Regional Space–time Differences and Dynamic Evolution Law of Real Estate Financial Risk in China

2.1 Identification of Real Estate Financial Risk

Selecting scientifically sound and reasonable risk assessment indicators is crucial for accurately identifying the level of real estate financial risks. Existing studies vary in their methods for identifying real estate financial risks, primarily due to differing definitions and interpretations of what constitutes such risks. Currently, there are three primary approaches to identifying real estate financial risks. The first approach defines real estate financial risk as the economic losses incurred by commercial banks during the provision of financial services, such as funding, financing, and clearing, for the real estate sector due to uncertain factors. Under this method, risk can be measured using indicators such as the ratio of real estate development loans to domestic loans or real estate investment or the non-performing loan ratio of commercial banks. A higher value of these indicators signifies more severe systemic financial risks caused by defaults of real estate enterprises that result in an inability to repay bank loans (Han et al., 2011; Ju et al., 2018; Liu et al., 2014; Zhang et al., 2023). The second approach considers real estate financial risk to be primarily driven by fluctuations in housing prices. When the housing price-to-income ratio for households exceeds a reasonable range or when the debt burden of residents in a region becomes excessive, the likelihood of real estate financial risks increases. Consequently, housing prices, housing price growth rates, and the housing price-to-income ratio can be used as key indicators to measure real estate financial risks (Ge & Li, 2022; Jiang et al., 2021; Romainville, 2017; Tang & Mao, 2020). The third approach views real estate financial risk as arising from the interactions among multiple market participants. In the interconnected development of these participants, imbalances in their own operations or external disturbances may lead to economic losses for one or more entities, which in turn could trigger financial losses for banks and other financial institutions. To address this complexity, a multidimensional risk assessment framework can be constructed by incorporating factors related to government agencies, financial institutions, real estate enterprises, and consumers. Methods such as principal component analysis, weighted average, and entropy can be employed to assign weights to each indicator (Bai et al., 2020; Ge & Sun, 2022; He & Cheng, 2016; Hu, 2017; Xu et al., 2022; Yuan & Jing, 2024; Zhou & Sun, 2019; Zhou & Zheng, 2007; Zhou et al., 2021).

2.2 Spatial Characteristics of Real Estate Financial Risk

Although the real estate market exhibits regional characteristics, regional markets are not entirely independent. Mutual contagion effects may arise between markets due to factors such as information spillover, economic development, financial linkages, and population movements. Currently, most research on the spatial characteristics of real estate financial risk focuses on analyzing the spatial spillover effects of housing prices or housing price bubbles. In studies of housing prices spatial spillover effects, many scholars argue that inter-regional transmission of housing prices occurs regularly across various regions. This phenomenon is attributed to factors such as geographical proximity, population migration, capital flow, spatial arbitrage, and other elements. Using methods such as spatial econometric models, social network analysis, directed acyclic graphs, and information spillover index approaches, studies have shown that housing prices across regions exhibit significant positive spatial spillover and linkage effects (Chen & Wang, 2012; Chen et al., 2016; Ding & Ni, 2018; Holmes, 2007; Liu et al., 2022; Wang et al., 2021). Regarding the spatial spillover effects of housing price bubbles, most studies suggest that housing price bubbles spread to other regions through factors such as population migration, geographical proximity, economic development linkage, and information transmission. This diffusion process exhibits prominent spatial agglomeration characteristics (Fry, 2009; Füss et al., 2011; Guo & Chen, 2018; Li & Zhang, 2019; Liu & Lv, 2018; Nneji et al., 2015; Wood, 2003; Zheng et al., 2021).

However, the spatial spillover of housing prices and housing price bubbles does not necessarily imply a spatial spillover of real estate financial risk (Ge & Li, 2022). Housing prices and housing price bubbles reflect risks arising from supply and demand dynamics, whereas real estate financial risk is influenced by a multitude of factors, such as financial policies, local government land policies, and external macroeconomic conditions. Therefore, the spatial spillover patterns of housing prices or bubbles cannot fully represent the spatial spillover patterns of real estate financial risk in China.

Currently, there is limited literature directly investigating real estate financial risk. Ju et al. (2018) examined the relationship between China’s real estate financial risk and housing vacancy rates (HVRs), finding a significant positive spatial correlation in real estate financial risk. Ge and Li (2022), in their study of real estate financial risk within the residential sector of the Beijing–Tianjin–Hebei region, identified a significant spatial spillover effect and highlighted that housing demand and income levels are critical factors influencing the spatial spillover of risk. Additionally, Xu et al. (2022) and Zhang et al. (2023) identified a significant spatial dependency in real estate financial risk across provinces in China.

In summary, current research on real estate financial risks has yet to establish a unified index evaluation system, and existing studies have not focused on identifying such risks from their root causes. In terms of research content, there is a notable lack of analysis on intraregional differences and spatial distribution characteristics of real estate financial risks across China’s regions from both temporal and spatial perspectives. Additionally, studies exploring the dynamic evolution laws of real estate financial risks from a temporal progression perspective remain limited. This study conducts a comprehensive exploration of the aforementioned issues and offers an empirical foundation for accurately preventing regional real estate financial risks.

3.1 Construction of Multidimensional Index System of Real Estate Financial Risk

Due to the involvement of multiple stakeholders, industries, and departments, as well as the strong regional characteristics of the market, it is challenging to comprehensively and accurately measure real estate financial risk levels using only a single variable. Therefore, a more systematic and comprehensive analysis is necessary. Following Ge and Sun (2022), this study establishes a real estate financial risk evaluation index system from three perspectives: internal risk factors, external risk factors, and interconnected risk factors in real estate financial activities. The specific index system is divided into six dimensions. Internal risk factors refer to elements directly associated with the real estate market itself that contribute to financial risks, including supply and demand dynamics. External risk factors refer to macro-control policies and other external elements that may trigger financial risks in the real estate sector. Interconnected risk factors encompass related sectors, such as local governments or financial institutions, that influence real estate financial risks. Table 1 provides the specific selection of indicators.

3.2 Decomposition of Theil Index

The Theil index is a well-established metric for measuring regional differences, particularly in economic and financial contexts. It allows for the decomposition of overall differences into within-group and between-group components, providing insights into the regional difference in real estate financial risks. Compared to alternative measures, such as standard deviation, the Theil index offers a more nuanced understanding of spatial heterogeneity. This study employs the Theil index to analyze the differences in China’s real estate financial risk levels within and across various regions, as well as to measure the overall national differences, intra-regional differences, inter-regional differences, and their respective contribution percentages. The value of the Theil index typically ranges between 0 and 1. A lower Theil index indicates reduced overall differences in the level of real estate financial risk, whereas higher values correspond to greater overall differences. The formulas for calculating the overall Theil index, the overall regional difference Theil index, and the intra-regional and inter-regional Theil index are provided in equations (8)–(12):

Here, T represents the overall difference of the national real estate financial risk level Theil index; q denotes province; n denotes the number of provinces; R q express the real estate financial risk level of province q ; and R ¯ represents the average of the national real estate financial risk level. T p denotes the total differences Theil index of region p ; n p is the number of provinces in region p . R p q represents the real estate financial risk level of q province in p ; R ¯ p represents the average value of real estate financial risk in region p . Additionally, the overall Theil index is further divided into two components: the intra-regional Theil index, represented by T w , and the inter-regional Theil index, represented by T b . The respective calculation formulas are provided in equations (10) and (11). T w / T and T b / T are recognized as the contribution percentages of intra-regional differences and inter-regional differences to overall differences (Zhu & Chen, 2023). ( R p / R ) × ( T p / T ) is the contribution percentage of each region to the overall intra-regional difference. R p represents the sum of the real estate financial risk levels across various regions within area p ; R represents the total national real estate financial risk level.

3.3 Moran’s Index

The Moran’s index, introduced in 1950, is the most commonly used method for measuring spatial correlation (Moran, 1950). Its application in this study is critical for identifying spatial dependencies in real estate financial risks, which are often influenced by geographic proximity. In contrast to simple regression models, Moran’s index reveals underlying spatial relationships that may otherwise be overlooked. The value of the Moran’s index typically ranges between −1 and 1. A positive value indicates a spatial positive correlation between real estate financial risks in various regions, meaning that high-risk areas tend to cluster with other high-risk areas (high–high [HH] agglomeration) and low-risk areas cluster with other low-risk areas (low–low [LL] agglomeration). Conversely, a negative value signifies a spatial negative correlation between real estate financial risks, indicating that high-risk areas tend to cluster with low-risk areas (HL agglomeration and LH agglomeration). A value of 0 suggests no discernible spatial correlation, implying that the distribution is random. The Moran’s index is classified into two categories: the global Moran’s index and the local Moran’s index. The global Moran’s index primarily assesses the overall degree of correlation across regional spaces. Its calculation formula is provided in equation (13):

The local Moran’s index examines the degree of spatial correlation among distinct regions. The formula is provided in equation (14):

Here, n represents the total count of provinces. R i is the real estate financial risk level of region i . R ¯ signifies the mean value of the national risk level. S is the sample variance, and w i j is the geographical weight matrix based on the neighboring relationship of provinces.

3.4 Kernel Density Function

The kernel density function provides a smoothed visualization of financial risk distributions, making it highly effective for analyzing temporal changes. Typically, the distribution of real estate financial risks evolves over time, which makes it challenging to determine its exact distribution function. Non-parametric kernel density estimation can be employed without the need to predefine the specific form of the distribution function. For real estate financial risks with random variables, the kernel density function is provided in equation (15):

where K ( ⋅ ) represents the kernel function, R 1 , … R n is the real estate financial risk level of each region, and the mean value is denoted by R , n signifies the total number of observations, and h is the window width. Considering that the Gaussian function, when used as a kernel function, provides a smooth and continuous density estimate that aligns well with the distribution of many economic phenomena, it proves highly effective in fitting real-world data distributions. Therefore, this study employs the high-precision Gaussian kernel function to estimate the dynamic distribution evolution characteristics of real estate financial risk in China.

3.5 Markov Chain Model

The Markov chain model is employed to analyze the dynamic evolution laws of financial risk states, focusing on transition probabilities and long-term steady states. This method is particularly suitable for capturing the progression of risk across different regions and time periods, providing insights into potential stabilization or escalation trends. The Markov chain model excels at exploring state transitions within a stochastic framework, offering an advantage over time-series models. The basic model is established as follows: Assume that ( X n , n ≥ 0 ) is a random process, S = ( 1 , 2 , … ) is its state space, for any i 0 , i 1 ⋯ , i n + 1 ∈ S , n ∈ N 0 and P { X 0 = i 0 , X 1 = i 1 … , X n = i n } > 0 , the following holds:

where the random sequence ( X n , n ≥ 0 ) is referred to as a Markov chain. According to the calculation results of real estate financial risk, this study divides the risk into three levels: low, medium, and high. It is assumed that the change in risk follows a Markov process, that is: ( R n , n ≥ 0 ) is a random process. According to the classification of risk, the state vector of China’s real estate financial risk at time n is S ( n ) :

Then, the one-step transition probability matrix of China’s regional real estate financial risk from time n to time n + 1 is:

Here, the elements on the principal diagonal of the probability matrix represent the probability that the three real estate financial risk levels remain unchanged, while the off-diagonal elements indicate the transfer probability. For instance, the element in the second column of the first row denotes the likelihood of a low-risk level transitioning to a medium-risk level.

3.6 Data Source

The data of this study are sourced from the China Real Estate Statistical Yearbook, China Statistical Abstract, Wind database, and the National Bureau of Statistics. To ensure data consistency and comparability over time, the study covers the period from 2006 to 2022. Due to the lack of data on Tibet, the sample includes 30 provinces, excluding Hong Kong, Macao, and Taiwan. The HVR is defined as the ratio of the housing area available for sale to the total completed housing area by real estate development enterprises over the past 3 years. The fiscal deficit ratio (FDR) is defined as the proportion of the absolute difference between fiscal revenue and expenditure to GDP. Data required for the calculation of the remaining indicators are directly sourced from the aforementioned databases. Additionally, to address the issue of missing data on the supply area of construction land, this study employs the land area purchased by real estate development enterprises as a proxy variable to fill the data gaps. The descriptive statistics and weights of each variable are shown in Table 2. It can be observed that among the internal risk factors, the weight of REI is the highest at 0.179, indicating that real estate development investment has a significant impact on real estate financial risk. Among the external risk factors, the weight of M2 is the highest at 0.104, suggesting that the money supply is a critical factor influencing real estate financial risk. For interconnected risk factors, REL, REDL, and FDR have relatively high weights of 0.146, 0.176, and 0.142, respectively. This indicates that interconnected risk factors are key contributors to real estate financial risk. As mentioned earlier, the emergence of real estate financial risk is closely tied to the reliance on a single financing channel in China’s real estate industry and the intervention of local governments in the land market.

4.1 Analysis of Risk Measurement Results

4.1.1 Temporal Evolution of China’s Real Estate Financial Risk Level

According to the calculation results of real estate financial risk, this part analyzes its temporal evolution and regional differences from the perspective of China’s four major regions.^[1] Figure 1 visually depicts the temporal evolution of real estate financial risk levels in these four regions.

Figure 1

The temporal evolution of real estate financial risk levels in four regions.

From a national perspective, China’s real estate financial risk levels rose rapidly during the 2008 financial crisis. Subsequently, under the stimulus of China’s “4 trillion” fiscal policy, the real estate market developed rapidly, and the risk decreased. Between 2010 and 2018, the fluctuation range of real estate financial risks remained relatively stable, with a gradual upward trend. However, after 2019, under the combined influence of macroeconomic structural adjustments, cyclical changes in the industry, and the impact of the COVID-19 pandemic, China’s real estate financial risks became increasingly prominent. In 2022, a sharp upward trend was observed, highlighting that major shifts in the macroeconomic landscape had altered market expectations. As a result, real estate demand dropped significantly, leading to a sluggish market facing heightened risks. Given these new circumstances, where the balance of market supply and demand has shifted dramatically, it is imperative to stimulate housing demand and prevent further risk escalation.

From a regional perspective, compared with the national average, the Northeastern region exhibits the highest level of real estate financial risk, followed by the Western region, while the Central and Eastern regions show the lowest levels. This difference suggests that a stronger economic foundation, greater financial resource endowments, and higher population density in the Central and Eastern regions have mitigated market risk to some extent. In contrast, the Western and Northeastern regions face relatively underdeveloped economies and financial markets, which exacerbate the challenges their real estate markets encounter under the new circumstances. Additionally, the sharp population decline in these regions has further intensified the severity of their real estate market conditions.

4.2 Regional Space–time Differences of Real Estate Financial Risk

4.2.1 Regional Differences

As shown in Figure 2, the Theil index illustrates an upward trend with fluctuations after 2008 in terms of overall differences. However, during the financial crisis and after 2019, there has been a noticeable decrease in overall differences. This indicates that, on the one hand, the country’s “4 trillion” fiscal policy, the housing regulation policy of “housing is for living in, not for speculation,” and measures to prevent and mitigate asset bubble risks have shown significant effectiveness. On the other hand, the overall strategic effect of China’s “14th Five-Year Plan,” which emphasizes “improving the framework for financial risk prevention, early warning, resolution, and accountability systems, and maintaining the bottom line of preventing systemic risks,” has gradually emerged, enhancing the country’s overall capacity to prevent real estate financial risks.

Figure 2

China’s real estate financial risk level Theil index.

Table 4 shows that, according to the decomposition results, the contribution of intra-regional differences exceeded 50% prior to 2020, reaching a particularly high level of 89.99% in 2013. Overall, the contribution rate of intra-regional differences surpasses that of inter-regional differences, indicating that the overall difference in China’s real estate financial risks mainly stems from intra-regional differences. This can be attributed to significant heterogeneity in provincial policies and efforts to mitigate real estate financial risks, including specific initiatives implemented in certain regions to address risks associated with high-quality prime real estate enterprises. Therefore, the coordinated prevention and management of real estate financial risks within regions urgently require attention.

Table 4

Thiel index and contribution rate of real estate financial risk from 2006 to 2022

Time	Total difference	Inter-provincial differences	Intra-provincial differences
			Total	Eastern region	Central region	Western region	Northeast region
2006	0.0203	0.0083 (40.58)	0.0121 (59.42)	0.0120 (17.26)	0.0052 (4.57)	0.0124 (24.94)	0.0216 (12.66)
2007	0.0219	0.0051 (23.29)	0.0168 (76.77)	0.0157 (22.41)	0.0110 (8.62)	0.0234 (43.44)	0.0047 (2.32)
2008	0.0070	0.0020 (28.29)	0.0050 (71.71)	0.0025 (10.76)	0.0036 (10.25)	0.0087 (48.58)	0.0015 (2.17)
2009	0.0266	0.0033 (12.38)	0.0233 (87.58)	0.0422 (48.08)	0.0312 (23.08)	0.0104 (15.01)	0.0032 (1.43)
2010	0.0364	0.0070 (19.35)	0.0294 (80.68)	0.0400 (32.13)	0.0087 (4.35)	0.0367 (41.61)	0.0083 (2.57)
2011	0.0249	0.0036 (14.64)	0.0212 (85.36)	0.0319 (38.63)	0.0089 (6.83)	0.0241 (38.31)	0.0036 (1.61)
2012	0.0331	0.0047 (14.22)	0.0284 (85.78)	0.0522 (48.00)	0.0044 (2.42)	0.0284 (34.11)	0.0036 (1.25)
2013	0.0320	0.0032 (10.01)	0.0288 (89.99)	0.0579 (55.61)	0.0024 (1.39)	0.0223 (27.67)	0.0156 (5.33)
2014	0.0289	0.0038 (13.08)	0.0251 (86.92)	0.0527 (54.88)	0.0087 (5.75)	0.0190 (25.86)	0.0011 (0.44)
2015	0.0357	0.0060 (16.80)	0.0297 (83.20)	0.0670 (55.84)	0.0056 (2.87)	0.0217 (24.45)	0.0001 (0.04)
2016	0.0378	0.0117 (30.90)	0.0261 (69.10)	0.0640 (47.56)	0.0026 (1.25)	0.0180 (19.59)	0.0021 (0.70)
2017	0.0378	0.0062 (16.34)	0.0316 (83.68)	0.0744 (58.51)	0.0038 (1.87)	0.0216 (22.37)	0.0028 (0.93)
2018	0.0302	0.0072 (23.80)	0.0230 (76.20)	0.0504 (49.10)	0.0014 (0.85)	0.0198 (26.17)	0.0002 (0.06)
2019	0.0273	0.0112 (41.05)	0.0161 (58.99)	0.0306 (32.65)	0.0020 (1.24)	0.0147 (22.00)	0.0065 (3.11)
2020	0.0265	0.0121 (45.61)	0.0144 (54.39)	0.0171 (18.20)	0.0052 (3.47)	0.0185 (28.43)	0.0086 (4.28)
2021	0.0310	0.0166 (53.61)	0.0144 (46.39)	0.0184 (17.31)	0.0053 (2.67)	0.0185 (24.76)	0.0038 (1.65)
2022	0.0224	0.0122 (54.65)	0.0101 (45.35)	0.0074 (9.65)	0.0035 (2.65)	0.0173 (31.05)	0.0033 (1.98)

Note: The numbers in parentheses represent the contribution rates of inter-regional and intra-regional differences to the overall difference, expressed as percentages (%).

In addition, since 2018, the contribution of inter-regional differences has exhibited an overall upward trend, rising to 54.65% in 2022. This suggests that inter-regional differences in real estate financial risks are becoming increasingly prominent. Further decomposition of intra-regional differences reveals that the average Theil index values for the eastern, central, western, and northeastern regions from 2006 to 2022 are 0.037, 0.007, 0.020, and 0.005, respectively, with corresponding average contribution rates of 36.27, 4.95, 29.31, and 2.5%. These findings indicate that the eastern region of China exhibits the highest level of difference and contribution rate, followed by the western region, whereas the central and northeastern regions show relatively smaller differences. The differences arise from the varying real estate market conditions, land issues, and population challenges faced by each province. The primary reason for the significant differences between the eastern and western regions lies in the uneven development of the real estate market within the provinces of these regions. In the eastern region, stratification and polarization are particularly evident, as the real estate markets in first-tier cities such as Guangzhou, Beijing, Shenzhen, and Shanghai contrast sharply with those in other areas. In the western region, differences are largely driven by the weaker positions of provinces such as Xinjiang, Qinghai, Inner Mongolia, and Guizhou.

4.2.2 Spatial Distribution

Figure 3 presents the global Moran’s index of real estate financial risk, which captures the spatial correlation across different regions. The graph shows that the global Moran’s index consistently remains above 0, exhibiting an overall trend of initial decline followed by an increase. Although the index reached a lower point in 2013, it still reflects a significant spatial agglomeration effect of China’s real estate financial risks. Since 2020, the index has surpassed 0.5, indicating a strong spatial correlation among regions.

Figure 3

Global Moran’s index from 2006 to 2022.

Table 5 illustrates the spatial distribution of the local Moran’s index for 2006 and the past 4 years in the sample. It is evident that China’s real estate financial risks exhibit a pronounced spatial correlation. In the HH-type region, Inner Mongolia, Jilin, Heilongjiang, Gansu, Qinghai, Ningxia, and Xinjiang are the “resident guests.” Liaoning and Yunnan have transitioned to this agglomeration type successively over the past 3 years. The remote locations, sparse populations, and lack of financial resources in these areas are critical factors contributing to their elevated real estate financial risks. It is urgent to develop a new model for real estate growth in these regions, seeking fresh impetus and innovative pathways for market development. In the LL-type region, Shanghai, Jiangsu, Zhejiang, Anhui, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, and Chongqing have consistently been part of this category, while Beijing, Tianjin, Fujian, and Shaanxi have joined over the past 3 years. This type is primarily located in the eastern and central regions of China. Superior geographical conditions, significant economic development, and abundant financial resources provide a solid foundation for mitigating real estate financial risks in these areas.

Table 5

Spatial distribution of local Moran’s index

Year	HH	LH	LL	HL
2006	Shanxi, Inner Mongolia, Jilin, Heilongjiang, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Liaoning, Sichuan	Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Anhui, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Guangxi, Hainan, Chongqing, Yunnan	Fujian, Guizhou
2019	Inner Mongolia, Jilin, Liaoning, Heilongjiang, Guangxi, Hainan, Yunnan, Gansu, Qinghai, Ningxia, Xinjiang	Tianjin, Hebei, Sichuan, Shaanxi	Shanxi, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Chongqing	Beijing, Guizhou
2020	Inner Mongolia, Jilin, Liaoning, Heilongjiang, Gansu, Qinghai, Ningxia, Xinjiang	Tianjin, Hebei, Sichuan, Shaanxi	Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Guangxi, Chongqing, Guizhou	Beijing, Shanxi, Hainan, Yunnan
2021	Tianjin, Inner Mongolia, Jilin, Liaoning, Heilongjiang, Yunnan, Gansu, Qinghai, Ningxia, Xinjiang	Hebei, Sichuan, Shaanxi	Shanxi, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Chongqing	Beijing, Guangxi, Hainan, Guizhou
2022	Inner Mongolia, Jilin, Liaoning, Heilongjiang, Guangxi, Hainan, Yunnan, Gansu, Qinghai, Ningxia, Xinjiang	Hebei, Sichuan	Beijing, Tianjin, Shanxi, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Chongqing, Shaanxi	Guizhou

Sichuan is one of the few provinces classified in the LH-type region. Apart from Chongqing municipality, Sichuan, as the only western region with a mega-city population, enjoys strong population-supported housing market demand, and its real estate market development is relatively stable compared to other western regions. Guizhou has frequently appeared in the HL-type region. Compared to neighboring areas such as Guangxi, Hunan, and Chongqing, Guizhou’s more remote geographical location, relatively underdeveloped economy, and scarcer financial resources are key factors contributing to its more pronounced real estate financial risks.

4.3 Dynamic Evolution Law of Real Estate Financial Risk

4.3.1 Dynamic Distribution Evolution Characteristics

Figure 4 illustrates the distribution dynamics of real estate financial risk levels. The analysis reveals the following key points:

Figure 4

The kernel density estimation results.

First, the distribution of real estate financial risk exhibits a single-peak evolution pattern with a pronounced right-tail trend. Over time, this right-tail phenomenon has intensified, indicating the presence of certain areas with high real estate financial risk and highlighting an evident “polarization” effect. Second, from 2010 to 2022, the peak of the kernel density curve gradually shifts to the right. Concurrently, the peak of the kernel density map broadens, suggesting that China’s regional real estate financial risks are on an upward trajectory and that the gap between regions is progressively widening. Third, the left tail of the kernel density map gradually shortens, forming a striking contrast with the pronounced right tail phenomenon. This indicates that, over time, the phenomenon of LL agglomeration in China’s regional real estate financial risks still persists but is gradually weakening, with a growing tendency toward HH agglomeration.

In summary, China’s regional real estate financial risks exhibit a dynamic evolutionary trend characterized by polarization, a rising overall risk level, widening inter-regional disparities, weakening LL agglomeration, and strengthening HH agglomeration.

5.1 Conclusions

Analyzing real estate financial risk from the perspective of its root causes is crucial. This study employs panel data from 30 provinces in China to construct a real estate financial risk evaluation framework based on internal risk factors, external risk factors, and interconnected risk factors. The study examines the regional space–time differences and dynamic evolution laws of real estate financial risks in China. The research finds that: (1) Interconnected risk factors are key contributors to real estate financial risk. (2) Significant regional differences in real estate financial risks exist across China. The western and northeastern regions are hotspots for risk, while the eastern region exhibits the most pronounced market stratification and polarization. (3) There is significant spatial autocorrelation in China’s real estate financial risks, with most provinces showing HH and LL clustering. HH clusters are primarily located in the western and northeastern regions, while LL clusters are more prevalent in the central and eastern regions. (4) The distribution of real estate financial risks follows a single-peak evolutionary pattern, characterized by the dynamic transition of “weakening LL clusters and strengthening HH clusters. (5) The dynamic evolution of China’s real estate financial risks exhibits strong “spatial stickiness” and “positive reinforcement.” Over time, the probability of regions transitioning to higher-risk types increases, demonstrating a trend toward escalating risk levels.

This study provides a comprehensive analysis of the regional space–time differences and dynamic evolution of real estate financial risks in China. However, there are still some limitations. First, this study is limited by its inability to conduct an in-depth analysis and discussion of the driving factors behind the evolutionary characteristics of China’s real estate financial risk. Second, the lack of counterfactual and shock analyses constrains the study’s ability to assess the robustness of its findings under alternative scenarios or external economic shocks. These limitations arise from methodological constraints and challenges in data acquisition, which go beyond the scope of the current research. Future studies should focus on exploring these complex driving factors and employing advanced statistical methods for counterfactual and shock verification. Despite these constraints, the findings provide valuable insights into regional disparities and risk dynamics, serving as a solid foundation for future research.

5.2 Policy Implications

This study offers both theoretical and practical insights for scholars and policymakers. For scholars, it provides a clearer understanding of the evolution laws of real estate financial risks in China. In this study, we propose a real estate financial risk evaluation framework comprising three perspectives: internal risk factors, external risk factors, and interconnected risk factors. This framework facilitates a comprehensive examination of real estate financial risks from the perspective of their root causes. For policymakers, the research highlights regional differences in real estate financial risks, which are critical for addressing regional market imbalances and achieving a balanced national strategy. Furthermore, analyzing the spatial distribution of real estate financial risks enables policymakers to evaluate the effectiveness of existing policies in mitigating risks in specific regions and to identify areas where policy adjustments are required. This study provides valuable theoretical insights for formulating targeted measures to resolve real estate financial risks across different regions.

To mitigate real estate financial risks effectively, policymakers should adopt targeted and precise strategies. First, based on our findings, interconnected risk factors are identified as key contributors to real estate financial risks. Therefore, we recommend that the Chinese government focus on the indicator factors within the dimensions of financial institutions and local governments. For instance, encouraging real estate developers to diversify their financing channels could help prevent risks from becoming overly concentrated within the banking system due to reliance on a single funding source. Second, China’s real estate financial risks exhibit significant spatial differences, with high risks primarily concentrated in the western and northeastern regions. These areas also demonstrate strong “spatial stickiness.” To address this, we suggest that the Chinese government focus on these high-risk clusters to prevent further deterioration and cross-regional contagion. For example, targeted risk prevention and mitigation initiatives should be implemented in the western and northeastern regions, including identifying specific risk points, categorizing risk issues, and applying tailored mitigation measures to enable early detection and resolution. These actions would reduce the “spatial stickiness” of high-risk areas and curb the spread and clustering of risks across regions. Third, the spatial correlation analysis indicates significant positive spatial dependence of real estate financial risks across regions in China. We recommend that the Chinese government leverage this characteristic by implementing inter-regional collaborative risk prevention policies. Strengthening coordinated efforts among regions could amplify the synergy of risk mitigation strategies and enhance overall policy effectiveness.