Putting Children First: Identifying solutions and taking action to tackle poverty and inequality in Africa.
Addis Ababa, Ethiopia, 23-25 October 2017
This three-day international conference aimed to engage policy makers, practitioners and researchers in identifying solutions for fighting child poverty and inequality in Africa, and in inspiring action towards change. The conference offered a platform for bridging divides across sectors, disciplines and policy, practice and research.
How to Save a Place: Get the Word Out Far And Wide
Putting Children First: Session 1.6.B Catherine Porter - Smarter social protection [23-Oct-17]
1. Smarter social protection? Impacts of Ethiopia’s
Productive Safety Net on Child Cognitive Outcomes
Marta Favara, Catherine Porter, Tassew Woldehanna
Putting Children First Conference, Addis Ababa
catherine.porter@hw.ac.uk
September 23rd, 2017
(Porter, HWU) Smarter social Protection September 23rd, 2017 1 / 17
2. Overview
1 Introduction
2 Data
3 Research outline
4 Results
5 Conclusions and next steps
(Porter, HWU) Smarter social Protection September 23rd, 2017 2 / 17
3. Motivation
Productive Safety Net Program in Ethiopia (PSNP) transfer
cash/food for work
(Porter, HWU) Smarter social Protection September 23rd, 2017 3 / 17
4. Motivation
Productive Safety Net Program in Ethiopia (PSNP) transfer
cash/food for work
Work requirement means there is a (+ve) income effect but also a
(poss -ve) substitution effect on time use
(Porter, HWU) Smarter social Protection September 23rd, 2017 3 / 17
5. Motivation
Productive Safety Net Program in Ethiopia (PSNP) transfer
cash/food for work
Work requirement means there is a (+ve) income effect but also a
(poss -ve) substitution effect on time use
Previous evaluations have shown improvements in food security,
consumption (IFPRI evaluation)
(Porter, HWU) Smarter social Protection September 23rd, 2017 3 / 17
6. Motivation
Productive Safety Net Program in Ethiopia (PSNP) transfer
cash/food for work
Work requirement means there is a (+ve) income effect but also a
(poss -ve) substitution effect on time use
Previous evaluations have shown improvements in food security,
consumption (IFPRI evaluation)
Porter and Goyal (2016) find +ve effect on child nutrition
(height-for-age) using YL data from 2009
(Porter, HWU) Smarter social Protection September 23rd, 2017 3 / 17
7. Motivation
Productive Safety Net Program in Ethiopia (PSNP) transfer
cash/food for work
Work requirement means there is a (+ve) income effect but also a
(poss -ve) substitution effect on time use
Previous evaluations have shown improvements in food security,
consumption (IFPRI evaluation)
Porter and Goyal (2016) find +ve effect on child nutrition
(height-for-age) using YL data from 2009
Does this extend through to educational achievements in 2013 ?
(Porter, HWU) Smarter social Protection September 23rd, 2017 3 / 17
8. Context and previous literature on cognitive outcomes
Evidence on improvement in enrollment, grade, school attendance
(Fiszbein and Schady 2009 review, Glewwe and Kassouf, 2012 Bolsa
Escola)
(Porter, HWU) Smarter social Protection September 23rd, 2017 4 / 17
9. Context and previous literature on cognitive outcomes
Evidence on improvement in enrollment, grade, school attendance
(Fiszbein and Schady 2009 review, Glewwe and Kassouf, 2012 Bolsa
Escola)
Baird et al (2012, Malawi) CCT outperforms UCT at school
enrolment (2013, systematic review): CCT and UCT both have
significant effect on enrolment and attendance
(Porter, HWU) Smarter social Protection September 23rd, 2017 4 / 17
10. Context and previous literature on cognitive outcomes
Evidence on improvement in enrollment, grade, school attendance
(Fiszbein and Schady 2009 review, Glewwe and Kassouf, 2012 Bolsa
Escola)
Baird et al (2012, Malawi) CCT outperforms UCT at school
enrolment (2013, systematic review): CCT and UCT both have
significant effect on enrolment and attendance
Less evidence for any impact on cognitive achievement
(Porter, HWU) Smarter social Protection September 23rd, 2017 4 / 17
11. Context and previous literature on cognitive outcomes
Evidence on improvement in enrollment, grade, school attendance
(Fiszbein and Schady 2009 review, Glewwe and Kassouf, 2012 Bolsa
Escola)
Baird et al (2012, Malawi) CCT outperforms UCT at school
enrolment (2013, systematic review): CCT and UCT both have
significant effect on enrolment and attendance
Less evidence for any impact on cognitive achievement
Akresh et al (2016, Burkina Faso) CCT outperforms UCT on
enrolment, neither have impact on cognitive.
(Porter, HWU) Smarter social Protection September 23rd, 2017 4 / 17
12. Context and previous literature on cognitive outcomes
Evidence on improvement in enrollment, grade, school attendance
(Fiszbein and Schady 2009 review, Glewwe and Kassouf, 2012 Bolsa
Escola)
Baird et al (2012, Malawi) CCT outperforms UCT at school
enrolment (2013, systematic review): CCT and UCT both have
significant effect on enrolment and attendance
Less evidence for any impact on cognitive achievement
Akresh et al (2016, Burkina Faso) CCT outperforms UCT on
enrolment, neither have impact on cognitive.
Barham et al (2013, Nicaragua) cognitive impacts for boys of CCT
(Porter, HWU) Smarter social Protection September 23rd, 2017 4 / 17
13. PSNP research findings overall context
Positive overall on food security and consumption:
Berhane et al (2014) improved food security (also Yablonsky, 2007)
However mixed on child time use:
(Porter, HWU) Smarter social Protection September 23rd, 2017 5 / 17
14. PSNP research findings overall context
Positive overall on food security and consumption:
Berhane et al (2014) improved food security (also Yablonsky, 2007)
Participant households more likely to consume 1,800 calories per day
than non-beneficiaries (Save the Children UK, 2008)
However mixed on child time use:
(Porter, HWU) Smarter social Protection September 23rd, 2017 5 / 17
15. PSNP research findings overall context
Positive overall on food security and consumption:
Berhane et al (2014) improved food security (also Yablonsky, 2007)
Participant households more likely to consume 1,800 calories per day
than non-beneficiaries (Save the Children UK, 2008)
Participants avoid selling assets during crisis (Alderman and Yemtsov,
2012)
However mixed on child time use:
(Porter, HWU) Smarter social Protection September 23rd, 2017 5 / 17
16. PSNP research findings overall context
Positive overall on food security and consumption:
Berhane et al (2014) improved food security (also Yablonsky, 2007)
Participant households more likely to consume 1,800 calories per day
than non-beneficiaries (Save the Children UK, 2008)
Participants avoid selling assets during crisis (Alderman and Yemtsov,
2012)
Positive effects on child nutrition (Porter and Goyal, 2016)
However mixed on child time use:
(Porter, HWU) Smarter social Protection September 23rd, 2017 5 / 17
17. PSNP research findings overall context
Positive overall on food security and consumption:
Berhane et al (2014) improved food security (also Yablonsky, 2007)
Participant households more likely to consume 1,800 calories per day
than non-beneficiaries (Save the Children UK, 2008)
Participants avoid selling assets during crisis (Alderman and Yemtsov,
2012)
Positive effects on child nutrition (Porter and Goyal, 2016)
However mixed on child time use:
Emirie et al (2009) increased school enrollment
(Porter, HWU) Smarter social Protection September 23rd, 2017 5 / 17
18. PSNP research findings overall context
Positive overall on food security and consumption:
Berhane et al (2014) improved food security (also Yablonsky, 2007)
Participant households more likely to consume 1,800 calories per day
than non-beneficiaries (Save the Children UK, 2008)
Participants avoid selling assets during crisis (Alderman and Yemtsov,
2012)
Positive effects on child nutrition (Porter and Goyal, 2016)
However mixed on child time use:
Emirie et al (2009) increased school enrollment
Increase in paid and unpaid work (Tafere and Woldehanna, 2012)
(Porter, HWU) Smarter social Protection September 23rd, 2017 5 / 17
19. PSNP research findings overall context
Positive overall on food security and consumption:
Berhane et al (2014) improved food security (also Yablonsky, 2007)
Participant households more likely to consume 1,800 calories per day
than non-beneficiaries (Save the Children UK, 2008)
Participants avoid selling assets during crisis (Alderman and Yemtsov,
2012)
Positive effects on child nutrition (Porter and Goyal, 2016)
However mixed on child time use:
Emirie et al (2009) increased school enrollment
Increase in paid and unpaid work (Tafere and Woldehanna, 2012)
Girls work longer hours (Camfield, 2014)
(Porter, HWU) Smarter social Protection September 23rd, 2017 5 / 17
20. Public Works and Child Cognitive Outcomes
Mixed results for child cognitive outcomes of India’s NREGA, the largest
workfare programme in the world:
Mani et al (2014) find strong positive effects on grade progression and
a number of cognitive skills tests for the Young Lives cohorts in 2013
(Porter, HWU) Smarter social Protection September 23rd, 2017 6 / 17
21. Public Works and Child Cognitive Outcomes
Mixed results for child cognitive outcomes of India’s NREGA, the largest
workfare programme in the world:
Mani et al (2014) find strong positive effects on grade progression and
a number of cognitive skills tests for the Young Lives cohorts in 2013
Shah and Steinberg (2015) find the opposite - NREGs hampers
progression and cognition due to time demands - though this is
mainly for children over 12
(Porter, HWU) Smarter social Protection September 23rd, 2017 6 / 17
22. PSNP Background
2005 Productive Safety Net Scheme introduced after long history of
responsive food aid and drought
Workfare programme for 80% of participants, UCT for others
(labour-poor hh)
Cash-first principle (tho in practice 60-40)
Co-ordination between donors, and with Government innovation
In 2013 7.2 million beneficiaries (roughly 10% of the national
population) in 8 of 10 regions.
Phase 3 from 2010-2015 attempted to improve timeliness and
predictability of transfers, strengthen public works and accountability
(Porter, HWU) Smarter social Protection September 23rd, 2017 7 / 17
23. Young Lives dataset
Younger Cohort 2,000 children born in 2000, Older cohort 1,000
children born in 1994
(Porter, HWU) Smarter social Protection September 23rd, 2017 8 / 17
24. Young Lives dataset
Younger Cohort 2,000 children born in 2000, Older cohort 1,000
children born in 1994
20 sentinel sites (random selection within sites)
(Porter, HWU) Smarter social Protection September 23rd, 2017 8 / 17
25. Young Lives dataset
Younger Cohort 2,000 children born in 2000, Older cohort 1,000
children born in 1994
20 sentinel sites (random selection within sites)
Multi-purpose survey including cognitive skills, nutrition, education,
non-cognitive skills, household background, program participation
(Porter, HWU) Smarter social Protection September 23rd, 2017 8 / 17
26. Young Lives dataset
Younger Cohort 2,000 children born in 2000, Older cohort 1,000
children born in 1994
20 sentinel sites (random selection within sites)
Multi-purpose survey including cognitive skills, nutrition, education,
non-cognitive skills, household background, program participation
4 rounds (2002, 2006, 2009, 2013)
(Porter, HWU) Smarter social Protection September 23rd, 2017 8 / 17
27. Young Lives dataset
Younger Cohort 2,000 children born in 2000, Older cohort 1,000
children born in 1994
20 sentinel sites (random selection within sites)
Multi-purpose survey including cognitive skills, nutrition, education,
non-cognitive skills, household background, program participation
4 rounds (2002, 2006, 2009, 2013)
Round 3 and 4, sibling information collected (for one younger sibling)
(Porter, HWU) Smarter social Protection September 23rd, 2017 8 / 17
28. Young Lives dataset
Younger Cohort 2,000 children born in 2000, Older cohort 1,000
children born in 1994
20 sentinel sites (random selection within sites)
Multi-purpose survey including cognitive skills, nutrition, education,
non-cognitive skills, household background, program participation
4 rounds (2002, 2006, 2009, 2013)
Round 3 and 4, sibling information collected (for one younger sibling)
Attrition rate around 2.2 percent (very low)
(Porter, HWU) Smarter social Protection September 23rd, 2017 8 / 17
30. Cognitive Tests
Peabody Picture Vocabulary Test
Administered to YC, siblings but not the OC in round 4
Maths test in R1-4 for OC and YC (not siblings)
Difficulty of using raw scores across rounds/ages - we use comparable
Item-Response-Theory scores (Leon and Singh, 2017)
Therefore results interpreted slightly differently
(Porter, HWU) Smarter social Protection September 23rd, 2017 10 / 17
31. Framework: Cognitive development of children
Following Glewwe and Muralidharan (2016)
(Porter, HWU) Smarter social Protection September 23rd, 2017 11 / 17
32. Framework: Cognitive development of children
Following Glewwe and Muralidharan (2016)
Households invest in children subject to their preferences and
constraints- there is a “production function” for child cognitive
development, which depends on nutrition, learning activities inside
school, home.
(Porter, HWU) Smarter social Protection September 23rd, 2017 11 / 17
33. Framework: Cognitive development of children
Following Glewwe and Muralidharan (2016)
Households invest in children subject to their preferences and
constraints- there is a “production function” for child cognitive
development, which depends on nutrition, learning activities inside
school, home.
Social protection is external to the household - with income and
time-use effects
(Porter, HWU) Smarter social Protection September 23rd, 2017 11 / 17
34. Framework: Cognitive development of children
Following Glewwe and Muralidharan (2016)
Households invest in children subject to their preferences and
constraints- there is a “production function” for child cognitive
development, which depends on nutrition, learning activities inside
school, home.
Social protection is external to the household - with income and
time-use effects
Not all children have the same underlying abilities, and much is
unobserved (despite rich dataset)
(Porter, HWU) Smarter social Protection September 23rd, 2017 11 / 17
35. Framework: Cognitive development of children
Following Glewwe and Muralidharan (2016)
Households invest in children subject to their preferences and
constraints- there is a “production function” for child cognitive
development, which depends on nutrition, learning activities inside
school, home.
Social protection is external to the household - with income and
time-use effects
Not all children have the same underlying abilities, and much is
unobserved (despite rich dataset)
Todd and Wolpin (2003, 2007) note that under certain assumptions
then a “lag value added” model can be specified: the immediate lag
of achievement serving as a proxy for all previous inputs, and ability.
(Porter, HWU) Smarter social Protection September 23rd, 2017 11 / 17
36. Framework: Cognitive development of children
Following Glewwe and Muralidharan (2016)
Households invest in children subject to their preferences and
constraints- there is a “production function” for child cognitive
development, which depends on nutrition, learning activities inside
school, home.
Social protection is external to the household - with income and
time-use effects
Not all children have the same underlying abilities, and much is
unobserved (despite rich dataset)
Todd and Wolpin (2003, 2007) note that under certain assumptions
then a “lag value added” model can be specified: the immediate lag
of achievement serving as a proxy for all previous inputs, and ability.
Vector of controls includes other exogenous household controls
(household size, shocks, prices).
(Porter, HWU) Smarter social Protection September 23rd, 2017 11 / 17
37. Framework: Cognitive development of children
Following Glewwe and Muralidharan (2016)
Households invest in children subject to their preferences and
constraints- there is a “production function” for child cognitive
development, which depends on nutrition, learning activities inside
school, home.
Social protection is external to the household - with income and
time-use effects
Not all children have the same underlying abilities, and much is
unobserved (despite rich dataset)
Todd and Wolpin (2003, 2007) note that under certain assumptions
then a “lag value added” model can be specified: the immediate lag
of achievement serving as a proxy for all previous inputs, and ability.
Vector of controls includes other exogenous household controls
(household size, shocks, prices).
Also used e.g. by Andrabi et al (2011)
(Porter, HWU) Smarter social Protection September 23rd, 2017 11 / 17
38. Empirical Strategy (1)
Estimate reduced-form: cognitive development in 2013, as a function
of child and household characteristics
(Porter, HWU) Smarter social Protection September 23rd, 2017 12 / 17
39. Empirical Strategy (1)
Estimate reduced-form: cognitive development in 2013, as a function
of child and household characteristics
YL child is aged 12 years in 2013.
(Porter, HWU) Smarter social Protection September 23rd, 2017 12 / 17
40. Empirical Strategy (1)
Estimate reduced-form: cognitive development in 2013, as a function
of child and household characteristics
YL child is aged 12 years in 2013.
PSNP enters directly into estimating equation with exogenous
controls - dummy variable if a HH member participated in either
public works or direct support
(Porter, HWU) Smarter social Protection September 23rd, 2017 12 / 17
41. Empirical Strategy (1)
Estimate reduced-form: cognitive development in 2013, as a function
of child and household characteristics
YL child is aged 12 years in 2013.
PSNP enters directly into estimating equation with exogenous
controls - dummy variable if a HH member participated in either
public works or direct support
Include lag (pre-prog) child achievement as per conceptual model, to
control for ability and child-hh unobservable factors. Lagged
(previous) ability is measured at age 8
(Porter, HWU) Smarter social Protection September 23rd, 2017 12 / 17
42. Empirical Strategy (1)
Estimate reduced-form: cognitive development in 2013, as a function
of child and household characteristics
YL child is aged 12 years in 2013.
PSNP enters directly into estimating equation with exogenous
controls - dummy variable if a HH member participated in either
public works or direct support
Include lag (pre-prog) child achievement as per conceptual model, to
control for ability and child-hh unobservable factors. Lagged
(previous) ability is measured at age 8
Cluster fixed effects to control for village unobservable characteristics
(school quality etc)
(Porter, HWU) Smarter social Protection September 23rd, 2017 12 / 17
43. Empirical Strategy (1)
Estimate reduced-form: cognitive development in 2013, as a function
of child and household characteristics
YL child is aged 12 years in 2013.
PSNP enters directly into estimating equation with exogenous
controls - dummy variable if a HH member participated in either
public works or direct support
Include lag (pre-prog) child achievement as per conceptual model, to
control for ability and child-hh unobservable factors. Lagged
(previous) ability is measured at age 8
Cluster fixed effects to control for village unobservable characteristics
(school quality etc)
Remaining challenge is non-random targeting
(Porter, HWU) Smarter social Protection September 23rd, 2017 12 / 17
44. Empirical Strategy (2): Selection of Comparison Group
Selection of comparison group: never treated may be systematically
different from beneficiaries
Table: PSNP groups defined by years of treatment
PSNP Groups Obs Percentage
Non-beneficiaries 561 (190) 59.2% (33.0%)
2009 PSNP only 140 14.8% (24.3%)
2009 & 2013 PSNP 246 25.9% (42.7%)
Total 947 (576) 100.0%
(Porter, HWU) Smarter social Protection September 23rd, 2017 13 / 17
45. Empirical Strategy (2): Selection of Comparison Group
Selection of comparison group: never treated may be systematically
different from beneficiaries
We create “restricted comparison group” of more comparable
households
Table: PSNP groups defined by years of treatment
PSNP Groups Obs Percentage
Non-beneficiaries 561 (190) 59.2% (33.0%)
2009 PSNP only 140 14.8% (24.3%)
2009 & 2013 PSNP 246 25.9% (42.7%)
Total 947 (576) 100.0%
(Porter, HWU) Smarter social Protection September 23rd, 2017 13 / 17
46. Empirical Strategy (2): Selection of Comparison Group
Selection of comparison group: never treated may be systematically
different from beneficiaries
We create “restricted comparison group” of more comparable
households
Households receiving food aid/PW in 2005 (a criterion of inclusion in
PSNP)
Table: PSNP groups defined by years of treatment
PSNP Groups Obs Percentage
Non-beneficiaries 561 (190) 59.2% (33.0%)
2009 PSNP only 140 14.8% (24.3%)
2009 & 2013 PSNP 246 25.9% (42.7%)
Total 947 (576) 100.0%
(Porter, HWU) Smarter social Protection September 23rd, 2017 13 / 17
47. Empirical Strategy (2): Selection of Comparison Group
Selection of comparison group: never treated may be systematically
different from beneficiaries
We create “restricted comparison group” of more comparable
households
Households receiving food aid/PW in 2005 (a criterion of inclusion in
PSNP)
Households who reported being shortlisted for PSNP (but due to
budget constraints were not included)
Table: PSNP groups defined by years of treatment
PSNP Groups Obs Percentage
Non-beneficiaries 561 (190) 59.2% (33.0%)
2009 PSNP only 140 14.8% (24.3%)
2009 & 2013 PSNP 246 25.9% (42.7%)
Total 947 (576) 100.0%
(Porter, HWU) Smarter social Protection September 23rd, 2017 13 / 17
48. Empirical Strategy (2): Selection of Comparison Group
Selection of comparison group: never treated may be systematically
different from beneficiaries
We create “restricted comparison group” of more comparable
households
Households receiving food aid/PW in 2005 (a criterion of inclusion in
PSNP)
Households who reported being shortlisted for PSNP (but due to
budget constraints were not included)
These households have much more similar baseline characteristics but a
smaller group
Table: PSNP groups defined by years of treatment
PSNP Groups Obs Percentage
Non-beneficiaries 561 (190) 59.2% (33.0%)
2009 PSNP only 140 14.8% (24.3%)
2009 & 2013 PSNP 246 25.9% (42.7%)
Total 947 (576) 100.0%
(Porter, HWU) Smarter social Protection September 23rd, 2017 13 / 17
49. Headline results: positive effects on cognitive outcomes
though some reservations
PSNP overall effect on maths scores using restricted sample only
We split PSNP into a) continuing beneficiaries b) graduates
(Porter, HWU) Smarter social Protection September 23rd, 2017 14 / 17
50. Headline results: positive effects on cognitive outcomes
though some reservations
PSNP overall effect on maths scores using restricted sample only
No difference across gender
We split PSNP into a) continuing beneficiaries b) graduates
(Porter, HWU) Smarter social Protection September 23rd, 2017 14 / 17
51. Headline results: positive effects on cognitive outcomes
though some reservations
PSNP overall effect on maths scores using restricted sample only
No difference across gender
No difference comparing with the Older Cohort
We split PSNP into a) continuing beneficiaries b) graduates
(Porter, HWU) Smarter social Protection September 23rd, 2017 14 / 17
52. Headline results: positive effects on cognitive outcomes
though some reservations
PSNP overall effect on maths scores using restricted sample only
No difference across gender
No difference comparing with the Older Cohort
No effect on language acquisition
We split PSNP into a) continuing beneficiaries b) graduates
(Porter, HWU) Smarter social Protection September 23rd, 2017 14 / 17
53. Headline results: positive effects on cognitive outcomes
though some reservations
PSNP overall effect on maths scores using restricted sample only
No difference across gender
No difference comparing with the Older Cohort
No effect on language acquisition
We split PSNP into a) continuing beneficiaries b) graduates
Math score significant effect for graduates (only)
(Porter, HWU) Smarter social Protection September 23rd, 2017 14 / 17
54. Headline results: positive effects on cognitive outcomes
though some reservations
PSNP overall effect on maths scores using restricted sample only
No difference across gender
No difference comparing with the Older Cohort
No effect on language acquisition
We split PSNP into a) continuing beneficiaries b) graduates
Math score significant effect for graduates (only)
In all (math) models, no apparent difference in effect by cohort
(Porter, HWU) Smarter social Protection September 23rd, 2017 14 / 17
55. Table: PSNP Impact on Maths scores at age 12
Broad control group Restricted control group
(1) (2) (3) (4) (5) (6)
PSNP 0.051 0.161*
(0.075) (0.090)
2009 beneficiaries only 0.136* 0.129* 0.250** 0.221**
(0.076) (0.073) (0.085) (0.085)
2009 & 2013 beneficiaries -0.004 -0.033 0.101 0.057
(0.076) (0.078) (0.093) (0.088)
Math (z-score, age 5) 0.131*** 0.130*** 0.120*** 0.121** 0.119** 0.115**
(0.040) (0.040) (0.038) (0.042) (0.043) (0.042)
Socio-economic status (R2) x x x x x x
Demographics x x x x x x
Shocks (R3) x x
Shocks (R4) x x
Cluster fixed effect x x x x x x
Constant -2.384*** -2.337*** -2.219*** -1.633 -1.524 -1.554
(0.669) (0.665) (0.655) (1.253) (1.251) (1.086)
Observations 947 947 947 576 576 576
R-squared 0.118 0.120 0.151 0.116 0.119 0.161
(Porter, HWU) Smarter social Protection September 23rd, 2017 15 / 17
56. Mechanisms of impact?
We compare the control group, continuing beneficiaries and graduated
households
Graduating households do have significantly higher expenditure on
both food and non-food than continuing - Graduation criteria adhered
to in this sample?
Table: Year of graduation for “2009 only” PSNP beneficiaries
Year of graduation Frequency Percent
2007 1 1.92
2008 4 7.69
2009 1 1.92
2010 6 11.54
2011 9 17.31
2012 26 50.00
2013 5 9.62
Total 52 100.00
(Porter, HWU) Smarter social Protection September 23rd, 2017 16 / 17
57. Mechanisms of impact?
We compare the control group, continuing beneficiaries and graduated
households
Graduating households do have significantly higher expenditure on
both food and non-food than continuing - Graduation criteria adhered
to in this sample?
Graduating households spend longer time in school than either
continuing or control group households - plausible mechanism of
maths improvement?
Table: Year of graduation for “2009 only” PSNP beneficiaries
Year of graduation Frequency Percent
2007 1 1.92
2008 4 7.69
2009 1 1.92
2010 6 11.54
2011 9 17.31
2012 26 50.00
2013 5 9.62
Total 52 100.00
(Porter, HWU) Smarter social Protection September 23rd, 2017 16 / 17
58. Mechanisms of impact?
We compare the control group, continuing beneficiaries and graduated
households
Graduating households do have significantly higher expenditure on
both food and non-food than continuing - Graduation criteria adhered
to in this sample?
Graduating households spend longer time in school than either
continuing or control group households - plausible mechanism of
maths improvement?
Again no differences by gender of effects
Table: Year of graduation for “2009 only” PSNP beneficiaries
Year of graduation Frequency Percent
2007 1 1.92
2008 4 7.69
2009 1 1.92
2010 6 11.54
2011 9 17.31
2012 26 50.00
2013 5 9.62
Total 52 100.00
(Porter, HWU) Smarter social Protection September 23rd, 2017 16 / 17
59. Summary
With a restricted control group - we find effects of PSNP on maths
scores of 12 year old boys and girls
(Porter, HWU) Smarter social Protection September 23rd, 2017 17 / 17
60. Summary
With a restricted control group - we find effects of PSNP on maths
scores of 12 year old boys and girls
For both groups effects seem to be greater for those who have
graduated by 2013 (in 2012, mainly) - selection effect?
(Porter, HWU) Smarter social Protection September 23rd, 2017 17 / 17
61. Summary
With a restricted control group - we find effects of PSNP on maths
scores of 12 year old boys and girls
For both groups effects seem to be greater for those who have
graduated by 2013 (in 2012, mainly) - selection effect?
No impact on vocabulary
(Porter, HWU) Smarter social Protection September 23rd, 2017 17 / 17
62. Summary
With a restricted control group - we find effects of PSNP on maths
scores of 12 year old boys and girls
For both groups effects seem to be greater for those who have
graduated by 2013 (in 2012, mainly) - selection effect?
No impact on vocabulary
Time use - schooling higher (work not lower)
(Porter, HWU) Smarter social Protection September 23rd, 2017 17 / 17
63. Summary
With a restricted control group - we find effects of PSNP on maths
scores of 12 year old boys and girls
For both groups effects seem to be greater for those who have
graduated by 2013 (in 2012, mainly) - selection effect?
No impact on vocabulary
Time use - schooling higher (work not lower)
Latest round of data collection 2016 - children aged 15 years to
analyse
(Porter, HWU) Smarter social Protection September 23rd, 2017 17 / 17
64. Summary
With a restricted control group - we find effects of PSNP on maths
scores of 12 year old boys and girls
For both groups effects seem to be greater for those who have
graduated by 2013 (in 2012, mainly) - selection effect?
No impact on vocabulary
Time use - schooling higher (work not lower)
Latest round of data collection 2016 - children aged 15 years to
analyse
Comments most welcome...
(Porter, HWU) Smarter social Protection September 23rd, 2017 17 / 17