Presentación de Julieta Trias, The Institute for Fiscal Studies and the University College London, durante el Tercer Seminario de Transferencias Condicionadas de Ingresos, realizado en Santiago de Chile el 01 y 02 de Diciembre de 2008.

1. Old and newOld and new
welfare: their relative effectwelfare: their relative effect
on child nutritionon child nutrition
Julieta TriasJulieta Trias
(joint with Orazio Attanasio –
Marcos Vera-Hernandez)
FAO – Chile, December 1 2008

2. InterventionsInterventions
 Interventions that can potentially improve nutrition: a)Interventions that can potentially improve nutrition: a)
conditional cash transfers (CCT), b)unconditional cash transfers,conditional cash transfers (CCT), b)unconditional cash transfers,
c) price subsidies, d)distribution of nutritional supplements, andc) price subsidies, d)distribution of nutritional supplements, and
e) childcare centres where children are fed and taken care.e) childcare centres where children are fed and taken care.
 CCT programs have become very popular tools forCCT programs have become very popular tools for
governments to relieve poverty and increase human capitalgovernments to relieve poverty and increase human capital
accumulation. There is a consensus that they are effectiveaccumulation. There is a consensus that they are effective
tools at increasing the uptake of preventive care (Lagardetools at increasing the uptake of preventive care (Lagarde
2007) but there is also controversy about their effect on2007) but there is also controversy about their effect on
nutritional status.nutritional status.
 For instance, PROGRESA increased the height of childrenFor instance, PROGRESA increased the height of children
under 36 months by 1 cm (Behrman and Hoddinott 2005,under 36 months by 1 cm (Behrman and Hoddinott 2005,
Gertler 2004, and RiveraGertler 2004, and Rivera et alet al 2004). Honduran CCT2004). Honduran CCT
programme was not found to achieve any improvement inprogramme was not found to achieve any improvement in
nutritional status (IFPRI 2003).nutritional status (IFPRI 2003).

3. ObjectiveObjective
 The objective of the paper is to compare the relativeThe objective of the paper is to compare the relative
merits of CCT programs with more traditionalmerits of CCT programs with more traditional
programs to improve children nutritional status.programs to improve children nutritional status.
 As far as we know, there has been no researchAs far as we know, there has been no research
comparing these type of programs.comparing these type of programs.
 We will compare the effect ofWe will compare the effect of Familias en Acción,Familias en Acción, a CCTa CCT
program with the effect ofprogram with the effect of Hogares Comunitarios,Hogares Comunitarios, aa
childcare and feeding program in Colombia onchildcare and feeding program in Colombia on
nutritional outcomesnutritional outcomes (HAZ, WAZ, WHZ, chronic,(HAZ, WAZ, WHZ, chronic,
global and acute malnutrition and risk of malnutrition)global and acute malnutrition and risk of malnutrition)
andand morbidity outcomesmorbidity outcomes ((diarrheadiarrhea (EDA) and acute(EDA) and acute
respiratory infection (ARI) ).respiratory infection (ARI) ).

4. ““Hogares Comunitarios”Hogares Comunitarios”
 Nutrition and childcare program introduced all overNutrition and childcare program introduced all over
Colombia in the mid 1980s.Colombia in the mid 1980s.
 The program is targeted to poor children between 0-6 yearsThe program is targeted to poor children between 0-6 years
old.old.
 Parents are required to pay a monthly fee about $US 4 perParents are required to pay a monthly fee about $US 4 per
month per child, although there is considerable variation inmonth per child, although there is considerable variation in
the amount across towns.the amount across towns.
 Children attending to the nurseries receive a lunch and twoChildren attending to the nurseries receive a lunch and two
snacks that include a nutritional beverage calledsnacks that include a nutritional beverage called bienestarinabienestarina..
Children attending to HC should receive the 70% ofChildren attending to HC should receive the 70% of
recommended daily intake.recommended daily intake.

5. ““Familias en AcciFamilias en Acción”ón”
 Large-scale welfare program introduced in 2001. In 2002, theLarge-scale welfare program introduced in 2001. In 2002, the
program registered 365,000 and currently involves more thanprogram registered 365,000 and currently involves more than
1.5 million households.1.5 million households.
 The program gives a monetary transfer to mothers providedThe program gives a monetary transfer to mothers provided
their children are up to date with growth and developmenttheir children are up to date with growth and development
monitoring visits and attend school regularly.monitoring visits and attend school regularly.
 Nutritional subsidy: $CO 46,500 (U$20) monthly per familyNutritional subsidy: $CO 46,500 (U$20) monthly per family
with children under 7. Primary School ($CO 14,000 ($US 5)with children under 7. Primary School ($CO 14,000 ($US 5)
pcm) and Secondary School ($CO 28,000 ($10 pcm)).pcm) and Secondary School ($CO 28,000 ($10 pcm)).
 EligibilityEligibility. Families with children under 18 classified as being. Families with children under 18 classified as being
in the lowest level of the official socio-economic classificationin the lowest level of the official socio-economic classification
(Sisben level 1).(Sisben level 1).
 No nutritional supplementation. Program perceived as anNo nutritional supplementation. Program perceived as an
alternative toalternative to HCHC for families with children 0-6.for families with children 0-6.

6.  Other issues:Other issues:
 FA operates by transferring money to the motherFA operates by transferring money to the mother
(unclear how much of this transfer ends up(unclear how much of this transfer ends up
benefiting the child). HC should provide foodbenefiting the child). HC should provide food
directly to the child while the child is in the HCdirectly to the child while the child is in the HC
centre.centre.
 FA program is relatively easy to expand to otherFA program is relatively easy to expand to other
households or municipalities but HC requires settinghouseholds or municipalities but HC requires setting
up the logistic of food purchase, providing trainingup the logistic of food purchase, providing training
to the child carers, monitoring to prevent the foodto the child carers, monitoring to prevent the food
from being resold or used by individuals differentfrom being resold or used by individuals different
from the children for whom it is intended, as well asfrom the children for whom it is intended, as well as
to have adequate premises for the children.to have adequate premises for the children.

7. OutlineOutline
 DataData
 Basic StatisticsBasic Statistics
 MethodologyMethodology
 ResultsResults
 Summary & concluding remarksSummary & concluding remarks

8. Data - ColombiaData - Colombia
 Survey collected to evaluate FA program +Survey collected to evaluate FA program +
administrative dataadministrative data
 First WaveFirst Wave: 2002. Collected in 122 communities, 57 are: 2002. Collected in 122 communities, 57 are
targeted by FA. (9.080 children 0-6 / 3.940 households)targeted by FA. (9.080 children 0-6 / 3.940 households)
 Second WaveSecond Wave: 2003 (8.880 children 0-6 / 2.760 households): 2003 (8.880 children 0-6 / 2.760 households)
 Third WaveThird Wave: 2005/2006 (not used): 2005/2006 (not used)
 FA surveyFA survey also collected data on the participation of thealso collected data on the participation of the
children on HCchildren on HC as well as on variables that are importantas well as on variables that are important
determinants of the participation in HC.determinants of the participation in HC.
 Information: socio-demographic characteristics,Information: socio-demographic characteristics,
anthropometric variables, distances to important places in theanthropometric variables, distances to important places in the
town such as nearest health centre and school, distance fromtown such as nearest health centre and school, distance from
the household to the nearest HC centre, current andthe household to the nearest HC centre, current and
retrospective information on participation in the HCretrospective information on participation in the HC
program.program.

9. 0.1.2.3.4
Density
0 1 2 3 4 5 6 7 8
number of payments
Payments Distribution
Familias en Accion - 2003
Basic StatisticsBasic Statistics

10. Age
Ever attended
to HC
None Ever attended
to HC
Only HC FA + HC in the
past
Only FA None
0 2.4 97.6 2.0 0.5 1.5 78.5 19.5
1 9.9 90.1 12.8 2.8 10.0 69.4 17.8
2 27.9 72.1 24.2 0.9 23.3 69.9 6.0
3 34.1 65.9 38.2 1.1 37.1 61.8 0.0
4 43.7 56.3 37.3 0.9 36.4 62.7 0.0
5 40.9 59.1 47.2 0.4 46.8 52.8 0.0
6 43.8 56.2 42.8 0.6 42.2 57.2 0.0
Age
2002 2003
Ever attended
to HC
None Ever attended
to HC
None
0 3.3 96.7 4.2 95.8
1 19.6 80.4 22.0 78.0
2 43.2 56.8 44.8 55.2
3 52.0 48.0 55.6 44.4
4 53.8 46.2 62.2 37.8
5 56.0 44.0 61.6 38.4
6 58.4 41.6 58.0 42.0
Municipalities where FA was implemented
2002 2003
Municipalities where FA was not implemented

11. Prevalence of chronic malnutrition (z-score HAZ <-2 s.d)
FA Treatment Control
Age 2002 2003 2002 2003
0 13.48 8.87 13.52 10.50
1 30.57 29.89 25.44 35.81
2 23.08 21.89 22.57 20.28
3 22.93 27.87 25.15 23.61
4 28.00 24.66 25.14 25.09
5 23.97 26.08 24.66 23.29
6 23.79 21.81 23.68 23.40
Total 24.06 23.71 23.52 23.54
Prevalence of global malnutrition (z-score WAZ <-2 s.d)
FA Treatment Control
Age 2002 2003 2002 2003
0 7.83 8.87 8.54 6.30
1 16.60 19.16 15.38 22.30
2 15.73 8.42 12.86 13.80
3 10.22 13.22 10.02 11.58
4 11.25 9.73 10.85 8.05
5 9.36 8.84 11.11 10.86
6 10.28 8.45 9.94 9.98
Total 11.34 10.56 11.10 11.38

12. Empirical SpecificationEmpirical Specification
 YYihmtihmt : nutritional status of child
i of household h, living in
municipality m, in year t.
 EFAEFAihtiht: nutritional subsidy per
child in household h until time
t (or exposure for child i at until t)
 EHCEHCihtiht: months in HC for
child i until time t (or exposure
for child i until t)
 XXihtiht:contains variables that are
specific to child i and
household h including a
dummy whether or not
household h has a child eligible
for the FA program.
 TTtt: dummy variable for each
wave
 θθmm : municipality fixed effects
Problem: EFA and EHC are endogenous variables
Strategy: IV approach with municipality FE
Identification: βFA relies on the availability of the program.
βHC relies on some households living closer to a HC centre
than others (Attanasio and Vera-Hernandez (2006)).
yihmt= βFA (EFA)ihmt+βHC(EHC)ihmt+ βXXiht +βTTimet+θm+εihmt

13. Treatments and InstrumentsTreatments and Instruments
 Treatments:Treatments:
 FA: accumulated conditional payments per child and lifeFA: accumulated conditional payments per child and life
exposure (#months FA/ child’s age), exposure scaledexposure (#months FA/ child’s age), exposure scaled
#children (#months FA/children under 17)#children (#months FA/children under 17)
 HC: # months child ever attended and life exposure.HC: # months child ever attended and life exposure.
 Instruments:Instruments:
 FA: potential conditional payments per child, potential lifeFA: potential conditional payments per child, potential life
exposure and potential exposure. Treatment in theexposure and potential exposure. Treatment in the
municipalitymunicipality
 HC: distance to the nearest HC (controls for other distance)HC: distance to the nearest HC (controls for other distance)
 Non linear prediction of months in HC (PHC). InteractionNon linear prediction of months in HC (PHC). Interaction
between PHC with FA treatment in the municipality.between PHC with FA treatment in the municipality.

14.  Child characteristicsChild characteristics: age, birth order, gender.: age, birth order, gender.
(inverse of age)(inverse of age)
 Mother’s characteristicsMother’s characteristics: age, education, height,: age, education, height,
marital status (single)marital status (single)
 Household characteristics:Household characteristics:
 Children under 7 in the household potentially eligible for FAChildren under 7 in the household potentially eligible for FA
(0 or 1 for controls and treatment)(0 or 1 for controls and treatment)
 # children 8-12, # children 13-17# children 8-12, # children 13-17
 household head’s age and education.household head’s age and education.
 Location:Location: area, travel time to health center, school andarea, travel time to health center, school and
town center.town center.
ControlsControls

15. Results – First StageResults – First Stage
 Distance to HCDistance to HC: 30 min. of extra travel to the HC increases the: 30 min. of extra travel to the HC increases the
subsidy per child by $CO 1,232 and reduces the time attending asubsidy per child by $CO 1,232 and reduces the time attending a
HC by 9.8 days (21.6 days considering distance at each wave).HC by 9.8 days (21.6 days considering distance at each wave).
 FA programFA program (treatment + 1 year pot. subsidy): reduces the(treatment + 1 year pot. subsidy): reduces the
attendance to HC in 6.1 months for children at the average ageattendance to HC in 6.1 months for children at the average age
(48 months) and increases the transfer in $CO 502,000.(48 months) and increases the transfer in $CO 502,000.
 Households with schooling age childHouseholds with schooling age child: An additional sibling in: An additional sibling in
secondary school age increases the nutritional subsidy by $COsecondary school age increases the nutritional subsidy by $CO
2000 per child and reduces the attendance to HC by 6.2 days.2000 per child and reduces the attendance to HC by 6.2 days.
 Single Mothers:Single Mothers: increases child’s attendance to HC by 28 daysincreases child’s attendance to HC by 28 days
and reduces the transfer per child by $CO 2,000.and reduces the transfer per child by $CO 2,000.
 Distance to health facilitiesDistance to health facilities: 30 min. of extra travel reduces the: 30 min. of extra travel reduces the
transfer by $CO 234 per child and increases the attendance totransfer by $CO 234 per child and increases the attendance to
HC by 3.4 days. However, those effects are not significant.HC by 3.4 days. However, those effects are not significant.

16. First Stage for program effect on HAZ
Negative
Binomial
First Stage FA First Stage
HC
Months in HC Conditional
Money per
Child¹ (mill)
Months in HC
FA treatment available in the community 0.525** 0.049*** 0.599
[0.234] [0.013] [0.891]
Potential conditional money per child¹ (mill) -4.198* 0.168 0.209
[2.177] [0.119] [5.552]
Potential conditional money per child¹ (mill) ^2 3.117 -0.014 8.961
[2.198] [0.192] [6.759]
age_m x potential conditional money per child¹ (mill) -0.02 0.006** -0.139
[0.071] [0.003] [0.160]
age_m^2 x potential conditional money per child¹ (mill) 0.001 0.000 -0.001
[0.001] [0.000] [0.002]
travel time to the nearest HC in minutes (at the time of
the first wave)/100 -1.508*** 0.007 -1.252
[0.294] [0.008] [1.407]
[travel time to the nearest HC in minutes (at the time of
the first wave)/100] ^2 0.252* -0.004 0.421
[0.145] [0.004] [0.542]
travel time to the nearest HC in minutes/100 -2.134*** 0.003 -3.152**
[0.305] [0.009] [1.341]
(travel time to the nearest HC in minutes/100)^2 0.663*** -0.001 0.807
[0.159] [0.004] [0.527]
travel time to HC at wave 1 x FA treatment available in
the community -1.165*** -0.005 -0.568
[0.254] [0.010] [0.910]
travel time to HC x FA treatment available in the
community 0.522** -0.018 0.969
[0.238] [0.011] [0.916]
prediction of months in HC 0.000 0.576***
[0.000] [0.066]
prediction of months in HC^2 0.000 -0.004***
[0.000] [0.001]
prediction of months in HC x treatment available in the
community 0.000 0.058
[0.000] [0.071]
Observations 8640 8640 8640
R-squared 0.56 0.36
Test instruments 638.19 91.03 30.13
Prob inst 0.00 0.00 0.00

17. Program effect HAZ
Chronic
Malnutrition
Risk of
Chronic
Malnutrition WAZ
Global
Malnutrition
Risk of
Global
Malnutrition
Conditional money pc (mill) ¹ 0.620*** -0.101 -0.329*** 0.978*** -0.187*** -0.367***
[0.225] [0.086] [0.110] [0.237] [0.066] [0.113]
Months in HC 0.010** -0.003* -0.004** 0.005 -0.001 -0.001
[0.004] [0.002] [0.002] [0.004] [0.001] [0.002]
Observations 8640 8640 8640 8640 8640 8640
R-squared 0.22 0.14 0.16 0.14 0.07 0.10
Effect of one year FA 0.09 -0.01 -0.05 0.14 -0.03 -0.05
t-test 2.75 -1.18 -3.00 4.13 -2.82 -3.25
Effect of one year HC 0.12 -0.04 -0.04 0.06 -0.01 -0.01
t-test 2.44 -1.72 -2.00 1.20 -0.46 -0.40
F Test- Same effect of one year
program 0.38 0.96 0.00 2.20 1.52 2.67
Prob 0.54 0.33 0.94 0.14 0.22 0.10
Robust standard errors in brackets (standard errors clustered at household level)
* significant at 10%; ** significant at 5%; *** significant at 1%
¹ Only includes nutritional componet
One year program effectOne year program effect

18. Program effect WHZ
Acute
Malnutrition
Risk of
Acute
Malnutrition EDA IRA
Conditional money pc (mill) ¹ 0.837*** -0.098*** -0.119 -0.216** -0.254*
[0.238] [0.030] [0.081] [0.092] [0.142]
Months in HC -0.003 0.001 0.000 -0.002* -0.004***
[0.004] [0.000] [0.001] [0.001] [0.002]
Observations 8640 8640 8640 9315 9314
R-squared 0.10 0.02 0.05 0.04 0.06
Effect of one year FA 0.12 -0.01 -0.02 -0.03 -0.04
t-test 3.51 -3.24 -1.46 -2.35 -1.79
Effect of one year HC -0.04 0.01 0.00 -0.03 -0.05
t-test -0.94 1.33 0.32 -1.91 -2.66
F Test- Same effect of one year
program 9.24 9.34 1.53 0.01 0.41
Prob 0.00 0.00 0.22 0.91 0.52
Robust standard errors in brackets (standard errors clustered at household level)
* significant at 10%; ** significant at 5%; *** significant at 1%
¹ Only includes nutritional componet

19.  HAZHAZ: one year in FA program for a child with 3: one year in FA program for a child with 3
siblings under 17, increases the z-score by 0.09,siblings under 17, increases the z-score by 0.09,
reduces prevalence of chronic malnutrition (cn)reduces prevalence of chronic malnutrition (cn)
by 1% and the risk cn by 4%. For the case ofby 1% and the risk cn by 4%. For the case of
HC the effect is 0.12, -4% and -5%, respectively.HC the effect is 0.12, -4% and -5%, respectively.
 F-Test equal effect FA and HC: no rejection.F-Test equal effect FA and HC: no rejection.
 WAZWAZ: FA increase z-score by 0.14 and reduce: FA increase z-score by 0.14 and reduce
the probability of global malnutrition by 3%the probability of global malnutrition by 3%
while for HC is 0.06 sd and 1%, respectively.while for HC is 0.06 sd and 1%, respectively.
 EDA:EDA: reduction in about 3% in both programsreduction in about 3% in both programs
 IRA:IRA: reduction in 4% for FA and 5% for HC.reduction in 4% for FA and 5% for HC.

20. Summary and ConclusionsSummary and Conclusions
 Both programs improve the nutritional status and morbidityBoth programs improve the nutritional status and morbidity
outcomes of children under 7 and there is not significantoutcomes of children under 7 and there is not significant
difference in their impact. This result is consistent withdifference in their impact. This result is consistent with
previous studies where FA program improves the quality ofprevious studies where FA program improves the quality of
the food consumed. It also complements studies of thethe food consumed. It also complements studies of the
effectiveness of HC.effectiveness of HC.
 This resultThis result doesn’tdoesn’t imply that the programs are substitutes.imply that the programs are substitutes.
Different groups of the population may prefer differentDifferent groups of the population may prefer different
programs. Our estimates provide someprograms. Our estimates provide some insight about theinsight about the
characteristics that are relevant in the choice of FA vs HC –characteristics that are relevant in the choice of FA vs HC –
for instance, single mothersfor instance, single mothers prefer HC program.prefer HC program.
 Further research on the potential complementarities of bothFurther research on the potential complementarities of both
programs should be carried out.programs should be carried out.