The effect of loan approval decentralization on microfinance institutions' outreach and loan portfolio quality

Journal of Business Research 94 (2019) 1–17 Contents lists available at ScienceDirect Journal of Business Research journal homepage: www.elsevier.co...

Download PDF

464KB Sizes 0 Downloads 36 Views

Report

PDF Reader
Full Text

Journal of Business Research 94 (2019) 1–17

Contents lists available at ScienceDirect

Journal of Business Research journal homepage: www.elsevier.com/locate/jbusres

The eﬀect of loan approval decentralization on microﬁnance institutions' outreach and loan portfolio quality☆

T

⁎

Hubert Tchakoute-Tchuigouaa, , Issouf Soumaréb a

Department of Finance and Accounting, KEDGE Business School, 680 Cours de la Libération, 33 405 Talence cedex, France Department of Finance, Insurance & Real Estate, & Laboratory for Financial Engineering (LABIFUL), Faculty of Business Administration, Université Laval, Building Palasis-Prince, Oﬃce 1204, 2325, rue de la Terrasse, Quebec, Quebec G1V 0A6, Canada b

A R T I C LE I N FO

A B S T R A C T

JEL classiﬁcations: G21 G32 G39

We study the impact of loan approval decentralization on MFI portfolio quality and outreach, and the eﬀects of alignment mechanisms when loan oﬃcers combine information production and decision functions. Using an independently pooled cross-section of 374 MFI-year observations for 280 MFIs in 70 countries, we ﬁnd that eﬀective incentive schemes and internal control systems help mitigate agency problems within MFIs, and thus increase the outreach of MFIs without altering the quality of their loan portfolio. Our results are robust after controlling for alternative portfolio risk and outreach measures, outreach threshold eﬀect, crisis period, selection bias and endogeneity.

Keywords: Loan approval Decentralization Loan portfolio quality Loan portfolio risk Microﬁnance Outreach

1. Introduction Microﬁnance institutions (MFIs) are double-bottom-line organizations and hybrid organizations (Battilana & Dorado, 2010; D'Espallier, Hudon, & Szafarz, 2013). They are now part of the ﬁnancial landscape of most developing and emerging countries, and their primary goal is to provide ﬁnancial services to low income people and to small and informal businesses. As reported by some recent studies (Armendáriz de Aghion & Morduch, 2010; Dixon, Ritchie, & Siwale, 2007; Tchakoute Tchuigoua, 2012), decision-making authority is allocated to the loan oﬃcer in many MFIs. Merging resource allocation and information production functions provides the loan oﬃcer with incentives to produce and use soft information when approving loans (Stein, 2002). The loan oﬃcer, whose role in the loan approval process is now evidenced in microﬁnance literature (Agier, 2012; Agier & Szafarz, 2013), becomes more powerful when information production and loan approval decisions are concentrated in his or her hands. In this study, we ﬁrst investigate whether allocating decisional authority to the loan oﬃcer improves outreach without deteriorating loan portfolio quality. However, a decentralized credit decision or a powerful loan oﬃcer can expose the MFI to a principal-agent problem and thus induce

agency conﬂicts between the loan oﬃcer and the lending organization (Berger & Udell, 2002; Stein, 2002). A loan oﬃcer may make credit decisions contrary to the interests of the MFI or without complying with current loan approval procedures. Implementing appropriate incentive schemes (better human resource management), strong internal controls, and audit procedures are thus crucial for microcredit portfolio performance and the overall ﬁnancial health of MFIs (Basel Committee on Banking Supervision (BCBS), 2010). Appropriate incentive mechanisms and internal control procedures may ensure that loan oﬃcers who approve loans undertake allocation decisions that improve MFI eﬃciency. The eﬀect of loan approval decentralization on risk and outreach will be especially important if the agent who received authority is subject to incentive systems and controls that would tend to limit the risk associated with decentralization. The second objective of this study is to investigate whether incentive schemes and internal control systems are eﬀective in reducing agency conﬂicts within an MFI. To date, the microﬁnance literature has not suﬃciently addressed the issue of agency costs induced by delegation. More speciﬁcally, the question of whether well-designed incentives and an eﬀective internal control system – whether perceived or assessed as such by a third party – could limit the risk associated with loan approval decentralization is

☆

We are grateful to participants at the seminar organized by the Department of Finance, Insurance and Real Estate and the Research Chair in Governance of Laval University in April 2016 for their helpful comments and suggestions, especially Jean Bédard. All errors and omissions are the authors' sole responsibilities. ⁎ Corresponding author. E-mail addresses: [email protected] (H. Tchakoute-Tchuigoua), [email protected] (I. Soumaré). https://doi.org/10.1016/j.jbusres.2018.09.021 Received 9 May 2017; Received in revised form 19 September 2018; Accepted 20 September 2018 0148-2963/ © 2018 Elsevier Inc. All rights reserved.

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

not well documented.1 Some recent microﬁnance papers investigate the risk associated with decentralization in microﬁnance lending in developing countries, including subjective preferences of the loan oﬃcer and the impact of loan oﬃcer behaviors on risk and MFI outreach (Agier & Szafarz, 2013; Labie, Méon, Mersland, & Szafarz, 2015; Sagamba, Shchetinin, & Yusupov, 2013). Agier (2012) emphasizes the role of credit oﬃcers on microloan performance and Tchakoute Tchuigoua (2012) examines the eﬀect of decentralization on loan contract terms. This empirical literature on microﬁnance then suggests a link between decentralization and adverse selection and provides evidence that granting authority to loan oﬃcers may bias the selection process because of subjective loan oﬃcer preferences. The eﬀectiveness of existing alignment mechanisms in MFIs can therefore enable MFIs to overcome this principal-agent problem arising from decentralizing the loan decision, and also align loan oﬃcer interests with those of the MFI. To analyze the eﬀectiveness of alignment mechanisms on the risk and outreach of MFIs, we study an independently pooled cross-sectional sample of 374 MFI-year observations for 280 MFIs from 2001 to 2012 across 70 countries and make at least two main contributions to the existing microﬁnance literature. First, we extend upon previous studies on loan oﬃcer role and behavior in MFIs. Contrary to Agier (2012) who limits the role of the loan oﬃcer to information production, we assume loan oﬃcer duality and the resulting principal-agent problem between the MFI and the loan oﬃcer who both produces information and allocates loans. By assessing the mitigating eﬀect of incentive schemes and internal control mechanisms on loan oﬃcer behavior, our contribution also goes beyond the previous literature, which focuses on loan oﬃcer subjectivity (Agier & Szafarz, 2013; Labie et al., 2015; Sagamba et al., 2013). These studies assume or demonstrate that discriminatory behavior of a loan oﬃcer is partly because of inexistence of a control system or failures of existing internal control systems. The quality of the internal control system and the quality of incentive mechanisms are not taken into account from an empirical point of view. Our study accounts for this missing piece by considering the existence of eﬀective internal control systems and incentive mechanisms. We measure the eﬀectiveness of the internal control system and the incentive system using the rating scores produced by a rating agency, Planet Rating in this case, while controlling for the associated selection bias. Second, by focusing on loan oﬃcer-MFI agency problems and on the eﬀectiveness of alignment mechanisms within MFIs, we extend upon the previously mentioned microﬁnance studies and contribute to the existing microﬁnance corporate governance literature. Indeed, some studies have focused on the agency conﬂicts between MFIs executives and owners and on the eﬀectiveness of incentive schemes and control mechanisms implemented by MFIs in order to align the interests of these two groups (e.g., Hartarska, 2005; Hartarska & Mersland, 2012; Mersland & Strøm, 2009; Tchakoute Tchuigoua, 2014). Based on previous studies in nonﬁnancial organizations (Adams, Almeida, & Ferreira, 2005), MFIs (Galema, Lensink, & Mersland, 2012) and banks (Pathan, 2009), some others have placed particular emphasis on CEO power, that is, the merging of Chairman and CEO positions into a single position, and examined its impact on performance and risk. Other studies focus on agency problems between the interest of borrowers and the interest of MFIs, and assess the eﬀectiveness of incentive design for aligning the borrower and MFI preferences (Armendáriz de Aghion &

Morduch, 2010; Stiglitz, 1990). However, except for a few theoretical studies that have examined the incentives designed for loan oﬃcers (Aubert, de Janvry, & Sadoulet, 2009; Besley & Ghatak, 2005; Conning, 1999), little is known from an empirical standpoint about a principalagent problem involving MFIs and loan oﬃcers, and the eﬀectiveness of alignment mechanisms designed to avoid information problems within MFIs. Our empirical evidence supports the hypothesis that providing the loan oﬃcer with incentives such as performance-based pay and putting in place an eﬀective internal control system may contribute to aligning the interest of the loan oﬃcer with that of the institution. Hence, implementation of human resource management practices and internal control systems, which are perceived as eﬀective, mitigates the possible deterioration of loan portfolio quality following loan approval decentralization, without altering MFI outreach. This result links our study to the literature on organizational architecture (Berger & Udell, 2002; Stein, 2002) and to microﬁnance-speciﬁc literature which focuses on designing incentives for loan oﬃcers (Aubert et al., 2009; BCBS, 2010; Labie et al., 2015). We account for the eﬀect of MFI ownership type, given that managerial discretion and proﬁt distribution constraints are likely to vary across diﬀerent types of ownership among MFIs. We ﬁnd that incentive mechanisms and internal control system eﬀectiveness increase MFI outreach without altering loan portfolio quality when loan approval is decentralized, in both not-for-proﬁt and proﬁt-oriented MFIs; the eﬀects are much stronger in not-for-proﬁt organizations and even tend to reduce risk. The remainder of the article is organized as follows. Part two gives an overview of prior literature and develops the research hypotheses. Part three explains the research design. Part four presents the results and robustness tests, and part ﬁve concludes with an acknowledgment of research limitations and avenues for future research. 2. Background 2.1. Why is loan approval decentralization an important issue for microlending? The corporate governance literature provides an explanation of why ﬁrms grant decision-making authority to agents (loan oﬃcers). Incentive-based theories (Aghion & Tirole, 1997; Stein, 2002) suggest that transferring the decision-making authority to agents who produce soft information is a way to recognize and reward their expertise in this ﬁeld and enable the loan oﬃcers to make appropriate decisions. Sah and Stiglitz (1986) explain the choice of a decentralized decisionmaking structure in terms of the cost of acquiring and communicating information. Given information asymmetries between the person who gathers and processes information, and the one with the authority to make decisions, the data communicated by the former to the latter can be either partial or contaminated, thereby leading to ﬂawed decision making. Investigating decentralization of the loan approval process is of particular interest in the microﬁnance sector for at least two reasons. First, close ties between MFIs and their clients are one of the main features of microﬁnance (Stiglitz, 1990). Close proximity to the poor not only makes it easier for MFIs to understand their clients' needs but also enables them to develop trust with the communities in which they operate and develop and oﬀer products and services in line with the ﬁnancial needs of their intended target markets (see, for example, Ledgerwood, Earne, & Nelson, 2013). Proximity also enables loan ofﬁcers to produce soft information and to grant loans eﬃciently. Given that loan oﬃcers usually live in the same local community as their borrowers and maintain direct and personal contact with them, they may build privileged ties with other small businesses and individuals who hold relevant information about potential borrowers and their businesses in the local community (Berger & Udell, 2002). In addition, daily interactions and personal relations between loan oﬃcers and local

1 The existing banking literature emphasizes lending practices of large and small banks and their eﬀects on credit availability, risk and proﬁtability (Berger & Black, 2011; Berger, Cowan, & Frame, 2011; Berger, Miller, Petersen, Rajan, & Stein, 2005); loan oﬃcer's rotation policy in commercial banking and its eﬀects on moral hazard, that is, the oﬃcers' reporting behaviors (Hertzberg, Liberti, & Paravisini, 2010); hierarchical distance of information use in large multinational banks (Liberti & Mian, 2009); and gender bias in bank lending markets (Beck, Behr, & Guettler, 2013; Bellucci, Borisov, & Zazzaro, 2010).

2

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

qualitative information that could lead to the rejection of credit applications that he or she defends (Berger & Udell, 2002). The loan ofﬁcer can thus voluntarily select borrowers with low ability to repay, which may result in nonperforming loans and thus deteriorate the MFI's loan portfolio quality. According to one stream of the governance literature (Berger & Udell, 2002; Brickley, Linck, & Smith, 2003; Stein, 2002) and the microﬁnance-speciﬁc literature (Aubert et al., 2009; Jeon & Menicucci, 2011; Labie et al., 2015), providing the loan oﬃcer with incentives such as performance-based pay and an eﬀective internal control system may thus contribute to aligning the interests of the institution with the loan oﬃcer's interest, and thus avoid information asymmetry between the principal (the MFI's management or its credit committee) and the agent (the loan oﬃcer). Aubert et al. (2009) developed an ex ante information acquisition model and show that setting up an incentive mechanism will tend to align the interest of credit oﬃcer with that of the MFI. The credit oﬃcer will thus select borrowers adequately based on their ability to repay and their poverty levels. Assuming an agency conﬂict between the MFI and the biased loan oﬃcer with loan decision-making authority, Labie et al. (2015) show how designing wage incentives in a nonproﬁt MFI may deter its loan oﬃcers from discriminating. They document that oﬀering incentives in socially oriented MFIs mitigates discrimination by loan oﬃcers. Jeon and Menicucci (2011) demonstrate how incentive schemes help avoid moral hazard and constrain loan oﬃcer behavior. Incentive schemes limit the loan oﬃcer's ability to embezzle borrower repayments. These theoretical studies suggest that well-designed incentive schemes allow an MFI to improve outreach, limit MFI discrimination against disabled micro-entrepreneurs, and help improve the MFI's loan portfolio quality. From an empirical standpoint, except for exploratory studies,2 analysis of the eﬀectiveness of incentive mechanisms for loan oﬃcers and internal control systems in reducing agency conﬂicts within the MFI has received very little attention in the microﬁnance literature. Based on the previously described literature, we may expect that providing incentives to the loan oﬃcer and a better internal control system are likely to mitigate the eﬀects of powerful loan oﬃcers on MFI performance. We thus expect that incentive schemes and internal control systems may ensure that loan oﬃcers who approve loans make allocation decisions that improve the MFI's outreach and the loan portfolio quality, hence the following Hypothesis 2.

borrowers provide easy access to soft information. Second, according to the BCBS (2010), most MFI failures stem from deterioration in the quality of loan portfolios. Christen, Lauer, Lyman, and Rosenberg (2012) suggest that there may be agency problems in loan allocation and that deterioration of an MFI's loan portfolio quality may result from the ineﬀectiveness of the allocation process and the internal governance system. The 2014 Centre for the Study of Financial Innovation (CSFI) survey, Microﬁnance Banana Skins, ranks credit risk (that is, risk that poor lending practices lead to loan losses) second among risks facing MFIs and ﬁnds strong links with weaknesses in management and internal control systems. In addition, the microcredit market has become increasingly competitive, and conclusions of the CSFI (2014) show that microﬁnance experts rank competition third among the risks facing MFIs. The increased competition among MFIs in the microcredit market pushes them to expand their credit portfolio and to increase their market share. Competition enables MFIs to expand their outreach but seems to exacerbate agency problems in loan allocation. Some recent empirical studies evidence the negative eﬀect of competition on MFI loan portfolios (Assefa, Hermes, & Meesters, 2013; Baquero, Hamadi, & Heinen, 2018; Guha & Chowdhury, 2013; Tchakoute Tchuigoua, 2016). To address the loan portfolio eﬀect of competition, MFIs implement human resource policies among which staﬀ incentive systems to ensure that the lending activity will not jeopardize the sustainability and survival of the MFI. 2.2. Hypotheses development 2.2.1. The eﬀect of Loan oﬃcer approval on MFI outreach and loan portfolio quality Recent studies using household survey data in developing and emerging countries document that proximity allows MFIs to produce soft information and improves access to ﬁnance (Allen et al., 2014; Allen, Demirgüc-Kunt, Klapper, & Martinez-Peria, 2016; Brown, Guin, & Kirschenmann, 2016). Lending decisions based on the use of soft and hard information enable MFIs to better screen borrowers. Simultaneous use of soft and hard information may lead to accepting loan applications that could have been rejected solely on the basis of hard information, which is not necessarily reliable or relevant in the microﬁnance sector. Decentralization thus helps to increase access to credit and improve loan conditions. We may thus expect that outreach increases when the loan oﬃcer approves loans, which leads to the following Hypothesis 1.a.

H2. The eﬀects of loan approval decentralization on loan portfolio quality and outreach are attenuated when alignment mechanisms are eﬀective.

H1a. Loan approval decentralization is positively associated with MFIs outreach. Using soft information as a complement to hard information thus reinforces the ability of MFIs to select borrowers eﬃciently based on their repayment capacity, which de facto limits the risk that the credit portfolio will deteriorate. Decentralization contributes to limiting the risk in the credit decision and may have a positive eﬀect on the quality of the MFI loan portfolio. We may thus expect the riskiness of the loan portfolio to decrease when the loan oﬃcer approves loans (Hypothesis 1.b).

2.2.3. The eﬀect of MFI ownership types MFIs oﬀer ﬁnancial services to the poor across several institutional forms including cooperatives, credit unions and non-governmental organizations (NGOs), usually grouped in the category of nonproﬁt MFIs, and banks and nonbank ﬁnancial institutions (NBFIs), considered as privately owned and proﬁt-oriented MFIs. Since the transformation movement of MFIs was initiated in the early 1990s, the debate over the impact of choosing an institutional form and the impact of the evolution of ownership type remains a subject of concern among practitioners and scholars in the ﬁeld of microﬁnance. Some studies have attempted to answer the question of whether the ﬁnancial and social performance of

H1b. Loan approval decentralization is positively associated with MFIs loan portfolio quality. 2.2.2. The mitigating eﬀect of alignment mechanisms Decentralization of the credit decision is not a suﬃcient condition to overcome informational problems in MFIs. Indeed, a powerful loan oﬃcer can create a principal-agent problem. Strong relationships between the loan oﬃcer and the MFI's clients (borrowers) and the resulting lending decentralization gives considerable leeway to loan ofﬁcers, especially when incentive schemes and internal control systems are lacking, and the existing ones ineﬀective. For subjective reasons, the loan oﬃcer may voluntarily grant non-performing loans by ignoring

2 An international survey on staﬀ incentive scheme practices in the microﬁnance industry (McKim & Hughart, 2005) shows that the percentage of MFIs implementing staﬀ incentive schemes increased for > 12 years. They show that the percentage of MFIs using staﬀ incentive schemes grew from 6% to 63% between 1990 and 2003. In 2005, they found 72% of MFIs had an incentive scheme, with some diﬀerences across regions, by legal status and by MFI characteristics. Incentive schemes for loan oﬃcers are individual monetary schemes (bonuses) and are found in 83% of MFIs.

3

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

MFIs varies across MFIs ownership type, and ﬁnd consistent evidence. Results do not evidence diﬀerences in accounting performance across MFI ownership types and thus yield inconsistent evidence with the transformation thesis. Mersland and Strøm (2008) and Tchakoute Tchuigoua (2010), among others, show that NGOs are not more socially oriented than privately owned MFIs, nor are privately owned MFIs more proﬁtable than NGOs. This is evidenced as well in a more recent study by D'Espallier, Goedecke, Hudon, and Mersland (2017), which applies an event study methodology to compare the performance of 66 transformed MFIs over the period 1993–2011, and investigates how transformation inﬂuences the MFI's overall business model, including the MFI's proﬁt function, the mix of funding sources and the scope of services provided. MFIs clients' characteristics are one of the main distinctive features of microﬁnance ownership type (Jansson, Rosales, & Westley, 2004; Tchakoute Tchuigoua, 2010). As noted by Galema et al. (2012), microﬁnance NGOs borrowers tend to be low-income entrepreneurs, work in informal sectors, with no physical capital and have limited formal documentation. In these microﬁnance NGOs, loan approval is mostly decentralized and based on the use of soft lending technology and on a ﬁeld evaluation of the client's ability to pay. Conversely, privately owned MFIs target wealthier clients with greater ability to repay their loans, and are more centered on the use of hard lending technology to assess risk and the borrower's ability to repay. This would result in better credit portfolio quality and larger average loan size in for-proﬁt MFIs than in other institutional forms. This seems to be conﬁrmed by some existing empirical studies. Indeed, D'Espallier et al. (2017) ﬁnd a signiﬁcant increase in average loan size, both in absolute terms and scaled by GNI per capita for transformed MFIs. Tchakoute Tchuigoua (2010) ﬁnds that private microﬁnance companies have better loan portfolio quality than nonproﬁt MFIs. He also ﬁnds no diﬀerence among nonproﬁt MFIs, that is, between cooperatives and NGOs. The existing diﬀerences between for-proﬁt MFIs and not-for-proﬁt ones may be explained by the fact that, privately owned MFIs adopt management procedures and practices similar to those of conventional banks, and they may be expected to implement similar sophisticated risk management practices, as opposed to those used by cooperatives and NGOs. This ﬁnding may also be driven by diﬀerences in staﬀ incentives and internal control systems within MFIs. In addition, as documented by Galema et al. (2012) and Servin, Lensink, and Van den Berg (2012), each MFI ownership type is speciﬁc in terms of managerial discretion (the level of ties between ownership and control) and proﬁt-distribution constraints (those who receive proﬁts earned by MFIs). NGOs are characterized by a non-distribution constraint and cooperative/credit unions and privately owned MFIs distribute proﬁts, respectively, to members and owners. Privately owned MFIs apply market-based principles that may enable them to implement mechanisms, such as market or corporate control or performance-based compensations, to eﬀectively inﬂuence opportunistic behaviors of loan oﬃcers. Perilleux, Hudon, and Bloy (2012) and Hudon and Périlleux (2014) also provide supportive evidence for an existing diﬀerence in proﬁt distribution constraints across MFIs ownership type. They use a stakeholder-centered approach and answer the question of how MFIs allocate their surplus to stakeholders and whether the allocation process varies depending on the MFI ownership type. They document that the surplus distribution is more in favor of providers and employees in cooperatives, whereas, NGOs and shareholderMFIs keep their surplus internally as a self-ﬁnancing mechanism. Moreover, according to Aubert et al. (2009), designing incentives to align the interests of employees with the MFI's objectives is more costly in socially oriented MFIs than in for-proﬁt oriented MFIs. This may even results in a trade-oﬀ between ﬁghting discrimination against disabled borrowers and extending outreach to the poor in nonproﬁt MFIs (Labie et al., 2015). These latter two studies are consistent in the way that, in the presence of agency costs, introducing incentives improves social and ﬁnancial performance in for-proﬁt MFIs.

The above literature review allows us to conclude that for-proﬁt MFIs have better outreach and better loan portfolio quality than nonproﬁt MFIs. However, as argued previously, allocating the loan decision-making authority to the loan oﬃcer can lead the loan oﬃcer to prioritize his subjective preferences and loan expansion can be seen as a result of the loan agent's subjective preferences. In this case, it can be expected that in for-proﬁt MFIs, designing eﬀective alignment mechanisms will help to mitigate the discretionary power of the loan ofﬁcer and limits the eﬀect of the loan oﬃcer managerial discretion on the loan portfolio quality and the outreach of the MFI. We thus formulate the following Hypothesis 3. H3. The attenuating eﬀect of eﬀective alignment mechanisms will be stronger for privately owned MFIs than for not-for-proﬁt MFIs. 3. Research design 3.1. Model The main question addressed in this study is whether incentive schemes and internal control systems limit agency conﬂicts between the MFI and the loan oﬃcer when the loan approval process is decentralized. Incentive schemes and internal control systems will be eﬀective if the riskiness of the loan portfolio decreases or remains the same while outreach increases. To address this research question, we ﬁrst model MFI performance (risk and outreach) as a cross-sectional function of the credit approval decision, internal control eﬀectiveness, and human resource management eﬀectiveness after controlling for some MFI-level variables, country-level variables, year ﬁxed eﬀects and heteroscedasticity. The estimated model is as follows:

MFI performance = α0 + αi Xi + βi Yi + γj ICVj + δt + ε,

(1)

where i indexes MFIs, j indexes country and t indexes year. Xi is the vector of our main MFI-speciﬁc variables: Loan oﬃcer approval, Governance eﬀectiveness, Internal control eﬀectiveness, and the cross products “Loan oﬃcer approval ∗ Governance eﬀectiveness” and “Loan oﬃcer approval ∗ Internal control eﬀectiveness”; Yi is the vector of other MFI-level variables: Proﬁtability, Loan portfolio growth, Individual lending, and the cross product “Loan oﬃcer approval ∗ Individual lending,” and MFI ownership type. ICVj is the vector of country-level variables: the Kaufmann, Kraay, and Mastruzzi (2010) corruption index and GDP growth. δt is the year ﬁxed eﬀects and ε is the error term. To answer the question of whether the eﬀects of governance eﬀectiveness and internal control eﬀectiveness when MFIs decentralize the loan approval process vary by MFI ownership type, we split the sample into two subgroups (for-proﬁt MFIs versus not-for-proﬁt MFIs) and reestimated Eq. (1) for each subgroup. 3.2. Variables 3.2.1. Dependent variables Our ﬁrst dependent variable is MFI loan portfolio risk, which we measure by three indicators. The main indicator is the Portfolio at risk at 30 days, which measures the portion of the loan portfolio that is 30 days past due. For a robustness test, we use two alternative measures to proxy loan portfolio risk, namely, Write-oﬀs ratio and Loan loss provision expenses. We use three outreach indicators as dependent variables. The main indicator is the Breadth of outreach, measured by the number of active borrowers. The second outreach indicator is the Yield on loan portfolio (real) which proxies the annualized interest rate charged on loans by MFIs (Cozarenco, Hudon, & Szafarz, 2016; Cull, Demirgüç-Kunt, & Morduch, 2007; D'Espallier et al., 2013). The third measure of outreach is the Depth of outreach, which measures the outreach to the poor. 4

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 1 Description of variables. Identity of variable

Deﬁnition and measurement

Outreach

Breadth of outreach

Yield on loan portfolio

Loan portfolio quality indicators

Depth of outreach Portfolio at risk 30 days

Write-oﬀ ratio

Loan loss provision expense ratio Decentralization

Loan oﬃcer approval

Type of loan contract

Credit committee approval Governance eﬀectiveness Internal control eﬀectiveness Individual lending

Proﬁtability MFIs growth Ownership type

Return on assets Loan portfolio growth For-proﬁt MFIs

Country level variables

Country corruption index

Governance ratings Internal control ratings

Country economic growth

Natural logarithm of the number of active borrowers Number of active borrowers (NAB) If NAB > 30,000, high outreach; If 10,000 ≤ NAB ≤ 30,000, medium outreach; if NAB < 10,000, low outreach (Yield on Gross Portfolio (nominal) − Inﬂation Rate) / (1 + Inﬂation Rate) Yield on Gross Portfolio = Interest and Fees on Loan Portfolio / Loan Portfolio Average loan size per borrower scaled by the per capita gross national income (GNI). (Outstanding Balance on Arrears over 30 days + Total Gross Outstanding Reﬁnanced (restructured) Portfolio)/ Total Gross Portfolio Measurement of portfolio quality. It shows the part of the portfolio aﬀected by outstanding payments when there is a risk that they might not be repaid. The threshold is < 10% given that ﬁnancial guarantees in microﬁnance are not always suﬃcient. Write Oﬀs/Loan Portfolio, gross, average Total amount of loans written oﬀ during the period. A write-oﬀ is an accounting procedure that removes the outstanding balance of the loan from the Loan Portfolio and from the Impairment Loss Allowance when these loans are recognized as uncollectable. Net loan loss provision expense/Average gross outstanding portfolio Net loan loss provision expense = Loan loss provision expense and write-oﬀ minus Recovery from Loans written oﬀ Dummy: 1 if loans are approved by the loan oﬃcer, and zero otherwise (the credit decision is taken at the branch committee level or at headquarter). Dummy: 1 if loans are approved by the MFI credit committee, and zero otherwise. Value comprised between 1 and 5; 5 being the highest (better governance). Value comprised between 1 and 5; 5 being the highest (better internal control). Individual loans as a percentage of the outstanding loan portfolio. Loan is granted to a single borrower. Net Operating Income/Assets, average Relative change of the gross loan portfolio Privately owned MFIs included microﬁnance banks and non-bank ﬁnancial institutions (NBFI), Binary variable: 1 if the MFI is a privately owned MFI; 0 otherwise. Control of corruption from the World Governance Indicators (WGI): The index reﬂects perceptions of the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as “capture” of the state by elites and private interests. Estimate of governance (ranges from approximately −2.5 (weak) to 2.5 (strong) governance performance). Source: Kaufmann et al. (2010); World Bank. Annual growth rate of the real GDP per capita

Table 2 MFIs sample distribution by region and year. This table provides the distribution of MFIs sample by region and by year. Panel A: MFIs distribution by region Region

Number of MFIs

% of the sample

Africa & Middle East Europe & Central Asia Latin America & Caribbean South Asia, East Asia & the Paciﬁc Total

149 56 141 28 374

39.84 14.97 37.70 7.49 100

Panel B: MFIs distribution by year Year

Number of MFIs

% of sample

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Total

2 11 24 28 54 60 46 36 33 37 30 13 374

0.53 2.94 6.42 7.49 14.44 16.04 12.30 9.63 8.82 9.89 8.02 3.48 100.00

5

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 3 Summary statistics. This table displays summary statistics for the variables used in the regressions. We provide deﬁnitions of the variables in Table 1.

Portfolio at risk 30 days Loan Loss provision expense Write-oﬀ ratio Breadth of outreach Depth of outreach Yield on loan portfolio Loan oﬃcer approval Credit committee approval Governance eﬀectiveness Internal control eﬀectiveness Individual lending Proﬁtability: return on assets Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth

Number of observation

Mean

Standard deviation

Minimum

Maximum

Median

373 367 326 374 322 343 374 374 374 374 371 374 334 374 373 373

0.05 0.02 0.03 21,983 0.75 0.38 0.52 0.28 3.43 3.37 0.61 0.03 0.50 0.38 −0.46 0.06

0.06 0.04 0.06 44,882 1.28 0.19 0.50 0.45 0.82 1.00 0.41 0.09 0.97 0.49 0.43 0.03

0.00 −0.19 0.00 23.00 0.00 0.02 0.00 0.00 1.00 1.00 0.00 −0.68 −0.74 0.00 −1.51 −0.05

0.45 0.34 0.56 477,267 9.52 1.10 1.00 1.00 5.00 5.00 1.00 0.52 15.94 1.00 1.47 0.17

0.03 0.01 0.01 10,284 0.32 0.33 1.00 0.00 3.00 3.00 0.84 0.03 0.35 0.00 −0.46 0.05

approval takes a value of 1 if the loan approval decision is made by the branch credit committee and 0 otherwise.

3.2.2. Explanatory variables Loan oﬃcer approval: When the allocation decision is decentralized, MFIs face at least two agency conﬂicts: (1) one between the MFI and the loan oﬃcer when he or she approves loans and (2) another one between headquarters and the branch when loans are approved at the branch level. We initially focused on the ﬁrst agency conﬂict, the one between the MFI and the loan oﬃcer. To build the variable Loan oﬃcer approval, we conduct an exploratory study of rating reports to answer the question of whether the loan oﬃcer has decision-making authority or not. The exploratory analysis of rating reports enabled us to identify three situations. In the ﬁrst situation, the loan oﬃcer has the authority to make the ﬁnal credit approval decision. Resource allocation and information production are associated. Some cases, such as Beneﬁt and Mikroﬁn in Bosnia & Herzegovina, Miselini in Mali, JV MFO Microinvest in Moldova and MFO Alliance in Georgia, illustrate this situation. For example, in the case of JV MFO Microinvest in Moldova, the loan oﬃcer approves loans up to US$7000. The loan approval process in this case is fully decentralized and the loan oﬃcer is the decision maker. In the second situation, the branch-level credit committee that comprises the branch manager and loan oﬃcers approves loans. Loan oﬃcers who act as an interface between the credit committee and borrowers produce information, and the ﬁnal authority to approve loans rests with the credit committee. By participating in the credit committee at the branch level, it is more likely that the loan oﬃcer inﬂuences the credit decision. The loan approval in this case is semidecentralized, and the loan oﬃcer is associated with the loan approval decision. The case of JV MFO Microinvest in Moldova illustrates this situation. The branch credit committee approves loan of between US $7000 and US$30,000. In the third situation, the headquarters approves loans. The loan approval decision is centralized, that is, all loan applications are ﬁrst completed by loan oﬃcers, reviewed by the branch manager, and then sent to the relevant oﬃcer at headquarters toward credit committee approval. Nor Horizon in Armenia, Nachala in Bulgaria, Narodnyi Kredit in Russia and LEAD in Egypt are illustrative cases of this third situation. We code the variable Loan oﬃcer approval as 1 if the loan oﬃcer both uses information production and also approves credit, and 0 otherwise. This corresponds to the ﬁrst case described. As a robustness test for the decentralization of the loan approval decision, we use the alternative measure Branch-level credit committee approval stemming from the second situation described above, where the loan oﬃcer is associated with the loan approval decision taken at the branch level. In this case, the variable Branch-credit committee

3.2.2.1. Internal control eﬀectiveness. We proxy the internal control system eﬀectiveness using the aggregate scores provided by Planet Rating. Indeed, Planet Rating uses a ﬁve-point alphabetical rating scale (from e to a) to assess each of the main areas of rating: governance, information, risk management, activity, funding and liquidity, and eﬃciency. We thus measure the eﬀectiveness of MFI internal control systems using the risk management ratings provided by Planet Rating. The risk management ratings cover three main areas: risk management systems, internal controls and procedures, and internal audits. We match the ratings (e to a) with values between 1 and 5, with the value of 1 corresponding to grade e, 2 for grade d, 3 for grade c, 4 for grade b, and 5 for grade a. The higher the value the better the perceived eﬀectiveness of the MFI's risk management and internal control system. 3.2.2.2. Governance eﬀectiveness. Here also, we use Planet Rating governance eﬀectiveness scores as a proxy for the eﬀectiveness of human resource management. The corporate governance score provided by Planet Rating, in addition to the eﬀectiveness of the governance structure and decision-making processes, captures the relevance of the strategy, the eﬀectiveness of the management team and the eﬀectiveness of human resource management practices (notably the recruitment process, training of loan oﬃcers and existence of a well-designed and motivating remuneration scheme). The governance eﬀectiveness score is measured on the same scale as the internal control system and takes values between 1 and 5, with 5 being the highest value (better governance). 3.2.3. Other control variables MFIs use either individual loan contracts or joint liability contracts, which consist of group solidarity lending and village banking lending. Following previous studies that link the choice of joint liability contract to MFI social eﬃciency and loan portfolio quality (Cull et al., 2007; Mersland & Strøm, 2009; Tchakoute Tchuigoua, 2012), we account for the type of lending contract applied by the MFI. We include Individual lending as an indicator of MFI lending methodology, and the cross product “Individual lending ∗ Loan oﬃcer approval” to assess whether loan decentralization inﬂuences the eﬀect of the lending methodology on MFIs' risk and outreach. We measure Individual lending as average loan size approved under individual loan contracts as a percentage of the overall gross loan portfolio. Because of the lack and insuﬃciency of data, we were unable to account for the loans granted by product type (consumption loans, micro-enterprise loans) or by industry (agricultural loans, services loans, trade loans and manufacturing loans). 6

Journal of Business Research 94 (2019) 1–17

1.00 −0.14⁎⁎

Corruption index

1.00 0.04 0.01

For proﬁt MFIs Loan portfolio growth

Data used in this study comes from two diﬀerent sources. Our main source of data is Planet Rating's website (www.planetrating.org). Organizational structure data, such as governance, internal control and loan approval decisions, and also other MFIs level variables, come from assessment reports produced by Planet Rating, a rating agency specialized in rating microﬁnance institutions. We chose Planet Rating data for at least two reasons. First, alternative data sources such as the Microﬁnance Information Exchange (MIX) database do not enable us to gather core and useful data to conduct this kind of study on the choice of decentralized versus centralized loan approval decisions, the eﬀectiveness of internal MFI control systems, and the eﬀectiveness of MFI governance practices. Planet Rating reports provide detailed information about the loan approval process and MFI organizational structure. Second, among rating agencies specialized in MFI ratings, Planet Rating seems to be the only rater that assigns scores to each of the areas assessed. Indeed, Planet Rating, through its GIRAFE (governance, information, risk management, activity, funding and liquidity, eﬃciency) methodology rates and assigns scores to internal MFI control systems (risk management) and governance practice eﬀectiveness (including human resources management). Rating reports for rated MFIs between 2003 and 2015 are available on the Planet Rating website.3 During the 2003–2015 period, we count about 435 rating reports available on the Planet Rating website. However, we were unable to extract some of the reports because they remained conﬁdential and were not freely available, speciﬁcally, recent rating reports published between 2013 and 2015. Additionally, some other rating reports by Planet Rating were not available on the website of Planet Rating. The Rating Fund website (www.ratingfund2.org) enables us to exploit assessment reports produced by Planet Rating but unavailable on the Planet Rating website, such as rating reports during the period 2001–2003, and thus complements data gathered from the Planet Rating website. We exclude rating reports in which precise information about the loan approval process was missing. We end up with a ﬁnal sample comprised of 374 assessment reports for 280 MFIs in 70 countries from 2001 to 2012. Country-level data come from the World Bank's World Development Indicators (WDI)4 and Worldwide Governance Indicators (WGI)5 databases. Table 2 presents the distributions of the sample by region (panel A) and by year (panel B). As panel A of the table shows, the sample includes MFIs from the following four main regions: Africa and the Middle East (149 rating reports); Eastern Europe and Central Asia (56 rating reports); Latin America and the Caribbean (141 rating reports); South Asia, East Asia, and the Paciﬁc (28 rating reports).

1.00 0.02 −0.04 0.11⁎⁎ 0.03

Proﬁtability

1.00 0.02 −0.06 0.16⁎⁎⁎

3.3. Sample

0.11⁎⁎ 0.10⁎⁎ −0.02 0.28⁎⁎⁎ 0.16⁎⁎⁎ −0.08

3

www.planetrating.com/FR/rating-girafe.html. http://data.worldbank.org/data-catalog/world-development-indicators. 5 http://info.worldbank.org/governance/wgi/index.asp. 4

⁎⁎

⁎

p < 0.10. p < 0.05. ⁎⁎⁎ p < 0.01.

0.14 0.14⁎⁎⁎ −0.05 0.25⁎⁎⁎ 0.15⁎⁎⁎ −0.07 0.00 0.05 0.09 −0.04 0.10⁎ −0.15⁎⁎⁎

−0.08 −0.03 −0.04 0.01 −0.04 0.14⁎⁎⁎

⁎⁎⁎

1.00 0.04 0.09⁎

Loan oﬃcer approval Credit committee approval Governance eﬀectiveness Internal control system eﬀectiveness Individual lending proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth

1.00 −0.66⁎⁎⁎ 0.02 −0.02

1.00 0.69⁎⁎⁎

1.00

1.00 0.01 0.02 0.15⁎⁎⁎ 0.04 −0.03

Individual lending Internal control system eﬀectiveness Governance eﬀectiveness Credit committee approval Loan oﬃcer approval

Table 4 Correlation matrix. This table presents the Pearson matrix of the explanatory variables used in the regressions. We provide deﬁnitions of the variables in Table 1.

Ownership type: We measure Ownership type as a dummy that takes a value of 1 if the MFI is proﬁt-oriented (for-proﬁt) and zero otherwise. Proﬁt-oriented MFIs include microﬁnance banks and nonbank ﬁnancial institutions (NBFI) (Tchakoute Tchuigoua, 2010). We measure Proﬁtability as the return on assets (ROA) and include Loan portfolio growth to account for the growth of the MFI. Country-speciﬁc variables: Based on previous country case studies (Patten, Rosengard, & Johnston Jr., 2001) and existing empirical papers such as Ahlin, Lin, and Maio (2011), who associated country macroeconomic conditions with MFI social performance and eﬃciency, we control for the country's economic growth or economic performance (GDP growth measured as the growth rate of the real per capita GDP). We also control for the institutional framework using the Country corruption index. Table 1 summarizes and describes the variables used in the study.

1.00

Annual GDP growth

H. Tchakoute-Tchuigoua, I. Soumaré

7

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 5a Baseline results. This table reports the results of the pooled OLS with risk measures as dependent variables. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row. Governance eﬀectiveness

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness

Internal control system eﬀectiveness

Portfolio at risk at 30 days

Write-oﬀs ratio

Loan loss provision

Portfolio at risk at 30 days

Write-oﬀs ratio

Loan loss provision

−0.0762⁎⁎ (−2.041) −0.0315⁎⁎⁎ (−3.671) 0.0201⁎⁎ (2.062)

−0.0717⁎⁎ (−2.493) −0.0188⁎⁎⁎ (−2.973) 0.0200⁎⁎⁎ (2.755)

−0.00345 (−0.168) −0.00191 (−0.378) −0.000168 (−0.0308)

−0.0455⁎ (−1.693)

−0.0392 (−1.498)

−0.00160 (−0.0756)

−0.0210⁎⁎⁎ (−3.433) 0.0110 (1.599) 0.0168⁎⁎ (2.564) 0.000206 (0.925) −0.182⁎⁎⁎ (−5.325) 0.00585 (1.055) −0.000181 (−0.0305) 0.0111 (1.534) −0.138 (−1.419) 0.105⁎⁎⁎ (3.691) Yes 330 0.224 0.171 10.84

−0.0123⁎⁎⁎ (−2.909) 0.0107 (1.610) −0.00899 (−0.954) −0.000354 (−1.509) −0.165⁎⁎⁎ (−3.466) −0.00110 (−0.585) 0.00210 (0.329) 0.00789 (1.194) −0.134 (−0.865) 0.0720⁎⁎⁎ (3.304) Yes 319 0.193 0.136 .

−0.00715 (−1.633) −0.000808 (−0.147) 0.000516 (0.112) −0.0000129 (−0.0604) −0.139⁎⁎⁎ (−3.086) 0.00494⁎⁎⁎ (3.891) 0.0117⁎⁎ (2.399) 0.00891⁎ (1.801) −0.123 (−1.302) 0.0504⁎⁎ (2.495) Yes 328 0.202 0.147 5.069

Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth _cons Year ﬁxed eﬀects Number of observations R-sq Adj. R-sq F

0.0181⁎⁎⁎ (2.743) 0.0000986 (0.405) −0.180⁎⁎⁎ (−5.604) 0.00513 (1.099) −0.000993 (−0.169) 0.0118 (1.555) −0.115 (−1.177) 0.142⁎⁎⁎ (3.928) Yes 330 0.241 0.189 10.55

−0.00842 (−0.923) −0.000374 (−1.629) −0.166⁎⁎⁎ (−3.504) −0.00155 (−0.896) 0.00162 (0.260) 0.00730 (1.057) −0.115 (−0.759) 0.0893⁎⁎⁎ (3.168) Yes 319 0.199 0.142 .

−0.000416 (−0.0861) −0.0000231 (−0.103) −0.144⁎⁎⁎ (−3.102) 0.00494⁎⁎⁎ (3.356) 0.00848⁎ (1.752) 0.00748 (1.524) −0.110 (−1.143) 0.0245 (1.131) Yes 328 0.173 0.117 10.25

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

4. Empirical results and discussion

4.2. Multivariate analysis

4.1. Summary statistics

The distribution of MFIs per year, given in panel B of Table 2, shows that the data structure looks like a panel in that the data is observed over 12 years (from 2001 to 2012). However, the sample size is not the same from one year to another, and the data contains diﬀerent statistical units at diﬀerent points in time. The data structure is thus an independently pooled cross-section (IPCS). No MFI renewed its rating 12 times during the sample period. Among the 280 MFIs in our sample, only two MFIs renewed their ratings four times; ﬁve MFIs renewed three times; nine MFIs renewed twice; and ﬁnally only 53 MFIs renewed once. Therefore, we use the pooled sample robust OLS estimation technique with controls for year eﬀects as suggested by Wooldridge (2010). Tables 5a and 5b report the results of the pooled OLS with robust standard errors and control for year ﬁxed eﬀects for the whole sample. We assess whether governance eﬀectiveness and a better internal control system mitigate agency problems within an MFI, especially when the MFI decentralizes the loan approval process. We assess the eﬀect of these mitigating instruments on risk (portfolio at risk 30 days) and outreach (breadth of outreach). The table also presents results for the alternative risk measures (write-oﬀ ratio and loan loss provision) and the alternative outreach measures (yield on loan portfolio and depth of outreach). The results show that allowing the loan oﬃcer to have decisionmaking authority has a signiﬁcant reducing eﬀect on portfolio risk, and

Summary statistics presented in Table 3 show that the average value of loans over 30 days past due is 0.05, below the 0.1 threshold (Bruett, 2005). We can conclude that our sample loan portfolio is healthy. In 52% of the cases, the loan oﬃcer approves loans. Our sample average governance eﬀectiveness exhibits a value of 3.43, indicating that the MFIs in our sample obtain a governance rating grade between c and b. The average value of internal control system ratings is 3.37, indicating that the average sample grade is also between c and b. On average, outstanding individual loans account for 61% of loans granted by MFIs. Privately owned MFIs (NBFIs and microﬁnance banks) represent 38% of the observations. Prior to estimations, we assessed the presence of multicollinearity among our explanatory variables by computing the correlation between our variables (Table 4). Our diagnostic reveals that, except for the correlation between ratings of governance eﬀectiveness and the internal control eﬀectiveness (r = 0.69; p < 0.01) and the correlation between loan oﬃcer approval and branch credit committee approval (r = −0.66; p < 0.01), the intensity of the relationship among the other explanatory variables is relatively weak. To avoid multicollinearity, we include the indicators of governance eﬀectiveness and internal control eﬀectiveness separately in the regressions.

8

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 5b Baseline results. This table reports the results of the pooled OLS with outreach measures as dependent variables. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row. Governance eﬀectiveness

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness

Internal control system eﬀectiveness

Breadth of outreach

Yield on loan portfolio

Depth of outreach

Breadth of outreach

Yield on loan portfolio

Depth of outreach

2.978⁎⁎⁎ (4.626) 1.024⁎⁎⁎ (7.251) −0.656⁎⁎⁎ (−3.700)

−0.186⁎⁎ (−2.077) −0.0282 (−1.423) 0.0506⁎⁎ (2.109)

0.508 (0.655) −0.306⁎⁎ (−2.002) −0.102 (−0.515)

1.985⁎⁎⁎ (3.876)

−0.112 (−1.537)

0.176 (0.320)

0.736⁎⁎⁎ (6.163) −0.362⁎⁎ (−2.517) −1.049⁎⁎⁎ (−5.949) 0.00425 (0.729) 1.047 (1.518) −0.156⁎⁎⁎ (−2.839) −0.000421 (−0.00270) −0.270 (−1.490) −3.960 (−1.374) 7.116⁎⁎⁎ (12.74) Yes 331 0.299 0.251 9.765

−0.0202 (−1.315) 0.0300 (1.550) −0.113⁎⁎⁎ (−4.163) 0.00138⁎ (1.779) −0.0438 (−0.276) 0.0387⁎⁎⁎ (3.524) 0.0296 (1.391) 0.0299 (1.582) −0.806⁎⁎ (−2.077) 0.584⁎⁎⁎ (7.888) Yes 329 0.187 0.132 8.254

−0.219⁎⁎ (−2.103) −0.0153 (−0.109) 1.156⁎⁎⁎ (6.587) −0.00251 (−0.889) −0.564 (−0.580) −0.0308 (−0.888) 0.168 (1.243) −0.266 (−1.273) −1.395 (−0.499) 1.828⁎⁎⁎ (3.985) Yes 309 0.224 0.167 .

Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth _cons Year ﬁxed eﬀects N R-sq Adj. R-sq F

−1.082⁎⁎⁎ (−6.134) 0.00776 (1.206) 1.039 (1.489) −0.133⁎⁎⁎ (−3.116) 0.0542 (0.356) −0.275 (−1.631) −4.825⁎ (−1.669) 6.172⁎⁎⁎ (9.593) Yes 331 0.303 0.256 12.09

−0.113⁎⁎⁎ (−4.178) 0.00145⁎ (1.872) −0.0492 (−0.311) 0.0379⁎⁎⁎ (3.214) 0.0282 (1.352) 0.0260 (1.383) −0.772⁎⁎ (−1.986) 0.598⁎⁎⁎ (6.662) Yes 329 0.192 0.137 7.961

1.181⁎⁎⁎ (6.740) −0.00519 (−1.613) −0.355 (−0.410) −0.0373 (−1.127) 0.178 (1.318) −0.218 (−1.045) −1.637 (−0.587) 2.389⁎⁎⁎ (3.945) Yes 309 0.246 0.191 .

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

adequately. The implementation of good governance (better human resource management) practices and internal control systems, perceived as effective, improves the social performance of the MFIs (outreach) while mitigating portfolio quality deterioration. Analyzing the cross product of the loan approval variable and governance or internal control indicators, our results show that better governance quality and internal control systems improve MFI eﬃciency in selecting borrowers, and hence improve outreach. Indeed, we observe that better governance and internal control systems have an attenuation eﬀect on credit expansion when the loan oﬃcer approves loans, which explains why the eﬀect on risk is non-signiﬁcant or reducing. This result suggests that the implementation of appropriate performance pay, audits, and incentive schemes constrains the loan oﬃcer's behavior, limiting the risk of selecting bad borrowers, thereby improving the quality of screening. This evidence thus supports Hypothesis 2. Providing the loan oﬃcer with incentives such as performance-based pay and putting in place an effective internal control system may contribute to aligning the interest of the institution with the loan oﬃcer's interest when loan approval decisions are decentralized. This result is in line with the governance literature (Berger & Udell, 2002; Stein, 2002) and BCBS (2010) and Jeon and Menicucci (2011). For further analysis and to test Hypothesis 3, we divide our sample by ownership type to explore whether the eﬀects diﬀer depending on the commercial orientation of the MFIs. Tables 6a and 6b present the regression results by ownership type (proﬁt-oriented MFIs versus not-

thus has a positive eﬀect on the quality of the loan portfolio. However, this eﬀect is only observed for two of those measures (portfolio at risk 30 days and write-oﬀs ratio). The eﬀect on the MFIs' outreach is positive and signiﬁcant. Indeed, the relationship between decentralization of loan approval and the breadth of outreach is positive and signiﬁcant in the governance and internal control models, suggesting that access to credit is greater in MFIs when the loan oﬃcer approves loans. This is conﬁrmed when we use the alternative outreach measure, the yield on loan portfolio. This ﬁnding is consistent with Hypothesis 1a and seems to support the assumption that proximity improves ﬁnancial inclusion and access to loan (Allen et al., 2014; Allen et al., 2016; Brown et al., 2016). In addition, the eﬀect on the quality of the portfolio is positive for the two risk measures portfolio at risk 30 days and write-oﬀs, and non-signiﬁcant for the risk measure loan loss provision. Therefore, the quality of the loan portfolio is either not aﬀected or improves, thus supporting Hypothesis 1.b. This result suggests that decentralizing loan approval decisions or allocating decision-making authority to the loan oﬃcer increases the outreach but does not necessarily alter the loan portfolio quality, especially when there are better governance mechanisms and internal control systems in place. The deterioration of the loan portfolio quality that may have resulted from credit expansion to more borrowers, in our sample, is mitigated by the eﬀectiveness of the governance mechanism and the effectiveness of the internal control system. These ﬁndings support Aubert et al. (2009) and Labie et al. (2015), who show that setting up these incentives may enable loan oﬃcers to select borrowers 9

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 6a Ownership type results. This table reports the results of the pooled OLS with the loan portfolio at risk at 30 days measures as dependent variable. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row.

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth Corruption index Annual GDP growth _cons Year ﬁxed eﬀects N R-sq Adj. R-sq F

For proﬁt MFIs

Not for proﬁt MFIs

−0.0109 (−0.179) −0.0148 (−1.026) 0.00512 (0.305)

−0.116⁎⁎ (−2.255) −0.0395⁎⁎⁎ (−3.603) 0.0345⁎⁎ (2.461)

−0.0201 (−0.305)

Table 6b Ownership type results. This table reports the results of the pooled OLS with Breadth of outreach as dependent variable. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row.

−0.0566⁎ (−1.722)

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending

−0.0208⁎⁎⁎ (−3.018) 0.0170⁎ (1.957)

−0.0224 (−1.393) 0.00766 (0.430)

0.0211 (1.494) −0.0301 (−1.168)

0.0245⁎ (1.683) −0.0338 (−1.362)

0.0263⁎⁎⁎ (3.080) 0.000306⁎ (1.687)

0.0243⁎⁎⁎ (2.867) 0.000450⁎⁎⁎ (2.627)

−0.150⁎⁎ (−2.617) −0.00920 (−1.549) 0.0114 (0.961) −0.143 (−1.224) 0.149⁎⁎ (2.272) Yes 129 0.252 0.121 .

−0.155⁎⁎ (−2.345) −0.0105⁎ (−1.795) 0.0111 (0.961) −0.161 (−1.430) 0.196⁎⁎⁎ (2.673) Yes 129 0.295 0.172

−0.209⁎⁎⁎ (−4.706) 0.00740⁎⁎ (2.355) 0.0150⁎ (1.688) −0.0487 (−0.308) 0.151⁎⁎⁎ (3.319) Yes 201 0.311 0.235 17.74

−0.216⁎⁎⁎ (−4.657) 0.00862⁎⁎ (2.176) 0.0171⁎ (1.786) −0.0707 (−0.443) 0.0869⁎⁎⁎ (2.686) Yes 201 0.270 0.189 15.68

Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth Corruption index Annual GDP growth _cons Year ﬁxed eﬀects N R-sq Adj. R-sq F

For proﬁt MFIs

Not for proﬁt MFIs

3.685⁎⁎⁎ (3.855) 1.066⁎⁎⁎ (4.556) −1.027⁎⁎⁎ (−3.738)

2.062⁎⁎ (2.430) 0.988⁎⁎⁎ (5.305) −0.392 (−1.600)

0.452 (0.451)

2.266⁎⁎⁎ (3.580)

0.850⁎⁎⁎ (6.050) −0.464⁎⁎ (−2.494)

0.340 (1.596) −0.0788 (−0.287)

−1.395⁎⁎⁎ (−3.182) 0.974⁎ (1.682)

−1.235⁎⁎⁎ (−2.722) 0.636 (1.102)

−1.132⁎⁎⁎ (−5.055) 0.00513 (0.697)

−1.140⁎⁎⁎ (−5.096) −0.00120 (−0.179)

0.823 (0.857) −0.0531 (−0.235) −0.288 (−0.939) −2.240 (−0.556) 5.707⁎⁎⁎ (4.952) Yes 129 0.291 0.168 .

0.128 (0.117) −0.0422 (−0.175) −0.283 (−0.807) −0.730 (−0.175) 7.894⁎⁎⁎ (6.948) Yes 129 0.205 0.066 .

1.234 (1.091) −0.134⁎⁎⁎ (−2.997) −0.224 (−0.963) −9.587⁎ (−1.953) 6.307⁎⁎⁎ (8.061) Yes 202 0.386 0.318 14.04

1.238 (1.177) −0.172⁎⁎⁎ (−4.415) −0.314 (−1.248) −9.097⁎ (−1.949) 6.997⁎⁎⁎ (9.903) Yes 202 0.419 0.355 14.40

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

for-proﬁt MFIs). This table reports the results of the pooled OLS with robust standard errors and controls for year ﬁxed eﬀects. We assess whether the eﬀect of better internal control systems and better human resource management on MFI performance – in terms of loan portfolio quality (panel A) and outreach (panel B) – diﬀer by ownership type when the loan approval process is decentralized. The results in this table show that loan approval decentralization has a positive eﬀect on the outreach of not-for-proﬁt as well as proﬁtoriented MFIs (panel B). Moreover, as in the full sample case, having eﬀective governance mechanisms and internal control systems in place has an attenuating eﬀect on a powerful loan oﬃcer's impact on MFI outreach in both subgroups (proﬁt-oriented MFIs and not-for-proﬁt MFIs). However, the interaction eﬀect between the loan oﬃcer approval and the governance eﬀectiveness is signiﬁcant only among proﬁt-oriented MFIs, whereas the cross product loan oﬃcer approval ∗ Internal control eﬀectiveness is signiﬁcant only among notfor-proﬁt oriented organizations. As for the loan portfolio quality (panel A), we ﬁnd signiﬁcant differences across MFI ownership types with respect to the direct impact of loan approval decentralization. Indeed, loan approval decentralization reduces risk only among the not-for-proﬁt MFIs, whereas its coeﬃcient is non-signiﬁcant in the proﬁt-oriented subgroup. Eﬀective governance and internal control systems signiﬁcantly improve not-for-proﬁt MFIs' loan portfolio quality; the coeﬃcients are not signiﬁcant in the forproﬁt MFIs subgroup. Thus, in nonproﬁt-oriented MFIs and proﬁt-

oriented MFIs, although loan approval decentralization increases MFI outreach, it does not alter the loan portfolio quality. The existence of eﬀective governance mechanisms and internal control systems even contributes to signiﬁcantly reducing loan portfolio risk in not-for-proﬁt microﬁnance organizations. Hence, with respect to our third hypothesis (H3), we ﬁnd the eﬀectiveness of incentive controls and governance mechanisms to positively impact both nonproﬁt MFIs and for-proﬁt MFIs, with no speciﬁc attenuating intensity for proﬁt-oriented MFIs as posit by our hypothesis. The other control variables have the expected signs. For instance, individual lending, as opposed to joint liability contracts, reduces the outreach and the portfolio quality, and more so among not-for-proﬁt MFIs. The presence of a loan oﬃcer with loan authorization power tends to slightly reduce (10% signiﬁcance level) the negative impact of individual loan contracts on outreach in for-proﬁt oriented MFIs. Therefore, joint liability contracts can be eﬀective as risk management tools in MFIs. Proﬁtability, measured by the return on assets (ROA), signiﬁcantly reduces the loan portfolio risk and has no signiﬁcant impact on outreach. High loan portfolio growth reduces portfolio quality and outreach in not-for-proﬁt MFIs. High levels of corruption discourage the MFI from allocating credit to more borrowers and are detrimental to portfolio quality, especially among not-for-proﬁt oriented MFIs. Finally, GDP growth improves the loan portfolio quality, and corruption worsens it.

10

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 7a Robustness checks. Analyzing the crisis eﬀect. This table reports the results of the pooled OLS with portfolio at risk at 30 days as dependent variable. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row.

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness

Pre-crisis period Y01–Y06

Crisis period Y07–Y09

Post-crisis period Y10–Y12

Pre-crisis period Y01–Y06

Crisis period Y07–Y09

Post-crisis period Y10–Y12

−0.0786 (−1.443) −0.0356⁎⁎⁎ (−2.644) 0.0209 (1.491)

0.0590 (1.177) -0.00593 (−0.570) −0.0172 (−1.359)

−0.214⁎⁎ (−2.288) −0.0543⁎⁎ (−2.151) 0.0553⁎⁎ (2.119)

−0.0540 (−1.343)

−0.0189 (−0.323)

−0.106 (−1.511)

−0.0189⁎⁎ (−2.257) 0.0126 (1.367) 0.0205 (1.419) −0.00179 (−0.100) −0.162⁎⁎⁎ (−3.518) 0.00838⁎⁎⁎ (2.911) −0.0136⁎⁎ (−2.149) 0.0129 (1.018) 0.0622 (0.395) 0.109⁎⁎ (2.403) 154 0.238 0.184 5.315

−0.0238 (−1.572) 0.00409 (0.264) 0.00812 (0.683) 0.000642⁎⁎⁎ (4.127) −0.190⁎⁎ (−2.517) −0.0158⁎ (−1.666) 0.0180 (1.477) 0.0159 (1.344) −0.194 (−1.372) 0.140⁎⁎ (2.374) 99 0.286 0.205 6.147

−0.0391⁎ (−1.788) 0.0267 (1.292) 0.0338 (1.136) 0.00532 (0.172) −0.0993⁎⁎ (−2.444) −0.0339 (−1.603) −0.000223 (−0.0194) −0.00488 (−0.389) −0.0779 (−0.346) 0.184⁎⁎⁎ (2.671) 76 0.224 0.104 5.486

Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth _cons N R-sq Adj. R-sq F

0.0269⁎⁎ (2.083) −0.00394 (−0.236) −0.145⁎⁎⁎ (−3.403) 0.00722⁎⁎⁎ (3.064) −0.0106⁎ (−1.660) 0.0171 (1.256) 0.127 (0.859) 0.157⁎⁎⁎ (2.699) 154 0.288 0.238 6.438

0.00605 (0.514) 0.000529⁎⁎⁎ (2.853) −0.188⁎⁎⁎ (−2.786) −0.0177 (−1.614) 0.0170 (1.330) 0.0201⁎ (1.663) −0.181 (−1.289) 0.0840⁎ (1.838) 99 0.253 0.168 4.978

0.0360 (1.227) 0.00722 (0.233) −0.0872⁎⁎ (−2.233) −0.0167 (−1.241) −0.00293 (−0.253) −0.00454 (−0.367) −0.0347 (−0.170) 0.236⁎⁎ (2.643) 76 0.294 0.186 4.379

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

(2010−2012). The results provided in Tables 7a and 7b show that there are diﬀering results depending on the sub-period considered. Indeed, the results for the pre-crisis period and the post-crisis period are consistent with our ﬁndings above for the whole sample period. However, over the crisis period, eﬀective governance mechanisms and internal control systems had positive impacts on outreach, and no signiﬁcant eﬀect on portfolio risk. Thus, overall, our results are robust even after controlling for the crisis eﬀects, since the results show that loan oﬃcer approval and eﬀective governance mechanisms and internal control systems increase outreach without reducing portfolio quality. We were expecting this, because no special behaviors were observed in terms of the coeﬃcients of the year dummies for the crisis years in the full sample regressions presented in Tables 5a and 5b above.

4.3. Robustness checks 4.3.1. Use of alternative portfolio quality and outreach measures To strengthen the validity and robustness of our results, we conducted a number of robustness analyses. First, we use two alternative loan portfolio quality or risk measures in addition to our main risk measure, portfolio at risk 30 days, namely write-oﬀs ratio and loan loss provisions. Using these two additional risk measures to run our regressions does not contradict our main ﬁndings mentioned above and shown in Tables 5a and 5b. Indeed, we ﬁnd similar signs for the coefﬁcients of the write-oﬀs ratio measure as in the case of our main risk measure, portfolio at risk 30 days. With the loan loss provisions variable, although the coeﬃcients of interest are not signiﬁcant, the ﬁndings do not contradict our results. As alternative measures for outreach, we use the yield on loan portfolio, measured as the total interest and fees on the gross loan portfolio, and the depth of outreach, obtained as the average loan size per borrower scaled by the per capita gross national income (GNI). The results obtained and presented in Tables 5a and 5b are consistent with previous ﬁndings. In sum, granting loan approval authority to the loan oﬃcer combined with eﬀective governance mechanisms and internal control systems increases MFI outreach without jeopardizing loan portfolio quality.

4.3.3. Controlling for selection bias and endogeneity Third, we control for the endogeneity of governance and internal control ratings, and of the loan approval decentralization. Indeed, only using data on rated MFIs in our study can create potential selection bias, in that ratings are only assigned if the MFI decides to be rated. The same holds for loan approval decentralization, as this can be a selfselection phenomenon. Hence one may argue that only MFIs that are conﬁdent about the indicators covered in the rating will go for a rating; and, those only feeling strongly about governance mechanisms and internal control will implement them eﬀectively. Therefore, pooled OLS estimations may yield biased and inconsistent results. To address the selection bias and endogeneity issues, we implement the two-step procedure developed by Heckman (1979). In the ﬁrst step of this procedure, we estimate two probit selection models (a rating

4.3.2. Controlling for the 2007–2009 crisis period eﬀect Second, to control for the potential eﬀect of the 2007–2009 crisis period, we split the sample into three subsamples: the pre-crisis period (2001–2006), the crisis period 2007–2009 and the post-crisis period 11

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 7b Analyzing the crisis eﬀect. This table reports the results of the pooled OLS with Breadth of outreach as dependent variable. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row.

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness

Pre-crisis period Y01–Y06

Crisis period Y07–Y09

Post-crisis period Y10–Y12

Pre-crisis period Y01–Y06

Crisis period Y07–Y09

Post-crisis period Y10–Y12

2.886⁎⁎⁎ (3.719) 1.102⁎⁎⁎ (5.944) −0.671⁎⁎⁎ (−2.847)

1.746 (1.440) 0.816⁎⁎⁎ (2.645) −0.342 (−0.944)

3.449⁎ (1.818) 1.198⁎⁎⁎ (3.934) −1.004⁎⁎ (−2.168)

1.857⁎⁎⁎ (2.682)

1.264 (1.228)

2.606 (1.616)

0.644⁎⁎⁎ (4.313) −0.338⁎ (−1.790) −0.663⁎ (−1.773) 0.111 (0.226) 0.0573 (0.0500) −0.165⁎⁎⁎ (−3.741) 0.220 (0.922) −0.0794 (−0.313) −5.290 (−1.253) 6.936⁎⁎⁎ (11.31) 154 0.314 0.266 11.19

0.662⁎⁎ (2.458) −0.168 (−0.501) −1.340⁎⁎⁎ (−4.564) 0.00558 (1.010) 1.439 (1.070) 0.0667 (0.318) −0.321 (−1.083) −0.144 (−0.411) 0.996 (0.270) 7.266⁎⁎⁎ (8.085) 100 0.308 0.230 9.740

1.128⁎⁎⁎ (3.976) −0.813⁎ (−1.954) −2.297⁎⁎⁎ (−4.477) 1.582⁎⁎ (2.079) 0.313 (0.174) −0.0255 (−0.0593) −0.143 (−0.421) −0.962⁎⁎ (−2.625) −3.963 (−0.621) 6.235⁎⁎⁎ (4.956) 76 0.356 0.256 7.645

Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth _cons N R-sq Adj. R-sq F

−0.861⁎⁎ (−2.247) 0.199 (0.398) −0.265 (−0.234) −0.127⁎⁎⁎ (−3.219) 0.188 (0.805) −0.170 (−0.806) −7.679⁎ (−1.778) 5.678⁎⁎⁎ (9.024) 154 0.358 0.313 16.44

−1.306⁎⁎⁎ (−4.314) 0.00848 (1.576) 1.635 (1.281) 0.111 (0.540) −0.225 (−0.816) −0.180 (−0.548) 0.668 (0.192) 6.606⁎⁎⁎ (5.544) 100 0.297 0.217 10.32

−2.191⁎⁎⁎ (−4.098) 1.453⁎ (1.838) 0.0587 (0.0277) −0.480 (−1.298) −0.0989 (−0.302) −0.973⁎⁎ (−2.586) −4.310 (−0.697) 5.907⁎⁎⁎ (4.557) 76 0.364 0.267 4.711

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

choice decision model and a loan approval decentralization decision model) with robust standard errors. Ideally, controlling for the endogeneity of corporate governance eﬀectiveness and internal control systems requires building a control sample. To do that, we use a control sample of 561 MFI-year observations from 59 unrated MFIs with unbalanced and reliable data, over 2001–2012. This sample was extracted from the MIX database. Only MFIs rated four diamonds or higher by the MIX are considered. Financial statements in this category are certiﬁed by auditors, and for some of them, by the Big Four accounting ﬁrms (PwC, KPMG, Ernst & Young, and Deloitte). Based on the existing literature on rating choice (Adams, Burton, & Hardwick, 2003; Hartarska & Nadolnyak, 2008), we model the rating as a function of some MFI-speciﬁc characteristics, such as maturity (proxied by age), size of the loan portfolio, number of active borrowers, asset size, performance, ownership type and portfolio risk. The estimated probit rating decision model is as follows:

Loan approval decentralization decision = β0 + β1 ln(Assets) + β2 Average loan size per borrower + β3 Depth of outreach + β4 Cooperatives + β5 Privately owned + β6 Group lending + β7 Village banking lending + β8 Annual GDP growth + ε.

We use the results of Eqs. (2) and (3) to compute the inverse Mills ratios (IMR), which we introduced as control variables in Eq. (4) below to control for selection bias and endogeneity eﬀects. We then obtain the following model:

MFI performance = α0 + αi Xi + βi Yi + γj ICVj + IMR + δt + ε,

(4)

where i indexes MFIs, j indexes country and t indexes year. Xi is the vector of our main MFI-speciﬁc variables and Yi is the vector of other MFI-level variables described previously in Section 3. ICVj is the vector of country-level variables, δt is the year ﬁxed eﬀects and ε is the error term. The results for Eqs. (2) and (3) are reported in Table 8 (stage 1). We observe that the chi-squared are statistically signiﬁcant, which suggests that the decisions to seek the rating or to decentralize the loan approval decision are signiﬁcantly associated with age of MFI, outstanding loans, MFI ownership type, number of active borrowers, asset size, ROA, portfolio risk, average loan size per borrower, depth of outreach and economic performance, and therefore support the ﬁndings of previous studies in microﬁnance (Hartarska & Nadolnyak, 2008; Tchakoute Tchuigoua, 2012). We use the results of Eqs. (2) and (3) to compute the inverse Mills ratios, which we introduced as control variables in Eq. (4)

Rating decision = β0 + β1 Age + β2 Gross loan portfolio + β3 Portfolio at risk 30 days + β4 ROA + β5 Cooperatives + β6 Privately owned + β7 ln(Active borrowers) + ε.

(3)

(2)

To estimate the loan approval decentralization decision model, we follow Berger and Udell (2002), Stein (2002), Beck, Demirgüç-Kunt, and Martinez-Piera (2011) and Berger, Klapper, Martinez-Piera, and Zaidi (2008) who consider variables such as size and ownership type, and country macroeconomic variables, as determinants of the choice to decentralize the loan approval process. The estimated probit decentralization decision model is as follows: 12

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 8 Results of the Heckman two steps procedure estimation. Controlling for the endogeneity of ratings and Loan decentralization to the loan oﬃcer. Stage 1.A. Estimation of the probit rating decision model with robust standard errors. Dependent variable: rating decision

Coeﬃcients

t-Stat

⁎⁎⁎

Constant Age (maturity) Gross loan portfolio Portfolio at risk at 30 days Return on assets (proﬁtability) Ln (active borrowers) Cooperative Privately owned Pseudo R2 Chi2 Log-likelihood value Number of observations

9.50 −6.55 −4.53 −1.68 3.30 −5.41 3.68 −2.09

3.47 −0.49⁎⁎⁎ −1.51⁎⁎⁎ −0.97⁎ 2.09⁎⁎ −0.13⁎⁎⁎ 0.61⁎⁎⁎ −0.20⁎⁎ 14.15% 154.25⁎⁎⁎ −515.0896 881

Stage 1 B. Estimation of the probit loan approval decision model with robust standard errors. Dependent variable: the choice of decentralization

Coeﬃcients

t-Stat

Constant Size Cooperative Privately-owned Average loan size per borrower Depth of outreach Group lending Village banking lending Annual GDP growth Pseudo R2 Chi2 Log-likelihood value Number of observations

0.28 0.31⁎⁎⁎ 0.76⁎⁎⁎ 0.23 −0.24⁎⁎ 0.02⁎⁎ 0.20 0.18 5.07⁎ 12.31% 44.79⁎⁎⁎ −191.51457 316

0.33 4.36 3.13 1.28 −2.07 2.38 0.65 0.62 1.79

Stage 2.A Governance eﬀectiveness

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness

Internal control system eﬀectiveness

Portfolio at risk at 30 days

Write-oﬀs ratio

Loan loss provision

Portfolio at risk at 30 days

Write-oﬀs ratio

Loan loss provision

−0.0220 (−0.690) −0.0187⁎⁎⁎ (−2.823) 0.00571 (0.685)

−0.0495 (−1.456) −0.0149⁎ (−1.907) 0.0137⁎ (1.658)

−0.00667 (−0.311) −0.00452 (−0.831) 0.000838 (0.147)

−0.00978 (−0.383)

−0.0208 (−0.726)

−0.00359 (−0.146)

−0.0126⁎⁎ (−2.253) 0.00171 (0.250) 0.0169⁎⁎⁎ (2.606) 0.000184 (0.892) −0.196⁎⁎⁎ (−4.373) 0.00695 (1.387) −0.0000913 (−0.0149) 0.00920 (1.313) −0.148 (−1.370) 0.00524 (0.552) 0.00100 (0.0825) 0.0683⁎⁎⁎ (3.083) Yes 304

−0.00955⁎ (−1.914) 0.00543 (0.784) −0.00641 (−0.685) −0.000327 (−1.200) −0.141⁎⁎⁎ (−3.227) −0.00138 (−0.611) 0.00290 (0.414) 0.00926 (1.458) −0.0947 (−0.516) 0.0121 (0.855) −0.00375 (−0.476) 0.0527⁎⁎ (2.158) Yes 292

−0.0108⁎⁎ (−2.037) −0.000459 (−0.0713) −0.000221 (−0.0441) 0.0000228 (0.122) −0.150⁎⁎⁎ (−2.976) 0.00634⁎⁎⁎ (4.054) 0.00973⁎ (1.957) 0.00950⁎⁎ (1.979) −0.126 (−1.209) 0.0166⁎ (1.955) 0.00277 (0.245) 0.0543⁎⁎⁎ (2.601) Yes 302

Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth Inverse Mills ratio 1 (rating) Inverse Mills ratio 2 (decentralization) _cons Year ﬁxed eﬀects N

0.0180⁎⁎⁎ (2.720) 0.0000903 (0.408) −0.190⁎⁎⁎ (−4.546) 0.00663 (1.472) −0.000776 (−0.129) 0.0108 (1.497) −0.143 (−1.323) 0.00625 (0.641) 0.00134 (0.105) 0.0898⁎⁎⁎ (3.483) Yes 304

−0.00566 (−0.621) −0.000364 (−1.322) −0.140⁎⁎⁎ (−3.310) −0.00175 (−0.897) 0.00228 (0.334) 0.00934 (1.423) −0.0826 (−0.453) 0.0117 (0.791) −0.00478 (−0.565) 0.0664⁎ (1.943) Yes 292

−0.00116 (−0.221) −0.0000368 (−0.192) −0.155⁎⁎⁎ (−2.901) 0.00588⁎⁎⁎ (3.210) 0.00684 (1.340) 0.00857⁎ (1.737) −0.117 (−1.068) 0.00959 (1.118) 0.00152 (0.127) 0.0282 (1.274) Yes 302

(continued on next page) 13

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 8 (continued) Stage 2.A Governance eﬀectiveness

R-sq Adj. R-sq F

Internal control system eﬀectiveness

Portfolio at risk at 30 days

Write-oﬀs ratio

Loan loss provision

Portfolio at risk at 30 days

Write-oﬀs ratio

Loan loss provision

0.221 0.157 8.541

0.159 0.087 .

0.191 0.124 5.528

0.213 0.148 9.238

0.156 0.084 .

0.236 0.172 10.18

Stage 2.B Governance eﬀectiveness

Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness

Internal control system eﬀectiveness

Breadth of outreach

Yield on loan portfolio

Depth of outreach

Breadth of outreach

Yield on loan portfolio

Depth of outreach

1.878⁎⁎⁎ (2.882) 0.680⁎⁎⁎ (4.623) −0.481⁎⁎⁎ (−2.701)

−0.208⁎ (−1.928) −0.0315 (−1.274) 0.0579⁎⁎ (2.055)

0.0333 (0.0417) −0.268⁎ (−1.746) 0.00215 (0.0107)

1.655⁎⁎⁎ (3.034)

−0.138 (−1.446)

−0.452 (−0.757)

0.540⁎⁎⁎ (4.449) −0.404⁎⁎⁎ (−2.741) −0.512⁎⁎⁎ (−2.910) 0.00136 (0.281) 2.456⁎⁎⁎ (3.743) −0.0715 (−0.987) −0.0962 (−0.728) −0.344⁎⁎ (−2.474) −7.886⁎⁎⁎ (−3.328) 1.971⁎⁎⁎ (9.369) −1.552⁎⁎⁎ (−6.188) 8.232⁎⁎⁎ (12.70) Yes 304 0.504 0.463 17.32

−0.0240 (−1.182) 0.0382 (1.564) −0.124⁎⁎⁎ (−4.158) 0.00119 (1.240) −0.128 (−0.690) 0.0368⁎⁎⁎ (3.556) 0.0443⁎⁎ (2.085) 0.0291 (1.540) −0.881⁎⁎ (−2.004) −0.0628⁎ (−1.750) 0.00671 (0.187) 0.634⁎⁎⁎ (6.761) Yes 302 0.214 0.148 7.782

−0.192⁎ (−1.655) 0.131 (0.885) 1.286⁎⁎⁎ (6.630) −0.0102⁎⁎⁎ (−2.868) −0.816 (−0.854) −0.0915⁎⁎ (−2.156) 0.205 (1.478) −0.279 (−1.406) −2.125 (−0.707) −0.779⁎⁎⁎ (−2.894) −0.645⁎⁎ (−2.592) 2.626⁎⁎⁎ (4.534) Yes 300 0.274 0.214 .

Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth Inverse Mills ratio 1 (rating) Inverse Mills ratio 1 (decentralization) _cons Year ﬁxed eﬀects N R-sq Adj. R-sq F

−0.539⁎⁎⁎ (−2.999) 0.00339 (0.688) 2.196⁎⁎⁎ (3.370) −0.0596 (−1.044) −0.0821 (−0.655) −0.364⁎⁎⁎ (−2.619) −8.017⁎⁎⁎ (−3.414) 1.899⁎⁎⁎ (8.729) −1.600⁎⁎⁎ (−6.471) 7.798⁎⁎⁎ (11.13) Yes 304 0.504 0.463 19.16

−0.124⁎⁎⁎ (−4.163) 0.00123 (1.316) −0.128 (−0.679) 0.0357⁎⁎⁎ (3.204) 0.0437⁎⁎ (2.081) 0.0254 (1.344) −0.846⁎ (−1.920) −0.0651⁎ (−1.853) 0.00671 (0.193) 0.647⁎⁎⁎ (6.210) Yes 302 0.218 0.153 7.425

1.305⁎⁎⁎ (6.755) −0.0113⁎⁎⁎ (−3.036) −0.547 (−0.615) −0.0899⁎⁎ (−2.231) 0.227 (1.645) −0.239 (−1.197) −2.345 (−0.774) −0.651⁎⁎⁎ (−2.663) −0.571⁎⁎ (−2.431) 3.035⁎⁎⁎ (4.344) Yes 300 0.287 0.228 .

Stage 1.A table reports the results of the rating decision model (probit). From this model, we generate the Inverse Mills ratio 1 (rating). Stage 1.B table reports the results of the decentralization decision model (probit). From this model, we generate the Inverse Mills ratio 2 (decentralization). t-Statistics are in parentheses. Stage 2.A reports the results of the second stage with risk measures as dependent variables. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row. t-Statistics are in parentheses. Stage 2.B reports the results of the second stage with outreach measures as dependent variables. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row. t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

internal control mechanisms have an attenuating eﬀect on excessive credit growth. Overall, the results show that granting loan decisions to loan oﬃcers increases outreach without altering portfolio quality when more eﬀective governance mechanisms and internal control systems are in place.

to control for selection bias and endogeneity eﬀects. The results of the re-estimated models reported in Table 8 (stage 2) yield consistent evidence and improve the reliability of our previous results; that is, granting loan approval decision making to the loan ofﬁcer increases the MFI's outreach without altering the loan portfolio quality. With this selection bias correction model, we observe that loan oﬃcer approval has a non-signiﬁcant direct impact on portfolio risk, but a positive signiﬁcant impact on outreach, in all the regressions. However, eﬀective internal control systems and governance mechanisms have a risk-reducing eﬀect. In addition, better governance and

4.3.4. Alternative measure of loan approval decentralization Fourth, as we have discussed above, it may be the case that the loan approval decision is made at the branch credit committee level with the participation of the loan oﬃcer in the decision process. Loan oﬃcers 14

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 9 The credit committee approved loans. This table reports the results of the pooled OLS with loan portfolio at risk and Breadth of outreach as dependent variables. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row. Portfolio at risk at 30 days Credit committee approval Governance eﬀectiveness Credit committee approval ∗ governance eﬀectiveness

0.00178 (0.0515) −0.0185⁎⁎⁎ (−3.441) −0.00543 (−0.592)

Internal control system eﬀectiveness Internal control system eﬀectiveness ∗ credit committee approval Individual lending Individual lending ∗ credit committee approval Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth _cons Year ﬁxed eﬀects N R-sq Adj. R-sq F

0.0129 (1.463) 0.0188 (1.213) −0.176⁎⁎⁎ (−5.635) 0.00560 (1.176) −0.00197 (−0.325) 0.0135⁎ (1.741) −0.164 (−1.588) 0.100⁎⁎⁎ (3.453) Yes 330 0.225 0.172 4.735

Breadth of outreach −0.00276 (−0.0811)

−0.0144⁎⁎⁎ (−3.729) −0.00279 (−0.302) 0.0134 (1.536) 0.0144 (0.957) −0.179⁎⁎⁎ (−5.183) 0.00607 (1.108) −0.000715 (−0.122) 0.0118 (1.583) −0.176⁎ (−1.716) 0.0809⁎⁎⁎ (3.710) Yes 330 0.215 0.161 4.999

−0.849 (−1.041) 0.614⁎⁎⁎ (5.080) 0.188 (0.820)

−1.105⁎⁎⁎ (−4.865) −0.229 (−0.615) 0.887 (1.226) −0.131⁎⁎⁎ (−2.964) 0.0681 (0.429) −0.269 (−1.505) −3.795 (−1.276) 8.000⁎⁎⁎ (12.42) Yes 331 0.235 0.183 9.912

−0.883 (−1.336)

0.530⁎⁎⁎ (6.172) 0.152 (0.831) −1.127⁎⁎⁎ (−4.968) −0.100 (−0.277) 0.924 (1.284) −0.146⁎⁎ (−2.467) −0.00520 (−0.0330) −0.233 (−1.250) −3.302 (−1.107) 8.381⁎⁎⁎ (17.10) Yes 331 0.243 0.192 8.639

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

5. Conclusion

who act as an interface between the credit committee and borrowers produce information, and the ﬁnal authority to approve loans rests within the credit committee. However, by participating in the credit committee at the branch level, it is more likely that the loan oﬃcer inﬂuences the credit decision. The loan approval in this case is semidecentralized, and the loan oﬃcer is associated with the loan approval decision. We therefore use the branch-credit committee approval as an alternative measure of loan approval decentralization. The regression results given in Table 9 with this alternative loan approval decentralization indicator support our ﬁndings.

In this article, we examine the eﬀect of powerful loan oﬃcers on MFI outreach and loan portfolio quality, and given that powerful loan oﬃcers may exacerbate a principal-agent problem, we also investigate whether alignment mechanisms—incentive schemes and internal control systems—in place are eﬀective mitigating tools when loan oﬃcers combine information production and decision functions. We use an independently pooled cross-sectional sample of 374 MFIyear observations from 2001 to 2012 for 280 MFIs active in 70 countries. Our results suggest that decentralizing loan approval decisions or allocating decision-making authority to the loan oﬃcer increases the outreach but does not necessarily alter the loan portfolio quality, especially when there are better governance mechanisms and internal control systems in place. We also ﬁnd that incentive schemes and internal control systems help avoid agency problems within MFIs and thus increase the outreach of MFIs without altering the quality of their credit portfolio. These ﬁndings remain valid irrespective of the MFI ownership type. These results are robust after controlling for alternative portfolio risk and outreach measures, outreach threshold eﬀect, crisis period eﬀect, selection bias and endogeneity. The results are also conﬁrmed when we use branch-level credit committee approval as alternative measures of loan approval decentralization. Overall, we have shown empirically that the alignment mechanisms, namely, the eﬀectiveness of the governance and internal control systems, are eﬀective and tend to constrain the discretionary power of loan oﬃcers in MFIs. However, the results obtained must be interpreted with caution for at least two reasons. First, the analysis was based on MFI data, which did not allow us to explore whether the loan oﬃcer could,

4.3.5. Accounting for the non-linearity of outreach Fifth, above we have assumed outreach to be continuous, but it is possible that outreach is non-linear and there may be threshold eﬀects. We control for possible threshold eﬀects by ﬁrst running a multinomial regression, where high outreach (NAB > 30,000) is coded 1, medium outreach (10,000 ≤ NAB ≤ 30,000) is coded 2 and low outreach (NAB < 10,000) is coded 3. The unreported results obtained do not show any signiﬁcant diﬀerences relative to our main results.6 Additionally, we split the sample into subsamples of high outreach (NAB > 30,000), medium outreach (10,000 ≤ NAB ≤ 30,000) and low outreach (NAB < 10,000) and rerun the regressions on each subsample. The results provided in Table 10, although similar to the general trend observed with the whole sample, seem to indicate a much stronger eﬀect in the low outreach subsample. 6 For conciseness, we do not report those results tables, but they are available from the authors upon request.

15

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

Table 10 Threshold level of outreach. This table reports the results of the pooled OLS with Breadth of outreach as dependent variable. Standard errors are heteroskedastic robust. We control for year ﬁxed eﬀects. All independent variables are deﬁned in Table 1. We report estimated coeﬃcients in the ﬁrst row. Governance eﬀectiveness Low outreach Loan oﬃcer approval Governance eﬀectiveness Loan oﬃcer approval ∗ governance eﬀectiveness

⁎

1.401 (1.940) 0.553⁎⁎⁎ (3.582) −0.350 (−1.637)

Internal control eﬀectiveness

Medium outreach

High outreach

Low outreach

Medium outreach

High outreach

0.286 (0.864) 0.156⁎⁎ (2.210) −0.0622 (−0.682)

−1.528 (−0.826) −0.359 (−0.923) 0.252 (0.623)

0.729 (1.199)

0.337 (1.412)

−0.488 (−0.619)

0.252⁎ (1.903) −0.0913 (−0.532) −1.189⁎⁎⁎ (−4.682) 0.00379 (0.0111) 0.689 (0.907) −0.0735⁎ (−1.821) 0.0704 (0.381) 0.191 (1.017) −7.970⁎⁎ (−2.015) 8.940⁎⁎⁎ (12.96) Yes 159 0.344 0.249 5.502

0.118⁎ (1.943) −0.0745 (−1.121) 0.0133 (0.160) 0.00295 (1.504) 0.430 (1.507) −0.0199 (−0.292) 0.0000780 (0.00117) −0.0205 (−0.207) −1.856⁎⁎ (−2.109) 9.207⁎⁎⁎ (26.41) Yes 115 0.162 −0.016 3.214

0.134 (0.668) −0.0321 (−0.159) −0.557 (−0.827) 0.535 (0.751) 0.783 (0.445) 0.230 (0.790) −0.617⁎⁎ (−2.157) −0.113 (−0.777) −0.762 (−0.146) 10.57⁎⁎⁎ (11.12) Yes 57 0.389 0.050 2.628

Internal control system eﬀectiveness Loan oﬃcer approval ∗ internal control system eﬀectiveness Individual lending Loan oﬃcer approval ∗ individual lending Proﬁtability Loan portfolio growth For proﬁt MFIs Corruption index Annual GDP growth _cons Year ﬁxed eﬀects N R-sq Adj. R-sq F

−1.364⁎⁎⁎ (−5.195) 0.214 (0.609) 0.675 (0.964) −0.0621⁎⁎ (−2.064) 0.0682 (0.366) 0.180 (1.065) −7.334⁎ (−1.793) 7.839⁎⁎⁎ (9.293) Yes 159 0.386 0.297 8.918

0.0000673 (0.000866) 0.00407⁎⁎ (2.122) 0.270 (0.972) −0.0217 (−0.320) 0.00587 (0.0933) −0.0440 (−0.434) −1.885⁎⁎ (−2.097) 9.105⁎⁎⁎ (31.44) Yes 115 0.192 0.020 3.351

−0.245 (−0.516) 0.292 (0.546) 1.457 (0.851) −0.0331 (−0.123) −0.597⁎⁎ (−2.399) 0.0107 (0.0683) 0.760 (0.147) 12.65⁎⁎⁎ (6.583) Yes 57 0.405 0.074 12.34

t-Statistics are in parentheses. ⁎ p < 0.10. ⁎⁎ p < 0.05. ⁎⁎⁎ p < 0.01.

References

for subjective reasons, ration credit. Second, some MFIs seal the amounts when the credit decision rests with the loan oﬃcer. This limit varies from one MFI to another in our sample, so there could be some typologies/levels of credit allocation authority that we did not take into account. These limitations are good avenues for future research. Finally, another limitation is related to the measurement of the eﬀectiveness of governance mechanisms and the eﬀectiveness of internal control systems. We use ratings and assume them to be objective measures of alignment mechanisms. However, we were unable to control for raters' behavior (their independence) and thus assume zero collusion between rating agencies and rated MFIs, which is not always the case. The solution to that would have been to use ratings from other rating agencies. To that end, we investigated whether alternative aggregate measures of microﬁnance governance and internal control mechanisms exist. Unfortunately, such data is not available in the microﬁnance sector for at least two reasons: (1) Although governance and internal control are central in the rating process of most rating agencies specializing in microﬁnance, Planet Rating is to date the only one that assigns a score to each of the main rating areas. (2) Very few MFIs are rated by more than one rating agency during a year. For those who seek ratings from at least two diﬀerent rating agencies, the governance and internal control scores were not available when the MFI rating was assigned by a rating agency other than Planet Rating. Unfortunately, these data are not available in the microﬁnance sector. The lack of data does not enable us to conduct additional sensitivity analyses.

Adams, M., Burton, B., & Hardwick, P. (2003). The determinants of credit ratings in the United Kingdom insurance industry. Journal of Business Finance and Accounting, 30(3–4), 539–572. Adams, R. B., Almeida, H., & Ferreira, D. (2005). Powerful CEOs and their impact on corporate performance. Review of Financial Studies, 18(4), 1403–1432. Aghion, P., & Tirole, J. (1997). Formal and real authority in organizations. Journal of Political Economy, 105, 1–29. Agier, I. (2012). The role of credit oﬃcers in the performance of microloan. Evidence from Brazil. The Economics of Transition, 20(2), 271–297. Agier, I., & Szafarz, A. (2013). Subjectivity in credit allocation to micro-entrepreneurs: Evidence from Brazil. Small Business Economics, 41, 263–275. Ahlin, C., Lin, J., & Maio, M. (2011). Where does microﬁnance ﬂourish? Microﬁnance institutions performance in macroeconomic context. Journal of Development Economics, 95, 105–120. Allen, F., Carletti, E., Cull, R., Qian, J., Senbet, L., & Valenzuela, P. (2014). Improving access to banking – Evidence from Kenya, CEPR discussion paper no. DP9840. Allen, F., Demirgüc-Kunt, A., Klapper, L., & Martinez-Peria, M. S. (2016). The foundations of ﬁnancial inclusion: Understanding ownership and use of formal accounts. Journal of Financial Intermediation, 27, 1–30. Armendáriz de Aghion, B., & Morduch, J. (2010). The economics of microﬁnance (2nd ed). Cambridge, MA: MIT Press. Assefa, E., Hermes, N., & Meesters, A. (2013). Competition and the performance of microﬁnance institutions. Applied Financial Economics, 23(9), 767–782. Aubert, C., de Janvry, A., & Sadoulet, E. (2009). Designing credit agent incentives to prevent mission drift in pro-poor microﬁnance institutions. Journal of Development Economics, 90, 153–162. Baquero, G., Hamadi, H., & Heinen, A. (2018). Competition, loan rates, and information dispersion in nonproﬁt and for-proﬁt microcredit markets. Journal of Money, Credit and Banking, 50(5), 893–937. Basel Committee on Banking Supervision (BCBS) (2010). Microﬁnance activities and the core principles for eﬀective banking supervision. Basel, Switzerland: Bank for International Settlements. Battilana, J., & Dorado, S. (2010). Building sustainable hybrid organizations: The case of

16

Journal of Business Research 94 (2019) 1–17

H. Tchakoute-Tchuigoua, I. Soumaré

paper no. 5430. Labie, M., Méon, P.-G., Mersland, R., & Szafarz, A. (2015). Discrimination by microcredit oﬃcers: Theory and evidence on disability in Uganda. The Quarterly Review of Economics and Finance, 58, 44–55. Ledgerwood, J., Earne, J., & Nelson, C. (2013). The new microﬁnance handbook a ﬁnancial market system perspective. Washington D.C: The World Bank. Liberti, J. M., & Mian, A. R. (2009). Estimating the eﬀect of hierarchies on information use. Review of Financial Studies, 22(10), 4057–4090. McKim, A., & Hughart, M. (2005). Staﬀ incentive schemes in practice: Findings from a global survey of microﬁnance institutions. MicroFinance network. Available at: http://www.mfnetwork.org/working_groups/staﬀ_incentives.html. Mersland, R., & Strøm, R.Ø. (2008). Performance and trade-oﬀs in microﬁnance organisations—Does ownership matter? Journal of International Development: The Journal of the Development Studies Association, 20(5), 598–612. Mersland, R., & Strøm, R.Ø. (2009). Performance and governance in microﬁnance institutions. Journal of Banking & Finance, 33(4), 662–669. Pathan, S. (2009). Strong boards, CEO power and bank risk-taking. Journal of Banking & Finance, 33, 1340–1350. Patten, R. H., Rosengard, J. K., & Johnston, D. E., Jr. (2001). Microﬁnance success amidst macroeconomic failure: The experience of Bank Rakyat Indonesia during the East Asian crisis. World Development, 29(6), 1057–1069. Perilleux, A., Hudon, M., & Bloy, E. (2012). Surplus distribution in microﬁnance. Diﬀerences among cooperative, nonproﬁt, and shareholder forms of ownership. Nonproﬁt and Voluntary Sector Quarterly, 41(3), 386–404. Sagamba, M., Shchetinin, O., & Yusupov, N. (2013). Do microloan oﬃcers want to lend to the less advantaged? Evidence from a choice experiment. World Development, 42, 182–198. Sah, R., & Stiglitz, J. (1986). The architecture of economic systems: Hierarchies and polyarchies. American Economic Review, 76, 716–727. Servin, R., Lensink, R., & Van den Berg, M. (2012). Ownership and technical eﬃciency of microﬁnance institutions: Empirical evidence from Latin America. Journal of Banking & Finance, 36, 2136–2144. Stein, J. C. (2002). Information production and capital allocation: Decentralized versus hierarchical ﬁrms. Journal of Finance, 57(5), 1891–1921. Stiglitz, J. (1990). Peer monitoring and credit markets. World Bank Economic Review, 4, 351–366. Tchakoute Tchuigoua, H. (2010). Is there a diﬀerence in performance by the legal status of microﬁnance institutions? The Quarterly Review of Economics and Finance, 50(4), 436–442. Tchakoute Tchuigoua, H. (2012). Active risk management and loan contract terms: Evidence from rated microﬁnance institutions. The Quarterly Review of Economics and Finance, 52, 427–437. Tchakoute Tchuigoua, H. (2014). Performance of microﬁnance institutions: Do board activity and governance ratings matter? Finance (Review of the French Finance Association), 35(3), 7–52. Tchakoute Tchuigoua, H. (2016). Buﬀer capital in microﬁnance institutions. Journal of Business Research, 69(9), 3523–3537. Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data (2nd edition). Cambridge, London: MIT Press.

commercial microﬁnance organizations. Academy of Management Journal, 53, 1419–1440. Beck, T., Behr, P., & Guettler, A. (2013). Gender and banking: Are women better loan oﬃcers? Review of Finance, 17, 1279–1321. Beck, T., Demirgüç-Kunt, A., & Martinez-Piera, M. S. (2011). Bank ﬁnancing for SMEs: Evidence across countries and bank ownership types. Journal of Financial Services Research, 39, 35–54. Bellucci, A., Borisov, A., & Zazzaro, A. (2010). Does gender matter in bank-ﬁrm relationships? Evidence from small business lending. Journal of Banking & Finance, 34, 2968–2984. Berger, A. N., & Black, L. K. (2011). Bank size, lending technologies, and small business ﬁnance. Journal of Banking & Finance, 35, 724–735. Berger, A. N., Cowan, A. M., & Frame, W. S. (2011). The surprising use of credit scoring in small business lending by community banks and the attendant eﬀects on credit availability, risk, and proﬁtability. Journal of Financial Services Research, 39, 1–17. Berger, A. N., Klapper, L. F., Martinez-Piera, M. S., & Zaidi, R. (2008). Bank ownership type and banking relationships. Journal of Financial Intermediation, 17, 37–62. Berger, A. N., Miller, N. H., Petersen, M. A., Rajan, R. G., & Stein, J. C. (2005). Does function follow organizational form? Evidence from the lending practices of large and small banks. Journal of Financial Economics, 76, 237–269. Berger, A. N., & Udell, G. F. (2002). Small business credit availability and relationship lending: The importance of bank organizational structure. Economic Journal, 112, 32–53. Besley, T., & Ghatak, M. (2005). Competition and incentives with motivated agents. American Economic Review, 95(3), 616–636. Brickley, J. A., Linck, J. S., & Smith, C. W. (2003). Boundaries of the ﬁrm: Evidence from the banking industry. Journal of Financial Economics, 70, 351–383. Brown, M., Guin, B., & Kirschenmann, K. (2016). Microﬁnance banks and ﬁnancial inclusion. Review of Finance, 20(3), 907–946. Bruett, T. (2005). Measuring performance of microﬁnance institutions: A framework for reporting analysis, and monitoring. USAID: SEEP Network. Christen, R. P., Lauer, K., Lyman, T., & Rosenberg, R. (2012). A guide to regulation and supervision of microﬁnance: Consensus guidelines. CGAP. Conning, J. (1999). Outreach, sustainability and leverage in monitored and peer monitored lending. Journal of Development Economics, 60(1), 51–77. Cozarenco, A., Hudon, M., & Szafarz, A. (2016). What type of microﬁnance institutions supply savings products? Economics Letters, 140, 57–59. CSFI (2014). Microﬁnance banana skins. 2014. Cull, R., Demirgüç-Kunt, A., & Morduch, J. (2007). Financial performance and outreach: A global performance of leading microbanks. Economic Journal, 117, F107–F133. D'Espallier, B., Goedecke, J., Hudon, M., & Mersland, R. (2017). From NGOs to banks: Does institutional transformation Alter the business model of microﬁnance institutions? World Development, 89, 19–33. D'Espallier, B., Hudon, M., & Szafarz, A. (2013). Unsubsidized microﬁnance institutions. Economics Letters, 120, 174–176. Dixon, R., Ritchie, J., & Siwale, J. (2007). Loan oﬃcers and loan delinquency in microﬁnance: A Zambian case. Accounting Forum, 31, 47–71. Galema, R., Lensink, R., & Mersland, R. (2012). Do powerful CEOs determine microﬁnance performance? Journal of Management Studies, 49(4), 718–742. Guha, B., & Chowdhury, P. R. (2013). Microﬁnance competition: Motivated micro-lenders, double-dipping and default. Journal of Development Economics, 105, 86–102. Hartarska, V. (2005). Governance and performance of microﬁnance institutions in Central Eastern Europe and the Newly Independent States. World Development, 33(10), 1627–1643. Hartarska, V., & Mersland, R. (2012). Which governance mechanisms promote eﬃciency in reaching poor clients? Evidence from rated microﬁnance institutions. European Financial Management, 18(2), 218–239. Hartarska, V., & Nadolnyak, D. (2008). Does rating help microﬁnance institutions raise funds? Cross-country evidence. International Review of Economics and Finance, 17(4), 558–571. Heckman, J. (1979). Sample selection bias as a speciﬁcation error. Econometrica, 47, 153–161. Hertzberg, A., Liberti, J. M., & Paravisini, D. (2010). Information and incentives inside the ﬁrm: Evidence from loan oﬃcer rotation. Journal of Finance, 65(3), 795–828. Hudon, M., & Périlleux, A. (2014). Surplus distribution and characteristics of social enterprises: Evidence from microﬁnance. The Quarterly Review of Economics and Finance, 54(2), 147–157. Jansson, T., Rosales, R., & Westley, G. (2004). Principles and practices for regulating and supervising microﬁnance. Washington, DC: Inter-American Development Bank. Jeon, D. S., & Menicucci, D. (2011). When is the optimal lending contract in microﬁnance state non-contingent? European Economic Review, 55(5), 720–731. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2010). The worldwide governance indicators: A summary of methodology, data and analytical issues. World bank policy research working

Hubert Tchakoute-Tchuigoua is Professor of corporate ﬁnance at KEDGE Business School. He holds a degree in economics and ﬁnance (Catholic University of Central Africa), a Ph.D. in management sciences (Bordeaux University), and an “Habilitation à Diriger des Recherches” (qualiﬁcation to supervise doctoral dissertation) from Sorbonne Graduate Business School, University Paris 1-Panthéon-Sorbonne. His research activities are in the ﬁeld of development ﬁnance with a speciﬁc focus on the oﬀer side, that is, on very small banking organizations in developing and emerging countries also called microﬁnance institutions. He is the author or co-author of more than twenty articles published in national and international high quality journals. His current publications cover four main areas namely, ﬁnancing policy, ratings, ﬁnancial information quality, and loan contract. Issouf Soumaré is Professor of Finance and the Director of International Relations of the Faculty of Business Administration at Université Laval in Canada. He is also the Director of the Laboratory for Financial Engineering of Université Laval. His research and teaching interests include international ﬁnance, risk management, ﬁnancial engineering and numerical methods in ﬁnance. His theoretical and applied ﬁnancial economic works have been published in leading international economics and ﬁnance journals. He worked for the African Development Bank (ADB) at the former Risk Management Unit from 1996 to 1998. Prof. SOUMARÉ holds a PhD in Business Administration in Finance from the University of British Columbia (UBC, Canada) and an MSc in Financial Engineering from Université Laval (Canada).

17

The effect of loan approval decentralization on microfinance institutions' outreach and loan portfolio quality

The effect of loan approval decentralization on microfinance institutions' outreach and loan portfolio quality

Recommend Documents