Radical Urbanization and Economic Growth Quality: Evidence from Nighttime Light and FDI Flow Dynamics

Abstract

This study systematically examines the impact of radical urbanization on the quality of economic growth using city-level data from 290 major prefecture-level cities in China during 2003–2019. A comprehensive indicator system for economic growth quality is constructed using PCA, capturing multiple dimensions of efficiency, stability, and sustainability. Nighttime light data obtained from the NOAA is extracted and calibrated, and the ratio of urban built-up area to nighttime light intensity is employed to measure the degree of radical urbanization. Empirical results reveal a divergence between economic quantity and quality effects: while radical urbanization promotes economic expansion, it significantly inhibits the quality of economic growth. To address potential endogeneity concerns, the change in FDI relative to changes in built-up area is used to capture FDI flow direction, with its one-period lag serving as an instrumental variable. Mechanism analysis, based on an interaction-based identification framework, shows that radical urbanization suppresses growth quality primarily through two transmission channels: reduced fiscal output efficiency and declining land use efficiency. Further analysis indicates that radical urbanization crowds out science and education expenditures, weakening fiscal effectiveness and reinforcing the identified transmission mechanisms. These findings provide objective evidence for evaluating urbanization strategies and offer policy insights for promoting quality-oriented and sustainable urban development.

1. Introduction

Rapid urbanization has become one of the most transformative forces shaping economic development in developing economies. However, urban expansion does not always proceed in coordination with population concentration and industrial upgrading. This paper focuses on a specific development pattern termed radical urbanization, defined as a structural mismatch between spatial urban expansion and the evolution of demographic and economic fundamentals. Radical urbanization, as a distinct phenomenon within rapid urbanization processes, is characterized by the misalignment between urban spatial expansion and the pace of population agglomeration and industrial development. Specifically, it manifests as a disconnect between land urbanization and population or industrial urbanization, where new urban areas expand while urban population growth slows and industrial development lags behind [1].

Unlike conventional urbanization driven by labor migration and productivity gains, radical urbanization is primarily policy-led and investment-driven, emphasizing physical expansion of built-up land rather than endogenous economic transformation. In recent years, many developing countries have experienced rapid economic growth and expanding trade volumes, yet the quality of economic growth has not improved proportionally with quantitative expansion, revealing a notable inconsistency between economic quantity and quality.

It is important to distinguish radical urbanization from related concepts widely discussed in urban economics. Urban sprawl typically refers to low-density spatial expansion driven by market forces and residential preferences, while premature urbanization describes population concentration occurring before industrialization. Similarly, the “ghost city” phenomenon captures localized outcomes of excess housing supply. In contrast, radical urbanization captures a systemic development pattern in which government-led land expansion advances faster than both population absorption and industrial restructuring, generating persistent resource misallocation rather than temporary spatial imbalance. The existing literature presents divergent views on the economic effects of such development patterns.

On one hand, Fallah et al. [2] argue that inefficient public service provision in government-led land development reduces the allocation efficiency of social capital. The radical urbanization model leads to extensive land idling and resource waste, suppressing total factor productivity and ultimately undermining economic growth efficiency. Similarly, Xu et al. [3] contend that radical urbanization diverts substantial credit resources to infrastructure investment rather than productive sectors. Such overinvestment may crowd out non-productive public expenditures, hindering the cultivation of new drivers for sustainable growth. On the other hand, Davis and Dingel [4] offer a contrasting perspective, emphasizing a selection effect whereby urbanization-induced demand expansion attracts capital and technology flows that may foster industrial upgrading and long-term structural transformation.

These competing perspectives suggest that the consequences of radical urbanization cannot be fully understood through aggregate growth indicators alone. While spatial expansion may stimulate short-term economic output, its implications for growth quality remain theoretically ambiguous and empirically underexplored. The existing literature has predominantly focused on quantitative economic growth, whereas analyses of growth quality remain largely theoretical and lack systematic empirical validation. Research examining the mechanisms linking radical urbanization and growth quality remains fragmented, with limited attention to intermediate transmission channels.

In reality, urbanization influences economic development through both quantitative expansion and qualitative transformation, yet these dimensions often evolve asynchronously. In many rapidly urbanizing developing economies, improvements in infrastructure and construction activity coexist with declining efficiency, fiscal pressure, and environmental constraints. We therefore contend that radical urbanization exhibits an inconsistent dual impact: while it enhances quantitative growth, it may simultaneously suppress improvements in economic growth quality.

Two primary transmission mechanisms underlie this phenomenon. First, the fiscal distortion channel arises when fiscal interventions prioritize urban construction investment, distorting capital costs and marginal returns. Repeated investments and resource waste diminish growth efficiency, while capital misallocation undermines macroeconomic stability. Second, the land-use efficiency channel emerges from poorly coordinated land development, where redundant industries monopolize land resources and reduce land-use intensity. The lagged adjustment of production factors constrains their ability to support high-quality output, ultimately impeding sustainable development. Together, these mechanisms explain how policy-driven spatial expansion may generate short-term growth momentum while weakening long-run economic quality.

China’s distinctive urbanization trajectory and its policy transition from high-speed growth toward high-quality development provide an ideal empirical context. This study analyzes city-level data from 290 Chinese prefecture-level cities during 2003–2019, incorporating calibrated NTL imagery from NOAA and fixed-effects estimation methods. China represents a particularly relevant case because land urbanization has systematically outpaced population urbanization, producing observable human-land mismatches with broader relevance for developing economies.

The National New Urbanization Plan (2014–2020) explicitly acknowledges that land urbanization has outpaced population urbanization, with the land-to-population urbanization ratio rising from 0.5 in 2001 to 0.7 by 2020. This divergence manifests in imbalanced real estate markets and spatial mismatches between land allocation and population mobility, producing patterns described as urbanization without industrialization or strong production bases with weak urban functionality.

A key methodological challenge in studying radical urbanization is measurement. This paper does not directly analyze nighttime light (NTL) intensity as an outcome variable; instead, it employs NTL brightness as a proxy for the spatial distribution and intensity of economic activity, allowing urban land expansion to be evaluated relative to its economic base. Satellite-based NTL data provide a consistent and objective indicator of human economic activity intensity across regions and time, overcoming limitations of administrative statistics. The NTL dataset is calibrated through three steps: (i) cross-satellite harmonization to ensure comparability across regions and years; (ii) correction of multi-satellite observations to remove light spillover and unstable noise sources; and (iii) adjustment of abnormal fluctuations to reduce measurement errors. Radical urbanization is then constructed as the logarithm of the ratio between urban built-up area and calibrated NTL brightness, capturing the extent to which spatial expansion exceeds underlying economic activity intensity. Following Kang et al. [5], this approach interprets excessive land expansion relative to economic luminosity as evidence of spatial-economic mismatch.

NTL intensity is strongly correlated with economic activity, investment expectations, and capital flows. Foreign direct investment (FDI) responds to anticipated economic vitality, yet its adjustment typically occurs with a temporal lag. Exploiting this relationship, the study constructs an instrumental variable using the lagged ratio of changes in international direct investment flows to changes in urban built-up area (L.invflows). This variable reflects external investment expectations linked to economic activity intensity while remaining plausibly exogenous to contemporaneous growth-quality outcomes, thereby strengthening causal identification.

This study utilizes city-level data from 290 prefecture-level cities from 2003 to 2019. Through Principal Component Analysis (PCA), we construct a comprehensive economic growth quality index encompassing three dimensions: growth efficiency, stability, and sustainability, based on 19 indicators. Using ArcGIS 9.2, NTL data from DMSP/OLS and NPP/VIIRS systems are calibrated to generate a unified long-term luminosity dataset.

Empirical results show that radical urbanization significantly inhibits overall economic growth quality. Disaggregated analysis indicates negative effects on growth efficiency and stability but a positive effect on sustainability. Mechanism tests reveal that fiscal output decline and reduced land-use efficiency are key transmission channels, amplifying resource misallocation. Heterogeneity analysis shows stronger negative impacts in cities with higher marketization levels, greater population density, larger land quotas, and intensive infrastructure investment. Exclusion tests confirm that non-productive public expenditure improves growth quality but is crowded out under radical urbanization.

Compared with the existing literature, this study makes four primary contributions. First, it advances urban economics research by empirically linking radical urbanization to economic growth quality rather than aggregate growth levels, providing direct quantitative evidence on qualitative development outcomes. Second, it proposes a novel measurement framework that combines calibrated NTL data with spatial land expansion indicators to capture structural mismatches between economic activity and urban expansion. Third, the study identifies and empirically verifies fiscal distortion and land-use efficiency as causal transmission mechanisms, clarifying how spatial policies translate into macroeconomic quality outcomes. Fourth, by constructing a multidimensional growth-quality index using PCA at the city level, the study offers a replicable methodological framework applicable to other developing economies undergoing rapid urban transformation.

The remainder of this paper is organized as follows. Section 2 reviews the literature. Section 3 presents the empirical methodology and variable construction. Section 4 reports baseline regression results. Section 5 provides further analysis, and Section 6 concludes with policy implications and research contributions.

2. The Literature Review

2.1. Economic Growth, Quality, and Measurement

The conceptualization of economic growth quality has evolved from narrow welfare-based interpretations toward multidimensional analytical frameworks. Early studies, such as Thomas et al. [6], defined growth quality primarily in terms of welfare improvement and equitable distribution outcomes. Barro [7] expanded this perspective by incorporating broader social development indicators, including education attainment, healthcare conditions, and institutional environments. While these approaches enriched the concept normatively, they also introduced excessive breadth, making empirical operationalization difficult.

Recent studies have therefore shifted toward evaluating the intrinsic characteristics of economic growth itself. Chao and Ren [8] proposed a framework emphasizing growth structure, stability, outcomes, and efficiency, which has been widely adopted in subsequent empirical research [9]. This stream of the literature reorients growth quality toward the sustainability of the growth process rather than solely its distributive consequences. Following this approach, the present study defines growth quality through three core dimensions: efficiency, stability, and sustainability, examining whether economic expansion is productive, resilient, and environmentally viable over time.

Regarding measurement methodologies, existing studies generally adopt either single-indicator proxies or comprehensive evaluation systems. Glaeser and Kahn [10] used total factor productivity as a proxy for growth quality, while Henderson et al. [11] employed the investment–output ratio. However, single indicators capture only partial aspects of economic performance and may obscure multidimensional trade-offs across growth processes. Consequently, multi-indicator evaluation systems have become dominant. Fan et al. [12] constructed composite indices using covariance structures, and Zhou et al. [13] incorporated sustainability indicators into broader evaluation frameworks. Given the multidimensional nature of growth quality, composite indices derived from PCA provide a more comprehensive and empirically robust measurement strategy [14].

2.2. Radical Urbanization and Spatial Measurement

Radical urbanization, a distinctive phenomenon within rapid urbanization processes, refers to the systematic divergence between spatial urban expansion and the evolution of population concentration and industrial development. Peng and Lu [15] describe it as a development pattern in which the growth rate of built-up areas exceeds that of permanent residents. Unlike market-driven urban expansion, radical urbanization is typically associated with policy-led land development and infrastructure investment, generating spatial growth that outpaces underlying economic fundamentals.

The existing literature often conflates radical urbanization with related concepts such as urban sprawl or premature urbanization. However, radical urbanization differs in both mechanism and outcome: it emphasizes institutional incentives and fiscal dependence on land expansion rather than demographic decentralization or early population urbanization. This distinction is essential for empirical identification and theoretical interpretation.

Measurement approaches have evolved substantially. Earlier studies relied mainly on statistical yearbook indicators, which suffer from inconsistencies across regions and reporting biases. The emergence of satellite remote sensing data has fundamentally transformed urbanization measurement by enabling spatially consistent observation of human activity.

Elvidge et al. [16] pioneered the use of NTL data as a proxy for economic activity intensity, while Kang et al. [5] quantified radical urbanization by matching luminosity with population spatial distribution. NTL data have since become an established empirical tool because light emissions strongly correlate with electricity consumption, production intensity, and urban economic activity.

A growing body of research demonstrates the validity of NTL data across contexts. Henderson et al. [11] showed that nighttime lights provide reliable proxies for regional economic performance. Storeygard [17] employed NTL data to measure urban revenue patterns in Sub-Saharan Africa, while Zhou et al. [18] applied NTL data to examine urban expansion dynamics in Vietnam. Studies across South Asia [19], the pan-Pacific region [20], and the Taiwan Strait region [21] further confirm that luminosity data effectively capture spatial economic dynamics and urbanization heterogeneity.

Importantly, this paper does not treat NTL intensity as a direct outcome variable. Instead, it interprets luminosity as a benchmark for economic activity intensity against which spatial expansion can be evaluated. By comparing built-up land expansion with calibrated light brightness, the study identifies spatial-economic mismatches that characterize radical urbanization.

NTL data have also been widely applied to analyze urban infrastructure transitions [22], urban spatial boundaries [23], city-center detection [24], population distribution [25], and socioeconomic dynamics [26]. Studies focusing on China further validate their usefulness. Liu et al. [27] analyzed urban space quality in Northeast China using NTL-based spatial models, while Nie [28] demonstrated that corrected luminosity measures reliably proxy city-level economic activity after addressing top-coding problems. Collectively, the literature establishes NTL data as a credible and increasingly indispensable measurement tool for spatial economic analysis.

2.3. FDI Dynamics and Urbanization

Existing research also highlights the role of foreign direct investment (FDI) in shaping urban development trajectories. FDI flows respond to expected economic activity rather than contemporaneous outcomes, making them closely linked to anticipated urban economic vitality. Cao and Duan [29] show that FDI is strongly associated with industrial restructuring and urbanization dynamics in coastal Chinese provinces. Their findings suggest that industrialization and urbanization jointly influence investment inflows.

Because NTL intensity reflects realized economic activity while FDI captures forward-looking investment expectations, the two variables exhibit correlation with temporal lag structures. This property provides an empirical basis for using lagged FDI-related indicators as instrumental variables. In empirical urban economics, lagged capital flows are frequently used to address endogeneity arising from reverse causality between development outcomes and investment allocation.

2.4. Economic Effects of Radical Urbanization

The economic consequences of radical urbanization remain debated. One strand of the literature emphasizes efficiency losses. Fallah et al. [2] argue that government-led land development reduces social capital allocation efficiency through diseconomies in public service provision. Vu [30] highlights excessive land concentration and weak spatial planning, which reduce land-use intensity and generate underutilized urban space. Xu et al. [3] further show that infrastructure-oriented credit allocation crowds out productive investment and weakens long-term growth drivers.

Conversely, Davis and Dingel [4] identify potential positive effects through demand expansion and selection mechanisms, whereby urbanization attracts capital and technology flows that may promote industrial upgrading and structural transformation. These conflicting findings suggest that urban expansion may simultaneously generate short-term growth stimulus and long-term efficiency challenges, reinforcing the need to evaluate growth quality rather than output alone. Table 1 below synthesizes the existing literature.

Table 1. Synthesizes representative studies relevant to this research.

Study	Focus	Method/Data	Key Finding	Limitation
Henderson et al. [11]	Economic activity measurement	Nightlight data	NTL proxies GDP	Limited quality dimension
Storeygard [17]	Urban spatial growth	NTL data	Transport affects city growth	No growth-quality analysis
Fallah et al. [2]	Land development	Panel analysis	Efficiency loss	Mechanisms incomplete
Davis & Dingel [4]	Urban demand effects	Structural analysis	Possible upgrading effects	Quality outcomes unclear
Cao & Duan [29]	FDI and urbanization	Panel models	FDI linked to urban structure	No spatial mismatch analysis
Yun et al. [21]	Spatiotemporal urbanization dynamics	Nighttime light (NTL) remote sensing data	Urbanization shows significant spatial clustering and temporal expansion patterns	No direct economic growth, quality, or mechanism analysis

The synthesis reveals three major limitations in the existing literature. First, studies on urbanization primarily focus on growth quantity rather than growth quality outcomes. Second, research using NTL data typically measures economic activity but rarely integrates multidimensional growth-quality frameworks. Third, transmission mechanisms linking spatial expansion to fiscal performance and land-use efficiency remain insufficiently tested empirically.

2.5. Research Gap and Analytical Framework

Despite substantial progress, the literature remains fragmented along three dimensions. First, empirical studies largely emphasize aggregate growth performance, leaving growth quality insufficiently tested using rigorous quantitative frameworks. Second, research on radical urbanization and growth quality has evolved independently, with limited integration of spatial data and multidimensional economic indicators. Third, heterogeneity and causal transmission mechanisms remain underexplored.

This study addresses these gaps by integrating calibrated NTL data with a multidimensional growth-quality index at the city level. Radical urbanization is measured as the logarithmic ratio between urban built-up area and economic luminosity, capturing spatial–economic mismatch. Fixed-effects estimation is combined with an instrumental-variable strategy based on lagged FDI-flow dynamics to mitigate endogeneity concerns.

In summary, existing research establishes important foundations but lacks a unified empirical framework capable of linking spatial urban expansion, economic activity intensity, and multidimensional growth quality through identifiable causal mechanisms. Building on these gaps, the present study develops an integrated empirical strategy that combines satellite-derived spatial indicators, composite growth-quality measurement, and mechanism testing focused on fiscal output efficiency and land-use productivity. The following section introduces the data, variable construction, and econometric methodology employed to implement this framework.

3. Materials and Methods

3.1. Measurement Model Setting

This study employs a fixed-effects model to assess the impact of radical urbanization on economic growth quality across Chinese prefecture-level cities, with the baseline specification formulated as follows:

$$qualit{y}_{it}=c_0+{\alpha }_{1}rad\text{\_}u{r}_{it}+{\beta }_{1}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(1)

In the specified model, quality_it represents the composite index of economic growth quality, while ${rad\text{\_}ur}_{it}$ serves as the core explanatory variable measuring radical urbanization. The control variables ($control$) primarily include trade dependence ($trade$), non-agricultural sector proportion ($non\text{\_}ag$), urban scale ($scale$), freight volume ($ft\text{\_}vol$) and human capital accumulation level ($university$), with subscripts $i$ and $t$ denoting city and year, respectively.

Given that the dependent variable ${quality}_{it}$ constitutes a comprehensive growth quality indicator encompassing three-dimensional sub-indices-growth efficiency (${efficiency}_{it}$), stability (${stability}_{it}$), and sustainability (${sustainability}_{it}$). We extend Model 1 by substituting these component indicators to establish distinct econometric specifications. This approach yields the following refined Models (2–4) capturing radical urbanization’s heterogeneous effects across different quality dimensions:

$$efficienc{y}_{it}=c_0+{\alpha }_{2}rad\text{\_}u{r}_{it}+{\beta }_{2}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(2)

$$stabilit{y}_{it}=c_0+{\alpha }_{3}rad\text{\_}u{r}_{it}+{\beta }_{3}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(3)

$$sustainabilit{y}_{it}=c_0+{\alpha }_{4}rad\text{\_}u{r}_{it}+{\beta }_{4}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(4)

3.2. Variable Explanation

3.2.1. Economic Growth Quality Index

Building upon the methodological framework established by Sui [8], this study adopts a growth-oriented perspective to examine systemic economic improvements with substantive economic significance. Specifically, we construct a comprehensive economic growth quality evaluation system comprising: (1) a composite growth quality index, (2) three categorical indicators (growth efficiency, growth stability, and growth sustainability), (3) nine secondary indicators, (4) nineteen underlying metrics, and (5) twenty-two proxy variables, as systematically presented in Table 2.

Table 2. Evaluation system for the quality of urban economic growth in China.

Classification Metrics	Sub-Indicators	Basic Indicators	Proxy Variables	Unit	Metric Properties
					Positive	Negative	Moderate
Growth efficiency Growth Stability Growth Sustainability	Factor productivity Production organization efficiency Market efficiency Factor supply stability Output stability Structural stability Forward-growth drivers Resource constraints Ecological cost	Capital productivity Labor productivity Total factor productivity Product market efficiency Factor market efficiency Fluctuations in capital supply Fluctuations in labor supply Fluctuations in output Price fluctuations Structure of the national economy Industrial structure Structure of income and expenditure Knowledge capital level Research and development investment Human capital level Resource utilization efficiency Degree of air pollution Sewage emissions Solid waste emissions	Capital productivity Labor productivity TFP Retail sales of social goods/gross output value of industry and agriculture Amount of deposits and loans of financial institutions/GDP Number of employment agencies per 10,000 people Technology market turnover/industrial output Actual interest volatility Volatility of average wages in urban employment Volatility in the value added of outputs Inflation rate Non-State-owned enterprises/GDP Secondary industry output/GDP Tertiary industry output/GDP (Imports–exports)/GDP Number of patents per capita Research and development spending as a percentage of GDP Average number of years of education for the population aged six and over Energy-to-energy ratio per unit of output Degree of air pollution per unit of output, Number of sewage discharges per unit of output Emissions per unit of solid waste	% % % % % Multiple % % % % % % % % % Multiple % % Multiple Multiple Multiple Multiple	√ √ √ √ √ √ √ √ √ √ √ √ √ √	√ √ √ √	√ √ √ √

3.2.2. Radical Urbanization

This study measures the degree of radical urbanization through the collection, calibration, and analysis of NTL data across Chinese regions, using urban light intensity per unit area as the core metric. Specifically, the indicator is calculated as the ratio of built-up area to NTL brightness. In constructing this radical urbanization measure, we apply a logarithmic transformation to the area-to-light ratio. While this transformation preserves the essential characteristics of the variable and maintains its correlation relationships with other factors, it effectively compresses the data scale to facilitate more reliable parameter estimation in subsequent econometric analysis.

Given the temporal discontinuity between first-generation (DMSP/OLS) and second-generation (NPP/VIIRS) NTL data, inconsistencies across multiple imagery sources, and potential measurement or recording errors, this study enhances data reliability by constructing a continuous panel dataset for robust empirical analysis. Building upon the methodological framework established by Wu et al. [31], we implement a three-stage calibration and correction procedure utilizing ArcGIS with Stata 17 integration.

The initial step involves recalibrating satellite observation data to ensure comparability across different regions. The first-generation NTL data were collected by six satellites, with F18 selected as the calibration standard. This study uses Hegang City in Heilongjiang Province as the reference sample due to its ideal observational characteristics, with other cities’ light brightness values being adjusted according to this reference benchmark.

$$D{N}_{r,hg,s-t}= {a\times D{N}^b}_{r,hg,18\text{-}2003}+\epsilon$$

(5)

Equation (5) uses F19-2003 data as the reference where i represents cities other than the Hegang reference point, and r denotes the brightness values of individual grid cells in other cities. This yields adjusted grid-cell brightness values for each city i under specific “year-satellite combinations”. The DN values of grid-cell brightness in the F18-2003 reference area serve as independent variables, combined with the corresponding reference estimates for each “year-satellite combination”, denoted by a and b, to calculate the adjusted grid-cell brightness values for other cities under given “year-satellite combinations”.

$$D{N}_{r,i,s-t}= {a\times D{N}^b}_{r,i,18\text{-}2003}+\epsilon$$

(6)

Equation (6) utilizes F19-2003 as the baseline to recalculate each brightness grid for individual cities under specific “year-satellite combinations”, thereby calibrating luminosity values to ensure comparability of light data across different regions while eliminating confounding factors such as satellite sensor degradation and observational errors.

The second step involves correcting errors caused by light spillover or noise from unstable sources in cases where multiple satellites provide data for the same region and time period. For years covered by multiple satellites, the brightness value (DN) for each frame must be calculated based on the first step’s results, where ${DNA}_{r,t}$ and ${DN}_t$ represent the recalculated brightness values of two frames from step one. If in a given year at least one of the two satellites fails to measure stable visible light for a particular grid, then D for that grid in that year is set to 0, which may result from light spillover or failure to eliminate noise from unstable sources.

$${0 if{DN}_{r,t}^A=0 or {DN}_{r,t}^B=0; DN}_{r,t}=\left\{{DN}_{r,t}^A\right.+{DN}_{r,t}^B/2,\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{w}\mathrm{i}\mathrm{s}\mathrm{e}$$

(7)

The third step corrects abnormal fluctuations in the data to avoid unreasonable measurement errors. When a grid cell appears luminous in the first year but not in the second year, or when the DN brightness value in the second year is lower than that of the previous year, such “abnormal fluctuations” require consideration. Given China’s recent economic development and widespread urban population growth, a decrease in grid-cell brightness or its disappearance in the following year constitutes an anomaly likely caused by measurement errors, necessitating data correction. However, these two types of anomalies have distinct causes and require different treatments. The first case involves alternation between luminous and non-luminous states, potentially caused by excessive illumination or measurement errors. For this scenario, where no actual grid exists, all years’ data for such grid cells are adjusted to non-luminous status. The second case, where brightness decreases, likely results from sensor underestimation or special circumstances like earthquakes, abandoned factories, or fires that reduce subsequent years’ illumination. Here, the following year’s DN brightness value is corrected to match the previous year’s value. Grid cells not exhibiting “abnormal fluctuations” remain unadjusted. The calculation results are shown in Equation (8).

$${DN}_{rr,tt}=\left\{\begin{array}{c}0,{DN}_{rr,tt+1}=0\\ {DN}_{rr,tt-1},\\ {DN}_{rr,tt}, Otherwise \end{array}\right.\mathrm{when} {DN}_{rr,tt+1}> 0 \mathrm{and} {DN}_{rr,tt-1}>{DN}_{rr,tt}$$

(8)

Through the aforementioned three-step calibration process, most measurement errors potentially existing in the original first-generation NTL data, including recording errors, have been eliminated. The saturation values, along with the accuracy, authenticity, and comparability of the data, have been significantly improved. Background noise and outliers in the second-generation nighttime imagery were removed using threshold methods, while a logarithmic transformation was applied to compress the dynamic range of the radiance values. The spatial resolution was standardized, making its brightness distribution more consistent with the first-generation nighttime imagery. The detailed process of calibrating the second-generation data based on the first-generation nighttime lights as a reference will not be further elaborated here.

By applying a fitted logarithmic regression model to calibrate NPP/VIIRS data to the DMSP/OLS scale, we obtained a continuous long-term NTL imagery series (1992–2019) for China after consistency correction between DMSP/OLS and NPP/VIIRS images. The calibration results achieved continuity correction of NTL data, which holds significant importance for extending and deepening the application research of long-term NTL imagery. This process enabled the integration of first-generation and second-generation datasets into a complete nighttime dataset suitable for computation. From this merged dataset, we extracted city-level data for 2003–2019, providing a reliable data foundation for the empirical analysis in this study.

3.2.3. International Direct Investment Flow

The change in FDI flows reflects the movement patterns of outward investments. Built-up areas represent fixed urban characteristics, specifically referring to contiguous urban spaces within administrative boundaries that possess comprehensive public services and infrastructure. The change in urban built-up area reflects variations in urbanization levels. This study measures the direction of FDI flows using the ratio between these two change values, capturing how international investment flows shift with urbanization level changes. This measurement serves as an instrumental variable to identify and analyze the impact of radical urbanization on economic growth quality.

3.3. Descriptive Statistics

This study focuses on prefecture-level cities in China during 2003–2019, using panel data from 290 cities (4794 observations). Table 3 presents the descriptive statistics of the main variables. Variables show substantial variation across cities and over time. rad_ur averages 1.746 (S.D. 1.050), indicating moderate dispersion in urbanization. Efficiency and stability have near-zero means but high variability, reflecting differences in efficiency and fluctuations in stability. Sustainability and quality are also centered near zero with notable spread. Trade openness (trade) is generally low but reaches 2.491 in some cities, while non-agricultural employment (non_ag) and city scale (scale) vary widely, indicating diverse economic structures. Innovation indicators, such as ft_vol and university, also differ markedly, suggesting heterogeneous capacities across regions. Overall, these statistics highlight significant heterogeneity, supporting the use of panel models that capture both cross-sectional and temporal variation.

Table 3. Descriptive statistics of key variables.

Variable Name	(1) Sample Size	(2) Average Value	(3) Standard Deviation	(4) Minimum Value	(5) Maximum Value
rad_ur	4794	1.746	1.050	−1.918	4.128
eficiency	4794	0.006	1.347	−7.496	15.471
stability	4794	0.002	0.983	−24.74	35.383
sustainability	4794	−0.008	1.138	−2.036	3.397
quality	4794	0.009	1.268	−4.527	12.333
trade	4794	0.046	0.114	0.001	2.491
non_ag	4794	0.352	0.226	0.044	1
scale	4794	5.849	0.753	−3.219	8.154
ft_vol	4794	8.828	1.260	−5.449	13.226
university	4794	1.606	0.932	0.000	4.543

3.4. Data Source

The variables quality, efficiency, stability, and sustainability report the weighted calculation results of growth quality and its sub-indicators, serving as proxy variables for economic growth quality. The fundamental data for FDI flow direction and control variables (trade, non_ag, scale, ft_vol, university) primarily originate from sources including the China City Statistical Yearbook, as detailed in Table 4. The variable rad_ur represents the standardized radical urbanization indicator, calculated as the logarithm of the ratio between built-up area and NTL intensity. The NTL imagery data were mainly obtained from the official website of the National Oceanic and Atmospheric Administration (NOAA) at https://www.ngdc.noaa.gov/eog/download.html (accessed on 11 February 2025), with supplementary data during processing and correction coming from the National Geomatics Center of China. The study period covers 2003–2019 to ensure consistency between calibrated NTL datasets and harmonized prefecture-level statistical indicators. After 2019, structural disruptions associated with the COVID-19 pandemic and adjustments in statistical reporting practices introduced abnormal fluctuations in economic activity and NTL intensity, which may affect comparability and bias long-run estimation results. Therefore, the analysis focuses on the longest balanced panel with reliable and methodologically consistent data availability. Due to occasional missing indicators, minor variations exist in sample sizes across variables, though overall consistency is maintained. For limited missing data points, this study employed equation estimation and interpolation methods for imputation.

Table 4. Data source description.

Data Type	Year	Data Source
City-level key economic variables	2003–2019	China Statistical Yearbook; China City Statistical Yearbook; China County Statistical Yearbook; China Population and Employment Statistical Yearbook; Statistical Bulletin of China’s Outward Foreign Direct Investment; China Urban Construction Statistical Yearbook, etc. Statistical Communique on National Economic and Social Development of Corresponding Prefecture-level Cities (2004–2020)
DMSP/OLS nighttime light imagery	2003–2013	Official website of the National Oceanic and Atmospheric Administration (NOAA) (https://www.ngdc.noaa.gov/eog/download.html) (accessed on 11 February 2025)
NPP/VIIRS nighttime light imagery	2012–2019	The official website of the National Oceanic and Atmospheric Administration (NOAA) (https://www.ngdc.noaa.gov/eog/download.html) (accessed on 11 February 2025)
Vector data	2003–2019	National Geomatics Center of China (https://www.ngcc.cn) (accessed on 11th February 2025)

4. Results and Discussion

4.1. Baseline Regression

Prior to examining the impact of radical urbanization on economic growth quality, this study first verifies its effect on economic growth quantity. Using data from 2003 to 2019, the dependent variables for economic growth quantity are proxied by gross domestic product (gdp) and per capita GDP (pergdp). The explanatory variable is radical urbanization, measured by taking the logarithm of the ratio between built-up area and NTL intensity (rad_ur) to facilitate observation of regression results and mitigate the influence of magnitude differences. The NTL data were derived from calibrated first-generation DMSP/OLS and second-generation NPP/VIIRS datasets. The baseline regression model is specified as follows:

$$gd{p}_{it}=c_0+{\alpha }_{5}rad\text{\_}u{r}_{it}+{\beta }_{5}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(9)

$$pergd{p}_{it}=c_0+{\alpha }_{6}rad\text{\_}u{r}_{it}+{\beta }_{6}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(10)

The results based on Models (9) and (10) are presented in Table 5. Columns (1) and (2) display the regression results of radical urbanization’s effects on GDP and per capita GDP, respectively. To better illustrate the quantitative impact on economic development, we first took the logarithm of GDP and standardized it as gdp. The regression coefficient of radical urbanization (rad_ur) on GDP is 0.2862, significant at the 1% level, indicating that radical urbanization substantially promotes aggregate economic growth. This effect likely arises because radical urbanization accelerates the circulation of capital, labor, and social resources, generating positive output effects. For per capita GDP (pergdp), the coefficient of rad_ur is 0.3937, also significant at the 1% level and higher than the GDP coefficient. This suggests that radical urbanization has a stronger impact on per capita output, reflecting the uneven distribution of growth and potential widening of income disparities across regions. Column (3) examines economic growth quality, measured by quality. Here, rad_ur has a negative and significant coefficient (−0.1658), implying that while radical urbanization boosts quantitative growth, it may reduce the qualitative aspect of economic development. This underscores that rapid urbanization can generate growth that is unbalanced, environmentally unsustainable, or socially uneven.

Table 5. Regression results of radical urbanization on economic growth quality.

Variables	(1)	(2)	(3)
	gdp	Pergdp	Quality
rad_ur	0.2862 ***	0.3937 ***	−0.1658 ***
	(20.37)	(24.65)	(−3.87)
trade	0.0564 *	−0.0483	0.1105
	(1.90)	(−1.43)	(1.22)
non_ag	0.1490 ***	−0.0025	0.7555 ***
	(4.89)	(−0.07)	(8.12)
scale	0.6307 ***	−0.1865 ***	0.3329 ***
	(15.58)	(−4.05)	(2.69)
ft_vol	0.0067 **	0.0105 ***	0.0199 **
	(2.57)	(3.54)	(2.50)
university	−0.0147	−0.0281 **	0.0218
	(−1.47)	(−2.46)	(0.71)
Cons_	11.7800 ***	10.3325 ***	−2.1300 ***
	(50.10)	(38.65)	(−2.97)
Year FE	yes	yes	yes
City FE	yes	yes	yes
N	4794	4794	4794
R-squared	0.982	0.952	0.876

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, with standard errors in parentheses.

Other control variables also show notable effects. Non-agricultural employment (non_ag) positively affects GDP and growth quality, highlighting the role of industrialization and urban labor transition. City scale (scale) strongly promotes GDP but has a negative effect on per capita GDP, suggesting that larger urban centers may experience efficiency losses or increased inequality. Foreign trade volume (ft_vol) positively contributes to all three outcomes, emphasizing the benefits of trade openness. University presence shows a negative effect on per capita GDP, possibly reflecting regional mismatches between human capital and local labor markets. Overall, these results highlight that radical urbanization fosters quantitative growth but may compromise growth quality and equity, reinforcing the need for policies that ensure more balanced and sustainable urban development. The findings are consistent with Liu et al. [32], who show that while urbanization generally promotes economic growth through agglomeration, single-center economic agglomeration may impede progress and produce regional disparities.

Based on preliminary evidence indicating a negative association between radical urbanization and economic growth quality, this study incorporates both city and year fixed effects to formally test the research hypothesis. Building on Model (3–1), baseline regressions are estimated, and Column (3) of Table 5 (1) reports the results for the composite economic growth quality index. The coefficient of radical urbanization is −0.1658, significant at the 1% level, confirming a statistically significant negative effect on growth quality. This suggests that rapid urban expansion, characterized by the spatial mismatch between built-up areas and permanent urban populations, reduces economic efficiency and coordination. The findings support theoretical expectations that uncoordinated urbanization leads to fragmented economic activity, inefficient resource allocation, and widening inequality. Overall, while radical urbanization promotes quantitative growth, it undermines growth quality, highlighting the importance of coordinated urban planning for inclusive and sustainable development. These results are consistent with Zheng et al. [33], who emphasize integrated land-use planning and sustainable growth target management to curb disorderly expansion, and Li et al. [34], who advocate region-specific urbanization strategies to enhance sustainable development outcomes.

4.2. The Test Results of the Growth Quality Classification Indicators

The baseline regression results indicate that radical urbanization negatively affects urban economic growth quality, raising an important question: whether this decline results from simple factor accumulation or from changes in efficiency, stability, and sustainability. Since economic growth quality is measured through these three dimensions, further analysis helps identify the transmission mechanisms through which radical urbanization influences overall growth quality.

Following the same analytical framework, Model (1) is extended by replacing the composite growth quality index with its three components, namely growth efficiency, stability, and sustainability, producing Models (2)–(4). The regression results in Table 6 reveal heterogeneous effects across dimensions. Column (1) shows that radical urbanization significantly reduces growth efficiency (−0.3464, 1% level), suggesting resource misallocation and declining productivity when spatial expansion outpaces industrial and population development. Column (2) reports a stronger negative effect on economic stability (−0.5995, 1% level), indicating that rapid and uneven urban expansion increases economic volatility and inequality.

Table 6. Regression results of radical urbanization’s impact on efficiency, stability, and sustainability.

Variables	(1) Eficiency	(2) Stability	(3) Sustainability
rad_ur	−0.3464 ***	−0.5995 ***	0.3005 ***
trade	(−6.34)	(−6.68)	(10.49)
	−0.3379 ***	1.2150 ***	0.0671
non_ag	(−2.94)	(6.42)	(1.11)
	0.4938 ***	0.1530	0.6152 ***
	(4.17)	(0.78)	(9.90)
scale	−0.4699 ***	−0.0170	0.0875
	(−2.98)	(−0.07)	(1.06)
ft_vol	0.0297 ***	0.0096	0.0041
university	(2.93)	(0.57)	(0.78)
	−0.0659 *	0.0316	0.1047 ***
	(−1.69)	(0.49)	(5.11)
cons_	3.0439 ***	0.9035	−1.4673 ***
Year FE	(3.33)	(0.60)	(−3.06)
	yes	yes	yes
City FE	yes	yes	yes
N	4794	4794	4794
R-squared	0.821	0.091	0.926

Note: *, *** indicate significance at the 10% and 1% levels, respectively, with standard errors in parentheses.

In contrast, Column (3) shows a positive and significant impact on economic sustainability (0.3005, 1% level), implying that urbanization may support long-term development through infrastructure investment, technological upgrading, and greener urban systems. These findings suggest that radical urbanization generates trade-offs: while it may enhance long-run sustainability, it simultaneously weakens efficiency and stability, particularly during early urbanization stages. This pattern aligns with Henderson et al. [35], who argue that urbanization can stimulate growth without necessarily improving economic quality.

Overall, the results highlight the need to coordinate urban expansion with industrial upgrading and social infrastructure development to mitigate instability and efficiency losses, ensuring that urbanization contributes to balanced and sustainable economic growth.

4.3. Endogeneity Treatment

This study employs standardized measures of built-up areas and NTL imagery, whose calibrated data provide relatively objective measurements that help reduce measurement error and estimation bias. However, because radical urbanization is a dynamic process without universally accepted measurement standards, estimating its impact on economic growth quality remains subject to endogeneity concerns. Specifically, simultaneity (economic growth influencing urbanization) and omitted variables may bias baseline estimates, a common issue in urbanization research. Therefore, instrumental variable techniques are adopted to address potential endogeneity.

FDI flows are closely associated with urbanization dynamics but exhibit lagged responses. Accordingly, the study constructs an instrumental variable using changes in FDI flows relative to changes in urban built-up areas (invflows) and employs its one-period lag (L. invflows) to capture the temporal effect while satisfying exogeneity conditions.

Using this instrument, NTL data are estimated through a two-stage least squares (2SLS) framework to identify the causal effects of radical urbanization. The results reported in Table 7 show that, after controlling for endogeneity, the direction and significance of radical urbanization’s effects on economic growth quality and its three sub-dimensions remain unchanged. All coefficients remain significant at the 1% level, and magnitude differences are minor, confirming the robustness of the baseline conclusions.

Table 7. Endogeneity treatment results.

	(1)	(2)	(3)	(4)	(5)
Variables	The First Stage	The Second Stage	The Second Stage	The Second Stage	The Second Stage
	Rad_ur	Quality	Eficiency	Stability	Sustainability
rad_ur		−0.7798 ***	−1.1950 ***	−0.3861 **	0.4491 ***
		(−8.17)	(−9.98)	(−1.99)	(7.25)
L. invflows	0.6062 ***
	(33.51)
trade	−0.1716 ***	−0.2005 **	−0.6149 ***	1.3580 ***	−0.0492
	(−6.06)	(−2.17)	(−5.30)	(7.21)	(−0.82)
non_ag	−0.0107	0.7220 ***	0.5249 ***	0.1981	0.4728 ***
	(−0.34)	(7.11)	(4.12)	(0.96)	(7.17)
scale	0.1544 ***	0.5480 ***	−0.0939	−0.0596	−0.0156
	(3.81)	(4.14)	(−0.57)	(−0.22)	(−0.18)
ft_vol university	0.0119 ***	0.0354 ***	0.0487 ***	0.0058	0.0015
	(4.81)	(4.37)	(4.79)	(0.35)	(0.29)
	0.0242 **	0.0477	−0.0350	0.0645	0.1075 ***
	(2.33)	(1.43)	(−0.84)	(0.95)	(4.97)
Year FE	yes	yes	yes	yes	yes
City FE	yes	yes	yes	yes	yes
N	4403	4403	4403	4403	4403
Cragg-Donald Wald F statistic Anderson-Rubin Wald test	1122.824 [16.38] 0.000

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively, with standard errors in parentheses.

4.4. Robustness Test

4.4.1. Double Tail Reduction by 1%

The descriptive statistics in Table 3 indicate large disparities in extreme values for growth efficiency and stability, suggesting the presence of outliers. Following standard practice, this study applies 1% winsorization by replacing observations below the first percentile with threshold values and re-estimating the models to improve robustness.

The results in Table 8 show that radical urbanization continues to exert a statistically significant negative effect on overall economic growth quality (−0.0727, 5% level), confirming that the mismatch between spatial expansion and population or industrial development reduces economic quality. For the three sub-dimensions, coefficients remain significant at the 1% level, and although the magnitudes for efficiency and stability slightly decline after winsorization, the direction and significance remain unchanged, demonstrating strong robustness.

Table 8. The test result of a double-tailed reduction by 1%.

Variables	(1) Quality	(2) Eficiency	(3) Stability	(4) Sustainability
rad_ur	−0.0727 **	−0.2882 ***	−0.5026 ***	0.3005 ***
trade	(−2.27)	(−6.48)	(−31.61)	(10.49)
	−0.1428 **	−0.3700 ***	−0.1183 ***	0.0671
non_ag	(−2.11)	(−3.95)	(−3.53)	(1.11)
	0.5098 ***	0.3395 ***	0.0082	0.6151 ***
	(7.32)	(3.52)	(0.24)	(9.90)
scale	0.0281	−0.3299 **	0.1346 ***	0.0875
	(0.30)	(−2.57)	(2.94)	(1.06)
ft_vol	0.0206 ***	0.0300 ***	0.0152 ***	0.0041
university	(3.45)	(3.63)	(5.14)	(0.78)
	0.0247	−0.0657 **	0.0052	0.1047 ***
	(1.08)	(−2.07)	(0.46)	(5.11)
cons_	−0.4381	2.1657 ***	−0.0393	−1.4669 ***
	(−0.82)	(2.91)	(−0.15)	(−3.06)
Year FE	yes	yes	yes	yes
City FE	yes	yes	yes	yes
N	4794	4794	4794	4794
R-squared	0.919	0.862	0.272	0.931

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively, with standard errors in parentheses.

These findings confirm that radical urbanization has mixed effects, improving sustainability while weakening efficiency and stability, consistent with Rana [36] and Anwar et al. [37]. Rapid and uncoordinated expansion often causes inefficient land use, resource misallocation, and economic volatility, whereas effective planning can support sustainable development. Similarly, Dongfeng et al. [38] emphasize flexible and region-specific urbanization strategies combining inclusive regional policy with sustainable planning.

Among control variables, trade negatively affects quality, efficiency, and stability but positively influences sustainability, suggesting that international integration introduces competition while facilitating sustainable practices through technology transfer [39,40]. The non-agricultural sector (non_ag) positively contributes to quality, efficiency, and sustainability, highlighting the benefits of structural transformation, though its effect on stability is insignificant. Firm size (scale) negatively affects efficiency, indicating potential organizational inefficiencies in larger entities. Foreign trade volume (ft_vol) positively affects all outcomes, reflecting gains from technology diffusion and market access. University education (university) improves quality and efficiency but shows weaker effects on stability and sustainability, suggesting a stronger role in productivity than in resilience.

Overall, high R-squared values indicate strong explanatory power, although the relatively lower fit for efficiency suggests that additional factors may influence productivity outcomes.

4.4.2. Replace the Explanatory Variable

This study employs the ratio between the growth rate of built-up area in municipal districts (rad_ur2) and the population growth rate in municipal districts as a proxy variable, substituting for the core explanatory variable, with the measurement formula specified in (Equation (11)). Here, $r_{it}$ represents the degree of radical urbanization, $s_{it}$ denotes the built-up area of municipal districts, $p_{it}$ indicates the population of municipal districts, t stands for the year, $t_0$ refers to the base year (in this study, the base year $t_0$ is 2003 according to our dataset).

$$r_{it}={\displaystyle \frac{\raisebox{1ex}{$s_{it}$}\!\left/ \!\raisebox{-1ex}{$s_{i,t_0}$}\right.}{\raisebox{1ex}{$p_{it}$}\!\left/ \!\raisebox{-1ex}{$p_{i,t_0}$}\right.}}$$

(11)

To further verify robustness, the explanatory variable for radical urbanization is replaced with an alternative measure (rad_ur2), and the baseline regressions are re-estimated. The results in Table 9 remain consistent with previous findings. Column (1) shows that rad_ur2 continues to exert a significantly negative effect on overall economic growth quality (−0.0715, 5% level), indicating that the adverse impact of radical urbanization is robust to alternative measurements.

Table 9. The test result of replacing the explanatory variable.

Variables	(1) Quality	(2) Eficiency	(3) Stability	(4) Sustainability
rad_ur2	−0.0715 **	−0.1141 ***	−0.1116 **	0.0848 ***
trade	(−2.44)	(−3.03)	(−2.04)	(4.37)
	−0.0020	−0.3009 **	−0.0177	−0.1324 **
non_ag	(−0.02)	(−2.42)	(−0.10)	(−2.07)
	0.7878 ***	0.5223 ***	−0.0203	0.6112 ***
	(8.41)	(4.33)	(−0.12)	(9.83)
scale	0.3401 ***	−0.5620 ***	0.0269	0.1958 **
	(2.68)	(−3.44)	(0.11)	(2.32)
ft_vol	0.0193 **	0.0233 **	−0.0003	0.0094 *
university	(2.36)	(2.22)	(−0.02)	(1.74)
	0.0572 *	−0.0498	0.0174	0.1506 ***
	(1.79)	(−1.21)	(0.29)	(7.08)
cons_	−2.2365 ***	3.4700 ***	0.3014	−2.0466 ***
	(−2.99)	(3.60)	(0.22)	(−4.12)
Year FE	yes	yes	yes	yes
City FE	yes	yes	yes	yes
N	4592	4592	4592	4592
R-squared	0.872	0.799	0.074	0.934

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, with standard errors in parentheses.

Across sub-dimensions, growth efficiency remains significantly negative (Column 2), while economic stability also shows a negative effect, significant at conventional levels (Column 3). In contrast, sustainability retains a significantly positive coefficient (Column 4), confirming that radical urbanization generates trade-offs by weakening efficiency and stability while supporting long-term sustainability. Control variables largely maintain their expected signs: non_ag and ft_vol promote development outcomes, whereas larger city scale reduces efficiency but supports quality and sustainability.

Overall, the stable coefficient signs and significance levels confirm the robustness of the main conclusions regarding the multidimensional effects of radical urbanization.

4.4.3. Eliminate Omitted Variables

To address potential omitted variable bias, this study incorporates additional control variables by including previously overlooked factors that may influence the outcomes. Given the significant impact of China’s natural resource endowment on economic performance, we add resource endowment as a control variable to verify whether the signs and statistical significance of the coefficients for overall growth quality and its three sub-indicators remain consistent with the original model. Considering data availability and comparability, we measure resource endowment using per capita resource output (energy). The research identifies natural gas (Gas), petroleum (Oil), and coke (Coke) as China’s most utilized primary energy sources. Using city-level data from the China Energy Statistical Yearbook, we convert the per capita output levels of these three primary energy sources into standard coal equivalent units to calculate per capita resource output, as specified in Formula (12).

$$energy=({\displaystyle \frac{Coke\times 0.0714}{Kilogram}}+{\displaystyle \frac{Oil\times 1.4862}{Kilogram}}+{\displaystyle \frac{Gas\times 1.33}{Cubicmeter}})/population$$

(12)

Table 10 presents regression results after controlling for potential omitted variables. The coefficient of radical urbanization (rad_ur) remains significantly negative for overall economic growth quality (−0.1291, 1% level), confirming that the baseline findings are robust and not driven by omitted factors.

Table 10. The test results for eliminating omitted variables.

Variables	(1) Quality	(2) Eficiency	(3) Stability	(4) Sustainability
rad_ur	−0.1291 ***	−0.3152 ***	−0.6339 ***	0.3608 ***
trade	(−2.61)	(−5.16)	(−6.09)	(10.75)
	0.0214	−0.4312 ***	1.2256 ***	0.0460
non_ag	(0.24)	(−3.90)	(6.51)	(0.76)
	0.2596 *	0.1189	0.0871	0.3221 ***
	(1.91)	(0.71)	(0.30)	(3.48)
scale	−0.0609	−0.7905 ***	0.1623	−0.2874 ***
	(−0.39)	(−4.10)	(0.49)	(−2.72)
ft_vol	0.0246 ***	0.0347 ***	0.0040	0.0073
university	(2.97)	(3.39)	(0.23)	(1.29)
	0.0153	−0.0486	0.0842	0.0891 ***
	(0.41)	(−1.07)	(1.09)	(3.56)
energy	0.3641	5.0261 ***	−0.0663	0.8651
	(0.40)	(4.46)	(−0.03)	(1.40)
cons_	−0.1291 ***	−0.3152 ***	−0.6339 ***	0.3608 ***
Year FE	(−2.61)	(−5.16)	(−6.09)	(10.75)
	yes	yes	yes	yes
City FE	yes	yes	yes	yes
N	3887	3887	3887	3887
R-squared	0.891	0.853	0.106	0.932

Note: *, *** indicate significance at the 10% and 1% levels, respectively, with standard errors in parentheses.

Across dimensions, radical urbanization significantly reduces growth efficiency (−0.3152) and stability (−0.6339), while exerting a positive effect on sustainability (0.3608), indicating persistent trade-offs between short-term economic performance and long-term development outcomes. Control variables show broadly consistent patterns: non_ag improves quality and sustainability, scale weakens efficiency, and ft_vol supports growth quality and efficiency, reflecting benefits from structural transformation and external integration. University presence mainly promotes sustainability, while energy consumption shows limited influence across outcomes.

Overall, the stable coefficient signs and significance levels further confirm the robustness of the conclusion that radical urbanization negatively affects economic growth quality despite contributing to sustainability.

4.4.4. Exclude the Influence of Term Change

In China, local government leadership cycles linked to the National Party Congress may influence policy priorities and urban development strategies. To control for this institutional factor, this study constructs a key conference year dummy (deem_year), assigning 0 to Party Congress years (2007, 2012, and 2017) and 1 to all other years. Excluding these transition years reduces the sample size from 4794 to 4243 observations. The regression results in Table 11 show that radical urbanization (rad_ur) remains significantly associated with economic outcomes across all specifications. The coefficient for overall growth quality becomes more negative (−0.1852), indicating that the adverse effect strengthens after removing leadership transition periods. This suggests that policy adjustments during term-change years may temporarily moderate the negative impacts of radical urbanization.

Table 11. The test results exclude the influence of term change.

Variables	(1) Quality	(2) Eficiency	(3) Stability	(4) Sustainability
rad_ur	−0.1852 ***	−0.3749 ***	−0.6026 ***	0.2718 ***
trade	(−4.08)	(−6.49)	(−6.11)	(9.09)
	0.1330	−0.3737 ***	1.2598 ***	0.0551
non_ag	(1.44)	(−3.17)	(6.25)	(0.90)
	0.7924 ***	0.5310 ***	0.1532	0.6276 ***
	(8.22)	(4.33)	(0.73)	(9.88)
scale	0.3116 **	−0.4039 **	−0.0701	0.1081
	(2.37)	(−2.41)	(−0.24)	(1.25)
ft_vol	0.0188 **	0.0333 ***	0.0093	0.0007
university	(2.17)	(3.02)	(0.49)	(0.13)
	0.0355	−0.0604	0.0508	0.1245 ***
	(1.10)	(−1.47)	(0.72)	(5.85)
cons_	−2.0460 ***	2.6144 ***	1.1798	−1.5868 ***
Year FE	(−2.68)	(2.69)	(0.71)	(−3.15)
	yes	yes	yes	yes
City FE	yes	yes	yes	yes
N	4243	4243	4243	4243
R-squared	0.870	0.816	0.098	0.930

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively, with standard errors in parentheses.

Across sub-dimensions, radical urbanization continues to reduce growth efficiency (−0.3749) and stability (−0.6026), while positively affecting sustainability (0.2718). The slightly larger coefficients imply that the long-term structural effects of urban expansion persist beyond short-term political cycles. Control variables remain broadly consistent with earlier results: the non-agricultural sector (non_ag) significantly improves quality and sustainability, city scale weakens efficiency, and foreign trade volume (ft_vol) supports efficiency and overall performance. University presence primarily promotes sustainability through human capital accumulation. Overall, excluding leadership transition years does not alter the main conclusions, confirming that the negative relationship between radical urbanization and economic growth quality is structurally driven rather than politically cyclical.

4.4.5. Exclude the Influence of Other Policies

To control for potential macroeconomic policy interference, this study excludes observations from 2009 to 2010, when China implemented the four trillion yuan stimulus package aimed at economic stabilization following the global financial crisis. The regressions are re-estimated using the reduced sample, and the results are reported in Table 12. Radical urbanization (rad_ur) remains statistically significant at the 1% level across all specifications, with a negative effect on overall growth quality (−0.1417), growth efficiency (−0.2923), and stability (−0.5964), while maintaining a positive effect on sustainability (0.3008). The similarity in coefficient signs and magnitudes relative to baseline estimates indicates that the main findings are not driven by short-term expansionary policies.

Table 12. The test results exclude the influence of other policies.

Variables	(1) Quality	(2) Eficiency	(3) Stability	(4) Sustainability
rad_ur	−0.1417 ***	−0.2923 ***	−0.5964 ***	0.3008 ***
trade	(0.0467)	(0.0582)	(0.0992)	(0.0307)
	0.1174	−0.3235 ***	1.2258 ***	0.0683
non_ag	(0.0949)	(0.1183)	(0.2016)	(0.0623)
	0.8145 ***	0.5769 ***	0.2111	0.5982 ***
	(0.0995)	(0.1241)	(0.2114)	(0.0654)
scale	0.3878 ***	−0.4510 ***	−0.0384	0.1381
	(0.1323)	(0.1649)	(0.2810)	(0.0869)
ft_vol	0.0195 **	0.0297 ***	0.0105	0.0032
university	(0.0084)	(0.0105)	(0.0179)	(0.0055)
	0.0241	−0.0524	0.0293	0.1021 ***
	(0.0330)	(0.0411)	(0.0700)	(0.0217)
cons_	−2.4492 ***	2.8732 ***	0.9983	−1.7131 ***
Year FE	(0.7670)	(0.9561)	(1.6292)	(0.5038)
	yes	yes	yes	yes
City FE	yes	yes	yes	yes
N	4227	4227	4227	4227
R-squared	0.8725	0.8211	0.1000	0.9314

Note: **, and *** indicate significance at the 5%, and 1% levels, respectively, with standard errors in parentheses.

Control variables exhibit consistent patterns: the non-agricultural sector (non_ag) continues to enhance quality, efficiency, and sustainability, reflecting gains from structural transformation; city scale reduces efficiency but supports overall development; and foreign trade volume (ft_vol) positively contributes to performance outcomes. University presence mainly strengthens sustainability through human capital accumulation. Overall, excluding stimulus-policy years does not alter the empirical conclusions, confirming that the negative relationship between radical urbanization and economic growth quality reflects structural dynamics rather than temporary macroeconomic shocks.

4.5. Mechanism Analysis

Based on the literature review and theoretical analysis, this study posits that radical urbanization affects economic growth quality through two primary transmission channels: fiscal output and land use efficiency. The fiscal output channel reflects whether fiscal expansion associated with rapid urban development is effectively transformed into productive economic outcomes, while the land use efficiency channel captures how urban spatial expansion influences resource allocation efficiency and production performance. To examine these mechanisms, we employ fiscal output (fin_output) and land use efficiency (land_perform) as mediating variables while controlling for city and year fixed effects, constructing mechanism test models. Methodologically, although fiscal output and land use efficiency are theoretically transmission channels, the empirical strategy adopted here follows an interaction-based mechanism test that is conceptually closer to a moderation-effect framework rather than a traditional stepwise mediation analysis. Classical mediation approaches [41] decompose total effects through sequential regressions, whereas the objective of this study is to examine whether the impact of radical urbanization on economic growth quality varies systematically with these channel variables.

To avoid issues of over-identification and misuse, the analysis directly investigates the mechanisms of radical urbanization using a one-step moderation effect approach, building upon established causal relationships from the existing literature. Specifically, interaction terms between radical urbanization and the proposed channel variables are introduced into the regression model. A statistically significant interaction coefficient indicates that the marginal effect of radical urbanization depends on the level of the corresponding channel variable, thereby providing direct empirical evidence for the transmission path of radical urbanization → channel variable → growth quality. This interaction-based identification strategy allows mechanism verification within a unified specification and avoids potential biases associated with multi-step mediation procedures [42]. The test models are specified in Equations (13) and (14).

$$\begin{array}{l}qualit{y}_{it}=c_0+{\alpha }_{1}rad\text{\_}u{r}_{it}+{\alpha }_{2}financ{e}_{it}+{\alpha }_{3}rad\text{\_}u{r}_{it}\times fin\text{\_}outpu{t}_{it}\\ +{\beta }_{1}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}\end{array}$$

(13)

$$\begin{array}{l}qualit{y}_{it}=c_0+{\alpha }_{1}rad\text{\_}u{r}_{it}+{\alpha }_2\mathrm{land}\text{\_}perfor{m}_{it}+\\ {\alpha }_{3}rad\text{\_}u{r}_{it}\times land\text{\_}perfor{m}_{it}+{\beta }_{1}contro{l}_{it}+v_i+u_t+{\epsilon }_{it}\end{array}$$

(14)

Here, quality represents the composite economic growth quality, while fin_output and land_perform serve as mediating variables denoting fiscal output and land use efficiency, respectively. The three categorical indicators of economic growth quality-efficiency, stability, and sustainability—can substitute for the dependent variable quality in the aforementioned models, representing growth efficiency, stability, and sustainability correspondingly. Consistent with the baseline regression, these specifications are not separately presented here. The control variables maintain identical definitions to those previously established, and ε__it denotes the random disturbance term. This interaction-based approach is widely applied in empirical economics to identify influence channels through conditional effects. For example, Acemoglu et al. [43], in a study in the American Economic Review, employ interaction structures to examine how institutional factors condition the relationship between income and democracy, thereby identifying causal mechanisms within a single empirical framework. Similar applications demonstrate that mechanism identification can be achieved through interaction effects without requiring step-by-step decomposition [44,45].

4.5.1. Mechanism Analysis of Fiscal Output

This study contends that radical urbanization adversely affects economic growth quality by reducing fiscal output. Given the dominant role of government expenditure in China’s urbanization investments and fiscal channels as the primary government funding source, the input indicator employs government fiscal expenditure as the proxy variable. As gross regional product most directly reflects economic activity, following the established literature practices, the output indicator uses GDP as the proxy. Fiscal output is measured by the ratio between input and output indicators, specifically, the ratio of government fiscal expenditure to GDP.

The regression results validate the logical pathway of “radical urbanization → declining fiscal output → suppressed economic growth quality improvement.” As shown in Table 13, the coefficient for the rad_ur × fin_output interaction term is −0.7893, statistically significant at the 5% level with a negative sign. The results support the hypothesis that radical urbanization places substantial pressure on public finances, consistent with urban economics and public finance studies. Rapid urbanization increases demand for infrastructure, housing, transportation, and public services without proportionate fiscal capacity, leading to resource inefficiencies [46]. Specifically, radical urbanization primarily reduces growth efficiency, as reflected by the interaction coefficient of −0.8646 (5% level). Efficiency here refers to the economy’s ability to allocate resources without waste. Unplanned urban expansion causes congestion, resource depletion, and overburdened infrastructure, reducing productivity, in line with prior studies [47,48,49].

Table 13. The test results of the mechanism of action.

Variables	(1) Quality	(2) Quality	(3) Quality	(4) Eficiency	(5) Eficiency	(6) Eficiency
rad_ur	−0.1658 ***	−0.2248 ***	−0.2276 ***	−0.3464 ***	−0.3848 ***	−0.3801 ***
	(−3.87)	(−5.04)	(−5.22)	(−6.34)	(−6.71)	(−6.82)
finance		−3.0020 ***			−1.9298 ***
		(−6.07)			(−3.04)
rad_ur × finance		−0.7893 **			−0.8646 **
		(−2.42)			(−2.06)
landp			−0.0831 ***			−0.0939 ***
			(−6.97)			(−6.16)
rad_ur × landp			−0.0061 *			−0.0096 **
			(−1.80)			(−2.24)
trade	0.1105	−0.0645	−0.0196	−0.3379 ***	−0.0133	0.0198
non_ag	(1.22)	(−0.33)	(−0.10)	(−2.94)	(−0.05)	(0.08)
	0.7555 ***	0.7579 ***	0.7642 ***	0.4938 ***	0.5229 ***	0.5414 ***
	(8.12)	(8.87)	(8.96)	(4.17)	(4.76)	(4.97)
scale	0.3329 ***	0.4555 ***	0.3746 ***	−0.4699 ***	0.0103	−0.0799
	(2.69)	(3.41)	(2.89)	(−2.98)	(0.06)	(−0.48)
ft_vol	0.0199 **	−0.0045	−0.0112	0.0297 ***	−0.0175	−0.0232
university	(2.50)	(−0.31)	(−0.80)	(2.93)	(−0.96)	(−1.30)
	0.0218	0.0884 ***	0.0746 **	−0.0659 *	−0.0337	−0.0365
	(0.71)	(2.96)	(2.56)	(−1.69)	(−0.88)	(−0.98)
cons_	−2.1300 ***	−2.7460 ***	−1.5193 **	3.0439 ***	0.4649	1.8864 *
Year FE	(−2.97)	(−3.55)	(−1.98)	(3.33)	(0.47)	(1.92)
	yes	yes	yes	yes	yes	yes
City FE	yes	yes	yes	yes	yes	yes
N	4794	3968	3980	4794	3968	3980
R-squared	0.876	0.864	0.865	0.821	0.816	0.821
Variables	(1) Stability	(2) Stability	(3) Stability	(4) Sustain	(5) Sustain	(6) Sustain
rad_ur	−0.5995 ***	−0.4660 ***	−0.5619 ***	0.3005 ***	0.1509 ***	0.1363 ***
	(−6.68)	(−4.66)	(−5.75)	(10.49)	(5.51)	(5.08)
finance		−0.5996			−0.4586
		(−0.54)			(−1.51)
rad_ur × finance		1.2892 * (1.76)			0.4606** (2.30)
landp			0.0335 (1.26)			−0.0201 *** (−2.74)
rad_ur × landp			−0.0176 ** (−2.33)			0.0035 * (1.71)
trade	1.2150 ***	−0.1642	−0.2899	0.0671	0.1742	0.1621
non_ag	(6.42)	(−0.38)	(−0.67)	(1.11)	(1.47)	(1.37)
	0.1530	−0.0731	0.0121	0.6152 ***	0.5651 ***	0.5715 ***
	(0.78)	(−0.38)	(0.06)	(9.90)	(10.79)	(10.90)
scale	−0.0170	−0.0260	0.1124	0.0875	0.2105 **	0.2416 ***
	(−0.07)	(−0.09)	(0.39)	(1.06)	(2.57)	(3.03)
ft_vol	0.0096	0.0183	0.0140	0.0041	0.0090	0.0081
university	(0.57)	(0.58)	(0.45)	(0.78)	(1.04)	(0.94)
	0.0316	−0.0010	−0.0042	0.1047 ***	0.1615 ***	0.1466 ***
	(0.49)	(−0.01)	(−0.06)	(5.11)	(8.81)	(8.17)
cons_	0.9035	0.8206	−0.1726	−1.4673 ***	−2.2866 ***	−2.2324 ***
Year FE	(0.60)	(0.47)	(−0.10)	(−3.06)	(−4.82)	(−4.73)
	yes	yes	yes	yes	yes	yes
City FE	yes	yes	yes	yes	yes	yes
N	4794	3968	3980	4794	3968	3980
R-squared	0.091	0.087	0.087	0.931	0.942	0.942

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, with standard errors in parentheses.

Second, fiscal expenditure functionality positively influences growth stability, with the interaction coefficient significant at the 10% level, consistent with Keynesian theory that government spending can smooth economic fluctuations. Third, radical urbanization improves the investment conversion efficiency of fiscal funds, evidenced by a significantly positive coefficient (5% level). Urban agglomeration enhances infrastructure, economic networks, and access to labor and capital, enabling more efficient investment, consistent with Glaeser [49] and Fang and Yu [50], who emphasize cities as engines of growth through specialization and economies of scale.

4.5.2. Mechanism Analysis of Land Efficiency

This study argues that radical urbanization impedes economic growth quality by suppressing land use efficiency, examining this mediation mechanism through changes in land concession fees (landp). Due to missing values in the land concession fee data, partial data imputation was performed for the absent entries.

The regression results in Table 13 reveal a relatively weak alignment between radical urbanization and land use efficiency, demonstrating the mechanism of “radical urbanization → declining land efficiency → suppressed economic growth quality improvement.” The coefficient for the interaction term rad_ur × landp is −0.0061, significant at the 10% level. Compared with the moderating effect of fiscal output, the land efficiency channel exhibits a relatively weaker influence. The results indicate that rapid and unplanned urban expansion reduces land-use efficiency, though to a smaller extent than other urban dynamics such as fiscal pressure. This reflects urban sprawl, where disorganized growth leads to inefficient land allocation, underutilization, and low-density development, consistent with Bloom et al. [51], who found no evidence that urbanization level directly affects economic growth rates.

First, radical urbanization impairs economic growth efficiency through land-use inefficiency, as shown by an interaction coefficient of −0.0096. Second, inefficient land use negatively affects economic stability, significantly at the 5% level, by raising transportation costs, reducing connectivity, and creating inefficiencies in labor and capital markets. Third, radical urbanization generates adverse selection effects, forcing land-intensive enterprises to exit or transform operations, consistent with the urban growth and economic stability literature. Inefficient land allocation undermines productive capacity, pushing firms reliant on optimal land use, such as manufacturers or agricultural businesses, to adapt or leave the market.

5. Further Analysis

This study posits that radical urbanization fails to achieve expected outcomes in practice, resulting in disproportionate fiscal input–output ratios that may crowd out other types of public expenditures, thereby generating negative impacts on economic growth quality. This crowding-out effect can be understood as a specific manifestation of the fiscal output transmission channel identified in the mechanism analysis. When fiscal resources are increasingly directed toward short-term urban expansion and infrastructure investment, local governments may reduce expenditures that support long-term productivity and human capital accumulation, thereby weakening fiscal output efficiency and ultimately lowering economic growth quality. Consequently, when assessing these negative effects, we must examine whether radical urbanization displaces alternative public spending categories. Using regional government investments in science and education as a proxy for non-productive public expenditure (NPS) and building upon the research framework established by Sui et al. [52], we construct an analytical model to investigate non-productive public spending, enabling further analysis of radical urbanization’s impact on economic growth quality. Thus, further analysis extends rather than introduces a new mechanism by examining how fiscal allocation distortion operates within the fiscal output channel.

Firstly, this study examines whether the hypothesized relationship among radical urbanization, non-productive public expenditure, and economic growth quality holds empirically. Considering data availability and accuracy, the non-productive public expenditure indicator is measured by the sum of science and education shares within fiscal expenditures, with logarithmic transformation applied for standardization, thereby quantifying local governments’ investment intensity in non-productive public spending. The test model is specified as follows:

$$np{s}_{it}=c_0+{\gamma }_{1}rad\text{\_}u{r}_{it}+\mu contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(15)

$$qualit{y}_{it}=c_0+\rho qualit{y}_{i,t-1}+{\gamma }_{2}np{s}_{it}+\mu contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(16)

${\gamma }_1$ represents the effect of radical urbanization on non-productive public expenditure, while ${\gamma }_2$ denotes the effect of non-productive public expenditure on growth quality. If both ${\gamma }_1$ and ${\gamma }_2$ demonstrate statistical significance with non-zero values, this conclusively establishes the validity of the exclusionary analysis mechanism; otherwise, the mechanism does not hold.

To further investigate the moderating effects of radical urbanization on different dimensions of economic growth quality, this study substitutes the dependent variable quality with the three indicators-efficiency, stability, and sustainability-for empirical testing, as specified in Models (17), (18), and (19):

$$efficienc{y}_{it}=c_0+\rho qualit{y}_{i,t-1}+{\gamma }_{2}np{s}_{it}+\mu contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(17)

$$stabilit{y}_{it}=c_0+\rho qualit{y}_{i,t-1}+{\gamma }_{2}np{s}_{it}+\mu contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(18)

$$sustainabilit{y}_{it}=c_0+\rho qualit{y}_{i,t-1}+{\gamma }_{2}np{s}_{it}+\mu contro{l}_{it}+v_i+u_t+{\epsilon }_{it}$$

(19)

Secondly, the actual effects of the aforementioned analysis require further verification through Model (20):

$$\begin{array}{l}qualit{y}_{it}=c_0+\rho qualit{y}_{i,t-1}+{\gamma }_{2}np{s}_{it}+\theta rad\text{\_}u{r}_{it}\times np{s}_{it}\\ +\mu contro{l}_{it}+v_i+u_t+{\epsilon }_{it}\end{array}$$

(20)

Here, θ represents the moderating effect of urbanization on growth quality. If θ > 0, it indicates that the interaction term between urbanization and non-productive inputs has a positive effect; conversely, the interaction term exerts a negative influence on growth quality.

The regression results are presented in Table 14. Column (1) displays the results from Model (15), showing that radical urbanization crowds out non-productive inputs, with a coefficient γ_1 of −0.06443. This indicates a statistically significant negative effect of radical urbanization on non-productive public expenditure at the 1% level, suggesting that a one percentage point increase in radical urbanization leads to a 0.064 percentage point reduction in non-productive public spending. Column (2) presents the results from Model (16), where the coefficient ${\gamma }_2$ of 0.0954 is significantly positive at the 5% level, demonstrating that non-productive public expenditure enhances economic growth quality through positive output effects in areas like research and education. Disaggregated analysis reveals that non-productive public spending significantly improves growth stability at the 5% level (coefficient: 0.1951) and sustainability at the 1% level (coefficient: 0.1420). This occurs because non-productive public expenditures represent long-term investments that require time to translate into economic growth rather than directly affecting current growth efficiency. Such investments provide stable long-term growth momentum, ultimately exerting a positive moderating effect on overall economic growth quality.

Table 14. Exclusive analysis of non-productive public expenditure.

Variables	(1) nps	(2) Quality	(3) Eficiency	(4) Stability	(5) Sustain	(6) Quality
rad_ur	−0.0644 *** (−3.65)
nps		0.0954 **	−0.1573 ***	0.1951 **	0.1420 ***
		(2.36)	(−3.04)	(2.17)	(5.79)
rad_ur × nps						−0.1814 *** (0.0452)
trade	0.0593	−0.0247	0.0257	−0.2942	0.1574	0.1104
non_ag	(0.75)	(−0.13)	(0.10)	(−0.68)	(1.34)	(0.0925)
	0.2904 ***	0.6915 ***	0.5260 ***	−0.1037	0.5378 ***	0.8139 ***
	(8.31)	(8.01)	(4.76)	(−0.54)	(10.26)	(0.0964)
scale	0.9177 ***	0.2640 *	0.0231	−0.1670	0. 1368 *	0.3355 ***
ft_vol	(17.09)	(1.93)	(0.13)	(−0.55)	(1.65)	(0.1278)
	0.0341 ***	−0.0235 *	−0.0307 *	−0.0128	0.0065	0.0196 **
university	(5.88)	(−1.66)	(−1.70)	(−0.41)	(0.76)	(0.0082)
	0.0282 **	0.0832 ***	−0.0324	−0.0065	0.1581 ***	0.0118
	(2.29)	(2.76)	(−0.84)	(−0.10)	(8.63)	(0.0324)
cons_	6.5908 ***	−3. 180 7***	1.6598	−1.2605	−3.3390 ***	−2.0812 ***
Year FE	(21.04)	(−3.92)	(1.60)	(−0.70)	(−6.78)	(0.7414)
	yes	yes	yes	yes	yes	yes
City FE	yes	yes	yes	yes	yes	yes
N	3971	3971	3971	3971	3971	3971
R-squared	0.979	0.849	0.799	0.002	0.937	0.874

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, with standard errors in parentheses.

The comprehensive regression results across all five columns demonstrate statistically significant non-zero coefficients, confirming the validity of the mechanism through which radical urbanization affects growth quality via non-productive public expenditure. Given the significantly positive impact of non-productive public spending on economic growth quality, we conclude that radical urbanization crowds out such expenditures, thereby generating negative moderating effects. Radical urbanization diverts fiscal funds away from direct production processes while significantly displacing non-productive public investments in science, education, culture, and healthcare, while simultaneously stimulating infrastructure investment and administrative expenditures. Influenced by government preferences in public goods procurement, infrastructure spending, and non-productive public expenditures exhibit an inverse correlation under fixed annual fiscal budgets.

Column (6) presents the regression results of rad_ur × nps’s impact on overall growth quality, revealing that the interaction term coefficient θ is significantly negative at the 5% level. A negative θ indicates that radical urbanization suppresses the effectiveness of non-productive public expenditure. Under given fiscal budget constraints, radical urbanization squeezes non-productive public spending represented by science and education investments, thereby indirectly impairing economic growth quality. Furthermore, the regression coefficient of the interaction term rad_ur × nps is −0.1814, exceeding the coefficient from the baseline regression in Column (3) of Table 5, demonstrating an amplified negative moderating effect after interaction. Consequently, this additional verification of the mechanism’s actual effects confirms that radical urban expansion indirectly harms growth quality by crowding out non-productive public expenditures.

6. Conclusions

Radical urbanization emerges as a byproduct of China’s rapid urbanization process, characterized by imbalanced expansion of built-up areas and population disparities, leading to regional economic sluggishness and insufficient NTL intensity. With China’s ongoing urban development and spatial expansion, the economic growth effects of radical urbanization demand urgent academic attention. This study integrates city-level economic growth quality data with NTL imagery across China from 2003 to 2019, measuring radical urbanization through the logarithm of urban light intensity per unit area after systematic collection, calibration, and analysis of regional NTL data. We construct a comprehensive evaluation system for China’s urban economic growth quality and employ fixed-effects models to assess radical urbanization’s impact. To address potential endogeneity concerns, the study utilizes the one-period lagged direction of FDI flows as an instrumental variable, building upon the logical connection between international investment mobility and urban development patterns.

The baseline regression yields estimation results for overall economic growth quality and its three sub-indicators, demonstrating that radical urbanization exerts statistically significant inhibitory effects on economic growth quality. Higher degrees of radical urbanization consistently undermine regional high-quality economic development. Disaggregated analysis reveals that radical urbanization primarily suppresses growth quality through impairing economic efficiency and stability. To further validate these baseline results, robustness checks were conducted through bilateral 1% winsorization, explanatory variable substitution, characteristic fact verification, omitted variable exclusion, and policy impact mitigation, all confirming the findings’ reliability.

The underlying mechanism originates from radical urbanization’s amplifying effect on fiscal resources and land allocation imbalances. Mechanism analysis confirms that radical urbanization undermines both fiscal output efficiency and optimal land use productivity. Specifically, excessive urban expansion increases fiscal pressure on local governments, reducing the productivity of fiscal expenditures and weakening fiscal output efficiency, while disorderly spatial expansion lowers land allocation efficiency and diminishes the economic returns generated from urban land resources. These two channels jointly constitute the dual-channel mechanism through which radical urbanization affects economic growth quality. Extended examination reveals non-productive public expenditure’s positive moderating effect on growth quality and its three sub-dimensions, while radical urbanization crowds out such expenditures, thereby indirectly impairing economic growth quality. Importantly, the crowding-out effect provides a direct linkage between further analysis and the dual-channel mechanism: reductions in science and education spending weaken human capital accumulation and innovation capacity, thereby lowering fiscal output effectiveness, while insufficient investment in knowledge-intensive sectors reduces the efficiency with which urban land is utilized for high-value production activities. Consequently, the displacement of non-productive public expenditure reinforces both the fiscal output and land use efficiency channels, forming a unified transmission framework.

Based on these findings, several policy implications emerge. First, urban development policies should shift from expansion-oriented growth toward quality-oriented urbanization by aligning land expansion with population agglomeration and industrial upgrading. Second, fiscal governance reforms should prioritize expenditure efficiency rather than scale expansion, ensuring that fiscal resources are allocated toward productivity-enhancing sectors. Third, local governments should safeguard investments in science and education, as these expenditures mitigate the negative effects of radical urbanization by strengthening innovation capacity and improving long-term growth quality. Fourth, urban land management policies should promote intensive land use and technological upgrading to enhance land productivity and prevent inefficient spatial expansion.

Despite these contributions, this study has several limitations. The measurement of radical urbanization relies primarily on NTL data, which, although widely used, may not fully capture qualitative dimensions of urban governance or institutional differences across cities. In addition, the empirical framework focuses on city-level aggregate relationships and does not explicitly model micro-level behavioral responses of firms or households. Furthermore, because the analysis is based on China’s institutional environment, fiscal system, and urban development model, the policy recommendations derived from this study may not be directly transferable to urbanized regions in other countries. Their applicability outside China should therefore be interpreted with caution and may require contextual adaptation to different institutional and economic settings. Future research may address these limitations by incorporating micro-level datasets, spatial econometric approaches, or alternative urbanization indicators combining demographic, institutional, and environmental dimensions. Moreover, future studies could explore dynamic nonlinear mechanisms and regional heterogeneity in greater depth to better understand long-term adjustment processes associated with urban transformation.

Overall, this study provides systematic evidence that radical urbanization does not merely influence economic outcomes directly but operates through interconnected fiscal and spatial allocation mechanisms. By integrating mechanism analysis and further analysis within a unified dual-channel framework, the findings contribute to a more comprehensive understanding of how urbanization patterns shape high-quality economic development and offer policy guidance for achieving sustainable urban growth in rapidly developing economies.