What a micro lender does after giving a loan matters too
Microfinance was created in part because banks were not meeting the needs of the large segments of the population. One of the barriers to reaching those segments was the fact that formal credit records for those clients were nonexistent, leading microfinance institutions to consider other information instead. And for decades, the microcredit industry has considered two main factors when doing credit analysis: ability and willingness to pay.
The industry has changed a lot since the great Pancho Otero, the visionary behind the birth of Bancosol in Bolivia, said that you know a person is a good credit risk when you see the twinkle in their eye. Nowadays, you go to a microfinance conference and you are surrounded by discussions on big data instead.
Based on the experiences of larger banks and consumer lending companies in developed countries, new enterprises are promising new scoring models that make use of a variety of data left by clients both online and offline. In other words, there’s hope big data might change financial inclusion drastically. For example, much is being said about mining cellphone usage of potential clients as a proxy to assess credit worthiness. This is the attractive premise behind Cignifi, a US based company that was selected during the MIF´s first call for ideas for the Technologies for Financial Inclusion program in 2010. However, access to a range of client data will not translate easily into the way microcredit is done in Latin America or the Caribbean. A lot of trial and error will be needed.
In my view, the hypothesis behind using nontraditional client data for credit scoring in microfinance has challenges. In the field, after the assessment of ability and willingness to pay is done, the likelhood of repayment from a particular client will depend mainly on what the microlender does to recover the loan ex-post, not on what the microlender knew before making the loan.
This becomes clear when you see that the same clients may repay the loans to one entity but not to another. In some cases, risky clients say that they repay institutions which make a better effort to visit them and follow up on the loan payments, while they fall delinquent with other entities. A study of loan delinquency in 49 MFI’s in Kenya, shows that low loan delinquency is correlated to institution-specific activities that take place after origination, such as weekly borrower group meetings which build up financial discipline.
Credit risk is not a set condition; it depends on what is done after the credit is approved. In fact, in many cases village banking clients who received loans without a credit risk analysis are often better clients than individual clients that were subjected to detailed credit risk analysis. In fact, average loan delinquency rates for village banking clients in LAC tends to be lower than that of individual lending clients.
My proposal for microfinance institutions it is to seek better credit data -- even big data!-- as a way to inform decision making, while implementing different strategies on a client-by-client basis depending on available information. For example, allowing a less frequent schedule of loan officer visits for those clients whose data provides a better assurance of repayment. In other words, better credit information would be the first step for a successful strategy, but after that, high touch loan recovery tactics might still be needed.