe. Child Labor and Poor Rural Households in Bolivia. Vol. 36, Issue 11, Schooling Responses to Access to Microcredit in Rural World Development, 2440-2455. Elsevier, 2008. Bangladesh. Vol.51, Issue 1, Economic Inquiry, 46-61. Western Economic Association International, 2013. Metre, Kanika. Using Mobile Banking Services to Improve Financial Access for the Poor: Lessons from Islam, Asadul, and Pushkar Maitra. Health Shocks and Kenya, the Philippines, the United States, Haiti and Consumption Smoothing in Rural Households: Does India. Vol. 18, Policy Perspectives, 6-24. Washington, Microcredit Have a Role to Play? Vol. 97, Issue 2, DC: The George Washington University, 2011. Journal of Development Economics, 232-243. Elsevier, 2012. McKernan, Signe-Mary. The Impact of Microcredit Programs on Self-Employment Profit: Do noncredit Karlan, Dean, and Martin Valdivia. Teaching Program Aspects Matter? Vol. 84, Issue 1, Review of Entrepreneurship: Impact of Business Training on Economics and Statistics, 93-115. Boston: MIT Press Microfinance Clients and Institutions. Vol. 93, Issue for the John F. Kennedy School of Government, 2002. 2, Review of Economics and Statistics, 510-527. Boston: MIT Press for the John F. Kennedy School of MIMOSA. Microfinance Index of Market Outreach and Government, 2011. Saturation: Cambodia. MIMOSA Project, 2016. Khandker, Shahidur R., and Hussain A. Samad. Dynamic Morduch, Jonathan. Does Microfinance Really Help Effects of Microcredit in Bangladesh, Policy Research the Poor? New Evidence from Flagship Programs in Working Paper. Washington, DC: World Bank, 2014. Bangladesh. 1998. Khandker, Shahidur R. Microfinance and Poverty: Phim, Runsinarith. The Impacts of Microcredit on Poverty Evidence Using Panel Data from Bangladesh. Vol. 19, Reduction – Evidence from Cambodian Rural Villages. World Bank Economic Review, 263-286. Washington, Vol. LI, No.2, The Philippine Review of Economics, 121- DC: World Bank, 2005. 150. Quezon: University of the Philippines School of Economics and the Philippine Economic Society, 2014. Khandker, Shahidur R. Fighting Poverty with Microcredit: Experience in Bangladesh. New York: Oxford University Pitt, Mark M., and Shahidur R. Khandker. The Impact Press, 1998. of Group-based Credit Program on Poor Households in Bangladesh: Does the Gender of Participants Matter? Klapper, Leora, Annamaria Lusardi, and Peter van Issue 106, Journal of Political Economy, 958-996. Oudheusden. Financial Literacy Around the World: Chicago: University of Chicago Press, 1998. Insights from the Standard and Poor’s Rating Services Global Financial Literacy Survey. 2014. Pitt, Mark M. and Shahidur R. Khandker. Household and Intrahousehold Impact of the Grameen Bank and Similar Kumar, Kabir, Claudia McKay, and Sarah Rotman. Targeted Credit Programs in Bangladesh. Washington Microfinance and Mobile Banking: the Story So Far. DC: World Bank, 1996. Washington, DC: Consultative Group to Assist the Poor (CGAP), 2010. Pitt, Mark M., Shahidur R. Khandker, Signe-Mary McKernan, and M. Abdul Latif. Credit Programs for Maldonado, Jorge H., and Claudio Gonzάlez-Vega. the Poor and Reproductive Behaviors in Low Income Impact of Microfinance on Schooling: Evidence from Countries: Are the Reported Causal Relationships Microfinance and Household Welfare | 27 the Result of Heterogeneity Bias? Vol. 36, No. 1, Demography, 1-21. 1999. Pitt, Mark M., Shahidur R. Khandker, and Jennifer Cartwright. Empowering Women with Micro Finance: Evidence from Bangladesh. Vol. 54 No. 4, Economic Development and Cultural Change, 791-831. Chicago: University of Chicago Press. 2006. Roodman, David, and J. Morduch. The Impact of Microcredit on the Poor in Bangladesh: Revisiting the Evidence. Washington, DC: Center for Global Development, 2009. Roth, Vathana, Abdelkrim Araar, Bopharath Sry, and Dalis Phann The Dynamics of Microcredit Borrowings in Cambodia. Partnership for Economic Policy, Working Paper 2017-17, Nairobi: 2016. Seng, Kimty. Rethinking the Effects of Microcredit on Household Welfare in Cambodia. In Journal of Development Studies. 2017. Vietnam Microfinance Working Group. Bulletin Microfinance Regulation in Vietnam, No. 19. 2013. World Bank. Finance for All? Policies and Pitfalls in Expanding Access. Washington, DC: World Bank, 2008. World Bank. Improving Macroeconomic and Financial Resilience, Cambodia Economic Update. Washington, DC: World Bank, 2016. World Bank. The Global Findex Database: Measuring Financial Inclusion and the Fintech Revolution. Washington, DC: World Bank, 2017. Cambodia Policy Note | 28 Annex 1: Regression Results Table 3. The Characteristics of Formal and Informal Borrowers (Probit model) 1=formal borrower Cambodia Urban Rural Household size 0.070*** 0.094*** 0.063*** Dependent ratio -0.048** -0.114** -0.036 Household gender (1=male) -0.060 -0.040 -0.060 Household age (years) 0.007*** 0.003 0.008*** Household marital status (1=married) 0.139** 0.150 0.143* Household ethnic (1=Khmer) 0.613*** 0.231 0.690*** Education (reference category=no schooling) Household education (1=primary incomplete) -0.239** -0.287** -0.168 Household education (1=primary completed) -0.089 -0.214 -0.003 Household education (1=secondary completed) -0.147 -0.195 -0.084 Occupation (reference category=unemployed/retired) Household occupation (1=agriculture) 0.031 -0.287** 0.130** Household occupation (1=industry) 0.223*** -0.103 0.322*** Household occupation (1=service) 0.280*** -0.110 0.422*** Log of agricultural land (ha) -0.028 0.030 -0.031 Other control variables Urban-Rural dummy Yes Geographical dummy Yes Yes Yes Year dummy Yes Yes Yes Note: * = significant at 10%, ** significant at 5%, *** significant at 1%. Source: Authors’ calculation using CSES 2004, 2007-2016. Microfinance and Household Welfare | 29 Table 4. Summary Estimate Results of Welfare Effects (one-to-five matching) 2009 2012 2014 2015 2016 Household enterprise (loan access) % households engaged in non-agricultural activities 5.68*** 6.62*** 8.11*** 4.93** 5.88*** Household shock (loan access) Health expenditure per capita (ill members) (%) 11.55** 19.53*** 25.89*** 29.60*** 21.76*** Welfare (loan access) Food consumption per capita per day (%) 2.2 -3.04 -3.52*** 3.1 1.37 Nonfood consumption per capita per day (%) 8.06*** 5.2 8.03*** 9.56*** 11.14*** Total consumption per capita per day (%) 3.70** -1.05 4.57 5.88*** 5.89*** Poverty rate (%) -2.67* -3.1 -0.11 -4.89*** -2.77** Poverty gap (%) -1.56*** -0.83 -0.26 -1.32*** -1.06*** Agricultural activities (actual loan) Agricultural inputs (%) 4.65*** 12.87* 11.98*** 16.56** 28.33*** Agricultural income per capita per month (%) 0.3 1 -10.08*** -1.5 -3.46 Note: * = significant at 10%, ** significant at 5%, *** significant at 1%; one-to-one matching and kernel matching is also statistically significant. Source: Authors’ calculation using CSES 2009-2016. Table 5. International Comparison of Efficiency Efficiency (median) Operating Expense/GLP Cost per borrower (USD) 2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017 World 16.2 17.0 18.3 17.4 15.1 N/A 156 167 181 169 156 N/A South Asia 10.9 11.5 10.9 9.2 9.3 N/A 18 19 21 21 21 N/A East Asia and the 12.5 18.3 18.1 17.0 16.3 N/A 64 87 105 98 131 N/A Pacific Cambodia - MDIs 12.2 11.0 9.7 7.9 8.5 6.7 107 111 126 125 145 141 Cambodia - MFIs 14.1 13.7 11.2 12.5 13.0 14.3 290 241 204 296 526 635 Note: GLP=Gross Loan Portfolio. N/A= Not available. Source: Authors’ calculations using Mix Market and National Bank of Cambodia data. Cambodia Policy Note | 30 Figure 9. Efficiency among MDIs have increased over the past two years, whereas it declined for MFIs Efficiency score, weight average CRS-Technical Eefficiency, weight average 100% 100% 80% 80% 60% 60% 40% 40% 2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017 Technical efficiency Pure technical efficiency MDIs MFIs Total Scale efficiency Source: Authors’ calculation using National Bank of Cambodia data. Note: CRS: Constant return to scale. Table 6. The Determinants of Microfinance Efficiency – Zero Inflated Beta Regression Model Proportion One Inflate CRS_TE VRS_TE SCALE CRS_TE VRS_TE SCALE Foreign ownership -0.003 -0.002 -0.002 0.006 0.006 0.006 Loan per staff 2.966*** 4.274*** 0.405 7.837*** 9.563*** 7.837*** Loan size (log) -0.623*** -0.592*** -0.406*** -2.651*** -2.512*** -2.651*** Asset (log) 0.508** 0.556** 0.09 1.566** 2.369*** 1.566** Equity (log) 0.694* 0.775* 0.461 2.600* 2.530* 2.600* Interest expense (log) 0.002 -0.035 0.080** -0.544*** -0.568*** -0.544*** Operating expense (log) -0.701*** -0.860*** 0.035 -1.548** -2.642*** -1.548** Borrower density 1.003 1.742*** -1.223 4.921*** 5.145*** 4.921*** Constant 3.982*** 4.166*** 4.257*** 19.133*** 17.066*** 19.133*** Note: Dependent variable is technical efficiency scores based on constant return to scale. Similar results are obtained when estimates are based on variable returns to scale technical efficiency scores. Source: Authors’ calculation using National Bank of Cambodia data. Microfinance and Household Welfare | 31 Figure 10. Loan by sector reported by MFIs/MDIs 100% Transportation 80% Trade and Commerce Services 60% Other 40% Household 20% Construction Agriculture 0% 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Source: Authors’ calculation using National Bank of Cambodia data. Table 7: The effects of interest rate cap on microfinance sector by MDIs and MFIs MDI+MFI MDI MFI (1) (2) (3) (1) (2) (3) (1) (2) (3) Commune averages per week Number of loans <500 USD -4.77*** -4.85*** -4.85*** -5.92*** -6.25*** -6.25*** 1.21*** 1.25*** 1.25*** Number of loans 500-1000 USD 1.99*** 1.96*** 1.96*** 1.85*** 1.79*** 1.79*** 0.29*** 0.31*** 0.31*** Number of loans 1000-2000 0.62*** 0.63*** 0.63*** 1.08*** 1.12*** 1.12*** 0.08*** 0.07*** 0.07** USD Number of loans >2000 USD 1.10*** 1.11*** 1.11*** 1.51*** 1.54*** 1.54*** 0.03 0.05 0.05 Number of loans - all -1.22** -1.75*** -1.75*** -2.07*** -2.42*** -2.42*** 1.37*** 1.42*** 1.42*** Total loan amounts (USD) 3310** 3299** 3299** 3446*** 3212*** 3212*** 1216 1388* 1388 Individual loan characteristics Average loan size (USD) 746*** 739*** 739*** 1171*** 1190*** 1190*** 253** 240* 240* Loan duration (days) 205*** 208*** 208*** 306*** 310*** 310*** -19* -16 -16 Loan types Share of loan <500 USD -0.18*** -0.18*** -0.18*** -0.22*** -0.22*** -0.22*** 0.11*** 0.12*** 0.12*** Share of loan 500-1000 USD 0.09*** 0.09*** 0.09*** 0.10*** 0.10*** 0.10*** 0.03*** 0.03*** 0.03*** Share of loan 1000-2000 USD 0.03*** 0.03*** 0.03*** 0.04*** 0.04*** 0.04*** -0.01* -0.01* -0.01 Share of loan >2000 USD 0.06*** 0.06*** 0.06*** 0.10*** 0.10*** 0.10*** -0.11*** -0.11*** -0.11*** Source: Authors’ calculation using Credit Bureau Cambodia Data. Note: The control variables included borrower characteristics such as age, gender, and self-employed; urban area;a and the number of holidays. The results presented in Table 1 are those with polynomial order 2 option. Those with polynomial order 3 option are not reported. The main text discusses those results that are consistent. (1) = Conventional. (2) = Bias-corrected. (3) = Robust a. Urban area is defined at the district level within each province, i.e. one district per province. This definition is not in line with that of the National Institute of Statistics (NIS), which defines urban/rural area at the commune level. If the analysis used the NIS definition to identify urban/rural, a certain number of communes were unidentified, and those communes accounted for 35 percent of total loan portfolio in each year. This amount was too big to exclude from the analysis. Cambodia Country Office Exchange Square Building Floor 10th IBRD and 11th IFC Streets 51-61 and streets 102-106 Sangkat Wat Phnom, Khan Daun Penh Phnom Penh, Cambodia www.worldbank.org/cambodia