Spatial Characteristics of the Abandonment Degree of Residential Quarters Based on Data of the Housing Sales Ratio—A Case Study of Kunming, China

Abstract

The abandonment degree of an urban residential building can reflect the popularity of residential areas. This study uses this idea as a basis for proposing the concept of using residential quarters’ abandonment rate to measure the abandonment degree of an area. The spatial pattern of the abandonment rate and its clustering characteristics were obtained by taking 2517 residential quarters in Kunming’s Main Urban District as research object, and using their listing for sale ratio data. Thereafter, curve estimation was used to explore the influencing factors of abandonment rate. The results are as follows. (1) The abandonment rate of the four circles in Kunming’s Main Urban District increases from inside to outside, showing the pattern of the core area–second ring area–third ring area–new urban district, with evident “core–edge” characteristics. (2) The relationship between distance from the city center, housing ages, and abandonment rate can be well fitted using a quadratic function and shows an inverted “U”-shaped “rising–declining” trend. The relationship between housing prices and abandonment rate is fitted by the inverse function, showing an evident “up” trend. This study is a reference for managers of relevant departments and urban planners in formulating reasonable urban housing development policies.

1. Introduction

The abandonment degree of urban residential buildings (i.e., which areas are undesirable) is a popular topic in urban geography. The abandonment degree of residential quarters represents the willingness of residents to move out of the housing, which is the opposite direction of residential choice (i.e., residential migration). Residents’ dissatisfaction with their current residence and their desire to improve it create a “push–pull” effect of residential mobility [1] and promote the phenomena of suburbanization [2] and gentrification [3]. In addition, housing itself is an asset, and, for housing investors, the sale of housing assets (most likely without owner-occupation) also determines the housing abandonment degree. Therefore, studying intra-urban abandonment degrees is an important reference for understanding the trends of residential mobility and transfer of housing assets.

The urban spatial pattern of the abandonment degree of housing varies in different countries. For example, inner cities are often less popular in the US. One study has found that, from 1990 to 2010, an increase in suburban density became a common feature of all counties and cities in Wisconsin [4], which is a reflection of the phenomenon that suburbs are more popular than inner cities. In the two major cities of the Netherlands, namely Amsterdam and Rotterdam, most middle-class families choose to live in the suburbs, while only a minority of families choose to live in the city center [5,6]. By contrast, inner cities in China have not undergone the processes of decline and migration and are still the preferred living space for residents. Zhang et al. used a large sample of Hangzhou to analyze spatial differences in housing; their results showed that the affluent and emerging middle class prefer to live in the traditional central district because of its excellent public facilities and convenient transportation [7]. Cui et al. used questionnaires and found that Shanghai residents’ migration location remains mostly concentrated in the central city before and after the local population’s migration [8]. Wang et al. showed the spatial pattern of the residential location of Guangzhou’s highly educated population from the perspective of the residential location choice, thereby indicating the characteristics of mainly core areas [9]. Li and Mao’s study on Guangzhou also showed that households with higher education levels, higher socio-economic status, or working in government departments and public institutions are more likely to settle in core areas [10]. Qiang and Hu’s study on population mobility in metropolitan Beijing found that both the native population and migrant population tend to prefer core areas and functional areas in terms of residence choice [11]. Zheng et al. studied households’ location choices in five major cities in China and found that high-income households tend to live in the more expensive central district [12]. In addition, locales such as the Netherlands, Dublin in Ireland, Ghana, Gyeongsangnam-do in South Korea, and Mexico show characteristics of housing abandonment. Beckers and Boschman found that a group of highly skilled foreign workers in the Netherlands tends to settle in inner-city neighborhoods with high income and urban atmosphere [13]. In Dublin, older workers mainly choose to live in suburban areas with convenient traffic links to the city center or their workplaces [14]. Acheampong found through a questionnaire that as places in Ghana continue to urbanize and incomes increase, more households tend to realize their housing needs in the suburbs [15]. In Gyeongsangnam-do, housing abandonment occurs mainly in less developed areas, such as at the border between two administrative districts [16]. Monkkonen found, through the collection of demographic and census data from 100 large cities in Mexico, that the phenomenon of high vacancy rates in Mexican housing is strong in both central and peripheral urban areas [17]. In summary, spatial characteristics of the abandonment degree of residential quarters are studied mainly based on residential mobility data and the proposed willingness to move. However, these data are costly to obtain and do not necessarily achieve full coverage of large samples. Therefore, a markedly clear and explicit estimation method that is easily accessible to large samples should be explored immediately to analyze the spatial pattern of the abandonment degree in an urban area.

This study adopts a new research perspective: to study the pattern of the abandonment degrees of urban residential buildings based on the spatial pattern of the proportion of residential quarters listed for sale. The proportion of urban residential quarters listed for sale can be used as an important indicator to characterize the abandonment degrees of residential quarters. Although real estate transactions (property sales) cannot be a reliable, or the only, indication of a process, it can be an important and viable way to assess the abandonment degree of the residential quarter. The reason for this is that the listing of housing sales is reflects that owners do not want to own property in the area. That is, for various reasons owners intend to abandon their houses, do not want to live in them, or do not want to hold the asset. The higher the percentage of listed housing sales in secondary residential quarters, the higher the degree of homeownership for sale. In addition, the higher the abandonment rate of the residential quarters, the less popular it is with existing residents.

Accordingly, the following question must be answered: What types of residential quarters are more widely abandoned? In general, neighborhoods with older buildings and dilapidated houses are prone to abandonment. This finding has been demonstrated by Jeong-II Park [18]. Residential quarters with smaller housing areas are not favored by residents [19,20]. Super high-rise and high-rise residential buildings are likely to be abandoned in the future because of the high floor area ratio, which affects living experience and safety [21]. Residential quarters with poor location conditions are also prone to abandonment [22]—the possibility of abandonment increases if a residential quarter is distant from nearby facilities (e.g., schools, convenience stores, hospitals) and has poor transportation [16]. Residential quarters with a high elderly population have a higher probability of abandonment [22,23,24]. Residential quarters with poor neighborhood environmental characteristics (i.e., poor resources and services, disharmonious emotional relationships between neighbors, and even the presence of negative factors such as high crime rates [25]) are prone to abandonment [26], leading to a lack of attachment to the community and prompting residents to move out. In terms of housing prices, houses with high prices are prone to be abandoned because they are easily profitable as assets, while houses with low prices are also prone to be abandoned [27]. The public transportation provision can influence the residential choice behavior of residents and indirectly affect the degree of housing abandonment [28]. Yin and Silverman studied the City of Buffalo (US) and Kanayama and Sadayuki studied Toshima City (Tokyo, Japan) and demonstrated that the likelihood of abandonment is influenced by vacant or abandoned properties in surrounding neighborhoods [27,29]. Additionally, another study by Yin and Silverman of the City of Buffalo in the United States shows that abandoned properties have a strong relationship with race, as the concentration of abandoned properties in the city has been limited to predominantly African-American neighborhoods on the east side [30]. The types of residential areas with high levels of abandonment have been shown to vary across countries and types of cities. Therefore, additional research cases are needed to clearly address the preceding issues. Using Kunming, China as a case city, this study analyzes the spatial pattern of the residential quarters’ abandonment rates, and further investigates the types of residential quarters markedly prone to be abandoned. In other words, the current research focuses on the correlation between residential quarter abandonment rates and the housing year, location conditions, and housing prices. This study aims to answer the following questions: Is a residential quarter more likely to be abandoned the older it is? Is a residential quarter more likely to be abandoned the poorer its location is? What are the effects of housing prices on the abandonment rate of a residential quarter?

Kunming is largely a real estate development-driven city. If we define a city’s economic dependence on real estate by the indicator of the proportion of real estate investment to GDP, the real estate dependence of the urban area of Kunming is as high as 35.15% in 2019, ranking it second among China’s provincial capitals (just below Nanning). From 2018 to 2022, the total floor area of new residential housing in Kunming’s Main Urban District is as high as 59.34 million m², accounting for 30.91% of the total floor area of all current residential quarters in Kunming. In theory, in a city with such a large number of new residential quarters, housing is more likely to be oversupplied than overdemanded, and the residential quarters will be more likely to be abandoned. Therefore, Kunming is suitable as a case city.

The remainder of this paper is organized as follows. Section 2 describes the data and methods of this study, including the research design, data sources, new concepts proposed, and methods used. Section 3 presents the results of this study, focusing on the spatial pattern and characteristics of the abandonment rate in Kunming’s Main Urban District, and a brief analysis of the factors influencing it. Lastly, Section 4 concludes this research and, accordingly, discusses the characteristics and causes of the “abandonment rate.”

2. Data and Methods

2.1. Research Design

To investigate the housing abandonment degree in Kunming, we constructed a framework for this study the specific process of which is as follows. First, we propose the concept and calculation method of abandonment rate. Second, we analyze the spatial pattern and spatial autocorrelation of the abandonment degree of residential quarters, and identify the location of spatial clustering by taking 2517 residential quarters within Kunming’s Main Urban District as a sample. Third, we analyze the factors influencing the abandonment degree of residential quarters in Kunming’s Main Urban District from three perspectives: location, housing prices, and housing ages. Lastly, this study explores the relationship between the abandonment degree of residential quarters and the three influencing factors. The research framework is shown in Figure 1.

2.2. Study Area and Data Sources

Kunming is located in Southwest China, in the middle of the Yunnan–Guizhou Plateau, at east longitude 102°10′–103°40′ and north latitude 24°23′–26°33′. Kunming is one of the most important central cities in Western China and the gateway city to China’s opening up to the southwest. Moreover, Kunming is known as the “Spring City” owing to its warm climate all year, and it is located in Lake Dian, the sixth largest freshwater lake in China, with pleasant scenery. Hence, Kunming can be used as a typical case city in urban geography to study housing issues. Kunming’s Main Urban District, as a strong urban core area, is representative of its urban housing problems and therefore ideal for the study. The delineation of Kunming’s Main Urban District refers to the description of the main urban area in the Kunming City Master Plan (2011–2020), according to the administrative boundaries of subdistricts or neighborhoods as criteria for delineating the boundaries of the study area and referring to the previous delineation by Wang et al. [31]. The area is divided into four circles from inner to outer in the following order: core area, second ring area, third ring area, and new urban district (Figure 2).

The 2517 residential quarter data in Kunming’s Main Urban District were used as the basic research unit. We referred to the Baidu map to draw the boundaries of the residential quarter polygon data; the period was November 2021. Attribute data, such as the total number of housing units, housing sales data, listed sales price, and housing year of the residential quarter, were obtained from the Anjuke website (https://km.anjuke.com/, accessed on 18 November 2021). Some data with missed attribute are supplemented by real estate websites, such as Fangtianxia (https://km.fang.com/, accessed on 9 February 2022) and Beike (https://km.ke.com, accessed on 9 February 2022), and were obtained in 2022. Thereafter, the residential quarters with a missed attribute for total households were removed, and 2517 residential quarters were retained. They were specifically distributed as follows: core area, 505; second ring area, 664; third ring area, 719; and new urban district, 629.

In this study, the intersection of Qingnian Road and Renminzhong Road in the core area was defined as the city center, and the straight-line distance from each residential quarter to this intersection was calculated to obtain the distance from the city center as an influencing factor. Housing price data were used to the listed sales price in 10,000 yuan/m². Housing age was referred to as 2021 minus the year when the housing was built.

2.3. Concept and Methodology

2.3.1. New Concept in Evaluating Abandonment Degree: Abandonment Rate

We are confronted with the question of how the abandonment degree of the residential quarter, which is a reflection of the willingness of residents to move and the properties of their housing assets, can be measured. This undertaking requires a quantitative evaluation of the indicator. Accordingly, we propose the new concept of abandonment rate to quantify the abandonment degree of residential quarters in the inner city. The abandonment rate of a residential quarter can be defined as the ratio of the number of houses listed for sale to the total number of houses in the neighborhood. On the basis of this definition, if residential quarter P has a total of S houses, and N of them are listed for sale, then the abandonment rate in residential quarter P is calculated as follows:

ARRQ _P = N/S_P × 100%

(1)

2.3.2. Spatial Autocorrelation Analysis

1.

Global Moran’s I

Global Moran’s I is used to measure the global spatial characteristics of the abandonment degree of residential quarters in Kunming’s Main Urban District with the following expressions [32,33]:

$$I=\frac{{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{W}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{S^2{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{W}_{ij}}$$

(2)

$$S^2=\frac{{{\displaystyle \sum }}_{i=1}^n{\left(x_i-\overline{x}\right)}^2}{n}$$

(3)

where I denotes the global autocorrelation index, which is used to measure the degree of spatial autocorrelation; x_i and x_j represent the abandonment rate of the ith and jth residential quarters, respectively; and w_ij is the weight matrix of each residential quarter. We specifically use the fixed distance method to create the weight matrix. Distance threshold is set to 1 km. In other words, the neighboring elements of residential quarters within 1 km are equipped with a weight value of 1 and have considerable influence on the residential quarter, while the elements of the residential quarters beyond this distance have a weight value of 0 and do not have any influence on the calculation of the quarter. The global spatial autocorrelation index I takes the range of [−1, 1]: when I is negative, there is a negative correlation between the abandoned rate of residential quarters; when I is positive, the abandonment rate of residential quarters has a positive correlation distribution. The closer the absolute value of I to 1, the greater the spatial autocorrelation. When I is 0, there is no correlation between the abandonment rates of residential quarters, and they are randomly distributed. The Z-value is a standardized statistic of I, which is used to determine the degree of clustering of the abandonment rate of residential quarters. The Z-value can be expressed as follows:

$$Z\left(I\right)=\frac{I-E\left(I\right)}{\sqrt{Var\left(I\right)}}$$

(4)

where Var(I) denotes the number of variances and E(I) is the mathematical expectation of the abandonment rate of a residential quarter. When the absolute value of Z is larger, it indicates that the spatial positive (negative) correlation of the abandonment degrees of residential quarters is more significant. The absolute value of Z tends to 0, which means that the result is not significant and the abandonment rate of the residential quarter is randomly distributed.

2.

Getis-Ord Gi*

Getis and Ord proposed the Gi* index as a method of local spatial autocorrelation analysis [34,35], which is introduced here to measure the local spatial correlation characteristics of the abandonment rate of residential quarters in Kunming’s Main Urban District. The Getis-Ord Gi* index can be used to accurately identify the spatial distribution of cold and hot spots of the abandonment rate of residential quarters in the region. Moreover, this index is commonly used for the analysis of the spatial pattern characteristics of the study objects in the region. The calculation formula is as follows:

$$G{i}^{*}\left(d\right)=\frac{{{\displaystyle \sum }}_{i=1}^n{W}_{ij}\left(d\right)X_j}{{{\displaystyle \sum }}_{i=1}^n{X}_i}$$

(5)

where d is the distance between residential quarters i and j and W_ij(d) is the distance weight between i and j. Z(Gi*) can also be obtained by normalizing the Z-values for Gi* with the following equation:

$$Z\left(G{i}^{*}\right)=\frac{G{i}^{*}-E\left(G{i}^{*}\right)}{\sqrt{Var\left(G{i}^{*}\right)}}$$

(6)

If Z(Gi*) is positive under the condition of significant p-value (p < 0.1), at confidence levels of 99%, 95%, and 90%, then the abandonment degrees of residential quarters in the region are distinguished into the following order: hotspot, sub-hotspot, and marginal hotspot, respectively. That is, residential quarter i has a high abandonment degree and the surrounding residential quarters have a high abandonment degree as well. Conversely, the order is coldspot, sub-hotspot, and marginal coldspot.

2.3.3. Curve Estimation

The fitting curve is a scatter plot of the dependent variable y and independent variable x. Thereafter, the curve is generated to reflect the relationship between the two variables based on the existing function model. Fitting analysis does not require the curve to pass through all discrete points but instead requires only that y = f (x) reflect the general trend of the discrete points without local fluctuations. In this study, abandonment rate of residential quarters was used as the dependent variable, while the distance from the city center, housing prices, and housing ages were used as independent variables. Meanwhile, five models, namely linear, logarithmic, quadratic, inverse, and power functions, were constructed to fit. On the bases of the model fitting results (i.e., R², F-value, and p-value) using SPSS19.0, we filtered the best-fit curves for each of the three influencing factors. Given that the data of the selected sample points were large and not easy to fit, we used the quantile method in ArcGIS to classify the three influencing factors into 15 categories and calculated the mean value of each category and abandoned rate. Each independent variable obtained 15 points for fitting.

3. Results and Analysis

3.1. Characteristics of Abandonment Degree in Kunming’s Main Urban District

On the bases of the concept and calculation method of abandonment rate, we obtained the abandonment rate of residential quarters in Kunming’s Main Urban District. Accordingly, we first calculated the abandonment rate of each circle, and thereafter analyzed the abandonment degree pattern of the residential quarter in Kunming’s Main Urban District, as shown in

Table 1. Descriptive analysis of the abandonment rate in four circles in Kunming’s Main Urban District.

	Core Area	Second Ring Area	Third Ring Area	New Urban District
Houses for sale (sets)	1023	4872	8925	12653
Total number of houses (sets)	134,069	414,462	611,336	736,022
abandonment rate (%)	0.76%	1.18%	1.46%	1.72%

. The core area in the inner circle has the lowest abandonment rate, 0.76%, while the new urban district in the peripheral area has the highest abandonment rate, 1.72%. Note that the abandonment rate of the four circles of Kunming’s Main Urban District increases from inside to outside, showing the core area–second ring area–third ring area–new urban district pattern, with the evident feature of “core-edge”.

Figure 3 shows the boxplot of the abandoned rate distribution of the residential quarter in each circle, thereby enabling us to specifically understand the abandonment rate of each circle district in Kunming’s Main Urban District. To improve the display, we removed the data from three residential quarters with abandoned rates above 30%. Figure 3 further indicates that the abandonment rate of the four circles shows a sequential increasing trend, which is consistent with the preceding results.

3.2. Spatial Pattern and Spatial Association Characteristics of Abandoned Rate

We used ArcGIS to visualize spatial pattern (Figure 4). First, we used the manual method to set the threshold values of 0%, 1%, 2%, 5%, and 10%, and the abandonment rate of residential quarters was divided into six levels from low to high, as shown in Figure 4. Figure 4 illustrates that the abandonment rate in Kunming’s Main Urban District generally shows an evident pattern of increasing from inside to outside. Moreover, the abandonment degree of the new urban district is higher than that of the inner circle area. Areas with higher abandonment rates are mainly located in two major areas, namely, Dianchi Road Area in the southwest and Jinchen–Hongyun–Expo Area in the northeast. Both areas are located in the new urban district in the outer circle.

To further investigate the spatial characteristics of the abandonment rate in Kunming’s Main Urban District, we introduced spatial autocorrelation analysis to measure its spatial correlation characteristics globally and locally. First, we calculated Moran’s I in ArcGIS10.7, and the result was Moran’s I = 0.055416 with a p-value of 0.00, Z-score was 15.7658, and the Z-value substantially exceeded the critical value of 1.96 at the 5% confidence level. This result indicates a significant global spatial autocorrelation in the abandonment degree of the residential quarter. Furthermore, residential quarters with the same characteristics tend to cluster in Kunming’s Main Urban District.

To measure the local spatial clustering characteristics of the abandonment degree of residential quarters and to accurately identify the locations of the hot and cold spots of abandonment degree, we introduced the Getis–Ord Gi* index with the distance threshold set at 1000 m. As shown in Figure 5, at the 90% confidence level, the cold spots of the abandonment degree of a residential quarter are mainly clustered in the core area and its outer edges, as well as in the southwest part of the second ring area (i.e., the inner circle area of Kunming’s Main Urban District), showing a distribution pattern contiguous with evident clustering characteristics. Meanwhile, hot spots of abandonment are mainly distributed in the new urban district of the outer circle, primarily in the southwest direction adjacent to the Dianchi area, such as Dianchi Resort Area, Fubao Area, Dashanghui, south of Hailun Guoji, northwest of High-tech Zone and Huangtupo, and northeast of Beichen Wealth Center and Jiangdong Garden, North Downtown, Golden Temple Area and the Expo Area in the northeast. The result fully demonstrates that the abandonment degree of Kunming’s Main Urban District shows an increasing pattern from the core area–second ring area−third ring area–new urban district. This result is consistent with the previous spatial pattern result display.

3.3. Influencing Factors of Abandonment Degree

From the curve fitting results (

Table 2. Comparison of the relationship between the three influencing factors and the abandonment rate of the residential quarter.

	Models	Model Summary			Parameter Estimates
		R2	F	p	Constant	b1	b2
Distance	Linear function	0.443	10.323	0.007	0.931	0.84
	Logarithmic function	0.670	26.387	0.000	0.768	0.421
	Quadratic function	0.848	33.474	0.000	0.334	0.370	−0.024
	Inverse function	0.586	18.383	0.001	1.632	−0.867
	Power function	0.733	35.622	0.000	0.779	0.368
Housing prices	Linear function	0.326	6.289	0.026	0.773	0.474
	Logarithmic function	0.503	13.142	0.003	1.137	0.995
	Quadratic function	0.679	12.706	0.001	−1.389	3.064	−0.677
	Inverse function	0.685	28.331	0.000	2.722	−1.675
	Power function	0.403	8.761	0.011	0.910	1.152
Housing ages	Linear function	0.451	10.698	0.006	2.045	−0.045
	Logarithmic function	0.195	3.144	0.100	2.207	−0.354
	Quadratic function	0.736	16.719	0.000	1.152	0.101	−0.005
	Inverse function	0.031	0.410	0.533	1.150	1.039
	Power function	0.233	3.948	0.068	2.898	3.348

), the fitting relationship between distance from the city center and abandonment rate has statistical characteristics, and the quadratic function fits best, with the highest goodness of fit R² reaching 0.848, F value of 33.474, and significance p-value of 0.00, all three of which meet the statistical requirements. Evidently, the indicator of distance from the city center is closely related to the abandonment rate. The fitted curve shows an inverted “U”-shape (i.e., a “rising–declining” trend), and the sample observations are evenly distributed on both sides of the curve. Figure 6 shows that the relationship between the abandonment degree and distance from the city center point is divided into two main phases, with the turning point occurring at about 7.71 km. In the first phase, abandonment rate has a significant upward trend with increasing distance, which is consistent with the theoretical perception that abandonment rate is significantly positively correlated with distance, and the abandonment degree of a residential quarter becomes higher with increasing distance. However, in the second stage, this positive relationship takes a turn, and the relationship between abandonment rate and distance turns to a downward trend (i.e., abandonment degree decreases with increasing distance).

The relationship between housing prices and abandonment rate can be fitted by the inverse function (Figure 7), which results in R² of 0.685, F-value of 28.331, and p-value of 0.00, at a 0.01 significant level, with strong statistical characteristics, which can prove that housing prices and abandonment rate are closely related. The curve shows a clear upward trend (i.e., an increase in housing prices leads to an increase in abandonment rate in residential quarters), but the rate of increase is different and has a gradual slowing trend. When housing prices are low, the rise in housing prices has a more significant effect on the abandonment rate. However, as housing prices become higher, the extent of this effect decreases, but still maintains a simultaneous upward trend.

After comparing the results of the models of housing ages and abandonment rate, note that the quadratic curve fits better, with R² of 0.736, F-value of 16.719, and p-value of 0.00, thereby meeting the statistical requirements (Figure 8). The curve is similar to the locational characteristics, with an inverted “U”-shape, in which abandonment rate is the first to show an increasing trend with housing ages. After reaching a peak (at about 10.1 years), abandonment rate turns around and shows a decreasing trend. Therefore, it is not the case that the older the residential quarter, the more likely it is to be abandoned, as in the case of some houses located in the old area, which, although relatively old, still have a significantly high occupancy rate.

4. Discussion

This study proposes the concept of residential quarters’ abandonment rate to evaluate the abandonment degree of residential quarters as a reflection of the popularity of urban spaces. In previous studies, data for residential mobility have mainly come from sampling, questionnaires, and face-to-face interviews, among others. Although this method can obtain high-precision data, there are still some limitations, such as high data collection costs, limited numbers of interviewees, and an inability to obtain large sample data, which seriously restrict the scope of the study, thereby affecting the research results. Compared with these traditional methods, we use housing sales data from real estate websites to study the abandonment degree of residential quarters, which is an easy and fast way to obtain data, with a large sample coverage but also markedly objective. Moreover, information from the housing market can be used to reflect and even predict trends in residential disparities over time, thereby effectively compensating for the time lag inherent in social statistics [36].

By fitting the curves of distance and abandonment rate, we found that abandonment in Kunming’s Main Urban District showed an inverted U-shaped trend of “rising–declining”, which is different from those in existing studies and is a new finding of the current research. Previous local and international studies have shown two opposite spatial patterns. Some studies have found that the abandonment rate of residential quarters tends to increase with distance (i.e., houses in central urban districts are more popular, while houses in suburban areas tend to be more abandoned, which is mostly found in some Chinese cities, such as Hangzhou and Guangzhou). Zhang et al. and Wang et al. demonstrated that people tend to choose to live in the urban district [7,9]. Another study indicated that houses in suburban areas are more popular compared with those in core areas, and this urban pattern is represented by foreign cities, such as Amsterdam and Rotterdam in the Netherlands [5,6] and Dublin in Ireland [14]. However, regardless of the relationship presented between abandonment degree and distance, it is always consistent with local conditions and urban development strategy.

Housing prices are one of the key factors influencing residential mobility decisions [37]. Furthermore, housing prices affect residents’ willingness to abandon and retain housing property. When housing value is high, homeowners choose to sell their property for profit acquisition. Conversely, when housing value is low and unpopular in the market, it remains vulnerable to abandonment. This study elaborates on the relationship with an inverse function curve, which reflects a simultaneous upward trend between the two, similar to the findings of Vakili-Zad and Hoekstra [38].

Housing ages, as the most typical factor in housing physical environment characteristics [18], affect the functionality and living experience of houses, thereby reducing their economic value. Therefore, the general belief is that the older the residential quarter, the higher the abandonment rate. Park et al. and Morkel found that residential quarters with older housing stock are prone to abandonment [39,40]. However, their findings are inconsistent with previous research results. That is to say, we found an inverted U-shaped relationship between housing ages and abandonment rate by curve estimation (i.e., an increasing trend followed by a decreasing trend). First, abandonment rate increases when housing age increases, while abandonment rate decreases when housing age reaches about 10.1 years. The former phase is consistent with theoretical perceptions and previous studies. For the latter phase, the abandonment rate is probably because the older residential quarters are mainly located in the inner circle of Kunming’s Main Urban District, mainly in the core and second ring areas, while the third ring and new urban areas are less widely distributed. This result is also consistent with the conclusion of the influence of location on the abandonment degree in this study.

Second, abandonment degree, spatial differences, and spatial characteristics of residential quarters are the result of a combination of factors. For example, some studies have claimed that abandoned neighboring properties cause residential quarters to be abandoned [27,29,41,42]. Yin and Silverman studied the City of Buffalo (US) and found that abandonment rate is negatively affected by surrounding vacant or abandoned properties [29]. It has also been claimed that residential quarters suffer from inadequate conditions such as a lack of infrastructure and basic services, which in turn lead to abandonment [43]. In addition, existing studies have found that the direction of residential mobility, either edge–center or center–edge migration, depends on people’s characteristics [44].

Although the abandonment degree of residential quarters is influenced by several factors, traditional locational factors still dominate [45]. The current study’s argument is consistent with those of previous findings that distance from the city center has a significant effect on residential abandonment, as demonstrated by Hillier et al. and Joo and Lee [16,22]. Whether moving out or moving in, purchasing or renting, location is consistently a prerequisite and plays a dominant role in residential mobility, as well as in location choice. Moreover, people of any class tend to choose areas with superior location conditions. On the one hand, location is a comprehensive reflection of a community’s infrastructure support (e.g., life security, public transportation, culture and education, health care, sports, and entertainment) and is a balance between various resources associated with a given location [38]. On the other hand, it is an important factor affecting the economic value of a house, and the best location can bring the greatest benefit to the property owner. These two aspects are precisely the key factors that influence the abandonment rate of residential quarters. Hence, location is a vital factor that cannot be disregarded when studying the abandonment degree of residential quarters.

Nevertheless, there are still some shortcomings in our study. First, beyond real estate transactions (property sales), the phenomenon of abandonment should be considered in a broader aspect. Second, we did not subdivide housing types (e.g., apartments, villas, ordinary houses), which may be a factor influencing the abandonment rate. For example, the abandonment rate of apartments may be higher than that of ordinary houses. Third, we only analyzed the pattern and characteristics of the abandonment rate of the residential quarters spatially, without considering the development trend over time. Fourth, we only considered the distance from the city center as a factor to characterize the relationship between location characteristics and housing abandonment degree. However, location is not only the distance from a city center, so this is not a comprehensive factor for location characteristics. Lastly, the causes of the abandonment degree of residential quarters are complex and multifaceted. Apart from the three factors in this study, additional factors should also be considered.

In a future study, we will explore the influencing factors from multiple perspectives based on abandonment rate, such as the physical environment characteristics of housing, neighborhood characteristics, families’ willingness to move, housing asset attributes, urban development strategies, and planning policies. We will construct an index system of multiple factors and use spatial regression models to explore the influencing factors and their spatial heterogeneity. We can also analyze the reasons for housing abandonment from different occupational and class characteristics and further reveal the formation mechanism. On the basis of the new concept of “abandonment rate”, we can conduct a comparative analysis of the spatial differences and spatial characteristics of housing abandonment in other cities to explore the similarities and differences. This aspect is proof of the universality of the method and also an effective reference and basis for the formulation of housing planning policies in the case cities.

5. Conclusions

This study examines the abandonment degree of residential quarter in Kunming’s Main Urban District. The main contribution of this study is to propose a new concept (i.e., abandoned rate) to measure abandonment degree and further reflect the popularity of urban residential quarters. The combination of the uniqueness of the Chinese housing market and progress of current research makes the current study significant. By using Kunming’s Main Urban District as the study area, the spatial pattern of the abandoned rate in Kunming’s Main Urban District is obtained using the data of the proportion of 2517 residential quarters listed for sale. Thereafter, global and local spatial characteristics are explored by conducting spatial autocorrelation analysis. Thereafter, the influencing factors of abandoned rate are analyzed by performing curve fitting analysis. On this basis, we obtain some findings that the abandoned rate of four circles in Kunming’s Main Urban District increases from inside to outside, showing the pattern of core area–second ring area–third ring area–new urban district, with evident “core–edge” characteristics. We found that the relationship between distance from the city center and housing age and abandoned rate can be considerably fitted using a quadratic function. Meanwhile, the relationship between housing prices and abandoned rate can be fitted using an inverse function. Moreover, the relationship between distance from the city center and housing age and abandoned rate shows an inverted “U”-shaped “rising–declining” trend. Meanwhile, the relationship between housing prices and abandoned rate shows a clear “up” trend.

This study can provide effective reference values for managers of relevant departments and urban planners to formulate reasonable urban housing development planning policies. Given the current situation of abandonment degree in Kunming’s Main Urban District, we propose the following policy recommendations: (1) reasonably balance education and medical resources and public transportation services in the inner and outer city; (2) reasonably optimize high-quality resources as an important means to address high abandonment degree in residential areas; and (3) advocate the smooth development of the real estate market through the improvement of public resources and balance of services.