International Critical Thought

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/rict20

Urbanization and Exclusion: A Study on Indian Slums

Somenath Ghosh & Saumya Chakrabarti

To cite this article: Somenath Ghosh & Saumya Chakrabarti (2021) Urbanization and
Exclusion: A Study on Indian Slums, International Critical Thought, 11:3, 450-479, DOI:
10.1080/21598282.2021.1966820

To link to this article: https://doi.org/10.1080/21598282.2021.1966820

Published online: 19 Sep 2021.

Submit your article to this journal

Article views: 304

View related articles

View Crossmark data

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rict20
INTERNATIONAL CRITICAL THOUGHT
2021, VOL. 11, NO. 3, 450–479
https://doi.org/10.1080/21598282.2021.1966820

Urbanization and Exclusion: A Study on Indian Slums

Somenath Ghosh and Saumya Chakrabarti
Department of Economics and Politics, Visva-Bharati University, Santiniketan, India

ABSTRACT ARTICLE HISTORY

The processes of modern urbanization across the Global South have Received 17 April 2020
generated an intense debate. While mainstream researchers view it Revised 29 January 2021
as largely inclusive and advocating for slum development, critics Accepted 30 January 2021
argue that this process has an inherent tendency to displace
KEYWORDS
slum. Given this perspective, the authors show that there is, in Slum location; exclusionary
fact, a change in the location of Indian slums and the slum urbanization; urban
population from city centers to city fringes, where fringes are inequality; slum
found to be significantly and consistently unprivileged in terms of infrastructure; center–fringe
an array of infrastructural facilities in and around the slums. This inequality
paper also argues that the typical process of urbanization and
increasing urban inequality is inducing this changing location of
slums. The analyses indicate a forced relocation of the slum
population away from the city centers towards an inferior
standard of living on the city fringes. This paper uses Indian slum-
level data for 2001–2012, undertakes various advanced statistical
analyses, and presents a basic theoretical framework.

1. Introduction
In a developing country like India, slum has become a major challenge for “inclusive
urban development.” Despite a high rate of economic growth originating from urban
areas, the slum population has increased rapidly with this urbanization (Office of the
Registrar General and Census Commissioner of India 2001, 2011).1 Moreover, such a
growth of the slum population in the cities stands as evidence that “poverty is urbaniz-
ing” (Piel 1997) in the country.
Previously, it was thought that slum is a temporary phenomenon, as poor people who
enter into slums would gradually be shifted to proper housing with other facilities (Fran-
kenhoff 1967; Glaeser 2011; The World Bank 2009; Turner 1969). However, later on, it
was observed that slum would not wither away, because a lack of opportunities for
employment in the formal sector, the inability of the poor to get organized sector jobs,
and the improper distribution of growth benefits have withheld the disappearance of
slum (Harris and Todaro 1970; Stokes 1962). So, to take action on this, Government pol-
icies during the 1970s and 1980s have emphasized forced or voluntary resettlement of
slums to the fringes. But the implementation of such policies has produced a negative
impact on the economy of the cities as well as the conditions of the slum residents (Buck-
ley, Singh, and Kalarickal 2007). Moreover, these policies have not turned out to be

CONTACT Somenath Ghosh [email protected]

© 2021 Chinese Academy of Social Sciences
 INTERNATIONAL CRITICAL THOUGHT 451

successful in resolving the problem of slum as such. Hence, during the early 1990s, pol-
icies on upgrading existing slums rather than shifting them have been recommended by
the United Nations (UN) (UN-Habitat 2003). The UN suggested that, through such pol-
icies, slum residents could be included within the cities and a process of building inclus-
ive cities would be ensured.
However, fundamental critiques have been put forward against this recommendation,
proposing the view that slum would never be part of the proposed inclusive growth pro-
cess. Slum is always facing the threat of eviction and pushed towards the periphery with
the redevelopment of cities (Davis 2004; Harvey 2008). The underlying idea of such a
counter-argument is that more and more investment by big corporations in urban rede-
velopment would tend to expropriate slum lands by evicting the slum population by
using economic and political power. Thus, Harvey (2012) noted that a process of displa-
cement through urban development would engender a capturing of the land of the low-
income population and push it towards the periphery of the city with inferior living
conditions.2
This argument encourages us to look into the possibility of the displacement of slums
and the change in slums’ condition due to it, in the Indian context.
The theoretical debate commenced when the change in the location of the urban poor
had been identified by a group of (Chicago School) scholars, with evidence collected from
several Western cities during the 1920s and 1930s. The poor occupied the inner-city
spaces and the rich classes encircled them in the suburbs. There existed the availability
of comfortable spaces in the suburbs, with improved transport facilities for the privileged,
while the expansion of the factories in the city core attracted the poor laboring classes
(Burgess’s “spatial schema” specifies a “zone of transition” which was inhabited by the
working class [Burgess 1925]). This thought had also been reiterated by the Neo-Classical
School (Alonso 1960), by means of studying the trade-off between the cost of acquiring
the inner-city space and the cost of travelling to the suburbs. However, modern cities are
different—different to what had been conjectured by the spatial micro-economists (those
who have shaped spatial interaction models, spatial choice theory, and others). Contem-
porary cities have been transformed into goliaths, accumulating surpluses for the pro-
duction of new socio-cultural-geographical city spaces that are raising the chances of
the displacement of the poor through the process of gentrification. This can be elucidated
by quoting Harvey:
Surplus absorption through urban transformation has an even darker aspect. It has entailed
repeated bouts of urban restructuring through “creative destruction,” which nearly always
has a class dimension since it is the poor, the underprivileged and those marginalised
from political power that suffer first and foremost from this process. Violence is required
to build the new world on the wreckage of the old. . . .

A process of displacement and what I call “accumulation by dispossesion” lie at the core of
urbanisation under capitalism. It is the mirror-image of capital absorption through urban
redevelopment, and is giving rise to numerous conflicts over the capture of valuable land
from low-income population that may have lived there for many years. (Harvey 2008,
33–34)

Likewise, the “socio-spatial exclusion,” as termed by Wacquant (2010), is embedded in

the very nature of the modern urbanization process in developing countries, and operates
452 S. GHOSH AND S. CHAKRABARTI

through the process of “urban polarisation.” In this process, the elite class appears to be
distancing itself from the poor by confining itself in enclave societies and through the
adaptation of a new form of urban cultural life (mostly centering around consumerism):
“Ghettoization” is happening at the periphery. Even in the cities of the developed
countries, “urban polarisation” has become a pertinent issue that has drawn attention.
However, unlike the urban regeneration approach, Bhattacharya and Sanyal (2011)
describe the cities of India that have adopted a “bypass approach” for the rich to exclude
the poor-class. These debates insinuate a new form of urban transformation that requires
empirical investigation.
So far, many studies have been undertaken on slums, but little emphasis has been
placed on studying the displacement or eviction of slums at the macro level. During
the 1970s and 1990s, the “index of segregations” had been frequently applied by several
scholars (who widely used factorial ecology techniques) to identify the level of dissimi-
larity across spaces in urban areas (Lebowitz 1977), but the possibility of unequal treat-
ment towards the rich and the poor through re-location has not been explored
sufficiently using quantitative methods. Given this shortcoming in the literature in gen-
eral, this paper aims to observe the pattern of change in the location of slum (from city
center to city periphery) in India and intends to find out the probable factors influencing
it. We also seek to see the difference in infrastructural conditions between the fringe (city
periphery) and non-fringe (city center) slums to understand whether the change of
location marks deterioration in the living conditions of the slum dwellers or not.
This study is done using Indian state level as well as unit (slum) level data over time.
On the one hand, we have taken data for 14 major states3and India as a whole, and, on the
other, a total of 1573 sample slums, for two rounds: 2002 and 2012. Although slum is our
main unit of analysis, we have even considered slum households in some parts of our
study. For our analysis, we have focused on the following variables:

(a) For the first part of the study on the location of slum, we take the main target vari-
able: the share of slums on the fringe areas out of the total number of slums. Along
with this, we take the number of slum households in fringe and non-fringe areas and
their growth rates. We also select the percentage of slum households situated at
fringe areas out of the total number of slum households as another variable to under-
stand the shift of slums. Along with this, satellite images of Kolkata City (see Figure
A1. [a] and Figure A1. [b]) have been used in this study to understand the change in
the structure of urbanization.
(b) For the second part of the study, the (target) variables are: types of housing structure;
types of sources of drinking water; types of roads within and approaching the slums;
and types of drainage, latrine, sewerage, and garbage disposal facilities that are con-
sidered to indicate the difference in infrastructural conditions between fringe and
non-fringe slums.
(c) For the third part of the study, to analyze the factors influencing slum location, we
have taken three main (explanatory) variables: share of the urban population com-
pared with the total; consumption inequality in urban areas per square kilometer;
and per capita Net State Domestic Product of the urban area.4 Along with this,
rural–urban migration and average household expenditure are taken as two control-
ling factors in the empirical analysis.
 INTERNATIONAL CRITICAL THOUGHT 453

In our analysis, we mostly used diagrams for visual representations. In addition, in the
first part, we did a paired t-test to identify the extent of the shift of slums to the fringe of
the city. Moreover, 2 × 2 factorial design (two-way ANOVA [variance analysis]) and
ordered logit regression have been done to observe if differences in the conditions
between fringe and non-fringe slums are significant and, subsequently, whether these
conditions and their differences are changing significantly over time. Finally, in the
third part, we use correlations and binary logit regression to find out the effect of the
probable exogenous variables on location change for slums.
In the paper, we deal first with the issue of the change in the location of slums and find
that there is an increase in the concentration of slums on the fringes of the cities. Next, we
take up a comparison of the infrastructural conditions of the slums situated at the fringe
and the center of the city. This is to understand whether this shifting of the slum from the
non-fringe to the fringe of a city, as mentioned earlier, is an improvement in living con-
ditions or not. This exercise shows that the conditions of the slums on the fringe areas are
significantly worse compared to that of the non-fringe areas. Further, we show that, over
time, there is an overall deterioration of the standard of living of slum people. In the next
section, we look at the probable factors for this change of location towards the fringe and
identify the extent of urbanization and inequality as two crucial aspects. Finally, in the
concluding section, we summarize the findings of the paper and discuss its political
and economic implications. This article has tried to knit three different parts of the
study with one logical string to comprehend the ongoing processes of exclusionary
urbanization in the context of India, where a course of urban polarization is taking
place. It may help in contemplating new ways of urban/city planning favoring the motion
for a “Right to the City” (following Henri Lefevre’s ideas).

2. Change of Location of Slum across Indian States from Inner City to City
Fringe
This section is devoted to estimating the increased concentration of slums at the fringe
areas of the cities to assess the extent of the displacement of slum to the periphery.
The transformation to post-Fordist cities is observed to have hollowed spaces at the
center for consumption attractions (parks, business centers, shopping malls, and so
on), whereas fragmented suburbs have been occupied by the distressed, which Davis
(2004, 260) had rightly identified: “The classic slums are decaying inner-city, the new
slums are more typically located on the edge of urban spatial explosions.” Our analysis,
too, has shown a similar pattern of increase in the share of slums at the fringe areas of the
city over time, in India as a whole, and in its states. The major states like Andhra Pradesh,
Bihar, Chhattisgarh, Gujarat, Haryana, Maharastra, Punjab, Rajasthan, Orissa, and Tamil
Nadu have shown an increase in the share of slums at the periphery of the cities (see
Figure 1). However, there is a reverse trend in the states of Karnataka, Madhya Pradesh,
and Uttar Pradesh and a very marginal decrease in the state of West Bengal. Despite this,
a paired t-test confirms a significant increase in the average percentage of slum in the
fringe areas (see Table 1).
Along with the increased concentration of slums at the fringe areas of the cities, it is
also evident that the number of slum households has increased in most of the states as
well as for India as a whole (see Table 2). We find a negative decadal growth rate of
454 S. GHOSH AND S. CHAKRABARTI

Figure 1. Percentage of slums situated at fringe areas.

Note: The NSSO (National Sample Survey Organisation) belongs to the Ministry of Statistics & Programme Implementation
of the Government of India. Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and
Census Commissioner of India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India
(2018).

Table 1. Results of the mean difference test of the percentage of slums situated in fringe areas.
Mean Standard deviation Significance
2002 23.22 16.83 0.022
2012 34.20 19.63
Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).

non-fringe slum households in the states of Bihar, Chhattisgarh, Maharashtra, Tamil

Nadu, West Bengal, and Uttar Pradesh, which reveals a decreasing number of slum house-
holds in the city centers. However, in other states, we find positive growth in the number
of slum households in both fringe and non-fringe areas: the number of slum households at
the fringes have increased at a higher rate compared to those of the non-fringe. This may
indicate either the concentration of the migrant rural poor in the periphery region, unable
to enter into the cities, or the shifting of slums from the city center to the city periphery.
For example, the states of Maharashtra and Gujarat are highly urbanized, at around 40%.
The average economic growths of these states are higher relative to those of other states,
but the percentages of slums in the city fringes, in these states, have increased at higher
rates. One of the reasons can be the proliferation of industries in the fringe locations,
and hence the poor slum residents are pulled to the city fringes (Kumar 2018). However,
there are many slum eviction experiences due to infrastructural development inside the
city (Doshi 2012; Sato 2017). On the other hand, urbanization in Tamil Nadu is spatially
dispersed and there has been the emergence of small towns and agglomerations due to
substantial improvement in infrastructural facilities (Kolappan 2016), although evictions
in the large cities of Tamil Nadu are quite evident (Saharan, Pfeffer, and Baud 2018),
which may induce the shifting of slums to the fringe. Land prices in West Bengal are
not so high compared to the urban locations of other states, so that can be a reason for
the sluggish growth of slums on the fringes there. However, the explosion of the slum
population at the fringes of Kolkata City has been shown through satellite images during
the periods 2002 and 2020 (see Appendix 1).
Table 2. Number of slum households in fringe and non-fringe locations and their decadal growth rate.
2001 (in 1000 households) 2011 (in 1000 households)
Percentage of percentage of
fringe slum fringe slum Decadal growth Decadal growth
Slum Slum households out of Slum Slum households out of rate of fringe rate of non-
household household Total slum total slum household household Total slum total slum slum fringe slum
State fringe non-fringe households households fringe non-fringe households households households households
Andhra 215 914 1129 19.1 1052 1218 2270 46.3 388.8 33.3
Pradesh
Bihar 41 28 69 59.1 67 11 78 86.1 63.2 -62.1
Chhattisgarh 42 68 110 37.9 22 67 89 24.9 -46.7 -1.9
Gujarat 38 120 158 23.8 101 254 355 28.5 168.7 111.3
Haryana 1 21 22 4.3 21 47 67 30.6 2041.2 117.4
Karnataka 68 416 484 14.1 200 427 627 31.9 193.6 2.7
Madhya 48 261 308 15.4 154 347 502 30.8 225.0 33.2
Pradesh

INTERNATIONAL CRITICAL THOUGHT

Maharashtra 209 2973 3183 6.6 930 2433 3363 27.7 344.6 -18.2
Orissa 0 18 18 0.0 52 211 263 19.9 1056.2
Punjab 5 4 9 51.9 9 161 170 5.2 89.5 3644.7
Rajasthan 3 54 57 4.8 198 207 405 48.9 7193.7 282.3
Tamil Nadu 130 489 620 21.0 241 410 651 37.1 85.1 -16.2
Uttar 58 170 228 25.3 2901 167 3068 94.6 4938.7 -2.2
Pradesh
West Bengal 712 819 1531 46.5 291 539 830 35.1 −59.1 -34.2
India 1631 6599 8230 19.8 6396 6920 13316 48.0 292.2 4.9
Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The
Reserve Bank of India (2018).

455
456 S. GHOSH AND S. CHAKRABARTI

3. Difference in Conditions between Fringe and Non-fringe Slums

3.1. Difference in Infrastructure Conditions across Fringe and Non-fringe Slums
One may put forward an argument for location choice, following the view on “Central
Place Theory” (King 2020). The lower-order markets as mentioned by Christaller in
1933 might be prevailing in the periphery region, which are perhaps enticing slum resi-
dents to choose the periphery locations in a city to access such markets for better living.
Therefore, analyzing the differences in living conditions between fringe and non-fringe
slums could be a way to verify this argument of whether the choice of location leads
to better living conditions. To understand the differences across the fringe and non-
fringe slums, we have selected some variables pertaining to infrastructure. These vari-
ables are ordinal, with ordering ranging from better to worse conditions.
We observe that slums having houses made up of “puccka”5 material are found more
in non-fringe areas than fringe areas. Nearly 60% of the slums in non-fringe areas have
houses made with “puccka” material, compared to less than 50% of slums in fringe areas.
In addition, the percentages of slums having “semi-puccka”6 and “serviceable-katcha”7
houses are greater in fringe areas than in non-fringe areas (see Figure 2). Similarly, trea-
ted tap water is mostly available in non-fringe areas. More than 70% of the slums in non-
fringe areas have tap-water facilities as a major source of drinking water, whereas less
than 65% of slums in fringe areas do. We observed that tube wells are the major source
of drinking water available mostly in the slums situated in the fringe locations (see
Figure 3). However, the percentage of slums with a “puccka” road inside the slums is
higher in the fringe than in the non-fringe areas (see Figure 4). Almost the same percen-
tage of slums has a motorable metal road approaching the slums in both the fringe and
non-fringe locations. Although the percentage of slums that has (non-motorable) metal
roads is higher in the non-fringe locations, the slums with (motorable) non-metal roads
are extensive in the fringe areas compared to in the non-fringe locations (see Figure 5). In
the case of latrine facilities, the fringe areas have a higher percentage of slums without
latrine facilities. Moreover, the percentage of slums having public latrines is higher in
non-fringe areas (see Figure 6). Further, a higher percentage of slums in non-fringe
areas has sewerage systems and underground drainage systems. On the contrary, a higher
percentage of slums in fringe areas has no drainage facilities at all (see Figures 7 and 8).8

Figure 2. Type of housing structure.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).
 INTERNATIONAL CRITICAL THOUGHT 457

Figure 3. Source of drinking water.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).

Figure 4. Types of roads within slums.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).

Figure 5. Types of roads approaching the slums.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).
458 S. GHOSH AND S. CHAKRABARTI

Figure 6. Types of latrine facilities.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).

Figure 7. Sewerage system facilities.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).

Figure 8. Types of drainage facilities.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).
 INTERNATIONAL CRITICAL THOUGHT 459

Figure 9. Arrangements for garbage disposal.

Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).

Regarding garbage disposal facilities, the number of slums having garbage disposed of by
governments is higher in non-fringe areas. Around 60% of the slums in non-fringe areas
have garbage disposal facilities arranged by the governments, while around 40% of the
slums in fringe areas have no garbage disposal facilities at all (see Figure 9).
By summarizing the analysis given here, it is observed that the infrastructure con-
ditions, except roads, are comparatively better in the slums situated in non-fringe
areas than those which are situated in the fringe location of the cities. This may indicate
that slums situated inside the cities are enjoying better infrastructural facilities than the
slums located on the fringes. Moreover, there is a possibility that the conditions of slums
have deteriorated due to their re-location to the fringe. Therefore, factorial designs have
been undertaken to observe whether the difference in the infrastructural condition of
slums is significant between the fringe and non-fringe locations and whether or not
the change in location with time is bringing any variation in conditions.

3.2. Results of Factorial Design (Two-Way ANOVA)

The factorial experiment here has been built upon a 2 × 2 model of factorial design with
“replications.” A factorial design is a kind of ANOVA, used for studying the effect of two
or more factors. In this statistical technique, all possible combinations of the level of the
factors are investigated in each replication (Pannerselvam 2013). For this model, we have
considered two main factors: “location” and “time.” Under each of these main factors,
there are two sub-factors or levels. Non-fringe and fringe are two sub-factors under
“location,” whereas 2001 and 2011 are the years, considered as two sub-levels or sub-fac-
tors under “time.” The replications are the values for each slum, out of the ordering cat-
egories, taken under each variable. So, the model is illustrated here:
Yijk = m + Ai + Bj + ABij + eijk (1)
Equation 1 represents 2 × 2 factorial design, where Yijk is the dependent variable. In this
case, the dependent variables are the different aspects of infrastructure, i.e., housing
structure, types of road within and approaching slums, types of drinking water, and so
 460
Table 3. Ranks and values of infrastructural conditions.

S. GHOSH AND S. CHAKRABARTI

Types of housing structure Sources of drinking water Types of roads within slums
Rank/value for Dummies for ordinal Rank/value for Dummies for ordinal Rank/values for Dummies for ordinal
Criteria ANOVA regression Criteria ANOVA regression Criteria ANOVA regression
Puccka 1 3 Tap water 1 3 Puccka 1 1
Semi-puccka 2 2 Tube well 2 2 Katcha 2 0
Serviceable katcha 3 1 Protected well 3 1
Unserviceable katcha 4 0 Unprotected well 4 0
Types of roads approaching the slums Type of latrine facilities Sewerage system facilities
Criteria Rank/values for Dummies for ordinal Criteria Rank/values for Dummies for ordinal Criteria Rank/values for Dummies for ordinal
ANOVA regression ANOVA regression ANOVA regression
Motorable metal road 1 3 Owned latrine 1 2 Yes 1 1
Motorable non-metal 2 2 Public latrine 2 1 No 2 0
road
Non-motorable metal 3 1 No latrine 3 0
road
Non-motorable non- 4 0
metal
Types of drainage facilities Arrangements for garbage disposal
Criteria Rank/values for Dummies for ordinal Criteria Rank/values for Dummies for ordinal
ANOVA regression ANOVA regression
Underground 1 4 Arranged by 1 2
government
Covered puccka 2 3 Arranged by slum 2 1
residents
Open puccka 3 2 No arrangements 3 0
Open katcha 4 1
No drainage 5 0
Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The
Reserve Bank of India (2018).
 INTERNATIONAL CRITICAL THOUGHT 461

on; µ is the overall mean of the model. Ai and Bj are two independent variables, which are
location and years (time) respectively, where “i” is two types of location, namely “fringe”
and non-fringe, “j” is two time periods, and k is the number of replications. ABij is the
interaction term which will denote the interaction effects of the independent variables
on the dependent variables. eijk is the error term.
In turn, to do the experiments, the ordering of categories in each variable are valued or
ranked accordingly from good to bad. The lowest value has been imposed on the best
category and, similarly, the highest value has been imposed on the worst category
among them. This is explained in Table 3.
With the help of Table 3, the average values of replications across location and time
have been represented through plot diagrams (see Figures 10–17).

Figure 10. Estimated marginal means of housing structures.

Figure 11. Estimated marginal means of drinking water.

462 S. GHOSH AND S. CHAKRABARTI

Figure 12. Estimated marginal means of roads within slums.

Figure 13. Estimated marginal means of roads approaching slums.

The vertical axes in all of the diagrams (see Figures 10–17) represent the average of
values or ranks or replications and the horizontal axes represent the years. There are
two kinds of lines in each diagram. One line represents changes in the condition
(from 2001 to 2011) of a particular infrastructure occurring in fringe areas (the line
named “Fringe”) and another line is the same for the non-fringe areas (the line named
“Non-fringe”). The outcomes of this analysis are discussed as follows:
In terms of the average values, the conditions of the overall infrastructure are better
(except for infrastructure roads within and approaching the slums) in non-fringe
locations than fringe areas.
In the case of the condition of housing structures, the average values are lower in non-
fringe locations (i.e., values 1.7 and 1.5) compared to those in the fringe (i.e., values 1.9
 INTERNATIONAL CRITICAL THOUGHT 463

Figure 14. Estimated marginal means of latrine facilities.

Figure 15. Estimated marginal means of sewerage systems.

and 1.6) in both the years, which indicates better housing conditions in non-fringe
locations than in fringe. However, the average values dropped at a higher rate in fringe
locations from 2001 to 2011 (see Figure 10), which implies a narrowing of the fringe–
non-fringe gap.
As per the average values, the condition of drinking water is inferior in slums situated in
fringe areas in 2001. However, the increase in average values in both the locations in 2011
signifies the deteriorating conditions of drinking water facilities in the slums of both places,
and the condition has deteriorated at a higher rate in the slums of fringe areas (see Figure 11).
However, the condition of roads within and approaching the slums are found to be
inferior in slums situated in non-fringe locations, but the conditions of those have gradu-
ally improved in 2011 (see Figures 12 and 13).
464 S. GHOSH AND S. CHAKRABARTI

Figure 16. Estimated marginal means of drainage facilities.

Figure 17. Estimated marginal means of garbage disposal.

In 2001, the conditions of latrine facilities were found to be better in the slums of
fringes compared to the slums in non-fringe areas, but the condition has deteriorated
in fringe areas more rapidly than in non-fringe areas (see Figure 14).
The average values which signify the conditions of the sewerage, drainage, and garbage
disposal facilities are also found to be in better condition in non-fringe locations than in
fringe locations (see Figures 15–17).
In addition to the explanations of the diagrams given here, Table 4 illustrates the
results of a two-way ANOVA, which reveals whether the difference in conditions across
locations and over time are significant or not. We find that there is a significant difference
in the condition of housing, drinking water facilities, internal roads, sewerage facilities,
 INTERNATIONAL CRITICAL THOUGHT 465

Table 4. Results of factorial design (two-way ANOVA).

Independent variables Location Time Interaction
Dependent variables F-ratio Significance F-ratio Significance F-ratio Significance Observations
Types of structures of housing 9.870 0.002 37.580 0.000 2.290 0.130 1573
Types of sources of drinking 7.705 0.006 9.603 0.002 1.844 0.175 1573
water
Types of roads within slums 8.418 0.004 13.642 0.000 25.774 0.000 1573
Types of roads approaching 2.130 0.145 6.288 0.012 0.167 0.683 1573
slums
Types of latrine facilities 1.839 0.175 61.720 0.000 12.330 0.000 1570
Types of sewerage facilities 26.830 0.000 0.539 0.463 0.437 0.509 1573
Types of drainage facilities 29.827 0.000 0.686 0.408 0.005 0.945 1573
Mode of garbage disposal 14.460 0.000 4.278 0.039 1.334 0.248 1573
arrangements
Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of
India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).

drainage facilities, and garbage disposal facilities across the locations. The conditions of
all the infrastructures, except sewerage and drainage facilities, have significantly varied
over time. The results of the interaction part have explained the variation of infrastruc-
tural conditions with a change in location and time, and are illustrated in Table 4. It
shows that the differences in conditions between two locations have not changed signifi-
cantly with time. Therefore, the result of ANOVA signifies that the conditions of most of
the infrastructures are significantly inferior in slums situated in fringe areas and this
difference is persisting significantly over time. These results are thus refuting an argu-
ment in favor of the location choice of the slum residents and, on the contrary, show
that they are not moving away from the center to the city periphery by their choice. Per-
haps there are a variety of push factors that are operational in shifting them to inferior
conditions in city fringes, and that such inferior conditions have persisted over time in
the city fringes.
Apart from applying ANOVA, regression can be used for more robust results in
measuring the effects of the independent variables such as location, time, and their
interaction (as in ANOVA) on the conditions of infrastructure. Therefore, the next
section discusses an application of ordered logit regression technique for estimating
the effect of the independent variables on the ordered categorical dependent
dummy variables.

3.3. Results of Ordered Logit Regression

Here, in this section, we perform ordered logit regressions with the same eight aspects
of infrastructure (viz. the nature of housing structures; types of roads within and
approaching the slums; the conditions of drinking water facilities; and the types of
drainage, sewerage, latrine, and garbage disposal facilities) related to slums. These
aspects are considered as the dependent variables in the regressions to assess the
effect of location and time and their interaction on the conditions of each of these
infrastructural facilities. The ordered logit regression is undertaken mainly to predict
ordered categorical variables given one or more independent variables. We use this
kind of regression model because the dependent variables of interest are categorical
and ordinal in nature. The dummies of the categories under each of these variables
466 S. GHOSH AND S. CHAKRABARTI

have been ordered from better to worse where the best among the categories have
been marked with the highest values or scores; the worst among the categories is con-
sidered as the benchmark and valued with zero (see Table 3). The independent vari-
ables—“location,” “time,” and their interactions—are also binary categorical variables
taking only values “0” and “1.” The variable location takes the value 0 for slums situ-
ated in non-fringe areas and the value 1 for slums situated in fringe locations. The
variable “time” takes the values 0 and 1, and has been denoted for the years 2001
and 2011 respectively. Along with these, the variable—number of slum households
—has been taken as a covariate to control the size of slums. The ordinal regression
model used here is explained with the following equation:

Y = Ln(Øj ) = Aj + B1 Xlocation + B2 Xtime + B3 Xinteraction + B4 Xslum household no. + ei (2)

where θj = Prob(score ≤ j) / Prob(score > j), and θ is the odd ratios and j is the num-
ber of categories of each infrastructural aspect. Here Aj is the threshold value for each
j. B1, B2, B3, and B4, are the coefficients of the independent variables location, time,
interaction, and slum household numbers respectively.
As we study the results of the eight ordered logit regressions of Table 5 we see that all
the above-mentioned regressions are significant and that among eight of these
regressions, seven regressions have shown the significant impact of location on the con-
dition of infrastructure. The result of Equation 1 shows that location is negative and sig-
nificantly affects the condition of housing, which implies the change in the location of the
slums, from non-fringe/center to fringe, has deteriorated housing conditions. That is,
fringe slums have a proportionately higher number of semi-puckka, serviceable katcha,
and unserviceable kutcha houses compared to that of non-fringe slums. However, with
the change in time, during 2001 and 2011, the condition of housing improved signifi-
cantly in the slums, across the locations. There is no significant impact of interaction
(of location change and time periods) and number of slum households on the dependent
variable.
The result of regression 2 shows that the condition of drinking water in terms of its
source has significantly deteriorated with the shift of slums to fringes; moreover, over
time too, there is a significant deterioration. However, the probable impacts of the inter-
action variable and the number of slum households are insignificant.
In regression 3, we see that the location of slums has a significant positive impact on
the condition of roads within the slums; moreover, time too has a significant positive
impact. However, in regression 4, except time, none of the independent variables has a
significant impact on the conditions of roads approaching slum. Time has a positive
and significant impact on the dependent variable, which implies an improvement in
the condition of approach roads to slums due to the change in time over 2001–2011.
In regressions 5, 6, 7, and 8, we find a negative and significant impact of location on
the condition of the latrine, sewerage, drainage, and garbage disposal facilities of slums
respectively in terms of their types. It means that the conditions of latrine, sewerage, drai-
nage, and garbage disposal facilities in slums have deteriorated due to a shift of slums to
fringe areas. However, except in regression 5, in all three regressions, time has no signifi-
cant effect on the dependent variable. In regression 5, time has a positive and significant
effect on the condition of latrine facilities in the slums, which reveals an improvement in
Table 5. Results of ordinal logit regressions.
Coefficient of independent variable (odd ratios)
Equation Interaction of location and Slum household Number of Pseudo R-
number Dependent variable Location Time time number observations square
1 Type of housing structures .6062*** (.096, -3.15) 1.778*** (.211, 4.18) 1.295 (.278, 1.20) 1.000 (.000, 0.44) 1573 0.0166
2 Sources of drinking-water .5041*** (.099, -3.48) .6472*** (.095, -2.94) 1.163 (.291, 0.60) .999 (.000, -0.98) 1573 0.0157
facilities
3 Types of roads within 2.154*** (.455, 5.09) 2.73*** (.342, 8.03) .304*** (.072, -5.02) .999 (.000, -0.63) 1573 0.0345
4 Types of roads approaching .9671 (.161, -0.20) 1.334** (.166, 2.32) 1.234 (.283, 0.91) .999 (.000, -1.35) 1573 0.0042
5 Types of latrine facilities .5550*** (.090, -3.59) 1.884*** (.209, 5.69) 1.025 (.222, 0.12) .999 (.000, -0.04) 1573 0.0207
6 Types of sewerage facilities .3768*** (.099, -3.70) .9795 (.138, -0.15) 1.359 (.450, 0.93) 1.000 (.000, 1.58) 1573 0.0196
7 Types of drainage facilities .5534*** (.087, -3.76) 1.094 (.121, 0.81) 1.039 (.214, 0.19) 1.000 (.000, 1.10) 1573 0.0077

INTERNATIONAL CRITICAL THOUGHT

8 Arrangements for garbage .5869*** (.097, -3.21) 1.095 (.134, 0.75) 1.134 (.251, 0.57) .999 (.000, -0.42) 1573 0.0069
disposal
Note: ** and *** are the sign of 5% and 1% level of significance respectively. In the regressions, the coefficients are the odd ratios. We also put the standard error and Z score within the bracket
with each coefficient.
Source: Author’s estimate based on Pal and Ghosh (2007), Office of the Registrar General and Census Commissioner of India (2001, 2011), NSSO (2001, 2003a, 2003b, 2010, 2012, 2014), and The
Reserve Bank of India (2018).

467
468 S. GHOSH AND S. CHAKRABARTI

latrine facilities during 2001 and 2011, across locations. However, we again find no sig-
nificant impact of the interaction term and that of the number of slum households.
So, the above regression results confirm the hypothesis that the relocation of slum resi-
dents to the city peripheries has deteriorated their condition and this inequality has per-
sisted over time. However, this induces us to raise another question: what are the factors
which are influencing the change in the location of slums? Hence, in the following sec-
tion, we try to find out the probable factors which may affect the change in the location of
slums, through the use of correlations and regression.

4. Probable Factors
We propose that urbanization, economic inequality, and economic growth could be
some of the determining factors which might be displacing slums to the peripheries of
the cities. There is a popular view that infrastructural facilities are limited relative to
the congestion of the urban population (FICCI [The Federation of Indian Chambers
of Commerce and Industry] 2011), which can prevent slum growth in the center. How-
ever, this approach is unable to consider a variety of crucial aspects.
Firstly, urbanization may involve the modernization of public space and the better-
ment of infrastructural facilities and public services. However, this may lead to an
increase in the cost of living, which may push away inner-city slum.
Secondly, economic inequality may generate unequal capacity to access urban land. This
may lead to the concentration of inner-city land in the hands of the rich, and poor slum
people may be dispossessed through economic and political processes. Increases in income
inequality within the urban economy may generate stark inequality in terms of urban ser-
vices and infrastructural facilities too: the rich may not be willing to finance the services
and infrastructural facilities to be used as public goods (for example, roads, health facilities,
garbage disposal, and so on). The rich may privately organize these schemes for themselves
and bargain for the lowering of public expenditure and hence efficacy of these services and
facilities. Because of the lack of public funding, the condition of these public services
deteriorates drastically. This may lead to a crisis for the urban poor, pushing them towards
the fringes (Bhattacharya, Saha, and Banerjee 2016).
Finally, modern economic growth is heavily biased towards the asset rich and the
skilled and discriminates against the poor and unskilled. Consequently, the benefits of
growth are poorly accessed by the already disadvantaged slum population and this rise
in income inequality makes many of the basic necessities of urban life inaccessible.
This may compel the inner-city slum population to shift to the fringes with a relatively
lower cost of living.

4.1. Line Diagrams and Correlation Results

We have tried to check whether these aspects (urbanization, inequality, and growth)
could be responsible for the movement of slum from center to periphery. The outcomes
of our diagrammatic and correlation analyses are as follows:

a) We have observed a positive and significant relationship between the variables

share of slums at the fringe areas and the share of urban population to the
 INTERNATIONAL CRITICAL THOUGHT 469

Figure 18. Line diagram showing the relation between the share of the number of slums situated at
the fringes (y-axis) and the share of urban population to total (x-axis).
Note: “Fitted values” denotes the prediction of the mean values of the deviation of y variable with respect to increase in x
variable.

total with the help of fitted lines and correlation (see Figure 18 and Table 69).
This result may indicate that urbanization could be a factor which may displace
slums to the city fringes.
b) Moreover, a positive and significant relationship between the variables share of slums
at the fringes to total slums and urban consumption inequality per square kilometer
signifies the eviction of the poor from inside the city as inequality rises (Figure 19 and
Table 6). Here we have normalized the value of inequality by dividing it with the area
of the states. This has been done to take care of the issue of the degree of concentration
of inequality having an impact on people’s relative decision making.
c) On the other hand, we find no relation between the share of the number of slums in
fringe areas and per capita urban NSDP (Per Capita Net State Domestic Product). This
result may indicate that the kind of economic growth originating in urban areas has no
significant role to play in benefiting the poor (see Figure 20 and Table 6).

Table 6. Correlation results.

Share urban population out Urban consumption inequality per Per capita urban
of total square kilometer NSDP
% of slums situated in .3998* 0.3167* −0.0597
fringe areas
Note: * indicates a 10% level of significance.
Source: Author’s estimate based on The Reserve Bank of India (2018).
470 S. GHOSH AND S. CHAKRABARTI

Figure 19. Line diagram showing the relation between the share of the number of slums situated at
the fringes (y-axis) and urban consumption inequality per square km (x-axis).

Figure 20. Line diagram showing the relation between the share of the number of slums situated at
the fringes (y-axis) and per capita urban NSDP (x-axis).
Note: The unit of the values mentioned for x axis in the figure should be Indian Rupees (INR), which has been included
along with the name of the variable of x axis in the figure.
 INTERNATIONAL CRITICAL THOUGHT 471

Along with correlations, we intend to use regression to confirm the effects of the prob-
able factors. So, in the following section, we use the binary logit regression model for pre-
dicting the location change due to the independent variables, as specified.

4.2. Results of Binary Logit Regression

Here we use binary logistic regression to estimate the effects of the independent variables
on location change. Binary logit regression is used when the outcome variable takes only
the values in 0 and 1. In our case, the dependent variable takes the value 1 when slums are
situated on fringes and is 0 otherwise. State-level data for the independent variables have
been repeated for slum level. It is illustrated here in an equation format,
Li = Ln[Pi /(1 − Pi )] = f (Share of urban-population to total,
Urban consumption inequality per sq-km, Per capitaurban NSDP,
(3)
Share of migrants from rural to urban out of total migrants, MPCEurban ,
Time-dummy)
where Pi is the probability of slums situated on city fringes and Li is the log of odd
ratios of the probability of slums in fringe to non-fringe areas. Along with the targeted
explanatory variables, the share of migrants from rural to urban (data source: NSSO
2001, 2010) has been taken as a covariate to control the influence of rural to urban
migration, as there is a high possibility that the increase in slums on the city fringes
may be due to rural–urban migration. We have also considered another control vari-
able i.e., Monthly Per Capita Expenditure (MPCE) for the urban households (2003a,
2014), because increases in household expenditure may lead to an increase in demand
for domestic services which, in turn, may attract the poor to gather around the wealthy
citizens inside the cities.
The regression result (Table 7) shows that an increase in the size of the urban popu-
lation and urban consumption inequality can predict the shift of slums to city peripheries
from city centers significantly. However, urban per capita income and rural–urban
migration have not significantly affected the change in the location of slums. On the

Table 7. Result of the binary logit regression.

Number of observations 1394
Wald chi2 18.24
Prob > chi2 0.0057
Pseudo R 2 0.011
Independent variables Coefficient B (S.E, Z) Significance
Share of urban population to total .030 (.014, 2.12) 0.034
Urban consumption inequality per square kilometer 94671 (50918, 1.86) 0.063
Per capita urban NSDP -6. 02e (4.34e, -1.39) 0.166
Monthly Per Capita Expenditure (MPCE) of urban -.000 (.000, -1.83) 0.067
share of migrants from rural to urban out of total migrants .008 (.010, 0.78) 0.437
Time dummy 1.00 (.333, 3.01) 0.003
Constant -1.85 (.791, -2.35) 0.019
Note: “S.E” stands for Standard Error and “Z” stands for “Effect Size” of the independent variable on the dependent vari-
able. Z-value is regression coefficient divided by standard error.
Source: Author’s estimate based on Office of the Registrar General and Census Commissioner of India (2001, 2011), NSSO
(2001, 2003a, 2003b, 2010, 2012, 2014), and The Reserve Bank of India (2018).
472 S. GHOSH AND S. CHAKRABARTI

other hand, we find out, MPCE-urban has a negative and significant impact on the
dependent variable and a change in the time from 2001 to 2011 significantly shifts the
location of slums, from city center to city peripheries.
The result given here can be explained in the following manner: the incidence
and pattern of urbanization and rising economic inequality could be important
reasons behind the shift of slums to city peripheries, and, surprisingly, rural–
urban migration is found to be insignificant, in this context. In addition, the signifi-
cant negative outcome of household expenditure may mean that there is a demand
for domestic and a variety of household workers from rich people living inside the
cities, which may be inducing the poor to cluster around the wealthy households
inside the cities.

5. In Lieu of Conclusion
5.1. Summary Findings
Summarizing the findings of the paper, we could arrive at the following propositions:

(a) First, we find that, although in India the number of slums is reducing over time
(during 2001–2011), the size of the slums on average, in terms of slum population,
is expanding. There is now a smaller number of larger slums.
(b) There is increasing concentration of slum on the city fringes compared to that of the
city-centers, in terms of both number of slums as well as overall slum population
(households).
(c) This might be because of a movement of the slum population from the city core to
the city fringes. We do not deny the possibility that the expansion of fringe slums is
perhaps due to rural–urban migration, as the poor rural migrants migrating to
urban areas could not enter the city and are pushed to settle at the city’s fringes—
but the movement of the slum population from city center to city periphery may
be more important. So, larger and larger proportions of slum households are con-
centrating at the fringes. The proportion of fringe slum households is increasing,
even if we consider the overall urban household.
(d) Next, we posited the question of whether this movement of slum population is bet-
terment for the migrant or a deterioration of living standards. To answer this ques-
tion, we compared the infrastructural conditions of the slums situated in both city
cores and on city fringes.
(e) We found that the infrastructural conditions of the peripheral slums is significantly
inferior compared to those of the city centers. Not only that, this gap is persisting
over time.
(f) From the findings given here as summarized through (c), (d), and (e), we could
argue that the movement of the slum population away from the city center to the
fringes is due to the displacement of the slum population of the city center. We
argue that the movement of the slum population to the periphery is because of
the operation of certain push factors working at the core of the cities.
(g) We propose, through urban development, that there is not only direct displacement
of slum from city core to the fringes, but that there is also increasing inequality (in
 INTERNATIONAL CRITICAL THOUGHT 473

terms of opportunities and access to land, infrastructural facilities, public services,

and so on), which is pushing the population out of the city cores.
(h) Thus summarizing all the findings presented here, we could say that in neo-liberal
India, contrary to the proposed doctrine and policy of building inclusive cities, we
are perhaps witnessing a process of exclusionary urbanization.

5.2. In Lieu of Conclusion: Some Political-Economic Implications

The results given here indicate that the contemporary processes of urbanization are such
that the people who are in a disadvantageous position and stay in slums are adversely
affected. These processes of urbanization lead to a kind of “elite capture” of not only
urban land (private land as well as commons), but also of urban services, infrastructural
facilities, high-skill based service-sector oriented job opportunities, and so on. Thus, such
a process of urbanization having distinct bias in favor of the privileged not only reduces
the employment and income opportunities of the disadvantaged slum population, but
also increases the cost of decent living. Thus the slum population faces income decline
as well as an increase in the cost of living. This suffocates and pushes the population
towards the periphery of the modern cities (an unproductive location), with a lower
price for land on the one hand, and cheap but poor infrastructural facilities as well as
other services on the other. Thus, this typical process of urbanization pushes and
finally excludes the slum population. In this context, we present a brief framework par-
tially explaining the process noted here (Appendix 2).
Of course, over and above such processes of dispossession through income decline
and increasing costs of living, there are direct processes of the eviction of the slum popu-
lation as well, through a variety of methods of urban redevelopment. The very nature of
urbanization in the contemporary Global South is thus a stark contrast to what has been
proposed for inclusive cities. As per evidence, a fact sheet published by the Housing and
Land Rights Network in 2018 shows that government authorities in India forcefully
evicted more than 33,326 slum residents by demolishing 6943 homes in 2017 (Housing
and Land Rights Network 2018).
Further, not only are there instances of elite capture through direct and indirect
methods, as mentioned earlier, but also pressure on the government from the elite for aus-
terity and the channeling of government funds into non-basic luxury services like urban
beautification. The elite, either through price mechanisms or by direct (political or admin-
istrative) methods, dispossess the slum people; not only that, the elites pressurize the gov-
ernment not to spend on basic minimum infrastructure for the disadvantaged groups at
subsidized rates. The elites demand sophisticated services with high user charges and
restrict the government from providing subsidized support to the slum population. To
put pressure on the government, the elites bargain for tax rates and so on, on the pretext
of curbing wasteful expenditure by the government on the slum population.
Therefore, this study indicates a typical process of urban transformation which is pro-
ducing a new form of modern cultural space within the expanding urban premises. These
alluring modern spaces are demanded by the urban elites for the conspicuous consump-
tion of “commodities to value social relations” (as highlighted by Appadurai [1986] in The
Social Life of Things) and they are instinctively restricting the entry of the poor to such
474 S. GHOSH AND S. CHAKRABARTI

spaces by increasing the cost of its access. This deepens the socio-economic inequality
between the two classes—the rich and the poor—engendering displacement of the latter
to the unproductive spaces of the city (mostly at the growing city periphery). On the other
hand, the rising value of these elite spaces encourages corrupt political processes, even-
tually causing the eviction of slums. Such a process of urban transformation is spreading
like plague and has been jeopardizing every global consensus of inclusive city planning.

Notes
1. One interesting fact is that the number of slums in India decreased from 51,688 in 2002 to
33,510 in 2012 (NSSO 2003a, 2014), but the slum population increased from 42.5 million to
65.5 million during that period. During this period the slum/urban population share
increased from 14% to 17%. This may reveal that the existing slums are becoming bigger,
with a migrated population, and at the same time that small slums are amalgamating to
become bigger slums. This tendency may imply that the inner city slums are getting relo-
cated to the periphery, because larger slums may not be accommodated within the main
city area.
2. Vision Mumbai is a plan submitted by Mckinsey & Company for restructuring and redeve-
loping Mumbai to make it an international city (Government of Maharashtra 2004).
According to one researcher (Roy 2013), these kinds of plans and policy programmes
(such as RAY slum-free-city-planning) generally result in the eviction of many slum resi-
dents from inner-city spaces towards the fringes and precarious conditions.
3. Although this analysis deals with Indian states, it could have larger implications. Many of
the Indian states are very large, having distinctly different languages and various other
socio-economic traits. Many are even larger than most of the countries of the world. If
we rank Indian states in terms of total population and hypothetically consider them as inde-
pendent countries, they would take the following positions: Uttar Pradesh (5th), Maharash-
tra (11th), Bihar (12th), and West Bengal (13th).
4. Here, urban per capita Net State Domestic Product (NSDP) has been estimated by multiply-
ing per capita NSDP with the ratio of total urban to rural consumption expenditure.
5. Pucca is an Indian word. It means concrete/brick made.
6. Semi-pucca means a combination of concrete/brick and temporary materials.
7. Serviceable katcha defines houses made of mud but which are serviceable.
8. Open-pucca defines drains that are concrete but uncovered.
9. Very few outliers have been detected through box plots, and those data points have been
excluded during analysis both in the line diagrams and correlations.

Acknowledgements
The authors would like to thank Professor Cheng Enfu, Dr Wang Zhen, and the two anonymous
reviewers of this paper for their valuable comments, suggestions, and support. The authors are
also grateful to Harsha Tiwary, Ridhee Ghosh, and Pallabi Seth for help in editing and getting appro-
priate data. Lastly, the first author thanks Dr Sudakshina Gupta, the discussant of this paper presented
at the Department of Economics, University of Calcutta, in 2018. The usual disclaimer applies.

Disclosure Statement
No potential conflict of interest was reported by the author(s).
 INTERNATIONAL CRITICAL THOUGHT 475

Notes on Contributors
Somenath Ghosh is pursuing a PhD in Economics at the Department of Economics and Poli-
tics, Visva-Bharati University, India, under the supervision of Professor Saumya Chakrabarti.
He currently works as Assistant Manager at Lutheran World Services India. He has published
articles in titles such as Journal of Urban and Regional Analysis, Social Sciences and Huma-
nities Open, and others. He has also authored a chapter in a book entitled Social, Health,
and Environmental Infrastructures for Economic Growth published by IGI Global.
Saumya Chakrabarti is Professor of Economics at Visva-Bharati University, Santiniketan, India.
He has also taught at St. Xavier’s College, University of Calcutta, and Presidency University. He
has been a visiting fellow at Brown University, USA. At present, he is Head of the Department
of Economics and Politics, Visva-Bharati University. He was Honorary Director of the Agro-
Economic Research Centre (Government of India), Santiniketan. He has published in journals
such as the Cambridge Journal of Economics, Review of Radical Political Economics, Economic
and Labour Relations Review, Economic and Political Weekly, Indian Journal of Agricultural
Economics, and Indian Journal of Labour Economics, among others, and has written books pub-
lished by Prentice Hall and Oxford University Press. He has travelled across several countries of
the Global South and North and has regularly contributed to popular journals and vernacular
dailies.

References
Alonso, W. 1960. “A Theory of the Urban Land Market.” Papers in Regional Science 6: 149–157.
doi:10.1111/j.1435-5597.1960.tb01710.x.
Appadurai, A. 1986. The Social Life of Things: Commodities in Cultural Perspective. New York:
Cambridge University Press.
Bhattacharya, S., S. Saha, and S. Banerjee. 2016. “Income Inequality and the Quality of Public
Services: A Developing Country Perspective.” Journal of Development Economics 123 (C):
1–17. doi:10.1016/j.jdeveco.2016.07.003.
Bhattacharya, R., and K. Sanyal. 2011. “Bypassing the Squalor: New Town and Immaterial
Labour and Exclusion in Post-Colonial Urbanisation.” Economic and Political Weekly 46
(31): 41–48.
Buckley, R. M., M. Singh, and J. Kalarickal. 2007. “Strategizing Slum Improvement in India: A
Method to Monitor and Refocus Slum Development Programs.” Accessed March 20, 2021.
https://globalurban.org/GUDMag07Vol3Iss1/Buckley%20PDF.pdf.
Burgess, E. W. 1925. “The Growth of the City: An Introduction to a Research Project.” In The City,
edited by R. E. Park, E. W. Burgess, and R. D. McKenzie, 47–62. Chicago: The University of
Chicago Press.
Davis, M. 2004. “The Planets of Slums.” New Left Review 26: 259–263.
Doshi, S. 2012. “The Politics of the Evicted: Redevelopment, Subjectivity, and Difference in
Mumbai’s Slum Frontier.” Antipode 45 (4): 844–865.
FICCI (The Federation of Indian Chambers of Commerce and Industry). 2011. “Urban
Infrastructure in India.” Accessed January 9, 2021. http://ficci.in/spdocument/20122/Urban_
infra.pdf.
Frankenhoff, C. A. 1967. “Elements of an Economic Model for Slums in a Developing Economy.”
Economic Development and Cultural Change 16 (1): 27–36.
Glaeser, E. 2011. Triumph of the City: How Our Greatest Invention Makes Us Richer, Smarter,
Greener, Healthier and Happier. New York: Penguin.
Government of Maharashtra. 2004. “Vision Mumbai: Transforming Mumbai into a World-Class
City.” Accessed January 9, 2021. http://www.mumbaidp24seven.in/reference/taskforce.pdf.
Harris, J. R., and M. P. Todaro. 1970. “Migration, Unemployment and Development: A Two-
Sector Analysis.” The American Economic Review 60 (1): 126–142.
Harvey, D. 2008. “The Right to the City.” New Left Review 53: 23–40.
476 S. GHOSH AND S. CHAKRABARTI

Harvey, D. 2012. Rebel Cities: From the Right to the City to the Urban Revolution. London: Verso.
Housing and Land Rights Network. 2018. “Forced Evictions in India in 2017: An Alarming
National Crisis.” Accessed March 20, 2021. https://www.hlrn.org.in/documents/Forced_
Evictions_2017.pdf.
King, L. J. 2020. “Central Place Theory.” In Web Book of Regional Science, edited by R. Jackson, 19–
30. Morgantown: Regional Research Institute, West Virginia University.
Kolappan, B. 2016. “What Drives Urbanisation in Tamil Nadu.” The Hindu, April 1. Accessed
March 20, 2021. https://www.thehindu.com/news/national/tamil-nadu/what-drives-
urbanisation-in.
Kumar, R. 2018. “Leave Us Alone: India’s Villagers Rebel against Urbanisation.” The Guardian,
February 12. Accessed March 20, 2021. https://www.theguardian.com/cities/2018/feb/12/
india-villagers-rebel-urbanisation-gujarat.
Lebowitz, M. D. 1977. “A Critical Examination of Factorial Ecology and Social Area
Analysis for Epidemiological Research.” Journal of the Arizona Academy of Science 12 (2):
86–90.
NSSO (National Sample Survey Organisation). 2001. “Migration in India 1999–2000. NSS 55th
Round (July 1999–June 2000), Report No. 470 (55/10/8).” Accessed January 9, 2021. http://
mospi.nic.in/sites/default/files/publication_reports/470_final.pdf.
NSSO (National Sample Survey Organisation). 2003a. “Household Consumer Expenditure and
Employment-Unemployment Situation in India, 2001–2002. NSS 57th Round, Report No.
481(57/1.0/1).” Accessed January 9, 2021. http://mospi.nic.in/sites/default/files/publication_
reports/481_final.pdf.
NSSO (National Sample Survey Organisation). 2003b. “Condition of Urban Slums Salient Features
NSS 58th Round., NSSO Report no. 486(58/0.21/1), July–December.” Accessed January 9, 2021.
http://mospi.nic.in/sites/default/files/publication_reports/486_final.pdf.
NSSO (National Sample Survey Organisation). 2010. “Migration in India 2007–2008. NSS 64th
Round, Report No. 533(64/10.2/2).” Accessed January 9, 2021. http://mospi.nic.in/sites/
default/files/publication_reports/533_final.pdf.
NSSO (National Sample Survey Organisation). 2012. “Urban Slums in India NSS 69th Round,
NSSO Report no. 561(69/0.21/1), July–December.” Accessed January 9, 2021. http://mospi.
nic.in/sites/default/files/publication_reports/nss_report_561_19dec14.pdf.
NSSO (National Sample Survey Organisation). 2014. “Household Consumer of Various Goods
and Services in India, 2011–2012. NSS 68th Round, Report No. 558(68/1.0/2).” Accessed
January 9, 2021. http://mospi.nic.in/sites/default/files/publication_reports/Report_no558_
rou68_30june14.pdf.
Office of the Registrar General and Census Commissioner of India. 2001. “Census of India 2001
(Provisional) Slum Population in Million Plus Cities (Municipal Corporations): Part A.”
Accessed November 20, 2018. http://www.censusindia.gov.in/Census_Data_2001/Census_
Data_Online/Census.
Office of the Registrar General and Census Commissioner of India. 2011. “Primary Census
Abstract Data for Slum.” Accessed November 20, 2018. http://www.censusindia.gov.in/
2011census/population_enumeration.html.
Pal, P., and J. Ghosh. 2007. “Inequality in India: A Survey of Recent Trends.” Accessed January 9,
2021. https://www.un.org/esa/desa/papers/2007/wp45_2007.pdf.
Pannerselvam, R. 2013. Design and Analysis of Experiment. Delhi: PHI learning.
Piel, G. 1997. “The Urbanization of Poverty Worldwide.” Challenge 40 (1): 58–68.
Roy, A. 2013. “The Inclusive City: A New Paradigm of Urban Planning in India.” In
Inclusive Urban Planning: State of the Urban Poor Report 2013, edited by Ministry of
Housing and Urban Poverty Alleviation Government of India, 134–148. Oxford: Oxford
University Press.
Saharan, T., K. Pfeffer, and I. Baud. 2018. “Shifting Approaches to Slums in Chennai: Political
Coalitions, Policy Discourses and Practices.” Singapore Journal of Tropical Geography 39 (3):
454–471.
 INTERNATIONAL CRITICAL THOUGHT 477

Sato, Y. 2017. “Coping with the Threat of Evictions: Commercialisation of Slum Development,
Marginalisation of NGOs and Local Power Play in Ahmedabad.” Accessed January 9, 2021.
https://cept.ac.in/UserFiles/File/CUE/Working%20Papers/Revised%20New/33CUEWP33_Cop
ing%20with%20the%20Threat%20of%20Evictions%20_Commercialisation%20of%20Slum%
20Development,%20Marginalisation%20of%20NGOs%20and%20Local%20Power%20Play%
20in%20Ahmedabad.pdf.
Stokes, C. J. 1962. “A Theory of Slums.” Land Economics 38 (3): 187–197.
The Reserve Bank of India. 2018. “Handbook of Statistics on Indian Economy 2017–2018.”
Accessed January 9, 2021. https://rbi.org.in/SCRIPTS/AnnualPublications.aspx?head=
Handbook+of+Statistics+on+Indian+Economy.
The World Bank. 2009. The World Development Report: Reshaping Economic Geography.
Washington, DC: The World Bank.
Turner, J. 1969. “Uncontrolled Urban Settlement: Problem and Policies.” In The City in Newly
Developing Countries: Reading on Urbanism and Urbanization, edited by G. W. Breese, 507–
534. Englewood Cliff, NJ: Prentice Hall.
UN-Habitat. 2003. The Challenges of Slums. London: Earthscan.
Wacquant, L. 2010. “Designing Urban Seclusion in the Twenty-First Century.” Accessed January
9, 2021. https://loicwacquantorg.files.wordpress.com/2018/02/lw-2010-designing-urban-seclusi
on-in-the-twenty-first-century.pdf.

Appendices
Appendix 1: Two Images of Kolkata City: One Non-fringe Area of the City (Maa
Flyover) and One Fringe (Anandapur) Area of the City
The image of the area beside the Maa Flyover shows the eventual conversion of land which used to be
slum areas in 2002 into high rise areas in the year 2020 (see Figure A1. [a]). However, the image of
Anandapur highlights the gradual change from fallow to slum at the periphery of the city (see Figure
A1. [b]).

Figure A1. (a) Location map of Maa Flyover: A center part of Kolkata City.
478 S. GHOSH AND S. CHAKRABARTI

Figure A1. (b) Location map of Anandapur: A fringe part of Kolkata City.

Appendix 2: Model: Demand and Supply Side Interaction of Urban Services

We have tried to develop a model to illustrate the differences in the capacity of slum and non-slum
dwellers regarding the purchase of urban services. Through this model we can understand how
this difference may enhance the cost of living of the slum dwellers. In this model, we assume
two separate demand curves for urban services: one for the slum residents and another for the
non-slum residents. In Figure A1, illustrations of two demand curves have been presented (i.e.,
AA1 is the demand curve for non-slum residents, BB1 is the demand curve for slum residents,
and AC1 is a kinked joint demand curve), which shows that the demand of non-slum residents
is greater compared to the demand of slum residents. This is because the capacity of non-slum
residents to demand urban services is assumed to be higher compared to that of the slum residents.
Next, we illustrate the supply of composite urban services with the help of a supply curve SS1 in the
figure. It shows two equilibrium points for two separate demand curves. At the lower equilibrium
point, for the demand curve of slum residents, they can pay the reservation price P2 for only get-
ting the quantity of composite urban services Q2. On the other hand, at the upper equilibrium
point, the non-slum dwellers are ready to pay a price of P1 and get a higher amount of better-

Figure A2. Demand and supply of urban services.

 INTERNATIONAL CRITICAL THOUGHT 479

quality service Q1, where P1 > P2 and Q1 > Q2. So, Figure A2 reveals that the capacity of non-slum
dwellers is higher compared to the slum dwellers and that they can purchase proper urban
services at higher prices inside cities. By this phenomenon, it can be understood that the prices
of urban services will remain high and that the cost of living of slum dwellers will also be high.
As per the capacity of the slum residents, they cannot afford this high price within the city and,
hence, are compelled to move to border areas where they can enjoy only low cost, low-quality
urban services.
As soon as urban income inequality increases, the gaps between the two demand curves will rise
(when the demand curve shifts from AA1 to XX1). This indicates that an increase in the capacity of
non-slum dwellers compared to slum dwellers increases the demand and, hence, prices of urban
services inside the city. This, in turn, increases the cost of living of the slum dwellers and they are
forced to shift to fringe areas.