Deindustrialization in cities of the Global South

Recent research by economists has shown that deindustrialization is more severe in Sub-Saharan Africa and Latin America than it ever was in the Organisation for Economic Co-operation and Development (OECD). Nevertheless, most research on deindustrialization is focused on the former centres of Fordist manufacturing in the industrial heartlands of the North Atlantic. In short, there is a mismatch between where deindustrialization is researched and where it is occurring, and the objective of this paper is to shift the geographical focus of research on deindustrialization to the Global South. Case studies from Argentina, India, Tanzania and Turkey demonstrate the variegated nature of deindustrialization beyond the North Atlantic. In the process, it is demonstrated that cities in the Global South can inform wider theoretical discussions on the impacts of deindustrialization at the urban scale.


INTRODUCTION
Scholarship on deindustrialization has come of age in the past four decades and has evolved into a rich interdisciplinary field. It has expanded thematically from an initial focus on political economy to include less tangible impacts of deindustrialization on people and communities 'beyond the ruins' (Cowie & Heathcott, 2003). Although deindustrialization scholarship has become thematically rich and theoretically sophisticated, its geographical imagination has remained strikingly limited. A recently edited publication, The Deindustrialized World: Confronting Ruination in Postindustrial Places (High, MacKinnon, & Perchard, 2017, p. 6), notes that 'the centrality of the US experience … has effectively anchored the international field of research' since the early 1980s. One ostensible objective of the book is to introduce 'into the conversation different national contexts' and 'reveal the comparative threads drawing together the particularities of industrial decline in different geographies ' (p. 8). Although the book's 15 chapters manage to avoid some of the most studied examples of deindustrialization, such as Detroit and Pittsburgh, they regrettably focus exclusively on cities in the Organisation for Economic Co-operation and Development (OECD).
The Global South looms large in the background of deindustrialization scholarship, but it remains a non-descript destination for runaway factories. The implicit assumption embedded in much of this scholarship is that the South has industrialized at the expense of cities and communities in the North Atlantic and elsewhere in the OECD. This assumption could not be farther from the truth. While there has undoubtedly been a reorganization of industrial production on a global scalewhat Fröbel, Heinrichs, and Kreye (1980) termed the 'new international division of labour'-recent research demonstrates that industrial offshoring was a highly uneven process. Baldwin (2016, p. 2) shows that 'only six developing nations … saw their share of world manufacturing rise by more than three-tenths of one percentage point since 1990'. Over roughly the same period, many countries in Latin America and Africa experienced deindustrialization that, at least in terms of percentage of gross domestic product (GDP), exceeds the level experienced by countries in the North Atlantic that are typically the focus of deindustrialization scholarship (Rodrik, 2016). Indeed, Dani Rodrik shows that 'the brunt of globalization and the rise of Asian exporters has been borne by other developing countries, rather than the advanced economies' (Rodrik, 2016, p. 16). In summary, there is a mismatch between where deindustrialization is researched and where it is occurring (however, see Hassink, Hu, Shin, Yamamura, & Gong, 2018).
This paper broadens the focus of deindustrialization scholarship. The findings from Rodrik, Baldwin and other economists provide a useful starting point because they demonstrate the need to shift our focus and explore the emergent geography of deindustrialization in the Global South. However, their use of country-level data has limitations: most notably, it problematizes deindustrialization as a national/developmental issue that can be addressed with industrial and trade policy. It does not inform our understanding of how deindustrialization unfolds in cities and communities, yet this is precisely the scale at which deindustrialization is most intensively and viscerally experienced. Indeed, deindustrialization in the North Atlantic is typically framed as a city/regional problem, and most research is situated at subnational scales which confirms that it is not only an economic phenomenon. This paper takes its cue from this rich body of research and focuses on how deindustrialization unfolds in cities in the South. It presents a broad range of cases from India, Tanzania, Turkey and Argentina. The objective is two-fold. The first is to recover the experience of people beyond the North Atlantic whose histories have been erased in scholarship on deindustrialization. The second is to highlight the variegated modes of deindustrialization in cities in the South, and implicitly compare these processes with assumptions informed by the experience from the North Atlantic.
The paper is structured as follows. The next section situates this paper in research on deindustrialization and postcolonial urbanism. It demonstrates that the former failed to account for events beyond the North Atlantic, while deindustrialization has remained a blind spot in the latter. The third section introduces the case studies. The final section concludes by drawing out comparative insights and identifying avenues of future research at the intersection of scholarship on deindustrialization and urban studies.

SITUATING DEINDUSTRIALIZATION BEYOND THE NORTH ATLANTIC
At the end of the Second World War, industrial activity was highly concentrated. The United States had 7% of the global population, yet it controlled more wealth than the rest of the world and its manufacturing capacity dwarfed foreign rivals (Adas, 2006, p. 223). In the following decades, the United States launched a programme of reconstruction whose objective was to restore the industrial capability of Western Europe and Japan to pre-war levels (Ekbladh, Deindustrialization in cities of the Global South 3 AREA DEVELOPMENT AND POLICY 2009). In response to the perceived threat from the Soviet Union, the United States embraced a Fordist-Keynesian model whereby the state sought to impose a wage bargain and contain conflict between capital and labour. 1 The result was a golden era in the North Atlantic, and according to Esping-Andersen (1996, p. 1), 'prosperity, equality and full employment seemed in perfect harmony'. While Esping-Andersen overlooks the fact that discrimination in many forms was hard-wired into the post-war regime, his glowing appraisal does highlight the remarkable longevity and stability of this growth episode (Piketty, 2014). This regime entered a period of crisis in the late 1960s from which it never recovered. It remains unclear whether the erosion and ultimate breakdown of the post-war economic order was caused by in-built contradictions or exogenous shocks (Piore & Sabel, 1984), but what is clear is that many places in the North Atlantic that had powered the post-war economic miracle entered a prolonged period of decline. Bluestone and Harrison (1982, p. 4) lamented that '[t]he system that seemed so capable of providing a steadily growing standard of living during the turbulent 1960s had become totally incapable of providing people with a simple home mortgage, a stable job, or a secure pension'. The 1970s' economic crisis disrupted the wage bargain in the United States and reactivated dormant hostilities between capital and labour. Industrial capital sought to 'outrun' labour by relocating from its traditional bases of operation in the American Northeast and Midwest to the Southern states where labour was less organized (Bluestone & Harrison, 1980). Capital's sojourn in the American South was a brief layover in a much longer journey overseas to so-called 'Third World' countries (Bluestone & Harrison, 1980;Dicken, 2015;Fröbel et al., 1980). Offshoring accelerated in the 1980s and 1990s as advancements in information and communication technology and international institutions allowed multinational corporations to develop global value chains (Baldwin, 2016;Castells, 1996;Gereffi & Korzeniewicz, 1994). Cities that had been centres of production and affluence were profoundly impacted by disinvestment and descended into 'paroxysms of devaluation and deindustrialization' (Harvey, 1989, p. 185). The speed of this decline was remarkable, and in many instances, it occurred before policy-makers could devise a meaningful response. Storper and Walker (1989, p. 90) noted that '[w]ithin a few years, or even months, the economic base of a community can disappear, leaving the empty hulks of obsolete factories and decaying infrastructure'.
The socio-spatial impacts of deindustrialization in the North Atlantic were explored at the urban scale by geographers and sociologists who sought to understand how the various trajectories of the post-industrial transformation of cities contributed to growing inequalities through the processes of gentrification and abandonment. From the 1970s, gentrification research was characterized by a debate between consumption-and production-side approaches that explained gentrification either as a product of the growth of a 'new middle class' employed in service industries (Leys, 1996) or as the reinvestment of capital in previously disinvested inner-city neighbourhoods (Smith, 1979). Despite disagreement over the specific causes of gentrification, both approaches understood gentrification as the outcome of structural changes associated with the loss of industry (Ghertner, 2015). Against the background of this debate, researchers have explored the urban impacts of deindustrialization by documenting processes of neighbourhood change and class-based displacement in post-industrial cities such as New York and London (Curran, 2004;Davidson & Lees, 2005). In contrast to gentrification researchers' focus on the displacement that results from the reinvestment of capital, other researchers have focused on places that continue to be characterized by high levels of abandonment. Wacquant's( 2008) comparative study of 'advanced marginality' in the post-Fordist city argued that deindustrialization and 'deproletarianisation' have resulted in the concentration of the urban poor in stigmatized 'neighborhoods of relegation', such as Chicago's poorest neighbourhoods and Parisian banlieues.
A field of scholarship emerged to capture the broad-based impacts of industrial decline, that is referred to here as deindustrialization studies. Its initial focus on political economy has expanded and it has become truly interdisciplinary. The field broadened in line with general trends in social science, most notably the rise of postmodernity and cultural studies in the 1980s (Strangleman & Rhodes, 2014). Scholarship focused on plant closures and struggles between unions and corporations gave way to the reappraisal of deindustrialization as a 'more socially complicated, historically deep, geographically diverse, and politically perplexing phenomenon than previously thought' (Cowie & Heathcott, 2003, p. 2). Scholars engaged with the complexity and far-reaching consequences of deindustrialization, not least of which is the profound transformation of place and the disintegration of communities and ways of life (Cowie & Heathcott, 2003;High & Lewis, 2007;Linkon, 2018;Mah, 2012). Although deindustrialization studies has evolved thematically, its geographical imaginary has curiously remained stunted. As noted above, deindustrialization in Latin America and Sub-Saharan Africa has been severe; in the case of the former, the 'magnitude of adverse employment effects … is even larger than in developed economies' (Rodrik, 2016, p. 19). However, deindustrialization studies has remained steadfastly rooted in the industrial heartlands of the North Atlantic, and has only recently begun to account for deindustrialization elsewhere in the OECD (High et al., 2017). The Global South remains a vague construct, which, when it appears at all, is as a homogenized recipient of offshored industry.
The failure of deindustrialization studies to account for events in the Global South is mirrored by the neglect of deindustrialization in postcolonial urban studies. The latter has critiqued the Western-centric character of mainstream urban studies, with many scholars heeding Roy's( 2009, p. 820) call to 'blast open theoretical geographies, to produce a new set of concepts in the crucible of a new repertoire of cities'. Whereas high-profile debates around world/global cities privileged a small number of wealthy cities in discussions of urban modernity, scholars such as Robinson (2006) and Roy (2009) have argued for a postcolonial approach to urban studies that flattens hierarchies among cities and incorporates the experiences of diverse urban contexts into the generation of new theories of urbanity. In this context, there is a growing enthusiasm for comparison as a tool for challenging and rethinking established theories, as well as generating new concepts from locations previously marginalized in urban studies (McFarlane, 2010;Robinson, 2016). However, this body of scholarship has tended to highlight aspects of cities in the South that render them different from their OECD counterparts (Bhan, 2019;Lawhon & Truelove, 2019;Parnell & Robinson, 2012;Schindler, 2017;Simone, 2010). Indeed, it is the existence of these differences that justifies the very notion of 'postcolonial' or 'Southern' urban studies, yet the imperative to highlight difference has meant that postcolonial urban scholars have avoided interrogating phenomena that are typically associated with cities in the North Atlantic. In this paper deindustrialization is recognized as a common process occurring across the North and South, while also remaining sensitive to the importance of 'historical difference' in shaping how its varying trajectories manifest at the urban scale (Roy, 2016, p. 203).
By shifting the geographical focus of research on deindustrialization to the South, this paper speaks to ongoing debates in urban studies surrounding the comparability of cities (Peck, 2015;Roy, 2016;Scott & Storper, 2015). Deindustrialization in the Global South tends to appear in social science scholarship as a background context for other processes, such as urban transformation or the erosion of working conditions, rather than a phenomenon worth researching in its own right (Bremen's, 2004, study of industrial decline in Ahmedabad is an important exception). For example, Harris (2008Harris ( , p. 2421 finds that gentrification in Mumbai has occurred in the context of 'local deindustrialisation processes', such as the relocation of textile mills to the hinterland. Elsewhere, deindustrialization is portrayed as a precondition for widespread precarity in post-apartheid Johannesburg (Barchiesi, 2011) and Deindustrialization in cities of the Global South 5 the Zambian Copperbelt (Ferguson, 1999). Meanwhile, the decline of state-owned enterprises in north-east China, and the disappearance of the basket of social benefits that they provided, serves as the basis for claims-making among retrenched workers (Kwan Lee, 2007). What remains unclear, however, is the extent to which these local episodes can generate novel insights into the variegated character of global deindustrialization processes. Rodrik's( 2016) research provides a tantalizing clue. He notes that in contrast to deindustrialization in the OECD, deindustrialization in the South is not necessarily accompanied by increased productivity in the service sector. As the case of Brazil demonstrates, it can even be driven by a reprimarization of the economy as capital is disinvested from manufacturing and reinvested in the extraction of resources and agriculture (Gorenstein & Ortiz, 2018;Jenkins, 2015). Following Robinson's(2016, p. 20) encouragement to theorize from 'anywhere' the identification of these local variations raises an important question: How can cities in the Global South inform wider theoretical discussions of the impacts of deindustrialization at the urban scale?
The case studies presented below speak to this question by showing how deindustrialization has unfolded in cities in India, Turkey, Tanzania and Argentina.

CASE STUDIES OF DEINDUSTRIALIZATION
The series of case studies are situated at the intersection of deindustrialization studies and urban studies from India, Turkey, Tanzania and Argentina. Different methods were used in each case, such as the analysis of land-use planning documents, analysis of urban political economies and ethnography. Each case employs a lens that enables an appreciation of different aspects of deindustrialization as it unfolds in the Global South: the case of India demonstrates how the drivers and dynamics of deindustrialization can differ between cities even within the same country; the case of Turkey explores how the relationship between deindustrialization and urban transformation is mediated by national politics; the case of Tanzania highlights the impact of deindustrialization on labour, livelihoods and capabilities; and the case of Argentina draws attention to the importance of iterative processes of deindustrialization and reindustrialization for analysing the changing geography of cities beyond the North Atlantic. Taken together, they demonstrate that deindustrialization serves as a useful entry point for understanding the variegated nature of 21st century globalization and urbanization.

Variegated deindustrialization in urban India
Much has been written about the liberalization of the Indian economy in 1991, and the impacts that this has had on urban regions (Ahluwalia, 2000;Shaw, 1999;Shaw & Satish, 2007). Following liberalization, public sector industries were seen as bureaucratic, inefficient and badly managed (Khanna, 2015), resulting in disinvestment and an imperative to attract investment from the private sector. In many Indian metros there has been a gradual dismantling and/or relocation of industrial infrastructure and manufacturing units. However, a full understanding of the drivers and implications of deindustrialization on urban space and residents requires looking beyond national-level data to study deindustrialization within specific cities. Here recent research in Mumbai and Bangalore was drawn upon to illustrate different techniques through which one can learn about the patterns, processes and implications of deindustrialization. City-level economic and employment data demonstrate the extent of deindustrialization in Mumbai. Manufacturing's share in district GDP fell from 39.10% in 1993/94 to 25.04% in 2009/10, while that of services increased across the period from 59.64% to 73.94% (Pethe, Nallathiga, Gandhi, & Tandel, 2014). The city's employment structure has also experienced a parallel shift towards services. Manufacturing employment as a proportion of total employment fell from 36.0% to 28.5% between 1980 and 1990, while in the same period the trade, finances and services sector increased from 52.1% to 64.3% (Banerjee-Guha, 2002).
Deindustrialization in Mumbai was orchestrated by governments at various scales in pursuit of a decades-old national policy commitment to relocate industries to underdeveloped areas beyond the city centre. This imperative accelerated after the political defeat of Mumbai's working class in the failed textile mill strike of 1982-83, and the emergence of a Indian urban regime that has been called the 'cabal city' (Heller, Mukhopadhyay, & Walton, 2016). The argument is that a cabal is a small group of actors who collude across state and market boundaries to maximize returns to scarce resources, and in this particular case their focus is on urban land. Logan and Molotch's( 1988) 'growth machine' is geared towards the augmentation of land value 'subject to coherent governance and an overall economic logic' that results in growth that is 'sustainable and inclusive, if not equitable'. By contrast, the primary objective of the cabal is to extract land-centric rent through the management of spatial transformation. The drive to maximize the extraction of rent from urban land means that the development schemes they pursue are rarely aligned with long-term economic growth or competitiveness (Heller et al., 2016). Thus, officialdom in Mumbai and powerful elites in the private sector pursue urban development strategies that prioritize increasing the value of land in the short term, rather than establish the conditions for long-term economic prosperity or competitiveness.
Mumbai's cabal is enabled by increasing centralization of power in the state government and the disempowerment of local government, even as civil society is selectively empowered. Thus, key planning and policy decisions have become the domain of the Urban Development Department of the government of Maharashtra (Patel, 2015). The Chief Minister of Maharashtra, who is typically an elected representative of a distant rural constituency, also chairs key parastatals that determine redevelopment and infrastructure building initiatives, and ultimately wields greater power over the built environment in Mumbai than the elected municipal corporation. The vision for Mumbai's future that has guided urban development since the early 2000s has been the city's 'Shanghaification', to which both central and state governments were committed. These commitments emerged from the famous 'Vision Mumbai' document produced by McKinsey & Co. in 2003 at the request of Bombay First, a non-governmental organization (NGO) floated by the local business community in 1995, modelled on London First. Mumbai's manufacturing sector was to be relocated to the urban periphery, while the plan envisioned the tertiarization of the city centre.
In practice, the city's transformation is guided by instrumental concerns of those with the power to determine land use. Mumbai's cabal overrides the disciplinary logic and practice of 'technical planning' by trained professionals in state organizations, and instead it leverages 'sovereign planning' by fiat from the highest level of the subnational government (Krishnankutty, 2018). The resulting urbanism is marked by a tension between, on the one hand, an aspirational and discursive ideal that may be called 'engine urbanism' (Burte, 2016-17) and, on the other, the realized urbanism founded on land-centric rent extraction favouring the cabal. Indeed, it is through the transformation of urban spacethat is, from derelict textile mills to gated communitiesthat rent is generated. However, urbanism founded on landcentric rent extraction favours a small elite whose strategic positions allow them to capture the revenue generatedformally and informallyby the transformation of the urban fabric (cf. Chattaraj & Walton, 2017;Nijman, 2000). An illustrative example is the 1996 intervention by the government of Maharashtra (of which Mumbai is the state capital) to override the municipal corporation's more equitable proposal for urban development, which allowed textile mill owners to retain land leased to them for industry and redevelop it for non-industrial purposes (Adarkar & Phatak, 2005). This enabled inequitable land rent extraction following deindustrialization, as industrial land was converted to real estate.
The repercussions of the cabal city are experienced in Mumbai in many forms: by its unsystematic urban development that favours land-rent extraction over other social or even capitalistic objectives; in its prioritization of interventions with short-term efficacy but longterm unsustainability that are implemented with scant local political oversight through parastatals: in the four years from 1997, for example, approximately 40 flyovers were built in the city, flying in the face of any transportation planning logic; in the opaque and often contradictory logic embraced by officials situated at various scales; in the splintering of classes and communities that has been deepened on the ground (Anand, 2006;Zérah, 2008); and in the violence it has enacted on the poor and in the scattered practices of coping and contestations that it has catalysed (Bhide, 2013;Burte & Kamath, 2017). The shift in priority from a model of urban development in which industry played a role, however small, in the city's economy, to a regime geared towards land-rent extraction means that sovereign planning in Mumbai has consisted of selective interventions largely designed to augment lucrative real estate and infrastructure projects that are necessarily out of the reach of the majority of the population, including retrenched workers, thereby deepening Mumbai's housing crisis. Meanwhile, propoor projects that may reduce inequalities, such as improvement of the uneven and inadequate water-supply systems and sanitation provision across the city, have been largely postponed or even ignored (Zérah, 2008) as they are less monetizable. For example, transportation planning has moved in the opposite direction from inclusive mobility through the rapid expansion of car-friendly road infrastructure (including a sea-link and an upcoming coastal road). This infrastructure has opened up new peri-urban lands for real estate development and also reinforced the centrality of the central business district (CBD) and South Mumbai, contradicting the state's own historical commitment to deconcentration.
Bangalore's economy exhibits a shift from industrial to post-industrial sectors reminiscent of processes played out in cities in the OECD. Municipal authorities are involved in the real estate sector (Goldman, 2011), but in contrast with Mumbai's cabal whose efforts are often at odds with stated planning objectives, their efforts to capture rents have complemented the city's very real shift to post-industrial sectors and its emergence as India's premiere high-tech hub. The impacts on the city's manufacturing base have been significant. An analysis of land use in Bangalore's spatial planning documents highlights a clear shift from public investment in manufacturing to private investment in services and real estate.
Industrial transformation has had a significant impact on Bangalore's urban form and the urban economy, visible through an analysis of successive master plans and changing land-use plans. Post-independence Bangalore's urban form was determined largely by the creation of five large public sector units (PSUs) that focused mainly on manufacturing for defence, national security and other sensitive industries. These industries not only are employers but also are industrial townships on the urban periphery that provided housing, schools, healthcare and other amenities to employees and their families. These PSUs encouraged the development of auxiliary micro-, small and medium-sized enterprises (MSMEs) throughout the city. As Bangalore's economy has become increasingly service orientedprimarily driven by information technology companiesits PSUs and MSMEs are gradually shutting down or moving out of the city. These shifts have had profound impacts on the city's built form. Large tracts of vacant former industrial land are increasingly being converted to high-end residential and/or commercial uses. One illustration of this is the case of the Hindustan Machine Tools (HMT) watch factory that was shut in 2016. The factory land has been parcelled into smaller units, with 10 acres converted into mixed-use real estate and 210 acres sold to the Indian Space Research Organization.
Analysis of land use of the 2004, 2015 and 2031 Bangalore master plans shows that land allocated to industrial use within the city's boundaries has been declining over the last two decades, with industrial land gradually being converted to (primarily) residential land ( Table 1).
Analysis of specific parcels on the northern and eastern peripheries of Bangalore reveals that nearly 2500 hectares of industrial land has been converted primarily into residential use. Industries are being relocated to the urban periphery where Karnataka Industrial Areas Development Board (KIADB) land banks are being consolidated in ways that favour formal capital-intensive industrial production. These reallocations are consistent with broader national government plans to build industrial corridors and new industrial settlements, though it is unclear whether these settlements will house existing plants that are being relocated from within the city or new emerging industries. Initial analysis therefore indicates that manufacturing is declining within the city, particularly with reference to SMEs and the informal sector.
The current trend of dismantling industry and manufacturing in Bangalore seems to sit in contradiction with the national and regional governments' push to promote manufacturing and establish new industrial settlements throughout industrial corridors. Furthermore, it exacerbates a mismatch between the city's labour force and the kinds of investments that are being made as part of the new wave of urban development which favours capital-intensive high-tech industry and globally oriented information-technology enabled services. There is also a growing presence of global actors in the development of these new industrial settlements, as investors, developers as well as tenants, which has further implications for how peri-urban industrialization takes place.
The two case studies presented here indicate that deindustrialization can be driven by state disinvestment in manufacturing and industry, tied to a rebalancing of the economy towards finance and service sectors, and tied up in intricateand informalcoalitions of rent-seeking state and non-state actors prioritizing profit over more equitable development. There is also a clear trend towards spatial transformation that is emerging in these cities that favours particular types of urban development, focusing largely on real estate development. It is increasingly apparent that deindustrialization in its current form will leave a legacy, because many urban residents will neither benefit from employment opportunities within the new economy nor be able to afford housing in redeveloped areas.

State-coordinated urban transformation in Turkey
Turkish industry is highly competitive in some sectors, and its profile is increasingly shifting from low-to medium-skilled sectors (Hallward-Driemeier & Nayyar, 2018). However, as a percentage of GDP manufacturing peaked at 23% in 1989 and remained more or less constant until 1998 when it began to fall. 2 The downward trend continued until 2010 when manufacturing comprised 15% of GDP and it then began inching upwards. While Turkey'sglobalshareof manufacturing increased between 1994 and 2014, the share of manufacturing value-added in GDP decreased (Hallward-Driemeier & Nayyar, 2018). Government spending has sought to counter lagging demand, particularly since the coup attempt in 2016, and this has led to unsustainable levels of borrowing (OECD, 2018). The national government has channelled this money into the built environment, housing and infrastructure. This section focuses on deindustrialization in Turkish cities in general, given the fact that state-coordinated efforts designed to catalyse urban transformation are situated at the national scale. Thus, while the reverberations of the Justice and Development Party's (AKP) ambitions are most evident in İstanbul (Kuyucu & Ünsal, 2010;Lovering & Türkmen, 2011), it is important to note that analogous processes are playing out in less dramatic fashion in Turkey's largest cities.
From the early 2000s onwards, public spending targeted infrastructure and housing projects, and this fostered a boom in the construction sector, which, as a percentage of GDP, increased from 11.5% in 2008 to 17.0% in 2017 (OECD, 2018). The centrality of the construction sectorwhich is often included in 'manufacturing' databelies a broader shift in Turkey's political economy that favours urban transformation at the expense of manufacturing. Indeed, the rapid transformation of Turkish cities is a cornerstone of the ruling AKP's claim to legitimacy. Since coming to power in 2002 under the leadership of Recep Tayyip Erdoğan in the wake of an economic crisis, the AKP has consolidated a national development strategy geared towards urban transformation. Its claim to success rests on its ability to manage fundamental urban transformation across Turkey in ways that incorporate a diverse range of stakeholders. This national-led urban development strategy extends the promise of new forms of urban citizenship to conservative and poor populations in Anatolia, and especially to those in metropolitan cities who the AKP (and its predecessor, the National Vision Movement) 3 has claimed to be marginalized by the secular regime since the inception the republic in 1923. Leaving behind the anti-West/globalization/neoliberalism attitude of its predecessor, the regime established by the AKP is anchored by a politics of hope whereby it speaks to the aspirations of Turkey's urban poor and the middle class, and extends the promise of what Brand and Wissen (2018) refer to as an 'imperial mode of living'.
The AKP's urban transformation strategy enjoyed broad-based support, and its power was rooted in the formation of a coalition after the 2002 elections that included the conservative urban poor, Anatolian bourgeoisie, political Islamists and Westward-looking secular middle class (Somel, 2011;Türkay, 2011). It was financed by an inflow of capital from global investors in the form of cheap bank credit (Balaban, 2011;Savran, 2013), which prioritized urban development schemes over investment in capital goods and manufacturing. The result was a booming urban property market that encouraged industrialists and other fractions of capital to invest in the (re)development of urban land. This regime is anchored by a carefully constructed notion of urban citizenship designed to appeal to populations that were marginalized before the AKP's rise to power. Rather than one's status as a waged labourer, membership in the political economy of Turkey's largest cities became based on property ownership, and for the first time, poor conservative populations were able to enter the formal real estate market en masse (Doğan, 2016;Karaman, 2013;Şengül, 2013). For this rather precarious population, the ownership of assets (e.g., real estate) offset the negative consequences of deindustrialization, at least in the short term. The rents generated by Turkey's booming statesupported real estate market were distributed to the AKP's constituents through patronage networks, while the bourgeoisie was able to earn handsome profits through its connection to the construction sector (Doğan, 2016). Elsewhere the central government was directly involved in the production of housing (Balaban, 2012), and it established a delicate balancing act in which the proceeds of urban transformation were (re)distributed through negotiation among diverse actors at multiple scales (government ministries, municipalities, investors, residents, businesses, religious charities) (e.g., Bedirhanoğlu, 2015;Penpecioğlu, 2016). Thus, in contrast to the traumatic effects of deindustrialization in the West (Strangleman & Rhodes, 2014), the AKP's urban strategy fostered deindustrialization, while simultaneously offering hitherto excluded sections of society a 'modern' urban (Islamic) lifestyle via newly built spaces of habitation and consumption (Bayırbağ & Penpecioğlu, 2017).
The AKP's response to the global economic crisis of 2008 was to deepen and widen its efforts to transform urban areas, which was enabled further by quantitative easing and low interest rates in the United States and Europe. It was as if the AKP were riding a bicycle ever faster in order to maintain its balance, and the result was a deepening reliance on financialization of the urban built environment while manufacturers struggled as demand from world markets declined (Aalbers, 2013;cf. Akçay & Güngen, 2014, p. 200;Bayırbağ, 2013b). Slackening demand was offset by a dramatic increase in public spending on large-scale infrastructure projects, which initiated a wave of urban transformation projects across the country and opening of new areas for the construction sector (Kayasü & Yetişkul, 2014;Şengül, 2013;Topal, Yalman, & Çelik, 2019). The cracks in this regime began to show in the late 2000s as short-term gains were unable to offset the long-term costs of (1) excessive public spending, (2) prioritization of investment in the built environment over more productive sectors of the economy and (3) the controversial impacts of total urban transformation. 4 Thus, following the 2011 elections, the AKP's urban development regime increasingly relied on coercion rather than consent. It was apparent that Turkey's household and private-sector debt was unsustainable, and to this discontent was added the colonization of public and common spaces by construction firms friendly with the AKP.
Unrelenting pressure on public space in Turkish cities fostered a tense environment, and public anger exploded in İstanbul over the proposed redevelopment of Gezi Park in 2013 (Bayırbağ, 2013b;Gündoğdu, 2013;Şengül, 2015). Furthermore, Turkey's political and economic crises have mounted since the coup attempt in 2016, and the constitution and the politico-administrative system of the country have been radically altered to institutionalize authoritarianism in its aftermath. National politics play out at the urban scale given the centrality of the transformation of cities in the AKP's political platform. Given that the AKP emerged from a nationwide urban hegemonic project led by the National Vision Movement (Bayırbağ, 2013a), most of the major figures of the AKP first entered the national political arena in 1994 as mayors from the Welfare Party (most notably Erdoğan as Mayor of İstanbul). The AKP's candidates lost municipal elections in most metropolitan cities in 2019 (including their stronghold, İstanbul), and although it is too soon to tell whether this signals the beginning of the end of its grip on power at the national scale, it does seem to indicate that its regime of urban transformation has reached its limits. Any new model of development will likely require a concerted effort to reindustrialize Turkey's cities and rebalance manufacturing capacity vis-à-vis the construction sector.
The case of Turkey offers several lessons on the spatio-temporality of deindustrialization. First, this case demonstrates that urban transformation can act as a driving force, rather than simply an outcome, of deindustrialization. Whereas Smith (1979) and Marcuse (1985) found that deindustrialization in US inner cities created the conditions for subsequent processes of gentrification, Curran (2004) demonstrates that gentrification can also contribute to the displacement of industry from urban areas. Similarly, deindustrialization in Turkey has occurred due to an influx of investment in the built environment, especially financial capital from abroad. Indeed, state-coordinated urban transformation in Turkey has been shown to trigger, or at least accelerate, deindustrialization. Second, the term 'deindustrialization' often refers to a historical rupture where industrialization is replaced by urbanization as a mode of wealth creation and (re)distribution. However, the example of Turkey indicates that it is important to detect continuities between those two moments of capitalism, and their coexistence takes the spatial form of heterogeneous geographies of accumulation, lending support to a point made by Henri Lefebvre that 'so-called underdeveloped countries are now characterized by the fact that they undergo the rural, the industrial, and the urban simultaneously' (Lefebvre, 2002, p. 32). Finally, departing from the caveat that '[i]t is crucial that we resist the temptation to explain urbanization as a straightforward manifestation of global capitalism' (Schindler, 2017, p. 6), this case demonstrates the importance of accounting for the variegated ways in which global capitalism is translated by national political-economic regimes that are manifested in cities.

Precarity, work and intergenerational tension in Tanzania
East Africa experienced an industrial boom in the decade following the Second World War, and Tanganyika (Tanganyika and Zanzibar established a union in 1964 and became Tanzania) was no exception, with its exports growing from £8.6 million to £48. 3 million between 19453 million between and 19523 million between (Burton, 2005. Tanzanian manufacturing was largely state owned, and most subsectors peaked in the early to mid-1970s (Coulson, 2013). The slowdown was caused by a combination of factors including mismanagement, drought and a costly war with Uganda, and by the early 1980s there was a full-blown crisis (Tripp, 1997). In spite of continued public investment, Tanzanian industry shrank by 11.2% in 1981 (Tripp, 1997), and by the late 1990s most large-scale textile mills had been closed (Government of Tanzania, 1996). Workers suffered tremendously during the crisis, as real wages fell by 83% between 1974 and 1988 (Tripp, 1997).
Industry in Tanzania was initially clustered almost exclusively along Pugu Road in Dar es Salaam, from where it expanded to other parts of the city and other cities (Iliffe, 1979). Several parastatals were established in Arusha and became national champions. Despite enjoying more reliable water and energy supply than their counterparts in Dar es Salaam, many did not survive the crisis (de Valk, 1996). Indeed, empty and derelict factories are testament to the scarcity of opportunities for Arusha's working class. By 2012, only 12% of Arusha's formalsector workforce was employed in the manufacturing sector (National Bureau of Statistics (NBOS), 2015). Three key issues around the lingering effects of the economic crisis in Arusha are notable: its influence on youth livelihood outcomes; intergenerational tensions; and a deterioration of working terms and conditions. Arusha's young men and women find themselves in increasingly competitive labour markets, rarely able to find stable work and reliant upon informal, irregular and insecure work opportunities. While young people constitute nearly half of Arusha city's population, they are overrepresented in unemployment rates. These stand at 8.4% for young people and 3.3% for adults (NBOS, 2016). In the disadvantaged ward where the present research took place, rates of youth unemployment were significantly higher: 37% and 53% for young men and women, respectively (Banks, 2016).
This has serious implications for sociality and livelihoods. Work is key to social acceptance and recognition, allowing young people to achieve other important cultural markers of adulthood: getting married, establishing their own house, having children and becoming recognized as legitimate citizens. Land ownership and the establishment of a family have served as avenues to respectability in Tanzania, yet these are unattainable by the youth who lack stable livelihoods. Youth under-and unemployment has also fuelled widespread stereotypes of young people as criminals, or 'wahuni' in Swahili (Burton, 2005), further limiting the institutional support afforded to them from the community. Households in this disadvantaged ward traditionally stop supporting their young people after finishing school. Where they cannot find work to meet their needs, assumptions are made that they must be supporting themselves through crime, including theft and prostitution.
The inability of parents and elders to help young people navigate this precarious labour market exacerbates these impacts. Many made the transition to adulthood in the 'golden age' of Arusha's industrial past. Community leaders spoke of how it was 'easy and smooth' to get a job before all the factories closed. While small business was still a preferred form of work, large numbers were employed in factories, and this fuelled a thriving local economy. Limited opportunities for wage labour and widespread poverty have left parents ill-equipped to guide their children in these challenging labour markets. Parents had put their faith in the education system, hoping that it would equip their children with labour market-relevant skills and knowledge. However, amidst large youth populations and tight labour markets, the payoffs from education are ambiguous. Intergenerational tensions emerge as young people are frustrated that their parents have not provided them with the guidance, education or skills investment they can use to find gainful employment. Meanwhile, parents claim that un(der) employed youth are distracted by leisure opportunities offered by the city and 'stop listening'. They interpret this as a rejection of traditional values, further exacerbating tensions. Local leaders said that young people have 'chosen globalization to be their parents' so that they can 'copy and paste' the lives they admire further afield.
This large-scale shortage of jobs also has repercussions on the wages and working conditions of remaining factory work. Factory work is far from a desirable job for young people, who voice concerns around the low pay, the arduous work, the long hours and its impacts on health. One young man who had to leave a job at a cement factory due to respiratory problems explained that once you have left for a 'personal problem' a factory will not take you back. Not only are wages low but also young people tend to be offered short-term contracts and are paid less than their senior colleagues. After seeing multiple older workers being recruited onto permanent contracts over the two years he had worked at a factory, one young man got together with some friends to ask the manager why their positions remain tenuous and impermanent. The manager took umbrage and they were sacked. Despite repeated attempts to secure compensation through formal channels, they were unsuccessful. 'We saw such pain in having to fight for it,' the young man stated, 'but we wanted to know why they were not treating us like human beings.' Amidst the scarcity of formal-sector work, there is little incentive for employers to offer better working conditions or more stable contracts. Where one young person is unhappy with their conditions, another will be willing to take his or her place. Indeed, each week on the recruitment day of one of Arusha's few remaining factories, large crowds wait outside factory gates in the hope they will be selected. As one young man explained, 'There may be 10 or 15 getting in on a good day, but there will be 70 people waiting at the gate!' Thus, deindustrialization has not only disrupted social reproduction but also resulted in intergenerational conflict and rivalry.
Processes of deindustrialization have fundamentally changed the economic and social landscape of Arusha. The oversupply of youth labour in combination with its stagnant labour market has resulted in insecure and exploitative working conditions. While the social and cultural changes accompanying urbanization are not driven solely by deindustrialization, the widespread poverty and insecurity left in its wake has exacerbated social problems and limited the ability of parents and elders to support young people. Frustrations, limited ambitions and a 'transitional limbo' characterize the experience of youth in Arusha.

Crisis and contested deindustrialization in Argentina
One of the first countries in the Global South to experience large-scale industrial development, Argentina also constitutes a significant case of premature deindustrialization (Castillo & Martins Neto, 2016). Argentina has a long history of industrial decline interspersed with periods of reindustrialization (Basualdo, 2001;Cohen, 2012;Waisman, 1987). Cycles of deand re-industrialization have been closely related to the national development strategies pursued by successive Argentinian governments in the latter half of the 20th century (Rock, 2002). Import substitution under successive post-war governments gave way to exportoriented industrial growth under Carlos Menem in the 1990s (Cohen, 2012;Waisman, 1987). This chapter of Argentine history ended with one of the most severe economic crises recorded in recent history (Epstein & Pion-Berlin, 2006;Varesi, 2010). Between 2003 and 2015, Nestor and Cristina Kirchner cultivated a demand-driven recovery of industry by protecting local industries and workers from global competition and subsidizing domestic consumption. In addition to these shifts in trade and industrial policy, successive governments have responded to deindustrialization by incentivizing speculative investments in urban-based sectors such as property development, consumer services and the creative industries. As a result, Argentina's largest cities have been transformed into mosaics of socioeconomic inequality, characterized by fragmented land-use and chaotic territorial governance regimes.
Argentina is once again experiencing a major economic crisis that threatens to reverse the industrial gains made in the long decade of Kirchnerism following the neoliberal debacle. Although industry benefited from increased domestic demand during the protectionist Kirchner years, export commodities such as soy crops were the primary driver of national economic growth. As such, the Kirchners' development strategy failed to establish the conditions for more sustainable and inclusive manufacturing growth (Teubal & Palmisano, 2015). Furthermore, economic recovery was cut short by the imposition of a new programme of economic liberalization and conservative policy reforms under President Mauricio Macri , whose administration aspired to establish a new model of accumulation with a set of policies including the lifting of previous currency exchange controls, economic deregulation and fiscal reform. The intention of this model was to integrate the Argentine economy more thoroughly, and not only through primary exports, into global value chains. While the foreign investments that the Macri government sought to attract have failed to materialize, Argentina is once again on the brink of economic crisis and experiencing deindustrialization (Natanson, 2019). Particularly hard hit are the SMEs located in metropolitan Buenos Aires that serve the domestic market and face shrinking aggregate demand.
The deepening transnationalization of Argentina during the Macri years may not have ushered in an era of investment-driven industrial competitiveness and innovation, but it did have important implications for the geography of metropolitan Buenos Aires. One dimension of Macri's strategy to attract foreign investment featured infrastructure upgrading, particularly focused on the transport and energy sectors. In fact, during its temporary presidency of the G20, Argentina identified infrastructure investments as one of the overarching aims for the year's agenda. Thereby, the country sought to participate in the emerging global model of infrastructure-led development and capture some of the North to South capital flows focused on infrastructure projects . In Buenos Aires, this has translated into a thorough revamping of logistics connectivity. The city's single largest infrastructure project is a controversial US$700 million underground roadway corridor stretching 7 km, and linking the southern and northern highways with the city centre (Vallejos, 2017). While industrial parks in outer-ring suburbs are likely to benefit from improved connectivity with the city's port, disused factories and other industrial spaces in the central city and inner suburbs have either been redeveloped for commercial and residential uses or simply left derelict (Dinardi, 2019).
Macri's defeat in the October 2019 elections may signal a return to more protectionist approach reminiscent of the Kircher era. It is also important to recognize that the model described above has provoked grassroots contestation. For example, popular struggles against deindustrialization have animated worker-led occupations of shuttered factories and the establishment of democratically managed cooperatives. Furthermore, an urban environmental movement has emerged to demand state action to address the hazards that the city's residents face as a result of inadequately regulated industrial production, such as exposure to floods and air pollution (Abrutzky, Dawidowski, Murgida, & Natenzon, 2014;Auyero & Swistun, 2009). Proposals to improve environmental conditions through, for example, increasing the amount of green/blue infrastructure in the city envision abandoned industrial spaces as opportunity areas to implement such innovations. In this sense, Buenos Aires is not only an important case study to analyse the drivers and consequences of deindustrialization in Latin America but also a site from which to learn how grassroots' mobilization could lead to the realization of morejust and more-than-industrial urban futures.

CONCLUSIONS
The diverse cases discussed in this paper illustrate that processes of deindustrialization have significantly different influences on the geography and political economy of cities in Africa, Asia, Latin America and the Middle East. As such, it demonstrates that there is a pressing need for researchers to pay greater attention to the relationship between deindustrialization and urban transformation in cities of the Global South. This research can make a significant contribution to the ongoing project to shift the focus of urban studies away from a small number of privileged cities in the North (Roy, 2009). Indeed, by exploring the urban dynamics of deindustrialization outside of cities traditionally understood as 'post-industrial', this research on cities in the South has the potential to disrupt outmoded epistemological distinctions between 'modern' and 'developing' cities (Robinson, 2006). In the process, it can build on the comparative turn in urban research in order to question and rethink established urban theories, and to generate new theoretical accounts in the process (McFarlane, 2010;Robinson, 2016).
Although a systematic comparison is beyond the scope of this paper, when taken together, the cases indicate several fruitful avenues for future research. First, the case of India illustrates that the relationship between deindustrialization and urban development can vary between cities within the same country. While both Mumbai and Bangalore have undergone what Shatkin (2016) refers to as a 'real estate turn', Mumbai's is driven by a cabal, while Bangalore's transformation is underpinned by an unmistakable shift to a post-industrial economy. Second, the paper indicates the need for comparative research to explore the variegated relationship between economic change and urban transformation in different cities, both across the South and between OECD and Southern cities. Although the deindustrialization of US cities often precipitated urban processes of capital reinvestment and gentrification (Marcuse, 1985;Smith, 1996), the case of Turkey indicates that investment in the built environment may in some instances precipitate industrial decline (although see Curran, 2004, for a similar phenomenon in the United States). As such, a more globally oriented deindustrialization studies can contribute new insights into the relationship between primary and secondary circuits of capital in particular urban contexts (Shin, 2014).
The relationship between processes of deindustrialization and dynamics of social inclusion/ exclusion is not predetermined and will be contingent on the political-economic dynamics of specific urban contexts. Whereas deindustrialization in Tanzania has contributed to the exclusion of young people from formal wage labour, the AKP's regime of urban citizenship in Turkey demonstrates that its shift of emphasis from industrialization to urbanization can potentially be mobilized to create new forms of inclusion on the basis of asset ownership. Further research should explore whether new modes of citizenship are emerging in contexts where participation in the formal labour force is no longer a realistic aspiration (Ferguson, 2015), and where urbanization is likely to proceed without significant industrialization. Finally, the case of Argentina demonstrates that deindustrializationand periods of reindustrializationis a process that can be contested and subverted. In Buenos Aires there is a history of contestation, as with the occupied factories movement and more recent environmental movements. As such, future scholarship on the urban politics of deindustrialization should attend to the agency of city dwellers who may not necessarily be workers and the emergence of new sites of value creation and struggle. For example, Gago (2017) showed how a market on the periphery of Buenos Aires where land ownership was indeterminate became a hive of transnational economic activity that allowed many retrenched workers to create and capture value during the 2001 crisis. The cases discussed in the present paper indicate possible methodological avenues for researching deindustrialization at the urban scale, including: political-economic analysis to understand how state actors have sought to respond to and shape processes of economic change; analysis of spatial planning documents and remote sensing to track changes in the built environment over time; and ethnographic methods to understand city dwellers' lived experience of deindustrialization.
In conclusion, scholarship on deindustrialization has become thematically and theoretically sophisticated, yet it has failed to keep up with the 21st century's evolving geography of deindustrialization. Meanwhile, urban scholars have begun to craft a body of scholarship focused on the novelty of urban processes in the Global South, in the world's largest and most dynamic cities. However, this scholarship has paid insufficient attention to deindustrialization. In this paper, initial steps have been taken to bridge these two bodies of scholarship. Hopefully, future research will build on this paper to create a more globally oriented deindustrialization studies.

DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.