ME

ED

a

ARRAA

KMSAH

1

caasvwtmnac2mpasm

0h

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

Contents lists available at SciVerse ScienceDirect

Structural Change and Economic Dynamics

jou rna l h omepa g e: www.elsev ier .com/ locate /sced

icrofinance and households access to credit:vidence from Côte d’Ivoire

dith Leadaut Togba ∗

epartment of Economics and Management, University of Cocody-Abidjan, PO Box 2761, Abidjan 04, Côte d’Ivoire

r t i c l e i n f o

rticle history:eceived January 2010eceived in revised form August 2012ccepted August 2012vailable online 19 August 2012

a b s t r a c t

Evidence on microfinance services these days ironically shows a great preference for savingsproducts rather than credit products by households. For some authors, this phenomenon isexplained by the fact that microfinance products, and especially loans, from formal micro-finance institutions do not fit the households demand. This paper first presents evidence

eywords:icrofinance

ocial capitalccess to crediteckman two steps

on the observed phenomenon in the Ivorian microfinance sector. Second, it analyses theIvorian credit market so as to understand the determinants of the choice for credits fromformal sources versus informal sources. The results reveal the size of the loan, agriculturalpurpose, the geographical area where households live and ethnicity as factors influencingthe choice for formal sources.

© 2012 Elsevier B.V. All rights reserved.

. Introduction

Microfinance is made up of a set of small size finan-ial products which include savings, credit and insurance,nd which suit people with low incomes who weret first excluded from the classic or formal bankingystem (Soulama, 2005). Microfinance products are pro-ided by the intermediation of a multitude of institutionshich vary according to size, the degree of organiza-

ion and the legal status, and include NGOs, associations,utual insurance companies/cooperatives, limited compa-

ies, banks, financial institutions, but also more informalnd unregulated institutions such as tontines, usurers,hange keepers, loans between friends, etc. (Soulama,005, www.lamicrofinance.org). For some years now,icrofinance has been found a vital tool to eradicate

overty among the vulnerable by the provision of products

nd banking services similar to those delivered by clas-ic institutions (Brau and Woller, 2004). Observations wereade on the growth of these institutions all over the world.

∗ Tel.: +225 04 44 63 48.E-mail address: togbedith@yahoo.fr

954-349X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.strueco.2012.08.002

In fact, according to the report of the 2012 campaign pub-lished by the microcredit summit on December 31, 2010,there existed 3652 institutions who took care of about 200million customers.

In spite of their recent development and strong prox-imity to the poor, it is important to note that formalmicrofinance institutions only bring a very partial answerto the poor households’ need for finance. Nimal (2008)reveals that although microfinance institutions are presentin Asia and the pacific regions, more than 300 millionhouseholds suffer from the lack of access to financial prod-ucts offered by the formal and semi-formal sector. Insub-Saharan Africa in 2008, the number of active borrowersand savers in percent of population living under nationalpoverty line has reached a rate of penetration of about3% for credits and 5% for savings (MIX and CGAP, 2010).These rates confirm the irony observed around microfi-nance today which is the preference of households forsavings products, in comparison to loans (Meyer, 2002).

Yet, credit is important to finance the start of activi-

ties and allows fixed capital investments in poor rural andurban economies where it is difficult to save, as confirmedby Guirkinger (2008) for Peruvian rural zones. Facing theseconstraints, lots of households continue to rely on informal

Econom

474 E.L. Togba / Structural Change and

sources of finance to increase their production capacity,to diversify risk, and to smooth their consumption duringtheir life cycle. Informal finance refers to all transactions,loans and deposits occurring outside the regulation ofa central monetary authority (Atieno, 2001). Besley andBurgess (2001) reported that 82% of loans are contractedin the informal sector against only 12% for the formal sec-tor in Nepal. In a survey on the credit market in Côted’Ivoire, Azam et al. (2001) noted that small entrepreneursin Côte d’Ivoire prefer to borrow from parents andneighbours.

Therefore, there seems to be a need to understand thebehavior and preferences of the poor in terms of finan-cial services (Matin et al., 2002). Given the often informalnature of the financial services used by the poor, such ananalysis requires integrating the presence of social net-works and social capital,1 not only as a source of choicesof credits but also as an explanatory factor of the choice.In developing countries, networks are made up of thecommunity and ethnic groups. The first network gener-ally constitutes a source for competitive credit. Consideringethnic network analysis, studies on their interaction withthe credit market are scarce. Some research documentsexamine the interaction between ethnic groups, credit andenterprises in Africa but they do not explore the diver-sity 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 indigenousdiversity. This paper tries to fill this gap, by includingethnic diversity in the indigenous African population inthe analysis of preferences and use of financial instru-ments. The paper focuses in the first place on estimatinga model that explains what determines whether or nothouseholds in Côte d’Ivoire make use of loans. Next thepaper tries to determine the factors influencing house-holds’ decision to obtain loans from informal versus formalsources.

The structure of the paper is as follows: in the next sec-tion, we introduce the microfinance sector in Côte d’Ivoireby describing the characteristics of the microfinance insti-tutions concerned. Section 3 provides a short literaturereview which is the basis for the empirical analysis of thechoice for particular financial sources. Section 4 deals withthe methodology, and Section 5 gives basic information onthe data which were used. Finally, Section 6 presents theresults of the study, and the final section summarizes theargument.

1 Social capital refers to the norms and networks that enable peopleto act collectively (Woolcock and Narayan, 2000). Fox (1996) definessocial capital as a social organization, relationship of cooperation andreciprocity, networks and leadership that facilitate collective action. How-ever, according to the level considered, the definition differs. Social capitalcould be the degree of trust in government or other societal institutions(Fukuyama, 1995 cited in Okten and Osili, 2004), social cohesion, reci-procity and institutional effectiveness. Or as stated by Grootaert and vanBastelaer (2002a), social capital is broadly the institutions, the relation-ships, the attitudes, and values that govern interactions among people andcontribute to economic and social development. This definition depictsclosely the developing countries’ situation.

ic Dynamics 23 (2012) 473– 486

2. The microfinance sector in Côte d’Ivoire

Côte d’Ivoire is classified 164th of 177 countries inthe Human Development index with 49% of the 21.1 mil-lion citizens living under the poverty line (World Bank,2009). According to same source, poverty is most severein the savanna of the north (54.6%) and the rural forestof the East (46.6%), followed by the urban regions (33.8%)apart of Abidjan, the western rural forest (24.5%) and Abid-jan (11.1%). Considering the poverty of its population, thegovernment of Côte d’Ivoire decided to make poverty alle-viation a priority in its socio-economic programs. Manystrategies were defined to achieve this goal. The creationof microfinance institutions for households was one of thestrategies. A microfinance institution (MFI) is an organiza-tion that provides financial services to the poor. This verybroad definition includes a wide range of providers thatvary in their legal structure, mission, and methodology.However, all share the common characteristic of provid-ing financial services to clients who are poorer and morevulnerable than traditional bank clients.

Microfinance institutions are now observed in all sec-tors of the economy where long term financing is not apriority. In Côte d’Ivoire, the available alternatives for ahousehold include banks and financial institutions, com-panies of framing, credit unions, social funds, ROSCAs,moneylenders, cooperatives and other.

Formal providers are defined as those that are sub-ject not only to general laws but also to specific bankingregulation and supervision (development banks, savingsand postal banks, commercial banks, and non-bank finan-cial institutions, credit unions). Semi-formal providersmay also be registered entities subject to general andcommercial laws but not under bank regulation andsupervision (companies of framing and social funds). Infor-mal 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 institutionsaims at protecting customers and allowing access throughlimiting the price for credit, which is lower than those prac-ticed by non-regulated institutions. It also aims at securingfinancial operations by requesting to respect managerialnorms like prudency and demanding for operational auton-omy. The effects of this regulation are felt at the levelof access or demand for credits and financial productsbecause conditions for borrowing from these institutionsmay become difficult to fulfill by households. Indeed, thisis displayed in the form of prescribed minimum loanamounts, complicated application procedures and restric-tions on credit for specific purposes (Schmidt and Kropp,1987). On the contrary, the service from informal sourcesis based on flexible arrangements to adjust to changingeconomic circumstances, and on reducing the transactioncosts to borrowers, who respond by maintaining disci-pline in order to sustain their access to credit (Atieno,2001). Unlike formal sources, informal lenders often attach

more importance to loan screening than to monitoringthe use of credit. Screening practices often include groupobservation of individual habits, personal knowledge by

E.L. Togba / Structural Change and Econom

Table 1Distribution of formal MFIs’ funding by economic sectors in Côte d’Ivoire.

Sectors 2004 (%) 2005 (%)

Trade 39.08 30Crafts industry 0.62 13.5Agriculture 9.32 10.1Construction and housing 18.15 0.1Industry 0.03 0.0Transportation 0.10 5.7Catering industry 6.25 0.1Education and health 21.00 1.4

S

ictcrm

evtriteaireledi

eiw2aota

iit(btfmmotrtmT

and Mushinski (1999) showed that heavy transaction costs

Other sectors 5.45 39.2

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

ndividual moneylenders, recommendations by others andreditworthiness. Therefore, low income households haveo choose between borrowing from formal sources, whereredit is cheaper, but where their loan application is usuallyejected, or resorting to informal sources where funding isuch more expensive.The microfinance sector is not only of use to informal

conomic activities, but also extends its area to civil ser-ants, and private sector employees. Statistics provided byhe National Commission of Microfinance (NCM) (2005)eveal that a large share of the formal microfinance fundss used to finance economic activities outside agricul-ure. However, it can be noticed that in a country whoseconomies are founded on agriculture, just little funds arellocated to this sector. Statistics show that in 2005, trad-ng activities received more funds than any other sector,epresenting about 30% of the funds. Consumer credits forducation and health represent 1.4% in 2005. There is aot of variation over time: the construction and housing,ducation and health and catering industries show a sharpecline in 2005. Yet, sectors like crafts and transportation

ndustries have observed a relative increase (Table 1).The supply of financial services from formal MFIs is

ssentially centered on savings and credits. It is clearly anmportant industry in terms of mobilizing savings. Savings

ere estimated to be 88,679 million FCFA in 2008 (MEF,009). Compared to these savings, 33,013 million FCFA, orbout 37%, were supplied as outstanding discounted billsf credit by formal MFIs. Fig. 1 shows these proportions forhe period 1998–2005, revealing a rapid growth in savingst the detriment of credits in MFIs.

These data reveal a great interest by the populationn products proposed by MFIs, but especially so in sav-ngs. Two factors might explain this situation. The first ishat there is effectively a stronger preference for savingse.g., as observed by Meyer, 2002). This preference coulde explained by the need for secured savings even thoughhe interest rate is low (Yaron et al., 1994). The secondactor is based on the notion of transaction costs and infor-

ation asymmetry between contracting parties. Since theicrofinance institution is incapable to control all actions

f borrowers due to incomplete and expensive informa-ion, it will formulate terms of contracts that attract lessisky borrowers and will be in favor of the MFI. Transac-

ion costs incurred in obtaining credits are then considered

ore important than the utility derived (Atieno, 2001).herefore there exist significant obstacles in transforming

ic Dynamics 23 (2012) 473– 486 475

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 forcredit.

It could thus be possible that the great increases in sav-ings are due to the terms of concession or conditions ofobtaining credit. The initial requirements to get a loan froma MFI are: (i) a minimal deposit is required to those request-ing a loan; (ii) adherence of at least six months with regularsavings is required; (iii) a member has the right to demandfor a loan equivalent to twice his initial deposit: (iv) othermembers to be considered as guarantors must have suffi-cient funds in his/her savings account to cover the amountof loan requested. (v) Sometimes, the need for a guaran-tor who is not a member of the microfinance institution isrecommended. All these conditions are operationalized ina set of documents whose filling is difficult for householdswho are mostly financial illiterate. In addition, loan con-tracts include specifications regarding the interest rate, theamount of the loan, the modalities of repayment, and theguarantees, creating possibly additional obstacles to formalcredit.

The set of conditions and formalities constitute a mainfactor which creates differences between microfinanceinstitutions. They also give orientation for the choice ofthe households. In effect, the conditions and formalitiesrequired by the formal microfinance institutions are notnecessary as far as informal institutions are concerned.Generally, for the informal institutions, it is necessary tohave objects to serve as guarantee when the money bor-rowed is not given back. Besides the absence of an explicitcontract and more flexible terms of repayment (i.e., a pos-sibility of spreading the term) lead households to moreoften look for credit with informal microfinance institu-tions. This suggests that the household is rational in sensethat it makes choices that maximize its direct utility subjectto constraint on expenditure.

In summary, it seems obvious that there is a big prefer-ence for savings in formal MFIs, but this could be explainedby the barriers to getting access to formal credits. Thiscould also explain the coexistence of informal credit mar-kets with formal credit establishments. If this is the case,then which factors determines the choice between the twosources? In the next section, we will investigate the liter-ature on the determinants of access to formal credit, andtest them on Ivorian data in the subsequent section.

3. Literature review

Credit rationing theoricians give an explanation of theborrowers’ choice between formal and informal sources.In fact, according to Stiglitz and Weiss (1981), the pres-ence of problems of asymmetries of information (adverseselection and moral hazard) at the level of credit mar-kets and the problems of enforcement of contract lead tocredit rationing through the formal sources. Chung (1995)

in the formal sector can discourage some households totake loans. In such a situation borrowers need to considerinformal sources as last resorts or as alternatives to the

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

ns from

Fig. 1. Evolution of savings and loaSource: compiled by the author from several monographs BCEAO.

formal sector in developing countries. According to Kochar(1992), informal loans, especially those from friends andparents, would be less expensive as compared to the for-mal loans and therefore preferred by borrowers. Guirkingerand Boucher (2007) add to this that informal lenders haveaccess to local information, allowing them to write downcontracts that are less risky for borrowers. The relativelylow information and transaction costs, the simplicity andflexibility in financial procedures ease access to low incomepersons.

The microfinance literature demonstrated how theproblem of information asymmetry can be tackled by theuse of social capital (Rankin, 2002; Gomez and Santor,2001; Besley and Coate, 1995). Social capital refers to thenorms and networks that enable people to act collectively(Woolcock and Narayan, 2000). Fox (1996) defines socialcapital as a social organization, relationship of cooperationand reciprocity, networks and leadership that facilitate col-lective action. Social capital could be the degree of trust ingovernment or other societal institutions (Fukuyama, 1995cited in Okten and Osili, 2004), social cohesion, reciprocityand institutional effectiveness. Or as stated by Grootaertand van Bastelaer (2002a), social capital is broadly the setof institutions, the relationships, the attitudes, and valuesthat govern interactions among people and contribute toeconomic and social development. This definition depictsclosely the developing countries’ situation. Social capitalhelps to correct the issue of incomplete information byensuring default payments through social sanctions. Socialcapital can be a tool for the diffusion of information onthe sources of finance and consequently will influence thedifferent choices (Okten and Osili, 2004). This explains

the inclusion of variables like ethnic groups and religiousadherence further in this model. In the same way, theirpresence can constitute a non-negligible competitor for theMFIs.

formal MFIs on period 1998–2005.

The working hypothesis that will guide the empiricalwork below is that the microfinance institutions in Côted’Ivoire can help solve some of the market imperfectionsthat exist in the financial sector. Following the short lit-erature review above, the main way in which this willbe possible is by lowering the transaction costs by meansof offering loans through the informal sector. It is alsohypothesized that social capital plays a role in solvingthe market imperfections. In practical terms, we will putthese hypotheses to the test by estimating an equationfor whether or not a household will take a loan throughthe formal or the informal sector, and including householdincome, loan size, and social capital variables as explana-tory variables. If microcredit helps solve market failure, weexpect household income, the size of the loan and socialcapital all to have a negative impact on loans through theformal sector.

There are also a number of variables that need to betaken into account as controls. Borrowers can base theirchoice on a particular feature of the loan which is theinterest rate. However, focusing on this component aloneis not sufficient to explain the choices made by house-holds (Nguyen, 2006). Many studies demonstrated thatthis component is not the most important factor, but thatinstead factors not directly linked to the prices and ser-vices play an important role (Diagne and Zeller, 2001). Infact, if reimbursement modalities, the required collateral,and the availability of additional services are not in accor-dance with the needs of the borrower, the latter does notmanifest any demand for credit from microfinance insti-tutions. The type of financial institutions and borrowingpolitics put in place will determine the choice made by

households (Schmidt and Kropp, 1987) and brings to theforefront the question of what are the characteristics andfeatures that enable households and individuals to borrowfrom formal sources.

Econom

tsao(oasicuiTwtGris

otSidi

ip2cdncTsctdbp

4e

cm1baeirf

ffom

E.L. Togba / Structural Change and

Zeller (1994) found that the age, the level of educa-ion, salaries and gender have an influence on the choice ofources of finance. For Nguyen (2007), the size of the housend the type of activities exercised has a positive effectn making as choice formal sources of finance. Guirkinger2008) added factors like the wealth level, the presence ofther source of income; the number of persons in his chargend the geographical location of the household are equallyusceptible to influence their choice. The wealth and/orncome seem to be the basic one among the importantharacteristics of household. In fact, as Madestam (2007)nderlined, the wealth level of the economic agent is an

mportant factor determining the credit sector he chooses.hus, when an economic agent has got a sufficient level ofealth, he will choose to borrow with the formal institu-

ions. This is confirmed by the results of studies such asuirkinger (2008), Crook (2006) and Zeller (1994), which

eveal that the household’s wealth has positive and signif-cant effect on the loan demand and the choice of financeources.

The noticeable positive effect of the size of householdsn the demand for loans is confirmed in some contribu-ions (Guirkinger, 2008; Nguyen, 2007; Okurut et al., 2004;wain, 2002). The role of household’s size can be seenndirectly. The larger the household the greater is its expen-iture. Thus, the household would apply for a (larger) loan

n order to smooth his consumption.At the empirical level, some authors think that it is

mportant to make the distinction between access andarticipation to credit (Zeller, 1994; Diagne and Zeller,001). Access to credit is an essential phenomenon con-erning the supply of credit, due to the fact that the lenderecides on who can borrow or not. Participation is a phe-omenon linked to demand. A household has access to aertain source if it can in principle borrow from this source.he household participates if it actually borrows from thisource. Consequently, a household can have access andhoose not to borrow. Taking into account this distinction,he decision to choose one particular source of finance isone from an analysis of the demand for credit expressedy households and that is the approach developed in thisaper.

. The household choice of sources of credit:mpirical approach

An analysis of households decisional choices in theredit market are usually carried out using discrete choiceodels (Nguyen, 2007; Duong and Izumida, 2002; Zeller,

994). In fact, according to Atieno (2001), it is impossi-le to identify a program for the demand for loans usingn amount of observed loans since this only reflects thexisting supply. The credit demand function can only benterpreted from the decision to participate by the bor-ower, that is the decision to borrow or not to borrow androm which sector.

Zeller (1994) used a univariate probit model to estimate

actors determining the borrowing decisions of individuals,rom the point of view of their participation in the formalr informal credit market in Madagascar. The author treatsarket segments separately to identify the similarities and

ic Dynamics 23 (2012) 473– 486 477

differences between the sectors in terms of demand forcredits and their rationing. Nguyen (2007) also separatesthe sources of loans while expecting that the determinantsof credit participating be different as the eligible require-ments are different between sources. This paper adoptsZeller’s approach (1994) that the market must be seg-mented 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 ahousehold are:

1: Bank and financial institutions.2: Companies of framing.3: Credit unions.4: Social funds.5: ROSCAs.6: Moneylenders.7: Cooperatives.8: Others.

In order to take into account the major differencesbetween these alternatives, they are grouped as (1) for-mal institutions for those who are regulated, includingbanks and financial institutions, companies of framing,credit unions, and social funds; and (2) informal institu-tions, including moneylenders, ROSCAs, cooperatives andothers.

Assuming that the household is rational in the sensethat it makes choices that maximize its direct utility sub-ject to constraint on expenditure, it is possible to derivean indirect utility function (Maddala, 1983). We define anunderlying latent variable y* to denote the indirect utilitylevel associated to the direct utility y. The observed variabley is defined as:

y = 1 if y∗ > 0

y = 0 otherwise(1)

However, there are many errors in this maximizationbecause of imperfect perception and optimization, as wellas the inability to measure exactly all relevant variables.McFadden (1974a) suggested using a random functionwhere the random term comes in an additive manner. Con-sequently, this indirect utility function y* will be written asfollows:

y∗ = ˇx1 + ε1 (2)

where x1 is a vector of observable attributes specific to thehousehold; ε1 is the random component of utility that rep-resents the unobserved household i’s idiosyncratic taste forchoosing a source. It is assumed to be independently andidentically distributed.

Choice of the source of borrowing is a two-step processwhich requires that households demand a loan at the firststage, and at the second stage they choose the source wherethey want to borrow. Since the second stage is a conditionalon the first stage, it is likely that the second stage sam-

ple is non-random, which could create a sample selectionbias. Indeed, Nagarajan et al. (1995) think that estimatesof loan demand or choice of credit source are often biasedbecause they use models that do not adequately correct for

Econom

478 E.L. Togba / Structural Change and

selectivity bias. Therefore, it is important to correct for thissample bias in order to obtain consistent estimates. Theidea that factors affecting selection into the sample maysimultaneously affect the binary outcome of interest hasbeen the motivation for the introduction of the probit sam-ple selection model (van De Ven and van Praag, 1981). Inour case, it is believed that the decisions of choosing asource of borrowing and that of expressing a demand ofloan are correlated (both decisions are binary). In effect,the data set specifies a binary variable that identifies theobservations for which the dependent is observed/selectedor not observed.

The underlying structural framework is a householdproduction model with utility maximizing households,who demand credit (demand = 1) if a loan is expectedto increase utility, and they do not demand credit(demand = 0) in the opposite case.

The dichotomous demand selection equation is givenby:

d ={

1 if d∗ > 0

0 otherwise(3)

The latent equation is given as follows:

d∗ = ıx2 + ε2 (4)

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

y ={

1 if y∗ > 0 and d = 1

0 y∗ ≤ 0(5)

where the latent equation for outcome equation is

y∗ = ˇx1 + ε1 (6)

It is assumed that the latent errors are bivariate normaland independent of the explanatory variables. The probitmodel with sample selection can be expressed as follows:

y∗i

= ˇx1 + ε1

yi ={

1 if y∗i

> 0 and d = 1

y∗i

≤ 0d∗

i= ıx2 + ε2

di = 1 if di∗ > 0 and 0 otherwise

(7)

Heckman (1990) has shown that selection bias can beovercome by including the inverse Mills ratio from thesample selection equation in the equation of interest. Inthis approach, the selection into the sample of those whodemand credit is first modeled. Then, the inverse Millsratio (lambda) from this regression is incorporated into theequation of interest.

We also encounter a problem of endogeneity of someexplanatory variables. In fact, information on interest rateswhich represent the prices of loan from each source is

missing in the data set. Since prices and income are thekey variables explaining the demand for credit, the non-inclusion of the prices variable could create a correlationbetween the variable income and the error term. That could

ic Dynamics 23 (2012) 473– 486

lead to an endogeneity bias, making the estimator inconsis-tent. Rivers and Vuong (1988), cited in Wooldridge (2001),provide a simple test to verify endogeneity in the caseof a binary model. They suggest to model a continuousendogenous variable as a linear function of the exogenousvariables and some instruments. Predicted values fromthis regression are then used in the second stage probit.Therefore, in order to find an instrumental variable for theincome, Rivers and Vuong‘s approach is used.

Finally, to test hypotheses in line with the previous sec-tion, we specify the demand of loan as a linear functionof household characteristics including income, gender ofhousehold head, age of household head, number of house-hold members, religion, education level of the householdhead, geographical location, and socio economic group ofthe household head, etc.

The empirical model to be estimated is presented asfollows:

dloani = ˛0 + ˛1income∗i + ˛2mastati + ˛3Endowni

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

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

+ ˛11agei + ε1i (8)

dloan denotes the loan demand equation. income denotestotal household income. Age denotes the age of householdhead, mastat denotes the marital status of the householdhead, endown denotes whether the household head hasa house or not, and land or not, Socioecogrp denotes thecategorical socio-economic group to which the householdhead belongs; male denotes the head household is male;hhsize denotes the number of persons in the householdArea denotes the area where lives the household, noprojectdenotes the household does not plan to extend its activity;education denotes the level of the education of the house-hold head; Religion denotes the religion of the householdhead, and εi1 denotes the error term assumed to be nor-mally distributed.

We specify also the choice of credit as a linear functionof household characteristics including gender of householdhead, age of household head, number of household mem-bers, religion, education level of the household head, ethnicgroup, geographical location, and variables related to thecredit contract, including time of repayment, loan size, typeof activity funded, etc.

formali = ˇ0 + ˇ1incomei + ˇ2assetindexi + ˇ3malei

+ ˇ4agei + ˇ5age2i + ˇ6hhsizei + ˇ7religioni

+ ˇ8schoolingi + ˇ9ethnici + ˇ10areai

+ ˇ11timrepaymi + ˇ12loansizei + ˇ13useloani

+ ˇ14� + ε2i (9)

formal denotes the choice for formal sources. Assetindexrepresents a measure of wealth; schooling denotes whetherthe household head is illiterate or not, Ethnic denotes the

household head ethnicity, Area denotes where lives thehousehold lives, timrepaym denotes the time of repay-ment of the loan, loansize denotes the amount of loandemanded by the household head, useloan denotes the

Econom

ptad

5

iSthtaitoheldhthttwtscTo

hinoAmec3atrptmo8

pvidavacsd

E.L. Togba / Structural Change and

urpose for which the loan has been taken, � representshe inverse Mills ratio fitted from loan demand equation,nd εi2 denotes the error term assumed to be normallyistributed.

. Data

The data used in this research is the households Liv-ng Standard Survey conducted in 2002 by the Nationaltatistics Institute of Côte d’Ivoire. The research unit ishe household and the people who live in it. The 2002ouseholds Living Standards Survey is a nationwide, multi-opic household survey with modules covering numerousspects of living standards. The survey contains detailednformation on households from all regions of the coun-ry. The household survey has 12 sections, gathering datan education, health and employment status of house-old members, household economic activities, income andxpenditure, household size and housing, borrowing andending activities. It covers 10,800 households living in Côte’Ivoire. Out of 10,800 households surveyed, 1392 house-olds have demanded a loan. They represent 12.88% ofhe overall sample. This sample is composed of those whoave applied for and who have received the total amount,hose who have received a part of amount of loan, andhose whose application has been refused. Among thoseho have demanded a loan, 85.06% expressed a demand

o informal sources. Table 2 gives a description of this sub-ample of households demanding a loan and presents theonstructed variables’ summary statistics. In the appendix,able A1 further specifies the definition and measurementf the variables.

The demographic profile of the 1392 respondent house-old indicates that the average age of a household head

s 42.59 years and about 84.63% of them fall in the eco-omically active population (ages 18–59). The majorityf household heads (53.59%) have no formal education.pproximately 19.40% and 22.49% of them have low andedium education respectively. Christians and adher-

nts to non-traditional religion and those without religiononstitute about 36.78% and 24.14% respectively, against8.15% of Muslims. 30.68% of households live in other urbanreas. The socio economic categories define the broad sec-or of employment of the household head. The statisticseveals that the majority of households heads work in therivate services sector for the choice of for formal insti-utions, and the agricultural sector for informal loans. The

ajority of the household heads are married, which is a signf household stability. Male-headed households constitute2.33%.

The average household size in the entire sample is 5.66ersons per household. Concerning the income, the averagealue of income for formal sources is inferior to those of thenformal sources suggesting low income households preferemanding loans from the formal sources. The asset vari-ble is a combination of the asset data which are all dummyariables, indicating whether households own a particular

sset or not. Principle component analysis has been used toreate the asset index, to proxy wealth and capture owner-hip of tangible assets. The assets considered are consumerurables goods. Ownership of these assets determines the

ic Dynamics 23 (2012) 473– 486 479

choice of formal source since those assets could be used ascollateral. Then, the formal institutions favour house own-ing households as is evident from the higher share of formalborrowing in the category of households owning a house.On the contrary, a land owning household is predominantfor informal sources.

Social capital or social networks play an importantrole in the Ivorian context. In line with the definitionby Grootaert and van Bastelaer (2002a), social capitalrepresents the institutions, relationships, attitudes, andvalues that govern interactions among people. Ethnic andreligious backgrounds play an important role in thesevariables. Within these groups, potential borrowers sharecultural similarities, facilitating access to credit to theirmembers (Azam et al., 2001).

From Table 2, we notice that Christians and Muslims arethe religious backgrounds that are associated to a higherdemand for loans. The Christian households are the oneswho choose more informal sources for their loans applica-tions, whereas the Muslim households go more towardsformal institutions. At the ethnic group level, we noticethat some ethnic groups like the informal sources more.It is mainly the case of Akan group, South Mande andother African ethnic groups. On the contrary, the Kru groupand North Mande prefer formal sources whereas for theremaining groups the preference is not so clear.

Considering the factors related to loan contracts, thetime of loan repayment ranges from 1 to 75 months withmean about 2.74 months in the survey. The mean of for-mal credit time of loan repayment is 3.48 months, whilethe mean of informal credit is 2.61 months. The time ofrepayment varies from 1 to 36 months in the formal insti-tutions. In the informal institutions, the time of repaymentranges from 1 to 75 months. The modal value of time ofloan repayment for all types of sources is one month. Thefirst objective of loan is to create revenue from an activity inorder to improve living conditions of the household. Mostdesigns of loans product in MFIs concern production pur-poses. Indeed, about 91.82% of total borrowing from theformal sources was for production purposes (agricultureand trade). Loans from informal sources were also mainlyused for production. The statistics give about 90.87%. How-ever, it has been noticed that agricultural activities remainthe first activities for which each source is chosen. Loandemands for agricultural activities have, about a proportionof 58.17% and 60.81% respectively for formal and infor-mal sources. That is due to the fact that the majority ofhouseholds lives in rural areas, where informal sources areprefered.

6. Empirical results of borrowing sources

As stated earlier, Heckman’s two-step approach is usedto estimate the determinants of choosing a credit pro-gram. Before proceeding to the regression analysis, let usanalyze the stated endogenous problem. As stated ear-lier, Rivers and Vuong’s approach allows making a simple

test on the residuals from income regression. This testreveals that income is correlated with the error term inthe demand equation. That means that there is endogene-ity 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 2Summary statistics.

Variables Full sample of loandemanding households

Demanding formal loan Demanding informal loan

Characteristics of the household headAge composition (%)Age of the household head 42.59 (14.46) 41.1 (13.68) 42.85 (14.58)Age (less than 18 years) 0.79 0.48 0.33Age (from 18 to 59 years) 84.63 86.53 84.29Age (+60 years) 15.01 12.98 15.37Gender (%)Male 82.33 80.76 82.60Female 17.67 19.24 17.40Marital status (%)Married 70.50 66.83 71.19Unmarried 16.6 19.71 16.13Other (divorced, separated, widow, widower) 12.8 13.46 12.66Education (%)No education 53.5 56.25 53.12Low education 19.4 19.71 19.34Medium education 22.5 20.19 22.88High education 4.5 3.84 4.64Area (%)Abidjan 17.2 18.75 16.89Other urban areas 30.7 35.09 29.89Eastern rural forest 17.2 17.78 17.06Western rural forest 18.4 12.98 19.42Rural savannah 16.5 15.38 16.72Socio-economic group (%)Agricultural worker 22.9 18.75 23.65Public services 5.4 4.32 5.57Private formal services 20.2 22.59 19.76Own business 16.7 15.86 16.89Other occupation 17.0 17.78 16.89Characteristics of the householdHousehold size 5.66 (4.01) 4.88 (3.46) 5.35 (3.87)Income (in 1000 FCFA) 60.4049 (141.891) 37.8188 (53.579) 64.3727 (151.873)Assets (between 0 and 1) .0442 .2299 .0115Owning house (%) 49.14 50 48.98Owning land (%) 54.38 45.67 55.91Social capital variablesReligion (%)Christian 39.22 35.57 39.86Muslim 36.64 41.82 35.72Other religion (traditional, other religions, no religion) 24.14 22.59 24.41Ethnicity (%)Akan 29.74 27.88 30.06Kru 14.15 15.38 13.96Mande south 8.91 5.76 9.45Mande north 12.43 16.82 11.65Voltaic 13.58 13.94 13.51Other African ethnic 21.19 20.19 21.36Characteristics of the loanTime of repayment (in months) 2.74 (4.12) 3.48 (5.13) 2.61 (3.87)Loan size (in FCFA, in log.) 11.507 (1.857) 12.74 (1.805) 11.29 (1.781)Purpose of loan (%)Trade activities 30.6 33.65 30.06Agricultural activities 60.42 58.17 60.81Transport activities 0.43 0.48 0.42

Other activities 8.55

Source: own computation from 2002 INS Survey Data.

instrumental variable is used. This is the predicted value ofincome from the income regression.

Concerning the demand of loan equation, the convec-tional Wald test statistic is significant at 1%. It rejects the

null hypothesis that all coefficients are zero. Knowing thesign of the parameter is enough to determine whether thevariable has a positive or negative effect on the demandequation.

7.69 8.69

The following variables have been found relevant toexplain the demand for a loan: income, owning land,having an own business, the household size, having nodevelopment project, and geographical location. The effect

of income is positive and significant for demanding aloan. That demonstrates that a household demands aloan when its income is higher. Owning land increasesthe probability of demanding a loan. The explanation is

E.L. Togba / Structural Change and Economic Dynamics 23 (2012) 473– 486 481

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

Explanatory variable Coefficients Standard error Marginal effects Standard error

Predicted Income 1.11e−07** 4.73e−08 1.39e−08 .000Matrimonial statusMarried −.041 .0622 −.0052 .008Unmarried .0077 .0724 .0009 .0091EndowmentHouse .0561 .0391 .0071 .005Land .2012*** 0392 .0255 .0057Socioeconomic groupAgricultural worker −.060 .0466 −.0073 .0056Public services .0517 .0879 .0067 .0118Private formal services −.0185 .0517 −.0023 .0064Own business −.1401*** .0511 −.0163 .0057Male .0112 .0531 .0014 .0066Household size .0186*** .0052 .0023 .0007AreaOther urban areas .1569*** .0498 .0207 .0073Eastern rural forest .1008* .0605 .0132 .0084Western rural forest .0638 .0613 .0082 .0082Rural savannah .1003 .0643 .0132 .0089No project −2.464*** .3017 −.164 .004EducationLow education .0091 .0462 .0011 .0058Medium education −.0114 .0484 −.0014 .006Higher education −.0805 .0901 .0095 .0101ReligionChristian .0365 .0447 .0046 .0057Muslim −.0323 .0451 −.004 .0056AgeAge less or equal 24 years .0338 .0834 .0043 .011Age between 25 and 39 −.0267 .0569 −.0033 .0071Age between 40 and 59 −.029 .0530 −.0036 .0065Constant −1.204*** .0848

Number of obs. =10,800 Prob > Chi2 =.000Wald Chi2(25) =160.22 Pseudo R2 =.1013Pseudo log likelihood =−3729.5996

Note: z denotes z-statistics.***

th

hhdtbtmf

thtostfdaia

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

hat the land could be used as collateral by the house-old.

Regarding the socio-economic groups to which theousehold head belongs, it can be seen that the head ofousehold who is doing its own business is least likely toemand a loan. This negative effect is explained by the factake the loans from these sources require the provision ofusiness registration, procedure enterprises, or complexax procedure and the collection of public revenue, docu-

ents they do not have sometimes and necessitate moneyor their establishments.

The household size has a significantly positive effect onhe probability of borrowing. A greater number of house-old members imply higher expenses. Most of the time,he budget cannot cover the expenses of the all membersf the household. Therefore, in order to smooth their con-umption, households have to borrow. In addition, the facthat the head of the household has no development projector his activities has a negative effect on the probability of

emanding a loan. Marital status, education, religion andge of the household head do not appear to significantlynfluence the probability that the household is demanding

loan (Table 3).

In the second step of the estimation, we try to deter-mine the factors influencing the choice of formal versusinformal sources. Again, the conventional Wald test statis-tic is significant at 1%, rejecting the null hypothesis that allcoefficients are zero. The predicted probability of choos-ing formal source is 11.56%. That confirms the fact there ispreference for informal sources in Côte d’Ivoire (Azam etal., 2001).

The following variables have been found relevant toexplain the choice of formal source. Income has a negativeeffect on the likelihood that a formal source is chosen. Thus,informal sources of credit are preferred by low incomegroups. This is one of the main variables in our analysis, andwe take this result as an indication microcredit may helpsolve some part of the market failure in the financial sec-tor in Côte d’Ivoire. Low income households who will not beable to get a loan in the formal sector do get access to creditthrough the informal sector. This is particularly relevant incombination with the result for the loan size. The size of

the loan is positively and significantly related to the prob-ability of choosing the formal credit program. Indeed, theprobit results display a positive and significant effect of theloan size for the choice of formal source. That supports the

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

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

Explanatory variable Coefficients Standard error Marginal effects Standard error

Income −2.68e−06*** 6.69e−07 −5.22e−07 .0000Assetindex .229*** .075 .0447 .0149Male .012 .122 .0022 .0236Age .032* .0187 .006 .0036Age squared −.0003* .0002 −.000 .00004Household size −.001 .0148 −.0002 .0029Schooling −.048 .107 −.009 .0208ReligionChristian −.079 .121 −.015 .023Muslim .184 .158 .037 .033AreaOther urban areas −.188 .141 −.035 .025Eastern rural forest −.256 .162 −.045 .025Western rural forest −.402** .174 −.067 .024Rural savannah −.381** .170 −.063 .023EthnicAkan .208 .156 .042 .033Krou .371** .183 .084 .047Mande north .241 .163 .052 .038Mande south −.030 .213 −.005 .040Voltaic .112 .163 .022 .035Time of repayment .011 .0108 .002 .002Loan size .224*** .026 .043 .005Use of loanTrade activities .099 .165 .019 .033Agricultural activities .273* .153 .051 .028Inverse Mills ratio .905 1.458 .176 .283Constant −4.591 .659

Number of obs. =1392 Prob > Chi2 =.000Wald Chi2(22) =126.84 Pseudo R2 =.1279Pseudo log likelihood =−511.94384

*** Significant at 1%.** Significant at 5%.

* Significant at 10%.

presumption in the literature that informal institutions arefar more effective at financing small borrowers than theformal institutions. Therefore, when the loan size is largerthe household will choose formal sources. The higher trans-action costs that are typically related to borrowing fromformal sources may in the case of larger loan sizes becomerelatively smaller. A 100% increase of the loan size is asso-ciated with an increase in the probability of choosing theformal sources of 4.37%.

The importance of personal wealth is confirmed hereby the positive effect of the variable asset index, proxy ofwealth, on the choice of the formal sources. In addition, theage of the household head raises the probability of borrow-ing from formal sources.

Compared to other ethnic groups, the Krou are mostlikely to choose formal sources for borrowing. Indeed, theprobit results display a positive and significant sign, ascompared to the reference group of other ethnic group.The marginal effect is an increase of 8.4%. The ethnic net-work plays an important role in relations between thehouseholds in Côte d’Ivoire through the imposition of socialsanctions for misconduct of a member of the network. But

it also imposes the respect of strong kinship ties requir-ing acute sense of forgiveness for the person that wentwrong has. Therefore, depending on whether the house-hold comes from an ethnic group where social sanctions

are higher, the use of formal loans will be chosen for fearof losing its reputation. That explains why the householdsfrom the ethnic groups as Kru and Akan, North Mande orVoltaic prefer the formal sources.

The effect when the household head lives in westernrural forest and rural savannah comparatively to Abidjanis negative on the probability of borrowing from formalsource. In fact, households living in such regions havedifficulties to access to facilities, higher transaction costsand as such they do not choose formal institutions. Thisresult is in line with the unequal distribution of MFIs onthe whole territory, as stated by the national commis-sion of microfinance (NCM, 2005). The magnitude of theseeffects when the household head lives in western rural for-est 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 theborrower lives has an impact on the choice of source ofborrowing.

The purpose of the loan, such as to finance trade activi-ties or agricultural activities, shows that a positive effect isobserved for the choice of formal sources when it concernsagricultural loans, as compared to other activities. Loansfor agricultural activities are 5.16% more likely to be taken

from formal sources, corroborating the results by Nguyen(2007) (Table 4).

Econom

7

tcgiiarmhat

savaghsstptwt

TR

E.L. Togba / Structural Change and

. Conclusion

Understanding the socioeconomic factors influencinghe determinants of households’ choice of microfinanceredit program is useful for future policy designs, as it canive an indication of how well the credit market performsts function. Formal and informal sources of credit co-existn the market, with informal sources providing lower trans-ction costs. Because of information asymmetries, creditationing may exist in the market, and transaction costsay be high. This may especially affect low income house-

olds, who may have a higher preference for small loansnd thus face disproportionately high transaction costs inhe formal financial sector.

This paper attempted to clarify the role of microcredit inolving market failure by testing the impact of three vari-bles on the source (formal or informal) of credit. Theseariables were the household income, the size of the loan,nd social capital, which was measured by ethnic back-round and religion. It was indeed found that low incomeouseholds tend to prefer informal sources of credit, andmaller loans are also performed through the informalector. This confirms our hypothesis that informal loanshrough microcredit institutions have a particular role to

lay in the financial sector in Côte d’Ivoire. We also foundhat the choice of source of credit depends in ethnic net-orks, although religious backgrounds did not matter for

he source of credit.

able A1eports of the descriptive statistics (the means and standard deviations) of the su

Variable Description

SourceFormal =1 if the household borrows from formal sAssetindex =a proxy of wealthGender =1 if head of household is male

Demand of loan =1 if the household answer he has been boEducationNo education =1 if head of household has no education

Low education =1 if head of household has some primaryMedium education =1 if head of household finished primary s

continued to secondary schoolHigh education =1 if head of household completed secondSchooling =1 if the head household head has some liSocioeconomic groupAgriculture =1 if head of household is employed in AgPublic service =1 if head of household is employed in puPrivate formal service =1 if head of household is employed in prOwn business =1 if head of household is doing its own bOther occupation =1 if head of household is doing other occAreaAbidjan =1 if household lives in Abidjan

Other urban areas =1 if household lives in the other urban arEastern rural =1 if household lives in rural eastern foresWestern rural =1 if household lives in rural western foreSavannah =1 if household lives in rural savannah

Ethnic groupAkan =1 if the head of household has a native toKru =1 if the head of household has a native toMande south =1 if the head of household has a native toMande North =1 if the head of household has a native toVoltaic =1 if the head of household has a native toOther African ethnic =1 if the head of household has a native to

ethnic

ic Dynamics 23 (2012) 473– 486 483

Thus, we conclude that microfinance institutions havean important role to play in the economy of develop-ing countries. In the descriptive work that preceded thestatistical analysis, it was also found that microfinanceinstitutions in Côte d’Ivoire collect a much larger amount insavings than what they give out in loans. This is an interest-ing feature that demands further theoretical and empiricalwork to clarify the exact role of microfinance in developingcountries.

Appendix A. Appendix

A.1. Estimation of income

Consider the income model:

incomei = ˇ0 + ˇ1income2i + ˇ2texpi + ˇ3typhi + ˇ4mstai

+ ˇ5ostai + ˇ6sgrpi + ˇ7gendi + ˇ8areai

+ ˇ9landowneri + ˇ10noprojecti + ˇ11Educi

+ ˇ12religi + ˇ13agei + 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 dummyvariable: =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).

rveyed households.

Mean St. dev.

ources .149 .356−4.29e−09 .631.823 .381

rrowed .1288 .335

.535 .498 schooling .194 .395chooling or .225 .417

ary or higher. .045 .208teracy level .448 .497

riculture .229 .420blic service .054 .226ivate formal service .202 .401usiness .167 .373upation .170 .375

.172 .377eas .306 .461t .171 .377st .184 .388

.165 .371

ngue Akan .292 .454ngue Kru .137 .344ngue South Mande .088 .283ngue North Mande .131 .337ngue voltaic .138 .345ngue other African .212 .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 the time of repayment of the loan 2.676 4.119Loan size The amount of loan demanded by the household head 1,161,877 6,790,604Matrimonial statusMarried =1 if head of household is married .705 .456Unmarried =1 if head of household is no married .166 .372Other matrimonial status =1 if others (separated, widow) .128 .334ReligionChristians =1 if head of household is Christian .367 .482Muslim =1 if head of household is Muslim .381 .486Other religions =1 if head of household is other religion .109 .313Household size Number of household members 5.664 4.009Age Age of head household 42.59 14.461Income Total income of household head (FCFA/month) 60,404.9 141,891.7Type of houseVilla =1 if household lives in a villa, apartment .123 .328Set of house =1 if household lives in a set of house .506 .500Detached house =1 if household lives in a detached house .146 .353Other (hut, shack) =1 if household lives in an hut, a shack .224 .417Occupation status of houseOwner =1 if household owns the house .489 .500Rented =1 if household rents the house .340 .474Other occupation =1 for other occupation .170 .225No project =1 if the head of household does not plan to extend his activity .038 .193Use of loanTrade activities =1 if the head of household demanded the loan for the trade

activities.296 .456

Agricultural activities =1 if the head of household demanded the loan for theagricultural activities

.622 .481

Transport activities =1 if the head of household demanded the loan for thetransport activities

.003 .057

Other activities =1 if the head of household demanded the loan for the otheractivities

.078 .268

Source: Own computation from the INS Survey 2002.

Table A2Estimation results of Income. Dependent variable: income.

Explanatory variables Coefficients Standard error P > t

Income2 1.0026 .0033 .000Expenditure 1.072 .2391 .000Type of houseVilla 20,938.15 6936.23 .003Set of house −7325.27 3537.16 .038Detached house 985.32 3718.51 .791Matrimonial statusMarried 12,586.63 3287.22 .000Unmarried −15,960.12 4724.95 .001Occupation statusOwner 4143.46 3476.11 .233Rented 1906.61 5510.41 .729Socioeconomic groupAgriculture 22,632.5 3548.63 .000Public business 103,113.6 12,530.24 .000Private service 78,475.86 5040.45 .000Own service 62,440.79 5484.64 .000Gender 14,670.1 3067.02 .000Household size 6182.55 1088.39 .000AreaOther urban areas −7531.69 7906.35 .341Eastern rural forest 4676.04 8782.74 .594Western rural forest −8152.87 8798.97 .354Rural savannah −11,501 7618.77 .131Landowner 3829.54 3455.02 .268No project 8003.81 4238.71 .059

Table A2 (Continued)

Explanatory variables Coefficients Standard error P > t

EducationLow education −2813.69 3709.56 .448Medium education 7907.56 5912.28 .181Higher education 103,143.8 13,726.18 .000ReligionChristian −4380.94 4001.46 .274Muslim −4655.13 3638.96 .201AgeAge less or equal 24

years5545.83 7388.89 .453

Age between 25 and 39years

−1853.89 6028.51 .758

Age between 40 and 59years

16,557.58 5924.59 005

Constant −75,870.16 16,887.18 .000

Number of obs. =10,800 Prob > F .000F(29, 10,770) =3949.173 R2 .6228

E.L. Togba / Structural Change and Econom

Table A3River and Vuong’s endogeneity test. Dependent variable: demand of loan.

Explanatory variable Coefficients Standard error P > t

Income −1.84e−07 1.20e−07 .127Matrimonial statusMarried −.0419 .0621 .500Unmarried .0078 .0724 .914Houseowner .0561 .0391 .151Landowner .2014*** .0391 .000Socioeconomic groupAgriculture −.0605 .0466 .195Public business .0509 .0878 .562Private service −.0206 .0517 .690Own service −.1410*** .0511 .006Gender .0111 .0531 .833Household size .0186*** .0052 .000AreaOther urban areas .1578*** .0498 .002Eastern rural forest .101* .0605 .092Western rural forest .064* .0613 .295Rural savannah .101 .064 .116No project −2.4633*** .300 .000EducationLow education .0089 .0462 .846Medium education −.0115 .0484 .811Higher education −.0857 .0900 .341ReligionChristian .0361 .0447 .418Muslim −.0321 .04517 .477AgeAge less or equal 24 years .0339 .0834 .684Age between 25 and 39 −.0256 .0569 .653Age between 40 and 59 −.0294 .0530 .579Residuals fitted 2.91e−07** 1.29e−07 .024Constant −1.203*** .0847 .000

Number of obs. =10,00 Prob > Chi2 .000Wald Chi2(25) =163.00 Pseudo R2 .1016Pseudo log likelihood =−3728.5969

Note: z denotes z-statistics.

R

A

A

A

BBBBBBB

B

B

B

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

eferences

ryeetey, E., 1996b. Rural finance in Africa: institutional devel-opments and access for the poor. In: The World Bank AnnualConference on Development Economics, Washington, DC,25–26 April.

tieno, R., 2001. ‘Formal and Informal Institutions’ Lending Policies andAccess to Credit by Small-Scale Enterprises in Kenya: An EmpiricalAssessment, AERC Research Paper 111.

zam, J.P., Biais, B., Dia, M., Maurel, C., 2001. Informal and formal creditmarkets and credit rationing in Cote d’Ivoire. Oxford Review of Eco-nomic Policy 17 (4).

CEAO, 2005. Situation microfinance. Monographie Côte d’Ivoire.CEAO, 2004. Situation microfinance. Monographie Côte d’Ivoire.CEAO, 2003. Situation microfinance. Monographie Côte d’Ivoire.CEAO, 2002. Situation microfinance. Monographie Côte d’Ivoire.CEAO, 2001. Situation microfinance. Monographie Côte d’Ivoire.CEAO, 2000. Situation microfinance. Monographie Côte d’Ivoire.esley, T., Burgess, R., 2001. The Political Economy of Government Respon-

siveness: Theory and Evidence from India, CEPR Discussion Paper No.2721. Centre for Economic Policy Research, London.

esley, T., Coate, S., 1995. Group lending, repayment incentives and socialcollateral. Journal of Development Economics, 46.

iggs, T., Raturi, M., Srivastava, P., 2002. Ethnic networks and access to

credit: evidence from the manufacturing sector in Kenya. Journal ofEconomic Behavior and Organization, 49.

rau, J.C., Woller, G., 2004. Microfinance: a comprehensive review of theexisting literature. Journal of Entrepreneurial Finance and BusinessVentures 9 (1).

ic Dynamics 23 (2012) 473– 486 485

Chung, I., 1995. Market choice and effective demand for credit: the rolesof borrower transaction costs and rationing constraints. Journal ofEconomic Development 20 (2).

Crook, J., 2006. The demand and supply for household debt: a cross countrycomparison. In: Bertola, G., Disney, R., Grant, C. (Eds.), The Economicsof Consumer Credit. MIT Press, Cambridge.

Diagne, A., Zeller, M., 2001. Access to Credit and Its Impact on Welfarein 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 Develop-ment 30 (2).

Fafchamps, M., 2000. Ethnicity and credit in African manufacturing. Jour-nal of Development Economics 61 (1).

Fafchamps, M., 2003. Ethnicity and networks in African trade. Contribu-tions to Economic Analysis & Policy 2 (1).

Fisman, R., 2003. Ethnic ties and the provision of credit: relationship-levelevidence from African firms. Advances in Economic Analysis & Policy3 (1).

Fox, J., 1996. How does civil society thicken? The political construction ofsocial capital in rural Mexico. World Development 24 (6).

Gomez, R., Santor, E., 2001. Membership has its privileges: the effect ofsocial capital and neighbourhood characteristics on the earnings ofmicrofinance borrowers. The Canadian Journal of Economics 34 (4).

Guirkinger, C., 2008. Understanding the coexistence of formal and infor-mal credit markets in Piura, Peru. World Development 36 (8).

Guirkinger, C., Boucher, S.R., 2007. Risk, wealth and sectoral choice in ruralcredit markets. American Journal of Agricultural Economics 89 (4).

Grootaert, C., van Bastelaer, T. (Eds.), 2002a. The Role of Social Capital inDevelopment: An Empirical Assessment. Cambridge University Press.

Heckman, J., 1990. Varieties of selection bias. American Economic Review80.

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, Work-

ing Paper. Bocconi University, Milan, Italy.Matin, I., Hulme, D., Rutherford, S., 2002. Finance for the poor: from

microcredit to microfinancial services. Journal of International Devel-opment, 14.

McFadden, D., 1974a. Conditional logit analysis of qualitative choicebehavior. In: Zarembka, P. (Ed.), Frontiers of Econometrics. AcademicPress.

Meyer, R., 2002. The demand for flexible microfinance products: lessonsfrom Bangladesh. Journal of International Development 14 (3).

Microfinance Information Exchange (MIX) et Groupe Consultatifd’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 Côted’Ivoire: Un secteur en plein essor. Le trésorier 24.

Mushinski, D.W., 1999. An analysis of offer functions of banks and creditunions 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, 6–9August 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, Discussionpaper 2007/03.

Nguyen, T.T.P., 2006. Factors influencing access to credit of households inrural areas of Vietnam case study of Tan Linh Commune, Ba Vi District,Ha Tay Province. Master Thesis in Rural Development with Special-ization in Livelihoods and Natural Resource Management SwedishUniversity of Agricultural Sciences.

Nimal, A.F., 2008. Managing Microfinance Risks Some Observations andSuggestions. 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 andcredit rationing in the informal financial sector in Uganda. Paper to the

DPRUT/Tips/Cornell conference on African development and Povertyreduction: The Macro–Micro Linkages.

Rankin, K.N., 2002. Social capital, microfinance, and the politics of devel-opment. Feminist Economics 8 (1).

Econom

of Six Rural Finance Institutions in Sub-Saharan Africa, World Bank

486 E.L. Togba / Structural Change and

Soulama, S., 2005. Micro-finance, pauvreté et developpement. desArchives contemporaines et en partenariat avec l’Agence Universitairede la Francophonie.

Schmidt, R.H., Kropp, E., 1987. Rural Finance Guiding Principles. GTZ,Eschborn.

Stiglitz, J., Weiss, A., 1981. Credit rationing in markets with imperfectinformation. American Economic Review 71 (3).

Swain, R.B., 2002. Credit rationing in rural India. Journal of EconomicDevelopment 27 (2).

van De Ven, W.P.M.M., van Praag, B.M.S., 1981. The demand of deductiblesin private health insurance: a probit model with sample selection.Journal of Econometrics, 17.

ic Dynamics 23 (2012) 473– 486

Woolcock, M., Narayan, D., 2000. Social capital: implications for develop-ment theory, research, and policy. World Bank Research Observer 15(2).

Wooldridge, J.M., 2001. Econometric analysis of Cross Section and PanelData, 1st ed. The MIT Press.

Yaron, J., Gurgand, M., Pederson, G., 1994. Outreach and Sustainability

Discussion Papers 248.Zeller, M., 1994. Determinants of credit rationing: a study of informal

lenders and formal credit groups in Madagascar. World Development22 (1).