4. IGNORING INEQUALITY

The distribution or division of the desirable things in any society—such as wealth, income,

good health, status, opportunity, upward mobility, and access—depends on the depth or

extent of the divisions in society and its geography. Or, as I have written elsewhere, “on the

joint outcome of changes to social inequality and spatial inequality.”1 Social inequality refers

to the differences in average life conditions and opportunities that are associated with social

identity. Brahman and Dalit are different not because of their biology—their genes cannot be

distinguished in a lab—but because, on average, they have unequal starting points and

opportunities in life. Similarly, one’s place of birth creates unequal starting points. Being

born in Gurgaon district in Haryana, for instance, provides an average starting point that is far

ahead of a birth in Nabrangpur district in Odisha. This gap can be thought of in terms of

spatial or geographical inequality.

In this chapter, I show that the most important features of material reality and

inequality in India—about income, wealth, and social mobility—are effectively unknown.

We have some basic information about geographical inequality, but little that is useful about

income or wealth from government data. These conditions are unknown because they are not

measured, or measured poorly, or disputed, or denied, or ignored. There are solid indicators

from non-government sources, and they show that inequality in India is very high and

increasing. This is true of all forms of inequality—between families, between social groups,

and between places. The level of income inequality in India may even be the highest in the

world. But the available indicators are often deliberately misread by experts and not used at

all by politicians.

I argue that as meticulously as the social divisions in India were established through

information gathering and categorization in British and independent India, just as zealously

have the most meaningful manifestations of these divisions not been counted in either British

or independent India. These are the very same inequalities that have been the source of the

most significant social divisions (between Forward and Backward castes and tribes, and

between religions) and have given rise to the most extensive social policies (on reservations)

and politics (the caste- and religion-based party formations that dominate much of India). If

it were not real, this situation—in which we appear to know least about the very thing we

profess to care most about—would be considered farcical.

*****

India’s social and geographical divisions arguably have more dimensions and are

deeper than in any other country. Most countries are not divided by religion, and neither are

they as divided by language; in fact, religious and/or linguistic homogeneity are often the key

bases along which national identity is created. And no other country is divided by caste or

anything resembling caste. At the same time, geographical differences in the quality of life in

India—between states (like Bihar and Goa, for instance) or districts (between Nabrangpur

and Gurgaon)—are arguably larger than in any other country.

As a result, it is possible that India is the most divided or diverse country in the world,

and in some ways, the most unequal. (This is not an overstatement, as I show later in this

chapter.) Most citizens are aware of India’s diversity. Some feel pride in it, some are

antagonistic. But most citizens are either unaware of or oblivious to the other side of the coin

of diversity—that is, inequality. From the miserable backwaters of predominantly Adivasi

districts in eastern and central India to the gleaming plushness of parts of Mumbai,

Bangalore, and Delhi, the differences in wealth, income, consumption, health, and education

are so vast as to be unmatched. They are unmatched geographically (that is, in comparison to

other countries in the world) and through time (because this is without doubt the peak of

inequality in Indian history; it has never been this acute, not even under foreign rule).

The social divisions of India—by religion, caste, and tribe—have become

institutionalized. There is a vast bureaucratic and legal apparatus at the central and state

government levels that assigns social identities and fine tunes the rights available to those

social identities. This apparatus is by no means a finished product because the politics of

identity, especially at the state level, is oriented primarily toward negotiating both these

identities and rights. This is exactly the reason for the Jat arakshan sangharsh and the

Kanhaiya Kumar Dalit andolan that created the uproar in Delhi in February, 2016, where this

book began.

But, those struggles over rights and their fine-tuning by governments takes place with

little to no knowledge of their effects. As we shall see in this chapter, we do not know what

individuals or households earn in India because income has never been measured by the

government. So, there are no official data on Dalit, Brahman, Muslim, or Adivasi incomes.

Though there are official data on wealth, they are so inadequate as to be less than useless;

they are misleading and counterproductive. What little knowledge there exists on income,

wealth, and inequality is confined to tiny expert circles and, at the same time, disputed among

them. As a result, there is very little official or agreed upon knowledge about the true extent

of income or wealth or social inequality today. There is even less knowledge on how these

inequalities have changed in recent decades while the population grew well over three-fold

after independence and the per capita gross domestic product grew six-fold.

This chapter is an exposition and indictment of this paradoxical condition in which

the rhetoric on social inequality is far in excess of information on its manifestations. For

example, so paltry is the information on inter-caste inequality—say on the difference in

income or wealth between Brahmans or upper caste groups and Dalits—that the discourse on

social inequality often becomes one about humiliation and dignity. A prominent recent case

in point is Ramnarayan Rawat and K. Satyanarayana’s compendium Dalit Studies (and Gopal

Guru’s essay in the same volume), that begins by locating both Indian nationalist and Dalit

political consciousness at the same source—“the categories of humiliation and dignity.”2

Little is known about the extent of inter-caste inequality of income or wealth or any other

measure with a more objective standard than humiliation or dignity. Even less is known

about intra-caste (or within-caste) inequality. Whereas it is obvious that all Brahmans are not

well-off nor are all Dalits poor, there is, as far as I can tell, almost no statistical accounting of

this reality—that is, how many Brahmans are poor and how many Dalits are well-off (the so-

called “creamy layer”).3 The same state of ignorance exists in the domain of inter- and intra-

religion inequality.

There is a reality of inequality in India. Just because much of it appears to be

unknown does not mean it is not real. In fact, it is possible to piece together an incomplete

but reasonable account that shows the extent of these inequalities. That is, it is possible to

dig deep into expert domains, especially non-governmental sources, and unearth some

indicative information (as I do in this chapter and Appendix 3). But, there is no agreement on

this information among experts; and among non-experts this information does not seem to

have any existence. This unaccounted reality, I argue, is the source of much right-wing

political mobilization that includes the demand for reservations by dominant groups like Jats

in Haryana and Patels in Gujarat and the systematic efforts by middle and upper caste groups

to subvert Dalit politics.

An explanation is needed. Why would a nation that appears to care much for social

inequality—a concern that is demonstrated openly in its policies and politics—care so little to

find out how much inequality there is or whether its supposedly progressive redistributive

policies are working? That is, whether reservations and other social policies are doing the job

they are meant to do? Whether the benefits of economic growth are reaching all social

groups more or less equally? Whether the post-liberalization growth of the economy has

been “inclusive?” The fact that we do not know the answers to these questions raises the

larger question: Why do we not know the answers? Why remain in this state of ignorance?

What purpose or whose agenda does this ignorance serve? Is there a deep conspiracy at work

or is there something about the nature of information or the nature of politics that explains

this curious absence of what should be vital political information?

Later in this chapter I show that the nature of inequality information may have a lot to

do with ignorance about it. The inequality information, as it is currently available, may be

too complicated to use (which may make this chapter too complicated to read). The right

kind of inequality information—that is simple, at the right scale, and usable by non-experts—

is not available. But that does not let the politics of information off the hook. I argue that

this state of ignorance is not accidental, neither is it the result of lack of government capacity

or competence, nor because it is too difficult to obtain this information, but because it serves

a political purpose. The absence of information allows every interested party to make

whatever claim they wish to make. It is convenient for them to not have the facts because the

absence of facts allows them to appeal to whatever constituency they wish to target.

In short, purposeful ignorance on inequality in India serves the political purpose of all

political actors. The same reason is behind the purposeful avoidance of collecting caste

demographic data; and having collected them, for the first time in 80 years in 2011, refusing

to divulge them. In this era of increased political competition, true information on the

economic conditions attached to social identity is a powder keg. If it explodes it can smash

indiscriminately; no one is safe because no one controls the narrative. On the other hand, the

absence of information is an opportunity for all to shape whatever narrative serves their

purpose. Ignorance on inequality is political bliss for all.

A Primer on Inequality

Before we proceed further, and at the risk of explaining things that are known or obvious to

many readers, let us begin with some basic ideas on inequality. Inequality is a

multidimensional phenomenon that is also conceptualized in several distinct ways. As a

result, the broad swathe of the multiple dimensions and conceptualizations of inequality

forms what is very likely the biggest subject of analysis among social scientists. In order not

to get bogged down in these fundamental issues, they are placed as a separate discussion in

Appendix 3. Readers interested in these basics should find Appendix 3 useful. A summary

of some of the key ideas below should be sufficient for us to move forward with the main

arguments of this chapter.

Inequality is another word for disparity or unevenness. We understand inequality by

measuring outcomes on variables that matter. In other words, inequality is a

multidimensional and multi-conceptual phenomenon that only becomes real when the

conception is operationalized—that is, when a fuzzy idea is converted to quantifiable

phenomena and the phenomena are then measured. Inequality becomes real through

quantification or measurement. If a dimension cannot be quantified—such as happiness—it

is not possible to analyze inequality for that dimension.4

Quantification produces information in the form of data. Because of this, almost all

inequality research takes the form of data collection and analysis. These masses of data have

to be simplified in order for all people—from the researchers themselves to other interested

parties—to make sense of it. In fact, the need to reduce complexity in inequality research is

no less important than in any of the other phenomena we have given attention to so far. As a

result, inequality researchers from different fields have developed what may be called simple

measures of inequality. We will pay attention to some of the simplest of these measures in

the following pages.

A word about the different conceptualizations of inequality may be useful here (more

details are in Appendix 3). These different conceptualizations exist largely because of

differences in the knowledge systems and methods used in the different social science

disciplines. There is some overlap, of course, because the boundaries between academic

disciplines are not watertight, and many methodologies are common between them. But, by

and large, it is possible to associate specific academic disciplines with specific

conceptualizations of inequality. To simplify, let us think of three distinct conceptualizations

and their associated academic disciplines: income distribution or income inequality in

economics, social inequality in sociology and anthropology, and spatial inequality in

geography.

Perhaps the best description of income inequality is the one provided by Jan Pen in

his parade of dwarfs and a few giants.5 Let us say that it was possible to arrange a parade of

all income earners in a society where each person’s height is proportional to her income; that

is, an average income earner would be of average height, say about five and a half feet. If

such a parade were to last for one hour, starting with the lowest income earner and ending

with the highest, one would “see” the income distribution of a given territorial space in

dramatic light. The parade would begin with individuals walking on their hands, representing

negative income earners. Using 1978-79 data for the United Kingdom, Anthony Atkinson

summarizes the rest of the parade:6

Next come old age pensioners (with) the height of the pensioners not much over a

foot. After them come low paid workers, with…the rule of women first for each

occupation… The slowness with which the height increases is one of the striking

features of the parade… It is only when we pass the average income (with twenty-four

minutes to go) that events begin to speed up, but even when we enter the last quarter

hour (the top 25 percent), the height of marchers is only some 7’. But then they begin

to shoot up. Police superintendents are 11’ tall. The average doctor or dentist is some

14’. Around 20’ come senior civil servants, admirals and generals. The chairman of

a medium sized company may be 35’ and for larger companies his height could be 35

yards. Indeed, the highest paid directors are…over 70 yards tall. They are not,

however, the last, since the final part of the parade is made up of people of whom Pen

says ‘their heads disappear into the clouds and probably they themselves do not even

know how tall they are’.

Keeping that vivid image in mind, consider an illustration of the different

conceptualizations of inequality in Figure 4.1 which combines the “Pen’s parade” insight

with different ways of organizing information about a society that is divided into two groups.

Let us call the groups “grey” and “black.” One can imagine these two groups in any way one

likes—Forward and Backward caste, Hindu and Muslim, vegetarian and non-vegetarian, etc..

Let us also assume, like Pen and Atkinson, that the height of each individual is proportional

to his or her income. Figure 4.1a shows a random arrangement of 50 individuals—25 each

from the groups grey and black. Because they are randomly arranged, it is not possible to say

much about the overall distribution other than what is obvious: that both the grey and black

groups have some tall (or high income) individuals, some short (low income) individuals, and

some individuals of medium height (middle income).

When we sort these individuals by height and arrange them by rank (in Figure 4.1b),

we are able to see Pen’s Parade. This curve represents inequality in this full population. The

properties of this curve—such as, how much it sags away from the diagonal—can be

estimated (using methods that range from simple to complicated) and summary calculations

of inequality derived from it. This curve is analogous to income inequality in the full

population of grey and black individuals in this hypothetical distribution. Economists are

primarily interested in this distribution.

Now, the same exercise can be done with the grey and black populations separately.

We can sort and rank the black group (Figure 4.1c) and grey group (Figure 4.1d) separately

and estimate the inequality within these groups by analyzing their separate curves of

inequality. These can be thought of as “within-group” inequalities (analogous to inequality

within Forward castes and within Backward castes separately). Now, each group (grey and

black) has an average height (or income). In this illustration the grey average is higher than

the black average. The difference between these averages is analogous to “between-group”

inequality; that is, the inequality between Forward and Backward castes (or, as I show below,

between Forward and Backward states or districts).

So, the distribution of income can be studied using an abstract method in which

everyone in India—from the most destitute to Mukesh Ambani—is ranked without reference

to anyone’s social identity (this is the common method used by economists). Or, it can be

done by grouping society by social identity and looking at the differences within and, in

particular, between groups.

This “within” and “between” distinction is important. We know that an average (or

mean) is merely one representation of a group. This is illustrated by the “Bill Gates walks

into a bar” story: before he enters the bar, the average wealth of its occupants may be USD

100,000; after he enters it could be a billion dollars or more (depending on how many people

are in the bar). All groups have internal differences—highs and lows within the groups that

are not captured by an average. So, it goes without saying, that all Brahmans do not have a

higher income or a better starting point than all Dalits; conversely, all Dalits do not have a

lower income or inferior starting point than all Brahmans.

This complication is captured by the idea that group inequality can be conceptualized

along two dimensions—between-group inequality and within-group inequality. The former

—between-group inequality—is what is typically what we mean by social inequality: these

are the differences in averages between pairs like Hindu vs. Muslim, or Forward vs.

Backward caste. But the average tells us nothing about the “poor Brahman” or “rich Dalit”

situation. There are ways to calculate this. Economists have developed a number of

“decomposable” measures of inequality (such as the Theil Index and the decomposable Gini)

which calculate the contribution of between-group and within-group inequality to total

inequality. As a general rule, within-groups inequalities contribute more to total inequality

than between-group inequalities.7 But, as I show later in this chapter, there is little useful

information on within-group inequality: that is, inequality between Dalits or between

Muslims, etc.. So, important as it is, we are unable to investigate this in any detail.

Geographical Inequality

Let us begin our exploration of inequality in the domain in which we have more information.

Geographical (or spatial) inequality is a distinct form of group inequality. Here, the groups

are not organized by social identity but by location. In some ways, this is the most obvious

form of inequality and its most obvious manifestation is when the location (or scale) is the

nation. The one unquestionable fact of international development is that there is a steep

hierarchy of national incomes: the averages range from below USD 500 per year in some

landlocked countries of central Africa to USD 60K in the U.S. to USD 100K in Luxembourg.

This difference in average incomes may be the driving force of politics and economics in the

world.

Location matters. The social identity of a person at birth frequently combines with the

location of that birth to have extraordinary influence on how the rest of that person’s life will

go. To give an international example:8

A child born in a village far from Zambia’s capital, Lusaka, will live less than half as

long as a child born in New York City—and during that short life, will earn just $0.01

for every $2 the New Yorker earns. The New Yorker will enjoy a lifetime income of

about $4.5 million, the rural Zambian less than $10,000.

The range in India is not quite as large as that (after all, the variance inside India

cannot be larger than the variance in the world as a whole), but India has deep spatial

divisions. They could be deeper than in any other country. One reason for it is India’s size—

because the bigger a country, the larger the range of possibilities in it. But the variation in

living standards in India go beyond what could be considered “normal” for a large country

(like China or Brazil).

Geographical inequality in India refers to the fact that spatial units such as states,

districts, and cities have different average incomes, so their residents have different average

opportunities. As with social inequality, geographical inequality too has between-group and

within-group components. For example, the average resident of Goa has an income that is

seven times higher than his counterpart in Bihar; but at the same time, many residents of

Bihar (from the upper end of Bihar’s income distribution) have incomes higher than many

residents of Goa (from the lower end of Goa’s income distribution). Despite an average

seven-fold difference, all Goans are not richer than all Biharis; some Biharis are richer than

some Goans.

This idea that a geographical average does not capture the range of possibilities within

a geographical space is especially true of large spaces, like big cities. There are great

numbers of people who live far above and far below the averages of such places. The

average income of a metropolis in India includes incomes of the wealthy owner of multiple

flats and his maid, cook, driver, durwan, and nanny. People who can pay crores of rupees for

an apartment live alongside people who defecate in the open, sometimes just outside the

walls of the gated estates in which these apartments are ensconced. The latter clearly do not

have the same starting point as the former.

At the scale of the state there are massive and, in many cases, growing differences in

average income (more accurately, the Net State Domestic Product per capita), poverty, and

other measures of welfare.9 For example: as mentioned above, the average income difference

between the highest-income and lowest-income states (Goa to Bihar) is more than seven-fold.

This gap between the top and bottom has grown significantly after independence. The

leading states then (West Bengal and Punjab) had incomes that were 2.5 times higher than

Bihar’s; by the late 1990’s this ratio had grown to 4, and has kept increasing thereafter.

Average farm size is about twenty-fold higher in Punjab than Kerala (over nine acres in the

former, and barely 0.5 acres in the latter), and female literacy rates are almost twice as high in

Kerala than Rajasthan or Bihar (close to 100 percent in Kerala and around 50 percent in the

latter two). The poverty rate in the mid-2000’s in Odisha and Bihar was five times larger

than in Punjab (around 45 percent compared to 8 percent); by the mid-2010’s, despite the fact

that overall poverty had declined in the country, perhaps quite significantly, the poverty rate

in states like Jharkhand and Chhattisgarh was about eight times higher than in Goa (37-40

percent compared to 5 percent).10

If the state-level differences are high, the district-level differences are considerably

higher. For instance, in Nabrangpur district in Odisha, which the Indian Express named

“District Zero” (as the least developed in the country), the poverty level in the mid-2000’s

was over 80 percent.11 There are very significant differences at the scale of districts for

poverty and other indicators of welfare (such as infant mortality, longevity, maternal

mortality). In fact, just as the language of being Backward is deeply embedded in the

discussions of caste and social inequality, the same language is part and parcel of the

language of district-level development. The Planning Commission created lists of Backward

districts on an irregular basis: in 2002 there was a list of 100 and in 2005 a list of 177 such

districts. Individual states have their own lists of Backward districts and create incentives,

quite unsuccessfully, to attract private investment into them. Bibek Debroy and Laveesh

Bhandari created a list of 69 lagging districts using their own metrics, and Jyostna Jalan and

Martin Ravallion have written extensively about “spatial poverty traps” in Indian districts.

My own work on industrialization has identified clusters of districts that receive little or no

industrial investment.12

There is no doubt that variance in development indicators (on income or poverty or

any of the other variables mentioned above) is considerably higher at the district level than at

the state level. This is to be expected, but the scale of difference is remarkable. For instance,

in 2010-11, the per capita income of the richest district in Haryana (Gurgaon at Rs. 4.5 lakh)

was ten times higher than that of the poorest district in the state (Mewat at Rs. 46,000). A

ten-fold difference existed within the same small state. Across states, Gurgaon’s average

income was 30 times higher than in District Zero, Nabrangpur in Odisha (Rs. 15,000).13 It is

worth noting that in the 2011 census, of Nabrangpur’s 1.22 million residents, 56 percent were

categorized as scheduled tribe and 15 percent as scheduled caste; that is, over 70 percent of

the population belonged in the category of marginalized (or Backward) minorities. In

Gurgaon, on the other hand, only 13 percent of the 1.5 million residents were categorized as

scheduled caste and there was not a single person classified as scheduled tribe (because there

is no official recognition or schedule of tribes in Haryana).

This is in line with the conclusion of Sonalde Desai and her associates that “a poor,

illiterate Dalit labourer in Cochi or Chennai is likely to be healthier, and certainly has better

access to medical care than a college graduate, forward caste, large landowner in rural Uttar

Pradesh.”14 The simple data shown here starkly illustrate how inequality in India is

manifested by the intersection of location and social identity. When both are classified as

“backward,” as in Nabrangpur, the combination yields the most abject living conditions in the

country.

The question arises: why use the label Backward—which is suggestive of a condition

that is ancient and unchangeable—instead of a term like “lagging”—which suggests a

condition that is temporary and changeable. To the best of my knowledge, the term

Backward is not used in any other country to identify either its regions or social groups that

are measurably behind the leading regions or groups. The term “backward region” is simply

not used anywhere other than India.

Large countries like Brazil and China have large regional differences, but they do not

use the word Backward to describe their low income regions. In other countries that are

divided by social identity (like South Africa, Brazil, and the U.S.), the condition of being low

on the development or income scale is associated with skin pigmentation, hence the language

of inequality tends to be racialized—leading to the use of census categories like branco

(white), pardo (brown), preto (black), and amarelo (yellow) in Brazil, or black, colored,

white, and Indian in South Africa. It is impossible to imagine that any of these groups or

American blacks could be officially classified as “backward.” The demand for a status or

label that gives a group preferential access to government patronage is not limited to India, of

course. But it is only in India that lagging social groups and regions are called “backward.”15

The use of this language may signal a deeply paternalistic and patronizing attitude

among the elite—the government leaders who create categories and labels—but it does not

appear to bother the groups who demand to be categorized as Backward. It is possible that

the word has lost its original bite through overuse and normalization. That is, in India,

backward no longer means what it does in the rest of the English-speaking world: which is

retarded, stupid, ignorant. Like “passed out” or “good name” or “history-sheeter,” backward

in India may have created its own meaning, which is probably something like “deprived”

(more so than “depressed” which was the label used in the early twentieth century by the

British Indian government). Hence, the purpose of reservations for “backward classes” or

special policies for “backward districts” is to mitigate deprivations. The question is: have

these policies worked? The answer, which I outlined in Chapter 1 and explain now, is that

we do not know for sure (because we do not know what would have happened in the absence

of these policies), but the likely answer is negative.

Economic and Social Inequalities

In this section I discuss the reality of inequality in India using the best available information

and data. First I consider economic inequality and the three different ways it is

conceptualized: by expenditure (what people spend), by income (what people earn), and by

wealth (what people own). Following that, I consider the available information on social

inequality; that is, inequality between social groups. The sources of the analyses include

official data (produced by the government) and unofficial data (produced by non-government

institutions).

The data presentation itself is in Appendix 3. Some of the material is technical

(though I have attempted to simplify it as much as I can) and may not be of interest to all

readers. The discussion in the following pages incorporates some of that data presentation,

primarily by summarizing the key findings. To keep the data discussion simple, the only

measure of inequality used is the Gini Index. It is not a perfect measure, but there is no

perfect measure of inequality (there is a brief explanation for it in Appendix 3). It is

nonetheless the most widely used measure of inequality, most likely because it is intuitively

easy to understand. It is a number between 0 and 100 (or 0.0 and 1.0 for purists) in which

higher numbers indicate higher inequality. 0 means that everyone has an equal amount (of

income, wealth, land, or whatever distribution is of interest), 100 means that one person (or

unit) has all of it (income or wealth or land, etc.). Therefore, a Gini Index of 40 indicates

higher inequality than a Gini index of 30. The number 40 also means that 40 percent of the

resource being studied (income or wealth or land etc.) has to be redistributed to make the

Gini Index 0, that is, equal.

To put the magnitude of Gini income inequality in perspective: the lowest Gini

indexes for income in the world are in the mid to high 20’s. These low inequalities can be

found in countries reputed for their high tax and high redistribution regimes (as in

Scandinavian countries like Iceland, Finland, Sweden, and Norway) or in post-Soviet

societies in central Europe (like Ukraine, Slovenia, Slovakia, the Czech Republic, and

Belarus) that have retained some or much of the egalitarian ideology and apparatus of the

Soviet years. The highest Gini indexes in the world are in the lows 60’s. The most

egregiously high levels are in southern Africa (specifically South Africa, Namibia, and

Botswana), in regimes that are deeply divided, especially by racial groups or extractive

classes where the key is control of gems and precious minerals.16

Broadly, the story of inequality in India that emerges from the available resources and

studies is one of high and growing economic inequality, a story that is at odds with the

official narrative on inequality in India—that it is low and unchanging. The argument I make

is not an isolated one. It is one that is supported by all serious scholars of inequality in India.

Why then is there such a fundamental difference between the official and scholarly

conclusions? The simple answer is that the official position in India is based on information

on expenditure, whereas the rest of the world studies income (and, increasingly, wealth).

There are other, deeper explanations, but we can discuss those only after we have gone over

the basics.

Branco Milanovic, one of the leading scholars of inequality in the world, writes:

“How unequal is India? The question is simple, the answer is not.”17 That is largely because,

in India, we can say nothing about income inequality from official data because income has

never been officially measured. This seems like an outrageous statement, but it is true. This

is not because the Indian government does not measure social conditions. Quite the contrary.

The Indian system for gathering social statistics—led by the National Sample Survey

Organization (NSSO)—is considered among the most sophisticated and professional in the

developing world.18 But the NSSO does not estimate income in any of its national surveys. It

estimates consumption or expenditure. That is, it estimates what households spend rather

than what they earn. As a result, the estimates of inequality in India are for expenditure

rather than income.

Expenditure inequality is, however, not considered an adequate measure of inequality

of condition. Households at lower income levels tend to spend all they earn; in fact, they

often have to borrow to meet unexpected expenditures (like illness), or sell assets (like land

and gold, if they have any), or rely on remittances (money sent by close relatives working

somewhere else). Higher income households, on the other hand, are able to save; that is, they

do not spend all they earn, and instead put the additional money into assets like stocks, gold,

and property. Their unspent income is converted into wealth.

As a result, expenditures do not capture the true range of quality of life conditions,

and expenditure inequality does not provide a good sense of the true inequality of quality of

life (or opportunity or access to value-producing resources). Expenditure, by definition, is

narrower in range than income, and, by definition, expenditure inequality is lower than

income inequality. Some analysts have estimated the gap between income and expenditure

inequality for the Gini Coefficient/Index to be around 5-6 points.19 As we shall see, the gap

in India is considerably larger. It is so large that the measurement of expenditure inequality

may be meaningless in India.

The origins of this choice (to measure expenditure rather than income) goes back to

the early post-independence years when basic decisions were being taken on a number of

issues (including this one). The focus then was more on poverty than inequality. In fact,

inequality did not become a serious issue to study or fight until after the mid-1970’s, after

some development economists began to discover that economic growth did not automatically

mitigate poverty or improve the lives of populations at the bottom of the income

distribution.20 At very low levels of income (as India had in the post-independence years),

expenditure (rather than income) was rightly considered to be the superior measure of

poverty. As a result, from its very first surveys in 1951, the NSS (as it was named then) was

geared to measuring how much people spend (to understand, among other things, how many

calories they intake), in order to understand the depth and breadth of poverty in the country.

The expectation was that policies to mitigate poverty would be based on these data. That

method (of measuring expenditure rather than income) continues to be used to the present

day.21

*****

As detailed in Appendix 3, the magnitude of expenditure inequality in rural India is in

the high 20’s (using the Gini Index) and appears to be more or less unchanged in four

decades. The magnitude of expenditure inequality for urban and all-India is roughly 35-36

(using the Gini Index); this is possibly a little higher now than it was in the early-2000’s

(when the Gini was in the low 30’s).22 If these figures were true, that is, if they represented

the reality of distribution, then inequality in India would be among the lowest in the

developing world and among the most stable and unchanging.

In international comparisons of inequality, the low official Gini Indexes of the NSSO

are usually taken at face value. In the absence of official data on income in India, there is a

widespread conflation between income and expenditure inequality. They are assumed to be

the same—which leads to the misleading conclusion that India is a low inequality country

with a stable Gini hovering in the low to mid-thirties for decade after decade. The confusion

is evident in many international documents: for example, in the World Development Report

of the World Bank which mentions that “India had fairly low income inequality,” in the

United Nations Development Program which reports that the “income gini coefficient” in

India is 33.9, and in policy papers by the International Monetary Fund which use the same

figures.23 Today, in early 2018, the websites of the World Bank and IMF that list inequality

for all countries show India’s income Gini Index to be 35.1, which we know is India’s

expenditure (not income) inequality level.

This problem that official surveys in India do not report income have been tackled in

two different ways that have led to different income inequality estimates, both of which are

significantly higher than the official expenditure inequality estimates. First, income data

have been collected and analyzed by the India Human Development Survey (IHDS, details in

Appendix 3) for 2004-5 and 2011-2; the income Gini Index for both years is around 54.24

Second, S. Chandrasekhar and K. Naraparaju and I have studied two surveys of the NSSO in

which income data were collected, but for the agricultural sector alone (but not the urban

sector, nor all-India), and calculated the Gini Index to be around 60 between 2003 and 2013.25

Other analysts have gone further based on the justifiable argument that household

surveys almost always fail to capture the very top end of the income distribution. Hence,

inequality calculations based on household surveys always underestimate inequality. This

happens because survey personnel are often denied access to upper income households. This

problem is quite acute in India. For example, in the IHDS 2004-5 survey, the individual with

the highest income out of 41,000 families earned less than Rs. 22 lakh per year (about USD

48,000 at the exchange rate at that time). It seems obvious that the IHDS survey missed the

top one percent of earners. Even more troubling are the NSSO expenditure surveys. For the

2011-2 round, their highest spending group, the top five percent of urban India, averaged

expenditures of Rs. 123,000 per year (less than USD 2,300). This is roughly what

government college professors earn per month. It is clear again that the NSSO also missed

more than the top one percent (perhaps the top 3-5 percent) of consumers.

This means that the NSSO surveys severely underestimate expenditure inequality to

begin with; had the NSSO tried to measure income, it would have also failed to get

information on the highest income households. The main reason is that survey data are

useless to investigate the upper tail of income or wealth. Surveyors are never able to enter the

houses and gated apartments in which the Upper and Proto Upper Class live and ask them

about their income or wealth. Even if by some miracle some survey did manage to do so,

there is no reason to expect that they will be told the truth.

How to get income information on the high income household without having access

to them? One attempt has been made by Luke Chancel and Thomas Piketty. They

supplement household survey data (from the NSSO and IHDS) with tax data to conclude that

the top one percent of income earners captured 22 percent of the national income in 2012, the

highest share since income taxes have been collected in India.26 Laurence Chandy and Brina

Seidel use a different approach (that utilizes the gap between survey data and national

accounts statistics) to calculate India’s income Gini Index in 2012 to be 56 (rather than 36, as

calculated from NSSO’s expenditure surveys).27

Wealth inequality is expected to be higher than income and expenditure inequality

everywhere and the best available evidence shows that to be true in India too. Ishan Anand

and Anjana Thampi estimate the Gini Index of assets and net worth to be 74 and 75

respectively in 2012, having risen from 65 and 66 in 1991 (and about the same levels in

2002).28 These estimates are based on the NSSO’s All India Debt and Investment Survey

(AIDIS) which suffers from serious problems that significantly underestimate wealth

inequality. First, the NSSO is unable to get asset information on the richest households (just

as it is unable to get expenditure information from them). Second, the NSSO uses an

inadequate method of estimating the value of land and buildings (which make up 85 percent

of total assets according to their own calculations). The problems are discussed in detail in

Appendix 3. Some corrections to these problems have been made in reports from Credit

Expen

diture

(NSSO)

Income (IHDS)

Income (Chan

dy & Se

idel)

Wealth (N

SSO)

Wealth (C

redit S

uisse)

0

10

20

30

40

50

60

70

80

90

36

54 56

7583

Figure 4.2 Expenditure, Income, and Wealth Inequality in the 2010's

Gini

Inde

x

Suisse which show the Gini Index of wealth inequality in India to be 83 in 2016, among their

list of the most unequal in the world.29

The condition of inequality (as calculated from available data) in India is summarized in

Figure 4.2. NSSO survey based Expenditure inequality, which is often cited as a “true” measure

of inequality in India, is low by global standards. Income inequality—following the IHDS data

(in which income is measured, unlike the NSSO data) and corrections to it using national

accounts—is considerably higher. If correct, this would place India’s income inequality in a

cluster of high-inequality countries (many in Latin America), but not the very highest in the

world. Wealth inequality is even higher than income inequality (as is to be expected) and

increasing. If correct, this would place India among countries with the most unequal wealth

distribution (a little less unequal than countries like the U.S. and Switzerland on the one hand

and Gabon and Central African Republic on the other). However, it is quite likely that because

of inadequacies of household survey methods—including limited access to high income

households, erroneous assumptions about stocks and land, and a general opacity about the

identity, income, and wealth of the top one percent—all of these calculations of expenditure,

income, and wealth underestimate the true condition of inequality in India.

*****

The condition of social inequality (that is, the gaps between the averages of the Forward

and Backward groups) is not systematically studied in India (more on which soon), but it is

possible to collate a range of diverse works and sources on the subject. The conclusion are stark.

By all measures—expenditure, income, and wealth—the gaps between Forward and Backward

groups is very large. Moreover, the gaps have been growing in recent decades for all the

variables for which comparable temporal data are available.

Consider the evidence (the details are in Appendix 3). The average urban individual who

was neither Dalit nor Adivasi spent almost twice as much as the average rural Dalit or Adivasi in

1983. A quarter century later the former (the urban non-Dalit, non-Adivasi person) spent about

2.3 times as much as the latter (the rural Dalit or Adivasi). All the other gaps on expenditure

widened during the same period: between the rural marginalized and the rural majority, and

between the urban marginalized and the urban majority. There is unambiguous evidence of a

large and growing gap in expenditure between the socially marginalized and the rest of the

population.

In income data from the agriculture sector (from the NSSO) we see large gaps between

the non-marginalized and Backward groups, and a growing gap in income between Dalits and

the non-marginalized. The income data from IHDS show large gaps between Forward and

Backward group averages. Brahman average incomes were twice as large as average Dalit and

Adivasi incomes. The average incomes of OBC and Muslim families were about 20 to 30

percent higher than Dalit and Adivasi incomes. Other studies show that the Forward castes

progress up the income ladder most rapidly. There is income growth among Dalit and Adivasi

households too; but Dalits had the least upward mobility (experienced by 30 percent of Dalit

families) and the largest downward mobility (experienced by 41 percent of Dalit families).30 In

short, there is “higher occupational mobility among forward castes than among SCs and STs…

[and] a much higher prevalence of sharp descents among SC and ST sons.”31

The wealth scenario is even more stark and deteriorated sharply in 2002-2012. The Dalit

and Adivasi share of national wealth had each been roughly half their population share till the

early 2000’s but dropped to 40 percent in 2012. The per capita wealth of the general population

(non-Dalit and non-Adivasi) in 2012 was almost five-fold higher than that of the Backward

population. The wealth gap between the Backward and non-marginalized populations had

roughly doubled in two decades. These numbers are quite remarkable.

If we look at other important issues—such as education, poverty, and health—there are

vast and often growing gaps between the Forward and Backward groups. For example,

Brahmans, the most educated group, have twice as many years of education, are four-fold as

likely to matriculate from school, and seven-fold more likely to hold a college degree than the

least educated group (Adivasis). Adivasis are half as likely to be in college as non-marginalized

groups, and Muslims are even further behind, only one-fourth as likely to be in college as the

non-marginalized Hindu groups. Rural poverty was three- and two-times higher in the Adivasi

and Dalit populations compared to non-marginalized groups. Urban poverty was about three-

times higher for both. Malnutrition was almost twice as high for Adivasis compared to “upper”

castes, and in the 1990’s, had declined more slowly; that is, the gap was growing larger.32

*****

Location is an explanation for many of these gaps. For example, Backward groups are

more likely to live in Backward regions. These are typically rural settings where low incomes

are common (because agriculture does not pay; it is the lowest value-added activity in India and

the world) and land is valued less (because “backward” region land is in least demand).

Therefore location alone would lower the income and wealth of the Backward groups even if

they had as much land (Adivasis have more land per head, but of poor quality; Dalits have the

least land of all social groups). Location, just by itself, would therefore also increase poverty.

These same “backward” rural places also have inferior education and health infrastructure. That

would lead to inferior outcomes on education and health. We can speculate on the effects of

location, but, absent analyses that begins from a clear understanding of inequality, we cannot do

much more than guess at this point.

Finally, it is necessary to give some attention to the subject of within-group inequality.

The evidence is clear that there are significant between-group differences when we compare the

averages of marginalized or Backward groups with dominant or Forward groups. But what of

the distributions inside these Backward and Forward groups? Recall that this question is at the

heart of the political agitations by leading caste groups like Jats in Haryana and Patels in Gujarat;

a version of “poor Brahman” problem—the argument being that all Brahmans are not well-to-do

and therefore deserve special opportunities.

The data we have access to seems to show no pattern in these internal distributions within

groups like Brahmans, Forward caste, Backward caste, etc. Within-group inequality levels for

all social groups tend to correspond to the money variable being studied—they are lowest for

expenditure, high for income, and highest for wealth. This is true for both within-backward and

within-forward groups.33 One would expect that inequalities within Forward groups would be

higher than within Backward groups, and it is quite possible that if good data were available on

the top of the distribution (which is undoubtedly occupied by Forward groups) there would exist

undeniable evidence on higher inequalities within Forward groups. But with the information and

analyses available now it is not possible to make a strong claim on this issue. The bottom line is:

there are significant levels of inequality within Forward and Backward groups with little

discernible difference between them in the available data.

Ignorance is Bliss

These are the facts of inequality in India as best as they can be identified from the existing

data and studies:

Very little is known “officially” because the official statistics estimate either expenditure

(a variable that is quite inadequate to study inequality) or wealth (a variable that is

appropriate for studying inequality but is poorly surveyed and calculated). Government

(and non-government) surveys have generally been unable to capture the top end of

India’s income and wealth distribution. To remedy these inadequacies, several attempts

have been made to piece together official and “unofficial” data—a patchwork quilt of

sorts—to generate more accurate or representative profiles of inequality in India.

These patched together data suggest that income inequality in India is very high and

growing rapidly. It is certainly among the highest in the world, and, if realistic data from

the top one percent were incorporated, may even be the very highest. Wealth inequality

is significantly underestimated because of inadequacies in surveying and calculating.

Despite these flaws, India’s wealth inequality estimates are among the highest in the

world and growing rapidly.

India’s social inequalities—the gaps between the marginalized and non-marginalized

groups—are also very large, and to the extent they can be measured over time, appear to

be growing. The expenditure gap and wealth gap between the Forward and Backward

groups have grown in recent decades: this is demonstrably true of Dalits and Adivasis,

but not so for OBC’s. The income gaps are also very large and growing. And there are

massive gaps in educational attainment, poverty, and health indicators (like malnutrition).

However, there are significant inequalities within every group, Forward or Backward,

and all groups include families that are far above and far below the group averages.

These findings—of high and rising income and wealth inequalities—summarize the

strongest work done by scholars who study inequality in India. But, among the thought-

leaders of the Indian state, there is either little acknowledgment or outright denial of both

realities—that the inequalities that matter (of income and wealth) are both very high and

increasing. The position on social inequality is more complicated, and I will deal with that

separately, a few pages later.

The denial of the reality of inequality of income and wealth is not limited to any one

ideology or political party. Experts identified as left-of-center are as likely to deny it as those

identified to be right-of-center. Consider the words of Montek Singh Ahluwalia, who worked

at the World Bank and International Monetary Fund and was Deputy Chairman of the

Planning Commission of India under the Congress-led UPA regime. It is not far-fetched to

suggest that Mr. Ahluwalia was one of the principal architects of Congress economic policy

for a decade, if not longer. As Deputy Chairman of the Planning Commission he wrote:34

The perception of concentration of wealth and widening disparities is sharpened by

the tendency of the media, including especially the electronic media which now has

very wide reach, to publicise success at the top end, including the conspicuous

consumption with which it is often associated, while simultaneously focusing

attention on the depth of poverty at the other end. Both extremes are understandably

viewed as newsworthy, but in focusing disproportionately on them, the steady

improvement in living standards of the very substantial population in the middle, and

the associated rise of a growing middle class receives much less attention than it

should.

Dr. Surjit Bhalla, a highly accomplished economist and important policy figure inside

the Delhi Ring Road, both when the Congress-UPA was in power and when it was not (as

member of the Prime Minister’s Economic Advisory Council under the BJP-NDA), is just as

dismissive about concerns about inequality. He wrote:35

Often in the polemical debate about poverty and policy, and the poverty of policy, the

facts (unfortunately) become irrelevant…what is revealing is that to-date, there has

been little variation in real inequality in India…While comparative data needs to be

explored, it is likely the case that this near constancy is unusual given the “buzz” of

the conventional wisdom that inequality increases with growth and/or that Indian

inequality has sharply worsened.

And Professor Jagdish Bhagwati, a renowned economist at Columbia University who

is strongly associated with the BJP-NDA regime, wrote:36

The fact is that several analyses show that the enhanced growth rate has been good for

reducing poverty while it has not increased inequality measured meaningfully, and

that large majorities of virtually all underprivileged groups polled say that their

financial situation has not worsened and significant numbers say that it has improved.

To paraphrase these experts: inequality in India is neither high nor increasing because

the expenditure data say so; even if it has grown a bit recently, the people do not mind

because they told us so; and all of this has been blown up by the media because they only

juxtapose the extremes of conspicuous consumption and poverty. Let us say we accept that

media has a propensity to focus on extremes, but to propose that the Indian media focuses

“disproportionately” on inequality seems to suggest that there is another media out there that

I do not have access to. There is more to say on the media in the next chapter and we will

tackle the issue of what is covered by it and why at that point.

But, Ahluwalia, Bhalla, and Bhagwati are bona fide experts and should know better.

In fact, they do know better. Their stellar track records and demonstrated mastery of the

subject of inequality prove that they know better.37 Actually, one does not have to be an

expert economist at their level to know that expenditure inequality tells us almost nothing

about inequality of economic condition. One does not have to be an expert economist at their

level to know that a society in which everyone is becoming better-off may, at the same time,

be turning more unequal. That is the very point of paying attention to inequality—because a

more progressive distribution provides more welfare at the same level of national income or

growth. That is precisely why growing inequality is a matter of serious concern in very high

income societies. Getting out of absolute, caloric poverty is not the issue in those societies,

justice is, and fairness.

Consider that the poverty line for a household of four is about USD 25,000 per year in

the U.S., which is roughly fifteen-fold India’s GDP per capita by exchange rates; which

means that almost no one in the U.S. is poor by Indian standards, but almost everyone in

India is poor by American standards. This does not mean that there is no discourse of

inequality in the U.S. Quite the contrary. It is hard to believe that these experts do not know

all this, or are deceived by what “official” expenditure statistics say, or are completely

unaware of the studies of income and wealth. So the question arises: why do accomplished,

eminent people make claims that they must know are incorrect?

The most likely explanation, I believe, is ideology, which I have shown (in Chapter 1

and Appendix 1) to be a version of confirmation bias. Let us recall that definition here:

“Confirmation Bias, also called Myside Bias (to underline its self-serving property), is the

tendency to look for, interpret, favor, and remember information (‘selective recall’ or

‘confirmatory memory’) so as to confirm one’s preexisting beliefs, while being dismissive of

or denying information that is contradictory or could offer different explanations and

possibilities (to avoid ‘cognitive dissonance,’ which the human mind finds hard to handle).”

It is doubtful that any of these experts ordinarily suffers from “cognitive dissonance.” On the

other hand, it is very likely that they, like everyone else, tend to “look for, interpret, favor,

and remember information” that supports what they already believe or what is convenient for

them.

The ideology these experts from the left and right share, their common belief (which

happens to be convenient for their personal and professional ambitions) is support for

economic growth. Let me be clear that this is a very common condition: the belief in or

desire for economic growth is one of the most widely-shared features among politicians,

experts, and laypersons the world over. In the minds of many, growth is ephemeral, even

magical; it is not guaranteed nor fully understood; if by some chance or action it happens, one

should ride it—like a tiger by its tail—as long as possible, without asking too many

questions, without disturbing the flow of magic. Sustained growth is transformative: in one

generation it can reduce absolute poverty to single digit levels in a very poor society; in two

generations it can transform a low income developing nation into a developed one. Witness

China.

This line of thinking—that growth and egalitarianism are enemies, that redistribution

is a drag on strong economic performance, that inequality is inevitable with growth—is one

that has been in existence in decades. It has proven impossible to kill, despite the almost

unanimous conclusion of professional economists that it is wrong. Arthur Okun argued that

there is a tradeoff between equality and efficiency, and that redistribution was akin to

carrying water from the rich to the poor in a “leaky bucket.” Simon Kuznets suggested that

inequality increases in the early decades of development and declines later; this became the

famous Kuznets inverted-U curve of development. These ideas have been empirically

examined dozens of times, including by Montek Ahluwalia, and have been found so wanting

that Gary Fields wanted to give them a “decent burial.” Other scholars like Alberto Alesina

and Dani Rodrik have argued for the reverse causality—that inequality itself is a drag on

growth. Yet, the regressive ideas persist. Surjit Bhalla’s quote above includes a statement

about “the conventional wisdom that inequality increases with growth.” He knows, as does

Ahluwalia, that there is no such conventional wisdom.38

For some, it may be difficult to admit that inequality is increasing, as if

acknowledging that fact would delegitimize growth and the policies and political parties that

are associated with growth. For others, it may be useful to conflate social identities and

geographies: if India as a whole is growing, then one need not worry about whether Dalit and

Adivasi incomes (or Bihari or Rajasthani incomes) are growing apace or catching up. “Grow

first, redistribute later.” This conflation between India and all its social groups and regions

may be politically necessary so that the “left behinds” and other dissidents do not begin to

make electoral gains.

Is it coincidental that the expert class in India is almost exclusively comprised of

members from dominant social groups—“upper castes,” Brahmans, Jains, Sikhs (with

perhaps some representation from selected OBC communities in recent years)—the ones that

have benefitted “disproportionately” from economic growth in recent years? Is it surprising

that the groups that get to “speak” and create “text” (books, papers, policies) also interpret

reality in ways that benefit themselves? That they see what they wish to and ignore what is

inconvenient. We have seen in Chapters 2 and 3 how India’s social structure was constructed

through “text” by groups with the power to create or interpret them. I suggest that the current

obsession with the growth of the Indian economy in expert circles (and the media) is a

continuation of similar forces at work. The “official” data on (low and stable) expenditure

inequality may simply happen to be convenient for deflecting or redirecting attention away

from unpleasant and inconvenient distributional issues.

*****

But, that is not a sufficient explanation for why the statistical information on

inequality is not visible in the political discourse in meaningful ways. After all, what

Ahluwalia, Bhalla, and Bhagwati write (or I do) is only accessible by a miniscule section of

Indian society. In a political sense, what they write (or I do, or almost any scholar cited in

this book does) does not matter. It might as well be gibberish. This is expert discourse that

has not been simplified for the masses. It has not gone through the process of what I called

“second-order simplification” in Chapter 1. There I wrote that “second-order simplification,

however, is rarely done by experts. Very few technical experts have the translation skill—the

‘common touch’—that is needed to simplify expert knowledge for non-expert understanding.

Others do this work of translation. Politicians, journalists, public intellectuals, priests, and

teachers.”

Where are those politicians, journalists, public intellectuals, priests, and teachers that

should be talking about the truth of inequality—if not income and wealth inequality, at least

social inequality? These translators should exist. The Indian system of representative

democracy has seats reserved for socially marginalized groups. Relatively new political

formations like the Bahujan Samaj Party and Samajwadi Party have emerged in north India

and been electorally successful for exactly that reason. In states like Maharashtra and Tamil

Nadu, Dalit politics are less monolithic but have deep roots. Adivasis constitute between

one-fifth and one-third of the populations of large states like Odisha, Madhya Pradesh,

Jharkhand, and Chhattisgarh.

One would imagine that the measured reality of social inequality would be of great

interest to these groups, a mobilizing principle. One would imagine that there would be

political demands for a proper accounting of income and wealth by marginalized social

groups and that on finding out that they were far behind to begin with (which they knew

already) and have fallen further behind (which they suspect but do not know for sure), and

that Forward castes have five-fold the wealth they hold (and that too is likely to be an

underestimate), there would be outrage and political consequences. A delusional rationalist

could even imagine that there would also be some critical examination of the fact that there is

very high inequality within the Dalit population.

But there is none of this. To the best of my knowledge, the “facts” of social

inequality derived from official and unofficial statistics never make it to the public speeches

of Dalit or Adivasi political leaders, nor are they discussed or debated in parliament or state

assemblies by their elected representatives. In fact, these figures—even the easily available

(if grossly inadequate) expenditure data—are hard to find in the highest-quality academic

texts written by leading Dalit scholars.39 As I wrote in the beginning of this chapter, for many

leading Dalit scholars, the focus is squarely on “humiliation,” not statistical inequality. Other

scholars have studied symbolic changes on status and social distance—such as diet, marriage

ostentation, seating arrangements, etc.—to examine the question of inequality.40 The

question that rises for us is: why do the numerical “facts” of inequality seem not to matter to

the groups at the bottom of the ladder? This is serious question and I submit three

possibilities as answers.

The first possibility is that non-expert stakeholders are largely uninformed about the

statistical facts of inequality. This could happen because the inequality information has not

been sufficiently simplified for it to provide cognitive utility among the general populace.

Given that I felt compelled to have a “primer on inequality” in this chapter (which means that

I thought it was needed) and have had to devote many pages to lay out the evidence on

inequality (underlining many gaps and caveats in the evidence), most of which I have placed

in an appendix rather than the main body, it is probably not hard to conclude that the

discourse on statistical inequality remains confined to the expert domain. The “second-order

simplification” of this multidimensional and complex issue has not been done yet, at least

with statistics. As a result, this is not yet, and perhaps never will be, the stuff of the street

theater, parody, and musical comedy that one sees in Dalit political meetings in Mumbai.

A second possibility is that the available inequality information is at the wrong scale

(national) and that there is little or no usable inequality information at the needed or

appropriate scale (local). The difficulty with making political use of inequality statistics is

compounded by the fact that data are never available at the scale that most people can

comprehend or that matters to them. If inequality is itself an abstract idea that people have

difficulty with, scaling it up to the nation or world makes it even less substantial. It is a view

from far above. It has little relationship to the ground, the few square kilometers around their

living space that most people can see (in a social and political sense) and seek to understand

or change. For the Dalit in Mumbai, the person attending a musical revue making fun of

Brahmans all dressed up in their caste marks and superstitions, does it matter what the

Forward caste average income is in Bengal or Andhra or even Nagpur? What relevance does

it have to his life or political identity?

This problem of scale can take on ominous dimensions when it is compounded by the

reality that every group in India—Forward and Backward—includes very large numbers of

poor. The relatively high average income and wealth of the Forward caste group is likely to

provide little comfort to the poor from the Forward castes who may feel, or made to feel, that

reservations for Backward groups discriminate against them. Consider an American parallel:

in 2015, more than one-third of Black households (about 6 million in number) earned less

than USD 25,000, at the same time that less than one-fifth of White households (about 16

million in number) did the same. That is, Blacks were significantly overrepresented in the

low income population, but low income Whites were significantly more numerous than low

income Blacks. It is precisely this reality about inequality—that low income is not perfectly

matched to race or caste or religion regardless of the histories of oppression and

discrimination—that enables a political backlash. Like Trumpism in the U.S., the backlash is

based on identity-based mobilization of the low income among the forward groups.

These political mobilizations are distinctly geographical (for example, the red state-

blue state dyad in the U.S.) because the manifestation of this other dimension of inequality

(the “backward among the forward”) is clearly visible at local scales. People can see or

instinctively understand that there is great inequality within all groups: Forward and

Backward, Upper and Lower. All Dalits (or American Blacks) are not poor, nor are all

Forward caste families (or Whites in America) well-to-do. Therefore, even if they are

known, the statistical facts of social inequality—that some groups have been systematically

deprived and are significantly worse off—have little or no political meaning for the low

income among the “upper” groups.

In short, people choose the inequalities that matter to them. Inequality exists by

reference, through comparison. Experts may refine these comparison mechanisms as best

they can using the most sophisticated tools they possess, but people choose the comparisons

that matter to their lives.

This leads to the third possible explanation for why information on statistical

inequality does not seem to matter. It may be because the absence of simple and agreed upon

inequality information benefits all political agents and parties; because the information

vacuum allows all agents and parties to make claims that are convenient for them. Consider

the issue of “reservations”. The basic claim in India is that reservations provide benefits for

the reserved groups. Therefore, those that have reservations should seek to keep or expand

them and those that do not should seek to get them. This, in essence, is one of the core

principles of Indian politics.

Rarely is the question asked: How many specific individuals or families benefit from

reservations, or what proportion of the reserved groups actually receives a reservation

benefit? These too are statistical questions without satisfactory answers. Let us try to

generate some rough estimates. In 2011, there were 17.5 million public sector jobs in India;

if 20 percent were held by Dalits and Adivasis, there were 3.5 million jobs for them at the

same time that there were about 305 million people classified as Dalit or Adivasi (201 million

Dalit + 104 million Adivasi). If we assume that not a single Dalit or Adivasi person would

have received a public sector job without reservations, we can conclude that about 1.1 percent

of the Dalit and Adivasi population were direct beneficiaries of employment reservations. If

each direct beneficiary was from a different family—that is, there was no “creamy layer”

problem or nepotism or cronyism or corruption in getting public sector jobs—they could each

have created four more indirect beneficiaries (usually family members). Using these rather

generous assumptions it is possible that up to 5 percent of the Dalit and Adivasi population

currently benefits from employment reservations. Is this a figure a political leader can boast

about to his followers? Is a one-in-hundred chance of getting a public sector job worth

setting oneself or one’s public transportation system on fire? Or do ordinary people even

know what the odds are of getting a public sector job through reservation?

One has to conclude that they do not. Certainly there is little incentive for the

established leaders of Backward groups to acknowledge that their primary demand—for

reservations—has failed to deliver on many of its promises. In general, statistics and

quantitative information on reservations appear to have little relevance for the affected people

and their political leaders. Facts—which are valid, reliable, and verifiable information—may

have nothing to do with belief. It is possible to launch many a theoretical missile to attack

this patent problem of irrationality, but not if the very foundation of rationality is shaky. And

following the discussions in Chapter 1 (and Appendix 1 and 2)—Daniel Kahneman’s fast and

slow thinking brain, the human tendency to cognitive ease and confirmation bias, and the

principle of simple information—we should be skeptical about rationality itself.

We should be most deeply skeptical about the idea that rational individuals process all

information fully and objectively. This is an impossible burden because it is clear that in the

real world many decisions—perhaps most political decisions—are taken without any

information in the form of “facts” whatsoever. There is information, alright, but not what

passes for such among experts. Information exists in the form of categories, labels,

stereotypes, and stories—but not data. In fact, as I have argued above, data may be

unnecessary or useless. In the absence of data it is possible to stick ever more closely to

categories, labels, stereotypes, and stories—that is, what one already knows, one’s comfort

zone, the lazy “system 1” part of the brain that is self-affirming and doubt-free. The more

information there is, the more facts there are, the more they bombard the brain, the more

comfort and ease there is in ignoring them or slotting them into predetermined categories,

labels, stereotypes, and stories.

We are left in a very troubling position. The best available data and analyses from

independent scholars suggest that income and wealth inequality levels in India are very high

and increasing. If properly measured, they are the highest in the world or close to it. Yet, in

official and quasi-official expert circles there is a strong tendency to deny this reality by

pointing at other things that seem relevant but actually are not—such as, the low and steady

expenditure inequality, declining poverty, and, worst of all, opinion polls. The existence of

comforting information, especially on expenditure inequality, provides some plausible

deniability about the truth about inequality in India. That deniability is strengthened by the

failure of expert discourse on inequality to produce usable information for the general

population. This failure serves the purpose of all political parties, which, in theory, should

represent the interests of all sections of Indian society, including its marginalized groups.

But, these political agents do not have much use for inequality information either. All the

while, the truth about inequality in India is disagreeable and getting worse. Ignorance about

it—real or feigned—benefits everyone.

This book began by identifying two key features of India’s existential debate. The

first is the struggle over social identity—heterogeneity vs. homogeneity, complexity vs.

simplification. The second is about material reality, which, I argue, is best understood

through the concept of inequality. Whether or not Indian society is heterogeneous or

homogenous is best understood not by making unverifiable claims about religion and identity

but examining the evidence. Are there gaps in opportunity and achievement between India’s

social groups and among the citizenry in general? How big are they? Have they been

growing or closing in recent decades? The answers to these questions speak more clearly to

the issue of heterogeneity vs. homogeneity than bombastic claims by politicians. We have

seen the best available answers to these questions in this chapter, and they should be deeply

worrisome to most people. But, what may be even more worrisome is the manner in which

this vital information is received—with ignorance, obfuscation, or denial. As a result, India

has entered the information age—and its politics of polarization—without much information

on a fundamental feature of politics: its inequalities.

INEQUALITY DATA

Measuring Inequality

Inequality is another word for disparity or unevenness. It is a multidimensional phenomenon.

Scholars study inequality of income, wealth, education, health, access, assets, housing, and other

variables. Inequality is also conceptualized in several distinct ways in the different disciplines

that study the issue. In economics, the focus is often on studying distributions in whole

populations; in sociology and anthropology, the focus is on studying groups and their

differences; in geography, the focus is on differences between spatial units (like nations, states,

cities etc.). We understand inequality by measuring outcomes on dimensions or variables that

matter. Inequality measurement is a vibrant and active sub-field in economics, as is, in

sociology, the measurement of sociological conceptualizations of inequality (such as segregation,

isolation, etc.). There are hundreds of measures of inequality. However, only a handful of

measures are used in practice; as a result, the choice is not as difficult as it could be.

Among the multiple dimensions along which inequality is studied—income, wealth,

assets, educational attainment, health outcomes (longevity, infant mortality, maternal mortality,

etc.)—the primary focus here is on income with a secondary focus on wealth. Income has a

direct relationship to welfare and opportunity and as a result it is doubtless the most commonly

studied variable among inequality researchers. Wealth is also important, but is generally much

more difficult to measure because the wealthy have many ways to hide and obfuscate their

holdings. Some analysts—especially those associated with the Human Development approach—

argue that the focus on income takes attention away from other important markers of welfare,

such as education and health.41 I do not dispute that education and health are very important, but

suggest that income is most important because it is the primary determinant of education and

health outcomes and it is income inequality (along with government failures to provide adequate

public goods) that leads to inequalities in education and health. I do provide some information

on education later in this appendix, but as I show there, these figures probably hide as much as

they reveal. In fact, the clinching argument in favor of focusing on income is that so little is

known about it despite its overwhelming significance. Chapter 4 has been written precisely

because so little is known about the different inequalities of income.

Economists tend to analyze the world in terms of the individual (person, firm, or

institution), whereas other social scientists, especially from sociology and anthropology,

typically think in terms of groups. Mark Granovetter, a prominent sociologist, suggested that

economics as a discipline is “undersocialized” whereas sociology is “oversocialized.”42 A

nation, in the economic framework, is a collection of individuals, each one serving his own

individual interest. Their social identities or spatial locations do not matter in this “abstract”

form of inequality. But in the other social sciences, the most important unit of analysis is usually

not the individual but the group or location. As a result, the social world is understood through

the concepts of in-group cooperation and out-group derogation or conflict (see Appendix 2).

Depending on the context, group identity and interest can either be less important or significantly

more important than individual identity and interest. Consider, for example, the contrasting self-

and group-interests of Wall Street bankers (“greed is good”) vs. soldiers (“band of brothers”) or

Bollywood stars vs. the builders of the sets on which they frolic. Let us think of group identity

in terms of social identity. It is fair to say that investigations of social identity and inequality

form the core of the field of contemporary sociology.

For example, if a society is composed of two groups—black and white, or Forward caste

and Backward caste, or Hindu and Muslim—the only way to understand whether they differ as

groups is to measure things that say something about the quality of their lives and see whether

there is any difference, and, if there is, how much it is. In other words, the extent of division in

any social system is understood by measuring or quantifying the extent of difference or

inequality between the divisions. The difference should be over something that matters. To say

that black (or Backward caste) has darker skin pigmentation than white (or Forward caste) is

beside the point. The question is, whether meaningful outcomes for the group called black (or

Backward caste) are measurably different from the group labeled white (or Forward caste) on

scales that most reasonable people can agree on? There are intricacies of measurement and

making meaning from measurements—and some of those are discussed below—but the basic

point must be clear. If social divisions are real, then at some level they are measureable. They

will show up as differences in things like income, wealth, assets, longevity, infant mortality,

years of education, and so on. The extent of difference is social inequality.

In economics, the primary area of interest is in the distribution of income and the

distribution of human capital (simply: education); wealth distribution is also studied, but to a

lesser extent, because it is harder to track and crack. Some of the most important contributions

to our understanding of income and human capital inequality have come from notable

economists like Anthony Atkinson, Gary Becker, Ronald Bénabou, Gary Fields, Branco

Milanovic, Thomas Piketty, and Amartya Sen, who have discussed ways of measuring

inequalities in income distributions, the ideology and ethics of different distributions and their

measurements, and the meanings and consequences of such inequalities for growth and

economic development.43

The key question economists ask is how income (or wealth or education) is distributed in

a population? An useful visual analog for an income distribution was provided by Jan Pen (that

is described in Chapter 4).44 He imagined a parade in which every individual in a society walks

in order of his or her income and where their heights are proportional to their incomes. This

Pen’s Parade traces a curve of income distribution, from the lowest-income microscopic people

(who have to walk on their hands because they have negative incomes) to giants with their heads

soaring above the clouds. There is much interest among economists in calculating the properties

of the curve traced by this parade and to create summary measures—simple measures—that

capture in a single number a sense of the inequality in a distribution.45

Often these calculations are done by grouping the population into equal sizes—for

example, broken into 10 segments of 10 percent of the population each (called deciles) or five

segments of 20 percent of the population each (called quintiles) ranked by income. How much

of the national income does the poorest decile earn? How much does the richest quintile earn?

What is the ratio of the income share of the richest (decile or quintile) to the poorest? What is

the income share of the superrich—the top one percent, or the top one percent of the top one

percent?

There are two alternative approaches in comparing different income distributions—

whether to include the complete distribution (including all income earners) or simply compare

the top and the bottom of the distribution. The former approach accounts for everyone whereas

the latter approach is useful for investigating changes at the extremes of a given distribution.

Using the latter suggests that the investigator is interested in studying income polarization. If the

full distribution is to be used, certain measurement properties are considered desirable.

Discussions on these desirable properties are available at many sources, the most accessible of

which is on the World Bank’s website.46 In general there are five key axioms or principles that

inequality measures should follow: The Pigou-Dalton transfer principle, the axiom of income

scale independence, the principle of population, the axiom of anonymity, and the principle of

decomposability.

These axioms are not, however, value-free. Consider the second axiom of income scale

independence: that if every individual’s income increases by the same proportion (say everyone

receives a five percent increase in income), a proper inequality measure should not change.

However, since the rich will receive higher absolute income increases than the poor, this is at

best a status quo condition that can also be considered regressive. If we believe the utilitarian

argument that each successive marginal income increase produces less utility or welfare (since

the first lakh rupee one earns is valued more highly than say the tenth lakh), then an equal

proportional increase in all incomes produces less overall utility or welfare than when the same

total income increase is distributed more heavily among the lower income groups. Partly in

response to such normative anomalies in supposedly value-free inequality measures, a group of

explicitly normative or welfarist measures have been created. Among inequality scholars,

Anthony Atkinson’s measure based on explicit choices of “inequality aversion” is well known.47

In keeping with the spirit of this book, we will avoid these complicated measures.

Instead, for a summary measure, we will use only the Gini Coefficient. It is an useful visual

analog of both the Pen’s Parade and the distribution of income by groups like deciles or

quintiles. There is much information on the Gini Coefficient on the net. The Wikipedia page is

as useful as any.48 As used in this Appendix, the Gini can take a value between 0 and 100.

When 0, everyone has the same income (or wealth, or whatever is being measured); when 100,

one rich person has everything. Wherever possible, I will use information that is even simpler

than the Gini.

Expenditure, Income, and Wealth Inequality

Expenditure

Table A3.1 lists the Gini Index estimates of inequality of expenditure or consumption in rural,

urban, and all India from three sources (Himanshu, Subramanian and Jayaraj, and NSSO) for the last

four decades.49 The estimates are not identical because different analysts tend to use slightly

different assumptions and methods for calculating the Gini Index; but the underlying data for all

three sets of estimates are the same: all were collected by the NSSO. Let us not focus on the minor

differences between the different estimates (which are meaningless), nor the more important finding

that expenditure inequality in urban India is consistently higher than in rural India (it is not

particularly meaningful because the phenomenon of higher urban than rural inequality is seen all

over the world).

Let us focus instead on the magnitude of inequality and its consistency. The magnitude of

Gini inequality in rural India is seen to be in the high 20’s and it appears to be more or less

unchanged in four decades. The magnitude of Gini inequality is roughly 35-36 for urban and all-

India, with a possible small uptick from the low 30’s after the early-2000’s. If these figures were

true, that is, if they represented the reality of distribution, then inequality in India would be among

the lowest in the developing world and among the most stable and unchanging.

Income

But, as explained in Chapter 4, there are few serious analysts of inequality who would consider the

NSSO expenditure data and the Gini Indexes calculated from them to represent the reality of

inequality in India. Consider what we know from one of the most important alternative sources of

large scale survey data—the India Human Development Survey (IHDS)—that is also the one major

“unofficial” but reliable source of income inequality data in India. The IHDS is a nationally

representative survey of about 41-43K households that has been carried out in two rounds so far: in

2004-5 and 2011-12.50 The IHDS calculations show that income inequality is considerably higher

than expenditure inequality: in the range of Gini 54 in 2004-5 and 2011-2.51 These results bolster

the innovative findings of Luke Chancel and Thomas Piketty who combine household survey data

(from NSSO and IHDS), national accounts statistics, and tax data to argue that income inequality in

India is very high, perhaps the highest it has ever been, primarily because the share of national

income accruing to the top one percent of income earners is 22 percent of the total income, the

Table A3.1: Expenditure Inequality in India, 1973-74—2011-12

Source: Himanshu Source: Subramanian & Jayaraj

Source: NSSO

Rural Gini

Urban Gini

All-India Gini

Rural Gini

Urban Gini Rural Gini

Urban Gini

1970-71 28.9 34.71972-73 30.7 34.51973-74 28.1 30.21977-78 34.2 34.8 33.6 34.51983 27.1 31.4 29.8 31.6 33.9 29.7 32.51987-88 30.2 35.71993-94 25.8 31.9 30 28.6 34.4 28.2 34.01999-2000 26.3 34.7 26.0 34.22004-05 28.1 36.4 34.7 30.5 37.6 26.6 34.82009-10 28.4 38.1 35.8 29.9 39.3 27.6 37.12011-12 28.7 37.7 35.9 28.0 36.7

highest level in a century, far above the 6 percent it was in the early 1980’s; a visual representation

of Chancel and Piketty’s findings is shown later in this chapter in Figure A3.2.52

Along with two colleagues (S. Chandrasekhar and Karthikeya Naraparaju), I have studied

some aspects of income distribution over the last decade in rural India.53 We analysed the Situation

Assessment Surveys of Farmers/Agricultural Households undertaken by the NSSO in 2003 and

2013. We found that there was a very large difference between the two measurement concepts—

income vs. expenditure inequality—where the Gini Indexes of per capita income and expenditure

were around 60 and 30 respectively during this study period. We argue that while our findings are

narrow in coverage (being limited to the agricultural sector, that covers roughly half the population)

that narrowness itself leads to greater robustness. Therefore, the startling gap of 30 Gini points

between expenditure and income inequality should be taken seriously. Added to the findings of

IHDS and Piketty and his associates, these findings should conclusively burst the mythical balloon

of low inequality in India.

In fact, I argue that the true level of income inequality in India is higher than anything

calculated by any analyst so far. There are several reasons for taking this position. First, most

inequality calculations are unlikely to include the very top and bottom ends of the income

distribution. For example, our own work on rural India misses the population that has little or no

income from agricultural activities; much of this group is likely to be the landless population that

may comprise more than 40 percent of rural households.54

The far bigger problem is that most income data derived from surveys are likely to miss or

have unreliable figures on the very top end of the income distribution. The Indian upper middle

class is notoriously difficult to survey. Even if a survey team can make it to their doors (which is

very hard to do in the gated housing estates in which the upper middle class tends to live), it is

usually refused entry. The upper class is, of course, well and truly beyond questioning by anyone.

For example, in the IHDS 2004-5 survey, the individual with the highest income out of the 41,000

plus families surveyed earned less than Rs. 22 lakh per year. It seems obvious that the IHDS survey

missed the top one percent of earners. Even more troubling are the NSSO expenditure surveys. For

the 2011-2 round, their highest spending group, the top five percent of urban India, averaged

expenditures of merely Rs. 1.2 lakh per year. This is roughly what government college professors

earn per month. I have no doubt that the NSSO also missed the top one percent (perhaps the top 2-3

percent) of consumers. On top of this is the well-known tendency of the poor to over-report and the

rich to under-report their incomes.55

These problems with survey-based inequality calculations are beginning to become widely

recognized. Laurence Chandy and Brina Seidel write: “Missing top incomes in household surveys is

a long established problem in both developed and developing economies…The more new

information we uncover about top incomes, the less faith we have in traditional survey-based

inequality measures, and the less knowledge we can claim to have about the distribution of income

across an economy’s entire population.” They “use the missing income between surveys and

national accounts as a proxy for missing top incomes in surveys” following a method suggested by

Christoph Lakner and Branco Milanovic.56 The new calculations of Chandy and Seidel show large

increases in Gini Indexes for several countries—the average increase is from 39 to 48. One of the

largest increases is for India, where the Gini goes from 36 (calculated from official expenditure data)

to 56 for the early 2010’s.

That too may be an underestimate. If the Gini Index of agricultural income alone is 60 (as

my work with Chandrasekhar and Naraparaju has shown), there is almost no doubt that the Gini

index is significantly higher at the national scale. There are two reasons to justify this claim. First,

we know that urban inequality is higher than rural inequality by 5-8 Gini points even using the

flawed NSSO expenditure data. The gap between urban and rural inequality is likely to be higher

with income data. Second, we know that average urban incomes are at least twice as high as

average rural incomes for every size subgroup (decile or quintile) of the population.57 Hence, if we

add the two distributions—rural and urban—and it is possible to assess the income of the top one to

two percent and bottom decile of households with any reasonable accuracy, a strong argument can

be made that income inequality in India is among the most extreme in the world. It would not be a

surprise if the true level of income inequality in India was in the range of Gini 65, on par with or

higher than the highest known level of inequality in South Africa.

Wealth

The recently published figures on wealth inequality in India strongly suggest that the worst-case

scenarios may indeed be true. There have been a spate of such publications in recent years, spurred

by the annual Global Wealth Reports produced by Credit Suisse beginning in 2010. The tone of the

Credit Suisse reports is largely celebratory, but the U.K.-based NGO Oxfam produces an annual

Global Inequality report (based on the same wealth data) whose tone is anything but. For example,

Oxfam’s 2017 report argued that the richest eight billionaires in the world (Bill Gates, Amancio

Ortega, Warren Buffett, Carlos Slim Helú, Jeff Bezos, Mark Zuckerberg, Larry Ellison, and Michael

Bloomberg) had as much wealth between themselves as the poorest 50 percent of the world’s

population put together, and that the richest one percent of the world had as much wealth as the

remaining 99 percent. The situation was “beyond grotesque,” the Oxfam report said. For India, the

Credit Suisse report stated that the richest 10 percent possessed 73 percent of the nation’s wealth,

whereas Oxfam stated that 73 percent of the wealth generated in 2016-7 in India went to just the

richest one percent. According to the latest available Credit Suisse report, the Gini Index of wealth

inequality in India is 83, among their list of the highest in the world.58

It is obvious that neither Credit Suisse nor Oxfam has the resources or ability to actually

study wealth in India by themselves…and they do not. The primary data source for both is the

decennial All India Debt and Investment Survey (AIDIS) carried out by the NSSO (last undertaken

in 2012-3). One should be hesitant to rely on distant sources like Credit Suisse and Oxfam which

may be tweaking the raw data from NSSO in ways that are not visible to observers (which would

seem to be the case if new findings are generated every year though no new AIDIS data are

available). Their audience is global whereas we need to stay closer to the ground. Therefore, it may

be better to look at the findings of scholars who have looked at the AIDIS data directly and

carefully. The most recent of these is a paper by Ishan Anand and Anjana Thampi, in which the

Gini Index of assets and net worth are shown to be 74 and 75 respectively in 2012, having risen

from 65 and 66 in 1991 (and about the same levels in 2002).59

Note that the AIDIS data itself is open to serious criticism. It suffers from some of the main

problems of the NSSO expenditure surveys; most notably, the difficulties with getting good data on

the top of the distribution. For example, in the period that the Indian stock market boomed (the last

decade), the NSSO data show that the weight of shares/stocks actually went down to 0.13 percent of

total wealth in its survey sample. That is simply not credible. Given that the market capitalization

of all stocks on the BSE had almost equaled the country’s gross domestic product in early 2018 (Rs.

135 trillion in stocks compared to Rs. 150 trillion in GDP), the AIDIS sample clearly has missed

almost all of India’s upper middle class, and, of course, the entire upper class.

Figure A3.2: Change in Expenditure, Income, and Wealth Inequality over Time

Early 1990's Early 2000's Early 2010's20

30

40

50

60

70

80

A. Expenditure, Income, and Wealth Inequality

Expenditure (NSSO) Income (IHDS)Income (Chandy & Seidel) Wealth (NSSO)

Gini

Inde

x

19511954

19571960

19631966

19691972

19751978

19811984

19871990

19931996

19992002

20052008

20112014

0

5

10

15

20

25

B. Income Shares of the Top 1 % and Bottom 50%, 1951-2014

Top 1% Bottom 50%

Shar

e of

Nati

onal

Inco

me,

%

Sources: A. As shown in figure; B. Calculated from data in Luke Chancel and Thomas Piketty, 2017, Indian Income Inequality, 1922-2014.

In addition, the AIDIS has an unusual finding: more than 90 percent of India’s wealth is

shown to be in land and buildings. About 70 percent of rural wealth and a little under half of urban

wealth is shown to be in land alone. As a result, much of the calculations (of wealth and inequality)

depend on how accurately land is valued. There is serious case to be made that it is generally

undervalued, especially given the five-fold increase in land prices across the country in the period

2000-2013, and is evidenced by the AIDIS calculation that in urban areas the value of land is

roughly the same as the value of buildings. That too is simply not credible. Depending on the city

and location, the value of land in total property is much above 50 percent, and for the upper class it

easily surpasses 95 percent.60 Therefore, it is very likely that the very high levels of wealth

inequality calculated from AIDIS data are nonetheless significant underestimates because the survey

was unable to capture the two main sources of wealth for the Indian upper middle and upper classes

—stocks and land.

That possibility is highlighted by the findings of Chancel and Piketty in the second part of

Figure A3.2, that show the long-term trajectories of income earned by the top one percent and the

bottom 50 percent of families. If correct, this should be a severe indictment of, if nothing else, the

absence of a serious discourse on inequality by government after government (more on this in

Chapter 4).

Social Inequalities

As discussed earlier in this Appendix (and Chapter 4), social inequality is conceived, in a

sociological sense, as the average difference between social groups. In our case, the social groups

under consideration are those that have been identified as marginalized from pre-independence India

(Scheduled Castes and Scheduled Tribes, to be called Dalit and Adivasi in the remainder of this

discussion), new groups that have been brought into consideration for reservation or affirmative

action after the Mandal Commission recommendations (Other Backward Classes), and others (who,

depending on the data available, may be called “Forward Castes” or “Brahmans” plus “Other

Forward Castes”). The explicit assumption of the Indian system of reservations is that there are

sizable gaps between the Backward and Forward groups, and the explicit goal of the reservations is

to narrow those gaps. So the question before us is: What do we know about: (a) how far apart these

groups are from each other, and (b) whether the gaps between them now are narrower or wider than

before?

These are questions of fact and can only be answered with data. As we have seen above, the

official data-gathering system in India does not collect some critical information for anyone (that is,

income) and what it does collect surely does not include households at the top, which are very likely

to be dominated by the Forward groups. We do not even know how much of the “top” is missing in

surveys; the top one percent almost certainly, and perhaps as much as the top 2-3 percent. As a

result, we know little about their income or wealth (that is, their land and stocks). Perhaps just as

crucially, we do not know their social identities either (that is, what religion or caste they belong to).

So, if our goal is to measure the gap from the “low” groups to the “top” groups, it is necessary to

recognize from the very outset that it cannot be done, at least not easily, and not without violating

some privacy barriers (such as those that protect the identities of tax payers from public scrutiny).

It is possible to tease out some indicators of what may have been happening to social

inequality using the available official data from NSSO and unofficial data from IHDS. Several of

the scholars who have been cited above have included sections on caste inequality in their larger

studies of inequality. These studies are not uniform because they all use different formulations of

social groups: in some studies, Dalits (Scheduled Castes) and Adivasis (Scheduled Tribes) are

combined; some studies separate out Other Backward Classes or Forward castes or Brahmans, some

do not; some allow the identification of religious identity; most do not. Therefore, these studies are

not comparable. They do not have the same definitions of low, middle, and high, do not use the

same data over the same time period, and, above all, are flawed for the same reasons that all these

studies are flawed—they measure inequality without knowing much about the top of the

distribution. It is not the researchers’ fault. They have to work with what they have, and what they

have is flawed.

These indicative figures on social inequality are combined in a set of graphics in Figure A3.3

that provide some information on expenditure, income, wealth, and education over some period of

time. This allows us to see the extent of the gaps between Forward and Backward groups and the

changes in the conditions and their trajectories over recent decades. Note that if we had access to

information on the uppermost section, these gaps would likely have been larger, and, crucially,

growing over time.

As it is, the data show that the gaps between the averages of the Forward and Backward

groups are considerable. Moreover, they have been growing over time for all the variables for

which comparable temporal data are available. Consider expenditure (in Figure A3.3a-A), which we

know is the primary welfare information collected by the NSSO and have seen earlier is the least

meaningful marker of quality of life as far as inequality is considered. The highest-spending group

is, as expected, the urban non-Dalit non-Adivasi population and the lowest-spending is the rural

Dalit and Adivasi population. The ratio of their expenditures has increased from 1.9 to 2.3 from

1983 to 2010. That is, the average urban non-Dalit non-Adivasi person spent almost twice as much

as the average rural Dalit or Adivasi in 1983; a quarter century later the former spent about 2.3 times

as much as the latter. All the other gaps on expenditure widened during the period: between the

rural Backward and the rural majority and between the urban Backward and the urban majority.

There is unambiguous evidence of a large and growing gap in expenditure between the socially

marginalized and the rest of the population.

Figure A3.3a: Expenditure and Income for Social Groups

1983 20100

50

100

150

200

250

300

350

A. Expenditure per capita

Rural: SC & STRural: OthersUrban: SC & STUrban: Others

Aver

age

Expe

nditu

re p

er m

onth

2003 2013600

800

1,000

1,200

1,400

1,600

1,800

2,000

B. Income in the Agriculture Sector

Scheduled Caste

Scheduled Tribe

Other Backward Classes

All groups

Others

Aver

age

Inco

me

per m

onth

Brahmans High Caste OBC Scheduled Caste

Adivasi Muslim All Groups0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

90,000

100,000

C. Annual Household Income

Rs.

Figure A3.3b: Wealth by Social Group

1991 2002 20120

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

A. Wealth Distribution, 1991-2012

Scheduled TribeScheduled CasteOther Backward ClassesGeneral

Ratio

of W

ealth

Sha

re/P

opul

ation

Sha

re

Hinduism Islam Christianity Sikhism Jainism Buddhism0

1

2

3

4

5

6

7

8

B. Assets by Religion

2002 2012

Asse

t Sha

re /

Pop

ulati

on S

hare

Figure A3.3c: Education by Social Group

Brahmans High Caste OBC Scheduled Caste

Scheduled Tribe

Muslim All Groups0

2

4

6

8

10

12

14

16

0

5

10

15

20

25

30

35

40

45

A. Educational Attainment, 2004-5

Years of education College graduate "Matric"

Year

s of E

duca

tion;

Col

lege

Gra

duat

e, %

"Mat

ric",

%

Muslim Scheduled Tribe Scheduled Caste All groups All groups minus minorities

0

0.5

1

1.5

2

2.5

3

3.5

B. Share of Total Population in College, 2015-6

Shar

e, %

The income data from the agriculture sector (also from NSSO surveys) are more fine-grained

and show somewhat better outcomes for the marginalized. Here, it is possible to differentiate

between Dalit and Adivasi incomes, and between OBC and everyone else. Again we see large gaps

between the non-marginalized “Others” and the Backward groups, but the gap is smaller than for

expenditure (above). We see a growing gap between Dalit income and “Others”, but the gap

between “Others” income and both Adivasi income and OBC income, though large, narrowed

between 2003 and 2013.

The income data from IHDS shown in Figure A3.3a-C are available for a single year, and

they show, again, large gaps between Forward and Backward group incomes. Brahman average

incomes (identifiable only in IHDS data) are twice as large as average Dalit and Adivasi incomes.

The average incomes of OBC and Muslim families is about 20 to 30 percent higher than Dalit and

Adivasi incomes.

Because the IHDS surveyed the same set of households at two time periods (2004-5 and

2011-2), it has become possible to analyze change—or income mobility—at the household-level (in

addition to the usual population-level). The surveys cover a short period (7 years), but that was also

a time of great economic change. The findings in a study by Ranganathan, Tripathi, and Pandey are

generally negative.61 Forward castes are of course heavily represented in the top income group (two

to three times more heavily than their population weight) and Backward castes least represented.

The progress of Forward castes up the income ladder is also the most rapid. There is income growth

among the Dalit and Adivasi households too; close to one-third experienced upward mobility. But

among all social groups studied in the paper, Dalits had the least upward mobility (30 percent of

families) and most downward mobility (41 percent of families). Using the same IHDS data sets,

Iversen, Krishna, and Sen find that there is “higher occupational mobility among forward castes than

among SCs and STs…[and] a much higher prevalence of sharp descents among SC and ST sons.”62

The wealth scenario (in Figure A3.3b) is even more stark and appears to have deteriorated

sharply in the last decade. The Dalit and Adivasi share of national wealth had each been roughly

half their population share till the early 2000’s but dropped to 40 percent in the last debt and

investment survey of the NSSO. The wealth share of the OBC also dropped from 90 to 80 percent

of population share in the same time. In contrast, the wealth of the general (non-marginalized)

population was 20 percent above its population share in 1991 and almost 90 percent above in 2012.

The general (non-Dalit non-Adivasi) population’s wealth per capita in 2012 was almost five-fold

higher than that of the Backward population. The wealth gap between the Backward and non-

marginalized populations was large to begin with and had roughly doubled in two decades.

It is important to remember that almost much of this “wealth” is notional rather than real; it

is derived from land ownership and the assumed value of land. Hence, it is possible that what these

data really reveal are differences between where people live—the marginalized on marginal/remote

and less valuable land, the non-marginalized on more urban and generally more valuable land. It is

also possible that since the NSSO has seriously undervalued urban land and has almost no account

of the stock market, the wealth gap (notional or real) between the marginalized and non-

marginalized is considerably higher than five-fold.

The graphic on the distribution of assets by religion shows the affluence of the small

minorities (Jains, Sikhs, Christians) and the poverty of the large minority (Muslims). Not only do

the small minorities have significantly greater assets than average, but their shares grew over the

decade 2002-12. Jains, already the wealthiest religious group by far, saw their asset share more than

double in a decade, during the same time that Muslims saw their asset share shrink measurably.

Finally, we look at some information on educational attainment by social group. The

reasoning is simple. Education is “the hard core of the ‘hard core’ of human capital.”63 It is the key

to income generation, intergenerational mobility, and social status, not to mention citizenship and

awareness of self and rights. If the educational gaps between social groups do not close, the

material gaps between them will not either. Education and educational inequality in India are big

subjects. Contributions to it range from articles in well-known journals like Nature to technical

expert analyses. This minimal discussion here does no more than scratch the surface of a deep

problem in which the major issues include access, quality, and cost (by social identity, location, and

income class).

There is general agreement that some aspects of educational inequality have improved in the

preceding decades. Notably, there have been big gains in literacy and school attendance among the

young (including girl children) in all segments of society, and a general surge in college attendance

(which nonetheless remains biased toward Forward castes). At the same time, many analysts

recognize that the education market has become increasingly segmented, which means there are

significant differences in quality (all “literates” are not the same, and neither are all college degrees)

and that the Backward continue to fall behind in quality (even if they are catching up in quantity,

having started from a very low base).

The graphics in Figure A3.3c highlight the significant differences in educational attainment

by social identity among the adult population in India. There are vast differences between the most

educated group (Brahmans) and the least educated (Adivasis): the former have twice as many years

of education, are four-fold as likely to matriculate from school, and seven-fold more likely to hold a

college degree. Dalits and Muslims are also very far behind Brahmans and other “high caste”

groups. A snapshot of current college enrollees shows that, while some gaps may be closing, very

large differences remain between social groups. Adivasis are still half as likely to be in college as

non-marginalized groups, and the Muslim population is far behind, only one-fourth as likely to be in

college as the non-marginalized Hindu groups.

These differences in averages exist all along the economic and social spectrum. A recent

article in The Economist graphed the gaps between India’s social groups on poverty and

malnourishment. They called the gaps “unconscionable.” In 2010-11, the poverty rates for

Forward, OBC, Dalit, and Adivasi groups were respectively 12.5 percent, 20.7 percent, 29.4 percent,

and 43 percent. Rural poverty was three- and two-times higher in the Adivasi and Dalit populations

respectively compared to non-marginalized groups. Urban poverty was about three-times higher for

both.64 Malnutrition was almost twice as high for Adivasis compared to “upper” castes, and in the

1990’s, had declined more slowly; that is, the gap was growing larger. In an innovative new paper,

Diane Coffey, Payal Hathi, Nidhi Khurana, and Amit Thorat document, among other issues, the

extent of prejudice against Dalits—more than half their rural survey respondents (in Rajasthan and

Uttar Pradesh) practiced untouchability and were in favor of having laws banning inter-caste

marriages.65 The numbers speak for themselves. No editorial commentary is needed.

1

CHAPTER 4

Sanjoy Chakravorty, 2006, Fragments of Inequality: Social, Spatial, and Evolutionary Analyses of

Income Distribution, New York: Routledge, p. 12. 2 Ramnarayan S. Rawat and K. Satyanarayana, 2016, Introduction: Dalit Studies, p. 3. Gopal

Guru’s essay in the same volume is titled: The Indian Nation in its Egalitarian Conception, pp. 31-

52. Also see Gopal Guru (Editor), 2011, Humiliation: Claims and Context.3 There are estimates of the numbers of rich and poor Brahmins (but not their incomes) from the

Center for the Study of Developing Societies (CSDS) based on their own surveys. These are not

official data, of course, and their reliability is unclear. See

https://www.outlookindia.com/magazine/story/brahmins-in-india/234783. 4 This has not stopped the emergence of a new subfield of “happiness research,” which should not

be surprising given that there is a much longer history of measuring the opposite of happiness—

depression—which is just as impossible to measure. 5 Jan Pen, 1971, Income Distribution, Hammondsworth: Allen Lane.6 Anthony B. Atkinson, 1983, The Economics of Inequality. Second edition. Oxford: Clarendon

Press, p. 14-15.7 For instance, in the US, where the white vs. black gaps in wealth, income, and education are quite

significant, the contribution of these differences in averages to total inequality is significantly less

than the contributions of within-group (within-black and within-white) inequality. See Sanjoy

Chakravorty, 1996, Urban Inequality Revisited: The Determinants of Income Distribution in U.S.

Metropolitan Areas, Urban Affairs Review 31: 759-777. One of the interesting findings of that

paper, that is now found more generally, is that inequality within the black population was higher

than within the white population. For a recent example see Frédéric Chantreuil and Thérèse

Rebière, 2016, Decomposition of Income Inequality by Attributes: Does the Race Matter in the US?

Mimeo. Available at https://conf-tepp2016.sciencesconf.org/99419/document. 8 World Bank, 2009, Reshaping Economic Geography, World Development Report 2009,

Washington DC: The World Bank, p. 1. 9 The correct term for state-level “income” is Net State Domestic Product (NSDP) per capita. This

is strictly not income, but since there is nothing better to use, it is considered close enough.

Individual income is officially not measured in India, which is a key theme of this chapter.10 For data on state-level indicators, see Sripad Motiram and Vamsi Vakulabharanam, 2011, Poverty

and Inequality in the Age of Economic Liberalization, India Development Report, Ed. D. M.

Machane, New Delhi: Oxford University Press, pp. 59-69; and Sanjoy Chakravorty, 2012. Regional

Development in India: Paradigms Lost in a Period of Great Change. Eurasian Geography and

Economics 53: 21–43. The most recent poverty data are from the Reserve Bank of India: Table

162, Number and Percentage of Population Below Poverty Line. Reserve Bank of India,

Government of India. 2013. Available at

https://en.wikipedia.org/wiki/Indian_states_and_territories_ranked_by_poverty#cite_note-1. 11 Harish Damodaran, 2015, District Zero Nabarangpur: Why This is the Heart of a Changing India,

http://indianexpress.com/article/india/india-others/district-zero-nabarangpur-why-this-is-the-heart-

of-india-changing/. 12 Planning Commission, 2005, Report of the Inter-Ministry Task Group on Redressing Growing

Regional Imbalances, New Delhi, India: Planning Commission, 2005, available at

http://planningcommission.nic.in/aboutus/taskforce/inter/inter_reg.pdf. Bibek Debroy and Laveesh

Bhandari, 2003, District-Level Deprivation in the New Millennium. New Delhi, India: Konark

Publishers, 2003. Jyotsna Jalan and Martin Ravallion, 1997, Spatial Poverty Traps? Washington,

DC: World Bank, Development Research Group Working Paper No. 1862. Sanjoy Chakravorty,

2000, How Does Structural Reform Affect Regional Development? Resolving Contradictory

Theory with Evidence from India, Economic Geography 76, 4:367–394. Sanjoy Chakravorty and

Somik Lall, 2007, Made in India: The Economic Geography and Political Economy of

Industrialization. New Delhi: Oxford University Press.13 Haryana data reported in http://www.hindustantimes.com/chandigarh/haryana-s-per-capita-

income-tops-charts-thanks-to-gurgaon-5-other-districts/story-4Bqn4HWeHnDul0bw1ewtQK.html.

Odisha data available from the state’s annual Economic Survey. For perspective: in the US, the

counties (which are approximate equivalents of Indian districts) with the highest per capita incomes

(New York County, which is Manhattan, and counties like Arlington and Fairfax in Virginia, in the

suburbs of Washington DC) were about seven-fold greater than the lowest income counties (like

Oglala Lakota in South Dakota and Wheeler in Georgia). The highest levels were around USD

62,000 per capita, and the lowest around USD 9,000 per capita. All these figures are from the

American Community Survey (ACS) of 2009-13 carried out by the U.S. Census Bureau.14 Sonalde Desai, Amaresh Dubey, Brij L. Joshi, Mitali Sen, Abusaleh Shariff, and Reeve

Vanneman, 2010, Human Development in India: Challenges for a Society in Transition, New Delhi:

Oxford University Press, p. 5.15 A few weeks before I finished writing this book, the NITI Aayog (the body that replaced the

Planning Commission) came out with a list of 101 “aspirational districts” that had been called

“backward” a few months earlier. One wonders whether the government will have the chutzpah to

rename “Other Backward Classes” to “Other Aspirational Classes.”16 These countries are said to suffer from a “resource curse”, which is the counterintuitive effect in

which being “blessed” with rich natural resources often turns into a “curse” for the very people in

whose lands those resources are located. It is not hard to see how the “resource curse” thesis can

easily be applied to the India’s poorest regions (southern Bihar, Jharkhand, Chhattisgarh, and

Odisha) which also happen to be its most resource-rich.17 Branco Milanovic, 2016, The question of India’s inequality, http://glineq.blogspot.in/2016/05/the-

question-of-indias-inequality.html. 18 The NSSO was created in 1950 (initially it was called NSS). It has so far conducted 71 rounds of

surveys (for which the data are available). Its major surveys in a ten year cycle include: Consumer

Expenditure and Employment & Unemployment (twice); Social Consumption (health, education

etc.) (twice); Un-organised Manufacturing (twice); Un-organised services (twice); and Land &

Livestock holdings. The NSSO also undertakes special surveys, such as the 70th round titled the

Situation Assessment Survey of Agricultural Households, All India Debt and Investment & Land

and Livestock Holdings (whose findings I will refer to later in this chapter). It also conducts the

Annual Survey of Industries.19 See P. B. Coulter, 1989, Measuring Inequality: A Methodological Handbook, Boulder: Westview

Press; Hongyi Li, Lyn Squire, and Hengfu Zhou, 1998, Explaining International and Inter-temporal

Variation in Income Inequality, The Economic Journal 108: 26-43.20 Among the most influential studies that returned attention to inequality was Hollis B. Chenery

and Moises Syrquin, 1975, Patterns of Development: 1950-1970, New York: Oxford University

Press.21 The use of this method has continued despite a rising awareness of its problems and findings. As

Milanovic points out: from the early 1990s, “the survey numbers began to diverge more and more

from National Accounts statistics: NSS was showing consistently lower rates of growth, and higher

poverty than many people thought it should be given India’s fast growth.” Moreover, it was

increasingly clear that the top earners were not being captured by the surveys, which is probably

why the NSS survey averages were low and did not match with the National Accounts statistics.22 That expenditure inequality in urban India is consistently higher than in rural India is not

particularly meaningful because the phenomenon of higher urban than rural inequality is seen all

over the world.23 World Bank, 2007, World Development Report—Agriculture for Development, The International

Bank for Reconstruction and Development/The World Bank, p. 46. The UN data are available at

http://hdr.undp.org/en/content/income-gini-coefficient. These calculations based on consumption

data continue to be used and conflated with income data on a regular basis. A good example from

the IMF is in Rahul Anand, Volodymyr Tulin, and Naresh Kumar, 2014, India: Defining and

Explaining Inclusive Growth and Poverty Reduction, IMF Working Paper WP/14/63.

24 Mehtabul Azam, 2016, Income Inequality in India 2004-2012: Role of Alternative Income

Sources. Economics Bulletin, 36(2), 1160-69.  Also see Reeve Vanneman and Amaresh Dubey,

2013, Horizontal and Vertical Inequalities in India, in Janet Gornick and Markus Jantti (eds.),

Income Inequality: Economic Disparities and the Middle Class in Affluent Countries, Stanford CA:

Stanford University Press. More papers that use IHDS data are available at https://ihds.umd.edu/. 25 Sanjoy Chakravorty, S. Chandrasekhar, and Karthikeya Naraparaju, 2017, Income Generation

and Inequality in India’s Agricultural Sector: The Consequences of Land Fragmentation. Available

at http://www.iariw.org/India/chandrasekhar.pdf. 26 Luke Chancel and Thomas Piketty, 2017, Indian Income Inequality, 1922-2014: From British Raj

to Billionaire Raj. WID.world Working Paper Series No. 2017/11. This report was widely covered

by the Indian media.27 Laurence Chandy and Brina Seidel, 2017, How Much do we Really Know About Inequality

Within Countries Around The World? Adjusting Gini Coefficients for Missing Top Incomes,

https://www.brookings.edu/opinions/how-much-do-we-really-know-about-inequality-within-

countries-around-the-world/.28 Ishan Anand and Anjana Thampi, 2016, Recent Trends in Wealth Inequality in India, Economic

& Political Weekly, December 10, 2016 51(50):59-67. A longer time series (beginning in 1961-2)

is available in Sreenivasan Subramanian and Dhairiyarayar Jayaraj, 2015, The Evolution of

Consumption and Wealth Inequality in India: A Quantitative Assessment, Journal of Globalization

and Development 4(2): 253–281. The latter are for rural and urban data separately; no national

estimates are presented. Similar analyses are available in Arjun Jayadev, Sripad Motiram and

Vamsi Vakulabharanam, 2007, Patterns of Wealth Disparities in India during the Liberalisation Era,

Economic & Political Weekly 42(38):3853–63.29 Credit Suisse, 2017, Global Wealth Databook 2017, available at http://publications.credit-

suisse.com/index.cfm/publikationen-shop/research-institute/global-wealth-databook-2017-en/. The

adjustments made by Credit Suisse to the AIDIS data are not publicly available, hence it is not

possible to judge how accurate their estimates are and whether all the major problems identified in

Appendix 2 have been accounted for.30 Thiagu Ranganathan, Amarnath Tripathi, and Ghanshyam Pandey, 2017, Income Mobility among

Social Groups, Economic & Political Weekly, 52(41): 73-6.31 Vegard Iversen, Anirudh Krishna, Kunal Sen, 2017, Rags to Riches? Intergenerational

Occupational Mobility in India, Economic & Political Weekly 52 (44): 107-114.32 The sources for the findings on education, poverty, and malnutrition are detailed in Appendix 3. 33 Consider, for instance, the wealth inequality data provided in Ishan Anand and Anjana Thampi,

2016, Recent Trends in Wealth Inequality in India. Their 2012 wealth Gini Indexes for the Adivasi

population are 61 in rural and 76 in urban areas; the corresponding figures for the general (non-

marginalized) population are 70 and 77.34 Montek Singh Ahluwalia, 2011, Prospects and Policy Challenges in the Twelfth Plan, Economic

& Political Weekly, 46 (21):88–105, p. 92-3. This and the next two quotes are all available in

Subramanian and Jayaraj, The Evolution of Consumption and Wealth Inequality in India. The

introduction to their paper is well worth reading.35 Surjit S. Bhalla, 2011, Inclusion and Growth in India: Some Facts, Some Conclusions, LSE Asia

Research Centre Working Paper 39, pp. 12-13, 10.36 Jagdish Bhagwati, 2011, Indian Reforms: Yesterday and Today, in Growth and Poverty: The

Great Debate, P. S. Mehta and B. Chatterjee, eds., Jaipur: Cuts International, p. 8. 37 Montek Ahluwalia is well-versed in the inequality debate. He wrote two of the first papers when

the subject of inequality became important again in the 1970’s (Montek S. Ahluwalia, 1976, Income

Distribution and Development: Some Stylized Facts. American Economic Review 66:128-35;

Montek S. Ahluwalia, 1974, Income Inequality: Some Dimensions of the Problem, in

Redistribution with Growth. H. B. Chenery et al. eds. London: Oxford University Press.) Surjit

Bhalla too has deep knowledge about inequality. He has written an insightful and methodologically

sophisticated book on the subject (Surjit S. Bhalla, 2002, Imagine There's No Country: Poverty

Inequality and Growth in the Era of Globalization, Washington D.C.: Institute for International

Economics). And Professor Bhagwati’s contributions to and advocacy of international trade are

legion.38 Arthur M. Okun, 1975, Equality and Efficiency: The Big Trade Off, Washington, D.C.: The

Brookings Institution. Simon Kuznets, 1955, Economic Growth and Income Inequality, American

Economic Review 45:1-28. Alberto Alesina and Dani Rodrik, 1994, Distributive Politics and

Economic Growth, Quarterly Journal of Economics 108:465-90. Gary A. Fields, 2001,

Distribution and Development: A New Look at the Developing World, Cambridge, Mass.: The MIT

Press. 39 One is more likely to find a discussion on intra-Dalit divisions in say Andhra Pradesh than any

mention of the relative expenditures by caste groups in the same state. See Sambaiah Gundimedha,

2016, Dalit Politics in Contemporary India, New Delhi: Routledge. 40 For accounts of positive social changes for Dalits see Devesh Kapur, Chandra Bhan Prasad, Lant

Pritchett, and D Shyam Babu, 2010, Rethinking Inequality: Dalits in Uttar Pradesh in the Market

Reform Era. Less positive outcomes are seen in both Ghanshyam Shah, Harsh Mander, Sukhdeo

Thorat, Satish Deshpande, and Amita Baviskar, 2006, Untouchability in Rural India, New Delhi:

Sage, and Ira N. Gang, Kunal Sen, and Myeong-Su Yun, 2016, Is Caste Destiny? Occupational

Diversification among Dalits in Rural India, The European Journal of Development Research 29

(2): 476-92.41

APPENDIX 3

The human development approach is detailed in the annual Human Development Report, first

published in 1990. It came about as a result of dissatisfaction among leading economists from the

Indian subcontinent like Mahbub ul Haque and Amartya Sen about the dominating focus on income

among inequality researchers and the inadequacies of income to fully explain the opportunities

available to individuals. There are now several editions of a separate India Development Report

and also multiple development reports at the state level in India.42 Granovetter, M. 1985. Economic Action and Social Structure: The Problem of Embeddedness.

American Journal of Sociology 91:481-510.43 Anthony B. Atkinson, 1970, On the Measurement of Inequality, Journal of Economic Theory

2:244-63; and 1983, The Economics of Inequality. Second edition. Oxford: Clarendon Press. Gary

S., 1962, Investment in Human Capital: A Theoretical Analysis, Journal of Political Economy 70:9-

49. Ronald Bénabou, 1996, Equity and Efficiency in Human Capital Investment: The Local

Connection, Review of Economic Studies 63: 237-264. Gary A. Fields, 1980. Poverty, Inequality,

and Development, Cambridge: Cambridge University Press; and 2001. Distribution and

Development: A New Look at the Developing World. Cambridge, Mass.: The MIT Press. Branco

Milanovic, 1998, Income, Inequality, and Poverty during the Transition from Planned to Market

Economy. Washington DC: World Bank. Thomas Piketty, 2014, Capital in the Twenty-first

Century, trans. Arthur Goldhammer, Cambridge, Mass.: Harvard University Press. Amartya K. Sen,

1973, On Economic Inequality. Oxford: Clarendon Press; and 1992, Inequality Reexamined, New

York and Oxford: Russell Sage Foundation and Clarendon Press.44 Jan Pen, 1971. Income Distribution.

45 Simple measures of inequality often miss some key features that interest inequality researchers.

So, there are also many complex measures of inequality. In fact, a thriving subfield of inequality

research is that of research on inequality measurement. Amartya Sen (1973, On Economic

Inequality), in a single footnote, cited over 100 articles and books on the measurement of inequality

in economics (counting nothing from sociology or geography). 46 See http://www.worldbank.org/poverty/inequal/methods/. Also see P. B. Coulter, 1989,

Measuring Inequality: A Methodological Handbook, Boulder: Westview Press; and Sanjoy

Chakravorty, 2007, Fragments of Inequality.47 Anthony B. Atkinson, 1970, On the Measurement of Inequality.48 https://en.wikipedia.org/wiki/Gini_coefficient.

49 Himanshu, No date, Inequality in India, available at

http://india-seminar.com/2015/672/672_himanshu.htm. Sreenivasan Subramanian and

Dhairiyarayar Jayaraj, 2015, The Evolution of Consumption and Wealth Inequality in India; also

see S. Subramanian and D. Jayaraj, 2015, Growth and Inequality in the Distribution of India’s

Consumption Expenditure, 1983 to 2009-10, WIDER Working Paper 2015/025. NSSO figures

reported in

http://planningcommission.nic.in/data/datatable/data_2312/DatabookDec2014%20106.pdf.50 The IHDS is a joint undertaking by researchers at the University of Maryland and the National

Council of Applied Economic Research (NCAER), New Delhi.51 There are several estimates of Gini using the IHDS data and they all vary slightly based on

assumptions and adjustments made by the specific analyst. These figures are from Mehtabul Azam,

2016, Income Inequality in India 2004-2012.  Also see Reeve Vanneman and Amaresh Dubey,

2013, Horizontal and Vertical Inequalities in India. More papers that use IHDS data are available at

https://ihds.umd.edu/. It is important to note that the IHDS data allow both expenditure and income

inequality to be estimated. The 2005 expenditure Gini estimate from IHDS is around 38, roughly

equivalent to the NSSO survey based estimate of 36 for the same time. In short, the IHDS roughly

captures the same population that NSSO does and is as reliable (or unreliable) as the latter.52 Luke Chancel and Thomas Piketty, 2017, Indian Income Inequality, 1922-2014.53 Sanjoy Chakravorty, S. Chandrasekhar, and Karthikeya Naraparaju, 2017, Income Generation

and Inequality in India’s Agricultural Sector. 54 Vikas Rawal, 2008, Ownership Holdings of Land in Rural India: Putting The Record Straight,

Economic and Political Weekly: 43-47.55 Jeffrey B. Nugent, 1983, An Alternative Source of Measurement Error as Explanation for the

Inverted Hypothesis, Economic Development and Cultural Change 31: 385-396.56 Laurence Chandy and Brina Seidel, 2017, How Much do we Really Know About Inequality

Within Countries Around The World? Christoph Lakner and Branko Milanovic, 2013, Global

Income Distribution: From the Fall of the Berlin Wall to the Great Recession, World Bank Policy

Research Working Paper, https://doi.org/10.1596/1813-9450-6719. 57 Sonalde Desai, Amaresh Dubey, Brij L. Joshi, Mitali Sen, Abusaleh Shariff, and Reeve

Vanneman, 2010, Human Development in India.58 Credit Suisse, 2017, Global Wealth Databook 2017. 59 Ishan Anand and Anjana Thampi, 2016, Recent Trends in Wealth Inequality in India. A longer

time series (beginning in 1961-2) is available in Subramanian and Jayaraj, The Evolution of

Consumption and Wealth Inequality in India, but for rural and urban data separately; no national

estimates are presented. Similar analyses are available in Arjun Jayadev, Sripad Motiram and

Vamsi Vakulabharanam, 2007, Patterns of Wealth Disparities in India during the Liberalisation Era.60 Sanjoy Chakravorty, The Price of Land.61 Thiagu Ranganathan, Amarnath Tripathi, and Ghanshyam Pandey, 2017, Income Mobility among

Social Groups.62 Vegard Iversen, Anirudh Krishna, Kunal Sen, 2017, Rags to Riches? Intergenerational

Occupational Mobility in India.63 G. S. Sahota, 1978, Theories of Personal Income Distribution: A Survey, Journal of Economic

Literature 16:1-55, p. 12.64 The Economist, 2018, Unconscionable: Low-caste Indians are better off than ever—but that’s not

saying much, Jan 25, Asia Edition. Poverty data from Arvind Panagariya and Vishal More, 2013,

Poverty by Social, Religious & Economic Groups in India and Its Largest States, 1993-94 to 2011-

12, Working Paper No. 2013-02,

http://indianeconomy.columbia.edu/sites/default/files/working_papers/working_paper_2013-02-

final.pdf; also see R, Radhakrishna, 2015, Well-being, Inequality, Poverty and Pathways out of

Poverty in India, Economic & Political Weekly, Vol.50, No.41. Additional malnutrition findings

from Michele Gragnolati, Meera Shekar, Monica Das Gupta, Caryn Bredenkamp and Yi-Kyoung

Lee, 2005, India’s Undernourished Children: A Call for Reform and Action, HNP Discussion

Paper, World Bank. 65 Diane Coffey, Payal Hathi, Nidhi Khurana, and Amit Thorat, 2018, Explicit Prejudice: Evidence

from a New Survey, Economic & Political Weekly 53(1): 46-54.