Shifting approaches to slums in Chennai: Political coalitions, policy discourses and practices

Slums pose a persistent challenge for fast growing urban areas in the global South, despite several decades of policy intervention. While Chennai has adopted several strategies ranging from upgrading to reconstruction, the city has been unable to deliver its target of ‘clearing’ slum settlements. Through an analysis of four enumeration reports and a look at the evolving political contexts and subsequent practices, we illustrate the evolution of slum policy approaches in Chennai since the 1970s. The analysis shows slum practices in Chennai continue to be characterized by an underlying continuity, with relocation as the dominant mode of operation since the nineties. However, approaches to slums have also evolved from paternalistic socialism with in-situ development in the seventies, to approaches characterized by affordability and cost recovery in the eighties, to the aesthetics of global cities in the nineties, to the technology driven, to slum-free ‘smart city’ discourse currently in vogue.


Introduction
With more than one billion people living in slums, Sub-Saharan Africa and South Asia experience the 'worst shelter deprivations' (UN Habitat, 2014: 5). India, the largest economy in South Asia, has shown a relative decrease in its slum to urban population, with the percentage of slum dwellers dropping from 23.1 to 17.3 per cent in the last census decade. This decrease is attributed to the exclusionary practices of the Indian state resulting in large-scale evictions, rather than as a result of successful government schemes (Arabindoo, 2013;Bhan & Jana, 2013;Kundu, 2014). In the same census period, however, the absolute number of 22.8 million slum residents has increased (GOI, 2005;. Nevertheless, the last two slum censuses in India are difficult to compare as the number of cities enumerated has changed. While in 2001, slum blocks were identified in towns having a population of 20,000 or more at that time, in 2011, they were delineated in all statutory towns irrespective of their population size. Due to this change, the number of towns reporting slums has increased from 640 in 2001 to 2543 cities/towns in 2011. Despite including all the towns reporting slum population, the 2011 census data overlooks settlements with less than 60 households and therefore does not capture ground realities comprehensively (Bhan & Jana, 2013).
The enumeration of slums in India is not only a technical process but an exercise tinted with discourses of state power. For instance, Ghertner (2010) illustrates that the enumeration of slums in New Delhi was done without numbers by using aesthetic criteria that superseded evidence from statistics. Similarly, Roy (2003) shows how the state exerts power through the absence of maps in Calcutta, naming it the logic of 'unmapping', which is used as a tool of regulation. Richter and Georgiadou (2016) argue that the enumeration of slums is a process characterized by informal negotiations between the state and the slum population. Making someone 'visible' or 'invisible' involves an exercising of decisional power; determining who has legitimate access to resources or who is excluded from urban development processes (Polzer & Hammond, 2008). This power is embedded not only in the data, but also lies with the one who controls access to and analysis of that data (Taylor & Broeders, 2015). Building on the discussions of enumerating slums, this article sheds light on how the approaches to slums over the last four decades in Chennai were configured by analysing the slum enumeration reports.
Discourses underlining dominant slum approaches have witnessed large shifts over the past decades as a result of practices which do not pan out. Despite claiming a comprehensive strategy towards slum settlements, the policies and practices of the state commonly do not converge and are 'acted upon and often contested by multiple actors' (Sutherland et al., 2016:52). This can be particularly seen for the Tamil Nadu Slum Clearance Board (TNSCB 1 ), the nodal agency for slums in Chennai, who proclaims 'a three-pronged' strategy for the redevelopment of slums-infrastructure improvement, and tenement construction either in-situ or on a relocated site (TNSCB, n.d.: 1). However, the research shows that the co-existence of practices was a result of continuity of funds despite changes in discourses over time. Furthermore, there never seemed to be an equal thrust on the diverse practices of TNSCB as one practice dominated over others in different periods of its institutional history. These shifting discourses and continuity of practices were driven by changing political coalitions. In this article, we argue that though TNSCB claims a comprehensive approach towards slums in Chennai, its efforts were diluted by the dominance of a singular practice steered by shifting policy discourses and political coalitions in different historical periods. Practices here comprise of institutional enactment of a combination of TNSCB's policies which can be both constitutive and constituted by the representations of slum settlements in the four enumeration reports discussed in the article. This is based on Foucauldian (Foucault, 1972) understanding that is drawn on a range of institutional priorities, economic dimensions and political concerns which together define the rules and corpus of knowledge for understanding and improving slum settlements.
The article begins with the theoretical approaches informing the research. This is followed by a description of the research context and explanation of the methodology. The empirical analysis provides a genealogy of the slum approaches in Chennai within the context of political coalitions, policy discourses and practices since the 1970s. Finally, the conclusion links the theoretical debates outlined above to the slum practices in Chennai.

Configuring approaches to slums
The focus of this research relates to several debates. The first concerns the political coalitions on international and domestic scale levels that reveal the wider power structures within which approaches to slum policy are carried out. The frameworks for slum policies are subject to actors and arrangements at several scales extending beyond city institutions (Milbert, 2006). Global discussions regarding quality of life improvements of the urban poor in developing countries often influence domestic policies (Pugh, 1992;2001). As hegemonic actors, international organizations such as the World Bank led and defined trajectories of urban development processes in developing nations (Ramsamy, 2006). This was practised through specific contractual conditions about policy and administration that the Bank negotiated in the loan agreements with national governments (Pugh, 1990). Raman (2011) has for instance described how the World Bank profoundly influenced state practices in the 1970s and 1980s in Madras. In addition, the utopian idea of 'cities without slums' supported by the United Nations was used to legitimize slum eradication in housing policy rhetoric (Huchzermeyer, 2010). The international policy norms allied with national and regional actors often resulted in constraining local bodies which are limited by the 'legislation, policy and funding mechanisms' regulated by multiple institutions (Sutherland et al., 2016:50). These fluid coalitions between different institutions have the ability to influence slum policy and practice at city and provincial levels.
In political coalitions, the role of civil society in its interactions with state and private sectors is crucial for negotiating services for the urban poor (Mitlin & Satterthwaite, 2004). Moreover, the use of electoral politics as a negotiation tool by the urban poor is discussed extensively in urban theory (Benjamin, 2004;Chatterjee, 2004;. Many scholars have elaborated on slums in Chennai within the context of Dravidian politics (Arabindoo, 2011;de Wit, 1993;Pugh 1988;1989;1990;Raman, 2011;Subramanian 1999;Wiebe, 1981;Wyatt, 2013b). Yet there has been little work which analyses the way in which political shifts have impacted slum policy approaches. Drawing on a set of TNSCB's slum enumeration documents, this article analyses the mandates of dominant actors who have contributed to shifting slum approaches in Chennai.
The second debate relates to the construction and evolution of policy discourses in urban studies. The concept of discourse with its multiple interpretations has been used across various disciplines with different foci among scholars (Jacobs, 2006;Bacchi, 2000). When policy is considered as a discourse, power struggles are represented in the way texts are combined in a certain manner to exclude or displace other combinations (Ball, 1993). Framing 'problems' and their impact on possible action is analysed by deconstructing policy discourses and identifying the coalitions of actors who produce it (Bacchi, 2000). In policy research, Hajer (2009: 60) defines discourse as 'an ensemble of notions, ideas, concepts, and categorizations through which meaning is ascribed to social and physical phenomena, and that is produced in and reproduces in turn an identifiable set of practices'.
Most discourse analysis related to urban policies is based on the works of Foucault and Fairclough (Jacobs, 2006). While Fairclough focuses on textual analysis, Foucault adopts a historical approach and analyses power as a central object of concern. In his work, the genealogy of knowledge is explored through power relations that are the result of repression and domination (Foucault, 1972). He argues that discourses exemplify the continuous relation between regimes of truth and systems of power that produce and sustain it. He states that the wider discursive struggle often creates new 'regimes of truth', which mark the discontinuity of past discourses and the reconstruction of fresh strategies. One illustrative example is the case of Delhi: here the discourse of 'nuisance' provided legal justification for evicting slum dwellers from the city (Ghertner, 2008). In a similar vein, the legitimacy of squatter settlements in Cape Town was challenged through an ecological discourse contributing to the racial division of urban space (Dixon et al., 1994). While discussions on slum discourses are often limited to one period in history, this article explores how slum policy discourses in Chennai have shifted over time.
Finally, the third debate focuses on the processes and materialities embodied in practice and, more specifically, their evolution as a result of changing discourses and actor coalitions. 'Clearance' was the main solution for slums in nineteenth century Paris and London (Gilbert, 2007: 707). In the sixties and early seventies of the twentieth century, state-led housing construction-inspired by the clearance approach of the North-was adopted widely in the global South (Wiebe, 1981). By the seventies, the World Bank endorsed Turner's concept of 'self-help housing' and supported 'upgrading and sites and services' programs for developing urban poor settlements in the global South (Pugh, 2001). In the nineties, De Soto (2000) argued that land titling for urban poor could help them use their assets as collateral for generating income. The Inter-American Development Bank (IADB), World Bank and numerous governments across the world approved and sought his advice, though his ideas were, according to some, 'methodologically weak, his calculations were both dubious and opaque' (Gilbert, 2009: 42). With the new millennium, the World Bank backed the 'cities without slums' strategy of the Cities Alliance that was translated into UN's MDG Target 11 as an obligation to rid cities of slums (Huchzermeyer, 2014). Whether viewed from the policy perspective of clearance, state-led construction, or from the 'cities without slums' initiative, slums have always been considered a 'problem' to be fixed by the state (Gilbert, 2007). This construction of 'slums as problem' has witnessed an enormous shift from the conditions of slums depicted as a 'problem' in the legal discourse underlying nuisance law in Delhi (Ghertner, 2008) to equating illegality with informality and therefore delegitimizing the right to shelter of the urban poor (Dupont & Ramanathan, 2008).
Taking these strands of debates into account, the article explores how political coalitions, policy discourses and practices configure approaches to slums in Chennai, and how these dimensions influenced approaches over the last four decades.

Research context: Chennai
This research is based in Chennai, 2 the provincial capital and largest city in the Indian State of Tamil Nadu. The urban population of Chennai grew from 2.64 million in 1971 to 4.68 million in 2011; the city boundary expanded from 68 km 2 in 1901 to 426 km 2 in 2011 (Krishnamurthy & Desouza, 2015). Although the city has witnessed visible growth in the information technology industry in the last two decades, almost threefourth of its population is engaged in the informal sector (Kennedy et al., 2014). Official estimates indicate that the slum population in Chennai doubled from 0.7 million in the seventies to 1.3 million in 2011.

Methodological considerations for tracing slum approaches in Chennai
There is a lack of coherent policy on urban issues in India that impacts the state's intervention in the urbanization processes as well as the policies related to marginalized populations living in cities (Patel, 2009). Moreover, in the Indian constitution, urban development is a provincial subject; the TNSCB, established in 1970, is the state level institution responsible for slum interventions in Tamil Nadu. In the absence of slum policy documents, all enumeration reports prepared or commissioned by TNSCB since the seventies, totalling four (Table 1), were selected as they embody the dominant Shifting approaches to slums in Chennai policy shifts towards slums. These reports provide a unique opportunity to analyse and better understand the approaches to slums in Chennai. Using these reports as departures, slum approaches in Chennai are explored through the following specific questions: a. How have the mandates of dominant actors influenced approaches to slums in Chennai since the seventies? b. How are slums discursively constructed in the enumeration reports? c. How have the dominant practices of slum redevelopment by the TNSCB evolved?
For the current study, we limit ourselves to the settlements listed in slum enumeration reports of the TNSCB. Therefore, the term 'slum' is used as an administrative category and not in a rhetorical sense. To complement the policy document analysis, the research drew on twenty-one semi-structured interviews with strategic actors such as state officials, politicians, NGO staff members, social activists, researchers and private consultants, as well as on published and unpublished documents. The interviews varied considerably in length ranging from half an hour to two hours at times. They were audio-taped with verbal consent of the interviewee with exceptions when taping was intrusive or the respondent expressed disagreement.
The reports were analysed according to shifts in conceptualization, categorization, instruments used for enumeration and the actors involved in order to understand how slums are discursively constructed in official enumeration documents. This was coupled with the analysis of interviews that revealed the past and present issues related to slums in Chennai to construct the approaches.

Genealogy of slum approaches in Chennai
In this section, shifts in slum approaches are constructed using a historical lens. The genealogy maps the actors and coalitions involved as well as the rationales behind evolving policy discourses and practices. The analysis is divided into three historical periods, along with the latest enumeration process (Figure 1).

Paternalistic tenement construction towards a Madras 'clear' of slums
The Tamil Nadu Slum Clearance Board was established in 1970 by the erstwhile regional government of Dravida Munnetra Kazhagam (DMK). The main practice of TNSCB was in-situ construction of tenements for slum dwellers with the goal of 'clearing' slums from Madras (TNSCB, 1975: 6). The socio-economic survey enumerating slums in Madras was one of the initial steps towards achieving this enormous goal. This survey was presented in the first TNSCB report published in 1975 that underscored the role of the board as a 'provider' of housing for the poor (TNSCB, 1975: 6). This discourse heavily rested on the paternalist politics of DMK originating in the Dravidian movement of Tamil Nadu. In the pre-Independence era, the Dravidian movement was rooted in anti-Brahmin sentiments driven by the Justice Party, which was replaced in 1919 by Dravidian Kazhagam (DK). The DK split in 1949 resulted in the formation of DMK. As an election strategy, DMK distanced itself from the anti-Brahmin slogans of DK and appealed to a more universal Tamil identity, crossing caste and religious boundaries (Subramanian, 1999). Madras was considered DMK's stronghold and 'clearance' of slums from the city was part of their election manifesto (Srivathsan, 2012). Within a year of DMK's electoral victory in Tamil Nadu, TNSCB was established in line with DMK's policy to reach out to the urban poor. The discourse of socialism was reflected through the anti-eviction strategy of the TNSCB that was discussed as a key tool for development of slums in Madras (TNSCB, 1975: 6-9). The Tamil Nadu Slum Clearance Areas Act 1971 was passed with the salient feature of eliminating the ability to evict slum dwellers without alternative accommodation from private lands (de Wit, 1993;TNSCB, 1972). This attempt to eradicate the eviction of slum dwellers was designed to promote faith in TNSCB as an institution. TNSCB argued for an equal recognition of slum dwellers' contributions to the city compared to those of other residents in Madras. Nevertheless, the discursive construction of social services for the poor rested on both the notion of the paternal state as well as the narrative of 'deficient' slum resident (Wiebe, 1981: 54). For instance, the negative framing of slum settlements is evident in the excerpt below: Slums generally present the most unhygienic, ugliest, nauseating scene. During rainy season, the whole area gets flooded, the pathways become swampy and the entire colony becomes a fertile breeding place for mosquitoes, exposing the slum dwellers living in the area to all sorts of diseases. During summer, the thatched huts are prone to fire accidents. Thus, the slum dwellers life is the most miserable one, devoid of all basic amenities (TNSCB, 1975: 5).
Under the rationale of paternalistic socialism, the main practice to achieve slum clearance was through in-situ construction of tenements for slum dwellers, with the DMK party having powerful 'control' over the budget and policies of TNSCB during its initial years (Raman, 2011). A seven-year target date was set for the TNSCB to 'clear' all slums in the city through in-situ tenement construction based on the model of hirepurchase in the Slum Clearance Scheme. A total of 1202 slums in the Madras city area were enumerated with a population of 737 531 (Table 2) and were depicted in a hand drawn map showing their location in the city (TNSCB, 1975). Because it demanded a high investment of time and money, tenement construction was the dominant, but not the only practice employed by TNSCB (de Wit, 1993;Wiebe, 1981). Infrastructural upgrading projects funded by the central government, such as the Environment Improvement Scheme (EIS), were also initiated in 1972 to provide interim services alongside the clearance programme (de Wit, 1993;Wiebe, 1981). The socialist rationale in the slum discourses matched the dominant practice of state funded in-situ tenement construction by TNSCB.
Urban development through affordability and cost recovery by World Bank Because of high financial costs, slow pace of construction and the poor level of services in the project, TNSCB fell short of its slum clearance goal through state subsidized tenement provision in Chennai (Wiebe, 1981). After the Emergency period in which democratic processes were suspended , the DMK lost the elections to its newly founded political rival Anna Dravida Munnetra Kazhagam 3 (ADMK  (Pugh, 1990: 173). The World Bank intervention in Chennai ran parallel to the ADMK reign in Tamil Nadu. Unlike its political rival DMK, MG Ramachandran attracted votes from the poor and women, particularly in rural areas and smaller towns (Subramanian, 1999). The populist strategy of ADMK differed considerably from DMK, which concentrated on 'jobs and housing' (Wyatt, 2013a: 44). ADMK facilitated administrative and legislative reforms to assist new policies of affordable service provision administered by the World Bank. In return, ADMK harvested political advantage through their association with such urban development projects in Chennai (Pugh, 1992). MUDP shifted the focus of urban poor policies from tenement construction to sites and services and upgrading by investing in the latter and emphasizing a reduced role of TNSCB in tenement construction (Raman, 2011). However, the central government-funded Environmental Improvement Scheme (EIS) for slum improvement continued under a new name: the Accelerated Slum Improvement Scheme (de Wit, 1993). Although the dominant policy discourse was affordable service provision, TNSCB continued building tenements with state funds, albeit on a much smaller scale (World Bank, 1977). During this period, the discourse based on affordability and cost recovery was dominant, however, existing practices from the seventies were also maintained.
Economist Group, a private consultant, enumerated slums on behalf of TNSCB together with the MMDA, and the results were published in 1987. The slum enumeration suggested ambiguity in the count of the settlements in Madras through several measures. First, though the report stated a total of 1413 settlements (Table 2) in Madras Urban Agglomeration, the survey identified 996 slum settlements that needed intervention. Moreover, 434 slum settlements covered under World Bank's MUDP and clearance schemes by TNSCB were listed in the slum directory but these settlements did not directly form a part of the enumeration process (EG, 1987). Second, while the report lacked any definition for slums in Madras, it introduced a new set of classifications such as 'objectionable and unobjectionable slums'. Objectionable slums were categorized as 'encroachments without land ownership', tainted with illegality, while unobjectionable slums consisted of 'gram natham, cheri natham and town natham' 4 Shifting approaches to slums in Chennai characterized by legality through land rights (EG, 1987: 27). This ambiguity in numbers and categories contrasted with the technical presentation of the data. For instance, the report consisted of computerized data analysis with a map showing discrete locations and other relevant details of the settlements in Madras. All three parts of the report represented slums as quantifiable entities. A lack of photographs and individual cases reinforced the technical tone of the report.
World class city rationale and peripheralization of urban poor In its dream to be a world-class city, beautification schemes 5 became political goals of Tamil Nadu's ruling government. Unlike the first two phases when either DMK or ADMK dominated state politics for longer periods, after MGR's death, the two leading Dravidian parties were constantly competing and alternated power with every election. The 'fluid and unstable electoral alliances' have marked the regional politics of Tamil Nadu since the nineties, with rising numbers of smaller political parties (Wyatt, 2002: 736). This resulted in the emergence of 'programmatic policies' with universal appeal (Wyatt, 2013a: 28). In tandem with competitive politics, the increasing involvement of private sector reduced the stakes of the urban poor in Chennai. The Dravidian parties shifted their focus from the socialist development of the poor to the neoliberal strategies of attracting foreign direct investment (Adaikalam, 2010). Several mega-projects in the city, such as the MRTS 6city train, Chennai Metro and the Information Technology expressway to the South-became part of the populist ideologies of successively elected Dravidian parties. Politicians associated mega-projects with 'visibility' and 'appeal to modernization' as a means to attract global capital for infrastructural development (Kennedy et al., 2014: 17). The growth of mega-projects in the new millennium was linked with discourses on 'beautification and development' and justified slum evictions from Chennai's urban centre (Coelho & Raman 2010: 19). In addition, high population densities and poverty rates in Chennai slums, and the distribution of cash among voters in the settlements during election periods became common strategies of the Dravidian parties (The Hindu, 2011). Vote purchasing with cash limited the engagement of slum dwellers in negotiations as part of political society (Chatterjee, 2004). Marked by a larger stake from the central government in slum programs of Tamil Nadu, the TNSCB relied on funds from several sources such as the flood alleviation program, the Tenth and Eleventh Finance Commission and the Jawaharlal Nehru National Urban Renewal Mission (JNNURM), which were all federal programmes and equally supported by both dominant Dravidian parties when in power: Dravidian politics did not have any role to play in the policy of relocation and reconstruction because both parties were interested in construction. All the contractors are politicians. Supposing if DMK is in power, they select only DMK contractors. So the whole channel is set and nobody will object. Whether it is DMK or ADMK, they act as if they are against eviction, but in practice both support it. Even in the last period if you see ADMK was against RAY, but now they are in power so naturally they are also encouraging the project. (Retired Government Official, pers. comm., 2014) JNNURM was a landmark initiative as it brought the urban agenda to centre stage of Indian development policy. The program had two sub-missions, namely 1) Urban Infrastructure Governance (UIG) and 2) Basic Services to Urban Poor (BSUP), although the spending for the poor decreased over time and they became 'cosmetic' partners of the program (Kundu, 2014). Interventions for the urban poor were often repressive, tainting the program with exclusionary traits dominated by slum evictions. For Chennai, the BSUP component in the state-level report of the JNNURM shows that 85 per cent of the funds were used for relocation projects and the remaining fifteen for upgrading (GOI, 2014). Central government funded programs resulted in large-scale housing construction for resettlement colonies at the periphery of the metropolitan area (Coelho & Raman, 2010;HLRN, 2014). These initiatives coming under Rehabilitation and Reconstruction (R&R) Schemes became the dominant practice of the TNSCB, while tenement construction and upgrading continued in reduced forms. The World Bank also formed part of the political coalition through the Emergency Tsunami Reconstruction Project (ETRP). It helped create a Community Development Wing in TNSCB to reduce political meddling in MUDP; this Wing helped the TNSCB in evicting slums and implementing their R&R strategy through the networks they developed with slum communities (Raman, 2011). There was a wide gap between the discourse of governance, investment, and infrastructure present in the world-class city rhetoric and the ground realities of large-scale eviction and peripheralization of the urban poor. Regulation, development and right to the city were often used as strong arguments for eviction and relocation of slum dwellers, as stated below: Legally, the people who come to Chennai and built their hutments along the side of canals do not have rights to the place. They don't purchase that land for building their houses. They just settle there. If you legally question then they have no rights. It is the sympathy of the government or patronage towards that section of people that aids them to live in the space they occupied illegally. The government does not want to go there and tell-'no you do not live here'. If there is widening of roads or something which the government wants to do in terms of development, then they try to shift the people to some other place in pucca building. (Politician, pers. comm., 2014) Liberalization of the Indian economy laid the ground for transforming its metropolitan centres into global cities, while the 74th Amendment to the Constitution pushed for decentralized governance structures (Dupont, 2011). Globalization underlining the world-class city rhetoric of Indian metropoles has further exacerbated socio-spatial inequalities by increased marginalization of urban poor and polarizing city spaces (Banerjee-Guha, 2002). Making a 'world class' city, redevelopment and beautification through slum demolitions provided the authorities an opportunity to 'clean up' urban spaces (Dupont, 2008). Parallel to such aspirations were the normative dimensions of governance that dominated the rationale of infrastructure development in slums of Chennai. Along with new private consultants, semi-private institutions such as Tamil Nadu Urban Infrastructure Financial Services Limited 7 (TNUIFSL) became stakeholders in the slum enumeration process in Chennai in 2005. In this fresh endeavour of mapping slums, settlements were projected as sites of 'investment' with due focus on 'demand assessment' that included detailed budgets based on infrastructural deficits and socio-economic settlement characteristics (IN-RIMT & TCG, 2005). The enumeration was marked with technological advancements with regard to city and settlement level maps made using remote sensing techniques and Auto Cad software, and developing new categorizations. 242 settlements were categorized as 'undeveloped' slums requiring investment within the Municipal Corporation of Chennai area; they were differentiated from 'developed' slums that had been improved (Table 3). Focusing on 'governance, investment and infrastructure', the enumeration report endorsed the enablement policy of the World Bank and marked a changing role of the state from 'provider' in the early seventies to 'enabler' in the present-day neo-liberal regime (IN-RIMT and TCG, 2005).

Resurgence of 'slum-free city' agenda in Chennai
Marking a decisive turn from eviction-based practices, the erstwhile Congress government announced the Rajiv Awas Yojana 8 (RAY) or the 'Slum Free City' initiative in 2009 driven by the notion of inclusive development (Bhan, 2011;Mathur, 2009). This new policy was seen as an extension of the ongoing urban renewal program-JNNURM-with the prospect of tenure security as a new deal for the urban poor. The shift towards a more dominant role of the central government in funding and determining programmes continued from the previous period. However, when the national government changed in 2014, the JNNURM/RAY program was replaced with Pradhan Mantri Awas Yojana-Housing for All (Urban) and the 'Smart Cities Initiative' by the Modi Government. With a focus on governance and infrastructure, this urban development program focuses on city competitiveness. Although AIADMK won the state-level election in 2016, the current situation is volatile and it is difficult to make any sort of prediction of how it is going to pan out.
In the latest slum enumeration in Chennai, done under the RAY scheme in 2014, slums are depicted as 'diseased', with 'curative and preventive strategies' proposed to address the 'problem of slums in a holistic manner' by private consultants Darashaw. Titled 'Slum Free City Plan of Action', the report enumerated 1131 slums with a population of 1 155 025. In contrast to the previous reports, this document presents new categorizations of 'tenable and untenable slum' aligned to planning guidelines with tenable slums situated in areas earmarked for residential use, following environmental standards, and untenable slums as settlements located in 'unhealthy and environmental risk areas' (Darashaw, 2014: 73). Though the report discusses new categorizations, the slum definition is classic, using the same words as the first TNSCB report from 1975. The strategies are situated in a technical discourse of 'deficiency matrix', based on parameters related to infrastructure and vulnerability. Normative dimensions of 'best' and 'worst' slums are extracted from the deficiency matrix. An eight-year plan is proposed to phase out slums from Chennai, with delisting as an option for settlements scoring very low on the deficiency matrix. Those with the highest scores were to be addressed immediately. Technologies such as GIS and Management Information Systems are used, together with statistical analysis, to formulate the matrix and assign strategies and phasing plans to individual slums.
Under the Modi government the Central Ministry has shifted the focus towards Smart Cities. Recently, the Corporation of Chennai won the India Smart Cities Challenge competition with the goal of improving infrastructure, increasing efficiency of services and enhancing communication with citizens at a citywide scale. Focusing on technology in e-governance and smart city programs induces structural discrimination as not every resident can access technology or afford the instrument leading to participation (Pfeffer et al., 2015). The call for retrofitting the city underlines the shift from quantity to quality, with a changing focus from slum to non-slum populations of Chennai. As this fourth period is ongoing, it is too early to assess how the discourse of portraying slums as a 'disease' shapes the actual practices of TNSCB in Chennai. However, there is a growing distrust among non-governmental actors with respect to the practices that will mark the resurge in 'slum-free' agenda, as illustrated below: We are against the argument of slum free city in Chennai. We are not freeing the city of slums but freeing the city of slum dwellers. This is an orchestrated effort. The government of Tamil Nadu says that we do not have lands in the city to construct housing for the poor, but they have not conducted vacant land mapping for the RAY project, so how do they even know. I am not sure if it is the land question or the lack of will to give land to the urban poor. (Housing Activist, pers. comm., 2014)

Discussion and conclusions
This paper discussed how slum policy approaches have been configured since the seventies in Chennai. First, the shift is closely linked to the changing mandates of the dominant actors in political coalitions. In the seventies, the coalitions were mainly local, with DMK as the dominant actor in its alliance with TNSCB to drive the Slum Clearance scheme consisting of tenement construction. By 1977, the World Bank started to dominate with its global capital loans and ADMK gained political mileage through their coalition with the World Bank. The MMDA functioned as a key organization for MUDP, ending TNSCB's role as the sole institution catering to the urban poor. The nineties shifted towards increased eviction and construction through projects funded from World Bank and central government programs. The populist politics of past decades were transformed into programmatic politics of alternating Dravidian parties in power, consisting of slum relocation programmes (Wyatt, 2013a). Finally, the 'smart city initiative' by the Modi government since 2014 has altered the focus from slums to technology-based solutions, which may cater more to middle and upper middle-income populations. This analysis of political coalitions shows up an absence of civil society in Chennai as drivers of slum discourses in any of the periods. While the World Bank, as an international organization, has ended direct involvement in slum related projects, the neoliberal discourse continues to drive slum approaches. The analysis of political coalitions has revealed increasing influence from international and national actors; and private sector actors in the approaches to slums at the city scale.
Second, slum policy discourses evolved from paternal socialism in the seventies to the rise of affordability and cost-recovery in the eighties, to the focus on good governance and infrastructure in the nineties; and finally, the technology driven slum-free agenda that characterizes current approaches. These discourses illustrate three transforming dimensions: 1) the use of new vocabulary in slum classifications without any legal binding, 2) an increasing stake of the private sector in the knowledge construction process, and 3) a change in technology. The growing use of geo-technologies in mapping slums has led to greater contributions by the private sector in knowledge construction (Pfeffer et al., 2015). The technical expertise of private sector undermines the capacity of local governments in processes of knowledge construction and therefore decision-making. Private consultants as new knowledge producers, incorporated a fresh vocabulary regarding classifications (Table 3). However, these classifications are not legally binding because they were not included in the Tamil Nadu Slum Clearance Act 1971. Additionally, slum categories highlight the relationship between informality and illegality, with 'objectionable and undeveloped' representing illegal and informal settlements and 'un-objectionable and developed' referring to legal, yet informal areas. Slum strategies were thus built around these new categorizations, marking a discontinuity with past discourses and creating new 'regimes of truth' (Foucault, 1972). Unlike the continuous evolution of categories, the general and most negative definition of slums has been preserved over time.
Third, the analysis of slum approaches over time shows an underlying continuity of practices. Although the TNSCB claims to have a 'three-pronged strategy' (TNSCB, n.d.: 1) of in-situ upgrading, tenement construction and rehabilitation, a single approach has continued as the norm over the years, highlighting the ambiguous and informal nature of the state (Roy, 2009). The seventies focused on large-scale investments in state subsidized tenement construction as a paternalistic scheme of DMK, though infrastructural upgrading was used as interim relief for settlements waiting to be housed. User-paid service upgrading was the dominant practice during the World Bank's era in Chennai, though TNSCB never discontinued tenement construction. A large-scale investment in relocation projects has been the dominant practice since the nineties. Tenement construction and maintenance is still practised, yet receives negligible budgets when compared to construction in peripheral locations. Finally, neither the slumfree policy nor the idea of beautification is new in Chennai. Nevertheless, these discourses have re-emerged in the new millennium and the practices to achieve them have been transformed. Discourses of socialism and affordability in the first two phases matched the dominant slum practices of state subsidized tenement construction and user-paid service provision. This is especially revealing when compared to the last two periods where eviction has been at the forefront of the global and smart city strategies put forward by national government programmes.
Shifting political coalitions, policy discourses, and dominant slum practices contributed to new approaches in each of the periods. With regards to the slum-free agenda currently in vogue, slums are mapped and made visible. However, this visibility in enumeration jars with the negative discourse of depicting slums as a 'disease' and calling for preventive and curative strategies of action. Indian slum discourses are currently taking new shape with the rise of geo-and ICT-technologies as dominant tools in the smart cities initiative of the present day Modi government. This brings us to future research challenges involving the exploration of how an increased dominance of digital technology will influence slum discourses, how new actor coalitions will be formed, and perhaps most importantly, what type of slum practices will result from this new configuration.
Combining discourses with the type of actor coalitions and their actual practices reveal the lack of congruence between the various dimensions of a configuration at any given period. Furthermore, taking a temporal perspective on the approaches to slums deepens our understanding of the complexity underlying slum development processes. By analysing approaches to slums, this article reveals the cleavages between slum policies and institutional practices in each of the four time periods. Therefore, this article builds a case for a critical analysis of slum policies, taking into consideration discourses, political coalitions and practices dynamically in order to understand the outcome of large increases in informal settlements across the Global South despite several decades of policy intervention by national and international development agencies.

Acknowledgements
This research was carried out in the context of the EU-FP7 project Chance2Sustain: Urban Chances, City Growth and the Sustainability Challenge (Project no. 244828), with EADI as the lead partner of the project. We would like to thank those Chance2Sustain project members who have provided valuable comments on this research. We also acknowledge Jorge Izquierdo Cubero's design assistance for the figure. We are also thankful to insightful comments given by Federico Savini, Dennis Rodgers, Yves van Leynseele and Elisabeth Peyroux on earlier drafts of this article as well as the reviewers for their thorough reviews. Finally, we also thank all the respondents in Chennai for their valuable time and insightful discussions.

Notes
1 Please refer to the annexure at the end of the manuscript for the abbreviations used for actors and projects discussed in the article. 2 The city of Madras was renamed Chennai in 1996. Because of the historical nature of this article, the names are used interchangeably and refer to the capital city of Tamil Nadu. 3 ADMK (Anna Dravida Munnetra Kazhagam) political party was formed in 1972 by MG Ramachandran because of bifurcation of DMK and later the prefix AI (for All India) was tagged to the party name. In this paper, ADMK and AIADMK are used interchangeably as both the names signify the same political party. For further details of the party history, please refer to http://aiadmk.com/all-india-anna-dravida-munnetra-kazhagam/.
4 The EG report states that the unobjectionable slums located on gram, cheri and town nathams are not considered encroachments because the revenue authorities have granted land ownership rights to the residents of these settlements. Natham in Tamil language refers to land with common ownership among dwellers in the case of urban or rural settlements that can be only used for residential purposes.
5 Schemes such as 'Singara Chennai' (translated as Beautiful Chennai) by the DMK in 2006 (Ellis, 2012) were re-packaged as 'Ezhilmigu Chennai' (translated as More Elegant Chennai) by the AIADMK in 2011 (Kotteswaran, 2013). 6 MRTS (Mass Rapid Transit System) in Chennai is an elevated railway line that was conceived in the seventies and has been in operation since 1997. 7 TNUIFSL is a public-private partnership involving the Government of Tamil Nadu and financial institutions active in the large-scale property and infrastructure development sector, which hold a majority of the shares in the organization (For more details, refer to Adaikalam, 2010).
8 Since the election of Modi's government, RAY has been replaced with Pradhan Mantri Awas Yojana.