Our sample consists of 321 group loan contracts granted by various second tier MFIs in Cameroon for the period 2007–2014.
3.1 Explanations and coding of socio-economic variables of group loan borrowers
The collective contract here represents the file of a loan granted to a collective and/or group of individuals generally grouped together within an association or a company (SARL or SA) for a well-defined period. The variables identified on the various loan files and loan officers' follow-up sheets are grouped into two categories: the socio-economic variables of the group or collective and the variables related to the loan and the borrower's repayment behaviour.
3.1.1 Socio-economic variables of group loan borrowers
The analysis of the files selected for our study shows the following variables: The age of the group refers to the time that has passed since the creation of the collective or the enterprise until the moment of the loan application. We observed that for each group loan applicant, the age is exclusively quantitative and in all the files analysed the ages range from less than one year to less than ten years. The applicant must provide information on his location plan, which must then be verified by the loan officer and/or the loan officer. Thus, during the analysis of the files, we observed two phenomena in particular: clients who did not specify their location plan (number 0 for code) and those who filed it (number 1 for code). The sector of activity specifies the activities carried out by the credit applicant. The activities listed in the various files have been grouped into two poles, namely the trade pole (services; sales, etc.) for which the number 0 is used as the code and the production pole (livestock, agriculture, etc.) for which the number 1 is used as the code. Wealth here refers to the assets held by the applicants. Generally, for reasons of reliability, group loan applicants mention their assets and their values. The observation of the files shows that the applicants who have mentioned their assets are coded as 0 and those who have not mentioned their assets or have no assets are coded as 1. Income here means the assets or income of the applicants (employed or not). The source of income characterises the permanence or otherwise of the loan applicants' income. Two types of applicants are identified: those with permanent income (coded as 0) and those with non-permanent or transitory income (coded as 1). The new project designates the nature of the activity for which the loan is requested. Here, we seek to ascertain whether or not the activity that requires financing is already being carried out by the applicants. Two cases are identified: those whose project is new (new activity) with the number 1 as the code; those whose activity is not new with the number 0 as the code. The variable presence of MFIs characterises the competitive environment of microfinance activities. Thus, this variable explains the nature of the client's environment, which may or may not allow him to have several lenders. The number 0 is retained for the code of applicants whose presence of other microfinance institutions is noted; and that of 1 for the code of other applicants. Guarantee refers to the joint and several guarantee offered by the loan applicant to the lender in case of non-repayment. Two cases have been identified, namely: applicants who do not offer any joint guarantee (code number 1) and applicants who have a joint guarantee (code number 0).
3.1.2 Group lending variables.
The follow-up sheet of the managers and/or loan officers provides us with information such as: the amount of the loan granted; the study time; the duration of the loan; the interest rate; the credit rationing; the number of loans already granted to the applicant. It should be noted that only accepted applications are made available to us. The study time represents the number of days taken by the credit committee for the acceptance or not of the loan application formulated by a client. On the follow-up sheets, we note that the time taken for the acceptance of credit applications varies from 01 to 28 days. This time has been grouped into 3 classes: the first class is made up of the values 01 to less than 15 days with the number 1 as the code; the number 2 represents the code of the second class with the interval from 15 to less than 24 days; the interval from 24 to less than 30 days represents the third class with the number 3 as the code. Credit rationing occurs when the amount of the loan received is different from the amount requested by the borrower. In terms of coding, the number 1 is for cases where rationing is observed and the number 0 for the opposite case. Loan duration means the number of loan repayment schedules granted to the borrower. Thus, the monitoring sheets count the number of repayment schedules ranging from 02 to 08 months for all the files analysed. Observations show that a repayment schedule is defined as either daily or monthly, depending on the case. The number of schedules were grouped into two classes: schedules ranging from 02 to less than 6 months with the number 1 as the code and the interval from 06 to less than 10 months represented by the number 2 in the coding. Interest rate represents the cost of the loan granted. Here it is determined by negotiation between the lender and the borrower on a principle based on the risk level of the lender. This essentially monthly rate follows an order from 12–28%. For the sake of completeness, the rates have been grouped into two groups in chronological order. Thus, the first group (code number 1) is composed of rates ranging from 12% to less than 24%, the second group ranging from 24% to less than 30% with code number 2. The number of loans granted to the borrower characterises the number of loan contract renewals. Here, the aim is to present the history of the relationship between the applicant client and the borrower. This number varies from 01 to 05 for all the files analysed in our study. The amount granted represents the monetary value received by the client as a loan. In all the files retained for our study, these values range from 100,000 to 1,600,000FCFA. The value of the amounts was grouped into 3 classes: the first class made up of values from 100,000 to less than 500,000FCFA with the number 1 as the code; the number 2 represents the code for the second class with the interval from 500,000 to less than 1,000,000FCFA; the interval from 1,000,000 to less than 1,700,000FCFA represents the third class with the number 3 as the code.
3.2 Results and interpretations
3.2.1 Overall significance of the model and interpretation of the coefficients
This study has developed a method of measuring default risk without revealing bank secrecy; the notion of pre-default. Any deadline not met on the day in question is a pre-default, revealing that information problems have not been properly resolved without imputing the incident to one of the parties. At the same time, this tool helps to prevent the actual failure. The following main results present the characteristics of the regression model used and an analytical description of them.
3.2.1.1 Characteristics of the regression model
Since Probit regression is only applicable to large sample sizes, checking for the absence of a multi-colinearity problem is of paramount importance. The Pearson correlation matrix shows that all correlation coefficients are significantly smaller than 0.8, which is the limit drawn by Kennedy (1985), at which one usually starts to have serious multi-colinearity problems.
Pr (Y = 1) = Ø (𝛽0+ 𝛽1 acti+ 𝛽2 npce+ 𝛽3ration + 𝛽4time+ 𝛽5rising + 𝛽6duration+ 𝛽7 wealth + 𝛽8 collat+ 𝛽9income + 𝛽10 age group+ 𝛽11new project + 𝛽12contact+ 𝛽13 local + 𝛽14presence+ 𝛽15Interest) +\(\epsilon\)
Number of obs = 321 Wald chi2(15) = 78.6 Prob > chi2 = 0.0000
Log pseudolikelihood = -180.82904 Pseudo R2 = 0.1870
We can see that the R2 is 0.187. It represents the variance explained by the model (explanatory power). In other words, in this case, the model expresses 18.8% of the variance of the dependent variable. This rate may seem low, but it can be explained by the relatively exploratory nature of the present study. Similarly, it can be attributed to the diversity of the variables which are related to the borrower, his activity and the institution. The Chi-square test tests the null hypothesis that all explanatory variables are not significant in aggregate. The table shows a Chi-square of 78.63 for 15 degrees of freedom. This value is significant at the 1% level. This finding allows us to state that our model is significant as a whole.
3.2.1.2 regression results: estimated model coefficients
Before moving on to the interpretation of the estimated coefficients of the model, the overall quality or significance of the model is questioned. In terms of checking the strength of association of the model, the results are as follows: Among the 15 explanatory variables included in the model, only 06 variables are statistically significant. The following table illustrates these results.
Table: Variables in the group lending equation
|
dy/dx
|
Odds (Exp(B))
Exp(B) ratio
|
P>|z|
|
Sector_ of activity
|
− .3809829
|
.2062825
|
0.000*
|
Npce
|
− .020941
|
.9403938
|
0.628
|
Rationing_credits
|
.4451118
|
9.173488
|
0.000*
|
Time
|
− .0117354
|
.946952
|
0.345
|
Amount
|
5.38e -08
|
1
|
0.572
|
Duration
|
.0520138
|
1.216818
|
0.133
|
Interest rate
|
.0186287
|
1.08333
|
0.034**
|
Guarantee
|
.405486
|
5.590876
|
0.000*
|
Contact
|
.01330223
|
1.80844
|
0.261
|
Wealth
|
− .0294996
|
.891409
|
0.673
|
Source_of_income
|
.1479667
|
1.953977
|
0.149
|
Age of the group
|
.0033587
|
1.013015
|
0.856
|
Location
|
.2075775
|
2.316239
|
0.004**
|
New _project
|
− .245789
|
.3419977
|
0.000*
|
Presence_ imf
|
− .0372027
|
.8442209
|
0.596
|
Constant
|
− .0372027
|
.7898405
|
0.596
|
*significant at 1% level, ** significant at 5% level |
Source: Our results |
This table allows us, on the one hand, to observe the variables that have been included in the equation and, on the other hand, to examine their significance (box P>|z|.). When the variables are significant, we proceed to the interpretation of the odds ratios (Exp(B), or "Odds Ratios"). These odds ratios correspond to the number of times the group membership is increased when the value of the independent variable increases by 1 (i.e. in the case of binary variables when moving from one state to another). A odds ratio greater than 1 indicates an increase in the likelihood of being in the pre-default group, while a ratio less than 1 indicates a decrease in the likelihood of being in this group.
3.2.2 Interpretation of results
The results from the model indicate that the amount (size) of credit extended to a borrower is associated with a significant and positive coefficient. This is because the amount of credit granted to the borrower is determined by the loan officer based on the borrower's estimated probability of default. As a result, the borrower will tend to modify his or her project; this generally causes a disruption in the realisation of the project. These results are in line with those of Sharma and Zeller (1997) and Godquin (2004): the difficulty of repaying a loan increases logically with its amount. The commercial and service activities carried out by the borrowers generate sufficient income to maintain a good credit relationship with the microfinance institution. This finding allows us to conclude that the sector of activity and more specifically the commercial and service activities carried out by the borrowers reduce the chance of being a pre-default borrower by 38.09%. This result can also be attributed to the variability of production income such as agriculture and livestock which makes this industry risky. Being rationed is positively related to the probability of default. There are several reasons for this. Firstly, when the amount requested is different from the amount granted, the borrower will tend to modify his project, which generally leads to a disruption in its realisation. Secondly, the willingness of the microfinance institution to test borrowers with small loans increases the probability of default to the extent that the amount received does not cover all the costs of the borrower's project. Finally, it is often the opposite that is presented insofar as the borrower is concerned about the relationship with the credit agency, credit rationing will motivate the latter to increase these efforts and consequently reduce the probability of default. The significant coefficient associated with the credit rationing variable indicates that it is likely to increase the chance of pre-default by 44.51%. It is clear from the results of our study that the presence of a joint and several guarantee is positively correlated with the risk of default. It increases the chance of joining the group of pre-default borrowers by 40.54%. This result stems from the divergence of individual and collective interests on the one hand and the difficulty of managing the group on the other. These results are contrary to the results of Sengupta and Aubuchon (2008), who found that joint and several guarantees favoured a repayment rate of almost 90%. This is still debatable as late payment is not synonymous with default. The location of the group's places of activity by the institution is likely to increase the chance of being a pre-defaulter and qualified as a bad borrower by 20.75%. This location does not favour a methodical monitoring of the group by the lender, which will push the borrower to behave less well, as they are not always the managers of the project and/or the company. Generally, the lender's visits to the business are sometimes unsuccessful, as the managers of the company refuse to give information about the poor financial health of the entity. But they present better situations to lenders. The fact of having the contact of the representative of the collective or group by the institution increases by 13.30% the chance of having a delay in the payment of loans. This is justified when the representative does not have good information about the group and/or collective. The presence of other microfinance institutions in the same geographical area that offer the same service as the institution in question is likely to decrease the chance of the group being classified as pre-default by 3.72%. The probability of borrowers falling into arrears decreases with the presence of alternative sources of finance. This is justified by the fact that the borrower is more concerned about financing since his chances of financing are improved. Thus, even if there is a delay with the first microfinance institution, the borrower risks being rejected by the second microfinance institution in his or her region because of the diffusion of information through the proximity link. This is especially important as microfinance institutions through the proximity link share information on borrowers. The risk of insolvency increases when the borrower (the group) has a permanent income (salary, pension). This stable income should be an alternative to be exploited in case of repayment problems, but the opposite is observed. That it can be justified by the fact that generally the group made up of wage-earners relying on their wages or pensions is no longer concerned with the repayment of credit knowing that it will have the members' wages. For a group whose members are wage earners, we observe a 14.19% higher chance of being a pre-default borrower. New projects are more at risk than old ones. This result hardly justifies the negative correlation with late payment risk. According to the results of our study, this variable decreases by 2.45% the chance of repaying late and joining the group of pre-default borrowers. This degree of risk associated with new projects can be attributed to the search for a good reputation of new entrepreneurs as well as to the limited likelihood of risk taking. The risk of pre-default decreases if the borrower has wealth. This wealth, which is generally stable, is used as an alternative to exploit in case of repayment problems. This can be justified by the fact that generally the borrower with assets is more concerned about repaying the loan. For a group with assets, we observe a 29.49% lower chance of being a pre-default borrower. Our investigation shows that the age of the collective increases by 3.35% the chance of obtaining loans that will be in arrears. The interest rate set by the microfinance institution is positively correlated with the risk of late payment. This is unexpected since, as the interest rate is usually integrated in the overall amount of the debt, the borrower seeks to repay on time because otherwise he will be obliged to pay the penalties based on the said amount. And as sometimes this rate seems to be high, if there is a delay, the borrower will pay a high amount of penalty. The time it takes to apply for a loan is negatively correlated with the risk of late repayment. This is because when the application time is spread over a long period of time, lenders tend to take a good look at all the parameters on the stability of the group. It decreases the probability of late repayment by 11.73%. The results from the model indicate that the number of loans granted to the group measuring the long-term relationship is negatively related to the probability of pre-default. There are several possible reasons for this finding. These results support the view that the long-term relationship measured by the number of loans granted to a borrower reduces the risk of pre-default. New projects are more at risk than old ones. This result justifies the positive correlation with default risk. According to the results of our study, this variable increases by 1.891 the chance of repaying late and joining the group of bad borrowers. This degree of risk associated with new ventures can be attributed to the lack of experience of new entrepreneurs as well as the limited likelihood of risk taking. The results of our study show that the risk of late repayment increases with the duration of the group loan. This result can be justified by the fact that the spreading of a loan over a long period of time causes a race to individual interests in the group, which will then dominate those of the collective. This is a source of the group's fragility, as we observe a 5.20% chance that long-term loans are in arrears.