Abstract

Identifying eligible beneficiaries for social programs, a process known as “targeting,” can be a challenging and costly process for development and humanitarian organizations. Many widely-used targeting strategies were developed for rural environments and may not work as well in dynamic and densely populated urban centers. One potential new technique is “decentralized targeting,” a process that relies on information from socially knowledgeable members of a community. In Liberia, researchers conducted a randomized evaluation to measure the effectiveness of decentralized targeting in reaching poor households and households that have experienced an economic or health shock. Preliminary results find that both the proxy means test and decentralized targeting were prone to error – the majority of households identified by both were not the poorest.

Policy Issue

Targeting beneficiaries for social programs in urban areas is increasingly important as urban populations grow and poverty or emergency relief programs become more common in densely populated settings. However, current targeting strategies and tools may not be best suited for these dynamic urban environments. For example, the tools for targeting social programs often rely on methods developed in rural settings. These rural programs often leverage pre-existing social and political institutions to target beneficiaries. The effectiveness of these structures may break down in dynamic, urban environments. Another popular beneficiary targeting tool is the proxy means test which measures household wealth. While proxy-means tests are promoted as a quick option for assessing program eligibility, it requires regular updates to calibrate the means testing and does not extend outside of welfare-based eligibility.

This research aimed to provide evidence and recommendations for comprehensively identifying beneficiaries for social programs in urban areas, and specifically on how leveraging social connections compares to other targeting methods.

Context of the Evaluation

Liberia experienced prolonged conflict in the late 20^th and early 21^st century, in many ways it is both simultaneously growing and re-building. As the country became more stable and peaceful in the recent decades, the doors opened for many international donors, re-introducing the need for effective targeting of social protection programs on a large-scale. Additionally, when the Ebola crisis hit in 2014, health institutions and inter-personal support systems were pushed to their breaking points. With this background, the urban areas of Liberia are a prime – and highly relevant – location to test innovative ways to identify beneficiaries for a variety of social programs.

This study was conducted in Liberia’s capital city, Monrovia.

Details of the Intervention

Researchers evaluated how a decentralized targeting method, used for selecting beneficiaries of a cash transfer program, compared to more traditional methods in its effectiveness and cost-effectiveness in identifying needy households, and in how beneficiaries used and benefited from the transfer.

The research team first canvased 13 community blocks (neighborhoods) in three areas of Monrovia and gathered data on poverty ranking (through a proxy means test), recent economic or health shocks in the household, and who in the community may be best suited, based on their knowledge or connectedness, to identify needy households. The research team then invited 436 community members (drawn from 15 percent of the households in each community block) to one-on-one interviews, based on one of the following criteria:

• A community consensus said they were more knowledgeable about community welfare than other.
• The community indicated they were highly connected within the community’s social network.
• Some were selected randomly to offer a comparison to the above two groups.
• Leaders from the community were also invited to participate

During the interviews, participants provided input on how an unconditional cash transfer would be distributed among households living within their community block. Community members were asked about their perceptions of the poverty status and relative wealth of households in their community block.

Then, based on the information and data collected, three groups of households were selected to received one-time cash transfers of about US$80:

1. Eighty households receive a cash transfer if nominated by a leader within their community block.
2. One-hundred and twenty households received a cash transfer if nominated by a non-leader within their community block.
3. Eighty households were randomly selected from the poorest households, determined by a proxy means test (derived from the Poverty Probability Index® (PPI®)’s methodology) within the community block.

This design enabled the research team to assess if certain members of an urban community have better access to information about households that are most likely to benefit from a social program; if certain members are better positioned to share information about a social program; how cash transfer recipients share the money that they receive with other members of their social network; as well as if social network targeting mechanisms are useful tools for identifying participants for programs aimed at reducing intimate partner violence.

Results and Policy Lessons

Preliminary results:

Overall, both the proxy means test and decentralized targeting were prone to error - the majority of households identified by both were not the poorest. However, the results suggest that decentralized targeting increases constituent voice with only modest losses in targeting the poorest households.

Effectiveness of targeting methods:

• Proxy means test: The proxy means test was effective in identifying moderately poor households; however, the proxy means test was prone to significant errors of exclusion––65% of households classified as poor by this targeting strategy were not among the poorest quintile.
• Decentralized targeting: Asking members of the community to target a cash transfer also led to high error rates––76% of households nominated for a cash transfer were not among the poorest quintile.
• Nominated community members: Compared with community leaders, nominated community members were more likely to distribute cash transfers to households affected by a health shock and less likely to distribute cash transfers to households affected by a wealth shock.

Benefits of cash based on targeting method: The cash transfers increased business activity, but had no detectable impact on measures of household welfare, such as per capita expenditure, number of meals per day, subjective welfare or happiness.

These preliminary results suggest further optimization of the targeting tools studied here would be needed to reach the neediest households with cash transfers programs.

Analysis is still ongoing on the cost-effectiveness of targeting methods; results are forthcoming.