The risk of exit by borrowers from a microlender in Bolivia

Publication Date: 1999
Published by: St. Louis, USA: Microfinance.com

View this Document   Email this Document

Analysing the risk of delinquency

A credit scoring model reveals that exit risk is greater for newer borrowers, women, manufacturers, and those with more past arrears

Risk also depends on the amount disbursed, the loan officer and the branch, and the time since the first loan

Although the knowledge of quantitative characteristics cannot replace knowledge of qualitative character, the model predicts well enough out-of-sample to help lenders know where investment in relationships with specific borrowers is most likely to curb the risk of exit

Highlights:

• models that predict risk from traits of the borrower, the lender, and the loan are a common way to reduce the costs of credit-card lenders, the microlenders of high-income countries
• the model here pinpoints a host of factors that affect the likelihood of exit for a microlender in Bolivia. Furthermore, in out-of-sample tests, the model predicted exit much better than naïve models
• in microlending, knowledge of characteristics from computers will not replace knowledge of character and cash flows from personal contact. Scoring does, however, show some promise as a way to mark which cases to check first

• much work on scoring for microlending remains
• lenders will likely want to model default, and they may also want to estimate the risk that a borrower x days in arrears now will remain delinquent for at least y more days
• no one foresaw that scoring would replace individual analysis for consumer loans in high-income countries (Lewis, 1990)
• further work will show the extent to which scoring can help cut the costs of microlending in low-income countries