This presentation provides information about The Transfer Project and describes findings from a recent evaluation of the Kenya Cash Transfer Program for Orphans and Vulnerable Children.
Impact of the Kenya Cash Transfer for Orphans and Vulnerable Children on safe transitions to adulthood
1. Tia Palermo, Ph.D. & Amber Peterman, Ph.D.
UNICEF Office of ResearchâInnocenti
Research Seminar, Universita degli studi Firenze (DISEI)
May 2015
2. ⢠Adolescence = critical juncture in life
ďś Transitions have implications for later education, health, poverty, autonomy,
intimate partner violence, HIV risk (among others).
⢠Early transitions in Africa
ďś In Africa, 41% women ages 20 to 24 were first married (or in union) before 18
(UNICEF 2011)
ďś Globally, 14 million children are born to girls aged 15 to 19 annually, > 91% in
low and middle income countries (Kennedy et al. 2011).
⢠Reviews of âWhat worksâ to prevent adolescent childbearing and & HIV in
low-middle income countries: social cash transfers (SCTs) are promising
interventions (McQueston et al. 2013; Pettifor et al. 2012)
ďś However, evidence is drawn from few, geographically and programmatically
diverse studies.
Motivation
3. ⢠Who we are: Community of research, donor and implementing
partners
ďśUNICEF, FAO, UNC-Chapel Hill, AIR, National Governments
⢠What we do: Provide rigorous evidence on of government-run
large-scale (unconditional) SCTs
⢠Why SCTs: Our work is premised on the knowledge that
ďśIncome poverty has highly damaging impacts on human
development
ďśCash empowers people living in poverty to make their own
decisions on how to improve their lives
⢠Where: Ghana, Kenya, Lesotho, Malawi, South Africa, Zambia and
Zimbabwe (Tanzania?)
Transfer Project: Overview
4. ⢠Experimental and quasi-experimental study designs
⢠Longitudinal, mixed methods data collection to understand
impacts and pathways
⢠Focus on adolescents: How do cash transfers affect
structural determinants of safe transition to adulthood?
ďśOne-on-one interviews (up to 3 adolescents per
household) with same sex enumerators in Kenya,
Malawi, Zambia and Zimbabwe
Transfer Project: Research
5. Country Households
Sample Size
Adolescent
Age Range
Adolescent
Sample Size
Survey Years Design
Zambia MCTG 3078 13-17 2098 2011, 13,14 RCT
Zimbabwe HSCT 3063 13-20 1170 2013, 15, 16 District Matched
Case Control
Malawi SCT 3200 13-19 2109 2013, 14, 15 RCT
Kenya CT-OVC 1913 15-25 2223 2007, 09, 11 RCT
Tanzania PSSN TBD TBD TBD 2015, 17 RCT
Transfer project: Data collection (ongoing)
6. 1. Explore whether the Government of Kenyaâs Cash Transfer for
Orphans and Vulnerable Children (CT-OVC) has impacts on:
ďś Early pregnancy and marriage among sample of girls aged 12 to 24:
Handa, Peterman et al. 2015.
ďś Sexual debut and risky sex among sample of boys and girls aged 15 to
24. Handa, Halpern et al. 2014 and Handa, Palermo et al. 2015.
2. Examine pathways through which the Kenyan CT-OVC may affect
outcomes, including increased education and through mobility of
adolescents.
3. Draw parallels with other Transfer Project countries to inform
government policy.
Kenya case study: Aims
7. Conceptual Model: Cash transfers and HIV Risk
Figure Source: UNICEF et al. (2015). âSocial Protection Programs Contribute to HIV Prevention.â
http://strive.lshtm.ac.uk/system/files/attachments/Social%20protection%20programmes%20contribut
e%20to%20HIV%20prevention%20brief.pdf
8. Conceptual framework: SCTs and safe transitions?
⢠Reducing economic insecurity and negative coping strategies
ďś Buffer against short-term shocks, decreasing likelihood of need to marry
off girls, reduce household size and obtain brideprice payments
ďś Girls less likely to engage in intergenerational sex/ sexual exploitation
⢠Increase investment in girls education
ďś Schooling could reduce exposure to risky environments
ďś Increase health knowledge
ďś Perception that marriage and fertility come after education is finished
ďś Delay sexual debut or risky sexual activities
ďś Increase aspirations, positive future outlook
⢠Improving access to health case
ďś Increase in uptake of HIV treatment and care; reducing the vulnerability
to infections to those exposed or affected by virus
9. ⢠Mexico, Honduras and Nicaragua, government CCTs: Among women aged <20,
programs had negative but insignificant effects on fertility (Stecklov et al. 2006).
However, data from urban expansion in Mexico (2002-2004) showed significant
impacts on delayed marriage and pregnancies (Gulematova 2009).
⢠Pakistan Female School Stipend Program, CCT: Targeted girls 6th to 8th grade over
six years, increased age at marriage by 1.2 to 1.5 years using a quasi-experimental
case-control design. No impact on fertility outcomes (Alam et al. 2010).
⢠Malawi Zomba, UCT/CCT: Reduced pregnancy in last year and early marriage
among girls aged 13 to 22 at baseline over a 14-month period. Impacts driven by
girls who were either out of school at baseline or recent dropouts (Baird et al.
2010).
⢠South Africa Child Support Grant, UCT: Reduction in pregnancy found in analysis
using eligibility and enrollment criteria and propensity score matching (Heinrich et
al. 2012)
Evidence: SCTs, early marriage and pregnancy
10. ⢠Malawi Zomba, UCT/CCT: Transfers reduced prevalence of HIV
and Herpes Simplex 2, but results sensitive to weighting (Baird et al.
2012); reduced frequency of sexual activity & probability of partner
>1 year older (Baird et al. 2010)
⢠South Africa Child Support Grant, UCT: Transfers reduced
incidence and prevalence of transactional sex & age-disparate sex
among adolescents 10-18 years - PSM (Cluver et al. 2013); reduced
sexual debut (females only) and number of partners (males and
females) among adolescents aged 15-17
⢠Mexico Oportunidades, CCT: IV approach no impacts on sexual
debut & condom use among urban youth aged 12-24 (Galarraga &
Gertler 2009)
Evidence: SCTs and HIV risk
11. ⢠Governmentâs flagship social protection program started in mid-
2007 after small pre-pilot in 2004-2006.
⢠Reaches >135,000 households and over 280,000 OVC (aged <18
years).
⢠Targets households who are ultra-poor and contain an OVC.
⢠Flat monthly transfer of ~21 USD (Ksh 1500).
⢠Transfer given to caregiver, 80% of which are female: Care and
protection of the resident OVC is householdâs responsibility for
receiving the cash payment.
⢠Currently no punitive sanctions for noncompliance.
Kenya Cash Transfer Program for Orphans and Vulnerable
Children (CT-OVC)
12. 7 pilot districts,
4 Locations per
district
1. Kisumu
2. Migori
3. Homa Bay
4. Suba
5. Nairobi
6. Garissa
7. Kwale
13. ⢠Cluster randomized longitudinal design â 2 Locations per district
randomized to delayed-entry control group:
ďś Baseline (2007): 1,542 (T) and 755 (C) households (a ratio of 2:1).
ďś Wave 2 (2009): 1,325 (T) and 583 (C) households
ďś Wave 3 (2011): 1,280 (T) and 531 (C) households
⢠Household level attrition high between 2007 â 2009 coinciding with post-
election violence. Attrition tests confirm few differential attrition between
treatment and control with respect to covariates.
Sample Design and Data
14. ⢠Wave 3 only contained fertility module administered to females
aged 15 to 49.
⢠Sample restricted to girls aged 12 to 24 at wave 3, who had never
given birth previous to the baseline survey (N = 1,547).
⢠Probit regression, standard errors clustered at the household level.
Model 1: Pr(Yij )= β0 + βij(Treatj)+ βij(Xij)+ βj(Xj)+ ξij
Model 2: Pr(Yij )= β0 + βij(Treatj) + βij(Zij)+ + βij(Xij)+ βj(Xj)+ ξij
⢠Vector of X controls (baseline): Individual level [age (2011), gender,
child/grandchild of household head), and household level (age of
head, schooling attainment of head, living in Nairobi district].
Methodology
16. Pregnancy and marriage outcomes
Lowess plot of ever been
pregnant, girls (N = 1,547)
Means: All (0.15), T (0.13), C (0.19)***
Lowess plot of ever been married
or co-habited, girls (N = 1,547)
Means: All (0.07), T (0.06), C (0.08)
Pregnancy and marriage outcomes
17. Impact estimates
Probit models with marginal effects. Robust t-statistics in parentheses clustered at the household level.
* p<0.1; ** p<0.05; ** p<0.05. All control variables from 2011 round with the exception of household head characteristics.
18. Kenya CT-OVC: Sexual debut
⢠Adolescent module introduced in 2011 (Wave 3)
⢠N=2210 youth aged 15 to 25 years interviewed
⢠Excluded youth new to the household since 2007 or sexually
debuted before baseline survey (N=1429 final analysis
sample)
⢠Mediators
ďś Education =1 if currently enrolled or completed at least grade 12
(reported in 2009)
ďś Not depressed = 1 if CES-D scale â¤10 (reported in 2011)
ďś Hope = 1 if Hope score âĽmedian of 22 (reported in 2011)
ďś Economic well-being = total household per capita consumption
expenditures on food and non-food items (reported in 2009)
19. Outcome by study arm in analytic sample, Kenyan adolescents,
Ages 15-25 years, 2011 (N=1429)
Full Sample
(N=1429)
Intervention
(N=1001)
Control
(N=428)
p-
value*
Mean Mean Mean
Sexual debut 0.383 0.357 0.444 0.002
Kenya CT-OVC: Sexual debut
20. Impacts on Sexual debut Youth, 15-25 (N=1429)
(1) (2) (3) (4) (5) (6)
Treatment -0.084*** -0.081*** -0.082*** -0.083*** -0.087*** -0.084***
(-2.779) (-2.660) (-2.702) (-2.731) (-2.872) (-2.760)
Currently
enrolled/completed
Standard 12 -0.190*** -0.189***
(-5.287) (-5.206)
Score 10 or below on
CESD -0.052* -0.037
(-1.746) (-1.163)
Hope>median -0.025 -0.003
(-0.958) (-0.124)
Log per capita
expenditures 0.021 0.033
(0.782) (1.237)
Probit models with marginal effects. Robust t-statistics in parentheses clustered at the household level.
* p<0.1; ** p<0.05; ** p<0.05. All control variables from 2011 round with the exception of household head characteristics.
21. (1) (2) (3) (4) (5) (6)
Treatment -0.121** -0.140*** -0.120** -0.122** -0.122**
-
0.143***
(-2.488) (-2.836) (-2.457) (-2.503) (-2.501) (-2.889)
Currently
enrolled/completed
Standard 12 -0.310***
-
0.300***
(-4.812) (-4.647)
Score 10 or below on
CESD -0.096* -0.054
(-1.947) (-1.031)
Hope>median -0.085** -0.055
(-1.975) (-1.248)
Log per capita
expenditures 0.009 0.035
(0.196) (0.725)
Sexual debut: Females, 15-25 (N=552)
Probit models with marginal effects. Robust t-statistics in parentheses clustered at the household level.
* p<0.1; ** p<0.05; ** p<0.05. All control variables from 2011 round with the exception of household head characteristics.
22. Sexual debut: Males, 15-25 (N=877)
(7) (8) (9) (10) (11) (12)
Treatment -0.072* -0.064 -0.070* -0.073* -0.078* -0.070*
(-1.813) (-1.604) (-1.754) (-1.826) (-1.954) (-1.775)
Currently
enrolled/completed
Standard 12 -0.122*** -0.125***
(-2.664) (-2.682)
Score 10 or below on
CESD -0.033 -0.031
(-0.863) (-0.764)
Hope>median 0.005 0.020
(0.133) (0.553)
Log per capita
expenditures 0.034 0.039
(1.018) (1.194)
Probit models with marginal effects. Robust t-statistics in parentheses clustered at the household level.
* p<0.1; ** p<0.05; ** p<0.05. All control variables from 2011 round with the exception of household head characteristics.
23. ⢠Reduce HIV risk factors
ďś Delay sexual debut among adolescents 15 - 25 years (8 pp)
ďś Delay first pregnancy among females 12 - 25 years (5 pp)
ďś Lower risk of depressive symptoms, especially among adolescent males (5 pp)
⢠Impacts on behavioral risk
ďś No significant impact on early marriage, sexual partner characteristics or other
risky behaviors (condom use).
⢠Pathways
ďś Remain unclear - Program impacts on mental health and education in this
sample limited to males
ďś Education protective males and females; mental health protective for females
Summary of Findings: Kenya CT-OVC
24. ⢠Evidence that addressing upstream structural factors can have
impacts on HIV-risk related behaviors
⢠Less conclusive evidence on conditional transfers and short-
term motivations
⢠Causal pathways still unclear
⢠Promise of investing in national integrated and
comprehensive programs - HIV prevention as an important
resulting dividend
⢠Similar large scale national âunconditionalâ or âsocialâ transfer
programs exist in Zambia, Zimbabwe, Malawi, South Africa,
Mozambique
ďśCurrently examining impacts
Program and policy implications
25. Emails: tmpalermo@unicef.org, apeterman@unicef.org
⢠Download briefs at:
http://www.cpc.unc.edu/projects/transfer/publications/briefs
⢠New (as of Feb 2015):
ďś Methodological brief on adolescent modules
ďś Methodological brief violence
ďś Brief on education impacts
The authors are grateful to members of the OVC-CT Evaluation Team, in particular to
the Technical Working Group of the Childrenâs Department, Ministry of Gender,
Children and Social Development, Government of Kenya, and to Daniel Musembi, for
support to the follow-up study.
⢠This study received funding from the U.S. National Institute of Mental Health
through Grant Number 1R01MH093241 and by Eunice Kennedy Shriver National
Institute of Child Health and Development R24 HD050924 and was approved by
the Kenya Medical Research Institute Ethics Review Committee (Protocol #265)
and the Institutional Review Board of the University of North Carolina.
Acknowledgements and more information
26. Alam A, Baez JE & Del Carpio SV (2010). Does cash for school influence young womenâs
behavior in the longer term? World Bank: Washington D.C.
Baird S, Chirwa E, McIntosh C & Ăzler B. (2010). The shortâterm impacts of a schooling
conditional cash transfer program on the sexual behavior of young women. Health
economics, 19(S1), 55-68
Cluver L, Boyes M, Orkin M, Pantelic M, Molwena T, & Sherr L. (2013). Child-focused
state cash transfers and adolescent risk of HIV infection in South Africa: a propensity-
score-matched case-control study. The Lancet Global Health, 1(6), e362-e370.
Baird SJ, Garfein RS, McIntosh CT, & Ozler, B. (2012). Effect of a cash transfer
programme for schooling on prevalence of HIV and herpes simplex type 2 in Malawi: a
cluster randomised trial. Lancet, 379(9823), 1320-1329.
Gulematova-Swan, M. (2009). Evaluating the Impact of Conditional Cash Transfer
Programs on Adolescent Decisions about Marriage and Fertility: The Case of
Oportunidades. Doctoral dissertation in Economics, University of Pennsylvania.
Handa S, Tucker C, Halpern C, Pettifor A & Thirumurthy, H. 2014. "The Government of
Kenya's Cash Transfer Program Reduces the Risk of Sexual Debut among Young People
Age 15-25." PloS one 9(1):e85473.
GalĂĄrraga O, & Gertler PJ. (2009). Conditional Cash & Adolescent Risk Behaviors:
Evidence from Urban Mexico. Policy Research Working Paper. December.
References (1 of 2)
27. Heinrich C, Hoddinott J. & M Samson. (2012). Reducing Adolescent Risky Behaviors in
a High-Risk Context: The Effects of Unconditional Cash Transfers in South Africa.
Working paper.
Kennedy E, Gray N, Azzopardi P, & Crieati M. (2011). Adolescent fertility and family
planning in East Asia and the Pacific: A review of DHS reports. Reproductive Health,
8(11).
McQueston K, Silverman R, & A Glassman. (2013). The Efficacy of Interventions to
Reduce Adolescent Childbearing in Low- and Middle-Income Countries: A Systematic
Review. Studies in Family Planning, 44(4): 369-388.
Pettifor A, Macphail C, Nguyen N, & Rosenberg M. (2012). Can Money Prevent the
Spread of HIV? A Review of Cash Payments for HIV Prevention. AIDS Behav.
Stecklov G, Winters P, Todd J & Regalia F. (2006). Demographic externalities from
poverty programs in developing countries: Experimental evidence from Latin America.
Department of Economics Working Paper Series. Washington, DC. American
University.
UNICEF. (2011). Child Marriage: Progress.
http://www.childinfo.org/marriage_progress.html
Walker, J. A. (2012). Early Marriage in Africa â Trends, Harmful Effects and
Intervention. African Journal of Reproductive Health. 16(2), 231.
References (2 of 2)
Editor's Notes
Aims: understand program impacts & pathways
Structural factors poverty, poor education, gender & power inequalities are determinants of HIV risk for young women (Pettifor and McCoy)
Reducing risky sexual behaviour by addressing structural drivers of HIV risk,
Reducing economic insecurity, increasing school enrolment and attendance, promoting gender equality, and through other complex pathways, and
Improving access to healthcare such as uptake of HIV treatment and care, which reduce the vulnerability to infection of those exposed to or affected by the virus.
Economic security: reduce transactional sex & indirect effects through increase in school attendance
Education: delay sexual debut, homogenous sexual and social networks, exposure to HIV education, enhance ability to act on HIV information. Zimbabwe HSCT increases use of condom at first sex; evidence from Kenya that CT may reduce transactional sex among girls currently in school. A quasi-experimental study using the introduction of universal primary education policies in Malawi and Uganda as a natural experiment found that a one-year increase in schooling decreased the probability of testing positive for HIV among adult women by 3 to 6 percentage points (Behrman 2014).Summary of pathways: Simmons, and Donald Bundy. 2008. "Education and vulnerability: the role of schools in protecting young women and girls from HIV in southern Africa." Aids 22:S41-S56
Mental Health and stress: existing evidence indicates poor mental health linked to HIV transmission
Gourlay, Annabelle, Isolde Birdthistle, Gitau Mburu, Kate Iorpenda, and Alison Wringe. 2013. "Barriers and facilitating factors to the uptake of antiretroviral drugs for prevention of mother-to-child transmission of HIV in sub-Saharan Africa: a systematic review." Journal of the International AIDS Society 16(1).
Donenberg, Geri R, Erin Emerson, Fred B Bryant, Helen Wilson, and Eryn Weber-Shifrin. 2001. "Understanding AIDS-risk behavior among adolescents in psychiatric care: Links to psychopathology and peer relationships." Journal of the American Academy of Child & Adolescent Psychiatry 40(6):642-53
4. Expectations: increase opportunity cost of childbearing, delay sexual debut
5. Violence: New evidence (see notes for details) on robust relationship between IPV and HIV risk in Africa:
Durevall, Dick, and Annika Lindskog. 2014. "Intimate partner violence and HIV in ten sub-Saharan African countries: what do the Demographic and Health Surveys tell us?" The Lancet Global Health.
Other cash transfer programs show potential for CTs in reducing IPV [e.g., Give Directly in Uganda (Haushofer and Shapiro 2013), WFP in Ecuador (Hidrobo, Heise, Peterman 2014), etc.]
Further, sexual violence in childhood/adolescence is a risk factor for risky sexual behaviors. Note: we canât test this with Zim data because of low sample size of individual panel and low % experiencing forced sex (other sexual violence only measured at follow-up).
Transfer project provides evidence of how CT programs can be sustainable & delivered at scale. Existing evidence on CTs and HIV often drawn from smaller, localized CTs.
The Transfer Project is now in a unique position to examine these hypotheses and pathways via adolescent modules being collected in various countries.
Cluver, Lucie, Mark Boyes, Mark Orkin, Marija Pantelic, Thembela Molwena, and Lorraine Sherr. 2013. "Child-focused state cash transfers and adolescent risk of HIV infection in South Africa: a propensity-score-matched case-control study." The Lancet Global Health 1(6):e362-e70.
Pettifor, A., C. MacPhail, N. Nguyen, and M. Rosenberg. "Can money prevent the spread of HIV? A review of cash payments for HIV prevention." AIDS Behav 16(7):1729-38.
UNDP discussion paper: Cash transfers and HIV-prevention (October 2014) http://www.undp.org/content/undp/en/home/librarypage/hiv-aids/discussion-paper--cash-transfers-and-hiv-prevention/
Bor 2014: Hazard of teen pregnancy (ages 14-17) is 30 lower among girls in birth cohorts eligible for the grant.
Pettifor et al. Lancet comment: âanti-poverty programs can later the context of sexual decision making and thus HIV riskâ
Jury is still out, but evidence to date suggests that reductions in HIV risk-related poverty due to addressing poverty & economic conditions,
and not from changing motivations.
Tanzania RESPECT Study, CCT: Transfers conditional on testing negative for 3 STIs associated with marginally significant reduced odds of combined STI measure (ages 18-30 and spouses aged 16+) (de Walque et al. 2012)
Note: TZ study ages 18-30 & spouses 16 and up; conditional on testing negative for three STIs (C trachomatis, N gonorrheoe and T vaginalis
12 combined prevalence OR=0.73 (0.47, 0.99)
Zomba: 2 years of cash transfers conditional on 80% attendance; examined HIV differences by CT and UCT arms COMBINED v. control
Combined treatment v. control:
Baseline schoolgirls: OR=0.36 (0.14, 0.91)
Baseline dropouts: OR=1.37 (0.72, 2.61)
-CCT: OR=0.29 (0.09-0.98)
-UCT: OR=0.47 (0.14, 1.59)
HSV-2: UCT OR=0.08 (0.01, 1.58); CCT not significant
Could be updated
% girls that have been preg/married
Education: mediates 4% of program impacts
Depression: mediates 2% of impacts
Hope scale: no evidence of mediation
Economic well-being: no evidence of mediation