1
The impact of the current economic and financial crisis on microfinance
Msc. Thesis Corporate Finance, University of Groningen Sascha Huijsman1
January 18th, 2010 Supervisor: Robert Lensink
I. Abstract
This thesis provides a preliminary study of the impact of the current economic and financial crisis on microfinance institutions (MFIs). Results from our survey with 82 MFI managers worldwide show that as a consequence of the crisis MFIs face higher financing costs, less availability of funding, changed demand for loans and a deterioration of clients’ repayment behavior. Using a panel of monthly financial data of 57 MFIs over the period January 2007-August 2009, we investigated the impact of the current financial crisis by studying structural breaks in the times series of MFI performance indicators related to profitability, growth and portfolio quality. Results show that all performance indicators experience a significant negative shift, but that the timing of these shifts differs across the performance indicators. Profitability and growth are the first to experience a significant downward shift in the last quarter of 2008. Earliest signs of an adverse impact on clients’ repayment behavior are also visible late 2008, but a structural break in actual loan losses does not occur until August 2009. MFIs in Eastern Europe & Russia are most affected whereas MFIs in South America show the highest resilience to the crisis. The extent of the impact depends on the funding structure of MFIs. MFIs attracting savings are significantly less affected.
1 The author thanks SNS Asset Management for financial support
2 II. Preface and Acknowledgments
As early as November 2008 I talked with Theo Brouwers (SNS Asset Management) about potential research topics relevant for the microfinance sector and for the SNS Institutional Microfinance Fund (SIMF) which I could investigate for my master’s thesis.
We discussed current issues dominating the microfinance sector: the commercialization of the microfinance sector and the potential impact of the financial crisis on MFIs. We decided to investigate the latter. At that time we could not foresee the scale and scope of the crisis and how long it would last, a fact which, one year and a half later, is still undetermined. SNS Asset Management announced in their SIMF annual report of 2007 that they expected MFIs not to remain insulated from the effects of the credit crunch.
Although at the time many believed that microfinance was decoupled from events on mainstream markets, SNS turned out to be right: our research shows a clear adverse impact of the current financial crisis on MFIs.
I would like to show my gratitude to Theo for giving me the chance to combine my master thesis with an internship at SNS Asset Management. For me this was a unique opportunity. Access to the expertise of the SIMF team was very inspiring and essential for developing the ideas for my research. Also being allowed to join the IC meetings was a very nice experience and taught me a lot about how investment decisions are made.
Bart, thank you for supervising me during my internship at SNS. I really appreciate the time you have spent on discussing the setup and the content of my research with me.
Your feedback was very constructive. Ruben and Sinisa, thank you both for sharing your expertise with respect to finance and microfinance and for your input which was very useful for my research. Alexander, thank you for your helpful feedback on the MFI reports. I learned a lot from all of you.
I would like to thank Developing World Markets and Triple Jump for participating in this research and for enabling us to approach the MFIs in their portfolios for the questionnaire. Vivek Pradhan (DWM): I very much appreciate your involvement in the development of the questionnaire and for sending all the financial data sheets. Josefina
3
Aguilera and Steven Evers (Triple Jump): thank you very much for your feedback and for the cooperation in sending the questionnaire. Gerlof de Korte (Visionfund): I am thankful that you were willing to test the survey and for your useful comments which enabled me to improve it.
To all MFIs participating in the Survey: I would like to show my genuine gratitude for the insights you have provided through responding to the questionnaire. I very much appreciate the time you have taken to fill out the survey. Without your contribution it would have been impossible to write this thesis.
I would like to thank Damian von Stauffenberg (Microrate) and Harry Clemens (Hivos), for sharing their observations and experiences from the field regarding the impact of the current financial crisis on microfinance institutions.
Robert, thank you for supervising me during this research project. I am thankful for your supervision, guidance and close involvement during this process. It was a very good experience to write the Dutch article together. I learned a lot from your expertise in microfinance.
4
Table of contents
I. Abstract ... 1
II. Preface and Acknowledgments ... 2
1. Introduction ... 6
2. Background ... 9
2.1 Theoretical background: the relation between microfinance and formal markets .... 9
2.2 Empirical background: the relation between microfinance and formal markets .... 11
2.3 Macroeconomic conditions and microfinance in the current context ... 15
2.4 Recent studies on the impact of the current financial crisis on MFIs ... 18
3. Theoretical framework, Data, Methodology and Sample description ... 22
3.1 Theoretical framework and hypotheses ... 22
3.1.1 Theoretical Framework ... 22
3.1.2 Overview of the expected effects ... 25
Profitability ... 25
Portfolio quality ... 26
Growth ... 26
Across regions ... 27
Across MFIs ... 27
3.2 Samples, Data and Methodology ... 28
3.2.1 General remarks ... 28
3.2.2 Perception survey ... 30
3.2.2.1 Sample description ... 30
3.2.2.2 Survey description ... 31
3.2.2.3 Methodology for analyzing the survey results ... 33
3.2.3 Financial performance study ... 34
3.2.3.1 Sample description ... 34
3.2.3.2 Description financial dataset ... 35
3.2.3.3 Methodology for analyzing the financial data ... 40
4. Results ... 43
4.1 General effects ... 43
4.2 Profitability ... 45
5
4.3 Portfolio quality ... 47
4.4 Growth ... 48
4.5 Across regions ... 50
4.6 Across MFIs ... 56
5. Main Conclusion ... 62
6. Further Research ... 64
Figuur 1 Broad framework impact financial and economic crisis on MFI performance .. 22
Figure 2 Dsitribution of no. of MFIs across months ... 36
Figure 3 Survey results: Main Effects of the current crisis observed by 82 MFIs ... 44
Figure 4 Survey results: Reasons for deterioration of portfolio quality in 2008/2009 ... 47
Table 1 Description of performance indicators ... 38
Table 2 Abbreviations of MFI characteristics ... 39
Table 3 Results of the financial performance study: Effects of the crisis ... 43
Table 4 Results of the financial performance study: Impact across regions ... 51
6 1. Introduction
Microfinance provides small entrepreneurs, who do not have access to formal financial markets, with financial services. It is built on the premise that it will create a more inclusive financial sector, increasing the efficiency and professionalism in delivering financial services to the poor. Microfinance is by many perceived as one of the most effective tools to fight poverty as it promotes the economic development of the poor presumably leading to a higher standard of living. It is estimated that there are approximately 10.000 microfinance institutions (MFIs) worldwide serving more than 155 million clients (Microfinance Summit Campaign, 2009).
This paper focuses on the impact of the current financial and economic crisis on microfinance institutions. Some expected microfinance to be highly resilient to the current crisis (Microcapital, 2008) based on the common understanding that the performance of MFIs exhibits little correlation with performance of formal financial and economic markets. Empirical studies partly support this latter assumption, but results are ambiguous (Ahlin, 2006; Gonzalez, 2007; Hermes and Meesters, 2009). Moreover, the few available case studies specifically focusing on MFI performance during times of economic distress show that most MFIs are not immune to regional and domestic financial and economic crises (Marconi and Mosley, 2006; Jansson, 2001; Mcguire and Convoy, 1998).
Insight into the exposure of MFIs to adverse market shocks has become more relevant due to: (1) the growing importance of microfinance as part of local financial systems of developing countries (Gonzalez 2007) and (2) the increased involvement of commercial investors in the microfinance sector in recent years. The latter are typically more concerned about the risk of their investment than the traditional non-profit driven microfinance investors.
The current financial crisis provides an opportunity to test the resilience of MFI performance during a period of global market distress and to investigate the linkages
7
through which today’s microfinance sector is integrated into formal economies and financial systems. To date, no quantitative empirical investigation of MFI performance during times of economic distress has been carried out, neither during financial crises in the past, nor during the current financial crisis. In this paper we aim at filling this gap by empirically analyzing a panel of monthly financial data of 57 MFIs worldwide2 covering the period from January 2007 through August 2009. Fixed effects regressions are used to study structural breaks3 in MFI performance indicators related to profitability, growth and portfolio quality. This empirical analysis is complemented with the results of a survey among 82 MFI managers over the period April 2009 through May 2009. This survey explores the effects of the crisis on MFIs observed or expected by MFI managers.
It attempts to grasp the underlying mechanisms through which the current financial crisis affects MFI performance. Lastly, based on both the survey among the 82 MFI managers and financial data of the 57 MFIs we examine whether the impact differs across MFIs in different regions and with different characteristics.
Considering the limited period over which the data were collected, this study should be perceived as a preliminary analysis of the impact of the current crisis on microfinance. At this point in time it is not clear how the crisis will evolve, let alone what its final consequences for the microfinance sector will be.
The paper is organized as follows. We will first provide some background by discussing related literature and the current context. Secondly, we provide a framework describing the channels through which the crisis is expected to affect MFIs’ performance. Thirdly, we describe the sample, the data and the methodology used to measure and to analyze the
2 All MFIs investigated for this paper are part of the portfolios of SNS Asset Management, Developing World Markets and Triple Jump. These are all social commercial investors which in general invest in the best managed MFIs. Despite this bias, the types of MFIs in our sample serve the vast majority of microfinance clients worldwide (see chapter 3.3 for further elaboration on the sample). In the financial dataset no MFIs from Africa and the Middle East are included.
3 A structural break is defined as the unexpected shift in a time series.
8
impact of the crisis. Lastly, we will discuss the results and we will finalize with a conclusion.
9 2. Background
2.1 Theoretical background: the relation between microfinance and formal markets
Muhammad Yunus, perceived as ‘the founder’ of microfinance and winner of the Nobel Prize in 2005, stated in august 2008 that “despite the turmoil on financial markets, microfinance still works” (Microcapital, 2008). This statement confirms the common understanding that MFIs exhibit little correlation with formal markets which is considered to be one of the reasons why commercial investors have been increasingly attracted to investments in microfinance (Deutsche Bank Research, 2007) and why equity of microfinance institutions should deserve a premium over traditional finance (JP Morgan, 2009).
There are several explanations behind the common assumption of a low exposure of MFI performance to financial- and economic market movements4
4 As the focus of our study is the impact of an adverse market shock on MFIs, we discuss theories about the exposure of MFI performance to financial and economic markets mainly in the context of downside market risk.
. Firstly, MFIs were traditionally capitalized through social funding provided by institutions such as government agencies, development banks and NGOs. Availability and cost of this type of funding is generally expected to have a lower correlation with business cycles and to provide MFIs with continuous access to funding. For example, Fonseca (2004) found international assistance to MFIs not to be curtailed during the Latin American banking crisis because international financial institutions acknowledged the importance of microfinance to local economies. Besides, Walter and Krauss (2009) characterize MFI shareholders by having long term strategic interests and by being less driven by market forces, suggesting that MFIs do not have to fear sudden reversals of international capital
10
in times of unfavorable domestic market conditions. Secondly, the typical short maturities of microfinance loans in combination with the generally long maturities of MFIs’
liabilities, create a favorable maturity mismatch (JP Morgan, 2009). This positive mismatch gives MFIs the flexibility to adjust in times of tight liquidity by restricting the outstanding loan portfolio in order to meet debt obligations. Also when interest costs increase, MFIs can quickly adapt lending terms of outstanding loans due to the short maturities of these assets. Thirdly, Walter and Krauss (2008) add the generally low financial and operational leverage of MFIs compared to those of mainstream banks resulting in low fixed costs. The latter implies that operations can be scaled back at a low cost in times of economic downturn or in times of tight liquidity.
Lastly, for several reasons MFIs’ portfolio quality is often assumed to be resilient to adverse market shocks. First of all, short periods between installments, considered to be typical for microfinance loans, give MFIs the chance to early signal bad repayment behavior and hence to restrict credit to clients who fail to meet debt obligations or to adjust lending practices accordingly (Walter and Krauss, 2009). Second of all, microfinance clients have high incentives to repay5
5 Incentives of microfinance clients are high as a result of: (1) unconventional lending techniques typical for microfinance such as group-lending (2) close ties between MFIs and clients (3) access to (larger and better) future microfinance loans on the condition that clients show good repayment behavior and (4) related to the previous incentive; microfinance often represents the only and most reliable option for microfinance clients to borrow, which makes clients even more motivated to maintain access to funding from the MFI.
making the ‘price’ of non-repayment high to borrowers. Clients are therefore expected to put more at stake in order to repay and hence to show solid repayment also when they are negatively affected during periods of economic downturn. Thirdly, microfinance clients are considered to be quick adapters to changed conditions (Fonseca, 2004). The relatively low fixed cost at which the typical small business operates, enables it to cut prices in times of economic downturn or to switch to less cyclical sectors. Fourthly, microfinance clients are often assumed to be relatively isolated from the state of the economy as they are typically employed in the informal sector which is considered to be less responsive to business-cycles (Galema,
11
Lensink and Spierdijk, 2009). Demand for products from microenterprises may even flourish in times of economic recession as consumers possibly buy domestically produced cheaper products as a substitute for luxury and imported goods (Patten, Rosengard and Johnston, 2001). This would provide growth opportunities for microenterprises, in its turn increasing the demand for microfinance loans and improving repayment rates. In addition, in times of economic downturn fired workers from the formal sector may start income generating activities in the informal sector, also increasing the demand for microfinance loans. On the other hand, as Hermes and Meesters (2009) point out, macroeconomic conditions may affect clients in many different ways. During times of economic recession overall lower demand for products could also reduce business opportunities for small enterprises, in its turn diminishing demand for microfinance loans and deteriorating repayment capacity of borrowers.
2.2 Empirical background: the relation between microfinance and formal markets
The few academic studies focusing on the relation between MFI performance and performance of formal markets show mixed results. All empirical studies other than case studies use data from the web-based dataset mixmarket. This dataset is publicly available and contains annual financial data from around 1700 MFIs worldwide. The dataset is considered to contain the largest and most profitable MFIs for the reason that it requires a certain level of reporting ability and willingness of the MFIs to publicly disclose private information (Gonzalez, 2007).
Using fixed effects regressions, Walter and Krauss (2009) test the correlation between several MFI performance measures -such as return on equity, profit margin, loan portfolio growth and indicators for asset quality- and the state of the domestic economy and several global and emerging market indexes. They use a panel of mixmarket data of 325 MFIs during the period 1998-2006 and find MFIs to be significantly less exposed to global market risk than mainstream banks in emerging markets. Regarding the exposure to domestic risk, hardly any significant differences between MFIs and traditional banks
12
are found. MFIs’ profit margins even tend to be more sensitive to domestic risk than other asset classes. The only performance indicator of MFIs showing significantly less exposure to domestic economic risk compared to emerging commercial banks is portfolio quality, implying a relatively low correlation of repayment behavior of microfinance borrowers with formal domestic markets compared to clients of mainstream banks.
However in absolute terms repayment rates are substantially exposed.
Gonzalez (2007) investigated whether changes in gross national income per capita affect portfolio quality by using both fixed- and random effects panel regressions. He used the mixmarket data for 639 MFIs in 88 countries covering the period 1999-2005. In line with the results of Walter and Krauss (2009), he finds risk of default, measured by the proportion of the loan portfolio with payments behinds schedule for more than 30 days, to decrease when economic growth per capita increases. However he does not find a significant impact on actual default rates and concludes that eventual loan recovery does not depend on economic conditions.
In contrast to Gonzalez (2007), Ahlin (2006) does find a negative correlation between domestic economic growth and lower default rates. He uses data from 112 MFIs from 48 countries during the period 1996-2004. Between and within panel regressions are performed to test the relation between the following MFI performance measures:
operational self-sufficiency, the write off ratio, portfolio at risk, the cost per borrower, growth of active borrowers, and measures for the state of the macro economy (GDP per capita, inflation, foreign direct investment as a percentage of GDP and manufacturing as a percentage of GDP). In addition to the finding of a positive relation between economic growth and default rates, Ahlin (2006) also shows evidence of a positive impact on MFIs’
ability to cover costs and hence on profit margins. He finds a trade-off between indicators for the level of formalization of an economy -such as workforce participation- and borrower growth. This may suggest that in times of economic recession, when unemployment rises, more people are in need for a microfinance loan, creating an opportunity for MFIs to expand their client base. Ahlin (2006) also tests for reverse
13
causality, i.e. whether good performance of MFIs will promote higher economic growth.
Their results do not support such a relationship.
Hermes and Meesters (2009) investigate to which extent macroeconomic growth (and also the financial-, institutional- and political environment) affects cost-efficiency of MFIs. They use data from 435 MFIs in 63 countries over the period 1997-2007. Cost- efficiency is measured by the extent to which the actual cost of lending of the MFI approximates the cost of a best-practice MFI which produces the same output under the same conditions. Interest expenses and labor expenses are used as input costs and gross loan portfolio is the measure for output. They find macroeconomic growth to be significantly positively related to MFIs’ cost efficiency. Based on their results they raise the concern that the current financial crisis may have a serious negative impact on MFIs.
Several case studies have investigated MFI performance during the currency crisis in South East Asia. Patten, Rosengard and Johnston (2001) show the microfinance division of an Indonesian bank to be highly resilient to the economic crisis and loan delinquency rates of microfinance loans even to decline. They attribute the latter to the countercyclical demand for microenterprise products, high incentives to repay among clients and loans tailored to clients’ cash flows. On the other hand, a study on the performance of MFIs during the same financial crisis, but including MFIs from the entire South East Asian region, finds that MFIs do have to deal with borrowers fallen into arrears and with a decline in outstanding loan portfolio as a result of the crisis (McGuire and Conroy, 1998).
In contrast to Patten et al. (2001) this shows a clear adverse impact of the crisis on clients.
In addition, case studies were conducted on MFIs performance during the banking crises in South America. Jansson (2001) found MFIs in Bolivia, Peru and Colombia to be negatively affected during the Latin American banking crisis with respect to loan portfolio growth, delinquency rates and return on assets. However, on average MFIs were significantly less affected than traditional financial institutions. A case study by Marconi and Mosley (2006) also found MFIs in Bolivia to be substantially negatively affected by the regional financial crisis. The authors attribute part of this vulnerability to the state of
14
the microfinance sector in Bolivia which was characterized by a high degree of competition and by over-indebted microfinance borrowers. Especially the for-profit MFIs, which were largely involved in consumer lending were affected whereas MFIs with more integrated business models, i.e. offering additional services besides financial services, and MFIs with a high proportion of female clients performed well during the crisis. Fonseca (2004) shows anecdotal evidence of high resilience of Argentinean MFIs and their clients during the currency crisis facing Argentina in 2002. Clients showed a high ability to adapt to the changed conditions which led to the resilience of repayment rates. Moreover, through effectively renegotiating loan terms corresponding to their changed payment capacity, both MFIs and clients were able to meet debt obligations.
In conclusion, empirical studies testing the relation between MFI performance and formal markets show mixed results. Some studies support the common assumption of the low exposure of MFI performance to international markets (Walter and Krauss, 2009) and to domestic macroeconomic markets (Gonzalez, 2007). However, other studies reject the latter and found a significant relationship between MFI performance and domestic economic growth (Ahlin, 2006; Hermes and Meesters, 2009). Nevertheless, these studies do not test for periods of distress on economic and financial markets. Case studies which did investigate this often show MFIs to be negatively affected. However, the impact differs across types of MFIs and some MFIs do indeed show resilience to adverse macroeconomic shocks (Patten et al., 2001).
Based on the academic studies and theories discussed in section 2.1 and 2.2, it is hard to determine how the current financial crisis affects MFIs. In particular the direction in which clients’ demand for loans and repayment capacity are affected is difficult to predict as in this respect no unambiguous theory exists and evidence from previous crises differs across cases.
15
2.3 Macroeconomic conditions and microfinance in the current context
In addition to previous case studies showing that the commonly assumed low exposure of MFIs to macroeconomic risk is often not valid (Marconi and Mosley, 2006; McGuire and Conroy, 1998), the resilience of microfinance is expected to be even more challenged in the current context.
The current financial crisis is different from any crisis facing MFIs before. Being marked as the most global crisis of the past 60 years it affects financial systems and real economies worldwide at a time that markets in developing countries have been more integrated into global markets than ever before (IMF, 2009b). The crisis on financial markets in developed countries contaminated financial markets in developing countries affecting both MFIs’ domestic and international funding sources. Secondly, as economies all over the world are facing economic downturn, not only the commercial sector but also the so-called highly dedicated social funding sources such as NGOs, governments and other development agencies may have to deal with budget squeezes. Availability of funding from development agencies is expected to further come under pressure as priority may be given to allocations to those who would need their aid in this difficult time more urgently than the microfinance sector, which is assumed to be able to better cope with periods of economic distress (CGAP, 2009a).
The fact that the current crisis affects all actors worldwide seems to make it impossible for microfinance clients to escape the effects of the crisis. Next to the fact that their businesses may be hit by spillovers of the economic recession to the demand for their products (unless businesses from microfinance clients indeed turn out to be detached from formal markets or even to flourish in times of crisis), also development aid and remittances from immigrated family members may decline as a result of the global contraction.
In addition to the unique nature of the current global crisis, also today’s microfinance sector is different than a few years ago. The sector has largely evolved in recent years,
16
which has possibly increased its integration into formal markets. Firstly, the funding landscape of the micro finance industry changed in the past few years. Private and institutional investors have become increasingly interested in investing in microfinance and in lending to MFIs. During the period 2005-2008 the stake of these investors in microfinance increased from 14% to 41%, whereas the proportion from international development institutions decreased from 36% to 19% (USAID, 2009). Commercial funding has made a major contribution to the recent growth and to the professionalization of the microfinance sector, but its involvement may have its downsides as well. Next to the concern that the involvement of commercially oriented investors will lead to mission drift6, the increased integration of MFIs in formal domestic and international financial markets is expected to reduce the advantages of stable funding sources as the cost and availability of commercial funding is subject to business cycles and to investor attitudes towards microfinance (Walter and Krauss, 2009). Moreover, the abundance of available capital and the eagerness of MFIs to grow may have relaxed risk management of liabilities, increasing average leverage ratios and foreign currency funding in recent years. Higher leverage ratios imply higher fixed costs for MFIs and hence reduce MFIs’
advantage of the high capacity to adapt to changed conditions (see section 2.1).
Moreover, foreign-currency funding exposes MFIs to currency risk.
Secondly, typical microfinance practices and characteristics on which many of the explanations behind the low exposure of MFIs to downside market risk are based (see section 2.1) -such as group lending7
6 A drift away from MFIs’ traditional mission of poverty reduction
, the provision of short term loans and serving clients whose business activities are decoupled from the formal economy- may not longer apply or at least not to the entire sector. Many serve higher segments with larger individual
7 The typical microfinance modality is group lending implying that group members are liable for each other: if one member fails to repay the other group members have to recover this loss in order for all group members to maintain access to future funding. This provides incentives for borrowers within one group to screen and monitor each other and to enforce contracts, in which they are expected to be more effective than the MFIs due to social ties between group members (Hermes and Lensink, 2007)
17
loans8. Next to microenterprise loans, MFIs are increasingly lending to small and medium enterprises (SMEs). These SMEs are more likely to be integrated into domestic and international formal markets and may have less ability to adapt to changing economic conditions as opposed to microenterprises.
Microfinance was always known for its excellent portfolio quality. However, concerns are raised about the deterioration of this quality due to MFI policies in favor of fast growth, such as the relaxation of credit standards (USAID, 2009). The microfinance sector has shown impressive growth rates in recent years9 which could be reached by the abundance of available capital. In order to expand their loan portfolio, MFIs tapped into new riskier markets when the demand of the most creditworthy clients had been met. In addition, due to increased competition different MFIs started to target the same clients resulting in clients with multiple loans. Some claimed that MFIs were aggressively selling their loans in order to reach growth targets and to encourage people to borrow beyond their capacity to repay (Microfinance Banana Skins, 2009). For-profit microfinance banks were accused of recklessly lending to poor people (Economist, 2009).
A direct parallel was drawn between microfinance loans and the sub-prime loans in the United States. In the context of the current global recession, many fear that loan portfolios will show less resilience than during past crises as they do not longer exclusively contain the best clients and as portfolios are considered to be ‘polluted’ with over indebted clients (Microfinance Banana Skins, 2009).
In this section we showed that many of the premises on which the assumed resilience of the microfinance sector was based are rejected in the current context. Firstly, the unique nature of the current crisis, affecting all parties worldwide, does not make it very likely that MFIs and its clients can escape impact of the financial crisis. Secondly, the evolvement of the microfinance sector in recent years resulting in (1) higher integration
8 This is often defined as mission drift and as discussed before possibly encouraged by pressure in recent years of commercial funders to scale up operations in order to reach financial sustainability.
9 It is estimated that between 2001 and 2007 the outstanding loan portfolio increased from $4 billion to $36 billion, which implies a growth of 800% over 6 years (USAID, 2009)
18
of MFIs’ liabilities and assets in formal financial and economic markets and (2) the recent deterioration of portfolio quality has presumably exposed the sector to a higher extent to market shocks than in the past.
2.4 Recent studies on the impact of the current financial crisis on MFIs
When the liquidity crisis started to affect financial markets in developed countries in 2007, many expected that MFIs would remain unaffected based on the commonly assumed limited integration of MFIs into formal financial markets. However, as early as 2007, SNS Asset Management, a manager of two institutional microfinance funds, started to express concerns about the potential impact of the liquidity crunch on microfinance institutions in their annual microfinance fund report of 2007 (SNS Asset Management, 2007). As the liquidity crisis became apparent to the public in the second half of 2008 and spilled over into the real economy, there were very few microfinance practitioners, experts and investors who thought that the microfinance sector would not feel the impact of the crisis. Main concerns were a direct impact of the crisis on MFIs’ funding, deterioration of asset quality and higher exchange rate losses. However, experiences and expectations about the timing, extent and consequences of the impact differed.
The first broad inquiry into the impact of the crisis on MFI performance was held by the Consultative Group Assisting the Poor (CGAP), an independent research and policy center. It organized a virtual conference in which opinions and experiences of 600 MFI managers, investors, central bankers and advisers were collected. The most direct impact was perceived on the funding side through higher interest costs and lower availability of funding (CGAP, 2008). Especially for non-deposit taking MFIs the effects were expected to be substantial. Among MFI managers concerns existed about problems meeting refinancing needs and devaluating currencies which would increase cost of funding even more. In some countries pressure on clients was perceived through higher inflation and early signs of deterioration of economic conditions, but especially through lower remittances. Investors managing institutional microfinance funds did not experience early
19
withdrawals of funding but they expected it to be harder to attract new funding. In March 2009, CGAP held an opinion survey among 400 MFI managers (CGAP 2009a). 70% of the respondents reported portfolio quality to be negatively affected by market conditions.
Moreover loan portfolio growth slowed down dramatically for which the credit crunch and economic recession were held accountable.
USAID, an US agency providing development assistance, published a paper in March 2009, presenting a framework along which the future impact on microfinance can be assessed. This study is based on interviews with 19 MFIs, 26 microfinance investors and 2 rating agencies and on desk research to macro economic data. The impact on the liability side is expected to depend on the extent to which the MFI’s country is integrated in the global financial system, as well as MFI specific characteristics such as the reliance on commercial funding and deposits. On the asset side (loans to clients), the key question is how the current crisis is going to affect microfinance clients and hence their repayment capacity. Clients are expected to be affected if their countries feel effects of the crisis in the form of credit tightening, commodity price volatility, a decline in remittances, current account pressures and political risk. The ability of the country to implement countercyclical policies and social safety nets is perceived as an important factor determining the extent to which microfinance clients will be affected in the long term.
The microfinance rating agency Microrate published a study in March 2009 based on interviews with 52 MFIs, 23 microfinance investment vehicles and two microfinance networks in combination with the rating agency’s own observations (Microrate, 2009).
This study exclusively focuses on the microfinance sector in Latin America. Through the second half of 2008 MFI portfolio growth slowed down dramatically. Reasons were that MFIs in this region faced tight domestic credit as the financial crisis was quick to hit Latin America. Moreover, MFIs applied stricter lending criteria and reduced loan sizes due to concerns about a deterioration of portfolio quality as a result of the financial crisis putting downward pressure on growth. Despite these concerns about lower repayment rates, by May 2009 portfolio quality of most MFIs in this region had still not seriously
20
declined (Von Stauffenberg,D., May 200910). Nevertheless, through the impact of the financial crisis (e.g. on the availability and cost of funding) many MFIs in Latin America were perceived to be less creditworthy and received lower rating grades than they had ever obtained in the past (Von Stauffenberg, May 200911).
Loncar, Novak and Ciemil (2009) focus on MFIs in Eastern Europe & Russia and claim that MFIs in this region are most affected. Main effects observed are: flat or even declining growth rates of outstanding loan portfolio in the period 2008 through 2009 and repayment rates falling to 80% in 2009 coming from 98% in 2007. The high exposure of the microfinance sector in this region is attributed to the weak macro economic situation in East-European economies pre-existing the crisis, characterized by high inflation, slow economic growth, an increasing budget deficit and a depreciating currency. This situation worsened even more due to the global credit crunch. Other explanations for the cyclical reaction of MFI performance in this region are a number of distinct features of Eastern European MFIs: a high degree of individual lending as opposed to group- lending, high leverage ratios and the provision of relatively large loans. The latter can be inferred through the fact that 27% of the global loan portfolio concentrates in Eastern Europe whereas only 2% of clients worldwide is being reached. In addition, some MFIs in this region experienced a massive withdrawal of deposits. This effect is attributed to the traumatic experience of the collapse of the public banking system in Eastern Europe a few years ago, to the global sense of instability and to a loss of confidence.
Microfinance Banana Skins (2009), a report published by the Centre for the study of Financial Innovation, discusses the risks facing the microfinance industry. Results are obtained from a survey conducted in April and May 2009 among 430 microfinance investors, practitioners and regulators from 82 countries worldwide. Credit risk, liquidity risk and macroeconomic trends are perceived by respondents as the highest risks of the microfinance sector. The authors conclude that the current financial crisis has completely
10 Email contact (May 2009) with Damian von Stauffenberg, Microrate.
11 Email contact (May 2009) with Damian von Stauffenberg, Microrate.
21
changed the perception of risks as the highest ranking risks were only considered to be minor risks in the same survey conducted last year.
Similar to our study almost all studies published focus on the absolute impact of the crisis on MFIs. They make no comparison with the impact on other institutions operating on emerging markets to investigate the relative impact. CGAP (2009b) makes an attempt by comparing the performance of microfinance funds with the performance of emerging market funds. Returns of microfinance funds significantly outperformed emerging market indexes during 2008. However, it must be noted that the high return of microfinance funds can be largely attributed to higher credit spreads on MFI debt and the fact that most investments are in hard currency debt. Although this is beneficial for microfinance funds it imposes higher costs on MFIs and puts the burden of foreign exchange losses on their shoulders. Therefore this study does not provide much support that MFIs significantly outperforms emerging markets.
This section shows that the impact of the current financial crisis on microfinance has already been intensively analyzed and that the crisis clearly affects or is expected to affect MFIs. Our study contributes to the existing studies by quantitatively measuring the extent and the timing of the impact and by systematically exploring the mechanisms through which the crisis affects MFI performance worldwide. The previously published studies are an important part of our research as it contributed to the development of our theoretical framework.
22
3. Theoretical framework, Data, Methodology and Sample description
3.1 Theoretical framework and hypotheses
Figure 1 presents a simplified version of reality and roughly reflects the channels through which the crisis is expected to have an impact on MFIs. It describes the causes and effects from the origin of the crisis to the potential impact on MFIs. After having presented this framework we will zoom into the different determinants of MFI performance (profitability, growth and portfolio quality) and discuss how we expect these to be affected by the current crisis.
3.1.1 Theoretical Framework
Figure 1 Broad framework impact financial and economic crisis on MFI performance
23
On the basis of the crisis lay the sub-prime loans in the United States, which arguably could exist through flaws in the Anglo-Saxon financial systems (e.g. lack of supervision), abundant liquidity and financial institutions which lost track of integrity (1). Mortgage loans and consumer credit were extended to people with very bad credit histories or with insufficient income to afford future repayment of the loan. Mortgages were extended in the belief that house prices would remain increasing. When American house prices collapsed in 2007 (crisis timeline, 200912
12 For the events of the current financial crisis we relied on a compilation of different sources available at
), mortgages lost significant value and left financial institutions with debt backed by collateral of which value was substantially lower than the loan. Both American banks and banks worldwide, which had invested in these risky loans through the purchase of the mortgage backed securities, the so-called
“toxic assets”, had to make large write offs and had to be saved by national governments.
The financial crisis was a fact (2). It was not transparent which institutions had been investing in these ‘toxic’ credits resulting in a loss of trust among financial institutions.
As a consequence interbank lending dried up affecting financial markets worldwide (1).
In first instance, most developing countries showed to be fairly resilient to the financial crisis, due to limited cross-border linkages of their banking systems. However, financial systems in some developing countries had become increasingly integrated into international capital markets in recent years through a high degree of foreign ownership, exposing them to a direct impact of the credit crunch through tighter financing conditions and increased interest rates (IMF, 2009a) (3). The combination of higher cost of funding, less availability of funding and higher risk aversion of both international and domestic financial institutions led to more prudent lending practices. Also for MFIs there may be less funding available and costs may increase (7). Companies found it harder or even impossible to obtain funding at agreeable conditions, impeding investments. Consumer confidence reached a historical low in October 2008. The financial crisis had turned into an economic crisis (4). In December 2008 the NBER declares that the US had already been in a recession since December 2007. The recession in the United States and in other
http://www.scribd.com/doc/16770503/Crisis-Timeline
24
advanced economies in which financial systems had destabilized, spilled over to the rest of the world, for example through lower external demand. Due to the global economic contraction, aid budgets of government (which often depend on GDP), budgets of NGOs and other social investors are expected to deteriorate, leading to a decrease in social funding (7). Economies in developing countries have become increasingly integrated into the global economy through aid, trade and foreign direct investment. While for many developing countries the effects of the global recession lagged the rest of the world, its eventual impact may be severe for developing countries, especially given their limited scope for countercyclical policies (IMF, 2009a). They will likely feel the impact of slower economic growth in the developed world through lower exports, spillovers to domestic demand, a decline in commodity prices and a reduction in the inflow of capital (IMF, 2009a). Depending on the extent to which the economic activities of microfinance clients are integrated into the formal economy, lower economic growth will have its impact on clients’ income e.g. through lower demand for their products, higher unemployment or falling commodity prices. Fewer business opportunities may arise causing the demand for productive loans to decrease13
13There could also be a reverse impact on microentreprises, increasing growth opportunities and leading to higher demand for productive loans (see section 2.1). In additions, clients’ fewer available sources for funding and fired workers from the formal sector may increase loan demand. Therefore the direction in which loan demand is affected is hard to predict.
(9). Lower demand for productive loans restricts growth of MFIs’ loan portfolios (11). Besides, growth rates come under pressure due to the reduced availability of funding (7) impeding MFIs to finance growth (11). The economic crisis in developed countries also directly affects microfinance clients through a reduction in remittances. Many poor people in developing countries depend on remittances, which are cash flows from family members who immigrated to more developed countries. These immigrated family members were exactly the ones who lost their jobs first when the recession started to take effect (ILO, 2009) and hence lost their ability to send back money to their home countries. The IMF (2009a) estimated that remittances stagnated in the second half of 2008 and shrank in 2009 (9). There will be pressure on local currencies of developing countries due to lower demand for products from these countries and due to capital flight out of ‘risky currencies’ (IMF, 2009d) (10).
25
This will result in an impact on MFIs’ cost of hard currency loans, defined as foreign exchange losses (7). Additionally, depreciating currencies may affect clients receiving loans in hard currency or clients depending on foreign inputs for production (9).
As a consequence of the financial crisis, MFIs take measures in order to limit the impact on their institutions. Due to the application of stricter risk policies, in order to control portfolio quality, there may be a downward pressure on growth rates of the outstanding loan portfolio and of the client base (11). Moreover, in the short term measures may increase operational costs and as costs need to be spread over a smaller loan portfolio (due to the expected negative impact on portfolio growth), operational costs margins will likely be higher. This in its turn puts pressure on profitability (12).
3.1.2 Overview of the expected effects Profitability
From an accounting perspective net interest margins, net operating income, non operating income and the loan loss provisioning expense make up the determinants of the profitability of financial institutions (following the model of Gerlach et al., 200314
14 𝑁𝑁𝑁𝑁𝑁𝑁 𝑇𝑇𝑇𝑇 = 𝑇𝑇𝑇𝑇𝑁𝑁𝑁𝑁+ 𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 −𝑁𝑁𝑂𝑂𝑇𝑇𝑇𝑇−𝑃𝑃𝑃𝑃𝑁𝑁𝑂𝑂𝑇𝑇𝑇𝑇 , where 𝑇𝑇𝑇𝑇 is total assets, 𝑁𝑁𝑁𝑁𝑁𝑁 is net operating income, 𝑁𝑁𝑁𝑁 is interest revenue minus interest expense, 𝑁𝑁𝑁𝑁𝑁𝑁 the non operating income, 𝑁𝑁𝑂𝑂 overhead and 𝑃𝑃𝑃𝑃𝑁𝑁𝑂𝑂 loan provisioning.
). We discuss the effects of the crisis on these components. The most important driver of MFI profitability is the net interest margin, which is the difference between interest revenue and interest expense divided by total assets, and generally assumed to be similar to the gross margin of non-financial companies (Gerlach et al., 2003). Higher interest costs facing MFIs during the current crisis in combination with the inability (due to regulation) or reluctance (in conflict with social mission of MFIs) to pass these higher costs on to clients, makes us expect MFIs’ net interest margins to come under pressure affecting profitability. Secondly, MFIs are expected to increase the loan loss provisioning expense in anticipation to observed or expected higher loan losses in the face of the economic
26 crisis15
Portfolio quality
. Depreciating currencies of developing countries are expected to put pressure on profitability of MFIs borrowing in hard currency in the form of higher foreign exchange losses. The operational cost margin (overhead margin) is expected to be negatively affected due to costs incurred by the measures implemented in response to the crisis aimed at mitigating the impact of the crisis and due to a reduced scale of operations.
We expect the repayment capacity of microfinance clients to be directly negatively affected through a decline in remittances as a result of the contraction of the economies in advanced countries. As the global crisis spilled over to emerging markets and developing countries, we expect a negative impact on microfinance clients’ income from business activities. Although case studies and theories claim ambiguous effects on MFIs’
repayment rates of MFIs to adverse market movements in the past, we expect the overall impact on portfolio quality to be negative due to the increased integration of microfinance assets into formal markets and due to the nature of the current crisis.
Growth
We expect growth to be affected through an impact of the crisis on both the demand for and the supply of loans. The reduced availability of funding is expected to constrain MFIs ability to expand its loan portfolio and client base. It is harder to make predictions on the direction in which demand from clients for loans is affected. Arguments that demand will be lower due to a reduction in clients’ business opportunities are highly plausible. On the other hand, arguments supporting the expectation that demand will increase due to fewer available funding sources for (potential) clients and fired worker from the formal sector for example also sound likely.
15 Appendix table 3.1 shows loan loss provisioning expense to be significantly and positively correlated to indicators of portfolio quality (with PAR30 at a 1% level)
27 Across regions
Depending on how much the region is affected by the crisis (in its turn depending on the extent its financial sector and economy is integrated into the global economy); MFIs in different regions are expected to be affected to different extents by the crisis. We expect Eastern Europe & Russia to be most affected due to the direct impact on Russia and the features of the microfinance sector in this region (Loncar et al., 2009).
Across MFIs
Institution specific characteristics, such as funding structure and lending practices, vary widely among MFIs. Based on these differences we expect the exposure of MFI performance to the crisis to differ. We expect MFIs attracting savings and being not very dependent on commercial funding to be less affected. Savings are considered to be a stable funding source, whereas commercial funding is expected to be most directly affected by the financial crisis. The impact of the crisis on repayment rates is expected to depend on the type of clients the MFI serves, sectors in which clients are employed, lending techniques used and loan products provided. These characteristics are expected to determine the extent to which clients are integrated into formal markets or the level of incentives to repay and hence decide on the robustness of repayment rates during the current crisis.
28 3.2 Samples, Data and Methodology
We investigated the impact of the current financial and economic crisis on MFIs’
performance through two studies: (1) a survey among 82 MFI managers (a perception survey) and (2) a quantitative analysis of financial data of 57 MFIs (a financial performance study). The two studies are complementary: the empirical analysis of the financial data investigates whether performance of MFIs significantly changed during the current financial and economic crisis, while the survey among MFI managers enables us to analyze how the financial crisis is perceived to affect MFI performance. In addition, both studies allow for comparing differences of the impact across regions and MFIs.
3.2.1 General remarks
One of the key issues of research in the field of microfinance is representativeness. No clear-cut statistics about the entire microfinance population are available, making an accurate estimation of the representativeness of a sample of MFIs impossible. Rough estimates about population characteristics exist but vary widely among sources. For example estimates for the total number of borrowers fluctuates from 40 million (Walter and Krauss, 2009) to 155 million (Microcredit Summit Campaign, 2009). Taking into account that the latter estimate actually counted the number of borrowers of 3552 MFIs and that the estimate of the total number of MFIs worldwide is 10.000 (Deutsche Bank Research, 2007), the total number of microfinance borrowers is likely to be even higher than 155 million. This high degree of uncertainty about what the entire microfinance population consists of, makes it impossible to determine the extent to which samples of MFIs reflect the entire population. In addition, it is hard to obtain data from all types of MFIs. Much information, especially of the smaller MFIs, is not publicly disclosed. The available MFI data is often biased as it stems from MFIs which are able and willing to report their private data. This applies for example for data from the often used
29
microfinance database mixmarket 16 but also for data from other widely used sources17. For this reason most datasets are considered to be random samples of the best managed and larger MFIs who have the resources to afford internal information systems and whose performance is good enough to be willing to publicly disclose it (Gonzalez, 2007).
We acknowledge that the MFIs in both our samples, being part of the portfolios of social commercial investors, are subject to the same bias as the mixmarket set. MFIs in our samples meet the reporting standards and are able to generate an acceptable return, both requirements for becoming eligible for investment. In both samples the smallest so-called tier-4 MFIs, which have an asset value of less than 3 million and which are often assumed to be unprofitable startups (Deutsche Bank research, 2007), are absent. We need to accept these limitations and admit that our results should be interpreted as a reflection of the impact on the population of larger and more profitable MFIs. However, these larger MFIs are considered to represent the majority of the microfinance sector in terms of clients served and total assets (Deutsche Bank Research, 2007). This is reflected in the high degree of concentration: worldwide 9% of the MFIs is estimated to reach 75% of all borrowers (Gonzalez and Rosenberg, 2006)18. The importance of large MFIs is also reflected in the fact that the largest MFI in a country has a median market share of one third of the entire domestic microfinance market and the five largest MFIs have a median total market share of more than 80% (Gonzalez and Rosenberg, 2006). Hence, the impact of adverse market shocks on these large MFIs is important for the overall domestic microfinance markets.
In section 3.2.1.1 and 3.2.2.1 the representativeness of the survey sample and the sample investigated in the financial performance study are discussed within the boundaries of a population of larger MFIs reflected by two benchmarks. We assess the
16 Publicly available on the internet and provided by the microfinance information exchange (the MIX).
www.mixmarket.org
17 E.g. the Microcredit Summit Campaign database (http://www.microcreditsummit.org).
18 Another estimate claims the largest 2% of the MFIs to serve almost 90% of the poorest microfinance clients (Microcredit Summit Campaign, 2009)
30
representativeness of our samples based on MFI type, size and region, which are characteristics most often used to categorize MFIs. The benchmarks we use are (1) the Microfinance Banking Bulletin (MBB) Benchmark 2008 provided by the leading microfinance information provider the Microfinance Information Exchange (the Mix) and (2) Consultative Group Assisting the Poor (CGAP), the leading microfinance research center. The MBB Benchmark contains around 1100 MFIs and is considered to be a random sample of the largest MFIs as it excludes the smaller often unprofitable MFIs which are not capable or willing to report to this institution. We rely on CGAP to obtain the distribution of MFIs across institutional sizes as this figure is not included in the MBB benchmark. The CGAP benchmark excludes the smallest MFIs and is therefore also considered as a random sample of the largest MFIs.
3.2.2 Perception survey 3.2.2.1 Sample description
The sample investigated in this part of the study is derived from the databases of SNS Asset Management, Developing World Markets and Triple Jump. We approached 160 MFI managers of which 84 responded. However, 2 MFIs returned the survey twice and hence 2 responses were eliminated leaving 82 MFIs for the analysis. It contains MFIs from all regions where microfinance plays a significant role. However, compared to the benchmarks (appendix table 1.1) MFIs from Asia and from Africa and the Middle East are underrepresented and MFIs from Eastern Europe & Russia and Central Asia & the Caucasus are overrepresented. We acknowledge that our average results may be somewhat exaggerated due to the high exposure of these overrepresented regions to the current financial crisis (Loncar et al., 2009). Regarding institutional type19, credit unions are underrepresented whereas non-bank financial institutions (NBFIs) and Non- Governmental Organizations (NGOs) are overrepresented. It is difficult to predict whether and how this biases our results20
19 In appendix section 1 definitions of the different institutional types are provided.
. Being highly dependent on international
20 In chapter 4 we test whether institutional type matters for the impact of the financial crisis.
31
funding and not allowed to take deposits, NGOs are expected to be the first to feel the credit crunch in advanced countries (CGAP, 2008). However, primarily driven by social motives, operations may be less efficient and internal control systems less professionalized leading to a higher adverse impact on portfolio quality in times when clients are possibly cut in their income. On the other hand, NGOs in the past have shown to create loyalty among their customers and hence repayment rates may be more robust during times of economic downturn. Our survey sample perfectly follows the benchmark distribution across institutional sizes.
Lastly, the survey sample may be biased compared to the total group of MFIs to which the survey was sent. MFIs were promised to receive a report of the results (appendix section 6) on the condition that they completed the questionnaire. MFIs which do not perceive or expect the financial crisis to have an impact on their institution may be less interested in such a report and might not be triggered to fill out the survey. Our survey results may therefore somewhat exaggerate the average impact of the financial crisis.
3.2.2.2 Survey description
Survey set up
The questionnaire (section 7 of the appendix) was designed in collaboration with Developing World Markets21
21 Developing World Markets (DWM) is an investment bank which structures transactions that provide MFIs with access to international capital markets. DWM is the investment manager of the institutional microfinance fund.
(DWM) and SNS Asset Management. It contains 47 questions drawn from effects in previous studies about MFI performance during the current financial crisis and during previous crises (see section 2). For the development of the questions financial experts managing the Institutional Microfinance Fund of SNS Asset Management provided useful insights on the determinants driving MFI performance. Developing World Market, being in direct contact with the MFIs, was in a good position to assess the understandability of the questions and to anticipate the
32
willingness of MFIs to fill out a survey of a particular length. It was decided to make all questions non-mandatory in order to make it more accessible to MFIs resulting in a higher response rate. The questionnaire was designed in both English and Spanish in order to avoid language problems as half of the MFIs we approached are located in Latin America. A pilot was performed and the general director of the MFI Vision fund, Gerlof de Korte tested the survey.
The questionnaire was sent by email from the accounts of DWM and Triple Jump, depending on the portfolio to which the MFI belonged. In the introductory email to the survey MFIs was assured that the data would be treated confidentially (i.e. anonymity was guaranteed and MFI names were only disclosed to the researchers). Responses were automatically sent to an online database allowing us to easily retrieve the results.
Participants were allowed to take as much time as they needed and could interrupt their response to the questionnaire at any time to continue at any point of time in the future.
The questionnaire was sent in April 2009 and the last results were received in May 2009.
The average time taken by MFIs was 40 minutes. Spanish speaking MFIs needed less time to fill it out as they could use their native language. Response rates were not different across MFIs based on language. In June 2009 the MFIs which had responded to the questionnaire received a report of the survey results (appendix section 6).
Survey questions
Most questions were multiple choice questions. There were a few open answers and one open question22. We will briefly describe the contents of the survey23
22 Open answers are reported in the report sent to the survey respondents (appendix section 6)
. The following categories were distinguished: general effects, funding problems as well as effects on growth, operational expense ratio, loan loss ratio and net interest margins. The survey schedule was designed on the basis of these categories in 8 sections. The first section contains general questions about the effects of the crises, including measures taken by the MFIs in response to the crisis, prior conditions from which the MFI suffered before the
23 For full survey see appendix section 7
33
start of the crisis and the impact of government programs. The latter two were included in order to test whether conditions other than the crisis are also affecting MFI performance.
Section 2 contains questions to inquire whether the MFI experienced funding problems and the type of funding problems. It also poses some general questions on the funding structure of MFIs to test issues such as whether MFIs dependent on social funding less often encounter particular effects than MFIs which receive mainly funding from commercial sources. Section 3 treats the impact on MFIs' loan portfolio- and client growth targets and the impact on the underlying forces determining growth, i.e. whether and how the crisis affects supply and demand for loans. Section 4 contains questions about the impact on interest rates, both borrowing-, lending- and savings rates to estimate the impact on net interest margins. Section 5 includes questions about general MFI characteristics in order to test the impact across MFIs. Section 6 treats the impact on operational costs and section 7 the impact of the crisis on clients, their repayment capacity and their incentives to repay. The final question is an open question capturing effects which were not dealt with by the questions in our survey.
3.2.2.3 Methodology for analyzing the survey results
The survey results (appendix section 5) are analyzed by (1) computing respondents positively answering to a particular answer as a percentage of the total number of responses to the question24 or (2) as a percentage of a certain subgroup. Moreover, we performed cross-tab analysis for some variables to analyze whether a significant relationship existed between effects felt by MFIs and particular characteristics of MFIs.
For example, whether MFIs reporting an adverse impact on repayment rates are significantly more often located in particular regions. Significance was tested by performing a chi-square test, which shows whether different answers given by MFIs are significantly correlated.
24 For the reason that none of the survey questions was mandatory, the number of responses differs across questions.
34 3.2.3 Financial performance study
3.2.3.1 Sample description
MFIs investigated in the financial performance study are all part of the SNS Institutional Microfinance Fund. Larger MFIs (Tier-1) are overrepresented in this sample compared to the benchmark and compared to the survey sample (appendix table 1.2). To make the results comparable to the results of the survey and in order to approach the distribution of the benchmark based on institutional size25 we adjusted our financial dataset. We used a method based on post-stratification. The latter implies that after the sample data is selected, sample weights are adjusted to make weighted sample moments equal finite population moments (Carrington et al., 2000). We adjusted all observations to make the distribution of the sample based on institutional size, equal to that of the benchmark (and to that of the sample which follows the same distribution as the benchmark). We obtained this by multiplying all variables in the dataset by 𝜋𝜋𝑖𝑖 /𝑛𝑛𝑖𝑖 , which is the population (benchmark) weight per strata 𝜋𝜋𝑖𝑖 divided by the sample proportion of each strata 𝑛𝑛𝑖𝑖 26. This implies that all observations of Tier-1 MFIs are multiplied by 24/44=0,55 and observations of Tier-2 and Tier-327 are multiplied by 76/56=1,36 (see appendix table 1.2 for sample and population proportions). Results discussed in the main section are based on the weighted sample data (appendix section 3)28
25The sample was not adjusted based on institutional type as this information is in some cases missing.
Hence, weighting our sample based on types would have created many missing values. In addition, it was not adjusted based on regions since the region Africa and the Middle East is missing in the financial data sample and hence the survey sample and benchmark proportions would need to be rescaled (leaving out this region) in order to compute the weights.
.
26 This is an experimental design for correcting for sample selection bias and has been thoroughly discussed with Lammert Jan Dam who has a profound background in econometrics.
27 Tier 2 and tier 3 are defined as one group in the CGAP benchmark.
28 In appendix section 4 the results of the untransformed data are presented.
