Spatio-Temporal Evolution Pattern and Driving Forces of Carbon Lock-In in the Yangtze River Delta Region

Abstract

Addressing carbon lock-in is essential for facilitating economic transformation and sustainable, low-carbon growth in the Yangtze River Delta (YRD) region. This study establishes a multidimensional evaluation framework to assess carbon lock-in levels and explores its spatio-temporal evolution as well as key drivers within the YRD urban agglomeration. Findings indicate a general decline in carbon lock-in across the region, with diminishing disparities among cities. While industrial lock-in, technological lock-in, and institutional lock-in have shown a weakening trend, social behavioral lock-in has intensified. Initially, higher levels of carbon lock-in were concentrated in less developed cities, though this concentration has steadily decreased, whereas more developed cities consistently exhibited lower lock-in levels. The carbon intensity of fixed assets and carbon emission intensity have emerged as the primary barrier factors contributing to carbon lock-in. Additionally, socio-economic factors and digital technology innovations are the main influences on carbon lock-in. These insights provide guidance for policy efforts to mitigate carbon lock-in and support for advancing green integrated development strategies in the YRD region.

1. Introduction

Swift economic expansion and industrialization, largely fueled by fossil fuels like natural gas, coal, and oil, have caused carbon emissions to soar and worsened global warming. This high-carbon development model has not only escalated greenhouse gas emissions but has also triggered severe environmental challenges, jeopardizing global ecosystem stability and undermining sustainable development pathways for humanity. In response to these escalating environmental concerns, 195 countries signed the Paris Agreement, committing themselves to cap the global temperature rise to below 1.5 °C above pre-industrial levels. This collective effort aims to protect the planet and ensure a sustainable future for all. Likewise, the Chinese government has also established ambitious “dual carbon” targets of peaking carbon emissions by 2030 and achieving carbon neutrality by 2060. These targets were emphasized in the 20th National Congress of the Communist Party of China, signaling the nation’s strong commitment to sustainable development and climate action. Despite these ambitious carbon goals, the International Energy Agency (IEA) reported that China’s carbon emissions reached 10,613.171 metric tons in 2022, representing a staggering 223% increase since 2000. This figure accounts for 31.1% of global emissions, making China the world’s largest emitter. This situation is deeply rooted in China’s economic growth model, which remains heavily reliant on a carbon-based energy system, with fossil fuels dominating its energy consumption, a structure unlikely to shift in the near future [1]. However, the path to addressing climate change hinges on a crucial transition from a fossil fuel-dependent economy to a decarbonized, sustainable one [2]. Both empirical evidence and theoretical arguments suggest that carbon lock-in presents a significant obstruction to this transition. It constrains the development of green technologies, low-carbon industries, and sustainable institutions and societies, leading to the continued accumulation of carbon emissions. Therefore, in the context of low-carbon sustainable development, a comprehensive evaluation of carbon lock-in levels, an exploration of its spatio-temporal evolution, and the identification of influencing factors hold significant practical importance for promoting the transition of the economy toward a low-carbon economy.

The concept of carbon lock-in, first introduced by Unruh (2000) [3], characterizes the entrenchment of industrial economies within conventional, carbon-intensive energy systems, constrained by the technological and structural limits of carbon-based regimes. This phenomenon arises from the complex interactions between industry, technology, institutions, and social behavior, which collectively reinforce carbon-intensive systems through increasing returns to scale, sunk costs, and behavioral inertia, leading to the formation of “techno-institutional complexes” characterized by path dependence and structural stability, which impede a green and low-carbon transition. As carbon lock-in has the potential to globalize, it poses a significant constraint on climate change mitigation efforts [4]. Since the introduction of the concept of the technology–institution complex, academics have further enriched the connotations and formation mechanisms of carbon lock-in in various ways. In terms of connotation, scholars have further enriched the connotation of carbon lock-in from different perspectives, such as technological, institutional, and normative behaviors [5]; the level of resource endowment and industrial structure [6]; and the interplay of technological, economic, and institutional factors [7]. In terms of the formation mechanism, it mainly includes the co-evolution of government behavior and market mechanisms [8,9], the mutual promotion of sunk cost and path dependence [5], and deep-rooted social behaviors and norms [10]. These studies collectively enrich our understanding of both the conceptual underpinnings and the mechanisms that sustain carbon lock-in, offering valuable insights into the complexities of transitioning to a low-carbon economy.

Meanwhile, a large number of studies have been conducted to estimate carbon lock-in levels. Some studies have used single indicators, such as carbon overloading rate [9,11], carbon-intensive infrastructure [12], carbon-intensive assets [13], or the ratio of the number of coal-fired power plants to their capacity [14,15]. Other studies have constructed comprehensive evaluation indicators based on the sources of carbon lock-in [16,17,18]. Although established studies have evaluated carbon lock at multiple levels, they have not revealed the dynamic evolution of carbon lock under the integrated city cluster development model, and the richness of the connotation of carbon lock requires a more comprehensive evaluation index system. Moreover, the factors influencing carbon lock-in levels are multifarious. On the one hand, economic growth [19], industrial structure [20], and energy intensity [21] all play a part in it. On the other hand, green technologies [22] and well-thought-out government policies [23,24] are the keys to decoupling economic growth from carbon emissions. Green finance [25] and infrastructure optimization [26] are also critical for breaking the carbon lock-in. It can be seen that most of the studies reveal the influence mechanism of carbon locking level from the perspective of economics, but fewer of them combine economics and geography to discuss the influence factors of carbon lock-in.

Internationally, research on carbon lock-in has predominantly concentrated on mature industrialized countries and regions, such as the United Kingdom, Sweden, and Japan, where government policy shifts and bottom-up technological innovations have played significant roles in alleviating institutional and industrial lock-in [27,28,29]. However, in megacities such as Los Angeles and Moscow, the transition to a low-carbon economy is hindered by the continued emphasis on coal and oil studies in their universities, which shape people’s beliefs and behavioral inertia, thereby maintaining high-carbon lifestyles [30]. Notably, these cases often overlook the complex interplay between regional integration policies and the multidimensional nature of lock-in in rapidly urbanizing, administratively complex regions.

Therefore, the focus of this study is the YRD region. The YRD region is one of the most economically vibrant areas in China, with a high level of development and strong innovation capabilities [31], but also one of the regions with high energy consumption and carbon emissions [32]. In 2024, the YRD region contributed 24.5% of the GDP with only about 3.7% of the national land area. But at the same time, it produced nearly 20% of the carbon emissions, and the total carbon emissions in the YRD region in 2030 are projected to reach 1580.70 million tons [33]. To promote regional integration, China issued the “Outline of the Plan for the Integrated Development of the Yangtze River Delta Region” in 2019, elevating this integration to a national strategy. Subsequently, the “Yangtze River Delta Eco-Green Integrated Development Demonstration Zone Overall Plan” was launched, laying the groundwork for advancing eco-friendly growth in the region. Despite these efforts, the YRD’s industrial and energy structures remain heavily reliant on the secondary sector and fossil fuels, creating a high-carbon economic model that is difficult to shift. The region faces a structural dilemma: balancing the growing energy demands of economic expansion with the imperative to reduce carbon emissions. Under the promotion of dual carbon targets and the requirements of green integrated development, this research focuses on the YRD urban agglomeration. It examines the spatio-temporal evolution of carbon lock-in, identifies its influencing factors, and explores practical pathways for carbon unlocking. The findings provide critical insights for promoting the ecological and green development of the YRD. Meanwhile, this research focuses on the following key research questions: (1) How can a comprehensive evaluation framework be developed to systematically quantify the extent of carbon lock-in and its sub-dimensions? (2) What are the spatial and temporal evolution patterns of carbon lock-in and its sub-dimensions, and are there spatial correlations across regions? (3) What key factors influence the extent of carbon lock-in, and what strategies can effectively facilitate carbon unlocking?

The following are the contributions of this study to existing research. Firstly, this paper offers a systematic analysis of the mechanisms driving carbon lock-in. Building on the foundational connotation of carbon lock-in, the study examines its underlying dynamics within four aspects: industrial lock-in, technological lock-in, institutional lock-in, and social-behavioral lock-in. Based on these dimensions, a comprehensive measurement framework for assessing carbon lock-in levels is constructed to facilitate deeper insights into its persistence and mitigation pathways. Secondly, PCA is applied to evaluate the carbon lock-in level and its sub-dimensions in the YRD region, and the spatial and temporal evolution patterns and spatial characteristics of the overall carbon lock-in and its sub-dimension lock-in levels are comprehensively analyzed. Thirdly, this study examines the internal and external factors affecting carbon lock-in within the YRD region. They offer solid evidence for advancing the high-quality integrated development of this area and its transition to a green, low-carbon economy. The research process described in this paper is shown in Figure 1.

2. Research Methodology

2.1. Constructing the Evaluation System of Carbon Lock-In

The principles of systematicity, objectivity, representativeness, and data availability were used in selecting the indicators, based on the theoretical explanation of carbon locking and the underlying logic of its development. By referencing existing research [17,26,34], 18 specific indicators were identified across four dimensions, including industrial lock-in, technological lock-in, institutional lock-in, and social behavior lock-in. Using these indicators, a comprehensive measurement system for carbon lock-in in the YRD region was constructed (

Table 1. Carbon lock-in measurement index system.

Guideline Layer	Indicator Number	Indicator Layer	Unit	Nature
Industrial lock-in	(a1)	Carbon intensity of fixed assets (carbon emissions/investment in fixed assets of the whole society)	tonnes/ million	Positive
	(a2)	Value added to industry/value added to secondary industry	%	Positive
	(a3)	Foreign trade structure (export volume/total import and export volume)	%	Positive
	(a4)	Percentage of employment in tertiary industry	%	Negative
	(a5)	ESG score	-	Negative
Technological lock-in	(b1)	Energy intensity (total combined energy consumption/GDP)	tonnes of standard coal/million	Positive
	(b2)	Carbon intensity (carbon emissions/GDP)	tonnes/million	Positive
	(b3)	Number of green patents granted	piece	Negative
	(b4)	Urban Innovation Index	-	Negative
Institutional lock-in	(c1)	General public budget expenditure on science and technology	million	Negative
	(c2)	General public budget expenditure on energy conservation and environmental protection	million	Negative
	(c3)	Government public environmental concern	-	Negative
	(c4)	Artificial forestation area	hectare	Negative
Social behavioral lock-in	(d1)	Population density	per square kilometer	Positive
	(d2)	Private car	vehicle	Positive
	(d3)	Passenger turnover	ten thousand kilometers	Positive
	(d4)	Public environmental concern	-	Negative
	(d5)	Level of human capital	population with undergraduate degrees or above/total resident population of the city	Negative

).

Industrial lock-in, technological lock-in, institutional lock-in, and social behavioral lock-in do not exist in isolation but are mutually reinforcing through complex synergies, path dependencies, and feedback loops, which together maintain the stability of the high-carbon system and form a solid carbon lock-in structure. High-carbon industries rely on specific technology systems, and thecontinuous investment of enterprises in mature technologies creates technological lock-in. Technological lock-in further consolidates the competitiveness of high-carbon industries by reducing production costs and increasing scale effects, forming industrial path dependence. Institutional preferences and policy inertia promote the diffusion of carbon-based technologies, which further leads to institutional lock-in and ultimately the formation of a “technology–institution” complex. At the same time, the proliferation of carbon-based products in society has profoundly affected consumers’ psychological preferences. This sustained consumer demand, through market feedback, has exacerbated the embeddedness of high-carbon technologies and ultimately led to social behavior lock-in.

Industrial lock-in refers to the extent to which an economy or system relies on carbon-intensive industries [35]. This phenomenon often arises from path dependence on high-carbon industries during regional economic development, creating systemic barriers to transitioning to greener industries and policies [22]. The secondary industry is particularly vulnerable to this lock-in, as substantial investments in fixed assets with long life cycles entrench societies in carbon-intensive emission pathways that are both difficult and costly to alter [5]. Additionally, the strong upstream and downstream linkages in industrial sectors foster agglomeration, making it harder to shift toward lower-carbon technologies due to the “one-shot effect.” Cities heavily invested in high-carbon industries and fixed assets face higher risks of carbon lock-in [36]. The degree of industrial lock-in can be gauged by the carbon emission intensity associated with fixed assets, as well as the proportion of industrial value added in the secondary sector. Additionally, the degree of openness to foreign trade, reflected by the proportion of exports, provides opportunities to introduce advanced, low-carbon technologies. A growing tertiary sector, indicated by an increasing share of employment, also aids in reducing carbon emissions. Furthermore, enterprises play a critical role in breaking this lock-in through strong ESG performance, promoting sustainable practices, and directly contributing to carbon reduction. These actions are vital in facilitating the transition to a low-carbon economy.

Technological lock-in refers to the entrenchment of high-carbon technologies resulting from sustained investments and continuous updates by enterprises, driven by increasing returns to scale [11]. This phenomenon makes transition to low-carbon technologies increasingly challenging and costly, leading to a prolonged reliance on existing high-carbon systems [37]. Early in technological development, various technologies compete to meet market and societal demands. Once a high-carbon technology reaches “critical capacity” in the market [38], it becomes dominant, and as market penetration increases, complementary carbon-based technologies and assets emerge, reinforcing the high-carbon system [39]. Unlike the classical economic assumption of diminishing returns, technological systems often experience increasing returns to scale, which enhances the competitive advantage of certain technologies. This dynamic reduces production costs, accelerates adoption, and ultimately locks out alternative low-carbon solutions. Therefore, technological lock-in can be assessed through energy and carbon emission intensity, reflecting the prevalence of high-carbon technologies and the effectiveness of low-carbon substitutes. In addition, green technology innovation plays a significant role in carbon unlocking, so this article measures the combination of technological progress and economic development using indicators such as the number of green patents and the Urban Innovation Index, which reflect efforts to mitigate technological lock-in and reduce carbon emissions.

Institutional lock-in refers to the way policies and institutions enforce carbon-intensive practices, thereby impeding the shift to low-carbon alternatives. This concept emphasizes the critical role that government decisions play in shaping carbon emissions [2,16]. According to North (1990), institutional change follows a path of increasing rewards and self-reinforcement, where established policies tend to perpetuate themselves [40], and these policies can lead to either rapid optimization or entrenched inefficiencies, creating a “locked” state that is hard to escape without significant external intervention. During early industrialization, competitive local governments prioritized industrial expansion, frequently at the cost of environmental sustainability, thereby locking in high-carbon development models. Institutions shape techno-economic activities by offering incentives (e.g., subsidies, government procurement) and enforcing regulations (e.g., emission standards, environmental laws). Once these institutional structures are entrenched, path dependence may result in a carbon lock-in that is challenging to overcome. To measure institutional lock-in, indicators like public budget expenditures on science, technology, energy conservation, and environmental efforts, as well as forestation initiatives, are used. Additionally, government attention to environmental governance, such as energy conservation and emission reduction, is critical [41].

Social behavioral lock-in refers to how human behaviors and social practices contribute to sustained high carbon emissions [35]. As carbon-based technologies become embedded in society, they shape consumption patterns, usage habits, and social norms, creating a culture of carbon-intensive living [10,25]. This phenomenon, often referred to as “social embeddedness,” further strengthens carbon lock-in by aligning societal systems with carbon-based technological frameworks [42]. In this context, population density, consumption patterns, and transportation habits are key drivers of carbon emissions [17]. First, densely populated areas built on carbon-intensive infrastructures face challenges in transitioning to low-carbon systems due to the high economic and time costs involved. Second, transportation remains a major source of global emissions, with private car ownership and travel behaviors contributing substantially to this issue. Some scholars argue that altering travel behavior could reduce transport-related CO₂ emissions by as much as 50% by the end of the century [43]. Third, education and human capital improve environmental awareness, encouraging green investments and the adoption of sustainable technologies [44]. Therefore, indicators like population density, private car ownership, travel behavior, public environmental concern, and human capital are used to assess social behavioral lock-in.

2.2. Data Sources

The data used in this study are from a panel dataset covering 41 cities in the YRD region, spanning from 2000 to 2022. The data regarding carbon emissions are derived from the Emissions Database for Global Atmospheric Research (EDGAR). The Urban Innovation Index is derived from the China Urban and Industrial Innovation Index [45]. The ESG score originates from HuaZheng ESG Ratings. It excludes specific sectors like special financial and insurance industries, firms under Special Treatment (ST), and those with significant data gaps. Governmental environmental concern is derived from the annual governmental work reports of each region. Public environmental concerns are obtained from the Baidu Index search. Data for the remaining variables are sourced from past editions of the China Urban Statistical Yearbook, the Cathay Pacific database, as well as the EPS database.

2.3. Research Methods

2.3.1. Principal Component Analysis (PCA)

PCA is a multivariate statistical analysis method that employs dimensionality reduction to transform multiple variables into a small number of composite variables (principal components) by determining weights based on the inherent characteristics of the data [46]. As carbon lock-in permeates and integrates various aspects of industry and society, it is challenging to encapsulate it with a single quantitative index. Additionally, given that some measurement indicators are not inherently metric [47], the analysis involves transforming the inverse of certain indicators to ensure a consistent directionality. The original data are then standardized to achieve dimensionless processing, facilitating an accurate and consistent assessment [48].

2.3.2. Kernel Density Estimation (KDE)

KDE does not rely on any parametric assumptions and allows for the exploration of data distribution characteristics, and this method facilitates the estimation of the probability density function directly from the data, providing a continuous density curve that describes the shape of the distribution [49]. For a specified density function of a random variable, the probability density at a particular point (x) can be written as follows:

$$f(x)={\displaystyle \frac{1}{Nh}}{\displaystyle \sum _{i=1}^{N}K(\frac{X_i-x}{h}})$$

(1)

where $N$ denotes the quantity of observations; $h$ is value of the bandwidth; $X_i$ stands for the count of independent and identical data points; and $x$ is the mean. $k$ is the kernel function, which is presented in Equation (2):

$$k(x)={\displaystyle \frac{1}{\sqrt{2\pi }}}ex\rho (-{\displaystyle \frac{x^2}{2}})$$

(2)

2.3.3. Spatial Autocorrelation Analysis

Under the promotion of regional integration policies, the carbon lock-in level of YRD cities is not only affected by spatial geographic distance but also has certain economic correlations. Therefore, this paper draws on Fingleton and Gallo (2008) [50] to construct an economic distance spatial weights matrix to test the spatial autocorrelation of carbon locking in the YRD region, which can effectively reveal the compounding mechanism of geographic proximity and economic interaction on the spatial pattern of carbon locking. The calculation formula is shown in (5):

$$W_{ij1}=\left\{\begin{array}{c}{\displaystyle \frac{1}{d_{ij}^2}},(i\ne j)\\ 0,(i=j)\end{array}\right.$$

(3)

$$W_{ij2}=\left\{\begin{array}{c}{\displaystyle \frac{1}{\left|GD{P}_i-GD{P}_j\right|}},(i\ne j)\\ 0,(i=j)\end{array}\right.$$

(4)

$$W_{ij}=W_{ij1}\times W_{ij2}$$

(5)

Moran’s index is employed to detect the spatial correlation structure pattern across the entire study area [51,52]. It assesses the degree of spatial dependence between each decision-making unit and its neighboring units, thereby enabling the investigation of whether carbon lock-in is spatially clustered across different units. Spatial autocorrelation analysis can be categorized into two types. Global spatial autocorrelation is employed to evaluate the overall degree of spatial clustering of the study object, while local spatial autocorrelation identifies the presence of significant high and low values in specific areas [53,54]. The formulas for the global Moran index and the local Moran index are presented in Equation (6) and Equation (7), respectively:

$$I={\displaystyle \frac{n{\displaystyle {\sum }_{a=1}^n{\displaystyle {\sum }_{b=1}^n{w}_{ab}({\partial }_a-\stackrel{-}{\partial })({\partial }_b-\stackrel{-}{\partial })}}}{{\displaystyle {\sum }_{a=1}^n{\displaystyle {\sum }_{b=1}^n{w}_{ab}{\displaystyle {\sum }_{a=1}^n{({\partial }_a-\stackrel{-}{\partial })}^2}}}}}$$

(6)

$$I_a={\displaystyle \frac{{\partial }_a-\stackrel{-}{\partial }}{({\displaystyle {\sum }_{a=1}^n\frac{{\partial }_a-\stackrel{-}{\partial }}{n}{)}^2}}}{\displaystyle {\sum }_{b=1}^n{w}_{ab}({\partial }_b-\stackrel{-}{\partial })}$$

(7)

where $n$ indicates the aggregate quantity of the areas that have been observed, representing 41 cities in the YRD region used in this paper; $a \mathrm{a}\mathrm{n}\mathrm{d} b \mathrm{r}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{d}\mathrm{i}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{c}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{e}\mathrm{s}, \mathrm{t}\mathrm{h}\mathrm{a}\mathrm{t} \mathrm{i}\mathrm{s}$, $a\ne b$; ${\omega }_{ab}$ is the spatial weighting value; ${\partial }_a$ and ${\partial }_b$ denote the sample indexes of a and b, respectively; that is, the carbon lock-in levels of City a and City b; and the value of $\overline{\partial }$ is the mean value of the carbon lock-in levels for sample cities during the observation period.

2.3.4. Barrier Degree Analysis

The barrier degree model is employed to evaluate the extent of impediments at the level of individual indicators and guidelines, aiming to pinpoint the key to lowering the carbon lock-in level. The relevant formulae are presented below:

$$Q_{ij}={\displaystyle \frac{(1-X_{ij}^{\prime })\times {\omega }_j\times 100\%}{{\displaystyle \sum _{j=1}^n(1-X_{ij}^{\prime })\times {\omega }_j}}}$$

(8)

$$Q_i={\displaystyle \sum _{j=1}^n{Q}_{ij}}$$

(9)

where $Q_{ij}$ is the degree of barrier that a single indicator poses to alleviation of the carbon lock-in level; $Q_i$ stands for the degree of the barrier that the ith criterion layer presents to the reduction in the carbon lock-in; $X_{ij}^{\prime }$ means the normalized value of the jth secondary indicator and ${1-X}_{ij}^{\prime }$ represents the extend to which the indicator deviates; and ${\omega }_j$ is the weight assigned to the jth indicator.

2.3.5. Geodetector Analysis

Various methods exist to explore the mechanisms influencing geographic phenomena, such as spatial econometric models and geographically weighted regression models. However, these methods often fall short of effectively capturing the similarity of factors within the same region and the differences between different regions [55]. In contrast, the geodetector model is designed to detect spatial dissimilarities and reveal the driving forces behind them [56]. Therefore, in this paper, factor detection along with interaction detection are employed to identify the crucial influencing factors and the nature of their interactions that impact the spatial distribution pattern of carbon lock-in within the YRD region. The formula for factor detraction is as follows:

$$q=1-{\displaystyle \frac{{\displaystyle \sum _{h=1}^L{N}_h{\sigma }_h^2}}{N{\sigma }^2}}$$

(10)

where $q$ represents the explanatory degree of the influencing factor (n) with respect to carbon lock-in; ${\sigma }^2$ is the total sample size and variance, respectively; $N_h$ and ${\sigma }_h$ represent the sample size and sample variance of stratum h ($h=1,2,...,\mathrm{L}$), respectively; and $q$ takes the value with the range of (0, 1).

Factor interaction detection primarily serves to determine whether the combined influence of two distinct factors will amplify or diminish their impact on the spatial pattern of carbon sequestration. Additionally, it helps to ascertain whether these two factors exert independent effects on this spatial pattern. The evaluation method is to interact the calculated q values of the factors influencing carbon lock-in Y two by two and then compare their interaction values with their q-values. The specific relationship between two-factor detection can be referred to in Wang et al. (2017) [56].

3. Results

3.1. Temporal Evolution Trend of Carbon Lock-In

Figure 2 illustrates the trend of carbon lock-in in YRD cities from 2000 to 2022, showing that the degree of carbon lock-in has been decreasing since 2000. Additionally, the disparity in urban carbon lock-in has also diminished over this period. This trend is influenced by factors such as the integrated development strategy, energy consumption, and economic development of YRD cities. To gain further insight into the dynamic evolution of carbon lock-in over time, kernel density analysis was conducted for the years 2000, 2011, and 2022. As depicted in Figure 2, the kernel density curve has progressively shifted leftward, indicating a gradual reduction in the overall level of carbon lock-in and a concentration at lower levels. Over the years, the rightward extension of the curve has also shortened, signifying a growing number of cities with lower levels of carbon lock-in. Regarding the changes in the curve’s shape over time, the kernel density curve transformed dramatically from a “broad and low” configuration in 2000 to a “narrow and high” profile by 2022. This shift reflects substantial differences in carbon lock-in among cities, with an increasing concentration of cities experiencing lower levels of carbon lock-in as time progresses.

Figure 3, Figure 4, Figure 5 and Figure 6 depict the trends in the sub-dimensions of carbon lock-in in the YRD region. Figure 3, Figure 4 and Figure 5 reveal that the temporal trends of industrial lock-in, technological lock-in, and institutional lock-in in YRD cities aligned closely with the overall carbon lock-in trend, with a weakening carbon lock-in intensity since 2000. And the kernel density curves for these sub-dimensions progressively shifted leftward, signifying a gradual reduction in lock-in levels and a concentration at lower levels. Additionally, the shortening of the right-hand tails of the curves suggests that more cities are transitioning to states of reduced lock-in. However, Figure 6 shows a different trend for social behavioral lock-in, which rose from 2000 to 2012 and then fluctuated to a relatively stable level. In 2000, the kernel density curve for social behavioral lock-in shifted to the right, indicating an increasing and high concentration of social behavioral lock-in over time. This highlights the challenge of breaking the high carbon lock-in pattern in social behavior, which becomes a critical issue to address. Addressing this challenge requires the development of targeted strategies to fundamentally alter high-carbon behavioral patterns, a key factor in mitigating carbon lock-in in the future.

3.2. Spatial Pattern of Carbon Lock-In

To investigate the shifts in the geographical distribution pattern and the degree of the carbon lock-in phenomenon, each dimension was classified into five categories: very high, high, medium, low, and very low. Spatial visualizations covering 2000, 2011, and 2022 were created with the ArcGIS 10.8 software. As shown in Figure 7, the spatial differentiation in carbon lock-in across the YRD region had generally narrowed over time. In 2000, the region displayed a generally high carbon lock-in level. Lower-value areas were primarily concentrated in core cities such as Nanjing, Suzhou, and Hangzhou, while peripheral cities, particularly in the western and northern parts of Anhui, exhibited higher levels of lock-in. By 2011, a downward trend in carbon lock-in levels became apparent, with many cities in southern Zhejiang, northern Jiangsu, and Anhui shifting from high to medium or relatively low levels, displaying a “high in the west and low in the east” trend. This pattern continued, with all cities achieving lower levels of carbon lock-in by 2022. Shanghai, Hangzhou, and Nanjing have seen further reductions in their carbon lock-in. Overall, the carbon lock-in level in the YRD region had decreased, consistently showing a “high in the west and low in the east” distribution.

The YRD region’s various dimensions of lock-in show marked spatial distribution differences. Industrial, technological, and institutional lock-in generally decreased from west to east, while social behavioral lock-in increased from north to south. As seen in Figure 8a–c, in 2000, most YRD cities exhibited medium to high levels of industrial (95.12%), institutional (65.85%), and technological (97.56%) lock-in, with a clear “high in the west, low in the east” distribution. By 2011, these lock-in levels had decreased, and the “high in the west, low in the east” pattern weakened, reflecting more balanced regional development. By 2022, overall lock-in levels had further reduced, indicating an integrated development pattern across the YRD. However, as shown in Figure 8d, social behavioral lock-in rose in comparison, with cities like Hangzhou, Ningbo, Nanjing, and Suzhou showing high levels early on. By 2011, most cities in Zhejiang, Jiangsu, and Anhui provinces experienced high social behavioral lock-in (90.24%), especially in provincial capitals like Hefei, Nanjing, and Hangzhou. By 2022, the level of social behavioral lock-in in the YRD region has further increased.

3.3. Spatial Association Analysis of Carbon Lock-In

This paper further employed an economic distance spatial weights matrix and utilized the Stata 17 software to estimate the global Moran index of carbon lock-in (

Table 2. The global Moran I index of carbon lock-in.

Year	I	Z	p-Value	Year	I	Z	p-Value
2000	0.304	4.433	0.000	2012	0.206	3.127	0.001
2001	0.295	4.815	0.000	2013	0.311	4.576	0.000
2002	0.316	4.832	0.000	2014	0.238	3.707	0.000
2003	0.371	5.497	0.000	2015	0.235	3.612	0.000
2004	0.253	3.800	0.000	2016	0.227	3.419	0.000
2005	0.306	4.586	0.000	2017	0.292	4.237	0.000
2006	0.289	4.261	0.000	2018	0.275	3.956	0.000
2007	0.266	4.082	0.000	2019	0.206	3.056	0.001
2008	0.253	3.883	0.000	2020	0.204	3.085	0.001
2009	0.286	4.385	0.000	2021	0.193	2.927	0.002
2010	0.252	3.875	0.000	2022	0.142	2.326	0.010
2011	0.198	3.066	0.001

) and its sub-dimensions (

Table 3. Global Moran’s I for the carbon lock-in sub-dimension.

Year	Industrial Lock-In	Technological Lock-In	Institutional Lock-In	Social Behavior Lock-In
2000	0.500 ***	0.341 ***	0.115 *	0.108 *
2001	0.450 ***	0.219 **	−0.092	0.129 *
2002	0.418 ***	0.237 ***	−0.012	0.125 *
2003	0.417 ***	0.164 **	0.061	0.129 *
2004	0.352 ***	0.132 *	0.012	0.123 *
2005	0.284 ***	0.282 ***	0.273 ***	0.100 *
2006	0.259 ***	0.312 ***	0.390 ***	0.170 **
2007	0.176 **	0.305 ***	0.453 ***	0.144 *
2008	0.157 **	0.296 ***	0.385 ***	0.119 *
2009	0.128 *	0.426 ***	0.391 ***	0.087 *
2010	0.162 **	0.370 ***	0.387 ***	0.082 *
2011	0.158 **	0.357 ***	0.357 ***	0.020
2012	0.216 **	0.334 ***	0.317 ***	0.005
2013	0.104 *	0.323 ***	0.179 **	−0.049
2014	0.070	0.120 *	0.071 *	−0.014
2015	0.064	0.125 *	0.099 *	0.022
2016	0.068	0.122 *	0.082 *	0.020
2017	0.118 *	0.296 ***	0.001	0.040
2018	0.082 *	0.303 ***	−0.032	0.049
2019	0.063	0.284 ***	−0.93	0.066
2020	0.105 *	0.191 **	0.029	0.089 *
2021	0.054	0.272 ***	−0.045	0.072
2022	0.151 *	0.406 ***	−0.067	0.210 **

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

) in the YRD region from 2000 to 2022. The global Moran index of carbon lock-in during the study period remained above 0, indicating a strong positive spatial correlation. This suggests that the carbon lock-in level in a city not only influenced its surrounding cities but was also affected by neighboring cities, exhibiting spatial clustering of similar values. The global Moran index shows a trend of “high and low turnover,” with a gradual weakening of carbon-locked inter-city linkages after 2018. The above findings provide the basis for the empirical analysis of this study using spatial econometric modelling.

Table 3 shows the different spatial correlations of industrial lock-in, institutional lock-in, technological lock-in, and social behavioral lock-in of the YRD region during 2000–2022. Industrial lock-in exhibits a spatial agglomeration trend in most years, although small fluctuations indicate a weakening of this agglomeration over time, reflecting a decline in its overall intensity. In contrast, the Moran index of technological lock-in shows an overall upward trend, indicating a significant spatial agglomeration feature of technological lock-in. In contrast, the global Moran index for institutional lock-in was initially insignificant and shifted to significant from 2005 to 2016, suggesting that the spatial correlation has strengthened over time. However, it is worth noting that institutional lock-in shows a negative index in 2018 and beyond. The Moran index for social behavioral lock-in went through a process of significance to non-significance.

Figure 9 illustrates the local spatial pattern of carbon lock-in within the YRD region for 2000, 2011, and 2022. The spatial distribution of carbon lock-in within the region is characterized by the predominance of high–high and low–low agglomeration types, which are relatively balanced. Over time, the high–high agglomeration type had decreased and dispersed. Moreover, high–high agglomerations had formed in the periphery of relatively underdeveloped areas, including cities in southwestern Zhejiang and western and northern Anhui. These areas are distant from the core economic activities of the YRD and primarily rely on industrial development, leading to greater resistance to low-carbon transformation. Conversely, low–low agglomerations were mainly concentrated in Shanghai, southern Jiangsu, and northern Zhejiang, forming cluster-like spatial agglomerations. This is caused by the YRD core area’s strong economic ties, which drive industrial structure upgrading and green technology development, promoting a carbon-based technological system transition and reducing carbon lock-in.

3.4. Barrier Diagnosis of Carbon Lock-In

This paper evaluated the individual indicator barriers to carbon lock-in in the YRD region using a barrier degree model; but, to improve the brevity and readability of the results, only key barrier indicators (the top five barriers per year) are reported, and the timeframe was summarized into three periods: 2000–2007, 2008–2015, and 2016–2022. As shown in

Table 4. Barrier indicators for carbon lock-in.

Period (Year)	Top Barriers (Indicator Codes)	Average Barrier Degree (%)
2000–2007	b2, a1, c2, b3, d4	26.59, 14.86, 9.28, 5.97, 4.98
2008–2015	b2, a1, b1, c1, c2	31.42, 16.25, 6.84, 6.14, 5.11
2016–2022	b2, a1, b1, c1, c2	33.29, 16.84, 8.27, 4.46, 3.97

, it is worth noting that carbon emission intensity (b2) and carbon intensity of fixed assets (a1) had consistently ranked among the top barriers on average in each period, and both were on an upward trend, although the year-on-year growth rates had decreased. Specifically, the barrier level for carbon emissions intensity increased from 26.59% to 33.29%, and the barrier level for carbon intensity of fixed assets increased from 14.86% to 16.84%. In addition, the barriers of energy intensity (b1), number of green patents (b3), energy conservation and environmental protection expenditures (c2), science and technology expenditures (c1), and public concern for the environment (d4) were larger in comparison with other indicators but showed a decreasing trend with the YRD region’s efforts to progressively break the dependence on high-carbon paths in the dimensions of technology, policy, and social behavior.

In Figure 10, it is evident that the average barrier degree of technological lock-in was the highest at 41.49%, followed by institutional lock-in and industrial lock-in with average barrier degrees of 31.72% and 16.68%, respectively, while social behavioral lock-in had the lowest average barrier degree at 9.12%. The declining trend in the barrier degrees of institutional lock-in and social behavioral lock-in suggests improvements in the areas, driven by increased environmental awareness and enhanced carbon reduction and emission control systems. In contrast, the barrier degree of technological lock-in fluctuated from 36.48% in 2000 to 44.50% in 2022 and overtook institutional lock-in after 2004, with an average annual increase of approximately 0.36%. The reason for this trend may be that the technological density of the YRD region shows a clear “core-periphery” structure, with a “Z”-shaped technology-intensive zone formed with Hefei, Shanghai, Hangzhou, and Ningbo as the core. Although this structure promotes the technological development of the core region, it also leads to excessive concentration of resources, and the technological development of the peripheral regions lags behind, which increases the difficulty of technological diffusion and forms an obstacle to technological lock-in. The extent of barriers to technological lock-in tends to be more complex and more difficult than institutional lock-in to eliminate. Technology lock-in is not only dependent on the complexity of the technology itself but also on the ability of technology diffusion in the region, the innovation capacity of enterprises, and the synergy capacity of universities and research institutions. For example, the “proximity effect” of technology diffusion suggests that the diffusion of technology within a region is limited by geographical distance and knowledge gaps. Barriers to industrial lock-in also showed a fluctuating upward trend, climbing from 13.77% in 2000 to 17.77% in 2022. Although the growth was modest, it was closely related to the early concentration of secondary industries in the YRD region, which had led to the predominance of traditional heavy industries and energy-intensive industries with high carbon emissions and low energy efficiency. Overall, the main barriers to carbon lock-in are technological lock-in and institutional lock-in in the YRD region, mainly due to the fact that high-carbon technological systems and the existing institutional framework reinforce each other, forming a “technology–institutional complex” that hampers the spread of low-carbon technologies.

3.5. The Geodetector Analysis of Driving Factors of Carbon Lock-In

Drawing on the work of related scholars, this paper examines the carbon lock-in levels of cities as the dependent variable and establishes an influencing factors indicator system based on six dimensions: economic development, digital technological innovation, industrial structure, urbanization level, development of transport infrastructure, and informatization level (

Table 5. Indicators for impact factor detection.

Detection Factor	Representative Indicator	Unit
Level of economic development	Per capita GDP (x1)	yuan/CNY
Digital technological innovation	Digital patent applications (x2)	piece
Industrial structure upgrading	Output value of tertiary industry/output value of secondary industry (x3)	-
Urbanization level	Urban population/total population (x4)	%
Informatization level	Postal and telecommunication services per capita (x5)	CNY 10,000
Level of green finance	Entropy measurements (x6)	-

). To analyze the spatial variability of the region’s carbon lock-in patterns, the study employs the geodetector model, focusing on both factor detection and interaction detection. Economically developed cities tend to have a more diversified industrial base, increasing the likelihood of adopting industries with lower carbon footprints [35], and are characterized by GDP per capita [57], which reflects not only the stage of economic development but also the associated energy consumption patterns. Digital technological innovations enable the integration and intelligence of urban systems, improve urban operational efficiency, reduce environmental impacts [58], and foster carbon reduction through spatial synergy effects, such as the “siphon effect” and “demonstration effect,” with spillover benefits reducing carbon intensity in neighboring urban agglomerations of similar economic levels [59]. The number of patent applications serves as an indicator of digital technological innovation [32,60]. Optimizing industrial structure, measured by the ratio of tertiary to secondary industry output, can help eliminate carbon lock-in [61]. Additionally, rapid urbanization, indicated by the proportion of urban population to total population, can increase energy consumption and carbon emissions, thereby intensifying the carbon lock-in effect [62]. And the level of informatization, which enables real-time monitoring and management of carbon dioxide emissions, is measured by per capita telecommunication services [63]. Finally, green finance contributes to alleviating carbon lock-in both directly and indirectly by promoting green technological innovation [25].

The K-means clustering method of SPSS stat 29 was used to transform the numerical variables into discrete variables, and the geographic detector analysis yields the influence value (q-value) of each factor on the spatial pattern of carbon lock-in in the YRD region from 2000 to 2022 (Figure 11). The findings indicate that the influencing factors of the carbon lock-in spatial pattern in descending order are x1 > x4 > x2 > x5 > x6 > x3. This suggests that economic development, urbanization processes, and digital technology innovation are the main drivers of the evolution of carbon locking patterns in the YRD region. In addition, industrial structure upgrading, informatization levels, and green finance development also have an important influence on the formation of carbon locking spatial patterns.

An examination of the q-values reveals that both socioeconomic factors and the development of digital technology have played significant roles in carbon lock-in in the YRD region. Economic development ranked as the top influence, highlighting it as the key force driving regional carbon lock-in. Since the initiation of China’s reform and opening up, the YRD region has exhibited notable spatial disparities in economic development. More developed cities have better access to advanced technologies and greater financial resources for supporting green initiatives, accelerating their transition to a low-carbon economy and reducing carbon lock-in. In contrast, less developed cities face more significant challenges in achieving these outcomes. Digital technology advancements also exerted a greater impact. Digital technologies can help improve energy efficiency by optimizing the energy production and consumption chain, while they may help mitigate carbon lock-in as they can pinpoint carbon emission sources and optimize carbon-neutral pathways. While urbanization typically increases carbon lock-in due to higher energy consumption, this effect has slowed in recent years, likely due to improvements in urban planning practices. The development of information technology provides additional ways for public participation in environmental protection, enhances perception and understanding of environmental issues, and facilitates the monitoring of environmental violations. Information technology development also had a greater influence on carbon lock-in. In addition, green finance also has a strong influence on carbon lock-in. The growth of green finance has fostered the development of the carbon trading market, enhancing liquidity and pricing efficiency through financial instruments such as carbon futures, carbon options, and other innovative products. This not only provides economic incentives for carbon emission reduction in enterprises but also promotes the low-carbon transition of the whole society through the market mechanism.

The detection results of the interactions showed that the interactions of two independent influencing factors, one factor nonlinearly attenuated and the other nonlinearly attenuated, were not found, and there were only different degrees of nonlinear enhancement or two-factor enhancement. This suggests that carbon lock-in in the YRD region results from the joint influence of multiple factors, and the interaction between the influencing factors will enhance the explanatory power of carbon locking evaluation.

4. Discussion

As an important engine of China’s economic development and urbanization, the YRD region is one of the regions with high energy consumption and carbon emissions, and its transformation to low-carbon development has an important role to play in demonstrating and leading the nation and the world. This study extends the carbon lock-in theoretical framework by integrating four sub-dimensions (industrial, technological, institutional, and social-behavioral lock-ins), which is consistent with the multidimensional perspective of recent studies [64,65] but diverges from earlier studies that focused solely on technological or institutional dimensions [3,7]. Notably, the YRD region has demonstrated a steady decline in carbon lock-in over the years, which is mirrored in the downward trends of industrial, technological, and institutional lock-ins, which can be attributed to technological advancements and structural optimizations within the YRD economy, reflecting a shift away from traditional high-carbon practices. As one of China’s most developed regions, the YRD has also been proactive in implementing policies, such as the “Implementation Plan for Peak Carbon Achievement in the Yangtze River Delta Eco-Green Integrated Development Demonstration Zone,” aimed at accelerating green and low-carbon development through regional cooperation and policy alignment.

However, the social behavioral lock-in dimension has shown a persistent upward trend, likely driven by the region’s economic prosperity, which fosters energy-intensive lifestyles, including increased private car use. This highlights the need to address high-carbon social behaviors, as well as the influence of cultural diversity and high population mobility, which may hinder the spread of green development norms. Enhancing public participation and using behavioral incentives to guide lifestyle changes toward sustainability are crucial steps in this direction. The study also finds strong positive spatial correlations in carbon lock-in across the YRD, reflecting the clustering of high-carbon activities in areas with concentrated economic output. The “high in the west, low in the east” spatial pattern of carbon lock-in underscores the need for region-specific approaches. Intensifying emission reduction efforts in the high-carbon western areas while optimizing low-carbon strategies in the east will be effective. Additionally, fostering regional cooperation through mechanisms like carbon trading, shared technology initiatives, and collaborative green supply chain innovations can further mitigate spatial carbon disparities.

Different spatial trends are observed in the carbon lock-in sub-dimension. It is worth noting that the spatial clustering effect of industries has weakened over time. This may be due to the fact that with the optimization and upgrading of the industrial structure, the clustering effect of traditional industries has gradually weakened. And there are differences in the degree of industrial structure upgrading among regions, leading to a more dispersed distribution of emerging industries. On the contrary, the spatial agglomeration of technology lock-in is increasing. Although technology diffusion promotes regional innovation to a certain extent, the technological advantages of core cities still dominate, such as Shanghai, Nanjing, and Hangzhou. The technological innovation capacity of these cities has been increasing, forming the agglomeration effect of technological lock-in. In addition, the continuous adjustment of inter-regional competition and cooperation may be the reason why the spatial relevance of institutional lock-in shows different characteristics at different stages. Social behavioral lock-in shows limited spatial relevance, suggesting random and localized effects, which may be influenced by the shift from high-carbon to low-carbon industries.

Despite the weakening degree of technological and institutional lock-in, technological lock-in remains a major barrier to mitigating carbon lock-in, as switching from high-carbon to low-carbon technologies requires substantial capital investment and time costs, especially in energy and industry, where the inertia of the technological system makes switching more difficult. In addition, although the YRD region has made some progress in regional integration, there are still deficiencies in the synergy of low-carbon policies. Policy differences and administrative barriers in different regions limit the cross-regional diffusion of low-carbon technologies, so institutional lock-in is also a major obstacle to mitigating carbon lock-in. Conversely, social behavioral lock-in, though rising, poses a smaller barrier, suggesting that behavioral transformations may be more feasible with lower associated costs. While economic development has spurred increased carbon emissions, it also provides pathways for green transformation through innovation and structural upgrades. The adoption of digital technologies further aids in carbon unlocking, leveraging spillover effects that enhance efficiency and reduce emissions. Achieving the “dual carbon goal” will require sustained effort across multiple domains. Internally, this involves industrial restructuring, targeted emission reductions in key sectors, and technology development. Externally, economic growth, digital innovation, urbanization, and informatization are pivotal for reducing carbon lock-in. Expanding digital applications, fostering smart urban growth, and promoting low-carbon awareness among residents are vital steps toward a sustainable, low-carbon future in China.

This study has some limitations that deserve further attention. First, the evaluation indicator system for carbon lock-in still needs to be improved. Due to the lack of consensus on a unified indicator set for carbon lock-in and limited data availability, key representative indicators were selected for this study. However, with the continuous enrichment of the connotation of carbon lock-in, a wider range of indicator systems should be included to achieve a more comprehensive and multidimensional assessment. Second, there is a need to increase the expansiveness of the study’s conclusions. Since this paper only discusses the spatial and temporal evolution patterns of carbon lock-in and the influencing factors of representative urban agglomerations in China, it may lead to a lack of expandability of the research results. Future research will consider using more extensive data to increase the general applicability of the findings.

5. Conclusions and Policy Implications

5.1. Conclusions

This study applied PCA to calculate the composite scores of carbon lock-in for 41 cities in the YRD region from 2000 to 2022 and used KDE to capture the spatiotemporal evolution of carbon lock-in and its sub-dimensions, as well as the Moran index to analyze the spatial autocorrelation. The results show that the overall carbon lock-in level in the YRD region has shown a stable downward trend during this period. From the perspective of each dimension, the trend of carbon lock-in in the industrial, technological, and institutional dimensions has weakened, while the carbon lock-in in the social behavior dimension has slightly increased. In terms of spatial distribution, the level of carbon lock is relatively high in the west of the region and relatively low in the east. As measured by the Moran index, there is a significant positive spatial autocorrelation within the region, indicating that cities with high or low carbon lock tend to cluster together within the region. In addition, the main barriers and influencing factors of carbon lock-in were identified using the barrier degree model and the geodetector model. At the indicator level, the carbon intensity of fixed assets and carbon emission intensity barriers are the most prominent, and at the criterion level, technological lock-in and institutional lock-in have a more pronounced role in constraining carbon lock-in. Finally, the level of carbon lock-in in the YRD region is influenced by the interaction of various factors, among which the influence of economic development and digital technology adoption is more prominent. Factors such as the improvement of information technology, the development of green finance, and the upgrading of industrial structure also play a role, suggesting that the region needs a multidimensional strategy to mitigate carbon lock-in.

5.2. Policy Implications

While the overall trend of carbon lock-in in the YRD region is positive, the lock-in effect in the social behavior dimension remains significant, exhibiting spatial differentiation and correlation. In response, policies must adopt a multi-dimensional and synergistic approach. Firstly, it is essential to strengthen strategies for guiding social behavior. This can be achieved by incentivizing low-carbon lifestyles, enhancing public engagement in policymaking, and breaking the inertia that leads to behavioral lock-in. Secondly, given the significant regional differences, high-carbon regions in the west should focus on deep emissions reductions, while eastern regions should prioritize optimizing existing low-carbon models. Concurrently, regional collaboration must be strengthened through joint carbon market development, shared low-carbon technologies, and the promotion of green supply chain coordination to address the spatial interconnectivity of carbon emissions. Finally, the fundamental path lies in accelerating industrial and energy structure transformation, vigorously supporting the upgrading of high-carbon industries toward low-carbon technologies, cultivating clean energy industries, and promoting energy-saving technologies to enhance energy efficiency. Additionally, the integration of digital technologies will play a pivotal role in driving the low-carbon transformation and enhancing overall sustainability in the region.