Experimental Design
We conducted three experiments: Chennai, India in 2007, Cagayan de Oro, Philippines in 2007, and Cagayan de Oro, Philippines in 2010 (from different markets than in 2007). Each experiment took place in urban market settings. Going stall to stall, we identified individuals eligible for the study based on the following criteria: (1) they were the decision-maker of the business, (2) they borrowed consistently from a professional moneylender (defined as interest rate at or above 5% per month) for the past five years, and (3) they had an outstanding balance of US$100 or below in the Philippines or US$50 or below in India with moneylenders.
Both 2007 experiments included the same four equal-sized treatment arms: debt payoff; financial education; debt payoff and financial education; and, control. We implemented a brief financial training in small groups of roughly 25 vendors by a professional survey team. The training emphasized two messages: (1) that lending at high rates from moneylenders led to a large loss of money and thus consumption every year compared with other alternatives, and (2) that savings and consumption discipline could help vendors avoid having to take out loans at all. The training included interactive activities, discussion, and lecture.
In the 2010 Philippines experiment, participants were randomized into one of four groups: debt payoff; savings account; debt payoff and savings account; and, control. All three treatment groups also received a 5-10 minutes financial education lesson (i.e., slightly briefer than the 2007 financial education, but no longer separately tested as its own treatment arm). Because of problems with insufficient compliance with account opening requirements (specifically, providing necessary documentation and identification), only a few savings accounts were opened, and thus there is nothing to analyze with respect to the savings account treatment arms. For analysis, we collapse this wave to two groups: debt payoff and financial training (with and without the offered savings account) and control.
In all three sites, the financial training was conducted prior to the announcement of the debt payoff. Several days after the financial training, surveyors went to those selected for debt payoff and informed them they had won a prize through the lottery. In the Philippines, as debt collectors generally collect repayments daily in this area, surveyors waited until the collectors arrived and paid off the debt in view of the vendors. In India, individuals were instructed to come to a central location to pickup money if they were in a debt payoff treatment group. The amount paid was equal to the amount the debt collector said was owed, up to 5,000 pesos (~US$100) in the Philippines or 3000 rupees (~$50) in India.
The 2007 Philippines and India experiments did not employ stratification. The 2010 Philippines experiment stratified on quintile of preexisting debt level. Baseline surveys measured (a) the history of their savings and debt over the past few years, (b) basic cognitive skills and educational level, (c) mental health that could be linked to ability to fulfill plans, (d) business information, e.g. assets and revenues, and (e) demographic data.
We conducted four follow-up surveys, starting 4-6 weeks after the debt payoff, and ending at the latest 2 years after the debt payoff. Follow-up surveys measured (a) whether the vendors had moneylender and other types of debt, (b) what changes have occurred in the business (i.e., if the business has expanded significantly or not) (c) any shocks the household has experienced, and how they coped, (d) some components of consumption, (e) savings.