Is it the food Nutrition Interventions and Children Height in Rural Colombia

1
Is it the food? Nutrition Interventions and
Children Height in Rural Colombia

• Orazio P. Attanasio Marcos Vera-Hernandez
• IFS and UCL

2
The importance of malnutrition

• One third of less than five years old children
  are stunted in growth (Onis et al 2000)
• Poor health and nutrition play a central role in
  underdevelopment (Dasgupta 1993, Fogel 1994)
• Inadequate malnutrition in childhood affects long
  term physical development (Martorell and Habicht
  1986, Baker 1990), as well as the development of
  cognitive skills (Brown and Pollit, 1996 and
  Balazs et al 1986).

3
The importance of malnutrition

• Policy-makers have increasingly promoted early
  childhood nutrition programs as way to raise
  living standards in developing countries (World
  Bank, 1993 Young 1996)
• Improving child nutrition would increase school
  enrollment (Alderman et al 2001) and improve
  learning productivity (Glewwe et al)

4
Variety of interventions

• INCAP (supplementation with nutritious beverage
  in Guatemala) increased child height and
  improvements in nutrition
• Deworming drugs in schools or pre-schools improve
  health and participation rates (Miguel and Kremer
  2004) and Bobonis et al (2003)
• Iron supplementation improves academic outcomes
  (Nokes et al 1998)
• Conditional cash transfers
• Pre-school provision of child care and food

5
Pre-school provision of childcare and food

• They have been implemented in several developing
  countries to improve nutrition and learning
  (Myers, 1995).
• Little is known about their effectiveness though
  the WB and other agencies have devoted increasing
  resources to support such efforts (Behrman, Cheng
  and Todd 2001)

6
An example of Pre-school provision of childcare
and food PIDI in Bolivia

• Behrman, Cheng and Todd 2001
• Assumption of Selection on Observables.
• They find negative effects on nutritional
  outcomes and they claim that their assumption of
  selection on observables is not justified.
• Conditional on participating on the program they
  find more exposure improves weight.
• They find positive effects on test scores
• They do not try to analyze the channels through
  which this program can operate.

7
Another example Hogares Comunitarios in
Guatemala City

• Ruel, Briere, et al 2002
• Assumption of Selection on Observables.
• They find participating children consume better
  and more nutrients, including the food they take
  at home.
• Beneficiaries mothers are more likely to have a
  salaried (and possibly more stable) employment
  than mothers who use other childcare
  arrangements.
• this results in higher wages and a larger number
  of employment benefits.

8
Which programs?

• Early years interventions are perceived to be
  particularly important given the evidence on the
  importance of those years in the process of human
  capital accumulation
• Given the scarcity of resources it is crucial to
  evaluate the effectiveness of each
• This is particularly true in the case of Colombia
  where the new program Familias en Acción is
  widely perceived as an alternative to Hogares
  Comunitarios de Bienestar Familiar
• Except for a descriptive study in 1998 I am not
  aware of any systematic evaluation of Hogares
  Comunitarios

9
Outline

• Hogares comunitarios details of the program
• Evaluation strategy
• Nutrition outcomes a production function
  approach.
• Data
• Results

10
Hogares comunitarios

• Hogares comunitarios is a nutrition/child care
  program introduced all over Colombia in the early
  1980s.
• The program is targeted to poor people.
• At the beginning of the program poor parents were
  encouraged to form parents committees that
  would elect one madre comunitaria
• The madre comunitaria would have to satisfy
  certain conditions (education, large enough
  house) and would received up to 12 children

11
Hogares comunitarios

• Parents would pay a low monthly fee (about 4
  dollars per month) per child which is used to pay
  a small stipend to the mother (and sometimes to
  an assistant)
• Children 0-6 can be sent to the hogar
• Moreover, the parents association receives funds
  to arrange the delivery of food to the hogar
  comunitario
• The food is kept in the madre comunitarias
  refrigerator and is used to feed the children
  lunch and a snack
• The program was adamant about giving parents the
  opportunity of choosing the specific food
• In addition, kids are fed a nutritional
  supplement called bienestarina

12
Hogares comunitarios

• Subsequently existing Hogares receive new
  parents.
• The mother (and therefore the location of the
  Hogar) can change
• Poor parents can send children to an existing
  Hogar

13
Hogares comunitarios

• The program is now very large there are about
  80,000 hogares comunitarios throughout Colombia.
• The program is run by the Instituto Colombiano de
  Bienestar Familiar (ICBF) and is financed with a
  0.3 tax on wages.
• There has never been a systematic evaluation of
  the program, except a descriptive study in the
  mid 1990s
• Some new programs, and in particular the
  Colombian version of PROGRESA, Familias en
  Acción, is widely perceived as an alternative to
  Hogares Comunitarios

14
Evaluation of a nutrition program

• The objective of the paper is to evaluate the
  effect of Hogares comunitarios on several
  outcomes
• Height per age
• Weight per age
• Leg length (stunting)
• long run outcomes education achievements
• The basic strategy is to compare children who
  have been attending Hogares comunitarios to
  children who have not.

15
Evaluation of a nutrition program

• The big problem, of course, is the endogeneity of
  participation
• Parents (mothers?) that can or want to benefit
  more from the program are those that might be
  sending the kids to the hogares
• Parents who are more aware of the benefits of a
  good nutrition might be sending the kids to the
  hogares but would also feed the kids properly
  anyway
• In the absence of a randomized experiment we use
  an instrumental variable approach
• We consider a variable that affects participation
  into the program and yet does not affect the
  nutritional outcome of interest

16
What instrument?

• We use distance of the household from the nearest
  hogar
• We also use the average minimum distance from a
  hogar in the municipality
• We present some evidence on the exogeneity of
  this instrument by looking at a
  pre-intervention variable birth weight
• Movers never indicate getting closer to a
  hogar comunitario as a reason for moving

17
Nutrition outcomes a production function
approach.

• A second objective of the paper is to understand
  the channel through which the program operates.
• For this we take a production function approach
• We consider the program as a possible input along
  with household food consumption.
• Finally we also study the effect that the program
  has on female labour supply.

18
Data

• DNP and FIP commissioned the consortium IFS-
  Econometria SEI the evaluation of the program
  Familias en Acción, a conditional cash transfer
  program.
• Education and nutrition
• Hogares comunitarios is perceived as the existing
  alternative to the nutrition component of
  Familias .
• We collected a baseline pre-programme database in
  122 communities, 57 of which targeted by
  Familias.
• We have a total of 11,500 households or nearly
  80,000 individuals

19
Data

• The data contains information on many variables,
  including current and past attendance to Hogares
  Comunitarios.
• We measure several anthropometric variables
  including weight, height, leg length.
• In this study we use only data from the control
  municipalities, that is municipalities where
  Familias does not exist

20

Main features of the population

• Average family size 7
• Average monthly consumption 150US
• Including consumption in kind
• Food consumption accounts for about 60 of
  consumption
• 85 of households report consumption in kind

21
Figure 1Age structure
800
700
600
500
400
300
200
100
0
0
4
8
12
16
20
24
28
32
36
40
44
48
52
56
60
64
68
72
76
80
84
88
92
96
100
22
Percentage of chronically malnourished children
23
Percentage of children attending Hogares
Comunitarios
24
Why do children not attend Hogares?
25
Distribution of Travel time to the nearest Hogar
in minutes
26
Defining Treatment and Instruments

• Current Attendance
• ExposureA / B.
• ANumber of months the kid has been attending a
  Hogar during her life
• BAge of the kid in months
• Travel_hou Travel distance from house to
  the nearest Hogar
• Travel_mun Average of Travel_hou in the
  municipality

27
First Stage Regressions. Children 1-5
28
First Stage Regressions. Children 1-5
29
First Stage Regressions. Children 1-5
Other included variables Sex, cubic in age,
birth order, head of household education, head of
households age, mothers age, urban/rural,
female headed household. Average travel distance
to health care centre in the municipality.
Presence of hospital. Students per teacher,
Index of municipality financial status and social
infrastructure, wages, altitude, price of rice
and surface, percentage of households with sewage
system, percentage of household with piped water.
30
Overall effect on Height per Age. Results on
Exposure. Children 1-5
31
Overall effect on Height per Age. Results on
Exposure. Children 1-5
32
Overall effect on Height per Age. Results on
Attendance. Children 1-5
33
Other interesting results from the Exposure
analysis
34
Other interesting results from the Exposure
analysis
35
Other included variables
Cubic in age and sex, head of household
education, head of households age, mothers age,
urban/rural . Average travel distance to health
care centre in the municipality. Students per
teacher, Index of municipality financial status
and social infrastructure, Altitude, Surface,
percentage of households with sewage system,
percentage of household with piped water.
36
Can we believe our results?

• Is our instrument correlated with household
  awareness of the importance of nutrition?
• Is our instrument correlated with mother
  nutritional status (resources)?
• We look at Child Birth Weight as it should be
  influenced by mothers resources and nutrition
  awareness but cannot be influenced by child
  attendance to Hogares Comunitarios

37
OLS Regressions of Birth Weight on Travel
Distance and other covariates
38
Can we believe our results?

• Now, we will repeat the IV strategy using Birth
  Weight as an outcome

39
Overall effect on Birth Weight. Results on
Exposure.
40
Overall effect on Birth Weight. Results on
Attendance.
41
Nutrition outcomes a production function
approach.
42
Nutrition outcomes a production function approach

• Program participation (for the problems discussed
  above) and food consumption are likely to be
  correlated with the residual term
• We, once again, take an IV approach and use
  distance from the hogar and average wages in the
  municipality as instruments for program
  participation and food.

43
Nutrition outcomes a production function approach

• Children might benefit not only from attending
  the program but also indirectly if the program
  somehow allows an increase in household
  consumption.
• This increase might come from additional income
  that could be generated by females being able to
  participate into the labour market
• For this reason we also look at the effect of the
  program on mothers labour supply

44
Production function. Children 1-5 Results on
Exposure.
45
Production function. Children 1-5 Results on
Attendance.
10, 5, 1
46
Is there a pattern?

• When we condition on household food consumption,
  it emerges that the contribution of HC to
  children Height is smaller than the overall IV
  effect.
• This is repeated across age groups analyzed all,
  15, 1-4.
• It could be that HC benefits children by some
  other source than food consumed in the HOGAR but
  positively correlated with household food
  consumption

47
A possible explanation

• If mothers labor supply increased because of
  Hogares, food consumption in the household could
  also increase, and the children could benefit
  indirectly
• Before we saw that child care was an important
  component of why not sending a child to a Hogar
• 35 of mothers worked last week if none of her
  children was attending a Hogar, 45 if at least a
  child was attending a Hogar.

48
Highlights of labour supply model for mother
49
Conclusions and future work

• We find that Hogares Comunitarios significantly
  increases height per age
• The effect is reduced when we control for food
  consumption
• Female labour supply increases with the program

50
Conclusions and future work

• Future work
• Effect on school achievement
• Effect on other outcomes
• Interaction effects
• Evaluate the program in other areas?
• Direct comparison with Familias en Acción