• No results found

Master Thesis Microfinance and Portfolio diversification: How does the financial crisis influence the addition of Microfinance Institutions to a benchmark portfolio?

N/A
N/A
Protected

Academic year: 2021

Share "Master Thesis Microfinance and Portfolio diversification: How does the financial crisis influence the addition of Microfinance Institutions to a benchmark portfolio?"

Copied!
25
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Master Thesis

Microfinance and Portfolio diversification: How does the financial crisis influence the

addition of Microfinance Institutions to a benchmark portfolio?

Abstract

Student number: s2056860

Name: B.P. Bosga

Study Program: MSc Finance

Supervisor: Prof. Dr. B.W. Lensink

Word Count: 9385

(2)

1.

Introduction

An explosive growth in Microfinance Institutions (MFI's) took place since the end of the 1990s. Mixmarket.com reports an MFI grand total of 163 in 1999 which increased to over 1300 MFIs present-day. These developments gained a lot of attention to the microfinance business. The popularity among investors severely increased because the accessibility to foreign investments suddenly became much easier. Various studies have been concerned with this phenomenon and mainly with the time period before the financial crisis took place; their findings will be described momentarily. This research will address the differences in investor behaviour towards microfinance between the period before and after the banking crisis. To be precise, this study will focus on the following questions: What are the characteristics of MFIs that makes them profitable for investors to invest in? How did the affection of investors towards MFIs change after 2007, so the period after the financial crisis took place? And last, are there differences between regions or different types of MFIs with respect to investor attractiveness?

This paper distinguishes itself because it actually tries to identify changes in investor behaviour after the financial crisis. Previous papers have described the process of incorporating MFIs to commercial portfolios but looked at other aspects. The upcoming paragraph describes a more detailed background of the research that has already been done.

The structure of this paper is as follows: paragraph 2 will provide a thorough explanation of the characteristics of microfinance and the path to commercialization of microfinance institutions. Paragraph 3 is concerned with the methodology and paragraph 4 displays the data that has been used. In paragraph 5 the results of a mean-variance spanning test and an OLS regression will be presented and paragraph 6 describes the conclusions and suggestions for further research.

2.

Theoretical background

Characteristics of Microfinance

Microfinance has experienced intense changes since the 90s, MFIs no longer offer just credit to starting entrepreneurs (microcredit) but nowadays it contains an entire package of financial services. For example, customers can deposit their savings and receive insurance (Cull, Kunt & Murdoch. 2007). Through these developments microfinance has gained a lot of attention from private and institutional investors. In the next paragraph we will describe behavioural patterns of commercialization of microfinance. Before that, we will turn deeper into the subject of commercialization, an extensive digression on the characteristics of microfinance and the total picture of what microfinance entails these days will be displayed.

(3)

different characteristics among the two. MFI loans are often short term, repeating and unsecured. Basically they are structured as a weekly repayment. This is due to the fact that their clients usually do not have a stable income1.

Moreover, the clients of MFIs mainly are not capable of providing collateral. This means that MFIs are rather unsure whether the clients can repay their loans, especially in a long term perspective. These uncertainties are often captured through high interest rates.

CGAP mentions in a consensus guideline on microfinance regulation (2012) that prudential regulation is not particularly necessary because the global financial system is barely affected by the system of MFIs, so protection against it does not contribute much to the stability of the financial system. However, some form of formalization is required to protect local saver- and borrower deposits. This basically means that there should exist reserve requirements. This is rather important when we look back at the origin of MFIs: MFIs are often dominated by the non profit NGO that created the MFI in the beginning. This traditional type of MFI differs from regular banks in the system by, for example, the fact that they are not contributing to interbanking lending systems. Furthermore, the intention of the NGO MFI is not to profit from providing credit, but was rather build for a more socially responsive purpose. Investors of MFIs are less profit driven which results in less focus on efficiency and the fact that it can be much more difficult to respond to sudden capital needs (Deutsche Bank research, 2007). That is why the reserve requirements can be of great benefit to these institutions. Besides the reserve requirements it is important to focus on non-prudential regulation so that customers are protected against fraud. This is possible by providing them a thorough understanding of contracts so it is rather difficult for malicious lenders to abuse their customers. Cull et al. (2009) find that the regulatory framework for microfinance institutions have objectives that are much alike. By composing regulations on MFIs the standard theory on banking regulatory can be used. The differences are more concerned with the operational design in accordance with the specific markets and client characteristics in the sector.

Altogether, these differences between MFIs and regular financial institutions can be observed by investors in several ways. Investors probably diverge in the sense on how they value the various characteristics on microfinance.

(4)

necessarily has an influence on the size of the losses. Their main conclusion is that microfinance attracts either private and institutional investors because in general, they become more socially responsible in how these organizations drive business. The authors suggest that adding MFIs to portfolios can provide greater diversification and reduce volatility. This paper will pursue whether these findings also hold after the credit crisis. According to Hasan et al. (2009), who conducted a case study in Bangladesh, MFIs need to focus on becoming more financially efficient and should decrease the reliance on subsidies which is in accordance with the findings of Campbell and Rogers (2012). Kraus and Walter (2008) reported that MFI subsidies can be compared to bailouts for commercial banks and from an investor perspective look like a too-big-to-fail situation. However, since the magnitude of these subsidies is rather uncertain, they will be neglected in this research. This is because it seems impossible to clearly define the influence a particular subsidy has on the respective MFI.

Lastly, a Deutsche Bank Research (2007) provides information on the type of investors who tend to engage in MFI activities within their portfolios. International financial institutions (IFIs) do not only invest in the equity side of MFI investments but they also significantly raised their debt investments. When MFIs are categorized within the bond department of a portfolio, it also seems to be a rather profitable investment. For the sake of the magnitude of this entire research, we will focus on composite benchmark portfolios which are well diversified.

Commercialization of Microfinance

After pointing out the essential components of microfinance we will continue our theoretical review by explaining the increasing popularity of private and institutional investing towards the microfinance area. As mentioned before, investors who tended to contribute to MFIs were usually interested in this business for socially responsible reasons. This is still the case and due to the increased attention socially responsible investing has attracted, affection with the microfinance sector rises as well. Moreover, investing in MFIs promises investors high returns and possible diversification benefits (Deutsche Bank research, 2007).

(5)

Brière and Sfafarz (2015) conducted a similar research to ours but focused on solely listed MFIs and neglected the MIVs. They mention that the advantage of their research is that microfinance equity is priced daily on stock exchanges in contrast to MIVs. Even though we agree on this statement, this paper will be based on historical data of MFIs because of the lack of extensive data on listed MFIs.

Returning to incorporating MFIs into private investor portfolios, it seems relevant to determine why microfinance is interesting for international investors. Krauss & Walter (2008) mention several reasons for the risen interest in microfinance: First of all, entrepreneurship with the focus on social responsible areas becomes more and more important. Altruistic motives induce investors to contribute against poverty problems around the globe. Investors find that when MFIs directly target the poor, they can actually help reducing poverty by turning around governmental inefficiencies in emerging countries. Secondly, there seems to be a huge opportunity when we consider the potential size of microfinance. Ghalib and Hailu (2008) say that in especially rural areas of the particular areas, up to four billion people lack access to decent financial services. Moreover, the rise of entrepreneurs worldwide is significant as well. An amount of 500 million micro-entrepreneurs has been estimated. This implies a large section of microfinance is yet uncovered by financial institutions. These gaps can be filled by pooling them together and selling them to international banks who can distribute them among their investors.

Except for the previous mentioned size potential, the returns on lending to micro-entrepreneurs seems rather high as well. Why is lending to micro-micro-entrepreneurs more profitable than lending in the traditional form? The return on assets in developing countries is higher than similar business in the western world. S.K. Mitra (2009) mentions that poor borrowers can afford much higher rates on their loans. In emerging and developing countries it is not uncommon for lenders to ask interest rates close to 20% which is equal to over 5000% a year. Since the interest elasticity of demand is high, this will, however, in many cases lead to a reduced demand. There are several opposite results in recent literature which contradict each other on this subject. A lot of literature has been written on the default rates on microfinance products. Krauss and Walter (2008) say that the default rates on microfinance are low. For instance, because of their group lending mechanisms, MFIs seem to be quite innovative in their lending techniques and the authors show that default rates are mostly below the 5% line. Together with the high interest rates this seems to be a rather profitable investment. On the contrary, specific cases show that there is evidence of troubles with repayment (Pradesh, 2012). Furthermore, low default rates can also be the effect of the still high subsidization of MFIs and yet a low client base; when an increase in client base arise, the quality of the average borrower is likely to decrease.

(6)

macro-economic influences and shocks. This results in low correlations with returns of other assets in the global economy. This is the main reason for investigating this matter because if this is the case, severe diversification benefits can be offered by MFIs to reduce portfolio variance. Galema et al. (2009) find a significant benefit of diversification in the area of Latin America. Brière and Szafarz (2014), however, find that listed MFIs are more closely related to the current international banking system and that diversification benefits have been reduced since the financial crisis. They state that microfinance in general has become less risky and show clear similarities to the global financial sector. Therefore they do not consider MFIs as a self-standing sector. According to the authors this is mainly the case because of the currency exposure which influences both microfinance and the global financial sector in the same proportions. They do argue, however, that a self-standing microfinance sector would ultimately present diversification opportunities.

Nevertheless we expect MFIs to still be profitable for outside investors; considering the potential returns, size and low default rates it seems like a proper investment. Therefore our first hypothesis is: Generally, MFIs are an attractive investment to incorporate in international portfolios. Mainly because of the changes in funding and even because several MFIs have gone public, we suspect several changes in the profitability of diversification when investing in MFIs. MFIs are more easily accessible than before to private investors. Even though we expect MFIs to be more closely related to the global economy we still presume there to be opportunities. Investors might prefer MFI incorporation in their portfolios even more after the financial crisis because they want a safe investment which is not entirely dependent on the world economy as a whole. On top of that, the opportunities in growth and returns are reaching high peaks. Hence our second hypothesis is: MFIs as an investment object has an increased attractiveness and profitability due to the financial crisis.

3.

Methodology

This research tries to identify the differences in the effect of incorporating Microfinance Institutions (MFI's) into a diversified portfolio. Are there differences between regions or types of MFIs, and did the financial crisis have a substantial influence on the attractiveness of investing in MFIs? In order to answer these questions, we need a timeframe of the entire time period in which MFIs are existent.

(7)

et al. (2009) and search for differences in MFI returns which the financial crisis might have caused. We will base our research on the entire time span and also use 1999 as our starting point. However, our research is extended by the fact that we can use a total timeframe from 1999 to 2014, which comes down to 16 annual observations. This means that the participating MFIs will not be examined separately; the sum of MFIs will be taken per subgroup and observations will be based on these amounts.

To account for differences in a portfolio by adding new assets to it, a mean-variance spanning test will be conducted. If the objective is to improve the minimum-variance frontier by adding a new set of assets, a spanning test is mainly used (Kan and Zhou, 2012). By using this test, is it possible to examine whether adding assets to the benchmark portfolio will reach a frontier with a higher mean and lower variance. Or to put it differently, to find the lowest variance for any fixed return.

De Roon and Nijman (2001) describe the intuition of the spanning test as follows: A distinction is made between the benchmark frontier (red line) and the mean-variance frontier of the added portfolio (blue line). The new assets (in this case the MFIs) will be added to the benchmark portfolio. When the mean-variance frontier and the benchmark portfolio intersect at just one point, it means that at least for one function it is not beneficial to add MFIs to the benchmark portfolio. Just one utility function cannot benefit from adding MFIs to his diversified portfolio. Moreover, if these two frontiers totally move along with each other, spanning occurs. This means that investors cannot earn a profit from adding MFIs to their portfolio. In case of spanning, the benchmark portfolio is the optimal portfolio possible regarding risk for any given return.

(8)

Graph 1. Graphical exhibit mean-variance spanning test, the axes represent the mean (µ) and the standard deviation (σ)

The first authors who came up with a mean-variance frontier research were Huberman and Kandel (1987). Subsequent to the previous mentioned authors, Huberman and Kandel described the mean-variance spanning test as a multivariate test where is hypothesized that the minimum-variance line K plus a set of added assets moves in accordance with a set of K benchmark assets. Kan and Zhou (2012) extend this test by using a step-down approach of spanning which determines the economic importance of the added test assets. In general, this is important because in this way it is possible to adjust in sizes between the added assets and the benchmark portfolio. In this research we use yearly portfolios of MFIs which are directly added to our chosen benchmark portfolios, it is in this case unimportant to adjust for weighting the components regarding differences in size. Similar to Galema et al. (2009), Briere and Szafarz (2015), use a mean-variance spanning test in order to find opportunities for investing in MFIs. The authors also use the Huberman and Kandel model to predict whether diversification benefits exist by adding MFIs to benchmark portfolios. Since this research uses a portfolios of assets for each year between 1999 and 2014, the spanning test seems highly appropriate.

The first test in this research will be an OLS regression where the entire sample of MFIs will be added to a benchmark portfolio in order to check for overall spanning. We simply want to investigate whether there is spanning for the entire period (1999 to 2014).

R1 = α + β*R2 + ε (1)

(9)

to test our null-hypothesis. When we follow the null-hypothesis of a regular spanning test, the value of Alpha should be equal to zero and the value of Beta is equal to one in order to conclude there to be spanning. In this case the two lines from Figure 1 intertwine completely which ultimately means that there are no profitable investments opportunities by adding the MFI portfolio.

Our second test is similar to the first one and has the same equation only here we will distinguish whether different regions are spanned to our benchmark portfolio. Furthermore, different types of MFIs are included in the sample and are spanned in this model to seek for variations. If spanning does not occur, the two lines will significantly differ from each other. Thereby said, the values of Alpha and Beta will significantly differ from respectively zero and one which ultimately drives the blue line (added MFI portfolio) leftwards from the red line. This represents a lower average standard deviation (σ) with respect to the mean returns (µ).

To account for these values of Alpha and Beta, we will conduct a Wald test.Brooks and Kurz (2012) mention that the Wald-test is perfect to test whether a Beta is equal to one. Moreover, Eviews allows us to easily add another parameter (in this case Alpha is equal to zero) so we are able to test the two parameters at once. An F-test will decide whether the null hypothesis of a=0 and B=1 will be rejected in order to conclude for no spanning and hence, diversification benefits. After we have distinguished whether spanning exists for the entire period, we want to address possible differences in diversification benefits before and after the financial crisis. In order to do so, we need to find out if a break point exists around the time where the financial crisis occurred (2007/2008). To account for this presumed split we are using a Chow-test in order to determine whether there exists an actual breaking-point in the data. According to Dufour (1982), a Chow-test assesses the reliability of the model by making sure the Beta is stable over time. Since we would like to prove a clear change in the Beta over time, we expect a rejection of the Chow-test null-hypothesis that the Beta is constant over time.

Secondly, after establishing whether a breaking point exists around the financial crisis, our second test will be concerned with the MFI subgroups (Regions and MFI types) which clearly showed a breaking point around the financial crisis according to our Chow-test. A breaking point is indicated when the according p-value of the Chow-test is lower than 0.10. The results section will elaborate more on this matter.

(10)

region Africa should be tested for spanning between the time period of 2007-2014 and be compared with the Alpha and Beta for the time period of 1999 until 2007. Again the null-hypothesis here is that spanning exists and diversification benefits are not present. All in all, we will start with a global spanning test for the entire period. Secondly a Chow-test accounts for possible breaking points and if that is the case, these subgroups are immediately subjected to another spanning test for two separate time periods.

4. Data Requirements

As previously mentioned, our dataset is covering the time series between 1999 and 2014. This data will be extracted from Mixmarket. Mixmarket is an organization founded by the CGAP and sponsored by, for instance, the Bill and Melinda Gates Foundation2. It provides an exchange where MFIs can interchange their data for professional usage. The data will be verified by Mixmarket itself through an auditing process where members have to provide documentation to the exchange.

Our data will be focused on continents with emerging and/or developing countries. This ultimately excludes samples from Western Europe and the Northern Americas. We will extract a sample from Mixmarket which includes the Return on assets and Return on equity of all the included MFIs. In addition, the dataset includes the Debt-to-equity ratio and the amount of MFIs per region. The dataset will be divided into subsets. Our first subset contains the returns, MFI count and D/E-ratio from the different regions. We will try to find out if investing in different regions may have significant effects on the results. Our chosen regions for this research are: Africa, Eastern Asia and the Pacific, Eastern Europe and Central Asia, Latina America and the Caribbean, the Middle East and North Africa and last, South Asia. Also here the differences between pre- and post-crisis data will be assessed. Additionally, we will allocate the participating MFIs to their type of origin. Thence we can observe the developments of different types of MFIs, for example, the difference in returns between MFIs in the form of Banks, Cooperatives, NGOs, Rural banks and financial institutions which were difficult to categorize among these first four categories which will be defined as 'Others'.

Attracting investors to invest in MFIs will require a stable return on equity (ROE). In order to sustain a decent ROE, an MFI must provide decent loans just like commercial banks. The loans of an MFI should be to credit-worthy borrowers so eventually they can become self-sufficient (Campbell and Rogers, 2012). This research will consider the ROE and ROA of many MFIs all around the globe to find out whether these institutions would be beneficial when incorporated in a portfolio. In present-day markets, we observe a tendency to more actively traded MFI's. For example,

(11)

the Dutch Triodos bank offers funds which are connected to Microfinance in emerging countries3. Research conducted by Briere and Szafarz (2015), make use of exclusively listed MFIs. Obviously, this will result in more accurate findings because the returns are based on contemporary data instead of historical data such as ROE and ROA. However, the amount of data available on listed MFIs is rather poor. Therefore, for the purpose of this research we will use the ROE and ROA of the data sample. Regarding the benchmark portfolios we use the S&P500 and the Morgan Stanley Capital International (MSCI) World since they are found to be the most solid portfolios for benchmark purposes and most accurately reflect market volatility.

5.

Results

Table 1 provides a brief overview of the descriptive statistics of the sample. A comparison has been made regarding the Return on Equity and Return on Assets per region and MFI type. Generally, we observe severe differences between the returns in the different sections. Return on Assets seem to be lower for every single region and MFI type. This is due to the fact that the assets of a company are always higher than the equity of the company unless the company does not have any liabilities at all. In that case the value of the assets and the value of equity should be equal and similar returns should be found. The returns of the benchmark portfolios S&P500 and the MSCI World are on average more similar to the Return on equity and thus, are categorized accordingly. Regarding the regions, the returns (both for equity and assets) particularly stand out for Africa, North Africa & the Middle East. These are compared to the other regions the most underdeveloped continents and respectively have lower returns. Concerning the various MFI types, rural banks stand out to have significantly higher returns (both for equity and assets) than the other types. This might have to do with the fact that, as mentioned previously, with the extraordinary high interest rates asked in rural areas. The MFI count is highest ranked for the regions Latin America and Africa and an average MFI seems to be an NBFI or an NGO. With respect to the Debt to Equity ratios; the region South Asia stands out with a D/E-ratio of 4.36. Apparently the MFIs use a lot of debt in comparison to their equity. This could partly explain the high return on equity and low return on assets (since they probably have a lot of value at the equity side as well). Similar results are found for (Rural) Banks which also have D/E ratios above 4. These MFI types also have a high return on equity and a low return on assets.

(12)

Table 1. Descriptive statistics per Region and MFI type

Return on Equity Return on Assets MFI Count D/E-ratio Benchmark portfolios:

S&P 500 0,0693

MSCI World 0,0705

Regions

Africa 0,0705 0,0049 3096 2.00

Eastern Europe & Central Asia 0,0877 0,0279 2589 1.93 Eastern Asia & the Pacific 0,1145 0,0261 1818 2.62 Latin America & the Caribbean 0,0962 0,0257 4033 2.43 Middle East & North Africa 0,0505 0,0188 582 0.62

South Asia 0,0924 0,0114 2486 4.36

MFI Type:

Bank 0,1043 0,0169 1418 4.28

Cooperatives & Credit Unions 0,0692 0,0153 2429 3.45

NBFI 0,0737 0,0194 4633 1.93

NGO 0,0692 0,0198 4985 1.59

Rural Bank 0,1492 0,0246 731 5.70

Other 0,0158 0,0090 192 1.40

(13)

Table 2. Global spanning test

Totals Alpha (α) Beta (β) F-test Probability

S&P500 Return on Equity 0.033808 0.452397 0.055946 0.9458 Return on Assets 0.051138 0.949976 0.534366 0.5975 MSCI World Return on Equity 0.056303 0.180933 0.080551 0.9230 Return on Assets 0.056818 0.715392 0.458882 0.6412 Regions:

Our next test will split up the total sample of MFIs into regional subsamples. An attempt will be made to make a distinction between the several Regions in order to conclude for spanning per single Region. Table 3 displays the merged outcomes of the test, focussing on Alpha’s, Beta’s and the p-values of the F-test. If the null hypothesis can be rejected, then one of the Regions (Africa, Eastern Europe & Central Asia, Latina America & The Caribbean, Middle East & North Africa and South Asia) significantly contributes to one of the two benchmark portfolios. Secondly, in the last column the results for the Chow-breaking point test are stated. A breaking point test has been conducted on the years 2007, 2008 and 2009. The reason for this is that the financial crisis started at the end of 2007 (which makes it already possible for changes to happen). However, the influences of the financial crisis might have some delay before they reflect on the returns. We decided to just include the results of the subgroups who had significant results from the Chow-breaking point test in order to account for more clarity.

(14)

insignificant at the 1%, 5% and also 10% level. This means that we cannot reject the null-hypothesis for spanning which in turn means that spanning exists for the time period of 1999 to 2014. However, we do find some significant results from the Chow Breaking Point test in the last column. Starting with the S&P500 we notice the first breaking point in 2009 for the Region Africa. A p-value of the F-test in the category Return on Assets has an outcome of 0.1057 which cannot be safely rejected to account for a breaking point. However, when we consider the year 2009 in the Return on Equity category we notice a p-value of the F-test of 0.0717 which can safely be rejected at the 10% level which in turn, implies that the chance of a breaking point in 2009 is present. Naturally, it is necessary to confirm these findings by checking if there are also breaking points existent when we would add the MFIs from Africa to the benchmark portfolio MSCI World. This does not seem to be the case and also for the years 2007 and 2008 no breaking points are found. Nevertheless we will take the region Africa into account when we conduct the last spanning test.

(15)

Table 3. Spanning test per Region, including the significant results from the Chow Breaking point test.

Regions α β F (prob.) Chow (year)

Africa S&P500 Return on Equity 0.099513 -1.023090 0.659887 (0.5323) 0.0717 (2009) Return on Assets 0.073605 -0.875512 1.120220 (0.3538) 0.1057 (2009) MSCI World Return on Equity 0.123157 -1.783483 0.935559 (0.4156) Return on Assets 0.097688 5.537835 1.294549 (0.3049)

Eastern Europe and Central Asia

S&P500 Return on Equity 0.091613 -0.254374 0.538692 (0.5951) Return on Assets 0.175919 -3.824678 1.015562 (0.3874) MSCI World Return on Equity 0.091547 0.240007 0.435806 (0.6552) Return on Assets 0.161133 -3.251423 0.720978 (0.5035)

Eastern Asia and the Pacific

S&P500 Return on Equity 0.101800 -0.283816 1.101094 (0.3597) Return on Assets 0.134342 -2.490584 0.715331 (0.5061) MSCI World Return on Equity 0.127799 -0.500485 1.095020 (0.3615) Return on Assets 0.151606 -3.106021 0.686012 (0.5197)

Latin America & the Caribbean

S&P500 Return on Equity 0.102136 -0.341291 0.344045 (0.7147) 0.0013 (2009) Return on Assets 0.159596 -3.513209 0.801039 (0.4684) 0.0371 (2008) MSCI World Return on Equity -0.013051 0.868565 0.117574 (0.8899) 0.0051 (2009) Return on Assets 0.094038 -0.915874 0.411326 (0.6705) 0.0542 (2008)

Middle East and North Africa

(16)

MFI Types:

Similar to the previous test, Table 4 elaborates the results of the spanning test where we subdivide the entire sample into subsamples based on the Type of MFI. We will pursue whether the MFI types (Bank, Cooperatives, NBFI's, NGO's, Rural banks or Others) significantly contribute to one of the two benchmark portfolios. Most of the results are very far off except the p-value of 'Rural Banks' in the category Return on Equity added to the MSCI World. Although even with this p-value we cannot reject our null hypothesis, this one seems to come most close-by.

The only remarkable outcome at this first test is the probability of the F-test in the category 'Others' at the 1% significance level. This counts for either the MSCI World portfolio as the S&P500 portfolio but solely when we based our predictions on the Return on Equity. Even though these results seem quite clear, the significance level can be due to the low MFI count in the category Others. Gathering from Table 1, we observe that the MFI count for ‘Other’ is explicitly lower than for the other categories, we cannot base our entire research on this outcome without knowing the exact characteristics of this small group. Because our research focuses on the financial crisis and its influences on MFI incorporation we will disregard this result and continue with the second part of Table 4 which is the Chow breaking point test. The Chow breaking point test tries to determine whether there are significant results existent for different types of MFIs. When the MFI type 'Bank' is added to the S&P500 benchmark portfolio there seems to be a breaking point in 2009 with an associated p-value of 0.0426 which can be rejected at the 5% level (when we observe the Return on Assets section). Accordingly, the Return on Equity category provides us with results that are alike. A breaking point is found in 2009 and is significant at the 1% level (p-value of the F-test of 0.0089). To confirm these results a comparison will be made with the MSCI World benchmark portfolio. Significant results are found for this benchmark portfolio as well. For the Return on Assets a p-value of the F-test is found to be 0.0598 and is rejected at the 10% level. In the Return on Equity category we observe a breaking point in 2009 as well. The p-value of the F-test can be rejected at the 2% level (p-value of 0.0183). The results for the MFI type 'Bank' seem rather strong and will be included into our second spanning test.

(17)

0.0251 which can be rejected at the 5% level. Regarding the Return on Equity, we observe similar results with p-values for 2007 and 2008 of respectively 0.0294 and 0.0276 which are significant at the 5% level. Cooperatives & Credit Unions seem to have strong results as well and will also be included into our next spanning test. Unfortunately, no significant results are found for other MFI types. Concluding, there are four groups which will be included in the next test. These are: Africa, Latina America & the Caribbean, Banks and Cooperatives & Credit Unions.

Table 4. Spanning test per MFI type, including Chow Breaking point test.

MFI types α β F (prob.) Chow (year)

Bank S&P500 Return on Equity 0.076732 -0.071177 0.442306 (0.6512) 0.0089 (2009) Return on Assets 0.122757 -3.165089 0.573406 (0.5763) 0.0426 (2009) MSCI World Return on Equity 0.085353 -0.142358 0.429433 (0.6592) 0.0183 (2009) Return on Assets 0.101728 -1.849192 0.554082 (0.5867) 0.0598 (2009)

Cooperatives & Credit Unions

(18)

MSCI World

Return on Equity 0.056670 0.103592 8.079284 (0.0052)

Return on Assets 0.055651 0.294891 1.109852 (0.3589)

Financial crisis:

Our last test is another spanning test and elaborates on the findings of the Chow breaking point test. We test, for the groups who indicated to have a clear breaking point, if spanning occurs for the period before and after the financial crisis. If the p-values are significant, spanning does not occur and the addition of MFIs to the benchmark portfolio should be beneficial before or after the financial crisis. We divided the sample for each group in samples from 1999 to 2007 and from between 2007 and 2014. Significant and remarkable results are displayed in Table 5. Starting with the Regions; we immediately find significant results for Africa. For the S&P500 benchmark portfolio we find, under the Return on Assets category, a p-value of the F-test of 0.0429 and under the Return on Equity category a p-value of 0.0066. These values are respectively significant at the 5% and 1% level. Moreover, the MSCI World benchmark shows to have benefits from the addition of MFIs. Also in this case, the period between 2009-2014 is rejected for spanning with a p-value of 0.0407 at the Return on Equity section and a value of 0.0867 on the Return on Assets section. Respectively these p-values are significant at the 5% and the 10% level. Region Africa seems to be significant at four different tests and is rather profitable to include in a portfolio after the financial crisis. The time periods until 2007 do not seem to find significant results.

The second region is Latin America & the Caribbean. Table 5 shows a significant result for the Return on Assets when they are added to the S&P500 portfolio. The associated p-value is 0.0335 which rejects the null-hypothesis of spanning at the 5% level. This value represents the time period 2008 to 2014. Interestingly, significant results are also found when the assets are added to the MSCI World portfolio when looking at the Return on Assets. In this case the p-value is 0.0488 which is significant at the 5% level. However, with regards to the Return on Equity for either the S&P500 as the MSCI World portfolio, the outcomes seem to be contradicting. In this case the period before the financial crisis (1999 – 2008) shows significant results. A p-value of the F-test of 0.0068 is found when added to the S&P500 and a p-value of 0.0162 is found when the MFIs are added to the MSCI World. This indicates that investing in Latin America & the Caribbean was profitable before the financial crisis and has become less profitable since the financial crisis occurred.

(19)

MSCI World they do not show significant results and hence spanning seems to exist. This also counts for the S&P500 when we observe the Return on Equity.

(20)

Table 5. Spanning test before and after the financial crisis

MFI types α β F Probability

Africa S&P500 (1999-2008) Return on Equity 0.058065 -2.540901 1.339920 0.3148 Return on Assets 0.036216 -9.566578 0.376121 0.6980 S&P500 (2009-2014) Return on Equity 0.504723 -7.363777 30.94976 0.0066 Return on Assets 0.284587 -13.97309 7.644335 0.0429 MSCI World (1999-2008) Return on Equity 0.085649 -2.975308 1.223650 0.3438 Return on Assets 0.061411 -11.63557 0.439257 0.6592 MSCI World (2009-2014) Return on Equity 0.516188 -8.244630 7.709259 0.0407 Return on Assets 0.346168 -25.30122 4.786525 0.0867

Latin American & the Caribbean

S&P500 (1999-2008) Return on Equity -1.293964 11.94520 9.924376 0.0068 Return on Assets -0.173508 6.834209 0.177182 0.8413 S&P500 (2009-2014) Return on Equity 0.406943 -3.103597 2.626369 0.1869 Return on Assets 1.132837 -60.58110 4.513284 0.0335 MSCI World (1999-2008) Return on Equity -1.448649 13.53843 7.203564 0.0162 Return on Assets -0.161460 7.235825 0.296036 0.7526 MSCI World (2009-2014) Return on Equity 0.521587 -5.005153 1.178934 0.3958 Return on Assets 1.187902 -65.52559 4.128109 0.0488 Bank S&P500 (1999-2008) Return on Equity -0.019695 0.243626 1.215317 0.3460 Return on Assets -0.033158 2.105522 0.024680 0.9757 S&P500 (2009-2014) Return on Equity 0.179594 -0.056873 1.419435 0.3421 Return on Assets 0.282404 -7.953804 6.121908 0.0600 MSCI World (1999-2008) Return on Equity -0.016663 0.389938 0.592998 0.5752 Return on Assets -0.037812 3.348814 0.030328 0.9702 MSCI World (2009-2014) Return on Equity 0.269722 -1.251700 0.885694 0.4804 Return on Assets 0.436272 -21.30110 3.857602 0.1166

Cooperatives & Credit Unions

S&P500 (1999-2007)

Return on Equity -0.067101 1.632864 0.083302 0.9211

Return on Assets -0.084808 7.780530 0.315029 0.7411

(21)

Return on Equity 0.845179 -10.99459 10.43081 0.0097 Return on Assets 0.788066 -51.11404 34.12321 0.0187 MSCI World (1999-2007) Return on Equity -0.080557 2.155766 0.053841 0.9480 Return on Assets -0.071111 8.334214 0.416079 0.6773 MSCI World (2008-2014) Return on Equity 0.837642 -11.18646 7.312132 0.0246 Return on Assets 0.762135 -50.72919 12.02458 0.0080

6. Conclusion

This research made an attempt to overtake the attractiveness of incorporating MFIs into private investor portfolios. Spanning tests as well as Chow Breaking point tests are used to conclude for the profitability of MFIs into said portfolios. This paper adds to existing literature since it distinguishes between periods of time and in particular focuses on the influence of the financial crisis. Furthermore, it uses a larger sample which generally excludes irregularities.

The first spanning test resulted in insignificant outcomes and were concerned with the overall attractiveness of MFIs over the time period from 1999 to 2014. Since the results were insignificant, spanning seemed to exist. This means that when we are adding a composite portfolio of MFIs of various types and regions to a benchmark portfolio such as S&P500 or MSCI World, this would not be beneficial to the benchmark portfolio in terms of risk and return. An explanation for this finding could be that the MFIs tend to be a more familiar investing object these days. This would imply that the MFI system might be converging to the regular financial system. This would also support the findings of Briere and Szafarz (2015) who find similar results. They extend this by exclusively using listed MFIs for their research to demonstrate the similarities between the MFI system and the global financial system.

Our research continues by subdividing the entire sample into Regions and secondly by splitting the sample based on MFI types. Joining the findings of the first spanning test, this second spanning test does not find significant findings for the entire time period of 1999 to 2014 for either a single region or MFI type. Dividing into specific subgroups does not change the attractiveness of MFIs. Our first hypothesis was: Generally, MFIs are an attractive investment to incorporate in international portfolios. After evaluating the results of the first tests we need to reject this hypothesis.

(22)

financial crisis for the subgroups Africa, Latina America & the Caribbean, Banks and Cooperatives and Credit Unions and continued with another spanning test to see if there actually are differences observable between the period before and after the financial crisis.

Starting with the utmost significant results, Africa and Credit Unions & Cooperatives are found to be significant when we consider the Return on Equity as well as the Return on assets. Moreover, the Robustness test of testing with two benchmark portfolios solely resulted in significant outcomes. When we look at the distribution among the MFIs in Africa, we notice that the majority of MFIs are Credit Unions & Cooperatives, especially after the financial crisis. This might imply that the most profitable MFIs are the Credit Unions & Cooperatives in the region Africa. But why does these MFI subgroups seem profitable, and why might the combination be suitable to invest in? A possible explanation to this matter is the distance between the global financial system and the financial system in Africa. Africa seems to be the continent in which, particularly in the rural areas, financial infrastructure is the lowest and therefore the interbanking system is less correlated to the global financial system. This indicates that shocks from the outside world have less influence on the African financial system. Moreover, when we observe the Credit Unions & Cooperatives, comparable reasons occur. Since Credit Unions are cooperatives which are democratically owned by their members, they basically constitute their own small financial system4. If combined, Credit Unions & Cooperatives in Africa might have severe diversification benefits due to their freestanding character compared to the global financial system.

Other remarkable results are found in the region Latina America & the Caribbean. Differences are observed between the Return on Assets and the Return on Equity categories. When we assume the Return on Equity is the right measure, it was profitable to invest in Latin America & the Caribbean MFIs before the financial crisis. Totally contradicting results are found when we assume the Return on Assets is the right measure. In that case, it is beneficial to invest in this region after the financial crisis. Relating to the paper of Galema et al. (2009) it is more likely that the Region of Latin America is beneficial before the financial crisis since that significantly matches their results. Nevertheless, these findings are so severely contradicting each other that we cannot conclude for investing benefits for Latin America & the Caribbean.

Last, the MFI type Bank does only partly find significant results and therefore we cannot totally predict a workable outcome on this subgroup.

(23)

hypothesis which implies a rather profitable investment opportunity in these subgroups.

These conclusions hold when investors purely base their investing on risk and return. However, in essence, the start-up of MFIs in emerging countries was meant to reduce poverty and increase the financial opportunities for the outreach of the local communities. When addressing the effect on outreach and thus local financial stability, Hermes et. al (2009) find conflicting effects. Starting with the positive effects; commercialization can improve funding availability which is definitely positive for local citizens since they have a more wide range of choice where they can attract funding from. This also leads to a broader assortment of funds which in turn results in more diversified sources of funding. Regarding the negative effects of commercialization on outreach are basically that the transaction costs of a small loan are relatively high. So in that case, striving for financial sustainability is not beneficial for the outreach. Finally, commercialization of microfinance can lead to a focus on wealthier clients in the developing countries. In essence this seems logical but the aim to reduce poverty is then no longer the main priority of MFIs anymore. We need to take into account the reasons for microfinance next to the benefits it can entail for private investors.

Even though we found significant results which could certainly add to the existing literature, a few limitations to this paper rise to the surface. For the sake of the magnitude of this research we did not use short sale constraints. Short sale constraints might make this research more accurate because it cancels out the usual investors going short which is not possible for non-listed MFIs. That directly brings us to the second limitation which is that we did not use listed MFIs. The reason for this is that the sample of listed MFIs is rather limited and because the sample is so low we cannot base conclusions on it. Furthermore, the data is not adjusted for subsidies and it is possible that the database of Mixmarket only include the better performing MFIs because the MFIs who are less performing probably will not release their data. Regarding future research on this subject, we also think there should be more focus on the composition of the MFI type Banks and the region Latin America & the Caribbean since the results of these subgroups were contradicting or simply not resolute. Moreover, when in the future there should be more available data on listed MFIs, it is advisable to focus more on listed MFIs rather than on historical data. Even though these limitations could affect the results, we still think this paper has found significant results which can be beneficial for further research and private investing.

7. References

(24)

Brooks, S.M. and Kurtz, M.J., Paths to Financial Policy Diffusion: Statist Legacies in Latin America’s Globalization. International Organization 66 (2012), 95-128

Campbell, N.D. and Rogers, T.M., Microfinance institutions: a profitable investment alternative? Journal of Developmental Entrepreneurships 17 (2012)

CGAP, A Guide to Regulation and Supervision of Microfinance: Consensus Guidelines, October (2012) Cull, R., Kunt, A. and Morduch, J., Microfinance and the Market, Journal of Economic Perspectives, October 18, (2007)

Cull, R., Kunt, A. and Morduch, J., Microfinance Tradeoffs: Regulation, Competition and Financing, October (2009)

Deutsche Bank Research, Microfinance: An emerging investment opportunity, Frankfurt am Main: Deutsche Bank, December 19, (2007), 1250024 1-10

Dufour, J.M. Generalized Chow Tests for Structural Change: A Coordinate-Free Approach, International Economic Review 23-3 (1982), 565-575

Galema, R., Lensink, B.W. and Spierdijk, L., International diversification and Microfinance, Journal of International Money and Finance 30 (2009), 507-515

Ghalib, A.K. and Hailu, D., Banking the Un-Banked: Improving Access to Financial Services, Policy research brief 9: (2008)

Hasan, M.M., Hassan, K.M. and Uddin, M.R., Local government investment outreach and

sustainability of microfinance institutions: A case study of BURO, Bangladesh. The Journal of Social, Political and Economic Studies, 34 (2009), 318–346.

Hermes, N. and Lensink, B.W., Microfinance: Its impact, outreach and sustainability, Worldschool, 2009

(25)

Krauss, N. and Walter, I., Can Microfinance Reduce Portfolio Volatility? SSRN working paper no. 943786 (2008).

Referenties

GERELATEERDE DOCUMENTEN

Zowel de inhoud als de vorm van de kuil bevatten geen specifieke elementen die ons zouden kunnen inlichten over de functie van het spoor?. Wel is duidelijk dat het

Oat onderzoek began in de industrie toen Pennings in speelse proeven erin slaagde uit oplossingen van polyetheen (PE) draadjes te trekken, die bijzonder sterk

The theory of strong and weak ties could both be used to explain the trust issues my respondents felt in using the Dutch health system and in explaining their positive feelings

However, with individual steering signals a more detailed pattern can be reached since multiple houses can react on one shared signal, as can be seen on the imbalance in the car

De bovengronden in het zuidwestelijk perceel (1001 en 1002) hebben een te hoge fosfaattoestand voor schrale vegetaties en hier zijn de perspectieven om deze

In the results section it is analyzed if the inclusion of Bitcoin to the global market portfolio lead to better risk return metrics in terms of the Kappa ratio compared to

Furthermore, all additional controls are negatively related to the return loss (and marginally significant): people with higher cognitive abilities, that are more risk

The first results are presented based on changes made in the product variety for the PVC-KG portfolio when one of the primary product configurations (in this case indicated