Field
Experimental Design (Public)
|
Before
Our partner organization in Nigeria has agreed to randomize loan applicants into a number of different treatment groups (irrespective of their credit score), which creates individual variation in the likelihood of loan approval, and also in the loan amount. Our control group comprises a random sample of 'business-as-usual' customers, who might not be able to avail of a loan in the event that their credit score is too 'low'. Between 4-20 weeks after the initial loan application, a phone survey is conducted with around 4,000 individuals (in treatment and control) that measures various aspects of welfare outlined earlier. We exploit the random variation in the likelihood of loan approval, and the random variation in loan amounts disbursed to estimate average and heterogeneous treatment effects (both intent to treat and treatment on treated) on the outcomes of interest.
|
After
Our partner organization in Nigeria has randomized loan applicants into a number of different treatment groups (irrespective of their credit score), which creates individual variation in the likelihood of loan approval, and also in the loan amount. Our control group comprises a random sample of 'business-as-usual' customers, who might not be able to avail of a loan in the event that their credit score is too 'low'. Between 4-20 weeks after the initial loan application, a phone survey is conducted with around 4,000 individuals (in treatment and control) that measures various aspects of welfare outlined earlier. We exploit the random variation in the likelihood of loan approval, and the random variation in loan amounts disbursed to estimate average and heterogeneous treatment effects (both intent to treat and treatment on treated) on the outcomes of interest.
|