More urbanization, more polarization: evidence from two decades of urban expansion in China

Rapid urban expansion in developing countries, particularly in mega-metropolitan areas, raises concerns about urban polarization—the increasing disparity in city size distributions. This uneven development can exacerbate socioeconomic disparities and potentially lead to social instability. The research community is divided on the existence and extent of this polarization. Some studies suggest significant polarization in past decades, while others indicate stable city size distributions. China, with its massive urbanization and planned development efforts, presents a critical case study. Between 1978 and 2015, the number of cities in China more than tripled, leading to numerous mega-cities and concerns about polarization. China's 2014 New Urbanization Plan aimed to mitigate polarization through spatially targeted interventions, coordinated urban agglomerations, and improved interregional economic equality. However, comprehensive comparative analyses of city-size development across all Chinese urban agglomerations are limited. This study addresses this gap by investigating whether polarization is occurring within and across urban agglomerations in China, considering the impact of the nation's urbanization policies. Traditional population-based measures of city size have limitations, such as underrepresenting temporary migrants. Therefore, this study utilizes built-up area (BUA) as a more robust indicator of urban size, leveraging satellite remote sensing data and machine learning techniques to assess BUA expansion from 1995 to 2018 across 23 urban agglomerations. The study employs the primacy index and Pareto exponent to analyze city-size dynamics and assess polarization.

Existing research on urban polarization and city size distributions offers conflicting conclusions. Some studies highlight increasing polarization, particularly in previous decades, emphasizing the disproportionate growth of mega-cities compared to smaller urban centers. These studies often point to factors such as agglomeration economies and national policies favoring uneven development. Other research, however, suggests that city size distributions have remained relatively stable over time, indicating a lack of significant polarization. The debate is further complicated by methodological differences, with studies employing various measures of city size (population versus built-up area) and different methodologies for analyzing city-size distributions. Studies examining China’s urbanization often focus on specific agglomerations, particularly large coastal ones, hindering a comprehensive national perspective. The current study aims to contribute to this debate by adopting a more holistic approach, examining city-size dynamics across all designated urban agglomerations in China and employing a novel measure of city size (built-up area). Furthermore, it considers the impact of China’s state-led urbanization policies on the observed trends, adding another dimension to the existing literature.

This study utilizes high-resolution satellite remote sensing data to analyze the built-up area (BUA) of 23 urban agglomerations in China from 1995 to 2018. The data are derived from Landsat imagery and nighttime light imagery, processed using a machine-learning algorithm (Random Forest classifier) trained with nighttime light data as a proxy for built-up areas. The algorithm classifies pixels as built-up or non-built-up based on spectral indices and various Landsat bands. The accuracy of the classification was validated using hand-labeled data from Google Earth imagery, yielding an overall accuracy consistently above 90%. The 23 urban agglomerations examined were those identified by the Chinese government in the 11th Five-Year Plan (2006-2010), categorized into coastal, northern inland, and southern inland groups based on their geographic location and economic linkages. The study employs two key metrics to analyze city-size distributions and polarization. The primacy index is used to measure the degree of dominance of the largest city within each agglomeration. A higher primacy index indicates greater polarization within the agglomeration. The Pareto exponent is used to analyze the rank-size distribution of cities within the three largest agglomerations (Yangtze River Delta, Pearl River Delta, and Beijing-Tianjin-Hebei), providing a more comprehensive measure of city-size distribution and polarization. A higher Pareto exponent signifies a more even city size distribution, while a lower exponent indicates greater polarization. The mean value of the rank (MVR) of cities within each of these three agglomerations is also calculated to track changes in the overall ranking of cities over time. The annual growth rate (AGR) of BUA is calculated for each agglomeration to assess the relative growth rates of larger versus smaller agglomerations. The analysis also accounts for potential spatial autocorrelation in city rankings to rule out spatial planning as a primary driver of city-size changes.

The study reveals a dramatic increase in China's total BUA, rising 287.60% from 1995 to 2018. This growth is even more pronounced within the 23 urban agglomerations examined, exhibiting a 280.90% increase in BUA. While overall BUA increased substantially, the study reveals a significant spatial disparity in growth patterns. Coastal agglomerations consistently exhibited significantly larger BUAs and higher annual growth rates compared to inland agglomerations. This gap widened over the study period, signifying increasing polarization *among* urban agglomerations. However, the analysis of city-size distributions *within* urban agglomerations reveals a different picture. The primacy index decreased over time in most agglomerations, suggesting a decline in the dominance of primate cities and a movement towards more uniform city-size distributions. This trend is supported by the Pareto exponent analysis of the three largest agglomerations. The Yangtze River Delta and Beijing-Tianjin-Hebei agglomerations showed increasing Pareto exponents, signifying a more even city-size distribution. In contrast, the Pearl River Delta showed a decreasing Pareto exponent, indicating increasing polarization within that specific agglomeration. Analysis of the mean value of the rank (MVR) of cities within these three agglomerations further supports these findings. The Yangtze River Delta and Beijing-Tianjin-Hebei showed decreasing MVR, reflecting improvements in the ranks of smaller cities, while the Pearl River Delta showed an increasing MVR, reflecting the dominance of larger cities. These findings indicate that while policies aimed at promoting balanced regional development might be successful at reducing polarization within agglomerations, they have not been entirely effective in addressing the growing gap between coastal and inland regions. The study notes that the differing growth trends across the three major agglomerations are possibly influenced by factors such as proximity to economic centers (Yangtze River Delta), political advantages (Beijing-Tianjin-Hebei), and the spillover effect from Hong Kong and Macao (Pearl River Delta).

The findings of this study provide valuable insights into the complex dynamics of urban development in China. The observation of increasing polarization *between* coastal and inland urban agglomerations despite decreasing polarization *within* individual agglomerations highlights the limitations of solely focusing on intra-regional policies for balanced urbanization. The contrasting trends in the three largest agglomerations further underscore the role of factors beyond national policies in shaping city-size distributions. Agglomeration economies, transportation improvements, differential infrastructure investment, and migration policies have all played crucial roles, as has the legacy of the historical Huhuanyong Line. The discrepancy between this study's findings on Pearl River Delta and previous research could be attributed to differences in data sources and methodologies used for assessing built-up areas. The use of high-resolution remote sensing data and a sophisticated machine-learning algorithm in this study allows for a more accurate and detailed assessment of BUA changes, capturing the nuances of urban expansion better than previous studies based on lower resolution data. The study's findings call for a more nuanced approach to urban planning in China, one that addresses both intra- and inter-regional disparities. Future policies should consider targeted interventions beyond individual agglomerations, promoting balanced development between coastal and inland regions.

This study demonstrates a complex interplay between national urban planning policies and the inherent dynamics of urban growth in China. While policies appear to have had some success in reducing polarization within urban agglomerations, they have been less effective in addressing the growing gap between coastal and inland regions. The differing patterns observed among the three largest agglomerations underscore the need for a more nuanced understanding of the factors influencing city size distributions. Further research should investigate the long-term impacts of current policies and explore alternative strategies to promote more sustainable and equitable urban development in China. Improving data resolution and incorporating socioeconomic factors into future research could further enhance the analysis and provide more comprehensive insights into this critical issue.

This study focuses solely on built-up area as a proxy for city size, neglecting other important dimensions such as population, economic activity, and social infrastructure. The reliance on satellite imagery and machine learning introduces potential biases and inaccuracies in the estimation of BUA. While the validation process achieved high accuracy, there might still be some degree of uncertainty associated with the classification. Furthermore, the study's temporal scope is limited to the period between 1995 and 2018. Extending this analysis to cover a longer period or incorporating future projections could provide a more comprehensive picture of the long-term trends of urbanization and its impact on city-size distributions in China. Finally, the study does not explicitly model the impact of specific policy instruments on urban expansion patterns. Future studies might benefit from incorporating policy variables into regression models to quantify policy effectiveness and assess their contribution to urban polarization.