https://link.springer.com/article/10.1007/s11111-019-00329-2
The measurement and characterization of urbanization crucially depends upon defining what counts as urban. The government of India estimates that only 31% of the population is urban. We show that this is an artifact of the definition of urbanity and an underestimate of the level of urbanization in India. We use a random forest-based model to create a high-resolution (~ 100 m) population grid from district-level data available from the Indian Census for 2001 and 2011, a novel application of such methods to create temporally consistent population grids. We then apply a community-detection clustering algorithm to construct urban agglomerations for the entire country. Compared with the 2011 official statistics, we estimate 12% more of urban population, but find fewer mid-size cities. We also identify urban agglomerations that span jurisdictional boundaries across large portions of Kerala and the Gangetic Plain.
The global rate of urban transition has been immense in the past half century, with much of that transition and associated population growth occurring across parts of Asia (Ellis and Roberts 2015; Schneider et al. 2015). In 1960, India and China had similar urban population percentages of 18% and 16%, respectively (World Bank 2018). Yet by 2016, according to the World Bank statistics, while the Chinese urban population was at 54%, Indian urban population was at 33% suggesting very different developmental trajectories. In fact, the World Bank, based on Census of India statistics, estimates that urban India is growing at a declining rate (3.8% in the 1970s to 2.7% in the 1990s and 2000s, to 2.4% in the 2010s) (World Bank 2018). Widely varying estimates of such rates can be found from other sources. United Nations figures rely on national statistics that themselves are generated by a wide diversity of definitions of urban, leading to incomparable estimates of urban population and urbanization rates across countries (Uchida and Nelson 2010). In addition, a long-running debate exists in the literature about the relationship between urbanization of a country’s population and its economic growth (Fay and Opal 2000; Henderson 2003; Spence et al. 2009). While higher levels of urbanization are observed in countries with higher per-capita GDP, the rates of urbanization have little correlation to economic growth (Bloom et al. 2008; Chen et al. 2014).
Yet much of this literature presumes that urbanization levels, along with the GDP, are measured consistently and appropriately in different contexts (Satterthwaite 2007). Cross-country consistency in urban definitions is necessary for the design and study of urban policies that may vary by nation, such as the organization of public services or the allocation of development finance towards meeting international development goals (OECD 2012). For example, the Sustainable Development Goal 11, to “Make cities and human settlements inclusive, safe, resilient and sustainable”, is associated with a number of indicators and targets, the measured values of which can change substantially when applying different definitions and delineations of cities (Klopp and Petretta 2017).
Definitional differences are not just a matter for comparative convenience; they have both theoretical and policy implications. Studies of agglomeration economies and the determinants of urban economic growth in India often use districts as units of analysis due to a lack of availability of consistent boundaries for metropolitan areas, which would be a more appropriate unit for such research questions (e.g., Desmet et al. 2015; Duranton and Puga 2004; Ghani et al. 2016). This problem could potentially lead to misleading conclusions in cross-country comparative work. For example, Chauvin et al. (2017) conclude that India does not conform to spatial equilibrium, a central idea in urban economics, in a comparative analysis of India, Brazil, China, and the USA. In this study, districts were the unit of analysis for India, while units more analogous to Metropolitan Statistical Areas in the other three countries were used. In contrast, Hasan et al. (2017) find evidence of relatively low agglomeration economies in India based on town and city-level data, but do not account for how such towns may be part of larger metropolitan regions in their analysis.
From a governance standpoint, the delineation of urban areas has consequences for the spatial distribution of infrastructure provision and related institutional arrangements. Urban areas are seen as engines of economic development and infrastructural and resources are concentrated on them (Indian Planning Commission 2011). Even so, urban infrastructure investment is often assessed to be inadequate in India (Ahluwalia et al. 2014). Underestimating the existence of dense population clusters only exacerbates this problem by limiting the political attention, governance reform, and finance necessary to build and maintain appropriate levels of infrastructures such as intra-city transportation, water, sanitation, and health in dense, yet officially rural areas. Areas with high population density require qualitatively different types of infrastructure and necessitate different institutions to govern them than lower-density areas, regardless of whether they are administrated as urban or rural units (Rakodi and Lloyd-Jones 2002).
In India, rapid urbanization that was expected to follow economic liberalization policies starting in the 1990s was predicted to hollow out rural areas in favor of large urban areas such as Bengaluru due to migration based on economic opportunity. In part, these conclusions are drawn from undercounting urban areas and ignoring the large in situ urbanization happening over time. Denis et al. (2012) argue that close to two-fifths of the population live in urban settlements and 35% of the urbanites do live in small towns below 100,000 in population. More importantly, the patterns of urban settlements are different regionally, which also lead to regional developmental imbalances. For example, the less developed states of West Bengal and Bihar have substantially more dense settlements in the Denis et al. (2012) approach than the official estimates. Accordingly, Kundu (2011) argues that when optimistic rural-urban migration predictions were not realized, there were adverse consequences for urban livelihoods in smaller towns, which contribute little to national productivity and command little political attention. Indeed, initiatives such as the Jawaharal Nehru National Urban Renewal Mission (JNNURM), one of the largest infrastructure programs ever undertaken by the Government of India, allocated funds disproportionately to large urban areas and may have caused stagnation in smaller towns and their surrounding rural areas (Khan 2016).
Underbounding metropolitan areas has a related policy consequence when combined with India’s federalist governance structure. The 73rd and 74th Constitutional Amendments of 1993 devolved many planning and infrastructure provision responsibilities from state to local governments, including urban local bodies (ULBs) for officially urban areas and gram panchayats for officially rural areas. This devolution in some circumstances allowed local communities to organize appropriate institutions and infrastructure packages (Hutchings 2018). However, it also raises barriers for coordination between communities in the provision of some public goods or the management of shared common-pool resources. For example, the highly administratively fragmented Kochi urban area saw many JNNURM projects delayed or applications rejected due to competing priorities and conflict between the Kochi Municipal Corporation and surrounding ULBs and gram panchayats in the region (Kamath and Zachariah 2015). Such phenomena highlight the potential gains to be had from more regional planning structures that incorporate all neighboring clusters of high-density jurisdictions (whether administered by ULB or panchayat) into related infrastructure needs, as suggested by Mukhopadhyay et al. (2017).
The lack of a georeferenced and consistently delineated dataset also poses a problem for studying urban change over time. Official estimates put change in Indian urban population at 3.3% between 2001 and 2011, with 29.5% of this urban growth due to reclassification of rural areas into Census Towns by the Census of India, rather than expansion or densification of existing urban areas. This is higher than the growth in urban population attributable to migration (Pradhan 2013). However, the significance of these invisible urban villages, classified as urban by the Census but administered as rural areas, is not readily apparent due to the unavailability of appropriate georeferenced datasets. Since no fine-grained geographic and demographic data are readily available, researchers have to look for clues in various census tables to locate and measure the extent of such in situ urbanization. In this paper, we aim to make this urbanization visible, so that appropriate political and economic institutions can be fashioned to meet their governance needs.
There is no consistent definition of what constitutes an urban area around the world (Buettner 2015; Cohen 2006; Satterthwaite 2007). Previous efforts to define consistent, global definitions of urban area relied on daytime satellite images (Angel et al. 2011), nighttime lights (Zhou et al. 2011), functional integration (OECD 2012), and population density combined with travel times to the nearest large city (Uchida and Nelson 2010). Others have followed a more hierarchical definition of classifying the urban areas based on density, the proportion of the population living in different density clusters, population size, and contiguity characteristics (Dijkstra and Poelman 2014).