Behavioral research suggests that self-control, procrastination, attention, and other behavioral factors may limit individuals’ ability to save for the long-term. New mobile money platforms in many developing countries are creating financial products that can help low- and moderate-income individuals overcome these barriers. Researchers partnered with a mobile money provider to test whether offering its employees the opportunity to automatically deposit a portion of their paycheck into a savings account increases long-term savings. Results showed that defaulting employees into a savings program and employer savings matches both increased savings. For employees identified as more likely to procrastinate, phone-based financial consultations also increased savings.

Policy Issue 
Savings enable people to accumulate small sums over time to make large purchases or investments, or cope with emergencies. In countries with no health insurance or social security, savings are all the more critical for the well-being of the poor. However, people may face several barriers to saving. Behavioral research suggests that lack of self-control, procrastination, and inattention can all prevent people from saving. These barriers, exacerbated by lack of access to appropriate financial services and information, may lead individuals to save less than they would like. Meanwhile, the rapid proliferation of mobile money is paving the way for financial services that are designed to meet the financial needs of low- and moderate-income individuals in developing countries. Increasingly, financial institutions and employers have the opportunity to use these platforms to develop products to help individuals save more. 
Research from high-income countries shows that default enrollment into automatic savings plans, which transfer funds from an employee’s paycheck into a long-term savings account, can be very effective at increasing deposits. This is the first study to test the impact of automatic transfers for retirement savings in a low-income country. 
Context of the Evaluation 
Afghanistan has one of the lowest bank account penetration rates in the world. An estimated 90 percent of the adult population does not have an account at a formal financial institution. The savings rate is also low: it is estimated that only one in 25 adults have saved any money in the past year. Mobile phone access, on the other hand, is quite high, with 75 mobile phone subscriptions for every 100 adults. 
Roshan, Afghanistan’s largest mobile communications provider, launched M-Paisa, a mobile payments system, in 2008. At the time of the study, M-Paisa had approximately 1.2 million registered users and was the largest mobile money network in Afghanistan. This evaluation targeted employees of Roshan located across seven field offices, in both rural and urban locations around the country. Employees who participated held a broad range of positions, including janitor, security guard, engineer and manager. 
Details of the Intervention 
Researchers partnered with Roshan to test multiple interventions designed to increase use of a mobile savings account available to all Roshan employees. This account, called M-Pasandaz, is linked to each employee’s existing M-Paisa mobile money account so that employees may deposit and withdraw funds to the M-Pasandaz account using the nationwide network of M-Paisa agents. Employees were randomly assigned to groups to test the impact of two main interventions:
Default contribution:  Employees were first randomly assigned to one of two groups. In the first group, five percent of their salaries were automatically deposited into the savings accounts. They could change their automatic contribution levels or opt-out of the automatic contribution plans at any point. The second group received the status quo, access to the M-Pasandaz account, but no portion of their salary was automatically deposited.
Employer savings-match incentive: Each of the two groups mentioned above were further divided into 3 sub-groups:
  1. 50 percent match: For employees in this group who made regular contributions to their M-Pasandaz account for at least 6 months, without making any withdrawals, Roshan matched half of what they saved, up to 10 percent of their salary.
  2. 25 percent match: For employees in this group who made regular contributions to their M-Pasandaz account for at least 6 months, without making any withdrawals, Roshan matched one quarter of what they saved, up to 10 percent of their salary.
  3. Comparison: Roshan did not match any portion of the savings for those in the third sub-group.

Two months after the initial intervention was underway, when voluntary contribution adjustments had slowed, the researchers implemented three additional interventions aimed at helping people overcome behavioral biases against saving:

  • Personal Consultation: The firm called a randomly selected half of employees in the study to offer a personal consultation providing more information about their account. In the meetings, company representatives estimated the payouts under different contribution rates, and offered employees a chance to change their contribution rate. This intervention aimed to alleviate nonfinancial costs (such as time and effort) to participation.
  • SMS reminders: Researchers randomly selected half of the employees in the study to receive a series of text messages, sent directly to employees by Roshan’s human resources office each month over several months. The text messages reminded employees of their current contribution rate and the phone number to call if they wished to change it. The text messages were designed to test if employees were not taking advantage of the savings accounts simply because they were unaware or had forgotten.
  • Monthly phone surveys: Half of the sample was randomly selected to receive brief monthly surveys on financial behaviors and understanding of the M-Pasandaz account. These surveys potentially also addressed inattention by reminding employees of the program. 

Baseline and endline surveys measured characteristics such as time inconsistency, the extent to which, in a hypothetical experience, participants would be willing to accept a smaller sum of money sooner or wait to receive a larger sum of money at a later date.

Results and Policy Lessons 
In the absence of either an automatic contribution or an employer match, savings levels were low, about one percent. However both of the primary interventions were effective at increasing contribution rates and savings. 
Effects of default contributions: Across match levels, two months after the interventions began employees who were assigned to have five percent of their salary automatically deposited into the M-Pasandez account were approximately 40 percentage points more likely to contribute than those were not. This effect of setting this default contribution was roughly equivalent to the employer offering to match half of the employee’s contribution. At six months, after the latter set of interventions, this difference remained 33 percentage points.  Over the six month study period, those in the default contribution group saved on average an additional 2,426 Afghanis (approximately US $40).
Effects of savings matching: When the employer matched one quarter of the employee’s savings, the number of employees saving through the program increased by about 25 percentage points. When the employer matched one half of the employee’s savings, participation jumped about 47 percentage points. 
Human Resources consultations: Approximately 11 percent of employees offered the consultations changed their contributions immediately after. Specifically, the consultations seemed to affect the employees who were identified as being more time inconsistent and who’d agreed to a later, rather than immediate, appointment time. This intervention seemed to work best for the population expected to procrastinate.
SMS reminders: On average, approximately three percent of employees who received text message reminders raised their contribution amount immediately after the messages were sent.
Monthly surveying: Being surveyed more frequently did not affect employee contributions.
Overall, the findings suggest that in Afghanistan, people face the same behavioral barriers to savings found in high-income countries, and that programs found to be effective in boosting savings can work there as well.