Recast(e)ing Inequality: Residential Segregation by Caste across City Size and Over Time in Urban India

Recast(e)ing Inequality: Residential Segregation by Caste across City Size and Over Time in Urban India Abstract This paper compares residential segregation by caste across small and large cities in India. It additionally explores how caste based segregation changes over time. To our knowledge, it presents the first systematic attempt to quantitatively examine the change in caste based segregation longitudinally, while offering a more nuanced understanding of how caste dynamics manifest differentially across city size classes in India. We focus on cities within two regions that have differing histories of caste politics: the northern state of Uttar Pradesh and the southern state of Karnataka. In both states, we find strong evidence for spatial inequality by caste across city size, with smaller cites consistently having higher levels of residential segregation by caste compared to larger cities in the same district. In Karnataka, we also find that residential segregation by caste decreases over time in majority of the cities whereas the results are quite mixed in Uttar Pradesh. Study Motivation and Statement of Problem India s transformation from a majority rural to an increasingly urban society has gained pace over the last few years. With 410 million urban dwellers, today India has the second largest urban population globally as compared to 263 million in United States (United Nations World Urbanization Prospects, 2014). Over the next three decades, India s urban population is expected to almost double, 1 making it the single most defining characteristic of India s development trajectory and with profound implications for existing and emerging forms of social inequality and stratification. Against the backdrop of India s rapid urbanization, this paper examines how enduring patterns of caste-based discrimination have evolved in an increasingly urban context. Caste system is a longstanding form of social stratification in India, but how caste shapes social and economic life in India from where people live to their educational, employment, social 1 Approximately 400 million more Indians are projected to join the urban population by 2050 (United Nations 2014) 1

network, and marriage opportunities evolves over time and is influenced by the changing social, political and economic context. This paper seeks to understand how residential segregation by caste varies across small and large city sizes in India. It further takes a longitudinal perspective to examine how residential segregation by caste has improved or worsened over time using data from Indian Census 2001 and 2011. In doing so, it marks the first attempt within social science research on caste in India to quantitatively examine the change in caste based segregation over time, while providing a more nuanced understanding of how this durable system of social inequality manifests differentially across city size categories. By examining patterns of residential segregation, this study takes a spatial approach to an examination of social inequality in urban India and builds upon two types of scholarship. First, it engages with the growing body of scholarship on caste-based segregation, including studies by Mehta (1968, 1969), Dupont (2004), Vithayathil and Singh (2012), Singh (2014), and Sidwani (2015) in the contemporary Indian context. Second, it borrows from more than a century of urban social research in the United States, which demonstrates that residential segregation serves as a good proxy for social stratification, and residential location in a city shapes individual as well as group opportunities and outcomes. 2 Theoretical Framework Urban theory suggests that as individuals and groups adapt to city life, traditional forms of social organization such as caste, family, and religion weaken and reorganize. The political, economic and social landscapes of cities are theorized to transform social relations and cause new forms of social organization and control to emerge. One reason for this is the increased proximity and interaction of different social groups in cities, which can serve to strengthen weak ties (or bridging networks), and attenuate stronger kinship ties (or bonding networks). However, the limited research currently available on caste in contemporary urban India finds that caste continues to structure lives, opportunities and outcomes, 3 while some of the extreme social indignities related to caste are becoming less prevalent or taking on new forms. For example, Thorat and Newman (2010) find that caste plays a dominant role in hiring decisions in the most market-driven and technologically-advanced sectors of corporate, urban India. Ethnographic 2 For example: Kim and White 2010; Iceland and Scopilliti 2008; Zorlu and Mulder 2008; White, Kim, and Glick 2005; Adelman 2004; Logan, Zhang, and Alba 2002; Massey and Denton 1993, Massy and Denton 1985. 3 For example: Munshi and Rosenzweig (2006); Thorat and Attewell (2007); Thorat and Newman (2010) 2

work on rural-urban migrants from the state of Jharkand concludes that the decision to move to the city for individuals often includes an expectation of liberation from rigid caste structures in villages (Shah 2006). Despite this perception of liberation, caste hierarchies and identities experienced by migrants do not necessarily disappear within cities, but are rather re-organized and re-produced as new collective identities that emerge along reformulated caste lines, unique to the urban context (Breman, 1995, 2003). These new collectivities engage in urban politics in unique ways, reshaping the nature of social and political life in contemporary urban India while also contributing to the patterns of social stratification within Indian cities in the future. Detailed ethnographic studies during the colonial period shape much of our global understanding of how caste structures social relations in India, as well as the extremely localized and hierarchical nature of the caste system. 4 The localized nature of caste system highlighted by these studies is an importance aspect, which not only has theoretical and analytical implications for the caste studies scholars but also has methodological implications that we return to later in this paper. This long-standing social science tradition of caste studies offers an important basis for understanding caste-based social structures in contemporary India. At the same time, this scholarship tends to mask mobility and social change within caste system and is deeply embedded in nineteenth century race theories (Bayly 1999). Breaking away from colonial constructions of caste, more recent scholarship on how caste operates in post-colonial India argues that caste categories and their saliency are fluid and vary situationally (Rudolph and Rudolph 1967; Kothari 1970; Bayly 1999; Shah 2004). Such contemporary scholarship emphasizes that caste operates in different ways in different spheres of life while still structuring the daily experiences of hundreds of millions of Indians. In addition, caste interacts with gender, class and religion to limit and shape individual life opportunities. All evidence from these social scientists underscores the point that caste matters and continues to structure social stratification. But understanding of how caste and other axes of social, political, and economic inequality transform, and are transformed by, urban spaces and urbanization is a newly emerging field of research that this paper contributes to. The focus of this study on patterns of case-based residential segregation is motivated by the rich body of empirical research on spatial segregation in cities of United States. This body of 4 For example: Hutton 1961; Srinivas 1962, 1966; Rudolph and Rudolph 1967; Ghurye 1969; Kothari 1970; Gupta 1991; Dupont 1990; Bandyopadhyay 1990; Bayly 1999; Weiner 2001; Thorat and Newman 2010. 3

social science research has unarguably established that social stratification manifests itself in physical spaces within the urban environments, particularly within residential locations (examples include, Kim and White 2010; Iceland and Scopilliti 2008; Zorlu and Mulder 2008; White, Kim, and Glick 2005; Logan, Zhang, and Alba 2002; Alba et al 2000, Massey and Denton 1985, among several others); and that spatial segregation within a city can be a source of disadvantage for the segregated (Massey and Denton, 1993; Wilson, 1987, Portes and Rumbaut 1996, 2001; Portes and Zhou, 1993; Alba, 2005). We also borrow established quantitative segregation measurement tools from this body of empirical research, which have been less utilized within the Indian scholarship. Further, unlike the United States, few studies on urban inequality in India have examined patterns of segregation across a multiple city sizes, with particularly little attention to how caste inequalities operate in small Indian towns. Based on preliminary insights by Desai and Dubey (2011), we expect higher levels of residential segregation in smaller cities where caste relations and patterns of inequality are more likely to be similar to developed village settings, and residential patterns remain highly structured by caste identities and relations. In contrast, in larger cities, we expect that the intermixing of diverse linguistic, ethnic and regional identities to begin the process of reconfiguring casting relations, leading to comparatively lower levels of residential segregation by caste. In addition, the structure and politics of caste varies across regions of India and we expect these region-specific histories to influence patterns of residential segregation by caste. With this in mind, we examine towns and cities within two large states with differing histories of caste politics: the northern state of Uttar Pradesh and the southern state of Karnataka. Context With a population of over 200 million people, 5 Uttar Pradesh (UP) is the largest state in the country and found within north India s Hindi speaking belt. Karnataka is located in south India and has a population of 64 million. The two states are located in regions with different caste profiles and histories of caste politics. In Uttar Pradesh, like much of the Hindi Belt, the caste system more closely resembles the varna model, with four castes (i.e., Brahmins, Kshatriyas, Vaishyas and Shudras) and the untouchables (or outcastes, later termed as Dalits that literally 5 Uttar Pradesh has a population slightly larger than Brazil with 1/35 th the geographic area. 4

translates as the downtrodden ). 6 Connecting the categories with the categories of the Indian state, Brahmins, Kshatriyas and Vaishyas, or twice-born castes are considered the upper castes; Shudras roughly map onto the category of Other Backward Classes; and the untouchables corresponds with the state category of Scheduled Castes. In UP, the percentage of upper castes is estimated around 19 percent, while the remaining population of the state is 41.5 percent Other Backward Classes, 21 percent Dalits, and 18.5 percent Muslims are 18.5 (Verma 2014). 7 Brahmins have historically dominated politics in Uttar Pradesh and the neighboring state of Bihar, although since the 1990s there has been a dramatic shift in political power as OBCs politicians and political parties have gained strong hold in both states. Against this backdrop, anthropologist Jeffrey Witsoe (2013: 12) discusses the role of caste in everyday life: The pervasive influence of caste networks explains why in getting a job or a loan, interacting with the police, dealing with mafia figures, interacting with the police, dealing with mafia figures, interacting with or negotiating a bribe with a government official, or even renting a flat or commercial space in the capital, caste mattered. It is no coincidence that government offices, colleges, shopping complexes, and criminal networks in Bihar were often populated with people sharing similar caste background. Witsoe s observations of how caste networks play a role in shaping where people live and work emerge from Bihar. But given its proximity to UP and coterminous trajectory of politics entrenched in caste dynamics, we anticipate similar patterns occurring in Uttar Pradesh and seek to systematically understand how people living in small towns and bigger cities are affected by their caste identity. In contrast to India s Hindi-belt, in much of the south including Karnataka, the Khatriyas and Vaishyas in the varna system are missing and the overall percentage of the upper castes is much smaller. In Karnataka, upper castes make up only 3.5 percent of the state population (compared to nearly 20 percent in Uttar Pradesh). Lingayats (comprising 15.3 percent of state population) and Vokkaligas (10.8 percent of state population) are Shudras in the varna model and classified as OBCs by the state, but are also considered dominant castes locally because of 6 While social hierarchies by caste predated the colonial period, the introduction of the varna view of caste into the practices of state administration occurred during the colonial period (Dirks 2001). 7 The last published census with caste data for all the major castes is from 1931. Researchers have also estimated the proportion of upper castes using NSSO data based on using leftover 5

their political and economic power (Shastri 2011: 431, 434). 8 A much earlier non-brahmin movement in the south meant that as early as 1921 affirmative action benefits were created for non-brahmins. The Lingayats and Vokkaligas, which were the main land-owning communities among the non-brahmins, organized and benefited from the non-brahmin movement. They have consolidated their political and economic power over the past 100 years and refused multiple efforts to remove them from the category of OBC. Dalits or Scheduled Castes in Karnataka continue to face considerable discrimination. Study Design and Methods We set out to explore three inter-related questions: I. First, does the degree of residential segregation by caste vary across city size, within each of the two particular ethno-linguistic regions of India? II. Second, are the same cities becoming more or less segregated over time? III. Third, based on the findings from the previous two questions, we explore if and how the diversity of caste dynamics and socio-political contexts across regions relate to different patterns of spatial inequality? Findings and preliminary discussion based on the first two questions is explored within this version of the paper whereas, detailed examination of the third question will be done in the final version of the paper. Data Overview: Overall, we answer these questions by examining residential segregation by caste across two city-size categories in the states of Uttar Pradesh and Karnataka in 2001 and 2011. We compare Class I cities (with a population of 100,000-1 million) and smaller towns (with a population of 30,000-50,000 population). We use ward level caste data from the 2011 and 2001 Decennial Censuses of India. The Census collects caste data for Scheduled Castes and Scheduled Tribes, while the remaining populations are categorized as other. According to the 2011 Census, Scheduled Castes accounted for 16.6% and Scheduled Tribes 8.6% of India s population 8.Lingayats consider themselves an independent religion from Hinduism, although they are classified as Hindu by the state. 6

(Chandramouli 2013). These caste data allow us to examine if the line of untouchability, which physically separated those castes that were literally considered untouchable by all other castes and therefore historically patterned where individuals and groups lived, continues to shape patterns of residence in urban spaces. Our measure for residential segregation by caste is the index of dissimilarity (D) and we compute its value based on splitting the population into two groups: Scheduled Caste and Scheduled Tribes versus all others (non-scheduled Castes and non- Scheduled Tribes). For each city, our geographic area of focus is bounded by the city limits, or the region in which the municipal or town corporations are responsible for providing services. Below we outline the methods used for each question. Question I: To answer the first question, we explore whether the degree of residential segregation by caste varies across city size. We do this by comparing the dissimilarity index for caste for small and large city pairs within the same district in Uttar Pradesh and Karnataka. We compare between the two city-size categories within a particular district to account for varying socio-political regimes within the same state. 9 As mentioned before, detailed ethnographic studies have highlighted the extremely localized nature of caste dynamics and how it structures social relationships. Districts serve as a good geographical proxy to control for the local context in this sense, given that that their origins were on the basis of historical and cultural similarities, alongside the sub-geographies present within states. We use only 2011 Decennial Census data in this portion of the analysis. In Karnataka, out of a total of 30 districts in 2011, we include 21 districts and 67 cities in these districts in our analysis, based on the eligibility criteria of city-size pairs within the same district (see details in the footnote). 10 Table 1 provides summary statistics 9 Karnataka, for example, was created by combining the state of Mysore with districts from the Kannada speaking regions of Madras, Hyderabad and Bombay. During the colonial period, these regions experienced distinct political histories. Since 1956, all of the districts within current day Karnataka have been governed by the same state administration, while the district level administration varies. 10 In Karnataka there are 30 districts in 2011; 19 of these districts have at least one city in both size categories and are therefore included in the analysis. Three districts (i.e., Chamarajanagar, Ramanagara and Yadgir Districts) are excluded from the analysis because they do not have cities in either size category. Eight districts were initially excluded because they have cities in one size category, but not the other (i.e., Bangalore, Kodagu, Uttara Kannada, Chikkaballapura and Bangalore Rural districts do not have a city in the 100,000 to 1 million category; Chitradurga, Dharwad, and Koppal districts do not have a city in the 30,000 to 50,000 size category). We then relaxed the conditions among the three districts with a large city but not a city in the 30,000 to 50,000 size category and included towns with a population between 25,000 and 30,000; this change allowed us to add 2 additional districts to the analysis (i.e., Dharward and Chitradurga). 7

for the Karnataka districts included in our analysis. In Uttar Pradesh, out of a total of 71 districts in 2011, we include 41 districts and 117 cities in these districts in our analysis (see details in the footnote). 11 Table 2 provides summary statistics for the districts in Uttar Pradesh included in our analysis. Question II: Second, we study how levels of segregation by caste are changing over time in these cities. To create a pool of eligible cities for this longitudinal analysis, we use 2001 and 2011 census data to determine a subset of cities from the first portion of the analysis that have a consistent number of wards in 2001 and 2011. In numerous instances the number of wards changes because an existing town has undergone a change in its designation, and/or because of a significant change in the city s population, the official area of the city expands. Often, the boundary changes are not just based on splitting of wards in a way that they can be easily grouped together over the two time periods but rather, the changes signal a more complex reorganization of ward boundaries, making it difficult to create integrity of ward boundaries over time. As such, we exclude these cities from our analysis. For Karnataka, we include 51 cities in this analysis and exclude 16 cities due to changes in the number of wards over time. The 16 cities in Karnataka excluded from our analysis are distributed across 14 districts (from a total of 21 districts included in the first analysis). In Uttar Pradesh, we include 67 cities in this analysis and exclude 50 cities due to changes in the number of wards. The cities in Uttar Pradesh excluded from our analysis are distributed across 35 (from a total of 41 districts included in part 1). For each city in this portion of the analysis, we compare the index of dissimilarity for caste in 2001 and 2011. 11 There are 71 districts in Uttar Pradesh in 2011. Thirty-four of these districts have at least one city in both size categories and are therefore included in the analysis. Nine districts are excluded from the analysis because they do not have cities in either size category and 28 districts were initially excluded because they have cities in one size category, but not the other. We then relaxed the conditions for the 12 districts that had a large city but not a small one to include small cities with a population between 25,000 and 30,000; this change allowed us to add 5 additional districts to the analysis. We then relaxed the conditions for the 16 districts that had a small town but not a large city to include large cities with a population between 1 and 2 million; this change allowed us to add 2 additional districts to the analysis. 8

Table 1: Selected Districts for Karnataka (2011) District Name No. of % Urban % SC % ST Literacy Total Pop. Households population Pop. Pop. Rate Bagalkot 361,149 1,889,752 31.6 16.9 5.1 68.8 Belgaum 983,854 4,779,661 25.3 12.1 6.2 73.5 Bellary 489,118 2,452,595 37.5 21.1 18.4 67.4 Bidar 319,937 1,703,300 25.0 23.5 13.8 70.5 Bijapur 408,806 2,177,331 23.1 20.3 1.8 67.1 Chikmagalur 276,085 1,137,961 21.0 22.3 4.0 79.2 Chitradurga 357,003 1,659,456 19.9 23.4 18.2 73.7 Dakshina Kannada 439,733 2,089,649 47.7 7.1 3.9 88.6 Davanagere 410,176 1,945,497 32.3 20.2 12.0 75.7 Dharwad 382,700 1,847,023 56.8 9.6 4.7 80.0 Gadag 219,096 1,064,570 35.6 16.4 5.8 75.1 Gulbarga 471,601 2,566,326 32.6 25.3 2.5 64.9 Hassan 433,453 1,776,421 21.2 19.4 1.8 76.1 Haveri 330,414 1,597,668 22.3 13.8 8.8 77.4 Kolar 333,348 1,536,401 31.2 30.3 5.1 74.4 Mandya 428,625 1,805,769 17.1 14.7 1.2 70.4 Mysore 700,968 3,001,127 41.5 17.9 11.1 72.8 Raichur 363,853 1,928,812 25.4 20.8 19.0 59.6 Shimoga 406,816 1,752,753 35.6 17.6 3.7 80.4 Tumkur 640,081 2,678,980 22.4 18.9 7.8 75.1 Udupi 253,078 1,177,361 28.4 6.4 4.5 86.2 Table 2: Selected Districts for Uttar Pradesh (2011) District Name No. of Households Total Pop. % Urban Pop. % SC Pop. % ST Pop. Literacy Rate Agra 710,566 4,418,797 45.8 22.4 0.2 71.6 Aligarh 611,371 3,673,889 33.1 20.6 0.0 67.5 Ambedkar Nagar 368,728 2,397,888 11.7 24.7 0.0 72.2 Baghpat 209,916 1,303,048 21.1 11.4 0.0 72.0 Bahraich 603,754 3,487,731 8.1 14.6 0.3 49.4 Ballia 480,268 3,239,774 9.4 15.3 3.4 70.9 Banda 319,963 1,799,410 15.3 21.6 0.0 66.7 Bareilly 756,784 4,448,359 35.3 12.5 0.1 58.5 Budaun 615,776 3,681,896 17.5 16.9 0.0 51.3 Bulandshahr 587,529 3,499,171 24.8 20.7 0.0 68.9 Deoria 468,346 3,100,946 10.2 15.1 3.5 71.1 Etah 290,683 1,774,480 15.1 15.8 0.0 70.8 Etawah 277,527 1,581,810 23.2 24.5 0.0 78.4 Faizabad 423,375 2,470,996 13.8 22.5 0.0 68.7 9

Farrukhabad 314,144 1,885,204 22.1 16.6 0.0 69.0 Fatehpur 472,238 2,632,733 12.2 24.7 0.0 67.4 Firozabad 414,266 2,498,156 33.4 19.0 0.1 71.9 Ghaziabad 850,676 4,681,645 67.6 16.5 0.1 78.1 Ghazipur 546,664 3,620,268 7.6 20.1 0.8 71.8 Gonda 541,247 3,433,919 6.6 15.5 0.0 58.7 Gorakhpur 692,960 4,440,895 18.8 21.1 0.4 70.8 Hardoi 730,442 4,092,845 13.2 31.1 0.0 64.6 Jaunpur 663,513 4,494,204 7.7 22.0 0.1 71.5 Jhansi 367,779 1,998,603 41.7 28.1 0.2 75.0 Jyotiba Phule Nagar 314,401 1,840,221 24.9 17.3 0.0 63.8 Kanshiram Nagar 237,903 1,436,719 20.1 17.7 0.0 61.0 Kheri 745,077 4,021,243 11.5 26.4 1.3 60.6 Mahamaya Nagar 260,860 1,564,708 21.3 24.8 0.0 71.6 Mainpuri 313,690 1,868,529 15.4 19.7 0.0 76.0 Mau 324,424 2,205,968 22.6 21.5 1.0 73.1 Mirzapur 394,925 2,496,970 13.9 26.5 0.8 68.5 Moradabad 796,170 4,772,006 33.0 15.3 0.0 56.8 Muzaffarnagar 676,642 4,143,512 28.8 13.5 0.0 69.1 Pilibhit 362,573 2,031,007 17.3 16.4 0.1 61.5 Rae Bareli 619,707 3,405,559 9.0 30.3 0.1 67.3 Rampur 393,736 2,335,819 25.2 13.2 0.0 53.3 Saharanpur 597,656 3,466,382 30.8 22.1 0.0 70.5 Shahjahanpur 527,501 3,006,538 19.8 17.7 0.0 59.5 Sitapur 801,764 4,483,992 11.8 32.3 0.0 61.1 Unnao 588,533 3,108,367 17.1 30.5 0.1 66.4 Varanasi 560,162 3,676,841 43.4 13.2 0.8 75.6 Findings We begin by examining patterns of residential segregation by caste across city size in Karnataka and Uttar Pradesh. In Karnataka, we find strong evidence that spatial inequality by caste varies by city size (Table 3). In general, the smaller cites have higher dissimilarity value compared to larger cities in the same district. Out of the 21 districts in our analysis, we find that city-pairs in 17 districts follow the pattern where residential segregation by caste is higher in small cities compared to larger ones. The four districts that are the exception to this pattern are Gulbarga, Kolar, Raichur and Chitraduga districts. In Karnataka, across all the districts included in the analysis, the median D for small cities is 0.45 while for larger cities the median D is 0.33. 10

In Uttar Pradesh, we also find strong evidence for spatial segregation by caste varying between the city size categories (Table 4). Out of the 41 districts in our analysis, we find that 40 districts follow the pattern where residential segregation by caste is higher in small cities compared to larger cities. The only district that deviates from this pattern is Mirzapur. In Uttar Pradesh, across all the districts included in the analysis, the median D for small cities is 0.56 and for larger cities it is 0.40. Therefore across both states, we consistently find residential segregation by caste is greater in small cities compared to big cities. The second portion of the analysis explores how spatial inequality is changing over time. In the case of Karnataka, we consistently find that the index of dissimilarity has decreased across almost all of the cities in our analysis (Table 5). Between 2001 and 2011, we find that in 46 out of 51 cities included in our analysis the level of residential segregation by caste decreases over time. For the remaining 5 cities, we find that residential segregation by caste increases over time. For the small cities in Karnataka included in our analysis, we find that the median D is.496 in 2001 and decreases to.467 in 2011. For the large cities included in our analysis, we find that the median D is.364 in 2001 and decreases to.323 in 2011. In contrast, in the case of Uttar Pradesh, we find mixed results (Table 6). In some cases, there is a decrease, but in many instances the change is negligible or in several cases there is actually an increase in patterns of segregation over time. Out of 67 cities included in the analysis, there is a decrease in index of dissimilarity for 25 cities, while for 27 cities there is an increase in the index of dissimilarity between 2001 and 2011. For 14 cities there is no/ minimal change over time. For the small cities in Uttar Pradesh included in our analysis, we find that the median D is.568 in 2001 and decreases to.552 in 2011. For the large cities included in our analysis, we find that the median D is.393 in 2001 and increases to.408 in 2011. Preliminary Discussion and Future Directions: Overall, the findings from these analyses underscore that caste-based residential segregation in modern India is not only a rural phenomenon or markings of a bygone era. Our results show that segregation based on caste continues to persist in Indian cities but at the same time, there are important differences in levels of segregation across the two city-size categories examined in this paper. As mentioned earlier in the paper, our decision to examine caste segregation in small and 11

large city sizes was motivated by preliminary findings by Desai and Dubey (2011) based on nationally representative survey data that indicate smaller cities show higher levels of income inequality faced by Dalits (Scheduled Castes) and Adivasis (Scheduled Tribes). Our findings confirm that this relationship also manifests itself in urban physical space in the form of residential segregation. A key reason for higher levels of caste-based residential segregation in smaller cities may be due to their earlier stage of urbanization where social relations and intercaste mixing still remain highly structured by caste identities that are similar to the dynamic more prevalent in rural India. In contrast, as cities grow and become denser, there are greater levels of interaction among different social groups, leading to intermixing of diverse linguistic, ethnic and regional identities. As a consequence, caste relations may be reconfigured to promote comparatively lower levels of residential segregation by caste. With respect to the analysis over time, our findings point to differences in results across the two socio-political regions of Karnataka and Uttar Pradesh. As highlighted in the context, the structure and politics of caste varies across regions of India and we expect these region-specific histories to influence patterns of residential segregation by caste. While a full discussion of how the socio-economic and political context has influenced caste-based segregation will be presented in the final version of the paper, a few observations are presented here. The consistent decline in the residential segregation by caste across both city-size categories in Karnataka between 2001 and 2011 has been accompanied by a fast pace of urbanization in the state. Between the two Census rounds, Karnataka has become the most urbanized state in the country with more than 38% of its population living in urban areas. In contrast, Uttar Pradesh s urban population has grown at a much slower pace in the same period with only 22% of the state s population living in urban areas. Uttar Pradesh has in fact slipped from 18 th position in 2001 to 23rd in 2011 in the level of urbanization among all Indian states. In addition, Uttar Pradesh continues to suffer from one of the lowest levels of literacy rates in India. Lower levels of literacy in the state may also retard the pace of changes in traditional caste dynamics. Finally, as highlighted in the Context section, Uttar Pradesh politics and Karnataka politics have both been deeply entrenched in caste relations. Yet, the politics of caste has not translated into the same degree of reduction of caste based residential segregation in Uttar Pradesh, when compared to Karnataka. The levels of residential segregation by caste for both small and large city-size 12

categories in Karnataka are smaller than the levels found in Uttar Pradesh. In addition, residential segregation by caste is consistently decreasing over time across both city size categories in Karnataka, while residential segregation by caste increases between 2001 and 2010 in the large city size category in Uttar Pradesh. The final version of this paper will delve deeper into the explanations for differences in the findings in the two states. In addition, we will also add two other Indian states for comparison an eastern state (West Bengal) and a western state (Gujrat) for comparison. Finally, the authors are currently negotiating access to GIS ward boundaries of selected pairs of cities included in this analysis in order to visually demonstrate the patterns of Scheduled Caste settlements within city-wards. These will be added to the final version of the paper. 13

Table 3: Caste Dissimilarity by District for Large and Small Cities in Karnataka (2011 Census Data) Mean D, Mean D, Karnataka Large Cities Small Cities Large Small District (100,000-1 million pop.) (30,000-50,000 pop.) Cities Cities Bagalkot Bagalkot Mahalingpur, Guledgudda, Badami 0.305 0.408 Belgaum Belgaum Bail Hongal, Athni, Chikodi, Ramdurg, Sankeshwar 0.327 0.496 Bellary Bellary, Hospet Kampli, Sandur 0.272 0.414 Bidar Bidar Homnabad, Bhalki 0.324 0.433 Bijapur Bijapur (formerly Vijayapur) Indi, Sindgi, Muddebihal, Basavana Bagevadi, Talikota 0.305 0.469 Chikmagalur Chikmagalur Tarikere, Kadur 0.348 0.361 Chitradurga Chitradurga Hosdurga 0.256 0.361 Dakshina Mangalore Bantval Kannada 0.268 0.303 Davanagere Davanagere Harapanahalli 0.342 0.473 Dharward Hubli-Dharwad Annigeri Gadag Gadag-Betigeri Lakshmeshwar, Nargund, Gajendragarh 0.352 0.508 Gulbarga Gulbarga Shahabad, Aland, Sedam, Chitapur, [Wadi] 0.497 0.491 Hassan Hassan Channarayapatna 0.303 0.336 Haveri Ranibennur Savanur, Byadgi 0.344 0.506 Kolar Kolar, Robertson Pet Bangarapet, Malur 0.485 0.327 Mandya Mandya Malavalli 0.392 0.708 Mysore Mysore Krishnarajanagara 0.321 0.505 Raichur Raichur Manvi, Lingsugur 0.407 0.307 Shimoga Shimoga, Bhadravati Shikarpur 0.238 0.299 Tumkur Tumkur Kunigal 0.235 0.484 Udupi Udupi Kundapura 0.330 0.431

Table 4: Caste Dissimilarity by District for Large and Small Cities in Uttar Pradesh (2011 Census Data) Uttar Pradesh District Large Cities (100,000-1 million pop.) Small Cities (30,000-50,000 pop.) Agra Agra (M Corp.) Fatehpur Sikri (NPP), Shamsabad (NPP) 0.36 0.60 Aligarh Aligarh (M Corp.) Khair (NPP) 0.52 0.61 Ambedkar Nagar Akbarpur (NPP) Jalalpur (NPP) 0.45 0.67 Baghpat Baraut (NPP) Khekada (NP) 0.60 0.84 Bahraich Bahraich (NPP) Nanpara (NPP) 0.43 0.60 Ballia Ballia (NPP) Rasra (NPP) 0.32 0.56 Banda Banda (NPP + OG) Atarra (NPP) 0.29 0.43 Bareilly Bareilly (M Corp. + OG) Nawabganj (NPP) 0.38 0.58 Budaun Budaun (NPP) Islamnagar (NP), Bisauli (NPP), Kakrala (NPP) 0.40 0.62 Bulandshahr Bulandshahr (NPP + OG), Khurja (NPP) Siana (NPP), Dibai (NPP), Shikarpur (NPP) 0.40 0.51 Deoria Deoria (NPP) Rudrapur (NP), Gaura Barhaj (NPP) 0.31 0.42 Etah Etah (NPP) Jalesar (NPP) 0.45 0.54 Etawah Etawah (NPP) Bharthana (NPP) 0.29 0.43 Faizabad Faizabad (NPP) Rudauli (NPP) 0.32 0.58 Farrukhabad Farrukhabad-cum-Fatehgarh (NPP) Kaimganj (NPP) 0.36 0.42 Fatehpur Fatehpur (NPP) Bindki (NPP), Khaga (NP) 0.26 0.44 Firozabad Firozabad (NPP),Shikohabad (NPP) Sirsaganj (NPP) 0.45 0.53 Ghaziabad Modinagar (NPP), Loni (NPP), Hapur (NPP) Dasna (NP), Garhmukhteshwar (NPP) 0.45 0.49 Ghazipur Ghazipur (NPP + OG) Mohammadabad (NPP), Zamania (NPP) 0.32 0.51 Gonda Gonda (NPP) Colonelganj (NPP) 0.34 0.39 Gorakhpur Gorakhpur (M Corp.) Sahjanwan (NP) 0.28 0.34 Hardoi Hardoi (NPP) Pihani (NPP), Mallawan (NPP) 0.38 0.45 Jaunpur Jaunpur (NPP) Machhlishahr (NP), Shahganj (NPP) 0.32 0.65 Jhansi Jhansi (M Corp.) Barua Sagar (NPP), Gursarai (NPP) 0.33 0.55 Jyotiba Phule Nagar Amroha (NPP) Dhanaura (NPP), Bachhraon (NPP), Naugawan Sadat (NP) 0.52 0.69 Kanshiram Nagar Kasganj (NPP) Ganj Dundawara (NPP) 0.48 0.60 Kheri Lakhimpur (NPP) Paliya Kalan (NPP), Mohammadi (NPP), Kheri (NP) 0.39 0.54 Mahamaya Nagar Hathras (NPP) Sikandrarao (NPP), Sadabad (NP) 0.44 0.58 Mean D, Large Cities Mean D, Small Cities 14

Mainpuri Mainpuri (NPP + OG) Bhogaon (NP) 0.34 0.51 Mau Maunath Bhanjan (NPP) Ghosi (NP), Kopaganj (NP), Muhammadabad (NP) 0.50 0.55 Mirzapur Mirzapur-cum-Vindhyachal (NPP) Chunar (NPP) 0.39 0.31 Moradabad (M Corp.), Sambhal (NPP), Thakurdwara (NPP), Bhojpur Dharampur (NP), Bilari (NPP), Moradabad Chandausi (NPP) Bahjoi (NPP) 0.53 0.68 Muzaffarnagar Shamli (NPP), Muzaffarnagar (NPP) Kandhla (NPP), Thana Bhawan (NP), Budhana (NP + OG) 0.41 0.61 Pilibhit Pilibhit (NPP) Puranpur (NPP) 0.49 0.65 Rae Bareli Rae Bareli (NPP) Jais (NPP) 0.26 0.51 Suar (NPP), Tanda (NPP), Bilaspur (NPP), Shahabad (NP), Milak Rampur Rampur (NPP) (NPP) 0.42 0.63 Saharanpur Saharanpur (M Corp.) Rampur Maniharan (NP) 0.47 0.66 Shahjahanpur Shahjahanpur (NPP) Katra (NP), Jalalabad (NPP) 0.24 0.54 Sitapur Sitapur (NPP) Khairabad (NPP) 0.32 0.63 Unnao Unnao (NPP) Bangarmau (NPP) 0.25 0.56 Varanasi Varanasi (M Corp.) Ramnagar (NPP) 0.35 0.38 15

Table 5: Comparing Dissimilarity Index (D) in 2001 and 2011 for Select Cities in Karnataka Total pop. in 2011 Prop. SC/ST in 2011 Median ward pop. in 2001 Median ward pop. in 2011 D D City District 2001 2011 Chikodi Belgaum 38307 0.22 1427 1448 0.44 0.39 Athni Belgaum 45858 0.23 1706 1650 0.72 0.60 Sankeshwar Belgaum 34637 0.19 1414 1573 0.60 0.54 Belgaum Belgaum 488157 0.11 6891 7230 0.40 0.33 Saundatti-Yellamma Belgaum 41215 0.09 1659 1711 0.46 0.49 Ramdurg Belgaum 33338 0.16 1384 1371 0.57 0.53 Mahalingpur Bagalkot 36055 0.17 1342 1561 0.47 0.46 Guledgudda Bagalkot 33382 0.12 1478 1396 0.47 0.41 Muddebihal Bijapur 34217 0.16 1227 1212 0.51 0.44 Bijapur Bijapur 327427 0.16 7044 8687 0.33 0.31 Indi Bijapur 38217 0.24 1369 1393 0.55 0.50 Sindgi Bijapur 37226 0.18 1206 1591 0.50 0.39 Talikota Bijapur 31693 0.11 1139 1037 0.51 0.47 Bhalki Bidar 40333 0.31 1526 1681 0.48 0.46 Bidar Bidar 214373 0.19 4939 5733 0.36 0.32 Raichur Raichur 234073 0.24 5926 6135 0.46 0.41 Manvi Raichur 46465 0.27 1635 1894 0.35 0.33 Nargund Gadag 36291 0.14 1416 1553 0.54 0.50 Gajendragarh Gadag 32359 0.23 1227 1231 0.65 0.56 Gadag-Betigeri Gadag 172612 0.14 4428 4042 0.43 0.35 Lakshmeshwar Gadag 36754 0.14 1453 1413 0.61 0.46 Hubli-Dharwad Dharwad 943788 0.13 11734 12892 0.29 0.26 Annigeri Dharwad 28267 0.14 1118 1226 0.46 0.36 Savanur Haveri 40567 0.11 1546 1588 0.57 0.54 Byadgi Haveri 30014 0.22 1116 1196 0.45 0.47 Ranibennur Haveri 106406 0.10 2891 3454 0.40 0.34 Hospet Bellary 206167 0.30 4693 5744 0.29 0.31 Kampli Bellary 39307 0.29 1538 1568 0.46 0.48 Bellary Bellary 410445 0.26 9050 11460 0.29 0.23 Chitradurga Chitradurga 140206 0.24 3506 3865 0.37 0.35 Shikarpur Shimoga 36015 0.12 1370 1511 0.37 0.34 Shimoga Shimoga 322650 0.15 7839 8055 0.26 0.24 Bhadravati Shimoga 151102 0.20 4590 4320 0.23 0.26 Udupi Udupi 125306 0.10 3232 3468 0.37 0.33 Kadur Chikmagalur 34151 0.20 1340 1362 0.45 0.40 Tumkur Tumkur 302143 0.17 7112 7775 0.28 0.23 Kunigal Tumkur 34155 0.12 1319 1349 0.56 0.48 Anekal Bangalore 44260 0.12 1442 1822 0.49 0.43 Mandya Mandya 137358 0.15 3748 3598 0.42 0.39 Malavalli Mandya 37601 0.22 1559 1594 0.68 0.71

Table 5: Comparing Dissimilarity Index (D) in 2001 and 2011 for Select Cities in Karnataka Hassan Hassan 133436 0.08 3331 3588 0.35 0.30 Channarayapatna Hassan 38792 0.12 1446 1641 0.43 0.34 Mangalore Dakshina Kannada 488968 0.06 6659 8098 0.31 0.27 Krishnarajanagara Mysore 35805 0.21 1332 1390 0.54 0.50 Mysore Mysore 893062 0.16 11621 12607 0.42 0.32 Aland Gulbarga 42371 0.14 1532 1584 0.52 0.54 Gulbarga Gulbarga 533587 0.17 7683 8445 0.51 0.50 Sedam Gulbarga 39341 0.24 1371 1392 0.58 0.47 Malur Kolar 40050 0.20 1209 1605 0.47 0.32 Bangarapet Kolar 44849 0.24 1683 1686 0.43 0.34 Robertson Pet Kolar 143233 0.58 4041 3951 0.45 0.47

Table 6: Comparing the Dissimilarity Index (D) in 2001 and 2011 for Select Cities in Uttar Pradesh City District Total Population 2011 D 2001 D 2011 Kandhla Muzaffarnagar 46796 0.67 0.67 Shamli Muzaffarnagar 107266 0.40 0.42 Thana Bhawan Muzaffarnagar 36669 0.65 0.55 Budhana Muzaffarnagar 39867 0.66 0.61 Thakurdwara Moradabad 44255 0.82 0.78 Bilari Moradabad 37567 0.65 0.66 Bahjoi Moradabad 37037 0.57 0.59 Chandausi Moradabad 114383 0.54 0.47 Suar Rampur 32158 0.69 0.66 Tanda Rampur 48059 0.78 0.79 Bilaspur Rampur 43908 0.54 0.52 Shahabad Rampur 38276 0.51 0.58 Milak (NPP) Rampur 30553 0.60 0.60 Dhanaura (NPP) Jyotiba Phule Nagar 30007 0.76 0.78 Bachhraon (NPP) Jyotiba Phule Nagar 31101 0.63 0.52 Naugawan Sadat (NP) Jyotiba Phule Nagar 32954 0.85 0.77 Baraut (NPP) Baghpat 103764 0.64 0.60 Garhmukhteshwar (NPP) Ghaziabad 46077 0.59 0.52 Siana (NPP) Bulandshahr 44415 0.37 0.39 Dibai (NPP) Bulandshahr 39818 0.57 0.60 Shikarpur (NPP) Bulandshahr 37969 0.57 0.53 Khurja (NPP) Bulandshahr 111062 0.47 0.48 Khair (NPP) Aligarh 35751 0.60 0.61 Sikandrarao (NPP) Mahamaya Nagar 46038 0.76 0.77 Fatehpur Sikri (NPP) Agra 32905 0.58 0.65 Shamsabad (NPP) Agra 33144 0.48 0.56 Shikohabad (NPP) Firozabad 107404 0.38 0.40 Sirsaganj (NPP) Firozabad 31377 0.61 0.53 Mainpuri (NPP + OG) Mainpuri 120400 0.37 0.34 Bisauli (NPP) Budaun 32780 0.49 0.50 Kakrala (NPP) Budaun 37986 0.72 0.78 Puranpur (NPP) Pilibhit 40007 0.62 0.65 Jalalabad (NPP) Shahjahanpur 38202 0.47 0.46 Paliya Kalan (NPP) Kheri 41126 0.59 0.50 Mohammadi (NPP) Kheri 44968 0.56 0.52 Kheri (NP) Kheri 33355 0.51 0.60 Khairabad (NPP) Sitapur 48538 0.58 0.63 Pihani (NPP) Hardoi 36014 0.39 0.45 Mallawan (NPP) Hardoi 36915 0.46 0.46 Bangarmau (NPP) Unnao 44204 0.56 0.56 Jais (NPP) Rae Bareli 26735 0.50 0.51 Kaimganj (NPP) Farrukhabad 34384 0.41 0.42 Bharthana (NPP) Etawah 44120 0.41 0.43

Table 6: Comparing the Dissimilarity Index (D) in 2001 and 2011 for Select Cities in Uttar Pradesh Gursarai (NPP) Jhansi 26869 0.56 0.55 Barua Sagar (NPP) Jhansi 25028 0.53 0.54 Atarra (NPP) Banda 47419 0.43 0.43 Bindki (NPP) Fatehpur 36926 0.53 0.52 Rudauli (NPP) Faizabad 43091 0.60 0.58 Jalalpur (NPP) Ambedkar Nagar 31972 0.71 0.67 Nanpara (NPP) Bahraich 48337 0.58 0.60 Colonelganj (NPP) Gonda 29435 0.54 0.39 Deoria (NPP) Deoria 129479 0.30 0.31 Rudrapur (NP) Deoria 34014 0.39 0.41 Gaura Barhaj (NPP) Deoria 36459 0.44 0.43 Rasra (NPP) Ballia 31765 0.55 0.56 Ballia (NPP) Ballia 104424 0.34 0.32 Shahganj (NPP) Jaunpur 26556 0.74 0.78 Ghazipur (NPP + OG) Ghazipur 110587 0.29 0.32 Mohammadabad (NPP) Ghazipur 38328 0.57 0.55 Zamania (NPP) Ghazipur 33243 0.42 0.46 Varanasi (M Corp.) Varanasi 1198491 0.36 0.35 Ramnagar (NPP) Varanasi 49132 0.34 0.38 Chunar (NPP) Mirzapur 37185 0.41 0.31 Etah (NPP) Etah 118517 0.52 0.45 Jalesar (NPP) Etah 38130 0.55 0.54 Kasganj (NPP) Kanshiram Nagar 101277 0.48 0.48

References: Adelman, Robert M. 2004. Neighborhood Opportunities, Race, and Class: The Black Middle Class and Residential Segregation. City and Community 3(1): 43-63. Alba, R. 2005. Bright vs. Blurred Boundaries: Second-Generation Assimilation and Exclusion in France, Germany, and the United States. Ethnic and Racial Studies, 28(1): 20-49. Alba, Richard D., John R. Logan, and Brian J. Stults. 2000. "The Changing Neighborhood Contexts of the Immigrant Metropolis." Social Forces 79(2):587-621. Desai, Sonalde and Amaresh Dubey (2011): Caste in 21st Century India: Competing Narratives Economic and Political Weekly. 46 (11): 40-49. Dirks,Nicholas. 2001. Castes of Mind: Colonialism and the Making of Modern India. Princeton: Princeton University Press Dupont, V. 2004. Socio-Spatial Differentiation and Residential Segregation in Delhi: A question of scale? Geoforum 35:157 175. Iceland, John and Melissa Scopilliti. 2008. Immigrant Residential Segregation in U.S. Metropolitan Areas, 1990-2000. Demography 45(1): 79-94. Kim, Ann H. and Michael J. White. 2010. "Panethnicity, Ethnic Diversity and Residential Segregation." American Journal of Sociology 115(5):1558-1596. Logan, John R., Wenquan Zhang, and Richard D. Alba. 2002. "Immigrant Enclaves and Ethnic Communities in New York and Los Angeles." American Sociological Review 67(2):299-322. Madheswaran S. and Paul Attewell. 2007. Caste Discrimination in the Indian Urban Labor Market: Evidence from the National Sample Survey. Economic and Political Weekly 42(41):46-53. Massey, Douglas S., and Nancy A. Denton. 1985. "Spatial Assimilation as a Socioeconomic Outcome." American Sociological Review 50(1):94-106. Massey, Douglas S., and Nancy A. Denton. 1993. American Apartheid: Segregation and the Making of the Underclass. Cambridge, Mass.: Harvard University Press. Mehta, S. 1969. Patterns of Residence in Poona, India, by Caste and Religion: 1822-1965 Demography 6(4): 473-491. Mehta, S. 1968. Patterns of Residence in Poona (India) by Income, Education, and Occupation (1937-65) The American Journal of Sociology 73(4): 496-508. Munshi, Kaivan and Mark Rosenzweig (2006): Traditional Institutions Meet the Modern World: Caste, Gender, and Schooling Choice in a Globalizing Economy. American Economic Review 96(4): 1225-1252. Portes, A. and Zhou, M. 1993. The new second generation: Segmented assimilation and its variants. The annals of the American academy of political and social science, 530(1), 74-96. Portes, Alejandro and Ruben G. Rumbaut. 1996. Immigrant America: A Portrait. Berkeley: University of California Press. Sidhwani, Pranav. 2015. Spatial Inequality in Big Indian Cities Economic and Political Weekly 50 (22): 55-62. Singh, Gayatri. 2014. Freedom to Move, Barriers to Stay: An Examination of Rural Migrants Urban Transition in Contemporary India. PhD Dissertation. Thorat, Sukhadeo and Paul Attewell. 2007. The Legacy of Social Exclusion Economic and Political Weekly 42(41): 41-45. Thorat, Sukhadeo and Katherine Newman. 2010. Blocked by Caste: Economic Discrimination and Social Exclusion in Modern India. Delhi: Oxford University Press. United Nations. 2014. World Urbanization Prospects: The 2014 Revision. Vithayathil, Trina and Gayatri Singh. 2012. Spaces of Discrimination Economic and Political Weekly, 47 (37): 60-66. 20

White, Michael J., Ann H. Kim, and Jennifer E. Glick. 2005. "Mapping Social Distance: Ethnic Residential Segregation in a Multiethnic Metro." Sociological Methods & Research 34(2):173-203. Wilson. W. J. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. Chicago: University of Chicago. Witsoe, Jeffrey. 2013. Democracy Against Development. Chicago: The University of Chicago Press. 21