A NIMAL S OURCED F OODS AND

C HILD S TUNTING
D EREK H EADEY , K ALLE H IRVONEN , AND J OHN H ODDINOTT

Stunting affects 160 million pre-school children globally with adverse life-long consequences. While
work within nutritional science suggests that stunting in early childhood is associated with low
intakes of animal-sourced foods (ASFs), this topic has received little attention from economists. We
attempt to redress this omission through an analysis of 130,432 children aged 6–23 months from 49
countries. We document distinctive patterns of ASF consumption among children in different
regions. We find evidence of strong associations between stunting and a generic ASF consumption
indicator, as well as dairy, meat/fish, and egg consumption indicators, and evidence that consuming
multiple ASFs is more advantageous than any single ASF. We explore why ASF consumption is low
but also so variable across countries. Non-tradable ASFs (fresh milk, eggs) are a very expensive
source of calories in low-income countries and caloric prices of these foods are strongly associated
with children’s consumption patterns. Other demand-side factors are also important, but the strong
influence of prices implies an important role for agricultural policies—in production, marketing and
trade—to improve the accessibility and affordability of ASFs in poorer countries.

Key words: Animal-sourced foods, dietary diversity, fisheries, livestock, nutrition, stunting.
JEL codes: I15, O15, Q18.

Approximately 160 million children under the causally linked to adverse later-life economic
age of 5 and living in developing countries are outcomes, including reduced schooling, poorer
considered chronically undernourished or cognitive skills, lower earnings and a higher
stunted (de Onis and Branca 2016). That is, likelihood of living in poverty (Hoddinott,
their height is at least two standard deviations Rosegrant, and Torero 2013).
less than the median height of a well-nourished Most stunting manifests in the first
child of the same age and sex. Nutritional status 1,000 days of life (Victora et al. 2010). In
is a component of health and thus reducing many countries children are born small, par-
chronic undernutrition has intrinsic value. In tially protected from infections and low nutri-
addition, chronic undernutrition in early life is ent diets by predominant breastfeeding in the
first few months after birth, but then subject
to accelerated growth faltering from
This article was invited by the President of the Agricultural & 6–24 months. Several things need to happen
Applied Economics Association for presentation at the 2018 an-
nual meeting of the Allied Social Sciences Association, after which in this 6–24-month period to ensure that
it was subjected to an expedited peer-review process. growth faltering does not take place. First,
Derek Headey and Kalle Hirvonen are with the International since breast milk can no longer provide all
Food Policy Research Institute (IFPRI); John Hoddinott is with nutrients required for healthy growth and de-
Cornell University. The authors thank seminar participants at the
Allied Social Science Association (ASSA) meetings in
velopment, starting around the age of
Philadelphia, January 2018, the editor, and three anonymous 6 months, children need to be introduced to
reviews for helpful comments and suggestions. This research was complementary foods that have to be
conducted under the “Advancing Research on Nutrition and
Agriculture (AReNA)” project, supported by The Bill & Melinda provided frequently (because of small gastric
Gates Foundation (Investment ID: OPP1112307) and the CGIAR capacity) and fed a nutrient-rich diet (or for-
Program on Agriculture for Nutrition and Health (A4NH).
Funding for this research was received from The Bill and Melinda tified complementary foods) to ensure that
Gates Foundation through the grant Advancing Research on they meet their nutrient requirements.
Nutrition and Agriculture (ARENA) Phase II (Investment ID: Second, because young children’s immune
OPP1177007). We thank Marie Ruel and two anonymous referees
for useful comments. The usual disclaimer applies. systems are in development, they need to be

Amer. J. Agr. Econ. 100(5): 1302–1319; doi: 10.1093/ajae/aay053

Published online July 31, 2018
C The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics
V
Association. This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any
medium, provided the original work is properly cited.
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1303

protected from both acute and chronic infec- is uniquely rich in calcium and its ability to
tions through preventative measures (e.g., stimulate the secretion of insulin-like growth
improved water, sanitation, hygiene, food factor I (IGF-I), a hormone that stimulates
safety, immunization) and optimal use of bone and tissue growth (Dror and Allen 2011).
health services for timely treatment. After Supporting this view is a smattering of
24 months, further growth faltering is less small-scale, somewhat idiosyncratic random-
commonplace as immune systems strengthen, ized control trials and observational studies.
but in developing country contexts there is Walker et al. (1991) provided a high-protein
also little evidence of systematic catch-up milk-based supplement to stunted children
growth in the first 5 years of life, and stunting aged 9–24 months in Jamaica, finding that the
tends to be highly persistent at a population growth in length was 0.89 cm higher in the
level (Leroy et al. 2014). treatment group. Likewise, several observa-
While there is broad agreement on the im- tional analyses of household surveys show
portant general interplay between diets, that children from cow-owning households
health, and care practices, thinking on the im- are taller than children whose households do
portance of specific nutrient dimensions of not own cows (Rawlins et al. 2014;
child undernutrition has evolved over time. Hoddinott, Headey, and Dereje 2015;
In the post-war period, substantial emphasis Choudhury and Headey 2017a) even after
was placed on protein deficiency as one of controlling for household wealth.1,2 Tang
the main causes of undernutrition in low-in- et al. (2014) conducted a cluster-randomized
come countries (Semba 2016), but an influen- controlled trial in China that provided a daily
tial article by McLaren (1974) argued that if a 60g-portion of minced pork to a treatment
child consumed an adequate amount of calo- group. The growth in length was 0.26 cm
ries, it was likely that they would also con- higher in the treatment group relative to the
sume adequate amounts of protein. control group who received rice. Iannotti
Coinciding with the 1974 global food (calo- et al. (2017) randomly assigned 6–9 month-
rie) crisis, McLaren’s arguments prompted a old children into treatment that provided
much greater emphasis on improving calorie one egg per day for a 6-month period and a
availability. This also suited the emphasis of control that followed a usual diet. The
economists and agronomists on increasing effects were large: length-for-age z score in-
the productivity of calorie-dense staple crops creased by 0.63 SD and stunting prevalence
through the Green Revolution. In recent decreased by 47% relative to the control
years, however, the 1970s revisionist view has group.
been called into question—there are three Finally, economic historians have been ex-
strands to this questioning. amining the links between access to ASFs,
The first comes from new work in metabolo- primarily meat and dairy, and adult stature.
mics (the study of small molecule chemicals Data from 19th century Bavaria show how
that result from metabolic processes) aimed at statures were greater in regions that had
understanding cellular growth processes. A rel- higher levels of per capita milk production
evant example is improvements in understand- (Baten and Murray 2000; Baten 2009).3
ing mTORs. These are catalytic proteins that
signal (or regulate) anabolic processes such as
cellular growth and differentiation. One type, 1
Still other studies review relationships between ASF intake
mechanistic target of rapamycin complex 1 and growth, but not among pre-school children, and only in a mix
of high- and low-income countries (Hoppe, Mølgaard, and
(mTORC1) regulates growth in chondral plates Michaelsen 2006; de Beer 2012).
(i.e., part of the bone where growth takes place) 2
Grasgruber et al. (2016) analyze data on adult male heights
and in skeletal muscle growth. Essential amino from 105 countries, which they link to dietary proxies from the
FAO Food Balance Sheets and genetic markers linked to lactose
acids are needed for the activation of mTORC1 intolerance. These authors show that adult male heights are most
(Semba 2016). These cannot be synthesized highly correlated with animal-sourced protein intake, which is in
turn strongly associated with more lactose-tolerant genetic
from scratch within the human body; they must profiles.
be obtained via diet; the best sources are animal 3
The absence of technology to refrigerate inhibited trade
source foods (ASFs) (meat, poultry, fish, and meant that local production and consumption of dairy were
highly correlated. In the United States, mechanical refrigeration
eggs). Plant sources also contain these, but typi- technology started to spread in the late nineteenth century, facili-
cally in much lower concentrations. In addition, tating market integration of perishable foods such as dairy
ASFs are dense in a wide range of micronu- (Goodwin, Grennes, and Craig 2002). Craig, Goodwin, and
Grennes (2004) estimate that the adoption of refrigeration post
trients linked to growth and cognitive develop- 1890s United States increased annual dairy consumption and an-
ment (iron, B12, choline, zinc), and cow’s milk nual protein intake by 1.7 % and 1.25 %, respectively. The
1304 October 2018 Amer. J. Agr. Econ.

Similarly, per capita cattle holdings are found an important role for agricultural policies—in
to be an important predictor of within- production, marketing and trade—to improve
country differences in female stature in post- the accessibility and affordability of ASFs in
colonial Africa (Moradi and Baten 2005). poorer countries.
Takahashi (1984) argues that the post-war
improvements in child heights in Japan are
strongly linked to the introduction of dairy Data and Methods
into the Japanese diet.
The common implication of these different
strands is that chronic undernutrition in de- We use multi-country household survey data
veloping countries may reflect low levels of from phases 5 and 6 of the Demographic
ASF intake. But the small-scale nature of the Health Surveys (DHS), covering surveys con-
few randomized control trials of ASFs, to- ducted between 2006 and 2014 (ICF
gether with the idiosyncratic nature of obser- International 2015). The DHS are widely
vational work, gives pause. To the best of our used in analyses of child nutrition, including
knowledge, there are no studies that system- the main dependent variables in our analysis,
atically assesses patterns of ASF consumption child stunting, which is defined as a height-for-
in the critical 6–24-month growth window age Z-score less than 2 standard deviations be-
across countries and regions, whether these low the World Health Organization (WHO;
consumption patterns are associated with 2006) growth standards. Since the mid-2000s
stunting, and what economic drivers underpin (phase 5), the DHS has implemented a highly
these patterns of ASF consumption. This pa- standardized dietary module in which mothers
per attempts to redress these knowledge gaps are asked to recall which of 14 disaggregated
by posing three objectives. food groups their youngest child (0–24 months
The first is descriptive. Using an unusually of age) consumed in the past 24 hours.4
large agglomeration of recent Demographic Table 1 lists these aggregated food groups
Health Surveys (DHS), encompassing along with the aggregations we use for our
130,432 children aged 6–23 months living in analysis. These data are available for 130,432
49 low- and middle-income countries, we doc- children aged 6–23 months from 49 countries in
ument complex regional patterns of ASF in- five regions, thus ensuring reasonable region-
take in 24-hour recall data. Second, we use specific sample sizes (see supplementary online
these data to conduct an age-disaggregated appendix table 1 for a full list of countries):
multivariate analysis of the associations be- Latin America & Caribbean (7 countries,
tween children’s food consumption patterns 29,387 children); Middle East & North Africa
and child height in models saturated with a (4 countries, 14,445 children); Central, South
host of relevant control variables. We assess & South-East Asia (9 countries, 24,721 chil-
the predicted decline in stunting due to con- dren); West & Central Africa (16 countries,
sumption of any ASF, multiple ASFs, specific 37,474 children); Eastern & Southern Africa
ASF groups (dairy, eggs, meat/fish), as well (13 countries, 24,405 children).
as regional variation in ASF-stunting rela- While the large sample sizes and broad
tionships. Third, we explore why ASF con- geographic coverage of these data are consid-
sumption is low, but also so variable across erable strengths, there are limitations. First,
countries and regions, even those at similar we only know whether a child consumed
levels of income. We link DHS data to na- some amount of these food groups in the past
tional level data on “cereal-relative caloric 24 hours. We do not have data on quantities,
prices” that tell us how expensive a source of nor do we know the frequency or duration of
calories each ASF is relative to the cheapest consumption. Second, where there are multi-
staple cereal in each country. Combined, ple children aged less than 24 months in the
these data can shed light on why consumption household, the DHS only collects dietary
of ASFs vary across countries and house- data for the youngest child. The youngest
holds. This analysis provides a potential entry children within this age range are not rele-
point into this issue for agricultural econo- vant to our analysis since: (a) children
mists because high relative ASF prices imply 0–5 months are (or should be) exclusively

4
We do note one idiosyncrasy: Peruvian DHS do not provide
authors further estimate that a large fraction of the improvement any disaggregated data on flesh foods, with fish and meat/organs
in adult statures in 1890s were due to refrigeration. aggregated together.
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1305

Table 1. Aggregating Food Groups in Phases 5 and 6 of the Demographic Health Surveys
Food Groups Found in DHS Data Aggregated Food Groups Aggregated Food Groups
Used in This Analysis Used in This Analysis with
ASFs Disaggregated
(1) Grains Starchy staples: (1) Grains, Starchy staples: (1) Grains,
(2) Roots/tuber (2) Roots/tuber
(2) Roots/tubers
(3) Legumes/nuts Legumes/nuts Legumes/nuts
(4) Vitamin A rich fruits Any fruit: (4) Vitamin A rich Any fruit: (4) Vitamin A rich
fruits; (6) Other fruits fruits; (6) Other fruits
(5) Vitamin A rich vegetables Any vegetables: (5) Vitamin Any vegetables: (5) Vitamin
A rich vegetables; (7) Dark A rich vegetables; (7) Dark
green leafy vegetables; (8) green leafy vegetables; (8)
Other vegetables Other vegetables
(6) Other fruits
(7) Dark green leafy vegetables,
(8) Other vegetables
(9) Fresh/powdered cow’s milk Any ASF: (9) Fresh/powdered Dairy: (9) Fresh/powdered
cow’s milk; (10) Infant cow’s milk; (10) Infant
formula; (11) Eggs; (12) formula;
Meat/organs; (13) Fish
(10) Infant formula
(11) Eggs Eggs
(12) Meat/organs Meat/fish: (12) Meat/organs;
(13) Fish
(13) Fish
(14) Fortified infant cereals

breastfed; and (b) one would not expect to the full set of controls listed in table 2—mostly
see instantaneous growth benefits from a re- drawn from (Danaei et al. 2016).5 In addition
cent introduction of complementary foods to controlling for child age (through age-in-
(e.g., in the 6–11-month range). Instead, the month dummies) and sex, our selection of con-
benefits of an improved diet would be trol variables is informed by the notion of a nu-
expected to accumulate throughout the first trition production function (Behrman and
two years of life, consistent with the growth- Deolalikar 1988) including: nutrients (the food
faltering patterns observed in Victora et al. groups described above); illness (diarrhea); ma-
(2010). Indeed, one recent analysis strongly ternal knowledge of good nutrition (maternal
suggests that analyses of the determinants of education); the health environment (the sanita-
child growth should typically focus on chil- tion and water variables); and genetic endow-
dren who have completed the first two years ments (maternal height measured in cm).6,7 We
of life (Alderman and Headey 2018). For this control for whether the mother was a teenager
reason, we focus on children aged (<20 years) when the child was born, which
6–23 months, but also disaggregate within this captures aspects of both maternal knowledge
age group to focus on 6–11, 12–17 and 18–23-
month age brackets. Consistent with
Alderman and Headey (2018), our expecta- 5
Our results are robust to using a probit model instead of lin-
tion is that the associations between dietary ear probability model.
6
Diarrhea can also be considered as an outcome variable in
indicators and stunting is strongest for the the sense that ASF consumption could increase or decrease diar-
18–23-month group. rhea risk among young children. However, our results are robust
to omitting this variable from the model.
A bivariate analysis of consumption of ASFs 7
Assets (or incomes) do not typically directly enter the nutri-
and stunting carries with it the obvious risks as- tion production function as they are thought to influence nutri-
sociated with the presence of confounding fac- tion outcomes indirectly through various inputs. Still, a
consistent empirical finding in the literature is that ASFs have
tors. Many factors that might determine ASF high-income elasticities (e.g., Pinstrup-Andersen and Caicedo
consumption might also affect nutritional status 1978; Colen et al. 2018), raising a concern that our ASF coeffi-
through other channels. For this reason, we es- cients are picking up wealth effects. Our main specification does
not include household assets, but our results are robust to their
timate linear probability models that include inclusion (see table S3 in the supplementary online appendix).
1306 October 2018 Amer. J. Agr. Econ.

Table 2. Summary Statistics for Key Variables

Variable Obs. Mean Std. Dev.
Stunting (0/1) 130,432 0.32 0.46
Grain/root/tuber (0/1) 130,432 0.76 0.43
Legume/nut (0/1) 130,432 0.25 0.43
Fruits (0/1) 130,432 0.36 0.48
Vegetables (0/1) 130,432 0.32 0.47
ASFs (0/1) 130,432 0.62 0.49
Dairy (0/1) 130,432 0.35 0.48
Eggs (0/1) 130,432 0.22 0.42
Meat/fish (0/1) 130,432 0.38 0.49
Red/white meat (0/1) 130,432 0.24 0.43
Fish* (0/1) 117,626 0.20 0.40
Fortified infant cereals (0/1) 130,432 0.10 0.30
Currently breastfed (0/1) 130,432 0.77 0.42
Infant formula (0/1) 130,432 0.10 0.30
No food yesterday (0/1) 130,432 0.12 0.33
Diarrhea (0/1) 130,315 0.23 0.42
Medical facility birth (0/1) 130,432 0.61 0.49
Improved latrine (0/1) 130,432 0.30 0.46
Unimproved toilet (0/1) 130,432 0.42 0.49
Improved water (0/1) 130,432 0.63 0.48
Teenage mother (0/1) 130,432 0.18 0.39
Mother’s height (cm) 130,432 155.8 7.15
Woman 9þ yrs education (0/1) 130,432 0.33 0.47
Partner 9þ yrs education (0/1) 130,432 0.38 0.48
Partner absent (0/1) 130,432 0.11 0.31
Child is boy (0/1) 130,432 0.51 0.50
Age (months) 130,432 14.2 5.13
Rural (0/1) 130,432 0.65 0.48
Note: These are DHS data from 49 countries listed in supplementary online appendix table 1. Asterisk * indicates fish consumption data are not available for
Peru.

of child care and nutrition practices as well as Headey et al. (2017). The underlying price
child endowments. We include survey dum- data pertain to nationally averaged prices of
mies, meaning that the regression coefficients 200 standardized food products collected by
represent within-survey associations with stunt- national statistical agencies for 177 countries
ing. We further control for subnational fixed under the auspices of The International
effects (states, provinces, et cetera) and Comparison Program (ICP) coordinated by
whether the child resides in a rural area. The the World Bank. These weight-based prices
inclusion of geographical fixed effects allows us were then converted to prices per calorie.
to capture, to a certain extent, regional patterns Specific foods were then allocated into
in food access, prices, incomes, health, and the groups (see table 3), and in each country the
health environment. While these control varia- cheapest staple cereal was selected as a
bles capture some aspects of confounding fac- numeraire. CCPs were then constructed by
tors, it would be a mistake to assume that they taking the ratio of the cheapest food in each
capture all inputs into the nutrition production group to the cheapest staple cereal in a coun-
function and for this reason we refer to our try. As an example, the CCP for eggs in
results as associations, not causal estimates. Bangladesh is the ratio of the price of 1 calo-
Our regression results are not population- rie of an egg relative to 1 calorie of rice.
weighted, but the coefficients do reflect implicit Table 3 indicates that most ASF categories
weights based on sample size and on within- involve good coverage of a range of specific
survey variation in stunting and dietary food items, although apart from tilapia, there
variables. is insufficient coverage of local varieties of
In the final part of the analysis, we supple- freshwater fish. Coverage of fresh and proc-
ment DHS data with national level data on essed dairy products is good (6 products), as
cereal-relative calorie prices (CCPs) from is eggs (medium and large).
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1307

Table 3. Classification of Cereals and

All Children

Note: Data pertain to children 6–23 months of age from DHS surveys from 49 countries. See supplementary online appendix table 1 for a list of countries and sample sizes. Asterisk * indicates that fish consumption data are not available for
18–23 mo
Specific ASF Products in the ICP 2011 Data

41.3%

69.8%
35.7%
24.0%
11.8%
40.0%
27.0%
46.3%
30.4%
24.0%
Food Group Number of Specific Products Used
products to Construct Minimum
Price

All Children
Cereals 13 Rice (5 types), bread

12–17 mo
products (5 types),

9.4%
32.8%

66.4%
37.0%
22.2%

37.4%
24.1%
41.8%
26.8%
21.8%
maize flour, maize,
tortilla
Cow’s milk, 2 Pasteurized fresh milk,
fresh unskimmed or low-fat
Cow’s milk, 3 Condensed milk,

All Children
6–11 mo
long-life powdered milk, UHT

21.6%

50.7%
34.6%
15.0%

29.2%
16.6%
26.4%
16.5%
13.7%
4.8%
Meat, fresh 20 Whole chicken (2
types), chicken
breast, chicken
leg; Beef/veal (7
varieties), Lamb/

& Southern

(6–23 mo)
mutton (4 varieties),

Eastern

Africa

3.2%
37.4%

49.3%
36.4%
12.0%

18.6%
13.1%
33.7%
17.4%
21.0%
Pork (4 varieties),
Goat (1 variety); all
unprocessed.
Chicken eggs, 2 Large brown eggs,
fresh medium brown eggs
(6–23 mo)
& Central

Fish, fresh 5 Fresh Carp, Mackerel or

Africa

4.4%
32.9%

52.0%
35.4%
13.5%

20.6%
12.1%
39.4%
15.4%
31.3%
West

Tilapia; canned
Sardines or canned
Tuna
Source: Headey et al. (2017).
& SE Asia
(6–23 mo)
Central
South,

4.3%
37.7%

56.8%
43.7%
13.8%

37.7%
15.5%
22.7%
13.2%
12.6%
Table 4. Stunting and Dietary Patterns by Region and Child’s Age

We use these national-level price data in

linear probability regressions where con-
sumption of individual ASFs is modelled as a
Middle East

(6–23 mo)
& North

function of child and household characteris-

Africa

25.9%

75.5%
38.2%
30.3%
11.0%
64.0%
30.2%
30.1%
23.6%
7.6%
tics from the DHS—including important pre-
dictors of ASF consumption such as
household wealth and parental education—
and national level prices (CCPs). The inclu-
America &
Caribbean
(6–23 mo)

sion of time-invariant prices for 2011 means

22.5%

82.9%
27.8%
36.2%
20.6%
52.1%
45.3%
55.9%
51.3%
Latin

that these regressions necessarily omit sur-

N/A

vey/country dummies, but we include a series

of additional national-level control variables
that might influence ASF prices and simulta-
All Children

neously determine ASF consumption pat-

(6–23 mo)

terns through other mechanisms. These

31.5%

62.0%
35.8%
20.3%

35.4%
22.4%
37.9%
24.3%
19.6%
8.5%

variables are sourced from the World Bank

and include GDP per capita, cereal yields,
and a conflict dummy variable.
Red/white meat
Child consumed:

1 ASF only

Dietary Patterns among Young Children

Meat/fish *
2 ASFs
3 ASFs
Any ASF

Fish
Dairy

Table 4 reports nutrition and dietary indica-

Stunted

Eggs

tors for children 6–23 months. Stunting rates

Peru.

are high at 31.5%. The regional

1308 October 2018 Amer. J. Agr. Econ.

disaggregation shows that stunting is much products to be introduced only at 12 months

higher in the two sub-Saharan African because of potentially adverse side effects on
regions and Asia—where roughly one-third fragile infant digestive systems (FAO 2013),
of children in this age range are stunted— in this sample about 20% of children
than in the two predominantly middle- 6 months of age consumed dairy products on
income regions. Consumption of ASFs fol- the previous day. This rises to 40% by
lows a similar pattern. While 61.8% of 18 months. The consumption of eggs, meat,
children consumed at least one ASF the pre- and fish are all less common than dairy and
vious day, this masks considerable heteroge- are characterized by similar levels of con-
neity across regions. Latin America and the sumption at all ages.
Caribbean region, with the highest propor-
tion of children consuming an ASF (82.9%),
has the lowest prevalence of stunting Associations between Stunting and Animal
(22.35%), while the Eastern and Southern Sourced Foods
Africa region has the lowest proportion of
children consuming an ASF (49.3%) and the Our parametric results are reported in table 5.
highest prevalence of stunting (37.4%). The control variables provided above (includ-
Table 4 also shows that consumption rates ing the sub-national fixed effects) are in-
among different ASFs are broadly similar in cluded in these regressions but their
the Latin America and Caribbean region, coefficients are not reported (see supplemen-
with roughly half of children consuming tary online appendix table 2 for full results).
dairy, meat/fish or eggs in the past 24 hours. In the first column, we look at the association
Moreover, 36.2% of children in this region between the consumption of any ASF and
consumed two ASFs in the past 24 hours, stunting. The coefficient is statistically signifi-
while 20.6% consumed all three ASFs. In cant and negative, and its magnitude implies
most other regions, one ASF tends to domi- that, on average, intake of an ASF is associ-
nate. Dairy dominated in the Middle East ated a 2.3-percentage point reduction in
and North Africa region (64.0%), although a stunting rate. Considering the overall stunt-
substantial share of children consumes multi- ing rate of 31.5% in this age-range, a 2.3-per-
ple ASFs there. Dairy also dominates in the centage points translates into a 7.2%
South, Central and South-East Asia region reduction in stunting. When we disaggregate
(37.7%), although in some countries fish is the data by child age (columns 2, 3, and 4),
more important (Bangladesh). Fish is also we see the expected result that the associa-
dominant in the West and Central Africa re- tions tend to strengthen for older age brack-
gion (31.3%), but no particular ASF is domi- ets. For children aged 18–23 months—our
nant in the Eastern and Southern Africa preferred sample—consumption of an ASF
region. It is also notable that relatively few the previous day is associated with a four-
children in these poorer regions consume percentage point reduction in the likelihood
multiple ASFs (<20%). of being stunted. Since the stunting rate
To contextualize age-specific patterns of among children in this age range is 41.3%
stunting and consumption of ASFs, figure 1 (see table 4), this translates into 9.7% reduc-
plots stunting against child age for 49 tion in stunting. Note that this association is
countries. Just after birth, roughly 18% of the larger than that observed for other non-staple
sample is stunted, but this changes little for foods. This includes legumes—an important
the first few months of life when most infants alternative source of protein, especially in
are predominantly breastfed. From 4– low-income countries (Messina 1999).
5 months onwards, stunting rates increase Moreover, the ASF and legume coefficients
precipitously until about 20 months of age are statistically different from zero.8 We also
where they level off at just over 40%. note that the coefficient on fruit is significant
Regional data show similar patterns of
accelerated growth faltering (supplementary
online appendix figure 1). Figure 2 plots con-
sumption of any ASF and some specific ASFs 8
We can reject the null hypothesis that the coefficients on
by child age. Dairy is the most commonly legumes and ASFs are equal. Consistent with this, adding con-
consumed ASF in this sample. While the sumption of legumes to our ASF variable (thus creating a vari-
able capturing the consumption of proteins from animal or plant
Food and Agriculture Organisation of the sources) yields a coefficient slightly smaller than the coefficient
United Nations (FAO) recommends dairy on ASFs. Results are available upon request.
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1309

Figure 1. Stunting prevalence by child age for children from 49 countries

Note: These are local polynomial plots of stunting against child age, with 90% confidence intervals, for 296,370 children aged 0–59 months

Figure 2. Consumption of any ASF and specific ASFs, by child age

Note: These are local polynomial plots of ASF consumption indicators against child age, with 90% confidence intervals, for 147,521 children aged 0–23
months

and negative in all samples, although it is In table 6, we re-run column 4 in table 5 on

again statistically smaller than the coefficient region-disaggregated data. The coefficient on
on ASFs in the 18–23-month sample.9 the ASF consumption variable remains nega-
tive in all sub-samples and is statistically sig-
nificant at the 5% level in four out five
9
The significant negative coefficient on fruit consumption is regions. Only in the Middle East & North
biologically plausible. Fruits are soft and highly palatable for Africa (MENA) region is the association be-
young children and require very little preparation time, implying
that fruit consumption might be associated with higher calorie in-
tween ASF intake and stunting not statisti-
take. Fruits are also rich in a wide array of micronutrients that cally significant. Elsewhere, consumption of
are beneficial for immune system strengthening. However, while at least one ASF is associated with a 2–9 per-
fruit consumption has an association with stunting, the coefficient
tends to be smaller than the ASF one (see table 5) and, unlike centage point reduction in the likelihood of
ASF, is not consistent across regions (see table 6). being stunted, with the strongest association
1310 October 2018 Amer. J. Agr. Econ.

Table 5. ASFs Are Negatively Associated with Stunting, Especially in the 18–23-month Age
Group
(1) (2) (3) (4)
Sample: 6–23 Month 6–11 Month 12–17 Month 18–23 Month
Children Children Children Children
Any ASF 0.023*** 0.016*** 0.011** 0.040***
(0.003) (0.005) (0.006) (0.006)
Grains/roots/tubers 0.015*** 0.023*** 0.001 0.013
(0.004) (0.006) (0.008) (0.009)
Legumes/nuts 0.006** 0.007 0.006 0.003
(0.003) (0.005) (0.005) (0.006)
Any fruit 0.019*** 0.015*** 0.014*** 0.018***
(0.003) (0.005) (0.005) (0.005)
Any vegetables 0.003 0.006 0.001 0.004
(0.003) (0.005) (0.005) (0.005)
Child and household level controls? Yes Yes Yes Yes
Rural dummy? Yes Yes Yes Yes
Survey dummies? Yes Yes Yes Yes
Sub-national fixed effects? Yes Yes Yes Yes
Coefficient on ASF is statistically different Yes No No Yes
to the coefficients on other food groups? (except fruit)
R-squared 0.149 0.104 0.141 0.167
Observations 130,315 45,316 45,026 39,973
Note: Unit of observation is a child and the dependent variable obtains a value 1 if the child is stunted, and zero otherwise. Standard errors clustered at the
DHS cluster level. Child and household level controls include: Child had a diarrhea (0/1), Medical facility birth (0/1), Improved latrine (0/1), Unimproved toi-
let (0/1), Improved water (0/1), Teenage mother (0/1), Mother’s height (cm), Woman 9þ yrs education (0/1), Partner 9þ yrs education (0/1), Partner absent
(0/1), child is boy (0/1), and child’s age (age-in-months dummies). Also included are survey dummies, subnational fixed effects (states, provinces, et cetera)
and whether the child resides in a rural area. Statistical significance denoted as ***¼ p < 0.01, **¼ p < 0.05, and *¼ p < 0.1.

Table 6. ASF Consumption Is Associated with Reduced Risks of Stunting in Children 18–
23 Months of Age in All Regions Except the Middle East and North Africa
(1) (2) (3) (4) (5)
Sample: Latin America Middle East & South, Central West & Eastern &
& Caribbean North Africa & SE Asia Central Africa Southern Africa
Any ASF 0.088*** 0.018 0.042*** 0.022** 0.045***
(0.017) (0.026) (0.013) (0.011) (0.013)
Grains/roots/tubers 0.039* 0.046* 0.015 0.024 0.004
(0.023) (0.027) (0.022) (0.017) (0.018)
Legumes/nuts 0.000 0.010 0.032** 0.012 0.007
(0.009) (0.017) (0.014) (0.012) (0.014)
Any fruit 0.013 0.026 0.028** 0.005 0.031**
(0.010) (0.017) (0.012) (0.011) (0.013)
Any vegetables 0.001 0.039** 0.002 0.002 0.004
(0.010) (0.018) (0.012) (0.010) (0.013)
Other food groups? Yes Yes Yes Yes Yes
Child and household Yes Yes Yes Yes Yes
level controls?
Rural dummy? Yes Yes Yes Yes Yes
Survey dummies? Yes Yes Yes Yes Yes
Sub-national Yes Yes Yes Yes Yes
fixed effects?
R-squared 0.223 0.138 0.171 0.123 0.104
Observations 9,507 3,857 7,831 10,662 7,480
Note: Unit of observation is a child, 18–23 months and the dependent variable obtains a value 1 if the child is stunted, and zero otherwise. Standard errors
clustered at the DHS cluster level. Controls are those listed in table 5. Statistical significance denoted as ***¼ p < 0.01, **¼ p < 0.05, and *¼ p < 0.1.
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1311

Table 7. Consuming More than 1 Type of ASFs is Associated with a Significantly Lower Risk
of Stunting in the 18–23-Month Age Group
(1) (2) (3) (4)
6–23 Months 6–11 Months 12–17 Months 18–23 Months
1 ASF 0.018*** 0.017*** 0.010* 0.037***
(0.003) (0.005) (0.006) (0.007)
2 ASFs 0.034*** 0.021*** 0.009 0.057***
(0.004) (0.007) (0.007) (0.008)
3 ASFs 0.045*** 0.009 0.014 0.061***
(0.005) (0.010) (0.009) (0.010)
Other food groups? Yes Yes Yes Yes
Child and household level controls? Yes Yes Yes Yes
Rural dummy? Yes Yes Yes Yes
Survey dummies? Yes Yes Yes Yes
Sub-national fixed effects? Yes Yes Yes Yes
P-value of H0: b(1 ASF) ¼ b(2 ASFs) 0.000*** 0.416 0.852 0.002***
P-value of H0: b(1 ASF) ¼ b(3 ASFs) 0.000*** 0.422 0.593 0.003***
P-value of H0: b(2 ASF) ¼ b(3 ASFs) 0.024** 0.182 0.512 0.762
R-squared 0.150 0.105 0.142 0.169
Observations 130,315 45,316 45,026 39,973
Note: Unit of observation is a child and the dependent variable obtains a value 1 if the child is stunted, and zero otherwise. Standard errors clustered at the
DHS cluster level. Controls are those listed in table 5. Statistical significance denoted at ***¼ p < 0.01, **¼ p < 0.05, and *¼ p < 0.1.

pertaining to the Latin America and consuming one additional ASF is associated
Caribbean sample, where ASF consumption with a further 2-percentage point reduction
is both high and diverse (table 4). (p < 0.01). Consuming three ASFs also
Tables 7 and 8 extend this core finding in implies a large reduction in stunting (6.1
two ways. Table 7 assesses whether the num- points), but the implied difference between
ber of different ASFs consumed is associated 2 ASFs and 3 ASFs is not statistically
with stunting. Biologically, one would expect significant.
diversity in ASF consumption to yield larger Table 8 disaggregates by type of ASF; all
growth benefits because different ASFs con- ASF types are associated with lower stunting
tain different growth-stimulating nutrients. among children aged 18–23 months, al-
Dairy, for example, is rich in calcium and though the coefficient on dairy is signifi-
contains a growth stimulating hormone (IGF- cantly larger than the coefficients on eggs
1) but is bereft of iron, whereas meat, fish, and meat/fish. The results in table 8 are consis-
and eggs are relatively rich in iron but low in tent with table 7 insofar as the implied benefits
calcium. And as we saw in table 4, there are of different ASFs are additive. For example,
substantial numbers of children who con- among children aged 18–23 months, the im-
sume multiple ASFs (over 40% in Latin plied reduction in stunting from consuming
America and MENA, and less than 20% in dairy, eggs, and meat/fish is a 6.6-point reduc-
other regions). The results in table 7 suggest tion (3.4 from dairy, 1.3 points from eggs, and
there are additional benefits to consuming 2.1 points from meat/fish).
multiple ASFs. In the 6–23-month sample, supplementary online appendix table 3 also
the implied reduction in stunting to one adds a potentially important confounding fac-
ASF is just 1.8 points, but this rises to 3.4 tor to the model (household wealth) mea-
points and 4.5 points for 2 ASFs and 3 sured via a principal components analysis of
ASFs, respectively, and all three coeffi- household assets (Filmer and Pritchett
cients are statistically different from one 2001). Wealth is likely an important deter-
another. The results are much weaker in the minant of ASF consumption but could also
6–11-month and 12–17-month sample, but independently influence stunting. However,
substantially stronger in the preferred 18– the results from table 5 are robust to the in-
23-month sample. In that sample consump- clusion of wealth terciles, even though the
tion of one ASF is associated with stunting coefficient on middle- and upper-wealth
being reduced by 3.7-percentage points, but terciles are themselves highly significant. In
1312 October 2018 Amer. J. Agr. Econ.

Table 8. All ASF Types Are Associated with Lower Stunting among Children 18–23 months,
although Dairy Has the Strongest Association
(1) (2) (3) (4)
6–23 months 6–11 months 12–17 months 18–23 months
Dairy 0.020*** 0.008 0.011* 0.034***
(0.003) (0.005) (0.006) (0.006)
Eggs 0.011*** 0.001 0.001 0.013**
(0.003) (0.005) (0.005) (0.006)
Meat/Fish 0.017*** 0.014*** 0.005 0.021***
(0.003) (0.005) (0.005) (0.005)
Other food groups? yes yes Yes yes
Child and household level controls? yes yes Yes yes
Rural dummy? yes yes Yes yes
Survey dummies? yes yes Yes yes
Sub-national fixed effects? yes yes Yes yes
P-value of H0: b(Dairy) > b(Eggs) 0.034** 0.184 0.076* 0.008***
P-value of H0: b(Meat) > b(Eggs) 0.092* 0.046** 0.199 0.152
P-value of H0: b(Dairy) > b(Meat) 0.266 0.202 0.225 0.056*
R-squared 0.150 0.105 0.142 0.169
Observations 130,315 45,316 45,026 39,973
Note: Unit of observation is a child and the dependent variable obtains a value 1 if the child is stunted, and zero otherwise. Standard errors clustered at the
DHS cluster level. Controls are those listed in table 5. Statistical significance denoted at ***¼ p < 0.01, **¼ p < 0.05, and *¼ p < 0.1.

the 18–23-month sample, for example, the consumers diversify away from starchy staple
coefficient on ASF drops from 0.040 to foods as incomes rise as per Bennett’s law
0.038 and remains statistically significant. (Choudhury and Headey 2017b), opting for
Thus, there is little indication that the esti- more expensive sources of calories as
mated benefits of ASFs simply represent incomes rise (Subramanian and Deaton
the greater economic status of ASF- 1996), including strong demand for ASFs
consuming households. (Colen et al. 2018). If incomes constrain ASF
consumption, this suggests they may be a
very expensive source of calories in low-in-
come countries. Certainly, many ASFs are
Constraints to ASF Consumption in imperfectly tradable, especially in environ-
Developing Countries ments with underdeveloped value chains
(e.g., poor transport, lack of cold storage, and
The evidence above suggests significant nutri- low levels of processing). Highly perishable
tional benefits from consuming ASFs, and yet foods also have associated food safety risks,
ASF consumption in poorer countries is gen- with semi-commercialized dairy systems be-
erally very low. Why is this? One perceived ing a notable example in both contemporary
barrier is a lack of nutritional knowledge. developing economies (Gizachew et al. 2016)
Many nutritional programs have attempted and early twentieth century industrialized so-
to improve nutritional knowledge, albeit it cieties (Gordon 2016). In turn, this draws at-
with mixed success and substantive concerns tention to the price of ASFs as a factor
on sustainability (Dewey and Adu-Afarwuah constraining children’s consumption of these
2008; Menon et al. 2015). Consistent with foods.
knowledge constraints, parental education Table 9 reports data on the price of ASFs
has been strongly linked to nutrition out- and other foods relative to the price to a
comes and dietary diversity even when apply- country-specific numeraire, the cheapest sta-
ing methods to reduce endogeneity bias ple cereal (i.e., rice, wheat, and maize prod-
(Alderman and Headey 2017). Gender bias ucts). For example, the value 6.1 for fish in
and low levels of maternal empowerment are Eastern & Southern Africa tells us that it
also widely-cited constraints (Jayachandran costs 6.1 times more to consume a calorie
and Pandi 2017), as are food taboos (Pach on from a fish product than it does a calorie
et al. 2007; Zerfu, Umeta, and Baye 2016). from a staple cereal. Across all regions, roots
Additionally, many studies show that and tubers are very cheap sources of calories.
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1313

Table 9. Cereal-relative Calorie Price Ratios for Various Foods, by Region

Vegetal sourced food prices
Roots Vitamin A-rich Dark Green Other Other Legumes
& Tubers Fruit and Veg. Leafy Veg. Veg. Fruit
High income 1.6 3.0 9.0 3.3 1.7 1.2
Latin America & 1.2 1.9 5.6 6.5 1.3 2.2
Caribbean
Middle East & North 2.1 2.5 6.1 5.3 3.3 2.1
Africa
South, Central & 1.5 1.9 6.2 6.0 3.1 2.0
South-East Asia
Western & Central 1.0 2.3 11.5 11.6 3.1 NA
Africa
Eastern & Southern 1.7 3.1 7.3 11.4 3.2 NA
Africa
Animal sourced foods & fortified baby cereal prices
Cow’s milk, Cow’s milk, Chicken eggs Meat Fish Fortified
fresh processed baby cereal
High income 3.2 2.2 3.0 2.0 4.3 5.0
Latin America & 3.9 3.0 4.9 3.2 3.4 9.6
Caribbean
Middle East & North 10.1 3.1 6.1 6.2 6.0 16.1
Africa
South, Central & 7.8 3.8 6.2 6.5 5.3 16.4
South-East Asia
Western & Central 16.5 4.0 9.9 5.3 5.0 23.4
Africa
Eastern & Southern 13.9 5.8 9.1 5.6 6.1 18.6
Africa
Note: Caloric prices are the ratio of the price of 1 calorie of a given food (e.g., eggs) relative to 1 calorie of the cheapest staple cereal in each country (e.g.,
rice, wheat, and maize products). See Headey et al. (2017) for details. NA refers to not available. ICP data did include some legume prices for sub-Saharan
Africa, but not widely consumed local varieties.

Strikingly, vitamin-A rich fruits and vegeta- consumption of dairy products among infants
bles and “other fruit” are also relatively af- and young children is widespread, with over
fordable, partly because many such fruits are half consuming dairy in the past 24 hours
highly suitable to tropical conditions, and be- (table 4). By contrast, fresh milk is moder-
cause some (e.g., mangos, papayas, bananas) ately expensive in South, Central and South-
are relatively easy to store and transport. By East Asia (though relatively cheap in India,
contrast, vegetables, which are highly perish- where dairy consumption is high), and ex-
able and low in calorie density, are an expen- tremely expensive in sub-Saharan Africa. In
sive source of calories. Legumes, which are Western and Central Africa, for example,
reasonably high in lower-quality protein, are fresh milk is 16.5 times as expensive as the
an affordable source of calories in most cheapest cereals. Processed milk is highly
regions, but less so in Africa. storable, so it is of little surprise that it is
Price ratios of ASFs, however, reveal strik- more affordable in all regions. Despite this, it
ing differences across income levels and would appear that processed milk is not
regions. Both perishable fresh milk and widely consumed in these regions.
highly storable processed milk are relatively Price variation for eggs is even more strik-
cheap sources of calories in high-income ing. Eggs are a very cheap source of calories
countries but also in Latin America and the and protein in high-income countries
Caribbean. In North Africa and Middle East (Iannotti et al. 2014). Yet in the Middle East
region, processed milk is very cheap, but & North Africa region, they are at least 6
fresh milk is very expensive. It is notable that times as expensive as staple cereals, and in
in both these middle-income regions, Africa they are 9–10 times as expensive.
1314 October 2018 Amer. J. Agr. Econ.

Table 10. Linear Probability Model Regressions of Consumption of ASFs by Children aged
6-23 months as a Function of Calorie Price Ratios, and Various Child-, Household-,
Community-, and National-level Characteristics
(1) (2) (3) (4)
Dairy consumption Egg consumption Meat consumption Fish consumption
Log own priced 0.103*** 0.153*** 0.066*** 0.052***
(0.005) (0.006) (0.005) (0.004)
Log cereal yieldsd 0.066*** 0.015*** 0.081*** 0.064***
(0.005) (0.004) (0.004) (0.005)
Log GDP per capitad 0.075*** 0.029*** 0.065*** 0.084***
(0.004) (0.004) (0.004) (0.005)
Log urbanization rated 0.002 0.204*** 0.062*** 0.114***
(0.007) (0.006) (0.007) (0.009)
Conflict dummyd 0.008 0.070*** 0.053*** 0.041***
(0.005) (0.004) (0.004) (0.004)
Wealth, middle tercileb 0.063*** 0.053*** 0.024*** 0.001
(0.004) (0.003) (0.003) (0.004)
Wealth, upper tercileb 0.162*** 0.077*** 0.066*** 0.040***
(0.006) (0.005) (0.005) (0.006)
Mother 9þ yrs of schoolb 0.030*** 0.032*** 0.044*** 0.005
(0.004) (0.004) (0.004) (0.004)
Father 9þ yrs of schoolb 0.009** 0.004 0.012*** 0.023***
(0.004) (0.003) (0.003) (0.004)
Open defecation, villagec 0.019*** 0.061*** 0.042*** 0.010*
(0.006) (0.004) (0.004) (0.006)
Unimproved water, villagec 0.049*** 0.028*** 0.012*** þ0.028***
(0.006) (0.004) (0.004) (0.005)
Women’s autonomyb 0.018*** 0.012*** 0.005 0.013***
(0.004) (0.004) (0.004) (0.004)
Currently breastfeda 0.078*** 0.008** 0.050*** 0.013***
(0.004) (0.004) (0.004) (0.004)
Ruralc 0.000 0.008** 0.022*** 0.027***
(0.004) (0.004) (0.004) (0.004)
Father absentb 0.055*** 0.033*** 0.039*** 0.016***
(0.007) (0.005) (0.005) (0.006)
Hospital/clinic accessb 0.020*** 0.034*** 0.012*** 0.046***
(0.004) (0.003) (0.003) (0.004)
R-squared 0.177 0.146 0.104 0.115
Observations 98,840 100,991 91,793 89,896
Note: Unit of observation is a child and the dependent variable obtains a value 1 if the child consumed from the food group, and zero otherwise. Asterisks in-
dicate the following: ***¼ p < 0.01, **¼ p < 0.05, and *¼ p < 0.1. Superscript a indicates child-level indicators; b¼household-level indicators; c¼ community/
cluster-level indicators; d¼ national-level indicators. Regressions also control for short birth interval, number of children, teenage motherhood, child gender.
These coefficients had small and often insignificant coefficients.

This provides prima facie evidence that the Caribbean, but also—compared to other
highly perishable nature of eggs and the low ASFs—relatively affordable in sub-Saharan
productivity of egg production in lower in- Africa where meat/fish calories are just 5–6
come countries would appear to account for times as expensive as staple cereals. Notably,
the low levels of egg consumption observed fortified baby cereals—a potential substitute
earlier. for ASFs—are extremely expensive in all
The price story on flesh foods is quite dif- regions except high-income countries and
ferent. Both red/white meat and fish are rela- Latin America and the Caribbean (see
tively tradable, either because of the scope to Masters, Nene, and Bell 2017 for a more de-
trade live animals, or the ability to freeze, tailed anlaysis).
dry, smoke, or salt flesh foods. Meat and fish Table 10 reports linear probability models
are again notably cheap in high-income coun- explaining children’s consumption of four dif-
tries and in Latin America and the ferent ASFs. Own prices are always
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1315

significantly and negatively associated with relationship in developing countries, and lit-
consumption of each specific ASF. Dairy and tle previous work has systematically docu-
egg consumption are especially strongly asso- mented patterns of ASF consumption among
ciated with their relative prices. In the case of young children or attempted to explain why
dairy, a doubling of the price ratio for fresh these patterns exist.
milk—for example, the difference between In this paper we document that ASF con-
dairy prices in the United States and India— sumption among young children is relatively
predicts a reduction in dairy consumption of low in sub-Saharan Africa and most of Asia,
around 10 percentage points. Considering but is also characterized by some distinctive
that 35% of the children in our sample con- patterns. Most notable are: (a) low levels of
sumed dairy (see table 4), this translates into dairy consumption in most lowland areas of
a 29% drop in dairy consumption prevalence. Africa and much of Asia (Bangladesh,
The association for eggs is even stronger, sug- Cambodia), but relatively high levels of fish
gesting that the very high prices of eggs in consumption; (b) low levels of egg and red/
most regions (table 9) provides a strong ex- white meat consumption in Africa and Asia;
planation of why egg consumption is not (c) low levels of multiple ASF consumption
common among infants and young children in Africa and Asia (<20%). We show evi-
in poorer countries. Meat and fish consump- dence of strong associations between ASF
tion share somewhat more modest associa- consumption and child growth, particularly
tions with their own price ratios. In the case consumption of multiple ASFs and of dairy.
of fish this may be the result of some attenua- Finally, we show that high relative prices of
tion bias, as the ICP only measures a few fish ASF calories are a critical constraint restrict-
varieties relevant to lower-income countries. ing consumption of these foods, especially for
Amongst other national factors there are highly perishable products such as fresh milk
some interesting associations. Dairy and and eggs. While our econometric models are
meat consumption have strong associations saturated with a wide array of controls and
with GDP per capita, but in the case of dairy are robust to the inclusion of household
there is also an association with household- wealth, we emphasize that these are associa-
level wealth that is much stronger than the tions, not causal relations. That being said,
corresponding associations for other ASFs. there is also an important reason to suspect
Strikingly, fish consumption is negatively as- that we may be underestimating the associa-
sociated with GDP per capita and with tion between ASF consumption on stunting:
household wealth. The negative association the 24-hour recall is likely to be an imperfect
with national GDP per capita could partly re- proxy for usual diets (i.e., within-person er-
flect high levels of consumption in much of ror) leading to an attenuation bias (Thorne-
sub-Saharan Africa, as well as countries like Lyman, Spiegelman, and Fawzi 2014). This
Cambodia and Bangladesh, but the negative may also be more problematic for some ASFs
association with wealth likely reflect the fact than others since some (e.g., dairy) may be
that in many populations, fish is considered consumed in larger quantities and more fre-
an inferior good relative to red/white meat. quently than others (e.g., flesh foods, eggs).
Also notable is that the national urbanization All this points to the need for new studies
rate is very strongly associated with egg and that can illustrate the causal links between
fish consumption, but negatively associated quantities of ASF consumption and child
with meat consumption. These results might nutrition.
indicate that more commercialized large- For the agriculture and nutrition communi-
scale low-cost poultry production systems ties, these findings are of clear importance.
tend to emerge when large urban agglomera- Nutritionists have long promoted interven-
tions give rise to concentrated centers of de- tions designed to improve nutritional knowl-
mand (Narrod, Pray, and Tiongco 2008). edge on feeding practices, and/or to increase
homestead production of ASFs (Leroy and
Frongillo 2007; Dewey and Adu-Afarwuah
2008). The extremely high prices of ASFs
Conclusions suggests these interventions could have lim-
ited impacts, however. One reason is that
Despite plausible biological mechanisms link- high ASF prices provide strong incentives for
ing ASF intake to child growth, there is a sur- producers to sell ASFs rather than feed them
prising dearth of evidence on this to their own children. Behavioral change
1316 October 2018 Amer. J. Agr. Econ.

interventions are used to encourage house- 2006; Steinfeld et al. 2006). The amount of
holds to at least partially resist the financial GHG emissions are closely linked to size of
rewards of selling ASFs, but doing so is costly the livestock population as well as to the size
and potentially unsustainable, and will still of the animal, with non-dairy cattle being a
fail to increase ASF consumption among major source of methane emissions (O’Mara
non-producers. Potentially a more cost- 2011). Therefore, optimizing the size, produc-
effective and sustainable strategy is to em- tivity and composition of the national live-
ploy larger-scale investments in the livestock stock and fisheries sectors to maximize the
and fisheries sector and its associated value nutritional benefits whilst minimizing GHG
chains, and/or appropriate trade policies, to emissions is an important area of future
reduce the prices of ASFs for all consumers. research.
In the case of eggs—a highly non-tradable Additionally, a substantial proportion of
product—investments in domestic production the world’s undernourished population is
are clearly crucial. However, egg production vegetarian for cultural reasons. The DHS for
is characterized by huge economies of scale India—which accounts for around one-third
(Narrod, Pray, and Tiongco 2008), suggesting of the world’s population of stunted chil-
that investments in village production sys- dren—suggests that around one-third of
tems are unlikely to be economically effi- Indian adults are vegetarian or lacto-
cient. Moreover, a growing body of research vegetarian. Short of radical cultural shifts,
also suggests scavenging poultry systems are these populations will likely continue to rely
hazardous to the health of young children on legumes as their key source of protein.
(Ngure et al. 2013; Headey and Hirvonen While legumes were not associated with
2016). For dairy, the issue is much more com- stunting in our full sample results, we did ob-
plex since dairy powder is highly tradable but serve a significant association between le-
not obviously a close substitute for fresh gume consumption and child growth in the
milk, nor highly desirable in contexts where South Asian sample (table 6), suggesting that
water quality is a major concern. Many popu- legumes play a special role in this region that
lations in tropical regions also have little tra- they do not in other populations. Clearly, in-
dition of consuming milk, and adults face creasing consumption of ASFs in India and
lactose intolerance issues, suggesting there Nepal is challenging, and may require a mix
may be significant cultural issues to be over- of agricultural policies and tailored behav-
come (FAO 2008). Meat and fish are also ioral change designed to either increase ASF
complex sectors characterized by both inten- consumption (dairy particularly among lacto-
sive and extensive production modalities and vegetarians) or appropriately supplement
some degree of tradability. The importance diets with legumes or fortified infant cereals.
of fish in many African and Asian diets— The importance of ASFs for child nutrition
even those of young children—makes it a therefore has important ramifications for fu-
particularly interesting sector for future re- ture research and for international and na-
search and investment. One concern is that tional investments in agriculture. We see
extensive systems are facing major over- three lines of attack: (a) as economists, we
fishing problems in Africa especially, al- are used to thinking that large price differen-
though more sustainable fish farming systems tials present opportunities for arbitrage. In
have experienced explosive growth in much the case of ASFs, this is clearly not occurring,
of Asia and parts of Africa and may offer a suggesting significant barriers to trade. An
pathway to expanding fish consumption understanding of the relative roles played by
(Bene et al. 2015). policy decisions (such as tariff levels), tech-
While ASFs are critical for improving hu- nology (addressing perishability issues), or a
man nutrition, there are significant obstacles combination of the two (e.g., phytosanitary
to overcome. First, scaling up ASF produc- regulations) will help inform appropriate pol-
tion comes at a potential environmental cost icy changes and interventions that should
since livestock and fish production releases lower these barriers and in turn reduce these
methane (CH4) and nitrous oxide (N2O) to price differentials. (b) Understanding
the atmosphere, thus contributing to global whether improvements in ASF value chains
warming. Depending on the methodology, it can improve diets of pre-school children.
is estimated that the livestock sector contrib- Randomized control trials may be difficult to
utes between 8% and 18% of the total global implement but it might be possible to exploit
greenhouse gas (GHG) emissions (IPCC natural experiments such as improved access
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1317

to these value chains following infrastructure Hollis and T. N. Srinivasan, 631–711.

improvements, such as roads. (c) Increased Amsterdam: North-Holland.
investments in research that will lead to in- Bene, C., M. Barange, R. Subasinghe, P.
creased productivity in ASF sectors and, re- Pinstrup-Andersen, G. Merino, G.-I.
latedly, assessing whether such Hemre, and M. Williams. 2015. Feeding 9
improvements, possibly twinned with Billion by 2050—Putting Fish Back on
demand-side interventions such as nutrition the Menu. Food Security 7 (2): 261–74.
behavior change communication, can in- Choudhury, S., and D. Headey. 2017a.
crease pre-school child ASF consumption Household dairy production and child
and in turn reduce stunting.10 More generally, growth: Evidence from Bangladesh.
we see such a research portfolio as part of Economics & Human Biology
efforts to shift the international and national (forthcoming).
agricultural research and policy agendas ———. 2017b. What Drives Diversification
away from their dominant focus on food of National Food Supplies? A Cross-
quantity towards a more balanced portfolio Country Analysis. Global Food Security
that pays increased attention to diet quality. 15:85–93.
Colen, L., P. Melo, Y. Abdul-Salam, D.
Roberts, S. Mary, and S.G.Y. Paloma.
Supplementary Material 2018. Income Elasticities for Food,
Calories and Nutrients across Africa: A
Supplementary material are available at Meta-Analysis. Food Policy 77:116–132.
American Journal of Agricultural Economics Craig, L.A., B. Goodwin, and T. Grennes. 2004.
online. The Effect of Mechanical Refrigeration on
Nutrition in the United States. Social
Science History 28 (2): 325–36.
References Danaei, G., K.G. Andrews, C.R. Sudfeld, G.
Fink, D.C. McCoy, E. Peet, A. Sania,
Alderman, H., and D. Headey. 2018. The M.C. Smith, et al. 2016. Risk Factors for
Timing of Growth Faltering Has Childhood Stunting in 137 Developing
Important Implications for Countries: A Comparative Risk
Observational Analyses of the Assessment Analysis at Global,
Underlying Determinants of Nutrition Regional, and Country Levels. PLoS
Outcomes. PLoS One 13 (4): e0195904. Medicine 13 (11): e1002164.
———. 2017. How Important Is Parental de Beer, H. 2012. Dairy Products and
Education for Child Nutrition? World Physical Stature: A Systematic Review
Development 94: 448–64. and Meta-Analysis of Controlled Trials.
Baten, J. 2009. Protein Supply and Economics & Human Biology 10 (3):
Nutritional Status in Nineteenth Century 299–309.
Bavaria, Prussia and France. Economics de Onis, M., and F. Branca. 2016. Childhood
& Human Biology 7 (2): 165–80. Stunting: A Global Perspective. Maternal
Baten, J., and J.E. Murray. 2000. Heights of & Child Nutrition 12: 12–26.
Men and Women in 19th-Century Dewey, K.G., and S. Adu-Afarwuah. 2008.
Bavaria: Economic, Nutritional, and Systematic Review of the Efficacy and
Disease Influences. Explorations in Effectiveness of Complementary Feeding
Economic History 37 (4): 351–69. Interventions in Developing Countries.
Behrman, J.R., and A.B. Deolalikar. 1988. Maternal & Child Nutrition 4 (S1): 24–85.
Health and Nutrition. In Handbook of Dror, D.K., and L.H. Allen. 2011. The
Development Economics, ed. Chenery Importance of Milk and Other Animal-
Source Foods for Children in Low-
Income Countries. Food and Nutrition
10
Currently, the vast proportion of agricultural research and Bulletin 32 (3): 227–43.
development expenditures for lower-income countries has been IPCC 2006. 2006 IPCC Guidelines for
directed towards staple crops, with livestock and fisheries ac-
counting for approximately 12% of CGIAR (Consultative
National Greenhouse Gas Inventories, ed.
Group for International Agricultural Research spending), Eggleston H.S., Buendia L., Miwa K.,
according to David Laborde (personal communication). This Ngara T. and Tanabe K. Geneva:
does not include national agricultural expenditures; we suspect
that the share of the total agricultural budget devoted to live- National Greenhouse Gas Inventories
stock is unlikely to be higher than this. Programme.
1318 October 2018 Amer. J. Agr. Econ.

Filmer, D., and L. Pritchett. 2001. Estimating Potential for Improving Maternal and
wealth effects without expenditure data - Young Child Nutrition among the
or tears: an application to education en- World’s Poor. Nutrition Reviews 72 (6):
rollment in states of India. Demography 355–68.
38:115–32. Iannotti, L.L, C.K. Lutter, C.P. Stewart,
Food and Agriculture Organisation of the C.A.G. Riofrıo, C. Malo, G. Reinhart, A.
United Nations. 2013. Milk and Dairy Palacios, et al. 2017. Eggs in Early
Products in Human Nutrition. Japan: Complementary Feeding and Child
IGES; Rome: Food and Agriculture Growth: A Randomized Controlled
Organisation. Trial. Pediatrics 140 (1): e20163459.
Gizachew, D., B. Szonyi, A. Tegegne, J. ICF International. 2015. The Demographic
Hanson, and D. Grace. 2016. Aflatoxin and Health Surveys Program. Available
Contamination of Milk and Dairy Feeds at: https://dhsprogram.com (accessed
in the Greater Addis Ababa Milk Shed, Nov 1, 2017).
Ethiopia. Food Control 59: 773–9. Jayachandran, S., and R. Pandi. 2017. Why
Goodwin, B.K., T.J. Grennes, and L.A. Are Indian Children so Short? The Role
Craig. 2002. Mechanical Refrigeration of Birth Order and Son Preference.
and the Integration of Perishable American Economic Review 107 (9):
Commodity Markets. Explorations in 2600–29.
Economic History 39 (2): 154–82. Leroy, J.L., and E.A. Frongillo. 2007. Can
Gordon, R.J. 2016. The Rise and Fall of Interventions to Promote Animal
American Growth. Princeton, NJ: Production Ameliorate Undernutrition?
Princeton University Press. Journal of Nutrition 137 (10): 2311–6.
Grasgruber, P., M. Sebera, E. Hrazdıra, J. Leroy, J.L., M. Ruel, J.-P. Habicht, and E.A.
Cacek, and T. Kalina. 2016. Major Frongillo. 2014. Linear Growth Deficit
Correlates of Male Height: A Study of Continues to Accumulate beyond the
105 Countries. Economics and Human First 1000 Days in Low-and Middle-
Biology 21: 172–95. Income Countries: Global Evidence from
Headey, D., H. Alderman, C. Maitra, and P. 51 National Surveys. Journal of Nutrition
Rao. 2017. The Relative Prices of 144 (9): 1460–6.
Healthy and Unhealthy Foods in 177 Masters, W.A., M.D. Nene, and W. Bell.
Countries. Paper presented at 2017. Nutrient Composition of Premixed
Agriculture for Nutrition and Health and Packaged Complementary Foods for
Academy Week. Kathmandhu, Nepal, Sale in Low- and Middle-Income
10–13th July, 2017. Countries: Lack of Standards Threatens
Headey, D., and K. Hirvonen. 2016. Is Infant Growth. Maternal & Child
Exposure to Poultry Harmful to Child Nutrition 13 (4): e12421.
Nutrition? An Observational Analysis McLaren, D.S. 1974. The Great Protein
for Rural Ethiopia. PLoS One 11 (8): Fiasco. Lancet 304 (7872): 93–6.
e0160590. Menon, P., P.H. Nguyen, K.K. Saha, A.
Hoddinott, J., D. Headey, and M. Dereje. Khaled, T. Sanghvi, J. Baker, K. Afsana,
2015. Cows, Missing Milk Markets, and R. Haque, E.A. Frongillo, M.T. Ruel,
Nutrition in Rural Ethiopia. The Journal and R. Rawat. 2016. Combining
of Development Studies 51 (8): 958–75. Intensive Counseling by Frontline
Hoddinott, J.F., M.W. Rosegrant, and M. Workers with a Nationwide Mass Media
Torero. 2013. Investments to Reduce Campaign Has Large Differential
Hunger and Undernutrition. In B., Impacts on Complementary Feeding
Lomborg ed. Global Problems, Smart Practices but Not on Child Growth:
Solutions. Cambridge: Cambridge Results of a Cluster-Randomized
University Press. Program Evaluation in Bangladesh. The
Hoppe, C., C. Mølgaard, and K.F. Journal of Nutrition 146: 2075–2084.
Michaelsen. 2006. Cow’s Milk and Linear Messina, M.J. 1999. Legumes and Soybeans:
Growth in Industrialized and Developing Overview of Their Nutritional Profiles
Countries. Annual Review of Nutrition and Health Effects. American Journal of
26: 131–73. Clinical Nutrition 70 (3): 439s–50s.
Iannotti, L.L., C.K. Lutter, D.A. Bunn, and Moradi, A., and J. Baten. 2005. Inequality in
C.P. Stewart. 2014. Eggs: The Uncracked Sub-Saharan Africa: New Data and New
Headey, Hirvonen, and Hoddinott Animal Sourced Foods and Child Stunting 1319

Insights from Anthropometric Estimates. Long Shadow: Environmental Issues and

World Development 33 (8): 1233–65. Options. Rome: Food and Agriculture
Narrod, C., K. Pray, and M. Tiongco. 2008. Organisation.
Technology Transfer, Policies, and the Subramanian, S., and A. Deaton. 1996. The
Role of the Private Sector in the Global Demand for Food and Calories. Journal
Poultry Revolution. IFPRI Discussion of Political Economy 104 (1): 133–62.
Paper 00841. Washington DC: The Takahashi, E. 1984. Secular Trend in Milk
International Food Policy Research Consumption and Growth in Japan.
Institute (IFPRI). Human Biology 56 (3): 427–37.
Ngure, F.M., J.H. Humphrey, M.N. Mbuya, F. Tang, M., X.-Y. Sheng, N.F. Krebs, and K.M.
Majo, K. Mutasa, M. Govha, E. Mazarura, Hambidge. 2014. Meat as
et al. 2013. Formative Research on Complementary Food for Older
Hygiene Behaviors and Geophagy among Breastfed Infants and Toddlers: A
Infants and Young Children and Randomized, Controlled Trial in Rural
Implications of Exposure to Fecal China. Food and Nutrition Bulletin 35
Bacteria. American Journal of Tropical (4_suppl3): S188–92.
Medicine and Hygiene 89 (4): 709–16. Thorne-Lyman, A., D. Spiegelman, and
O’Mara, F.P. 2011. The Significance of W.W. Fawzi. 2014. Is the Strength of
Livestock as a Contributor to Global Association between Indicators of
Greenhouse Gas Emissions Today and in Dietary Quality and the Nutritional
the Near Future. Animal Feed Science Status of Children Being
and Technology 166–167: 7–15. Underestimated? Maternal & Child
Pachon, H., K.B. Simondon, S.T. Fall, P. Nutrition 10 (1): 159–60.
Menon, M.T. Ruel, C. Hotz, H. Creed- Victora, C.G., M. de Onis, P. Curi Hallal, M.
Kanashiro, et al. 2007. Constraints on the Blössner, and R. Shrimpton. 2010.
Delivery of Animal-Source Foods to Worldwide Timing of Growth Faltering:
Infants and Young Children: Case Revisiting Implications for Interventions.
Studies from Five Countries. Food and Pediatrics 125: 473–80.
Nutrition Bulletin 28 (2): 215–29. Walker, S.P., C.A. Powell, S.M. Grantham-
Pinstrup-Andersen, P., and E. Caicedo. 1978. McGregor, J.H. Himes, and S.M. Chang.
The Potential Impact of Changes in 1991. Nutritional Supplementation,
Income Distribution on Food Demand Psychosocial Stimulation, and Growth of
and Human Nutrition. American Journal Stunted Children: The Jamaican Study.
of Agricultural Economics 60 (3): 402–15. American Journal of Clinical Nutrition 54
Rawlins, R., S. Pimkina, C.B. Barrett, S. (4): 642–8.
Pedersen, and B. Wydick. 2014. Got Milk? World Health Organization. 2006. WHO
The Impact of Heifer International’s Child Growth Standards Based on
Livestock Donation Programs in Rwanda Length/Height, Weight and Age. Acta
on Nutritional Outcomes. Food Policy 44: Paediatrica Supplement 450: 76–85.
202–13. Zerfu, T.A., M. Umeta, and K. Baye. 2016.
Semba, R.D. 2016. The Rise and Fall of Dietary Habits, Food Taboos, and
Protein Malnutrition in Global Health. Perceptions towards Weight Gain during
Annals of Nutrition Metabolism 69 (2): Pregnancy in Arsi, Rural Central
79–88. Ethiopia: A Qualitative Cross-Sectional
Steinfeld, H., P. Gerber, T. Wassenaar, V. Study. Journal of Health, Population and
Castel, and C. de Haan. 2006. Livestock’s Nutrition 35 (1): 1–7.