Impact of Ahar Anudan Yojana on

Nutrition among Particularly Vulnerable Tribal Groups

in Madhya Pradesh

Dr B R Ambedkar School of Economics University, Bengaluru

Team:

Dr N R Bhanumurthy

Dr Rajesh B

Dr Sheetal Bharat

Dr Sumirtha Gandhi

For contact: [email protected]

Acknowledgements

The tribal population is among the most vulnerable in India, deprived of nutritious
food. Upliftment of these categories would change the face of the country. We gratefully
acknowledge the Atal Bihari Institute for Good Governance and Policy Analysis and Madhya
Pradesh Government for giving us the opportunity to study how the Ahar Anudan Yojana is
impacting the lives of people belonging to the marginalised sections of our society. For this,
we would like to specifically thank Dr Pallavi Govil, Principal Secretary, Tribal and SC Welfare
Department, Government of Madhya Pradesh, Mr. Sanjeev Singh, Commissioner, and Ms
Rita Singh, Deputy Commissioner for funding this project. We consider this as a great
opportunity to contribute meaningfully towards the welfare of the target populations and
also a contribution to academia. Ms Tanaya Mohanty has effectively coordinated between
our team and the Tribal Department, and we thank her for also making our trips to Bhopal
quite comfortable. During the preliminary visit, pilot survey and the main primary survey,
we met and benefited from conversations with several Tribal Department officers,
specifically, Mr Anand Rai Sinha, Mr Uday Singh Uikey and Mr S K Chaturvedi from Umaria
district. We are thankful for their generosity with their time and information.

We would like to thank our field investigators Ms Aditi Dutt, Mr Ankit Singh, Ms
Bhanvi Singh, Mr Gaurav Singh, Ms Ishita Verma, Ms Prajakta Gajbhiye, Ms Priyanka Uikey,
Ms Poonam Singh, Mr Ram Bahaddur Singh Parihar, Ms Rutvi Shukla, Mr Shikha Sharma,
and Ms Vaishnavi Raj, for their complete cooperation throughout the primary survey work
and the data cleaning phase afterward. We would also like to thank our own students, Mr
Sagar Khatri, Mr Aniket Sharma, Ms Putta Parimala, Mr Pranoi Rapheal Raju, Mr Arnab
Saha, Mr Aryan Kumar, Ms Vrushali, Mr Karanveer Singh Rathore, Mr Shreekar Durg, Mr
Anirudh Ravishankar, Ms Akshatha Nayak, and Ms Namratha Hubli, who joined us with
enthusiasm and helped us in cleaning up the data with utmost sincerity. Special thanks to
our student Ms. Samhitha for all the technical support she extended to us in data cleaning,
preparation of several illustrations, and computational help. We specifically thank her for
her speedy responses and the cleanliness of her work. We thank Mr Prakhar Rana for taking
care, in very good time, of the design and layout of this report.

Ms Anuapama S J Nair, the research assistant on this project, has been a pillar of
strength throughout. The team would not be so confident about the quality of the
background literature, questionnaire, and the primary data, had it not been for her
steadfastness and sincerity. I thank the faculty colleagues at our university who were
always willing to extend help on matters of coding or interpretation of results – specifically,
Dr Muhammad Rafi OPC, Dr Aritri Chakravarti, and Dr Bipin Sony. We also give a special
mention to Dr Michael Kuehlwein, a co-author in another project and genuine well-wisher
for helping with the interpretation of some results. Dr Bhabesh Hazarika and Mr Rohit Dutta
from the National Institute of Public Finance and Policy were most generous with their

2
guidance on the practicalities of organising primary field work and the technicalities of the
software Kobo. Senior paediatrician Dr Sarala Sundar, and Dr Vijaya Bharat, senior
cardiologist from Jamshedpur clarified some vital questions surrounding children’s nutrition
for us and we are thankful to her for sharing her time and knowledge. Dr Gauri Bharat and
Dr Neeraj Hatekar helped clarify some critical issues surrounding survey methodology. Last
but not the least we thank our university administrative staff who helped us in successful
completion of the project.

While all the persons listed have greatly helped improve the data collection,
cleaning, analysis, and interpretation of results we admit that any errors are our own.

3
Summary

The Tribal Welfare Department of the Government of Madhya Pradesh has been
electronically transferring ₹1,000 per month to the bank accounts of the female heads of
particularly vulnerable tribal households since 2018 under the Ahar Anudan Yojana or food
grant scheme. They commissioned an evaluation of the effect that this scheme has on the
health of the women and children, since these are the groups that are often most
vulnerable within a family.

There are three tribal groups that are included in this scheme – Saharia, mostly
found in the northern parts of the state, Baiga in the east and Bharia mostly in the south.
There are only 14 districts that make this scheme available to households belonging to these
tribes, though members of these tribes are spread wider. Given this layout of the scheme,
we chose to use a treatment-control group method of ascertaining the impact of the
scheme. For each tribe, we chose two districts adjacent to each other – one that offers the
scheme and the other that has the same tribespeople in it but does not offer the scheme.
Within these districts, we chose only the blocks that lie along the border between the two
districts. This proximity of the treatment and control group households helped ensure that
the other socio-economic and demographic characteristics would largely remain the same,
and therefore leave differences in their nutritional intake to be attributable to the scheme.

For reasons of data accuracy, we chose to follow the 24-hour dietary recall method
of ascertaining the nutritional status of the women (15 and over) and children (five and
younger). This involved a display of spoons, glasses and bowls of varying sizes so that the
respondents could indicate the volumes of the ingredients used in cooking all of the meals
prepared and consumed in the preceding 24 hours, the total volumes of items prepared,
and the volumes consumed by each member of the household. This data was used to
compute the level of various nutrients consumed by each woman and child. The list of
nutrients was taken from the National Institute of Nutrition website and included protein,
fat, polyunsaturated fatty acid, fibre, energy among the macronutrients. The vitamins
included were A (retinol and carotenoids), B1, B2, B3, B5, B6, B7, B9, C, D2, D3, E, K1 and K2.
The minerals included were calcium, chromium, copper, iron, magnesium, manganese,
molybdenum, phosphorus, potassium, selenium, sodium, and zinc.

We use a coarsened exact matching technique to ascertain the impact of the

scheme, controlling for the age of the household head, income, occupation, education, and
dependency ratio. All the reported coefficients carry a positive sign and are statistically
significant at the 5 per cent level. The results are presented for three nutrient groups –
macronutrients, vitamins, and minerals, all of which show a 15-18 per cent higher
consumption among the treatment group than in the control group.

4
 These results can be put into a more realistic
Treatment group
perspective by looking at the foods commonly
women consume, on
consumed by these families. The report proceeds to use
average, 15.5% more the results in physical units to give a picture of what the
vitamins than control difference in nutritional intake looks like when
group women. presented in terms of a few hundreds of grams of
various foods the households commonly consume.

The above results have been for women only. The similar analysis of nutritional
intake for children does not seem to yield such clear results. Possible reasons include
reliance on breastmilk, data for which is unavailable, and challenges in measuring the height
and weight of children during primary survey. Based on the positive results for the
interviewed women, we may extrapolate the benefits to the rest of the household.

Daily reports from the field made clear that the interviewees were desperately
lacking jobs, access to water, roads, sanitation facilities, education and health facilities, etc.
This led us to explore the relation between access to public goods and nutrition. We
created an index reflecting the level of public goods that each family had access to using five
variables: i. piped water connection, ii. electricity connection, iii. source of cooking fuel, iv.
time taken to reach the nearest government health facility, and v. average time taken to
reach the nearest ration shop and anganwadi as a general measure of road access. This
index is normalised to 1 and is, on average 0.44 in the three treatment districts pooled in
and 0.53 in the control districts pooled in.

A plot of the public goods index against the nutrition ratios for women for treatment
and control groups provides a new angle to the story on the factors determining nutrition
intake. The relation between the public goods index and nutrition ratios in the control
districts is rather ambiguous, whereas this relation in the treatment districts is distinctly
positive. Among the treatment districts, the households that have better access to public
goods have significantly better nutrition intake than households with poorer level access to
public goods. This points towards a strong complementarity between access to public
goods and receipt of the scheme money. A mean difference test confirms this theory.

The same relation is also explored using

regression analysis with coarsened exact matching
and control variables as earlier. The results match Treatment group women
the expectations – it is only when both public goods with high access to public
access and the scheme are available that the goods consume, on
treatment effect is statistically significantly different average, 30% more
from zero and high. The average treatment effect vitamins than control
for women with high level of access to public goods group women.
is 25 per cent for macronutrients, 30 per cent for

5
vitamins and 20 per cent for minerals. These effects are much higher than for the pooled
observations, indicating that public goods access is an important contributing factor. For
macronutrients and minerals, it appears that the complementarity of the scheme and public
goods access is strict, since treatment without access to public goods does not seem to
enhance their intake. In the case of vitamins, we find that treatment group women, even
those with low level access to public goods, are in fact consuming 15.7 per cent more than
the control group women. In this case, the scheme does in fact benefit the women even if
they have a low level of access to public goods.

Conceptually, this is easy to explain, though it is not something the literature on

nutrition has much considered. A claim can well be made that a family with better access to
public goods is likely overall to be more efficient in their routine endeavours, whether in
finding gainful employment, or accessing nutritious food for their families, or accessing
quality healthcare and education services. This claim comes directly from the role that
public goods are in general supposed to play in a market economy. In that sense, this claim
is not novel. But the literature on nutrition does not seem to have considered this
complementary relation between public goods access and nutrition intake.

This report closes with the suggestion

that the Ahar Anudan Yojana is indeed having The Ahar Anudan Yojana is
the intended effect on the nutritional intake of certainly enhancing the level
the beneficiaries. Even in the case of women the
of nutrients consumed by the
effect is differential and seems to depend closely
beneficiaries, but this impact
on the level of public goods that the household
can be increased substantially
has access to. If the scheme is to benefit all the
recipients they must have better access to all- by also providing access to
weather roads, potable water within their public goods like all-weather
homes, electricity connections and easier access road access, water and
to healthcare, educational services, market and electricity connections at
employment opportunities. Put another way, home, schools and hospitals,
several of the recipients of the monthly grant are anganwadis, etc.
unable to effectively use it to enhance their
nutritional intake, because they are simply too
removed from the mainstream of economic
activity, or because they are spending too much
of their energies on things that the additional income cannot help with – maybe fetching
water from a nearby river, or walking an hour to the ration shop or to the nearest
government hospital. Women engaged in microenterprises from their homes may also be
greatly benefited if they have better access to all-weather roads and other public goods.

Aside from the impact of the scheme on the health of the recipients, the
implementation of the scheme does seem to be satisfactory overall, based on the

6
responses. There were a few reports of agents demanding small fees for filling up the form,
and for helping with securing access to various documents for the application process, and
for using the ATM to withdraw the money received under the scheme, and so on, but these
instances seem to be few and far between. The beneficiaries do in fact receive the amount
every month, and any delays were brief due to the pandemic-induced budget shortages.

The general nutritional status of the PVTG tribes in Madhya Pradesh is poor, as is
evidenced from the much-below-one nutrition ratios (for almost all nutrients’ consumption
measured) computed for the control group women. If additional tribes or additional
districts are brought under the ambit of the scheme, the health of several thousands more
could well benefit, but before those benefits are to manifest in the intended way, the
shortage of public services must be addressed.

Contents

7
Chapter 1: Introduction..........................................................................................................11
1.1: Background............................................................................................................................11
1.2: Review of literature...............................................................................................................12
1.2.1: Malnutrition: definition, policy response and some trends..................................12
1.2.2: Tribal populations at risk of malnutrition..............................................................14
Chapter 2: Objectives and Conceptual Framework................................................................16
2.1: Objectives...............................................................................................................................16
2.2: Conceptual framework: Budget constraint approach.......................................................16
Chapter 3: Sampling and data................................................................................................19
3.1: Sampling procedure..............................................................................................................19
3.1.1: Stage 1 - Districts...................................................................................................19
3.1.2: Stage 2 - Blocks..................................................................................................... 21
3.1.3: Stage 3 - Gram Panchayats....................................................................................22
3.1.4: Stage 4 - Households.............................................................................................23
3.2: Data collection procedure....................................................................................................26
3.2.1: Competing methods..............................................................................................26
3.2.2: Training field investigators....................................................................................29
3.2.3: Computation of key variables...............................................................................32
Chapter 4: Summary statistics................................................................................................36
Chapter 5: Women’s nutrition................................................................................................41
5.1: Nutritional intake of sampled women................................................................................41
5:2: Nutrition ratio........................................................................................................................43
5.3: Empirical Strategy..................................................................................................................45
5.4: Results.....................................................................................................................................47
Chapter 6: Complementarity of the Ahar Anudan Yojana and public goods..........................50
6.1: Conceptual framework II: Public goods approach.............................................................50
6.2: Public goods in rural Madhya Pradesh...............................................................................52
6.3: Public goods and nutrition...................................................................................................58
Chapter 7: Implementation of the scheme............................................................................64
Chapter 8: Discussion and policy recommendations..............................................................67
References..............................................................................................................................70

8
Appendices.............................................................................................................................72
Appendix 1: The questionnaire...................................................................................................72
Appendix 2: Villages chosen based on Census 2011.................................................................81
Appendix 3: Histogram showing the nutrition ratio for each macronutrient, vitamin and
mineral for treatment (1) and control (0) districts...................................................................83
Appendix 4: Average treatment effect on nutrition intake, detailed results........................87
Appendix 5: Relation between nutrition ratio and public goods access................................88

List of tables
Table 1: Spread of PVTG beneficiaries across districts...........................................................19
Table 2: Purposively sampled taluks...................................................................................... 21
Table 3: Number of households interviewed from each zone and district.............................23
Table 4: Process of computing nutrient consumption from 24-dietary recall survey.............33
Table 5: Summary statistics....................................................................................................37
Table 6: Average consumption of nutrients of interviewed women against NIN
recommendations.................................................................................................................. 42
Table 7: Impact of inclusion in the Ahar Anudan Yojana on women’s NR..............................47
Table 8: Treatment effect in terms of commonly consumed food items...............................48
Table 9: Average public goods index by district.....................................................................52
Table 10: Distribution of observations by treatment and by access to public goods.............60
Table 11: Mean difference test for nutrition ratios across public goods access and treatment
............................................................................................................................................... 60
Table 12: Treatment effect by level of public goods access...................................................61
Table 13: Fund-release process..............................................................................................65

Table of Illustrations
Illustration 1: Microeconomic theoretical foundation of the scheme....................................17
Illustration 2: Geographical spread of the PVTGs Saharia, Baiga and Bharia beneficiaries....20
Illustration 3: Districts purposively selected as treatment and control groups......................22
Illustration 4: Geolocation of homes sampled in each zone...................................................24
Illustration 5: Handwashing practices....................................................................................38

9
Illustration 6: Ownership of durable assets............................................................................39
Illustration 7: Age distribution of all interviewed women......................................................41
Illustration 8: Control (0) and treatment (1) group nutrient ratio histograms ......................44
Illustration 9: Availability of public goods in urban and rural areas.......................................51
Illustration 10: Relation between the nutrition ratio and the public goods index..................58

List of photographs
Photograph 1: Survey equipment.......................................................................................... 28
Photograph 2: Primary survey training session in Tamia, Chhindwara..................................30
Photograph 3: Measuring the heights and weights of women and children..........................31
Photograph 4: 24-hour dietary recall interview with the use of vessels, glasses and spoons 32
Photograph 5: Poor access to clean water for drinking and cooking.....................................53
Photograph 6: No in-home electricity connection.................................................................54
Photograph 7: No access to clean fuel...................................................................................54
Photograph 8: Poor road-access............................................................................................ 55
Photograph 9: Anganwadi in Umaria district – outside and in...............................................56
Photograph 10: Not yet ODF..................................................................................................57
Photograph 11: Pradhan Mantri Aawas Yojana......................................................................57
Photograph 12: Registered beneficiaries have been receiving ₹1,000 per month regularly ..64
Photograph 13: A Bharia woman displaying her tattoo or Gudna..........................................66
Photograph 14: Home-based microenterprises.....................................................................68

10
Chapter 1: Introduction

1.1: Background

The diversity of India’s peoples makes for challenging policy formulation. Social,
cultural, economic, geographic, and historic experiences shape lives, and of these
multitudes of shapes, several require close policy attention. India’s tribal populations are
one such.

Madhya Pradesh houses the largest tribal population in absolute numbers and forms
the subject of this study. The Tribal Welfare Department of Madhya Pradesh has been
transferring since January 2018, under the banner of the Ahar Anudan Yojana, ₹1,000
electronically, directly and quite regularly to the bank accounts of women household heads
belonging to three particularly vulnerable tribal groups (PVTGs) across fourteen districts.
These PVTGs are Saharia, Baiga and Bharia and the districts across which they are spread
are Anuppur, Ashoknagar, Balaghat, Chhindwara, Datia, Dindori, Guna, Gwalior, Mandla,
Morena, Shahdol, Sheopur, Shivpuri, and Umaria. This policy, as the name suggests, is
expected to improve the health and nutritional intake of these tribal families, who, because,
in part, of their remoteness from the mainstream economy of Madhya Pradesh, are unable
take advantage of the benefits that specialisation and trade have brought to the rest of the
population. Having been in effect for nearly five years, this appears to be a good time to
evaluate the scheme against its stated aims.

The report proceeds with first presenting a survey of the literature on the issues
surrounding the health of women and children, and more specifically, among tribal
populations. A fairly simple conceptual framework is outlined to deal with the objective of
the project. A detailed description of the sampling and survey method follows, which
details out the variables computed. Some summary statistics are then presented, followed
by the empirical strategy and results on women’s nutrient intake. While these results are
strong, it is observed that the access to public goods varies widely within the sample, and
this provides an additional angle for enquiry. A second conceptual framework is presented
to address this theoretical angle and this is then followed by some more summary statistics
and results on the relation between public goods access and the treatment. The concluding
section presents some policy recommendations and directions for future research.

11
1.2: Review of literature

1.2.1: Malnutrition: definition, policy response and some trends

Malnutrition refers to deficiencies, excesses or imbalances in a person’s intake of

nutrients. In the case of these tribal families, the malnutrition is often glaring. Poverty is
often at the heart of malnutrition. 80 per cent of India’s population (about 250 million
people) was in poverty at the time of independence. The government has been trying
various corrective measures: reservation system for education and employment, cash
transfer schemes, employment schemes, and schemes providing nutrition to women and
children, among others. Other programmes such as the public distribution system through
fair price shops, the food for work programme, employment guarantee programmes etc.
have been initiated to make food affordable and accessible. While these measures have
gone a long way in ameliorating poverty, 20.8 per cent (346.3 million) of the population are
still living in poverty as of 2019 (GDC, UNICEF). India ranks 94 th among 107 countries in the
Global Hunger Index and 14 per cent of the population is estimated to be undernourished –
an unpardonably large share.

Women often do not have access to the basic rights of good health and well-being
even though they are responsible for the rest of the household. If the mother fails to secure
access to resources, the children suffer too. For this reason, women and children are
considered the subjects of studies in public health and nutrition, including the current study.

Cultural norms do tend to influence the nutritional status. The high fertility of
Indian women is one of the most detrimental influences on nutritional status because the
metabolic stresses of pregnancy and lactation may not be adequately compensated by
dietary intake before, during or even after these physiological processes. During pregnancy
women’s access to food may be even more restricted in traditional Indian household
through taboos and ritual observances (Chatterjee 1989).
Women in rural areas are less likely to receive antenatal care than women in urban
areas (Saha 2010), either because they may not seek care, or because trusted care is
unavailable or inaccessible (IIPS, 1998). Women in poor physical and mental health are
more likely to give birth to low-birth-weight infants. They are less likely to be able to
provide nutrition and healthcare for their children. High levels of infant mortality combined
with the strong son preference motivates women to bear high numbers of children to have
sons surviving into adulthood. Numerous pregnancies and closely spaced births erode a
woman’s nutritional status, which can negatively affect pregnancy outcomes and increase
the health risks for mothers (Jejeebhoy & Sathar 2001, ).
The literacy level of women can affect reproductive behaviour, use of
contraceptives, health and upbringing of children, hygiene practices, access to employment

12
and overall status of women in the society (Saha 2010). The negative effects of malnutrition
among women are compounded by heavy work demands, poverty, childbearing and rearing
and special nutritional needs of women, resulting in increased susceptibility to illness and
consequently higher morbidity (Chatterjee 1990, Sultana, Rahman, & Akter 2019, Shahid et
al 2022, Bangladesh Nutrition Profile, March 2014, Akter et al 2020, Mahmudiono et al
2018, Ghosh & Varerkar 2019).

Food security is a combination of access to food and its absorption by the body,
which depends on a number of non-food factors such as sanitation, access to clean drinking
water, access to health facilities, and so on. As families are already battling water crises,
energy shortage, price inflation and climate change, food security has become the most
intractable challenge in the development agenda (Ittyerah 2013). Food insecurity and
malnutrition rtesult in catastrophic amounts of human suffering. The World Health
Organization estimates that approximately 60 per cent of all childhood deaths in the
developing world are associated with chronic hunger and malnutrition. In developing
countries, the rural populace, particularly children, are vulnerable to malnutrition because
of low dietary intake, lack of appropriate care and inequitable distribution of food within the
household.

Children’s nutrient intake in the first two years of life is governed by appropriate
infant and young child feeding practices, comprising breastfeeding and complementary
feeding practices. In Madhya Pradesh, overall 90 percent of children continue to be
breastfed at one year and almost 73 percent at two years. The median duration of
breastfeeding is 33.2 months. Only about 38 per cent of children aged 6-23 months are fed
the recommended minimum number of food groups. Women’s poor nutrition is one of the
most important determinants of childhood stunting. The prevalence of thinness in women
has declined since NFHS-4 overall, but in Madhya Pradesh, 17 percent of women still are
undernourished according to NFHS-5.

Some focused policy interventions have caused a substantial decreasing trend in

stunting rate in some states, though these trends differ quite a lot. The share of children in
Madhya Pradesh who are underweight has improved since the NFHS-4 round, from 43
percent to 36 percent. The share of children who are wasted declined from 26 percent to
19 percent. That said, even during the first six months of life when almost all babies are
breast fed, 21 percent of children are still stunted, 27 percent are wasted, and 30 percent
are underweight according to NFHS-5. These findings are in fact corroborated in our field
work. The food and care requirements of a child vary with age. Children from 25 to 36
months of age were associated with higher odds of malnutrition. As the child’s age
increases, the probability of malnutrition also increased, as the reliance on breast milk must
go down.

Undernourished children get sick, are unable to complete school and fall further
into poverty in the aftermath of drought, disease or economic instability. Poor children are

13
also least likely to have access to safe water and adequate sanitation, to receive
preventative healthcare such as vaccinations, and when ill, are less likely to get adequate
medical care.
Poverty and malnutrition are about much more than just financial resources. For
many families, they are intertwined with social exclusion, discrimination and marginalization
driven by gender, disability, ethnicity, geographic remoteness and displacement. At the
individual level, such exclusion and inequity determine local access to goods and services,
including healthy foods. At the societal level, they can mean that the voices and needs of
poor and marginalized communities may not be heard in broader decision-making.
Disability can be both a cause and consequence of malnutrition. A lack of nutrients
can lead to blindness or neurological damage. At the same time, some physical disabilities –
such as intellectual and developmental disabilities or a cleft palate – can lead to reduced
nutrient intake. Stigma around disability can result in a new-born not being breastfed or
children being given less nutritious or smaller portions of food, or even not being fed at all.

1.2.2: Tribal populations at risk of malnutrition

Geographical isolation, primitive agricultural practices, socio-cultural taboos, lack of

formal education, poor infrastructure facilities, improper health seeking behaviour, poverty
etc. are known to lead to the development of various morbidities (Yadav et al 1999, Yadu et
al 2000, Khonngsdier et al 2002, Gogoi et al 2002, Bose & Chakraborty 2005).
Indigenous people and other disadvantaged ethnic groups face a greater risk of
malnutrition. Geographical remoteness affects the access a family has to essential foods,
and healthcare. Compared with their urban counterparts, children living in rural areas are
twice as likely to be stunted (UNICEF 2019). Deprivation, measured by the Household
Deprivation Index significantly impacts children’s nutrition (Srinivasan at al 2008). More
than half of socioeconomically deprived households had at least one child underweight or
stunted in their houses as compared to the non-poor counterparts (Panda et al 2020).
According to Census 2011, the total tribal population of India is 104 million, or 8.6
per cent of the total population. India is home to the largest tribal population in the world
and accounts for 705 ethnic / tribal groups under the Schedules Tribes group (Census,
2011). The largest concentration of tribal people is found in the seven states of north-east
India, and the so-called central tribal belt stretching from Rajasthan to West Bengal (IWGIA,
2020). The tribal people are the most socially and economically disadvantaged sections of
the society: the rate of literacy in India is 73 per cent, but the literacy among the scheduled
tribes is 59 per cent. These populations are often less aware about common measures to
improve health and nutrition, simple preventive measures for common diseases, types of
foods for general health etc.
Even though the Government of India has been trying to eradicate poverty, the
schemes are too often not reaching them effectively. A one-size-fits-all approach schemes
like the National Rural Livelihood Mission, The Mahatma Gandhi National Rural Employment

14
Guarantee Act, Pradhan Manthri Awas Yojana – Gramin, the public distribution system etc.
do not always reach the grass root level and may be less effective when it comes to tribal
people because of their low level of interaction with the money economy.
Tribal children in India are the most affected nutritionally among young population.
The consumption patterns of children actually reflected the eating patterns of their families.
Acute food insecurity in tribal households is due to the loss of their traditional dependence
on forest livelihood and the generally deepening agrarian crisis. Besides loss of livelihood,
systemic issues such as exclusions in public distribution system and weakening of public
nutrition programmes have aggravated the undernutrition problem (Ghosh et al 2019).
Tribal children lag far behind in school attainment above the primary level. Smaller
proportions of tribal children receive qualified medical treatment linked to physical
remoteness to health facilities but also due to deep-rooted cultural mistrust of conventional
medicine providers (Stiller et al 2020). This last observation made by Stiller was not
substantiated in the field work for the current project, where all the families did claim to
seek medical help from government hospitals and not locally available traditional healthcare
options.

This report proceeds to list the objectives of the study, then discuss the very basic
budget constraint-based conceptual framework for evaluating the effect of the AAY on the
health and nutrition of the women and children in the PVTG families in the sampled
districts. As the expected and unexpected aspects of the lives and livelihoods of the families
are revealed through the data, we proceed to explore other explanatory factors and
understand the overall effect of the scheme.

15
Chapter 2: Objectives and Conceptual Framework

2.1: Objectives

The Tribal Department of the Government of Madhya Pradesh, in the light of the
poor health records of the tribal populations, chose to resort to an unconditional cash
transfer scheme.

The Government of Madhya Pradesh has initiated several schemes to improve the
nutritional outcomes among women and children. The Tribal Affairs Department started
the Aahar Anudan Yojana, under which the head woman of each household belonging to
Baiga, Bharia and Saharia tribes receives ₹1,000. The 2018 state budget sanctioned s300
crores for the scheme, according to a Press Note by the Madhya Pradesh government (July
2020). Funds are being disbursed every month to eligible women. As per government
reports, in the year 2018-19, an amount over ₹17,469 lakh was distributed to 2,42,752 PVTG
beneficiaries and for the year 2019-20, ₹30,600 lakh was distributed to 2,24,058 PVTG
beneficiaries. During the month of July 2020 alone, almost ₹2,198 lakhs was distributed to
the identified families (based on internal notes provided by the Department). Even the
districts with the minimum number of eligible recipients have over 2,000 beneficiary
families. This suggests that the coverage of the scheme is quite vast across the tribal
districts of the state.

Based on the particulars of the case at hand, and based on the literature that exists
on this subject at this point, this study proceeds to address the following question:

 Has the Ahar Anudan Yojana, the unconditional money transfer scheme of the
Tribal Department of the Government of Madhya Pradesh succeeded in its aim of
improving the health and nutritional intake of the PVTG populations?
 Is the scheme is being implemented in the most efficient way possible, without
excessive leakages along the way and in such a way that the maximum benefits
are in fact being materialised among the intended beneficiaries?

2.2: Conceptual framework: Budget constraint approach

The simplest, almost simplistic conceptual framework for the way in which the
impact of the AAY can be understood is through the budget constraint approach. An
additional income of ₹1,000 per month is bound to push their budget constraint out and so
increase the quantities of goods and services they will be able to purchase. The additional

16
income may either be spent on nutritious foods, as is expected, or it may be spent on other
goods, maybe non-food items, as shown in Illustration 1.

Illustration 1: Microeconomic theoretical foundation of the scheme

Illustration 1 shows that with the AAY, the family’s budget constraint gets shifted out
and their budget set expands from the area under AB to that under CD. This affords the
family more of either food items, or non-food items, or some of both. Given as food is an
essential consumption item, and it has been observed that relatively poor families tend to
spend a greater share of their incomes on food and other necessary items, it is hoped that
the family will in fact increase their consumption of nutritious foods and the scheme will
have been vindicated.

As a parenthetical point, the reason in general unconditional cash transfer schemes

are more popular than conditional cash transfers or giving benefits in kind is that the entire
area under CD would no longer be available to the family if any conditions were imposed. If
there was somehow a condition imposed on the family that the money received should only
be spent on food items, then the new budget constraint would not be CD, but it would be
AED, thus cutting off access to the ACE part of the budget set. Under this scenario, the
family would have no option but to acquire AE amount of food. Since this is perceived as an
unnecessary infringement on the liberties of a family, and also because it might conceivably
be administratively burdensome to impose conditions, it is considered more prudent, more
efficient to make unconditional rather than conditional cash transfers.

Identification, Design, Implementation, and Impact are the four main pillars for
designing government programs in the form of transfers that will be effective to improve

17
the outcomes. The rationale of any government welfare programme crucially depends on
the main objective, target beneficiaries, and the context of the programme. Unconditional
cash transfers have in fact worked in areas such as increasing school enrolment (Ferreira et
al 2013), increasing household consumption, women empowerment (Johannes et al 2016),
reducing poverty, increase in the level of dietary diversity, likelihood of having been food
secure, decrease in level of illness (Pega et al 2022), decrease in criminality and drug use
amongst the youth (Marinescu 2018), increase in childhood nutrition (Aguero et al 2006)
even reduction in the levels of stress hormones lowering symptoms of depression among
the recipients (Angeles et al 2019). That said, they have shown little to no effect on
mortality risk, no increase in probability of working, and the effects on labour supply are
mixed. It also increases dependency on social welfare schemes and show lack of clear
positive effect in the long run.

In India, both the central and state governments have been running different forms
of transfers to improve the health and nutrition outcomes of pregnant and lactating
mothers and their children. Though the uptake of institutional deliveries in public
institutions has increased over time (Balasubramanian & Ravindran, 2012), evidence on
improvement in maternal and child health outcomes is uncertain (Narayanan & Saha, 2011,
2020, Gangopadhyay, Lensink, & Yadav, 2015).

The following section proceeds to explain the sampling strategy, and then on to the
process employed to ascertain the nutritional status of the women and children in these
PVTG families.

18
Chapter 3: Sampling and data

3.1: Sampling procedure

The project is ideal for employing a treatment-control group technique to measure

the impact of the scheme. We have chosen one treatment district for each of the three
tribes covered under the scheme. The control group has been chosen from among the
districts within Madhya Pradesh that border the chosen treatment districts. These districts
do not offer the scheme but have significant populations from these same PVTGs. The
control and treatment districts are adjacent each other to minimise any confounding
influences.

The sampling process is outlined in four stages as follows:

3.1.1: Stage 1 - Districts

The choice of districts is purposive. For each tribe that is covered under the scheme,
we have selected the district with the greatest share of the tribe in Madhya Pradesh. This is
Shahdol for Baiga tribe, Chhindwara for Bharia tribe, and Guna for Saharia tribe. It may be
noted from Table 1 that the districts with a greater share of Madhya Pradesh’s Saharia tribe
population than in Guna are Sheopur and Shivpuri, but we are unable to select these
districts because they do not share a border with any non-scheme districts within the state.
There would therefore be no suitable control group for those districts, as is clear from
Illustration 2.

Table 1: Spread of PVTG beneficiaries across districts

Districts ↓/ Tribes → Baiga Bharia Sahariya Total

Anuppur 6,064 2 6,066
Ashoknagar 1 2 16,369 16,372
Balaghat 5,722 5,722
Chhindwara 6 14,914 8 14,928
Datia 2,181 2,181
Dindori 10,134 13 10,147
Guna 1 1 18,105 18,107
Gwalior 3 9,196 9,199

19
 Mandla 11,509 71 1 11,581
Morena 2,209 2,209
Shahdol 24,796 2 24,798
Sheopur 1 9 39,227 39,237
Shivpuri 2 54,337 54,339
Umaria 22,998 22,998
Grand Total 81,234 15,017 1,41,633 2,37,884

Source: Tribal Department, Government of Madhya Pradesh

Illustration 2: Geographical spread of the PVTGs Saharia, Baiga and Bharia beneficiaries

Notes: Districts shaded in yellow house Baiga tribespeople, Purple is for Bharia, and Green,
for Saharia. Information on this spread is derived from the list of beneficiaries.

20
3.1.2: Stage 2 - Blocks

Our choice of blocks within the districts is also purposive and determined by the
location of the blocks along the border between the treatment and control districts. Two to
three blocks have been chosen from each of the treatment and control districts such that
the ones from the control district border the ones from the treatment district.

Table 2: Purposively sampled taluks

Treatment Control
Districts Blocks Districts Blocks
North zone – Guna: 12.8% of Raghogarh: Vidisha Lateri;
Saharia Saharias 13.6% Sironj
Aron: 7.2% of
Saharias
East zone – Shahdol: 30.5% Jaisinghnagar: Sidhi Rampur Naikin;
Baiga of Baigas 18.8% Majhauli; Kusmi
Beohari: 2.1% of
Baigas
South zone - Chhindwara: Tamia: 43.3% Narsimhapur Babai (Chichli);
Bharia 99.3% of Bharias Harrai: 32.5% of Kareli;
Bharias Narasimhapur

Notes: The percentages are computed from the list of AAY beneficiaries only. So this
information is unavailable for the control districts and blocks.

Source: The maps consulted for choosing the blocks are from the 2011 Census of India,
Administrative Atlas, Madhya Pradesh, and the various District Census Handbooks,
containing village directories.

21
Illustration 3: Districts purposively selected as treatment and control groups

Note: The areas marked in grey are treatment districts (T) for the three labelled tribes and
marked in green are the respective control districts (C) which border the chosen treatment
districts. The blocks listed in the Table 2 above are all along these borders.

3.1.3: Stage 3 - Gram Panchayats

From within all the blocks in a district taken together, we have selected the 20 Gram
Panchayats with the largest population of the specific PVTG that receives the benefits of this
scheme in that district (or, in the case of the control district, the same tribe that receives the
benefits in the neighbouring district). This list for the treatment group is available from the
list of beneficiaries provided by the Tribal Department, and for the control group it is
available from the 2011 census directory of town and villages.

22
 It may be noted that the population census includes all tribal and non-tribal
populations, but our survey concerns only specific tribespeople in each of the districts.
Villages in Madhya Pradesh are classified, unofficially, and customarily as Bharia villages,
Baiga villages, and Saharia villages. It was found, during the primary survey, that the
districts chosen from the northern part of the state had more Saharia people than any other
tribespeople, and similarly for the eastern (Baiga tribe) and southern (Bharia tribe) parts.
For this reason, using the population census figures for choosing the villages did not appear
to give a misleading choice of villages. Most of the tribespeople found in the villages in the
list were in fact from the tribe that was meant to be interviewed.

It may further be noted that in some cases the specific village listed was inaccessible,
not found or the specific tribespeople were not found. In such cases, the field investigators
went ahead and interviewed the appropriate tribespeople from the nearby village/s which
fall under the same Gram Panchayat.

3.1.4: Stage 4 - Households

From each village the field investigators were requested to interview 10 to 20

households from the specific PVTG of interest for that district, until we had 180 – 200
interviews per district pair, with the treatment - control sample ratio being approximately
3:2.

Parts of the borders between the treatment and control districts appear to be
devoid of interviews according to Illustration 4 because these are hilly terrains and therefore
devoid of habitation. In some cases, villages were extremely remote and inaccessible by
road.

Table 3: Number of households interviewed from each zone and district

Zone Control districts Treatment districts Total interviews

Nort Vidisha 8 Guna 11 190
h 0 0
East Sidhi 8 Shahdol 11 201
3 8
South Narsimhapur 6 Chhindwara 11 174
3 1
565

23
Illustration 4: Geolocation of homes sampled in each zone

[a] North zone

Control group interviews in

Vidisha blocks Lateri and Sironj

Treatment group interviews in

Guna blocks Aron and Raghogarh

[b] East zone

Control group interviews in Sidhi

blocks Rampur Naikin, Majhauli
and Kusmi

Treatment group interviews in

Shahdol blocks Beohari and
Jaisinghnagar

24
[c] South zone

Control group interviews in

Narsimhapur blocks Babai
(Chichli), Kareli, Narsimhapur

Treatment group interviews in

Chhindwara blocks Tamia and
Harrai

Notes:
- These images have been created using MS Excel – both the background map of the
district and the scatter plot of households interviewed. The locations of the
households are exact, according to the geolocation captured during the interview.
- The gaps along the border between the treatment and control districts in some cases
is due to there being no households to interview in hilly terrains. In some cases there
were a few villages that were extremely remote and inaccessible by road.

The strength of this sampling strategy lies in the following points:

i. Since the blocks in the treatment and control groups are adjacent, this strategy to
identify the impact of the scheme on the health and nutrition of the beneficiaries is
likely to be unsullied by factors such as differences in soil quality and climatic
conditions.
ii. Since the treatment and control groups are both members of the same tribe, we can
rest assured that the other socio-economic and demographic characteristics are
likely to remain largely the same.
iii. Since the control group is geographically separated, entirely, from the treatment
group, there are no spill-over effects of the scheme that need to concern us. This

25
 means that there is no chance that non-beneficiaries who live in proximity with the
beneficiaries may learn and adopt some healthy dietary habits.

3.2: Data collection procedure

3.2.1: Competing methods

The literature mostly recognises two major methods to ascertain the level of
nutrition of a person and change in it after an intervention: food frequency questionnaire
and 24-hour dietary recall.

There are several ready templates available for the food frequency questionnaire
that can be deployed on the field with some minor culture-specific adjustments to the list of
foods (Thompson & Subar 2013). They are structured to ascertain the general nutrients
consumed by a family. Often, they are used to check if a population is consuming a
particular set of nutrients and only include foods that contain those nutrients. The food
frequency questionnaires, broadly structured, are good at getting at the usual diet of the
family. They are also easier to administer and analyse than the 24-hour dietary recall data.
That said, the food frequency questionnaire requires a lot more planning and preparation to
validate the study for the specific population and their diet. It is also more suited to
ascertain the level of nutrition for a household rather than an individual (Shim et al 2014).
Due to the longer time periods involved, it is prone to measurement error, but because it
does not cover the entire year, it also has the disadvantage of including only seasonal foods
(Coates et al 2012).

The 24-hour dietary recall method has the advantage of using volumes of various
foods consumed and the volumes of the various ingredients used for preparation in the 24
hours preceding the interview (Gibson & Ferguson et al 2008). It is a more detailed process
that requires the use of measuring bowls, glasses, and spoons so that the respondents can
point out the volumes of various foods they have consumed and the volume of the
ingredients that have gone into the preparation of those foods. This same process
conducted a second time for at least a subset of the original sample helps further validate
the results achieved in the first round (Gibson et al 2017). Compared with the food
frequency questionnaire, the 24-hour dietary recall method conducted a second time has
been validated in the middle and lower-income countries as well. The level of detail that
this survey affords makes possible a wider range of questions to be studied. The time taken
to conduct this is its major drawback. Like the food frequency questionnaire, the 24-hour
dietary recall has the same disadvantage of not being representative of the foods that the
individual may consume in other parts of the year.

26
 The pilot survey, conducted in April 2022, included both elements – one block in the
questionnaire on frequency of consumption and expenditure on various food categories,
and another block on the 24-hour dietary recall. The experience of the pilot was interesting
in the aspects of tribal skills it revealed. When asked about the frequency of purchase of
staple items of consumption, their answers were casually and highly imprecise. For
instance, we asked the lady of the home how frequently they purchased potatoes
(something which most families seemed to be certainly consuming very frequently) – daily,
a couple of times a week, weekly, a couple of times a month, monthly, or even less
frequently. While the question did not really puzzle them, the answer that was
forthcoming, in all the surveys conducted in the pilot without exception, was such that we
were unable to enter an answer in any of those slots. Their answer in Hindi was the
equivalent of “if I happen to go to the market, and if I need potatoes, I buy them. This could
happen within a few days, or weeks or any time…”. This pattern was observed for every
item – even wheat, which is supposed to be purchased from the government-run ration
shops, from where they received a small, fixed amount of wheat, rice and sugar every
month. Since the amount of wheat received is small for the family, most families are in the
habit of purchasing a quintal (100Kg) or two of wheat from the market as and when needed.
They reported that wheat cost about ₹2,000-2,500 per quintal, but were unable, again, to
report the frequency of this large purchase. One would expect that an item of the family
shopping list that costs a significant share of their annual income would be planned and
purchased at a well-thought-out time of year or with a fixed, predictable frequency – but
this was not the case in a single family interviewed as part of the pilot.

We tried alternative ways of framing the question so that we may somehow get at
an answer in the interest of science. “When was the last time you bought potatoes?” or
“Did you buy potatoes when you went to the market yesterday/last week?” or several
variations of these. The answer was often “hōr kya”, which, loosely translated is “of
course!”, or “what else!”, and caused us to wonder if we were being offensive or silly in any
way. But the easy air in which it was said indicated that it was their way of saying simply
and unemphatically, “yes”. A couple of respondents thought we were daft for asking such
questions. Their position was that if food is required, and if there is money in their purse,
then they will purchase whatever is available at whatever price it is available. When food is
needed, it must be bought, and the question of frequency is meaningless. This worldview is
very different from that of a person who has been educated in the modern system and is
living in close association with the modern world that values measurability above almost all
else. Measurability makes a great cognitive demand, which these tribal families do not feel
obliged to commit to.

Also, the complete unwillingness to factor in current and future incomes in the
decision-making process about current expenditure also appeared to be near universal
among these families. Their incomes are so unpredictable and so uncertain that it would

27
never do to consider those sources in making their expenditure decisions, especially when it
comes to something as crucial as food.

These finding made it abundantly clear that the food frequency questionnaire in
tribal Madhya Pradesh would yield nothing for analysis. Fortunately, the 24-hour dietary
recall worked smoothly in the pilot. The ladies who prepared the foods were able to deftly
point at the bowl size indicating the volume of various ingredients they used, the total
volumes of items they prepared, and how much the family members each consumed. This
clarity, I do not believe is incongruent with their responses to the food frequency
questionnaire. They have spoons and bowls of different sizes in their own kitchens, and it is
a simple visual mapping that permitted them to point out the volumes of ingredients they
used in their meals’ preparation and the volumes they prepared for the household and
consumed personally. This is not cognitively burdensome. Surely there may have been
biases in their reporting of the various volumes, and each interview did take a good amount
of time, but this was the surest way to get at an accurate measure of the nutritional
consumption of these families. Given the financial, personnel and time constraints, we were
unable to carry out another 24-hour dietary recall for the same families a week or so apart,
as is often recommended. We hope this single survey of the 24-hour dietary recall is
adequate as has been deemed in the case of the NFHS as well.

Aside from the 24-dietary recall, which forms the most important part of the data
collected, we gathered only the most basic anthropometric data – height and weight. Due
to a shortage of resources and official access to skilled personnel, we were also unable to
collect blood samples of the respondents which would have given a wealth of information
on the state of their nutritional intake and absorption of those nutrients by their body. The
questionnaire also included questions on consumption on food and non-food items, basic
demographic information on the households, ownership of durable assets, their experiences
with the scheme (in the case of the treatment group) or their awareness about it (in the
case of the control group), among others. The entire questionnaire if reproduced in the
appendix.

Photograph 1: Survey equipment

Six sets of the following: 6 bowls, 3

glasses and 3 spoons, weighing
scale, plank of wood, spare
batteries, measuring tape, writing
pad, pen

28
3.2.2: Training field investigators

12 field investigators, who are all economics undergraduate or postgraduate

students or doctoral scholars in Madhya Pradesh were trained for two days in Tamia block
of Chhindwara district to conduct the survey. The training included:

 A broad layout of the project and its purpose, so that the field investigators
understand the importance of the work being planned and feel a valued part of the
team
 An explanation of the identification strategy – treatment and control groups had
different questionnaires prepared
 A run through of the various numerical codes used in each question in the survey
 Soft skills training on how to approach the families, what tone of voice to use to
encourage them to speak, how to be considerate of their lives and choices, how to
get their responses, rather than permit the interviewers to cast an unnecessary and
inappropriate shadow on the responders of their own guesses or opinions, etc.
 A distribution of survey equipment: writing pads, pens, weighing scales, extra
batteries, planks of wood on which to place the weighing scales because the ground
not being even may result in incorrect weight readings, measuring tapes, and
spoons, glasses and bowls of fixed sizes
 How to weigh children who are unable to stand yet – making the parent stand on the
scale with the child and then subtracting the weight of the parent
 How to measure the height of the women and children standing up straight: to
observe for the position of the feet, ensure that the Frankfurt horizontal plane is
attained, and the spine is erect
 How to measure the height of babies unable to stand yet: to ensure that their legs
are as straight as possible and the heigh is measured from the top of the head to the
heel of the foot
 How to administer the 24-hour dietary recall part of the questionnaire, and how to
record the responses such that decoding and analysis are easiest. Important here,
was to know in which cases to ask for volumes of raw ingredients, and in which cases
to ask for volumes of prepared meals. The interviewers were asked to proceed
reverse chronologically, from the last meal consumed, backward to the earliest meal
consumed within the last 24 hours. Meal options in the questionnaire were divided
into dinner, daytime meal and other meal, since it was observed in the pilot that the
families did not really have anything that can be called breakfast. The spoons and
glasses were in three sizes each and the bowls in six sizes. Various combinations of

29
 these volumes were adequate for the respondents to report all the ingredients and
foods they prepared and consumed.
 At the end of the training at the training venue, we travelled with the field
investigators to the first set of villages and conducted interviews in three homes with
them, for them to see directly how the interactions were proceeding.
 After this, the field investigators themselves took turns conducting the interviews in
the presence of the principle investigators for us to be sure they had grasped the
process.

Photograph 2: Primary survey training session in Tamia, Chhindwara

30
Photograph 3: Measuring the heights and weights of women and children

1. Since we expected the ground to be uneven, and the resultant weight readings to be
inaccurate, the field investigators were provided with planks of wood slightly larger
than the base of the weighing scales so that the scales could be placed on the plank
while in use. This way, the accuracy of the weight reading was ensured.
2. Some challenges were encountered in measure the length of babies unable to stand
yet.
3. Measuring the height of the adults posed fewer challenges.

31
Photograph 4: 24-hour dietary recall interview with the use of vessels, glasses and spoons

There were six bowls (volumes ranging from 200 to 3,000ml), 3 glasses (100, 250 and
300ml), and 3 spoons (5, 15 and 50ml) among which the respondents could choose the
volume of the ingredients they used for cooking and the volume of foods they prepared and
consumed.

3.2.3: Computation of key variables

The data collection along the lines discussed above were cleaned and the resultant
nutritional information was computed for each interviewee as shown in the detailed outline
below. Table 4 takes the example of one vegetable preparation and computes the level of
protein that one interviewee might have consumed as part of one meal.

32
Table 4: Process of computing nutrient consumption from 24-dietary recall survey

Steps followed Computational example

1 Items prepared in the Potato sabzi + roti for dinner

household for one meal

2 Ingredients used to Potato, onion, soyabean oil, salt, garlic, etc. in the sabzi
prepare one item

3 Volume of ingredients B3 of potato, etc. (B3, bearing volume 620ml, chosen

used among the bowls shown to the respondent)

4 Volume to weight 250ml of potato = 140.54g

conversion

5 Computation of various n
P
nutrients present in each
∑ 100ij ( v i j c i )
¿

i=1

ingredient, totalled across

all ingredients in the item where i identifies ingredients and runs from 1 to n, and j
identifies an item prepared. Pij is the protein content in
100g of potato, vij is the B3 volume, or 620ml, of potato
used, and ci is the volume (ml) to weight (g) conversion
factor for potato, in this case, (140.54g/250ml =
0.5622g/ml). The protein level is given in the National
Institute of Nutrition website in terms of per 100g of raw
edible portion. The computed protein values are summed
up for all ingredients i used to prepare item j.

6 Total volume prepared Say, B4 (1,020ml) of potato sabzi was prepared that meal

7 Volume consumed Say, the respondent consumed B1 (200ml) of the sabzi

8 Proportion of nutrients v j 1 n P ij
from the item prepared to
P j1 = ∑ (v c )
v jh i=1 100 i j i
¿

be assigned to a
respondent where vj1 is the volume of potato sabzi (j) consumed by
respondent 1 (200ml), and vjh is the total volume prepared
for the household (1,020ml), and Pj1 is respondent 1’s
protein intake from potato sabzi. The ratio of the total
protein present in the potato sabzi consumed by
respondent 1 is 200/1020 = 0.1961.

33
9 Total nutrient value to be
[ ]
m
v j 1 n P ij
P 1= ∑ ∑ (v c )
computed from all meals
¿

j=1 v jh i=1 100 i j i

and various snacks
consumed in 24 hours where j identifies the items prepared and runs from 1 to
m. j includes all the meals and snacks consumed in the 24
hours preceding the interview. P1 is the total protein
intake for respondent 1 in 24 hours.

10 Step 5-9 are repeated for The following nutrient values are computed for each
all the other respondent: macronutrients: protein, fibre, fat,
macronutrients, vitamins polyunsaturated fatty acids (in %); vitamins: A (retinol, and
and minerals to arrive at a total carotenoids), B1, B2, B3, B5, B6, B7, B9, C, D2, D3, E,
complete nutrient profile K1, K2; minerals: calcium, chromium, copper, iron,
for each respondent magnesium, manganese, molybdenum, phosphorous,
potassium, selenium, sodium, and zinc.

11 Nutrition ratio computed National Institute of Nutrition, India recommends a certain

for each respondent level of daily intake for most of the computed nutrients.
The Nutrition Ratio = level of nutrient consumed /
recommended intake. NR less than 1 indicates less than
required consumption of the nutrient, and NR more than 1
indicates adequate or more than adequate consumption
of that nutrient.

Note: The data was collected on the Kobo Toolbox mobile application and nutrient
computations were made using MS Excel and Python.

Some caveats on the above process must be mentioned proceeding with the
summary statistics and analysis:
 The nutrients have been computed for raw edible portion of each of the ingredients
used for cooking, because this is what the NIN data gives. When the raw foods are
cooked, surely, their nutrient content would change depending on how long it is
cooked for, how much of the nutrient is drained out with the water, or simply lost in
the heat, etc. We are unable to take into considerations these factors. We must
accept that the best approximation or actual nutrition consumed possible for us is to
compute the nutrition in the raw food. If the style of cooking is broadly uniform
across the households, then one may hope that there is no major error introduced
into the analysis.
 Another source of deviation is likely to be the estimation that the family provides of
the bowl sizes when talking about vegetables. The estimation is supposed to be
based on chopped vegetables, but the size of the pieces is likely to determine the

34
 volume that the vegetable occupies. Finely chopped potato pieces will occupy a
smaller volume than large chunks, for the same weight of potato. Since our
computations are based on volume, this is likely to cause some error.
 Since the 24-hour dietary recall questionnaire can be taxing in terms of time on both
the interviewer and the respondent, some amount of error could creep in due to
fatigue on both sides. The interviewers were required to ask the details of every
ingredient used in the preparation of every item in every meal in the last 24 hours,
the volume of each, the amount prepared in total, and the amount consumed by
each woman and child included in the study.
 To the extent that the pattern of information gathering may have differed from field
investigator to field investigator, there might be some variation in the data. But we
believe this is not something that may cause any serious error in our results because
it was the same team of field investigators who ran the interviews in each treatment
– control pair. Any minor errors in data collection are likely to have been present in
both treatment and control groups, therefore ensuring uniformity in the error.

Given the depth of information gathered, we do believe that the above-listed problems do
not post a great threat at all. The summary statistics show that the quality of the data is
well within an acceptable range.

35
Chapter 4: Summary statistics

Table 5 reports the summary statistics of certain variables across the treatment and
control districts. Except for the northern districts most of the households have electricity.
The major reason for the households of Vidisha
not having electricity connection is that several Almost all our respondents in
households seem to have the unfortunate all districts were daily wage
experience of having very few gadgets that use
workers, and gas cylinders are
electricity but having to pay very high monthly
not affordable to them even
bills, around ₹3,000 – 5,000 a month. And a
point worth noting is that they are not directly
with the subsidies. They
paying it to the government office, but the meter believed even though they
reader collects the amount from them were in receipt of the monthly
personally. This naturally plugs in the possibility scheme amount, having to
of come rent-seeking behaviour. spend almost all of it on gas
would leave nothing for food.
None of the households in Sidhi have an
LPG connection at home. Most respondents
were not literate enough to secure a well-paying
job in the nearby peri-urban centres. Regardless of skill, the employment opportunities
were perceived to be too few in their localities. These families felt that working on their
land was not adequate to satisfy their necessities.

There is a severe lack of sanitation

 Employment opportunities facilities. About 90 per cent of the sampled
are perceived to be too few households in Sidhi and Shahdol practice
open defecation. These numbers are slightly
 90% of the interviewed
better in the northern and southern zones.
households in the east zone
People are less aware about the importance
practice open defecation of hygiene. Most of the households clean
 Nearly all interviewees clean their house daily, but they do not wash their
their home daily but do not hands properly with soap and water. As
use soap to wash their hands shown in Illustrations 5a to 5c hand washing
 Poor access to markets has practices vary quite drastically among the
resulted in high reliance on regions. Overall, the practice of frequent
barter for food handwashing does not seem to be prevalent.
One reason worth pointing out is the severe
water shortage these districts face. In
Dhulaniya village it was observed that people

36
are bathing in river water that is visibly dirty. The highest percentage of people who are
accessing piped water for drinking is just 22.9 per cent, the lowest being 4.2 per cent.
Access to clean water could be a major reason for people being unable to care for hygiene.

84.7 per cent people in Chhindwara (highest) and 38.6 per cent in Sidhi (lowest)
own land but water shortage in the district does not encourage people there to engage in
farming. They are cultivating the minimum necessary for their household. Access to the
ration shop seems better in the south zone than in the other two, where most of the
respondents have to travel more than 15 minutes. Even though people buy ration have
some little access to markets in the nearby (relatively) villages, they still rely on barter for
several of their food needs. People in Vidisha district exchange food grains for vegetables.
In Guna district people do not spent much on pulses and cereals; they procure them from
their relatives through barter.

Most of the districts have anganwadis which are reasonably accessible – within 15
minutes, but too few respondents receive any services from these anganwadis. In the
ownership of durable assets the southern districts are better compared with the other two
zones. While the eastern and northern district show ownership of a few items only, the
southern districts show ownership of most of the items.

Table 5: Summary statistics

North zone East zone South zone

Vidisha Guna Sidhi Shahdol Narsimhapur Chhindwara
(C) (T) (C) (T) (C) (T)
Hh interviewed 80 110 83 118 63 111
Electricity 67.5 41.3 94 95.8 88.9 97.3
Piped drinking 22.5 22.9 9.6 4.2 17.5 5.4
water
LPG for cooking 38.8 37.6 0 2.5 20.6 35.1
Open defecation 76.3 58.7 90.4 89 28.6 18.9
Female HH head 6.3 17.4 8.4 10.2 7.9 16.2
Monthly income 42.5 15.6 55.4 -- 90.5 45.9
<₹5,000
Daily wage work 78.8 66.1 91.6 73.7 63.1 51.4
Own land 48.8 45.9 38.6 63.6 54 84.7
Own cattle 33.8 31.2 60.2 65.3 44.4 76.6
Yellow ration card 35.0 86.2 68.7 65.3 22.2 89.2
Ration shop in < 15 40 31.2 41 37.3 79.4 82.9
mins away
Goods, services 17.5 22 18.1 21.2 31.7 57.7

37
from Anganwadi
health facility < 15 23.8 22 39.8 21.2 38.1 48.6
mins away
Use soap 51.3 43.1 20.5 31.4 31.7 51.4
Clean home daily 87.5 83.9 84.3 67 50.8 77.5
Drink boiled water 0 3.7 0 2.5 0 13.5

Note: figures given are in percentages of total interviews from the district.

Illustration 5: Handwashing practices

[a] North zone

[b] East zone

[c] South zone

38
Illustration 6: Ownership of durable assets

[a] North zone

[b] East zone

39
[c] South zone

40
Chapter 5: Women’s nutrition

This chapter discusses the nutritional levels of the women interviewed as part of the
survey. A total of 459 women were interviewed from the three treatment districts together
and 285 from the three control districts. These women were at least 15 years of age and
the oldest woman interviewed was 90 years old. The nutritional level is computed based on
the 24-hour dietary recall method described in Table 4.

We use the regression method together with the coarsened exact matching
technique to isolate the treatment effect of the transfer. These results show that the
treatment group women do in fact consume more of the various nutrients than the control
group women when controlling for various potentially confounding variables.

Illustration 7: Age distribution of all interviewed women

5.1: Nutritional intake of sampled women

Table 6 shows the average nutritional intake of the sampled women in each
zone, together with the recommendations for each nutrient prescribed by the National
Institute of Nutrition, India. The purpose of this table is to show the general level of
nutritional intake of the sampled population in relation with what is recommended. In most
cases the consumption level is far below the recommended level. The only exceptions are
Vitamin B6, manganese, molybdenum and selenium.

41
Table 6: Average consumption of nutrients of interviewed women against NIN recommendations

Nutrient NIN recommendation North East South

Macronutrients
Protein 46g 34.29 27.69 37.17
(75%) (89%) (73%)
Fat (in g) -- 5.82 3.44 6.24
PUFA* (in mg) -- 2251.55 1171.78 2154.49
Fibre 30g 32.06 14.52 26.81
(47%) (96%) (65%)
Energy (in KJ) -- 3585.79 1135.89 2711.89
Nutrient NIN recommendation North East South
Vitamins
A (Retinol) (in µg) -- 0.14 0.07 0.19
Carotenoids (in µg) -- 3505.13 1798.00 2409.17
B1 1.7mg 1.08 1.05 1.35
(82%) (87%) (72%)
B2 2.4mg 0.43 0.31 0.44
(100%) (100%) (100%)
B3 14mg 6.53 3.34 5.91
(98%) (98%) (98%)
B5 5mg 2.83 2.09 2.79
(91%) (98%) (92%)
B6 1.9mg 7.52 2.05 2.20
(96%) (82%) (79%)
B7 40µg 4.53 2.17 3.48
(100%) (100%) (100%)
B9 220µg 198.57 107.84 136.62
(83%) (95%) (87%)
C 65mg 38.15 19.33 18.13
(86%) (93%) (94%)
D2 (in µ) -- 33.78 6.39 26.29
D3 (in µ) -- 0.03 0.01 0.02
E 7.5mg 1.16 0.49 0.72
(99%) (98%) (100%)
K1 55µg 44.35 44.17 35.78
(79%) (86%) (81%)
K2 -- 0.58 0.27 0.32
Nutrient NIN recommendation North East South
Minerals
Calcium 1,000mg 152.83 204.04 211.76

42
 (100%) (99%) (100%)
Chromium 50mg 0.03 0.01 0.02
(100%) (100%) (100%)
Copper 1.7mg 1.53 0.73 1.23
(60%) (94%) (78%)
Iron 29mg 11.49 4.65 9.18
(100%) (98%) (99%)
Magnesium 370mg 347.57 154.91 293.15
(57%) (97%) (75%)
Manganese 4mg 7.16 6.29 8.31
(25%) (29%) (17%)
Molybdenum 0.05mg 0.10 0.07 0.09
(26%) (43%) (30%)
Phosphorus 1,000mg 856.84 768.74 1029.03
(70%) (77%) (51%)
Potassium 3,500mg 1552.05 1225.1 1566.31
3
(97%) (98%) (95%)
Selenium 40µg 124.65 89.41 131.28
(17%) (8%) (10%)
Sodium 2,000mg 1057.44 115.19 568.23
(85%) (98%) (92%)
Zinc 13.2mg 7.32 4.01 6.87
(92%) (99%) (95%)

Notes:

 * The PUFA count shown excludes that derived from oils.

 Within brackets are percentages of women whose consumption of each nutrient
is less than the NIN recommended level.

5:2: Nutrition ratio

To explore this relation, presented below are the nutrition ratios (NR) for each
nutrient. This ratio is computed as the level of the nutrient consumed by a women divided
by the recommended level. An NR of 1 or over would indicate that the woman is consuming
the nutrient in adequate or greater amounts, and if it is less than 1, then she is not.

Protein consumed i
Protein NR i=
Protein consumption recommended by NIN
43
 For each nutrient, this measure standardises the consumption by setting a value of 1
as the frame of reference, regardless of the absolute value of NIN recommendation and the
actual level of consumption by a respondent. This transformation makes interpretation of
the regression results easier.

It may be noted that the medical literature dealing with nutrition considers this same
ratio as the nutrition adequacy ratio or NAR with the additional condition that the value is
capped at 1. This condition lends to the ratio the sense of nutrition adequacy. Our current
purpose does not require for this ratio to be capped at one, and so we retain the formula
without this condition and call it the nutrition ratio.

Illustration 8 below shows the histogram of NR for some representative nutrients,

for the treatment and the control groups. The histogram for each nutrient is presented in
Appendix 3. One of the observations that can directly be made from most of the nutrients’
histograms is that the treatment group histogram (coded in as 1) has several more right-tail
values than the control group. In addition, in the case of some of the nutrients, like protein,
vitamin B1, B7, C, E, chromium, magnesium, manganese, molybdenum, phosphorus,
potassium, selenium it is also clear that the overall treatment group distribution is shifted
slightly to the right compared with the control group distribution.

Illustration 8: Control (0) and treatment (1) group nutrient ratio histograms

0 1 0 1
20

15

82% 76% 82% 78%

15

10
Percent

Percent
10

5
5
0

0 1 2 3 4 0 1 2 3 4 0 1 2 3 0 1 2 3
Protein Vitamin B1
Graphs by Treatment Graphs by Treatment

0 1 0 1

64% 59% 99%

15

15

95%
10
10

Percent
Percent

5
5

0
0

0 1 2 3 4 0 1 2 3 4 0 .5 1 1.5 0 .5 1 1.5
Phosphorus Potassium
Graphs by Treatment Graphs by Treatment

44
Notes:

 The red vertical line through the histograms marks the point where the nutrient ratio
is 1, or where the woman is consuming the NIN-recommended amount of that
nutrient. Women to the left (right) of the vertical line are consuming less (more) than
recommended. The percentage of women consuming less than the required amount
is shown on each histogram.
 The horizontal and vertical scales on the histograms have been matched, thereby
facilitating a direct comparison of the control and treatment group distributions.

5.3: Empirical Strategy

To test this hypothesis in a more organised way, we use a matching technique to

ascertain the impact of the treatment on the NR of each of the nutrients. We ascertained
the causal effects of the scheme on maternal and child health by employing a binary logistic
regression on the matched dataset. The matching procedure was undertaken to avoid the
potential problem of differential characteristics between treatment and control groups,
which is visible in the summary statistics. According to Imbens (2015), since randomisation
is difficult, the researchers are discouraged from carrying out studies assessing the causal
impact of a particular programme and any estimation process carried out without
appropriate checks are thought to produce imprecise estimations. In this study, we
employed a coarsened exact matching (CEM) technique to assess the scheme’s impact. This
method coarsens each variable by recoding it to assign the same numerical values to the
groups that are substantially indistinguishable. Next, we applied an exact matching
algorithm to match the dataset and to prune the unmatched units and retain the values of
the matched data. Finally, we applied a binary regression model to obtain the average
treatment effect. We explain CEM and the logistic regression method in the following few
paragraphs.

In the broad category of the monotonic imbalance bounding (MIB) class of methods,
the process of coarsening asserts the maximum degree of imbalance between treated and
control groups. Also, the coarsening choice for any one variable does not affect the
imbalance bound for any other variable.

According to the congruence principle, there must be congruence between the data
space and analysis space. When this principle is violated, the inferences are less robust with
sub-optimal and counter-intuitive properties (Mielke & Berry 2007). Unlike other matching
procedures, in which a separate step is required prior to matching, in CEM, no separate step
is required. In fact, all observations requiring extrapolations are automatically removed.
This makes the process relatively easier.

45
 Compared to other matching methods like Mahalanobis matching and Genetic
matching, CEM preserves a greater percentage of units in both treated and control groups.
Hence, to some extent, the process of coarsening overcomes measurement issues, for the
pre-processing matching stage.

The process of coarsening enables one to control the bounds on the extent of model
dependence and thereby reduces model dependence. Reduction of model dependence is a
key property of any matching procedure. In other words, the matching procedure should
establish a situation where the influence on the estimate of the causal quantity should be
less if it was not matched (Ho et al 2007). This holds true for CEM, which belongs to the MIB
method class.

One of the interesting properties of the MIB class is that their tuning parameters
directly reduce the maximum possible causal estimation error. This is does not hold true for
EPBR class.

Since real datasets are highly correlated compared to the independent draws in
hypothetical cases, CEM tends to generate a reasonable matching. This was shown by Lacus
et al (2012) that if control units are larger in number, then the number of unmatched
treated units become 0. In other words, CEM normally generates a well-balanced dataset
with a reasonable number of observations and so the data is useful for making causal
inferences.

CEM is computationally efficient compared to other matching methods. Each

observation i with a vector of covariates Xi is stored as a record with only coarsened values.
Hence, the number of covariates does not affect the dimensionality of the coarsened
dataset. Finally, CEM procedure is empirically efficient, unlike the other matching
techniques, such as propensity score matching (King et al., 2011) which approximates
random matching and increases the imbalance.

The model run is as follows:

N Ri=β 0 + β 1∗Treatment dumm y i + β 2∗Control ¿ variablesi + ε i

where NR for a particular nutrient is defined for each woman i as above, the
Treatment_dummy variable takes a value of 1 for women in the treatment districts and 0 is
in the control districts, the Control_variables vector includes household income and size.
Including other control variables do not drastically change the magnitude or direction of the
results.

We complemented the regression analysis by employing coarsened exact matching.

The semi-parametric nature of this method avoids the misspecification of conditional
average of the outcome variable, and the arbitrary specification of the impact of
heterogeneity (Imbens 2004). We first found a perfect statistical twin for as many members

46
in the treated group as possible and then we ran a regression on it. It had been ascertained
that the selected matching covariates should tend to affect both the uptake of AAY and the
outcome indicator. The matching results were subject to change according to the inclusion
or exclusion of matching variables. Our matching variables were selected from the previous
literature (Lim et al 2010, Mazumdar et al 2012, Mohanan et al 2013). These variables
were: wealth group, districts, and education and age of the woman.

5.4: Results

Table 7 presents the brief results (detailed results are presented in Appendix 4).

The first thing to note is that all the reported coefficients carry a positive sign and are
statistically significantly different from zero at the 5 per cent level, as reported. This
indicates that compared with the control group average NR, the treatment group average
NR tends to be greater, controlling for the age of the household head, income, occupation,
education, and dependency ratio. Table 7 also reports the control group average NR which
serves as the comparison for the coefficient. This has been used to compute the average
increase in intake of the nutrient group in percentage terms (Appendix 4 gives the detailed
results which show the average treatment effect in physical units for each nutrient). These
are strong results – macronutrients on average are consumed 17.8 per cent more in the
treatment than in the control group. This figure is over 15.5 per cent for the vitamins and
minerals.

Table 7: Impact of inclusion in the Ahar Anudan Yojana on women’s NR

Nutrient group Coefficient Control group Average

(se) average NR Treatment Effect
Macronutrients 0.14** 0.78 17.8%
(0.06)
Vitamins 0.05** 0.35 15.5%
(0.03)
Minerals 0.14** 0.88 15.5%
(0.05)

Notes:

 ** indicates 5% level of statistical significance level.

 The macronutrients group includes only protein and fibre, because these are the only
two for which a recommendation exists in physical units. The vitamins group includes

47
 vitamins B1, B2, B3, B5, B7, and C. The minerals group includes chromium, copper,
iron, magnesium, manganese, phosphorus, selenium, sodium, and zinc.

It can further be deduced, based on the general consumption pattern of the

respondents, and looking at each nutrient separately, that these reported increases in
nutrients may be attributed, notionally, to the following extra portions of foods consumed
per woman per day. This list is provided for each nutrient, independently of the other
nutrients that those same foods may additionally provide.

Table 8: Treatment effect in terms of commonly consumed food items

Nutrient Foods that contain the amount of the nutrient identified as the average
treatment effect
Protein 200 g cauliflower, purslane leaves, okra, French beans, drumstick, or spinach
Fibre 100g coriander, amaranth leaves, 200g of spinach, raw papaya, or pumpkin
Vitamin B1 100g fenugreek leaves, yard long beans, chicken, 200g bathua, drumstick
leaves, bitter gourd, jackfruit, brinjal or potato
Vitamin B2 100g coriander leaves, cabbage, jackfruit, chicken, 200g raw mango, peas,
tomato, or banana
Vitamin B3 100g amaranth, coriander, fenugreek leaves, 200g spinach, cauliflower,
papaya, onion, rice, or coconut
Vitamin B5 100g bathua leaves, papaya, one roti, 200g wood apple, tomato, or spinach
Vitamin B7 100ml milk tea, 200g maize dalia, or matar dal
Vitamin C 100g pumpkin, 200g bottle gourd, or banana
Chromium 100g soya beans, onion, coconut, one roti, 200g moong dal, peas, or potato
Copper 100g bread, spinach, 200g raw mango, or mutton
Iron 100g peas, bread, 200g drumstick, okra, or chicken
Magnesium 100g French beans, peas, 200g brinjal, chicken, or mutton
Manganese 100g maize dalia, vermicelli, coconut, 200g jackfruit, or banana
Phosphoru 100g vermicelli, drumstick leaves, 200g fenugreek leaves, or drumstick
s
Potassium 100g cabbage, tomato, fenugreek leaves, or 200ml milk
Selenium 100g soya beans, or 200g of maize dalia
Sodium 25g instant noodles, 300g fried, salted, packaged snack, or less than 1g salt
Zinc 100g bathua leaves, 100ml curd, 200g spinach, okra, or French beans

These amounts of food are of course chosen simply from the list of items most often
consumed by the interviewed families, and so additional portions of these can be easily

48
justified as ‘treatment effect’ for each woman each day. The 100g or 200g of each item
listed are also approximately what one adult may consume as part of one meal, and this
serves as an additional justification of these foods as treatment effect per day.
The only additional point to be made about the above table is the row listing sodium.
The treatment effect is given as 25g of instant noodles, like Maggi, or 300g of a fried salted
snack such as Haldiram’s namkeen mixture, or an extra gram of salt. These items do seem
incongruous with the rest of the table, and so require explanation. There were only three
families in the entire set of interviews that listed these items of food as part of what they
consumed during the previous 24 hours. In addition to these, white bakery bread was also
listed by two families. These families all happened to be within the treatment group. It is
easy to see that these items are all much higher in their sodium content that the other
grains, pulses and vegetables that are consumed more routinely. It must also be noted that
the high level of sodium that comes with these foods brings the NR to much greater than 1 –
not particularly a sign of health. In the sodium histogram in Appendix 3, there is one woman
whose sodium NR is at around 17 – her sodium consumption appears to be 17 times the
recommendation. This is the only instance where a woman in the treatment group has
overdone her consumption of a particular nutrient, and so this need not reflect, in general,
negatively on the scheme.
This observation about sodium must also be used to highlight the fact that the
processed foods trend that has swept the urban world much earlier has also touched large
parts of rural India. The smallest and most meagre establishment, which may be no more
than a table and a few shelves with a tarpaulin cover propped up by bamboos, that may
barely even be called a shop, has supplies of packaged biscuits, chips and instant foods,
though not in great variety. The fact of the presence of such shops indicates that they are
adequately profitable and that their customer base is large enough for their continued
existence. During the pilot survey in Guna district, one family did mention that when their
three-year old son somehow got hold of a ten-rupee note, he smartly took himself to
neighbourhood shop and helped himself to a small glass bottle of a carbonated beverage.
One could surmise that consumption of such snacks that may not be considered particularly
healthy may only be at a level adequate for the survival of those shops, but not at a level
that causes a major concern for the general health of the population. It did, after all, not
show up in the 24-dietary recall interviews in a very significant way, except in the instances
mentioned.
As a parenthetical point, the very presence of these shops is a sign of the reach of
commercial interests in the rural recesses of the country. These are parts where even the
road network has not reached, public transport is absent, and where mobile networks, in
India, which is seeing stunning growth in mobile phone usage, are non-existent, and from
where ration shops, anganawadis and government hospitals are often half an hour or more
away. Some families did mention that they had to travel an hour or so on foot towards the
lower elevations and towards civilization to get reception on their mobile phones.
This observation leads us to the next angle that we explore in this paper – access to
public goods.

49
Chapter 6: Complementarity of the Ahar Anudan Yojana and public
goods

6.1: Conceptual framework II: Public goods approach

In our search for a suitable explanation for the results we had so far, the field reports
prepared by the field investigators and our own observations during the pilot and main
primary survey came into assistance. Looking back at the conceptual framework, one must
remember that this scheme is transferring money to the bank accounts of women who live
in remote areas with relatively sparce access to goods and services that may be bought with
money. That budget constraint assumes that the food and non-food items listed on the
axes are all accessible to these families, and that the money that is pushing their budget
constraint out is good for making all those additional purchases. In rural Madhya Pradesh
these assumptions do not hold by far. The daily reports submitted by our field investigators
mentioned frequently that the families they interviewed complained about not having
access to water, and being billed much more for electricity than was due, given their usage,
about not having access to employment opportunities locally, about having to go too far to
a hospital and often losing unwell members of their family, especially children, enroute,
about the bank security guard demanding a small bribe every time they went to withdraw
their scheme money, etc. Given these experiences it is imperative to address the issue of
public goods access as an integral part of the overall analysis of how the scheme is
operating. There are often hinderances to the effective use the AAY money.

Urban India tends to have a much richer menu of public goods available – most
homes are likely to have an electricity connection (even if illegal), piped water, accessible
healthcare services, pakka roads, and a gas connection. A population of the scheme
beneficiaries is likely to be able to take advantage of the money it receives with the
complementary assistance of all the public goods. Rural India, on the other hand is likely to
have a poorer spread of public goods, and as shown in Illustration 9b, it is very likely that a
home that has electricity may not have water, or a home that has piped water may not have
proper cooking fuel. A whole village may be so far away from the nearest market or ration
shop that it is largely cut off from the nearest economic centre and so also markets and
other facilities. As one or several of the essential public goods are missing, the family may
face hinderances in taking full advantage of the money it has on hand.

50
Illustration 9: Availability of public goods in urban and rural areas

[a] Urban area

Electricity connection

Piped water connection

Accessible healthcare services
Accessible anganwadi, ration shop
Pakka roads

[b] Rural area

Piped water connection

Electricity connection
Accessible healthcare services

Accessible anganwadi, ratio shop

Pakka roads

51
 It may also be argued that a greater connection with government personnel, either
through the anganwadi or the ration shop or for any other reason, could cause to filter into
the consciousness of the rural women the importance of certain foods for growing children,
or for pregnant or lactating women, even though sharing of this information is not, by
design, a part of the scheme. Awareness is a key element in the possibility for the scheme
money to enhance nutrition for women and children.

6.2: Public goods in rural Madhya Pradesh

If the above hypothesis were true, then the task at hand is to look for the effect that
the treatment has in the presence of various public goods. Towards this test, we first
construct a public goods index using five variables:

i. piped water connection,

ii. electricity connection,
iii. source of cooking fuel,
iv. time taken to reach the nearest government health facility, and
v. average time taken to reach the nearest ration shop and anganwadi as a
measure of road access.

This index is normalised to 1 and is, on average, 0.40 in the three treatment districts pooled
in and 0.58 in the control districts pooled in. The average public goods index in each district
is given in Table 9. The south zone has a better index than the other two.

Table 9: Average public goods index by district

Districts Public goods index

North zone Vidisha Control 0.57
Guna Treatment 0.34
East zone Sidhi Control 0.57
Shahdol Treatment 0.37
South zone Narsimhapu Control 0.61
r
Chhindwara Treatment 0.49

52
Photograph 5: Poor access to clean water for drinking and cooking

The clear drinking water carried

by the field investigators in
comparison with the murky
water used by the respondents
for drinking and cooking

53
Photograph 6: No in-home electricity connection

Photograph 7: No access to clean fuel

54
Photograph 8: Poor road-access

Top: Several villages in Guna, Vidisha and Narsimhapur were hard to reach. Middle: The
field investigators’ vehicle was stranded because it tried to pass a too-narrow road. Bottom:
The market at Bhaiskhoh village was the only market for several villages in the district.

55
Photograph 9: Anganwadi in Umaria district – outside and in

This anganwadi the team saw during the preliminary visit to Umaria organised by the local
officials of the Tribal Department. The lady managing the anganwadi got the children to
sing a song. During the primary survey, no anganwadis were seen in the villages selected for
the interviews.

56
Photograph 10: Not yet ODF

While the intention to make Madhya Pradesh open defecation-free is certainly there, the
reality is quite far. Public toilets constructed a few years earlier lie unused.

Photograph 11: Pradhan Mantri Aawas Yojana

As a result of this
scheme, most
families own their
homes and several
of them have pakka
homes constructed.

57
6.3: Public goods and nutrition

Illustration 10 shows the relation between the public goods index (on the horizontal
axis) and the nutrition ratio for each nutrient (on the vertical axis), separately for the control
and treatment groups for a small number of nutrients. The difference between the slopes
of the lines of best fit going through the control group scatter and the treatment group
scatter are obvious in almost all the nutrient plots (shown in the appendix). The lines of
best fit all have a positive slope for the treatment scatter, but this relation is rather
ambiguous for the control group. This indicates that the treatment is helping the women
enhance their level of consumption better when there are public goods available, rather
than when such public goods are scarce.

Illustration 10: Relation between the nutrition ratio and the public goods index

0 1 0 1
1

1.5

0 1
0 1
3
4

Vitamin B1Potassium
1
Calcium
3 .5

2
.5
Protein
2

0 1
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1
1

PGI PGI

Fitted values calcium Fitted values potassium

0

Graphs
0 by Treatment
0

Graphs by Treatment .5 1 0 .5 1
0 .5 1 0 .5 1 PGI
PGI
Fitted values b1
Fitted values protiens Graphs by Treatment

Notes: Graphs by Treatment

0 1
0 1
3
.6

2
Vitamin B5
Vitamin B2
.4

1
.2

0
0

0 .5 1 0 .5 1
0 .5 1 0 .5 1 PGI
PGI
Fitted values b5
Fitted values b2
Graphs by Treatment
Graphs by Treatment
0 1 0 1
.003

1.5
.002

1
Chromium

Iron
.001

.5
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1
PGI PGI
Fitted values chromium Fitted values iron
Graphs by Treatment Graphs by Treatment

58
  The x-axis measures the public goods index, constructed as a simple average of five
variables which may be considered as a reflection of access to public goods. These
are: i. electricity connection at home; ii. piped water connection at home; iii. use of
LPG as cooking fuel; iv. average of how much time it takes to reach the nearest
anganwadi and ration shop; and v. the time it takes to reach the nearest government
hospital. The values are normalised to 1, such that if a woman has the best access to
each of these, her score would be 1, and if a woman has access to none of these, and
takes the longest time to reach the hospital and anganwadi and ration shop, her
score would be 0.
 The y-axis measures the nutrition ratio for the listed nutrient. NR of 1 indicates that
the woman is consuming just as much as is recommended. This ratio is less than 1
for most nutrients and for most women in the sample.
 It is evident that the line of best fit drawn through each scatter has a positive slope in
the case of almost all the treatment group scatter plots, but the relation is more
ambiguous in the case of the control group scatter plots. This indicates that the
treatment is helping the women enhance their level of consumption better when
there are public goods available, rather than when such public goods are scarce.

Based on the conceptual framework presented above, the relation evident here is
exactly what would be expected – there is a positive association between the levels of
consumption of various nutrients by women and the level of public goods available to her.
A woman who has ₹1,000 to spend, will be in a better position to utilise it if she had proper
road access to the market, to the health facility, and to the ration shop. Access to
electricity, piped water and sanitation facilities at home enhance her overall productivity by
saving time that might otherwise be spent collecting water from a nearby well or river, or
employing her physical labour where a gadget might save time.

This suggestion which comes out clearly from the scatter plots must be tested more
rigorously. To this end, we use a simple mean difference test1 across treatment and control
groups after dividing the sample into women who have a low (index less than 0.5) and a
high (index 0.5 or more) access to public goods, regardless of treatment. The distribution of
the women across the groups is shown in Table 10. The number of treated women in each
group is lopsided because overall the level of public goods access in the treatment districts
is lower than in the control group.

Table 10: Distribution of observations by treatment and by access to public goods

Poor access High access Total

1
We thank Dr Aritri Chakravarti for making this suggestion.

59
 PGI = 0 to 0.49 PGI = 0.5 to
1
Control 116 165 281
Number of
Treatment 290 162 452
women
Pooled 406 327 733

The results of the mean difference tests for the nutrient groups are presented in
Table 11. For the group with PGI less than 0.5 and so poor access to public goods, there
does not appear to be a consistent difference between the nutrition ratio in the treatment
and control groups. For the macronutrients, there is no significant difference. For the
vitamins, the treatment group NR is greater than that for the control group. And for the
minerals the control group NR is greater. While this perverse result appears to fly against
reason, our attempts to test this same relation using regressions with propensity score
matching gives better results, since it controls for various possibly confounding factors.

Table 11: Mean difference test for nutrition ratios across public goods access and treatment

PGI < 0.5 PGI > 0.5

Macronutrients Control 0.79 0.76
Treatment 0.73 0.92
p-value -- 0.007

Vitamins Control 0.34 0.36

Treatment 0.38 0.44
p-value 0.06 0.001

Minerals Control 0.88 0.88

Treatment 0.80 1.00
p-value 0.06 0.02
Notes:

 Only p-values less than 0.1 are reported. The reported p-values correspond to either
the treatment group average nutrition ratio being greater than the control group
average or the other way round, as the case may be.
 The macronutrients group contains proteins and fibres, the vitamins group includes
vitamins B1, B2, B3, B5, B7, and C and the minerals group includes chromium, copper,
iron, magnesium, manganese, phosphorus, selenium, sodium, and zinc.

On the other hand, for the group with the PGI being at 0.5 or above, there appears a
consistent result that the treatment group NR is greater than the control group NR, and this
difference is statistically significant. Given the scatter plots shown in Illustration 10, this is
exactly what is expected.

60
 The above results can be confirmed through a simple OLS regression as well. The
model is the same as earlier, but instead of one treatment variable, we use an interaction of
treatment and a PGI dummy. The results are shown in Table 12.

Table 12: Treatment effect by level of public goods access

T=1 T=0 T=1 Mean Treatment Treatment

PGI<0.5 PGI>0.5 PGI>0. NR effect (T = 1, effect (T = 1,
5 (T = 0) PGI<0.5) PGI>0.5)
Macronutrien 0.04 0.02 0.19** 0.78 25.0%
ts (0.05) (0.06) *
(0.07)
Vitamins 0.06** 0.03 0.11** 0.35 15.7% 30.6%
(0.02) (0.03) *
(0.03)
Minerals 0.01 0.04 0.18** 0.88 20.1%
(0.05) (0.06) *
(0.06)

Notes:

 The coefficients listed in the table are the coefficients on variables representing
combinations of treatment (1 for treated, 0 for not) and public goods access (high
access – PGI from 0 to 0.5 and low access – PGI from 0.5 to 1).
 The macronutrients group contains proteins and fibres, the vitamins group includes
vitamins B1, B2, B3, B5, B7, and C and the minerals group includes chromium, copper,
iron, magnesium, manganese, phosphorus, selenium, sodium, and zinc.
 The asterisks take on the traditional meanings.

As is evident from the coefficients in Table 12, there appears to be no consistent

relation between treatment and the nutrition ratio for the subgroup having a low public
goods access. In sharp contrast, the subgroup having better access to public goods seems to
show a strongly positive average treatment effect. In comparison with the average nutrition
ratio for the high PGI control group, the high PGI treatment group women have a 25 per
cent higher consumption of macronutrients, a 30.6 per cent higher consumption of
vitamins, and a 20.1 per cent higher consumption of minerals. It is clear, based on the
earlier results that these effects are greater than for the pooled observations. This indicates
that the treatment is significantly stronger in the presence of adequate level of access to
public goods. It is only in the case of vitamins that the treatment seems to have an effect

61
even with low pubic goods access, indicating that there are situations in which the
treatment is helping, but the extent is help is certainly greater with public goods.
After a proposal of the theoretical possibility, and the illustration with the scatter
plots, this table is the final proof that such a relation exists, and strongly. The women who
receive the monthly transfer, are in fact unable to use it to any advantage if they do not
have reasonable public goods access. A high level of public goods access helps the women
effectively use their monthly transfer to enhance their consumption.
There are two routes through which this effect may be operating. One is the more
direct route – women who have better access to ration shops and angawadis have a place
to spend their money such that the expenditure results in an improvement in their
consumption. If a woman has to travel (walk, bullock cart, or borrowed vehicle) over an
hour to the nearest ration shop, even though she is receiving the scheme money, she has no
convenient way to use it.
The other is a more indirect channel of influence, which, though feasible, is hard to
empirically test with the currently collected data. A greater level of access to public goods
also means a greater level of connection with government personnel. The ladies running
the anganwadis, the staff manning the ration shops, hospitals, and the line officials
responsible for providing access to LPG cylinders, or electricity connection or piped water
connection are people that low PGI group women are almost unlikely to at all encounter.
The women who do have these services, and so do encounter some government personnel
at some frequency are most likely to hear from them a thing or two about a healthy
consumption pattern, the advantages of consuming a variety of fruits and vegetables rather
than relying on roti and potato curry very frequently, about some measures of basic hygiene
that maybe prevent frequent illnesses and therefore improve their appetite, and so on.
These indirect connections may be playing their role in the effects seen above, the broader
point here being that these tribal families need to be better incorporated into the
mainstream for all the benefits it can bring.
Some brief notes on the control variable coefficients are warranted:
- The coefficient on the age of the household head is throughout negative and
significant, indicating that an older person is less able to provide adequate nutrition.
- The education variable is used as a code. The reference group is household heads
who are completely illiterate. Other groups are people who are literate without
formal education, people who have studied till grade 8, and those who have studied
more. In most cases the coefficient on the group who are literate without formal
education is negative, though not statistically significant. For the group who have
studied till grade 8, the coefficients are mostly positive, and in some cases,
statistically significant. This shows that it is some amount of formal schooling, with
its formal instruction that has the potential to improve nutrition, rather than just the
ability to read and write.

62
- The coefficient on the livestock ownership variable comes in positive and significant,
indicating that a household that owns livestock is probably able to consume more
dairy, and are therefore doing nutritionally better.
- Monthly income also comes in as having a statistically significant positive effect on
nutrient intake, which is as one would expect.

63
Chapter 7: Implementation of the scheme

The scheme has been implemented since January 2018 reasonably well, with only a
brief disruption during the early months of the COVID-19 pandemic. Across the state, the
recipients had to wait three months to receive their pending transfers all at once. Aside
from this, it appears that the vast majority of the recipients do in fact receive the transfer
safely to their bank accounts every month.

Photograph 12: Registered beneficiaries have been receiving ₹1,000 per month regularly

The system of transferring the money to the women’s bank accounts directly has
done a lot to reduce if not eliminate rent-seeking behaviour. The administrative set up
supporting the implementation is vast and efficient, as shown in Table 13.

The process of registration of beneficiaries, though not as hassle free as one might
hope, is not totally chaotic for the eligible families either. There were relatively few reports
of local agents of various descriptions trying to capture part of the money that the families
are receiving or might receive in the future. In one instance, a boy was mentioned to have
stationed himself outside the ATM, so that he can demand ₹200 from the women who were
going there to withdraw their scheme money. In another, an agent had demanded a few
hundreds of rupees in payment for preparing the application paperwork: specifically, ₹100
for the jaati pramaan patra or caste certificate, and ₹300 for the application form. These
are clearly not the rates fixed by the government, as these forms and processes are meant
to be free of charge. In Pandupiparia and Dhulania, eligible women are having to pay
anywhere between ₹500 to 1,000 for the domicile and caste certificate. In one set of
villages in Shahdol district, the fixed rate for filling the AAY form appears to be ₹700.

64
Table 13: Fund-release process

Ste Entity Process

p
1 Female heads of
households
Resident of a
district offering Eligibility criteria
the AAY Documents needed: Caste certificate, Domicile certificate, Bank
Member of a passbook, Aadhar card (preferred)
PVTG
Has a bank
account
2 Village level
Check documents and start registration process at public service
assistant for
centres on Madhya Pradesh Tribal Affairs Automation System (MPTAAS)
registration
3 Village Sarpanch Complete registration process
4 Block + District +
Head office staff at Verificaiton of all registration documents
Bhopal
5 Block + District + Send back to lower levels of governments any cases showing
Head office staff at discrepancies, and retain complete applications. Applications with
Bhopal incomplete information may be set aside as “deactivated” – this process
may be repeated as many times as needed to clean up applications
6 Fund-release process starts with sending a request for every set of 500
Head office staff
beneficiaries
7 State Bank of India Fund-release for 40 sets of 500 beneficiaries is approved and sent to SBI
staff server
8 One-click fun-release for 40 sets of 500 beneficiaries (20,000
CM
beneficiaries)

The above instances appear to be the exceptions that prove the norm. In most
cases, the women are able to register for the scheme without severe harassment and are
then able to receive the money, access it and use it in their best judgement. 89 per cent of
the respondents in Guna, 91.5 per cent in Shahdol and 66 per cent in Chhindwara claimed
overall to be happy with the implementation of the scheme. Similar proportions also felt
that the scheme was in fact helping improve their nutrition. Perception is important,
together with the facts on the ground. It is because of these general perceptions, that some
families, probably the ones who were not too happy with the scheme, said that the amount
of ₹1,000 per month was too little. One irate woman said, when asked whether the amount
transferred every month was adequate, “I can’t even see at the bottom of my shopping bag,
all the things I can buy with this money”. One gentleman suggested, helpfully, that the
amount should be increased to ₹5,000 per month.

We additionally believe, based on the interviews that the beneficiaries are also
selected without types i or ii errors. If the female head of a household is already receiving

65
the transfer, then her daughter or daughter-in-law also is not permitted to register herself,
unless she lives separately and is able to prove this. Based on the names of the heads of
household, it is clear that only the PVTG families are in fact registering for the scheme, and
not other tribespeople. The identity of the people along the lines of caste is strong enough
that it is not possible for a family to seek a caste certificate saying that they belong to a
PVTG, if in fact they do not. The caste identity is central to these women, and they wear it
on their sleeves, literally.

Photograph 13: A Bharia woman displaying her tattoo or Gudna

66
Chapter 8: Discussion and policy recommendations

This paper has attempted to study the impact of the Ahar Anudan Yojana on the
health and nutrition of PVTG women and children in rural Madhya Pradesh. This is achieved
through a 24-hour dietary recall study which has helped ascertain with a high level of
accuracy the levels of various nutrients consumed by each woman, 15 and over, and to a
lower level of accuracy for each child, 5 or under. The results show clearly that the women
in receipt of the scheme money are in fact consuming greater quantities of almost all the
nutrients. The results are largely ambiguous in the case of children.

An important aspect of the study that has surfaced through the daily reports of the
field investigators is that the tribal families interviewed greatly lack access to various public
goods. The nutrition results turn out to bear a very important relation with the level of
access to public goods. Women who have poor access to public goods, though they may be
receiving the scheme money, are unable to utilise that money to enhance their levels of
consumption of various macro- and micro-nutrients. Women with a higher level of public
goods access appear to be doing well in converting their scheme money to enhanced
nutrition. The difference is striking.

About implementation, there appear to be only a few, not very major lapses in
protocol. There are no systemic issues that need correction. At the local level, enterprising
agents have captured the rent-seeking possibilities they observed in assisting eligible
beneficiaries to register for the scheme. Since most beneficiaries are not literate and not
aware of documentation protocol, it is easy for those with access to the local government
officials to control it for the rest. This is not something that can be easily addressed at the
central level but must be addressed by the Gram Panchayat officials themselves.

One line of thinking that has persisted since the beginning of modernisation is the
preservation of tribal cultures. The view is that participation of tribal populations in the
modern economy puts a distance between them and their culture and causes valuable
aspects of these cultures to forever disappear. Based on the conversations our team had
with these tribal populations, the sense we got is that the families are more concerned with
their immediate needs of food, education and employment and do not appear to be overly
concerned about the loss of culture. Given their limited access to the modern world, there
probably already is a wedge driven in between their current lifestyles and that of their
ancestors. The major presence they feel in their lives is the lack of access to facilities that
have the potential to alleviate at least some of their many discomforts. Though it might well
be true that connecting them in a bigger way to the economic mainstream might separate
them from their cultural roots even more, it is the only way to go.

67
 Based on the above findings, the following policy recommendation may be made:

 The scheme is doing a great service to the populations who are traditionally most
removed from the mainstream. But the money spent on the scheme can be made to
perform significantly better if it were accompanied by some measure of public
goods, like roads, access to electricity, piped water connection, more accessible
health services, LPG cylinder, etc.
 An additional angle not brought up earlier is vocational. Many women have taken up
home-based micro-enterprises which involve collecting leaves and berries for sale in
the health supplies market. These enterprises have the scope to enhance their
incomes and so help their health and that of their families. Since the aim of the Ahar
Anudan Yojana is to improve the health and nutrition of these families, this can also
be achieved by improving the road connectivity of the most remove villages with the
local market areas.

Photograph 14: Home-based microenterprises

1. Chironji seeds set for drying are a highly sought-after seed for ayurvedic preparations
2. Tendua leaves gathered from the local forest, and being organised for sale

68
  It will also help to get the local level health workers to share information with the
PVTG families on strategies to improve nutrition for the whole family, including
young children who have just started consuming solid foods using cheap and locally
available ingredients that are rich in protein and fat. This is currently a conspicuous
problem, as three-quarters of all respondents reported that they do not receive any
assistance from their local Anganwadi.
 There are a few instances of rent-seeking behaviour by private agents as well as GP
officials in the process of helping the eligible families register for the scheme. The
work of private agents cannot be directly controlled, but it can be addressed
indirectly if the process of application is made smoother and simpler for illiterate
women to handle without outside help.

Further research will need to focus on any possible differential impact by tribe,
which would require a larger sample size than collected in the current project. This would
facilitate analysis at the level of a single tribe and the impact of the treatment may in fact
vary by tribe and also by nutrient. These analyses were not possible with the small sample
size of the current project. Additional research will also be needed to study the impact of
the treatment on mortality and morbidity. Given the time and resources at our disposal, we
were unable to secure the assistance of medical personnel to help with these important
questions. The presence of medical personnel would have made it possible to gather blood
and urine samples, which would tell us not just how much of various nutrients were
consumed, but, more importantly, how much has been absorbed by the body – a number
that might be quite different from what was consumed.

As for the findings of the current project, it is adequately clear that the Ahar Anudan
Yojana has tremendous potential to bring the tribal populations to the fold of the
mainstream economy, and this potential is in fact realised for a part of the PVTG population.
Some measures need to be taken to expand the influence.

~~~

69
References
Acquah, E., Darteh, E., Amu, H., & Adjei, D. (2019). Predictors of underweight in children
under-five years in Ghana. Ghana medical journal, 53(1), 71–78.
https://doi.org/10.4314/gmj.v53i1.11
Adhikari, D., Khatri, R. B., Paudel, Y. R., & Poudyal, A. K. (2017). Factors Associated with
Underweight among Under-Five Children in Eastern Nepal: Community-Based Cross-
sectional Study. Frontiers in public health, 5, 350.
https://doi.org/10.3389/fpubh.2017.00350
Adhikari, T., Yadav, J., Tripathi, N., Tolani, H., Kaur, H., & Rao, M. V. V. (2021). Do tribal
children experience elevated risk of poor nutritional status in India? A multilevel
analysis. Journal of Biosocial Science, 53(5), 683-708.
Aguero, J., Carter, M., & Woolard, I. (2006). The impact of unconditional cash transfers on
nutrition: The South African Child Support Grant.
Angeles, G., de Hoop, J., Handa, S., Kilburn, K., Milazzo, A., Peterman, A., & Team, M. S. C. T.
E. (2019). Government Haushofer, J., & Shapiro, J. (2016). The short-term impact of
unconditional cash transfers to the poor: of Malawi's unconditional cash transfer
improves youth mental health. Social Science & Medicine, 225, 108-119.
Baird, S., Ferreira, F. H., Özler, B., & Woolcock, M. (2013). Relative effectiveness of
conditional and unconditional cash transfers for schooling outcomes in developing
countries: a systematic review. Campbell systematic reviews, 9(1), 1-124.
Bhatia, M., Dwivedi, L. K., Ranjan, M., Dixit, P., & Putcha, V. (2019). Trends, patterns and
predictive factors of infant and child mortality in well-performing and underperforming
states of India: a secondary analysis using National Family Health Surveys. BMJ
open, 9(3), e023875.
Coates et al 2012 – add reference
Dietary Assessment Primer. https://dietassessmentprimer.cancer.gov/
Gangopadhyay, S., Lensink, R., & Yadav, B. (2015). Cash or In-kind Transfers? Evidence from
a Randomised Controlled Trial in Delhi, India. The Journal of Development Studies, 51(6),
660-673.
Ghosh, S., & Varerkar, S. A. (2019). Undernutrition among tribal children in Palghar district,
Maharashtra, India. PloS one, 14(2), e0212560.
Gibson 2017 – add reference
Gibson & Ferguson et al 2008 – add reference
Haushofer, J., Ringdal, C., Shapiro, J. P., & Wang, X. Y. (2019). Income changes and intimate
partner violence: Evidence from unconditional cash transfers in Kenya (No. w25627).
National Bureau of Economic Research.
Ittyerah, A. C. (2013). Food security in India: Issues and suggestions for effectiveness. Indian
institute of public administration, New Delhi, 14.
Joshi, K., Moarana, G., & Jeeva, J. C. (2016). Assessment of nutritional status of tribal
children: a case of Bhumija Munda tribe of Mayurbhanj.
Kristal, A. R., Peters, U., & Potter, J. D. (2005). Is it time to abandon the food frequency
questionnaire?. Cancer Epidemiology and Prevention Biomarkers, 14(12), 2826-2828.
Leroy, J. L., Ruel, M., Frongillo, E. A., Harris, J., & Ballard, T. J. (2015). Measuring the food
access dimension of food security: a critical review and mapping of indicators. Food and
nutrition bulletin, 36(2), 167-195.

70
Lichtenstein, A. H., Petersen, K., Barger, K., Hansen, K. E., Anderson, C. A., Baer, D. J., ... &
Matthan, N. R. (2021). Perspective: Design and conduct of human nutrition randomized
controlled trials. Advances in Nutrition, 12(1), 4-20.
Marinescu, I. (2018). No strings attached: The behavioral effects of US unconditional cash
transfer programs.
Meh, C., Sharma, A., Ram, U., Fadel, S., Correa, N., Snelgrove, J. W., ... & Jha, P. (2022).
Trends in maternal mortality in India over two decades in nationally representative
surveys. BJOG: An International Journal of Obstetrics & Gynaecology, 129(4), 550-561.
Narayanan, S., & Saha, S. (2011, May 21-27). A Case for Reframing the Cash Transfer Debate
in India. Economic and Political Weekly, 46(21), 41-48.
Narayanan, S., & Saha, S. (2020). Take home rations (THR) and cash transfers for maternal
and child nutrition: A synthesis of evidence in India. Working Paper 039, 1-50.
National Research Council. (1981). Assessing changing food consumption patterns.
https://www.ncbi.nlm.nih.gov/books/NBK216696/pdf/Bookshelf_NBK216696.pdf
Pega, F., Pabayo, R., Benny, C., Lee, E. Y., Lhachimi, S. K., & Liu, S. Y. (2022). Unconditional
cash transfers for reducing poverty and vulnerabilities: effect on use of health services
and health outcomes in low‐and middle‐income countries. Cochrane Database of
Systematic Reviews, (3).
Rao, K. M., Balakrishna, N., Arlappa, N., Laxmaiah, A., & Brahmam, G. N. V. (2010). Diet and
nutritional status of women in India. Journal of Human Ecology, 29(3), 165-170..
Saha, U. C., & Saha, K. B. (2010). A trend in women’s health in India–what has been achieved
and what can be done. Rural and Remote Health, 10(2), 296-306.
Shim, J. S., Oh, K., & Kim, H. C. (2014). Dietary assessment methods in epidemiologic
studies. Epidemiology and health, 36, e2014009. https://doi.org/10.4178/epih/e2014009
SRD, R., Julie Armstrong, M. P. H., & SRD, R. A short review of dietary assessment methods
used in National and Scottish Research Studies.
Stiller, C. K., Golembiewski, S. K. E., Golembiewski, M., Mondal, S., Biesalski, H. K., &
Scherbaum, V. (2020). Prevalence of undernutrition and anemia among santal adivasi
children, Birbhum District, West Bengal, India. International Journal of Environmental
Research and Public Health, 17(1), 342.
Thompson & Subar 2013 or 2017? Confirm please.

Web resources consulted:

 https://ourworldindata.org/wasting-definition
 https://www.unicef.org
 https://www.who.int
 https://www.who.int/news/item/06-05-2021-the-unicef-who-wb-joint-child-
malnutrition-estimates-group-released-new-data-for-2021
 https://dhsprogram.com
 http://rchiips.org/nfhs/NFHS-5_FCTS/MP/
 https://gdc.unicef.org/resource/report-india-lifted-271-million-people-out-poverty-
decade
 ICMR-National Institute of Nutrition, India (nin.res.in)

71
Appendices
Appendix 1: The questionnaire

Block 1: Identification of Sample Household

1. District: Code 2. Block: Code
3. Village: 4. Interviewer’s ID: Code
5. HHID: 6. Phone No: Numeric
Numeric
Codes for Block 1:
Item 1: Shahdol-1, Chindwara-2, Guna-3, Sidhi-4, Narasimhapur-5, Vidisha-6
Item 2: Jaisinghnagar-01, Beohari-02, Tamia-03, Harrai-04, Raghogarh-05, Aron-06, Rampur
Naikin-07, Majhauli-08, Kusmi-09, Babai (Chichli)-10, Karkeli-11, Narasimhapur-12, Lateri-13,
Sironj-14
Item 4: East- E1, E2, E3, E4, South- S1, S2, S3, S4, North- N1, N2, N3, N4

Block 2: Household Characteristics

1. Religion: Code 2. Tribe: Code
3. Since when did you start living in this district: 4. Household size: Numeric
Code
5. House ownership: Code 6. Type of floor: Code
7. Type of roof: Code 8. Electricity Connection: Yes/No
9. Major source of drinking water: Code 10. Primary source of fuel for cooking:
Code
11.Type of toilet used: Code
Codes for Block 2:
Item 1: Hinduism -1, Islam -2, Christianity -3, Others -9
Item 2: Baiga -1, Bharia -2, Saharia -3
Item 3: Since birth -1, More than 5 years -2, Less than 5 years -3, Less than 3 years -4
Item 5: Owned -1, Rent -2, Lease -3, Others -9
Item 6, 7: Cemented/ Tiles -1, Kutcha-2
Item 9: Natural-1, Piped-2
Item 10: LPG -1, Other-9
Item 11: Flush latrine (piped sewer system) -1, Open pit latrine -2, open defecation – 3

Block 3: Demographic Particulars

1. Name of head of household: 2. Gender of Head of Household: Code
3. Age of Head of Household: Numeric 4. Education of Head of Household:
Numeric
5. Occupation of the Head of Household: Code 6. If others, please specify:
7. Total number of women members who are 8. Number of Children who are 5 and

72
15 and above the age of 15: Numeric below: Numeric
9. Highest educational attainment of the 10. Highest educational attainment among
household: Numeric women in the family: Numeric
11. Total members engaged in paid work: 12. Total number of women engaged in
Numeric paid work: Numeric
13. Average monthly income of the family:
Code
Codes for Block 3
Item 2: Male-1, Female-2, Other-9
Item 5: Farm worker -01, Small farmer -02, Domestic helper -03, Brick maker -04, Motor
transport -05, Rag picker/sewage cleaner -06, Autoricksaw wala -07, Wood worker -08,
Craftsmen/ artisan -09, Carpenter -10, Security personnel -11, Sweeper -12, Coolie -13,
Household level worker -14, Office worker -15, Factory worker -16, Shepherd– 17, home-
based micro enterprises -18, Others -19
Item 13: Below 5,000 -1, 5,000- 15,000 -2, 15,000- 25,000 -3, Above 25,000 -4, In kind-5

Block 4: Livelihood
1. Primary occupation of the household: 2. Do you own land?: Yes/No
3. If yes, land holdings (in acre): Numeric 4. Usage of the land: Code
5. If farming, mention crops grown: 6. Annual crop yield (weight):
7. Income generated from land per year: 8. Do you own cattle / livestock?: Yes/No
Code
9. If yes, ownership of cattle / livestock?: 10. Products generated from
Code cattle/livestock: Code
11. Income generated from cattle /
livestock: Code
Code for Block 4
Item-1: Farming -1, Family business -2, Daily wage workers -3, Private Job -4, Cattle rearing -
5, Government job -6, others -7
Item-4: Farming-1, Cattle rearing-2, Poultry farming-3, Nothing-4
Item-7, 11: Below 2,000 -1, 2,000- 5,000 -2, 5,000- 10,000-3, 10,000-15,000-4, 15,000-
20,000-5, More than 20,000-6, Zero -7
Item-9: Cows -1, Bulls -2, Buffaloes -3, Donkeys -4, Mules -5, Goat -6, Sheep -7, Pig -8,
Chicken -9, Duck -10, Horses -11, Other -12
Item-10: Milk -1, Eggs -2, Meat for consumption/sale -3, Leather -4, Wool -5, Transportation
-6, Manure -7, Used for farming -8

Block 5: PDS and Anganwadi

1. Ration card of the family: Code 2. Do you buy the ration regularly?: Yes/No
3. How long does it take to reach the 4. Do you have Anganwadi in your locality?:
ration shop from home?: Code Yes/No

73
 5. How long does it take to reach the 6. Has your household received any goods /
anganwadi?: Code services from anganwadi in the last 4 years?:
Yes/No
7. If yes, what are the goods/services your
household received from the anganwadi?:
Codes for Block 5
Item 1: Yellow, Antyoday -1, Blue, below poverty line -2, White, above poverty line -3, Token
-4, No -5
Item 3, 5: Reach within 15 minutes -1, About half an hour -2, About an hour or more -3.

Block 6: Food Security

1. In the past 4 weeks, did you worry that your household would not If yes, how often
have enough food?: Yes/ No did this happen?:
3. In the past 4 weeks, were you or any household member not able to Code
eat the kinds of food you preferred because of a lack of resources?:
Yes/ No
5. In the past four weeks, did you or any household member have to
eat a limited variety of foods due to a lack of resources?: Yes/ No
7. In the past four weeks, did you or any household member have to
eat some foods that you really did not want to eat because of a lack of
resources to obtain other types of food?: Yes/ No
9. In the past four weeks, did you or any household member have to
eat fewer meals in a day because there was not enough food?:
Yes/ No
11. In the past four weeks, was there ever no food to eat of any kind in
your household because of lack of resources to get food?: Yes/ No
13. In the past four weeks, did you or any household member go a
whole day and night without eating anything because there was not
enough food?: Yes/ No
Codes for Block 6:
Items 2, 4, 6, 8, 10, 12, 14: Rarely (once or twice in the past four weeks) -1, Sometimes
(three to ten times in the past four weeks) -2, Often (more than ten times in the past four
weeks) -3

Block 7: AAY - Awareness and Bank Accessibility (Treatment)

1. How did you come to know about the 2. How did you learn about the eligibility
scheme?: Code conditions and documents required for AAY?:
Code
3. Who did you approach to apply for the 4. What was the fee involved in the application?
scheme?: Code Numeric
5. Did you face any difficulty when you 6. If yes, what were the issues you faced?:

74
submitted the documents: Yes/No
7. How was the problem resolved? 8. How long did it take to complete the
registration process once you applied for the
scheme?: Code
9. Month and year in which you have 10. Month and year in which the first payment
registered for the scheme: was credited:
11. What is the amount you received as 12. Regularity of payment: Code
the first payment from the scheme?:
Numeric
13. If there was any delay, what was the 14. How do you spend the money that you
reason?: Code received from AAY?: Code
15. Do you save any amount from the 16. Do you know about the CM Helpline:
fund you receive?: Yes/No Yes/No
17. Have you used it anytime for AAY 18. Was the problem resolved?: Yes/No
scheme?: Yes/No
19. Have you used the CM helpline for 20. If yes, was your problem resolved?: Yes/No
any other purposes?: Yes/No
Codes for Block 7:
Item 1, 2: Newspaper-1, Radio-2, Television-3, Family / Relatives-4, Friends / Peer groups -5,
Government office / officials-6, Other-7
Item 3: Panchayat office -1, Panchayat member -2, Village head -3, Akshaya Centre -4, Asha
worker -5, Other -6
Item 8: Within a week -1, Within 2 weeks -2, Within a month -3, More than one month -4
Item 12: Yes, I receive the amount regularly -1, No, I haven't received the amount, but it was
credited later -2, No, I didn't receive the amount, and hasn't been credited yet -3
Item 13: Bank account issues -1, I had to reapply for the scheme because of the change in
family setting -2, Other -3
Item 14: On food-1, On clothes -2, On education -3, On durable assets for home -4, Health
(medicines) -5, Others -6

Block 7: AAY - Awareness and Bank Accessibility (Control)

1. Have you heard of Ahar Anudan Yojana?: 2. Where did you first hear about the
Yes/No scheme?: Code
3. Do you know the eligibility conditions?: 4. Do you know what documents are
Yes/No required for registration?: Yes/No
5. Do you know the CM Helpline for 6. Have you ever used it?: Yes/No
grievance redressal?: Yes/No
7. If used, for what purpose? 8. Was your problem resolved?: Yes/No
Codes for Block 7:
Item-2: Newspaper-1, Radio-2, Television-3, Family / Relatives-4, Friends / Peer groups -5,
Government office / officials-6, Other-7

75
Block 8: Health and Hygiene
1. Which is the nearest health facility you 2. How long does it take to reach the
have?: Code health facility?: Code
3. Do you always seek for treatment if a 4. If yes, where do you generally go for
member of household is mildly unwell?: treatment?: Code
Yes/No/Sometimes
5. If no, why?: Code 6. Do you seek for treatment/ medical aid
in case of major ailments?:
Yes/No/Sometimes
7. If yes, where do you go for treatment?: 8. If no, why?: Code
Code
9. What do you usually prefer for treatment?: 10. How often do you wash your hands in
Code a day?: Code
11. What do you use to clean when you wash 12. How often do you clean your
your hands?: Code house?: Code)
13. Do you boil water before drinking?:
Yes/No
Codes for Block 8
Item 1, 7: Government hospital -1, Private Hospital / Clinic -2, Traditional/ local healers -3,
Others -4
Item-2: Reach within 15 minutes -1, About half an hour-2, About an hour or more-3.
Item-4: No treatment -1, Govt. facilities -1, Private Facilities -3, Pharmacy -4, Home
treatment -5, Traditional healers -6, Others -7
Item-5: Financial constraints -1, Lack of access to treatment -2, Do not believe they help -3,
Others - 4
Item-8: Financial constraints -1, lack of access to facilities-2, do not believe they help-3,
others -4
Item 9: Allopathy -1, AYUSH -2, Traditional/ local healers -3, Home remedies- 4, Other -5
Item-10: When you wake up -1, Before handling food -2, After handling food -3, After getting
back home from outside -4, After using the toilet -5, Only when you feel that your hands are
dirty -6
Item-11: Soap and water -1, Plain water -2
Item-12: Daily -1, 2-3 days in a week -2, Once in a week -3, Once in a month or less
frequently -4

Block 9: Details of women

1. Woman code: 2. Age: Numeric
3. Marital status: Code 4. Education: Numeric
5. Occupation: Code 6. If others, specify:
7. Monthly earnings: Code 8. Height (in cm): Numeric

76
9. Weight (in kg): Numeric 10. Intake of toxicants: Code
11. Number of times hospitalised in the last 12 12. List symptoms from any chronic
months: Numeric ailments:
13. List symptoms from any ailments in the last 15 14. List symptoms from any ailments
days: yesterday:
15. Any health insurance scheme?: Yes/No 16. If yes, specify:
17. Has she been pregnant any time in the last 6 18. How many children does she
years: Yes/No have?: Numeric
19. Did she receive any vaccinations during 20. Did she take iron tablets during
pregnancy?: Yes/No pregnancy?: Yes/No
21. Pre-natal care received from: Code 22. Nature of pre-natal care: Code
23. Treatment preferred: Code 24. Outcome of pregnancy: Code
25. Type of delivery: Code 26. Delivery taken place at: Code
27. Who attended the delivery (procedure)?: Code 28. Visited hospital for post-natal
care: Yes/No
29. Did you receive cash benefits or any other 30. If yes, specify:
benefits from government after delivery?: Yes/No
31. Are you breast feeding now?: Yes/No 32. Age in months of the baby you
are breast feeding now: Code
Codes for Block 9
Item 3: Married-1, Umarried-2, Separated -3, Divorced - 4, Widow -5
Item 5: Occupation of the Head of Household: Farm worker -01, Small farmer -02, Domestic
helper -03, Brick maker -04, Motor transport -05, Rag picker/sewage cleaner -06,
Autoricksaw wala -07, Wood worker -08, Craftsmen/ artisan -09, Carpenter -10, Security
personnel -11, Sweeper -12, Coolie -13, Household level worker -14, Office worker -15,
Factory worker -16, Shepherd– 17, home-based micro enterprises -18, Others -19
Item 7: 0 -0, Below 5,000 -1, 5,000- 10,000 -2, 10,000- 15,000 -3,
Item 10: Beedi -1, Cigarettes -2, Alcohol -3, Betel Leaves-4, Powa -5, Other -6, I don’t use any
of these – 7
Item 21: Government hospital -1, Anganwadi -2, Private hospital -3, Home Treatment -4,
Other -5, None -6
Item 22: Nutrition supplements -1, Nutritious food -2, Regular check-up -3, Information on
healthy pregnancy and childcare -4, Other -5, None -6
Item 23: Allopathy -1, AYUSH -2, Traditional/local healers -3, Home remedies -4, Other -5
Item 24: Live birth -1, Stillbirth -2, Abortion -3, Pregnancy continuing -4, Miscarriage -5,
Other -6
Item 25: Normal Delivery -1, C-section -2
Item 26: Government hospital -1, Private hospital -2, Home -3, Other -4
Item 27: Doctor -1, Nurse -2, Family members -3, Dai -4, Others -5
Item-32: 0-6 months -1, 6-12 months -2, More than 12 months -3

77
Block 10: Details of children in the family 5 years or below
1. Child code: 2. Age (in months): Numeric
3. Gender: Code 4. Order of birth: Numeric
5. Full term birth: Yes/No/Don't know 6. Height (in cm): Numeric
7. Weight (in kg) : Numeric 8. Number of times hospitalised in the last
12 months?:
9. If suffering from any chronic ailments, list 10. If suffering from any ailments in the last
symptoms: 15 days, list symptoms:
11. If suffering from any ailments since 12. Has the child received all vaccinations?:
yesterday, list symptoms: Yes/No
13. Any health insurance?: Yes/No 14. If yes, specify:
15. Preferred treatment: Code 16. Is the child breast fed now?: Yes/No
17. Was the child breast fed within one hour 18. If no, why?:
of delivery?: Yes/No
19. Was the child fed with the first milk of 20. If no, why?:
the mother after delivery?: Yes/No
Codes for Block 10
Item-3: Male -1, Female -2, Other -3
Item-15: Allopathy -1, AYUSH -2, Traditional/local healers -3, Home remedies -4, Other -5

Block 11: Mortality

1. Have any women 15 years or over or children 2. Code: Numeric
5 years or under passed away in the last 5
years?: Yes/No
3. Gender: Code 4. Age at death (in
hours/days/weeks/months/years):
5. Reason for death: 6. Medical attention received before
death?: Yes/No
7. If no, why didn't you seek medical
treatment?:
Codes for Block 11
Item-3: Male -1, Female -2, Other -3
Item-17: Lack of availability -1, Affordability -2, Accessibility -3, Acceptabilty -4

Block 12: 24-Hour Dietary Recall

1. Meal prepared: Code 2. Item prepared/consumed directly:
3. Main ingredients: 4. Volume of ingredients used:
5. Total volume of food prepared: 6. Item and volume of food consumed by
woman 1:
7. Item and volume of food consumed by 8. Item and volume of food consumed by
woman 2: woman 3:

78
9. Item and volume of food consumed by 10. Item and volume of food consumed by
woman 4: woman 5:
11. Item and volume of food consumed by 12. Item and volume of food consumed by
child 1: child 2:
13. Item and volume of food consumed by 14. Item and volume of food consumed by
child 3: child 4:
15. Item and volume of food consumed by
child 5:
Codes for Block 12
Item 1: Day time meal -1, Dinner -2, Other -3

Block 13: Food Consumption Expenditure

1. Cereals:
3. Pulses:
5. Green Leafy Vegetables:
7. Other vegetables:
9. Meat:
11. Fish:
Items, quantity and expenditure of cereals
13. Egg:
purchased: text, numeric
15. Milk and milk products:
17. Fruits:
19. Oils/Ghee:
21. Spices:
23. Tea/Coffee:
25. Bakery:
Codes for Block 13
Daily -1, Weekly -2, Monthly -3, Never -4

Block 14: Non-food Consumption Expenditure of the Household

Expenditure over the last 12 months on:
1. Education 2. Medical
3. Clothing/ Footwear 4. Jewellery
5. Debt repayment
Monthly expenditure on:
6.Rent 7.Electricity
8.Water 9.Gas/ Kerosene/ Firewood
10.Beedi/ Tobacco/ Alcohol 11.Transportation
12.Telephone/ Mobile/ internet 13.Newspaper/stationary items

Block 15: Possession of consumer durables

1.Mattress: Yes/ No 2.Cot: Yes/ No

79
3. Refrigerator: Yes/ No 4.Washing Machine: Yes/ No
5.Chair: Yes/ No 6.Table: Yes/ No
7.Electric fan: Yes/ No 8.Radio: Yes/ No
9.TV: Yes/ No 10.Sewing machine: Yes/ No
11.Sofa: Yes/ No 12.Almirah: Yes/ No
13.Mixie: Yes/ No 14.Grinder: Yes/ No
15.Clock/ Watch: Yes/ No 16. Telephone: Yes/ No
17. Small Mobile Phone: Yes/ No 18. Bicycle (2-wheeler): Yes/ No
19. Motorcycle (2-wheeler): Yes/ 20. 3-wheeler: Yes/ No
No
21. Bullock cart: Yes/No 22. Car (4-wheeler): Yes/ No
23.Tractor (4-wheeler): Yes/ No 24. Gold: Yes/ No
25. Smart Phone: Yes/No 26. None of the above

Block 16: Respondent’s View on AAY (Treatment)

1. Are you satisfied with the procedures 2. How do you think you benefitted from AAY?
and implementation of the scheme?:
Yes/ No
3. Do you have any suggestions for further 4. In your opinion is AAY able to improve the
improvement of this scheme?: nutritional status of the beneficiaries?: Yes/ No
Notes: Upload Location: Auto

Block 16: Respondent’s View on AAY (Control)

1. Would you like to be included in the AAY scheme: Yes/ 2. Notes:
No
3. Location:

80
Appendix 2: Villages chosen based on Census 2011
A2.1 North zone:

Vidisha – Control Guna – Treatment

Blocks Villages Blocks Villages
1 Lateri Danwas Raghogarh Raghogarh
2 Lateri Bajna Raghogarh Umaria
3 Sironj Chhapu Raghogarh Isatoda
4 Lateri Narayanpur Aron Mudrakhurd
Kalan
5 Lateri Pathera Chandu Aron Aron
6 Lateri Dhirgarh Raghogarh Khairai
7 Lateri Esagarh Aron Bhora
8 Lateri Raipura Aron Dimariyai
9 Lateri Chamar Umariya Raghogarh Godia
10 Sironj Bhokari Raghogarh Bhadoudi
11 Lateri Tajpura Raghogarh Bhulai
12 Lateri Mukata Kherdi Raghogarh Malia
13 Lateri Arino Abad Raghogarh Majhera
14 Lateri Fatehgarh Raghogarh Naandner
15 Lateri Mohanpur Khurd Raghogarh Anandpur Moia
16 Lateri Chanderi Aron Sarai
17 Lateri Sojani Kheda Raghogarh Belka
18 Sironj Okhlikheda Raghogarh Jalalpur
19 Lateri Aliganj Aron Roriya
20 Sironj Tahvariya Raghogarh Barodakala

A2.2: East zone

Sidhi – Control Shahdol – Treatment

Blocks Villages Blocks Villages
1 Rampur Kushmahar Jaisinghnaga Bataudi
Naikin r
2 Majhauli Gijwara Jaisinghnaga Jaisinghnagar
r
3 Majhauli Gajri Jaisinghnaga Jamudi
r
4 Rampur Dhanaha Jaisinghnaga Karki
Naikin r
5 Majhauli Naudhiya Jaisinghnaga Bansa
r
6 Rampur Chakdaur Beohari Boddiha
Naikin
7 Majhauli Karmai Jaisinghnaga Devra
r

81
 8 Kusmi Umariya Jaisinghnaga Mohani
r
9 Majhauli Bakwa Jaisinghnaga Bansukli
r
10 Majhauli Chhuhi Jaisinghnaga Kanadi Khurd
r
11 Rampur Posta Jaisinghnaga Kothigad
Naikin r
12 Majhauli Nebuha Jaisinghnaga Pipri
r
13 Rampur Umariha Jaisinghnaga Chhuda
Naikin r
14 Rampur Ahirantola Jaisinghnaga Dhodha
Naikin r
15 Majhauli Chamradol Jaisinghnaga Gajwahi
r
16 Rampur Sanda Jaisinghnaga Tetka
Naikin r
17 Rampur Gaurdaha Beohari Jagmal
Naikin
18 Kusmi Medara Jaisinghnaga Dholar
r
19 Kusmi Tansar Jaisinghnaga Lakhnaoti
r
20 Kusmi Dhupkhad Jaisinghnaga Jora
r

A2.3: South zone

Narsimhapur - Control Chhindwara – Treatment

Blocks Villages Blocks Villages
1 Babai (Chichli) Chandan Harrai Chilak
Kheda
2 Narasimhapur Mungwani Harrai Bhaiskhoh
3 Narasimhapur Gorakhpur Harrai Atariya
4 Babai (Chichli) Raipur Harrai Madhi
5 Narasimhapur Patha Pipariya Tamia Ghatlinga
6 Kareli Hiranpur Tamia Bhodiyapani
7 Kareli Dilheri Harrai Bhoipar
8 Narasimhapur Deori Kalan Tamia Linga
9 Kareli Richhai Tamia Kareaam Rated
1 Narasimhapur Usari Harrai Kamthi
0
1 Narasimhapur Dudwara Tamia Tamia
1

82
1 Kareli Machamau Tamia Sajkuhi
2
1 Babai (Chichli) Ukasghat Tamia Harrakachar
3
1 Babai (Chichli) Badagaon Tamia Chakhla
4
1 Kareli Piparha Tamia Bodalkachar
5
1 Narasimhapur Bandhi Tamia Kapurnala
6
1 Kareli Kumhroda Tamia Doriyakheda
7
1 Kareli Khapa Tamia Pandupipariya
8
1 Narasimhapur Mehgaon Tamia Sidholi
9
2 Narasimhapur Bandroha Tamia Lotiya
0

83
Appendix 3: Histogram showing the nutrition ratio for each macronutrient, vitamin
and mineral for treatment (1) and control (0) districts
0 1
20
15
Percent
10
5
0

0 1 2 3 4 0 1 2 3 4
Protein
Graphs by Treatment

0 1
20
15
Percent
10
5
0

0 2 4 6 0 2 4 6
Fibre
Graphs by Treatment

0 1 0 1
15

15
10

10
Percent
Percent
5

5
0
0

0 1 2 3 0 1 2 3 0 .2 .4 .6 0 .2 .4 .6

Vitamin B1 Vitamin B2
Graphs by Treatment Graphs by Treatment

00 11
20
15
15
Percent
Percent
10
10
55
00

0
0 11 22 33 00 11 22 33
Vitamin
VitaminB5
B3
Graphs
Graphs by
by Treatment
Treatment

84
 0 1
0 1

30
100

20
Percent
Percent
50

10
0
0 .2 .4 .6 0 .2 .4 .6
0

0 100 200 300 400 0 100 200 300 400 Vitamin B7

Vitamin B6 Graphs by Treatment
Graphs by Treatment

0 1
0 1
80100
60
Percent
Percent
4050
20
00

0 10 20 30 0 10 20 30
0 5 10 0 5 10
Vitamin B9
Graphs by Treatment Vitamin C
Graphs by Treatment

0 1
40
30
Percent
20
10
0

0 .5 1 1.5 0 .5 1 1.5
Vitamin E
Graphs by Treatment

0 1
60
40
Percent
20
0

0 5 10 15 0 5 10 15
Vitamin K1
Graphs by Treatment

85
 20
15 0 1
Percent
10
5
0

0 .5 1 0 .5 1
Calcium
Graphs by Treatment

0 1
20
15
Percent
10
5
0

0 .001 .002 .003 0 .001 .002 .003

Chromium
Graphs by Treatment

0 1
15
10
Percent
5
0

0 1 2 3 4 0 1 2 3 4
Copper
Graphs by Treatment

0 1
15
10
Percent
5
0

0 .5 1 1.5 0 .5 1 1.5
Iron
Graphs by Treatment

86
 0 1
20
15
Percent
10
5
0

0 2 4 6 0 2 4 6
Magnesium
Graphs by Treatment

0 1
15
10
Percent
5
0

0 2 4 6 0 2 4 6
Manganese
Graphs by Treatment

0 1
20
15
Percent
10
5
0

0 .005 .01 .015 0 .005 .01 .015

Molybdenum
Graphs by Treatment

0 1
15
10
Percent
5
0

0 1 2 3 4 0 1 2 3 4
Phosphorus
Graphs by Treatment

87
 15
10 0 1
Percent
5
0

0 .5 1 1.5 0 .5 1 1.5
Potassium
Graphs by Treatment

0 1
15
10
Percent
5
0

0 5 10 0 5 10
Selenium
Graphs by Treatment

0 1
100
Percent
50
0

0 5 10 15 0 5 10 15
Sodium
Graphs by Treatment

0 1
20
15
Percent
10
5
0

0 1 2 3 0 1 2 3
Zinc
Graphs by Treatment

Notes:

88
  The red vertical line through the histograms marks the point where the nutrient ratio
is 1, or where the woman is consuming the NIN-recommended amount of that
nutrient. Women to the left (right) of the vertical line are consuming less (more) than
recommended.
The horizontal and vertical scales on the histograms have been matched, thereby facilitating
a direct comparison of the control and treatment group distributions.

89
Appendix 4: Average treatment effect on nutrition intake, detailed results

Nutrient Coefficient Control group Treatment Treatment

mean NR effect in effect in %
physical units

Protein 0.0934** 0.726902 4.2955g 12.85%

Fibre 0.1593*** 0.830898 4.7803g 19.18%

Vitamin B1 0.0672* 0.70197 0.1142mg 9.57%

Vitamin B2 0.0229** 0.168252 0.0550mg 13.62%

Vitamin B3 0.0497** 0.389058 0.6961mg 12.78%

Vitamin B5 0.0845*** 0.511324 0.4224mg 16.52%

Vitamin B7 0.0270*** 0.073274 1.0786mg 36.80%

Vitamin C 0.1307* 0.258137 8.4935mg 50.62%

Chromium 0.0001** 0.000352 0.0032mg 18.44%

Copper 0.0961* 0.71803 0.1633mg 13.38%

Iron 0.0547*** 0.308153 1.5852mg 17.74%

Magnesium 0.1133** 0.750217 41.9347mg 15.11%

Manganese 0.1746* 1.918144 0.6985mg 9.10%

Phosphorus 0.1052** 0.905294 105.2221mg 11.62%

Potassium 0.0615*** 0.395163 215.2605mg 15.56%

Selenium 0.4819*** 3.078333 19.2760µg 15.65%

Sodium 0.1707** 0.233038 341.3152mg 73.23%

Zinc 0.0708** 0.482412 0.9343mg 14.67%

Notes: *: 10%; **: 5%; ***: 1% level of statistical significance

 The coefficients for Vitamins B6, B9, E and K1, and calcium and molybdenum were
not statistically significantly different from zero: these results have not been reported

90
Appendix 5: Relation between nutrition ratio and public goods access

0 1
0 1
4

6
3

4
Protein

Fibre
2

2
1
0

0
0 .5 1 0 .5 1
0 .5 1 0 .5 1
PGI
PGI
Fitted values protiens
Fitted values fibre
Graphs by Treatment
Graphs by Treatment

0 1 0 1
3

.6
2

Vitamin B2
Vitamin B1

.4
1

.2
0
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1
PGI PGI
Fitted values b1 Fitted values b2
Graphs by Treatment Graphs by Treatment
0 1
3

0 1
3

0 1
Vitamin B5

30
2
Vitamin B3

20
1

Vitamin B9
0

10

0 .5 1 0 .5 1
0

PGI
0 .5 1 0 .5 1
PGI Fitted values b5
Fitted values b3 Graphs by Treatment
0

Graphs by Treatment 0 0 1 1
0 .5 1 0 .5 1
10

PGI
0 1
Fitted values b9
.6
400

Graphs by Treatment
300

Vitamin C
5 B7
.4
Vitamin
Vitamin B6
200

.2
100

0
0

0 0 .5 .5 1 10 0 .5 .5 1 1
PGI
0

PGI
0 .5 1 0 .5 1
Fitted
Fittedvalues
values c b7
PGI
Graphs by Treatment
Graphs by Treatment
Fitted values b6
Graphs by Treatment

91
 0 1
1.5 0 1

15
10
1

Vitamin K1
Vitamin E

5
.5

0
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1

PGI PGI

Fitted values e Fitted values k1

Graphs by Treatment Graphs by Treatment

0 1 0 1

.003
1

.002
Chromium
Calcium
.5

.001
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1
PGI PGI
Fitted values calcium Fitted values chromium
Graphs by Treatment Graphs by Treatment

0 1 0 1
1.5
4
3

1
Copper

Iron
2

.5
1
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1
PGI PGI
Fitted values copper Fitted values iron
Graphs by Treatment Graphs by Treatment

92
 0 1 0 1
6

6
4

4
Magnesium

Manganese
2

2
0

0
0 .5 1 0 .5 1 0 .5 1 0 .5 1
PGI PGI
Fitted values magnesium Fitted values manganese
Graphs by Treatment Graphs by Treatment

0 1 0 1
.015

4
3
.01
Molybdenum

Phosphorus
2
.005

1
0
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1

PGI PGI

Fitted values molybdenum Fitted values phosphorus

Graphs by Treatment Graphs by Treatment

0 0 1 1
10
1.5
Potassium
.5 Selenium1
5
0
0

0 0 .5 .5 1 10 0 .5 .5 1 1
PGI
PGI
Fitted values
Fitted values potassium
selenium
Graphs by Treatment
Graphs by Treatment

0 1 0 1
3
15

2
10
Sodium

Zinc
1
5
0

0 .5 1 0 .5 1 0 .5 1 0 .5 1
PGI PGI
Fitted values sodium Fitted values zinc
Graphs by Treatment Graphs by Treatment

93
Notes:

 The x-axis measures the public goods index, constructed as a simple average of five
variables which may be considered as a reflection of availability of public goods.
These are: i. electricity connection at home; ii. piped water connection at home; iii.
use of LPG as cooking fuel; iv. average of how much time it takes to reach the nearest
anganwadi and ration shop; and v. the time it takes to reach the nearest government
hospital. The values are normalised to 1, such that if a woman has the best access to
each of these, her score would be 1, and if a woman has access to none of these, and
takes the longest time to reach the hospital and anganwadi and ration shop, her
score would be 0.
 The y-axis measures the nutrition ratio for each nutrient the consumption of which
was computed, and for which a recommended consumption level for adult women
exists. NR of 1 indicates that the woman is consuming just as much as is
recommended. This ratio is less than 1 for most nutrients and for most women in the
sample.

It is evident that the line of best fit drawn through each scatter has a positive slope in the
case of almost all the treatment group scatter plots, but the relation is more ambiguous in
the case of the control group scatter plots. This indicates that the treatment is helping the
women enhance their level of consumption better when there are public goods available,
rather than when such public goods are scarce.

94