Nature’s Neighborhood: The Housing Premium of Urban Parks in Dense Cities

Abstract

Urban parks, a core component of urban landscapes, play a crucial role in mitigating the negative effects of rapid urbanization and achieving sustainable urban development. In densely populated urban environments, providing urban parks fairly and efficiently, taking social preferences into account, is an important challenge. In this regard, we take Hangzhou, China, as an example and use a hedonic pricing model with a difference-in-differences estimator to test the effect of different types of urban parks on housing prices, quantify their respective economic values, and reflect homebuyers’ preferences. The findings indicate that the construction of new urban parks leads to an overall increase in the value of the surrounding real estate to some extent. Specifically, the construction of comprehensive parks and large parks significantly enhances the value of nearby housing, while proximity to new medium-sized parks also results in a housing price premium, albeit to a lesser extent. In contrast, community parks and specialized parks have a less significant impact on housing prices. These findings provide valuable insights for equitable urban development and planning, optimizing the allocation of urban parks and determining investment priorities for different types of parks to enhance the sustainability of the urban environment and human well-being.

1. Introduction

Currently, more than half of the global population resides in urban areas, and it is projected that by 2050 the proportion of urban dwellers will reach 66% [1]. Urban parks offer a wide range of benefits to residents, including reducing air pollution, lowering noise levels, mitigating the urban heat island effect, and providing both physical and mental health advantages [2,3,4,5,6,7]. With the outbreak of the COVID-19 pandemic, people are paying more attention to outdoor activities and social interactions, which further highlights the critical role of urban parks in supporting residents’ well-being [6]. In addition, the green landscapes of parks are inherently attractive, and the recreational facilities they offer cater to people’s pursuit of leisure [8]. Consequently, urban parks are increasingly influencing residents’ location decisions, with their “implicit value” being reflected in housing prices [9,10]. These attributes provide nearby communities with highly valued environmental and recreational elements, potentially adding an additional premium to property values [11].

The introduction of the hedonic pricing model has made housing price data a key basis for evaluating the economic value of green spaces, and scholars have extensively investigated the impact of urban green spaces on real estate prices [12,13,14,15]. Nonetheless, research dedicated to specific green spaces, such as urban parks, remains limited. Much of the existing research focuses on the impact of a particular large or high-profile urban park on neighboring house prices. In China, much of the research has centered around parks built specifically for major international events, such as the Olympic Forest Park, which was constructed when Beijing hosted the Olympic Games in 2008, and the Shanghai Expo Waterfront, which was built when Shanghai hosted the World Expo in 2010 [16,17,18]. Many similar studies have been conducted in other countries, for example, a study from the United States specifically assessed the economic value of proximity to two prominent parks, Saguaro National Park West and Tucson Mountain Park [19]. However, it is difficult to apply the results of such studies of unique individual situations to provide comprehensive and supportive information for urban green space planning and subsequent intervention management.

In addition to the single-park scale of research, in more studies, scholars have compiled comprehensive urban green space data to be analyzed from an overall city-wide perspective. Such studies tend to interpret the capitalized value of urban green space in an overly broad manner, and in several existing hedonic pricing studies, different types of green spaces within cities, including urban parks, are categorized and analyzed to assess their effect on property values [20]. For example, a study conducted in Aalborg, Denmark, classified urban green spaces into eight categories, including parks, lakes, nature areas, churchyards, sports fields, common areas, agricultural fields, and green buffers, and analyzed the heterogeneity of the capitalized value of these green spaces [20]. These studies have generally demonstrated the ability of parks to inflate nearby property values; however, in such a unified perspective, urban green space has been viewed as a homogeneous commodity [8,21,22,23,24,25]. These studies can reveal the heterogeneity of different types of urban green spaces from a capitalized value perspective, but they may overlook the internal heterogeneity of urban parks themselves.

Indeed, within the category of urban parks as green spaces alone, the various types of parks may have different potentials to influence house prices, and these are worth specializing in. The division into various types of urban parks often implies that they differ in size, function, quality, and characteristic attributes and that people are able to perceive these differences, which can have a heterogeneous premium effect on house prices [24,26,27,28].

Compared to developed countries, the role of urban parks is particularly significant in developing countries. Developing nations often have higher population densities and face more rapid urbanization. This rapid growth not only further reduces the living space for urban residents but also presents severe environmental challenges, with the reduction in green spaces exacerbating these risks [29,30]. Therefore, the environmental and health benefits provided by urban parks are especially crucial in developing countries [4,31]. For example, in China, the world’s largest developing country, rapid economic growth has led to high levels of urban population concentration and environmental pollution. The government has increasingly emphasized urban green spaces in city planning, such as prioritizing the construction of greenbelts [32]. However, China’s per capita green space remains lower than that of developed countries. Thus, in developing countries like China, the effective allocation of urban parks is vital for balancing economic growth with environmental protection and improving the quality of life for urban residents.

However, under the constraints of economic pressure and limited urban land resources, it is challenging for developing countries to plan large-scale urban parks [33]. Existing research evidence suggests that people perceive and evaluate parks based on the services they provide, which also determines their frequency of park visits [34,35]. It is clear that urban residents differentiate between various types of parks; therefore, such distinctions should also be made in studies of park valuation. Identifying the heterogeneous impact of different types of parks on property values can assist policymakers in effectively allocating urban parks within the constraints of limited resources [36].

Against this background, we take the internal heterogeneity of urban parks as a starting point and conduct a hedonic pricing analysis based on the careful classification of parks. According to officially published standards, we classify urban parks twice—based on function and size—ultimately forming five types of parks. These categories reflect the different services urban parks can provide. In line with the theoretical concept of housing units, each individual housing attribute contributes to the formation of property prices [13,37]. Therefore, we hypothesize that different types of urban parks can be viewed as housing attributes, each exerting a distinct influence on property values. By identifying the heterogeneous capitalized value of each park type, we can not only assist policymakers in making informed decisions regarding park planning and resource allocation but also guide the equitable distribution of parks, addressing issues of social equity and environmental justice.

The paper is organized as follows: Section 2 introduces the study area. Section 3 describes the hedonic pricing method employed, the classification of urban parks, the data sources, and the definition of variables. Section 4 presents the estimation results and conducts robustness checks. Finally, Section 5 discusses the key findings, the limitations, and policy recommendations.

2. Study Area

This study focuses on the central urban area of Hangzhou, the capital of Zhejiang Province, China, encompassing five districts: Gongshu, Xihu, Binjiang, Shangcheng, and Qiantang (Figure 1). We believe that the real estate market in Hangzhou is both large and mature enough to reveal buyers’ preferences for various attributes when purchasing housing. The study period spans from 2015 to 2023, a time of rapid development in Hangzhou. During this period, urban construction surged, with the built-up area expanding from 551 km² in 2015 to 801 km² in 2022. Simultaneously, the city’s population grew from 9.01 million in 2015 to 12.52 million in 2023 [38,39]. This explosive pace of urbanization and the increasing population have significantly impacted the city’s environmental quality. In response, the Hangzhou government has been implementing proactive policies to combat environmental degradation, including large-scale greening initiatives aimed at restoring lost green spaces and enhancing urban beautification, which has also spurred the rapid development of urban parks [40,41]. Given the limited area available for urban parks amidst the city’s rapid expansion, optimizing the spatial configuration of these parks to maximize their benefits is a prudent choice. This necessitates an assessment of the economic value of different types of urban parks. Consequently, future urban planning in Hangzhou urgently requires monetized information on the benefits of green spaces to formulate strategies that harmonize urban development with green space conservation.

We selected 11 urban parks in Hangzhou’s central urban area that were built between 2017 and 2021, have clearly defined opening dates, and are surrounded by at least 15 residential complexes as the study sites (Figure 1). Ultimately, 451 residential complexes within a 3 km radius of these 11 newly constructed urban parks were selected as the subjects for housing price data collection. These residential complexes are all located within urban areas, close to roads, typically constructed by a single real estate developer, and feature varying infrastructure and public services [42]. The 3 km spatial extent is derived from the Chinese industry standard Urban Green Space Classification Standard, which determines the maximum service radius of park green space as 3 km [43].

3. Methodology and Data

3.1. A Hedonic Pricing Model with a Difference-in-Differences Estimator

Urban parks, as public goods, provide valuable public services to city residents. However, public goods do not have a direct market or price, making it challenging to measure their implicit value in housing [44]. The implicit value of urban parks in housing is commonly referred to as their “capitalization effect” in housing prices. To validate the capitalization effect of urban parks, this study employs a hedonic pricing model with a difference-in-differences estimator [45]. The hedonic pricing model is a non-market valuation technique that explains the price of goods or services [27,28]. However, when using this model to examine the causal relationship between green space and housing prices, researchers often encounter issues of spatial autocorrelation, which can lead to biased results [46]. This occurs because the sale price of a property is often influenced by the prices of surrounding properties. Employing spatial fixed effects, spatial lag models, or spatial error models can help correct for the influence of spatial autocorrelation.

Using a hedonic pricing model, the capitalization effect of urban parks on housing prices can be identified. Building on this, a difference-in-differences (DID) method is applied, comparing housing price samples from an experimental group within the influence zone of urban parks with those from a control group outside this zone. This analysis yields the net effect of newly constructed parks on nearby housing prices. We set an analysis time window of eight quarters before and after the construction of the parks, spanning from Q4 2015 to Q4 2023. Additionally, the treatment group consists of housing samples within a 1.2 km radius of the urban parks, while the control group includes housing samples within a 1.2–3 km radius. The 1.2 km threshold distance is determined based on previous studies [47]. This study hypothesizes that urban parks have a positive impact on housing prices within a 1.2 km radius. All regressions were conducted in Stata 16, and the regression model is specified as follows:

$$\begin{gathered} Ln\left({HP}_{i,t}\right)={\alpha }_0+{\alpha }_{1}DIN\_PARK+{\alpha }_{2}AFTER+{\alpha }_{3}DIN\_PARK\times AFTER+{\alpha }_{4}Xh+{\alpha }_{5}Xa+{\alpha }_{6}Xz+{\alpha }_{7}POP \\ +{\alpha }_{8}E+{\alpha }_{9}LS+{\alpha }_{10}POL+{\epsilon }_{i,t}+{\phi }_i+{\phi }_t \end{gathered}$$

(1)

where $Ln \left(HP\right)$ represents the natural logarithm of housing prices. DIN_PARK indicates whether a particular housing price falls within the spatial influence range of an urban park (distinguishing the treatment group from the control group). AFTER indicates whether the urban park has been constructed. The interaction term DIN_PARK×AFTER captures the additional change in housing prices for the treatment group relative to the control group before and after the park’s construction. Xh, Xa, and Xz represent variables related to structure characteristics, neighborhood characteristics, and location characteristics, respectively. POP, E, LS, and POL are variables set to control for regional development conditions within the city (see

Table 1. Statistical description of variables.

No. of Observations: 43,650
Variables	Descriptions	Mean	Min	Max	S.D.
Dependent variable
$Ln\left({HP}_{i,t}\right)$	Log of housing price	10.429	8.332	11.513	0.444
Urban park variable
DIN_PARK×AFTER	The additional housing price changes in the experimental group relative to the control group before and after the construction of the urban park (dummy)	0.124	0	1	0.329
DIN_PARK×AFTER×TYPEn	The additional housing price changes in the experimental group relative to the control group before and after the construction of a certain type of urban park (dummy)
TYPE1	Comprehensive park	0.032	0	1	0.176
TYPE2	Theme park	0.073	0	1	0.260
TYPE3	Community park	0.055	0	1	0.227
TYPE4	Medium park	0.067	0	1	0.250
TYPE5	Large park	0.095	0	1	0.293
Apartment-level variables
Structure characteristics (Xh)
Xh1	Parking ratio	1.031	0.02	93.8	4.506
Xh2	Green rate	0.307	0.1	0.7	0.071
Xh3	Plot ratio	2.137	0.4	4.8	0.664
Xh4	Building age	14.306	1	48	7.442
Neighborhood characteristics (Xa)
Xa1	No. of bus stops within 1.2 km	29.724	8	61	8.357
Xa2	Distance to the closest subway (km)	0.689	0.036	3.479	0.408
Xa3	No. of key primary schools	0.689	0	1	0.367
Xa4	No. of hospitals	0.753	0	3	0.879
Xa5	Is there a shopping center within 1.2 km (dummy: 1 = yes)	0.889	0	1	0.314
Xa6	Are there any scenic spots within 1.2 km (dummy: 1 = yes)	0.691	0	1	0.462
Location characteristics (Xz)
Xz1	Distance to West Lake (km)	6.77	0.215	21.148	4.494
Xz2	Distance to Qianjiang New City (km)	11.162	0.495	21.927	4.559
District-level variables
POP	Log of population density	8.925	7.76	9.514	0.395
E	Log of per capita GDP	12.002	11.154	13.052	0.557
LS	Log of land supply	4.928	3.745	6.151	0.659
POL	Has the housing purchase restriction policy been restarted (dummy: 1 = yes)	0.887	0	1	0.317

). ${\alpha }_0$ is the constant term, and ${\alpha }_1$ through ${\alpha }_{10}$ are the parameters to be estimated. ${\phi }_t$ represents time fixed effects, ${\phi }_i$ represents locational fixed effects, and ${\epsilon }_{i,t}$ is the random error term in the model.

Additionally, this study will further examine the differences in capitalization benefits across different types of urban parks, hypothesizing that various types of urban parks exhibit heterogeneous capitalization effects. The regressions were also conducted in Stata 16, and the regression model is specified as follows:

$$\begin{gathered} Ln\left({HP}_{i,t}\right)={\alpha }_0+{\alpha }_{1}DIN\_PARK\times AFTER+{\alpha }_{2}TYPEn+{\alpha }_{3}DIN\_PARK\times AFTER\times TYPEn+{\alpha }_{4}Xh+{\alpha }_{5}Xa \\ +{\alpha }_{6}Xz+{\alpha }_{7}POP+{\alpha }_{8}E+{\alpha }_{9}LS+{\alpha }_{10}POL+{\epsilon }_{i,t}+{\phi }_i+{\phi }_t \end{gathered}$$

(2)

where TYPEn represents the type of urban park. The interaction term DIN_PARK×AFTER×TYPEn captures the additional change in housing prices for the treatment group relative to the control group before and after the construction of a particular type of urban park. The other variable settings remain consistent with those in Equation (1), with specific definitions provided in Table 1.

3.2. Classification of Urban Parks

In the relevant literature, most researchers treat urban parks as a homogeneous good. In the few hedonic pricing studies, where urban parks are considered as heterogeneous goods, the differentiation is typically based on “objective” criteria such as park size, vegetation coverage, and green space density [28,48]. We believe that people are more likely to view urban parks as heterogeneous goods, engaging in different recreational activities based on the diverse services they provide [49]. Therefore, to further examine the differences in the capitalization value of various types of urban parks, this study categorizes urban parks accordingly (see

Table 2. Urban park types in Hangzhou distinguished in our study.

Urban Park	Characteristics and Typical Use	Quantity
Community park	The land is independent and has basic recreational and service facilities, mainly green spaces for residents within a certain community to carry out daily leisure activities and services nearby.	2
Theme park	Green spaces with a specific content or form, with corresponding recreational and service facilities, mainly including zoos, botanical gardens, historic gardens, heritage parks, and amusement parks, as well as children’s parks, sports and fitness parks, waterfront parks, commemorative parks, sculpture parks, scenic beauty parks, urban wetland parks, and forest parks.	5
Comprehensive park	A green space with rich content, suitable for various outdoor activities, and complete recreational and supporting management service facilities.	4
Medium park	Parks with an area of 2–20 ha, including 20 ha.	7
Large park	Parks with an area of over 20 ha.	4

).

The data on urban parks were sourced from Baidu Maps, through which we identified the names, locations, sizes, and construction dates of 11 newly added urban parks. First, according to the “Urban Green Space Classification Standards,” urban parks were classified by function into three categories: comprehensive parks, theme parks, and community parks [43]. Second, based on the “Urban Park Management and Facility Maintenance Manual,” parks were classified by size into two categories: large parks and medium parks [50]. The distribution of these parks is shown in Figure 2 and Figure 3.

3.3. Data

3.3.1. Apartment-Level Variables

The dependent variable in this study is housing prices, represented by the average listing price of second-hand apartments in residential complexes. The data were sourced from Anjuke, a leading real estate information service platform in China that covers 67 cities and is a prominent online real estate intermediary in Hangzhou [51]. Based on the geographical locations of all residential complexes in Hangzhou’s central urban area provided by the platform, we selected 451 complexes within a 3 km radius of newly constructed urban parks (see Figure 1). From October 2015 to October 2023, we collected the average listing prices of second-hand apartments in these complexes for a total of 97 months, resulting in 43,650 housing price data points for further analysis. It is important to note that these 451 residential complexes were selected through a screening process. First, we only included standard residential properties, excluding villas, serviced apartments, and other housing types. Additionally, residential complexes with insufficient characteristic variables were also excluded.

The selection of characteristic variables was based on data availability and prior research [28,52,53]. These variables were categorized into three groups: building characteristics, neighborhood characteristics, and locational characteristics. Among the building characteristics, the parking ratio (Xh1), green rate (Xh2), plot ratio (Xh3), and building age (Xh4) were provided by Anjuke. Additionally, this study considered the accessibility of residential complexes to various public facilities, including bus stops (Xa1), subways (Xa2), key primary schools (Xa3), hospitals (Xa4), shopping centers (Xa5), and scenic spots (Xa6). The geographical locations of these facilities were sourced from relevant departments in Hangzhou and were mapped and measured using ArcGIS 10.2. Finally, to account for the locational advantages of residential complexes within the broader urban context, we measured and controlled for two variables: the distance from the complex to West Lake (Xz1) and the distance from the complex to Qianjiang New City (Xz2). Detailed descriptions and summary statistics for all these variables are provided in Table 1.

3.3.2. District-Level Variables

We used population density (POP) and per capita GDP (E) to control for socioeconomic differences across the five districts of Hangzhou’s central urban area. Data for these variables were extracted from the “Hangzhou Statistical Yearbook” [31]. Inspired by the studies of Peng R and Wheaton W C., we also considered the impact of land supply (LS) on housing prices, with data sourced from Hangzhou’s state-owned land supply plans from 2015 to 2023 [54,55]. Lastly, government regulatory policies can affect housing prices [56]. During the study period, Hangzhou’s government reintroduced housing purchase restrictions in September 2016, affecting both new and second-hand housing, with all five districts of our research area included in the restrictions. Therefore, we defined the dummy variable POL to represent the reintroduction of this policy. Detailed descriptions and summary statistics for all variables are listed in Table 1.

4. Results

4.1. Impact of Newly Built Urban Parks on Housing Prices

The empirical analysis results of the impact of newly constructed urban parks on housing price changes are presented in

Table 3. Impact of newly built urban parks on housing prices.

	Model 1	Model 2 Controlled for Apartment-Level Variables	Model 3 Controlled for District-Level Variables
DIN_PARK×AFTER	0.0970 ***	0.0955 ***	0.0881 ***
	(0.0045)	(0.0044)	(0.0044)
Xh1		0.7850 ***	0.7674 ***
		(0.0718)	(0.0702)
Xh2		−2.3381 ***	−2.1252 ***
		(0.3577)	(0.3612)
Xh3		−0.0448	−0.0567 *
		(0.0304)	(0.0300)
Xh4		0.0077 **	0.0050
		(0.0034)	(0.0033)
Xa1		0.0198 ***	0.0191 ***
		(0.0053)	(0.0052)
Xa2		0.4723 ***	0.4264 ***
		(0.0910)	(0.0905)
Xa3		−0.1374 ***	−0.1331 ***
		(0.0360)	(0.0353)
Xa4		−0.0962 ***	−0.0859 ***
		(0.0284)	(0.0289)
Xa5		0.4412 ***	0.3863 ***
		(0.1210)	(0.1192)
Xa6		−0.1118 ***	−0.0921 ***
		(0.0358)	(0.0356)
Xz1		−0.0381 *	−0.0453 **
		(0.0212)	(0.0211)
Xz2		0.0141	0.0096
		(0.0294)	(0.0287)
POP			0.0297 ***
			(0.0042)
E			0.1631 ***
			(0.0109)
LS			−0.0104 ***
			(0.0022)
POL			−0.1349 ***
			(0.0183)
Constant	10.4172 ***	9.2775 ***	7.4229 ***
	(0.0011)	(0.3299)	(0.3504)
Dist FE	YES	YES	YES
Year FE	YES	YES	YES
Observations	43,650	43,650	43,650
R−squared	0.806	0.807	0.810

*** p < 0.01, ** p < 0.05, * p < 0.1.

. In Model 1, the fixed effects for location and time are controlled. Model 2 builds upon Model 1 by further controlling for residential complex-level variables. Model 3 further extends Model 2 by including regional-level variables. Overall, the models demonstrate strong explanatory power, accounting for up to 80% of the spatial variation in housing prices. Regardless of the inclusion of control variables, the interaction term DIN_PARK×AFTER remains significant at the 1% statistical level. When controlling for other factors, the estimated coefficient for DIN_PARK×AFTER is 0.0881, indicating that the opening of a new park leads to an 8.81% increase in housing prices within a 1.2 km radius.

4.2. Impact of Different Types of Newly Built Urban Parks on Housing Prices

The empirical analysis results regarding the impact of different types of newly constructed urban parks on housing price changes are presented in

Table 4. Impact of different types of newly built urban parks on housing prices.

	Model 4 Comprehensive Park	Model 5 Theme Park	Model 6 Community Park	Model 7 Medium Park	Model 8 Large Park
DIN_PARK×AFTER×TYPE1	0.0950 ***
	(0.0121)
DIN_PARK×AFTER×TYPE2		0.0046
		(0.0100)
DIN_PARK×AFTER×TYPE3			0.0031
			(0.0084)
DIN_PARK×AFTER×TYPE4				0.0176 *
				(0.0092)
DIN_PARK×AFTER×TYPE5					0.0848 ***
					(0.0070)
DIN_PARK×AFTER	0.0689 ***	0.0852 ***	0.0869 ***	0.0776 ***	0.0256 ***
	(0.0046)	(0.0095)	(0.0051)	(0.0085)	(0.0053)
Constant	7.0959 ***	7.4029 ***	7.4393 ***	7.2304 ***	7.4859 ***
	(0.3636)	(0.3546)	(0.3514)	(0.3604)	(0.3484)
Dist FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
Observations	43,650	43,650	43,650	43,650	43,650
R−squared	0.810	0.810	0.810	0.810	0.810

*** p < 0.01, * p < 0.1.

. This analysis controls for both residential complex-level and administrative district-level variables influencing housing prices. Overall, the models exhibit strong explanatory power, accounting for up to 80% of the spatial variation in housing prices.

In Models 4, 5, and 6, we analyzed the impact of newly constructed parks with different functions on housing price changes. Firstly, Model 4 evaluates the effect of newly built comprehensive parks on housing prices. The statistically significant empirical results indicate that new comprehensive parks have a notable and positive effect on residential prices within a 1.2 km radius. Specifically, the construction of a comprehensive park within this radius leads to a 9.5% increase in residential sale prices. Model 5 assesses the impact of newly built specialized parks on housing prices. The estimated results show a positive effect of new specialized parks on housing prices within a 1.2 km radius, but the effect is not statistically significant. Model 6 examines the impact of newly constructed community parks on housing prices. The results indicate a positive effect of new community parks on housing prices within a 1.2 km radius, but the effect is not statistically significant. Comparing the results from Models 4, 5, and 6 reveals that the three types of parks have heterogeneous impacts on housing prices.

Models 7 and 8 evaluate the relationship between the size of newly added parks and housing price changes. In Model 7, the empirical results show that newly built medium-sized parks have a significant and positive effect on residential prices within a 1.2 km radius. Specifically, the construction of a medium-sized park within this radius results in a 1.76% increase in housing prices. In Model 8, the empirical results show that newly built large parks have a significant and positive effect on residential prices within a 1.2 km radius. The construction of a large park within this radius leads to an 8.48% increase in housing prices. Comparing the results from Models 7 and 8, it is evident that large parks generate a greater increase in housing prices compared to medium-sized parks within a 1.2 km radius. See Appendix A for full regession results (see

Table A1. Impact of different types of newly built urban parks on housing prices (full regression results).

	Model 4 Comprehensive Park	Model 5 Theme Park	Model 6 Community Park	Model 7 Medium Park	Model 8 Large Park
DIN_PARK×AFTER×TYPE1	0.0950 ***
	(0.0121)
DIN_PARK×AFTER×TYPE2		0.0046
		(0.0100)
DIN_PARK×AFTER×TYPE3			0.0031
			(0.0084)
DIN_PARK×AFTER×TYPE4				0.0176 *
				(0.0092)
DIN_PARK×AFTER×TYPE5					0.0848 ***
					(0.0070)
DIN_PARK×AFTER	0.0689 ***	0.0852 ***	0.0869 ***	0.0776 ***	0.0256 ***
	(0.0046)	(0.0095)	(0.0051)	(0.0085)	(0.0053)
Xh1	0.7632 ***	0.7686 ***	0.7678 ***	0.7726 ***	0.7770 ***
	(0.0674)	(0.0702)	(0.0701)	(0.0717)	(0.0693)
Xh2	−2.2648 ***	−2.1364 ***	−2.1235 ***	−2.2391 ***	−2.2038 ***
	(0.3631)	(0.3627)	(0.3745)	(0.3608)	(0.3708)
Xh3	−0.0657 **	−0.0565 *	−0.0567 *	−0.0631 **	−0.0572 *
	(0.0282)	(0.0299)	(0.0306)	(0.0301)	(0.0294)
Xh4	0.0059 *	0.0051	0.0050	0.0054	0.0055
	(0.0033)	(0.0033)	(0.0034)	(0.0033)	(0.0033)
Xa1	0.0167 ***	0.0191 ***	0.0191 ***	0.0180 ***	0.0192 ***
	(0.0050)	(0.0052)	(0.0053)	(0.0052)	(0.0052)
Xa2	0.4285 ***	0.4280 ***	0.4261 ***	0.4302 ***	0.4314 ***
	(0.0839)	(0.0905)	(0.0908)	(0.0903)	(0.0889)
Xa3	−0.1656 ***	−0.1338 ***	−0.1326 ***	−0.1501 ***	−0.1403 ***
	(0.0360)	(0.0353)	(0.0370)	(0.0363)	(0.0349)
Xa4	−0.0876 ***	−0.0866 ***	−0.0860 ***	−0.0865 ***	−0.0914 ***
	(0.0273)	(0.0290)	(0.0290)	(0.0283)	(0.0302)
Xa5	0.4307 ***	0.3874 ***	0.3842 ***	0.4068 ***	0.3852 ***
	(0.1165)	(0.1191)	(0.1203)	(0.1187)	(0.1178)
Xa6	−0.1154 ***	−0.0932 ***	−0.0916 **	−0.1083 ***	−0.1000 ***
	(0.0369)	(0.0356)	(0.0376)	(0.0367)	(0.0353)
Xz1	−0.0317	−0.0446 **	−0.0453 **	−0.0352	−0.0382 *
	(0.0223)	(0.0211)	(0.0226)	(0.0217)	(0.0209)
Xz2	0.0307	0.0098	0.0090	0.0178	0.0105
	(0.0284)	(0.0286)	(0.0290)	(0.0296)	(0.0279)
POP	0.0333 ***	0.0300 ***	0.0295 ***	0.0295 ***	0.0264 ***
	(0.0041)	(0.0043)	(0.0041)	(0.0042)	(0.0041)
E	0.1687 ***	0.1638 ***	0.1626 ***	0.1643 ***	0.1660 ***
	(0.0109)	(0.0111)	(0.0111)	(0.0109)	(0.0109)
LS	−0.0076 ***	−0.0101 ***	−0.0105 ***	−0.0092 ***	−0.0094 ***
	(0.0023)	(0.0023)	(0.0023)	(0.0023)	(0.0022)
POL	−0.1362 ***	−0.1350 ***	−0.1346 ***	−0.1369 ***	−0.1342 ***
	(0.0164)	(0.0183)	(0.0184)	(0.0181)	(0.0177)
Constant	7.0959 ***	7.4029 ***	7.4393 ***	7.2304 ***	7.4859 ***
	(0.3636)	(0.3546)	(0.3514)	(0.3604)	(0.3484)
Dist FE	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
Observations	43,650	43,650	43,650	43,650	43,650
R−squared	0.810	0.810	0.810	0.810	0.810

*** p < 0.01, ** p < 0.05, * p < 0.1.

).

4.3. Robustness

4.3.1. Parallel Trend Test

According to the parallel trend assumption, if the housing price changes for the treatment group (residential samples within 1.2 km of the new park) and the control group (residential samples within 1.2–3 km of the new park) exhibit similar trends before the park’s opening, then the difference-in-differences estimates are considered valid. We set the month of the park’s opening as period 0, with periods −1 and 1 representing the months before and after the park’s opening, respectively. Due to the long time span of our housing price data, which cover 97 months, we only display the 10 months prior to and 15 months following the park’s opening, without affecting the validity of the test.

The results of the parallel trend test (see Figure 4) indicate that in the 10 periods before the opening of the new urban park, the coefficients of the control group’s dummy variables do not show significant differences from those of the treatment group, which supports the parallel trend assumption. Additionally, we observe a lagged effect of the park’s opening on housing prices, as significant changes in prices occur five months after the park’s opening (see Figure 4).

4.3.2. Placebo Test

To further assess the extent to which the regression results might be affected by omitted variables, we conducted a placebo test following the method of Li [57]. We randomly generated a “pseudo-treatment group” from the sample of residential complexes and created a dummy variable for this pseudo-treatment group. Next, we randomly selected a month from the sample’s park opening interval to serve as the pseudo-event occurrence time and generated a pseudo-event dummy variable. The number of pseudo-treatment groups was the same as the actual number, and we re-ran the regression Equation (1). This data generation and regression process was repeated 500 times.

Figure 5 shows the distribution of estimates from the 500 regression runs, where most pseudo-policy dummy variables are greater than −0.091, and most p-values are greater than 0.1, indicating that the randomly constructed variable does not have a significant impact on housing prices. Therefore, our regression results have passed the placebo test.

5. Discussion and Conclusions

5.1. Key Findings

This study estimates the capitalization value of newly constructed urban parks in the central urban area of Hangzhou, using a hedonic pricing model with a difference-in-differences approach to measure the impact of new urban parks on surrounding real estate prices. Additionally, by categorizing these new parks, we further analyze the heterogeneous effects of proximity to comprehensive parks, specialized parks, community parks, medium-sized parks, and large parks on real estate prices. Finally, our estimates have also passed robustness checks.

Firstly, the findings on the capitalization value of newly constructed urban parks indicate that real estate prices closer to urban parks are higher. Specifically, housing prices within 1.2 km of a park have increased by 8.81% compared to those located within 1.2–3 km of the park. Our results are consistent with those of previous studies. For instance, Crompton’s research highlights a 20% positive impact on property values due to proximity to urban parks [26]. Trojanek et al., using hedonic pricing methods along with OLS, GLS, and QR models, estimated that urban green spaces within 100 m of housing result in a 3–4% increase in property prices [58]. Similarly, Liu and Chen’s study in Urumqi, China, found that for every 1 km reduction in distance to a park, housing prices increase on average by RMB 21,840 [59].

Secondly, consistent with expectations, our research reveals differences in the capitalization value of different types of urban parks. Among the three categories of urban parks classified by function, we find that newly constructed comprehensive parks have a higher capitalization value, contributing to a 9.5% increase in housing prices within 1.2 km. Conversely, the results show that specialized parks and community parks do not have a significant impact on surrounding housing prices. A possible explanation is that different park functions imply variations in the natural environment, quality, and services, which provide varying degrees of health benefits to residents [60]. Although each park type serves multiple functions, their primary functions differ, and thus, from both ecological perspectives (like area, vegetation, and water bodies) and cultural perspectives (like tranquility and cultural value), they offer different services and experiences to urban residents [60,61,62]. Comprehensive parks have a more balanced and diverse range of functions, typically featuring a wider array of facilities, more carefully designed spaces, and more sophisticated maintenance practices. They provide socioeconomic value from multiple perspectives, including ecology, education, tourism, and disaster prevention. As a result, comprehensive parks are better positioned to enhance the environmental quality of nearby communities and meet the varied needs of residents, thereby increasing local property values. In contrast, community parks are usually smaller in size and contain fewer ecological elements. Their primary purpose is to provide recreational spaces for the elderly and children. Additionally, community parks often lack the serene environment of comprehensive parks, which can impact the health benefits they offer to residents [60]. Similarly, specialized parks, defined by their specific themes and the particular activities and services they provide, tend to have a more singular function compared to comprehensive parks, which may affect their capitalization value. We also found that the impact of park types on housing prices varies across countries. Studies from Europe and North America suggest that community parks and linear parks have higher capitalization values. For example, research by Espey and Owusu-Edusei in South Carolina found that community parks have the greatest impact on housing prices [63]. Hobden et al. analyzed the effect of different types of green spaces on housing prices in Surrey, emphasizing the capitalization value of linear green spaces [64]. However, in China, previous studies indicate that the capitalization value of community parks and linear parks tends to be lower [65]. This discrepancy may stem from differences in the value that urban residents in various countries place on the benefits provided by parks.

Among the two categories of urban parks classified by size, we find that newly constructed large parks have a significantly greater positive impact on housing prices compared to medium-sized parks. Large parks contribute to an 8.48% increase in housing prices within a 1.2 km radius, whereas medium-sized parks result in only a 1.76% premium for nearby housing within the same distance. The significant positive effect of large parks on real estate prices has been demonstrated in numerous hedonic pricing studies [25,66]. For instance, Yang investigated the impact of park size and proximity on housing prices, concluding that the closer and larger the park, the higher the property values [67]. Similarly, research on green gentrification indicates that larger parks are more likely to lead to gentrification, while smaller parks do not [68,69]. The likely explanation is that large urban parks typically offer more comprehensive facilities and a greater range of functions [70]. The diverse landscapes, amenities, and enhanced management attract a broader range of users (children, teenagers, adults, and elderly) [71]. As a result, properties located near large parks tend to command a higher premium.

Finally, in the model, the variable Xh4 which is the building age shows a significant positive effect on house prices. This is contrary to previous studies. Here, we try to give a possible interpretation of it. In recent years, China has devoted a great deal of attention and substantial financial support to the upgrading of old residential areas. The scope of the renovation covers residential areas built before 2005. As a model city for the transformation of old neighborhoods, the Hangzhou Municipal Government issued the “Four-Year Action Plan for Comprehensive Transformation and Upgrading of Hangzhou Old Neighborhoods (2019–2022)” in 2019 [72]. By the end of 2023, a total of 1349 renovations of old neighborhoods had been implemented, of which 1152 have been completed [73]. These renovations involve all aspects of residential areas, including infrastructure and public services. The renovation work has greatly enhanced the living environment of old residential areas and accordingly strengthened their market value. In addition to the improved living environment, the more complete supporting facilities in the vicinity of older neighborhoods are also an important reason for their higher market value. Older neighborhoods are often surrounded by richer educational resources. At the same time, basic amenities are also readily available. These conditions can provide residents with a more convenient life, making the purchase of real estate in older neighborhoods the choice of many people.

5.2. Potential Policy Implication

Our findings provide valuable insights for developing more targeted and equitable urban planning strategies. In China, in response to central directives on green development, local governments have made varying degrees of effort to develop urban parks. For example, according to the “Hangzhou Green Space System Special Plan (2021–2035),” the city aims to build over 480 parks by 2035 [74]. Given the large-scale demand for new parks, the estimation of capitalization values for different types of urban parks presented in this study can be used for cost–benefit analysis, the pre-evaluation of park projects, and providing references for optimizing urban park allocation models.

Additionally, our findings highlight the importance of considering the capitalization value of urban parks in planning. Local governments often plan urban parks with the intention of allowing residents to equally benefit from them. However, the resulting increase in property values can lead to the displacement of lower-income residents, ultimately resulting in these new green spaces serving primarily wealthier individuals [75]. This presents a significant challenge for urban planners. According to our results, newly constructed comprehensive and large parks significantly capitalize into the housing market, while newly built community and specialized parks do not affect housing prices. This finding is significant as it can assist urban planners in mitigating the green gentrification effects by improving the allocation of urban parks while still achieving the intended planning goals.

Finally, the heterogeneous capitalization value of different types of parks also reflects the demands and preferences of urban residents. To maximize the benefits of urban parks and enhance residents’ health and well-being, it is crucial to consider the needs of the wider community. Decisionmakers should aim to provide urban green spaces that meet residents’ needs and involve them in the planning process to ensure they have the opportunity to clearly express their preferences [69].

5.3. Research Limitations

This study has limitations that should be addressed in future research.

First, due to data constraints, our analysis excluded two park types—small parks and pocket parks—from the examination of park types’ impact on capitalization value. Pocket parks, a concept first introduced by landscape designer Robert Zion in 1963, differ from traditional urban parks in their smaller size, diverse forms, flexible location, high usage, and good accessibility [76]. In China, pocket parks generally refer to parks ranging from 0.04 to 1 hectare, including small playgrounds and micro-green spaces [41]. As low-cost green infrastructure suitable for high-density urban areas, pocket parks have received considerable attention and investment in Hangzhou in recent years [77]. Future research should further explore the impact of pocket park development on real estate values.

Second, in addition to categorizing the parks for discussion, the specific characteristics of the parks themselves may also have an impact on housing prices. Future research should incorporate richer research methods to explore the various characteristics of parks that influence house prices.

Finally, our consideration of control variables is still insufficient. For example, in our study interval, Hangzhou hosted an active event, the G20 summit [78]. The Hangzhou municipal government placed great importance on this major international event, investing heavily in urban and road infrastructure prior to the summit. Previous studies have shown that transportation infrastructure is one of the key drivers of the real estate market [79]. Moreover, significant economic investment in the host city often stimulates investment and consumption growth, drives industrial restructuring, and fosters economic development, which can lead to fluctuations in housing prices [80,81]. Therefore, the occurrence of major events in such cities should be considered as an important variable affecting house prices in future studies.