Environmental amenities of urban rivers and residential property values: A global meta-analysis

Science of the Total Environment 693 (2019) 133628

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Review

Environmental amenities of urban rivers and residential property values: A global meta-analysis Wendy Y. Chen ⁎, Xun Li, Junyi Hua Department of Geography, The University of Hong Kong, Pokfulam Road, Hong Kong

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• We conducted a global meta-analysis about urban river's amenity. • River view is associated with the greatest premium on housing price. • River restoration has driven up residents' environmental perception towards urban rivers. • Homeowners' valuation of urban rivers is not sensitive to macro-geographical locations.

a r t i c l e

i n f o

Article history: Received 7 May 2019 Received in revised form 26 July 2019 Accepted 26 July 2019 Available online 27 July 2019 Editor: Sergi Sabater Keywords: Urban river Hedonic price method Environmental amenity/disamenity Meta-analysis Global scale

⁎ Corresponding author at: Pokfulam Road, Hong Kong. E-mail address: [email protected] (W.Y. Chen).

https://doi.org/10.1016/j.scitotenv.2019.133628 0048-9697/© 2019 Elsevier B.V. All rights reserved.

a b s t r a c t Environmental amenities and disamenities of urban rivers and their capitalization in property prices have been a major subject of empirical investigation in the hedonic price method (HPM) literature for several decades. Primary studies across the globe have nonetheless adopted varying valuation scenarios and modelling approaches. And systematic variation has been shown in homeowners' marginal willingness-to-pay (WTP) for urban rivers' amenities and disamenities, ranging between −12.2% and 63.58% price premium. To identify which valuation scenarios, socio-economic variables, and modelling characteristics might affect the quantification of urban rivers' impacts on property values, we conducted a very first meta-analysis of existing evidence to extract additional information concerning the heterogeneity in WTP estimates pertaining to urban rivers' environmental amenities and disamenities. A total of 53 observations from 30 primary studies that adopted HPM to provide WTP estimates for three prominent valuation scenarios, i.e., proximity, view and water quality, were synthesized using a random effects model. Our meta-analysis results revealed several important factors in explaining the heterogeneity in empirical WTP estimates pertaining to urban rivers' environmental amenities/disamenities. First, while all three valuation scenarios could capture urban rivers' impacts on residential property values, river view was associated with the greatest premium, followed by river water quality, and river proximity the least. Second, we found that WTP estimates were significantly higher after the year of 2000, indicating the widespread and successful river restoration and rehabilitation projects in the 21st century has driven up homeowners' environmental perception and appreciation of urban rivers' amenities, especially their clear depreciation of negative environmental disamenities, to a high level. Third, our results showed that homeowners' valuation of urban rivers was not sensitive to the macro-geographical locations of their residences, suggesting a universal overall appreciation/depreciation of urban rivers across varying cultures and societies. Instead, household income level and population density should be systematically controlled if value transfer across countries is necessary. The findings of

2

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

this meta-analysis could help refine urban rivers' attributes to be incorporated into HPM studies so as to adequately quantify people's sophisticated valuation of intertwined amenities and disamenities. On the practical front, our results supported two arguments from a very utilitarian point of view. First, it appears that the visual impacts might be prioritized for river restoration projects, such as through careful revegetation of riparian areas using native species. This could harbor rich diversity of ecological functions and in the meantime maximize environmental amenities that homeowners would like to pay for. And second, cost-effective river restoration in urbanized contexts should be prioritized in densely populated areas over places with relatively low population densities. This approach might maximize the number of people who can enjoy rivers for a given budget. © 2019 Elsevier B.V. All rights reserved.

Contents 1. 2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . Method . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Data assembly . . . . . . . . . . . . . . . . . . . 2.2. Meta-analysis model specification . . . . . . . . . . 2.3. Dependent variable . . . . . . . . . . . . . . . . . 2.4. Valuation scenario variables: environmental amenity . 2.5. Methodological variables . . . . . . . . . . . . . . 2.6. Contextual variables . . . . . . . . . . . . . . . . 2.7. Econometric estimation . . . . . . . . . . . . . . . 3. Results and discussion . . . . . . . . . . . . . . . . . . . 3.1. Impacts of environmental amenity valuation scenarios . 3.2. Impacts of modelling variables . . . . . . . . . . . . 3.3. Impact of contextual factors . . . . . . . . . . . . . 4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Limitations of the study . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . References. . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

1. Introduction Urban rivers and streams as well as associated riverine environs are usually the most influential natural components and defining features of many cities where ever-increasing population are residing around the world (Lundy and Wade, 2011; Völker and Kistemann, 2013a, 2013b; Chen, 2017; Bolleter, 2018). They play crucial ecological and societal roles within urban systems (Francis, 2012), provide diverse amenities, and influence urban dwellers' well-being (Everard and Moggridge, 2012; Auerbach et al., 2014; Polyakov et al., 2017). Yet, years of unchecked industrialization and urbanization have significantly modified and deteriorated the hydro-morphology, aquascape, and quality and quantity of urban rivers worldwide (Paul and Meyer, 2001; Vörösmarty et al., 2010; Monk et al., in press). Resultantly, some environmental disamenities (e.g. irritating odor or dark color induced by water pollution) might be generated concurrently with positive amenities (e.g. recreational opportunities and aesthetic value of riparian vegetation and blue spaces) (Chen, 2017; Chen and Li, 2017). These environmental amenities/disamenities cannot be directly priced due to their non-consumptive nature. Even so, researchers have developed hedonic pricing method (HPM) to examine whether urban rivers' environmental amenities/disamenities can be explicitly capitalized into property prices in the form of selling or rental premiums/discounts. Thus the economic benefits of this important urban natural element can be quantified and investments (usually substantial) in relevant rehabilitation and restoration projects justified (Sander and Zhao, 2015; Garcia et al., 2016; Wen et al., 2017; Jarrad et al., 2018), as urban rivers continue to be the foci of both the physical and cultural landscapes of cities and represent desirable areas for urban (re)development and gentrification (Francis, 2012; Ginzarly and Teller, 2019). Based on transaction data from real estate markets, HPM decomposes overall property prices into their contributing factors, such as structural, locational, neighborhood, and environmental characteristics (Rosen,

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

2 3 3 3 4 5 5 5 5 5 6 6 7 7 8 8 8

1974; Malpezzi, 2003). It has a long history for estimating the impacts of spatially distributed environmental goods (amenities) and bads (disamenities) on residential property values (Walsh et al., 2011). There have been two prevailing valuation scenarios in HPM studies of urban rivers in the existing literature. Firstly, many research efforts have focused on estimating the benefits of urban rivers that are capitalized into the sale prices, such as recreational uses and aesthetic values that are usually captured by variables like proximity (e.g., Wen et al., 2017) and visibility (e.g., Chen and Li, 2018). And the second valuation scenario has focused on the impacts of degraded water quality of urban rivers, which can depreciate surrounding property values (Walsh et al., 2011; Chen and Li, 2017). Despite a general agreement that urban rivers' positive amenities are attributed to recreational and pleasant visual contacts (Åberg and Tapsell, 2013; Haeffner et al., 2017) and negative disamenities are associated with water pollution and associated health concerns (Cho et al., 2011; Walsh et al., 2011; Gomes and Wai, 2014; Chen, 2017), casual observation shows heterogeneity and inconsistency in the empirical estimates. For instance, some studies suggested that urban river pollution would not generate environmental disamenities as most water pollution (such as elevated levels of various chemical and microbial contaminants) is barely noticeable for homebuyers (Leggett and Bockstael, 2000) and residents may still recreate in or near polluted rivers due to the lack of substitutes in accessible areas (Gobster and Westphal, 2004), and thus potential disamenities cannot be adequately capitalized into property values. In contrast, other empirical studies have observed that the effects of river water pollution on property values were adequately perceived as negative externalities (disamenities) by urban dwellers (e.g., Poor et al., 2007; Cho et al., 2011; Netusil et al., 2014; Chen, 2017). To shed some light on the universality of HPM results pertaining to urban river's amenities/disamenities and extract additional information concerning which valuation scenarios, socio-economic variables, and modelling characteristics might lead to the heterogeneity in the

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

environmental externalities estimated from primary studies, a detailed meta-analysis of the HPM literature on urban rivers would be a useful endeavor. Meta-analysis is a quantitative synthesis and generalization of the existing primary studies containing study-specific estimates (which are conventionally referred to as the effect sizes in metaanalysis, representing the magnitude of a difference or the strength of a relationship, according to Koricheva et al., 2013), so that quantifiable assessments, rather than simple narrative review, could be generated (Cooper et al., 2009; Stanley et al., 2013; Hunter and Schmidt, 2015; Nelson, 2015). First applied in psychological study (Glass, 1976) and gradually expanded to behavioral, social, health, and economic sciences (Braden et al., 2011), meta-analysis has become an important tool for synthesizing empirical findings from multiple case studies in the field of environmental economics (Smith and Pattanayak, 2002; Nelson and Kennedy, 2009; Guignet et al., 2018). Some examples include neighborhood tree cover (Siriwardena et al., 2016a, 2016b), urban (green) open space (Brander and Koetse, 2011), wetland (Brander et al., 2006), air quality (Smith and Huang, 1995), transportation noise nuisance (Bristow et al., 2015), contaminated waste site (Braden et al., 2011), water quality (Van Houtven et al., 2007), and flood risk (Beltrán et al., 2018). To the best of our knowledge, this meta-analysis is the first one focusing on the environmental amenities and disamenities of urban rivers. Such an analysis is of timely significance in light of increasing investments in river restoration projects worldwide (Palmer and Allan, 2006; Chen et al., 2017; Jarrad et al., 2018), in that an absence of discernable effects would weaken the case for actions to promote urban river restoration and rehabilitation. Some important initiatives include those called for under Clean Water and Endangered Species Act in the United States, the Water Framework Directive in the European Union (Beechie et al., 2009), and Urban Watercourse Rehabilitation Directive in China (Ministry of Housing and Urban-Rural Development of the People's Republic of China, 2015). Moreover, a synthesis of the current literature with attention to HPM (particularly which valuation scenarios could adequately capture urban rivers' environmental amenities/disamenities) could help identify overlooked fields of inquiry, thereby assisting to chart a research agenda.

2. Method 2.1. Data assembly A six-stage process recommended by Tranfield et al. (2003) was followed to systematically identify original studies using HPM to empirically study urban rivers' environmental amenities/disamenities, from (1) scoping, (2) keyword identification, (3) evaluation of search results, (4) search criteria refinement, (5) title and abstract review, to (6) selection of articles. Firstly, two categories of most representative keywords were identified and chosen, namely “river OR watercourse OR stream OR creek” AND “hedonic”. Searches for relevant articles were conducted using relevant English academic databases such as Science Direct, Web of Science and Scopus. Google scholar was also used to conduct search. Grey literature on institutional websites was also searched, which can help reduce the influence of a potential publication bias in metaanalysis (Reynaud and Lanzanova, 2017), as peer-reviewed journal articles tent to report significant and high estimates, in comparison with working papers retrieved from grey literature (Brander et al., 2006; Stanley and Doucouliagos, 2014). The searches were limited to articles published between 1960 and 2018, as the earliest HPM applications were found in 1960s–1970s, according to Price (2017). More than 300 records from various electronic databases and their reference lists were scrutinized. Based on their abstracts, studies without any reference to any specific urban rivers or those that did not report primary economic estimations were disregarded. And then a total of 64 studies were retained for further inspection.

3

Two scholars of our team then independently screened article titles and abstracts for further consideration, and verified each other's work. Any discrepancies were resolved by discussion with a third scholar. We selected studies to be included into the meta-analysis using the following criteria. (1) Associations between any attributes of urban rivers (such as view of river, distance to river, waterfront, and water quality) and housing sale prices have necessarily to be captured using an econometrically estimated HPM. (2) HPM functions must be based on transactions of residential properties (i.e., those studies based on commercial housings and rental rates were excluded). (3) Estimates must be capable of being expressed, after standardization when needed, as a percentage change in house prices attributed to urban rivers' environmental amenities/disamenities (i.e., those studies, which provide a marginal price but no average house price in the sample is provided, were excluded). Adhering to these criteria, the number of primary studies included in the meta-analysis is whittled down to a final total of 30, which provide 53 effect sizes. This relatively low number of observations is not uncommon in meta-analysis research (e.g., Mohammad et al., 2013; Mattmann et al., 2016; Siriwardena et al., 2016a, 2016b). As pointed out by Mattmann et al. (2016), a trade-off has to be made between conceptual homogeneity of the data studied and the amount of data points available for meta-analysis. On the other hand, the relative scarcity of primary HPM studies about urban rivers' amenities/disamenities shows that this is still an under-investigated area which calls for further research. A total of 16 out of the 30 studies are from the United States, 5 from Europe, 4 from Australia, 4 from China, and 1 from Middle East region (Israel, which is then grouped into European region in the metaanalysis). In Table 1 we present all original studies included in the meta-analysis. It should be noted that there is large variation in the sample sizes (ranging from 217 to 1,011,831) used for estimating the marginal prices of environmental amenities, which might imply some disparities in the precision of effect size estimation amongst empirical studies (Brander and Koetse, 2011). 2.2. Meta-analysis model specification In the meta-analysis, we include three sets of factors pertaining to original case studies as meta-regressors, namely the environmental amenity/disamenity valuation scenarios, the contextual characteristics, and the methodological factors, so as to explain the heterogeneity in the environmental amenities generated by urban rivers. In this way, the explanatory power of the meta-analysis could be enhanced and additional information that is not included in original studies can be extracted (Stanley et al., 2013; Siriwardena et al., 2016a, 2016b; Reynaud and Lanzanova, 2017). A general meta-analysis model can be written as: P ij ¼ α 0 þ

X

βk Dij;k þ μ j þ εij

ð1Þ

where Pij, the dependent variable, refers to the i estimate of the estimate of environmental amenity obtained from a given study j, Dij, k is the meta-regressor k and βk is the parameter associated with metaregressor k, which measures its impact on the estimate of urban river's environmental amenities, α0 is the constant, μi is a study-specific effect, and εij is an error term. Meta-analysis can be performed using either a fixed-effect model or a random-effect model. In this study, the random-effect model is preferred due to the following reasons. Firstly, the fixed-effect model assumes that the true effect size is identical across primary studies, which is implausible. In contrast, the random-effect model does not

4

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

Table 1 List of HPM studies included in the meta-analysis. Publication

Study site

Sample size

Anderson and West (2006) Artell and Huhtala (2017) Bin et al. (2017) Bonetti et al. (2016) Chen and Li (2017) Chen (2017) Cho et al. (2011) Cho et al. (2006) Cohen et al. (2015) Colby and Wishart (2002) Garcia et al. (2016) Garrod and Willis (1994) Gibbons et al. (2014) Jim and Chen (2006) Mahan et al. (2000) McLeod (1984) Mooney and Eisgruber (2001) Morgan et al. (2010) Mukherjee and Caplan (2011) Netusil (2005) Orford (2002) Polyakov et al. (2017) Poor et al. (2007) Qiu et al. (2006) Rambaldi et al. (2013) Sander and Polasky (2009) Tapsuwan et al. (2015) Wen et al. (2015) Williamson et al. (2008) Wu et al. (2004)

St. Paul metropolitan area, US 24,862 Finland 2747 Martin County, Florida, US 1526 Milan, Italy 10,530 Guangzhou, China 217 Guangzhou, China 968 North Carolina, US 595 Knox County, Tennessee, US 15,500 Barkhamstead, Connecticut, US 398 Tuscon metropolitan area, US 7658 Tel Aviv metropolitan Area, Israel 826 London and Midlands, UK 2062 Nationwide, UK 1,011,831 Guangzhou, China 652 Multnomah County, Oregon, US 14,485 Perth, Australia 168 Western Oregon, US 757

1 1 1 4 4 4 1 1 1 1 1 1 1 1 2 2 1

Augusta County, Virginia, US Logan, Utah, US Portland, Oregon, US Cardiff, UK Perth, Australia Maryland, US St. Louis metropolitan area, US Brisbane, Australia Ramsey, US Murran-Darling Basin, Australia Hangzhou, China West Virginia, US Portland, Oregon, US

2 1 3 4 1 4 1 2 2 1 1 2 1

1252 46,000 30,014 700 16,553 1377 2433 3944 4918 31,706 2887 1608 14,485

No. of obs.

impose this common effect size constriction (Borenstein et al., 2010; Soon and Ahmad, 2015; Beltrán et al., 2018). Secondly, since several studies have more than one observations, we expect observations (Pij) from the same empirical study to be correlated. To account for this, we use cluster robust variance estimators to account for this nonindependence of effect sizes, as employed in recent meta-analyses (Nelson and Kennedy, 2009; Beltrán et al., 2018; Gerrish and Watkins, 2018). Particularly, the random-effect model is more appropriate, because it allows for both within and cross study variations that are attributed to some unobserved factors. In comparison, the fix-effect model assumes that there is a single common price change of houses with respect to urban rivers across all primary studies and the variations of effect sizes only come from sampling errors (Higgins and Thompson, 2004; Sirmans et al., 2006). Thus fixed-effect model tends to overlook different features of urban rivers and local housing markets, which might induce different price impacts (Klemick et al., 2018). Thirdly, a random-effect model helps produce consistent and efficient (unbiased and best possible) results, while a fixed-effect model can only yield consistent but not efficient results (Braden et al., 2011). And lastly, the Hausman test (the null hypothesis is that the preferred model is random-effect vs. the alternative the fixed-effect) is performed (Greene, 2008), justifying the use of the random-effect model (prob N χ2 = 0.973). 2.3. Dependent variable The dependent variable (Pij) in our meta-analysis is defined as a measure of relative price change-the percentage change in house prices attributed to environmental amenities/disamenities brought by urban rivers. When conducting a meta-analysis at the global scale, a critical requirement is that empirical studies included in the meta-analysis should satisfy a criterion of commodity consistency: a minimal consistency for the dependent variable across observations

(Reynaud and Lanzanova, 2017). In our present meta-analysis, this community consistency criterion has been addressed in the following five ways. (1) We include only estimates derived from HPM, and studies that apply other evaluation methods, such as travel cost method, contingent valuation method, and choice experiments, are excluded, although empirical evidence suggested that environmental amenities estimated using either stated or revealed preference methods are comparable (Mattmann et al., 2016). Focusing on HPM which measures a change in consumer Marshallian surplus, the environmental amenities/disamenities that can be capitalized in housing prices are thereby consistently constructed on the basis of the same theoretical and methodological framework (Smith and Pattanayak, 2002; Braden et al., 2011; Klemick et al., 2018). (2) As there is no theoretical guidance for the functional form of hedonic pricing models, generally four types of parametric functional forms have been widely adopted in the hedonic valuation literature, including linear, semi-log, log-log, and BoxCox specification (Taylor, 2013). To satisfy the condition that the estimated meta-analysis equation must be based on a dependent variable measuring a common effect size, two procedures are adopted to standardize the interpretation of environmental amenities. Firstly, all empirical findings have been standardized into the percentage change of house values. Following the procedures given by Beltrán et al. (2018), the effect sizes (Pij) and associated standard errors are extracted directly from the original studies when semi-log functional form is used. For linear specification, P ij ¼ α=P (α is the estimated coefficient for valuation scenario variable and P is the average selling price of the sample in the original study) and its standard error is given by sα =P (sα is the standard error of the estimated coefficient for valuation scenario variable). None of empirical studies have used either loglog or Box-Cox specifications (the transformation procedure for these functional forms can be found in Beltrán et al., 2018). Secondly, all units of measures are adjusted to the metric system (i.e., foot and mile are converted into meter and kilometer, respectively). Such a standardization procedure could ensure that a robust basis for meta-analysis is being made (DeCoster, 2009). (3) The natural logarithm of the absolute value of effect size is adopted, as this can mitigate potential heteroscedasticity (Soon and Ahmad, 2015). In this way, we can focus on the magnitude of urban rivers' impacts on residential housing values. (4) An innovative attempt in this study is that we define a dummy variable (negative) to distinguish between positive environmental amenities and negative disamenities. This variable equals 1 if the effect size is negative and zero otherwise. Then it is interacted with all explanatory variables to derive some insights about which valuation scenarios or study contextual factors might be associated with negative estimates of urban rivers' impacts. (5) For the estimates associated with river proximity, we standardize the impacts to average percentage changes of housing prices for the distance of within 150 m of urban rivers. This distance is assumed to exert significant impacts on housing values and consistent with the notion that being located immediately adjacent to an urban river (Papenfus, in press). When a dummy waterfront variable (a width defined by authors) is used in the original HPM studies, the estimated coefficient for this variable (or its standardized form as explained earlier in this section) is used directly. When a continuous variable is used to depict the distance/proximity, the average percentage change of housing price for a width of 150 m between urban river and residential houses is calculated based on the coefficient reported in the original studies.

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

2.4. Valuation scenario variables: environmental amenity A major issue of interest in this meta-analysis is which attributes of urban rivers might affect the effect sizes (WTP estimates of environmental amenities/disamenities from HPM studies). With regard to this point, three scenario variables are specified, including proximity, view, and water quality. And correspondingly three scenario variables pertaining to urban rivers' characteristics are considered in the metaanalysis (Table 2).

5

population density of each city where the primary study was conducted (sourced from online archives of local statistical data), average housing price of each studied city (provided in the original publications), and geographical zone of primary study (America, Europe/Israel, Australia, and China). Both GDP per capita and average housing prices are adjusted to 2018 USD using the international purchasing power parity (PPP) data (OECD, 2018) and the Consumer Price Index (U.S. Department of Labor, Bureau of Labor Statistics, 2018). 2.7. Econometric estimation

2.5. Methodological variables Additionally, a set of methodological variables has been used to control the impacts of hedonic pricing model specifications. Three dummy variables are selected. The first one is “spatial” (whether spatial effects are control in the HPM), Spatial effects (in the form of spatial dependence, spatial heterogeneity or both) have constituted a common issue in HPM studies (LeSage and Pace, 2009; Chen and Li, 2017). Overlook of spatial effects may produce inaccurate or even biased estimates. This dummy variable can test whether addressing for spatial effects in HPM studies could affect the estimates of urban river's environmental amenities. The second methodological variable is defined as “year2000”, denoting whether the study year is after the year of 2000. This variable could help to examine if the widespread river restoration projects with a focus on ecological dimension in the 21st century (Viswanathan and Schirmer, 2015; Lorenz et al., 2018) have affected homeowners' perception towards river's environmental condition and thus lead to changes in the estimates of rivers' environmental amenities. The third variable is “significant”, describing whether the estimates of urban rivers' amenity/disamenity are significant in the HPM, because empirical evidence suggests that publications tend to be biased towards significant estimations (Soon and Ahmad, 2015). Sample size is originally considered as a moderator variable but removed later, because the dependent variable has been weighted by the standard error which is a function of sample size, thereby implicitly taking it into account. 2.6. Contextual variables A set of contextual covariates, which are intrinsic and unaffected by researchers' judgments, has been included to capture potential influence on the effect size variation, including GDP per capita and

Three econometric models of meta-analysis are estimated. We start from a parsimonious Model I, in which only valuation scenario variables are included. Then the methodological variables are incorporated into Model II. In Model III, contextual variables are also added. In this way, the consistency of environmental amenity/disamenity variables, the impacts of methodological and contextual variables, and the overall robustness of econometric models can be affirmed. In the meanwhile, the interaction terms between explanatory variables and negative effect sizes (Negative) are included (Models I, II and III) to test in what circumstances a negative disamenity is likely to be detected. All econometric models were estimated using Stata version 14 (Stata Corp, College Station, TX, USA). 3. Results and discussion The meta-analysis results of three econometric models are presented in Table 3, which help better position us to extract additional information concerning observed heterogeneity in the reported estimates pertaining to urban river's environmental amenities across studies. The measure of the goodness-of-fit (R2) indicates that Model I can only explain 47.3% of the variation in the estimates of urban rivers' environmental amenities. And the value of R2 increases to 0.621 (62.1% explanatory power) and 0.883 (88.3% explanatory power) in Model II and Model III, respectively. This statistics suggests that the effect size considered in this study is not only affected by evaluation scenarios, but also characteristics of the study and broad contextual factors. With regard to the main effects revealed by the models, very similar and consistent results can be apparently observed for scenario variables and modelling variables when extra explanatory variables are added in stepwise fashion, indicating the robustness of our meta-analysis in general. Other goodness-of-fit criteria of these three econometric models are

Table 2 Definition and description of variables used in the meta-analysis. Variable Dependent variable P Ln(|P|)

Definition

Mean

The percentage change of house prices associated with urban rivers Natural log of the percentage change in house prices attributed to environmental amenity/disamenity of urban river

3.63 −1.21

Std. dev.

Min

Max

11.35 3.95

−12.20 −10.62

63.58 4.15

9.97 6.70 12.13 0.47 0.24 0.10 0.20

0.89 2.07 1.71 0.50 0.43 0.30 0.40

7.32 2.56 6.99 0 0 0 0

11.13 9.82 14.85 1 1 1 1

Contextual variables Ln(gdppc) Ln(popdensity) Ln(price) America Europe/Israel Australia China

Natural log of GDP per capita (2018 USD) Natural log of the population density of the study site Natural log of the average housing price of the study site (2018 USD) =1, if the study site is located in America =1, if the study site is located in Europe or Israel =1, if the study site is located in Australia =1, if the study site is located in China

Scenario variables Proximity View Quality

=1, if proximity is evaluated =1, if river view is evaluated =1, if water quality is evaluated

0.75 0.08 0.18

0.44 0.27 0.39

0 0 0

1 1 1

Modelling variables Year2000 Significant Negative

=1, if the study year is after 2000 =1, if urban rivers' environmental amenities/disamenities are significant =1, if negative environmental disamenities are reported

0.75 0.90 0.67

0.44 0.30 0.47

0 0 0

1 1 1

6

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

Table 3 Meta-analysis results. Model I Coeff. Constant Scenario variables View Quality Proximity Modelling variables Spatial Year2000 Significant Contextual variables Ln(gdppc) Ln (popdensity) Ln(price) America Europe/Israel Australia China Interaction with negative disamenities ×View ×Year2000 ×Significant ×Ln(gdppc) ×Ln (popdensity) ×Ln(price) ×America ×Europe/Israel ×Australia R2 AIC BIC ll

Model II S.E.

Coeff.

Model III S.E.

−8.347⁎⁎⁎ (0.959) −13.145⁎⁎⁎ (1.400) 10.209⁎⁎⁎ 6.491⁎⁎⁎ Base

Coeff.

S.E.

8.286

(14.190)

(1.238) 8.061⁎⁎⁎ (1.501) 4.863⁎⁎⁎ Base

(0.890) 9.952⁎⁎⁎ (0.890) 6.426⁎⁎⁎ Base

(1.052) (2.115)

0.436 2.471⁎⁎ 4.420⁎⁎⁎

(0.817) 0.524 (0.956) 2.027⁎⁎⁎ (0.918) 4.186⁎⁎

(1.919) (0.476) (1.730)

−1.448 (1.212) −1.409⁎⁎⁎ (0.488) 0.863⁎⁎⁎ 0.690 1.376 5.395 Base

1.509

(0.966) −1.636 3.585 −5.569

(2.117) −7.434⁎⁎⁎ (3.204) 3.710⁎⁎⁎ (3.416) −2.711 −14.375 7.821⁎⁎⁎ −0.398 49.150⁎⁎⁎ 41.629⁎⁎⁎ 43.453⁎⁎⁎

0.473 271.468 279.845 −131.734

0.621 261.670 280.519 −121.835

(0.213) (5.331) (5.525) (8.510)

(1.082) (0.791) (1.805) (3.579) (1.218) (0.348) (10.472) (8.663) (12.229)

0.883 167.171 201.943 −65.585

Interaction terms between Negative and Quality, Spatial, as well as House are omitted during the modelling process. ⁎⁎ p b 0.05. ⁎⁎⁎ p b 0.01.

also introduced and evaluated, including AIC, BIC, and log likelihood. Based on all criteria, Model III obviously outperforms. Therefore the discussion of findings below is based on Model III, a full model with all main explanatory variables and interaction terms included, which provides a remarkably good fit (explaining 88.3% of the variation of effect sizes) given the widely disparate set of original studies and the range of countries and time periods covered. 3.1. Impacts of environmental amenity valuation scenarios With regard to environmental amenity assessing scenarios, we first examined the impacts of urban rivers' attributes individually. The meta-analysis results (Supplementary Table 1, Models S1–S3) suggest that all three attributes of urban rivers, namely proximity to river, water quality, and river view, could significantly affect the estimates of percentage change in housing price. And compared with river view and water quality, a low relative value is documented for river proximity, even though the impact of river proximity on property value has received the widest attention in HPM literature (Price, 2017) and often found to have pronounced impacts on housing values (e.g., McLeod, 1984; Sander and Zhao, 2015). This result echoes empirical evidences that “proximity is less important” and “view itself matters” (Bourassa et al., 2004; Cavailhès et al., 2009). It might also suggest that overlooking other attributes of urban rivers in the HPM studies, besides river proximity, might impair the adequacy of estimates, and could sometimes

lead to biased results when estimating urban rivers' amenity/ disamenity values. In the final Model III (Table 3), we use river proximity as the baseline scenario. The meta-analysis results reveals that river view is the most significant influential factor (β = 9.952, p b 0.01), suggesting that the environmental amenity/disamenity associated with river view is valued approximately 9.952 times higher than river proximity. This result is in line with the existing literature on freshwater landscapes like lakes (Reynaud and Lanzanova, 2017). The sight of river could elicit a wide range of beneficial psycho-physiological and emotional responses like pleasure, fascination, and relaxation (Völker and Kistemann, 2013a, 2013b, 2015; Chen and Li, 2017; Haeffner et al., 2017), which might trigger a desire to visit the riverine site latter on. Especially, when river view is regarded as a negative disamenity, its impact on housing price would become much small, as suggested by the negative and significant interaction term between river view and negative disamenities (β = −7.434, p b 0.01). Water quality is also highly influential for housing prices (β = 6.426, p b 0.01), even though it has been argued that homeowners might have difficulties in recognizing and interpreting water quality metrics and indicators, such as pH, dissolved oxygen, fecal coliform, etc. (Leggett and Bockstael, 2000; Walsh et al., 2011, 2017). And in fact, irritating odor or dark color associated with heavy water could easily attract urban dwellers' attention and raise health concern (Chen and Li, 2018; Cai et al., 2019). Our meta-analysis result suggests that water quality of urban rivers can be explicitly capitalized into housing prices worldwide and water quality is more valued than river proximity. This finding is consistent with estimates from recent original studies (Liu et al., 2019; Papenfus, in press) that in general the change of distance (from farther distance to within 150 m) has much smaller impact on housing prices than water quality impairment. Hitherto most hedonic studies focus on the environmental amenities associated with physical proximity (Price, 2017), which tends to capture the overall environmental amenities (recreational uses and the change of being in vista) of urban rivers. Our meta-analysis results suggest that various nuanced attributes of urban rivers should be incorporated into HPM models so as to adequately capture their environmental amenities and disamenities. An omission of relevant valuation variables might lead to biased estimates in hedonic analyses (Leggett and Bockstael, 2000; Williamson et al., 2008; Kuminoff et al., 2010; Price, 2017). For example, if the level of amenities/disamenities and the distance to a river (the source of amenities/disamenities) are related, more variations might be introduced into homeowners' experiences and perceptions pertaining to the river's environmental externalities as the distance changes. The multidimensional nature of urban rivers' attributes (such as water quality and quantity, riverine landscape characteristics their spatial and temporal variations) can greatly influence their environmental externalities like the quality of river views and the feasibility of recreational activities. Therefore, which attributes can constitute meaningful link between homeowners' WTPs and the environmental amenity of interest should be carefully specified and taken into the utility function and HPM analysis of property markets. 3.2. Impacts of modelling variables In the similar vein, we find that two modelling variables, Year2000 and Significant, are statistically significant with positive sign, while the other modelling variable (Spatial) is not statistically significant. The significance of variable Year2000 shows that urban rivers tend to have a greater marginal effect on housing prices (a higher percentage of environmental amenities) after the year of 2000. A possible explanation for this result could be that worldwide the ecological restoration and rehabilitation of urban rivers mainly started from the end of the 20th century (Beechie et al., 2009; Jarrad et al., 2018; Lorenz et al., 2018) has successfully brought healthy river ecosystems back to human societies, and this could enhance people's understanding and

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

recognition of the importance and value of urban rivers to human wellbeing. Meanwhile, the densification and/or re-densification of urban areas in the 21st century, which has transformed many natural areas into other urban land uses and urban riverscape becomes increasingly scarce. This scarcity could also improve urban dwellers' appreciation of urban rivers as valuable natural elements in their living environs (Haase et al., 2018). These two factors intuitively point to the fact that more and more homeowners indeed have better and greater appreciation of urban rivers' recreation and aesthetical benefits and they have higher WTPs, which can be explicitly reflected by the capitalization of relevant environmental amenities into property values. For the other significant modelling variable (whether urban river's environmental amenities are significant in original studies), this result indicates that significant effect sizes tend to have larger magnitude than insignificant ones. In other words, the magnitude might be negligible if homeowners regard urban rivers' impacts are insignificant. This result is in line with our common sense intuition, indicating that the publication bias towards selective reporting of statistically significant results does not exist (Stanley and Doucouliagos, 2014; Soon and Ahmad, 2015). Yet, this question remains as we failed to identify and include relevant empirical studies such as working papers and reports from grey literature. Finally, it is interesting to find that whether spatial effects are controlled in HPM studies would not affect the estimates significantly, implying little difference between the estimates of urban rivers' relative impacts on housing values from different functional forms (Braden et al., 2011). Although it has been argued that failure to account for spatial effects in HPM studies might produce biased and inconsistent estimates (Anselin, 2003; Pace and LeSage, 2004; Chen and Li, 2018), this result is consistent with the empirical evidence that the difference of the estimates between spatial hedonic models and conventional ones tend to be very small (Mueller and Loomis, 2008; Zygmunt and Gluszak, 2015; Chen, 2017). Turning to the interaction terms, the significance of (disamenity × Year2000, β = 3.710, p b 0.01) reveals that urban rivers' negative disamenity have larger influence on housing value after the year of 2000. This result corroborates our previous finding (significant main effect of Year2000) that after 2000 people become more aware of the benefits brought by river restoration, which offers a contrast to those unrestored degraded urban rivers and thus might enable greater awareness of the negative impacts (Chen, 2017; Dunn et al., 2019; Moran et al., in press). Moreover, the interaction term between disamenity and significant estimates is statistically insignificant, which indicates that both positive and negative amenities, either significant or insignificant, are reported impartially in the literature. 3.3. Impact of contextual factors With respect to the socio-economic variables, we find that GDP per capita is not statistically insignificant, while the average housing price is significant (β = 0.863, p b 0.01) in the meta-analysis. These two variables have often been used as a proxy of the mean household income in meta-analysis studies (e.g., Brander and Koetse, 2011). This result supports the argument made by Soon and Ahmad (2015) as well as Brander and Koetse (2011) that the average house price is a more exact measure of household income level than the GDP figures. The significant and positive coefficient of average housing price suggests that household income is an important determinant of the relative value of urban rivers' environmental amenities/disamenities: a higher income increases the impact of urban rivers on housing prices. This result reveals that higher affluence is associated with stronger preferences and higher WTPs for urban rivers' environmental amenities and environmental amenities brought by urban rivers are normal goods (Coulson and Zabel, 2013; Fan et al., 2016). Another socio-economic variable, population density, reflects two crucial determinants of the value of urban open spaces, covering both

7

green spaces like parks and blue spaces like rivers (Nilsson, 2014; Sander and Zhao, 2015; Völker and Kistemann, 2015). First it represents the local demand for natural amenities pertaining to open spaces: the higher the population density and the more residents living in the vicinity of an open space, the greater the demand for its amenities (Choumert and Cormier, 2011). And secondly, population density also indicates the availability of urban natural spaces: more densely populated areas might have either less open spaces as more land is used for residential and infrastructure uses (Tang and Wong, 2008) or even have more open spaces when efficient land use planning can be adopted in a city (Cinyabuguma and McConnell, 2013; Chen and Hu, 2015). We find a significant and negative relationship between population density and the relative value of urban rivers' environmental impacts. This result contrasts with the meta-analysis for urban open space, mainly green spaces and parks (Brander and Koetse, 2011). A possible explanation is that unlike green spaces which can be created to satisfy residents' increasing demands, urban rivers normally can only be restored or rehabilitated, rather than created, and in general are in fixed supply. Thus the increasing population density of a city means the riverine recreational experience might be deteriorated by congestion and river view (which is the most valuable attribute as explained earlier) might be blocked by dense buildings for most homeowners. Consequently, an increase in population density will result in a decrease in the estimates of urban rivers' environmental amenities/disamenities. Additionally, the significant and positive coefficient of the interaction term between disamenities and population density (β = 7.821, p b 0.01) suggests that higher population density would result in greater awareness of urban rivers' negative impacts and higher deduction in housing prices. The last group of contextual variables are pertaining to the broader geographic location (in terms of continental areas) where original studies were conducted, which are supposed to have an impact on the household valuation of urban rivers' environmental impacts, but are not directly available from the primary studies. We find that the relative prices of urban rivers' environmental externalities for America, Europe/ Israel and Australia are not significantly different from studies for China. This may imply that households across different countries have very similar WTPs for urban rivers' amenities/disamenities. However, when the interaction terms between negative disamenities and continental variables are considered, it is interesting to note that the relative prices of urban rivers' environmental negative externalities are much higher in America, European countries, Israel and Australia: households in these continents would command a greater reduction in their housing prices if rivers nearby exert environmental disamenities. In other words, the relative prices of urban rivers' disamenities are much lower for Chinese households. This sharp contrast might be associated with the relatively high tolerance of Chinese people to the substantial environmental degradation accompanied by tremendous urbanization and economic growth (Jim and Chen, 2006; Antipova, 2018), and on the other hand an increasing scarcity of natural landscapes in highly compact Chinese cities, as well as a trust of local authorities in restoring the degraded rivers (as shown by Chinese government's commitment to urban river rehabilitation and restoration, according to Ministry of Housing and Urban-Rural Development of the People's Republic of China, 2015), might also weaken the negative impacts of urban rivers on housing prices in China. 4. Conclusion This article attempts to provide the very first synthetic meta-analysis of the empirical studies applying the HPM to estimate the environmental externalities of urban rivers, a vulnerable yet crucial component of many urban ecosystems (Francis, 2014). The existing literature has been shown to be diverse in terms of valuation scenarios considered, modelling approaches adopted and geographical regions where hedonic studies were conducted. And the estimates attached to urban rivers produced by the existent original studies are not essentially directly

8

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

comparable. This meta-analysis help identify important valuation scenario, socio-economic, and modelling characteristics that determine people's valuation of urban rivers, and find a systematic explanation for the variation in estimates of urban rivers' value. This meta-analysis provided a comprehensive overview. One of the key results from our meta-analysis is that river view and water quality produce larger estimates of environmental amenities/disamenities, compared with the conventional valuation attribute that still dominates studies on the impacts of urban rivers: physical proximity (Price, 2017). This result offers insight on how to delineate attributes for hedonic model building to reflect people's sophisticated valuation of intertwined amenities and disamenities generated by urban rivers. The evidence of the significant value attributed to river view and water quality holds much convincing message for resource and environmental managers when planning for value-added features of urban river restoration and rehabilitation projects. Moreover, people's overall valuation of urban rivers is not sensitive to macro-geographical locations, suggesting a universally consistent impact of urban rivers on housing markets across varying cultures and societies. Instead, household income level and population density should be systematically controlled if value transfer across countries is necessary. Furthermore, we find that empirical estimates vary over times, indicating the widespread river restoration and rehabilitation in the 21st century has driven up homeowners' environmental perception and appreciation of urban rivers' amenities to a rather high level. On the practical front, these findings support two arguments from a very utilitarian point of view. First, it appears that the visual impacts might be prioritized for river restoration projects, such as through careful revegetation of riparian areas using native species that can help urban dwellers develop a sense of place (Kueffer and Kull, 2017). This could harbor rich diversity of ecological functions and in the meantime maximize environmental amenities that homeowners would like to pay for. And second, costeffective river restoration should be prioritized in densely populated areas over places with low population densities. This approach might maximize the number of people who can enjoy rivers for a given budget.

4.1. Limitations of the study Several caveats, however, need to be addressed. First, although the hedonic studies gleaned from the literature and those selected for this meta-analysis are broadly distributed across the globe, some regions (Africa and Latin America) are under-represented. More original studies are therefore needed in these countries to improve the generalizability and universality of the statistical findings derived in the current metaanalysis. Second, we purposely exclude another important disamenity, i.e., flood risk, from our meta-analysis, as we trust diverse engineering and ecological solutions have been successfully implemented to minimize flood risk in densely-populated cities, even though we could not accurately isolate flood risk and associated engineering practices (which have also devalued the riverine ecosystems, according to Gomes and Wai, 2014) from other river-related disamenity values considered in this study. And a very comprehensive meta-analysis of hedonic estimates of WT for floodplain location is already available (Beltrán et al., 2018). Nevertheless, it is critical to attune the HPM application to all value-bearing attribute. Thirdly, we could not include rivers' hydromorphological attributes (such as the availability of riparian greening) as regressors in this meta-analysis, which might also affect the effect sizes. While the evidence is still incomplete, we envision this meta-analysis can be helpful in generating insights for calibrating HPM and justifying the efforts of environmental intelligence to restore the visual attractiveness and water quality of urban rivers. Supplementary data to this article can be found online at https://doi. org/10.1016/j.scitotenv.2019.133628.

Acknowledgements This work is supported by the General Research Fund from the Research Grants Council of the Hong Kong Special Administrative Region (HKU840911H and HKU759413H). References Åberg, E.U., Tapsell, S., 2013. Revisiting the River Skerne: the long-term social benefits of river rehabilitation. Landsc. Urban Plan. 113, 94–103. Anderson, S.T., West, S.E., 2006. Open space, residential property values, and spatial context. Reg. Sci. Urban Econ. 36 (6), 773–789. Anselin, L., 2003. Spatial externalities and spatial econometrics. Int. Reg. Sci. Rev. 26, 153–166. Antipova, A., 2018. Urban Environment, Travel Behavior, Health, and Resident Satisfaction. Palgrave Macmillan, Cham. Artell, J., Huhtala, A., 2017. What are the benefits of the Water Framework Directive? Lessons learned for policy design from preference revelation. Environ. Resour. Econ. 68 (4), 847–873. Auerbach, D.A., Deisenroth, D.B., McShane, R.R., McCluney, K.E., Poff, N.L., 2014. Beyond the concrete: accounting for ecosystem services from free-flowing rivers. Ecosyst. Serv. 10 (1–5). Beechie, T.J., Pess, G.R., Pollock, M.M., Ruckelshaus, M.H., Roni, P., 2009. Restoring rivers in the twenty-first century: science challenges in a management context. In: Beamish, R.J., Rothschild, B.J. (Eds.), The Future of Fisheries Science in North America. Springer, Dordrecht, pp. 697–717. Beltrán, A., Maddison, D., Elliott, R.J.R., 2018. Is flood risk capitalised into property values? Ecol. Econ. 146, 668–685. Bin, O., Czajkowski, J., Li, J., Villarini, G., 2017. Housing market fluctuations and the implicit price of water quality: Empirical evidence from a South Florida housing market. Environ. Resour. Econ. 68 (2), 319–341. Bolleter, J., 2018. Living suburbs for living streams: how urban design strategies can enhance the amenity provided by living stream orientated public open space. J. Urban Des. 23, 518–543. Bonetti, F., et al., 2016. Canals vs. streams: to what extent do water quality and proximity affect real estate values? A hedonic approach analysis. Water 8 (12). Borenstein, M., Hedges, L.V., Higgins, J.P., Rothstein, H.R., 2010. A basic introduction to fixed-effect and random-effects models for meta-analysis. Res. Synth. Methods 1, 97–111. Bourassa, S.C., Hoesli, M., Sun, J., 2004. What's in a view? Environ Plan A 36, 1427–1450. Braden, J.B., Feng, X., Won, D.-H., 2011. Waste sites and property values: a meta-analysis. Environ. Resour. Econ. 50, 175–201. Brander, L.M., Koetse, M.J., 2011. The value of urban open space: meta-analyses of contingent valuation and hedonic pricing results. J. Environ. Manag. 92, 2763–2773. Brander, L.M., Florax, R.J., Vermaat, J.E., 2006. The empirics of wetland valuation: a comprehensive summary and a meta-analysis of the literature. Environ. Resour. Econ. 33, 223–250. Bristow, A.L., Wardman, M., Chintakayala, V.P.K., 2015. International meta-analysis of stated preference studies of transportation noise nuisance. Transportation 42, 71–100. Cai, W., Li, Y., Shen, Y., Wang, C., Wang, P., Wang, L., Niu, L., Zhang, W., 2019. Vertical distribution and assemblages of microbial communities and their potential effects on sulfur metabolism in a black-odor urban river. J. Environ. Manag. 235, 368–376. Cavailhès, J., Brossard, T., Foltête, J.-C., Hilal, M., Joly, D., Tourneux, F.-P., Tritz, C., Wavresky, P., 2009. GIS-based hedonic pricing of landscape. Environ. Resour. Econ. 44, 571–590. Chen, W.Y., 2017. Environmental externalities of urban river pollution and restoration: a hedonic analysis in Guangzhou (China). Landsc. Urban Plan. 157, 170–179. Chen, W.Y., Hu, F.Z.Y., 2015. Producing nature for public: land-based urbanization and provision of public green spaces in China. Appl. Geogr. 58, 32–40. Chen, W.Y., Li, X., 2017. Cumulative impacts of polluted urban streams on property values: a 3-D spatial hedonic model at the micro-neighborhood level. Landsc. Urban Plan. 162, 1–12. Chen, W.Y., Li, X., 2018. Impact of urban stream pollution: a comparative spatial hedonic study of high-rise residential buildings in Guangzhou, south China. Geogr. J. 184, 283–297. Chen, W.Y., Liekens, I., Broekx, S., 2017. Identifying societal preferences for river restoration in a densely populated urban environment: evidence from a discrete choice experiment in central Brussels. Environ. Manag. 60, 263–279. Cho, S.H., Bowker, J.M., Park, W.M., 2006. Measuring the contribution of water and green space amenities to housing values; An application and comparison of spatially weighted hedonic models. J. Agric. Resour. Econ. 31 (3), 485–507. Cho, S.H., Roberts, R.K., Kim, S.G., 2011. Negative externalities on property values resulting from water impairment: the case of the Pigeon River Watershed. Ecol. Econ. 70, 2390–2399. Choumert, J., Cormier, L., 2011. The provision of urban parks: an empirical test of spatial spillovers in an urban area using geographic information systems. Ann. Reg. Sci. 47, 437–450. Cinyabuguma, M., McConnell, V., 2013. Urban growth externalities and neighborhood incentives; another cause of urban sprawl. J. Reg. Sci. 53, 332–348. Cohen, J.P., Cromley, R.G., Banach, K.T., 2015. Are homes near water bodies and wetlands worth more or less? An analysis of housing prices in one Connecticut town. Growth Chang. 46 (1), 114–132. Colby, B.G., Wishart, S., 2002. Quantifying the influence of desert riparian areas on residential property values. Apprais. J. 70 (3), 304–308.

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628 Cooper, H.M., Hedges, L.V., Valentine, J.C., 2009. The Handbook of Research Synthesis and Meta-analysis. Russell Sage Foundation, New York. Coulson, N.E., Zabel, J.E., 2013. What can we learn from hedonic models when housing markets are dominated by foreclosures? Ann. Rev. Resour. Econ. 5, 261–279. DeCoster, 2009. Mete-analysis notes. http://www.stat-help.com/notes.html. Dunn, R.J.K., Waltham, N.J., Huang, J., Teasdale, P.R., King, B.A., 2019. Protecting water quality in urban estuaries: Australian case studies. In: Wolanski, E., Day, J.W., Elliott, M., Ramachandran, R. (Eds.), Coasts and Estuaries: The Future. Elsevier, Amsterdam, pp. 69–86. Everard, M., Moggridge, H.L., 2012. Rediscovering the value of urban rivers. Urban Ecosyst. 15, 293–314. Fan, Q., Hansz, J.A., Yang, X., 2016. The pricing effects of open space amenities. J. Real Estate Financ. Econ. 52, 244–271. Francis, R.A., 2012. Positioning urban rivers within urban ecology. Urban Ecosyst. 15 (2), 285–291. Francis, R.A., 2014. Urban rivers: novel ecosystems, new challenges. WIREs Water 1, 19–29. https://doi.org/10.1002/wat2.1007. Garcia, X., Corominas, L.I., Pargament, D., Acuña, V., 2016. Is river rehabilitation economically viable in water-scarce basins? Environ. Sci. Pol. 61, 154–164. Garrod, G., Willis, K., 1994. An economic estimate of the effect of a waterside location on property values. Environ. Resour. Econ. 4 (2), 209–217. Gerrish, E., Watkins, S.L., 2018. The relationship between urban forests and income: a meta-analysis. Landsc. Urban Plan. 170, 293–308. Gibbons, S., Mourato, S., Resende, G.M., 2014. The Amenity Value of English Nature: A Hedonic Price Approach. Environ. Resour. Econ. 57 (2), 175–196. Ginzarly, M., Teller, J., 2019. Operationalizing the HUL recommendation in urban river corridors: challenges and perspectives. In: Pereira Roders, A., Bandarin, F. (Eds.), Reshaping Urban Conservation. Springer, Singapore, pp. 511–527. Glass, G.V., 1976. Primary, secondary, and meta-analysis of research. Educ. Res. 5, 3–8. Gobster, P.H., Westphal, L.M., 2004. The human dimensions of urban greenways: planning for recreation and related experiences. Landsc. Urban Plan. 68, 147–165. Gomes, P.I.A., Wai, O.W.H., 2014. Sampling at mesoscale physical habitats to explain headwater stream water quality variations: its comparison to equalspaced sampling under seasonal and rainfall aided flushing states. J. Hydrol. 519, 3615–3633. Greene, W.H., 2008. Econometric Analysis. 6th ed. Prentice Hall, Upper Saddle River, N.J. Guignet, D., Jenkins, R., Ranson, M., Walsh, P.J., 2018. Contamination and incomplete information: bounding implicit prices using high-profile leaks. J. Environ. Econ. Manag. 88, 259–282. Haase, A., Wolff, M., Rink, D., 2018. From shrinkage to regrowth: the nexus between urban dynamics, land use change and ecosystem service provision. In: Kabisch, S., Koch, F., Gawel, E., Haase, A., Knapp, S., Krellenberg, K., Nivala, J., Zehnsdorf, A. (Eds.), Urban Transformations. Springer, Cham, pp. 197–219. Haeffner, M., Jackson-Smith, D., Buchert, M., Risley, J., 2017. Accessing blue spaces: social and geographic factors structuring familiarity with, use of, and appreciation of urban waterways. Landsc. Urban Plan. 167, 136–146. Higgins, J.P.T., Thompson, S.G., 2004. Controlling the risk of spurious findings from metaregression. Stat. Med. 23, 1663–1682. Hunter, J.E., Schmidt, F.L., 2015. Methods of Meta-analysis: Correcting Error and Bias in Research Findings. SAGE Publications, Thousand Oaks, California. Jarrad, M., Netusil, N.R., Moeltner, K., Morzillo, A.T., Yeakley, J.A., 2018. Urban stream restoration projects: do project phase, distance, and type affect nearby property sale prices? Land Econ. 94, 368–385. Jim, C.Y., Chen, W.Y., 2006. Impacts of urban environmental elements on residential housing prices in Guangzhou (China). Landsc. Urban Plan. 78, 422–434. Klemick, H., Griffiths, C., Guignet, D., Walsh, P., 2018. Improving water quality in an iconic estuary: an internal meta-analysis of property value impacts around the Chesapeake Bay. Environ. Resour. Econ. 69, 265–292. Koricheva, J., Gurevitch, J., Mengersem, K. (Eds.), 2013. Handbook of Meta-analysis in Ecology and Evolution. Princeton University Press, Princeton. Kueffer, C., Kull, C.A., 2017. Non-native species and aesthetics of nature. In: Vilà, M., Hulme, P.E. (Eds.), Impact of Biological Invasions on Ecosystem Services, Invading Nature. Springer, Switzerland, pp. 311–324. Kuminoff, N.V., Parmeter, C.F., Pope, J.C., 2010. Which hedonic models can we trust to recover the marginal willingness to pay for environmental amenities? J. Environ. Econ. Manag. 60, 145–160. Leggett, C.G., Bockstael, N.E., 2000. Evidence of the effects of water quality on residential land prices. J. Environ. Econ. Manag. 39, 121–144. LeSage, J.P., Pace, R.K., 2009. Introduction to Spatial Econometrics. CRC Press, Boca Raton. Liu, H., Gopalakrishnan, S., Browning, D., Sivandran, G., 2019. Valuing water quality change using a coupled economic-hydrological model. Ecol. Econ. 161, 32–40. Lorenz, A.W., Haase, P., Januschke, K., Sundermann, A., Hering, D., 2018. Revisiting restored river reaches-assessing change of aquatic and riparian communities after five years. Sci. Total Environ. 613-614, 1185–1195. Lundy, L., Wade, R., 2011. Integrating sciences to sustain urban ecosystem services. Prog. Phys. Geogr. 35, 653–669. Mahan, B.L., Polasky, S., Adams, R.M., 2000. Valuing urban wetlands: A property price approach. Land Econ. 76 (1), 100–113. Malpezzi, S., 2003. Hedonic pricing models: A selective and applied review. In: O’Sullivan, T., Gibb, K. (Eds.), Housing Economics and Public Policy. Blackwell Science, Malden, MA, pp. 67–89. Mattmann, M., Logar, I., Brouwer, R., 2016. Hydropower externalities: a meta-analysis. Energy Econ. 57, 66–77. McLeod, P.B., 1984. The demand for local amenity: an hedonic price analysis. Environ Plan A 16, 389–400.

9

Ministry of Housing and Urban-Rural Development of the People's Republic of China, 2015. Urban Watercourse Rehabilitation Directive. http://www.mohurd.gov.cn/ wjfb/201509/W020150911050936.pdf (accessed January 2019). Mohammad, S.I., Graham, D.J., Melo, I.C., Anderson, R.J., 2013. A meta-analysis of the impact of rail projects on land and property values. Transp. Res. A 50, 158–170. Monk, W.A., Compson, Z.G., Choung, C.B., Korbel, K.L., Rideout, N.K., Baird, D.J., 2019. Urbanization of floodplain ecosystems: weight-of-evidence and network metaanalysis elucidate multiple stressor pathways. Sci. Total Environ. https://doi.org/ 10.1016/j.scitotenv.2019.02.253 (in press). Mooney, S., Eisgruber, L.M., 2001. The influence of riparian protection measures on residential property values: the case of the Oregon plan for salmon and watersheds. J. Real Estate Financ. Econ. 22 (2-3), 273–286. Moran, S., Perreault, M., Smardon, R., 2019. Finding our way: a case study of urban waterway restoration and participatory process. Landsc. Urban Plan. https://doi.org/ 10.1016/j.landurbplan.2016.08.004 (in press). Morgan, O.A., Hamilton, S.E., Chung, V., 2010. Water Quality and Residential Property Values: A Natural Experiment Approach. Working paper No. 10-12. Department of Economics, Appalachian State University. Mueller, J.M., Loomis, J.B., 2008. Spatial dependence in hedonic property models: Do different corrections for spatial dependence result in economically significant differences in estimated implicit prices? J. Agric. Resour. Econ. 212–231. Mukherjee, S., Caplan, A.J., 2011. GIS-based estimation of housing amenities: the case of high grounds and stagnant streams. Lett. Spat. Resour. Sci. 4 (1), 49–61. Nelson, J.P., 2015. Meta-analysis: statistical methods. In: Johnston, R.J., Rolfe, J., Rosenberger, R., Brouwer, R. (Eds.), Benefit Transfer of Environmental and Resource Values. Springer, Dordrecht, pp. 329–356. Nelson, J.P., Kennedy, P.E., 2009. The use (and abuse) of meta-analysis in environmental and natural resource economics: an assessment. Environ. Resour. Econ. 42, 345–377. Netusil, N.R., 2005. The effect of environmental zoning and amenities on property values: Portland, Oregon. Land Econ. 81 (2), 227–246. Netusil, N.R., Kincaid, M., Chang, H., 2014. Valuing water quality in urban watersheds: a comparative analysis of Johnson Creek, Oregon, and Burnt Bridge Creek, Washington. Water Resour. Res. 50, 4254–4268. Nilsson, P., 2014. Natural amenities in urban space-a geographically weighted regression approach. Landsc. Urban Plan. 121, 45–54. OECD, 2018. Purchasing power parities (PPP). https://data.oecd.org/conversion/purchasing-power-parities-ppp.htm, Accessed date: December 2018. Orford, S., 2002. Valuing locational externalities: a GIS and multilevel modelling approach. Environ. Plann. B. Plann. Des. 29 (1), 105–127. Pace, R.K., LeSage, J.P., 2004. Spatial statistics and real estate. J. Real Estate Financ. Econ. 29, 147–148. Palmer, M.A., Allan, J.D., 2006. Restoring rivers. Issues Sci. Technol. 22, 40–48. Papenfus, M., 2019. Do housing prices reflect water quality impairments? Evidence from the Puget Sound. Water Resour. Econ. https://doi.org/10.1016/j.wre.2018.12.001 (in press). Paul, M.J., Meyer, J.L., 2001. Streams in the urban landscape. Annu. Rev. Ecol. Syst. 32, 333–365. Polyakov, M., Fogarty, J., Zhang, F., Pandit, R., Pannell, D.J., 2017. The value of restoring urban drains to living streams. Water Resour. Econ. 17, 42–55. Poor, P.J., Pessagno, K.L., Paul, R.W., 2007. Exploring the hedonic value of ambient water quality: a local watershed-based study. Ecol. Econ. 60, 797–806. Price, C., 2017. Landscape Economics. Palgrave Macmillan, Cham. Qiu, Z.Y., Prato, T., Boehm, G., 2006. Economic valuation of riparian buffer and open space in a suburban watershed. J. Am. Water Resour. Assoc. 42 (6), 1583–1596. Rambaldi, A.N., Fletcher, C.S., Collins, K., McAllister, R.R., 2013. Housing shadow prices in an inundation-prone suburb. Urban Stud. 50 (9), 1889–1905. Reynaud, A., Lanzanova, D., 2017. A global meta-analysis of the value of ecosystem services provided by lakes. Ecol. Econ. 137, 184–194. Rosen, S., 1974. Hedonic prices and implicit markets: product differentiation in pure competition. J. Polit. Econ. 82, 34–55. Sander, H.A., Polasky, S., 2009. The value of views and open space: Estimates from a hedonic pricing model for Ramsey County, Minnesota, USA. Land Use Policy 26 (3), 837–845. Sander, H.A., Zhao, C., 2015. Urban green and blue: who values what and where? Land Use Policy 42, 194–209. Siriwardena, S.D., Boyle, K.J., Holmes, T.P., Wiseman, P.E., 2016a. The implicit value of tree cover in the U.S.: a meta-analysis of hedonic property value studies. Ecol. Econ. 128, 68–76. Siriwardena, S.D., Boyle, K.J., Holmrd, T.P., Wiseman, P.E., 2016b. The implicit value of tree covers in the U.S.: a meta-analysis of hedonic property value studies. Ecol. Econ. 128, 68–76. Sirmans, G.S., MacDonald, L., Macpherson, D.A., Zietz, E.N., 2006. The value of housing characteristics: a meta analysis. J. Real Estate Financ. Econ. 33, 215–240. Smith, V.K., Huang, J.C., 1995. Can markets value air quality? A meta-analysis of hedonic property value models. J. Polit. Econ. 103, 209–227. Smith, V., Pattanayak, S., 2002. Is meta-analysis a Noah's Ark for non-market valuation? Environ. Resour. Econ. 22, 271–296. Soon, J.-J., Ahmad, S.-A., 2015. Willingly or grudgingly? A meta-analysis on willingnessto-pay for renewable energy use. Renew. Sust. Energ. Rev. 44, 877–887. Stanley, T.D., Doucouliagos, H., 2014. Meta-regression approximations to reduce publication selection bias. Res. Synth. Methods 5, 60–78. Stanley, T.D., Doucouliagos, H., Giles, M., Heckemeyer, J.H., Johnston, R.J., Laroche, P., Nelson, J.P., Paldam, M., Poot, J., Pugh, G., Rosenberger, R.S., Rost, K., 2013. Metaanalysis of economics research reporting guidelines. J. Econ. Surv. 27, 390–394. Tang, B., Wong, S., 2008. A longitudinal study of open space zoning and development in Hong Kong. Landsc. Urban Plan. 87, 258–268.

10

W.Y. Chen et al. / Science of the Total Environment 693 (2019) 133628

Tapsuwan, S., Polyakov, M., Bark, R., Nolan, M., 2015. Valuing the Barmah–Millewa Forest and in stream river flows: A spatial heteroskedasticity and autocorrelation consistent (SHAC) approach. Ecol. Econ. 110, 98–105. Taylor, L.O., 2013. Hedonics. In: Shogren, J.F. (Ed.), Encyclopedia of Energy, Natural Resource and Environmental Economics. vol. 3. Elsevier, London, pp. 342–348. Tranfield, D., Denyer, D., Smart, P., 2003. Towards a methodology for developing evidence-informed management knowledge by means of systematic review. Br. J. Manag. 14, 207–222. U.S. Department of Labor, Bureau of Labor Statistics, 2018. Consumer Price Index Data. https://www.bls.gov/cpi/tables/supplemental-files/home.htm, Accessed date: December 2018. Van Houtven, G., Powers, J., Pattanayak, S.K., 2007. Valuing water quality improvements in the United States using meta-analysis: it the glass half-full or half-empty for national policy analysis? Resour. Energy Econ. 29, 206–228. Viswanathan, V.C., Schirmer, M., 2015. Water quality deterioration as a driver for river restoration: a review of case studies from Asia, Europe and North America. Environ. Earth Sci. 74, 3145–3158. Völker, S., Kistemann, T., 2013a. “I'm always entirely happy when I'm here!” Urban blue enhancing human health and well-being in Cologne and Düsseldorf, Germany. Soc. Sci. Med. 78, 113–124. Völker, S., Kistemann, T., 2013b. “I'm always entirely happy when I'm here!” Urban blue enhancing human health and well-being in Cologne and Düsseldorf, Germany. Soc. Sci. Med. 91, 141–152.

Völker, S., Kistemann, T., 2015. Developing the urban blue: comparative health responses to blue and green urban open spaces in Germany. Health Place 35, 196–205. Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P., Glidden, S., Bunn, S.E., Sullivan, C.A., Reidy Liermann, C., Davies, P.M., 2010. Global threats to human water security and river biodiversity. Nature 467, 555. Walsh, P.J., Milon, J.W., Scrogin, D.O., 2011. The spatial extent of water quality benefits in urban housing markets. Land Econ. 87, 628–644. Walsh, P.J., Griffiths, C., Guignet, D., Klemick, H., 2017. Modeling the property price impact of water quality in 14 Chesapeake Bay counties. Ecol. Econ. 135, 103–113. Wen, H.Z., Zhang, Y., Zhang, L., 2015. Assessing amenity effects of urban landscapes on housing price in Hangzhou, China. Urban For. Urban Green. 14 (4), 1017–1026. Wen, H., Xiao, Y., Zhang, L., 2017. Spatial effect of river landscape on housing price: an empirical study on the Grand Canal in Hangzhou, China. Habitat Int. 63, 34–44. Williamson, J.M., Thurston, H.W., Heberling, M.T., 2008. Valuing acid mine drainage remediation in West Virginia: a hedonic modeling approach. Ann. Reg. Sci. 42, 987–999. Wu, J.J., Adams, R.M., Plantinga, A.J., 2004. Amenities in an urban equilibrium model: Residential development in Portland, Oregon. Land Econ. 80 (1), 19–32. Zygmunt, R., Gluszak, M., 2015. Forest proximity impact on undeveloped land values: a spatial hedonic study. Forest Policy Econ. 50, 82–89.