As part of UNICEF Innocenti's workshop on social protection in humanitarian settings, Pascale Schnitzer from World Bank presented her working paper "How to Target Households in Adaptive Social Protections Systems? Evidence from humanitarian and development approaches in Niger".
For more on this workshop and to access the seven papers released at the event, visit: https://www.unicef-irc.org/article/1829-evidence-on-social-protection-in-contexts-of-fragility-and-forced-displacement.html
How to target households in adaptive social protection systems?
1. How to Target Households in Adaptive Social Protection
Systems?
Evidence from Humanitarian and Development Approaches
in Niger
Pascale Schnitzer
Social Protection and Jobs, World Bank
June 7, 2018
2. Adaptive Social Protection (ASP) and targeting
2
- ASP aims to reduce poverty and mitigate the impact of shocks
1. Reduces poverty and builds resilience before shocks occur
2. Quickly scales up interventions in response to shocks
- Limited budgets = > methods to select beneficiaries required.
1. Persistently poor households
2. Households suffering from shocks
- Development and humanitarian actors often operate with little coordination when
identifying beneficiaries. A better understanding of targeting methods and its
integration is considered a core element in improving coordination and policy
effectiveness.
3. Study overview
3
Temporary food
insecurity
Persistent
poverty
Proxy Means Test (PMT)
* Niger National Safety Nets -> persistent
poverty
Household Economy
Analysis (HEA)
* Niger humanitarian NGOs
-> food insecure
Geographical
targeting
* Persistent poverty
* Temporary food insecurity
3 TARGETING METHODS 2 WELFARE OBJECTIVES
1. What is the relative efficiency based on persistent poverty and temporary food
insecurity?
2. How these methods can be used as part of an ASP system?
4. Punchline
4
Traditional methods for targeting social protection
programs can be tailored and combined for program
allocation in humanitarian settings
5. Outline
5
• Literature - Targeting in ASP systems
• Targeting approaches
• PMT
• HEA
• Geographical and combined approaches
• Methodology
• Results
• Comparing PMT and HEA
• Comparing PMT with other ways to identify the poor
• Comparing HEA with other ways to identify the food insecure
• Implications
6. Literature - Targeting in ASP systems
6
- Significant literature but important gaps remain that this paper helps filling in:
- Number of studies on PMT and geographical poverty targeting but limited studies about HEA and
geographical food insecurity targeting approaches
- Limited understanding of how development and humanitarian targeting approaches compare
and potential synergies
- Significant literature on how to identify the persistently poor but less is known about methods to
identify households suffering from shocks
- Comparative studies are limited
7. Method 1 - Proxy means testing (PMT)
1. Econometric methods are used to develop a formula aiming to estimate
poverty based on characteristics that are relatively easy to observe and verify
2. Door-to-door household survey in project areas to measure the variables used
in the formula
3. Use data collected + formula to estimate a score for each household
4. Select households with lowest score
8. Method 2 – Household Economy Analysis (HEA)
8
▪ Livelihood-based framework developed to better anticipate and respond to shocks
affecting food access.
▪ Framework key feature: “HEA baselines” -> defines and
describe livelihood zones and wealth groups within each.
Usually conducted through qualitative data collection.
▪ Used in different ways, for multiple purposes,
including household level targeting.
▪ Formula: largely similar process as the PMT, except formula
is “qualitatively” developed based on HEA baseline and the
analysis of program targeting data from 2009-2012
9. Geographical and combined targeting
approaches
Geographical targeting
a. “Geographical poverty”:
• Selects areas with lowest welfare based on persistent poverty
b. “Geographical food insecurity”:
• Selects areas with lowest welfare based on temporary food insecurity
Within selected areas, every household is assumed to benefit.
Note: PMT and HEA users in Niger rely on different geographical methods
Combined targeting approaches
a. “Poverty approach”:
• Geographical poverty + PMT.
b. “Food insecurity approach”:
• Geographical food insecurity + HEA
10. Methodology
10
▪ Agro-pastoral zones LSMS-ISA 2011 household survey. Data collected twice a year:
lean season (July, round 1) and harvest season (December, round 2).
▪ Simulate the selection of households under the different methods keeping
budget/coverage constant
▪ Methods compared to random assignment
▪ Welfare benchmarks
▪ Persistent poverty: proxied through consumption per capita (average between
two rounds)
▪ Temporary food insecurity: proxied through the food consumption score (FCS)
during the lean season. Reflects quantity and quality of food.
11. Comparing PMT and HEA efficiency
Note: Inclusion error (IE)=share of beneficiaries not within
bottom 30% of respective welfare distribution.
Inclusion errors1. Limited overlap between households selected using
the two methods
2. HEA households present lower poverty but appear to
be “less resilient”
3. Survey instruments largely similar but not the same ->
harmonization of questionnaires with little cost
4. A combination of PMT and HEA may be effectively
used to identify ASP beneficiaries
5. HEA formula may be improved through econometric
techniques
70%
43%
72%70%
52%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Random PMT HEA
Persistent poverty Temporary food insecurity
12. 12
Comparing PMT with other ways to identify the
poor
1. The PMT and combined method
perform best.
2. But a simple geographical approach
still results in important efficiency
gains. Note: geographical targeting
may present additional benefits by
lowering logistic costs
70%
43%
54%
40%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Random PMT Geographical
Poverty
Combined Poverty
Performance - Persistent poverty
13. 13
Comparing HEA with other ways to identify the
food insecure
• A simple geographical targeting is
especially effective to identify food
insecure households
• Relative to poverty, food insecurity
presents relatively high variation
between, and low variation within
geographical units
70%
52%
40%
43%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Random HEA Geographical Food
Insecurity
Combined Food
Insecurity
Performance - Temporary food insecurity
14. Implications for policy
14
• There is no single best performing targeting method in absolute terms. The choice of
method should consider different factors
• Program-specific welfare objective
• Distribution of welfare (i.e. between and within the targeted areas)
• Costs
• Coverage rates…
• Rather than focusing on identifying one optimal targeting method, actors trying to
develop ASP systems may emphasize the consolidation of information across programs
through the harmonization of household questionnaires and the building of a unified
database.
• This could support better coordination and effective application of alternative targeting methods, and
reduce duplication of efforts, wastage of resources and lead to faster response.
15. Implications for research
15
• Simple geographical targeting approaches can perform well, especially during food crises.
• Learning more about cost-effective Early Warning Systems may be particularly helpful since they can
result in large efficiency gains
• How well do simulations match reality?
• Program impacts and cost effectiveness?
16. How to Target Households in Adaptive Social Protection
Systems?
Evidence from Humanitarian and Development Approaches
in Niger
Pascale Schnitzer
Social Protection and Jobs, World Bank
June 7, 2018