Growth is Good but Not Enough to Improve Nutrition

8/3/2019 Growth is Good but Not Enough to Improve Nutrition

1/44

Growth is Good, but is notEnouGh to improvE nutrition

ole Ecke, Clee bege, a Kal pa

2020 Conference Paper 7 February 2011

Aace Cy


2/44

2020 Cnfrnc: Lvraging Agricltr fr Imprving Ntrin and Halth

Fby 10-12, 2011; N Di, Indi

Adance Copy

OlIvIer ecker, clemeNs BreIsINger, aND karl Pauw

2020 Conference Paper 7

Growth is Good, but is not Enough toImprove Nutrition

Olivier Ecker i podoo fo, Clemens Breisinger fo, nd Karl Pauw podoo fo

in the Development Strategy and Governance Division of the Internaonal Food Policy Research Instute,

winon, Dc.

Many people have contributed to this paper. Thanks for the comments, suggesons, and prior work of threeanonymous reviewers; Shenggen Fan, Rajul Pandya-Lorch, Xinshen Diao, and Samuel Benin of the Internaonal

Food Policy Research Instute (IFPRI); James Thurlow of United Naons University World Instute for Develop-

ment Economics and Research (UNU-WIDER); Man Qaim of the University of Goengen; and others.


3/44


4/44

Foreword

In recent years we have learned a great deal about how agriculture, nutrion, and health are linked.

We experience that policies designed to increase agricultural producon can either improve or threat-

en peoples nutrion and healthand conversely that peoples nutrion and health status can contrib-

ute to or interfere with agricultural producon. Researchers have shed light on some of the specic

paths through which these links play out, but gaps in our knowledge remain. There is sll much to be

learned about how agriculture, nutrion, and health are connected, how they interact and, especially,

how we can use those interacons to advance goals in all three sectors.

To help close the knowledge gaps, the 2020 Vision Iniave of the Internaonal Food Policy ResearchInstute (IFPRI) has commissioned a series of papers as part of a larger global consultaon that in-

cludes the February 2011 internaonal conference Leveraging Agriculture for Improving Nutrion

and Health (hp://2020conference.ifpri.info/). Developing eecve policies in agriculture, nutrion,

and health will require a strong foundaon of evidence. It is our hope that these papers will help lay

ondo fo ndndin of i nd d o poii i nn y,

producve lives for all people.

snn Fn rj Pndy-lo

Director General Head, 2020 Vision Iniave


5/44

Contents1. Introducon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2. Economic Growth and the (Dis)Connecon to Nutrion . . . . . . . . . . . . . . . . 6

3. Linking Growth and Economic Policies to Nutrion Outcomes: A New Approach for

sy Pnnin nd anyi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.1 a conp Fo fo Food siy anyi . . . . . . . . . . . . . .10

3.2 Esmang the Nutrional Impacts of Growth and Economic Policies: The

Food siy Fo appid . . . . . . . . . . . . . . . . . . . . . . . . 13

4. Impact of Economic Policies on Nutrion: Lessons from Case Studies . . . . . . .15

4.1 The Nutrion Problem in Yemen and Malawi: A Comparave Overview . . . 15

4.2 Impact and Lessons from Policy Reforms Opons in Yemen . . . . . . . . . . 16

4.2.1 A business-as-usual scenario (no policy change) . . . . . . . . . . . .16

4.2.2 ai poiy fo . . . . . . . . . . . . . . . . . . . . . . . . . 17

4.2.3 Poiin o o poiy. . . . . . . . . . . . . . . . . . . . . .

194.2.4 copin i-d o i non-i o-d

o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

4.3 Ip nd lon fo ai Poii in mi . . . . . . . . . . . 23

4.3.1 P iz-d o p . . . . . . . . . . . . . . . . . . . . . . . . 25

4.3.2 rn o on-n o . . . . . . . . . . . . . . . . . . . . . . . . 26

4.3.3 Bod-bd i o . . . . . . . . . . . . . . . . . . . . . 28

4.3.4 Diversifying agricultural growth and nutrion intervenons . . . . . 29

5. Conclusions and Policy Recommendaons. . . . . . . . . . . . . . . . . . . . . . .

30rfn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

appndix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36


6/44

grOwth Is gOOD, But Is NOt eNOugh tO ImPrOve NutrItION 5

Growth is Good, but is not Enough toImprove Nutrition

OlIvIer ecker, clemeNs BreIsINger, aND karl Pauw

Astract

NUTRITION IS INCREASINGLY RECOGNIzED AS AN IMPORTANT DIMENSION OF

economic development, yet relavely lile is known about the relaonship among

development policies, economic growth, and nutrion outcomes. Advancing the knowledge

in this eld can help policymakers to design strategies and priorie acon for accelerang

o andimproving nutrion. To contribute to this process, this paper uses cross-country

analysis and an innovave analycal framework applied to two case study countries. We

nd that policies that foster growth are important for improving nutrion. Agricultural or

non-agricultural growth can contribute to improved nutrion, depending on the countrys

economic structure and the characteriscs of the malnourished people. Agriculture has a

strong potenal to contribute to the reducon of malnutrion in agriculture-based economies,

mi, in in o-bd oni i iid i

potenal, such as Yemen, industry and service sector-led growth leads to beer nutrion

outcomes. However, we also nd that growth is not enough for improving child nutrion

and reducing micronutrient malnutrion. Therefore, pro-growth policy reform needs to

be complemented by strategic health and educaon investments and targeted nutrion

po.

1. IntrodctionThere is broad consensus in the development economics literature that growth is a necessary condion for

poverty reducon (Ravallion and Chen 1997, Dollar and Kraay 2002, Kraay 2004). However, the extent to which

growth translates into reducons in poverty largely depends on the structure of growth and the characteriscs

of the poor (Ravallion and Chen 1997, Deininger and Squire 1998, Bourguignon 2002). Compared to growth in

other sectors, agricultural growth has been shown to be more eecve at reducing poverty, especially in coun-

tries with large agricultural sectors and a large concentraon of agricultural households among the poor (Delga-

do et al. 1998, Diao et al. 2010, World Bank 2007). Manufacturing and service-sector led growth becomes more

important for poverty reducon during the process of economic transformaon, when the share of agriculture

in the economy declines and people migrate to urban areas (Breisinger and Diao 2008, World Bank 2009).

While the connecon between growth and poverty reducon is well established, relavely lile is known

about the relaonship between economic growth and nutrion and, hence, how economic policies can be lev-

eraged for improving nutrion. Some household-level studies have looked at the relaonship between changes

in household incomes and calorie and micronutrient deciency (e.g., Abdulai and Aubert 2004, Ecker and Qaim2011, Skouas et al. 2009). Other studies (e.g., Haddad et al. 2003, Alderman 2001) have analyed the eects of

changes in incomes on child nutrion. Results of these household-level studies generally show a posive rela-


7/44

6 GROWTH IS GOOD, BUT IS NOT ENOUGH TO IMPROVE NUTRITION

onship between growing incomes and nutrion outcomes, albeit to various degrees. Even fewer studies exist

on the relaonship between economic growth and nutrion improvement. Headey (2011) uses cross-country

regressions to explain changes in childrens nutrional status as an outcome of economic growth and nds a

posive but weak relaonship between growth and reducons in child malnutrion. However, there is also evi-

dence that economic growth does not always translate into improved nutrion such as in India. Deaton (2010)

nds that per capita calorie consumpon in India has been falling for a quarter of a century in spite of high rates

of per capita income and consumpon growth in recent years. Indias annual GDP growth averaged 7.2 percent

over the period 1997-2007, and poverty dropped from 39 percent in 1998 to 30 percent in 2006 (according tothe internaonal $1.25-a-day poverty line), but undernourishment (measured on the basis of per capita calorie

availability) increased from 17 percent in 1997 to 21 percent in 2007 (WDI 2010).

Against this background, several important quesons remain, the answers to which may provide important

insights to policymakers on how to improve nutrion. First, are there other countries in addion to India where

growth and nutrion may be de-linked? Second, does the structure of growth maer (as in the case of poverty)

for nutrional outcomes and, if so, how? Finally, how do dierent policies aect nutrion and what can govern-

ments do to improve nutrion?

This paper aims at to provide evidence to shed light on these quesons. We present a comprehensive

conceptual framework that links economic growth and policies to their nutrion outcomes. We also propose an

innovave methodological approach that allows modeling the eects of overall and (sub)sectoral growth and

economic policies on peoples nutrional status, consistent with our conceptual framework. We then apply the

methodology to two case study countries to simulate the nutrional impacts of economic policies. We have

chosen Yemen and Malawi as two countries that are low-income with high levels of malnutrion. Yemen repre-

sents an oil-based economy with a relavely small agricultural sector and Malawi an agriculture-based economy

with limited diversicaon. These countries therefore span a broad range of possible growth-nutrion rela-

onships so that ndings are relevant for variety of other countries, parcularly in the Middle East and North

afin ion nd sb-sn afi.

The paper is structured as follows: Secon 2 uses cross-country analysis to explore the general relaon-

ships between economic and agricultural growth and the prevalence of malnutrion over me. It serves as a

background for the secons that follow and helps to understand the (dis)connecon between economic growth

and nutrion by country. Secon 3 presents the conceptual framework and the methodology linking economicgrowth and policies to nutrion outcomes. Secon 4 comprises the case studies, and Secon 5 presents conclu-

sions and policy recommendaons based on the analyses of the paper.

2. Economic Grot and te (Dis)Connection to NtritionTo explore the general relaonship between economic growth and nutrion outcomes in the process of de-

velopment, we draw on results from cross-country data analyses. In addion, given agricultures important

role in poverty reducon, we specically analye this sectors impact on nutrion over me. The analysis is

based on economic and nutrion data available from the World Development Indicators (WDI) database of

the World Bank (WDI 2010). We assess a countrys stage of economic progression based on its annual gross

domesc product (GDP) per capita (at constant 2000 prices), and the contribuon of the agricultural sector by

the sectors annual value added per worker (at constant 2000 prices). The nutrion outcome indicators that

we apply for this analysis are the prevalence of undernourishment in the countries populaons, esmated on

a per capita calorie availability basis by the Food and Agriculture Organiaon (FAO)s stascs division (FAO-

STAT 2010), and the prevalence of child stunngan anthropometric indicator measuring the physical growth

retardaon in children under ve caused by long-term malnutrion and poor health condionsprovided by

the World Health Organiaon (WHO) Global Database on Child Growth and Malnutrion (GDCGM 2010); both

indd in wDI db.

Figure 1 shows the general relaonship between undernourishment and naonal GDP across (almost) all

low-income and lower middle-income countries (the red curve) and the relaonships for selected countries

over a period of 15 years, from the early 1990s to the middle 2000s. Figure 2 shows the same relaonships for

agricultural value added instead of total GDP. The red curves in both graphs mark the ed lines using fraconalpolynomial esmaon. They can be interpreted as the global undernourishment-growth trend and the global


8/44


undernourishment-agricultural growth trend, respecvely, which a representave country tends to follow in

po of dopn.

Both gures show that economic growth leads to a reducon of undernourishment (Figure 1 and Figure 2).

The shape of the ed line in Figure 1 indicates that undernourishment declines rapidly with growing naonal

income during the early growth period, though the relaonship becomes much weaker when the annual GDP

per capita exceeds about US$500. Thus, growth is generally important for achieving a situaon in which most

people have enough food to be suciently supplied with calories and hence free of hunger.

However, Figure 1 also shows large dierences in the prevalence of undernourishment among countriesat a similar income level as well as in the countries progress toward eradicang undernourishment (Figure 1).

so oni xpind i onoi o nd nd o i fo din ndnoi-

ment substanally, such as Vietnam. Other countries grew at lower rates or stagnated in economic growth, but

achieved a signicant reducon in undernourishment such as Malawi and Bangladesh. Few other countries

such as Yemen and zimbabwe had a similar GDP in 1990 and 2005, and their situaon of undernourishment did

no n o n ond. mny o oni indin und diy b f

in translang growth into undernourishment reducon; and, few countries failed to reduce undernourishment

dpi i o Indi in n y nd g oo 1990.

Figr 1Rlanship btwn ndrnrishmnt and GDP

0

10

20

30

40

50

60

70

80

0 500 1,000 1,500 2,000 2,500 3,000

India

Egypt

Bangladesh

Uganda

Malawi

Guatemala

Zimbabwe

Vietnam

Yemen

Annual GDP per capita (constant 2000 US$)

Proportonofunderno

urishedpeople(%

ofpopulaton)

Note: The dots mark the proporon of undernourished people for selected countries in 1990-92, 1995-97, 2000-02, and 2005-07 and their annual GDP

per capita in the respecve years. The red curve is the ed line using fraconal polynomial esmaon and data of 111 countries with annual GDP per

capita below US$ 3,000 (403 observaons). Figure A1 of the Appendix shows the graph for all country observaons.

Source: WDI (2010).

Agricultural growth also reduces undernourishment, albeit at dierent rates (Figure 2). The shape of the

of indiid oni o i fo oni i p of

naonal GDP and employs a large proporon of the undernourished populaon, such as in Malawi, zimbabwe,

Bangladesh, and Vietnam. However, comparing the ed line in Figure 2 with that in Figure 1 reveals important

paerns on the potenal of agricultural growth against overall growth for reducing undernourishment. During

the early stages of transformaon, agricultural growth strongly corresponds to a reducon in undernourish-

n, b in inin o pidy n o fo o o; nd, i inini o, on o n ndnoin of 10 pn, ipyin n

of 10 percent of the populaon cannot be freed from undernourishment by a growing agricultural sector. Thus,


9/44


dopn po, ipon of i o fo din ndnoin din

absolutely and relave to the role of non-agricultural growth.

Figr 2Rlanship btwn ndrnrishmnt and agricltr val addd

0

10

20

30

40

50

60

70

80

0 500 1,000 1,500 2,000 2,500 3,000

India

Egypt

Bangladesh

Uganda

Malawi

Guatemala

Zimbabwe

Vietnam

Yemen

Annual agriculture value added per worker (constant 2000 US$)

Propo

rtonofundernourishedpeople(%

ofpopulat

on)

Note: The dots mark the proporon of undernourished people for selected countries in 1990-92, 1995-97, 2000-02, and 2005-07 and their annual agricul-

ture value added per worker in the respecve years. The red curve is the ed line using fraconal polynomial esmaon and data of 107 countries with

annual agriculture value added per worker below US$ 3,000 (375 observaons). Figure A2 of the Appendix shows the graph for all country observaons.

Source: WDI (2010).

The next set of gures (Figure 3 and 4) show the relaonships between child malnutrion (measured by

child stunng) and naonal GDP and agricultural value added, respecvely. As in Figure 1 and Figure 2, the red

curves mark the ed lines and can be interpreted as the general child malnutrion-overall growth trend and

the general child malnutrion-agricultural growth trend which a representave country tends following in the

po of dopn.

As expected, Figures 3 and 4 show that overall growth and agricultural growth lead to a reducon of child

malnutrion in general. The ed lines in the two graphs are almost idencal, implying that an agricultural

sector growing faster than the rest of the economy has no addional benecial eect on child nutrion,per

se. Compared to the undernourishment curves in Figure 1 and 2, the child malnutrion curves are much at-

ter and slow down at much higher child malnutrion levels (converging toward a base child malnutrion rate

of somewhat above 15 percent). This suggests that child malnutrion is less responsive to overall growth andi o oo po of onoi dopn, iin poinn o o of

non-income related factors, such as informaon and knowledge, individual health, and care, in reducing child

malnutrion, especially at later stages of development.


10/44


Figr 3Rlanship btwn child malntrin and GDP

0

10

20

30

40

50

60

70

80

0 500 1,000 1,500 2,000 2,500 3,000

India

Egypt

Bangladesh

Uganda

Malawi

Guatemala

Zimbabwe

Vietnam

Yemen

Annual GDP per capita (constant 2000 US$)

Proporonofstuntedchildren(%

ofchildrenunderfive)

Note: The dots mark the proporon of stunted children (as determined by the height-for-age measure) for selected countries from dierent years over

the past three decades and their annual GDP per capita in the respecve years. The red curve is the ed line using fraconal polynomial esmaon and

data of 95 countries with annual GDP per capita below US$ 3,000 (266 observaons). Figure A3 of the Appendix shows the graph for all country observa-

ons.

Source: WDI (2010).

The weaker relaonship between growth and child nutrion is also supported by the observaon of greater

variaons between countries in the growth-child nutrion relaonship versus the growth-undernourishment

relaonship. For example, despite relavely low growth, Bangladesh has achieved impressive results in cut-

ng back child malnutrion constantly over me. In contrast, Egypt has experienced relavely high and steady

o o p dd i o of poy nd ndnoin. ho, pn

of child malnutrion is largely unrelated to growth and even increased in recent years back to its early 1990s

. O piod 2000-2008, eyp gDP n of 4.8 pn p nn. Poy

(measured by the internaonal $1.25-a-day poverty line) and undernourishment aect less than 5 percent of

the populaon since the early 1990s, but child stunng increased from 23 percent in 2000 to 29 percent in 2009

(WDI 2010). In Guatemala, child malnutrion remains among the highest in the world in spite of the countrys

high GDP level compared to other countries with similarly adverse child nutrion situaons.1

1 We could not nd any published studies that convincingly explain Bangladeshs success in reducing child malnutrion and the worsening of child nutri-on in Egypt against the background of the countries economic development.


11/44


12/44


The conceptualiaon of our framework follows this denion of food security and expands exisng frame-

o.2 In consideraon of the serious impacts of the 2007/08 food price crisis at both macro- and micro-

economic levels, our framework extends the common perspecve of food security as primarily a household-

level issue by including macroeconomic factors. The framework considers the many factors that inuence food

security and explicitly accounts for sectors that are most relevant for peoples nutrional status, namely agri-

culture, trade and infrastructure, and health and educaon. As such, it emphasies the need for an integrated,

cross-sector approach for improving food security eecvely. It is parcularly designed for analying country

development strategies and specic intervenon opons in the form of policies, investments, and programsto reduce malnutrion. It also allows for assessing the impacts of external events (or shocks), such food price

i, i n, n di o pidi.

The framework builds upon the noon that success in naonal policies or other intervenons can be as-

sessed in terms their impact on peoples nutrional status, given that nutrion is a fundamental human need

with important long-term development implicaons (World Bank 2006). In determining food security, the avail-

ability of and access to food in terms of quanty andquality maers, with quality referring to adequate macro-

and micronutrients and types of food (including breast milk for infants). Due to a close relaonship between

individual nutrional status and health condion, factors that aect nutrion indirectly through changes of

indiid in oponn of o fo.

Fi 5 pn o onp fo. I oin py in food iy y o

which policies and external shocks translate into nutrion outcomes and the key factors determining the out-

comes. For analycal purposes, the framework dierenates food security at the naonal and regional (subna-

onal) levelthe macro leveland at the household and individual levelthe micro level. The factors at each

level are horiontally and vercally interconnected through various linkages; the major ones will be outlined in

the following paragraphs. Given the focus of this paper, the case studies in the next secon look at the nutrion

outcomes and do not analye the food security situaon at the macro and micro level in detail. Nonetheless, it

is important to understand the muldimensionality of the food security problem in search of eecve, integrat-

ed soluons which are key for improving nutrion.

Figr 5Cncptal framwrk f fd scrity fr cntry dvlpmnt stratgy analysis

Source: cond by o.

2 The framework draws on the United Naons Children Fund (UNICEF)s framework of the causes of child malnutrion and death (UNCEF 1990) and theconcept of macro-level food security by Dia-Bonilla et al. (2002).


13/44


Macro-level food security refers to the balance of food supply and demand for a parcular country in which

populaons food needs can be met through domesc producon, imports, orusuallya combinaon of

the two. Thus, macro-level food security refers to the availability of food for the populaon, which is a neces-

sary but insucient condion for the individuals access to food. Food imports have to be nanced through the

exports of other goods and services. Therefore, the structure of the domesc economy and the output of the

various sectors maers for macro-level food security.

t o in o io o poidin ood nd i nd i fo ,

where households earn incomes from factors of producon such as capital, labor, and land. Households thenspend their incomes on food and other goods and services. These vercal links can mutually contribute to

iin food iy o nd io . ai i y o fo iin food -

iy in y of onoi dopn nd py n ipon o in dopn po in ny

ways; however, the sectors relevance for food security changes during the process of transformaon. Depend-

ing on a countrys stage of development, agricultures primary role in contribung to household food security

can be in supplying food, employment, and/or income (Haddad 2000). Agricultural growth has strong linkage

eects thus driving overall growth and oen reduces food prices, which especially benets net food consumers

with a high share of spending on food (Diao et al. 2010, World Bank 2007). Agriculture also has a high poten-

al to cushion rising income inequalies oen observed during structural transformaon, which oen leads to

high adjustment costs and can become a binding constraint for development (Breisinger and Diao 2008). At the

macro-level, agricultural exports oen provide substanal export earnings, thus improving macro-level food

iy.

Other key sectors for achieving food security are trade and infrastructure and health and educaon. In the

course of labor diversicaon and market integraon driving a countrys economic development, an ecient

trading system, funconing market instuons, and physical infrastructure become increasingly important to

bi food ppy in o food o fo pod o on. rod nd

transportaon are also needed to enable peoples access to health and educaon services (Fay 2005, Smith et

al. 2005, Strauss and Thomas 1998). The health and educaon sectors are crical for peoples nutrional status,

since malnutrion is oen caused by preventable diseases and a lack of nutrional informaon, and knowledge

determines households food choice, intra-household food distribuon and childrens feeding pracces, among

others (Lope et al. 2006, Walker et al. 2007)t io opi ood nd i indiid b. mio- food iy i d-

mined by the access to sucient and nutrious food, to basic assets, and to (public) services that can promote

individuals good nutrion and health. Most cases of limited access are due to limited nancial resources, thus

explaining the strong associaon between income poverty and hunger or malnutrion. Thus, the most impor-

tantbut not solefactor aecng food access is household income and, in subsistence households, food

producon for household consumpon. In addion, the higher the income of a household, the larger the buer

against food volality. Food prices are another key factor for food security, parcularly for the poor in develop-

ing countries, who are oen net food buyers and oen spend more than half their income on food, High prices

reduce a households real income, and relave food price changes aect food consumpon paerns that might

lead to a reducon of the diet quality and give rise to increasing micronutrient deciencies (Ecker and Qaim

2011).

Beyond the availability of sucient and nutrious food in the household, individual food and nutrient

in i dind by of non-ino fo. Indiid o pop food i bj o in-o-

hold resource allocaon and care for the individual, both of which in turn depend on the characteriscs of the

decision maker and gender roles, informaon and educaon, cultural and social customs, and others (Frangillo

et al. 1997, Rogers 1996, Thomas and Frankberger 2002, Wolfe and Behrman 1982). Especially for the nutrion

and health of young children, mothers (or caretakers) health, capability and knowledge are crical (Block 2004,

Garre and Ruel 1999, Ruel et al. 1992). Since the childrens mother is usually also the person responsible for

meal preparaon, her knowledge also maers for the nutrion of all household members (Popkin 2003, Tanu-

mihardjo et al. 2007, WHO/FAO 2003).

Besides adequate food intake, an individuals health condion aects the nutrional status substanally. Forexample, highly prevalent infecous diseases such as malaria or parasite-caused diarrhea reduce the absorp-

on of nutrients, necessitang higher nutrient intake and thus more food to cover the losses, if such compen-

saon is possible at all. At the same me, poor nutrion weakens the human immune system and therewith


14/44


increases the risk of infecon with diseases (Murray 1997, Chandra 2002). Hence, peoples access to assets and

basic (public) services, including proper shelter, hygienic sanitaon, clean drinking water, and basic health care

such as disease treatment, vaccinaon, and mother and child care programs, aect peoples nutrion indirectly,

o in i .

Good nutrion is a fundamental need for individuals to realie both their maximum physical and intellectual

potenal. It is the basis for the well-being of individuals and households and for human capital formaon and,

as such, key for economic development and the formaon of sociees in communies and countries. Converse-

ly, poor nutrion and health have serious, long-lasng economic consequences at the micro and the macrolevel. Malnutrion and illness reduce household income earning ability, perpetuate poverty, and slow economic

growth through three routes: (1) direct losses in producvity from poor physical and mental performance or

death, (2) indirect losses from reduced working and cognive capacity and related decits in schooling, and

(3) losses in resources due to increased health care costs. Even temporary malnutrion can cause irreversible

health impairments, parcularly in children, liming the development potenal of future generaons (World

Bank 2006).

External shocks, such as global economic crises, climate change, conicts, natural disasters, and epidem-

ics, can threaten food security. While it is oen dicult or even impossible to prevent such external shocks,

ppopi i nn i ipon o n food iy din i. gin f

i ino oni po o n o pond o xn o, poii ppo d

o in n onib o i nn. O food iy i nn oo ind

physical grain storage facilies, hedging in internaonal markets, micro-insurance schemes, and social safety

n, on o.

3.2 Estimating te Ntritional Impacts of Grot and Economic Policies: Te Food SecrityFrameork Applied

The conceptual framework presented above serves as the basis of a quantave modeling approach for es-

mang the nutrional impacts of growth and economic policies. Economic growth can aect households

directly and indirectly. The extent to which households directly benet from growth depends on the factors of

producon they possess and the demand for these factors by growing sectors and/or regions. Indirect benets

work via redistribuon, where the households that do not benet or even lose out in the growth process areopnd by onn, fo xp o di nf nd o onn pndin.

Sound policies and public investments play an important role in promong and accelerang economic

growth. For example, a sound investment climate is key for aracng private (foreign and domesc) invest-

ments. Foreign direct investments (FDI) in parcular not only increase the capital stock, but oen also come

with important knowledge spill-over eects (Markusen and Venables 1999, Torvik 2001). Public investments

that support the private sector in increasing producvity, such as through improved infrastructure and public

services, also have been key in the transformaon process of most successful countries (Breisinger and Diao

2008). For example, the Green Revoluon not only reinforced the view that technology-led producvity growth

facilitated by public investments can transform tradional agriculture into a modern sector, but also showed

that agriculture helps accelerate the economywide transformaon process (Diao et al. 2010).

t nd o onoi poii nd inn n ino ood ino n o i-

ous linkages between sectors and instuons which are governed by factor and commodity markets. Subject

to a variety of individual and household-specic factors, household income changes might then lead to adjust-

ments of peoples food consumpon, asset stocks, and demand for nutrion-relevant services and therewith

their nutrional status. To quanfy this sequence of eects and esmate the nal nutrional impact of growth

and economic policies, a comprehensive and integrated model is needed that captures all interacons between

sectors and instuons of an economy and their mulplier eects and translates them into changes in house-

hold incomes and, in turn, links those to the nutrional status of the individual household members.

In this paper, we present such a model and apply it in two case studies (see next secon). Consistent with

the conceptual framework presented in the previous subsecon, this model approach treats peoples nutrion-

di oo of ood ino nd non-ino fo nd n indi oo of o,

economic policies, and other external factors. For esmang the eects of growth and policies on the economy

and the incomes of dierent household groups, we use dynamic computable general equilibrium (DCGE)

models. DCGE models are economy-wide models that are based on disaggregated social accounng matrices


15/44


(SAMs) which capture all economic linkages and interacons between all sectors and instuons within the domesc

economy and the relaonships with the rest of the world. SAMs are composed of various data sources, including bal-

ance of payments, trade stascs, naonal accounts, and naonally representave household surveys. For example,

the DCGE model used in the Yemen case study includes 65 sectors, of which 22 are in agriculture; and the DCGE model

employed for the Malawi case study is disaggregated into 36 sectors, of which 17 are in agriculture. The Yemen DCGE

model is described in detail in Breisinger et al. (2010), and the Malawi DCGE model (which SAM has been updated for

this study) can be found in Benin et al. (2008).

In both country studies, we use the DCGE models to esmate the impact of economic policies on growth andhousehold incomes. The simulaons take a forward-looking perspecve and run unl 2020, i.e. the DCGE models are

i dyni. t Dcge od ond oniny i noi n qiibi oy.

The theorecal background and the analycal framework of computable general equilibrium (CGE) models are well

documented in Dervis, de Melo, and Robinson (1982), while the detailed mathemacal presentaon of a stac CGE

model is described in Lofgren, Harris, and Robinson (2002). The recursive dynamic version of the CGE model is based

on this standard CGE model, with the incorporaon of a series of dynamic factors. The early version of this dynamic

CGE model can be found in Thurlow (2004), while its recent applicaons include Diao et al. (2007) and Breisinger, Diao,

and Thurlow (2009).

For assessing changes in peoples nutrion levels as a response to changes in their income levels (measured on the

basis of household total expenditure), we use an expenditure elascity-based approach that captures the percentage

change in certain nutrion indicators to a one percent change in household total expenditure. We use a variety of dif-

ferent nutrion indicators. Specically, the nutrion indicators in the Yemen study are the dietary energy status of all

people, measured in calories, and the height-for-age anthropometric measure for children under ve, which idenes

child stunnga manifestaon of long-term malnutrion (and poor health). In the Malawi study, we analye the im-

pacts on peoples micronutrient status in addion to their calorie suciency level. The chosen micronutrients are iron,

inc, vitamin A, and folate, which are essenal for physical and mental human development and for which deciencies

are widespread and parcularly crical from a development policy perspecve.3 t onoi od po-

duced these nutrion-income elascies, the theory underlying the models, the calculaon method of the nutrion

indicators, and the data used are presented and discussed in Ecker et al. (2010) for Yemen and Ecker and Qaim (2011)

for Malawi. The elascity esmates are presented in Table A1 and A2 in the Appendix. In addion to income variables

from which the nutrion-income elascies are derived, the econometric models include various sociodemographicvariables (and, in the case of Malawi, price variables for all food groups consumed). Hence, the econometric models

control for structural dierences between households in their gender and age composion, educaonal level, regional

locaon, and access to markets or health-relevant assets and services.

To esmate the nutrion-growth eects and to simulate the nutrional impacts of policy opons over me, we

combine the annual income changes obtained from the DCGE model simulaons with the nutrion-income elascity

esmates from the econometric models for each household individually. Assuming specic changes in dierent mac-

roeconomic parameters under dierent policy scenarios, we predict a new nutrion level for each person per annum,

subject to the esmated annual income changes. The nutrion simulaon equaon is:

N,i,j

= yN,i,j-1

(1 + EN

Ci,j),

N,i,jis a persons predicted nutrion level for the nutrion indicator N nd nio ind in yj,yN,i,j-1ithe nutrion level in the previous year, E

Nis the expenditure elascity with respect to the nutrion indicator N, nd

Ci,j

i nn ino n of ood pon bon o nd nio ind in yj. a pon n

nutrion level is then related to his/her individual requirement level to idenfy whether the person is malnourished

or adequately nourished according to the specic nutrion indicator. The individual requirement levels are calculated

based on the individual physiological nutrion needs using standard reference levels from WHO and FAO sources

(FAO/WHO/UNU 2001, WHO/FAO 2004, WHO/FAO 2006, WHO 2006). Finally, based on the persons new nutrional

status, we calculate the new prevalence rates of nutrional deciencies and the new number of malnourished people

in the countrys total populaon and the populaon of specic groups of individuals. For Yemen, the nutrion assess-

ment methodology is described in detail in Ecker et al. (2010) and Breisinger et al. (2010), and for Malawi in Ecker and

Qaim (2011).

3 See next secon for the juscaon of the choice of the nutrion indicators.


16/44


4. Impact of Economic Policies on Ntrition: Lessons from Case StdiesThe two case study countries are low income countries with high levels of malnutrion. Yemen represents an

oil-based economy with a relavely small agricultural sector and Malawi an agriculture-based economy with

limited diversicaon. While levels of malnutrion in both countries are among the highest in the world, the

nature of the nutrion challenges dier and span a broad range of nutrion-related issues that are typical for

many other developing countries, parcularly those in the Middle East and North African region and Sub-Saha-

ran Africa. Therefore, in addion to providing country-specic policy analyses, the results of the case studies arexpd o poid ipon ini n fo bod n of oni.

4.1 Te Ntrition Prolem in Yemen and Malai: A Comparatie Oerie

Food insecurity and malnutrion is widespread in both Yemen and Malawi. IFPRIs Global Hunger Index (GHI)

considers Yemens food insecurity situaon as alarming and ranks it at 73 of 84 analyed developing coun-

tries (von Grebmer et al. 2010), making it the most food insecure country in the Arab world. Although Yemens

naonal GDP is more than three mes higher than Malawis, Yemens food security situaon is worse, according

to the GHI.4 The GHI ranks Malawi 48th, which classies its food insecurity situaon as serious.

As a consequence of the recent food and nancial crises, the proporon of people suering from calorie de-

ciency in Yemen has increased by 3 percentage points from 2006 to 32 percent in 2009. This means that about

7.5 million Yemeni do not have enough food to be free from hunger. Calorie deciency is parcularly wide-spread in rural areas, where 37 percent of the populaon was decient in 2009, compared with 18 percent in

urban areas. The absolute number of calorie-decient people living in rural areas is more than ve mes higher

than in urban areas6.4 million rural people versus 1.1 million urban people in 2009 (Ecker et al. 2010). In Ma-

lawi, calorie deciency is also more prevalent in rural areas, but the dierences are less pronounced. Ecker and

Qaim (2011) report that 35 percent of the populaon was decient in calories in 2005, though the proporon

is expected to have declined in recent years due to Malawis impressive agricultural growth (discussed further

below). In 2004, 36 percent of the rural populaon, or about 4 million, were decient in calories, while the pro-

poron of calorie-decient people in urban areas was 25 percent, equivalent to about 0.4 million people. Thus,

calorie deciency is largely a rural problem in Malawi, where only 12 percent of the total populaon lives urban

areas or cies (Government of Malawi and World Bank 2007).

Diet quality is lower in Malawi than in Yemen. Using data from the food consumpon modules of the latest

household surveys from 2004/05 for Malawi and 2005/06 for Yemen, comparison of dietary diversity suggests

that the risk of micronutrient deciencies is higher in Malawi than in Yemen. The Food Variety Score (FVS)

calculated as the count of dierent food items consumed over seven daysequals 15.9 in Malawi and 17.3 in

Yemen (Ecker and Qaim 2011, Ecker et al. 2010). A more diversied diet typically contains more vegetables,

fruits, and animal products and therefore higher (bioavailable) amounts of essenal vitamins and minerals than

a staple food-dominated diet. Parcularly when animal-source foods are scarce in the diet, low consumpon

of vegetables and fruits is oen the main cause of micronutrient malnutrion (Ruel et al. 2005). In 2005, the

average Malawian household consumed only 50 grams of animal products (including meat, sh, eggs, and dairy

products) per capita per day, and the average consumed quanty of vegetables and fruits was less than 200

grams per capita per day (Ecker and Qaim 2011)signicantly less than half of the minimum recommendedintake of 400 grams for a healthy diet (WHO/FAO 2003). Accordingly, Ecker and Qaim (2011) esmate that 47

percent of the Malawian populaon suered from iron deciency in 2005, 55 percent from inc deciency, 66

percent from vitamin A deciency, and 37 percent from folate deciency.5

In contrast, child malnutrion, and parcularly the severe forms of its manifestaons, is by far more wide-

spread in Yemen than Malawi. In Yemen, 58 percent of all children under ve were stunted, 43 percent under-

weight, and 16 percent wasted in 2006 (Ecker et al. 2010); while in Malawi, 48 percent of all children of the

same age group were stunted, 22 percent underweight, and 5 percent wasted (NSO and ORC Macro 2005). Tak-

ing child stunng as an example, 35 percent of the Yemeni children were severely stunted, while the prevalence

rate of the severe form was less than half of that (16 percent) in Malawi (Ecker et al. 2010, NSO and ORC Macro

2005). Moreover, unlike in Malawi, childrens nutrional status in Yemen is considerably worse relave to the

4 In 2009, Yemens GDP equaled US$ 565 per capita and Malawis GDP US$ 166 per capita (at constant 2000 US$ prices) (WDI 2010).

5 Several previous studies (e.g., Hatloy et al. 1998, Steyn et al. 2006, Torheim et al. 2004) have shown that FVS is highly correlated with mineral andvitamin intakes and a good rst-cut indicator of micronutrient adequacy.


17/44


countrys economic development stage, as Figure 3 shows (in Secon 2). Malawi seems to reect the average in

the general relaonship between child malnutrion and naonal GDP.

Given these dierences in the nature of the nutrion problem, the Yemen case study uses the proporon of

stunted children under ve as one nutrion outcome indicator, whereas the Malawi case study looks parcular-

ly at micronutrient deciencies in the populaon at large. Stunng in children is idened by the height-for-age

measurea standard anthropometric indicator. Stunng, or gaining insucient height relave to age, reects a

process of failure to reach linear growth potenal as a result of long-term malnutrion or illness, or a combina-

on of both (WHO 1995). Thus, in the Yemen study, we parcularly consider the long-term nutrional statusof infants and young children, because they are most vulnerable to adverse nutrion (and health) condions.

The chosen micronutrients in the Malawi study are iron, inc, vitamin A, and folate. The deciencies of these

iin idpd oby nd pn io onqn fo pyi nd n dopn of

human beings and, by extension, a countrys economic and social development. They are also parcularly chal-

lenging to eradicate. In addion, in both countries we analye the impact on calorie deciency, which idenes

the primary dimension of the nutrion problem (i.e. lacking staple foods).6

4.2 Impact and Lessons from Policy Reforms Options in Yemen

The Yemeni economy is highly dependent on oil and has lile diversicaon. In 2009, the hydrocarbon sector

ond fo bo on-fo of gDP, nd oi xpo nd -fo of foin xn n-

ings and government revenues. By contrast, agricultures contribuon to GDP is about 10 percent, with about

30 percent of the rural populaon living o of farming. Moreover, agriculture is dominated by qat, i

uses about 40-50 percent of the countries dwindling water resources and which has negave side eects for

food security, including diverng household resources away from food and health expenditures.7 As in many

resource-rich developing countries, the manufacturing sector is relavely small and makes up about 10 percent

of the economy. Services is the biggest sector, constung about 50 percent of GDP. Economic growth has been

relavely low, averaging 3.8 percent over the past years and 0.8 percent per capita. Furthermore, growth did

not trickle down to food insecure people, especially those in rural areas (Breisinger et al. 2010).

to o nd o o po-food , poiy fo i ny ndd. t o-

ernment of Yemen has idened a set of policy acons to foster growth in both agricultural and non-oil sec-

tors. Key policies include a reformed agricultural policy that combines a reducon in qatwith the promoon ofalternave crops and their processing and markeng, and a policy that improves the business climate to aract

pi inn in nfin nd i o.

In the remainder of this secon, we analye the potenal eects of these policies on nutrion outcomes.

t o poiy nio dibd bo pn n i o non-i o

strategy, thus providing important insights into the debate on the role of agriculture for improving nutrion.

The results of both scenarios are presented against the counterfactual where no policy acon is taken. This is

referred to as a business-as-usual scenario, which serves as baseline against which growth eects and the

resulng nutrion impact will be measured.

4.2.1 A buSINESS-AS-uSuAL SCENARIO (NO POLICY ChANGE)

In the baseline scenario (which reects past growth paerns), Yemens economy connues to grow at 3.8 per-n nny fo 2010 o 2020, i nn o of 5.1 pn in nonydobon o nd

2.3 percent growth in the hydrocarbon sector, taking into account depleng oil resources. In 2009/2010, GDP

spiked briey due to the development of o-shore gas resources. Under this scenario, the share of the hydro-

bon o in onoy f fo bo 23 pn in 2009 o 12 pn in 2020.

In this baseline scenario, the proporon of calorie-decient people will decrease (Table 1). The prevalence

of calorie deciency will drop to pre-global food price crisis rates during the rst two years of the simulaon

6 Figure A5 in the Appendix shows the proporon of calorie-decient people by district in Yemen, Figure A6 the spaal distribuon of the calorie-decient populaon, and Figure A7 the proporon of stunted children under ve by governorate. Figure A8 gives a polical map of Malawi. Figure A9shows the proporon of calorie, iron, and inc-decient people by district in Malawi and Figure A9 the proporon of vitamin A and folate-decientpop.

7 The young, fresh leaves of the qat tree are narcoc and widely consumed as a smulant. Naonwide, 56 percent of males and 27 percent of femalesd n y nd od qat. Qatconsumpon poses serious health risks, and the treatment related illnesses increases public health expen-di. Qatchewing increases the risk of gum and stomach cancer (likely caused by the intake of pescides applied on the qatleaves) and can causepainful stomach sickness. The consumpon of the leaves is also likely to interfere with the absorpon of nutrients from food. Moreover, qatineases consumers appete, which might have negave eects on food preparaon and child feeding pracces; and it lower adults aenon and caregiven to their children (Ecker et al. 2010).


18/44


(2010 and 2011), and it will then slowly decline almost linearly. This decrease notwithstanding, the propor-

on of calorie-decient people will remain high, at 24 percent overall and 29 percent in rural areas by 2020. In

addion, due to connuing high populaon growth (of 3 percent per annum), the number of calorie-decient

people will increase to 7.8 million naonwide in 2020, up from 7.5 million in 2009. Consistent with historical

paerns, urban households will benet most from economic growth. The proporon of calorie-decient people

will decline at the lowest rates among the non-farm (landless) households, among whom poverty and malnutri-

on is most widespread (Ecker et al. 2010). Despite the modest naonal GDP growth and decline in calorie de-

ciency, child malnutrion will not improve markedly under a business-as-usual policy. The proporon of stuntedidn i din by n 2 pn poin o 62 pn in nd 46 pn in bn

by 2020.

Tabl 1Baslin scnari

2009 2015 2020

Grwth (prcnt)

Naonal GDP 6.6 3.9 3.6

hydobon 4.3 -2.3 -2.3

Nonydobon 7.3 5.2 4.5

ai 5.1 2.6 2.1

Indy 11.2 6.4 4.5

si 6.3 5.1 4.7

Malntrin

Proporon of calorie-decient people (percent) 32.1 25.3 24.3

r 37.3 31.0 29.7

F 33.4 26.7 25.4

Nonf 39.2 33.0 31.8

ubn 17.8 9.7 9.3

Number of calorie-decient people (in thousands) 7,483 7,038 7,831

Proporon of stunted children under ve (percent) 59.4 58.1 57.8

r 63.4 62.3 61.9

ubn 47.9 46.3 46.0

Source: cond by o.

4.2.2 AGRICuLTuRAL POLICY REFORM

Agricultural policy reform aims at accelerang agricultural growth and improving food security, thus increasing

agricultural output for rural income generaon. The reform plans take into account that the reducon ofqati

oy ind o non-qatagricultural development. This is because both expansion of agricultural producon

area and the increase of agricultural producvity are severely constrained by the availability of water.8 Current

annual water consumpon is almost 30 percent above the renewable water supply, and groundwater tables arefin pidy. tin ino on non-i , i ny ony pn 10 pn

of total use, is likely to dramacally increase in the future, water will be the single most important constraint for

i o in f.

Qatuses 40-50 percent of all water resources. Therefore, saving water in Yemen and using more water for

accelerang non-qatagricultural growth will only be possible through a signicant reducon ofqatonp-

on and producon. To this end, the agricultural policy reform scenario includes an excise tax on qat, onin

with the governments reform plans. In addion to the unsustainable use of groundwater for qatproducon,

several externalies ofqatconsumpon relevant for improving nutrion jusfy the implementaon of a qatx.

Byond di nd indi ipin, qatdi of ino y fo ood

could otherwise be spent on food and child nutrion.

8 Many households lack sucient water for drinking, washing, and cooking, and the available water is oen polluted. More than one-third of the popu-laon (37 percent) suers from water scarcity. And, only 41 percent of the populaon is connected to a water network (of whom three-fourths live inurban areas). Poor hygiene and a high risk of infecon with water-borne diseases are consequences (Ecker et al. 2010).


19/44


Consistent with the naonal plan, tax revenues are used to promote agricultural alternaves to qatpo-

ducon, including coee and cereal producon in parcular. Yemens potenal for boosng high-quality coee

producon and markeng is high (considering that the Mocha coee bean originates in Yemen). Components of

poiy ind xpndin i xnion i, inn in i nd d-

opment, introducon of improved seeds (especially drought-resistance cereals), investment in water harvest-

ing and irrigaon systems, and promoon of farmers unions and eecve processing and markeng systems.

These measures are assumed to increase agricultural producvity (TFP) by 5 percent annually in cereals, coee,

trade and related food processing sectors from 2011 to 2020through higher yields, more eecve input use,reduced post-harvest losses, and more ecient supply chains. To ease the introducon of the qatx, -

sume its implementaon in two steps. The inial tax is set at an equivalent of 14 percent ofqat in 2011

and 20 percent aerwards. The results of the simulaon are summaried in Table 2.

Tabl 2Agricltral plicy rfrmScnari smmary

2009 2015 2020

Grwth (prcnt)


hydobon 4.3 -2.3 -2.3


Malntrin




r 63.4 61.9 61.6

ubn 47.9 46.1 45.6

Source: cond by o.

The implementaon of a qatxi x in i pi fo on, i iy i o dnd.

ti i d o dd qatproducon, freeing up land and water resources for alternave use. Furthermore,

the tax will reduce the protability ofqatrelave to other crops, and its producon will aract less private

investment as a consequence. Alternave crops will become more compeve and more aracve for farmers

to grow. However, during the rst year of implementaon, the qattax will have a negave eect on agricultural

growth (Figure 6). Agriculture will grow by about one percentage point below the baseline level in the rst year.

Aer that, agricultural growth will accelerate and exceed baseline growth levels as a result of the producvity-

enhancing investments in the targeted sectors. Manufacturing and service sectors also will benet from agricul-

ture-led growth due to linkage eects and producvity-enhancing eects in processing and markeng sectors.

go in o biiz bo on pn poin bo bin , i

o i o 2.1 pn poin bo bin in 2020. t i poiy fo i

in gDP o in of bo 0.9 pn poin fo ond y of fo ond.

Figr 6Agricltral plicy rfrmGrwth chang frm baslin lvls (in prcntag pints)

Naonal GDP

2009-1

0

1

2

3

2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Hydrocarbon Agriculture Industry Services

Source: cond by o.


20/44


Figr 7Agricltr plicy rfrmChang f th prprn f calri-dcint ppl frm baslin

lvls (in prcntag pints)

-7

-6

-5

-4

-3

-2

-1

0

1 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Natonal

Rural farm

Rural nonfarm

Urban

Source: cond by o.

The income growth resulng from the reform will reduce the proporon of calorie-decient people naon-

wide and among farmers in parcular. In 2020, calorie deciency will aect 2 percent less of the populaon

than under a business-as-usual policy (22 percent versus 24 percent). The prevalence rate among farmers will

drop by almost 7 percentage points to 20 percent, but calorie deciency among non-farm households in rural

areas will barely decline, so that 30 percent of the rural nonfarm populaon will remain calorie-decient by

2020. The number of calorie-decient people in 2020 will be lower than in 2009 by about 400,000 people. In

addion, as with the business-as-usual scenario, childrens nutrional status will not notably improve under the

i poiy nio.

The benets of reduced qatconsumpon will go beyond the nutrional impacts translated through house-

hold income changes, as modeled in this simulaon. They might include lower cancer rates, lower public health

costs, and beer care given to the nutrion and health of children. Yet these impacts, as well as the result-

ing long-term consequences for economic and social development, are dicult to quanfy. Thus, the posive

ip of qatreducon policy on child nutrion might be underrated here. Finally, it is important to note that

ny fo din qatconsumpon will require a careful and credible communicaon strategy, given qatconsumpon is deeply rooted in the society, with inuenal policymakers and other key stakeholders

on i bi on.

4.2.3 PROMISING SECTOR GROwTh POLICY

To promote growth in manufacturing and service sectors, the Government of Yemen has declared the improve-

n of inn i i pioiy. r fo wod Bn Doin Bin poj -

gests that there is scope to do so; according to the 2010 report, Yemen ranks 99 out of 183 countries in creang

a favorable investment climate (World Bank 2010). More specically, several key indicators are signicantly

below the internaonal average. Although Yemen ranks relavely favorably (53) in terms of starng a business,

dealing with construcon permits (50), and registering property (50), there is much scope for improvement in

the provision of access to credit (ranking 150), investor protecon (132), taxaon (148), and cross-border tradeprocedure (120). The unfavorable investment climate is exacerbated by the tenuous security situaon, which

further discourages domesc and foreign investors. Improving the investment climate is widely expected to at-

tract addional investments, thus raising producvity and creang jobs. Creang jobs is especially important for

improving nutrion, given the high and growing unemployment rate and the fact that more than half of all rural

households and all food-insecure households draw their main income from o-farm employment.

Several sectors have been idened as having high potenal for aracng investment and accelerang

growth. Sectors are considered promising if they have high potenal to grow by at least 5-6 percent annu-

ally. They include (non-mineral) mining, food processing, tourism, and the transportaon and communicaon

service sectors (Breisinger et. al. 2010). While it is important to note that picking winner strategies have al-

most always failedgrowth in successful countries has mostly been driven by the private sectorgovernments

have always played an important indirect role by providing infrastructure and services important for iniang

the transformaon and growth process (Breisinger and Diao 2008). In this context, we design a scenario where

policy reform and investments improve the business climate and spark growth in promising sectors. Speci-


21/44


cally, we assume that producvity in the promising industry and service sectors will increase from 2 to 7 percent

between 2010 and 2020. We also assume that policy reform will aract FDI inows, which we assume to rise by

5 pn nny din onidd piod.

Table 3 shows that growth in promising sectors will signicantly contribute to naonal GDP growth. Driven

by o in o nd i in o o o, gDP o i in by o n

3 percent annually above baseline levels from 2013 onward (Figure 8). The nonhydrocarbon sectors will grow by

8.5 nny bn 2010 nd 2020, i i 3.6 pn o p y bn 2013 nd 2020 n nd

bin-- nio; o in indy nd i o i 10.6 nd 8.3 pn nn-

ally. Despite higher food demand resulng from higher household incomes and higher demand from the food

poin o, i o i in y onind by n o, o o

growth will accelerate only by about 1 percentage point above baseline growth (Figure 8).

Tabl 3Prmising sctr grwth plicyScnari smmary

2009 2015 2020

Grwth (prcnt)


hydobon 4.3 -2.3 -2.2


Malntrin




r 63.4 61.2 59.7

ubn 47.9 44.7 42.5

Source: cond by o.

This promising sectors-led growth translates into higher household incomes and substanally contributes to

improving nutrion as measured by calorie suciency, primarily in rural areas. Given that many promising sec-o inin, food poin, nd oi n ino nd poyn in nd

especially among the landless nonfarm households, among whom malnutrion is most widespread, this non-

agricultural growth has high benecial eects for reducing calorie deciency. Model results show that growth

in promising sectors reduces malnutrion among rural non-farm households by about 11 percentage points by

2020 compared to the baseline (Figure 9). Rural farm households also benet from this type of growth, mainly

due to the availability of o-farm employment, but also by the increase in agricultural growth driven by an

increased demand for their products from the rural and urban populaon. Growth in promising sectors will not

lead to similar reducon rates of calorie deciency among urban households, largely because the inial propor-

on of urban calorie-decient people is already low. This result is consistent with the cross-country empirical

evidence presented in Figures 1-4, which show that the less prevalent malnutrion becomes, the more dicult

it is to further reduce its spread by economic growth alone. The remaining malnourished are oen unable toparcipate in the labor market or do not possess the right skills to reap the benets of growth.


22/44


Figr 8Prmising sctr grwth plicyGrwth chang frm baslin lvls (in prcntag pints)

0

1

2

3

4

5

6

2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Naonal GDP

Hydrocarbon

Agriculture

Industry

Services

Source: cond by o.

Figr 9Prmising sctr grwth plicyChang f th prprn f calri-dcint ppl frm

baslin lvls (in prcntag pints)

-12

-10

-8

-6

-4

-2

02009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Natonal

Rural farm

Rural nonfarm

Urban

Source: cond by o.

It is important to keep in mind that, despite the sharp drop of the prevalence of calorie deciency in Ye-

n nd poiin o o poiy, nonf ood i in o

calorie-decient group. According to our simulaon, the proporon of calorie-decient people among the

nonfin ood i in 21 pn, opd o 14 pn on f nd 6 pn

among the urban populaon by 2020. Naonwide, the prevalence of calorie deciency will reach a rate of 9

pn poin bo nd bin-- nio, o of 15 pn in bo

.

Yet, the great improvement in nutrion in terms of calorie deciency is not matched by reducons in child

malnutrion. Results show that growth driven by non-agricultural sectors will only translate to a modest im-provement of childrens nutrional status. From 2009 to 2020, the proporon of child stunng will decline by

only about 4 percentage points naonwide and slightly above 5 percentage points in urban areas. Thus, child

malnutrion will remain at extremely high levels, with 55 percent of the under-5 children stunted by 2020.

4.2.4 COMPARING AGRICuLTuRE-LED GROwTh wITh NON-AGRICuLTuRE SECTOR-LED GROwTh

Empirical evidence has shown that the structure of growth and the mulplier eects of dierent sectors maers

for improving peoples nutrion. According to the simulaons in previous secons, the Yemeni economy grows

4.0 pn nd bin-- nio o piod 2010-2020, i i-d o

and promising sector-led growth accelerate overall growth to 4.9 percent and 6.8 percent, respecvely. To make

these scenarios comparable and to evaluate which policy is most eecve in improving nutrionagricultural

reform versus business climate improvementwe calculate the average annual reducon rate of the prevalenceof calorie deciency and child stunng in percentage points when subject to one percent GDP growth over the

simulaon period (Table 4).


23/44


Tabl 4Avrag annal rdcn f th prprn f calri-dcint ppl and stntd childrn ndr

v (prcntag pints) with n prcnt GDP grwth

Bsinss-as-sal

scenario

Agricltral plicy

reform

Promising sector

grwth plicy

Calorie deciency -0.176 -0.191 -0.224

Child stunng -0.036 -0.036 -0.055

Source: cond by o.

These esmates suggest that policies fostering growth in promising sectors (mining, food processing, tour-

ism, transportaon, and communicaon) reduce calorie deciency and child stunng more than agricultural

policy reform (accelerang agricultural growth through cereals and coee producon and processing and

dd qatproducon). An addional one percent annual growth in GDP driven by the promising sectors leads

to an average annual reducon of the proporon of calorie-decient people by 0.22 percentage points and of

the proporon of stunted children by 0.06 percentage points. The annual reducon rate for calorie deciency

i 21 pn i n nd bin-- poiy nd 15 pn i n nd i

policy reform scenario; for child stunng, the reducon rate under the promising sectors scenario is higher by

bo on-id n nd o o nio. conqny, poiy foin o in poiin in-

dustry and service sectors is more benecial for improving nutrion because the absolute impact is higher, and

the structure of growth enables the malnourished populaon to reap greater gains from the growth compared

o i o.

The lower nutrional impact of agricultural growth can be mainly explained by the role of Yemens agricul-

ture for income generaon and domesc food supply. Due to the water scarcity, the growth potenal of the

agricultural sector is limited. Even if agriculture grows, the eects on reducing malnutrion are limited because

the majority of the populaon draws their income from nonfarming acvies, and the nonfarming rural popu-

laonnot farmersare the most malnourished populaon group. In addion, most foods and parcularly

staple foods, which constute a large share of household budgets, are imported, thereby liming the potenal

consumer benet of decreased food prices that typically would accompany agricultural producvity growth.

That said, neither agriculture nor promising sector-led growth can signicantly reduce child malnutri-

on. Table 4 underlines the growth inelascity of child stunng. Reducon rates of the proporon of stuntedchildren account for only one-h to one-fourth of the reducon rates of the proporon of calorie-decient

people. Most of that is explained by the low elascity of household income to childrens anthropometrics. Ye-

ni idn i-foin by ony 0.09 pn i on pn in in ood

expenditure on average, with rural areas slightly lower than urban areas (Table A1 in the Appendix). This inelas-

city can be explained by a combinaon of at least three possible causes. First, since anthropometrics measure

a childs physical growth, there is a me lag between the appearance and the measurable impact of a nutrion

improvement. Second, and related to that, child anthropometrics, and especially stunng, is inuenced by the

nutrion history of a child as well as factors such as health condion, knowledge, and ingrained habits (see Sec-

on 3.1) that do not change with a marginal increase in income. Third, the distribuon of available resources

within the household maers. This might be of parcular importance in the case of Yemen with its male-domi-

nated society, where tradionally the (male) household head and the oldest son are seen as deserving the most

aenon. Moreover, the high consumpon ofqatnd oid o i y ony od

otherwise be invested in ensuring the opmal nutrion and health of young children. In fact, the elascies of

GDP to child nutrion in Yemen are signicantly below the internaonal average. Using cross-country regres-

sions, Headey (2011) nds an average elascity for the relaonship between GDP per capita and child stunng

of -0.12, which is more than two and three mes higher than our esmates, depending on the policy scenario.

Given the limited and insucient impacts of growth on nutrion, especially on child malnutrion, addi-

onal and more targeted measures are needed for reducing child malnutrion. These measures must include

awareness and educaon campaigns and targeted nutrion and health programs. Given the scale and the

nature of the nutrion problem in Yemen, these campaigns need to be naonal and at the highest polical level

(for example, as presidenal campaigns), accessible to the general public, and targeted toward the malnour-ished populaon. The lack of informaon and nutrion and health-related knowledge is especially high in rural

areas, where more than 80 percent of women and more than half of all men never aended school or dropped

out before compleng the primary educaon level, and where tradional customs are most deeply rooted


24/44


(Ecker et al. 2010).9 Key topics of these campaigns should be womens empowerment, child feeding pracces

(especially breaseeding) and household nutrion, hygiene and disease prevenon, qatconsumpon, and fam-

iy pnnin.

The evidence clearly shows that gender inequality goes hand in hand with malnutrion. Therefore, eorts

are needed to ensure womens empowerment as a means to improved nutrion. Iniaves to this end could

include improving womens educaonal aainment, economic parcipaon, health status, and social and poli-

pon. sin n ypiy ood diion, i i ipon o ino on

with women in any gender, nutrion, and health-related measure to gain their support. To achieve a large-scale,sustainable impact, a credible communicaons strategy is crucial. The strategy should provide widespread infor-

maon frequently throughout the country, involve local communies, and be backed by policymakers and other

inuenal people. The religious leaders, Friday prayers, and the local tradional leaders (sheiks) are important

venues for communicaon, beyond convenonal communicaon channels such as radio and television. Impact

through public awareness and educaon campaigns can be achieved at relavely low costs. Close collaboraon

among the government, non-governmental organiaons (NGOs), and internaonal donors also can help reduce

public expenditure, in addion to allowing for buying in of experse (Breisinger et al. 2010).

Intervenon programs must focus on the nutrion and health of pregnant and lactang women and chil-

dren under two years of age, spanning the window of opportunity during the life cycle (Victoria et al. 2008,

von Grebmer et al. 2010). Programs are parcularly needed for pregnancy and aer-birth care, birth assistance,

immuniaon, early child growth monitoring, and educaon on breaseeding, infant feeding pracce, hygiene,

and disease and illness prevenon and treatment (Breisinger et al. 2010). Launching these programs at eecve

scale requires signicant public investment, but the rates of return are high. Behrman et al. (2004) report that,

for example, the benet-cost rao for programs promong breaseeding in hospitals varies between 5 and 7

and for integrated child care programs between 9 and 16. Beyond the cost eecveness of these programs,

there is also a strong intrinsic argument for their populaon-wide implementaon. These programs provide

bi od b fy ib by ybody, indin x poo, nd y od

be paid from government revenues. To generate sucient revenues for nancing this public good, economic

growth is indispensible. While some of these measures are specic for Yemen (such as a qatreducon cam-

paign), most of these intervenons are crical in all countries with an economic growth-nutrion disconnecon.

4.3 Impact and Lessons from Agricltral Policies in Malai

The Malawian economy expanded at a modest rate between 1990 and 2005. Naonal GDP grew at an annual

average of 2.8 percent over this period (Benin et al. 2008). This is below the long-run populaon growth rate

of over 3 percent per annum, thus suggesng that per capita GDP levels declined over the period. The agricul-

ture sector performed beer, growing at 4.6 percent per year during the same period. However, agricultural

growth has been volale, with the agricultural sector contracng during four of the 15 years between 1990 and

2005 (Benin et al. 2008). This volality also aected the naonal GDP growth rate, as the agricultural sector has

historically contributed up to 40 percent to naonal GDP. Currently the sector contributes about one-third to

GDP (WDI 2010), which means the naonal growth rate is sll highly suscepble to growth uctuaons in the

agricultural sector. Growth in the agricultural sector, in turn, is strongly inuenced by the maie and tobacco

bo. miz onib on-q o i gDP nd i on oy by od fo on

consumpon. Tobacco is Malawis major export crop, contribung a further 14.6 percent to agricultural GDP

and accounng for almost one-third of export revenue. Combined, these two agricultural subsectors contribute

almost 15 percent to naonal GDP, which suggests that growth volality can oen be traced directly back to

volality in maie producon or tobacco export prices.

The volality of the agricultural sector can also be linked to the frequent occurrence of droughts and oods.

Less than 5 percent of the culvated land is irrigated, which makes agricultural output and food supply heavily

dependent on rainfall levels. Between 1990 and 2005 Malawi suered at least three severe droughts and four

major ood events (Pauw et al. 2010). During the widespread 1994 drought the agricultural sector contracted

by 29 pn. r poo obind i poo nn of in o din ood y

meant that Malawi has undergone at least two major food decits since the turn of the millennium, leading to a

famine in 2002 and a serious food emergency situaon in 2005 (Denning et al 2009; Devereux 2007).

9 In urban areas, about half of all women and about one-third have no primary educaon (Ecker et al. 2010).


25/44


In response to parcularly severe food supply problems experienced during the 2004/05 harvest, the

government of Malawi launched its Farm Input Subsidy Program (FISP) prior to the start of the 2005/06 grow-

ing season. FISP is a large-scale subsidy program aimed at lowering the cost of ferlier and modern maie seed

variees to resource-poor farmers. Despite legimate concerns about the scal sustainability of the program,

which costs about 5 percent of GDP and is considered one of the most ambious and expensive programs in

Malawis history (Bue & Atolia 2009), it has been lauded for its success in raising maie yields. Maie produc-

on almost tripled in the rst two years since incepon (MOAFS 2010). Because maie is a primary staple in

Malawiabout 60 percent of households calories are obtained from maie dishesthe strong growth in maieproducon is thought to have had a major impact on caloric intake (Ecker and Qaim 2011).

Dorward et al. (2008) aribute at least some of the gains to the favorable growing condions experienced in

the 2005/06 and 2006/07 seasons. Producon growth has nevertheless been remarkable, with output increas-

ing from about 1.2 to 3.2 million tons over three years. The most recent crop esmates for 2009/10 suggest that

naonal maie output remains at the same level achieved in 2006/07 (about 3.2 million tons of maie) (MOAFS

2010). Over the 2004/05 to 2009/10 period, the maie producon area increased only marginally, from 1.51 to

1.64 million hectares. Chibwana et al. (2010) argue that at least some of this increase in land allocaon resulted

from a displacement of other crops such as groundnuts, soybeans, and dry beans. Naonal data (MOAFS 2010)

, o, op in bn din y by on o in yid, i iz yid

increasing from as lile as 0.81 tons per hectare in 2004/05 (this was low compared to average maie yields of

around 1.2 tons per hectare achieved during the 1990s) to around two tons per hectare in 2009/10.

While the reliability of recent maie producon stascs has been quesoned by some observers, FISP

undeniably has had a major impact on agricultural and naonal growth. Preliminary esmates reported by the

Ministry of Development Planning and Cooperaon (MODPC) show that naonal GDP grew by about 7.9 per-

cent in 2006. The growth rate has increased steadily since, reaching 8.7 percent in 2009 (MODPC 2009).

Aer a period of rapid maie-led growth, Malawian policymakers are now grappling with the queson of

o o inin on. en if FIsP bd in in, i i niy iz yid o

rates, which averaged 15-20 percent per annum during 2004/05 to 2009/10 seasons, can be sustained. The

i o no f dn of nin o i on-n o p of bo 3-4 pn. ti

is well below the 6 percent target set under Comprehensive African Agricultural Development Program (CAADP)

of which Malawi is a co-signee and is insucient to ensure posive per capita GDP growth unless the nonag-ricultural sector grows strongly. That said, FISP has had some important economywide eects; nonagricultural

growth improved dramacally from historical levels of 3-4 percent (Benin et al. 2008), with early esmates

suggesng an average annual growth rate in excess of 7 percent during 2005-2010 (NSO 2010). Nonagricultural

growth now contributes signicantly to overall growth, signifying a steady decline in the importance of the

agriculture sector, a process that is likely to connue as the economy develops and modernies. Growth in the

construcon sector has been parcularly impressive (11.1 percent), which could indicate that at least some of

p nd din piod of iz-d o bn ind in onoy, i in n

od p o inin n i of noni o.

wi noni o i boin o ipon, onn od no o fo on

agricultural sector. Much of the sector remains primive in terms of farming methods, the level of mechania-

on, the use of modern inputs, and irrigaon; hence there is sll scope for rapid expansion in the long run. The

fo of i poiy, o, od b bodnd o ind o i bo. cn -

ricultural policy promotes maie producon through seed and ferlier subsidies with the aim of achieving food

self-suciency. It also focuses on tobacco, the countrys major export crop, through ferlier subsidies.

This narrow focus on maie and tobacco is potenally dangerous from both a growth and household wel-

fare perspecve, especially since the two crops represent such a signicant poron of agricultural (40 percent)

and naonal GDP (16 percent) (Benin et al. 2008). Malawi has limited cereals storage capacity. A connued

focus on maie will require the governments markeng agency to consider exporng maie (which unl now it

has been reluctant to do), not ideal with maie being a bulky, low-value crop. Furthermore, widespread adop-

on of ferlier subsidy programs elsewhere in southern and eastern Africa (Tanania, Kenya, zambia, and Mo-

ambique; see Sanche et al. 2009) means that trade opportunies in the region might be limited due to over-supply and weak prices. The tobacco sector is also under threat. At the World Health Organiaons Framework

Convenon on Tobacco Control in November 2010 in Uruguay, all members agreed to ban the use of addives

and sweeteners commonly used in blended burley tobacco products. Although no meline has been aached


26/44


o i n, mi by obo o, on of in od in of xpo o, i

set to suer major losses once the ban is put into eect (Diao et al. 2002, NKC 2010). A more broad-based agri-

cultural growth strategy will ensure that the economy is less vulnerable to economic shocks or policies aecng

the maie and/or tobacco sectors in parcular.

Agricultural policy should also ideally shi away from its current emphasis on the ferlier subsidy program

to more sustainable opons such as shoring up rural and agricultural infrastructure, providing eecve exten-

sion services, and spending on research and development. The growth potenal under FISP has arguably been

exhausted, having realied most of the gains in 2006 and 2007 shortly aer its introducon. While the policy iseecvely maintaining producon at its current high levels, it is unlikely to lead to further producvity gains.

tfo, io nd xpnion iz op o i iy o n. Oid of iz, o min

farmers connue to produce crops for own consumpon only, partly because of primive producon methods,

nd py b y nb o . ti nd o b ddd nd id i

poiy.

A broader spending focus is in fact provided for under Malawis Agriculture Sector Wide Approach (ASWAp),

which draws on elements of the Malawi Growth and Development Strategy (MGDS) in seng a priority invest-

ment strategy for the agricultural sector (MOAFS 2010). Through promoon of research and development,

piy bidin, nd i inf inn, aswap i o i 6 pn i

growth target of CAADP. About 40 percent of the ASWAp budget will be spent on technology generaon and

disseminaon (research and extension services), as well as instuonal strengthening and capacity building in

public and private sectors; the remainder is allocated directly to investment projects (MOAFS 2010). The largest

single component of ASWAp is the Greenbelt Iniave (GBI), a large-scale irrigaon scheme that also incorpo-

rates elements such as infrastructural development and rehabilitaon, land administraon and environmental

management, technology development and disseminaon, instuonal development and capacity building,

agro-processing and markeng development, and monitoring and evaluaon. The scheme is movated by the

fact that Malawi has abundant sources of unulied water, with Lake Malawi constung one of worlds largest

bodi of f . t o of gBI on i on fo o n 78 pn of aswap bd in

the rst year of operaon (started in June 2010). The budget allocaon declines in subsequent years, but will

sll contribute over 43% to the total cost of ASWAp over the rst four years of implementaon. In addion to

crop producvity gains brought about by improved irrigaon, the iniave also hopes to contribute to livestockand sheries producon, improve value chain linkages, and promote value addion through processing of raw

materials. While ambiousthe GBI budget will supposedly be double that of FISPit is a truly broad-based

agricultural iniave.

Most analysts agree that the ferlier subsidy has accelerated growth and that agriculture will remain

ipon fo o o in mi fo y o o. ho, i i no ip bidy

policy has had on nutrion and how agricultural growth may help to improve nutrion in the future. To shed

light on these quesons, we look at the nutrional impact of Malawis historical and possible future growth

paths. We rst simulate the countrys maie-led growth path during 2005-2010. We then consider two scenarios

fo 2010-2020. In bin nio onoy o on, i i nd non-

agricultural growth returning to the long-run growth trends idened by Benin et al. (2008). We compare these

results against a broad-based agricultural growth path in which cereals growth sll slows down as expected, but

overall agricultural growth is maintained through promoon of a larger range of subsectors under ASWAp and

the GBI. For the nutrion outcome indicators, we focus on the deciencies of calories and key micronutrients,

as discussed Secon 4.1.

4.3.1 PAST MAIzE-LED GROwTh PATh

mi o pfon fo 2005-2010 pn d ipon o pio on nd

half decades. Preliminary esmates of MODPC (2009) suggest that naonal growth averaged more than 8 per-

n, din y by on o in i, i in n din by iz yid o of bo 20

percent per annum (MOAFS 2010). Ocial gures also show a much improved nonagricultural growth perfor-

n, i o in bo 7 pn p nn bn 2005 nd 2010.

w Dcge od fo mi i b-y 2004 o oy pod i o p o

period 2005-2010, although we assume a slightly more conservave growth trajectory than what preliminary

naonal accounts esmates suggest. Thus, over the period 2005-2010, we allow naonal GDP to grow at 7.2


27/44


percent. This is driven largely by strong growth in the cereals subsector (16.5 percent), which allows agricul-

tural GDP to expand at 9.1 percent per annum. In reproducing naonal accounts growth

Growth is Good but Not Enough to Improve Nutrition

Documents

Transcript of Growth is Good but Not Enough to Improve Nutrition