Community-based targeting to combat Covid-19-induced poverty
The Covid-19 pandemic presents an unprecedented risk of global recession and threatens to push millions of people back into poverty as governments have enacted lockdowns to restrict movement and curb the spread of the virus. In order to provide emergency economic assistance to vulnerable citizens, beginning in March 2020, the Government of Indonesia (GoI) decided to distribute cash transfers to protect individuals not covered under existing social safety net programs from falling back into poverty; however, open questions remained regarding how to most quickly and effectively target program recipients.
A previous randomized evaluation from Indonesia found that using community members to identify the poorest households to receive cash transfers (a process known as community-based targeting) greatly improved local satisfaction and led to smoother disbursal processes. Informed by the results of this study, the Indonesian Ministry of the Villages adopted community-based targeting for the rollout of its Covid-19 relief program. As of December 2020, the cash transfers had been distributed to nearly 75,000 villages and over eight million recipients, including 2.5 million female breadwinners who had not previously received any social safety net programs.
Social protection programs can provide critical support to those suffering the economic impacts of Covid-19, but quickly and accurately identifying those who need assistance can be difficult.
The Covid-19 pandemic presented an unprecedented risk of global recession, as countries have instituted measures to curb the spread of the virus, restricting many people’s ability to work and thereby reducing income. Governments around the world have responded to the crisis with large economic stimulus packages to support both individuals and businesses. However, questions have remained around how best to quickly and effectively identify and deliver support to those who need it.
Different identification methods have unique challenges—for instance, having people apply for assistance can be time-consuming to process while preexisting data may be outdated or incomplete. Leveraging local knowledge within communities to determine who receives the cash transfer may speed up the disbursal of aid without compromising accuracy.
In Indonesia, the government has a long history of disbursing assistance through large-scale social protection programs, including conditional cash transfers through the Family Hope Program (PKH) and electronic food vouchers through the Non-cash Food Assistance (BPNT) program. In March 2020, in response to the economic crisis triggered by Covid-19, the GoI increased the number of beneficiaries reached and the amount of assistance distributed through these programs. Nonetheless, a large number of people who had been economically impacted by the pandemic remained unenrolled in these existing social protection programs.
To address this gap, the Ministry of Village, Development of Disadvantaged Regions, and Transmigration (MoVDT) decided to reallocate 20-35 percent of the existing Village Fund1 to provide cash transfers worth 600,000 rupiah (about US$42.70) per month for three months to cushion the economic shocks of Covid-19.2 Later, the Minister of Finance (MoF) opened up additional resources so that village governments could reach more beneficiaries.
The cash transfers were intended to complement existing social protection programs by targeting those who were ineligible for Indonesia’s PKH or BPNT programs and thus not yet receiving assistance.3 The GoI aimed to provide relief to those most affected by the pandemic as quickly as possible, but was faced with the challenge of balancing the speed of disbursements with the accuracy of targeting.
A randomized evaluation completed nearly a decade before the onset of the pandemic found that identifying cash transfer recipients through community-based targeting greatly improved local satisfaction, better matched community members’ own concepts of poverty, and led to smoother disbursal processes.
From 2008-2009, J-PAL affiliates Abhijit Banerjee (MIT), Rema Hanna (Harvard), and Benjamin A. Olken (MIT), together with Vivi Alatas (CEO, Asakreativita) and Julia Tobias (CDC Group), worked with the Indonesian Central Bureau of Statistics, the NGO Mitra Samya, and the World Bank to test the impact of three different methods to improve the targeting of a cash assistance program.
To target one-time cash transfer recipients, sub-villages were randomly assigned to use either a proxy means test (PMT), a community-based approach, or a hybrid targeting approach. For the PMT approach, surveyors collected simple information on households’ and individuals’ assets and consumption to predict each household’s income; for the community-based approach, residents ranked households from richest to poorest at a community meeting led by trained facilitators; the hybrid approach combined the community ranking meeting with PMT verification. The poorest households determined by each method received the cash transfer.
When defining poverty as those who reported living on US$2 a day or less, the PMT was the most accurate targeting method, correctly classifying about 70 percent of households, while the community-based and hybrid methods correctly classified about 67 percent of households. Yet, while slightly less accurate overall, the community-based method correctly targeted a greater percentage of households who had self-identified as poor, thereby producing beneficiary lists that were more in line with the community’s own concept of poverty.
Residents assigned to the community-based method also reported higher satisfaction with the targeting process, cash transfer program, and beneficiary lists than residents exposed to the other methods. Higher satisfaction also led to a smoother disbursal process in villages assigned to the community-based method. For example, the facilitators who delivered cash payments were less likely to report difficulties during distribution in the community-based and hybrid methods, and facilitators were more likely to distribute cash payments in open meetings rather than through more time- and labor-intensive door-to-door visits in the community-based method.
The researchers suggested that the large benefits of community-based targeting in terms of community satisfaction outweighed its small costs in terms of accuracy, especially given projections that the PMT and community-based targeting would ultimately have similar impacts on Indonesia’s poverty rate.
For more information about this research, see the evaluation summary.
From Research to Action
Evidence from the earlier community targeting study informed the targeting and rollout strategy of Indonesia’s Covid-19 relief cash transfer program, which has reached more than eight million recipients previously unenrolled in social assistance programs.
After allocating funds to provide cash transfers to protect against the economic shocks caused by Covid-19, the GoI still needed to decide how to target and disburse the aid. Initially, the Ministry of Village (MoVDT) planned to select Village Fund cash transfer beneficiaries based on the Unified Database (DTKS) used to target other social assistance programs in Indonesia. The DTKS is intended to include information about the lowest 40 percent of the population in terms of welfare and socio-economic status; however, the database has known flaws and the unprecedented shock from the pandemic was likely to impact households who were not previously tracked in this system.
Realizing this, the team working on targeting at the National Team for the Acceleration of Poverty Reduction4 decided to implement the lessons from previous community-based targeting experiences, including the study above, into the government’s Covid-response plan. Their team drafted a policy memo for the Director General of Village and Community Empowerment and Development of MoVDT to promote the use of community-based targeting, drawing heavily on the findings of the community targeting study. Leveraging existing evidence can be a powerful tool to inform policy, and the memo presented a concrete and context-relevant solution to the targeting challenge the GoI faced during the pandemic.5
Informed by the presentation of this memo, in mid-April 2020, the GoI decided to begin involving communities in disbursing village cash transfers. Specifically, village volunteers were specially appointed at the neighborhood level to assist in targeting the distribution of the cash transfers. Furthermore, the beneficiary lists were finalized through village discussions, similar to the community method applied in the community targeting study.
According to MoVDT, as of December 2020, the cash transfers had been distributed to nearly 75,000 villages and more than eight million recipients. Among the recipients, 88 percent were farmers and 2.5 million were female breadwinners who had not previously received any social safety net programs. In 2020, the GoI disbursed approximately 23.74 trillion rupiah (US$ 1.64 billion) worth of cash transfers, distributing an average of US$200 to each household.
The GoI continued the program in 2021, distributing approximately 20.24 trillion rupiah (US$1.4 billion) worth of cash transfers to more than five million families across the 75,000 villages. Furthermore, the GoI announced that the cash transfer will continue in 2022 and has budgeted to distribute 27.2 trillion rupiah (US$1.89 billion) worth of cash transfers in 2022. Additionally, from December 2021 through February 2022, the GoI committed to providing an additional assistance of 300 thousand rupiah (approximately US$21) to nearly 700,000 of the poorest families already receiving the cash transfer.
Alatas, Vivi, Abhijit Banerjee, Rema Hanna, Benjamin A. Olken, and Julia Tobias. 2012. "Targeting the Poor: Evidence from a Field Experiment in Indonesia." American Economic Review 102(4): 1206-1240. http://dx.doi.org/10.1257/aer.102.4.1206
Abdul Latif Jameel Poverty Action Lab (J-PAL). 2021. "Community-based targeting to combat Covid-19-induced poverty.” Last modified February 2022.