Indeed, an old practice born from the mismatch of supply and demand on the formal credit market, microcredit was theorized at the end of the 20th century with significant success in the 1990s and 2000s. The 2006 award of the Prize Nobel Peace Prize to the pioneer of microfinance Muhammad Yunus gave a favorable response to the use of this financial tool for the fight against poverty in developing countries.

Microcredit is a local loan adapted to the needs of a small amount, significantly lower than the credit that a business or household can request from a bank ( Lelart, 2005 ). It is a loan for the creation or development of very small income-generating activities to people who do not have access to the traditional financial circuit due to lack of real guarantee. It is also a small amount loan with a short repayment period to the poor who do not have material guarantees to carry out an economic activity. Dedicated in its most common sense to helping to finance income-generating activities, microcredit appears in the field of sanitation as an innovative financing relay, intended to enable the improvement of the sanitation situation. in regions where limited tax revenues and insufficient subsidies do not yet make it possible to generate the public funds necessary for community sanitation ( Assié, 2013 ). It appears to be a still emerging response to the crucial question of sanitation in developing countries. According to Mehta the potential market for microcredit for sanitation in the years to come will expand. It is with this in mind that public or associative sanitation promoters offer programs in favor of sanitation including the use of microcredit. These microcredit programs, which improve access to sanitation for populations in developing countries, can be grouped into three large families: loans to households, loans to small entrepreneurs and solidarity mechanisms to improve overall sanitation situation, particularly in urban areas ( Mehta, 2008 ).

Indeed, direct loans to households to facilitate access to sanitation are the most developed. This type of credit is better adapted to the current capacities of microfinance institutions (MFIs) and could be subject to the creation of standardized types of products. However, low savings in certain areas can be a limiting factor, or even lead to the need for initial external financing. Additionally, for business loans Mehta sees the potential for lending to small entrepreneurs in the sanitation sector as more limited. Finally, grouped individual loans for access to collective infrastructure take place mainly in urban or peri-urban neighborhoods. The role of an MFI can then be to mobilize the savings of community members, and possibly to provide loans to households with stable incomes but low savings capacity, who build collective latrines together. This type of mechanism leads to a reduction in management costs borne by the MFI ( Assié, 2013 ).

The determinants of microcredit repayment in microfinance.

Define abbreviations and acronyms the first time they are used in the text, even after they have been defined in the abstract. Abbreviations such as IEEE, SI, MKS, CGS, sc, dc, and rms do not have to be defined. Do not use abbreviations in the title or heads unless they are unavoidable.

The literature on the determinants of repayment performance in a microfinance institution highlights the existence of information asymmetries. Viganò (1993) analyzed the risk factors for repayment default which are the ability to repay, the willingness to repay, the external economic and environmental conditions, the quality of the information which serves as a basis for decision for the granting of loans. credit, the ability of the lender to ensure the goodwill of the borrower through an optimal contract. Boot (2000) notes that the long-term relationship helps to reduce the problems of adverse selection and moral hazard which can result in particular from new borrowers. Guarantees also play an important role in the analysis of a loan repayment default. From a theoretical point of view, guarantees are seen as mitigating elements of the adverse selection problem faced by the lender ( Besanko & Thakor, 1987 ). The guarantees then act as a signal allowing the lender to reduce or eliminate the adverse selection problem caused by the existence of information asymmetries between lenders and borrowers. However, the nature and evaluation of real guarantees poses a problem, because the evaluation made of guarantees is most often erroneous and lacks objectivity. Using group credit is one solution. Several authors have demonstrated the performance of this type of credit. However, these loans too are prone to problems. Simtowe and Zeller (2006) summarize these problems into four categories and provide theoretical solutions to this type of loan. Firstly, the authors mention that peer selection makes it possible to fight against adverse selection, because members will avoid associating with those whose risk profile they do not know well. This principle allows the formation of homogeneous groups in which all members know perfectly the characteristics of their partners relating to the risk of the project and the level of solvency ( Van Tassel, 1999 ; Ghatak & Guinnane, 1999 ). The second category is related to peer monitoring to reduce the choice of a very risky project and the diversion of funds for other purposes. It constitutes an important element enabling the success of group credits ( Stiglitz, 1990 ; Aghion, 1999 ). Wenner (1995) , however, observes that the group must be small. The third category relates to the failure of the project for exogenous reasons or which are beyond the control of the group member. Intra-group solidarity, however, makes it possible to ensure repayment on time ( Huppi & Feder, 1990 ). The last category is linked to mutual pressure to avoid default in repayment. This peer pressure makes it possible to reduce moral hazard ex-post and avoid the risk for the group of being deprived of future credit ( Wydick, 1999 ). However, Diagne et al. (2000) report that pressure can have a negative or weak impact on repayment. Of these four elements, other factors have been mentioned in the literature. These include social capital and external credit options. These arguments therefore make the group with joint guarantee a more efficient instrument than individual credit.

This research took place in the Bouafle department, located in the west center of Côte d’Ivoire in the Marahoue region of which it is the capital, with an area of 4214.5 km² ( Conseil Régional de la Marahoué, 2017 ). It is located 60 km from the autonomous district of Yamoussoukro, the political and administrative capital and 310 km from Abidjan, the economic capital. Bouafle is limited to the east by Yamoussoukro, to the west by the department of Daloa, to the north by the department of Zuenoula, to the south by the department of Sinfra ( Conseil Régional de la Marahoué, 2017 ). The relief is composed of low plateaus with some lowlands and hills with an average altitude of 260 meters. The hydrographic network is made up of the Marahoue or Red Bandama River which crosses the department from North to South, rivers and lakes. The climate of Bouafle is Baoulean type. It has two rainy seasons and two dry seasons. The average temperature is 25.30˚ celsius and precipitation varies from 1800 to 2000 mm ( Conseil Régional de la Marahoué, 2017 ). The population of the Bouafle department amounts to 300,305 inhabitants including 161,857 men and 138,448 women with an average population density of 71.26 inhabitants/km² ( Institut National de la Statistique (INS) et ICF International, 2012 ).

The study population is all rural households residing in the Bouafle department which benefited from the CLTS-ECOSAN approach of the Millennium Water and Sanitation Program.

Inclusion criteria

Non-inclusion criteria

This is a longitudinal study with an analytical aim, which took place over the period from 2014 to 2017. The first pre-project survey took place from January 5 to January 31, 2014. The second took place took place from January 5 to January 31, 2015, the third survey from January 5 to January 31, 2016. The fourth survey from January 5 to January 31, 2017.

Let N be the size of the population, with N = 173 and n the size of the sample, we set the confidence threshold at 95% with a precision level of ±5%.

We therefore write the following formula:

$n=N/\left(1+N{e}^2\right)$.

n = 173/(1 + 173 × 0.05 × 0.05) = 121 households receiving microcredit for TSDU endowment.

To avoid data loss, we increased the size of our sample to 10% (that is 12 households), n = 133 households receiving TSDU.

The 133 households in the sample were distributed proportionally to the respective weight of each locality over all 10 localities by the following formula:

${n^{\prime }}_i=n\ast P_i$,

with:

n: sample size (n = 133),

P_i: The weight of locality i in terms of TSDU beneficiary households, with P_i = n_i/N,

n_i: the number of households in village i,

${n^{\prime }}_i$: the sample size in village i,

N: total number of households benefiting from TSDU from the study (N = 173).

Depending on the inclusion criteria defined, ten localities (village, camp) were targeted as part of our study. Following the population census in these localities, we counted 173 heads of households receiving microcredit for the acquisition of TSDU distributed as follows in Table 1 below.

Source: Author.

It concerns data collected from rural households during the implementation of PHAM in Côte d’Ivoire in the department of Bouafle. To collect data, we carried out three household surveys using a questionnaire with heads of households receiving TSDU or representatives in 2015; 2016; 2017.

A pre-survey was carried out on December 12, 2014 with 14 heads of households in the village of Kouakoublekro, in order to test the questionnaire, correct all errors and ambiguities. This pre-survey allowed the investigators to familiarize themselves with the different questions, have a common understanding of the questionnaire and understand the survey methodology.

Three household surveys were carried out as part of the analysis of the determinants of microcredit reimbursement intended for the acquisition of TSD.

The first survey took place from January 5 to 31, 2015, the second survey from January 5 to 31, 2016 and finally the third survey from January 5 to 31, 2017. These surveys covered all ten study locations and only concern heads of households benefiting from TSDU or their representatives. Accompanied by a translator, the information collected concerns the socio-economic characteristics of the households benefiting from TSDU, the cost of production of TSDU, the microcredit mechanism set up by the EAA agency, the monitoring by the village committee, the monitoring carried out by collection agents, production of biofertilizers, reimbursement rate. The interview lasted on average 20 minutes per household and the different responses of the respondents are recorded on the questionnaire form.

For data processing we used Excel and STATA 14 software. After a descriptive statistical analysis to study the socio-economic characteristics of TSDU beneficiary households and the repayment performance of the TSDU microcredit, we evaluated the determinants of the repayment rate TSDU microcredit.

1) Econometric Modeling

The endogenous variable is quantitative and the explanatory variables are quantitative and qualitative. We will use linear regression to evaluate the influence of each explanatory factor on the reimbursement rate.

Consider the following model:

$T_i={\beta }_0+{\beta }_i{X}_i+{\mu }_i$. (1)

With T_i the endogenous variable “Reimbursement rate of household i” and X the following exogenous variables:

X_1i: Marital status,

X_2i: Occupancy,

X_3i: Level of education of the head of household,

X_4i: Number of workers in the household (Active),

X_5i: Number of follow-ups by the PHAM facilitator in the household,

X_6i: Number of follow-ups by the village committee in the household,

X_7i: Quantity of hygienized urine applied in agriculture,

X_8i: Quantity of hygienized feces applied in agriculture,

X_9i: Cocoa production per hectare,

X_10i: Cocoa production by labor,

X_11i: Daily income per capita,

X_12i: Leadership of the chieftaincy committee,

X_13i: Leadership of the youth committee,

X_14i: Leadership women’s committee.

β₀, β₁…, β₁₄ are the coefficients of the model to be estimated and µ_i is the error term.

2) Robustness of test

We performed statistical tests to ensure the validity of our multiple linear regression model. The first test evaluates the overall significance of the regression coefficients using the Fisher test. Two hypotheses are formulated. H0: the coefficients are generally insignificant, against the alternative hypothesis H1: the coefficients are generally significant for a P-value less than 0.05.

The second test is that of Breusch-Godfrey, it is used to check the autocorrelation of errors. Two hypotheses are used. We have H0: autocorrelation of errors against H1: the residuals are uncorrelated.

The third test is that of Breusch-Pagen. It is used to check the homoscedasticity of errors with the following hypotheses: H0: the hypothesis of homoscedasticity of errors against the alternative hypothesis H1 of heteroscedasticity of errors.

The descriptive analysis of the data to know the profile of households equipped with Ecosan latrines reveals the following results:

The heads of households benefiting from TSDU are almost men (97%) and live as a couple (95%). They are agricultural (98) and more than half are not educated (57%), only a quarter of the heads of household have reached primary school, then the rest have at least secondary level (17%). The characteristics of the housing of TSDU beneficiaries indicate that the housing is largely built with local materials (clay, earth brick). The roofs are mostly made of sheet metal (86%). Many households use lighting sources specific to rural areas, including 47% for solar panels and 41% for other forms (oil, candle, battery). Only 11% of households are connected to the national electricity network (CIE). Regarding sanitation, only a quarter of households practice open defecation and a significant portion of the population has an improved latrine (83%). As part of the agronomic valorization of sanitation by-products by households, we note an increase in the use of biofertilizers (sanitized urine and feces) by households during the PHAM with a strong application in cocoa cultivation, resulting in an increase in cocoa production yield (see Tables 2-4 ).

Source: Our calculations using STATA.

Table 3. Average production of biofertilizers and cocoa crop yield in 2015.

Source: Our calculations using STATA.

Table 4. Average production of biofertilizers and cocoa crop yield in 2016.

Source: Our calculations using STATA.

In terms of household resources, we note on average an increase and stability respectively in income and consumption per capita during the implementation of PHAM (see Tables 5-7 below). Only a quarter of TSDU beneficiaries have a daily per capita income of more than 737 CFA franc. The analysis of household debt relating to the granting of microcredits for the acquisition of TSDU reveals a low level of repayment of TSDU beneficiaries although the overall repayment rate is 24.84%.

Source: Our calculations using STATA.

Table 6. Return on TSDU microcredit repayment and income per capita in 2015.

Source: Our calculations using STATA.

Table 7. Return on TSDU microcredit repayment and income per capita in 2016.

Source: Our calculations using STATA.

The stratification of the TSDU debt indicates that half of the households have liabilities of more than 45,000 FCFA, representing more than three quarters (3/4) of the cost of the TSDU (57,000 CFA franc) and only 5% of households were able to pay off their debt. TSDU debt.

We use the Variance Inflation Factor test at the threshold of 10 or tolerance 0.1.

The Variance Inflation Factor (VIF) test indicates values less than 10, so there is an absence of multicollinearity of the independent variables (see Table 8 ).

Source: Our calculations using STATA.

To estimate the vector of parameters β we will use the ordinary least squares (OLS) method. The estimation results are reported in Table 9 .

Source: Our calculations using STATA, ***p < 0.01, **p < 0.05, *p < 0.1.

1) The multiple adjusted determination coefficient ${\bar{R}}^2$

${\bar{R}}^2$ = 0.835: the multiple adjusted determination coefficient is close to 1 (SEE Table 9 above), the regression of the endogenous variable (TSDU microcredit repayment rate) on the exogenous variables best explains the variance in the repayment rate TSDU microcredit.

2) Overall significance of the regression coefficients (fisher test)

Let us test the null hypothesis H0: the coefficients are globally insignificant, i.e. all the coefficients of our multiple regression are zero, against the alternative hypothesis H1: the coefficients are globally significant, i.e. say all the coefficients of our multiple regression are globally different from 0.

We note that the p-value = 0.0000 of the Fischer test is less than 0.05 (see Table 9 above), therefore we reject H0, the coefficients of the multiple regression are overall significant at the 5% threshold.

3) Error autocorrelation test

The Breusch-Godfrey test is used to check the autocorrelation of errors. We formulate the following hypotheses:

H0 the errors are autocorrelated, against the alternative hypothesis H1 of absence of autocorrelation of the errors.

The test provides us with a p-value = 0.1658 > 0.05 (see Table 10 below), we reject hypothesis H0 of autocorrelation of errors, therefore the residuals are uncorrelated.

Source: Our calculations using STATA: p-value = 0.1658 > 0.05 we reject hypothesis H0 of autocorrelation of errors.

4) Homoscedasticity test

We used the Breusch-Pagen test with the following hypotheses:

H0: the hypothesis of homoscedasticity of errors against the alternative hypothesis H1 of heteroscedasticity of errors. The test indicates a p-value = 0.1158 > 0.05 (see Table 11 below), we accept the hypothesis H0 of homoscedasticity of errors.

Number of observations = 133, F (14, 118) = 1.52, Prob > F = 0.1158, R-squared = 0.1524; Adj R-squared = 0.0518. Source: Our calculations in Stata. p-value = 0.1158 > 0.05, we accept the hypothesis H0 of homoscedasticity of errors.

Considering the signs and values of the coefficients of the explanatory variables obtained from the linear regression at the 5% threshold, we note that:

The increase in cocoa productivity of one additional unit in the household generates an increase of 0.015 points in the TSDU microcredit repayment rate, all else being equal. An increase of one additional unit in the daily income per capita in the household induces a reduction of 0.008 points in the reimbursement rate of the microcredit granted for the acquisition of the TSDU, all other things being equal. An increase of one unit in the frequency of monitoring by the local (village) committee in the TSDU beneficiary household increases the reimbursement rate of the microcredit granted for the acquisition of the TSDU by 2 points, all other things being equal. A unit increase in the number of follow-ups by the PHAM facilitator in the TSDU beneficiary household increases the reimbursement rate of the microcredit granted for the acquisition of the TSDU by 3 points, all else being equal. An additional unit of liter of hygienized urine applied to crops in the TSDU beneficiary household increases the reimbursement rate of the microcredit granted for the acquisition of the TSDU by 0.007 points, all other things being equal. An additional unit of kg of hygienized feces applied to crops in the TSDU beneficiary household increases the reimbursement rate of the microcredit granted for the acquisition of the TSDU by 0.01 points, all other things being equal. Very active participation of the chieftaincy committee in the implementation of the PHAM leads to an increase of 16 points in the reimbursement rate of the microcredit granted for the acquisition of the TSDU, all other things being equal.