Does India Really Suffer from Worse Child Malnutrition Than Sub-Saharan Africa?
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
SPECIAL ARTICLE
Does India Really Suffer from Worse Child
Malnutrition Than Sub-Saharan Africa?
Arvind Panagariya
A common continuing criticism of the economic reforms 1 Introduction
A
in India has been that despite accelerated growth and widely-held view among nearly all development ex-
perts familiar with India is that despite accelerated
all-around poverty reduction, the country continues to
growth in the last three decades, the country contin-
suffer from worse child malnutrition than nearly all ues to suffer from worse child malnutrition than virtually
Sub-Saharan African countries with lower per capita every Sub-Saharan African country with lower per capita
incomes. This paper argues that this narrative, nearly income. According to this view, India is also guilty of having
made little or no progress in bringing malnutrition levels down.
universally accepted around the world, is false. It is the
A prominent example is The Economist magazine, which stated
artefact of a faulty methodology that the World Health in an article in its 23 September 2010 edition, “Nearly half of
Organisation has pushed and the United Nations has India’s small children are malnourished: one of the highest
supported. If appropriate corrections are applied, in all rates of underweight children in the world, higher than most
countries in sub-Saharan Africa. More than one-third of the
likelihood, India will be found to be ahead of
world’s 150m malnourished under-fives live in India.”
Sub-Saharan Africa in child malnutrition, just as in other In addition to this high level of child malnutrition, the article
vital health indicators. repeated the common claim that India had made very little
progress in combating child malnutrition. It noted, “Almost as
shocking as the prevalence of malnutrition in India is the coun-
try’s failure to reduce it much, despite rapid growth. Since 1991
gross domestic product (GDP) has more than doubled, while
malnutrition has decreased by only a few percentage points.”
In January 2011, even India’s otherwise measured Prime Minis-
ter Manmohan Singh went on to lament, “The problem of mal-
nutrition is a matter of national shame”, while releasing the
much publicised Hunger and Malnutrition (HUNGaMA) Report.1
Reforms critics had originally argued that reforms had not
helped the poor or that they had left the socially disadvan-
taged groups behind. Evidence has now decisively established,
however, that reforms have been accompanied by a decline in
poverty across all social groups, including the scheduled
castes and scheduled tribes (Mukim and Panagariya 2012),
that this reduction accelerated between 2004-05 and 2009-10
with the acceleration in growth, and that the gap in poverty
I thank an anonymous referee of this journal for several helpful ratios between the socially disadvantaged and other groups
suggestions. The paper has also benefited from criticism by Jishnu Das, has finally begun to come down (Thorat and Dubey 2012; Pan-
numerous very helpful discussions with Reuben Abraham and agariya and Mukim 2013). Therefore, the critics have now
comments by Prashant Reddy, Ursula Schwartzhaupt, and Rajitha
shifted to arguing that the reforms have done precious little
Swaminathan. It was the basis of the Chandrasekaran Memorial Lecture
at the International Institute of Population Studies, Mumbai, on for India’s malnourished children, with the country lagging
8 November 2012, and was originally written for the Program on Indian behind even much poorer Sub-Saharan Africa.
Economic Policies at Columbia University, funded by a grant from the This paper rejects this claim, arguing that it stems from
John Templeton Foundation. The opinions expressed do not necessarily child malnutrition estimates based on a flawed measurement
reflect the views of the John Templeton Foundation.
methodology. The central problem with the current methodo-
Arvind Panagariya (ap2231@columbia.edu) is Professor of Economics logy is the use of common height and weight standards around
and the Jagdish Bhagwati Professor of Indian Political Economy at the world to determine malnourishment, regardless of differ-
Columbia University, the United States.
ences that may arise from genetic, environmental, cultural,
98 may 4, 2013 vol xlviiI no 18 EPW Economic & Political WeeklySPECIAL ARTICLE and geographical factors. Though medical literature recog- prominently in poor gains in height, weight, and circumfer- nises the importance of these factors, the World Health ences of head and mid-upper arm. Other physical symptoms Organisation (WHO) totally ignores them when recommending such as skin peeling, abdominal distension, liver enlargement, globally uniform height and weight cut-off points against and sparse hair as well as behavioural characteristics such as which children are compared to determine whether they suf- anxiety, irritation, and attention deficit may also accompany fer from stunting (low height for age) or underweight (low protein deficiency. Micronutrient deficiency results from inad- weight for age) problems. equate levels of iron, folate, iodine, and various vitamins, It is important to point out at the outset that it is not my inten- including A, B6, D, and E, in the body. These deficiencies lead tion to downplay the seriousness of the child malnutrition prob- to anaemia, goitre, bone deformities, and night blindness. lem in India. Just like vital health statistics such as life expect- Given these many dimensions involved in identifying mal- ancy, infant and child mortality rates, and maternal mortality nutrition, only a thorough medical check-up can properly de- ratio, which need continued improvement, child malnutrition termine whether a child is malnourished or not. But few glo- must be brought down and eliminated. The contention of the bally comparable large-scale surveys rely on extensive medical paper, instead, is that the current globally uniform height- and check-ups to measure malnutrition in children. Moreover, weight-based measures of child malnutrition, which place India after the United Nations (UN) introduced the Millennium behind nearly every Sub-Saharan African country, are premised Development Goals (MDGs), which prominently included rapid on invalid assumptions and therefore need correction. The progress in combating child malnutrition as a key goal, the paper proposes to establish a strong presumption that once pressure to come up with estimates of the proportion of chil- health experts and economists come together to devise a better dren suffering from malnutrition grew. As a result, height and methodology of measurement, we will find that as in the case of weight, which are easy to measure and require no specialised indicators such as life expectancy, infant and child mortality medical skills, quickly became the focus of attention. This was rates, and maternal mortality ratio, India does not suffer worse further aided by the WHO, which provided the common stand- child malnutrition than Sub-Saharan Africa. ards of height and weight by age and gender to determine Some may argue that the debate on precisely measuring whether a child was stunted or underweight. child malnutrition is counterproductive since as long as a large Today, virtually all headline figures on child malnutrition, proportion of children are malnourished, the effort required including the ones that led Manmohan Singh to declare the to combat it is the same regardless of precise numbers.2 There phenomenon a national shame, are based on height and are at least three objections to this argument. First, proper weight. In view of the multidimensional nature of child mal- measurement and determination of progress have serious im- nutrition, this singular focus on low height and weight should plications for the allocation of scarce revenue resources among itself be a source of concern. But this is not the direction of the competing social objectives, especially in a poor country with critique in this paper. limited revenues. Should more be spent on combating child Instead, the focus of the paper is narrower. Accepting poor malnutrition or on improving elementary education? Or on height and weight gains as the principal manifestations of providing guaranteed employment or on alleviating adult malnutrition, I ask whether we are correctly identifying stunted hunger? Should India spread the limited resources available and underweight children. The underlying question is about for combating child malnutrition over half its children or just a the validity of applying uniform height and weight norms quarter of them? These are real choices a poor country must around the world as the basis for determining whether a given make in setting its budget priorities. child is well-nourished or malnourished. This focus does not Second, if a child is already receiving a balanced diet but is deny in any way the importance of a full medical examination misclassified as malnourished because an erroneous standard is to determine whether or not a child is malnourished.3 But it applied to evaluate her nutrition status, we may prescribe an in- addresses the deficiencies of the measures that are the source crease in her diet when such an adjustment is uncalled for. At the of virtually all discussion on child malnutrition in the public other extreme, what if we misclassify a malnourished child as policy space. well-nourished? In the former case, we run the risk of turning a healthy child obese and in the latter that of ignoring malnutrition. 3 Planting the Seeds of Doubt Finally, truth in numbers is an essential element in serious I begin by offering three comparisons that challenge the nar- intellectual discourse. Else, we would be tempted to falsify all rative that India has more stunted and underweight children other indicators such as those relating to poverty alleviation, than Sub-Saharan Africa. These comparisons should at least growth, life expectancy, infant mortality rate, and maternal give the reader a reason to pause and entertain the possibility mortality ratio on the premise that this is an effective way of that something is wrong with the news headlines depicting attracting public attention. The ends rarely justify the means. India as one of the worst performers in child nutrition. 2 Why the Focus on Height and Weight 3.1 India versus Chad Malnutrition is a multidimensional phenomenon. In broad Let us begin by comparing a set of commonly-used health indi- terms, it may be divided into protein energy malnutrition and ces for the child and the mother in India to those in Chad and micronutrient deficiency. The former manifests itself most the Central African Republic, two of the poorest countries in Economic & Political Weekly EPW may 4, 2013 vol xlviiI no 18 99
SPECIAL ARTICLE
the world. As Table 1 shows, Chad has just 48 years of life ex- 3.3 India versus the 33 Poorer Sub-Saharan African
pectancy against India’s 65 years; an infant mortality rate Countries
(IMR) of 124 against India’s 50; an under-five mortality rate of The above comparisons are not isolated examples. A comparison of
Table 1: India Compared to Chad and Central African Republic India with nearly every one of the 33 Sub-Saharan
Indicator India Chad Central African Chad as % CAR as % African countries with lower per capita incomes in
Republic (CAR) of India of India
2009 in current dollars shows the same pattern. I
Life expectancy (2009) 65 48 48 74 74
Infant mortality per 1,000 live births (2009) 50 124 112 248 224
demonstrate this with the help of Figures 1-3 and
Under-five mortality per 1,000 live births 66 209 171 317 259 Figures 4-7 (p 101) with each figure comparing India
Maternal mortality per 1,00,000 live births (2009) 230 1,200 850 522 370 to the 33 countries in Sub-Saharan Africa along one
Per cent children below 5 stunted (2000-09) 47.9 44.8 44.6 94 93 health indicator. I arrange the countries from left to
Per cent children below 5 underweight (2000-09) 43.5 33.9 21.8 78 50
right in order of increasing per capita incomes.
Source: WHO World Health Statistics, 2011.
Figure 1: Life Expectancy in India and 33 Poorer Sub-Saharan African Countries
209 relative to India’s 66; and a maternal mortality ratio Life Expectancy (2009): Countries in Rising Order of Per Capita GDP from L to R
70
(MMR) of 1,200 compared to India’s 230. Yet, Chad has dispro- 66
60 59 65 59 60 57 6058 62 60
65
60 56 57
portionately fewer stunted and underweight children than 50 49 49
54 52 49
48 48 52 49 52 49 48
53
48
55 50 54 51
50 47
India. The comparison with the Central African Republic is
40
equally stark. 30
20
3.2 Kerala versus Senegal and Mauritania
10
Next, I compare the Indian state of Kerala with two other coun- 0
Liberia
Madagascar
Burkina Faso
Mauritania
Burundi
Dem Rep of Congo
Malawi
Sierra Leone
Ethiopia
Niger
Eritera
Guinea
Mozambique
Chad
Gambia, The
Togo
Zambia
Central African Republic
Zimbabwe
Guinea-Bissau
Nigeria
Uganda
Rwanda
Mali
Kenya
Benin
Lesotho
Comoros
Tanzania
Senegal
Ghana
Cote d’lvoire
Cameroon
India
tries from Sub-Saharan Africa, Senegal and Mauritania. Of the
28 states in India, I choose Kerala to bring out the absurdity of
the current child malnutrition indicators as sharply as possible.
The conventional vital health statistics in Kerala are the high-
est among all Indian states and rival those observed in China. Figure 2: Infant Mortality Rates in India and 33 Poorer Sub-Saharan
Among the largest 17 Indian states, it ranks fourth in terms of African Countries
Infant Mortality per 1,000 Live Births (2009): Rising Per Capita GDP from L to R
per capita income. In terms of per capita income, Senegal and 140
126 123 124
Mauritania are among the better-off countries in Sub-Saharan 120
112
115
100 101 101
Africa but both lag behind India and Kerala with the gap being 80
96
91
95
88 86 83 86
especially large with respect to the latter.4 60
80 76 78 79 75 75
74 68
69 67 70
Table 2: Comparing Kerala in India with Senegal and Mauritania in Africa 50 64 56 55 61
40 51 47 50
Indicator Kerala Senegal Mauritania 39 40
30
Life expectancy 74 62 58 20
Infant mortality per 1,000 live births 12 51 74 10
Under-five mortality per 1,000 live births* 16 93 117 0
Burundi
Dem Rep of Congo
Liberia
Malawi
Sierra Leone
Ethiopia
Niger
Eritera
Guinea
Mozambique
Gambia,The
Togo
Madagascar
Zambia
Central African Republic
Uganda
Zimbabwe
Burkina Faso
Guinea-Bissau
Rwanda
Chad
Mali
Kenya
Benin
Lesotho
Comoros
Mauritania
Tanzania
Senegal
Nigeria
Ghana
Cote d’lvoire
Cameroon
India
Maternal mortality per 1,00,000 live births (2009) 95 410 550
Per cent children below 5 stunted (2000-09) 25.0 20.0 24.2
Per cent children below 5 underweight (2000-09) 23.0 15.0 16.7
Source: National Family Health Survey 2005-06 for Kerala. All other data from World Health
Statistics, 2011.
According to Table 2, Senegal, which has 4.25 times the in- Figure 3: Under-Five Mortality Rates in India and 33 Poorer Sub-Saharan
African Countries
fant mortality rate of Kerala, almost six times Kerala’s under- Under-Five Mortality per 1,000 Live Births (2009): Rising Per Capita GDP from L to R
250
five mortality, and 4.3 times Kerala’s maternal mortality ratio,
199 209
193 191
has lower rates of stunting and underweight children. Chil- 200 192
166 171
160 166
dren in Senegal, better nourished as per malnutrition esti- 142142 141 138154
150
mates, die at rates many times those in Kerala. A comparison 112110
128
111 118 117 118
104 103 98 104 108
93
with Mauritania yields the same picture. 100 89 84 84
69 66
58
A higher incidence of child malnutrition in Kerala than 50
55
Senegal and Mauritania is even more puzzling given its signifi-
cantly higher female literacy rate. The state has had a long his- 0
Burundi
Liberia
Dem Rep of Congo
Malawi
Sierra Leone
Ethiopia
Niger
Eritera
Guinea
Mozambique
Gambia,The
Togo
Madagascar
Zambia
Central African Republic
Uganda
Zimbabwe
Burkina Faso
Guinea-Bissau
Rwanda
Chad
Mali
Kenya
Benin
Lesotho
Comoros
Mauritania
Tanzania
Senegal
Ghana
Cote d’lvoire
Nigeria
Cameroon
India
tory of educating its women and its female literacy rate at 92%
in 2011 is among the highest in the developing world. In addi-
tion, women have traditionally enjoyed high social status in
Kerala with many communities following the matrilineal tra-
dition. In contrast, at 29%, Senegal has one of the lowest The life expectancy at birth in India at 65 exceeds those in all
female literacy rates in the world. Mauritania does better at but two of the 33 Sub-Saharan African countries (at 66 years,
51%, but it also lags far behind Kerala. Eritrea edges out India, while at 65 Madagascar ties with it).
100 may 4, 2013 vol xlviiI no 18 EPW Economic & Political WeeklySPECIAL ARTICLE
Figure 4: Still Birth Rates in India and 33 Poorer Sub-Saharan The maternal mortality ratio per 1,00,000 live births in In-
African Countries
Stillbirth per 1,000 Live Births (2009): Rising Per Capita GDP from L to R
dia at 230 is lower than those in every one of the 33 Sub-Saha-
45 42 ran African countries.
40
34
But this pattern collapses when it comes to child malnutrition.
35
30 29 30 30 29 The proportion of children under five years of age classified as
28 27 26 28 26 26 26 27 27 26 27 26
25 24 25 24 25
25 24 23
21
24
21
23
23 22 22 22
stunted (low height for age) at 47.9% is higher in India than all but
20
20 six of the poorer Sub-Saharan African countries (Burundi, Malawi,
15 Ethiopia, Niger, Madagascar, and Rwanda have stunting rates of
10
63.1%, 53.2%, 50.7%, 54.8%, 49.2% and 51.7%, respectively).
5
0 Figure 7: Percentage of Children Underweight in India and 33 Poorer
Liberia
Madagascar
Burkina Faso
Mauritania
Burundi
Dem Rep of Congo
Malawi
Sierra Leone
Ethiopia
Niger
Eritera
Guinea
Mozambique
Gambia, The
Togo
Zambia
Central African Republic
Zimbabwe
Guinea-Bissau
Rwanda
Chad
Cote d’lvoire
Nigeria
Cameroon
India
Uganda
Mali
Kenya
Benin
Lesotho
Comoros
Tanzania
Senegal
Ghana
Sub-Saharan African Countries
Percentage of Children below 5 Underweight (2000-09): Rising Per Capita GDP from L to R
50
43.5
45
38.9 39.9
40 37.4
34.6 34.5 33.9
35
28.2
Figure 5: Maternal Mortality Ratio in India and 33 Poorer Sub-Saharan 30 27.9 26.7
25.0
African Countries 25 20.4 21.3 21.2 20.5 21.8
18.0 20.2
20.8 16.7
Maternal Mortality per 1,00,000 Live Births (2009): Rising Per Capita GDP from L to R 20 15.5 15.8 14.9 16.414.0 17.4 16.4 16.6 16.7 14.5 16.7 16.6
14.3
1400 15
1200
1200 10
970 990 970 1000 5
1000
820 850 830 0
Liberia
Madagascar
Burkina Faso
Mauritania
Burundi
Dem Rep of Congo
Malawi
Sierra Leone
Ethiopia
Niger
Eritera
Guinea
Mozambique
Central African Republic
Gambia, The
Togo
Zambia
Zimbabwe
Guinea-Bissau
Rwanda
Chad
Tanzania
Uganda
Mali
Kenya
Benin
Lesotho
Comoros
Senegal
Ghana
Cote d’lvoire
Nigeria
Cameroon
India
790 790 840
800
670 680
550 560 540 530 530 550 600
600 510 470 470 470
400 350440 430 410
340
410
350
400
280
230
200
0
The proportion of children under five years of age classified as
Liberia
Madagascar
Burkina Faso
Mauritania
Burundi
Dem Rep of Congo
Malawi
Sierra Leone
Ethiopia
Niger
Eritera
Guinea
Mozambique
Gambia, The
Togo
Zambia
Central African Republic
Zimbabwe
Guinea-Bissau
Rwanda
Chad
Tanzania
Uganda
Mali
Kenya
Benin
Lesotho
Comoros
Senegal
Ghana
Cote d’lvoire
Nigeria
Cameroon
India
underweight (low weight for age) at 43.5% is higher in India
than every one of the 33 poorer Sub-Saharan African countries.
4 Can Superior Health Infrastructure Be the Explanation?
Figure 6: Percentage of Children Stunted in India and 33 Poorer Sub-Saharan When confronted with this evidence, some proponents of the
African Countries current malnutrition indicators respond that the lower rates of
Percentage of Children below 5 Stunted (2000-09)
70 infant and child mortality in India relative to Sub-Saharan
63.1
60
53.2 54.8
51.7
Africa despite higher malnutrition rates reflect its superior
50.7 47.0 49.2 47.7 46.9 47.9
50 45.8 43.7 45.8 44.6 44.5 44.8 44.7 45.2 44.4 medical infrastructure. In addition to contributing to low mor-
39.4 40.0 41.0
40.1
40 37.4 38.7
35.8
38.5
35.2 36.4 tality rates, the latter contributes to increased malnutrition by
30
27.6
26.9 24.2
28.6 helping save many malnourished infants and children.5
20
20.1 Though a logical possibility, the authors of this explanation
10
provide no concrete empirical evidence in its support. What-
ever evidence we can gather points in the opposite direction –
0
countries that have better medical infrastructure also nourish
Liberia
Madagascar
Burkina Faso
Mauritania
Burundi
Dem Rep of Congo
Malawi
Sierra Leone
Ethiopia
Niger
Central African Republic
Eritera
Guinea
Mozambique
Gambia, The
Togo
Zambia
Zimbabwe
Guinea-Bissau
Uganda
Rwanda
Chad
Mali
Kenya
Benin
Lesotho
Comoros
Tanzania
Senegal
Ghana
Cote d’lvoire
Nigeria
Cameroon
India
Figure 8: Malnutrition among Children Above 1 and Below 5 Years in Rural
Areas of Nine States under the NCHS 1977 Standard
90
77 79
80
70 69
The infant mortality rate per 1,000 live births in India at 50 62
65
58
is lower than those in all but three of the 33 Sub-Saharan Afri- 60
55
52
can countries (Eritrea, Madagascar, and Ghana have infant 50
mortality rates of 39, 40, and 47, respectively). 40
The under-five mortality rate per 1,000 live births in India at
30
66 is lower than those in all but two of the 33 Sub-Saharan
African countries (Eritrea and Madagascar have under-five 20
mortality rates of 55 and 58, respectively). 10
The stillbirth rate per 1,000 births at 22 in India is lower 0 1975-79 1988-90 1996-97 2003-06 1975-79 1988-90 1996-97 2003-06
than those in all but five of the 33 Sub-Saharan African coun- Proportion of 1-5 years old children underweight Proportion of 1-5 years old children stunted
tries (Eritrea, Madagascar, Zimbabwe, Kenya, and Ghana have Source: Authors’ construction based on NNMB (1999), Report of Second Repeat Survey-Rural,
Indian Council of Medical Research, Hyderabad, Table 19 and NNMB Fact Sheet 2003-06 at
stillbirth rates of 21, 21, 20, 22 and 22, respectively). http://www.nnmbindia.org/downloads.html (accessed 27 June 2011).
Economic & Political Weekly EPW may 4, 2013 vol xlviiI no 18 101SPECIAL ARTICLE
their children better. Developed countries enjoy low infant Three important indicators of child malnutrition are con-
and child mortality rates and maternal mortality ratios as well ventionally reported – the proportion of children stunted,
as low rates of malnutrition. Likewise, declining rates of mal- those underweight, and those wasted. As already indicated,
nutrition typically accompany declining infant and child mor- stunting and underweight refer to low height and low weight
tality rates and maternal mortality ratios. for age, respectively. Wasting refers to low weight for a given
India’s own experience points to improving malnutrition side- height, regardless of age. Since wasting has received little at-
by-side with declining infant and child mortality rates and ma- tention in the public policy discourse, I will not focus on it in
ternal mortality ratio. Figure 8 (p 101) shows the average pro- the rest of this paper.6 Indeed, for clarity of exposition, I will
portions of children classified as underweight and stunted in present much of the discussion with respect to the determina-
rural areas in nine Indian states during four time periods – 1975- tion of stunting.
79, 1988-90, 1996-97, and 2003-06. The proportion of children The height of an individual can vary for both genetic and
stunted as well as those underweight declines between every nutritional reasons. Without detailed medical examination,
successive pair of periods. If improved ability to save the infants one cannot conclude whether an individual is short because of
and children as reflected in declining infant and child mortality malnourishment or because of genetic factors. This makes
rates are expected to result in worsening child nutrition per- identifying stunting by referring to just height, an imperfect
formance, we should observe rising rates of malnutrition in exercise. Nevertheless, this is the current practice.
India. But we see precisely the opposite trend in Figure 8. The current approach is to define a height norm for children
Figure 9: Child Malnutrition and Under-Five Mortality of a given age and gender. All children of the same age and
70 gender in a given population who are below this norm are
Percentage of under-five children classified as stunted. The critical remaining step then is to
underweight, 2005-06
identify the height norm. In the strictest sense, the norms cur-
60
rently used are premised on the following key assumption – all
Percentage of children under-five underweight, 2005-06
differences in height between populations of children of a
50
given age and sex occur due to differences in nutrition.
y = 0.3024X + 19.145
R2 = 0.51528 This assumption implies that populations of children from
entirely different races, ethnicities, cultures, time periods, and
40 geographical locations would look identical in terms of height
distribution provided they are given the same nutrition. That
is to say no differences in the proportion of children below or
30 above any pre-specified height would exist between two popu-
lations provided they are given identical nutrition.
Linear (Percentage of under-five children Suppose we can identify the population of children of a
20
under weight, 2005-06) given age and sex that is the healthiest possible. Although the
heights of children within this population will differ due to
10 genetic differences, taken as a whole, the population repre-
sents the best attainable distribution. Then, given the above
assumption, any deviations in the distribution of height in a
0 population of children of the same age and sex from this popu-
0 20 40 60 80 100 120
Under-five mortality per thousand live births, 2001-05 lation would be attributable to malnutrition. This is the es-
Source: Author’s construction based on data from National Family Health Survey 2005-06. sence of the approach underlying the measures of malnutri-
This same pattern is observed in cross-state data in India. tion currently in use worldwide.
Figure 9, which shows the proportion of children who are un- The first step in making the approach operational is to iden-
derweight against the under-five mortality rate in the 15 larg- tify the healthiest populations of boys and girls of different
est states, illustrates this graphically. The scatter-plot shows an ages. Once this is accomplished, a certain percentage of chil-
upward trend, which is confirmed by the fitted linear trend. dren at the bottom of the distribution of this population by
On average, states with superior outcomes in child mortality height are defined as stunted. Based on statistical considera-
rates also exhibit superior nutrition outcomes. Once again, on tions, the conventional cut-off point for this purpose is set at
average, a greater ability to save infants and children does not 2.3%.7 The height of the child at the 2.3 percentile in the
translate into a higher incidence of malnutrition. healthiest population serves as the norm against which all
children are measured to determine their stunting status.
5 Measuring Malnutrition: Methodology What is required then is the identification of the healthiest
It is the contention of this paper that the answer to the para- population of children of a given age and gender, or what is often
doxical behaviour of stunting and underweight indicators in called the “reference population”. The US first adopted such a
the India-Sub-Saharan Africa comparison lies in the method- reference population in 1977. The National Center for Health
ology underlying the measurement of these indicators. This is Statistics (NCHS) of the Centers for Disease Control (CDC) deve-
what is developed and defended in the rest of this paper. loped the height and weight distributions of children by age
102 may 4, 2013 vol xlviiI no 18 EPW Economic & Political WeeklySPECIAL ARTICLE
and sex using longitudinal data collected in Yellow Springs, Figure 10: Hypothetical Cumulative Height Distributions of Five-Year-Old
Boys in India
Ohio between 1929 and 1975 by the Fels Research Institute Percentage of population
(Roche 1992). The NCHS 1977 distributions remained in use to below the height shown
measure malnutrition among children in the US until 2000.
Beginning in the late 1970s, the WHO encouraged other countries
to adopt this same reference population to measure malnutrition. Distribution 1 Distribution 2
In the 1990s, the CDC concluded that the Fels reference pop- 100
ulation data came from a sample that was quite limited in
genetic, geographic, cultural, and socio-economic variability Distribution 3
(Kuczmarski et al 2002: 2-3). It therefore replaced the NCHS
50
1977 charts by CDC 2000 charts that were based on a nation- Distribution 4
ally representative sample in which infants came from a 30
broader spectrum of racial/ethnic groups, socio-economic 15
backgrounds, and modes of infant feeding. 2.3
The discussions surrounding this change led the WHO to deve- O S Height in inches
lop its own height and weight standards on the basis of a more Key: (i) Distribution 1: Actual height distribution of five-year-old boys in a given year
(ii) Distribution 2: Height distribution of the same five-year-old boys achievable in the
diverse reference population. It collected data on 8,440 healthy same year with balanced diet for all (iii) Distribution 3: The best height distribution
breastfed infants and young children from Brazil, Ghana, India, achievable for five-year-old boys of a future generations of these children
(iv) Distribution 4: Height distribution of five-year-old boys in the reference population.
Norway, Oman, and the US and adopted new standards in
2006. Almost all developing countries, including India, now Figure 10 illustrates the above points with the help of four
use these WHO 2006 standards to measure malnutrition. The strictly hypothetical population distributions of five-year-old
estimate that approximately half of Indian children are mal- boys. On the horizontal axis, we measure height in inches. On
nourished is based on an application of these standards. the vertical axis, we measure cumulative population below the
height shown on the horizontal axis. The curve labelled “Dis-
6 The Key Elements of the Critique tribution 4” represents the tallest population anywhere in the
Central to the present critique is a challenge to the assumption world and serves as the reference population. Since 2.3% of
that the provision of a fully balanced diet will eliminate the this population is below the height labelled S, the height at
height and weight differences between the population of Indian point S serves as the norm.
children and the healthiest existing population of children “Distribution 1” gives the observed height distribution of a
any where, which is currently represented by the WHO refer- hypothetical population of five-year-old boys in a given year.
ence population. Potentially, the failure can occur at two levels. As shown, 50% of these boys have heights below point S and
First, we have what has been called the “gradual catch-up” are therefore classified as malnourished. “Distribution 2”
hypothesis whereby it takes several generations of balanced shows the height distribution that the population can achieve
diet for a population of children to achieve its full potential if every boy in it is given a fully balanced diet. It shows that
height and weight.8 Put differently, even if a fully balanced even after every boy is given a balanced diet, we would clas-
diet replaces the status quo diet of an entire cohort of mal- sify 30% of them as stunted. Finally, “Distribution 3” shows the
nourished children, it can achieve only so much improvement “maximum-height” distribution that the future generations of
in height and weight outcomes.9 Weak and malnourished this population can achieve after eliminating entirely the
mothers give birth to children with height and weight disad- “catch-up” deficit through a sustained balanced diet. If the
vantages that a balanced diet cannot fully eliminate. Prema- genetic potential of this population is below that of the refer-
ture births and lack of proper care during pregnancy give rise ence population, “Distribution 3” will lie above “Distribution
to similar problems. Therefore, what we may call the “catch- 4”, and if not, it will coincide with or lie below the latter. As
up” deficit takes several generations to eliminate. shown, “Distribution 3” is strictly above the reference popu-
Second, there is the possibility that a specific population of lation distribution with 15% of the boys still classified as
children is genetically shorter than the children in the refer- malnourished.
ence population. This means that even after the population There is broad agreement that cumulative height “Distribu-
has fully eliminated the “catch-up” deficit after several genera- tion 2” in India lies above “Distribution 4”. That is to say, even
tions of a balanced diet, it will still fall short of reproducing the if the entire current population of children were given a bal-
reference population. An example, discussed at length later in anced diet, it still would not achieve the height distribution of
the paper, is that Japanese adults have grown much taller the reference population. But almost all analysts explicitly or
on average over the generations. They have, nevertheless, implicitly see nothing wrong with the current approach under
remained 12 to 13 centimetres shorter than their Dutch coun- which “Distribution 2” would lead us to classify 30% of India’s
terparts. Poor nutrition and the “catch-up” deficit cannot ex- children as malnourished.10 At least from the viewpoint of policy
plain this height difference between Japanese and Dutch formulation, we need to make a distinction between the 20%
adults unless one argues that the Japanese are still in the in this example who can cross the threshold after being given a
“catch-up” process. balanced diet and the remaining 30% who are classified as
Economic & Political Weekly EPW may 4, 2013 vol xlviiI no 18 103SPECIAL ARTICLE
malnourished despite receiving such a diet. Without such a with a separate room used as kitchen and whose family owns a
distinction, we would run the risk of biasing policy towards car, colour television, telephone, and refrigerator”. This nar-
obesity for this 30% of the population. rowing down shrinks the sample to a mere 212 elite or privi-
The dominant view around the world is that “Distribution 3” leged children in India. Continuing to apply WHO 2006 growth
coincides with “Distribution 4”. That is to say, a balanced diet charts, even in this group, 20% children remain stunted and
over several generations will lead to height distribution of 9.4% remain underweight.
every population of children becoming coincident with the A follow-up report by the Government of India (2009) anal-
distribution of the tallest population in the world. In this view yses the data from NFHS-3 using an even stricter definition of
there are no genetic differences between populations as far as elite children. It defines them as children “whose mothers and
height and weight are concerned. fathers have secondary or higher education, who live in house-
I will document below substantial evidence from medical holds with electricity, a refrigerator, a TV, and an automobile
and other literature pointing to genetic differences across or truck, who did not have diarrhoea or a cough or fever in the
populations. But as a preliminary point, it may be noted that two weeks preceding the survey, who were exclusively breast-
when recommending the switch from NCHS 1977 standards to fed if they were less than five months old, and who received
CDC 2000 standards, even the CDC cited limited genetic, geo- complementary foods if they were at least five months old”
graphic, cultural, and socio-economic variability in the former (GOI 2009: 10). Applying WHO 2006 standards, the report esti-
sample as a key reason for its recommendation (Kuczmarski et al mates the proportion of stunted children among these elite
2002: 2-3). The argument rationalising the shift, thus, in essence children to be approximately 15%.
acknowledged the relevance of genetic factors. Against this Prima facie, these findings imply that even if the popula-
background, we must also ask whether the WHO 2006 sample, tions of children underlying the NFHS-2 and NFHS-3 were pro-
collected from countries as diverse as Brazil, Ghana, India, vided a balanced diet and other amenities that lead to good
Norway, Oman, and the US, adequately represents the popula- height and weight outcomes, 15% to 20% of them will remain
tion of India or any other country in terms of their genetic, stunted by the WHO 2006 standards. It can be further argued
geographical, cultural, and socio-economic backgrounds. that even these percentages understate the extent of stunting
despite a healthy diet due to two possible selection problems.
7 Evidence from Indian Data First, given that wealth persists over generations, the elite
The two standards that the WHO has recommended over the children identified in Tarozzi (2008) and GOI (2009) are prob-
decades – NCHS 1977 and WHO 2006 – lead to substantially dif- ably farther along the “catch-up” curve than the rest of the
ferent levels of measured malnutrition. Using the sample of population. Therefore, even if the non-elite children were
children under 3 years of age in the second round of the given the same diet and other amenities, they would exhibit a
National Family Health Survey (NFHS-2), Tarozzi (2008) esti- higher incidence of stunting and underweight than their elite
mates that the NCHS 1977 standard leads to classifying 42% of counterparts. Second, it is also possible that genetically taller
these children as stunted. But when the WHO 2006 standard is children are represented in disproportionately large numbers
applied to the same sample, the estimate rises to 48%. One can in the populations of elite children analysed by Tarozzi (2008)
imagine that over time populations in the same countries and GOI (2009). This may result, for example, from genetically
from which the WHO has drawn its sample will become taller individuals achieving success in disproportionately
healthier, yielding an even higher standard, which would turn larger numbers during the earlier part of India’s development
yet more children in the NFHS-2 sample from well-nourished process and holding on to their lead.
to malnourished. These findings and arguments show that the absence of a
Indeed, the problem turns out to be far deeper than what balanced diet alone cannot fully explain the estimates of
these remarks suggest. NFHS-2 data divide the families in the stunted and underweight children in India. The “gradual
sample into three wealth categories – high, medium and low – catch-up” hypothesis or genetic differences or both are at work
on the basis of a standard of living index (SLI) constructed as well. Deaton and Dreze (2009), who carefully review the
from ownership of a large number of assets and other wealth findings of Tarozzi (2008), reach the same conclusion. They
indicators. With the help of this index, Tarozzi isolates families discuss the possibility of genetic differences but favour the
in the high wealth category. This brings down the number of “gradual catch-up” hypothesis as the explanation for the high
children in the sample from tens of thousands to approxi- proportions of stunted and underweight children even among
mately 5,100. Measuring against WHO 2006 growth charts, the elite. They state, “The genetic potential hypothesis, al-
Tarozzi (2008: Table 4, last row) finds that both among boys though certainly not disproved, is becoming less accepted in
and girls in this group, approximately one-third remain the scientific literature, if only because there is a long history
stunted and one-quarter underweight. of differences in population heights that were presumed to be
Tarozzi (2008: 463) explores the issue further by “using genetic, and that vanished in the face of improved nutrition.”11
only information from families where malnutrition should be I will argue that while the “gradual catch-up” hypothesis is
unlikely”. Out of the 5,100 children in high SLI families, he definitely at work in India, it is insufficient to explain the
selects those “from urban areas, where both parents have at differences in the incidence of child malnutrition between
least a high school diploma, live in a house with a flush toilet India and Sub-Saharan Africa. Genetic differences remain a
104 may 4, 2013 vol xlviiI no 18 EPW Economic & Political WeeklySPECIAL ARTICLE
necessary part of the full explanation of these differences. I This evidence aside, even the references cited by Deaton
reinforce this conclusion by providing evidence of genetic dif- and Dreze (2009) – Cole (2003) and Nube (2008) – do not sup-
ferences across populations around the world. Table 3: Height Differences across High-income port the proposi-
Countries (cm) tion that the dif-
8 Height Differences Have Not Vanished: Adults Country Male Female Age Year
ferences between
Netherlands 183.2 169.9 20-30 2010
I noted above that the justification Deaton and Dreze (2009) heights of different
Sweden 181.5 166.8 20-29 2008
provide for the rejection of genetic differences across popula- populations vanish
Germany 181 168 18-25 2009
tions is that “there is a long history of differences in population with improved nu-
US 177.6 163.2 20-29 2003-06
heights that were presumed to be genetic, and that vanished in United Kingdom 177.1 164.4 16-24 2010
trition. Cole (2003:
the face of improved nutrition” (emphasis added). In providing Canada 176 163.3 25-44 2005 162) documents the
this justification, the authors do not specify if they have in South Korea 173.7 161.1 17-18 2011 steady increases in
mind here the differences in adult or child heights or heights at Portugal 173.7 163.7 21 2001 height over genera-
all ages. But since they refer to the contributions by Cole Japan 170.7 158 17 2011 tions in countries
(2003) and Nube (2008) in this context and these authors con- Singapore 170.6 160 17-25 2003 such as the US, the
Source: Excerpted from the table in http://en.wikipedia.
sider both adult and child nutrition, it is appropriate to con- org/wiki/Human_height (accessed 11 October 2012),
Netherlands, and
sider both populations. which also gives the original sources of the data. Japan but makes no
Begin with the evidence on whether improved nutrition claims that these
over several generations causes the differences in adult differences eventually vanish. Indeed, he explicitly notes,
heights to vanish. In his lively essay entitled “The Height Gap” “Height in the USA, the most affluent nation, currently lags
in the New Yorker, Burkhard Bilger (2004) traces the fascinat- behind that in Northern Europe”. He goes on to state, “These
ing history of the literature on the subject. Going by his differences are substantial”.
account, evidence supporting the hypothesis of improved Nube (2008) does not analyse the height dimension of nutri-
nutrition leading to the elimination of height difference remains tion and instead focuses on body mass index (BMI). He specifi-
the Holy Grail of researchers in this field. Bilger reports that cally focuses on south Asians living in various parts of the
US soldiers were two inches taller than the average German world and reaches the conclusion that genetic factors are par-
during the first world war. tially responsible for the low BMI among them. It is instructive
But sometime around 1955 the situation began to reverse [with Ger- to quote a key paragraph from his paper in its entirety.
mans surpassing the Americans in height]. The Germans and other Results from countries that are home to sizeable population segments
Europeans went on to grow an extra two centimetres a decade, and from different ethnic backgrounds, including people of Asian and Af-
some Asian populations several times more, yet Americans haven’t rican descent, reveal consistently higher prevalence rates of low BMI
grown taller in 50 years. By now, even the Japanese – once the shortest among people of South Asian descent. These differences cannot be
industrialised people on earth – have nearly caught up with us [Ameri- explained on the basis of indicators which relate to access to food, so-
cans], and Northern Europeans are three inches taller and rising cial status of women or overall standard of living. Apart from the pre-
(2004: 7; emphasis added). sented results on South Africa, Fiji and the USA, similar results are also
John Komlos, a professor of economics at the University of reported for England, although in these reports information on the
socio-economic status of the various ethnic population segments is not
Michigan and a pioneer in the field, has thoroughly analysed
presented. On the basis of these outcomes it is hypothesised that there
the data for signs of catch-up by US adults but found none. In exists among adults of South Asian descent an ethnically determined pre-
the words of Bilger, disposition for low adult BMI. This ethnic predisposition can be based
on both genetic and cultural factors (2008: 512; emphasis added).
But recently he [Komlos] has scoured his data for people who’ve
bucked the national trend. He has subdivided the country’s heights by Interestingly, in an earlier paper, Deaton (2007) himself
race, sex, income, and education. He has looked at whites alone, at analyses height data from 43 developing countries and finds
blacks alone, at people with advanced degrees and those in the highest that no variables, including those relating to nutrition as meas-
income bracket. Somewhere in the United States, he thinks, there ured by calorie intake, explain the differences across coun-
must be a group that’s both so privileged and so socially insulated that
tries. He finds the high stature of Africans the hardest to ex-
it’s growing taller. He has yet to find one (2004: 10).
plain, admitting, “Perhaps the major puzzle is why Africans
Adult height differences magnify as we expand the compari- are so tall” (ibid: 132-35). Variables such as per capita income
son to a larger group of high-income countries. Table 3 reports in childhood, incidence of infant and child mortality rates, per
comparable heights of men and women in several of these capita calorie availability, and mother’s education, conven-
countries with the countries arranged in declining order of tionally considered to correlate with height, all fail to explain
height of men. Male height in the Netherlands is shown to be the exceptionally tall stature of African men and women.
12.5 cm greater than in Japan. Even the gap between male Unable to resolve the puzzle, Deaton goes so far as to state,
heights in the Netherlands and Portugal is 9.5 cm. Similar dif- Given that Africans are deprived in almost all dimensions, yet are taller
than less-deprived people elsewhere (although not more than Europeans
ferences exist in female heights. In broad terms, both men and
or Americans), it is difficult not to speculate about the importance of
women in northern Europe are the tallest and those in Asia the possible genetic differences in population heights. Africans are tall de-
shortest. An interesting observation is that South Korea has spite all of the factors that are supposed to explain height (2007:132-36).
overtaken both Japan and Singapore even though its per capita But he stops short of accepting the genetic factor as an
income is still far below that of the latter. explanation, arguing that it does not explain the differences
Economic & Political Weekly EPW may 4, 2013 vol xlviiI no 18 105SPECIAL ARTICLE
between other populations. In effect, he leaves unexplained are at least two problems with this argument. First, as I
the tall stature of Africans despite greater deprivation relative have already documented, as a matter of general proposition,
to other countries. adult differences in heights persist across races and ethni-
cities. Second, the pattern found for AIA children by Alexander
9 Height Differences between Child Populations et al (2007) has also been observed for Japanese-American
It is puzzling that despite having discussed at length the inex- children.
plicably tall stature of African adults relative to those from In particular, Mor, Alexander, Kogan, Kieffer and Ichiho
other poor countries in Deaton (2007), Deaton and Dreze (1995) compare the birth outcomes of US-resident white and
(2009) makes no attempt to draw out its implications for the Japanese-American mothers using 1979-90 linked live birth
puzzle of lower incidence of stunting among children in nearly and infant death records from the state of Hawaii. The major-
all Sub-Saharan African countries than in India. One imagines ity of these Japanese-American mothers were born in Hawaii
that the two puzzles are intimately linked. But nowhere in the and the majority of the white mothers were born in the main-
paper do the authors mention this possibility. land US. Summarising their findings, the authors state,
A possible explanation of this oversight may be the belief After controlling for maternal socio-demographic and prenatal care
that differences in heights for reasons other than nutrition do factors with logistic regression, Japanese-American infants had sig-
not appear in childhood. But evidence fails to support this nificantly higher risks of low birth weight, preterm and very preterm
proposition as well. Height and weight differences can be birth and of being small-for-gestational age.
found even between populations of newborns who are other- It is difficult to attribute these differences to a “catch-up”
wise perfectly equally healthy. deficit among the Japanese-American mothers, especially
since the infant mortality rates for the Japanese children, like
9.1 Height and Weight Differences between Equally those for the American children, were reported by the authors
Healthy Populations of Newborns to be below the US Year 2000 Health Objective.
In a paper entitled “Birth Outcomes of Asian-Indian-Ameri-
cans”, Alexander, Wingate, Mor and Boulet (2007) compare 9.2 Older Children and Persistent Height Differences
infants born in the US to resident Asian-Indian-American Systematic studies of older children of migrant populations
(AIA) mothers to those born to resident non-Hispanic white settled in the developed countries provide additional evidence
and non-Hispanic African-American (AA) mothers. The sam- of persistent differences across populations despite improved
ple includes more than 1,00,000 AIA children, more than nutrition over some generations. Fredriks et al (2004) col-
three million white children, and more than one million AA lected cross-sectional growth and demographic data on 2,880
children. The authors are also able to control for the relevant children of Moroccan origin and 14,500 children of Dutch ori-
maternal characteristics. They summarise their key findings gin living in the Netherlands in the age range 0 to 20 years in
as follows. 1997. Their findings are consistent with our previous discus-
Compared to AAs or Whites, AIAs have the lowest percentage of births to sion. “Moroccan young adults were on average 9 cm shorter
teen or unmarried mothers and mothers with high parity for age or with than their Dutch contemporaries. …Height differences in com-
low educational attainment. After taking these factors into account, AIA parison with Dutch children increase from 2 years onwards.”
had the highest risk of LBW [low birth weight], small-for-gestational age
These authors find the differences so compelling that they rec-
(SGA) and term SGA births but a risk of infant death only slightly higher
than Whites and far less than AAs. Conclusions: The birth outcomes of
ommend drawing up separate growth charts for Moroccan
AIAs do not follow the paradigm that more impoverished minority popu- and Dutch children.
lations should have greater proportions of low birth weight and preterm Indeed, today, it is possible to find separate growth charts
births and accordingly greater infant mortality rates. for children of Moroccan and Dutch origin living in the Neth-
The authors speculate that the small body size and low birth erlands, making it possible to compare the two populations.12
weight of AIA children may be due to either “certain genetic Table 4 reports mean heights in centimetres in 2010 for these
factors related to the shortness or smaller size of the mother two populations. Differences are minimal at the first year but
caused by undernourishment occurring during childhood” or positive and rising from the second years onwards. By the
“a different body habitus among this ethnic group and maybe third year, the difference is a full centimetre and grows to 1.8
due to genetic factors, not suboptimal growth”. Whatever the cm for boys and 2.7 cm for girls in the fifth year. By the fifth
reason, the authors’ findings are that the AIA children are fully Table 4: Height of Moroccan and Dutch Children in the Netherlands
caught up with white and AA children of similar socio-eco- in 2010 (in cm)
nomic and demographic backgrounds in terms of infant mor- Age in Years Boys Girls
Moroccan Dutch Difference Moroccan Dutch Difference
tality but continue to exhibit higher incidence of low birth
1 76.1 76.7 0.6 75 75 0
weight and small size for gestational age.
2 87.7 88.4 0.7 86.5 87.1 0.6
Even so, it is tempting to invoke the “gradual catch-up” hy-
3 96.8 97.8 1 96 97 1
pothesis and argue that over several generations, Indian
4 104.5 105.5 1 103.5 104.9 1.4
mothers will catch up with the American mothers in height 5 111.4 113.2 1.8 110.2 112.9 2.7
and weight, thus bridging the size and weight gap between 21 177.8 183.8 6 162.8 170.7 7.9
the two sets of children that is currently observed. But there Source: www.tno.nl
106 may 4, 2013 vol xlviiI no 18 EPW Economic & Political WeeklySPECIAL ARTICLE
year, the gaps are thus almost a third of the gaps obtained at wrong for a number of reasons, some of which Tarozzi is him-
full adulthood – 6 cm for boys and 7.9 cm for girls. self careful to note.
Fredriks et al (2004) are not alone in finding persistent For starters, observe the qualification “prima facie” in the
differences between populations of children of migrants in de- statement. Tarozzi is tentative and by no means conclusive in his
veloped countries and those of local families. Smith et al tone. And there are good reasons for this caution. The sample
(2003), who compare the heights and weights of 6-12 year old with which he works is extremely inadequate to draw strong in-
Maya-American children using samples collected at two dif- ferences about the absence of genetic differences. Thus, for ex-
ferent points in time with the National Health and Nutrition ample, the sample of children under 2 years of age born to Indian
Examination Survey (NHANES) reference standards for US parents in the data set available to him is so small that he does
children, also find the height gap narrowing over time but not even attempt a comparison between them and children of
not vanishing. the same age born to white parents. For children 2 to 3 years old,
As many as half a million Guatemalan Maya, mostly from his sample has just 19 Indian children and for those between 2
rural areas, have migrated to the US since the civil war in Gua- and 5 years, it has 72 children. Such small samples are quite
temala in 1978. The bulk of this migration took place in the inadequate to measure even the average levels of stunted and
1980s. Smith et al compare the heights of Maya children living underweight children with any degree of precision, let alone the
in the US in 1992 and 2000 with their Guatemalan counter- entire distribution of the underlying population.
parts as well as the NHANES standard for American children. Moreover, even these small samples do not yield zero differ-
They find that 6-year-old Maya children living in the US in ences between stunting levels among children born to Indian
1992 were on an average 6 cm taller than their 1998 Guatema- parents and those to local white parents. The proportion of
lan counterparts. They had gained another 3 cm after eight Indian children, 2 to 3 years old, who are placed in the stunted
years in 2000. Nevertheless, they remained 5 cm shorter than category by the WHO 2006 definition is 5.3% compared with
the NHANES standard for American children in 2000. nil among white children. Surely, the difference between 5.3%
A defender of uniform worldwide norms for measuring nu- and 0% is not zero. Moreover, if we were to make the height
trition may argue that the reason Moroccan children in the norm against which stunting is evaluated even more demand-
Netherlands and Maya children in the US lag behind their host ing than the WHO 2006 norm, the proportion of Indian chil-
country counterparts in height is that they still have not had dren who are stunted would rise, whereas it may still remain
enough time to eliminate the “catch-up” deficit. It is possible that zero among white children.13
the remaining gap will be eliminated in another few generations. There are more qualifications to the conclusion by Tarozzi.
But this argument has two limitations. First, given that adult Even if it were true that the height gap between Indian children
height differences across developed country populations have born in the UK and their white counterparts is nil, it does not
persisted, as has the incidence of low birth height and low prove that at some point in time Indian children born and brought
birth weight between Japanese and American children born in up in India will also close the gap. There are at least two reasons
Hawaii, how can we be sure that the height and weight differ- for this conclusion. First, there may be a selection problem such
ences between children will vanish in due course? Indeed, the that Indian parents who migrated to the UK are disproportion-
weight of evidence remains in favour of the differences nar- ately drawn from a part of the population that is taller for genetic
rowing but not vanishing. Second, as previously stated, from a reasons. Those who chose to migrate may have on average en-
policy standpoint, what sense does it make to attribute dif- joyed some genetic advantage over the population left behind.
ferences in height and weight that can only be bridged over Tarozzi himself is aware of this possibility and is careful to
future generations to malnourishment? highlight it. Immediately following the conclusion quoted
above, he states,
9.3 Children of Indian Migrants in the UK Of course, these findings are not sufficient to disprove the claim that
At least some analysts who believe that height differences genetic factors play a role in explaining the relative disadvantage in
across populations of children can be eliminated by good nutri- growth pattern of children, such as those sampled within the NFHS,
who are born and raised in India. To argue that ethnic Indians who
tion have relied on a comparison of children born to Indian
live in the UK share the same genetic characteristics in terms of growth
(and Pakistani and Bangladeshi) parents settled in the UK with potential as their counterparts still living in India, one should argue
those born to white parents in the study by Tarozzi (2008). that migration to the UK is uncorrelated with growth potential. How-
This necessitates a close examination of his following finding. ever, there are reasons to suspect that correlation may exist, as mi-
grants are often taller (2008: 464).
Overall, these results [shown in his Table 6] provide some prima facie
evidence in support of the hypothesis that the growth performance of Second, even assuming that migrant parents are representa-
children of Indian ethnicity who live in the UK is comparable to that of tive of the Indian population, the possibility that the gap will
the reference population used to construct either the WHO-2006 or the persist in the case of children born and raised within India
CDC-2000 references (2008: 464; emphasis in the original).
cannot be ruled out. What if the UK geography, culture, and
A casual reader already unsympathetic to the possibility of environment are more conducive to height and weight deve-
genetic differences is likely to conclude from this statement lopment of children than the Indian geography, culture, and
that the key assumption underlying the WHO-sanctioned environment? Therefore, what is needed is evidence that some
methodology to measure malnutrition is valid. Yet, she will be sub-populations of children born and raised within India have
Economic & Political Weekly EPW may 4, 2013 vol xlviiI no 18 107You can also read
