Impact of Micronutrient Deficiencies on Pregnancy Outcomes in Developing Countries

Out of Africa: Human Capital Consequences of In Utero Conditions Victor Lavy University of Warwick and Hebrew University Analia Schlosser Tel Aviv University Adi Shany Hebrew University November 2014

Some Salient Facts About Micronutrient Deficiencies During Pregnancy in Developing Countries

Iron Deficiency Half of pregnant women in developing countries are anemic (WHO 2014). More than half of this anemia burden is due to Iron deficiency, the rest partly due to deficiency of folic acid, vitamin B12, vitamin A, and due to parasitic infections. Health consequences include poor pregnancy outcome, in particular impaired physical, brain and cognitive development. If iron supplementation starts after the first trimester of pregnancy it will not help prevent these poor birth outcomes.

Folic Acid Deficiency Recent review by WHO suggest that deficiencies of folic acid is a public health problem that affect many millions of people throughout the developing world. Folic acid deficiency in early pregnancy increases dramatically the chance of a spinal cord problem (Neural Tube Defect) or brain development problems. A NTD is an opening in the spinal cord or brain that occurs very early in gestation. Therefore FEolic acid supplement (folate) is advised for at least the first 12 weeks of pregnancy for all women - even if they are healthy and have a good diet. If Folic acid supplementation starts after the first trimester of pregnancy it will not help prevent these poor birth outcomes.

Iodine Deficiency (IDD) IDD during pregnancy is world s most prevalent cause of brain damage (WHO 2014). Thought to Matter most at time of fetal brain development. It is estimated that 1 billion people are at risk of brain damage from IDD worldwide. Comprehensive Handbook of Iodine (2009): Iodine deficiency is now recognized by WHO as the most common preventable cause of brain damage in the world today, with excess of 2 billion at risk from 130 counties. Iodine deficiency reduces intellectual capacity and cognitive development, with larger effect on girls. Humans require iodine for biosynthesis of thyroid hormone. In utero development of the central nervous system required for intellectual functioning depends critically on adequate supply of thyroid hormone, which influences the density of neural networks established in the developing of the brain (Bror-Axel Lamberg 1991)..

Micronutrient Deficiencies in India The prevalence of micronutrient deficiencies amongst pregnant women of New Delhi slum communities is high (Umesh et al, Indian Pediatrics 1999; 36: 991-998). Prevalence of anemia and Iodine deficiency was 78.8% and 22.9%, respectively. A 2000-2001 survey in a sample of rural villages in Haryana State show that 73.4, 26.3, and 6.4 percent of pregnant women were deficient in iron, folic acid and iodine, respectively (Pathak et al, Indian J Pediatrics. 2004 Nov). Padam Singh (2007): over all anemia among pregnant women in India is 85%.

Rresearch Question What will be the human capital and economic consequences of eliminating these Micronutrient Deficiencies among pregnant women, for example, bringing them to the level in developed countries? More general question, how would improvement of in utero environmental conditions affect later life cognitive and human capital outcomes? The challenge: Identify the casual effect of in utero conditions where children s family background correlate with in utero conditions.

Literature Medical The fetal origin hypothesis (Braker 1992): certain chronic conditions later in life can be traced to the course of fetal development. Evidence on the effect of malnutrition during pregnancy: Dutch famine - Neugebauer et al. (1999), Rooij et al. (2010) - severe maternal nutritional deficiency early in gestation is associated with inferior brain and cognitive development of off- springs. Evidence on effect of radiation exposure: Relatively short pulse of exposure to radioactive fallout between 8 and 25 weeks of gestation has long term impact on cognitive abilities of off-springs later in life. This is plausible given the nature of the developmental events occurring in the brain during this period of gestation (Nowakowski and Hayes (2008)).

Literature Economic Studies that use changes in local environment caused by negative in utero shocks and look at effect on health and education outcomes (review by Almond and Currie, 2011) Radiation shock (Almond et al. 2009 Chernobyl), Nutrition shocks (Almond et al. 2007 china famine), Economic shocks (Banerjee et al. 2010 19thcentury blight in French vineyards, Baten et al. food prices in Britain 18thcentury). Variation in infectious disease (Almond 2006 Influenza, Barreca 2009 malaria) Positive and policy driven events in early childhood: Food stamps program in USA that increasing family resources ((Hoynes et al. 2012) leads to reduction in incidence of obesity, high blood pressure, heart disease, diabetes, as well as an increase in reporting to be in good health. Migration to Sweden (Van Den Berg et al. 2012) at early childhood leads to improved outcomes at adulthood.

Literature Economic Field et al "Iodine Deficiency and Schooling Attainment in Tanzania." AEJ Applied (2009). Our findings suggest a large effect of in utero iodine on cognition and human capital: treated children attain an estimated 0.35-0.56 years of additional schooling relative to siblings and older and younger peers. Furthermore, the effect appears to be substantially larger for girls, consistent with laboratory evidence indicating greater cognitive sensitivity of female fetuses to maternal thyroid deprivation.

In This Research Project We use the immigration from a poor African country (Ethiopia) to a developed economy (Israel) in May 1991: Quasi-experimental variation Immigration was unexpected, in a short time and not self selected Random variation in the timing of pregnancy and birth Different environmental conditions during pregnancy Pregnant women experienced different environmental conditions because of immigration, especially in terms of micronutrient deficiencies. Natural Regression discontinuity Design Comparing cognitive and high school achievements of children who faced dramatic differences in prenatal conditions in utero based on their gestational age upon arrival to Israel in May 1991.

Unique in this Study Using unique identification based on a positive event of environmental differences caused by immigration. Investigating the effects of environmental conditions in utero in different stages of pregnancy on cognitive outcomes. Policy implications for developed and developing countries. Understanding of intergenerational effects of immigration

The immigration from Ethiopia to Israel In 1975 the Ethiopian Jews were recognized as Jewish by the state of Israel and allowed to immigrate to Israel under the Law of Return.

The immigration from Ethiopia to Israel Operation Moses - 6,000 immigrant airlifted from Sudan secretly between end of 1984 and 1985.

The immigration from Ethiopia to Israel "Operation Solomon" - On May 24th 1991 the Israeli government brought to Israel more than 14,000 Ethiopians Jews within 36 hours.

Operation Solomon May 24th1991

Why is it Natural Experiment? Unexpected Organized by the Israeli government as a rescue operation following unstable political situation in Ethiopia Completed in short time Airlift by Israeli aircrafts within 36 hours Not self selected Included almost all the Jews living in Ethiopia The quasi-experimental variation properties: The timing of immigration uncorrelated with the timing of the pregnancy and birth The decision to immigrate uncorrelated with unobserved factors like family background

Analysis Sample Students born in Israel between June 1991 and February 1992 to mothers who immigrated to Israel in "Operation Solomon : Their pregnancy incepted in Ethiopia but they were born in Israel Experienced the same conditions at birth and at later life Faced dramatic differences in prenatal conditions in utero based on their gestational age upon arrival to Israel in May 1991

Analysis Sample The birth distribution of all the "Operation Solomon" offspring:

Large environmental differences between Ethiopia and Israel that may have affected pregnant mothers Micro Nutrient Supplements: Ethiopia: No vitamins consumption (less then 1% of pregnant women age 15-49 took iron supplements for 90+ days DHS 2011). Israel: use of vitamins during pregnancy, mainly iron and Folid Acid (Granot et al. (1996)). Iodine: Ethiopia remains severely iodine deficient. Children in Northern Ethiopia are Iodine Deficient (Girma et al. 2014). Iodine level in Israel is adequate. Health Care and Pregnancy Monitoring: Ethiopia: Little or no medical care during pregnancy. In Israel: modern health care and medicine, regular monitoring. Leaving Conditions and Nutrition Income per capita in Israel 60 times higher, nutritional level and quality in Israel better than in Ethiopia. Probably not much difference in short term.

Environmental Differences Indicator Israel Ethiopia Source Israel (CBS 1993) Ethiopia (IFPRI 1993) 3,089 1,516 Calorie Supply Per Capita The world bank (1990/1) 10 120 Mortality rate, infant (per 1,000 live births) The world bank (1990/1) Mortality rate, under-5 (per 1,000 live births) 12 202 The world bank (2000/1) 8 15 Low birth-weight babies (% of births) The world bank (2000) 100 27 Pregnant women receiving prenatal care (%)

Data Israeli Ministry of Education Database 2007-2011: Students administrative records Birth date, Own and parental country of origin, Immigration date. Family characteristics parents schooling, number of siblings. High School data Completion of 12thgrade. Obtaining matriculation certificate by age 18. matriculation certificate is required for admission for academic post secondary schooling and for some jobs. Credit units of matriculation exams. Measure the quality of the matriculation program. Units in English and math, measures of inetability

Background Characteristics by Treatmetment First Trimester Second Third Trimester Trimester Mother's age at birth 30.76 (9.281) 30.91 (8.516) 30.59 (8.788) Parents's age gap 10.49 (7.143) 11.29 (7.749) 11.11 (6.987) Birth order 2.789 (1.966) 2.703 (1.883) 2.866 (1.895) Father's years of schooling 2.183 (3.675) 2.415 (4.079) 1.838 (3.552) Mother's years of schooling 1.972 (3.539) 2.415 (3.906) 2.317 (3.823)

Definition of Treatment The key variable: The gestational age of the student at immigration (measured by weeks) Treatment Definition by Trimester First trimester constitute the most critical time for negative effects of deficiencies in micro nutrients and iodine. Evidence from effect of famine on cognitive abilities later in life (Von Hinke Kessler Scholder et al. (2014), Neugebauer et al. (1999), Rooij et al. (2010)). Treatment Definition by the critical period for fetus brain Development: Week 8 to week 25 is viewed as the critical period of pregnancy for fetus brain and cognitive development (Nowakowski and Hayes (2008), Loganovskaja and Loganovsky (1999). E.X. damage caused by radiation to human fetus in this period can result in cognitive deficits that still manifest 16-18 years after birth it is the major neuron genetic period of the developing human neocortex. 1. 2.

Balancing Tests Testing the correlation between observable characteristics and timing of pregnancy based treatment definition by Trimesters: Number of sibiling Indictor for twins SES of first locality of Treatment Period Mother age at birth Parents age gap Birth order Mother schooling Father schooling trimester_1 0.165 (0.776) -0.628 (0.690) 0.000 (0.246) -0.079 (0.197) (0.033) 0.006 -0.454 (0.384) 0.239 (0.325) -0.020 (0.055) trimester_2 0.323 (0.826) 0.192 (0.674) -0.016 (0.205) -0.162 (0.206) (0.027) -0.017 -0.048 (0.381) 0.438 (0.295) 0.020 (0.051) Observations 624 624 624 624 624 624 624 572

Mean Outcomes by Treatment First Trimester Second Trimester Third Trimester Birth weight 3076.0 (491.2) 3087.4 (487.2) 3021.4 (468.8) Did not repeat 6th-12th grade 0.829 (0.377) 0.809 (0.394) 0.754 (0.432) Matriculation diploma 0.346 (0.477) 0.199 (0.400) 0.197 (0.399) Matriculation units 12.91 (11.35) 1.476 (1.538) 12.26 (11.38) 1.347 (1.549) 8.972 (10.15) 0.915 (1.355) Math units English units 2.224 (1.900) 2.025 (1.899) 1.620 (1.817) Math 5 units 0.0163 (0.127) 0.0127 (0.112) 0 (0)

The Baseline Model Treatment definition by trimesters: (1) ??= ?0+ ?1?????_??????????+?2??????_?????????+???+ ?? ??- Student ? school outcome ??- Vector of student ? characteristic The omitted category is Third trimester Treatment definition by weeks: (1) ??= ?0+ ?1???? (1 7)?+?2????(8 24)?+???+ ?? We expect that ?1> ?2> 0 Shortcomings: Not include cohort and month of birth effects. Estimates may be confounded by unobserved cohort effects or substantial seasonality in school performance by month of birth.

Controlling for Cohort and Month of Birth Effects (1) Adding a respective older cohort: Students born in Ethiopia between June 1990 and February 1991 and immigrated to Israel on "Operation Solomon": Post Treatment Primary sample June 1991 to February 1992 cohort Before Treatment Pre cohort June 1990 to February 1991 cohort (2) Adding two comparison groups for the two cohorts: Second-generation immigrants from "Operation Moses (parents immigrated before 1989) Conception and birth In Israel Control for cohort effects and seasonality in the timing of birth in Israel. Group A Ethiopian born students who immigrated between 1992 and 2000 Conception and birth In Ethiopia Control for cohort effects and seasonality in the timing of conception in Ethiopia. Group B

Controlling for Cohort and Month of Birth Effects Treatment definition by weeks: (2) ??= ?0+ ?1???? (1 7)?+?2????(8 24)?+???+ ????? + ?1????? ????+ ?2??? ????? ????+ ?3??????? + ?4???????+ ?? ???? - Month of birth fixed effects ????? ????- Students born between June 1990 and February 1991 ????- Students born in Ethiopia ????? ??- Students born in Israel to Operation Moses immigrants ????? ??- Students born Ethiopia and immigrated after 1991

Effect on Schooling Attainment

Effect on Repetition and Matriculation OLS Baseline Did not repeat 6th-12th Matriculation Treatment grade diploma No With No With Controls Controls Controls Controls Trimester_1 0.078** (0.037) 0.068 (0.041) 0.151*** (0.052) 0.137** (0.057) Trimester_2 0.054 (0.038) 0.051 (0.039) -0.000 (0.049) -0.005 (0.053)

Effect on Repetition and Matriculation Two Years Cohorts Sample Did not repeat 6th-12th A t t Matriculation Treatment grade diploma W i t No With Controls No With Controls Controls Controls Trimester_1 0.411*** 0.073*** 0.072*** (0.022) 0.097** (0.045) 0.094* (0.047) (0.108) (0.020) Trimester_2 0.135 0.058 (0.094) (0.035) 0.054 (0.035) 0.026 (0.032) 0.023 (0.030) Observations 2473 2473 2473 2473 2473

Effect on Quality of the Matriculation Program OLS Baseline Matriculation units Math units English units Math 5 units Treatment With No No With No With No With Controls Controls Controls Controls Controls Controls Controls Controls Trimester_1 3.387*** (1.034) 3.129*** (1.065) 0.444*** (0.152) 0.426** (0.163) 0.555*** (0.201) 0.472** (0.199) 0.016** 0.013** (0.007) (0.006) Trimester_2 2.613** (1.077) 2.677** (1.088) 0.249 (0.156) 0.243 (0.163) 0.420* (0.210) 0.401* (0.203) 0.009 (0.006) 0.006 (0.006) Observations 624 624 624 624 624 624 624 624

Effect on Quality of the Matriculation Program Two Years Cohorts Sample Matriculation units Math units English units Math 5 units Treatment With No No With No With No With Controls Controls Controls Controls Controls Controls Controls Controls Trimester_1 3.154*** (1.008) 3.096*** (0.950) 0.468*** 0.463*** (0.118) 0.455* (0.249) 0.436* (0.250) 0.021** 0.020** (0.009) (0.116) (0.009) Trimester_2 2.177** (0.861) 2.161** (0.759) 0.194** (0.092) 0.192** (0.085) 0.328 (0.203) 0.311 (0.194) 0.015* (0.008) 0.015* (0.008) Observations 2473 2473 2473 2473 2473 2473 2473 2473

Effect Size Effect on Matriculation diploma is about twice the effect of attending a high quality primary school versus a low quality primary school Gould, Lavy and Paserman, QJE May 2004: Immigrating tp Opportunity: Estimating the Effect of School quality Using A Natural Experiment on Ethiopians In Israel

Additional Results Heterogeneity Effect by Gender: Large and significant effect for girls. Small and insignificant effect for boys. Heterogeneity Effect by Family Education: Much larger and significant effect among the "high education" group. Negligible effect for the low education group. Robustness Checks: Results are similar when dropping from the sample students who born in the last two weeks of February. Placebo test: Estimating equation (2) for the comparison groups only.

Effect By Gender

Effect on Repetition and Matriculation Two Years Cohorts Sample, GIRLS Did not repeat 6th-12th A t t Matriculation Treatment grade diploma W i t No With Controls No With Controls Controls Controls Trimester_1 0.411*** 0.137*** 0.126*** (0.038) 0.164* (0.081) 0.157* (0.079) (0.108) (0.037) Trimester_2 0.1350.098** (0.094) (0.045) 0.089* (0.047) 0.093 (0.090) 0.089 (0.091) Observations 2473 1201 1201 1201 1201

Effect on Repetition and Matriculation Two Years Cohorts Sample, BOYS Did not repeat 6th-12th Matriculation Treatment grade diploma No With Controls No With Controls Controls Controls Trimester_1 0.011 (0.043) 0.013 (0.044) 0.064 (0.052) 0.063 (0.055) Trimester_2 0.015 (0.035) 0.012 (0.036) 0.037 (0.041) 0.035 (0.041) Observations 1272 1272 1272 1272

Effect on Quality of the Matriculation Program Two Years Cohorts Sample, GIRLS Matriculation units Math units English units Math 5 units Treatment With No No With No With No With Controls Controls Controls Controls Controls Controls Controls Controls Trimester_1 4.366*** (1.441) 4.112*** (1.407) 0.590*** 0.585*** (0.163) 0.661* (0.333) 0.626* (0.346) 0.007 (0.011) 0.008 (0.011) (0.163) Trimester_2 2.899* (1.498) 2.699* (1.440) 0.118 (0.179) 0.111 (0.170) 0.516 (0.312) 0.478 (0.338) 0.016 (0.011) 0.016 (0.011) Observations 1201 1201 1201 1201 1201 1201 1201 1201

Effect on Quality of the Matriculation Program Two Years Cohorts Sample, BOYS Matriculation units Math units English units Math 5 units Treatment With No No With No With No With Controls Controls Controls Controls Controls Controls Controls Controls Trimester_1 2.191* (1.086) 2.088* (1.078) 0.370* (0.206) 0.359 (0.208) 0.307 (0.260) 0.259 (0.245) 0.036*** 0.033** (0.012) (0.012) Trimester_2 1.599* (0.782) 1.581** (0.680) 0.271* (0.152) 0.253 (0.149) 0.172 (0.234) 0.141 (0.207) 0.014 (0.009) 0.013 (0.009) Observations 1272 1272 1272 1272 1272 1272 1272 1272

Why larger effect on girls? Evidence of larger impact for girls is consistent with related literature. 1. Baird et al (2011) find that in DC girls infant mortality is significantly more sensitive to aggregate economic shocks during pregnancy. 2. Field et al. (2009) find that delays in resupply of iodine for pregnant women in Tanzania has larger educational improvements for girls. 3. Oreopoulos et al. (2008) show that effects of infant health on reaching grade 12 by age 17 appear to be stronger for females than males. 4. Hoynes et al. (2012) find that increasing family resources during early childhood improve health at adulthood for both men and women but have positive significant effect on economic self-sufficiency only for women. 5. Gould et al. (2011) find that early childhood living conditions affected only girls short and long term outcomes (until age sixty).

Why larger effect on girls? Other studies, suggest that from conception and during pregnancy, males are more vulnerable than females. 1. Ravelli et al. (1999) show that in the Dutch famine during the Second World War the number of boys born fell in relation to the number of girls. 2. Eriksson et al. (2010) suggests that in the womb boys have a more dangerous growth strategy than girls. They grow more rapidly and invest less in placental growth, which puts them at greater risk of becoming undernourished. 3. Kraemer (2000) discusses how the human male is, on most measures, more vulnerable than the female, a vulnerability that is attributed in part to the biological fragility of the male fetus and argues that girls are more likely to survive adverse in utero health conditions.

Why larger effect on girls? If there are more spontaneous abortions of boys in utero during adverse conditions, as suggested in the literature, one possible explanation for our findings could then be that better environmental conditions early in utero enabled marginal boys to be born. This could explain the absence of an effect of early exposure to better environmental conditions among boys since we compare between relatively stronger boys born to mothers who immigrated during weeks 25+ and marginal boys born to mothers who immigrated during weeks 1-7.

Effect by Parental Education

Effect on Repetition and Matriculation Two Years Cohorts Sample, By Education Did not repeat 6th-12th grade Matriculation diploma Treatment No Controls With Controls Low Education No Controls With Controls Trimester_1 0.102*** 0.097*** 0.034 0.029 (0.026) (0.026) (0.028) (0.031) Trimester_2 0.088** 0.081** -0.019 -0.026 (0.031) (0.035) (0.041) (0.041) Observations 1489 1489 High Education 1489 1489 Trimester_1 0.008 0.017 0.220** 0.203* (0.066) (0.069) (0.099) (0.098) Trimester_2 -0.020 -0.020 0.112* 0.098 (0.075) (0.077) (0.059) -0.058 Observations 984 984 984 984

Effect on Quality of the Matriculation Program Two Years Cohorts Sample, By Education Matriculation units Math units English units Math 5 units Treatment With Controls No Controls W i No Controls With Controls No Controls With Controls No Controls With Controls Low Eduction Trimester_1 2.815*** (0.954) 2.823*** 3.096***0.457*** (0.863) (0.950) (0.138) 0.448*** (0.132) 0.391* (0.205) 0.400** (0.189) 0.036** (0.014) 0.036** (0.014) Trimester_2 2.019 (1.337) 2.035 2.161** 0.240 (1.311) (0.759) (0.204) 0.234 (0.200) 0.140 (0.226) 0.155 (0.210) 0.025** (0.012) 0.025** (0.012) Observations 1489 1489 2473 1489 1489 1489 1489 1489 1489 High Education Trimester_1 5.115* (2.676) 4.495 3.096***0.817*** (2.597) (0.950) (0.273) 0.762** (0.275) 0.729 (0.580) 0.661 (0.590) 0.003 (0.017) 0.002 (0.015) Trimester_2 4.109* (2.329) 3.706* 2.161** 0.633** (2.066) (0.759) (0.254) 0.595** (0.234) 0.612 (0.527) 0.555 (0.499) 0.014 (0.011) 0.012 (0.011) Observations 984 984 984 984 984 984 984 984

Why larger effect in educated families? These results are consistent with previews findings in the literature. Studies found that the negative impact of poor fetal health [Currie and Hyson (1999)] or exposure to negative shocks in utero [Almond, Edlund and Palme (2009)] on human capital accumulation is greater in low- education or low-income families. The explanation given for this finding is that higher educated families tend to compensate for negative shocks to fetal health or to poor birth health outcomes. Moreover, negative shocks have usually smaller effects on children in high income families because they are less vulnerable. In our analysis we evaluate the effect of positive shock so obtain the opposite result. The explanation can be that mothers with some formal education are able to take more advantage of a positive environmental shock by accessing more better medical technologies and nutrition.

Effect on Birth Weight

Effect on Birth Weight Only "Operation Solomon" "Operation Solomon" & "Operation Moses" Treatment No Controls With Controls No Controls With Controls 55.097 (64.204) 45.474 (62.415) 23.378 (70.271) 22.342 (65.017) trimester_1 68.656 (61.729) 62.457 (58.122) 1.283 (76.401) -5.066 (70.150) trimester_2 47.545 (50.822) 47.954 (54.760) Operation Moses 612 612 1082 1082 Observations