Subscribe to E-Bulletin Donate to CEGA

The Impact of Community Monitors on Savings

Development Challenge

Increasing a household’s capacity to save money can alleviate strain in economic downturns.  Though poor households wish to save, a number of barriers including lack of commitment to continue with a savings plan and lack of attention to savings reminders can prevent sustainable household saving.[1]  

At the same time, a growing body of evidence that suggests the level of “centrality” or connectedness of a person affects their ability to influence others in the community[2][4]. It has yet to be empirically examined, however, whether the centrality network members can be leveraged in policy-design. In this project, researchers examine whether more central network members can be utilized to encourage others to save, simply by revealing an individual’s progress towards her goal to the central member.

Context

In India, few citizens participate in financial institutions; only 35 percent have an account in a formal financial institution, and only 12 percent used the account to save in the past year.[3]  In Karnataka specifically, researchers collected a baseline survey that confirmed these same tendencies:  only 25 percent of survey participants used a savings vehicle, and only 10 percent of participant savings funds were kept in a formal bank account.  Most participants kept savings inside the house, or other informal financial institutions. 

Evaluation Strategy

This study evaluated how the selection of a savings monitor affected individuals’ savings behavior.  Researchers sampled a total of 3,000 individuals from 60 villages in Karnataka, India.  All villages had banks that offered interest-bearing “no frills” savings accounts with no minimum balances or transaction fees. Villages were chosen based on previously collected demographic and social network data.[4],[5] All potential participants were administered a baseline survey, and were then assigned to either a control or one of three savings mechanism treatments, as follows:  

Name

Description

Treatment 1 (T1)

Business Correspondent savings model (BC bundle)

Treatment 2 (T2)

BC Bundle + randomly assigned monitor

Treatment 3 (T3)

BC Bundle + endogenous monitor (chosen by participant)

Control

No account or treatment provided

BC Bundle

Treatment participants first picked an individual savings goal and outlined a six-month savings plan before being assigned to a savings mechanism.  Each savings participant was given a savings account and surveyors visited savers every two weeks to record their banks balances. This mechanically served as a reminder to save. 

Monitors

In the monitor treatment groups, the surveyor that visited the saver’s home relayed the account information to the participant’s monitor.  At the end of the study, monitors were rewarded 100 rupees (Rs, ~2 USD) if the savers they monitored reached at least half of the savings goal, and an additional 200 Rs (~4 USD) if their saver reached the full savings goal. The savers in T2 received randomly assigned monitors, whereas the savers in T3 chose their own monitors. The researchers made use of their pre-existing work – where they had detailed network data matched with a household level census in each village – to determine the closeness of the monitor to the saver (length of a path in the network from the monitor to the saver) as well as the centrality of the monitor and saver in the community network.[6]

At the end of the six-month savings period, researchers collected closing surveys and bank information from participants. 

Results and Policy Implications

Overall, rates of goal attainment were low among participants.  Notwithstanding, savings amounts increased across treatment groups, and monitors had significantly more success in producing savings than did the BC bundle alone.  In the standard BC bundle treatment, those who participated saved on average 25 percent more than the control group.  Despite improved savings effects, however, only seven percent of the BC bundle group met their savings goal.

The probability of reaching savings goals with a monitor increased six percentage points relative to the BC bundle treatment.  Savings balances in the randomly assigned monitor group increased by 35 percent compared to the BC bundle group.  Monitors chosen at random had a greater effect on savings than did monitors chosen by the saver.   Qualitative data from interviews suggested this may be because random assignment gave a sense of formality to the monitoring.  Those who chose their monitor also reported feeling socially limited as to whom they could choose, leading them to choose more friends and likely limiting the saver’s feeling of obligation.  Socially connected monitors created greater savings improvements.  Monitors of social distance one (direct connection to saver) increased goal attainment by 43 percent, relative to monitors of social distance two (indirect connection to saver). Similarly, monitors in the upper third of community-connected individuals generated a 20 percent increase in savings goal attainment. 

Results from the study suggest household savings accumulation can be significantly improved if monitors work with savers. Moreover, the efficacy of a monitor-saver match can be further increased if network relationships are taken into account. 

Timeline

2013


[1] Dupas, Pascaline, and Jonathan Robinson. (2013). "Why Don't the Poor Save More? Evidence from Health Savings Experiments." American Economic Review, 103(4): 1138-71.

[2] Kinnan, C., & Townsend, R. (2012). Kinship and financial networks, formal financial access, and risk reduction. The American Economic Review, 102(3), 289-293.

[3] World Bank. (2011).  Financial Inclusion Data.  http://datatopics.worldbank.org/financialinclusion/country/india

[4] Banerjee, A., A. G. Chandrasekhar, E. Duflo, and M. Jackson (2013): “The Diffusion of Microfinance,” Science, 341 (6144).

[5] Jackson, M. T. Barraquer, and X. Tan (2010). “Social Capital and Social Quilts: Network Patterns of Favor Exchange”.

[6] Researchers use “eigenvector centrality” to measure influence of a point or person in a group