Evaluating the Early Childhood Development (ECD) Program in the Philippines Jere Behrman (U. of Pennsylvania) Paulita Duazo (OPS, U. of San Carlos) Sharon Ghuman (U. of Michigan) Socorro Gultiano (OPS, U. of San Carlos) Elizabeth King (World Bank) The ECD Study Team (Office of Population Studies, University of San Carlos, Cebu) World Bank April 5, 2005
Background Health and nutrition at young ages affect important outcomes in childhood and later life ECD programs increasingly seen as ways to: Reduce malnutrition Reduce infectious disease Provide better pre-school learning environments Few evaluations of ECD programs in developing countries World Bank Research Fund Project with Harold Alderman PI on Uganda, Bolivia and the Philippines Behrman, Jere R., Yingmei Cheng and Petra Todd, 2004, “Evaluating Preschool Programs when Length of Exposure to the Program Varies: A Nonparametric Approach,” Review of Economics and Statistics 86:1,
2002 ECCD Act: Objectives To institutionalize a national program and a system for service delivery for children 0-6 To provide capability building for parents and caregivers To raise public awareness about the importance of early child care and development To mobilize resources and establish viable financing mechanisms for ECCD To ensure survival and promote the total development of children, particularly those who are most vulnerable and disadvantaged (formalized in R.A. 8980/2002)
Reduce by 30% under-5 mortality Decrease by 40% proportion of underweight children Reduce by 30% proportion of children under 6 with anemia Increase to 90% children aged months fully immunized Improve psychosocial & cognitive development of children Increase to 75% children ages 3-5 attending daycare centers Increase to 90% primary school completion rate for Grade 1 entrants Quantitative ECD Goals
Millennium Development Goals East Asia and Pacific Countries By 2015, reduce 1990 child mortality by two-thirds
Millennium Development Goals By 2015, halve 1990 child malnutrition rates and 90% of children immunized for measles
Millennium Development Goals By 2015, Net Primary Enrollment = 100%
The Filipino ECD Program Goal: improve the psycho-social development and nutrition of young children Began with pilot projects in 1999 Three regions Covers 10 provinces, 2.5 million households
ECD Program Mechanisms Support to all provinces for select MCH programs: Immunization Management of Child Illness Micronutrient Malnutrition Control Parent Effectiveness Seminars Grants to pre-selected municipalities to invest in service-provider packages Implementation Support
Service Provider Specific Components Key Providers: Midwife Day Care Worker Child Development Worker Day Care Mother Services Supported: Food and micronutrient supplementation Growth monitoring Child development monitoring and activities Educating parents Primary health care
ECD Evaluation Study Aim: To evaluate Filipino ECD program through collection of longitudinal data in program and non- program regions Baseline: representative sample of newborns and children below age 5 (N = 7,925) Two rounds of follow up data Evaluation analysis based on children who remained in same sample barangay (N = 6,693)
Figure 1. ECD Study Regions Region 6 Region 7 Region 8 Program Areas: Regions 6 and 7 Control Area: Region 8
The Evaluation Study Sample Region 6: 24 municipalities 96 barangays Region 7: 14 municipalities 96 barangays Region 8: 57 municipalities 96 barangays
Schedule of Data Collection ROUND 1: April November 2001 (8 months) Region 6Jun Oct 2001 Region 7Apr Oct 2001 Region 8Aug Nov 2001 ROUND 2: September March 2003 (7 months) Region 6Sep Mar 2003 Region 7Sep Mar 2003 Region 8Sep Feb 2003 ROUND 3: September January 2004 (5 months) Region 6Sep Jan 2004 Region 7Sep Jan 2004 Region 8Sep Dec 2003
Response Rates: Households NUMBER OF HOUSEHOLDSRESPONSE RATE ROUND 1 ROUND2 ROUND3 R1-2 R1-3 REG 6 1,456 1,377 1, REG 7 1,959 1,849 1, REG 8 1,909 1,699 1, TOTAL 5, , N.B. excludes migrants outside sample barangays (if incl., RP=92.8)
NUMBER OF CHILDREN RESPONSE RATE ROUND 1 ROUND2 ROUND3 R1-2 R1-3 REG 6 2,115 1,968 1, REG 7 2,913 2,694 2, REG 8 2,894 2,543 2, TOTAL 7,922 7,205 6, Response Rates: Children N.B. excludes migrants outside sample barangays (if incl., RP=92.1)
Evaluation Methodology Relation of Interest: ∆Y p -∆Y np = a(∆P p -∆P np ) + b(∆X p -∆X np ) + c(∆Z p -∆Z np ) + (∆e p -∆e np ) where ∆Y = Y 3 – Y 1, etc. Y = ECD outcomes P = receive ECD program NP = not receive ECD program X = child, family, community variables Z = unobserved variables e = error term
Participation in child feeding programs Percentage of children 0-4 years who participated in feeding programs Difference Program regions Control region Diff in Diff Impact a significance at p< a
Time After intervention Before intervention For example, participation in infant feeding program Legend: Region 8 Regions 6 & 7 Counterfactual Estimated diff-in- diff impact is
BUT NOT EXPERIMENT WITH RANDOM ASSIGMENT SO DIFFERENCES IN TREATMENT VS CONTROLS: PERCENT DISTRIBUTION OF PROGRAM MUNICIPALITIES BY INCOME CLASS There were 9 program municipalities in Region 7 during Round 1 (1NR).
PERCENT OF HOUSEHOLD HEADS WITH COLLEGE EDUCATION PERCENT OF HOUSEHOLD HEADS WITH COLLEGE EDUCATION N.B. Households present in all three rounds
PERCENT OF HOUSEHOLDS RESIDING IN URBAN AREAS N.B. Households present in all three rounds
PERCENT OF HOUSEHOLDS WITH HOUSES MADE OF STRONG MATERIALS PERCENT OF HOUSEHOLDS WITH HOUSES MADE OF STRONG MATERIALS N.B. Households present in all three rounds
PERCENT OF HOUSEHOLDS WITH REFRIGERATOR N.B. Households present in all three rounds