1. Heterogeneous Impacts of an Unconditional Cash
Transfer Programme on Schooling: Evidence from the
Ghana LEAP Programme
Richard de Groota, Sudhanshu Handaa, Michael Parkb, Robert Osei Darkoc, Isaac Osei-
Akotoc, Garima Bhallab, Luigi Peter Ragnod
a UNICEF Office of Research – Innocenti
b University of North Carolina at Chapel Hill, Department of Public Policy
c Institute of Statistical, Social and Economic Research, University of Ghana – Legon
d UNICEF Ghana Country Office
Adolescence, Youth and Gender: Building Knowledge for Change Conference
Oxford, September 8, 2016
2. Introduction
Findings from Ghana Livelihood Against Poverty
(LEAP) programme
What is the effect of an unconditional cash transfer
programme on schooling outcomes for children?
Do parents display compensating or reinforcing
behaviour towards lower ability children (‘marginal’
child)?
Is there a need for conditionality in this case? (i.e.
strict rules attached to receipt of transfer)
3. Ghana LEAP programme
Ghana’s flagship social protection programme
Initiated in 2008, currently reaching >150,000
households (and counting!)
Cash payments and health insurance
Three demographic groups:
Orphaned and vulnerable children (OVC’s), elderly and
people with a disability (PWD)
Soft conditions for OVC households
Transfer level: 7% of household consumption
4. Methods
Core: Difference-in-Difference (DD) with
Propensity Score Matching (PSM)
Treatment group (N=699) pulled from 2010
extension in 3 regions (Brong Ahafo, Central &
Volta)
Comparison group (N=914) pulled from national
survey
Identical survey instruments, teams and field work
methods Ideal conditions for PSM!
5. Methods
Inverse Probability Weighting (IPW)
IPW uses inverse of the propensity score:
𝑝𝑠
1−𝑝𝑠
Higher score, more similar to LEAP household
Covariate adjustment
Cluster fixed effects
Sample size (children 5 – 17 years):
Comparison Treatment
Baseline (2010) 1,239 979
Follow-up (2012) 1,076 869
6. Methods
Schooling indicators:
Current enrolment
Attendance (missed any school in last 7 days)
Analyze effect by subgroups:
Age group (5 – 12 and 13 – 17)
Sex
Cognitive ability (Raven’s test)
7. Ghana LEAP - Impacts on school enrollment
Impacts on enrollment large for older boys
-15
-10
-5
0
5
10
15
20
25
30
35
All children
5 - 17 years
All Boys Girls All Boys Girls
Children 5 - 12 years Children 13 - 17 years
-0.7
-4.9**
1.1
8.1**
20.3***
1.3
0.4
8. Ghana LEAP - Impacts on any missed school
Impacts on attendance for younger children and older girls
-25
-20
-15
-10
-5
0
5
10
15
All children
5 - 17 years
All Boys Girls All Boys Girls
Children 5 - 12 years Children 13 - 17 years
-8.5***
-10.5***
-13.0***
-8.3**
-5.4
0.4
-9.8*
9. Impacts on school enrollment by cognitive ability
Higher impact among older children with low
cognitive ability (the ‘marginal’ child)
-20
-10
0
10
20
30
40
All children 5 - 17
years
Children 5 - 12
years
Children 13 - 17
years
Boys 5 - 17 years Girls 5 - 17 years
Low cognitive ability High cognitive ability
-6.4**
22.0***
4.1*
8.5*
5.4*
10. Impacts on any missed school by cognitive ability
Strong impact observed among children with low
cognitive ability compensating behavior of
parents
-30
-25
-20
-15
-10
-5
0
5
10
15
20
25
All children 5 - 17
years
Children 5 - 12
years
Children 13 - 17
years
Boys 5 - 17 years Girls 5 - 17 years
Low cognitive ability High cognitive ability
-8.7**
-12.8*** -12.8**
11. Pathways of impact
How did LEAP create an increase in enrolment and
attendance?
Analyze effects on schooling inputs:
Significant increases in schooling expenditures (books,
uniforms, total schooling expenditures)
LEAP may have loosened constraints on out-of-
pocket educational expenditures
12. Conclusion
Ghana LEAP showed strong impacts on children’s
schooling
Important to move beyond average treatment effects
Strong impacts on adolescent boys positive
elasticity of demand for schooling
Significant increase in enrolment for older children
with low cognitive ability
Suggests that parents demonstrate compensating
behavior towards their children, if given the chance
13. School enrollment impacts among secondary age children
strong, equal to those from CCTs in Latin America
0
2
4
6
8
10
12
14
16
18
20
Notes: Bars represent percentage point impacts. Enrollment for primary aged children already high, thus
impacts reported at secondary age enrollment
Cross-country impacts on secondary age school enrollment
(percentage point impacts similar or above Latin American CCTs)
Zimbabwe: No impacts on enrollment (crowding out due to Basic Education
Assistance Module - BEAM). However, 7 pp increase in attendance.
Hinweis der Redaktion
Impacts at primary not large because primary rates already high
Impacts on secondary school age children are pronounced, and comparable to those of Latin American CCTs
Again, this result is important
It makes clear that poverty is an important barrier to school participation
Addressing poverty will lead to improved school participation, also in the absence of schooling conditions. Rigorous studies carried out as part of the TP make this case convincingly.