Corporate Governance in Microfinance Institutions

Размещено на http://allbest.ru

Corporate Governance in Microfinance Institutions

Introduction

corporate governance microfinance

The first microfinance institutions (MFIs) were established in 1980s in third world countries. These organizations provided the microloans to low-income population. This service became a phenomenon, because the ordinary banks did not lend money the poors. Soon MFIs became popular and started to extend over the developing countries all over the world. The microfinance institutions were developing for a long time and they begin to provide different kinds of financial services to the representatives of low-income families and micro-entrepreneurs. Financial services generally consist of credits and savings. However, some MFIs supply insurance and payment services. Moreover, a lot of MFIs provide social intermediation services, such as training in financial literacy, development of self-confidence and so on. That is why the MFI is not only the banking, it is also the development tool [J.Ledgerwood, 2000].

Providing the financial services indigent people the MFIs not only improve their financial position but also have a beneficial effect on economy, because giving money for development the micro-business or for consumer consumption can little by little improve the situation in economy. The social orientation of MFIs leads to the reduction of the level of poverty and unemployment in the country.

The social activity is aimed to give the opportunity to the vulnerable population to live a full life. The introduction people to the market of financial services increases the activity of citizens and leads to a certain social recovery, which allows removing some tensions in society.

But the microfinance institutions should take into account the economic aspect of the activity. They should maintain the financial sustainability that allows the MFI to continue fulfilling its social mission. Of course, some organizations are funds that exist due to charity and donations, government support and so on. But this type of organization also needs to allocate their resources properly, otherwise it cannot operate for a long time. That is why, the MFIs should find optimal strategy in financial and social performance. The competent policy in corporate governance helps to achieve this goal.

It should be noted that microfinance activity is to a large extent a business of female [Strшm et al., 2014]. The women are the largest market for MFIs. It is considered that they are good fighters with the poverty because they are more probable to reinvest their earnings in their families or business. Likewise, the microfinance is a business by women because the proportion of female directors or managers is higher than the proportion in other industry. Also it should be mentioned that the women's ability of risk-taking in decisions differs from the men's one. On account of specifics of the microfinance activity, the risks are quite different with the risks in another sector. That is why, we are interested in taking into account and analyzing this fact in our paper.

Summarizing all the points above, we set an aim of our thesis, which is to investigate how the corporate governance determined by women influences the MFIs' financial and social performance. The object of the research is the microfinance institutions. And the subject is the impact of the determinants of corporate governance on social and financial efficiency of MFIs.

In order to achieve the goal of the work, we assign the following missions:

· We will review the studies, which are devoted to the corporate governance in financial sector.

· We will define the hypothesis on the basis of our analysis of the previous research papers.

· We will collect specific data about the chosen MFIs.

· We will carry out regression analysis and obtain the nature of the relationship between the corporate governance and the performance.

· We will try to explain the relationships between the dependent and independent variables.

In our thesis we will build the model, which consists of 3 regressions based on the panel data, which involves 200 microfinance institutions from Eastern Europe and Central Asia, presented during 5 years (2010-2014). The data is provided by the non-profit organization the Microfinance Information eXchange (MIX). The first regression will analyze the relationships between the independent variables, which include proxies for women leadership and risks of MFIs. The second and the third regressions will contain, on the one hand, financial and social performance variables and, on the other hand, the proxies for the corporate governance, risks and female leadership.

The professional significance of the work is that the results received by the exploration can be used by current MFIs in considered countries in order to improve their effectiveness in the financial sustainability and social outreach taking into account the women's leadership and incurred risks by the organization.

This work have the following structure. In the first section we review existing literature. After that we develop the hypotheses, which we test in the empirical model from the second section. Also, we give there the description of the data. There are some explanations of obtained results in the third section of the work. And last but not least, it is the section of conclusion.

1. Literature Review.

1.1 The concept of the corporate governance

The concept of the corporate governance appeared when the special type of the organization developed. This type of the organization is characterized by the existence of the investors, managers and possible conflict between them. For example, Cheffins, the author of “The History of Corporate Governance” claims that the corporate governance goes back to the past in the 16^th and 17^th centuries, when the East India Company, the Hudson's Bay Company, the Levant Company and the other major chartered companies were founded [Brian R. Cheffins, 2012]. And the first official reference of the corporate governance was in the Federal Register, the official journal of the federal government, where the S.E.C. raised the issue about the managerial accountability [Ocasio and Joseph, 2005].

Why the concept of the corporate governance is still relevant for the firms? Because, according to principal-agent theory, the managers of a firm may have an opportunistic behavior; it means that the agents (managers) can pursue their interests instead of the interest of principals (owners, investors) [Jensen and Meckling, 1976]. Hence, investors want to prevent such behavior of managers. They need a special mechanism to monitor the agents. Such mechanism is called the corporate governance.

1.2 The difference between financial and non-financial firms

As we found out, the corporate governance exists in such type of the organization when there are the managers and the supervisory board. This type of the organization is inherent for the firms in non-financial and financial sector. Usually the researchers, exploring the corporate governance, are divided into 2 categories. The first group of investigators engages in the corporate governance in non-financial firm and the second one - in financial firms. There are several reasons, why this separation exists. For instance, R. Levine claims in his work [R. Levine, 2003] that the financial firms, especially banks, play the crucial role in the economy. If the management of the bank is at a high level, it will mean that the bank examines deeply the firms that it plans to fund and allocates the capital efficiently. Receiving money, the company invests them into the business, thus expanding their activity and improving the economic development of the country. On the contrary, if the corporate governance of the bank is at a low level and the managers run the bank according to their own interests instead of the shareholders' and debt holders' interests, then the bank will face the problem with the effective allocation society's savings and the destroying impact on the corporate governance over the firms they fund. So the bank can be in trouble. This distress is likely to expand to the relevant financial environment, which might lead to credit restrictions with a significant effect on other industries [Ukaegbu B. and Isaiah O., 2014]. Therefore, the corporate governance of banks influences not only the activity of the banks themselves, but also the whole economy.

This point of view is supported by M. Qian and B. Y. Yeung in their work [M. Qian and B. Y. Yeung, 2014]. They discover how the bank financing influences the corporate governance of firms, which take the credits for their own aims. They prove that inefficiency in banking leads to weak corporate governance of the firms and inability of the capital market to grow. The authors find that the failure of banks to monitor firms is a cause that the companies can easily raise a loan. This situation promotes that the majority shareholders becomes less interested in the reputation loss and the following higher cost of capital. Consequently, the minority shareholders are more likely to be infringed on the rights by the controlling shareholders. They also show that the bank loans are negatively associated with firm performance because of tunneling (the practice when a majority shareholder determines the business and the assets in a way that all gain will belong to him http://www.investopedia.com/terms/t/tunneling.asp#ixzz47UuFoQSe). And last but not least, the investigators define that inefficient loan allocation impairs the equity market. If the financial system is managed mainly by inefficient banks, the equity market will stagnate.

In most countries the bank operations are strictly regulated by the government. This is the second reason of distinguishing the consideration the corporate governance of financial firms from the manufacturing companies. Moreover, the significant quantity of banks has to be conformed to the international standards, such as Basel I, Basel II etc. That is why, a lot of rules change the corporate governance of financial firms and make it specific [R. Levine, 2003].

Furthermore, R. Levine states that the bank's activity and, therefore, financial reports are nontransparent. The investigator believes that the loan quality can be concealed because this aspect is not observable for external agents. It has been proven by Furfine that the information asymmetry is larger in banks in comparison with the firms from other industries [Furfine, 2001].

Additionally, R. Levine says that the alteration of the risk composition of the assets is less time-consuming for financial firms than for companies from other sectors. The article «The loans to customers» is one of the biggest item in assets in the statement of the financial position. Changing the policy of issuing the loans quite quickly, the bank alters the assets and its risk composition [R. Levine, 2003].

Considering all arguments that are given above, we will review the research papaers which focus mostly on the corporate governance in the financial sector.

1.3 The corporate governance in financial firms

The corporate governance plays the important role in financial companies because if the firm organizes the optimal corporate governance, it may achieve great results from the activity due to the clearly built system. That is why, the scholars explore this system trying to find the best composition of items of the corporate governance for the certain type of firms in order to lead the company to the higher performance.

We dwell upon some components of the corporate governance, which explorers commonly analyze for financial firms. First of all, we start reviewing the works, which consider the board of directors. The board of directors is a tool for owners to control managers and to make sure that the activity of the company submits to their interests [Haan and Vlahu, 2016]. Thus the board of directors has two functions. The first one is to monitor the managers in order to make them work effectively in line with the interests of the investors. The second one is to advise management in what direction the business should run.

Some works study how the size of the board of directors affects the activity of the firm. For instance, Dalton et al. report that the boards with a large number of directors increase the possibility to be beneficial because they enhance the variety of relevant knowledge and expertise and also the resources that are available to the firm [Dalton et al., 1999]. In contrast, there is another popular view that the large boards are ineffective. A big amount of people inhibit the board to solve current issues quickly because the board needs more time to come to the common decision. What is more, there is a possibility that the free-riding problems amongst directors can appear [Jensen, 1993].

There is a pool of investigations that is devoted to the way how the board size influences the bank performance. Let us consider some of them. Adams and Mehran analyze in their article 35 BHCs (bank holding companies) and conclude that more people in the board increase the performance of the bank if the directors from the main board also sit on the subsidiary board [Adams and Mehran, 2012]. The explanation of this phenomenon can be the following: being a member of the main and subsidiary board, directors comprehend better the situation at different stages of organization and can develop plan or strategy that is more appropriate for this bank.

Some investigators examine how the board size affects the firm performance during the recent financial crisis. Erkens et al. and Berger et al. conduct the research separately from each other and do not find the relation between the board size and the efficiency during the hard time to banks [Erkens et al., 2012; Berger et al., 2012]. At the same time Wang et al. discover that there is an inverse relationship between size of the board and the performance in BHCs [Wang et al., 2012]. The group of authors explains this finding as Jensen did in his work: if the board is becoming too big, it is quite difficult to coordinate it and to settle different types of issues. Hence, it is unable to carry out monitoring and advisory functions [Jensen, 1993].

There is a hypothesis that the relationship between board size and firm performance can be non-linear. Some authors focus on the study of this feature. Grove et al. expect that the board size have an impact on the financial performance by means of a concave relationship [Grove et al., 2011]. They think that, initially, when the board goes up, it might bring more expertise, however, after reaching the breakpoint, the growth of the board can lead to the increase in agency conflicts and bank performance can be impaired. The researchers find the confirmation of their hypothesis. Another group of investigators tests the same hypothesis, but they obtain the contradictory results. The efficiency of the bank decreases as the number of directors rises, after the defined point the effect changes: the performance goes up as the number of directors declines. It means that the small board is inefficient and the large one is efficient. This result goes against the existing theory [de Andres et al., 2008].

All previous articles described the developed market. Of course, some scholars pay their attention to the impact of the board size on the bank performance in the emerging markets. Tai gets the result that the board size positively affects the effectiveness of national banks in the Gulf Cooperation Council (GCC) countries [Tai, 2015]. GCC includes the following countries: Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and the United Arab Emirates (UAE). The same result is obtained by Onakoya et al. Nigerian banks [Onakoya et al., 2014].

Some researchers draw attention to the attendance of the directors at board meetings. The attendance is usually prescribed in the documents, which oblige the directors to visit a definite number of meetings. That is why, only few works consider this phenomenon. Adams and Ferreira analyze 5707 directorships during the years 1986-1999 and prove that the board size is positively related to problems with attendance [Adams and Ferreira, 2012]. The bigger the board, the more free-riding behavior may occur. Aebi et al. try to find the influence of the attendance on the bank performance, using in the regression the percentage of directors who attend less than 75 percent of board meetings [Aebi et al., 2012]. However, they do not find any significant relationship between these figures.

Let us turn to the literature, which considers the busyness of the board members. There are three strands of theory about the busyness of the directors [Grove et al., 2011]. The first one states that busy directors, who participate in the several supervisory boards, are often expected to be distracted, thus they may not be effective monitors. This view has been proved by Fich and Shivadasani, who showed that when the outside directors sit on three or more boards, the companies represent lower operating profitability and lower market-to-book ratios [Fich and Shivadasani, 2004]. The second strand notices that busy outside board members are significant sources of knowledge and expertise and they increases the performance of the firm. Harris and Shimizu provided evidence to this point of view, they showed that busy directors improve acquisition performance [Harris and Shimizu, 2004]. And last but not least, there is the third strand in the literature, where the investigators do not find the relationship between directors' busyness and firm performance. The example of such work is the article by Jagannathan and Pritchard [Jagannathan and Pritchard, 2003]. They find that serving on the different supervisory boards does not lead to the opportunistic behavior of directors.

Examining the foregoing literature, Grove et al. decide to test their hypothesis about the u-shaped relationship between the busyness of directors and financial bank performance. They assume that when the member is busy, it will be beneficial to the bank because he or she can provide the relevant, industry-specific expertise. However, too many busy directors in the board intensify the agency problems and reduces monitoring ability.

We have discussed above some articles, which consider how different characteristics of the supervisory boards influence the bank performance or other crucial items, which are inherent to the activity of the financial firms. We outline only several features of the board. The researchers also analyze the expertise, independence, age and other parameters of the board members. A special attention we will pay to the one more characteristic of the board, which is called the gender diversity.

Usually the literature, which analyzes the gender diversity in the board and management, assumes that women differ from men in leadership behavior, risk-taking and so on and tries to prove these statements applying the empirical models. One such study is the Nielsen and Huse research [Nielsen and Huse, 2010]. They emphasize that the women's effectiveness in the board depends on the performed tasks. The more female members there are in the board, the better board strategic control is, but there is no relationship with board operational control. Likewise, they obtain that the presence of women makes conflicts fade in the board, hence, there is a higher possibility to improve firm performance.

The significant number of works is devoted to investigating how women influence the financial decision. There is an opinion that women are more risk averse in investment decision making than men. This concept is proved by Barsky et al., Agnew et al. and Sundйn and Surette [Barsky et al., 1997; Sundйn and Surette, 1998; Agnew et al, 2003]. It also finds the support in Niederle and Vesterlund article, where the authors argue that women are less overconfident men [Niederle and Vesterlund, 2007]. However, some investigations get the opposite result. For example, Adams and Funk demonstrate that female directors are more prone to make risky decisions [Adams and Funk, 2012].

1.4 The corporate governance in microfinance organizations

We begin with the academic article, called “Governance mechanisms, social performance disclosure and performance in microfinance: Does legal status matter?” [Estap` E-Dubreuilґ and Torreguitart-Mirada, 2015]. The authors set a twofold goal. First, they investigate the difference in governance mechanisms between MFIs with diverse legal status (non-governmental organizations, regulated commercial financial institutions). Second, they study the relation between governance mechanisms and the results of the MFIs' missions, such as providing banking services to low-income families and micro-entrepreneurs and maintaining the financial sustainability. The researchers consider not only the financial performance, but also the social performance. They explore the social performance because the aim of microfinance institution besides receiving profit is to provide social benefits and raise the standard of life.

As for financial performance, it is assessed in terms of such economic indicators as return on assets (ROA), efficiency (through operational costs) and revenues (portfolio yield). Social performance includes the total number of borrowers of the MFI to measure its breadth of outreach, the average loan size to appraise its depth of outreach and the number of female borrowers.

In order to analyze how the corporate governance influences firm's performance; the authors considered 4 groups of economic indicators. In the study they consider ownership structure and board structure to analyze the effectiveness of governance framework by the following variables: a dummy variable identifying MFIs with NGO legal status, number of board members, percent of female board members and two variables related to the role of the board in managing and monitoring the social performance of the MFI. Moreover, the researchers take into account the MFI-customer dimension, which is justified by the extended governance paradigm, where stakeholder theory has been claimed to complement agency theory by offering a more inclusive approach to corporate governance [Solomon, 2007, Christopher, 2011]. That is why the variables are considered through fair practices related to employees and ethical codes related to clients. Also, the indicators of no little significance are parameters related to the structure of management (ratio of managers versus total staff, gender diversity of management) and the number of staff incentives offered by the MFIs. Additionally, the authors include three different and complementary variables in the external governance dimension: regulation usually introduced by government, deposits and donations. Last but not least, the authors turn to three variables to control for specific characteristics of the MFIs: a measure of risk, the size of the organization, and a dummy variable, which take into consideration the level of specialization of the organization in the microfinance sector.

Using econometric model, the researchers obtain some crucial findings. One of their results shows that corporate governance has a more profound effect on social performance than on financial performance. It is found that the instruments in the ownership-board dimension raise the enhancement of all measurements of social performance.

On the contrary, the governance mechanisms targeting the staff of the MFIs, such as incentives and fair practices in labour, have no impact either on its social performance or on its financial performance.

Finally, the authors stress the positive effect of the NGO status of an MFI. It is demonstrated in the article that social performance improves for this type of ownership whereas financial performance is not affected by it.

We move on to another research carried out by Mersland and Strшm and devoted to the same theme [Mersland and Strшm, 2009]. The authors consider the influence of different factors on social performance and financial performance. They use quite a similar group of factors as the work we have described above. As for characteristic of the board, they add such variables as the number of international members, having internal auditor, CEO/chairman duality and the gender of CEO. The results obtained reveal that the financial performance improves when the board has an internal board auditor and have local rather than international directors. As regards ownership type, it does not affect financial performance in contrast to the work of Estap` E-Dubreuilґ and Torreguitart-Mirada. They find that the microfinance institution is better served with a female CEO. Furthermore, the social performance rises with CEO/chairman duality because the number of credit clients increases. However, the outreach goes down with individual loans for both average loan size and the number of credit clients.

Two previous works are based on MIX (Microfinance Information Exchange) data from different countries from all over the world. But some researchers prefer to concentrate on the single country. For instance, Gohar and Batool investigate MFIs in Pakistan [Gohar and Batool, 2015]. They find that productivity of the MFI, firm size of the MFI, individual lending, MFI age, and bank regulations have positive significant impact on financial performance. However, board size and CEO/chair duality have negative effects on it. The female director, bank regulation, firm size and urban market influence social performance positively, and the board size affects negatively the outreach.

Some investigators examine the group of countries, which are quite similar, e.g. the activity of MFIs in East Africa is studied in the next article [Mori et al., 2013]. The authors explore three unique characteristics of MFIs: regulation status, international influence and founder management. They find that regulated MFIs have larger boards, higher board independence and less gender diversification. The same results are obtained for internationally influenced MFIs. It is also said that MFIs managed by founders have higher level of board gender diversity.

As we can see the results of previous studies are different or even controversial for different countries. That is why, we would like to contribute to the study of microfinance and make some points clear.

2.Methodology

2.1 Data

At first, we decided to conduct our analysis in the Russian MFIs, but after studying the market of microfinance institutions in our country we recognized the lack of the reliable data. For this reason we chose the region of Eastern Europe and Central Asia. This region includes Russia and neighboring countries, which have the alike history, development, economy and culture that help us to make the corresponding analysis and receive the relevant results.

In order to get the necessary output we have used the data from the microfinance institutions, which was obtained from the Microfinance Information eXchange (MIX) and was supplemented by the external information from the official web-sites of the MFIs. MIX is a non-profit organization, which was founded in 2002. The goal of this organization is to accumulate the information about the microfinance institutions, provided on a voluntary basis, in order to broaden transparency and information exchange about the microfinance activity. For the moment it provides the financial and social figures from more than two thousand MFIs all over the world http://www.mixmarket.org/about. MIX provides the reliable data because the MIX specialists collect data from the financial statements, following the International Financial Reporting Standards (IFRS), management reports and other documents, which contain necessary information. Furthermore, the data from MFIs is monitored constantly.

Initially, our sample contained the 734 microfinance institutions, which were present during 5 years (2010-2014). We choose such period of time for several reasons. First of all, the figures, which are presented for these years, are the latest available data about the MFIs at the time of the sample design. Secondly, this timeline ignores crisis year. This reason plays the important role for us, because we study the firm performance and it can differs strongly during the recession. So our analysis can show us the doubtful results. Thirdly, choosing the 5 years we increases our sample, hence, obtained outcomes will be more precise.

The sample of the 734 microfinance institutions includes a lot of missings. Some of them we filled in using the external sources, such as the official MFI's websites, audited financial statements following IFRS and annual reports. The microfinance institutions, which comprises the 3 and more gaps, are removed from the sample. Eventually, we retain only 200 organizations for 5 years, it means that our sample contains the 1000 observations.

2.2 Description of variable and making hypothesis

In order to achieve the goal of our thesis we will carry out the empirical analysis based on the model, which contains 3 groups of regressions. The first group refers to MFI's risk and the regression is the following:

,

where board_size_it - the size of the board of directors,

percent_women_board_it - percent of women in the board of directors,

women_ceo_it -Female CEO,

percent_women_managers_it - percent of women in the management,

percent_women_loan_officers_it - percent of women among loan officers,

average_loan_it - average loan balance per borrower,

total_expense_to_assets_it - total expense-to-assets ratio,

operating_expense_to_loan_portfo_it - operating expense to loan portfolio ratio,

capital_to_assets_it - capital-to-assets ratio,

offices_it- offices,

legal_status_it - legal status of the MFI,

age_it - age of the MFI,

firm_size_it - firm size,

GDP_per_capita_ppp_it - GDP per capita based on the purchasing power parity.

The second and the third group will help us to test the impact of independent variables on the financial and the social performance, respectively:

,

,

where board_size_it - the size of the board of directors,

percent_women_board_it- percent of women in the board of directors,

women_ceo_it -Female CEO,

percent_women_managers_it - percent of women in the management,

percent_women_loan_officers_it - percent of women among loan officers,

portfolio_risk90_it - portfolio at risk > 90 days ratio,

gross_loan_portfolio_to_total_as_it - gross loan portfolio-to-total assets ratio,

average_loan_it - average loan balance per borrower,

total_expense_to_assets_it- total expense-to-assets ratio,

operating_expense_to_loan_portfo_it - operating expense to loan portfolio ratio,

capital_to_assets_it - capital-to-asset ratio,

offices_it - offices,

legal_status_it - legal status of the MFI,

age_it - age of the MFI,

firm_size_it - firm size,

GDP_per_capita_ppp_it - GDP per capita based on the purchasing power parity.

You can see the detailed variables' description and the descriptive statistics in the appendix 1:

Now let us turn our attention to the dependent variables. They are divided into 3 groups depending on the type of the regressions. The first group refers to the MFI's risk, hence, the dependent variables are portfolio at risk > 90 days ratio and gross loan portfolio-to-total assets ratio. Using these variables we try to measure the credit risk of microfinance organizations. The portfolio at risk > 90 days ratio is the proportion of overdue loan (more than 90 day) to the loan portfolio. This coefficient demonstrates that the rise in the overdue credit leads to the increase in the possibility of the failure to return the loan. It means that if the figure goes up, the risk for MFI also increases. As for gross loan portfolio to total assets ratio, it show us the proportion of loan portfolio in assets. The loan portfolio is one of the main items under the assets side of balance sheet. It implies that the increase in the number of credits increases the risks of the organization, because the risk of loan default rises.

The second group of regressions deals with the financial performance. We decide to use the most common figures for MFIs [Strшm and Mersland, 2009; Gohar, 2015]. Consequently, dependent variables are return on assets, return on equity, operational self-sufficiency, profit margin, portfolio yield. There is an indicator, which also reflects the firm performance. This figure is called the Tobin's Q. We decided not to use it, because Tobin's Q reflects the expected future development and does not reflect the current situation in comparison with ROA, ROE and other chosen indicators [Fosu, 2013].

The third group considers the relationship between the social performance and independent variables. At first, let us consider what kinds of proxies for social outreach are identified by the scholars. Schreiner has emphasized six aspects of social outreach of microfinance activity: worth to clients, cost to clients, depth, breadth, length, and scope [Schreiner, 2002]. We would like to dwell on each aspect.

First of all, we consider the worth to clients. This figure is hard to be measured because it is defined as a desire to pay. It depends on the concluded financial contracts, the tastes and opportunities of clients.

The second aspect is the cost to clients. This indicator is the sum of the transaction and price costs. The transaction costs are non-price costs for non-cash opportunity costs (the time for the payments of the outstanding debt or the time to apply for a loan) and for indirect cash expenses (expenses for the transport, food or something else, which are indirectly related to the loan in the MFI). These costs can be measured by historic data about the minutes, miles and money that are spent in order to apply and follow the contract for the loan. Price costs are direct cash payments for the implementation of the loan agreement. The best calculated indicator for this type of costs is the internal rate of return, determined as the interest rate, which is measured by making the present value of the cash flows of a financial contract be zero.

We move on to the depth of outreach, which is defined as the estimated value of a net gain of a particular client. The indirect proxies are used oftener than the direct indicators of depth through income or wealth because of difficulty of the measurement. That is why, researchers prefer to use the sex, location, ethnicity, housing and access to public services. However, the most common proxy for depth is the loan size. The best way to measure it is to use the average amount outstanding. If the loan size is small, it means that the depth is great, because there is a greater possibility that the person is poor.

The next aspect is the breadth of outreach. It is measured by the number of clients.

We continue with the length of outreach. It is the time length of supply of microcredits. It is difficult to measure this indicator because it is associated with future.

And the last but not least is the aspect of the scope of the outreach. It is approximated by the number of the offered financial agreements.

The researchers usually prefer to use depth and breadth of the outreach, because these figures can be easily found, for instance see the work of Gohar and Batool [Gohar and Batool, 2015]. That is why, we also use these proxies for our regression.

Additionally, the investigators tend to use one more proxy for defining social performance. It is the percent of women borrowers. They use this indicator because it was proved that females are almost three times more likely to reinvest their earnings in the business and in their families than men [Noah M., 2010].

Summarizing all the information above we decided to use three proxies for social outreach, which are the average loan balance per borrower, number of borrowers and the percent of women borrowers.

Let us specify why we chose the following independent variables.

1) The board of directors (the size of board of directors and percent of women in board of directors).

The supervisory board is one of the most crucial part in the MFI, which determines the strategy and policy of the organization and monitors the managers' activity. Thus, the qualitative and quantitative composition of the Board of Directors influences the activity of the organization. That is why, in our regression we have two variables, which are interconnected with the board. They are the size of the board of directors and the percent of women in the board of directors.

Let us dwell on the relationship between the two variables and our dependent indicators. Firstly, we consider the size of the supervisory board. It has been proven by Pathan that the bank board size is negatively related to risk-taking [Pathan, 2009]. The confirmation of this fact was received by Minton et al. for non-crisis period [Minton et al., 2010]. The explanation might be the following: usually small banks have few directors in the board. These directors are more flexible in making decisions and more often go for risky financial schemes than the large supervisory board. We find this point view very convincing.

Hypothesis 1: The board size is negatively related to the risk of the MFI.

We have already discussed how the board size influences the financial performance. The majority of authors tend to use the same logic as Jensen [Jensen, 1993]. He claims that the bigger the board is the more difficult it is to coordinate it and to come to a decision, hence, it is unable to carry out monitoring and advisory functions. Therefore, the firm performance suffers.

Hypothesis 2: The board size is negatively related to the MFI's financial performance.

It was found in the previous literature that sometimes the relationship between the board size and social performance of the MFI is negative [Gohar and Batool, 2015]. We think that the explanation can be the same as the one on the impact of this variable on the financial performance.

Hypothesis 3: The board size is negatively related to the MFI's social performance.

Secondly, let us turn our attention to the percent of women in the board of directors. In the literature review we discussed that the most researchers reckon that women are more risk averse. We agree with this statement because in our opinion women are more careful and might be less overconfident than men [Niederle and Vesterlund, 2007].

Hypothesis 4: The percent of women in the board of directors is negatively related to the risk of the MFI.

We suppose that female members of the board are negatively related to the financial performance. The confirmation of this fact we see in the article by Strшm et al., they argue that female leadership is associated with weaker corporate governance, thus, the financial performance get worse [Strшm et al., 2014]. We expect this effect because women are more risk averse, hence, they can reject the risky but profitable projects.

Hypothesis 5: The percent of women in the board of directors is negatively related to the MFI's financial performance.

In the literature the researches found the positive relationship between the female membership in the board and social outreach [Strшm et al., 2014]. This may happened because women think more about the outreach than men and try to help indigent people.

Hypothesis 6: The percent of women in board of directors is positively related to the MFI's financial performance.

1) Management (female CEO, percent of women in the management, percent of women among loan officers).

Let us talk about female CEO, percent of women in the management, percent of women among loan officers. The whole management plays the crucial role in the activity of the organization, because they perform the business. For that reason we include it in the regressions.

It has been demonstrated that women in the management in different levels reduce firm risk and improve firm performance [Perryman et al.,2016]. We support this idea, because women are more diligent than men. Also we believe that women in the management have negative impact on social performance. We find proof of this fact in the literature. Estap` E-Dubreuilґ and Torreguitart-Mirada claims that female managers are inclined to rely on the clients who have already borrowed some money and returned them. [Estap` E-Dubreuilґ and Torreguitart-Mirada, 2015]

Hypothesis 7: The female CEO is negatively related to the risk of the MFI.

Hypothesis 8: The female CEO is positively related to the MFI's financial performance.

Hypothesis 9: The female CEO is negatively related to the MFI's social performance.

Hypothesis 10: The percent of women in the management is negatively related to the risk of the MFI.

Hypothesis 11: The percent of women in the management is positively related to the MFI's financial performance.

Hypothesis 12: The percent of women in the management is negatively related to the MFI's social performance.

Hypothesis 13: The percent of women among loan officers is negatively related to the risk of the MFI.

Hypothesis 14: The percent of women among loan officers is positively related to the MFI's financial performance.

Hypothesis 15: The percent of women among loan officers is negatively related to the MFI's social performance.

2) Risks (portfolio at risk > 90 days ratio and gross loan portfolio-to-total assets ratio).

We use the proxy for risks as the dependent variable in the first group of regressions. In the second and the third group we add them into regressions as the independent variables. We carry out the analysis of the regression on proxy for risks in order to find out whether women influence the risk of the MFI. After that we move them to the determinants, which contain the variables related to women and other indicators. We do this in order to clear up whether women influence the financial and social performance through the risks of the MFI or not. Using the common sense and the literature background we hypothesize the following. [Perryman et al.,2016]

Hypothesis 16: The portfolio at risk > 90 days ratio is negatively related to the MFI's financial performance.

Hypothesis 17: The portfolio at risk > 90 days ratio is negatively related to the MFI's social performance.

Hypothesis 18: The gross loan portfolio to total assets ratio is negatively related to the MFI's financial performance. (The gross loan portfolio to total assets ratio has inverse relationship with the risk of MFIs).

Hypothesis 19 The gross loan portfolio to total assets ratio is negatively related to the MFI's social performance. (The gross loan portfolio-to-total assets ratio has inverse relationship with the risk of MFIs).

3) Determinant from the accounting report (Total expense-to-assets ratio, operating expense-to-loan portfolio ratio, capital-to-asset ratio).

We take into account these variables because they help us to consider the financial figures through the proportions. These variables allow the regression to have more stable results.

4) Average loan per borrower

We take into account this variable in two regressions as independent variable because we suppose that this crucial indicator affects the risks and the financial performance. The increase in the average loan per borrower magnifies the risks, because the probability of the loan defaults goes up. As a consequence, the financial performance diminishes.

5) Offices

The MFIs are small organizations in comparison to other firms. Usually the firms with the investors, directors and managers have subsidiaries. The MFIs are too small to allow it. That is why, we decide to consider this variable as the subsidiaries for the big form. We expect that the large number of offices attracts new clients, so the social outreach improves. Furthermore, we suppose that with the appearance of the offices the expenses on the content rises. Thus, the financial performance gets worse and the risks grow.

6) Control variables (Firm size, age, legal status)

We include the control variables in our regression in order to take into account firm-specific effects. We use the typical control variables for MFI from different studies, such as the firm size, age and legal status [Strшm and Mersland, 2009; Mori et al., 2013]. We reckon that the large microfinance organizations are more attractive for investors. The main aim of participants of the corporate governance is to improve the welfare of the investors. Hence, the company tries to make a profit with fewer risks. We expect that the firm size is negatively related to the MFI's risk and its financial performance, but it is positively related to the social outreach, because more people are maintained by the large MFI due to the confidence in this organization. As for the age, we think that there are positive relationship with the financial and social performance, but there is the negative relation with the risks. We suppose that the greater age indicates the stability and the efficiency of the company. The weak companies are unable to work for a long time. Due to the stability of the organization, it can offer loans at quite a low rate. This action may attract new clients and he;p to continue partnership with regular customers. Therefore, the increase in the age decreases the risks. Considering the legal status, we can expect that banks (legal status equals 1) are better than the firms with other forms of the legal status in financial performance and in taking the risks. However, the social performance is higher in non-bank financial institution, credit union or non-governmental organization rather than in banks. As it was already mentioned, we consider different countries, that is why, we need to control country-specific effects. We do this by the means of GDP per capita based on the purchasing power parity.

2.3 Modeling

We will analyze which specification of regression is the most appropriate for each group models. For this reason we will compare 3 types of specification: pooled, fixed effect and random effect regressions. The pooled regression is the model, which does not take into account the panel nature of the data and use the method of ordinary least squares, whereas the fixed and random effect regression were invented deliberately for panel data. There is the intercept, which takes different values for each object in the sample in fixed effect model. The meaning of the intercept is to show the influence of the missing and unobservable variables, which characterize the individual features of the objects that do not change during the different time periods. The same implication has the intercept in the random effect model, but now the specific features, which are peculiar for every MFI, are random and they may change during the periods. It shoud be also mentioned that the fixed effect and random effect regressions use the least squares dummy variable and generalized least squares method, respectively.

We have the panel data and it is important for us to keep in mind the panel characteristics of the sample. Therefore, we do not consider the pooled regression because it does not meet our requirements. We will compare the fixed effect and random effect regressions.

Let us start with the first group regressions. At first we will analyze the regression, where the dependent variable is portfolio at risk > 90 days ratio.

These 2 models estimates 656 object. Herewith the coefficient of the determination in FE model is rather small in FE model (5.5%) and we cannot say something about R-squared in RE model, because this model uses the GLS method and the meaning of this figure disappears. Also, we can notice that indicators related to women are not significant in the models. There are only 4 factors significant in FE model (operating expense to loan portfolio ratio, capital to assets ratio, firm size and the intercept). The average loan balance per borrower, operating expense to loan portfolio ratio, legal status, firm size and the intercept are significant in RE model.

We conduct the Hausman test (see the appendix 1, test 1) and figure out that the null hypothesis (that the difference in coefficients is not systematic) is not rejected and the model with random effects is more appropriate for us. We should test the regression in order to find out if there are issues like multicollinearity, heteroscedasticity and autocorrelation.

We count the variance inflation factor (VIF) by means of STATA to test the multicollinearity. The mean VIF equals 5.17 and it is less than 10. It means that there is no need to adjust our regression to the multicollinearity (see the appendix 1, test 2).

There is no tests checking the heteroscedasticity for random effect models. However, we have a suspicion that the regression faces this problem. One of the reasons of the heteroscedasticity is the diversity of the firms in the sample. Our sample might deal with an issue of heteroscedasticity, because it contains MFIs, which are located in countries with different the standard of living, legislation and traditions. Thus, the results of the microfinance activity may differ strongly in selected states. That is why, we assume that there is the heteroscedasticity in the regression.