Nutrition and Economic Development EXPLORING EGYPT’S EXCEPTIONALISM AND THE ROLE OF FOOD SUBSIDIES Olivier Ecker, Perrihan Al-Riffai, Clemens Breisinger, Rawia El-Batrawy About IFPRI The International Food Policy Research Institute (IFPRI), established in 1975, provides research-based policy solutions to sustainably reduce poverty and end hunger and malnutrition. The Institute conducts research, communi- cates results, optimizes partnerships, and builds capacity to ensure sustainable food production, promote healthy food systems, improve markets and trade, transform agriculture, build resilience, and strengthen institutions and gover- nance. Gender is considered in all of the Institute’s work. IFPRI collaborates with partners around the world, including development implementers, public institutions, the private sector, and farmers’ organizations. About IFPRI’s Peer Review Process IFPRI books are policy-relevant publications based on original and innova- tive research conducted at IFPRI. All manuscripts submitted for publica- tion as IFPRI books undergo an extensive review procedure that is managed by IFPRI’s Publications Review Committee (PRC). Upon submission to the PRC, the manuscript is reviewed by a PRC member. Once the manuscript is considered ready for external review, the PRC submits it to at least two external reviewers who are chosen for their familiarity with the subject mat- ter and the country setting. Upon receipt of these blind external peer reviews, the PRC provides the author with an editorial decision and, when necessary, instructions for revision based on the external reviews. The PRC reassesses the revised manuscript and makes a recommendation regarding publication to the director general of IFPRI. With the director general’s approval, the manuscript enters the editorial and production phase to become an IFPRI book. Nutrition and Economic Development Exploring Egypt’s Exceptionalism and the Role of Food Subsidies Olivier Ecker, Perrihan Al-Riffai, Clemens Breisinger, and Rawia El-Batrawy Copyright © 2016 International Food Policy Research Institute. All rights reserved. Contact ifpri-copyright@cgiar.org for permission to reproduce. Any opinions stated herein are those of the author(s) and are not necessarily representative of or endorsed by the International Food Policy Research Institute. The boundaries and names shown and the designations used in the map or elsewhere in this publication do not imply official endorsement or acceptance by the International Food Policy Research Institute (IFPRI). International Food Policy Research Institute Library of Congress 2033 K Street, NW Cataloging in Publication Program Washington, DC 20006-1002, USA 101 Independence Avenue, S.E. Telephone: 202-862-5600 Washington, DC 20540-4283 DOI: 10.2499/9780896292383 Library of Congress Cataloging-in-Publication Data Names: Ecker, Olivier, author. | Al-Riffai, Perrihan, author. | Breisinger, Clemens, author. | El-Batrawy, Rawia, author. Title: Nutrition and economic development : exploring Egypt’s exceptionalism and the role of food subsidies / by Olivier Ecker, Perrihan Al-Riffai, Clemens Breisinger, and Rawia El-Batrawy. Description: Washington, DC : International Food Policy Research Institute, [2016] | Includes bibliographical references and index. Identifiers: LCCN 2016044755 (print) | LCCN 2016046490 (ebook) | ISBN 9780896292383 (pbk. : alk. paper) | ISBN 9780896292437 (e-book) Subjects: LCSH: Food industry and trade—Subsidies—Egypt. | Nutrition policy—Egypt. | Egypt—Economic policy. Classification: LCC HD9017.E32 E25 2016 (print) | LCC HD9017.E32 (ebook) | DDC 338.1/962—dc23 LC record available at https://lccn.loc.gov/2016044755 Cover Design: James Sample, IFPRI Project Manager: John Whitehead, IFPRI Book Layout: BookMatters v Executive Summary Egypt faces two major nutritional challenges of public health concern with critical implications for development and economic prosperity. These are the double burden of malnutrition and the growth-nutrition disconnect. This study documents and explores the two nutritional challenges and high- lights the importance of addressing them urgently and decisively. According to 2011 estimates, 31 percent of Egyptian children ages 6– 59 months are stunted—a prevalence rate that is usually seen only in develop- ing countries with much lower national income levels than Egypt’s. Egypt also ranks among the countries with the highest rates of female overweight and obesity in the world, with 73 percent of all (nonpregnant) women 20 years of age and older being overweight and 34 percent being obese. The prevalence of overweight among children (29 percent) is almost as high as the prevalence of child stunting. The double burden of malnutrition— that is, the coexistence of under- and overnutrition— occurs not only at the national level but also at the family level and even the individual level: 22 percent of the children who are stunted have a mother who is overweight, and 14 percent of children suf- fer from both stunting and overweight at the same time. Surprisingly, chronic child undernutrition, child and maternal overnutrition, and the double bur- den of malnutrition are prevalent at similar rates among the poorest and the richest income quintiles of the population as well as in urban and rural areas. Despite high economic growth, the prevalence rate of child stunting increased significantly and steadily throughout the 2000s— an atypical trend for a coun- try outside wartime. This study identifies four potential key drivers of Egypt’s two— probably interlinked— nutritional challenges, which, in combination, may have led to the country’s exceptionalism. These are the nutrition transition, economic crises and rising poverty, the food subsidy system, and insufficient nutrition-sensitive invest- ment. Although the high rates of female overnutrition may be partly attribut- able to the nutrition transition, and part of the increase in child stunting may be due to the succession of recent economic crises, these factors are insufficient to explain Egypt’s exceptionalism and observed patterns of malnutrition among the Egyptian population. The main hypothesis of this study is that Egypt’s large food subsidy system has been ineffective in reducing under nutrition; in fact, it may have contributed to sustaining and even aggravating both nutrition challenges. At least until the reform in 2014– 2015, both the Baladi bread and flour program and the ration-card program incentivized the consumption of calorie-overladen and unbalanced diets as a result of providing calorie-rich foods at very low and con- stant prices and with quotas much above dietary recommendations. In addition to these hypothesized direct nutritional effects, the food subsidy system may have another adverse, indirect effect on nutrition: the public budget allocated to food subsidies is unavailable for possibly more nutrition-beneficial spending, such as for child and maternal nutrition-specific interventions. To provide supportive evidence for the main hypothesis, the econometric analysis uses propensity score matching approaches for binary and continuous treatment variables and data from the 2010– 2011 Egypt Household Income, Expenditure, and Consumption Survey (HIECS) to investigate the direct effects of food subsidies on child and maternal nutrition, the double burden of malnutrition at the individual and the family level, and household diet qual- ity. The estimation results suggest that that the ration-card program (in place until May 2014) considerably affects under- and overnutrition mainly in urban areas. The Baladi bread and flour program has notable effects on overnutrition in both urban and rural areas. The nutritional effect of the ration-card program is generally larger than that of the Baladi bread and flour program. Consistent with the main hypothesis, there is no statistically significant indication that higher food subsidies lead to improved nutritional outcomes. On the con- trary, higher food subsidies increase the risk of malnutrition among both chil- dren and their mothers, particularly related to overnutrition. In urban areas, the probability of child overweight and the probability of maternal overweight increase with the subsidy levels that the families acquire from the ration- card program. And maternal overnutrition is more common among beneficiary families of the ration-card program than among nonbeneficiary families. Urban mothers’ risk of overweight also tends to increase with increasing Baladi bread and flour subsidies. Moreover, the estimated dose-response functions show a tendency toward increasing probabilities of child stunting and the double vi ExEcutivE Summary burden of child malnutrition with increasing ration-card-program subsidy lev- els in urban areas. The (binary) propensity score matching (PSM) estimations indicate that, on average, the risk of child stunting is still lower among urban beneficiary households of the ration-card program than among urban nonben- eficiary households. Thus, the ration-card program has no uniformly negative effect on child nutrition; it is rather the received subsidy level that matters for child malnutrition. However, regarding maternal overnutrition, the program has only adverse effects. The estimation results for household diet quality indi- cators confirm that the ration-card program seems to indeed adversely affect nutrition in urban areas through incentivizing diets that are unbalanced, espe- cially regarding the frequency of consuming micronutrient-rich foods. In conclusion, the findings of this book consistently suggest that— in addi- tion to the well-known economic rationale for reforming the Egyptian food subsidy system— there are strong reasons to reform food subsidies due to nutri- tion and public health concerns. From a nutritional perspective, primarily the ration-card program that was in place until May 2014 incentivized overcon- sumption of calorie-rich and unbalanced diets, especially among urban benefi- ciaries. In June 2014, Egypt’s new government began to fundamentally reform the food subsidy system with the aims of alleviating its considerable and rapidly growing fiscal burden on the country’s budget and of increasing its effectiveness as a social protection instrument. The recent changes mark important steps toward a voucher-based system and provide the basis for implementing a more targeted approach. Although nutrition concerns may have played no (decisive) role in the reform debate, the already implemented modifications, and espe- cially the changes to the ration-card program, can be expected to have positive dietary effects. They tend to reduce— but not fully remove— the incentives for overconsuming calorie-rich and micronutrient-poor diets. Given Egypt’s persistent and exceptional nutritional challenges, future food subsidy reform steps should consider nutritional implications from the onset. In fact, there may be scope to transform the current subsidy system into a key policy instrument in the fight against malnutrition. Follow-up impact evalu- ation studies can provide important evidence on income, dietary, and nutri- tional implications of different reform alternatives and thereby help to create a more nutrition-sensitive social safety net in Egypt. Keywords: propensity score, dose-response model, double burden of mal- nutrition, growth-nutrition disconnect, child stunting, overweight, food sub- sidy, food assistance, Baladi bread, ration card, Egypt ExEcutivE Summary vii Contents Executive Summary v Figures, Tables, and Boxes xi Foreword xix Acknowledgments xxi Chapter 1 Introduction 1 Chapter 2 Nutritional Challenges for Economic Development 9 The Double Burden of Malnutrition 10 The Growth-Nutrition Disconnect 24 Chapter 3 Drivers of Egypt’s Exceptionalism in Nutrition 31 Nutrition Transition and Implications for Malnutrition 31 Economic Crises and Poverty in Egypt 38 Food Subsidies and Nutritional Implications 46 Nutrition-Beneficial Investments in Egypt 78 Chapter 4 Analyzing the Nutritional Effects of the Egyptian Food Subsidy System 89 Methodology and Data 90 Estimation Results 122 Chapter 5 Conclusions 147 Summary of Main Findings 149 Policy Implications 153 Limitations and Research Implications 158 References 161 Appendix 183 Authors 253 Index 255 Figures, Tables, and Boxes Figures 2.1 Relationship between the prevalence of female overweight and child stunting 16 2.2 Relationship between the prevalence of female obesity and child stunting 17 2.3 Relationship between children’s BMIZs and HAZs in Egypt 20 2.4 Relationship between women’s BMIs and body heights in Egypt 21 2.5 Relationship between the prevalence of chronic child undernutrition and national income 25 2.6 Prevalence of child stunting in Egypt, by different subnational disaggregations, 2000–2011 29 3.1 Trends in food and calorie availability in Egypt, 1970– 2011 37 3.2 Costs of the Egyptian food subsidy system 49 3.3 Costs of the bread and flour subsidy program and the ration card program 53 3.4 Engel curves for total food consumption, consumption of cereals, and consumption of Baladi bread and flour in Egyptian families 71 3.5 Engel curves for total food consumption; consumption of rice, sugar, and vegetable oils; and consumption of subsidized rice, sugar, and cooking oil in Egyptian families 72 3.6 Engel curves for consumption of cereals, rice, and subsidized rice in Egyptian families 73 3.7 Engel curves for consumption of sugars, sugar, and subsidized sugar in Egyptian families 74 3.8 Engel curves for consumption of edible fats and oils, vegetable oils, and subsidized cooking oil in Egyptian families 75 3.9 Engel curves for consumption of vegetables and legumes in Egyptian families 76 3.10 Engel curves for consumption of meat and fish and milk and dairy products in Egyptian families 77 4.1 Dose- response functions for child HAZ 128 4.2 Dose- response functions for child stunting 129 4.3 Dose- response functions for child BMIZ 130 4.4 Dose- response functions for child overweight 131 4.5 Dose- response functions for mother’s BMI 133 4.6 Dose- response functions for maternal overweight 134 4.7 Dose- response functions for child stunting and overweight 135 4.8 Dose- response functions for child stunting and maternal overweight 136 4.9 Dose- response functions for child overweight and maternal overweight 137 4.10 Dose- response functions for HDDS 139 4.11 Dose- response functions for frequency of household vegetable consumption 140 4.12 Dose- response functions for frequency of household legume consumption 141 4.13 Dose- response functions for frequency of household meat and fish consumption 142 4.14 Dose- response functions for frequency of household milk and dairy products consumption 143 Tables 2.1 Prevalence patterns of the double burden of malnutrition in Egypt 18 2.2 Country comparison of child stunting reduction and economic growth 26 xii tablES 3.1 Per capita food and macronutrient availability in Egypt, and changes over time 36 3.2 Key macroeconomic indicators for Egypt, 2000–2011 40 3.3 Poverty and income inequality and annual average change in poverty, 2000–2011 43 3.4 Allocation of food ration cards by income quintile and poverty status in Egypt 59 3.5 Monthly quotas and prices of foods subsidized under the ration card program in Egypt 61 3.6 Per capita income and food consumption in urban and rural Egyptian families 69 3.7 Proportion of households in Egypt with improved drinking- water sources and sanitation facilities 80 3.8 Proportion of households by method of waste/trash disposal in Egypt 81 3.9 Prevalence of common infant-feeding practices in Egypt 85 4.1 Overview of estimation model specifications with type of outcome and treatment variables 113 4.2 Overview of PSM estimation results: Direction of causal effects 122 4.3 Estimated ATT of ration-card-program participation on child nutrition 125 4.4 Estimated ATT of ration-card-program participation on maternal nutrition 125 4.5 Estimated ATT of ration-card-program participation on the double burden of malnutrition and the coexistence of child and maternal overnutrition at the family level 126 4.6 Estimated ATT of ration-card-program participation on household diet quality 126 A.1 Prevalence of female overweight and annual average change in Egypt 184 A.2 Prevalence of female obesity and annual average change in Egypt 186 A.3 Prevalence of child stunting and annual average change in Egypt 188 tablES xiii A.4 Prevalence of child wasting and annual average change in Egypt 190 A.5 Prevalence of child underweight and annual average change in Egypt 192 A.6 Prevalence of child overweight and annual average change in Egypt 194 A.7 Prevalence of stunted children with overweight mothers and annual average change in Egypt 196 A.8 Prevalence of stunted and overweight children and annual average change in Egypt 198 A.9 Subsidized commodities, unit quantities, and unit prices under the ration-card program, as of January 2015 200 A.10 Prevalence of anemia among nonpregnant women and children and average annual change in Egypt 202 A.11 Descriptive statistics of estimation variables 204 A.12 Logistic regression results of the binary propensity score estimations for child nutrition indicators 206 A.13 Logistic regression results of the binary propensity score estimations for maternal nutrition indicators 208 A.14 Logistic regression results of the binary propensity score estimations for indicators of the double burden of malnutrition at the family level and coexisting child and maternal overweight 210 A.15 Logistic regression results of the binary propensity score estimations for household diet quality indicators 212 A.16 Balancing property test statistics of the binary propensity score estimations for child nutrition indicators 214 A.17 Balancing property test statistics of the binary propensity score estimations for maternal nutrition indicators 216 A.18 Balancing property test statistics of the binary propensity score estimations for indicators of the double burden of malnutrition at the family level and coexisting child and maternal overweight 218 A.19 Balancing property test statistics of the binary propensity score estimations for household diet quality indicators 220 A.20 GLM results of the generalized propensity score estimations for child nutrition indicators and ration-card-program subsidies 222 xiv tablES A.21 GLM results of the generalized propensity score estimations for maternal nutrition indicators and ration-card-program subsidies 224 A.22 GLM results of the generalized propensity score estimations for indicators of the double burden of malnutrition at the family level and coexisting child and maternal overweight and ration- card-program subsidies 226 A.23 GLM results of the generalized propensity score estimations for household diet quality indicators and ration- card-program subsidies 228 A.24 Dose-response function estimates for the effect of ration- card-program subsidies on child HAZ 230 A.25 Dose-response function estimates for the effect of ration- card-program subsidies on child stunting 230 A.26 Dose-response function estimates for the effect of ration- card-program subsidies on child BMIZ 231 A.27 Dose-response function estimates for the effect of ration- card-program subsidies on child overweight 231 A.28 Dose-response function estimates for the effect of ration- card-program subsidies on mother’s BMI 232 A.29 Dose-response function estimates for the effect of ration- card-program subsidies on maternal overweight 232 A.30 Dose-response function estimates for the effect of ration- card-program subsidies on child stunting and overweight 233 A.31 Dose-response function estimates for the effect of ration- card-program subsidies on child stunting and maternal overweight 233 A.32 Dose-response function estimates for the effect of ration- card-program subsidies on child overweight and maternal overweight 234 A.33 Dose-response function estimates for the effect of ration- card-program subsidies on household dietary diversity 234 A.34 Dose-response function estimates for the effect of ration- card-program subsidies on frequency of household vegetable consumption 235 tablES xv A.35 Dose-response function estimates for the effect of ration- card-program subsidies on frequency of household legume consumption 235 A.36 Dose-response function estimates for the effect of ration- card-program subsidies on frequency of household meat and fish consumption 236 A.37 Dose-response function estimates for the effect of ration- card-program subsidies on frequency of household milk and dairy product consumption 237 A.38 GLM results of the generalized propensity score estimations for child nutrition indicators and Baladi bread and flour subsidies 238 A.39 GLM results of the generalized propensity score estimations for maternal nutrition indicators and Baladi bread and flour subsidies 240 A.40 GLM results of the generalized propensity score estimations for indicators of the double burden of malnutrition at the family level and coexisting child and maternal overweight and Baladi bread and flour subsidies 242 A.41 GLM results of the generalized propensity score estimations for household diet quality indicators and Baladi bread and flour subsidies 244 A.42 Dose-response function estimates for the effect of Baladi bread and flour subsidies on child HAZ 246 A.43 Dose-response function estimates for the effect of Baladi bread and flour subsidies on child stunting 246 A.44 Dose-response function estimates for the effect of Baladi bread and flour subsidies on child BMIZ 247 A.45 Dose-response function estimates for the effect of Baladi bread and flour subsidies on child overweight 247 A.46 Dose-response function estimates for the effect of Baladi bread and flour subsidies on mother’s BMI 248 A.47 Dose-response function estimates for the effect of Baladi bread and flour subsidies on maternal overweight 248 A.48 Dose-response function estimates for the effect of Baladi bread and flour subsidies on child stunting and overweight 249 xvi tablES A.49 Dose-response function estimates for the effect of Baladi bread and flour subsidies on child stunting and maternal overweight 249 A.50 Dose-response function estimates for the effect of Baladi bread and flour subsidies on child overweight and maternal overweight 250 A.51 Dose-response function estimates for the effect of Baladi bread and flour subsidies on household dietary diversity 250 A.52 Dose-response function estimates for the effect of Baladi bread and flour subsidies on frequency of household vegetable consumption 251 A.53 Dose-response function estimates for the effect of Baladi bread and flour subsidies on frequency of household legume consumption 251 A.54 Dose-response function estimates for the effect of Baladi bread and flour subsidies on frequency of household meat and fish consumption 252 A.55 Dose-response function estimates for the effect of Baladi bread and flour subsidies on frequency of household milk and dairy product consumption 252 Boxes 4.1 Dilemma of the 2011 HIECS data: Household survey sampling 106 4.2 Dilemma of the 2011 HIECS data: Nutrition and dietary diversity indicators 108 4.3 Dilemma of the 2011 HIECS data: Food subsidy benefits 110 boxES xvii xix Foreword our knowledge about the importance of good nutrition for economic development has been growing rapidly in recent years. Research has shown that malnutrition during the early years of life leads to lower edu- cational achievements and lower incomes later in life, with negative impli- cations for societies as a whole. While progress has been made in reducing undernutrition in several countries, overweight, obesity, and related noncom- municable diseases have become growing concerns. Egypt is one of the countries that is grappling with both under- and over- nutrition at the same time. In Nutrition and Economic Development: Exploring Egypt’s Exceptionalism and the Role of Food Subsidies, the authors argue that the recent aggravation of the double burden of malnutrition in Egypt can be explained by a combination of several driving factors: economic growth has been largely disconnected from nutrition improvements; poverty has risen as a result of a succession of recent crises; nutrition-sensitive public invest- ments (such as in basic healthcare and safe drinking water and sanitation) and nutritional awareness and education programs have been insufficient; and the population has faced a rapid nutrition transition due to changing diets and lifestyle. Further, as one of this book’s main contributions to the discussion on nutrition and social protection in developing countries, the authors show that the Egyptian food subsidy system, as it was in place until May 2014, helped to aggravate the double burden of malnutrition. Higher food subsidies increase the risk of malnutrition among both children and their mothers, particularly related to overnutrition. Thus—in addition to the conventional arguments for food subsidy reforms related to their high fiscal costs and often untargeted nature—the authors provide an additional argument for subsidy reform. Since the preliminary results of this book were first presented in Cairo in early 2013, the Egyptian government has started to fundamentally reform the food subsidy system, in line with several of this book’s recommendations. In the long run, an important question for policy makers will be whether to follow the path of countries such as Algeria, Jordan, and Mexico, by replac- ing food subsidies with (conditional) income transfers, or to maintain (tar- geted) food subsidies as the United States and India, for example, have done. For the short run, the authors make concrete suggestions as to how to make the current system more nutrition-sensitive, including smarter subsidy target- ing, need-based selection of subsidized products, food fortification, and nutri- tional education. Follow-up analyses to monitor the economic and nutritional benefits and costs of the reforming system and the willingness of policy makers to learn and adjust will be important for success. Shenggen Fan Director General, IFPRI xx ForEword xxi Acknowledgments this study benefited from the comments and suggestions of many people. We are particularly thankful to Dina Magdy Armanious, Heba El-Laithy, and Sherine Al-Shawarby from Cairo University; Abraham Abatneh, Riham Abuismail, Gian-Pietro Bordignon, Nora Soliman, and Jane Waite from the World Food Programme (WFP); Nadim Khouri (formerly United Nations Economic and Social Commission for Western Asia); Jean-François Maystadt from the University of Lancaster; Michela Bia from the Luxembourg Institute of Socio-Economic Research; and Harold Alderman, Daniel Gilligan, Katrina Kosec, and Jef Leroy from the International Food Policy Research Institute (IFPRI). We highly appreciate the provision of poverty and inequal- ity estimates by Heba El-Laithy and the support of Khaled Maher from the Egypt Central Agency for Public Mobilization and Statistics (CAPMAS) in providing the applied datasets of the 2010– 2011 Egypt Household Income, Expenditure, and Consumption Survey. We thank Marc Nene from Tufts University for his research assistance. We are also very thankful to IFPRI’s Publications Review Committee, its Chair Gershon Feder, and two anony- mous reviewers for their critical and valuable comments and suggestions. We gratefully acknowledge the editing and design of this book by IFPRI’s visual design and production team, particularly Jamed Falik, James Sample, and John Whitehead, and the great support of IFPRI’s communications team, led by Katrin Park. We thank BookMatters for their work in producing the book. Preliminary results of our study were presented at the Hidden Hunger Congress in Stuttgart (Germany) in March 2013; in seminars in Cairo at CAPMAS in September 2014, the Faculty of Economics and Political Science of Cairo University in May 2013, and the United States Agency for International Development in May 2013 and January 2015; in seminars at IFPRI in Washington, DC, in August 2013 and at the Georg-August University of Goettingen (Germany) in February 2015; and in a joint WFP-CAPMAS-IFPRI workshop in Cairo in January 2013. We thank all participants in these events for their comments. We gratefully acknowledge financial support from the International Fund for Agricultural Development and the CGIAR Research Program on Policies, Institutions, and Markets. xxii acknowlEdgmEntS Good nutrition is widely regarded as one of the key factors for advanc- ing human well- being and economic prosperity.1 Recent research clearly shows that malnutrition— and especially undernutrition— is not only a consequence of poverty, food insecurity, and disease but also one of the rea- sons for the lack of progress in economic development throughout the devel- oping world (IFPRI 2014).2 Undernutrition slows economic growth and deepens poverty through productivity losses from poor physical performance and cognitive capacity (World Bank 2006). Productivity losses to individu- als are estimated at more than 10 percent of lifetime earnings, and produc- tivity losses to the gross domestic product (GDP) in developing countries are at least 2– 3 percent annually (Horton 1999; World Bank 2006). These eco- nomic costs vary considerably by country and may exceed 10 percent of GDP in countries with high prevalence rates of undernutrition and relatively high per capita workforce productivity (Horton and Ross 2003; IFPRI 2014). Losses in household income potential and GDP are due to impaired cog- nitive abilities, which are especially relevant in more advanced economies 1 Malnutrition is a chronic condition caused by under- or overconsumption of any or several essential macro- or micronutrients or by adverse health conditions affecting nutrient absorp- tion or storage in the human body. According to Mayer (1976), four forms of malnutrition can be differentiated: (1) protein- energy undernutrition (caused by dietary deficiencies in carbohy- drates and/or proteins and frequently referred to as “hunger”), (2) micronutrient undernutrition (caused by dietary mineral and vitamin deficiencies, frequently referred to as “hidden hunger”), (3) overnutrition (mostly resulting from overconsumption of carbohydrates), and (4) secondary malnutrition (that is, under- or overnutrition primarily caused by illness or disease). This book looks at both ends of the malnutrition spectrum and therefore differentiates terminologically between under- and overnutrition. 2 “Poverty is pronounced deprivation in well- being. . . . It includes low incomes and the inability to acquire the basic goods and services necessary for survival with dignity” (World Bank 2012, adapted from Haughton and Khandker 2009). “Food security exists when all people, at all times, have physical, social and economic access to sufficient, safe and nutritious food to meet their dietary needs and food preferences for an active and healthy life” (FAO 1996, par. 1). The “four pillars of food security are availability, access, utilization, and stability” and “the nutritional dimension is integral to the concept” (FAO 2009, fn. 1). INTRODUCTION Chapter 1 1 (Hoddinott et al. 2008; Selowsky and Taylor 1973). Chronically undernour- ished children tend to have lower intelligence quotient (IQ) scores, by 5 to 11 points, and worse school performance compared to their well- nourished peers (Caulfield et al. 2006; World Bank 2006). Undernutrition among young chil- dren and women of reproductive age is of particular concern from a societal perspective, because prevalence rates among them are highest; their nutri- tional status is most crucial for the prosperity of the next generation; and nutritional interventions are most effective during the window of opportu- nity in the life cycle, comprising the time of pregnancy and the first two years of life (Bryce et al. 2008; Engle et al. 2007; Victora et al. 2008). In addition to losses in GDP, both under- and overnutrition also increase healthcare costs (World Bank 2006), thus contributing an additional burden to often strained budgets and potentially drawing resources away from other urgently needed social or developmental expenditures. While undernutrition is still the main nutritional problem in the devel- oping world, overnutrition is rapidly on the rise in many countries. The global prevalence of obesity nearly doubled between 1980 and 2008, reach- ing 14 percent among women and 10 percent among men (Finucane et al. 2011; Stevens et al. 2012). Along with North America, two developing regions— the Middle East and North Africa (MENA) and Latin America and the Caribbean (LAC)— exhibit the highest prevalence of obesity, with rates among women of more than 30 percent (Finucane et al. 2011; Stevens et al. 2012). Rising rates of overnutrition often go along with a growing prevalence of noncommunicable diseases (NCDs), which entails substantially increas- ing healthcare costs and productivity losses to the individual and the soci- ety (Finkelstein, Fiebelkorn, and Wang 2003; Finkelstein, Ruhm, and Kosa 2005; Popkin et al. 2006; Trogdon et al. 2008). For example, the costs attributable to overweight and obesity in China are expected to rise from about US$50 billion in 2000 (or 4 percent of gross national product [GNP]) to about US$112 billion in 2025 (or 9 percent of GNP) (Popkin et al. 2006). Deaths related to NCDs are projected to increase by 15 percent worldwide between 2010 and 2020, with the greatest increases expected to exceed 20 percent in MENA, Africa south of the Sahara, and Southeast Asia (WHO 2011). The rapid rise of overnutrition, combined with the relatively slow decline of undernutrition, has led in recent years to a new nutritional challenge of growing public health concern in several developing countries. The coexis- tence of over- and undernutrition— often referred to as “the double burden of 2 Chapter 1 malnutrition”— has been particularly prevalent in middle- income countries and especially those in the MENA and LAC regions. A possible explanation for the rise of the double burden of malnutrition is a rapid “nutrition transition.” This phenomenon describes the shift in dietary patterns and physical activity levels that emerges from economic growth and transformation in combination with technological advances (especially in communications and transportation) (Popkin 1993, 1994). Rapid shifts in dietary patterns and eating habits such as toward more processed foods and eating outside the home, in combination with a reduced physical workload from increasingly sedentary economic activities and a lack of physical exercise have increased overweight and obesity at a faster rate than undernutrition has been reduced (Prentice 2006; Schmidhuber and Shetty 2005; Shrimpton and Rokx 2012). In fact, several MENA countries, including Egypt— the most populous country in the region— have been going through a substantial nutri- tion transition since the mid- 1970s, associated with rapid economic develop- ment (Galal 2002). There is growing evidence that in addition to the nutrition transition, eco- nomic and social policies and programs may contribute to the rapid rise of overnutrition and the double burden of malnutrition in developing coun- tries. For example, Asfaw (2006, 2007a, 2007b) finds that women’s body mass indexes (BMIs) and the probability of female overweight and obesity in Egypt increase as prices of the foods that are subsidized under the national food sub- sidy system fall. And Leroy et al. (2013) show that cash and in- kind transfers lead to excess weight gain in a population of women in rural Mexico with a high prevalence of overweight. These case study findings are supported by evi- dence from studies in high- income countries, and particularly studies on the effects of the United States’ Food Stamp Program on overweight and obesity (e.g., Chen, Yen, and Eastwood 2005; Meyerhoefer and Pylypchuk 2008; Ver Ploeg et al. 2007; Zagorsky and Smith 2009). This book contributes to the literature on the effects of social policies and public programs on contemporary nutritional problems, using Egypt as a case study country. Specifically, our study serves two objectives: First, it pro- vides a comprehensive overview of Egypt’s two major nutritional challenges— which are exceptionally pronounced in Egypt compared to other developing countries— and their potential key drivers. These two nutritional challenges are (1) the double burden of malnutrition and (2) the phenomenon of high, decades- long economic growth that, contrary to expectations, was not accom- panied by declining prevalence rates of chronic (child) undernutrition— a challenge referred to as the “ growth- nutrition disconnect.” Second, our study IntroduCtIon 3 econometrically investigates causal relationships between one of the poten- tial key drivers— consumer food subsidies— and nutritional outcomes. We hypothesize that Egypt’s large and long- standing food subsidy system has con- tributed to sustaining and even aggravating both nutritional challenges. To serve the first objective, we document that the double burden of mal- nutrition and the growth- nutrition disconnect are indeed exceptionally pro- nounced in Egypt compared to other developing countries, describe the respective patterns of malnutrition among the Egyptian population, and elab- orate on four— possibly interlinked— factors that often have been hypothe- sized to cause or to contribute to the high prevalence of malnutrition in Egypt. In addition to the global phenomenon of the nutrition transition, they include rising poverty resulting from a succession of economic crises, the food subsidy system, and insufficient nutrition- sensitive investments. The analysis in the first part of the book draws on a combination of lit- erature reviews and descriptive statistics that we derived from various cross- country databases, official data sources, and cross- sectional household surveys, including several rounds of the Demographics and Health Survey (DHS) (MOH, El- Zanaty and Associates, and Macro International 2008; MOHP, NPC, and ORC Macro 2000; MOHP et al. 2003; MOHP et al. 2005). In the second part, we turn to the second objective by focusing on the causal effects of the Egyptian food subsidies on nutrition and the double bur- den of malnutrition in particular. For the main empirical analysis, we use both a quasi- experimental design and unique cross- sectional household survey data- sets— compiled from the 2010– 2011 Egypt Household Income, Expenditure, and Consumption Survey (HIECS) (CAPMAS and WFP 2011)— to esti- mate the hypothesized direct effects of the food subsidy system on child under- and overnutrition, maternal overnutrition, the coexistence of under- and overnutrition in the same children and the same child-mother pairs, and household diet quality. Providing statistical evidence for the existence of the causal relationships between received food subsidies and nutritional outcomes is fundamental for the hypothesized role of the food subsidy system as a driver of the double burden of malnutrition. In addition, a potential adverse effect of food subsidies on chronic child undernutrition is likely to contribute to the observed growth- nutrition disconnect. We chose Egypt as a case study because the two nutritional challenges of the double burden of malnutrition and the growth- nutrition disconnect are much more pronounced in Egypt than in other developing countries and because addressing them through the reform of existing policies and programs can be expected to make a critical contribution to accelerating the country’s 4 Chapter 1 economic and social development. Probably the most important ongoing social policy reform in Egypt is a substantial revision of the food subsidy sys- tem. We hope that the findings from our study will be useful for informing the ongoing food subsidy reform process and policies related to the country’s social safety net in general. Because implications of food subsidies for nutri- tion and public health seem to have been hardly considered in past reforms, our study may offer a new perspective and an additional rationale for further changes and fundamental modifications. From a public health perspective, the double burden of malnutrition is of particular concern. Almost every third Egyptian child under five years of age is chronically undernourished (according to 2011 estimates)— a prevalence rate that is more characteristic of developing countries with much lower national income levels than Egypt’s. Egypt has also one of the highest female overweight rates in the world, affecting 78 percent of all (nonpregnant) ever- married women 15– 49 years of age, while almost 40 percent are obese ( El- Zanaty and Way 2009).3 In addition, contrary to the global trend of decreasing undernu- trition accompanying economic growth, chronic child undernutrition sig- nificantly increased over at least the first decade of the 2000s, despite high economic growth. A decade- average GDP growth of 4.8 percent was associated with an increase in the prevalence rate of child stunting, from 24.6 percent in 2000 to 31.2 percent in 2011. Although a few other developing countries have experienced increasing chronic child undernutrition in the face of economic growth in the 2000s, the magnitude of this growth- nutrition disconnect in Egypt is exceptional relative to other countries in the MENA region and other developing countries worldwide (as we will show in this book). We argue that although the nutrition transition is an underlying develop- ment that has facilitated Egypt’s contemporary nutritional challenges, it falls short of explaining Egypt’s exceptionalism in the double burden of malnu- trition. A common presumed explanation for the increase in the prevalence of child stunting is the cumulative impact of a succession of recent economic crises. These crises include the devaluation of the Egyptian pound (EGP) in 2003; the avian influenza epidemic in 2006; the global food, fuel, and finan- cial crises of 2007– 2009; and the macroeconomic instability caused by the revolution in the spring of 2011. Although these shocks have contributed to the continuous increase in (monetary) poverty, they fail to convincingly explain why child stunting increased most among the richest wealth/income 3 Adults are classified as overweight if their body mass indexes (BMIs) are 25 or larger, and as obese, if their BMIs are 30 or larger. IntroduCtIon 5 quintile of the Egyptian population and remained (nearly) stable among the poorest quintile (as we will show in this book). We further argue that Egypt’s large food subsidy system has been ineffec- tive in reducing child and maternal undernutrition, and hypothesize that it has contributed to sustaining and even aggravating both nutritional chal- lenges. The existence and the design of the food subsidy system may hence provide an explanation for the country’s exceptionalism in the global compari- son. The rationale for this hypothesis is twofold. First, the food subsidy system may discourage good nutrition and possi- bly contribute directly to malnutrition— and to the double burden of mal- nutrition in particular— through incentivizing overconsumption of cheap, calorie- rich foods and unbalanced diets. Ample availability of cheap calories through subsidization creates an incentive for their consumption in excess, causing overweight and obesity. Calorie consumption above physiological requirement levels does not improve children’s physical growth, while retarda- tion in child growth in Egypt has been caused by insufficient micronutrient intakes from inadequate diets (possibly in addition to poor health condi- tions). Rapid increases of overweight and obesity and slow reduction, stag- nation, or even increase of child stunting over past decades led to the double burden of malnutrition. In addition, inadequate reduction of— or even an increase in— child stunting despite high economic growth formed part of the growth- nutrition disconnect. The coverage and benefits of the Egyptian food subsidy system have been large and were even expanded in the 2000s— partly in response to the recent economic crises. Second, the food subsidy system constitutes a heavy burden on the pub- lic budget (in addition to the even more sizeable fuel subsidies), so funds are unavailable for possibly more nutrition- beneficial investments. In this way, the subsidy system may maintain and aggravate malnutrition indirectly. This potential indirect effect of the food subsidies is omitted from our main empirical analysis because of a lack of the data required to conduct such an investigation. Following this line of argumentation, the book is structured as fol- lows. The second chapter analyzes the double burden of malnutrition and the growth- nutrition disconnect by comparing Egypt’s situation with that of other developing countries and by exploring within- country differences between regions and population groups and over time. The third chapter investigates the potential key socioeconomic drivers of Egypt’s exception- alism. These potential drivers include the nutrition transition, the suc- cession of recent economic crises, the food subsidy system, and the lack of 6 Chapter 1 nutrition- sensitive investment. The fourth chapter presents the main empir- ical analysis of the hypothesized nutritional effects of the Egyptian food sub- sidy system. It explains the applied methodology, describes the data from the 2010– 2011 HIECS that were used, and presents the estimation results. The fifth chapter concludes the book by summarizing the main findings and dis- cussing implications for policy and research. IntroduCtIon 7 From an economic development perspective, malnutrition among women of reproductive age and among young children is of particular concern. This is because prevalence rates tend to be highest among these groups, their nutritional well- being is most crucial for future generations’ prosperity, and targeted interventions have been found to be most cost- effective during the period from pregnancy to when a child is 24 months of age (Bryce et al. 2008; Engle et al. 2007; Horton et al. 2010). Malnutrition among pregnant women can have serious long- term effects because of the biological intergenerational transmission of malnutrition in utero (see below). And malnourished mothers tend to be less able to pro- vide sufficient and nutritious breast milk and adequate care to their chil- dren, which, in turn, increases the children’s risk of being malnourished. Malnutrition during early childhood reduces cognitive and physical capacity and productivity in children’s later life, so children born from malnourished mothers tend to be economically disadvantaged from the beginning compared to their well- nourished peers. Due to the importance of maternal and child nutrition for economic and social development, our analysis focuses mainly on the nutritional status of preschool children and of women of reproductive age— especially the chil- dren’s mothers. Nutritional status is determined by anthropometric measures. Anthropometric measurements, particularly for children under five years of age, have been shown to be reliable nutrition indicators and, more broadly, critical development indicators for several reasons. First, focusing on young children, who are typically the weakest house- hold members, captures aspects of intrahousehold distribution of resources (including food and care time) that are ignored when using household- level indicators such as poverty (as measured by household income) or food and nutrient consumption derived from standard household consumption surveys. Second, anthropometrics are measurements of the human body and hence indicators of nutritional (and health) outcomes (unlike food and nutrient NUTRITIONAL CHALLENGES FOR ECONOMIC DEVELOPMENT Chapter 2 9 consumption/intake). Third, they aggregately capture adequacy of food intake in terms of macro- and micronutrients, health status, and the interac- tion of nutrient absorption and diseases in the human body. Fourth, young children’s nutritional status tends to be more responsive to changing living conditions and shocks than that of adults due to children’s high physiologi- cal nutrient requirements for growth, special dietary needs, and high vulnera- bility to diseases common in underdeveloped settings. Fifth, high prevalence rates of malnourished children are often associated with poor delivery of basic public services (especially in the health sector), poor quality of drinking water and sanitation infrastructure, low educational levels, gender inequality, and high population growth rates. Using selected anthropometric indicators, the following two sections examine the double burden of malnutrition and the growth- nutrition disconnect— Egypt’s two main nutritional challenges— in detail. The Double Burden of Malnutrition Under- and overnutrition can coexist at the population, the family, and even the individual levels. Common forms of the double burden of malnutrition at the population level are high prevalence rates of stunting among children and overweight among adults; at the family level, stunted children with over- weight mothers; and at the individual level, stunted but overweight children. Other common forms of the double burden of malnutrition include the coex- istence of micronutrient deficiencies and overweight. Because this study is concerned with chronic malnutrition and the link to (future) economic devel- opment, it focuses on the coexistence of stunting among children under five years of age and overweight among them and among their mothers. From a development policy analysis point of view, a situation where the double burden of malnutrition is prevalent at the population level but under- and overnutrition do not coexist in the same family or the same individual among a considerable share of the population may be of less concern compared to a situation where the double burden of malnutrition is also prevalent at the family and the individual levels. The former situation does not constitute a novel problem and may rather be an outcome of other, well- studied develop- ment challenges such as large inequalities in the distribution of wealth and access to public services. For example, one may imagine a situation in a coun- try where undernutrition affects almost exclusively the poor and overnutrition almost exclusively the rich (Corsi, Kyu, and Subramanian 2011; Subramanian, Perkins, and Khan 2009). Because the problems of undernutrition and 10 Chapter 2 overnutrition are separate here, some policy implications from analyses look- ing at under- and overnutrition independently of each other should be trans- ferable to this situation (given comparable settings). However, such transferability of findings is inappropriate and can even lead to counterproductive policy recommendations where the double burden of malnutrition is also prevalent at the family and the individual levels, as in the case of Egypt (as we will show below). In such cases, households may face circumstances that can simultaneously contribute to both under- and overnu- trition, and policies that influence these circumstances require careful consid- eration of their potential effects on both under- and overnutrition. Moreover, these circumstances must occur at large in order to result in prevalence rates of under- and overnutrition coexisting in families and individuals that are suffi- ciently high to be a public health concern. Physiological Pathways to the Double Burden of Malnutrition In the search for factors that cause and contribute to the double burden of malnutrition, it is important to note that there are genetic conditions that favor its manifestation and severity. Due to lack of suitable cohort data, we cannot estimate nutritional effects because of genetic factors and assess their importance relative to those of socioeconomic factors captured in our analyses (presented later in this book). Genetic conditions’ contribution to the preva- lence of malnutrition among the current generation, however, is likely to be of minor relevance for our study, which focuses on factors that can be influenced by policy changes in the present. Yet factors that cause and contribute to malnutrition in the current generation also increase the risk of malnutrition among future generations. The following literature review therefore describes such physiological pathways and draws particular attention to those that facil- itate the phenomena of overweight mothers having stunted children and of stunted children being overweight at the same time. The health literature provides robust evidence that under- and over- nutrition is partly predetermined in utero. It also provides comprehensive evidence for several interlinked pathways by which fetal nutrition affects patterns of physical growth and body composition and increases the risk of nutrition- related noncommunicable diseases (NCDs) in the individuals’ later lives. Most of these pathways build on or are consistent with Barker’s well- known “mismatch hypothesis” (Barker 1988)— also known as “thrifty phenotype hypothesis” (Hales and Barker 1992, 2001). The hypothesis offers an explanation for the intergenerational transmission of malnutrition in terms of both over- and undernutrition and related NCDs. According to NutritioNal ChalleNges for eCoNomiC DevelopmeNt 11 the hypothesis, poor nutrition during fetal life causes irreversible metabolic changes that are possibly designed to ensure survival in similar conditions as those faced in utero but increase the risk of nutritional diseases during adult- hood when facing different conditions (Barker 1988). To be specific, undernutrition during pregnancy in terms of macro- and micronutrient deficiencies is associated with higher risks of fetal growth retar- dation, premature birth, and low birth weight (Han et al. 2011; Haider and Bhutta 2012; Scholl and Hediger 1994; Yi, Han, and Ohrr 2013). Consistent with Barker’s hypothesis, growth retardation can be seen as a mechanism to compensate for insufficient nutrient supply without compromising the devel- opment of vital organs. Low birth weight increases the risk of cardiovascu- lar diseases, hypertension, and type 2 diabetes in adulthood— NCDs that are typically associated with overweight and obesity (Barker 2004). Although overweight and obese women are less likely to give birth to low- weight babies (1,500– 2,499 grams) than normal- weight women, they are more likely to give birth prematurely and to bear very low birth weight babies (< 1,500 grams), aggravating the risk of stillbirth (McDonald 2010). In addi- tion, women with excess weight gain and associated symptoms of cardiovas- cular diseases during pregnancy are at an elevated risk of having children with these birth abnormalities (Eriksson et al. 1999; Gluckman and Hanson 2008). Being overweight during pregnancy can affect placental growth, leading to impaired nutrient supply to the fetus and thus fetal growth retardation as well (Wu et al. 2004). It is worth noting here that low birth weight is an evident indicator of sub- optimal intrauterine growth, but its absence does not necessarily imply optimal development for the fetus. Fetal undernutrition can affect cognitive develop- ment by causing structural damage to the brain and by impairing infant motor development and exploratory behavior (Victora et al. 2008). A cohort study from the United States suggests that differences in mean IQs are directly asso- ciated with birth weight, while each 0.5 IQ- point difference in male siblings at seven years of age corresponds to a 100- gram difference in birth weight, even within the normal birth weight range (Matte et al. 2001). Moreover, a system- atic review study indicates that maternal obesity adversely affects breastfeeding success (Turcksin et al. 2014): compared with their normal- weight counter- parts, obese women are less likely to intend to breastfeed. Maternal obesity is also associated with a delayed onset of milk secretion, a less adequate milk sup- ply, and a decreased initiation and shortened duration of breastfeeding. Undernutrition in utero and during infancy may also lead to perma- nent metabolic changes that increase the risk of overweight and obesity and 12 Chapter 2 related NCDs in adulthood (Barker 2004; Uauy et al. 2008). Children born from undernourished mothers may have developed a survival mechanism that allows them to better draw dietary energy from food and to more readily store it in their body as fat. However, when these individuals face a nutritional envi- ronment in childhood or adulthood that encourages obesity, they tend to gain weight faster than their healthy peers because of the prenatal metabolic pro- gramming (Hales and Barker 2001). In addition, people who were small at birth are more susceptible to hypertension and type 2 diabetes than people who were big. The development patterns of the two disorders are identical and coincide with that of coronary heart disease (Barker 2004). The risk of NCDs declines with increasing birth weight and rises with rapid weight gain in early childhood (Barker 2004; Koletzko et al. 2012). Overnutrition during pregnancy may also cause permanent metabolic changes in the unborn that increase the risk of overweight and related NCDs later in life (Dabelea 2007; Dabelea et al. 2000; Freinkel 1980). The “ fuel- mediated teratogenesis hypothesis” (Freinkel 1980) proposes that excess glucose supply in utero can give rise to irreversible dysfunctions of the body’s control systems, including glucose intolerance, hunger regulation, and fat accumulation (Plagemann 2008). While maternal glucose is freely transferred to the fetus, maternal insulin does not cross the placenta, causing the fetus’s pancreas to absorb high glucose levels by producing high volumes of insu- lin (Freinkel 1980). Although evidence from human studies is still limited, existing data point to a higher risk of diabetes in persons exposed to diabe- tes in utero, independent of adiposity and genetic predisposition of diabe- tes (Dabelea and Crume 2011). Yet the mechanism through which diabetes exposure in utero affects cardiovascular diseases is still not fully understood. Impaired insulin secretion has been proposed as a possible mechanism, consid- ering that studies in newborns of diabetic mothers have revealed an enhanced insulin secretion in response to a glycemic stimulus. It is still unclear, how- ever, if this is a transitory phenomenon or a symptom of impaired glucose tol- erance later in life, when insulin resistance becomes important (Dabelea and Crume 2011). Nonetheless, there is compelling evidence that children from overweight women or women with excessive weight gain during pregnancy are at an increased risk of overweight and related NCDs, independent of genetic characteristics. Finally, in utero growth retardation and low birth weight are associated with higher risks of postnatal child stunting and retarded head circumference growth, while children with reduced head circumference are more likely to be delayed in psychomotor development (Bove et al. 2012). Although stunting NutritioNal ChalleNges for eCoNomiC DevelopmeNt 13 mainly manifests itself postnatally and usually peaks around 24 months (Victora et al. 2010), children born prematurely or small for gestational age are more likely to remain physically retarded in their growth compared to children born with normal stature and weight (Hediger et al. 1999; Karlberg and Albertsson- Wikland 1995; Leger et al. 1997; Strauss and Dietz 1998). Under suboptimal nutritional and health conditions, growth faltering can- not be caught up easily and may even accumulate over time, given the rapid development that children undergo, particularly during their first two years of life. Stunted children are more likely to become overweight in their later lives (Bove et al. 2012; El Taguri et al. 2009; Popkin et al. 1996; Schroeder, Martorell, and Flores 1999). Early evidence on the potential physiological mechanism involved in this process suggests that stunted children have a lower fat oxidation than their well- nourished peers and are therefore at higher risks of adiposity (Sawaya and Roberts 2003; Hoffman et al. 2000a, 2000b). In conclusion, “chronic diseases are not the inevitable lot of humankind; they are the result of the changing pattern of human development,” as Barker (2012; p. 185) puts it. Many children are malnourished and have an increased risk of NCDs in their later lives because their mothers were malnourished during pregnancy and lactation. In addition to the genetic loading, children with malnourished mothers have an increased risk of malnutrition because they tend to share the same or a similarly inadequate diet. Household diets are determined by household incomes, food prices, education, and many other factors— some of which can be influenced by economic and social policies. Egypt’s Double Burden of Malnutrition in the Global Comparison The double burden of malnutrition is most pronounced in two developing regions, namely the MENA and LAC regions (Garrett and Ruel 2005). While undernutrition is still common in both regions, overnutrition has been rising rapidly in the wake of high economic growth and structural transformation in recent decades. In fact, along with North America, the MENA and LAC regions exhibit the highest prevalence of overweight and obesity, with rates among women (20 years of age and older) of more than 60 percent for over- weight and 30 percent for obesity (Finucane et al. 2011; Stevens et al. 2012). The prevalence of female overweight and obesity and the prevalence of child stunting are inversely correlated across countries. Considering that the countries are at different stages of economic development, the inverse relation- ship may also mark the typical course of decreasing child stunting and increas- ing female overweight and obesity that an average country follows throughout 14 Chapter 2 its development process. Our estimated functions suggest that the relation- ship has a flattened S- shaped curve (Figure 2.1 and Figure 2.2). Yet several countries deviate from the general relationship. For example, relative to the prevalence of child stunting, the prevalence of female over- weight and obesity is extremely high in Egypt, Guatemala, Iraq, Lesotho, Mexico, Panama, South Africa, and Saudi Arabia. Among these countries, the prevalence of child stunting exceeds 30 percent in Egypt, Guatemala, Lesotho, and South Africa. The gap between the actual prevalence rate of female overweight and obesity and the expected one given the actual child stunt- ing rate is widest in Egypt, particularly for female obesity— the more severe form of overnutrition. Hence, according to this measure, the double burden of malnutrition (at the population level) in Egypt is most pronounced in the global comparison. Egypt’s prevalence of female overweight and obesity alone is among the highest in the world. Egypt ranks among the top 10 countries with the highest rate of obese women, ahead of the United States, Mexico, and Iraq and behind only Kuwait, Libya, Qatar, and some small Caribbean island states (Ng et al. 2014). Data from the Egypt DHSs in 2000, 2003, 2005, and 2008 suggest that a stable proportion of around 80 percent of all (nonpregnant) women 20– 49 years of age were overweight between 2000 and 2008 (MOH, El- Zanaty and Associates, and Macro International 2008; MOHP, NPC, and ORC Macro 2000; MOHP et al. 2003; MOHP et al. 2005). The prevalence rate of female obesity in 2008 was about 40 percent, roughly the same as in 2000.1 Considering a longer study period of 33 years, Ng et al. (2014) find the largest average increase in the female obesity rate to be in Egypt, followed by Saudi Arabia, Oman, Honduras, and Bahrain. Hence, the public health problem of chronic overnutrition and related NCDs in Egypt has increased considerably in severity in recent decades (Herman et al. 1995). In fact, Egypt has one of the highest death rates from cardiovascular diseases and diabetes in the devel- oping world (Alwan et al. 2010). Abegunde et al. (2007) estimate that Egypt lost a cumulative US$1.26 billion to chronic diseases between 2005 and 2015; most of the loss is due to cardiovascular diseases and diabetes. 1 See Tables A.1 and A.2 in the Appendix. NutritioNal ChalleNges for eCoNomiC DevelopmeNt 15 FIGURE 2.1 Relationship between the prevalence of female overweight and child stunting MEX(2006) PAN(1998) GTM(1999) ZAF(2004) LSO(2004) SAU(2005)IRQ(2006) EGY(2008) 0 10 20 30 40 50 60 70 Pr ev al en ce o f f em al e ov er w ei gh t ( % )a 0102030405060 Prevalence of child stunting (%)b Source: authors’ estimation based on data from World Bank (2014). Note: the sample includes the latest observations of the prevalence of female overweight and child stunting (measured within a period of less than five years) for 101 countries. the regions are North america and Western europe high- income countries (Ê); arab world (·); africa south of the sahara (); latin america and Caribbean (�); east asia and pacific (¡); and West, Central, and south asia and eastern europe (Ð). Countries with a high prevalence of female overweight relative to their prevalence of child stunting include egypt (egY), guatemala (gtm), iraq (irQ), lesotho (lso), mexico (meX), panama (paN), saudi arabia (sau), and south africa (Zaf). the estimated relationship between the prevalence of female overweight and the prevalence of child stunting is obtained in two steps. in the first step, a nonparametric regression is applied to the data, and the estimated function is graphed to gain evidence on the form of a parametric function that yields the best data fit. a locally weighted regression is run, using stata’s “lowess” (locally weighted scatter plot smoothing) command. (the chosen bandwidth of the lowess curve is 0.8, which is stata’s standard bandwidth. the results of the nonparametric regression are not reported.) Based on the found shape of the curve, a fractional polynomial regression of degree 2 and a robust estimator of variance are applied to the data in the second step to determine the specific functional form and plot the predicted line of the relationship. the estimated function is: y = −1.370 * (((x /10)^3) – 19.556) + 0.668 * (((x /10)^3) * ln(x /10) – 19.381) + 36.360, where y is a country’s female overweight rate, and x is its child stunting rate. statistical significance and explanatory power of the model are high (f = 56.65, r- sq. = 0.536). the sample means divide the sample into four quadrants, which group the countries according to the severity of the double burden of malnutrition. the color of the quadrant indicates the severity of the double burden of malnutrition, with green indicating low severity and red indicating high severity. a the prevalence of female overweight is defined as the proportion of women 15 years of age and older with a body mass index (Bmi) of 25 or higher. b the prevalence rate of child stunting is defined as the proportion of children under five years of age with height- for-age z- scores (haZs) below −2. 16 Chapter 2 FIGURE 2.2 Relationship between the prevalence of female obesity and child stunting MEX(2006) PAN(1998) GTM(1999) ZAF(2004) LSO(2004) SAU(2005) IRQ(2006) EGY(2008) 0 10 20 30 40 0102030405060 Pr ev al en ce o f f em al e ob es ity (% )a Prevalence of child stunting (%)b Source: authors’ estimation based on data from World Bank (2014). Note: the sample includes the latest observations of the prevalence of female obesity and child stunting (measured within a period of less than five years) for 100 countries. the regions are North america and Western europe high- income countries (Ê); arab world (·); africa south of the sahara (); latin america and Caribbean (�); east asia and pacific (¡); and West, Central, and south asia and eastern europe (Ð). Countries with a high prevalence of female obesity relative to their prevalence of child stunting include egypt (egY), guate- mala (gtm), iraq (irQ), lesotho (lso), mexico (meX), panama (paN), saudi arabia (sau), and south africa (Zaf). the estimated relationship between the prevalence of female overweight and the prevalence of child stunting is obtained in two steps. in the first step, a nonparametric regression is applied to the data, and the estimated function is graphed to gain evidence on the form of a parametric function that yields the best data fit. a locally weighted regression is run, using stata’s “lowess” (locally weighted scatter plot smoothing) command. (the chosen bandwidth of the lowess curve is 0.8, which is stata’s standard bandwidth. the results of the nonparametric regression are not reported.) Based on the found shape of the curve, a fractional polynomial regression of degree 3 and a robust estimator of variance are applied to the data in the second step to determine the specific functional form and plot the predicted line of the relationship. the estimated function is: y = 7.119 * (((x /10)^0.5) – 1.633) – 1.095 * (((x /10)^3) – 18.957) + 0.546 * (((x /10)^3) * ln(x /10) – 18.582) + 14.999, where y is a country’s prevalence rate of female obesity (in percent) and x is its prevalence rate of child stunting (in percent). statistical significance and explanatory power of the model are moderately high (f = 20.80, r- sq. = 0.375). the sample means divide the sample into four quadrants, which group the countries according to the severity of the double burden of malnutrition. the color of the quadrant indicates the severity of the double burden of malnutrition, with green indicating low severity and red indicating high severity. a the prevalence of female obesity is defined as the proportion of women 15 years of age and older with a body mass index (Bmi) of 30 or higher. b the prevalence rate of child stunting is defined as the proportion of children under five years of age with height- for-age z- scores (haZs) below −2. NutritioNal ChalleNges for eCoNomiC DevelopmeNt 17 Patterns of the Double Burden of Malnutrition within Egypt Data from the HIECS in 2011 (CAPMAS and WFP 2011) suggest that almost every third young child in Egypt is stunted. Hence, their physical development is retarded as a result of inadequate nutrition and poor health conditions, which will remain a life- long health impairment in most cases— with all its implications for individual well- being and economic development. In addition to chronic child undernutrition, child overnutrition is wide- spread. The 2011 HIECS data suggest that overweight among children ages 6– 59 months is almost as prevalent as stunting in this age group, while a large proportion suffer from stunting and overweight at the same time (Table 2.1). TABLE 2.1 Prevalence patterns of the double burden of malnutrition in Egypt DBM at population level DBM at family level DBM at individual level Child stunting Child overweight Female overweight Female obesity Stunted child with overweight mother Stunting and overweight in children Total (%) 31.2 29.2 72.6 33.9 22.3 14.0 By region and residential areaa (%) metropolitan 30.8 29.5 74.0 33.9 26.9 15.6 lower egypt 27.1 30.7 78.3 40.9 20.6 12.7 urban 24.7 27.1 78.5 39.7 20.4 11.9 rural 27.8 31.7 78.2 41.4 20.7 12.9 upper egypt 34.7 27.0 65.6 25.8 21.5 14.3 urban 39.2 28.1 71.5 29.4 25.6 16.2 rural 33.4 26.7 62.8 24.1 20.2 13.7 By income quintile and residential area (%) Quintile 1 (poorest) 34.0 27.9 66.2 27.6 20.0 14.6 Quintile 2 32.3 28.9 71.2 32.4 22.2 15.0 Quintile 3 29.2 28.7 75.1 36.6 22.8 11.9 Quintile 4 27.2 26.6 75.3 37.5 20.3 12.5 Quintile 5 (richest) 33.1 33.7 75.0 35.5 26.1 15.9 urban 31.9 28.8 74.3 34.2 25.3 15.1 Quintile 1 (poorest) 33.0 26.4 72.6 31.4 21.8 13.2 Quintile 2 32.3 29.9 77.0 36.7 25.5 16.6 Quintile 3 31.9 27.4 74.2 35.8 26.4 16.3 Quintile 4 33.0 30.4 76.1 36.2 29.9 14.1 Quintile 5 (richest) 29.4 30.1 71.6 30.7 22.9 15.4 18 Chapter 2 While 31.2 percent of the children are stunted, 29.2 percent are overweight.2 Of those children who are stunted, about 45 percent are overweight. In total, 14.0 percent of all children ages 6– 59 months are stunted and overweight. Children’s height- for-age z- scores (HAZs; indicating stunting) are distinctly negatively and linearly correlated with their body-mass-index-for-age z-scores (BMIZs; indicating overweight and obesity) (Figure 2.3); the correlation coef- ficient is −0.431. Hence, in Egypt shorter children tend to be fatter than their taller peers. Note that child wasting (indicating acute child undernutrition) 2 Children are classified as stunted if their height- for-age z- scores (HAZs) are below –2 and as overweight if their body-mass-index-for-age z- scores (BMIZs) are 2 or above. DBM at population level DBM at family level DBM at individual level Child stunting Child overweight Female overweight Female obesity Stunted child with overweight mother Stunting and overweight in children rural 30.8 29.3 71.3 33.8 20.6 13.4 Quintile 1 (poorest) 32.3 27.1 62.3 25.5 17.6 13.9 Quintile 2 33.8 28.2 69.6 31.0 22.4 14.5 Quintile 3 30.7 31.6 71.0 33.4 22.4 12.1 Quintile 4 26.8 25.9 74.9 37.3 18.8 11.7 Quintile 5 (richest) 30.3 34.0 78.8 41.8 21.9 14.7 Source: authors’ calculation based on data from Capmas and Wfp (2011). Note: DBm = double burden of malnutrition. the child stunting sample includes children ages 6–59 months with biologically plausible height-for-age z-scores (haZs) (−6 ≤ haZ ≤ 6) and has 3,852 observations. the child overweight sample includes children ages 6–59 months with biologically plausible body-mass-index-for-age z-scores (BmiZs) (−5 ≤ BmiZ ≤ 5) and has 3,631 observations. the female overweight/ obesity sample includes nonpregnant women 20–49 years of age with biologically plausible body mass indexes (Bmis) (5.2 ≤ Bmi ≤ 52.1) and has 9,778 observations. all others are subsamples of these samples with observations in both original samples. the child stunting and maternal overweight sample has 3,661 observations, and the child stunting and overweight sample has 3,577 observations. the prevalence rate of child stunting is defined as the proportion of children with haZs below −2, and the prevalence of child overweight is defined as the proportion of children with BmiZs of 2 or above. the prevalence of female overweight is defined as the proportion of women with Bmis of 25 or above, and the prevalence of female obesity is defined as the proportion of women with Bmis of 30 or above. the household disaggregation by wealth/income quintile in the Demographic and health surveys samples (moh, el-Zanaty and associates, and macro international 2008; egypt, mohp, NpC, and orC macro 2000; mohp et al. 2003; mohp et al. 2005) is based on the distribution of asset wealth (measured by a composite index), and the household disaggregation in the household income, expenditure, and Consumption survey sample (Capmas and Wfp 2011) is based on the distribution of household income per capita. a prevalence rates for frontier governorates are not reported because of insufficient observations. NutritioNal ChalleNges for eCoNomiC DevelopmeNt 19 and child underweight (indicating overall child undernutrition) have been low in Egypt, with national prevalence rates of below 8 percent since 2000.3 Moreover, the 2011 HIECS data (CAPMAS and WFP 2011) suggest that about 71 percent of all stunted Egyptian children have an overweight mother.4 Out of all Egyptian child- mother pairs, 22.3 percent are child- mother pairs with a stunted child and an overweight mother (Table 2.1). According to the HIECS data, 72.6 percent of (nonpregnant) Egyptian women 20– 49 years 3 Children are classified as wasted if their weight- for-height z- score (WHZ) is below –2 and as underweight if their weight- for-age z- score (WAZ) is below –2. For disaggregated prevalence rates of child nutrition indicators, see Tables A.3–A.6 in the Appendix. 4 The HIECS (CAPMAS and WFP 2011) does not explicitly indicate the child’s mother. However, we are able to identify the likely mother— or female caretaker— using information from the household member characteristics module. See Box 4.2 in the subsection “Survey Data and Estimation Variables” below. In the following, we omit “caretaker” for ease of readability. FIGURE 2.3 Relationship between children’s BMIZs and HAZs in Egypt –4 –2 0 2 4 –6 –4 –2 0 2 4 6 Ch ild B M IZ Child HAZ Source: authors’ estimation based on data from Capmas and Wfp (2011). Note: the sample includes children ages 6– 59 months with biologically plausible body-mass-index-for-age z-scores (BmiZs) (−5 ≤ BmiZ ≤ 5) and height- for-age z- scores (haZs) (−6 ≤ haZ ≤ 6) and has 3,577 observations. the relationship between children’s BmiZs and haZs is estimated nonparametrically, using a locally weighted regression estimation inbuilt in stata. the chosen bandwidth is 0.8, which is stata’s default bandwidth of the “lowess” (locally weighted scatter plot smoothing) command. 20 Chapter 2 of age are overweight, and 33.9 percent are obese.5 For these women, we also find a linear, negative relationship between their BMIs and absolute body heights (Figure 2.4), although the correlation, with a coefficient of −0.193, is less strong than that for children. Hence, shorter Egyptian women tend to be fatter than their taller peers. Note that stunting in women— which man- ifests itself mainly during early childhood— is not very prevalent in Egypt today. The 2011 HIECS data suggest that only 1.6 percent of all (nonpreg- nant) women 20– 49 years of age are shorter than 145 centimeters (cm). And female underweight is almost nonexistent in Egypt. According to the HIECS 5 The 2011 HIECS data (CAPMAS and WFP 2011) indicate somewhat lower prevalence of female overweight and obesity than the 2008 DHS data (MOP, El- Zanaty and Associates, and Macro International 2008; MOHP, NPC, and ORC Macro 2000; MOHP et al. 2003; MOHP et al. 2005) did. These differences may partly be due to different survey years and sam- ple populations. FIGURE 2.4 Relationship between women’s BMIs and body heights in Egypt 10 20 30 40 50 120 140 160 180 200 Fe m al e BM I ( kg /m 2 ) Female height (cm) Source: authors’ estimation based on data from Capmas and Wfp (2011). Note: cm = centimeters; kg/m2 = kilograms per square meter. the sample includes nonpregnant women 20– 49 years of age with biologically plausible body mass indexes (Bmis) (5.2 ≤ Bmi ≤ 52.1) and has 9,778 observations. the relationship between women’s Bmis and heights is estimated nonparametrically using a locally weighted regression estimation inbuilt in stata. the chosen bandwidth is 0.8, which is stata’s default bandwidth of the “lowess” (locally weighted scatter plot smoothing) command. NutritioNal ChalleNges for eCoNomiC DevelopmeNt 21 data, 0.8 percent of all (nonpregnant) women 20– 49 years of age are under- weight.6 DHS data from the survey rounds in 2000, 2003, 2005, and 2008 indicate that underweight affected less than 1 percent of all (nonpregnant) women throughout the 2000s (MOH, El- Zanaty and Associates, and Macro International 2008; MOHP, NPC, and ORC Macro 2000; MOHP et al. 2003; MOHP et al. 2005). These first results have several important implications: The double bur- den of malnutrition in Egypt is not only highly prevalent at the population level but also at the family and the individual levels. The findings that under- and overnutrition exist in child- mother pairs and in the same children at con- siderable prevalence rates, and that individual body height and weight are negatively correlated among young children and women of reproductive age, suggest that lack of sufficient staple foods can be ruled out as a cause of wide- spread undernutrition. In other words, chronic hunger should not be an issue of public concern. This notion is also supported by low prevalence rates of wasting and underweight among children and of underweight among women. Instead, these correlations point to a nutritional situation where calorie- rich foods are readily and cheaply available, which encourages overconsump- tion of these foods; where a large share of the population cannot afford or does not prefer a diversified diet dense in micronutrient- rich foods (such as meat, fish, dairy products, pulses, vegetables, and fruits); and where individual health conditions compromise optimal nutritional outcomes. The results also imply that the same circumstances in households’ environments, the same behavior of their members, and the same individual characteristics may indeed contribute to both under- and overnutrition at the same time. Furthermore, the HIECS data (CAPMAS and WFP 2011) show no dis- tinct patterns in the prevalence of undernutrition, overnutrition, and the double burden of malnutrition between urban and rural areas and between regions (Table 2.1). This might be surprising given that there are clear eco- nomic development gaps between urban and rural areas, the Metropolitan areas and the rest of Egypt, and Lower and Upper Egypt. The regional pat- terns of some forms of malnutrition appear to even be inconsistent with those of economic development. For example, child stunting is somewhat more prevalent in the Metropolitan areas than in urban Lower Egypt— although less prevalent than in urban Upper Egypt. In Upper Egypt, the prevalence of child stunting is higher in urban areas than in rural areas. Female overweight and obesity are more prevalent in both urban and rural areas of Lower Egypt 6 Adults are classified as underweight if their BMIs are below 18.5. 22 Chapter 2 than in the Metropolitan areas, whereas they are least prevalent in Upper Egypt. The double burden of malnutrition at the family and the individual levels is most prevalent in the Metropolitan areas and urban Upper Egypt.7 The absence of distinct spatial prevalence patterns of under- and overnutri- tion and unexpectedly high prevalence rates in some parts of Egypt may be explained by the coverage of the food subsidy system, the local health environ- ment, and differential impacts of the crises in the 2000s (Kavle et al. 2015a) to some extent. Breaking down the malnutrition prevalence rates by income quintiles reveals no clear income- dependent patterns for any considered form of malnu- trition— with the exception of female overnutrition in rural areas (Table 2.1).8 Thus, unlike most other developing countries, Egypt shows no typical ten- dency of high rates of undernutrition among poor rural households and of high rates of overnutrition among rich urban households. In contrast, under- nutrition and overnutrition are highly prevalent among both the poor and the rich at relatively similar rates. These results— obtained from the 2011 HIECS data (CAPMAS and WFP 2011)— are largely consistent with the patterns found in the DHSs conducted in the 2000s (MOH, El- Zanaty and Associates, and Macro International 2008; MOHP, NPC, and ORC Macro 2000; MOHP et al. 2003; MOHP et al. 2005).9 In addition, the HIECS data also do not reveal, either in rural or in urban areas, obvious income- dependent patterns of the double burden of malnutrition at the family and the individ- ual levels. Overall, these findings provide some initial evidence that household income (and monetary poverty) alone is unlikely to be the main driver of mal- nutrition in Egypt. The next section explores how the prevalence of chronic child undernutri- tion— one component of the double burden of malnutrition— changed over time in Egypt relative to national economic growth and compared to other developing countries, as well as between different regions in Egypt and among different income/wealth groups of the Egyptian population. 7 See Tables A.7 and A.8 in the Appendix. 8 This also holds true when using household expenditure instead of household income for defin- ing quintiles. In this study we use reported household incomes from the income section of the HIECS (CAPMAS and WFP 2011) because household expenditures are influenced by the amount of subsidies the households receive. We checked the robustness of all results using household expenditures instead of household incomes and did not find significant deviations. Note that the coefficient of the correlation of household income and expenditures is 0.906 across the entire sample (and 0.909 for urban households and 0.844 for rural households). 9 See Tables A.1–A.8 of the Appendix. NutritioNal ChalleNges for eCoNomiC DevelopmeNt 23 The Growth- Nutrition Disconnect Economic growth leads to reduced undernutrition— at least in the long term— through two main routes. First, economic growth increases house- hold incomes. Higher incomes allow households to purchase more and more nutritious food and to spend more on health, education, and housing that may result in improved nutrition indicators. Hence, the degree to which growth translates into reduced prevalence rates of undernutrition mainly depends on the proportion of the undernourished population that experiences income increases and the rates at which that population’s income grows— in other words, it depends on the inclusiveness of economic growth. Second, economic growth tends to increase the public budget through higher tax revenues, state- owned enterprises, and other revenue sources. These additional revenues provide fiscal space for income redistribution pol- icies (affecting individual nutrition through the first route), investments in nutrition- sensitive infrastructure and public services (such as water and san- itation and basic healthcare), and health and nutrition- specific interventions (such as micronutrient supplementation and fortification and nutrition infor- mation and education programs). In addition, policies often play a critical role in influencing consumers’ choices: they can either create an environment that encourages people to use increasing income for improving nutrition or that discourages them from nutrition- beneficial spending. An example of a pol- icy that can incentivize unhealthy diets is food subsidies— as hypothesized in this book. Thus, national policy can influence the degree to which economic growth translates into reduced prevalence rates of undernutrition through the amount of additional public budget generated from economic growth, the allocation and efficiency of the additional resources, the quality (and effectiveness) of the public services provided, and the design of policies enabling or interfering with nutritional improvement (Bryce et al 2008; Gillespie et al. 2013; Haddad et al. 2003; Smith and Haddad 2002). Egypt’s Chronic Child Undernutrition and Economic Growth in a Global Comparison Relative to the national income level, chronic child undernutrition has been considerably more prevalent in Egypt than in most other developing coun- tries— with the exception of a period in the late 1990s and early 2000s when it reached average prevalence rates (Figure 2.5). On a global level, countries’ child stunting rate has declined, at decreasing margins, with increasing GDP 24 Chapter 2 FIGURE 2.5 Relationship between the prevalence of chronic child undernutrition and national income 0 20 40 60 80 0 1,000 2,000 3,000 4,000 Pr ev al en ce o f c hi ld s tu nt in g (% )a GDP per capita (US$)b Egypt 1978 1992 2000 2003 2005 2011 Global trend Source: authors’ estimation based on data from World Bank (2014), complemented with data from other sources. Child stunting rates for egypt in 2011 are derived from Capmas and Wfp (2011) data. missing year observations of gross domes- tic product (gDp) per capita levels are replaced by computed gDp per capita levels, using gDp growth rates derived from imf (2014) and uNstat (2014) and population data from uN- Desa (2014). Note: the sample includes child stunting and gDp data for 141 countries with child stunting rates of 5 percent and above. the sample has a total of 628 observations, spanning the period from 1966 to 2012. the figure presents an excerpt, includ- ing countries with gDp per capita below us$4,000 (composing 89 percent of all observations). the estimation of the “global trend” graph is bivariate and cross- country (based on the full sample) and includes two steps. in the first step, a nonparametric regression is applied, and the results are plotted to gain evidence on the general functional form of the relationship between the prevalence of child stunting and gDp per capita levels. a locally weighted regression is run, using stata’s “lowess” (locally weighted scatter plot smoothing) command. (the chosen bandwidth of the lowess curve is 0.8, which is stata’s standard bandwidth. the results of the nonparametric regression are not reported.) given the shape of the curve, a fractional polynomial regression of degree 1 and a robust estimator of variance are applied to the data in the second step to determine the specific functional form and plot the predicted line of the relationship. the regression is weighted by population size. the estimated function is: y = 8.604 * (((x /10,000)^(–0.5)) – 2.385) + 21.220, where y is a country’s prevalence rate of child stunting (percentage) and x is its gDp per capita level in the same year. statistical significance and explanatory power of the model are high (f = 124.7, r- sq. = 0.625). We tested alternative specifications of the functional form that had gDp per capita levels lagged by one to five years but found no clear evidence for general time lags. the shaded area around the global trend line marks the 95 percent confidence interval. a the prevalence rate of child stunting is defined as the proportion of children under five years of age with height- for-age z- scores (haZs) below −2. b gDp per capita is measured at constant 2000 prices. NutritioNal ChalleNges for eCoNomiC DevelopmeNt 25 T A B L E 2 .2 C ou nt ry c om pa ris on o f c hi ld s tu nt in g re du ct io n an d ec on om ic g ro w th Co un tr y Ar ou nd 2 01 0 Ar ou nd 2 00 0 An nu al a ve ra ge c ha ng e in th e pr ev al en ce o f c hi ld st un tin g (p er ce nt ag e po in ts ) An nu al a ve ra ge GD P pe r ca pi ta g ro w th (% ) Ar c ch ild st un tin g- gr ow th el as tic ity c Pr ev al en ce o f ch ild s tu nt in g (% )a GD P pe r ca pi ta (U S$ )b Ye ar Pr ev al en ce o f ch ild s tu nt in g (% )a GD P pe r ca pi ta (U S$ )b Ye ar Eg yp t 31 .2 1, 97 7 20 11 24 .6 1, 47 6 20 00 0. 6 2. 7 0. 22 Co un tr ie s w ith s im ila r s tu nt in g ra te sd an go la 29 .2 56 3 20 07 61 .7 27 2 19 96 − 3. 0 6. 8 − 0. 43 Ca m er oo n 32 .6 66 6 20 11 36 .7 57 0 19 98 − 0. 3 1. 2 − 0. 26 ha iti 29 .7 38 5 20 06 28 .3 42 4 20 00 0. 2 − 1. 6 − 0. 15 ho nd ur as 29 .9 1, 35 4 20 06 34 .5 1, 15 0 20 01 − 0. 9 3. 3 − 0. 28 to go 29 .8 26 6 20 10 29 .8 28 3 19 98 0. 0 − 0. 5 0. 00 Zi m ba bw e 32 .3 32 3 20 10 33 .7 55 6 19 99 − 0. 1 − 4. 8 0. 03 Co un tr ie s w ith s im ila r G DP p er c ap ita le ve ls e al ge ria 15 .9 2, 11 5 20 05 23 .6 1, 79 4 20 00 − 1. 5 3. 3 − 0. 46 az er ba ija n 26 .8 1, 57 4 20 06 24 .1 65 5 20 00 0. 5 15 .7 0. 03 Ch in a 9. 4 2, 42 6 20 10 17 .8 94 9 20 00 − 0. 8 9. 8 − 0. 09 gu at em al a 48 .0 1, 85 6 20 09 50 .0 1, 71 7 20 00 − 0. 2 0. 9 − 0. 26 m or oc co 14 .9 1, 94 1 20 11 29 .0 1, 21 5 19 97 − 1. 0 3. 4 − 0. 30 sy ria 27 .5 1, 50 9 20 09 24 .3 1, 20 9 20 00 0. 4 2. 5 0. 14 Ot he r A ra b co un tr ie sf ira q 27 .5 71 0 20 06 28 .3 1, 06 3 20 00 − 0. 1 − 6. 5 0. 02 Jo rd an 8. 3 2, 57 7 20 09 12 .0 1, 87 1 20 02 − 0. 5 4. 7 − 0. 11 om an 9. 8 11 ,1 92 20 09 12 .9 8, 34 3 19 99 − 0. 3 3. 0 − 0. 10 26 Chapter 2 so m al ia 42 .1 28 8 20 06 29 .2 27 7 20 00 2. 2 0. 6 3. 48 su da n 38 .3 58 2 20 06 47 .6 44 5 20 00 − 1. 6 4. 6 − 0. 34 tu ni si a 9. 0 2, 77 6 20 06 16 .8 2, 24 5 20 00 − 1. 3 3. 6 − 0. 36 W es t B an k an d ga za 11 .8 98 4 20 07 16 .1 97 9 20 02 − 0. 9 0. 1 − 7. 54 Ye m en 59 .6 57 4 20 06 59 .3 50 8 19 97 0. 0 1. 4 0. 02 Ot he r l ar ge d ev el op in g co un tr ie sg Ba ng la de sh 41 .4 58 8 20 11 50 .8 36 4 20 00 − 0. 9 4. 5 − 0. 19 Br az il 7. 1 4, 29 8 20 07 13 .5 3, 62 8 19 96 − 0. 6 1. 6 − 0. 37 et hi op ia 44 .2 23 0 20 11 57 .4 12 4 20 00 − 1. 2 5. 8 − 0. 21 in di a 47 .9 62 2 20 06 51 .0 44 1 19 99 − 0. 4 5. 0 − 0. 09 in do ne si a 39 .2 1, 14 5 20 10 42 .4 77 3 20 00 − 0. 3 4. 0 − 0. 08 Ni ge ria 36 .0 56 6 20 11 39 .7 36 1 19 99 − 0. 3 3. 8 − 0. 08 pa ki st an 43 .0 67 2 20 11 41 .5 51 1 20 01 0. 2 2. 8 0. 05 ph ili pp in es 33 .6 1, 41 3 20 11 38 .3 1, 01 7 19 98 − 0. 4 2. 6 − 0. 14 vi et na m 23 .3 72 3 20 10 43 .4 40 2 20 00 − 2. 0 6. 1 − 0. 33 So ur ce : a ut ho rs ’ c al cu la tio n ba se d on d at a fro m W or ld B an k (2 01 4) , c om pl em en te d w ith d at a fro m o th er s ou rc es . C hi ld s tu nt in g ra te s fo r e gy pt in 2 01 1 ar e de riv ed fr om C ap m as a nd W fp (2 01 1) d at a; ch ild s tu nt in g ra te s fo r Y em en in 2 00 6 ar e de riv ed fr om C so (2 00 6) d at a. m is si ng y ea r o bs er va tio ns o f g ro ss d om es tic p ro du ct (g Dp ) p er c ap ita le ve ls a re re pl ac ed b y co m pu te d gD p pe r c ap ita le ve ls , us in g gD p gr ow th ra te s de riv ed fr om im f (2 01 4) a nd u Ns ta t (2 01 4) a nd p op ul at io n da ta fr om u N- De sa (2 01 4) . No te : t he s am pl e in cl ud es c ou nt rie s w ith p op ul at io ns a bo ve 5 m ill io n pe op le , a c hi ld h ea lth s ur ve y in th e pe rio d 20 05 –2 01 3, a c hi ld h ea lth s ur ve y in th e pe rio d 19 96 –2 00 4, a d iff er en ce o f a t l ea st fi ve ye ar s be tw ee n th e tw o su rv ey s, a nd g Dp d at a re po rte d fo r t he y ea rs o f t he c hi ld h ea lth s ur ve ys . a t he p re va le nc e ra te o f c hi ld s tu nt in g is d efi ne d as th e pr op or tio n of c hi ld re n un de r fi ve y ea rs o f a ge w ith h ei gh t- fo r- ag e z- sc or es (h aZ s) b el ow − 2; b g Dp p er c ap ita is m ea su re d in u s do lla rs a t c on st an t 20 00 p ric es ; c a rc e la st ic iti es a re p re se nt ed in g ra y if co un tri es e xp er ie nc ed n eg at iv e gD p pe r c ap ita g ro w th . th e sa m pl e in cl ud es 9 1 co un tri es . t he c ou nt rie s w er e se le ct ed a cc or di ng to th e fo llo w in g cr ite ria : d C ou nt rie s w ith c hi ld s tu nt in g ra te s in th e ra ng e of 2 p er ce nt ag e po in ts a ro un d eg yp t’s ra te ; e C ou nt rie s w ith g Dp p er c ap ita le ve ls in th e ra ng e of u s$ 50 0 ar ou nd e gy pt ’s g Dp ; f a ra b co un tri es th at d o no t f al l i nt o th e fir st tw o ca te go rie s, w he re a ra b co un tri es in cl ud e al l 2 1 ar ab l ea gu e m em be r s ta te s an d sy ria (w hi ch w as s us pe nd ed fr om th e le ag ue in 2 01 1) ; g C ou nt rie s w ith p op ul at io ns o ve r 8 0 m ill io n pe op le th at d o no t f al l i nt o th e fir st tw o ca te go rie s (e gy pt ’s p op ul at io n in 2 01 1 w as 8 2. 5 m ill io n) . NutritioNal ChalleNges for eCoNomiC DevelopmeNt 27 per capita levels. Egypt followed this global trend throughout the 1980s and 1990s and even experienced over- proportionate reduction rates in the second half of the 1990s. However, Egypt’s progress reversed in the early 2000s, and the prevalence of child stunting steadily increased throughout the 2000s. In fact, despite high economic growth in the 1990s and 2000s, child stunting in Egypt in 2011 (31.2 percent) was as prevalent as it was in 1992 (31.3 percent).10 Between 1992 and 2000, the prevalence of child stunting in Egypt declined by an average annual rate of 0.83 percentage points, so each 1 percent GDP per capita growth was associated with an average annual decrease in the child stunting rate of 0.29 percentage points. Globally, each 1 percent GDP per capita growth was associated with an average annual decrease in the child stunting rate of only 0.12 percentage points (at a GDP per capita level equivalent to Egypt’s average GDP per capita during this period). However, between 2000 and 2011, the prevalence of child stunting in Egypt increased by an average annual rate of 0.60 percentage points, and each 1 percent GDP per capita growth in Egypt was associated with an average annual increase of 0.22 percentage points in the prevalence of child stunting (Table 2.2). In con- trast, the global trend suggests an average annual decrease of 0.10 percentage points (at a GDP level equivalent to Egypt’s period- average GDP per capita). International comparisons confirm that Egypt’s nutrition- growth dis- connect is particularly pronounced (Table 2.2). Countries with prevalence rates of child stunting similar to Egypt’s have considerably lower GDP per capita levels. Conversely, countries with similar GDP per capita levels have substantially lower prevalence rates of child stunting, with the exception of Guatemala— another country facing a pronounced double burden of malnu- trition (Figure 2.1 and Figure 2.2). Further, only a few ot