Mitigation Strategy of Land Use Mix for Jobs-Housing Mismatch

Abstract

The jobs-housing mismatch phenomenon in urban China stems from the combined effects of housing commodification and the improvement of transportation infrastructure. These factors have contributed to the emergence of lengthy commutes and a range of urban challenges. This study examines the issue of jobs-housing mismatch in large cities, focusing on Hangzhou. It utilizes mobile signaling big data, geographically weighted regression, and spatial analysis to investigate the link between land mixed-use and this mismatch. The results reveal that Hangzhou faces a significant residential-employment mismatch, particularly in a ring-like pattern. Central urban areas are relatively balanced, while residential areas band around the center, and employment areas are scattered both centrally and on the outskirts. Land mixed-use impacts this mismatch spatially. In new developments, increased land use mix exacerbates the mismatch, while in ecological green spaces, it has a suppressive effect. Based on these findings, Hangzhou’s main urban area is divided into nine zones, each with tailored suggestions for balancing residential and employment spaces. This study demonstrates that mobile signaling data can precisely capture micro-level characteristics of residential and employment patterns. A multi-dimensional approach to land mixed-use offers a more comprehensive understanding than a single perspective. The zoning strategy helps establish spatial differences and balance residential-employment relations, providing valuable insights for urban renewal and land function optimization.

1. Introduction

One of the most salient characteristics of urban development in China is the phenomenon of a jobs-housing mismatch [1]. Since the reform and opening up, market-oriented reforms have adjusted land values, thereby promoting the differentiation of residential and employment space distribution [2]. The commodification of housing has dismantled the original housing allocation system, thereby influencing the distribution of the residential space through the operation of market mechanisms. The rapid increase in wealth and advancements in transportation methods have also made people more willing to bear greater commuting costs for better living conditions, which has ultimately resulted in the current jobs-housing mismatch situation [3]. The issue of a jobs-housing mismatch can give rise to a number of problematic consequences, including the necessity for long-distance commuting, traffic congestion, inefficient energy consumption and spatial pollution [4,5,6]. Consequently, the resolution of jobs-housing mismatch represents a significant area of interest for scholars.

The study of jobs-housing mismatch focuses on the phenomenon of spatial separation of the place of employment from the place of residence and its interaction with commuting costs and traffic congestion. Advances in transport technology and improvements in transport infrastructure have expanded people’s commuting range [7]. Efficient public transport systems enable people to commute over longer distances. Uneven distribution of land resources can also lead to jobs-housing mismatch. In the course of urban development, the distribution of residential and places of employment are overly concentrated, especially in China’s large cities, where people commonly live in high buildings, further increasing residential density. Measuring the jobs-housing mismatch is the first step in the study, so the quality of data is directly related to the validity and reliability of the measurement results of the mismatch. At present, scholars employ a range of data sources to assess jobs-housing mismatch, including questionnaire survey data, public transit card data, and mobile phone signal data [8,9,10]. The use of questionnaire data and public transit card data is limited by a number of factors, including the potential for strong subjectivity, high acquisition costs, and limited coverage. In contrast, mobile phone signal data is readily accessible and can be employed to portray the regional characteristics of the labour force and the residential population objectively and comprehensively [11].

Mixed land use is an organic combination of multiple land use types with different functions. Different scholars have different opinions on the definition of mixed land use. From a functional perspective, some scholars believe that mixed land use refers to the simultaneous inclusion of a variety of different land functions, such as residential, commercial, recreational, and entertainment, within a relatively compact geographical area [12]. While others emphasise that the land use in an area is such that residents’ daily travel destinations (e.g., workplaces, schools, supermarkets, etc.) are within walking, cycling or short-distance public transport reach [13]. At the same time, attention needs to be paid to compatibility, which is an extension of functionality and considers the relationship between functions [14]. The mixed land use has a profound effect on the distribution of residential and employment places [15,16]. However, the extent to which land use mix can alleviate jobs-housing mismatch remains a matter of debate among scholars. In 1989, Cervero first studied urban commuting behavior based on the concept of work-residence [17], inspired by this, the fields of economics, geography and planning have gradually focused on the study of urban commuting flows, leading to preliminary discussions on the effectiveness and methodology of work-life balance. Early scholars, in summarizing the urban growth experience, posited that land use mix accelerated urban sprawl at the macro level, promoted the extension of commuting distances, and thus intensified jobs-housing mismatch [18,19]. Nevertheless, studies conducted at the individual level or micro scale have demonstrated that living in neighbourhoods with a relatively high land use mix facilitates the balanced distribution of job opportunities and residential units, thereby reducing the probability of jobs-housing mismatch among the population [20,21].

The relationship between land use mix and jobs-housing mismatch has been a topic of debate among scholars. The reasons for this include, firstly, the issue of spatial scale, which is the primary factor contributing to the discrepancies or even diametrically opposed conclusions observed in research findings. The lack of sufficient coupling between micro-level urban resident population flow data and macro-level statistical data means that the latter is inadequate for supporting research into jobs-housing mismatch at the micro level. Consequently, micro-scale research is reliant upon subjective data, such as questionnaire surveys, which can readily result in distorted outcomes and conflict with macro-scale research conclusions. Secondly, the concept of land use mix is complex and multifaceted. However, many studies employ the diversity of land use types as the primary or even sole criterion for measuring land use mix [12,22]. The extent to which this characterization accurately reflects the actual state of land use mix remains a subject of debate. Furthermore, the relationship between land use mix and jobs-housing mismatch is characterized by spatial heterogeneity, and their underlying mechanisms are intricate. This makes it challenging to draw precise conclusions using conventional quantitative methods.

In conclusion, the lack of comprehensive data hinders the ability to ascertain the precise impact mechanism of land use mix on jobs-housing mismatch. The utilization of a big data system to quantify the extant situation of jobs-housing mismatch and land use mix, and to undertake a quantitative analysis of their relationship from the perspective of spatial heterogeneity, is of significant value in providing technical support for the alleviation of jobs-housing mismatch. Accordingly, this study employs a multi-source big data approach to quantify the micro-scale jobs-housing mismatch and land use mix in Hangzhou, China, subsequently analyzing their relationship. There are three possible contributions from this research: firstly, to assess the current state of jobs-housing mismatch and land use in Hangzhou; secondly, to elucidate the impact of land use mix on jobs-housing mismatch; and thirdly, to evaluate the relationship between land use mix and jobs-housing mismatch from a spatial heterogeneity perspective, and to propose zoning solutions to alleviate jobs-housing mismatch.

2. Materials and Methods

2.1. Research Area and Data Sources

2.1.1. Research Area

Hangzhou is one of the central cities in the Yangtze River Delta. As of 2023, the city has 13 districts. In order to facilitate the investigation of work-life segregation, eight districts with substantial populations were selected for analysis (Figure 1). According to data from the Hangzhou Statistical Yearbook 2023, the resident population and the number of employed people are shown in

Table 1. Population data statistics of Hangzhou city center.

District	Area (km2)	Resident Population (Ten Thousand People)	Number of Employed Persons (People)	Urbanization Rate (%)
Shangcheng	122	137.1	292,350	100
Gongshu	119	117.7	359,228	100
Xihu	263	116.7	391,714	97.4
Binjiang	73	53	421,649	100
Xiaoshan	931	211.0	525,511	81.1
Yuhang	940	136.4	322,991	74.1
Linping	282	110.8	251,658	89.1
Qiantang	338	79.7	301,593	88.6

. The permanent population of Hangzhou is concentrated in the four districts of Shangcheng, Gongshu, Binjiang, and Xihu, while the employment-population is concentrated in Binjiang and Gongshu. This indicates a significant issue of jobs-housing mismatch. The urbanization rate in areas such as Xihu and Xiaoshan District has reached almost 80%, yet the distribution of construction land suggests that their built environment still exhibits certain discrepancies compared to areas like Shangcheng. The distinctive distribution of the population and built-up areas in Hangzhou gives rise to a multi-center pattern, which presents a potential challenge to the development of jobs-housing mismatch. This study focuses on the central urban area of Hangzhou and analyses the characteristics of the population flow of this area, with a view to gaining a deeper understanding of the issues surrounding the spatial distribution of jobs and housing in Hangzhou. The findings of this study will provide a theoretical basis for alleviating jobs-housing mismatch.

2.1.2. Data Source

The mobile signaling data employed in this study is derived from anonymized data provided by Unicom’s Smart Footprint, including mobile signaling data pertaining to commuting within the urban area of Hangzhou in December 2020. This data set records commuting displacement start and end points, route duration, the number of people on the route, and travel modes, with a total of 26,591,398 records within the study area.

In consideration of the availability of the data in question, this article uses Point of Interest (POI) data to measure land use diversity. The POI data, sourced from Amap of Hangzhou in 2020, encompasses 533,754 entries, including dining establishments, commercial enterprises, retail outlets, and 13 additional subcategories. The data regarding the gross domestic product (GDP) is derived from the Chinese GDP spatial distribution kilometer grid dataset, while the benchmark land price is based on publicly accessible data from the Hangzhou Planning and Natural Resources Bureau.

2.2. Research Methods

2.2.1. Measurement of Jobs-Housing Mismatch

The initial step is to identify the population who commute. In order to minimize the influence of weather, public holidays and significant events on commuting patterns, it is recommended that one day with favorable weather conditions be selected from Monday to Friday, representing a total of five working days for the study. The jobs-housing index for each of the five selected dates should then be measured, with the result being the average of the five dates. This approach is designed to further mitigate the randomness caused by time issues. The selected dates are 7 December (Monday, 952,163 records), 10 December (Thursday, 879,544 records), 11 December (Friday, 867,538 records), 22 December (Tuesday,879,544 records) and December 23 (Wednesday, 823,698 records). The total number of records is 4,338,507, each record represents the commuting trajectory of a single individual on a given day, providing a comprehensive overview of the study area’s commuting patterns. Given that the population of Hangzhou in 2020 is estimated to be approximately 7.2 million, with a market share of Unicom of 19.13% (equivalent to 1.38 million individuals), and that this study focuses exclusively on the primary urban area of Hangzhou, the identified employment-population data is deemed to be relatively reliable. As the data set includes Origin-Destination commuting connections, the number of residents with work attributes is identified based on the employed population. This approach has the advantage of assuming that the residential population excludes the elderly and children from the employed resident group, thus better reflecting the essential characteristics of the commuting relationship under study.

Based on the above identification results, this article uses the jobs-housing mismatch index to characterize the degree of jobs-housing mismatch. The jobs-housing mismatch index was proposed by Cervero as a basic methodology to study the relationship between the number of residential units and employment units within a city [17]. The jobs-housing mismatch index refers to the ratio of the employed population to the resident population in a certain spatial unit divided by the ratio of the employed population to the resident population in the whole area, the calculation formula is as follows:

$$\begin{array}{c}{\mathrm{F}}_{\mathrm{j}}={\displaystyle \frac{{\mathrm{Y}}_{\mathrm{j}}/\mathrm{Y}}{{\mathrm{R}}_{\mathrm{j}}/\mathrm{R}}} \end{array}$$

(1)

In the formula, F_j is the jobs-housing mismatch of spatial unit j; Y_j is the number of employed population in spatial unit j; Y is the total number of employed population in the study area; R_j is the number of resident population in spatial unit j; R is the total number of resident population in the study area. A jobs-housing mismatch index equal to 1 indicates a balanced employment-residence situation, greater than 1 indicates a dominance of employment function, and less than 1 indicates a dominance of residential function.

2.2.2. Measurement of Land Use Mix

The degree of land use mix is mainly divided into three aspects: diversity, accessibility, and compatibility.

Based on the concept of a 15-min living area, a resident-friendly walking distance is usually 1000 m [23]. However, other studies suggest that the threshold of acceptable walking distance is around 1500 m [24,25]. Given that the benefits of less resident-friendly but still accessible land use types need to be taken into account when exploring mixed land use, a range threshold of 1500 m is preferred. In addition, mobile phone signalling data is generated by the base station receiving the mobile phone transmitting signals, so the base station signal coverage has a great impact on the accuracy of mobile phone signalling data. According to the Communication Industry Statistical Bulletin 2020 released by China’s Ministry of Industry and Information Technology, there are about 900,000 Unicom 4G base stations and 330,000 5G base stations in 2020, and considering that the signal coverage range of 4G base stations is about 1–3 km and that the coverage range of 5G signals is about 0.2–0.5 km, the base station signal penetration range is calculated to be 1.55 km according to the weighted number of base stations. Combining these two considerations, the grid size of the study is set to 1500 m.

(1)

Diversity

The study uses the Shannon-Wiener index to measure the diversity of mixed land use, which can simultaneously reflect the richness and evenness characteristics of land use practices, using the following formula:

$$\begin{array}{c}\mathrm{E}\mathrm{N}\mathrm{T}=-{\sum }_{\mathrm{j}=1}^{\mathrm{k}}{\mathrm{P}}_{\mathrm{j}}\ln \left({\mathrm{P}}_{\mathrm{j}}\right)\end{array}$$

(2)

In the formula, ENT is the Shannon-Wiener index, P_j represents the proportion of the j-th class of space; k represents land use type; the higher Shannon-Wiener index, the more diverse and evenly distributed the functional types are within the unit, indicating a greater diversity of land use.

(2)

Accessibility

Given that this approach fails to reflect the spatial heterogeneity of land use convenience, it is inadequate to rely on diversity indicators alone in order to assess the land use mix. Accordingly, this study incorporates accessibility into the index system for land use mix, utilizing the “opportunity accumulation model” to calculate the accessibility of mixed land use. The opportunity accumulation model was first proposed by Black et al. to study the ease with which residents in cities approach development opportunities [26]. The essence of this method is to ascertain the number of “opportunities” that can be accessed from a specific location, assuming a fixed travel cost of 1500 m. In this study, “opportunities” are defined as POI, with each POI representing a single “opportunity”. The formula is as follows:

$${\mathrm{A}}_{\mathrm{i}}=\sum _{\mathrm{j}}{\mathrm{O}}_{\mathrm{j}\mathrm{t}}$$

(3)

In the formula, A_i is the accessibility of area i, O_jt is the number of opportunities that node j can provide within threshold t (1500 m), the where j is the node whose cost to area i is less than threshold t.

(3)

Compatibility

Considering the externalities associated with land use, compatibility represents a pivotal aspect of mixed land use. To account for the interdependence between diverse land uses, the “Regional Land Use Compatibility Index” has been developed as a means of assessing the compatibility of mixed land use. Zhuo et al. first calculation of mixed land use compatibility and experimental validation in Guangzhou [27]. This index employs the following formula:

$$\mathrm{C}\mathrm{D}\mathrm{I}=1-{\sum }_1^{\mathrm{n}}{\displaystyle \frac{{\mathrm{n}}_{\mathrm{i}}}{\mathrm{n}}}$$

(4)

$$\mathrm{L}\mathrm{C}\mathrm{D}\mathrm{I}={\displaystyle \frac{{\sum }_1^{\mathrm{m}}\mathrm{C}\mathrm{D}\mathrm{I}}{\mathrm{m}}}$$

(5)

In Equations (4) and (5), CDI is a compatibility value of a certain POI point, n is the total number of POI points within the neighbourhood range (1500 m), n represents the compatibility assignment of a single POI point, n is the maximum value of compatibility assignments for all POI points. LCDI is the compatibility value of regional land mixed-use, which is the average of the compatibility values of all POI points in the region, and m is the total number of grid cells in the region.

In order to calculate compatibility values, it is first necessary to establish a compatibility judgement matrix, with existing research serving as a point of reference. In this matrix, compatibility is assigned a value of 0, conditional compatibility is assigned a value of 0.5, and incompatibility is assigned a value of 1. This process was carried out for 13 POI categories, resulting in the establishment of a compatibility judgement matrix (

Table A1. Compatibility judgment matrix part1.

	Catering	Companies	Shopping	Finance	Residential Areas	Science	Living Services	Medical
Catering	0
Companies	0	0
Shopping	0	0	0
Finance	0	0	0	0
Residential Areas	0	0	0	0	0
Science	0	0	0	0	0.5	0
Living Services	0.5	0.5	0.5	0.5	0.5	0.5	0
Medical	0	0	0	0	0.5	0	0.5	0
Entertainment	0	0	0	0	0	0	0.5	0
Government	0	0	0	0	0.5	0	0.5	0
Accommodation	0	0	0	0	0	0	0.5	0
Traffic Facilities	1	1	1	1	0.5	0	0.5	0
Scenic Spots	0.5	0.5	0.5	0.5	0	0.5	0.5	0.5

and

Table A2. Compatibility judgment matrix part2.

	Entertainment	Government	Accommodation	Traffic Facilities	Scenic Spots
Entertainment	0
Government	0	0
Accommodation	0	0	0
Traffic Facilities	1	0	1	0
Scenic Spots	0.5	0.5	0.5	0.5	0

).

Based on the above three results, the land use mix can be derived. It is postulated that diversity, accessibility, and compatibility exert a considerable influence on the extent of land use mix. Consequently, the weights have been set at 1/3 each, and the formula is as follows:

$$\mathrm{L}\mathrm{U}\mathrm{M}\mathrm{I}={\displaystyle \frac{\mathrm{N}\left(\mathrm{E}\mathrm{N}\mathrm{T}\right)+\mathrm{N}\left(\mathrm{A}\right)+\mathrm{N}\left(\mathrm{L}\mathrm{C}\mathrm{D}\mathrm{I}\right)}{3}}$$

(6)

In the formula, LUMI represents land use mix, N is the normalization method, ENT island use mix diversity, A is land use mix accessibility, LCDI is land use mix compatibility.

2.3. Construction of the Influence Mechanism Model

The spatial heterogeneity of the jobs-housing mismatch is influenced by various factors. However, traditional linear regression models start from the perspective of global assumptions that spatial variable relationships are fixed and do not change with spatial location. This premise assumption clearly violates the law of heterogeneity or non-stationarity of spatial relationships in the real world. Geographically Weighted Regression (GWR) is a local linear regression method based on the modelling of spatially varying relationships (Figure 2). Unlike traditional statistical analysis methods with global assumptions, GWR can effectively account for local spatial relationships and variable spatial heterogeneity by generating regression models that describe local relationships at each location in the study area [28]. Consequently, this study employs the GWR model to conduct a robust regression analysis, with the following formula:

$${\mathrm{y}}_{\mathrm{i}}={\mathsf{\beta }}_0\left({\mathrm{u}}_{\mathrm{i}},{\mathrm{v}}_{\mathrm{i}}\right)+\sum _{\mathrm{k}=1}^{\mathrm{n}}{\mathsf{\beta }}_{\mathrm{k}}\left({\mathrm{u}}_{\mathrm{i}},{\mathrm{v}}_{\mathrm{i}}\right){\mathrm{x}}_{\mathrm{i}\mathrm{k}}+{\mathsf{\epsilon }}_{\mathrm{i}}$$

(7)

In the equation, y is the value of the dependent variable at position i; x_ik (k = 1,2…m) is the value of the independent variable at position i; (u, v) are the coordinates of the regression analysis point i; β₀(u, v) is the intercept; β_k(u, v) (k = 1, 2…m) are the regression coefficients.

In this GWR model, the jobs-housing mismatch index is the dependent variable, whereas land use mix constitutes the independent variable. In light of the findings of relevant research, the principal factors influencing jobs-housing mismatch can be classified into three categories: traffic condition factors, public service factors, and socio-economic factors. For the factor of transport facilities, distance to bus stops and distance to metro stations are used as indicators, as a high level of bus and metro station coverage contributes to the formation of employment centers [29]; The distance to hospitals and distance to schools are employed as indicators, given their impact on residents’ housing choices and commuting travel patterns. Regional GDP incorporates benchmark land prices into the socio-economic indicator layer, given the significant impact these have on the distribution of residential space. The relative importance of each variable is indicated by the weights calculated using the entropy weight method. The Entropy weight method is a comprehensive evaluation method of multiple indicators developed on the basis of information entropy theory. The larger the entropy value of the indicator, the smaller the weight. This avoids the interference of human factors in the determination of indicator weights and makes the determination of weights more objective [30]. The results are shown in

Table 2. Index system.

System	Indicator	Source of Data and Description	Weight
Traffic Condition	Distance from transportation facilities	Amap POI data, Euclidean distance	35.41%
	Distance from subway stops	Amap POI data, Euclidean distance	64.59%
Public Service	Distance from hospitals	Amap POI data, Euclidean distance	56.59%
	Distance from schools	Amap POI data, Euclidean distance	43.41%
Socio-Economic Situation	GDP	GDP spatial distribution Kilometer grid dataset	39.06%
	Benchmark land price	Bureau of Planning and Natural Resources, Hangzhou	60.94%

.

3. Results

3.1. Results of Jobs-Housing Mismatch

Figure 3 illustrates the spatial visualization and numerical distribution of the jobs-housing mismatch index in the main urban area of Hangzhou. The numerical distribution of the jobs-housing mismatch index is approximately normal with an expected value of 1.00, indicating that the results of measuring the index using mobile signaling data are reliable. A jobs-housing mismatch index between 0.96 and 1.05 indicates a balance between residence and work. An index less than 0.95 indicates a dominance of residential functions. Conversely, an index greater than 1.05 indicates a dominance of employment functions.

The value of the jobs-housing mismatch index provides a quantitative basis and a key perspective for in-depth analysis of the spatial relationship between residence and employment in cities, and for optimising urban planning and resource allocation.

Firstly, the value of the jobs-housing mismatch index intuitively reflects the degree of matching between the place of employment and the place of residence. When the index value is close to 1, it indicates that there is a relative balance between jobs and residences, most residents can find a job within a short commuting distance, or there is an adequate supply of residences in the neighbourhood of jobs. For example, in residential-employment balance areas, such as some central urban areas and well-planned new districts, residents commuting time is relatively shorter and their living convenience is higher, which is conducive to reducing traffic congestion, lowering energy consumption and improving residents’ life satisfaction.

From the perspective of urban spatial structure, the differences in the jobs-housing mismatch index of different regions reveal the irrationality of the functional layout of the city. In areas dominated by residential functions, there is a belt-shaped distribution around the central city due to the double influence of the siphoning effect of the central city and the residents’ own economic conditions. These areas may attract a large number of residents due to their proximity to the central city, but employment opportunities are relatively limited, and residents often need to travel to the central city or other employment-concentrated areas to work, resulting in an increase in commuting distance, which reflects the spatial incongruity between the residential and employment functions. For employment-dominated areas, this means that the concentration of jobs far exceeds the distribution density of the residential population. For example, areas with employment concentrations are mostly located in the central city and remote suburbs, where a large number of employed people may come from the surrounding areas or even further afield, with commuting flows converging towards the area. This not only puts enormous pressure on the regional transport infrastructure, such as traffic congestion in the morning and evening peaks but also affects the work-life balance of the employed, reducing work efficiency and quality of life.

3.2. Spatial Pattern of Land Use Mix

The diversity, accessibility and compatibility of land use were measured separately, with standardized scores obtained for the three sub-measures and the degree of land use mix. The classification results based on the natural breaks method, are shown in Figure 4. The results demonstrate that there is considerable spatial variation in the degree of land use mix within the primary urban zone of Hangzhou. The areas with the highest value of land use mix are concentrated in the central urban area, new urban areas, and recently established economic and technological development zones. In contrast, the areas with the lowest value of land use mix are situated in urban ecological green space aggregation areas, while the median value areas are primarily located in transitional zones between the aforementioned areas.

In some emerging urban integrated functional zones, a good compatibility relationship has been formed between commercial land, residential land and land for public service facilities, with the prosperity of commercial businesses providing a convenient consumption environment for residents, the residential population providing a stable source of customers for commercial activities, and the public service facilities guaranteeing the basic needs of both residents and commercial operations, and zones with a high degree of compatibility tend to have a high degree of land-use mixing. On the other hand, in some traditional industrial areas or single-function residential areas, the compatibility is lower, and the land use mix is correspondingly lower. This fully demonstrates that land use compatibility is an important factor that cannot be ignored in contributing to the degree of land use mix.

At the same time, diversity-centred measurement methods have certain limitations in assessing the degree of land-use mixing. Such methods tend to focus only on the abundance of land use types while ignoring key elements such as spatial accessibility of land use and compatibility between different types of land. For example, in some urban fringe areas, although there may be a variety of land use types, due to the inconvenience of transport, the lack of effective links and interactions between various types of land, compatibility is poor, and the actual land use mixing effect is not ideal.

The method of weighted summation using the three seed measures can effectively compensate for this defect. By comprehensively analysing diversity, accessibility and compatibility, the real status of land use in different areas can be more accurately reflected. The contribution of accessibility may be relatively high in core business districts in the city centre, where convenient transportation is essential for business activities; while in some upscale residential districts, the importance of compatibility may be more prominent to reflect the comfort and coordination of the living environment. This weighted sum approach can comprehensively consider the characteristics of land use in different areas from the perspective of spatial heterogeneity, so as to express the land use characteristics more comprehensively and accurately, and provide a more scientific and practical basis for decision-making in the fields of urban planning and land resource management.

3.3. The Impact Mechanism of Land Use Mix on the Jobs-Housing Mismatch

In regard to the GWR model parameters, the study has elected to utilize the Gaussian function as the kernel function, the AIC method as the bandwidth selection criterion, and the Monte Carlo simulation method to assess the significance of variable spatial heterogeneity. The results of the GWR model, calculated using GWR4.0 software and the Monte Carlo test p-values, are presented in

Table 3. Regression results (***, ** and * are significant at 1%, 5% and 10% levels respectively).

Indicator	Jobs-Housing Mismatch Index
	Residential Dominant		Employment Dominant
	High Mismatch	Low Mismatch	High Mismatch	Low Mismatch
Diversity	0.041 **	0.003 **	0.072 **	−0.045 **
Availability	0.003 **	0.003 **	−0.045 ***	−0.023 ***
Compatibility	−0.013 ***	−0.002 ***	0.013 ***	0.026 ***
Traffic Condition	−0.180 ***	−0.061 ***	0.075 ***	0.108 ***
Public Service	−0.583 ***	−0.306 ***	0.627 ***	1.028 ***
Socio-Economic Situation	0.249 *	0.346 *	−0.004 ***	0.006 ***
R2	0.65		0.30

.

In general, the diversification of land use will result in an increase in the jobs-housing mismatch in residential areas, while a decrease will occur in areas with a high concentration of employment. This suggests that enhancing land use diversity will mitigate jobs-housing mismatch in these regions. The accessibility of land mixed use is found to be positively correlated with the degree of jobs-housing mismatch in residential-dominant areas and negatively correlated with that in employment-dominant areas. This suggests that enhancing land mixed-use accessibility may prove beneficial for achieving a more balanced jobs-housing ratio. The compatibility of land mixed use has a negative impact on the jobs-housing mismatch in residential-dominated areas and a positive impact in employment-dominated areas. This indicates that improving land mixed-use compatibility will promote jobs-housing mismatch. Figure 5. presents a visual representation of the spatial distribution of the GWR model regression results.

The goodness of fit (R²) of the GWR model in residential dominant areas is 0.65, indicating a strong explanatory power. In contrast, in employment-dominant areas it is only 0.3, suggesting that the impact is uncertain. Areas with a higher goodness of fit are concentrated in the central urban area, while areas with a lower goodness of fit are mainly distributed in urban ecological green space aggregation areas. This may be due to the economic and social conditions in that area leading to an unusual data distribution.

The regression results indicate that improvements in land mixed use diversity and accessibility will inhibit jobs-housing mismatch in residential dominant areas. Conversely, increased compatibility is associated with a promotion of jobs-housing mismatch. However, the situation in employment-dominant areas is somewhat different. In areas with high job mismatch, increased diversity will promote jobs-housing mismatch. When the regression results are combined with the visualization results, it is found that areas where land mixed-use promotes jobs-housing mismatch are mainly concentrated in newly built urban areas. In contrast, areas where jobs-housing mismatch is inhibited are mainly located in urban ecological land aggregation areas. It can be hypothesized that the increase in land use mix in newly built areas makes them local employment centers, while supporting facilities such as housing lag behind, thus exacerbating jobs-housing mismatch. In contrast, the source of increased land use mix in ecological green land aggregation areas is structural optimization during the land use adjustment process, which, to some extent, promotes the matching of residential and employment spaces. Therefore, the land use mix in these areas has a significant inhibitory effect on jobs-housing mismatch.

3.4. Strategies for Mitigating Spatial Differentiation in Jobs-Housing Mismatch

In order to gain a comprehensive understanding of the characteristics of land use layout in different areas, it is necessary to implement more targeted governance of urban spatial jobs-housing mismatch. The mismatch can be divided into five categories: high residential mismatch (≤0.5), low residential mismatch (0.51~0.95), residential-employment balance (0.96~1.05), low employment mismatch (1.06~1.5), and high employment mismatch (≥1.5). The mix is divided into low mix (≤0.5) and high mix (≥0.5). The intersection of the two results in nine spatial patterns of mix-mismatch zones, as illustrated in Figure 6.

Based on the nine Zoning categories obtained from the extraction, combined with the results of diversity, accessibility, compatibility, and environmental variable regression, a detailed description of each zone is provided below:

Employment dominant zone with low mix and high mismatch. This zone is mainly located outside the city’s third ring, dominated by industrial parks, with a small residential population, and with over-planned roads but insufficient related supporting facilities. To address this challenge, we propose a strategy to improve the accessibility of public resources to promote balanced development in this area. Specifically, new schools, hospitals, and commercial complexes should be planned along major transport routes, so that these resources can be closely linked to the employment-population in the region through convenient transport connections, thus promoting the balanced development of housing and employment in this area, and enhancing the comprehensive competitiveness of the region and the quality of life of its residents.

Employment dominant zone with low mix and low mismatch. This zone is distributed between the first and second ring and outside the third ring, with a relatively balanced land use structure and good accessibility. Our strategy for this area is to utilize its decentralization function to enhance the comprehensive use of the surrounding area. In detail, the region can be guided to extend some of its industrial functions to neighbouring regions with development potential, leading to employment growth and land development in the neighbouring regions; at the same time, the region can be encouraged to cooperate with its neighbours in such areas as infrastructure construction and the sharing of public services, so as to promote regional integration and development, and to further enhance the comprehensive benefits and development vitality of the region as a whole.

Residential dominant zone with low mix and high mismatch. Also located outside the third ring, this area is rich in land use types and carries the residential population from the neighbouring industrial parks and the central city, but the pressure on public facilities and traffic is high. Therefore, we propose strategies to enhance traffic diversion and resource support. These areas need to optimise the construction of more public transport routes and stations, and plan for the installation of tidal lanes to ease traffic congestion and improve the efficiency of residents’ travel. Large commercial complexes should be planned and constructed in close proximity to residential areas, while the development of community businesses, community hospitals, and schools should be encouraged to form a multi-level public service system.

Employment dominant zone with high mix and low mismatch. This zone is located between the first and third ring, and has a wider range of residential land and more employment opportunities compared to the central city, but with underdeveloped transport facilities. We propose to improve the localized transport conditions to promote a balance between employment and housing. Specific measures include the widening of roads, the construction of transport infrastructure such as viaducts and underpasses, the optimisation of the public transport network, the increase in the coverage of buses and metro lines, and the enhancement of transport capacity; and at the same time, increasing the supply of sheltered housing, flats for talented people, or rental housing, so as to reduce the pressure on traffic caused by the tidal flow of job seekers.

Employment dominant zone with high mix and high mismatch. This area is a city-wide economic agglomeration with poor site compatibility and high pressure for economic growth, making it difficult to match the job/residence relationship. Our strategy is to focus on synergy and complementarity with neighbouring districts to optimize land use and reduce population loss. These regions should strengthen their cooperation with neighbouring regions in terms of industrial division of labour and functional complementarities, transferring outward some of the industrial functions that are unsuitable for development in their own regions, and at the same time introducing functional elements lacking in the neighbouring regions, so as to achieve synergistic interregional development.

4. Discussion and Conclusions

4.1. The Spatial Distribution of Jobs-Housing Mismatch and Land Use Mix

The spatial distribution of jobs-housing balance and land use mix in Hangzhou shows significant differences. The study shows that the jobs-housing balance area is mainly concentrated in the central urban area, the residential functional area is distributed around the central area, and the employment functional area is distributed in the central area and its periphery. This distribution is consistent with some urban development trends in which jobs-housing balance is a key factor in sustainable urban planning [17]. The spatial distribution of the land use mix, especially in newly developed areas and eco-green spaces, has a profound impact on the separation of work and housing, highlighting the need for customized urban planning strategies that take into account local characteristics [21,31].

4.2. Effect of Land Use Mix on Jobs-Housing Mismatch and Partitioning Strategies

The impact of the land use mix on jobs-housing mismatch is multifaceted and exhibits spatial heterogeneity. The results of our GWR modelling suggest that the impact of land use diversity, accessibility and compatibility on jobs-housing mismatch varies across urban areas. The impact of the transport environment, especially metro stations, should not be ignored; in newly developed areas, an increase in land use structure exacerbates jobs-housing mismatch, whereas in eco-greenfield agglomerations, it mitigates jobs-housing mismatch. Similarly, the geographic characteristics of the city have a profound impact on the jobs-housing balance. These findings emphasize the complexity of the relationship between land use mix and jobs-housing mismatch and the importance of spatially sensitive policy interventions.

Based on measurements of jobs-housing mismatch and land use mix, targeted recommendations are made for achieving a differentiated jobs-housing balance. This approach is essential for implementing targeted urban spatial governance that takes into account the unique characteristics of each area. For example, areas with low mix and high mismatch may need to improve public resource allocation and accessibility, while areas with high mix and low mismatch may benefit from enhancing local transport infrastructure. At the same time, the application of new housing models offers new ideas for reducing the jobs-housing mismatch in certain areas [32].

4.3. Limitations and Further Studies

Our study has important implications for urban planning and policy-making. Our findings suggest that targeted land use strategies can help alleviate the jobs-housing mismatch, thereby promoting sustainable urban development and livability. However, the study still has limitations, firstly we focus on only one city, Hangzhou, which may limit the generalizability of our results. Second, we did not capture the long-term trends or dynamics of jobs-housing mismatch and the impact of certain special events [33,34,35,36], such as the impact of COVID-19 on commuting habits and land use dynamics. Therefore, future research may explore the long-term effects of land use policies on jobs-housing mismatch and consider the impact of emerging urban development trends on these patterns. In addition, research could benefit from comparative analyses of cities in different countries or at different levels of development to understand the broader applicability of our findings.

4.4. Conclusions

The analysis is based on the measurement of jobs-housing mismatch and land use mix in the city. This paper employs a GWR model to analyze the impact of the land use mix on jobs-housing mismatch in the main urban area of Hangzhou. It also proposes targeted zoning response plans. The principal conclusions are as follows: Firstly, the measurement of jobs-housing mismatch using mobile signal data is a highly informative and accurately positioned spatial analysis, which avoids the potential errors associated with small sample sampling. In comparison to results that commonly rely on “diversity” as the sole measurement, a comprehensive land use mix result that integrates three types of sub-measurement methods can more fully depict the functions, structures, and interactions of land use, thereby more accurately reflecting the impact of land mixed use on jobs-housing mismatch. Secondly, the impact of land use mix on jobs-housing mismatch is observed to exhibit spatial heterogeneity. An increase in land use mix in newly constructed areas is found to promote jobs-housing mismatch, whereas an increase in land use mix in areas characterized by the aggregation of ecological green space has an inhibitory effect on jobs-housing mismatch. Thirdly, the implementation of targeted urban jobs-housing mismatch governance based on the mix-mismatch zoning is conducive to the rational allocation of land use structure and spatial layout, the scientific planning of new transportation routes, and the promotion of the match between residential space and employment space.