Back to History

Fields Changed

Registration

Field Before After
Abstract Repayment rigidity has been shown to be particularly unfavourable for microfinance borrowers, especially in terms of investment potential: as most business activities take long to attain profitability, immediate repayments may be a great obstacle to entrepreneurship potential. Building on this premise, our intuition is that the provision of repayment flexibility will help innovation-oriented entrepreneurs who face serious growth constraints in the absence of an adequate repayment structure. To this end, the study has been set up as an RCT with a randomly selected group of microfinance borrowers under the individual-lending methodology being offered the possibility to choose between a flexible and a rigid repayment structure, although the rigid contract allowing for a lower degree of flexibility, will be cheaper and will require less collateral from the clients than the flexible one. The impact of the provision of a flexible schedule in terms of production growth, employment and income will be observed. With this experiment, we intend to study how preferences for repayment flexibility relate to customers’ characteristics and business performance. Repayment rigidity has been shown to be particularly unfavorable for microfinance borrowers, especially in terms of investment potential. This study proposes an innovative way for Microfinance Institutions (MFIs) to offer repayment flexibility in microfinance contracts, which uses contract price as a screening mechanism. The underlying intuition is that offering repayment flexibility as a more expensive contract option than the standard rigid contract can work as a screening instrument for lenders, with borrowers selecting into the contract that best suits their characteristics. This, in turn, can mitigate default rates. We design and set up a Randomized Controlled Trial (RCT) in Uttar Pradesh, India, to test this hypothesis: in treated branches, the lender offers a flexible repayment option along with the standard rigid contract, the former being more expensive than the latter. In control branches, only the standard, rigid contract is available. Our experimental design allows us to study two things: - By focusing on treatment branch only, we can look at borrowers’ selection into flexible vis-à-vis rigid contract by linking borrowers’ characteristics with contract choice - By comparing treatment (contract choice) with control (only standard contract offered), we can study the impact of offering a flexibility option on business outcomes and repayment rates. Our work specifically focuses on the design of microfinance contracts, and in particular on the introduction of flexible repayment schedules that can be profitable for both borrowers and the lenders. Second, it studies how borrowers select into different repayment schedules, based on their sensitivity to price and behavioral characteristics (time preferences, risk aversion, financial discipline), as well as on the value they give to continuing their relationship with the lender. In order to study borrowers’ selection, our experimental design does not “exogenously” assign borrowers to a flexible contract (as, for instance, in Field et al., 2013), but allows them to choose between a rigid and a flexible schedule, which are provided simultaneously by the lender.
Trial End Date June 30, 2017 September 30, 2019
JEL Code(s) O12, O16, D03
Last Published June 08, 2017 01:01 PM June 24, 2019 09:21 AM
Intervention (Public) Once clients are ascertained to be eligible for the individual, standard “rigid” loan, customers in treated branches are offered the opportunity to choose between the product they are expecting to receive (i.e. the rigid contract) at 24% interest and a flexible contract offered at an interest rate of 26%. This flexible loan gives the customers the opportunity to benefit from a three-month repayment holiday, to be exerted any time after the third month of the loan maturity. The first three instalments of the flexible contract require a monthly repayment. During the three-month repayment holiday, borrowers still needed to repay a small fee. Borrowers in control branches were only offered the standard rigid contract at 24%. This contract has been designed with our partner Microfinance Institution for the purpose of the study.
Primary Outcomes (End Points) The outcomes of the study are expected to be the impact the flexible schedule of payment has on production growth, employment and income. We intend to study how preferences for repayment flexibility relate to customers’ characteristics and business performance. In line with Barboni (2017), our primary outcomes of interest are repayment rates, business outcomes including sales, profits, investments (business assets), consumption (expenditures) as well as savings and borrowing. We also intend to study how preferences for repayment flexibility relate to customers’ characteristics, including risk aversion, time preferences, and personality traits.
Primary Outcomes (Explanation) We will construct a few outcome variables relating to customers’ behavioral traits. The main ones are time consistency and risk aversion. In the baseline and endline surveys, we play two sets of “lab-in-the-field games” to elicit respondent’s preferences. Borrowers’ attitude towards risk is measured with a standard Multiple Price List (MPL). The MPL protocol consists of presenting the subjects with two different lotteries, Lottery A and Lottery B, entailing six decisions. Payouts are constant but the probabilities of success change from one decision to the other, with Lottery B being riskier than lottery A. Until round three, lottery A gives a higher expected value than lottery B. Starting from round four, Lottery B yields a higher expected value. Therefore, subjects who stay with Lottery A longer than three rounds display increasing levels of risk aversion. Conversely, subjects switching to Lottery B in the earlier rounds would display increasing levels of risk-loving behavior In addition, customers’ intertemporal preferences are assessed using standard list choices. This protocol consists of two lotteries. In the first one, the respondent has to choose between a Rs. 200 sum to be paid the day after the interview and an equal or larger sum (Rs. 200, 240, 260, 280, and 300) to be paid one month later. The second lottery “shifted” the time horizon of the first lottery by three months. Combining the two lotteries not only allows one to estimate subjects’ discount rate, but also to detect any time inconsistency. If a subject preferred Rs. 260 one month later to Rs. 200 paid tomorrow, she should have also preferred Rs. 260 paid four months in the future to Rs. 200 paid three months in the future. This behavior is defined as “time consistent”. Still, preference “reversals” may emerge. For example, when a subject prefers Rs. 260 one month later to Rs. 200 paid tomorrow, but the choice is reverted for the later rewards, the subject is said to display hyperbolic discounting. Conversely, when a subject prefers Rs. 260 one month later to Rs. 200 paid tomorrow, but this choice is reverted for the earlier rewards, the subject is showing anti-hyperbolic discounting.
Experimental Design (Public) We intend to set up an exploratory study in which a sample of small entrepreneurs, borrowing for the first time at our partner MFI as individual borrowers, are offered the possibility to choose between a rigid, cheaper contract and a flexible, more expensive one. Our experiment is meant to get a first understanding on how the encouragement design works, and to bring preliminary evidence on the provision of a grace period. In this sense, we intend to achieve a proof of concept that could be used to design and implement the large-scale Randomized Controlled Trial. The latter will consist in randomizing our target population into two different treatments: one in which subjects are exogenously assigned a grace period, the other in which subjects are offered the choice between having or not a grace period. Subject borrowing in a rigid repayment schedule will be kept as control. The main objectives of the current exploratory study are therefore to carefully tailor the microfinance contracts (rigid versus flexible one), to assess the validity of the experimental design (whether there is enough heterogeneity among borrowers' types and subsequent choices), and to help making an accurate power calculation for the broad RCT. We randomize 28 branches of the partner institution in the state of Uttar Pradesh, India between control and treatment. Randomization of treatment and control branches was done at two stages – first for the 4 city branches and then the 24 non-city branches. The least squared distance method of randomization was adopted to assign branches to treatment and control. Additionally, the sampling distribution at each of the treatment and control branches was fixed in proportion to the average monthly number of loans disbursed at each of the branches. The randomization method adopted pairs branches together based on the minimum distance between two branches and randomly assigns them to treatment and control. Data is collected on a total of 800 households – 400 of whom have received the treatment intervention. Baseline household survey is administered to the treatment and control clients once their loan is disbursed. Regular follow-ups are conducted via telephonic surveys to capture customers’ delinquencies/defaults. This is further cross checked with the administrative data from the partner institution. A midline survey is conducted half-way through the loan period (12 months post loan disbursal), and an endline survey will be conducted at the end of the 24 month loan period.
Randomization Method The least squared distance method of randomisation was adopted to assign branches to treatment and control. The randomisation method adopted pairs branches together based on the minimum distance between two branches and randomly assigns them to treatment and control. The least squared distance method of randomization was adopted to assign branches to treatment and control. The randomization method adopted pairs branches together based on the minimum distance between two branches and randomly assigns them to treatment and control.
Randomization Unit Data will be collected on a total of 800 households – 400 of whom have received the treatment intervention. Randomization of treatment and control branches was done at two stages – first for the 4 city branches and then the 24 non-city branches. The least squared distance method of randomisation was adopted to assign branches to treatment and control. Additionally, the sampling distribution at each of the treatment and control branches was fixed in proportion to the average monthly number of loans disbursed at each of the branches. The randomisation method adopted pairs branches together based on the minimum distance between two branches and randomly assigns them to treatment and control. Randomization of treatment and control branches was done at two stages – first for the 4 city branches and then the 24 non-city branches. The least squared distance method of randomization was adopted to assign branches to treatment and control. Additionally, the sampling distribution at each of the treatment and control branches was fixed in proportion to the average monthly number of loans disbursed at each of the branches.
Was the treatment clustered? No Yes
Planned Number of Clusters 800 households- 400 of whom receive the treatment 28
Planned Number of Observations 800 households- 400 of whom receive the treatment 800 suggested, final sample size is 799.
Sample size (or number of clusters) by treatment arms 800 households- 400 of whom receive the treatment 400 treatment clients across 14 treatment branches
Power calculation: Minimum Detectable Effect Size for Main Outcomes We believe the sample size of 790 clients in 28 branches, out of which 14 are in treatment, is high enough to attain a minimum detectable effect size of 0.41 or less for a significance level of 0.05 and a power of 0.90. Assumptions Case 1 Case 2 Case 3 Alpha Level (α) 0.05 0.05 0.05 Two-tailed or One-tailed Test? 2 2 2 Power (1-β) 0.9 0.9 0.9 Rho (ICC) 0.1 0.1 0.1 P 0.5 0.5 0.5 R12 0.4 0.3 0.25 R22 0.3 0.2 0.2 g* 4 4 4 n (Average Cluster Size) 29 29 29 J (Sample Siz [# of Clusters]) 28 28 28 MDES 0.38 0.41 0.41
Additional Keyword(s) Microfinance, Adverse Selection, Repayment Flexibility, Screening
Back to top

Other Primary Investigators

Field Before After
Affiliation Post-Doctoral Research Fellow, Julis-Rabinowitz Center for Finance and Public Policy, Woodrow Wilson School, Princeton University Assistant Professor, Warwick Business School, University of Warwick
Back to top