Spatial Distribution Pattern and Influencing Factors of Homestays in Chongqing, China

Abstract

As an emerging business form of tourism development, homestays also play an important role in China’s rural revitalization and tourism transformation and upgrading, and has attracted increasing social attention. At present, Chongqing is the city with the largest number of homestays in China. Taking Chongqing as a case-study area, based on the homestay data of the Baidu map, this paper comprehensively uses the methods of spatial analysis, multiple regression and geographical weighted regression to thoroughly analyze the regional characteristics and influencing factors of homestay distribution in Chongqing. The results show that: (1) the nearest-neighbor index R of homestay distributions in Chongqing and all regions is one, which shows an obvious agglomeration type. (2) In addition to being highly concentrated in the central urban area, three secondary high-density areas are also formed in the surrounding areas of the central urban area, and there is a trend of concentration and contiguity. The spatial distribution densities of the two urban agglomerations in Southeast and Northeast Chongqing are very low, and the overall distributions are extremely uneven. (3) The factors, such as tourism resource endowment, economic development, service industry development, traffic location, consumption demand and social development conditions, have significant impacts on the distribution pattern of homestays, and the impacts of each factor on the layout of homestays has obvious spatial heterogeneity. Analyzing and revealing the temporal and spatial characteristics and dynamic mechanism of homestays has an important theoretical value and practical significance for better serving the new urbanization plan and implementing the strategy of urban–rural integration and rural revitalization.

1. Introduction

With the rapid development of China’s urbanization process and the increasing improvement of people’s living standards, major changes have occurred in the consumption demand and consumption structure for the majority of residents. Leisure tourism and vacation has gradually become normalized ordinary mass consumption. At present, people reflect more on the high-pressure life lived in the city [1], yearn for a quiet life in the countryside and personalized service experience, and the homestay business was created [2,3,4]. Being different from the general accommodation business, homestay pays more attention to tourists’ sense of experience and participation. It is a combination of leisure and tourism accommodation experience [5,6] and will develop into an important local tourism resource. In recent years, due to the demand orientation of the market and a series of incentive policies issued by the government [7], the development momentum of homestays in China has been rapid. The homestay industry has gradually become a new field in China’s current small- and medium-sized capital investments in tourism and living consumption, and has also played an important role in revitalizing social idle resources. Moreover, the development of homestays in rural areas can meet the new demands of the current rural construction movement and rural revitalization [8], bring considerable economic and social benefits, and effectively promote the integrated development of urban and rural areas [9].

Compared with ordinary hotels, homestays are more attractive because they focus on experiential personalized services, have smaller spaces, and are a private ‘home away from home’. Tourists can make new friends while being familiar with the new environment [10]. Homestays usually take the form of community- or village-based tourism [11]. Although homestays are becoming more and more important in the tourism market, the development of homestay tourism presents both advantages and disadvantages [12,13]. The development of homestays can provide sustainable income for local people struggling in remote areas but lacking employment opportunities, and improve the sense of ownership and responsibility of homestay operators to the local natural environment, so as to enable them to protect the local cultural and natural heritage [14,15]. However, the operation of homestays enables local residents to form a binary distinction between public hotels and private houses, residences and other places, commercial and non-commercial places, commercialization and authenticity, and work and residence, which increases the dependence of the local economy on tourism and reduces social cohesion and fairness [16]. The long-term development of tourism activities will also crowd out the economic activities loved by local people, and even new residents will replace the original residents [17]. The development of homestays in rural areas has promoted the commercialization of rural families. In this process, new capital flows into the countryside for construction or environmental reconstruction. New immigrants ‘take over’ local houses and promote the rural socio-economic transformation. However, the original rural social relations have been destroyed and capitalized, the competition between people in the same village has intensified, the gap between the rich and the poor has widened, and the social ties have weakened [18,19,20,21]. It also caused the relocation of some Aborigines, which was not only the apparent departure from the original houses, but also internal social and cultural ‘displacement’, especially for the nostalgic elderly.

Homestays have experienced a long process from their initial rise in Europe to their introduction to and prosperity in China. Due to the different background and forms of homestay businesses in various countries, there are differences in the understanding of the connotation of homestays in different countries and regions, but there are also some common points: (1) the operators and managers are usually owners, and the business carriers are idle folk houses [22]; (2) operating as a sideline [23]; (3) foreign tourists have certain exchanges with the hosts of the homestays, and the lodgers not only experience the accommodation, but also experience the lifestyle [24]; (4) small-business scale [25]; and (5) regional characteristics are the core attraction of homestays. Relying on the local unique resources, it provides a variety of characteristic experience services [26]. In the new edition of basic requirements and evaluations of tourism homestay inns published by the Chinese Ministry of Culture and Tourism in July 2019, the definition of tourism homestay inns is to make use of relevant idle resources, such as local houses, with operating rooms of no more than four floors and a building area of no more than 800 m². The host participates in the reception and provides tourists with small accommodation facilities to experience the local nature, culture, production, and lifestyle [27]. The research object of this paper is all homestays in Chongqing, including both urban and rural areas.

At present, the relevant research results mainly involve the quantitative distribution characteristics, spatial distribution types, distribution agglomeration degree, distribution balance, spatial distribution, and function coupling [28,29,30,31]. The research methods are mainly based on the geospatial analysis method of the ArcGIS geographic information software platform and the mathematical statistics method calculated by SPSS (Statistical Product Service Solutions) and other tools [32]. However, most of the research areas on the spatial layout of homestays are concentrated in small scales, such as cities and scenic spots, and most of the research areas are concentrated in the regions with rapid economic development and the developed tourism-service industry in eastern China [33,34,35,36]. There is little research conducted on Central and Western China, and the research content mostly focuses on qualitative representation research, lacking a systematic explanation of the spatial layout law and formation mechanism of homestays. It is necessary to use quantitative models to enhance the accuracy and scientificity of the analysis and elaboration. With the extension of China’s regional development strategy, it is necessary to conduct a detailed study on the distribution pattern, development process and trend and existing problems of typical regional homestays, so as to better serve the promotion of regional development strategies. Based on this, the current paper uses Chongqing as the research area. Firstly, it dissects the characteristics and patterns of the spatial layout of homestays, reveals the equilibrium of the spatial distribution of homestays in Chongqing by plotting the Lorenz curve, analyzes the types of spatial distributions of homestays in Chongqing through the nearest-neighbor index, and analyzes the agglomeration of the spatial distribution of homestays in Chongqing through a kernel density estimation. Then, the OLS (Ordinary Least Squares) global regression model and the GWR (Geographically Weighted Regression) local regression model are then used to analyze the factors influencing the distribution of the homestays in and areas outside the central city of Chongqing, and to conduct a comparative study of the two models to analyze the factors influencing the spatial distribution of homestays and their spatial heterogeneity, which is of great practical significance in promoting the optimization of the layout of the homestay industry, solving the problem of unbalanced and insufficient regional development.

2. Materials and Methods

2.1. Data

Chongqing (105°11′~110°11′ E, 28°10′~32°13′ N) is located in the southwest area of inland China (Figure 1). It is one of China’s four municipalities directly under the central government, and governs 26 districts and 12 counties (4 ethnic minority autonomous counties). Chongqing is an inland export commodity-processing base and a pioneer zone for expanding opening up to the outside world. It bears the important regional functions of economic, financial, scientific and technological innovations, and is a shipping and trade logistics center in the upper reaches of the Yangtze River. It is not only an important strategic fulcrum for Western development, but also an important link between the ‘the Belt and Road’ strategy and the Yangtze River Economic Belt [37]. However, for a long time, Chongqing has obvious characteristics of the urban–rural dual system. The central urban area (i.e., nine districts of the main city) is the main urban area in the sense of urbanization and sociology, which is different from other rural areas. They have significant differences in their urbanization rates and socio-economic development levels. Chongqing formulates the development pattern of ‘one district and two groups’ according to the differences in background elements of different regions (Figure 1). That is, the main urban metropolitan area (including 9 districts in the central urban area, 12 districts in the main urban new area and Wansheng Economic Development Zone), the urban agglomeration of the Three Gorges Reservoir Area in Northeast Chongqing (11 districts and counties), and the urban agglomeration of the Wuling Mountain Area in Southeast Chongqing (6 districts and counties). In view of this regional difference, when analyzing the influencing factors of the spatial distribution of homestays, this paper divided Chongqing into two parts: the central urban area and the area outside the central urban area.

The POI (Point of Interest) data of the homestay points in this paper were obtained on the API (Application Programming Interface) platform of the Baidu map website in September 2021, including the coordinates, price, name, detailed address, district, and county where the homestays were located. There were some duplicates in the preliminary data. Therefore, the obtained data were screened, the duplicate sample points with fuzzy attributes and missing key information were removed, and the coordinate deviation was corrected. In order to ensure the accuracy of the data, select some samples for field investigations and verifications or image comparisons to obtain the final homestay POI data.

Other socio-economic data presented in this paper, such as regional GDP (Gross Domestic Product), total tourism income, population, urbanization rate, highway mileage, per capita disposable income of residents, forest coverage, and air quality rate, were obtained from the statistical yearbook of Chongqing provided by the official website of the Chongqing Municipal Bureau of statistics and the statistical bulletin of national economic and social development of all districts and counties. Space vector data, such as administrative boundaries, were obtained from the 2016 national vector data package; DEM (Digital Elevation Model) data was obtained from LocaSpaceViewer (http://www.locaspace.cn/) (accessed on 18 September 2020) digital elevation data with medium GDEMDEM30M resolution. Finally, Arcgis10.2 was used as the operation platform; the Baidu map coordinate picker was used to select the control points, register POI point data, and unify the projection coordinate system of each data layer.

2.2. Methods

2.2.1. Average Nearest Neighbor

There are usually three spatial distribution types of point elements: condensed, uniform, and random. At present, the research method for analyzing the spatial distribution type of point elements often adopts the nearest-neighbor index method, also known as the R index, which can describe the mutual proximity of point elements in space [38], and its calculation formula is:

$$R=\frac{{\overline{r}}_{{}_i}}{{\overline{r}}_{{}_E}}$$

(1)

$$r_E=\frac{1}{2\sqrt{n/A}}$$

(2)

In Formulas (1) and (2), R is the nearest-neighbor index, that is, the ratio of the average to the theoretical nearest-neighbor distance, reflecting the spatial distribution characteristics of homestay points; $\overline{r_i}$ is the average nearest-neighbor distance; $\overline{r_E}$ is the theoretical nearest-neighbor distance; A is the area; and n is the number of points. It is generally believed that if R < 1, it indicates that the spatial distribution of point elements is clustered; if R > 1, it indicates that the spatial distribution of point elements is uniform; if, R = 1 it indicates that the spatial distribution of point elements varies with the model; and if R = 0, it indicates that all point features are concentrated at one point.

2.2.2. Kernel Density Estimation

Kernel density estimation mainly estimates the probability density value according to the distance between the elements to be estimated and the sample elements, and studies the spatial distribution characteristics by using the spatial attributes of the data samples [39,40], which can better show the concentration degree of the spatial distribution of the guest house. The formula is:

$$f\left(x\right)=\frac{1}{nh}{\displaystyle {\sum }_{i=1}^n{K}_n\left(\frac{x-x_i}{h}\right)}$$

(3)

In Formula (3), f(x) is the kernel density calculation function at the spatial position, h is the threshold value of the analysis range, n is the number of points within the analysis threshold range, and K is the default spatial weight kernel function. The density value is the highest at the central point of each analysis window and continuously decreases outward. When the distance from the center reaches a certain threshold value range, the edge function value is 0. This paper used ArcGIS10.2 to analyze the spatial distribution of homestays in Chongqing by kernel density.

2.2.3. Preparation of Variables and Model Building

As a form of tourism and accommodation, the impacts of many factors, such as economy, society, culture and ecology, on the layout of homestays in different research regions is different [41], and the influencing factors and intensity of homestay spatial distributions in different regions are also different (

Table 1. Influencing factors of homestays in Chongqing.

Types	Names	Paraphrase
Independent variable	Number of homestays	Number of homestays in districts and counties
Natural geographical conditions	Average altitude	Average altitude of the region
	Forest-cover rate	Regional forest cover
	Air-quality rate	Percentage of days with good air quality in the region
	A-level scenic spot	Number of A-level scenic spots in the region
Regional economic conditions	GDP	Regional annual gross domestic product
	Total tourism revenue	Total annual revenue of regional tourism
	Output value of primary industry	Regional agricultural annual gross product
	Output value of tertiary industry	Annual total revenue of regional service industry
	Per capita disposable income of residents	Per capita disposable income of local residents
	Residents’ per capita living consumption expenditure	Per capita living consumption expenditure of regional residents
	Accommodation and catering industry turnover	Regional accommodation and catering industry turnover
Social development conditions	Permanent residents	Regional resident population
	Urbanization rate	Urban population/total population
	Road network density	Total regional highway mileage/area
	Total number of visitors	The total number of tourists received by the regional tourism industry
	Metro station density	Number of metro stations/area
	College density	Number of regional colleges and universities

). According to the characteristics and conditions of the study area, this paper selected the indicators that were highly representative, easy to obtain, quantifiable, and practical for the analysis and research (Figure 2).

The average altitude, forest coverage, excellent air rate, and class A scenic spots were selected for the natural and geographical conditions. The altitude directly affects the local climate, temperature, light, precipitation, and other conditions. The area with a pleasant temperature, moderate light, and less extreme weather was more suitable for the location and layout of the homestay and the operation and development of the homestay [42]; the forest-coverage and air-quality rates were the most direct indicators of the quality of the ecological environment in the region. Compared with other ordinary accommodation formats, consumers were more frequently in pursuit of fresh air, clean water, and a green ecological environment [43]; there was no doubt about the impacts of scenic-spot resources on the layout of the homestays. Most of the homestay business forms were originally developed based on the ordinary accommodation business forms around the scenic spot. The unique natural or cultural landscape can cause considerable competitive advantages to the development of homestays in the region [44]. The higher-quality landscape resources can drive the agglomeration of homestays. Regional economic conditions mainly include regional GDP, total tourism income, tertiary output value, per capita disposable income of residents, per capita living consumption expenditure of residents, and turnover of accommodation and catering industry. A good level of regional economic development means that residents have higher disposable incomes, more diversified living consumption demands and strong tourism consumption demands, which can promote the development of new tourism formats, such as homestays [45]; good regional economic development can ensure the financial revenue of local governments. The government has the ability to promote the construction of local tourism infrastructures and service distribution facilities, has sufficient funds to invest in the development of the tourism industry, and implements more tourism development projects and publicity and promotion activities, which will help to improve the core competitiveness of the local tourism industry [46]; the developed service industry in the regional industrial structure can provide a good industrial environment for the development of homestay businesses. Social development conditions are selected from indicators, such as resident population, urbanization rate, road network density, and total number of tourists. The number of the regional resident population determines the potential market scale of homestay tourism, and stable passenger flow is an important prerequisite for the development and agglomeration of homestays [47]; the process of urbanization is a very important engine to stimulate the growth of social and economic consumption and investment [48]. At the same time, urban residents are facing many life pressures and urban diseases, and the demand for vacation and leisure time is stronger; regional accessibility is the key to the development of the tourism industry. Traffic conditions not only affect the development and utilization of homestays, but are also directly related to tourists’ tourism experiences [49]. The healthy development of the homestay industry needs to be backed by convenient traffic basic conditions.

Taking the county as the basic research unit, in order to break the limitations of the traditional linear regression model in the spatial characteristics of independent variables and their ‘global’ estimations, this paper comprehensively used the ordinary least squares method (OLS) and geographically weighted regression (GWR) model to calculate the impacts of the factors above on the spatial layout of homestays in the global and different regions, so as to more intuitively reflect the spatial imbalance and trend change in the driving mechanism of homestay spatial layouts [50]. The regression analysis model is as follows:

$$y_i={\beta }_0+{\displaystyle \sum _{i=1}^k{\beta }_i{x}_i+{\epsilon }_i}$$

(4)

In Formula (4), the classical linear regression model is usually composed of a random error term ${\epsilon }_i$ and a set of parameters, ${\beta }_0$ and ${\beta }_i$; i is the estimated value of equation parameters obtained by the minimum error smoothing method. It should be noted that the OLS model undertakes the function of eliminating invalid variables when running the GWR model, that is, the VIF value of the variable in the operation’s result should be less than 7.5; otherwise, the variable has a collinearity problem. The research conducted on spatial heterogeneity adopts the geographically weighted regression model (GWR), which is essentially an extension of the general linear regression global model. It regresses each observation point, generates the parameters of the corresponding region, and then embeds the geographical location of the data into the regression parameters to form a local rather than global parameter estimation [51]. The extended model is:

$$y_i={\beta }_0\left(u_i,v_i\right)+{\displaystyle \sum _{}^n{\beta }_k}\left(u_i,v_i\right)x_{ik}+{\epsilon }_i$$

(5)

In Formula (5), $y_i$ is the dependent variable, i.e., the number of homestays in the district and county; $x_{ik}$ is an independent variable, i.e., the individual natural, economic, and social condition indicators; $\left(u_i,v_i\right)$ is the geographic center coordinate of the ith sample space unit; ${\beta }_0\left(u_i,v_i\right)$ is the coefficient of the regression equation of sample i; ${\beta }_k\left(u_i,v_i\right)$ is the k-th regression function in the i-th sample space unit; and ${\epsilon }_i$ is a random error term with an independent distribution. If ${\beta }_k\left(u_i,v_i\right)$ has the same value at any point I in space, then the equation is a global regression model. Therefore, the GWR model can also be used as an interpolation tool for spatial data.

GWR models are generally weighted using a Gaussian function to construct the weighting function. The bandwidth is obtained using the AIC method with a kernel density estimation, and then the geographically weighted regression is calculated. The Gaussian function is used to determine the weighting function as:

$$W_{ij}=\exp \left[-{\left(\frac{d_{ij}}{b}\right)}^2\right]$$

(6)

In Formula (6), b is the bandwidth and d_ij is the direct distance between sample points i and j. If data obtained from i are observed, the weights of the other points decrease with the increasing distance d_ij based on a Gaussian curve. When given a bandwidth, b, the greater the distance d_ij, the lower the weight assigned to location j, and the weight at a far enough distance from point i tends to 0. The relationship between b and CV is as follows:

$$\mathrm{CV}={{\displaystyle \sum _{i=1}^n\left[y_i-y_{\ne i}\left(b\right)\right]}}^2$$

(7)

In Formula (7), y_≠i(b) is the fitted value of y_i, and when the CV is smallest, the corresponding b is the corresponding bandwidth. Different weighting methods calculate different bandwidths, but the following general guideline is used: when the AIC value is the smallest, then the bandwidth b is the optimal width. All operations in this paper were performed using Arcgis 10.2 software (Created by Esri, US).

3. Results

3.1. Spatial Distribution Characteristics of Homestays

The number of homestays distributed within different administrations in Chongqing varied considerably, with Yuzhong District having the highest number of homestays (525), accounting for 18.07% of the total number of homestays in the city; Zhongxian County had the lowest number of homestays, with only eight, accounting for only 0.28% of the total number of homestays in the city. Most of the districts and counties in Chongqing with a high number of homestays are located in the main city’s metropolitan area. HomestayBs offered more high-end services than the general accommodation sector and needed to be supported by a better economic development environment. To further explore the spatial distribution of homestays in Chongqing, a Lorenz curve was plotted to visualize the spatial distribution of the homestays. The mapping process was as follows: firstly, the regional entropy of the number of homestays in each district and county was calculated, i.e., the proportion of the number of homestays in each district and county in the total number of homestays in the city, and they were ranked according to the order of size. Then, the entropy of the location of homestays in each district and county was taken as the horizontal coordinate of the curve, and the cumulative proportion of the number of homestays in each district and county was taken as the vertical coordinate of the curve, which formed the Lorenz curve of the spatial distribution of homestays in Chongqing (Figure 3). As can be observed from the graph, the Lorenz curve is generally characterized by a downward concavity, with the curve increasing at a significantly faster rate after Banan District. The number of homestays in the first 11 districts and counties, including Banan District, accounted for 68.85% of the total number of homestays, while the number of homestays in the next 27 districts and counties only accounted for 31.15% of the total number of homestays, further illustrating the extremely uneven spatial distribution of the number of homestays in Chongqing.

In the process of analyzing the kernel density of homestays in Chongqing, because the number of homestays in the central urban area was close to eight times of that in the urban agglomeration of the Three Gorges Reservoir Area in Northeast Chongqing, there was a large gap in the number of homestays, and the analysis effect was poor. Therefore, in the process of analyzing the kernel density of homestays in Chongqing, this paper excluded the homestay points of nine administrative units in the central urban area, and created a kernel density analysis of the homestay points in the other three areas. The analysis results of the resident nucleus density of Chongqing citizens were obtained (Figure 4).

From the overall perspective concerning Chongqing, the spatial distribution density of homestays was generally high in the West and low in the East. In addition to the high concentration of homestay distributions in the central urban area, three high-density areas were formed around the central urban area, and there was a trend of concentration and contiguity. The spatial distribution density of the two urban agglomerations in Southeast and Northeast Chongqing was very low, forming a strong contrast with the high-density main urban areas, and the overall distribution was extremely uneven. Locally, except for the central urban area, the high-density centers of homestays were mainly distributed in Wulong District, Qijiang District and Yongchuan District, and there was a trend of agglomerations. Secondly, there were low-density centers in Wanzhou District, Shizhu County, Fengjie County, and Wushan County, but the agglomeration was not obvious.

Through ArcGIS10.2, the software calculated the nearest-neighbor index of homestays in Chongqing and different areas, and the results are presented in

Table 2. Nearest-neighbor index of homestays in Chongqing and its districts.

Names	Observed Mean Distance (m)	Expected Mean Distance (m)	Nearest Neighbor Ratio	Distribution Types
Chongqing	922.9240	3747.5492	0.246274	Agglomeration
Central urban area	318.5892	970.6503	0.328222	Agglomeration
Main city new District	1541.8764	3940.5550	0.391284	Agglomeration
The urban agglomeration of Wuling Mountain Area	1706.9458	5171.3703	0.330076	Agglomeration
The urban agglomeration of Three Gorges Reservoir Area	3384.9553	7545.0662	0.448631	Agglomeration

. Among them, the results passed the significance test of 1%. The nearest-neighbor index of the spatial distribution of homestays in Chongqing and all areas is far less than one, so the spatial distribution type of homestays is the agglomeration type. The agglomeration of homestays in space is easier to form a scale effect, which is conducive to the rational utilization and distribution of resources.

3.2. Spatial Heterogeneity of Factors Affecting the Spatial Distribution of Homestays

It should be noted that the OLS model undertakes the function of eliminating invalid variables before running the GWR model, that is, observe the magnitude of the VIF value. The closer the VIF value is to 1, the lighter the multicollinearity, and vice versa. When the VIF value exceeds 10, there is co-linearity, the higher the value the more severe the co-linearity. Redundant variables with cointegrations should be excluded as they are not meaningful for geographically weighted regressions. After running the OLS model for central and non-central urban areas and eliminating the relevant variables, the fitting operation results of the model were obtained (

Table 3. The fitting results of OLS model.

Model Parameters	Coefficient	t-Value	p-Value	VIF
Central Urban Area
A-level scenic spot	0.662	2.474	0.023 *	2.959
GDP	0.512	1.677	0.031 *	3.001
Tertiary output value	0.149	3.376	0.015*	2.223
Per capita living expenditure	0.648	4.069	0.017 *	1.987
Metro station density	0.818	4.106	0.009 **	1.108
Colleges and universities	0.203	5.936	0.004 **	1.107
The areas outside the central urban area
Average altitude	0.325	3.602	0.016 *	2.892
Forest cover	0.276	1.513	0.025 *	3.584
Air quality	0.401	1.251	0.046 *	5.468
A-level scenic spot	0.461	2.017	0.032 *	5.179
Urbanization	0.336	5.193	0.008 **	4.866
Total number of visitors	0.201	5.891	0.005 **	2.401

Note: ‘*’ means passing the significance level test of 5%, and ‘**’ means passing the significance level test of 1%.

). According to the calculation results, the influencing factors of the distribution of homestays in the central urban area mainly had six indicators: class A scenic spots, regional GDP, tertiary output value, per capita living and consumption expenditure of all residents, subway station density, and the number of colleges and universities. The factors affecting the distribution of homestays outside the central urban area mainly had six indicators: average altitude, forest coverage, air-quality rate, number of class A scenic spots, urbanization rate, and total number of tourists. Each index passed the test of significance levels of 1% or 5%.

The smaller the AICc (Akaike information criterion) value of the regression analysis model, the better the fitting effect of the model and the more consistent with the data characteristics observed in the study. Comparing the AICc parameter values of the two models in

Table 4. Comparison of fitting parameters between GWR and OLS models.

	R2	Adjust-R2	AICc
OLS in central urban area	0.57	0.45	116.88
GWR in central urban area	0.87	0.66	78.21
OLS the outside the central urban area	0.68	0.31	53.12
GWR the outside the central urban area	0.72	0.65	20.85

, the AICc value of the GWR model in the central urban area was 38.76 less than that of the OLS model, and the AICc value of the GWR model in areas outside the central urban area was 32.27 less than that of the OLS model, indicating that the fitting effect of the GWR model was significantly better than that of the OLS model, reflecting the superiority of the local regression model.

By comparing the action modes of various factors, it was observed that the regression coefficients of various variables were quite different, and the spatial heterogeneity was obvious (

Table 5. Statistical results of the GWR model coefficient values.

Independent Variable	Minimum	Upper Quantile	Median	Lower Quantile	Maximum	Average
Central Urban Area
A-level scenic spot	−0.356	−0.188	0.073	0.111	0.296	0.126
GDP	0.973	1.114	1.453	1.569	1.701	1.356
Tertiary output value	1.405	1.432	1.563	1.626	1.708	1.557
Per capita living expenditure	0.132	0.196	0.341	0.369	0.592	0.336
Metro station density	0.512	0.565	0.726	0.814	0.961	0.728
Colleges and universities	0.275	0.286	0.318	0.326	0.356	0.313
The Areas Outside the Central Urban Area
Average altitude	−0.991	−0.971	−0.752	−0.266	−0.264	−0.662
Forest cover	0.021	0.066	0.355	0.745	0.952	0.428
Air quality	0.103	0.137	0.176	0.277	0.316	0.202
A-level scenic spot	0.351	0.359	0.401	0.424	0.432	0.395
Urbanization	0.335	0.449	0.639	0.769	0.846	0.617
Total number of visitors	0.104	0.128	0.223	0.797	0.881	0.371

). From the average value of the absolute value of the regression coefficient, in the central urban area, the influence degrees of the output value of the tertiary industry, GDP, the density of subway stations, class A scenic spots, the number of colleges and universities, and the per capita living consumption expenditure of all residents decrease in turn. In terms of the areas outside the central urban area, the impacts of average altitude, forest coverage, A-level scenic spots, urbanization rate, air-quality rate, and the total number of tourists decrease in turn. From the positive and negative values of the regression coefficient, in the central urban area, only class A scenic spots had positive and negative effects, and other variables had positive effects on the spatial layout of homestays. In terms of the areas outside the central urban area, the average altitude had a negative impact, while class A scenic spots, forest coverage, air-quality rate, urbanization rate, and the total number of tourist receptions had a positive impact.

3.3. Spatial Differentiation of Influencing Factors

In order to more intuitively characterize the spatial heterogeneity of the role of various factors on land urbanization, the significance characteristics and action intensities of the variables in the two regions were integrated, and the indicators, such as population urbanization rate, per capita regional GDP, service industry development, road network density, and topographic conditions, were selected to describe the spatial patterns of the role of tourism resource endowment, economic development, service industry development, transportation location, consumption demand, and social development conditions on the layout of homestays in Chongqing (Figure 5 and Figure 6).

From the influencing factors of the spatial distribution of homestays in the central urban area, at the level of tourism resource endowment, the influence of the number of class A scenic spots on the homestay layout gradually weakened from the Northwest to the Southeast directions within the central urban area, among which Beibei District and Shapingba District had the strongest influence, followed by Yuzhong District, Yubei District, and Jiulongpo District. From the perspective of the economic development level, the GDP level had an obvious positive impact on the distribution of homestays in the central urban area, and the influence gradually weakened from the center to the surrounding areas. In particular, the distribution of homestays in Yuzhong District, Jiangbei District, Yubei District, and Jiulongpo District was strongly affected by GDP. From the perspective of the development level of the service industry, the positive impact of the output value of the tertiary industry on the distribution of homestays was the strongest in Jiulongpo District, Shapingba District, and Yubei District, and then gradually weakened towards the Northwest and Southeast. From the perspective of the consumption demand level, the per capita living consumption expenditure index of all residents had a positive impact on the overall spatial distribution of homestays, and the intensity of the effect gradually decreased from Shapingba District, Yuzhong District, and Nan’an District in the middle to the North–South direction. From the perspective of traffic location conditions, the influence intensity of the subway station density index on the distribution of homestays in the central urban area decreased from the middle to the North and South. From the perspective of social development conditions, the impact intensity of the index number of colleges and universities on the distribution of homestays in the central urban area presented scattered spatial characteristics, and the overall impact was positive.

In terms of the influencing factors of the spatial distribution of homestays outside the central urban area, at the level of physical and geographical conditions, the average altitude played a negative role in the spatial distribution of homestays, and the influence of the average altitude on the layout of homestays decreased step by step from the Southwest to Northeast. The quality of the ecological environment is a necessary condition for the development of the tourism industry. The forest-coverage and air-quality rates had a positive impact on the spatial distribution of the homestays; their influence showed the spatial layout gradually decreasing from East to West, forming a banded distribution from East to West. From the perspective of tourism resource endowment, the number of class A scenic spots can have a positive impact on the spatial distribution of homestays outside the central urban area of Chongqing, and its impact degree showed a decreasing trend from the Southwest to Northeast. From the perspective of the urban development level, the urbanization rate had a positive impact on the spatial distribution of homestays in Chongqing, and its impact degree showed a decreasing trend from the Southwest to the Northeast, and presented the characteristics of strip distributions from the Northwest to the Southeast. The influence of the total number of tourists on the spatial distribution of homestays gradually decreased from the Northwest to the Southeast, which was also characterized by a strip distribution.

4. Discussion

4.1. Complexity and Spatial Heterogeneity of Influencing Factors of Homestays Distributions

With the accelerating process of urbanization and the continuous promotion of tourism in Chongqing, the homestay industry is rapidly expanding. In general, the distribution of homestays showed a ‘one-pole multi-core’ development pattern around the main urban area, which fully confirmed the decisive role of location in the tourism and accommodation industry, and the spatial behavior law of tourists also played an important role. The vast majority of homestays were distributed in cities and towns and surrounding areas, which were more affected by economic factors. The highly concentrated comprehensive resources in the urban center were the core driving force for the development and concentration of urban homestays. They had stronger comprehensive service capacities, comprehensive advantages in transportation, scenic spots, commerce and other aspects, showing diversified functions. Urban homestays pay more attention to urban humanities and culture and modern accommodation services; consumer audiences also tend to be younger. In vast rural areas, homestays are scattered and less in number, which are mainly built by relying on local scenic-spot resources or government policies. Scenic spots in districts and counties, such as Wulong fairy mountain and Shizhu Huangshui National Forest Park, have a good brand effect. The development and improvement of scenic spots produces a large quantity of rigid demands for tourists and accommodation, which has become the core driving force of rural homestay agglomeration. Rural homestays pay more attention to the ecological environment and original ecological experience, but they show a unity of function, which fails to promote the development of the homestay industry to scale and agglomeration. The consumer audience also tends to travel with families with elderly individuals and children.

In addition, through the OLS quantitative simulation, it was observed that tourism resource endowment, economic development, service industry development, traffic location, consumption demand, and social development conditions had a significant impact on the layout of homestays. The test results of the geographically weighted regression model show that the regression coefficients of various variables have a significant spatial heterogeneity, indicating that there are significant differences in the action intensities of specific factors on different types of regions. For example, economic development plays the strongest role in promoting the development of homestays in the central urban area, while tourism resource endowment affects the regions outside the central urban area more, especially the new area of the main city. This not only verifies the complexity of the spatial differentiation law of homestays, but also increases the difficulty of analyzing its driving mechanisms.

4.2. Possible Problems of Homestay Development and the Path of Urban–Rural Integration

With the knowledge that the development momentum of homestays is still strong, the business form of homestays in Chongqing is excessively concentrated in the main urban area. Due to the lack of reasonable regional planning, unified industry associations, clear management norms, and a sound social service system, the problem of homogenization is serious, resulting in the internal consumption of resources and local vicious competition of homestays in urban areas, which is not conducive to the formation of scale effect and the shaping of the characteristics of local brands. While rural areas are blessed with good natural scenery, the lack of effective financing channels has resulted in inadequate infrastructure support, such as roads, water and electricity, car parks, fire-fighting facilities, and safety supervision systems, required for the development of homestays, as well as an inadequate system of social services, such as government laws and regulations, all of which are detrimental to the development of homestays. Although some areas have built and transformed a number of rural homestays with the help of the government, the overall industrial chain is imperfect, the added value is low, and there is a lack of long-term operation and management mechanisms, which leads to the lack of a sustainable development capacity of the rural homestay industry, most of which are short lived.

Under the background of the sustainable development and promotion of rural tourism, in order to solve the dilemma of inefficient and excessive homestay economic development, we should start from the perspective of urban–rural relations and insist on promoting the rational allocation of urban and rural factors with the focus on integrated urban–rural development. On the one hand, full play should be given to the key role of the market and government in resource allocation. The government formulates the long-term development plan of the homestay industry, and provides preferential policies to new business types, such as rural homestays and characteristic inns, focusing on solving the supporting problems of relevant infrastructures in rural areas. At the same time, the future direction of local industrial development is guided by market forces under the macro control of the government, so as to realize the free flow of factor resources, such as land, capital, and talents, in urban and rural areas. On the other hand, we should formulate homestay standards and management measures; standardize the business scale, construction facilities, fire safety, and operation management; strengthen quality control, business operation, safety, and environmental protection; and improve the standards and grades of homestays in rural areas.

4.3. Research Limitations

This paper comprehensively analyzed the development pattern and influencing factors of homestays in Chongqing, which is of practical significance for a greater understanding of the characteristics of the development process of the homestay industry. However, due to the limitations of data openness, this paper did not study the long-time series data of homestay distributions, and the data source channels lacked diversity. At the same time, it lacked the subdivision of homestay types. Different types of homestays certainly need different development conditions, and their spatial distribution laws will also be different. In the follow-up work, we will expand the model empirical and differential mechanism analyses of different types of areas, while simultaneously expanding the research on the spatial and social effects of homestay businesses, so as to identify the problem areas and regional problems of industry development, so as to better serve the urban–rural integration and rural revitalization of different types of areas and provide targeted and systematic reference information.

5. Conclusions

In this study, the nearest-neighbor index spatial analysis method was used to calculate the distribution types of homestays in Chongqing and different regions. The results were less than one, indicating that the distribution types of homestays in Chongqing and various regions were agglomeration types. At the same time, the kernel density estimation spatial analysis method was used to study the spatial distribution characteristics of homestays in Chongqing. It was observed that in addition to the high concentration in the central urban area, there were also three secondary high-density areas around the central urban area, and there was a trend of concentration and contiguity. The spatial distribution densities in Southeast Chongqing and Northeast Chongqing were very low, and the overall distributions were extremely uneven.

Multiple regression and geographically weighted regression analyses showed that the factors, such as tourism resource endowment, economic development, service industry development, traffic location, consumption demand, and social development conditions, had significant impacts on the distribution patterns of homestays, and the impacts of each factor on the layout of homestays had an obvious spatial heterogeneity. The central urban area and the new area of the main city had a high level of economic development and a good foundation for tourism development. The intensity of various influencing factors in the region was relatively strong, among which the urbanization rate was the strongest. The economic foundation of the urban agglomeration in the Three Gorges Reservoir area of Northeast Chongqing and the urban agglomeration in Wuling Mountain Area in Southeast Chongqing were relatively weak, and the tourism reception capacity of most districts and counties was weak. Although they have great advantages in ecological environment quality and tourism resource endowment, the development of the homestay industry is still relatively lagging behind.

The spatial characteristics of Chongqing’s homestays are the result of a combination of factors driven by natural endowments, transport, commerce, population levels, and tourism resources. Location is the basis for the concentration of urban homestays. In terms of the spatial layout, urban homestays can be chosen in areas that are central to the city and have more convenient transportation and shopping facilities. In terms of resource allocation, emphasis should be placed on the construction of transport and commercial facilities, which will facilitate the clustering and development of urban homestays. Regional branding is a booster for homestay clustering, especially for the rural homestay type. However, the establishment of a regional brand is a long-term process, and local governments should focus on the excavation and protection of local culture and on brand promotion strategies. Tourism resources are an early driver of homestay agglomeration. Under the agglomeration model of relying on natural attractions, emphasis should be placed on the development of tourist attractions and the improvement of the surrounding infrastructure, with targeted resource investment to maximize resource allocation efficiency. Road transport is a guarantee for the clustering and outward diffusion of homestays. In terms of the clustering and development of the homestay industry, the road transport network should be deployed in an integrated manner to strengthen the docking of reduced and heavy traffic and provide more convenient travel conditions for tourists.

The development of homestays in China has the general characteristics of tourism micro- and small enterprises, such as a reliance on tourism resources and tourism markets, limited capital and viability, privatization development goals, externalities, and agglomeration development trends. It is also characterized by the uniqueness of the stage of social development and regional culture, such as the diversified and differentiated development of homestays under the influence of globalization, the injection of cultural capital brought about by the emerging urban middle class moving to the countryside, the new investment and financing instruments accelerating the complete commercialization of homestays, and the intervention of government power in the development of the homestay industry under the new policy and economic context. The homestay industry has become an experimental segment in the structural reform of the supply side of tourism in some parts of China and a spatial carrier for the revitalization of the countryside, the re-creation of farming, the integration of agricultural tourism, and the integrated development of urban and rural areas. It can provide a reference of experience for the development of the tourism accommodation industry in less developed areas in general.