Targeting the Poor: Evidence from a Field Experiment in Indonesia Alatas, Banerjee, Hanna, Olken and Tobias American Economic Review 2012, 102(4):1206- 1240
Targeting the Poor: Evidence
from a Field Experiment in
Indonesia
Alatas, Banerjee, Hanna, Olken and Tobias
American Economic Review 2012, 102(4):1206-
1240
Agenda
Background
The Direct Cash Assistance (BLT)
program was launched in 2005 to
support poor households during
economic crisis.
How to find an appropriate approach
to target the poor.
Reports an experiment on three
approaches: proxy means tests
(PMT), community targeting, and a
hybrid.
Proxy Means Test (PMT)
Assets (as proxy) are used to predict consumption or income
Frequently being used to the selection of the beneficiaries in a targeted social safety net programs.
Only those with incomes below a certain threshold are eligible.
The presumption is that household assets are harder to conceal from government surveyors than income.
Community-Based
The government allows the community
or some part of it to select the
beneficiaries.
The presumption is that wealth is
harder to hide from one’s neighbors
than from the government.
Hybrid
Combines PMT with community-based
method.
Aims to take advantage of the relative
benefits of both methods.
I. Empirical Design
Setting Indonesia
Cash transfers US$10 (BLT Programs)
Combination of PMT with community-based
methods
Goal: the poorest one-third of households.
45% of funds were incorrectly provided to non
poor households; 47% of the poor were
excluded from the program in 2005-2006
(World Bank:2006)
I. Empirical Design
Sample: 3 provinces640 villages30% urban70% rural
Experimental design Treatment for PMT Treatment for Community Targeting Treatment for Hybrid
North Sumatra
Central Java
South Sulawesi
II. Describing the Data
Data Collection
Four main sources of data:Baseline data
•demographics
•Family networks
•Participation in community activities
•Relationships with local leaders
•Access to existing social transfer programs
•Households’ pecapita consumption
Data on treatment results
•Targeting rank list
Data on community
meeting
•Attendance list
•Questionnaire on perceptions of community’s interest and satisfaction level
Data on community satisfaction
•Suggestion boxes
•Subvillage head interviews
•Facilitator feedback
•Households interview
II. Describing the DataSummary Statistics
32%: incorrectly
targeted based
on consumption
20% of non poor
households
received cash
transfer
53% of the poor
excluded
III. Comparing the Methods
Results on Targeting
Performance and Satisfactiono Targeting Performance Based on Per Capita
ConsumptionPMT method outperforms both the community and hybrid treatment in terms of the consumption based error rate.
o Effects of Targeting Policy on Poverty Rate and Gap- The differences in targeting accuracy across the
three methods do not result in large differences in the measures of poverty under consideration.
- Community treatments do better at reducing the poverty headcount (from 15.64 to 13.68)
- The poverty gap is not significantly different among he three methods.
o Satisfaction
Data collection: end line household
survey, follow up survey of sub village
heads, and comment box, community’s
complaints, facilitator comments.
Results:
Individuals and sub village heads are
much more satisfied with the community
treatment.
Fewer complaints in the community
treatments.
IV. Elite Capture
Examining whether elite connected
households are more likely to be
beneficiaries when the elite have more
control of the process
Result: the elite capture is not the
reason that the targeting is worse
under the community method.
V. Community Effort
• Worse targeting in the community methods could result simply from fatigue as a the ranking progresses.
• To investigate, they randomized the order in which households were ranked and did regression.
• Results: the community treatment does slightly better than the PMT in the beginning but substantially worse towards the end.
VI. Testing the Maximand
The community is doing its best to identify the poor, but has a different concept of poverty.
Investigated by examining how the targeting outcomes compare not just against the government’s metric of welfare, but also against alternative welfare metrics (community, sub village head, and self-assessment survey ranks)
Results:Per capita consumption does not fully capture what
the community calls welfare.
The community and the elite broadly share the assessments of welfare
The community targeting methods are more likely to conform with individual’s self-identified welfare status.
Does the community lack
information to evaluate
consumption?
Community has residual information
about consumption beyond that
contained in the PMT score or even in
the PMT variables. Community has
most of the information in the PMT as
well, but choose to aggregate it in
different way.
A different view of individual welfare
People believe that there are household economies of scale, so that conditional on per capita consumption.
The community may know more about other households’ ability to smooth shocks.
No evidence that ethnic minorities are more likely to be ranked as poor.
Those who contribute money are likely to be ranked as richer.
Communities may try to provide the “right” incentives to households.
Village does not penalize those who spend a lot of money on smoking and drinking
Conclusion
The community seems to have widely shared objective function other than per capita consumption;
The objective function does not differ based on elite capture;
There are better understanding of factors that affect a household’s earning potential or vulnerability.
If targeting the poor based on consumption is the only objective, the PMT performs better than community methods.
Community does better at identifying the poor.
Thank You for Listening