[Survey Information in More Detail]
Number of Loan Cycles. This is the dependent variable—my measure for length of relationship.
Business Variables. In this model, average monthly profits are used to capture the individual’s economic return on her loan. It is assumed that individuals with more profitable businesses are more likely to stay in the relationship.
Loan Terms. Larger loan sizes diminish the probability of staying in the relationship, especially towards the beginning. This occurs because individuals sometimes find it difficult to pay back the loans when the loans are too big. Average loan size for the group will be used as the proxy for loan terms.
Group Dynamics. The quality of the borrower’s other group mates is a huge determinant of dropout. One is penalized if one member of the loan group fails to make the repayment, because everyone else must pay it for the defaulter (Kaffu 2005). This frustrating situation causes many to dropout. To capture the effect of other group members’ nonpayment a proxy covariate, group repayment problems, will be used. This covariate is a dichotomous variable, 1 if the individual’s group did experience repayment problems over the course of the relationship, and 0 otherwise.
How well the group members know each other is also a very important variable. The proxy for this relationship is called past borrowing experience with group members. Those who had prior borrowing experience with other group members are less likely to leave the relationship.
The third and last proxy for group dynamics is regularly held group meeting. Borrowers in groups that held meetings on a regular basis remain longer, their hazard of exiting is lower, relative to those borrowers in groups that did not.
Competition. It is hypothesized that the more competition there is for the MFI, the higher the likelihood of exit. There are numerous different sources of competition: formal financial services, e.g., banks, other MFIs, cooperatives, NGOs, etc., and informal financial services, e.g., family, local moneylenders, and friends (Kaffu 2005). To proxy this effect, I will use access to informal financial services and access to formal financial services in the first two periods of the banking relationship.
Idiosyncratic Shocks. Past literature all state the importance of external factors in determining the length of the banking relationship (Maximambali 1999, Murray 2001). Idiosyncratic shocks (individual and household level) shocks, such as births, deaths, chronic illness, ceremonies (weddings/baptisms), fire, theft, etc., played a significant role in the termination of the borrowing relationships. Past research has found that more idiosyncratic shocks lead to higher hazard. To measure this effect, the monetary value of household income shocks in period one and two is used.
Household Characteristics. The number of people in the borrower’s household could also affect the hazard rate of the borrower. When a household has a higher number of dependents, there is a lower likelihood of exit. The dependency ratio is defined as the number of those in the household who are under 15 and over 60.
Individual Characteristics. Past literature has also cited the importance of demographic characteristics to client exit rates. This study uses education and age. Education may positively influence longer banking relationships. Intuitively, higher age may also decrease hazard (Pagura 2003). Going a step further than Pagura, I will also include a variable for whether the individual is of an agricultural background or urban background. I will use a dummy variable agriculture, inputting a value of 1 if the individual is agricultural and 0 if otherwise.
