Abstract

Microcredit has been successful in bringing formal financial services to the poor, but given that many microcredit clients live in poverty, this success has sparked a debate surrounding the question of how to set interest rates. In Ghana, researchers set out to measure how different interest rates on individual loans affect demand for the loans and if and how different interest rates affect borrower profile. However, due to low participation rates, this study was unsuccessful in measuring the outcomes of interest. This “failure” is featured in the book Failing in the Field by Dean Karlan and Jacob Appel. 

Policy Issue

Now with more than 200 million borrowers, microcredit has been successful in bringing formal financial services to the poor. This practice has sparked a debate surrounding the question of "fair interest rates," particularly given the extreme poverty of many microfinance clients. Arguments in defense of higher rates include assertions that they are necessary to cover the high costs of issuing even small loans and that rates will drop as more institutions enter the market. But these arguments remained untested, and the question of what a "fair interest rate" is was unclear. unanswered. In light of this, the extent to which demand for credit changes in response to changes in the price of loans (in this case, interest rate shifts) is of key interest to policymakers and practitioners. This research sought to contribute evidence on this topic. 

Context of the Evaluation

Opportunity International Savings and Loans, Ltd (OISL), one of Ghana’s largest microfinance institutions, was interested in the implications of interest rate for both revenue and outreach. Before this study, OISL had offered joint-liability loans, but not individual liability loans. The study took place among potential clients of OISL—micro-entrepreneurs in busy urban areas of Accra, Ghana.

Details of the Intervention

Working with Innovations for Poverty Action, researchers set out to measure how different interest rates on loans affect overall demand and take-up according to borrowers’ poverty levels. Researchers randomly assigned 180 clusters (small geographic areas), spread across 12 busy commercial areas in Accra, to one of four interest rates: 24%, 31%, 38% (OISL’s normal rate), and 45%.

OISL credit officers conducted door-to-door marketing of the loans. They visited potential clients at their businesses, conducted a short six-question survey to estimate the person’s poverty level, presented a promotional flyer (which included the interest rate along with some marketing variations that were also part of the experiment), and briefly described the loan offer, a personal loan for micro-entrepreneurs.  

Applicants had to name a guarantor whose income was sufficient to cover the debt and who committed to pay if the applicant defaulted. To further establish creditworthiness, applicants had to provide information about their business assets and revenues, which had to be verified by a visit from a loan officer. Applicants could request loans with maturities from 3 to 12 months, and were expected to use the money to expand or improve their businesses.

Between door-to-door sales, special offers, and database software, the study added new routines to a variety of bank operations. To identify potential implementation challenges, the research team conducted a limited pilot in one branch prior to launch. The pilot ran just long enough to test the marketing and in-branch routines, but not long enough to observe people as they made their way through the loan application process.

Results and Policy Lessons

Due to low participation rates this study was unsuccessful in measuring the outcomes of interest (although “expressions of interest” were high, not enough people were able to complete the qualification requirements). This “failure” is featured in the book Failing in the Field: What We Can Learn When Field Research Goes Wrong by Dean Karlan and Jacob Appel.

Reasons for Failure

While demand initially appeared high—about 15 percent of business owners came to the branch to inquire about the loan, more than had done so in the pilot—a cumbersome and long loan application process proved to be a major barrier in actual take-up of the loan. The first and largest hurdle was the guarantor requirement. Most applicants had difficulty finding family or friends to commit to cover a large loan. Applicants also had to complete a large amount of paperwork, which required coming to the branch in person and providing extensive documentation to verify income or wealth. Moreover, the loan application process took on average one and a half months, which was too long for applicants with time-sensitive needs.

In the end, for every 100 business owners that received an offer, about 15 responded by visiting a branch. Of these, about five (4.7%) started an application, and only about two (1.8%) completed it. And of those two, just one (0.9%) actually took a loan. Thus the total lending of about 30 loans was immaterial to OISL from a bottom-line perspective and far too small to justify the large expenditure on door-to-door marketing.

In sum, there are two main reasons why participation rates were so low. The first related to the research setting: The personal loan product was developed just before the study began and had not gone through an appropriate tinkering phase, to iron out enough kinks such that the product was market ready. Second, the study placed too high a burden on OISL’s staff at the level committed to the product line. Because of this, staff were unable to process the applications in a timely manner.

Lessons Learned

• In hindsight, the research team should have followed the pilot clients all the way to loan disbursement; this might have uncovered problems that drove attrition in the full study. The product should have been through a more thorough “tinkering” phase, perhaps in a different geographic area, before embarking on the study.
• Plans that rely on full-time employees (in this case, partner staff) finding extra hours in their days to complete additional tasks are likely to cause delays.
• Stringent application processes can be a major barrier to take-up of financial products, which for research purposes means high attrition and, possibly, low statistical power.

Note: IPA worked with researchers to conduct a similar, successful study in Mexico.