- Email: [email protected]
Which types of microfinance institutions decentralize the loan approval process?
Accepted Manuscript Title: Which types of microfinance institutions decentralize the loan approval process? Author: Hubert Tchakoute Tchuigoua PII: DOI: Reference:
S1062-9769(17)30210-7 http://dx.doi.org/doi:10.1016/j.qref.2017.07.002 QUAECO 1051
To appear in:
The
Received date: Revised date: Accepted date:
16-7-2016 2-6-2017 3-7-2017
Quarterly
Review
of
Economics
and
Finance
Please cite this article as: & Tchakoute Tchuigoua, Hubert., Which types of microfinance institutions decentralize the loan approval process?.Quarterly Review of Economics and Finance http://dx.doi.org/10.1016/j.qref.2017.07.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Which types of microfinance institutions decentralize the loan approval process? Hubert TCHAKOUTE TCHUIGOUA KEDGE Business School Corresponding author: Hubert TCHAKOUTE TCHUIGOUA, KEDGE Business School, Department of Accounting, Auditing and Control, 680 cours de la Libération, 33405 Talence CEDEX, e-mail: [email protected] Highlights
We identify microfinance institutions characteristics that affect the choice of a decentralized loan approval process and link loan officer authority to MFIs outreach.
MFIs size, cooperative status, and solidarity lending affect the likelihood of choosing a decentralized loan approval process.
Allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers).
Allocating the decision-making authority does not affect the size of the loan portfolio and does not deteriorate MFIs’ loan portfolio quality.
Abstract: Which characteristics of microfinance institutions affect the choice of a decentralized loan approval process? That is the main question this article attempts to answer. A second concern is whether the choice of allocating the loan approval decision to the loan officer enables a microfinance institution to expand its number of loans and improve its loan portfolio quality after controlling for the endogeneity of the choice of decentralizing the loan approval. To achieve this goal, we study an independently pooled cross-section sample of 362 assessment reports for 267 MFIs from 2001 to 2012 across 67 countries. Results suggest that size, cooperative status, and solidarity lending are MFI-level variables that affect the likelihood of choosing a decentralized loan approval process. Allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers) but does not affect the size of the loan portfolio and does not deteriorate their loan portfolio quality. Keywords: decentralization, loan officer, loan approval, legal status, microfinance 1
JEL classification: G21, G32, G39 1. Introduction Microfinance is the provision of financial services to low-income people and entrepreneurs who are excluded from the conventional banking system. Microfinance institutions (MFIs) are hybrid or doublebottom-line organizations because they combine banking and development logics in running their businesses (Battilana & Dorado, 2010; D’espallier et al., 2013). In developing and emerging economies, these double-bottom-line institutions are more likely to grant loans to poor people and small businesses excluded from conventional financial services through different kinds of lending technologies. Lending is a core microfinance activity common to all MFIs, regardless of their legal status or commercial orientation. However, microcredit markets are imperfect, and MFIs face information problems in screening and monitoring borrowers’ behaviors. In credit markets, lending organizations may overcome information asymmetries by producing information about the borrower and using it in making credit decisions (Diamond, 1984). Close ties between MFIs and their clients are one of the main features of microfinance (Stiglitz, 1990). The existing microfinance literature documents that strong relationships between the MFIs and borrowers help to overcome existing information asymmetries and are beneficial for borrowers (Behr et al., 2011) and for the MFIs because they decrease loan defaults (Dorfleitner et al., 2017). Usually, loan officers live in the same local community as their borrowers and maintain a direct and personal contact with them. Therefore, 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). Daily interactions and personal relations between loan officers and local borrowers provide easy access to soft information. Loan officers are therefore better able to gather soft information and combine it with hard information when deciding whether to approve loans or not. Hence, one can state that loan officers play an important role in selecting potential borrowers in microcredit markets (Agier & Szafarz, 2013; Dixon et al., 2007). The delegation of authority to loan officers
2
may be facilitated by the proximity between MFIs and their clients and result in merging information production (expertise) and loan allocation functions. We argue that the proximity between MFIs and borrowers enables the former to delegate the decision-making authority to soft information holders. In most MFIs, the decision-making authority is allocated to the loan officer or is decentralized at the branch level (Armendáriz de Aghion & Morduch, 2010; Basel Committee on Banking Supervision (BCBS), 2010; Dixon et al., 2007; Tchakoute Tchuigoua, 2012). However, the microfinance literature so far has not focused on the types of MFIs that allocate authority to loan officers or at the branch level. To date, the empirical microfinance literature has examined issues related to the delegation of loan approvals to loan officers, for example, subjective preferences (Agier & Szafarz, 2013; Labie et al., 2015; Sagamba et al., 2013). Tchakoute Tchuigoua (2012) has investigated whether choosing a decentralized loan approval process benefits potential borrowers in terms of interest rates charged, loan size, and/or number of loans granted. To our knowledge, studies examining decentralization as a way to tackle informational problems in MFIs are scarce. The purpose of the present study is twofold: First, we identify the characteristics of MFIs that decentralize the loan approval decision. Second, we investigate whether the choice of allocating the loan approval decision to the loan officer enables microfinance institutions to increase the number of loans and improve their loan portfolio quality after controlling for the endogeneity of the choice of decentralizing the loan approval. Attempting to fill this empirical gap, we borrow from the existing literature in the banking sector, which identifies factors that are likely to drive the choice of an organizational architecture. The bank size and the bank ownership type are firm-level variables that may influence the choice of decentralizing the loan approval process (Berger & Udell, 2002). Except for Mersland (2009), who provides an overview of costs associated with ownership type in MFIs, the academic literature on microfinance addresses the question of the form of ownership, primarily in terms of governance, and pays little attention to the impact of ownership type on the decision-making process within MFIs. These studies have primarily focused on the relationship between ownership form and efficiency (Gutiérrez Nieto et al., 2007; Servin et al., 2012),
3
surplus distribution process among stakeholders (Hudon & Périlleux, 2014; Périlleux et al., 2012), accounting performance (Mersland & Strøm, 2008; Tchakoute Tchuigoua, 2010), and social performance (Roberts, 2013). We thus contribute to the literature by examining the relationship between MFI ownership and loan approval process in MFIs. In addition, the banking literature suggests that large banks use a hierarchical decision model, the allocation decision being made at their headquarters. Large banks are also headquartered at a substantial distance from potential relationship customers, which makes it difficult to gather and communicate soft information (Berger & Udell, 2002). In our article, we consider that large MFIs are complex in that they diversify by locating branches in areas where there are opportunities to expand their credit portfolio. The resulting proximity enables diversified MFIs to gather and use soft information. Similar to Tchakoute Tchuigoua (2012), we assume that the allocation of the loan approval decision to the loan officer can help mitigate informational problems in micro-credit contracts and thus limit their negative impact on MFIs’ loan portfolio quality, also enabling MFIs to expand their outreach and to improve the availability of credit. However, results obtained by Tchakoute Tchuigoua (2012) may be driven by MFIs’ incentives to allocate the credit decision authority to the loan officer. This previous study seems to neglect the endogeneity of the choice of allowing the decision-making authority to the loan officer, a factor that we take into account in this study. Our study focuses on an independently pooled cross-section sample of 267 MFIs rated from 2001 to 2012. Results indicate that size, cooperative status, and solidarity lending are MFI-level variables that affect the likelihood that an MFI will choose a decentralized loan approval process. Allocating the decisionmaking authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers) but does not affect the size of the loan portfolio and does not deteriorate the loan portfolio quality. The remainder of the article is organized as follows. Part 2 provides an overview of prior literature. Part 3 explains the research design. Further results and robustness tests are detailed in Part 4. Part 5 concludes with an acknowledgment of research limitations and suggestions for possible avenues of future work.
4
2. Prior literature Incentive-based theories (Aghion & Tirole, 1997; Stein, 2002) and those based on costs of acquiring and communicating information (Sah & Stiglitz, 1986) explain why firms and banking organizations transfer their decision-making authority to agents (loan officers). Sah and Stiglitz (1986) develop a model based on the costs of acquiring and communicating information. They argue that the way in which the decision-making process is organized within firms (hierarchy and architecture of the organization), whether financial or non-financial, may explain flawed decisions. Because of 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. We can expect that communicating soft information that is qualitative and unmeasurable will be more costly. In this case, in order to prevent contamination of information, thereby avoiding the risk of selecting bad borrowers, the organization will choose a decentralized decision-making structure. Considering the existence of asymmetric information between the principal, who has the formal authority, and the agent, who is the holder of the real authority, Aghion and Tirole (1997) develop a general model on how the decision authority gets allocated in organizations. In the information production function, the agent has an advantage over the principal given that the former has developed an expertise and the latter has not. Transferring decision-making authority to the agent will reward his or her expertise. Moreover, delegation provides incentives to the agent to be more involved in the collection and production of high-quality information that may facilitate better project screening. Hence, Aghion and Tirole (1997) recommend transferring the formal authority to the agent, although this delegation implies a loss of control for the principal. Stein (2002) developed a model based on agent incentives and soft information to help decide whether to extend credit or not. The model naturally applies to the banking industry, where information is critical to the activity of lending (Berger et al., 2005). According to this model, merging the allocation and 5
expertise functions ensures that the agent (loan officer) who develops expertise in producing soft information will use the gathered soft information to make the appropriate credit decision. Transferring the decision-making authority to agents who produce soft information is therefore a way to recognize and reward their expertise in this field. Decentralization is an organizational design that is best-suited using soft information. Allocating the decisional authority, that is, merging the resource allocation and information production functions to an agent (loan officer in the case of lending organizations) provides him or her with incentives to produce and use soft information when approving or denying loans. Hence, this transfer of authority may be a way for lending organizations to reduce information asymmetries in lending decisions. One empirical implication of Stein’s model (2002) is that large, hierarchical firms are at a comparative disadvantage when information about borrowers is soft. There is, however, little evidence on the characteristics of MFIs that may affect their probability of engaging in a decentralized loan approval process. The banking literature provides some arguments justifying why a bank should decentralize lending operations. Beck et al. (2011) examined whether the ownership type is associated with the likelihood of choosing a decentralized organizational architecture and found that privately owned banks are less likely to decentralize the loan approval process. Berger et al. (2008) addressed the links among ownership type, banking relationship, and multiple relationships in an emerging country such as India. They showed that privately owned banks tend to establish relationships with transparent firms, that is, large and listed firms that are more likely to produce hard information. These studies distinguish between private ownership and state ownership and, among privately owned banks, between foreign-owned banks and domestic-owned banks. Microfinance service providers include nongovernmental organizations (microfinance NGOs), financial cooperatives, micro-banks, and non-bank financial institutions (NBFIs), which are the main microfinance ownership types (Galema et al., 2012; Tchakoute Tchuigoua, 2010, 2015). Financial cooperatives are member-owned institutions that provide financial services (savings and loans) to their
6
members. NGOs are pro-poor entities with no owners. Their clients are primarily unbanked clients. Financial cooperatives and NGOs are usually grouped into the category of nonprofit MFIs. They offer loans to borrowers whose creditworthiness (assessed through conventional methods) is difficult to ascertain because of the poor quality of assets held and a lack of sufficiently valuable collateral. One may therefore expect that this category of MFIs will use collateral substitutes and primarily select their borrowers on the basis of soft information and soft lending methodologies. Micro-banks and NBFIs are privately owned MFIs that target unserved or underserved individuals and micro and small businesses that may pledge collateral and provide financial statements that help the loan officer to produce hard information. Furthermore, as shown in some empirical studies, there may be a trade-off between outreach and MFIs’ accounting performances (Cull et al., 2007, 2009) and efficiency (Hermes et al., 2011), in which financial sustainability and efficiency goals conflict with the outreach goal. The transformation of a nonprofit MFI (NGO) into a formalized, regulated financial institution (privately owned MFI) tends to accentuate such trade-offs and may lead to mission drift.1 Transformed MFIs therefore may tend to move away from their social mission and contract with creditworthy borrowers who are prepared to pledge collateral. Such MFIs may shift to hard lending practices when deciding whether to approve or deny a loan. Hence, one may expect privately owned MFIs (for-profit MFIs) to centralize the loan approval process given that it primarily depends on hard information about the borrower. The use of soft information seems to be a remedy against the incompleteness of contract in credit markets. Small size appears to be a feature of the banks that extensively use soft information and engage in relationship lending. Small banks are in a better position than large ones to collect and act on soft information and are more likely to lend to informationally opaque borrowers (Berger et al., 2005). Results of empirical studies in the banking sector suggest that smaller MFIs may favor a decentralized decisionmaking process. However, large MFIs are complex organizations that are geographically diversified
According to Armendáriz de Aghion and Szafarz (2011), “mission drift relates to a phenomenon whereby an MFI increases its average loan size by reaching out to wealthier clients. Mission drift in microfinance arises when an MFI finds it profitable to reach out to unbanked wealthier individuals while at the same time crowding out poor clients.” 1
7
because of market penetration strategies. Creating branches and sub-branches facilitates geographical proximity and close relationships between MFIs and their clients and enables loan officers to produce soft information, which is then used to select borrowers. This suggests that large MFIs, that is, those that are spatially diversified, tend to decentralize their credit decisions at the branch level.
3. Data and results 3.1. Sample MFI-level variables primarily come from assessment reports produced by Planet Rating, one of the agencies that specializes in the rating of microfinance institutions. Using rating reports, instead of the MIX database, helps gather information on the loan approval process in MFIs. Besides, as argued by Galema et al. (2012) and Hudon and Traca (2011), data collected and produced by rating agencies such as Planet Rating tend to be more reliable and representative of the microfinance industry than selfreported data, such as that obtained from the MIX database. From 2003 to 2015, about 435 rating reports were available on the Planet Rating website. Some rating reports were not exploited because they were confidential, specifically, those published between 2013 and 2015. Data before 2003 were gathered from Planet Rating reports available on the Rating Fund website (www.ratingfund2.org). We built a sample of 362 assessment reports for 267 MFIs from 2001 to 2012 across 67 countries. The sample included MFIs from the following four main regions: Africa and Middle East (138 rating reports); South Asia, East Asia, and the Pacific (28 rating reports); Eastern Europe and Central Asia (54 rating reports); and Latin America and the Caribbean (142 rating reports). The country-level variables, that is, countries’ rates of economic growth and corruption indexes, respectively come from the World Bank’s World Development Indicators (WDIs), available on the World Bank website (http://data.worldbank.org/data-catalog/world-development-indicators), and from the Worldwide Governance Indicators (WGIs) dataset (Kaufmann et al., 2011), available on the same website (http://info.worldbank.org/governance/wgi/index.asp).
8
3.2. Variables We base the construction of the outcome variable (credit approval decision) on incentives-based theories (Aghion & Tirole, 1997; Stein, 2002), which argue that allocating the decisional authority, that is, merging the resource allocation and information production functions to an agent (the loan officer in the case of lending organizations), provides him or her with incentives to produce and use soft information when approving or denying loans. This association reduces the risk of providing and using irrelevant information in the decision-making process. We measure the credit approval decision by a dummy, taking the value 1 if the loan officer has discretion to make the final credit approval decision, and 0 otherwise. The banking literature documents that the choice of a specific organization structure, that is, between a centralized versus a decentralized loan approval process, depends on firm-specific characteristics and countries’ legal institutions (Berger & Udell, 2002; Stein, 2002). We measure the ownership type variable by three dummy variables, namely, privately owned MFIs including micro-banks and NBFIs, member-owned MFIs (cooperatives), and not-owned entities (NGOs), in order to identify among nonprofit MFIs those that tend to decentralize the loan approval decision. Only two of these three modalities are taken into account in the regressions. The size of MFIs is measured by the natural logarithm of the total branches (SIZE). Following previous studies that link the microfinance entity’s choice of lending methodology to its social efficiency and loan portfolio quality (Cull et al., 2007; Mersland & Strøm, 2009; Tchakoute Tchuigoua, 2012), we account for the type of lending methodology applied by the MFIs. Indeed, MFIs make loans either under individual loan contracts or joint liability contracts, including village banking and solidarity/group lending. Unlike the previously mentioned studies that assess each lending methodology by dummies, we measure the type of loan by the total amount of loans approved under each type of lending methodology as a percentage of the overall gross loan portfolio. The loan officer size is measured by the percentage of loan officers.
9
Stiglitz (1990) and Berger and Udell (2002, 2006) argue that environmental factors, such as institutional environments in which MFIs operate, are likely to drive the choice between hierarchy and decentralization. Weaker or inadequate institutions are detrimental to contract enforcement and tend to exacerbate information asymmetries. The quality of the institutional environment raises the question of the reliability of collateral that borrowers own and can pledge and accounting information contained in the financial statements provided by microfinance clients such as very small enterprises (VSEs) and small and medium-size enterprises (SMEs). The low reliability of hard information in countries with weaker institutions suggests that MFIs should favor decentralization. We thus control for country governance quality measures. To this end, we assess the governance quality by estimating the country’s regulatory quality. Finally, we also control for the country’s rate of economic growth. Following Beck et al. (2011), we also control for the size of the country’s banking sector. Table 1 summarizes firm-level and country-level variables used in the study. (Insert Table 1 here) 3.3. Model Although the data structure looks like that of a panel data (267 MFIs and 12 years), it seems difficult to apply the principles of panel data analysis. No MFI renewed its ratings 12 times over the period studied. Four MFIs renewed their ratings four times, six thrice, nine twice, and five once. The data structure is thus an independently pooled cross-section. To identify which MFI-level variables may explain the choice of a decentralized loan approval process, we thus apply a pooled probit regression with control for country-level variables, region year-fixed effects, and heteroscedasticity. The estimated model is as follows: Probit (decentralization of the loan approval decision) = α0 + αi Xi + βi CVj + γi + δt + (1)
where i indexes the MFI, and j the country. Xi is the vector of MFI-specific variables: size (number of branches), joint liability contract, ownership type, percentage of loan officers. CVj is the vector of
10
country-level variables including the country’s regulatory quality and rate of economic growth. γi captures the region-fixed effects, and δt is the year-fixed effects. To analyze the effect of decentralization on the credit portfolio’s quality, the breadth of outreach and the credit availability, we estimate the following model (equation 2). Insofar as the relationship between decentralization and outreach may be driven by MFIs’ incentives to allocate the credit decision authority to the loan officer, there is an endogeneity bias, and more specifically a selection bias, that we correct by implementing the Heckman's two-step procedure.
Yi = α0 + αi Xi + βi CVj + λ IMRi + γi + δt +
(2)
where i indexes the MFI and j the country. Yi is the vector of dependent variables (breadth of outreach, loan portfolio quality, and size of the loan portfolio); Xi is the vector of MFI-specific variables: decentralization of the loan approval, age, square of the age, joint liability contract, size (number of branches). CVj is the size of the country’s banking sector; IMRi is the inverse Mills ratio, which we derived from equation (1) and introduced into equation (2) in order to control for selection effects (endogeneity of the choice of decentralization); γi captures the region-fixed effects, and δt is the year-fixed effects.
4. Results Summary statistics (Table 2) show that in 52% of the cases the loan approval process is allocated to the loan officer. On average, MFI networks are made up of 17 branches, and outstanding individual loans account for 61% of loans granted by MFIs. NGOs represent 45% of the sample, cooperatives 17%, and shareholder-based MFIs 38%. (Insert Table 2 here) Prior to estimations, we assessed the presence of multicollinearity among explanatory variables (Table 3). Our diagnostic reveals that, except for the correlation between individual lending and joint
11
liability contracts (r=–0.99; p=0.01), and between NGOs and privately owned MFIs (r=0.71; p=0.01), the intensity of the relationship between other explanatory variables is weak. (Insert Table 3 here) Multivariate results are reported in Tables 4 and 5. In all models, standard errors are heteroscedasticity robust. In each table, we report the results the baseline model and then check for robustness by replacing the number of branches by the natural logarithm of total assets (USD). In Table 4 we present the results of the probit estimation. The results are consistent with our expectations. Unlike Berger and Udell (2002) and Stein (2002), who found that large banks preferred a hierarchical and standardized decision-making model, we argue that large or complex MFIs are likely to decentralize the loan approval process. Indeed, the relationship between the size (number of branches) and the loan approval process is positive, indicating that large MFIs, those that are spatially diversified, tend to allocate authority to loan officers. MFIs that have been developed through the creation of local branches benefit from proximity to their clients and ties between loan officers and borrowers, allocating the decision to grant credit to loan officers in order to mitigate information problems in lending. (Insert Tables 4 and 5 here) The type of loan products provided by MFIs is also associated with the decentralization of the credit decision. Indeed, MFIs providing loans through joint liability are likely to allocate the decisional authority to loan officers. The cooperative status is associated with a decentralized loan approval process, suggesting that cooperative MFIs are more likely to decentralize the loan approval process. This result may be explained by the fact that most cooperatives expand mainly within their local communities (Mersland, 2009). Finally, MFIs in countries with weaker institutions are more likely to favor a decentralized loan approval process. Results of the pooled OLS regression are reported in Table 5, with control for the endogeneity of the decentralization of the loan approval. We estimate the effect of decentralization on MFIs’ breadth of outreach, loan portfolio size, and loan portfolio quality. The relationship between the decentralization of
12
the loan approval and the breath of outreach is positive and significant, suggesting that the proximity between the loan officer and the clients (related to the transfer of the decision-making authority to the loan officer) enables the MFI to extend credit to a larger number of clients. However, decentralization does not necessarily lead to an increase in the size of loans or to a deterioration in the loan portfolio quality, because the coefficient of decentralization is not significant either in the loan portfolio size or in the loan portfolio quality models. We thus conclude that allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (number of active borrowers) without affecting the size of the loan portfolio or deteriorating the loan portfolio quality. To check for robustness, we replaced the number of branches by the natural logarithm of the book value of assets (USD) and found them consistent (Tables 4 and 5). As an additional robustness check (the probit model) we distinguished whether the loan approval process was centralized or not. The credit approval decision is measured by a dummy taking the value 0 if the loan approval decision is centralized, that is, when all loan applications are first completed by loans officers (LOs), reviewed by the branch manager, and then sent to the head officer at headquarters for credit committee, and 1 otherwise, that is, when the loan officer has discretion to make the final credit approval decision, which is decentralized at the branch level in the credit committee comprising the branch manager and loan officers. By participating in the credit committee at the branch level, the loan officer may influence the credit decision. Summary statistics show that in 81% of cases, the loan officer directly approves credit or influences the loan approval process by participating in the credit committee at the branch level. The results of the probit regression, reported in Table 6, are consistent with those reported in Table 4. From the results of the probit model presented in Table 6, we generated the inverse Mills ratio and re-estimated equation (2). The results obtained are in accordance with those presented in Table 7. (Insert Tables 6 and 7 here)
5. Conclusion
13
This article addresses the question of whether the characteristics of MFIs affect the choice of a decentralized loan approval process. The secondary issue discussed is whether allocating the loan approval decision to loan officers enables microfinance institutions to increase the number of loans and improve their loan portfolio quality after controlling for the endogeneity of the choice of decentralizing the loan approval. To achieve this goal, we exploited a pooled sample of 362 assessment reports for 267 MFIs from 2001 to 2012 across 67 countries. Findings reveal that size, cooperative status, and solidarity lending are MFI-level variables that affect the likelihood of choosing a decentralized loan approval process. Allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers) but does not affect the size of the loan portfolio and does not deteriorate the loan portfolio quality. However, Stein (2002) and Berger and Udell (2002) argue that a decentralized credit decision process leads to agency conflicts between the loan officer and the organization. Loan officers may be tempted to follow their subjective preferences (Agier & Szafarz, 2013), depart from approved policies, or act opportunistically. An interesting direction for future research therefore is to investigate existing mechanisms put in place by MFIs to reduce agency conflicts inherent in a decentralized loan approval process and to limit fraud and opportunism, ensuring that loan officers make loan decisions that help improve MFIs’ efficiency.
14
References Aghion, P., & Tirole, J. (1997). Formal and real authority in organizations. Journal of Political Economy, 105, 1–29. Agier, I., & Szafarz, A. (2013). Subjectivity in credit allocation to micro-entrepreneurs: Evidence from Brazil. Small Business Economics, 41, 263–275. Armendáriz de Aghion, B., & Morduch, J. (2010). The economics of microfinance. Cambridge, MA: MIT Press. Armendáriz, B., & Szafarz, A. (2011). On mission drift in microfinance institutions. In the handbook of microfinance, B. Armendariz and M. Labie Ed. 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 commercial microfinance organizations. Academy of Management Journal, 53, 1419–1440. 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. Behr, P., Entzian, A., & Güttler, A. (2011). How do lending relationships affect access to credit and loan conditions in microlending? Journal of Banking and Finance, 25(8), 2169–2178. 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. (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. Berger, A. N., & Udell, G. F. (2006). A more complete conceptual framework for SME finance. Journal of Banking & Finance, 30(11), 2945-2966. 15
Cull, R., Demirgüç-Kunt, A., & Morduch, J. (2007). Financial performance and outreach: A global performance of leading microbanks. Economic Journal, 117, 107–133. Cull, R., Demirgüç-Kunt, A., & Morduch, J. (2009). Microfinance meets the market. Journal of Economic Perspectives, 23, 167–192. D'espallier, B., Hudon, M., & Szafarz, A. (2013). Unsubsidized microfinance institutions. Economic Letters, 120, 174–176. Diamond, D. W. (1984). Financial intermediation and delegated monitoring. Review of Economics Studies, 51, 393–414. Dixon, R., Ritchie, J., & Siwale, J. (2007). Loan officers and loan delinquency in microfinance: A Zambian case. Accounting Forum, 31, 47–71. Dorfleitner, G., Just-Marxa, S., & Priberny, C. (2017). What drives the repayment of agricultural micro loans? Evidence from Nicaragua. Quarterly Review of Economics and Finance, 63, 89–100. Galema, R., Lensink, R., & Mersland, R. (2012). Do powerful CEOs determine microfinance performance? Journal of Management Studies, 49, 718–742. Gutiérrez Nieto, B., Serra Cinca, C., & Mar Molinero, C. (2007). Microfinance institutions and efficiency. Omega, 35, 131–142. Hermes, N., Lensink, R., & Meesters, A. (2011). Outreach and efficiency of microfinance institutions. World Development, 39, 938–948. Hudon, M., & Périlleux, A. (2014). Surplus distribution and characteristics of social enterprises: Evidence from microfinance. Quarterly Review of Economics and Finance, 54, 147–157. Hudon, M., & Traca, D. (2011). On the efficiency effects of subsidies in microfinance: An empirical inquiry. World Development, 39, 966–973. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2011). The worldwide governance indicators: methodology and analytical issues. Hague Journal on the Rule of Law, 3(2), 220-246. Labie, M., Méon, P.-G., Mersland, R., & Szafarz, A., (2015). Discrimination by microcredit officers: Theory and evidence on disability in Uganda. Quarterly Review of Economics and Finance, 58, 44–55. 16
Mersland, R. (2009). The cost of ownership in microfinance organizations. World Development, 37, 469– 478. Mersland, R., & Strøm, R. Ø. (2009). Performance and trade-offs in microfinance organizations—Does ownership matter? Journal of International Development, 20, 598–612. Périlleux, 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. Roberts, P. W. (2013). The profit orientation of microfinance institutions and effective interest rates. World Development, 41, 120–131. 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 and 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? 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. Quarterly Review of Economics and Finance, 52, 427–437. Tchakoute Tchuigoua, H. (2015). Determinants of the governance quality of microfinance institutions. Quarterly Review of Economics and Finance, 58, 32–43.
17
Table 1: Description of variables Identity of variables Breadth of outreach Size of the loan portfolio
Number of active borrowers Gross loan portfolio
Loan portfolio quality Portfolio at risk at 30 days Loan approval process Size
Lending methodology or type of loan
Legal status Loan officers Maturity
Decentralization Number of branches Individual loan contracts Joint liability contracts NGO Cooperative Shareholderbased Percentage of loan officers Age Regulatory quality
Country’s variables
Country’s rate of economic growth
Size of the banking sector
Regions
LAC EECA AFRICA and MENA ASIA
Definition and Measurement The total number of active borrowers Gross loan portfolio/total assets Outstanding balance on arrears over 30 days + Total gross outstanding refinanced (restructured) portfolio)/Total gross portfolio Measurement of portfolio quality that shows the part of the portfolio affected by outstanding payments when there is a risk that they will not be repaid; the threshold is < 10% given that financial guarantees in microfinance are not always sufficient Dummy: 1 if the credit decision process is decentralized to the loan officer, 0 otherwise Natural logarithm of total branches Individual loans as a percentage of the outstanding loan portfolio Loan granted to a single borrower (group-lending loans + village banking loans) as a percentage of the outstanding loan portfolio Under group-lending methodology, loans are granted to individuals but the group (3–10 members) is jointly liable for credit. Group lending through village banking methodology is based on larger groups (over 10 people) Binary variable: 1 if the MFI is a nongovernmental organization, 0 otherwise Binary variable: 1 if the MFI is a cooperative, 0 otherwise Binary variable: 1 if the MFI is a shareholder MFI, 0 otherwise Number of loan officers/total number of staff members Number of years since inception The index reflects perceptions about the ability of the government to formulate and implement sound policies and regulations that help promote the private sector’s development. Estimate of governance (ranges from approximately −2.5 (weak) to 2.5 (strong) governance performance). Source: Kaufmann et al. (2011); World Bank. Annual growth rate of the GDP per capita Domestic credit provided by the banking sector as a percentage of the GDP. Domestic credit provided by the financial sector includes all credit to various sectors on a gross basis, with the exception of credit to the central government, which is net. The financial sector includes monetary authorities and deposit money banks, as well as other financial corporations where data are available (including corporations that do not accept transferable deposits but do incur such liabilities as time and savings deposits). Examples of other financial corporations are finance and leasing companies, money lenders, insurance corporations, pension funds, and foreign exchange companies. Dummy: 1 if the region is Latin America and the Caribbean, 0 otherwise Dummy: 1 if the region is Eastern Europe and Central Asia, 0 otherwise Dummy: 1 if the region is Africa and Middle East, 0 otherwise Dummy: 1 if the region is South Asia, 0 otherwise
18
Table 2: Descriptive statistics
Variables Breadth of outreach
Mean
Standard Deviation
Minimum
Maximum
Median
21,967.43
57,050.73
23.00
643,659.00
8,003.50
Size of the loan portfolio
0.74
0.17
0.05
1.03
0.76
Loan portfolio quality
0.05 0.52
0.07 0.50
0.00 0
0.54 1
0.03 1
25,900,000
87,300,000
43,275.60
1,010,000,000
5,219,776
Branches
17.36
34.61
0.00
410.00
8
Individual lending
0.61
0.41
0
1
0.84
Joint liability contracts
0.39
0.41
0
1
0.16
Shareholder-based
0.38
0.49
0
1
0
Cooperatives
0.17
0.37
0
1
0
NGOs
0.45
0.50
0
1
0
0.67
4.30
0.04
82.20
0.45
10.50
8.47
0.00
51.00
8.00
Decentralization Size
Percentage of loan officers Maturity Country’s rate of economic growth Regulatory quality
0.04
0.03
-0.07
0.14
0.03
−0.21
0.46
−1.21
1.50
−0.20
Size of the banking sector
0.42
0.31
−0.01
2.02
0.34
19
Table 3: Pearson’s correlation matrix explanatory variables Branches
Individual lending
Joint liability contracts
Privately owned
Cooperatives
NGOs
Percentage of loan officers
Regulatory quality
Branches
1.00
Individual lending
0.04
1.00
Joint liability contracts
−0.04
−0.99***
1.00
Privately owned
0.15***
0.14***
−0.14***
1.00
Cooperatives
−0.03
0.13**
−0.13**
−0.35***
1.00
−0.13**
−0.23***
0.23***
−0.71***
−0.41***
1.00
−0.01
−0.09
0.09
0.07
-0.03
−0.04
1.00
−0.07
0.02
−0.02
0.02
−0.06
0.03
−0.01
1.00
0.04
0.01
−0.01
0.12**
−0.04
−0.08
0.04
−0.11**
NGOs Percentage of loan officers Regulatory quality Country’s rate economic growth
of
Country’s rate of economic growth
1.00
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
20
Table 4: Multivariate results of the pooled probit with control for year-fixed effects and regions Baseline model Size is measured by the natural logarithm of the number of branches.
Robustness checks Size is measured by the natural logarithm of the number of the total book value of assets. Coefficients t-stat
Coefficients
t-stat
Constant
−0.31
−0.67
−1.56
−1.59
Branches
0.30***
4.39
0.12**
2.23
Joint liability contracts
0.39**
2.13
0.43**
2.37
Cooperatives
0.47**
2.21
0.44**
2.02
NGOs
−0.04
−0.23
−0.11
−0.61
Percentage of loan officers
−0.03***
−4.48
−0.04***
−4.98
Regulatory quality Country’s rate of economic growth Region dummies
−0.41**
−2.53
−0.39**
−2.40
0.03
0.11
0.09
0.04
yes
yes
Year-fixed effects
yes
yes
Pseudo R-squared
13.90%
11.32%
Chi 2
80.68***
66.28***
Log-likelihood value
−212.86
−219.23
Percent correctly predicted
66.39%
65.83%
N 357 * Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
357
21
Table 5: Results of the pooled OLS regression with control for year-fixed effects and regions. We analyze the effect of decentralization on MFI breadth of outreach, loan portfolio size, and loan portfolio quality.
Variable Constant Decentralization of the loan approval Age Age-squared
Baseline model Size is measured by the natural logarithm of the number of branches. Breadth of Size of the Loan portfolio outreach loan portfolio quality 6.76*** 0.59*** 0.02
Robustness check Size is measured by the natural logarithm of the number of the total book value of assets. Breadth of Size of the Loan portfolio outreach loan portfolio quality −0.45 0.44*** 0.12**
(9.59)
(4.94)
(0.93)
(−0.43)
(2.46)
(2.29)
0.43***
−0.01
−0.002
0.41***
−0.01
−0.00
(2.88)
(−0.69)
(−0.80)
(3.52)
(−0.67)
(−0.77)
0.05
0.01**
−0.002*
−0.00
0.01**
−0.00
(1.58)
(2.43)
(−1.69)
(−0.02)
(2.26)
(−1.43)
−0.00
−0.004**
0.001**
0.00
−0.001**
0.001*
(−1.22)
(−2.31)
(2.09)
(0.37)
(−2.26)
(1.92)
Joint liability contracts
0.97***
0.03
-0.01
1.00***
0.02
-0.01
(6.00)
(1.06)
(-1.46)
(6.28)
(0.83)
(-1.63)
Size
0.76***
0.03**
−0.01
0.64***
0.02**
−0.01***
(10.35)
(2.42)
(−1.37)
(10.49)
(2.28)
(−2.62)
0.34
0.01
−0.00
0.29*
0.01
−0.00
(1.52)
(0.33)
(−0.36)
(1.67)
(0.50)
(−0.15)
Size of the banking sector Mills ratio F r2 Adjusted_R² Number of observations
0.46
0.10**
0.01
−0.75***
0.04
0.02
(1.54)
(2.48)
(0.46)
(−3.08)
(1.28)
(0.90)
16.58*** 43.52%
2.071*** 14.96%
11.01*** 11.58%
65.62*** 59.03%
2.208*** 14.89%
4.299*** 13.26%
39.87%
9.45%
58.51%
56.36%
9.37%
7.64%
347
346
346
347
346
346
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
22
Additional robustness checks Table 6: Centralized versus decentralized multivariate results of the pooled probit with control for year-fixed effects and regions. The loan approval process is decentralized at the branch level: The loan officer has the final authority or participates in the credit committee at the branch level. He/ or she may thus influence the loan approval process. The credit approval decision is measured by a dummy taking the value 1 if the loan officer influences the final credit approval decision, and 0 otherwise, i.e. that is, if the loan approval decision is centralized. Coefficients
t-stat
Constant
–0.06
–0.10
Branches
0.30***
3.73
Joint liability contract
0.45**
2.07
Cooperatives
0.35
1.44
NGOs
–0.01
–0.06
Percentage of loan officers Regulatory quality Country’s rate of economic growth
0.02**
2.29
–0.54***
–2.83
–3.82
–1.24
Region dummies
yes
Year-fixed effects
yes
Pseudo R-squared
9.44%
Chi 2
36.08**
Log-likelihood value
–157.41
Percent correctly predicted
80.67%
N
357
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
23
Table 7: Results of the pooled OLS regression with control for year-fixed effects and regions. We analyze the effect of decentralization on MFIs’ breadth of outreach, loan portfolio size, and loan portfolio quality. The loan approval process is decentralized at the branch level: The loan officer has the final authority or participates in the credit committee at the branch level. He/ or she may thus influence the loan approval process. The credit approval decision is measured by a dummy taking the value 1 if the loan officer influences the final credit approval decision, and 0 otherwise, i.e. that is, if the loan approval decision is centralized.
Variable Constant Decentralization of the loan approval Age Age-squared Joint liability contracts Size (natural logarithm of the number of branches)
Decentralization of the loan approval Breadth of Size of the loan Loan portfolio outreach portfolio quality 7.50*** 0.64*** 0.03 (9.86)
(4.87)
(1.22)
0.42**
0.001
−0.01
(2.36)
(0.01)
(−1.25)
0.04
0.01**
−0.002*
(1.52)
(2.41)
(−1.68)
−0.002
−0.003**
0.004**
(−1.30
(−2.33)
(2.12)
0.82***
0.02
−0.01
(4.60)
(0.55)
(−1.25)
0.64***
0.02
−0.01
(7.74)
(1.32)
(−1.30)
0.45*
0.02
−0.00
(1.95)
(0.75)
(−0.49)
−0.47
0.06
0.02
(−0.80)
(0.68)
(0.57)
F
15.59***
1.924***
10.34***
r2
43.09%
13.7%
12.13%
Adjusted_R²
39.41%
8.11%
6.43%
347
346
346
Size of the banking sector Mills ratio
Number of observations
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
24
S1062-9769(17)30210-7 http://dx.doi.org/doi:10.1016/j.qref.2017.07.002 QUAECO 1051
To appear in:
The
Received date: Revised date: Accepted date:
16-7-2016 2-6-2017 3-7-2017
Quarterly
Review
of
Economics
and
Finance
Please cite this article as: & Tchakoute Tchuigoua, Hubert., Which types of microfinance institutions decentralize the loan approval process?.Quarterly Review of Economics and Finance http://dx.doi.org/10.1016/j.qref.2017.07.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Which types of microfinance institutions decentralize the loan approval process? Hubert TCHAKOUTE TCHUIGOUA KEDGE Business School Corresponding author: Hubert TCHAKOUTE TCHUIGOUA, KEDGE Business School, Department of Accounting, Auditing and Control, 680 cours de la Libération, 33405 Talence CEDEX, e-mail: [email protected] Highlights
We identify microfinance institutions characteristics that affect the choice of a decentralized loan approval process and link loan officer authority to MFIs outreach.
MFIs size, cooperative status, and solidarity lending affect the likelihood of choosing a decentralized loan approval process.
Allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers).
Allocating the decision-making authority does not affect the size of the loan portfolio and does not deteriorate MFIs’ loan portfolio quality.
Abstract: Which characteristics of microfinance institutions affect the choice of a decentralized loan approval process? That is the main question this article attempts to answer. A second concern is whether the choice of allocating the loan approval decision to the loan officer enables a microfinance institution to expand its number of loans and improve its loan portfolio quality after controlling for the endogeneity of the choice of decentralizing the loan approval. To achieve this goal, we study an independently pooled cross-section sample of 362 assessment reports for 267 MFIs from 2001 to 2012 across 67 countries. Results suggest that size, cooperative status, and solidarity lending are MFI-level variables that affect the likelihood of choosing a decentralized loan approval process. Allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers) but does not affect the size of the loan portfolio and does not deteriorate their loan portfolio quality. Keywords: decentralization, loan officer, loan approval, legal status, microfinance 1
JEL classification: G21, G32, G39 1. Introduction Microfinance is the provision of financial services to low-income people and entrepreneurs who are excluded from the conventional banking system. Microfinance institutions (MFIs) are hybrid or doublebottom-line organizations because they combine banking and development logics in running their businesses (Battilana & Dorado, 2010; D’espallier et al., 2013). In developing and emerging economies, these double-bottom-line institutions are more likely to grant loans to poor people and small businesses excluded from conventional financial services through different kinds of lending technologies. Lending is a core microfinance activity common to all MFIs, regardless of their legal status or commercial orientation. However, microcredit markets are imperfect, and MFIs face information problems in screening and monitoring borrowers’ behaviors. In credit markets, lending organizations may overcome information asymmetries by producing information about the borrower and using it in making credit decisions (Diamond, 1984). Close ties between MFIs and their clients are one of the main features of microfinance (Stiglitz, 1990). The existing microfinance literature documents that strong relationships between the MFIs and borrowers help to overcome existing information asymmetries and are beneficial for borrowers (Behr et al., 2011) and for the MFIs because they decrease loan defaults (Dorfleitner et al., 2017). Usually, loan officers live in the same local community as their borrowers and maintain a direct and personal contact with them. Therefore, 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). Daily interactions and personal relations between loan officers and local borrowers provide easy access to soft information. Loan officers are therefore better able to gather soft information and combine it with hard information when deciding whether to approve loans or not. Hence, one can state that loan officers play an important role in selecting potential borrowers in microcredit markets (Agier & Szafarz, 2013; Dixon et al., 2007). The delegation of authority to loan officers
2
may be facilitated by the proximity between MFIs and their clients and result in merging information production (expertise) and loan allocation functions. We argue that the proximity between MFIs and borrowers enables the former to delegate the decision-making authority to soft information holders. In most MFIs, the decision-making authority is allocated to the loan officer or is decentralized at the branch level (Armendáriz de Aghion & Morduch, 2010; Basel Committee on Banking Supervision (BCBS), 2010; Dixon et al., 2007; Tchakoute Tchuigoua, 2012). However, the microfinance literature so far has not focused on the types of MFIs that allocate authority to loan officers or at the branch level. To date, the empirical microfinance literature has examined issues related to the delegation of loan approvals to loan officers, for example, subjective preferences (Agier & Szafarz, 2013; Labie et al., 2015; Sagamba et al., 2013). Tchakoute Tchuigoua (2012) has investigated whether choosing a decentralized loan approval process benefits potential borrowers in terms of interest rates charged, loan size, and/or number of loans granted. To our knowledge, studies examining decentralization as a way to tackle informational problems in MFIs are scarce. The purpose of the present study is twofold: First, we identify the characteristics of MFIs that decentralize the loan approval decision. Second, we investigate whether the choice of allocating the loan approval decision to the loan officer enables microfinance institutions to increase the number of loans and improve their loan portfolio quality after controlling for the endogeneity of the choice of decentralizing the loan approval. Attempting to fill this empirical gap, we borrow from the existing literature in the banking sector, which identifies factors that are likely to drive the choice of an organizational architecture. The bank size and the bank ownership type are firm-level variables that may influence the choice of decentralizing the loan approval process (Berger & Udell, 2002). Except for Mersland (2009), who provides an overview of costs associated with ownership type in MFIs, the academic literature on microfinance addresses the question of the form of ownership, primarily in terms of governance, and pays little attention to the impact of ownership type on the decision-making process within MFIs. These studies have primarily focused on the relationship between ownership form and efficiency (Gutiérrez Nieto et al., 2007; Servin et al., 2012),
3
surplus distribution process among stakeholders (Hudon & Périlleux, 2014; Périlleux et al., 2012), accounting performance (Mersland & Strøm, 2008; Tchakoute Tchuigoua, 2010), and social performance (Roberts, 2013). We thus contribute to the literature by examining the relationship between MFI ownership and loan approval process in MFIs. In addition, the banking literature suggests that large banks use a hierarchical decision model, the allocation decision being made at their headquarters. Large banks are also headquartered at a substantial distance from potential relationship customers, which makes it difficult to gather and communicate soft information (Berger & Udell, 2002). In our article, we consider that large MFIs are complex in that they diversify by locating branches in areas where there are opportunities to expand their credit portfolio. The resulting proximity enables diversified MFIs to gather and use soft information. Similar to Tchakoute Tchuigoua (2012), we assume that the allocation of the loan approval decision to the loan officer can help mitigate informational problems in micro-credit contracts and thus limit their negative impact on MFIs’ loan portfolio quality, also enabling MFIs to expand their outreach and to improve the availability of credit. However, results obtained by Tchakoute Tchuigoua (2012) may be driven by MFIs’ incentives to allocate the credit decision authority to the loan officer. This previous study seems to neglect the endogeneity of the choice of allowing the decision-making authority to the loan officer, a factor that we take into account in this study. Our study focuses on an independently pooled cross-section sample of 267 MFIs rated from 2001 to 2012. Results indicate that size, cooperative status, and solidarity lending are MFI-level variables that affect the likelihood that an MFI will choose a decentralized loan approval process. Allocating the decisionmaking authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers) but does not affect the size of the loan portfolio and does not deteriorate the loan portfolio quality. The remainder of the article is organized as follows. Part 2 provides an overview of prior literature. Part 3 explains the research design. Further results and robustness tests are detailed in Part 4. Part 5 concludes with an acknowledgment of research limitations and suggestions for possible avenues of future work.
4
2. Prior literature Incentive-based theories (Aghion & Tirole, 1997; Stein, 2002) and those based on costs of acquiring and communicating information (Sah & Stiglitz, 1986) explain why firms and banking organizations transfer their decision-making authority to agents (loan officers). Sah and Stiglitz (1986) develop a model based on the costs of acquiring and communicating information. They argue that the way in which the decision-making process is organized within firms (hierarchy and architecture of the organization), whether financial or non-financial, may explain flawed decisions. Because of 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. We can expect that communicating soft information that is qualitative and unmeasurable will be more costly. In this case, in order to prevent contamination of information, thereby avoiding the risk of selecting bad borrowers, the organization will choose a decentralized decision-making structure. Considering the existence of asymmetric information between the principal, who has the formal authority, and the agent, who is the holder of the real authority, Aghion and Tirole (1997) develop a general model on how the decision authority gets allocated in organizations. In the information production function, the agent has an advantage over the principal given that the former has developed an expertise and the latter has not. Transferring decision-making authority to the agent will reward his or her expertise. Moreover, delegation provides incentives to the agent to be more involved in the collection and production of high-quality information that may facilitate better project screening. Hence, Aghion and Tirole (1997) recommend transferring the formal authority to the agent, although this delegation implies a loss of control for the principal. Stein (2002) developed a model based on agent incentives and soft information to help decide whether to extend credit or not. The model naturally applies to the banking industry, where information is critical to the activity of lending (Berger et al., 2005). According to this model, merging the allocation and 5
expertise functions ensures that the agent (loan officer) who develops expertise in producing soft information will use the gathered soft information to make the appropriate credit decision. Transferring the decision-making authority to agents who produce soft information is therefore a way to recognize and reward their expertise in this field. Decentralization is an organizational design that is best-suited using soft information. Allocating the decisional authority, that is, merging the resource allocation and information production functions to an agent (loan officer in the case of lending organizations) provides him or her with incentives to produce and use soft information when approving or denying loans. Hence, this transfer of authority may be a way for lending organizations to reduce information asymmetries in lending decisions. One empirical implication of Stein’s model (2002) is that large, hierarchical firms are at a comparative disadvantage when information about borrowers is soft. There is, however, little evidence on the characteristics of MFIs that may affect their probability of engaging in a decentralized loan approval process. The banking literature provides some arguments justifying why a bank should decentralize lending operations. Beck et al. (2011) examined whether the ownership type is associated with the likelihood of choosing a decentralized organizational architecture and found that privately owned banks are less likely to decentralize the loan approval process. Berger et al. (2008) addressed the links among ownership type, banking relationship, and multiple relationships in an emerging country such as India. They showed that privately owned banks tend to establish relationships with transparent firms, that is, large and listed firms that are more likely to produce hard information. These studies distinguish between private ownership and state ownership and, among privately owned banks, between foreign-owned banks and domestic-owned banks. Microfinance service providers include nongovernmental organizations (microfinance NGOs), financial cooperatives, micro-banks, and non-bank financial institutions (NBFIs), which are the main microfinance ownership types (Galema et al., 2012; Tchakoute Tchuigoua, 2010, 2015). Financial cooperatives are member-owned institutions that provide financial services (savings and loans) to their
6
members. NGOs are pro-poor entities with no owners. Their clients are primarily unbanked clients. Financial cooperatives and NGOs are usually grouped into the category of nonprofit MFIs. They offer loans to borrowers whose creditworthiness (assessed through conventional methods) is difficult to ascertain because of the poor quality of assets held and a lack of sufficiently valuable collateral. One may therefore expect that this category of MFIs will use collateral substitutes and primarily select their borrowers on the basis of soft information and soft lending methodologies. Micro-banks and NBFIs are privately owned MFIs that target unserved or underserved individuals and micro and small businesses that may pledge collateral and provide financial statements that help the loan officer to produce hard information. Furthermore, as shown in some empirical studies, there may be a trade-off between outreach and MFIs’ accounting performances (Cull et al., 2007, 2009) and efficiency (Hermes et al., 2011), in which financial sustainability and efficiency goals conflict with the outreach goal. The transformation of a nonprofit MFI (NGO) into a formalized, regulated financial institution (privately owned MFI) tends to accentuate such trade-offs and may lead to mission drift.1 Transformed MFIs therefore may tend to move away from their social mission and contract with creditworthy borrowers who are prepared to pledge collateral. Such MFIs may shift to hard lending practices when deciding whether to approve or deny a loan. Hence, one may expect privately owned MFIs (for-profit MFIs) to centralize the loan approval process given that it primarily depends on hard information about the borrower. The use of soft information seems to be a remedy against the incompleteness of contract in credit markets. Small size appears to be a feature of the banks that extensively use soft information and engage in relationship lending. Small banks are in a better position than large ones to collect and act on soft information and are more likely to lend to informationally opaque borrowers (Berger et al., 2005). Results of empirical studies in the banking sector suggest that smaller MFIs may favor a decentralized decisionmaking process. However, large MFIs are complex organizations that are geographically diversified
According to Armendáriz de Aghion and Szafarz (2011), “mission drift relates to a phenomenon whereby an MFI increases its average loan size by reaching out to wealthier clients. Mission drift in microfinance arises when an MFI finds it profitable to reach out to unbanked wealthier individuals while at the same time crowding out poor clients.” 1
7
because of market penetration strategies. Creating branches and sub-branches facilitates geographical proximity and close relationships between MFIs and their clients and enables loan officers to produce soft information, which is then used to select borrowers. This suggests that large MFIs, that is, those that are spatially diversified, tend to decentralize their credit decisions at the branch level.
3. Data and results 3.1. Sample MFI-level variables primarily come from assessment reports produced by Planet Rating, one of the agencies that specializes in the rating of microfinance institutions. Using rating reports, instead of the MIX database, helps gather information on the loan approval process in MFIs. Besides, as argued by Galema et al. (2012) and Hudon and Traca (2011), data collected and produced by rating agencies such as Planet Rating tend to be more reliable and representative of the microfinance industry than selfreported data, such as that obtained from the MIX database. From 2003 to 2015, about 435 rating reports were available on the Planet Rating website. Some rating reports were not exploited because they were confidential, specifically, those published between 2013 and 2015. Data before 2003 were gathered from Planet Rating reports available on the Rating Fund website (www.ratingfund2.org). We built a sample of 362 assessment reports for 267 MFIs from 2001 to 2012 across 67 countries. The sample included MFIs from the following four main regions: Africa and Middle East (138 rating reports); South Asia, East Asia, and the Pacific (28 rating reports); Eastern Europe and Central Asia (54 rating reports); and Latin America and the Caribbean (142 rating reports). The country-level variables, that is, countries’ rates of economic growth and corruption indexes, respectively come from the World Bank’s World Development Indicators (WDIs), available on the World Bank website (http://data.worldbank.org/data-catalog/world-development-indicators), and from the Worldwide Governance Indicators (WGIs) dataset (Kaufmann et al., 2011), available on the same website (http://info.worldbank.org/governance/wgi/index.asp).
8
3.2. Variables We base the construction of the outcome variable (credit approval decision) on incentives-based theories (Aghion & Tirole, 1997; Stein, 2002), which argue that allocating the decisional authority, that is, merging the resource allocation and information production functions to an agent (the loan officer in the case of lending organizations), provides him or her with incentives to produce and use soft information when approving or denying loans. This association reduces the risk of providing and using irrelevant information in the decision-making process. We measure the credit approval decision by a dummy, taking the value 1 if the loan officer has discretion to make the final credit approval decision, and 0 otherwise. The banking literature documents that the choice of a specific organization structure, that is, between a centralized versus a decentralized loan approval process, depends on firm-specific characteristics and countries’ legal institutions (Berger & Udell, 2002; Stein, 2002). We measure the ownership type variable by three dummy variables, namely, privately owned MFIs including micro-banks and NBFIs, member-owned MFIs (cooperatives), and not-owned entities (NGOs), in order to identify among nonprofit MFIs those that tend to decentralize the loan approval decision. Only two of these three modalities are taken into account in the regressions. The size of MFIs is measured by the natural logarithm of the total branches (SIZE). Following previous studies that link the microfinance entity’s choice of lending methodology to its social efficiency and loan portfolio quality (Cull et al., 2007; Mersland & Strøm, 2009; Tchakoute Tchuigoua, 2012), we account for the type of lending methodology applied by the MFIs. Indeed, MFIs make loans either under individual loan contracts or joint liability contracts, including village banking and solidarity/group lending. Unlike the previously mentioned studies that assess each lending methodology by dummies, we measure the type of loan by the total amount of loans approved under each type of lending methodology as a percentage of the overall gross loan portfolio. The loan officer size is measured by the percentage of loan officers.
9
Stiglitz (1990) and Berger and Udell (2002, 2006) argue that environmental factors, such as institutional environments in which MFIs operate, are likely to drive the choice between hierarchy and decentralization. Weaker or inadequate institutions are detrimental to contract enforcement and tend to exacerbate information asymmetries. The quality of the institutional environment raises the question of the reliability of collateral that borrowers own and can pledge and accounting information contained in the financial statements provided by microfinance clients such as very small enterprises (VSEs) and small and medium-size enterprises (SMEs). The low reliability of hard information in countries with weaker institutions suggests that MFIs should favor decentralization. We thus control for country governance quality measures. To this end, we assess the governance quality by estimating the country’s regulatory quality. Finally, we also control for the country’s rate of economic growth. Following Beck et al. (2011), we also control for the size of the country’s banking sector. Table 1 summarizes firm-level and country-level variables used in the study. (Insert Table 1 here) 3.3. Model Although the data structure looks like that of a panel data (267 MFIs and 12 years), it seems difficult to apply the principles of panel data analysis. No MFI renewed its ratings 12 times over the period studied. Four MFIs renewed their ratings four times, six thrice, nine twice, and five once. The data structure is thus an independently pooled cross-section. To identify which MFI-level variables may explain the choice of a decentralized loan approval process, we thus apply a pooled probit regression with control for country-level variables, region year-fixed effects, and heteroscedasticity. The estimated model is as follows: Probit (decentralization of the loan approval decision) = α0 + αi Xi + βi CVj + γi + δt + (1)
where i indexes the MFI, and j the country. Xi is the vector of MFI-specific variables: size (number of branches), joint liability contract, ownership type, percentage of loan officers. CVj is the vector of
10
country-level variables including the country’s regulatory quality and rate of economic growth. γi captures the region-fixed effects, and δt is the year-fixed effects. To analyze the effect of decentralization on the credit portfolio’s quality, the breadth of outreach and the credit availability, we estimate the following model (equation 2). Insofar as the relationship between decentralization and outreach may be driven by MFIs’ incentives to allocate the credit decision authority to the loan officer, there is an endogeneity bias, and more specifically a selection bias, that we correct by implementing the Heckman's two-step procedure.
Yi = α0 + αi Xi + βi CVj + λ IMRi + γi + δt +
(2)
where i indexes the MFI and j the country. Yi is the vector of dependent variables (breadth of outreach, loan portfolio quality, and size of the loan portfolio); Xi is the vector of MFI-specific variables: decentralization of the loan approval, age, square of the age, joint liability contract, size (number of branches). CVj is the size of the country’s banking sector; IMRi is the inverse Mills ratio, which we derived from equation (1) and introduced into equation (2) in order to control for selection effects (endogeneity of the choice of decentralization); γi captures the region-fixed effects, and δt is the year-fixed effects.
4. Results Summary statistics (Table 2) show that in 52% of the cases the loan approval process is allocated to the loan officer. On average, MFI networks are made up of 17 branches, and outstanding individual loans account for 61% of loans granted by MFIs. NGOs represent 45% of the sample, cooperatives 17%, and shareholder-based MFIs 38%. (Insert Table 2 here) Prior to estimations, we assessed the presence of multicollinearity among explanatory variables (Table 3). Our diagnostic reveals that, except for the correlation between individual lending and joint
11
liability contracts (r=–0.99; p=0.01), and between NGOs and privately owned MFIs (r=0.71; p=0.01), the intensity of the relationship between other explanatory variables is weak. (Insert Table 3 here) Multivariate results are reported in Tables 4 and 5. In all models, standard errors are heteroscedasticity robust. In each table, we report the results the baseline model and then check for robustness by replacing the number of branches by the natural logarithm of total assets (USD). In Table 4 we present the results of the probit estimation. The results are consistent with our expectations. Unlike Berger and Udell (2002) and Stein (2002), who found that large banks preferred a hierarchical and standardized decision-making model, we argue that large or complex MFIs are likely to decentralize the loan approval process. Indeed, the relationship between the size (number of branches) and the loan approval process is positive, indicating that large MFIs, those that are spatially diversified, tend to allocate authority to loan officers. MFIs that have been developed through the creation of local branches benefit from proximity to their clients and ties between loan officers and borrowers, allocating the decision to grant credit to loan officers in order to mitigate information problems in lending. (Insert Tables 4 and 5 here) The type of loan products provided by MFIs is also associated with the decentralization of the credit decision. Indeed, MFIs providing loans through joint liability are likely to allocate the decisional authority to loan officers. The cooperative status is associated with a decentralized loan approval process, suggesting that cooperative MFIs are more likely to decentralize the loan approval process. This result may be explained by the fact that most cooperatives expand mainly within their local communities (Mersland, 2009). Finally, MFIs in countries with weaker institutions are more likely to favor a decentralized loan approval process. Results of the pooled OLS regression are reported in Table 5, with control for the endogeneity of the decentralization of the loan approval. We estimate the effect of decentralization on MFIs’ breadth of outreach, loan portfolio size, and loan portfolio quality. The relationship between the decentralization of
12
the loan approval and the breath of outreach is positive and significant, suggesting that the proximity between the loan officer and the clients (related to the transfer of the decision-making authority to the loan officer) enables the MFI to extend credit to a larger number of clients. However, decentralization does not necessarily lead to an increase in the size of loans or to a deterioration in the loan portfolio quality, because the coefficient of decentralization is not significant either in the loan portfolio size or in the loan portfolio quality models. We thus conclude that allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (number of active borrowers) without affecting the size of the loan portfolio or deteriorating the loan portfolio quality. To check for robustness, we replaced the number of branches by the natural logarithm of the book value of assets (USD) and found them consistent (Tables 4 and 5). As an additional robustness check (the probit model) we distinguished whether the loan approval process was centralized or not. The credit approval decision is measured by a dummy taking the value 0 if the loan approval decision is centralized, that is, when all loan applications are first completed by loans officers (LOs), reviewed by the branch manager, and then sent to the head officer at headquarters for credit committee, and 1 otherwise, that is, when the loan officer has discretion to make the final credit approval decision, which is decentralized at the branch level in the credit committee comprising the branch manager and loan officers. By participating in the credit committee at the branch level, the loan officer may influence the credit decision. Summary statistics show that in 81% of cases, the loan officer directly approves credit or influences the loan approval process by participating in the credit committee at the branch level. The results of the probit regression, reported in Table 6, are consistent with those reported in Table 4. From the results of the probit model presented in Table 6, we generated the inverse Mills ratio and re-estimated equation (2). The results obtained are in accordance with those presented in Table 7. (Insert Tables 6 and 7 here)
5. Conclusion
13
This article addresses the question of whether the characteristics of MFIs affect the choice of a decentralized loan approval process. The secondary issue discussed is whether allocating the loan approval decision to loan officers enables microfinance institutions to increase the number of loans and improve their loan portfolio quality after controlling for the endogeneity of the choice of decentralizing the loan approval. To achieve this goal, we exploited a pooled sample of 362 assessment reports for 267 MFIs from 2001 to 2012 across 67 countries. Findings reveal that size, cooperative status, and solidarity lending are MFI-level variables that affect the likelihood of choosing a decentralized loan approval process. Allocating the decision-making authority to the loan officer improves MFIs’ breadth of outreach (the number of active borrowers) but does not affect the size of the loan portfolio and does not deteriorate the loan portfolio quality. However, Stein (2002) and Berger and Udell (2002) argue that a decentralized credit decision process leads to agency conflicts between the loan officer and the organization. Loan officers may be tempted to follow their subjective preferences (Agier & Szafarz, 2013), depart from approved policies, or act opportunistically. An interesting direction for future research therefore is to investigate existing mechanisms put in place by MFIs to reduce agency conflicts inherent in a decentralized loan approval process and to limit fraud and opportunism, ensuring that loan officers make loan decisions that help improve MFIs’ efficiency.
14
References Aghion, P., & Tirole, J. (1997). Formal and real authority in organizations. Journal of Political Economy, 105, 1–29. Agier, I., & Szafarz, A. (2013). Subjectivity in credit allocation to micro-entrepreneurs: Evidence from Brazil. Small Business Economics, 41, 263–275. Armendáriz de Aghion, B., & Morduch, J. (2010). The economics of microfinance. Cambridge, MA: MIT Press. Armendáriz, B., & Szafarz, A. (2011). On mission drift in microfinance institutions. In the handbook of microfinance, B. Armendariz and M. Labie Ed. 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 commercial microfinance organizations. Academy of Management Journal, 53, 1419–1440. 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. Behr, P., Entzian, A., & Güttler, A. (2011). How do lending relationships affect access to credit and loan conditions in microlending? Journal of Banking and Finance, 25(8), 2169–2178. 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. (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. Berger, A. N., & Udell, G. F. (2006). A more complete conceptual framework for SME finance. Journal of Banking & Finance, 30(11), 2945-2966. 15
Cull, R., Demirgüç-Kunt, A., & Morduch, J. (2007). Financial performance and outreach: A global performance of leading microbanks. Economic Journal, 117, 107–133. Cull, R., Demirgüç-Kunt, A., & Morduch, J. (2009). Microfinance meets the market. Journal of Economic Perspectives, 23, 167–192. D'espallier, B., Hudon, M., & Szafarz, A. (2013). Unsubsidized microfinance institutions. Economic Letters, 120, 174–176. Diamond, D. W. (1984). Financial intermediation and delegated monitoring. Review of Economics Studies, 51, 393–414. Dixon, R., Ritchie, J., & Siwale, J. (2007). Loan officers and loan delinquency in microfinance: A Zambian case. Accounting Forum, 31, 47–71. Dorfleitner, G., Just-Marxa, S., & Priberny, C. (2017). What drives the repayment of agricultural micro loans? Evidence from Nicaragua. Quarterly Review of Economics and Finance, 63, 89–100. Galema, R., Lensink, R., & Mersland, R. (2012). Do powerful CEOs determine microfinance performance? Journal of Management Studies, 49, 718–742. Gutiérrez Nieto, B., Serra Cinca, C., & Mar Molinero, C. (2007). Microfinance institutions and efficiency. Omega, 35, 131–142. Hermes, N., Lensink, R., & Meesters, A. (2011). Outreach and efficiency of microfinance institutions. World Development, 39, 938–948. Hudon, M., & Périlleux, A. (2014). Surplus distribution and characteristics of social enterprises: Evidence from microfinance. Quarterly Review of Economics and Finance, 54, 147–157. Hudon, M., & Traca, D. (2011). On the efficiency effects of subsidies in microfinance: An empirical inquiry. World Development, 39, 966–973. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2011). The worldwide governance indicators: methodology and analytical issues. Hague Journal on the Rule of Law, 3(2), 220-246. Labie, M., Méon, P.-G., Mersland, R., & Szafarz, A., (2015). Discrimination by microcredit officers: Theory and evidence on disability in Uganda. Quarterly Review of Economics and Finance, 58, 44–55. 16
Mersland, R. (2009). The cost of ownership in microfinance organizations. World Development, 37, 469– 478. Mersland, R., & Strøm, R. Ø. (2009). Performance and trade-offs in microfinance organizations—Does ownership matter? Journal of International Development, 20, 598–612. Périlleux, 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. Roberts, P. W. (2013). The profit orientation of microfinance institutions and effective interest rates. World Development, 41, 120–131. 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 and 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? 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. Quarterly Review of Economics and Finance, 52, 427–437. Tchakoute Tchuigoua, H. (2015). Determinants of the governance quality of microfinance institutions. Quarterly Review of Economics and Finance, 58, 32–43.
17
Table 1: Description of variables Identity of variables Breadth of outreach Size of the loan portfolio
Number of active borrowers Gross loan portfolio
Loan portfolio quality Portfolio at risk at 30 days Loan approval process Size
Lending methodology or type of loan
Legal status Loan officers Maturity
Decentralization Number of branches Individual loan contracts Joint liability contracts NGO Cooperative Shareholderbased Percentage of loan officers Age Regulatory quality
Country’s variables
Country’s rate of economic growth
Size of the banking sector
Regions
LAC EECA AFRICA and MENA ASIA
Definition and Measurement The total number of active borrowers Gross loan portfolio/total assets Outstanding balance on arrears over 30 days + Total gross outstanding refinanced (restructured) portfolio)/Total gross portfolio Measurement of portfolio quality that shows the part of the portfolio affected by outstanding payments when there is a risk that they will not be repaid; the threshold is < 10% given that financial guarantees in microfinance are not always sufficient Dummy: 1 if the credit decision process is decentralized to the loan officer, 0 otherwise Natural logarithm of total branches Individual loans as a percentage of the outstanding loan portfolio Loan granted to a single borrower (group-lending loans + village banking loans) as a percentage of the outstanding loan portfolio Under group-lending methodology, loans are granted to individuals but the group (3–10 members) is jointly liable for credit. Group lending through village banking methodology is based on larger groups (over 10 people) Binary variable: 1 if the MFI is a nongovernmental organization, 0 otherwise Binary variable: 1 if the MFI is a cooperative, 0 otherwise Binary variable: 1 if the MFI is a shareholder MFI, 0 otherwise Number of loan officers/total number of staff members Number of years since inception The index reflects perceptions about the ability of the government to formulate and implement sound policies and regulations that help promote the private sector’s development. Estimate of governance (ranges from approximately −2.5 (weak) to 2.5 (strong) governance performance). Source: Kaufmann et al. (2011); World Bank. Annual growth rate of the GDP per capita Domestic credit provided by the banking sector as a percentage of the GDP. Domestic credit provided by the financial sector includes all credit to various sectors on a gross basis, with the exception of credit to the central government, which is net. The financial sector includes monetary authorities and deposit money banks, as well as other financial corporations where data are available (including corporations that do not accept transferable deposits but do incur such liabilities as time and savings deposits). Examples of other financial corporations are finance and leasing companies, money lenders, insurance corporations, pension funds, and foreign exchange companies. Dummy: 1 if the region is Latin America and the Caribbean, 0 otherwise Dummy: 1 if the region is Eastern Europe and Central Asia, 0 otherwise Dummy: 1 if the region is Africa and Middle East, 0 otherwise Dummy: 1 if the region is South Asia, 0 otherwise
18
Table 2: Descriptive statistics
Variables Breadth of outreach
Mean
Standard Deviation
Minimum
Maximum
Median
21,967.43
57,050.73
23.00
643,659.00
8,003.50
Size of the loan portfolio
0.74
0.17
0.05
1.03
0.76
Loan portfolio quality
0.05 0.52
0.07 0.50
0.00 0
0.54 1
0.03 1
25,900,000
87,300,000
43,275.60
1,010,000,000
5,219,776
Branches
17.36
34.61
0.00
410.00
8
Individual lending
0.61
0.41
0
1
0.84
Joint liability contracts
0.39
0.41
0
1
0.16
Shareholder-based
0.38
0.49
0
1
0
Cooperatives
0.17
0.37
0
1
0
NGOs
0.45
0.50
0
1
0
0.67
4.30
0.04
82.20
0.45
10.50
8.47
0.00
51.00
8.00
Decentralization Size
Percentage of loan officers Maturity Country’s rate of economic growth Regulatory quality
0.04
0.03
-0.07
0.14
0.03
−0.21
0.46
−1.21
1.50
−0.20
Size of the banking sector
0.42
0.31
−0.01
2.02
0.34
19
Table 3: Pearson’s correlation matrix explanatory variables Branches
Individual lending
Joint liability contracts
Privately owned
Cooperatives
NGOs
Percentage of loan officers
Regulatory quality
Branches
1.00
Individual lending
0.04
1.00
Joint liability contracts
−0.04
−0.99***
1.00
Privately owned
0.15***
0.14***
−0.14***
1.00
Cooperatives
−0.03
0.13**
−0.13**
−0.35***
1.00
−0.13**
−0.23***
0.23***
−0.71***
−0.41***
1.00
−0.01
−0.09
0.09
0.07
-0.03
−0.04
1.00
−0.07
0.02
−0.02
0.02
−0.06
0.03
−0.01
1.00
0.04
0.01
−0.01
0.12**
−0.04
−0.08
0.04
−0.11**
NGOs Percentage of loan officers Regulatory quality Country’s rate economic growth
of
Country’s rate of economic growth
1.00
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
20
Table 4: Multivariate results of the pooled probit with control for year-fixed effects and regions Baseline model Size is measured by the natural logarithm of the number of branches.
Robustness checks Size is measured by the natural logarithm of the number of the total book value of assets. Coefficients t-stat
Coefficients
t-stat
Constant
−0.31
−0.67
−1.56
−1.59
Branches
0.30***
4.39
0.12**
2.23
Joint liability contracts
0.39**
2.13
0.43**
2.37
Cooperatives
0.47**
2.21
0.44**
2.02
NGOs
−0.04
−0.23
−0.11
−0.61
Percentage of loan officers
−0.03***
−4.48
−0.04***
−4.98
Regulatory quality Country’s rate of economic growth Region dummies
−0.41**
−2.53
−0.39**
−2.40
0.03
0.11
0.09
0.04
yes
yes
Year-fixed effects
yes
yes
Pseudo R-squared
13.90%
11.32%
Chi 2
80.68***
66.28***
Log-likelihood value
−212.86
−219.23
Percent correctly predicted
66.39%
65.83%
N 357 * Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
357
21
Table 5: Results of the pooled OLS regression with control for year-fixed effects and regions. We analyze the effect of decentralization on MFI breadth of outreach, loan portfolio size, and loan portfolio quality.
Variable Constant Decentralization of the loan approval Age Age-squared
Baseline model Size is measured by the natural logarithm of the number of branches. Breadth of Size of the Loan portfolio outreach loan portfolio quality 6.76*** 0.59*** 0.02
Robustness check Size is measured by the natural logarithm of the number of the total book value of assets. Breadth of Size of the Loan portfolio outreach loan portfolio quality −0.45 0.44*** 0.12**
(9.59)
(4.94)
(0.93)
(−0.43)
(2.46)
(2.29)
0.43***
−0.01
−0.002
0.41***
−0.01
−0.00
(2.88)
(−0.69)
(−0.80)
(3.52)
(−0.67)
(−0.77)
0.05
0.01**
−0.002*
−0.00
0.01**
−0.00
(1.58)
(2.43)
(−1.69)
(−0.02)
(2.26)
(−1.43)
−0.00
−0.004**
0.001**
0.00
−0.001**
0.001*
(−1.22)
(−2.31)
(2.09)
(0.37)
(−2.26)
(1.92)
Joint liability contracts
0.97***
0.03
-0.01
1.00***
0.02
-0.01
(6.00)
(1.06)
(-1.46)
(6.28)
(0.83)
(-1.63)
Size
0.76***
0.03**
−0.01
0.64***
0.02**
−0.01***
(10.35)
(2.42)
(−1.37)
(10.49)
(2.28)
(−2.62)
0.34
0.01
−0.00
0.29*
0.01
−0.00
(1.52)
(0.33)
(−0.36)
(1.67)
(0.50)
(−0.15)
Size of the banking sector Mills ratio F r2 Adjusted_R² Number of observations
0.46
0.10**
0.01
−0.75***
0.04
0.02
(1.54)
(2.48)
(0.46)
(−3.08)
(1.28)
(0.90)
16.58*** 43.52%
2.071*** 14.96%
11.01*** 11.58%
65.62*** 59.03%
2.208*** 14.89%
4.299*** 13.26%
39.87%
9.45%
58.51%
56.36%
9.37%
7.64%
347
346
346
347
346
346
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
22
Additional robustness checks Table 6: Centralized versus decentralized multivariate results of the pooled probit with control for year-fixed effects and regions. The loan approval process is decentralized at the branch level: The loan officer has the final authority or participates in the credit committee at the branch level. He/ or she may thus influence the loan approval process. The credit approval decision is measured by a dummy taking the value 1 if the loan officer influences the final credit approval decision, and 0 otherwise, i.e. that is, if the loan approval decision is centralized. Coefficients
t-stat
Constant
–0.06
–0.10
Branches
0.30***
3.73
Joint liability contract
0.45**
2.07
Cooperatives
0.35
1.44
NGOs
–0.01
–0.06
Percentage of loan officers Regulatory quality Country’s rate of economic growth
0.02**
2.29
–0.54***
–2.83
–3.82
–1.24
Region dummies
yes
Year-fixed effects
yes
Pseudo R-squared
9.44%
Chi 2
36.08**
Log-likelihood value
–157.41
Percent correctly predicted
80.67%
N
357
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
23
Table 7: Results of the pooled OLS regression with control for year-fixed effects and regions. We analyze the effect of decentralization on MFIs’ breadth of outreach, loan portfolio size, and loan portfolio quality. The loan approval process is decentralized at the branch level: The loan officer has the final authority or participates in the credit committee at the branch level. He/ or she may thus influence the loan approval process. The credit approval decision is measured by a dummy taking the value 1 if the loan officer influences the final credit approval decision, and 0 otherwise, i.e. that is, if the loan approval decision is centralized.
Variable Constant Decentralization of the loan approval Age Age-squared Joint liability contracts Size (natural logarithm of the number of branches)
Decentralization of the loan approval Breadth of Size of the loan Loan portfolio outreach portfolio quality 7.50*** 0.64*** 0.03 (9.86)
(4.87)
(1.22)
0.42**
0.001
−0.01
(2.36)
(0.01)
(−1.25)
0.04
0.01**
−0.002*
(1.52)
(2.41)
(−1.68)
−0.002
−0.003**
0.004**
(−1.30
(−2.33)
(2.12)
0.82***
0.02
−0.01
(4.60)
(0.55)
(−1.25)
0.64***
0.02
−0.01
(7.74)
(1.32)
(−1.30)
0.45*
0.02
−0.00
(1.95)
(0.75)
(−0.49)
−0.47
0.06
0.02
(−0.80)
(0.68)
(0.57)
F
15.59***
1.924***
10.34***
r2
43.09%
13.7%
12.13%
Adjusted_R²
39.41%
8.11%
6.43%
347
346
346
Size of the banking sector Mills ratio
Number of observations
* Significant at the 0.1 level; ** significant at the 0.05 level; *** significant at the 0.01 level.
24