Microfinance and households access to credit: Evidence from Côte d’Ivoire

Structural Change and Economic Dynamics 23 (2012) 473–486

Contents lists available at SciVerse ScienceDirect

Structural Change and Economic Dynamics journal homepage: www.elsevier.com/locate/sced

Microfinance and households access to credit: Evidence from Côte d’Ivoire Edith Leadaut Togba ∗ Department of Economics and Management, University of Cocody-Abidjan, PO Box 2761, Abidjan 04, Côte d’Ivoire

a r t i c l e

i n f o

Article history: Received January 2010 Received in revised form August 2012 Accepted August 2012 Available online 19 August 2012 Keywords: Microfinance Social capital Access to credit Heckman two steps

a b s t r a c t Evidence on microfinance services these days ironically shows a great preference for savings products rather than credit products by households. For some authors, this phenomenon is explained by the fact that microfinance products, and especially loans, from formal microfinance institutions do not fit the households demand. This paper first presents evidence on the observed phenomenon in the Ivorian microfinance sector. Second, it analyses the Ivorian credit market so as to understand the determinants of the choice for credits from formal sources versus informal sources. The results reveal the size of the loan, agricultural purpose, the geographical area where households live and ethnicity as factors influencing the choice for formal sources. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Microfinance is made up of a set of small size financial products which include savings, credit and insurance, and which suit people with low incomes who were at first excluded from the classic or formal banking system (Soulama, 2005). Microfinance products are provided by the intermediation of a multitude of institutions which vary according to size, the degree of organization and the legal status, and include NGOs, associations, mutual insurance companies/cooperatives, limited companies, banks, financial institutions, but also more informal and unregulated institutions such as tontines, usurers, change keepers, loans between friends, etc. (Soulama, 2005, www.lamicrofinance.org). For some years now, microfinance has been found a vital tool to eradicate poverty among the vulnerable by the provision of products and banking services similar to those delivered by classic institutions (Brau and Woller, 2004). Observations were made on the growth of these institutions all over the world.

∗ Tel.: +225 04 44 63 48. E-mail address: [email protected] 0954-349X/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.strueco.2012.08.002

In fact, according to the report of the 2012 campaign published by the microcredit summit on December 31, 2010, there existed 3652 institutions who took care of about 200 million customers. In spite of their recent development and strong proximity to the poor, it is important to note that formal microfinance institutions only bring a very partial answer to the poor households’ need for finance. Nimal (2008) reveals that although microfinance institutions are present in Asia and the pacific regions, more than 300 million households suffer from the lack of access to financial products offered by the formal and semi-formal sector. In sub-Saharan Africa in 2008, the number of active borrowers and savers in percent of population living under national poverty line has reached a rate of penetration of about 3% for credits and 5% for savings (MIX and CGAP, 2010). These rates confirm the irony observed around microfinance today which is the preference of households for savings products, in comparison to loans (Meyer, 2002). Yet, credit is important to finance the start of activities and allows fixed capital investments in poor rural and urban economies where it is difficult to save, as confirmed by Guirkinger (2008) for Peruvian rural zones. Facing these constraints, lots of households continue to rely on informal

474

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

sources of finance to increase their production capacity, to diversify risk, and to smooth their consumption during their life cycle. Informal finance refers to all transactions, loans and deposits occurring outside the regulation of a central monetary authority (Atieno, 2001). Besley and Burgess (2001) reported that 82% of loans are contracted in the informal sector against only 12% for the formal sector in Nepal. In a survey on the credit market in Côte d’Ivoire, Azam et al. (2001) noted that small entrepreneurs in Côte d’Ivoire prefer to borrow from parents and neighbours. Therefore, there seems to be a need to understand the behavior and preferences of the poor in terms of financial services (Matin et al., 2002). Given the often informal nature of the financial services used by the poor, such an analysis requires integrating the presence of social networks and social capital,1 not only as a source of choices of credits but also as an explanatory factor of the choice. In developing countries, networks are made up of the community and ethnic groups. The first network generally constitutes a source for competitive credit. Considering ethnic network analysis, studies on their interaction with the credit market are scarce. Some research documents examine the interaction between ethnic groups, credit and enterprises in Africa but they do not explore the diversity of the indigenous African population (Biggs et al., 2002; Fafchamps, 2000, 2003; Fisman, 2003). Azam et al. (2001) equally do not take into account this indigenous diversity. This paper tries to fill this gap, by including ethnic diversity in the indigenous African population in the analysis of preferences and use of financial instruments. The paper focuses in the first place on estimating a model that explains what determines whether or not households in Côte d’Ivoire make use of loans. Next the paper tries to determine the factors influencing households’ decision to obtain loans from informal versus formal sources. The structure of the paper is as follows: in the next section, we introduce the microfinance sector in Côte d’Ivoire by describing the characteristics of the microfinance institutions concerned. Section 3 provides a short literature review which is the basis for the empirical analysis of the choice for particular financial sources. Section 4 deals with the methodology, and Section 5 gives basic information on the data which were used. Finally, Section 6 presents the results of the study, and the final section summarizes the argument.

1 Social capital refers to the norms and networks that enable people to act collectively (Woolcock and Narayan, 2000). Fox (1996) defines social capital as a social organization, relationship of cooperation and reciprocity, networks and leadership that facilitate collective action. However, according to the level considered, the definition differs. Social capital could be the degree of trust in government or other societal institutions (Fukuyama, 1995 cited in Okten and Osili, 2004), social cohesion, reciprocity and institutional effectiveness. Or as stated by Grootaert and van Bastelaer (2002a), social capital is broadly the institutions, the relationships, the attitudes, and values that govern interactions among people and contribute to economic and social development. This definition depicts closely the developing countries’ situation.

2. The microfinance sector in Côte d’Ivoire Côte d’Ivoire is classified 164th of 177 countries in the Human Development index with 49% of the 21.1 million citizens living under the poverty line (World Bank, 2009). According to same source, poverty is most severe in the savanna of the north (54.6%) and the rural forest of the East (46.6%), followed by the urban regions (33.8%) apart of Abidjan, the western rural forest (24.5%) and Abidjan (11.1%). Considering the poverty of its population, the government of Côte d’Ivoire decided to make poverty alleviation a priority in its socio-economic programs. Many strategies were defined to achieve this goal. The creation of microfinance institutions for households was one of the strategies. A microfinance institution (MFI) is an organization that provides financial services to the poor. This very broad definition includes a wide range of providers that vary in their legal structure, mission, and methodology. However, all share the common characteristic of providing financial services to clients who are poorer and more vulnerable than traditional bank clients. Microfinance institutions are now observed in all sectors of the economy where long term financing is not a priority. In Côte d’Ivoire, the available alternatives for a household include banks and financial institutions, companies of framing, credit unions, social funds, ROSCAs, moneylenders, cooperatives and other. Formal providers are defined as those that are subject not only to general laws but also to specific banking regulation and supervision (development banks, savings and postal banks, commercial banks, and non-bank financial institutions, credit unions). Semi-formal providers may also be registered entities subject to general and commercial laws but not under bank regulation and supervision (companies of framing and social funds). Informal providers are non-registered non-regulated groups, such as rotating savings and credit associations (ROSCAs) and cooperatives, moneylenders and other (friends, family). The supervision or regulation of certain institutions aims at protecting customers and allowing access through limiting the price for credit, which is lower than those practiced by non-regulated institutions. It also aims at securing financial operations by requesting to respect managerial norms like prudency and demanding for operational autonomy. The effects of this regulation are felt at the level of access or demand for credits and financial products because conditions for borrowing from these institutions may become difficult to fulfill by households. Indeed, this is displayed in the form of prescribed minimum loan amounts, complicated application procedures and restrictions on credit for specific purposes (Schmidt and Kropp, 1987). On the contrary, the service from informal sources is based on flexible arrangements to adjust to changing economic circumstances, and on reducing the transaction costs to borrowers, who respond by maintaining discipline in order to sustain their access to credit (Atieno, 2001). Unlike formal sources, informal lenders often attach more importance to loan screening than to monitoring the use of credit. Screening practices often include group observation of individual habits, personal knowledge by

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486 Table 1 Distribution of formal MFIs’ funding by economic sectors in Côte d’Ivoire. Sectors

2004 (%)

2005 (%)

Trade Crafts industry Agriculture Construction and housing Industry Transportation Catering industry Education and health Other sectors

39.08 0.62 9.32 18.15 0.03 0.10 6.25 21.00 5.45

30 13.5 10.1 0.1 0.0 5.7 0.1 1.4 39.2

Source: Computed by the author from NCM report (2004, 2005).

individual moneylenders, recommendations by others and creditworthiness. Therefore, low income households have to choose between borrowing from formal sources, where credit is cheaper, but where their loan application is usually rejected, or resorting to informal sources where funding is much more expensive. The microfinance sector is not only of use to informal economic activities, but also extends its area to civil servants, and private sector employees. Statistics provided by the National Commission of Microfinance (NCM) (2005) reveal that a large share of the formal microfinance funds is used to finance economic activities outside agriculture. However, it can be noticed that in a country whose economies are founded on agriculture, just little funds are allocated to this sector. Statistics show that in 2005, trading activities received more funds than any other sector, representing about 30% of the funds. Consumer credits for education and health represent 1.4% in 2005. There is a lot of variation over time: the construction and housing, education and health and catering industries show a sharp decline in 2005. Yet, sectors like crafts and transportation industries have observed a relative increase (Table 1). The supply of financial services from formal MFIs is essentially centered on savings and credits. It is clearly an important industry in terms of mobilizing savings. Savings were estimated to be 88,679 million FCFA in 2008 (MEF, 2009). Compared to these savings, 33,013 million FCFA, or about 37%, were supplied as outstanding discounted bills of credit by formal MFIs. Fig. 1 shows these proportions for the period 1998–2005, revealing a rapid growth in savings at the detriment of credits in MFIs. These data reveal a great interest by the population in products proposed by MFIs, but especially so in savings. Two factors might explain this situation. The first is that there is effectively a stronger preference for savings (e.g., as observed by Meyer, 2002). This preference could be explained by the need for secured savings even though the interest rate is low (Yaron et al., 1994). The second factor is based on the notion of transaction costs and information asymmetry between contracting parties. Since the microfinance institution is incapable to control all actions of borrowers due to incomplete and expensive information, it will formulate terms of contracts that attract less risky borrowers and will be in favor of the MFI. Transaction costs incurred in obtaining credits are then considered more important than the utility derived (Atieno, 2001). Therefore there exist significant obstacles in transforming

475

a potential demand to an actual or real demand (Aryeetey, 1996b). In fact, on 912,959 individual customers in 2008, only 210,327 effectively demanded and received a credit, which is 23.04% of the customers expressing their need for credit. It could thus be possible that the great increases in savings are due to the terms of concession or conditions of obtaining credit. The initial requirements to get a loan from a MFI are: (i) a minimal deposit is required to those requesting a loan; (ii) adherence of at least six months with regular savings is required; (iii) a member has the right to demand for a loan equivalent to twice his initial deposit: (iv) other members to be considered as guarantors must have sufficient funds in his/her savings account to cover the amount of loan requested. (v) Sometimes, the need for a guarantor who is not a member of the microfinance institution is recommended. All these conditions are operationalized in a set of documents whose filling is difficult for households who are mostly financial illiterate. In addition, loan contracts include specifications regarding the interest rate, the amount of the loan, the modalities of repayment, and the guarantees, creating possibly additional obstacles to formal credit. The set of conditions and formalities constitute a main factor which creates differences between microfinance institutions. They also give orientation for the choice of the households. In effect, the conditions and formalities required by the formal microfinance institutions are not necessary as far as informal institutions are concerned. Generally, for the informal institutions, it is necessary to have objects to serve as guarantee when the money borrowed is not given back. Besides the absence of an explicit contract and more flexible terms of repayment (i.e., a possibility of spreading the term) lead households to more often look for credit with informal microfinance institutions. This suggests that the household is rational in sense that it makes choices that maximize its direct utility subject to constraint on expenditure. In summary, it seems obvious that there is a big preference for savings in formal MFIs, but this could be explained by the barriers to getting access to formal credits. This could also explain the coexistence of informal credit markets with formal credit establishments. If this is the case, then which factors determines the choice between the two sources? In the next section, we will investigate the literature on the determinants of access to formal credit, and test them on Ivorian data in the subsequent section. 3. Literature review Credit rationing theoricians give an explanation of the borrowers’ choice between formal and informal sources. In fact, according to Stiglitz and Weiss (1981), the presence of problems of asymmetries of information (adverse selection and moral hazard) at the level of credit markets and the problems of enforcement of contract lead to credit rationing through the formal sources. Chung (1995) and Mushinski (1999) showed that heavy transaction costs in the formal sector can discourage some households to take loans. In such a situation borrowers need to consider informal sources as last resorts or as alternatives to the

476

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

Fig. 1. Evolution of savings and loans from formal MFIs on period 1998–2005. Source: compiled by the author from several monographs BCEAO.

formal sector in developing countries. According to Kochar (1992), informal loans, especially those from friends and parents, would be less expensive as compared to the formal loans and therefore preferred by borrowers. Guirkinger and Boucher (2007) add to this that informal lenders have access to local information, allowing them to write down contracts that are less risky for borrowers. The relatively low information and transaction costs, the simplicity and flexibility in financial procedures ease access to low income persons. The microfinance literature demonstrated how the problem of information asymmetry can be tackled by the use of social capital (Rankin, 2002; Gomez and Santor, 2001; Besley and Coate, 1995). Social capital refers to the norms and networks that enable people to act collectively (Woolcock and Narayan, 2000). Fox (1996) defines social capital as a social organization, relationship of cooperation and reciprocity, networks and leadership that facilitate collective action. Social capital could be the degree of trust in government or other societal institutions (Fukuyama, 1995 cited in Okten and Osili, 2004), social cohesion, reciprocity and institutional effectiveness. Or as stated by Grootaert and van Bastelaer (2002a), social capital is broadly the set of institutions, the relationships, the attitudes, and values that govern interactions among people and contribute to economic and social development. This definition depicts closely the developing countries’ situation. Social capital helps to correct the issue of incomplete information by ensuring default payments through social sanctions. Social capital can be a tool for the diffusion of information on the sources of finance and consequently will influence the different choices (Okten and Osili, 2004). This explains the inclusion of variables like ethnic groups and religious adherence further in this model. In the same way, their presence can constitute a non-negligible competitor for the MFIs.

The working hypothesis that will guide the empirical work below is that the microfinance institutions in Côte d’Ivoire can help solve some of the market imperfections that exist in the financial sector. Following the short literature review above, the main way in which this will be possible is by lowering the transaction costs by means of offering loans through the informal sector. It is also hypothesized that social capital plays a role in solving the market imperfections. In practical terms, we will put these hypotheses to the test by estimating an equation for whether or not a household will take a loan through the formal or the informal sector, and including household income, loan size, and social capital variables as explanatory variables. If microcredit helps solve market failure, we expect household income, the size of the loan and social capital all to have a negative impact on loans through the formal sector. There are also a number of variables that need to be taken into account as controls. Borrowers can base their choice on a particular feature of the loan which is the interest rate. However, focusing on this component alone is not sufficient to explain the choices made by households (Nguyen, 2006). Many studies demonstrated that this component is not the most important factor, but that instead factors not directly linked to the prices and services play an important role (Diagne and Zeller, 2001). In fact, if reimbursement modalities, the required collateral, and the availability of additional services are not in accordance with the needs of the borrower, the latter does not manifest any demand for credit from microfinance institutions. The type of financial institutions and borrowing politics put in place will determine the choice made by households (Schmidt and Kropp, 1987) and brings to the forefront the question of what are the characteristics and features that enable households and individuals to borrow from formal sources.

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

Zeller (1994) found that the age, the level of education, salaries and gender have an influence on the choice of sources of finance. For Nguyen (2007), the size of the house and the type of activities exercised has a positive effect on making as choice formal sources of finance. Guirkinger (2008) added factors like the wealth level, the presence of other source of income; the number of persons in his charge and the geographical location of the household are equally susceptible to influence their choice. The wealth and/or income seem to be the basic one among the important characteristics of household. In fact, as Madestam (2007) underlined, the wealth level of the economic agent is an important factor determining the credit sector he chooses. Thus, when an economic agent has got a sufficient level of wealth, he will choose to borrow with the formal institutions. This is confirmed by the results of studies such as Guirkinger (2008), Crook (2006) and Zeller (1994), which reveal that the household’s wealth has positive and significant effect on the loan demand and the choice of finance sources. The noticeable positive effect of the size of households on the demand for loans is confirmed in some contributions (Guirkinger, 2008; Nguyen, 2007; Okurut et al., 2004; Swain, 2002). The role of household’s size can be seen indirectly. The larger the household the greater is its expenditure. Thus, the household would apply for a (larger) loan in order to smooth his consumption. At the empirical level, some authors think that it is important to make the distinction between access and participation to credit (Zeller, 1994; Diagne and Zeller, 2001). Access to credit is an essential phenomenon concerning the supply of credit, due to the fact that the lender decides on who can borrow or not. Participation is a phenomenon linked to demand. A household has access to a certain source if it can in principle borrow from this source. The household participates if it actually borrows from this source. Consequently, a household can have access and choose not to borrow. Taking into account this distinction, the decision to choose one particular source of finance is done from an analysis of the demand for credit expressed by households and that is the approach developed in this paper. 4. The household choice of sources of credit: empirical approach An analysis of households decisional choices in the credit market are usually carried out using discrete choice models (Nguyen, 2007; Duong and Izumida, 2002; Zeller, 1994). In fact, according to Atieno (2001), it is impossible to identify a program for the demand for loans using an amount of observed loans since this only reflects the existing supply. The credit demand function can only be interpreted from the decision to participate by the borrower, that is the decision to borrow or not to borrow and from which sector. Zeller (1994) used a univariate probit model to estimate factors determining the borrowing decisions of individuals, from the point of view of their participation in the formal or informal credit market in Madagascar. The author treats market segments separately to identify the similarities and

477

differences between the sectors in terms of demand for credits and their rationing. Nguyen (2007) also separates the sources of loans while expecting that the determinants of credit participating be different as the eligible requirements are different between sources. This paper adopts Zeller’s approach (1994) that the market must be segmented in order to capture all the features of every source, and models credit sources as products of substitution. In the case of this paper, the available alternatives for a household are: 1: 2: 3: 4: 5: 6: 7: 8:

Bank and financial institutions. Companies of framing. Credit unions. Social funds. ROSCAs. Moneylenders. Cooperatives. Others.

In order to take into account the major differences between these alternatives, they are grouped as (1) formal institutions for those who are regulated, including banks and financial institutions, companies of framing, credit unions, and social funds; and (2) informal institutions, including moneylenders, ROSCAs, cooperatives and others. Assuming that the household is rational in the sense that it makes choices that maximize its direct utility subject to constraint on expenditure, it is possible to derive an indirect utility function (Maddala, 1983). We define an underlying latent variable y* to denote the indirect utility level associated to the direct utility y. The observed variable y is defined as: y=1

if y∗ > 0

y=0

otherwise

(1)

However, there are many errors in this maximization because of imperfect perception and optimization, as well as the inability to measure exactly all relevant variables. McFadden (1974a) suggested using a random function where the random term comes in an additive manner. Consequently, this indirect utility function y* will be written as follows: y∗ = ˇx1 + ε1

(2)

where x1 is a vector of observable attributes specific to the household; ε1 is the random component of utility that represents the unobserved household i’s idiosyncratic taste for choosing a source. It is assumed to be independently and identically distributed. Choice of the source of borrowing is a two-step process which requires that households demand a loan at the first stage, and at the second stage they choose the source where they want to borrow. Since the second stage is a conditional on the first stage, it is likely that the second stage sample is non-random, which could create a sample selection bias. Indeed, Nagarajan et al. (1995) think that estimates of loan demand or choice of credit source are often biased because they use models that do not adequately correct for

478

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

selectivity bias. Therefore, it is important to correct for this sample bias in order to obtain consistent estimates. The idea that factors affecting selection into the sample may simultaneously affect the binary outcome of interest has been the motivation for the introduction of the probit sample selection model (van De Ven and van Praag, 1981). In our case, it is believed that the decisions of choosing a source of borrowing and that of expressing a demand of loan are correlated (both decisions are binary). In effect, the data set specifies a binary variable that identifies the observations for which the dependent is observed/selected or not observed. The underlying structural framework is a household production model with utility maximizing households, who demand credit (demand = 1) if a loan is expected to increase utility, and they do not demand credit (demand = 0) in the opposite case. The dichotomous demand selection equation is given by:



d=

1

if d∗

0

otherwise

>0 (3)

d = ıx2 + ε2

y=

1 if y∗ > 0 and d = 1 0

y∗ ≤ 0

(4)

(5)

where the latent equation for outcome equation is y∗ = ˇx1 + ε1

(6)

It is assumed that the latent errors are bivariate normal and independent of the explanatory variables. The probit model with sample selection can be expressed as follows: yi∗ = ˇx1 + ε1



yi =

1 if yi∗ > 0 and d = 1 yi∗ ≤ 0

+ ˛4 socioecogrpi + ˛5 malei + ˛6 hhsizei + ˛7 areai

+ ˛11 agei + ε1i

The outcome dependent variable y is observed only if y∗ > 0 and d = 1. In other words, the dependent equation can be written as follows:



dloani = ˛0 + ˛1 incomei∗ + ˛2 mastati + ˛3 Endowni

+ ˛8 noprojecti + ˛9 educationi + ˛10 religioni

The latent equation is given as follows: ∗

lead to an endogeneity bias, making the estimator inconsistent. Rivers and Vuong (1988), cited in Wooldridge (2001), provide a simple test to verify endogeneity in the case of a binary model. They suggest to model a continuous endogenous variable as a linear function of the exogenous variables and some instruments. Predicted values from this regression are then used in the second stage probit. Therefore, in order to find an instrumental variable for the income, Rivers and Vuong‘s approach is used. Finally, to test hypotheses in line with the previous section, we specify the demand of loan as a linear function of household characteristics including income, gender of household head, age of household head, number of household members, religion, education level of the household head, geographical location, and socio economic group of the household head, etc. The empirical model to be estimated is presented as follows:

(7)

di∗ = ıx2 + ε2 ∗

di = 1 if di > 0 and 0 otherwise Heckman (1990) has shown that selection bias can be overcome by including the inverse Mills ratio from the sample selection equation in the equation of interest. In this approach, the selection into the sample of those who demand credit is first modeled. Then, the inverse Mills ratio (lambda) from this regression is incorporated into the equation of interest. We also encounter a problem of endogeneity of some explanatory variables. In fact, information on interest rates which represent the prices of loan from each source is missing in the data set. Since prices and income are the key variables explaining the demand for credit, the noninclusion of the prices variable could create a correlation between the variable income and the error term. That could

(8)

dloan denotes the loan demand equation. income denotes total household income. Age denotes the age of household head, mastat denotes the marital status of the household head, endown denotes whether the household head has a house or not, and land or not, Socioecogrp denotes the categorical socio-economic group to which the household head belongs; male denotes the head household is male; hhsize denotes the number of persons in the household Area denotes the area where lives the household, noproject denotes the household does not plan to extend its activity; education denotes the level of the education of the household head; Religion denotes the religion of the household head, and εi1 denotes the error term assumed to be normally distributed. We specify also the choice of credit as a linear function of household characteristics including gender of household head, age of household head, number of household members, religion, education level of the household head, ethnic group, geographical location, and variables related to the credit contract, including time of repayment, loan size, type of activity funded, etc. formali = ˇ0 + ˇ1 incomei + ˇ2 assetindexi + ˇ3 malei + ˇ4 agei + ˇ5 agei2 + ˇ6 hhsizei + ˇ7 religioni + ˇ8 schoolingi + ˇ9 ethnici + ˇ10 areai + ˇ11 timrepaymi + ˇ12 loansizei + ˇ13 useloani + ˇ14  + ε2i

(9)

formal denotes the choice for formal sources. Assetindex represents a measure of wealth; schooling denotes whether the household head is illiterate or not, Ethnic denotes the household head ethnicity, Area denotes where lives the household lives, timrepaym denotes the time of repayment of the loan, loansize denotes the amount of loan demanded by the household head, useloan denotes the

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

purpose for which the loan has been taken,  represents the inverse Mills ratio fitted from loan demand equation, and εi2 denotes the error term assumed to be normally distributed. 5. Data The data used in this research is the households Living Standard Survey conducted in 2002 by the National Statistics Institute of Côte d’Ivoire. The research unit is the household and the people who live in it. The 2002 households Living Standards Survey is a nationwide, multitopic household survey with modules covering numerous aspects of living standards. The survey contains detailed information on households from all regions of the country. The household survey has 12 sections, gathering data on education, health and employment status of household members, household economic activities, income and expenditure, household size and housing, borrowing and lending activities. It covers 10,800 households living in Côte d’Ivoire. Out of 10,800 households surveyed, 1392 households have demanded a loan. They represent 12.88% of the overall sample. This sample is composed of those who have applied for and who have received the total amount, those who have received a part of amount of loan, and those whose application has been refused. Among those who have demanded a loan, 85.06% expressed a demand to informal sources. Table 2 gives a description of this subsample of households demanding a loan and presents the constructed variables’ summary statistics. In the appendix, Table A1 further specifies the definition and measurement of the variables. The demographic profile of the 1392 respondent household indicates that the average age of a household head is 42.59 years and about 84.63% of them fall in the economically active population (ages 18–59). The majority of household heads (53.59%) have no formal education. Approximately 19.40% and 22.49% of them have low and medium education respectively. Christians and adherents to non-traditional religion and those without religion constitute about 36.78% and 24.14% respectively, against 38.15% of Muslims. 30.68% of households live in other urban areas. The socio economic categories define the broad sector of employment of the household head. The statistics reveals that the majority of households heads work in the private services sector for the choice of for formal institutions, and the agricultural sector for informal loans. The majority of the household heads are married, which is a sign of household stability. Male-headed households constitute 82.33%. The average household size in the entire sample is 5.66 persons per household. Concerning the income, the average value of income for formal sources is inferior to those of the informal sources suggesting low income households prefer demanding loans from the formal sources. The asset variable is a combination of the asset data which are all dummy variables, indicating whether households own a particular asset or not. Principle component analysis has been used to create the asset index, to proxy wealth and capture ownership of tangible assets. The assets considered are consumer durables goods. Ownership of these assets determines the

479

choice of formal source since those assets could be used as collateral. Then, the formal institutions favour house owning households as is evident from the higher share of formal borrowing in the category of households owning a house. On the contrary, a land owning household is predominant for informal sources. Social capital or social networks play an important role in the Ivorian context. In line with the definition by Grootaert and van Bastelaer (2002a), social capital represents the institutions, relationships, attitudes, and values that govern interactions among people. Ethnic and religious backgrounds play an important role in these variables. Within these groups, potential borrowers share cultural similarities, facilitating access to credit to their members (Azam et al., 2001). From Table 2, we notice that Christians and Muslims are the religious backgrounds that are associated to a higher demand for loans. The Christian households are the ones who choose more informal sources for their loans applications, whereas the Muslim households go more towards formal institutions. At the ethnic group level, we notice that some ethnic groups like the informal sources more. It is mainly the case of Akan group, South Mande and other African ethnic groups. On the contrary, the Kru group and North Mande prefer formal sources whereas for the remaining groups the preference is not so clear. Considering the factors related to loan contracts, the time of loan repayment ranges from 1 to 75 months with mean about 2.74 months in the survey. The mean of formal credit time of loan repayment is 3.48 months, while the mean of informal credit is 2.61 months. The time of repayment varies from 1 to 36 months in the formal institutions. In the informal institutions, the time of repayment ranges from 1 to 75 months. The modal value of time of loan repayment for all types of sources is one month. The first objective of loan is to create revenue from an activity in order to improve living conditions of the household. Most designs of loans product in MFIs concern production purposes. Indeed, about 91.82% of total borrowing from the formal sources was for production purposes (agriculture and trade). Loans from informal sources were also mainly used for production. The statistics give about 90.87%. However, it has been noticed that agricultural activities remain the first activities for which each source is chosen. Loan demands for agricultural activities have, about a proportion of 58.17% and 60.81% respectively for formal and informal sources. That is due to the fact that the majority of households lives in rural areas, where informal sources are prefered. 6. Empirical results of borrowing sources As stated earlier, Heckman’s two-step approach is used to estimate the determinants of choosing a credit program. Before proceeding to the regression analysis, let us analyze the stated endogenous problem. As stated earlier, Rivers and Vuong’s approach allows making a simple test on the residuals from income regression. This test reveals that income is correlated with the error term in the demand equation. That means that there is endogeneity bias (Table A3). In order to deal with this problem, an

480

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

Table 2 Summary statistics. Variables Characteristics of the household head Age composition (%) Age of the household head Age (less than 18 years) Age (from 18 to 59 years) Age (+60 years) Gender (%) Male Female Marital status (%) Married Unmarried Other (divorced, separated, widow, widower) Education (%) No education Low education Medium education High education Area (%) Abidjan Other urban areas Eastern rural forest Western rural forest Rural savannah Socio-economic group (%) Agricultural worker Public services Private formal services Own business Other occupation Characteristics of the household Household size Income (in 1000 FCFA) Assets (between 0 and 1) Owning house (%) Owning land (%) Social capital variables Religion (%) Christian Muslim Other religion (traditional, other religions, no religion) Ethnicity (%) Akan Kru Mande south Mande north Voltaic Other African ethnic Characteristics of the loan Time of repayment (in months) Loan size (in FCFA, in log.) Purpose of loan (%) Trade activities Agricultural activities Transport activities Other activities

Full sample of loan demanding households

Demanding formal loan

Demanding informal loan

42.59 (14.46) 0.79 84.63 15.01

41.1 (13.68) 0.48 86.53 12.98

42.85 (14.58) 0.33 84.29 15.37

82.33 17.67

80.76 19.24

82.60 17.40

70.50 16.6 12.8

66.83 19.71 13.46

71.19 16.13 12.66

53.5 19.4 22.5 4.5

56.25 19.71 20.19 3.84

53.12 19.34 22.88 4.64

17.2 30.7 17.2 18.4 16.5

18.75 35.09 17.78 12.98 15.38

16.89 29.89 17.06 19.42 16.72

22.9 5.4 20.2 16.7 17.0

18.75 4.32 22.59 15.86 17.78

23.65 5.57 19.76 16.89 16.89

5.66 (4.01) 60.4049 (141.891) .0442 49.14 54.38

4.88 (3.46) 37.8188 (53.579) .2299 50 45.67

5.35 (3.87) 64.3727 (151.873) .0115 48.98 55.91

39.22 36.64 24.14

35.57 41.82 22.59

39.86 35.72 24.41

29.74 14.15 8.91 12.43 13.58 21.19

27.88 15.38 5.76 16.82 13.94 20.19

30.06 13.96 9.45 11.65 13.51 21.36

2.74 (4.12) 11.507 (1.857)

3.48 (5.13) 12.74 (1.805)

2.61 (3.87) 11.29 (1.781)

30.6 60.42 0.43 8.55

33.65 58.17 0.48 7.69

30.06 60.81 0.42 8.69

Source: own computation from 2002 INS Survey Data.

instrumental variable is used. This is the predicted value of income from the income regression. Concerning the demand of loan equation, the convectional Wald test statistic is significant at 1%. It rejects the null hypothesis that all coefficients are zero. Knowing the sign of the parameter is enough to determine whether the variable has a positive or negative effect on the demand equation.

The following variables have been found relevant to explain the demand for a loan: income, owning land, having an own business, the household size, having no development project, and geographical location. The effect of income is positive and significant for demanding a loan. That demonstrates that a household demands a loan when its income is higher. Owning land increases the probability of demanding a loan. The explanation is

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

481

Table 3 Probit estimation of the demand of loan. Dependent variable: demand of loan. Explanatory variable

Coefficients

Standard error

Marginal effects

Standard error

Predicted Income Matrimonial status Married Unmarried Endowment House Land Socioeconomic group Agricultural worker Public services Private formal services Own business Male Household size Area Other urban areas Eastern rural forest Western rural forest Rural savannah No project Education Low education Medium education Higher education Religion Christian Muslim Age Age less or equal 24 years Age between 25 and 39 Age between 40 and 59 Constant

1.11e−07**

4.73e−08

1.39e−08

.000

−.041 .0077

.0622 .0724

−.0052 .0009

.008 .0091

.0561 .2012***

.0391 0392

.0071 .0255

.005 .0057

−.060 .0517 −.0185 −.1401*** .0112 .0186***

.0466 .0879 .0517 .0511 .0531 .0052

−.0073 .0067 −.0023 −.0163 .0014 .0023

.0056 .0118 .0064 .0057 .0066 .0007

.1569*** .1008* .0638 .1003 −2.464***

.0498 .0605 .0613 .0643 .3017

.0207 .0132 .0082 .0132 −.164

.0073 .0084 .0082 .0089 .004

.0091 −.0114 −.0805

.0462 .0484 .0901

.0011 −.0014 .0095

.0058 .006 .0101

.0365 −.0323

.0447 .0451

.0046 −.004

.0057 .0056

.0338 −.0267 −.029 −1.204***

.0834 .0569 .0530 .0848

.0043 −.0033 −.0036

.011 .0071 .0065

=10,800 =160.22 =−3729.5996

Prob > Chi2 Pseudo R2

=.000 =.1013

Number of obs. Wald Chi2 (25) Pseudo log likelihood Note: z denotes z-statistics. *** Significant at 1%. ** Significant at 5%. * Significant at 10%.

that the land could be used as collateral by the household. Regarding the socio-economic groups to which the household head belongs, it can be seen that the head of household who is doing its own business is least likely to demand a loan. This negative effect is explained by the fact take the loans from these sources require the provision of business registration, procedure enterprises, or complex tax procedure and the collection of public revenue, documents they do not have sometimes and necessitate money for their establishments. The household size has a significantly positive effect on the probability of borrowing. A greater number of household members imply higher expenses. Most of the time, the budget cannot cover the expenses of the all members of the household. Therefore, in order to smooth their consumption, households have to borrow. In addition, the fact that the head of the household has no development project for his activities has a negative effect on the probability of demanding a loan. Marital status, education, religion and age of the household head do not appear to significantly influence the probability that the household is demanding a loan (Table 3).

In the second step of the estimation, we try to determine the factors influencing the choice of formal versus informal sources. Again, the conventional Wald test statistic is significant at 1%, rejecting the null hypothesis that all coefficients are zero. The predicted probability of choosing formal source is 11.56%. That confirms the fact there is preference for informal sources in Côte d’Ivoire (Azam et al., 2001). The following variables have been found relevant to explain the choice of formal source. Income has a negative effect on the likelihood that a formal source is chosen. Thus, informal sources of credit are preferred by low income groups. This is one of the main variables in our analysis, and we take this result as an indication microcredit may help solve some part of the market failure in the financial sector in Côte d’Ivoire. Low income households who will not be able to get a loan in the formal sector do get access to credit through the informal sector. This is particularly relevant in combination with the result for the loan size. The size of the loan is positively and significantly related to the probability of choosing the formal credit program. Indeed, the probit results display a positive and significant effect of the loan size for the choice of formal source. That supports the

482

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

Table 4 Probit estimation of borrowing from formal sources. Dependent variable: formal source choice. Explanatory variable

Coefficients

Standard error

Marginal effects

Standard error

Income Assetindex Male Age Age squared Household size Schooling Religion Christian Muslim Area Other urban areas Eastern rural forest Western rural forest Rural savannah Ethnic Akan Krou Mande north Mande south Voltaic Time of repayment Loan size Use of loan Trade activities Agricultural activities Inverse Mills ratio Constant

−2.68e−06*** .229*** .012 .032* −.0003* −.001 −.048

6.69e−07 .075 .122 .0187 .0002 .0148 .107

−5.22e−07 .0447 .0022 .006 −.000 −.0002 −.009

.0000 .0149 .0236 .0036 .00004 .0029 .0208

−.079 .184

.121 .158

−.015 .037

.023 .033

−.188 −.256 −.402** −.381**

.141 .162 .174 .170

−.035 −.045 −.067 −.063

.025 .025 .024 .023

.208 .371** .241 −.030 .112 .011 .224***

.156 .183 .163 .213 .163 .0108 .026

.042 .084 .052 −.005 .022 .002 .043

.033 .047 .038 .040 .035 .002 .005

.099 .273* .905 −4.591

.165 .153 1.458 .659

.019 .051 .176

.033 .028 .283

Number of obs. Wald Chi2 (22) Pseudo log likelihood

=1392 =126.84 =−511.94384

Prob > Chi2 Pseudo R2

=.000 =.1279

*** ** *

Significant at 1%. Significant at 5%. Significant at 10%.

presumption in the literature that informal institutions are far more effective at financing small borrowers than the formal institutions. Therefore, when the loan size is larger the household will choose formal sources. The higher transaction costs that are typically related to borrowing from formal sources may in the case of larger loan sizes become relatively smaller. A 100% increase of the loan size is associated with an increase in the probability of choosing the formal sources of 4.37%. The importance of personal wealth is confirmed here by the positive effect of the variable asset index, proxy of wealth, on the choice of the formal sources. In addition, the age of the household head raises the probability of borrowing from formal sources. Compared to other ethnic groups, the Krou are most likely to choose formal sources for borrowing. Indeed, the probit results display a positive and significant sign, as compared to the reference group of other ethnic group. The marginal effect is an increase of 8.4%. The ethnic network plays an important role in relations between the households in Côte d’Ivoire through the imposition of social sanctions for misconduct of a member of the network. But it also imposes the respect of strong kinship ties requiring acute sense of forgiveness for the person that went wrong has. Therefore, depending on whether the household comes from an ethnic group where social sanctions

are higher, the use of formal loans will be chosen for fear of losing its reputation. That explains why the households from the ethnic groups as Kru and Akan, North Mande or Voltaic prefer the formal sources. The effect when the household head lives in western rural forest and rural savannah comparatively to Abidjan is negative on the probability of borrowing from formal source. In fact, households living in such regions have difficulties to access to facilities, higher transaction costs and as such they do not choose formal institutions. This result is in line with the unequal distribution of MFIs on the whole territory, as stated by the national commission of microfinance (NCM, 2005). The magnitude of these effects when the household head lives in western rural forest and rural savannah are a decrease of 6.7% and 6.3% respectively. This result corroborates Guirkinger’s (2008) finding and Swain’s (2002) finding that an area where the borrower lives has an impact on the choice of source of borrowing. The purpose of the loan, such as to finance trade activities or agricultural activities, shows that a positive effect is observed for the choice of formal sources when it concerns agricultural loans, as compared to other activities. Loans for agricultural activities are 5.16% more likely to be taken from formal sources, corroborating the results by Nguyen (2007) (Table 4).

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

7. Conclusion Understanding the socioeconomic factors influencing the determinants of households’ choice of microfinance credit program is useful for future policy designs, as it can give an indication of how well the credit market performs its function. Formal and informal sources of credit co-exist in the market, with informal sources providing lower transaction costs. Because of information asymmetries, credit rationing may exist in the market, and transaction costs may be high. This may especially affect low income households, who may have a higher preference for small loans and thus face disproportionately high transaction costs in the formal financial sector. This paper attempted to clarify the role of microcredit in solving market failure by testing the impact of three variables on the source (formal or informal) of credit. These variables were the household income, the size of the loan, and social capital, which was measured by ethnic background and religion. It was indeed found that low income households tend to prefer informal sources of credit, and smaller loans are also performed through the informal sector. This confirms our hypothesis that informal loans through microcredit institutions have a particular role to play in the financial sector in Côte d’Ivoire. We also found that the choice of source of credit depends in ethnic networks, although religious backgrounds did not matter for the source of credit.

483

Thus, we conclude that microfinance institutions have an important role to play in the economy of developing countries. In the descriptive work that preceded the statistical analysis, it was also found that microfinance institutions in Côte d’Ivoire collect a much larger amount in savings than what they give out in loans. This is an interesting feature that demands further theoretical and empirical work to clarify the exact role of microfinance in developing countries. Appendix A. Appendix A.1. Estimation of income Consider the income model: incomei = ˇ0 + ˇ1 income2i + ˇ2 texpi + ˇ3 typhi + ˇ4 mstai + ˇ5 ostai + ˇ6 sgrpi + ˇ7 gendi + ˇ8 areai + ˇ9 landowneri + ˇ10 noprojecti + ˇ11 Educi + ˇ12 religi + ˇ13 agei + εi

(A.1)

where income2, secondary income from other activities (credit, remittance, etc.); texp, total expenditure of the head of household; typh, type of house captured by a dummy variable: =1 if villa; =1 if set of house; =1 if detached house; ostat, occupation status captured by a dummy: =1 if owner; =1 if rented; =1 if other occupation status (Tables A1–A3).

Table A1 Reports of the descriptive statistics (the means and standard deviations) of the surveyed households. Variable Source Formal Assetindex Gender Demand of loan Education No education Low education Medium education High education Schooling Socioeconomic group Agriculture Public service Private formal service Own business Other occupation Area Abidjan Other urban areas Eastern rural Western rural Savannah Ethnic group Akan Kru Mande south Mande North Voltaic Other African ethnic

Description

Mean

St. dev.

=1 if the household borrows from formal sources =a proxy of wealth =1 if head of household is male =1 if the household answer he has been borrowed

.149 −4.29e−09 .823 .1288

.356 .631 .381 .335

=1 if head of household has no education =1 if head of household has some primary schooling =1 if head of household finished primary schooling or continued to secondary school =1 if head of household completed secondary or higher. =1 if the head household head has some literacy level

.535 .194 .225

.498 .395 .417

.045 .448

.208 .497

=1 if head of household is employed in Agriculture =1 if head of household is employed in public service =1 if head of household is employed in private formal service =1 if head of household is doing its own business =1 if head of household is doing other occupation

.229 .054 .202 .167 .170

.420 .226 .401 .373 .375

=1 if household lives in Abidjan =1 if household lives in the other urban areas =1 if household lives in rural eastern forest =1 if household lives in rural western forest =1 if household lives in rural savannah

.172 .306 .171 .184 .165

.377 .461 .377 .388 .371

=1 if the head of household has a native tongue Akan =1 if the head of household has a native tongue Kru =1 if the head of household has a native tongue South Mande =1 if the head of household has a native tongue North Mande =1 if the head of household has a native tongue voltaic =1 if the head of household has a native tongue other African ethnic

.292 .137 .088 .131 .138 .212

.454 .344 .283 .337 .345 .409

484

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

Table A1 (Continued) Variable

Description

Mean

St. dev.

Time of repayment Loan size Matrimonial status Married Unmarried Other matrimonial status Religion Christians Muslim Other religions Household size Age Income Type of house Villa Set of house Detached house Other (hut, shack) Occupation status of house Owner Rented Other occupation No project Use of loan Trade activities

the time of repayment of the loan The amount of loan demanded by the household head

2.676 1,161,877

4.119 6,790,604

=1 if head of household is married =1 if head of household is no married =1 if others (separated, widow)

.705 .166 .128

.456 .372 .334

=1 if head of household is Christian =1 if head of household is Muslim =1 if head of household is other religion Number of household members Age of head household Total income of household head (FCFA/month)

.367 .381 .109 5.664 42.59 60,404.9

.482 .486 .313 4.009 14.461 141,891.7

=1 if household lives in a villa, apartment =1 if household lives in a set of house =1 if household lives in a detached house =1 if household lives in an hut, a shack

.123 .506 .146 .224

.328 .500 .353 .417

=1 if household owns the house =1 if household rents the house =1 for other occupation =1 if the head of household does not plan to extend his activity

.489 .340 .170 .038

.500 .474 .225 .193

=1 if the head of household demanded the loan for the trade activities =1 if the head of household demanded the loan for the agricultural activities =1 if the head of household demanded the loan for the transport activities =1 if the head of household demanded the loan for the other activities

.296

.456

.622

.481

.003

.057

.078

.268

Agricultural activities Transport activities Other activities

Source: Own computation from the INS Survey 2002.

Table A2 (Continued)

Table A2 Estimation results of Income. Dependent variable: income.

Explanatory variables

Explanatory variables

Coefficients

Standard error

P>t

Income2 Expenditure Type of house Villa Set of house Detached house Matrimonial status Married Unmarried Occupation status Owner Rented Socioeconomic group Agriculture Public business Private service Own service Gender Household size Area Other urban areas Eastern rural forest Western rural forest Rural savannah Landowner No project

1.0026 1.072

.0033 .2391

.000 .000

20,938.15 −7325.27 985.32

6936.23 3537.16 3718.51

.003 .038 .791

12,586.63 −15,960.12

3287.22 4724.95

.000 .001

4143.46 1906.61

3476.11 5510.41

.233 .729

22,632.5 103,113.6 78,475.86 62,440.79 14,670.1 6182.55

3548.63 12,530.24 5040.45 5484.64 3067.02 1088.39

.000 .000 .000 .000 .000 .000

−7531.69 4676.04 −8152.87 −11,501 3829.54 8003.81

7906.35 8782.74 8798.97 7618.77 3455.02 4238.71

.341 .594 .354 .131 .268 .059

Education Low education Medium education Higher education Religion Christian Muslim Age Age less or equal 24 years Age between 25 and 39 years Age between 40 and 59 years Constant Number of obs. F(29, 10,770)

Coefficients

Standard error

P>t

−2813.69 7907.56 103,143.8

3709.56 5912.28 13,726.18

.448 .181 .000

−4380.94 −4655.13

4001.46 3638.96

.274 .201

5545.83

7388.89

.453

−1853.89

6028.51

.758

16,557.58

5924.59

−75,870.16

16,887.18

.000

=10,800 =3949.173

Prob > F R2

.000 .6228

005

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486 Table A3 River and Vuong’s endogeneity test. Dependent variable: demand of loan. Explanatory variable

Coefficients

Standard error

P>t

Income Matrimonial status Married Unmarried Houseowner Landowner Socioeconomic group Agriculture Public business Private service Own service Gender Household size Area Other urban areas Eastern rural forest Western rural forest Rural savannah No project Education Low education Medium education Higher education Religion Christian Muslim Age Age less or equal 24 years Age between 25 and 39 Age between 40 and 59 Residuals fitted Constant

−1.84e−07

1.20e−07

.127

−.0419 .0078 .0561 .2014***

.0621 .0724 .0391 .0391

.500 .914 .151 .000

−.0605 .0509 −.0206 −.1410*** .0111 .0186***

.0466 .0878 .0517 .0511 .0531 .0052

.195 .562 .690 .006 .833 .000

.1578*** .101* .064* .101 −2.4633***

.0498 .0605 .0613 .064 .300

.002 .092 .295 .116 .000

.0089 −.0115 −.0857

.0462 .0484 .0900

.846 .811 .341

.0361 −.0321

.0447 .04517

.418 .477

.0339 −.0256 −.0294 2.91e−07** −1.203***

.0834 .0569 .0530 1.29e−07 .0847

.684 .653 .579 .024 .000

Number of obs. Wald Chi2 (25) Pseudo log likelihood

=10,00 =163.00 =−3728.5969

Prob > Chi2 Pseudo R2

.000 .1016

Note: z denotes z-statistics. *** Significant at 1%. ** Significant at 5%. * Significant at 10%.

References Aryeetey, E., 1996b. Rural finance in Africa: institutional developments and access for the poor. In: The World Bank Annual Conference on Development Economics, Washington, DC, 25–26 April. Atieno, R., 2001. ‘Formal and Informal Institutions’ Lending Policies and Access to Credit by Small-Scale Enterprises in Kenya: An Empirical Assessment, AERC Research Paper 111. Azam, J.P., Biais, B., Dia, M., Maurel, C., 2001. Informal and formal credit markets and credit rationing in Cote d’Ivoire. Oxford Review of Economic Policy 17 (4). BCEAO, 2005. Situation microfinance. Monographie Côte d’Ivoire. BCEAO, 2004. Situation microfinance. Monographie Côte d’Ivoire. BCEAO, 2003. Situation microfinance. Monographie Côte d’Ivoire. BCEAO, 2002. Situation microfinance. Monographie Côte d’Ivoire. BCEAO, 2001. Situation microfinance. Monographie Côte d’Ivoire. BCEAO, 2000. Situation microfinance. Monographie Côte d’Ivoire. Besley, T., Burgess, R., 2001. The Political Economy of Government Responsiveness: Theory and Evidence from India, CEPR Discussion Paper No. 2721. Centre for Economic Policy Research, London. Besley, T., Coate, S., 1995. Group lending, repayment incentives and social collateral. Journal of Development Economics, 46. Biggs, T., Raturi, M., Srivastava, P., 2002. Ethnic networks and access to credit: evidence from the manufacturing sector in Kenya. Journal of Economic Behavior and Organization, 49. Brau, J.C., Woller, G., 2004. Microfinance: a comprehensive review of the existing literature. Journal of Entrepreneurial Finance and Business Ventures 9 (1).

485

Chung, I., 1995. Market choice and effective demand for credit: the roles of borrower transaction costs and rationing constraints. Journal of Economic Development 20 (2). Crook, J., 2006. The demand and supply for household debt: a cross country comparison. In: Bertola, G., Disney, R., Grant, C. (Eds.), The Economics of Consumer Credit. MIT Press, Cambridge. Diagne, A., Zeller, M., 2001. Access to Credit and Its Impact on Welfare in Malawi. International Food Policy Research Institute, Washington, DC, Research Report 116. Duong, P.B., Izumida, Y., 2002. Rural Development finance in Vietnam: microeconometrics analysis of households surveys. World Development 30 (2). Fafchamps, M., 2000. Ethnicity and credit in African manufacturing. Journal of Development Economics 61 (1). Fafchamps, M., 2003. Ethnicity and networks in African trade. Contributions to Economic Analysis & Policy 2 (1). Fisman, R., 2003. Ethnic ties and the provision of credit: relationship-level evidence from African firms. Advances in Economic Analysis & Policy 3 (1). Fox, J., 1996. How does civil society thicken? The political construction of social capital in rural Mexico. World Development 24 (6). Gomez, R., Santor, E., 2001. Membership has its privileges: the effect of social capital and neighbourhood characteristics on the earnings of microfinance borrowers. The Canadian Journal of Economics 34 (4). Guirkinger, C., 2008. Understanding the coexistence of formal and informal credit markets in Piura, Peru. World Development 36 (8). Guirkinger, C., Boucher, S.R., 2007. Risk, wealth and sectoral choice in rural credit markets. American Journal of Agricultural Economics 89 (4). Grootaert, C., van Bastelaer, T. (Eds.), 2002a. The Role of Social Capital in Development: An Empirical Assessment. Cambridge University Press. Heckman, J., 1990. Varieties of selection bias. American Economic Review 80. INS, 2002. Enquête Niveau De Vie Des Ménages. Kochar, A., 1992. An Empirical Investigation of Rationing Constraints in Rural Credit Markets in India. Stanford University, Mimeo. Maddala, G.S., 1983. Limited Dependent and Qualitative Variables in Econometrics. Cambridge University Press, Cambridge. Madestam, A., 2007. Informal Finance: A Theory of Moneylenders, Working Paper. Bocconi University, Milan, Italy. Matin, I., Hulme, D., Rutherford, S., 2002. Finance for the poor: from microcredit to microfinancial services. Journal of International Development, 14. McFadden, D., 1974a. Conditional logit analysis of qualitative choice behavior. In: Zarembka, P. (Ed.), Frontiers of Econometrics. Academic Press. Meyer, R., 2002. The demand for flexible microfinance products: lessons from Bangladesh. Journal of International Development 14 (3). Microfinance Information Exchange (MIX) et Groupe Consultatif d’Assistance aux Pauvres (CGAP), 2010. Afrique subsaharienne 2009: benchmarking et analyse du secteur de la microfinance. Rapport. Ministère de l’Economie et des Finances (MEF), 2009. Microfinance en Côte d’Ivoire: Un secteur en plein essor. Le trésorier 24. Mushinski, D.W., 1999. An analysis of offer functions of banks and credit unions in Guatemala. Journal of Development Studies, 36. Nagarajan, G., Meyer, R., Hushak, L.J., 1995. Demand for agricultural loans: a theoretical and econometric analysis of the Philippine credit market. Paper presented at the AAEA Annual Meeting, Indianapolis, IN, 6–9 August 1995. National Commission of the Microfinance (NCM), 2005. Annual Report. National Commission of the Microfinance (NCM), 2004. Annual Report. Nguyen, C.H., 2007. Determinants of Credit Participation and Its Impact on Household Consumption: Evidence from Rural Vietnam, Discussion paper 2007/03. Nguyen, T.T.P., 2006. Factors influencing access to credit of households in rural areas of Vietnam case study of Tan Linh Commune, Ba Vi District, Ha Tay Province. Master Thesis in Rural Development with Specialization in Livelihoods and Natural Resource Management Swedish University of Agricultural Sciences. Nimal, A.F., 2008. Managing Microfinance Risks Some Observations and Suggestions. Asian Development Bank. Okten, C., Osili, U.O., 2004. Social networks and credit access in Indonesia. World Development 32 (7). Okurut, N., Schoombee, A., van der Berg, S., 2004. Credit demand and credit rationing in the informal financial sector in Uganda. Paper to the DPRUT/Tips/Cornell conference on African development and Poverty reduction: The Macro–Micro Linkages. Rankin, K.N., 2002. Social capital, microfinance, and the politics of development. Feminist Economics 8 (1).

486

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473–486

Soulama, S., 2005. Micro-finance, pauvreté et developpement. des Archives contemporaines et en partenariat avec l’Agence Universitaire de la Francophonie. Schmidt, R.H., Kropp, E., 1987. Rural Finance Guiding Principles. GTZ, Eschborn. Stiglitz, J., Weiss, A., 1981. Credit rationing in markets with imperfect information. American Economic Review 71 (3). Swain, R.B., 2002. Credit rationing in rural India. Journal of Economic Development 27 (2). van De Ven, W.P.M.M., van Praag, B.M.S., 1981. The demand of deductibles in private health insurance: a probit model with sample selection. Journal of Econometrics, 17.

Woolcock, M., Narayan, D., 2000. Social capital: implications for development theory, research, and policy. World Bank Research Observer 15 (2). Wooldridge, J.M., 2001. Econometric analysis of Cross Section and Panel Data, 1st ed. The MIT Press. Yaron, J., Gurgand, M., Pederson, G., 1994. Outreach and Sustainability of Six Rural Finance Institutions in Sub-Saharan Africa, World Bank Discussion Papers 248. Zeller, M., 1994. Determinants of credit rationing: a study of informal lenders and formal credit groups in Madagascar. World Development 22 (1).