Improving targeting in Rwanda in the context of broader Social Protection WB Safety-Net Course, Washington 12 February 2010 Annika, Crispus, Fidele, Heather, Nina
Objectives of presentation Evaluating benefits and challenges of current safety-net targeting system Exploring potential advantages and disadvantages of alternative targeting methods (particularly PMT) Challenges of integrating poverty-based targeting into broader social protection strategy development particularly harmonisation with categorical targeting methods Proposed next steps
Background poverty and vulnerability Population 9,544,011 Poverty line USD$160 Poverty headcount 56.9% Extreme poverty headcount 36.9% Poverty gap index (depth) 40.0% Poverty and vulnerability profile extreme poor (include high % older people and disabled), genocide survivors, returnees Evolution of national safety-net program: Reaction to being off-track against MDGs and PER findings (2006 - many, small SP programs some vulnerable groups assisted more than others)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Consumption, USD VUP safety-net program Vision 2020 Umurenge Programme (VUP) - objective is to reduce extreme poverty. Flagship program in EDPRS. Emphasis savings, investment, graduation Three components: 1) Public Works (PW): employment on community infrastructure projects 2) Direct Support (DS): unconditional cash transfers to those unable to work 3) Financial Services (FS): access to savings and credit for enhanced livelihoods 450 400 350 300 250 200 150 100 50 0 Extreme poor covered by VUP Targeting Extreme poor not covered by VUP Percentile of the population Consumption distribution Poverty line Extreme poverty line Poor not covered by VUP Limited to ~25% of households (scale of poverty and fiscal affordability)
Targeting Issues 1. Effectiveness of current safety-net program (VUP) targeting: * Community-based poverty targeting (superior information, participatory, ownership, only way to identify some groups) * But concerns about horizontal inequity * Some evidence of error but early stage of program; system has changed/improved since 2. Dynamic context development of social protection strategy: Likely move towards including: * Universal or tapered old age grant * Disability grant
Factors to consider in comparing different targeting options Sources and degree of error Design - Inherent to methodology Community-Based Horizontal inequity More information on all dimensions of poverty Appropriateness for poverty type Chronic, transitory Characteristics of poor Tend to exclude lazy and households certain types of poor Implementation Enumeration error Elite capture PMT Regression LSMS data quality Equivalence scales Homogeneity of poverty Chronic Greater exclusion of small HHs Enumeration error HH manipulation of responses Perverse incentives Financial cost Lower Higher Frequency Annual Every 3 years? Capacity requirements (time & skills for collection, supervision, entry, analysis) Lower Higher Community acceptance Transparency More Less (black-box) Community ownership More Less Prejudice/corruption safeguard Less More
Cost and economic factors to consider in comparing different targeting options Scenario 1: CBT of cash transfers and public works Scenario 2: PMT of cash transfers and public works Scenario 3: CBT/PMT combined cash transfers and public works Assuming: Budget constraint fixed benefit level and caseload Community-Based (20% error) Proxy Means Test (20% error) Proxy Means Test (40% error) Proxy Means Test (10% error) Program budget 150,000,000 150,000,000 150,000,000 150,000,000 Poor 1,860,000 1,760,000 1,320,000 1,980,000 Non-poor 465,000 440,000 880,000 220,000 Program caseload 2,325,000 2,200,000 2,200,000 2,200,000 Administrative cost 7,500,000 7,500,000 7,500,000 7,500,000 Targeting cost 3,000,000 10,500,000 10,500,000 10,500,000 Benefit level 60 60 60 60 Average poverty gap of poor 64 64 64 64 Share of poverty level 94% 94% 94% 94% Cost effectiveness of program 74% 70% 53% 79% % poor reached 34% 32% 24% 36%
Consumption, USD Combining poverty and categorical targeting 450 400 350 300 Non-poor old and disabled 250 200 Poverty line Extreme poverty line 150 100 50 0 Poor old and disabled Scenario 4: Old age and disability grants replacing cash transfers for households with old and disabled members Scenario 5: Old age and disability grants topped-up by cash transfers to extreme poor households
Scenario calculations issues to consider and information needed Will old age grants and disability grants replace cash transfers? Analyse overlap in household data Level of household benefits Financial cost and sustainability; Prioritisation of programmes? If overlap is large, what does this imply for targeting systems? Will we need to top-up? National database will be critical
Implications and issues to consider Targeting option considerations cost, capacity, error, community acceptance, transparency, impact Combining household and individual-based and poverty and categorical targeting systems Affordability and sustainability benefit levels, coverage, programme priorities Institutional arrangements enhancing coordination and collaboration between line ministries and local government MIS and national database all individuals/households and types of assistance Trade-offs: Keeping in mind feasibility and simplicity!
Priority Actions Make decision on timing of the VUP targeting pilot (LSMS/EICV existing data 2005/6 or new data 2011) Presentation of issues and propose the targeting options (CBT/PMT/hybrid) to the working group and find how they can be harmonized under the broader SP strategy. Discuss the benefits of moving towards single national registry/database/mis develop action plan.
MURAKOZE CYANE