ANALISIS DE DATOS

6.7.1 Selection of regression model

Table 6.7-1 depicts the multiple regression results for fixed-effect and random- effect approaches. The Hausman test result suggests that it is important to employ a fixed-effect model for Depth variable due to the rejection of the null hypotheses where p-values are significantly lower than the 0.05 level. The random-effect model has been employed for both Breadth and FemBorr variables because they fail to reject the null hypotheses.

Table 6.7-1: Selection of fixed or random effect model

Dependent Variables Prob>chi2 Hausman Test Result

Breadth 0.4844 Random-effect Model FemBorr 0.1883 Random-effect Model Depth 0.0083 Fixed-effect Model Note: This table presents the Hausman test results for outreach variables. The notations are defined in Table 4.6-1.

The Breusch and Pagan LM test for random effects has been used as a robustness check to see whether the choice of the random-effect model for Breadth and

167 suggests that a pooled regression model is not appropriate for Breadth and FemBorr

dependent variables as it has been rejected at 5% significance level. The alternative random-effect model is accepted as an appropriate model to represent the data.

6.7.2 Empirical results for outreach

Table 6.7-2 illustrates the empirical results of a multiple regression analysis of outreach variables in this study. These are from a regression of the whole dataset and controlling unobserved heterogeneity in the panel model. Even though most of the signs of the coefficients generated from the regression have the expected signs, very few are significant for outreach of Sri Lankan MFIs. However, interesting discoveries appear in both of these significant and non-significant regression results.

Female directors (FemDir) on the board are statistically significantly negatively correlated (t=-1.80, p=0.10) with Breadth of outreach, and significantly positively correlated (t=2.10, p=0.05) with percentage of female borrowers of total active borrows (FemBorr) in Sri Lankan MFIs. These findings indicate that the female directors appear to concentrate on gender inequality in the country and promote microfinance loans to more female clients. Males have a wide range of sources from which to access credit but most women receive their first loan from an MFI. Even though Mersland and Strøm (2009) argue that female CEOs (FemCEO) are better informed, which will result a greater outreach, they did not find significant coefficients for the relationship between female CEO and outreach. This study also finds no statistically significant relationship between those two variables.

The results for female directors on a board are opposite for a female chairperson

(FemChair) on a board. The female chairperson on a board is statistically significantly positively correlated (t=2.81, p=0.01) with Breadth of outreach but statistically significantly negatively correlated (t=-1.80, p=0.10) with female borrowers (FemBorr) in MFIs in Sri Lanka. Even though they are female leaders they appear to concentrate on increasing the number of active borrowers rather than increasing only women borrowers.

168

Table 6.7-2: The relationship between corporate governance and outreach of MFIs in Sri Lanka

Variables

Breadth FemBorr Depth

b/(p) [t] b/(p) [t] b/(p) [t] FemDir -0.376* [-1.800] 0.086** [2.094] 0.043 [1.383] (0.072) (0.036) (0.172) FemCEO -0.003 [-0.028] 0.010 [0.495] 0.015 [1.211] (0.977) (0.621) (0.231) FemChair 0.253*** [2.813] -0.032* [-1.795] -0.010 [-0.777] (0.005) (0.073) (0.441) Duality 0.034 [0.352] -0.026 [-1.363] -0.022 [-1.647] (0.725) (0.173) (0.106) IntDorDir -0.009 [-0.093] 0.047** [2.394] -0.006 [-0.600] (0.926) (0.017) (0.551) ClientDir -0.201* [-1.886] 0.034 [1.643] 0.021 [1.374] (0.059) (0.100) (0.175) IndDir -0.052 [-0.191] -0.028 [-0.524] 0.013 [0.347] (0.848) (0.601) (0.730) Bsize -0.099 [-0.874] -0.007 [-0.292] 0.003 [0.234] (0.382) (0.770) (0.816) IntAudit -0.051 [-0.581] 0.023 [1.302] 0.006 [0.701] (0.562) (0.193) (0.487) Regbank -0.201 [-0.287] -0.018 [-0.121] (0.774) (0.903) Fage -0.001 [-0.100] 0.005* [1.719] -0.013*** [-4.189] (0.921) (0.086) (0.000) Fsize 0.802*** [15.980] 0.024** [2.321] 0.015 [1.065] (0.000) (0.020) (0.292) Lev 0.237 [1.179] 0.037 [0.930] -0.022 [-1.012] (0.238) (0.352) (0.316) Constant -6.062*** [-7.281] 0.306* [1.786] -0.000 [-0.002] (0.000) (0.074) (0.998)

year dummies yes yes yes

organisation type dummies

yes yes no

firm fixed-effects no no yes

Number of observations 294 297 295 R-squared 0.413 0.157 0.243 F statistic 4.135*** Wald Chi-squared statistic 651.71*** 79.08*** Number of clusters 54

Note: This table presents the results of the relationship between corporate governance and outreach of MFIs in Sri Lanka. Asterisks indicate significance at 10% (*), 5% (**), and 1% (***). p-Values are presented in parentheses and based on robustness standard errors corrected for potential heteroskedasticity and serial correlation in the error term. t-Statistics are reported in brackets. The notations are defined in Table 4.6-1. Year dummy 2007 and organisation-type dummy private companies are treated as the benchmark categories to avoid the dummy variable trap. Year dummies and organisation-type dummies are unreported.

169 The international/donor directors (IntDorDir) have a statistically significant positive (t=2.39, p=0.05) association with female borrowers (FemBorr) which shows that directors who represent international and/or donor agencies are engaged in providing microcredit to women in Sri Lanka. This suggests that when international and/or donor representatives sit on MFI boards they are able to provide better monitoring and advisory services to improve women’s empowerment. However, the results of this study show that directors who represent clients

(ClientDir) are statistically significantly negatively (t=-1.89, p=0.10) associated with the number of active clients (Breadth) in MFIs in Sri Lanka. This is not astonishing as similar results were obtained by Mori and Mersland (2014) and Hartarska (2005). It is likely to infer from this finding that the representatives of clients on boards improve the MFI profitability at the expense of Breadth of outreach. Hartarska (2005, p. 1639) explains that these stakeholder representatives on MFI boards “may have engaged in rent-seeking behaviour, by promoting lending to wealthier borrowers”.

Other corporate governance variables such as Duality, non-executive directors on board (IndDir), number of board members (Bsize) and internal audit function

(IntAudit) are not statistically significantly impacting the outreach of MFIs in Sri Lanka. The number of MFIs that either have an internal audit function or are regulated under banking authority are very few and their impact on outreach is also negligible.

Interestingly, Depth of outreach does not have any significant relationship with corporate governance variables but does have a significant negative relationship with firm age (Fage), suggesting that as MFIs mature their attention toward outreach to poor people increases.

6.8

Conclusion

This chapter contains a discussion of the background information of MFIs in Sri Lanka and the empirical analysis of corporate governance and both financial performance and outreach of MFIs in Sri Lanka. This chapter emphasises guidance for selecting directors for MFI boards based on a board’s characteristics. In relation

170 to the traditional internal and external corporate governance mechanisms, this study finds only a few variables that influence the key financial performance and outreach of MFIs. Even though most of the signs of the coefficients generated from the regression analysis are consistent with theory, very few are statistically significant. Internal audit function, local directors, executive directors, female CEOs, female chair and client representatives on the board are found to positively influence the MFIs’ financial performance, while female directors, female chair, and international and donor directors are found to positively influence outreach of Sri Lankan MFIs.

The results also show a statistically significantly negative relationship between female directors on board, and financial performance and Breadth of outreach. Even though more than 30% of women are participating in Sri Lanka’s labour force (Department of Census and Statistics, 2011), very few have specialist managerial skills. Hence, the proportion of non-specialised female directors serving on MFI boards may higher than males (Hewa-Wellalage & Locke, 2013). It is noteworthy for MFIs’ to consider more training and mentoring activities for female directors to increase their involvement in improving financial performance and outreach of MFIs. However, women’s success in the workplace is always shaped by an array of cultural expectations, domestic responsibilities and self-perception (Women’s World Banking, 2013). In countries such as Sri Lanka, women often kept busy with family responsibilities and commitments. Therefore their impact on MFI financial performance and outreach is likely to be minimal. To build more effective and efficient diverse boards, MFIs need to consider these circumstances and develop women’s skills by having proper training and consultation.

Although there are only a small number of statistically significant results, it does point towards the view that corporate governance does matter for the financial performance of MFIs. Moreover, the evidence presented in this chapter should encourage MFIs to consider further significant governance factors which will improve and sustain the industry. Also, it would have been more appropriate to have MFI governance studies in different countries to validate the findings of this study. Therefore, in the next chapter, research is presented from Indian MFI data analysis to complement the strength this chapter’s conclusion.

171

7

CHAPTER SEVEN

CORPORATE GOVERNANCE AND

PERFORMANCE IN MFIS IN INDIA: AN

EMPIRICAL INVESTIGATION

7.1

Introduction

In Chapter 6 the discussion focused on how corporate governance practices impact the performance of MFIs in Sri Lanka. This chapter examines the Indian context. India has a large number of MFIs and may provide a useful comparison with Sri Lanka. An overview of the microfinance sector in India provides a starting point for the analysis. This is followed by an analysis of how corporate governance relates to the financial performance and outreach of the MFIs. The findings presented in Chapter 7 then compare with Chapter 6 findings to understand the differences. The chapter concludes with suggestions for changes in corporate governance practices that may favourably impact the performance of MFIs in India.