Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

14
The Impact of Experimental Nutritional Interventions on Education into Adulthood in Rural Guatemala: Preliminary Longitudinal Analysis Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI Reynaldo Martorell, Emory University Agnes Quisumbing, IFPRI Aryeh D. Stein, Emory University Second Meeting of the Social Policy Monitoring Network Health and Nutrition November 6-7, 2003 Sheraton Rio Hotel & Towers, Rio de Janeiro, Brazil

description

The Impact of Experimental Nutritional Interventions on Education into Adulthood in Rural Guatemala: Preliminary Longitudinal Analysis. Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI Reynaldo Martorell, Emory University Agnes Quisumbing, IFPRI - PowerPoint PPT Presentation

Transcript of Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Page 1: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

The Impact of Experimental Nutritional Interventions on Education into Adulthood in Rural

Guatemala: Preliminary Longitudinal Analysis

Jere R. Behrman, University of PennsylvaniaJohn Hoddinott, IFPRI

John A. Maluccio, IFPRIReynaldo Martorell, Emory University

Agnes Quisumbing, IFPRIAryeh D. Stein, Emory University

Second Meeting of the Social Policy Monitoring NetworkHealth and NutritionNovember 6-7, 2003

Sheraton Rio Hotel & Towers, Rio de Janeiro, Brazil

Page 2: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Introduction

• Education “matters”; it has both intrinsic and instrumental value

• The existing literature on the determinants of educational attainments is unsatisfactory in several respects:– They use a limited set of educational outcomes

– Analyses limited to school-aged populations cannot look at long-term outcomes;

– Analyses of adult attainments have limited data on childhood conditions

– Econometric analyses make strong assumptions to justify use of OLS-type estimators

– Most analyses neglect the cumulative nature of human capital formation, for example the links between determinants of nutritional status and subsequent schooling outcomes

Page 3: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Introduction, cont’d

• This paper presents PRELIMINARY results that attempt to redress these weaknesses by using data:– From a randomized community-level nutrition intervention in rural

Guatemala of children aged 0-15 in 1969-77 COMBINED with

– Data on parental and household characteristics at the time of the intervention AND

– Anthropological/historical data on school quality and “shocks” during and after the intervention AND

– An ONGOING re-survey of these individuals, now aged 25-40, with a variety of education-related outcomes

Page 4: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Methods

Estimate the following reduced form model:

Eia =f(Nai , Ni , Mi , Cf

i , Cvi , Fi , Uia)

Where:

Eia is education outcome

Nai is access to Atole nutritional supplement between ages 6-24 months

Ni is control for cohort effect

Mi is individual’s characteristics (age, sex)

Cfi is community fixed effect

Cvi is time varying community shocks

Fi fixed family background characteristics

Uia is disturbance term

Page 5: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Methods, cont’d

Given their distributions, we use the following estimators when examining the determinants of the following outcomes:

• Ever enrolled in formal schooling (probit)

• Ever passed the first grade of formal schooling (probit)

• Formal schooling completed by age 13 (ordered probit)

• Highest completed grade of (formal and informal) schooling (ordered probit)

• Educational achievement test results (literacy, vocabulary comprehension) in adulthood (ordered probit)

• Raven’s test results in adulthood (OLS)

Page 6: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Probit for being in sample: Selected Coefficient Estimates

San Juan -0.19 (4.4)

Conacaste -0.16 (2.9)

Age in years 0.018 (2.9)

Male -0.058 (4.4)

Cement Factory (age 7) -0.14 (3.5)

Atole exposure (6-24m) -0.031 (0.7)

Page 7: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Figure 1 – Formal grades completed by age 13 (913 observations)

010

020

030

0F

requ

ency

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14Formal grades attained by 13

Page 8: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Probit results (marginal effects) for attending school(t statistics in parentheses)

Ever attended school

Passed first grade

Atole 0.056

(2.6)

0.072

(2.1)

Page 9: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Figure 2 – All grades completed (913 observations)

010

020

030

0F

requ

ency

0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.0010.0011.0012.0013.0014.00High form/inform grade attained

Page 10: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Ordered probit results for grade attainment (t statistics in parentheses)

Grades completed by age 13

Highest grade attained

Atole 0.156(1.3)

0.295(4.6)

Atole x mother schooling

-0.056(1.1)

-0.063(1.7)

Atole x father schooling

-0.017(0.7)

0.028(1.2)

Atole x SES 0.244(4.5)

0.513(4.9)

Page 11: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Figure 3 – SIA test score – sum of vocabulary and comprehension scores (895 observations)

050

100

150

200

Fre

quen

cy

0 20 40 60 80Total sia score (compr+vocab)

Page 12: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Figure 4 – Ravens test score (874 observations)

0

.02

.04

.06

.08

Density

0 10 20 30 40 Total raven score (a+b+c) Kernel density

estimate Normal density

Page 13: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Ordered probits for Inter-American Reading testOLS for Raven’s Progressive Matrices

(t statistics in parentheses)

Inter-American Reading test

Raven’s Progressive Matrices

Atole 0.280

(5.5)

1.489

(1.8)

Atole x mother schooling

-0.107

(2.0)

-0.657

(2.8)

Atole x father schooling

0.078

(2.6)

0.306

(0.8)

Atole x SES 0.229

(2.4)

-0.085

(0.1)

Page 14: Jere R. Behrman, University of Pennsylvania John Hoddinott, IFPRI John A. Maluccio, IFPRI

Conclusions

• There are significantly positive and fairly substantial effects of the Atole supplement, received at age 6-24 months on educational outcomes measured 27-32 years later.

• These effects are larger in wealthier households, but somewhat lower when mothers have more schooling

• Transitory shocks (not reported in this presentation, but discussed in the paper) also affect these outcomes

• These results are PRELIMINARY as data collection is ongoing