The willingness to pay for green apartments: The case of Nanjing,China

Abstract

Faced with the challenge of developing sustainable cities, the Chinese government sets green construction as part of the national strategy to reduce energy consumption. However, the consumer market has shown limited response to such policies. To upscale green building, it is crucial to understand the market demands for green apartments. This article employs a conjoint model to estimate the willingness to pay for green dwellings versus accessibility to metros and jobs and neighbourhood quality by different socio-economic groups in Nanjing, China. Results show that the socio-economic status of homebuyers determines their willingness to pay for green attributes. Only the rich are prepared to pay for green apartments to improve their living comfort. To all, the notion of health is appealing as consumers are willing to pay for an unpolluted environment and for non-toxic construction materials used in buildings in good locations.

Keywords

green apartments health market segmentation socio-economic groups willingness to pay

Introduction

In the transition to a market-driven post-industrial urban economy, energy consumption of residential buildings in China has been increasing dramatically due to the improved living standard and the fast increase of the amount of urban households (Cai et al., 2009). To reduce energy consumption and develop sustainable cities, the Chinese government has adopted the green building policy as a national strategy (Guo et al., 2010; Malmqvist, 2008; Yuan et al., 2013). However, the consumer market has shown limited response to such policies, partly due to the fact that government subsidies cover only a part of the extra costs of green apartments. Whether the sustainability goal will be achieved not only depends on the ‘green’ supply, but also on substantial consumer demands. A deeper understanding of consumers’ preferences on green development is crucial to raise green demands (Yau, 2012b).

In the economic transition, the urban housing market in China has become more differentiated. Different socio-economic groups can only afford homes in a certain segment of the housing market. At present, green apartment buildings are mainly built in urban areas with a good location in terms of accessibility to metros and neighbourhood quality. The high price of green apartments is a symbol of luxury for the upper-middle class. However, the extra cost of building a green apartment is quite low in comparison to its inflated selling price (Cai et al., 2013). There is substantial profit embedded in green building (at least threefold). Yet, such extreme hikes in green housing prices deny ordinary consumers the opportunity to enjoy a living standard similar to that enjoyed by the rich (Vallance et al., 2011). To upscale green construction means to enlarge the market and satisfy various social groups of consumers. In addition, consumers’ green housing choices are not only driven by green dwelling attributes, but more by the affordability and the broad living environment, for instance, accessibility to activities and neighbourhood quality (Cole, 2000).

Currently, little is known about the green preferences of the local population in Chinese cities. Different geographical, political, social, and economic contexts might shape different green markets. This article attempts to answer the research question, ‘What are the key factors that influence the willingness to pay for green dwellings versus accessibility to metros and jobs and neighbourhood quality of different socio-economic groups?’ It analyses the green housing market in China to evaluate whether the current green policy is feasible. In the next section, a review is made of the existing literature related to the willingness to pay for green attributes in different contexts, followed by an overview of the study area. Next, the methodology, the research design, and the data collection are detailed. Finally, the empirical results are described and conclusions drawn.

Literature review

Green buildings are regarded as a solution to tackle environmental problems such as excessive energy consumption and air pollution (Retzlaff, 2009). Although in theory green building is related to economics, utility, durability, and comfort of both the building itself and its surrounding environment, in practice the designs and constructions of green building pay limited attention to the wider environmental issue (Leaman and Bordass, 2007). The major focuses are improving energy efficiency, reducing water consumption, using durable and non-toxic materials, and saving life-cycle costs of the building itself (Ali and Al Nsairat, 2009). Following existing studies on green building (e.g. Ali and Al Nsairat, 2009; Kibert and Grosskopf, 2007; Kientzel and Kok, 2011; Leaman and Bordass, 2007, Schweber, 2013), this article considers ‘green’ housing as environment-friendly, resource-efficient, energy-saving, health-improved and comfortable-living housing.

The current green development requires that residents’ expectations and preferences are taken into account (White and Gatersleben, 2011). Residents’ preferences can be measured by their willingness to pay for green buildings. Some research finds that green dwelling attributes which can directly reduce residents’ utility bills gain more willingness to pay due to economic incentives (Yau, 2012b). In Hong Kong, both green and conventional residents are willing to pay more for energy conservation than for indoor air quality improvement, noise level reduction, landscape area enlargement, or water conservation (Chau et al., 2010). In the UK it is argued that to attract prospective tenants, tangible cost savings from a more energy efficient building should be better assessed and communicated (McKinley, 2009). Besides cost saving, some research finds that green building buyers place premium value on better health conditions (Guo et al., 2010).

In addition to green dwelling attributes, attributes of living environment such as a good location, and the physical and social quality of the neighbourhood also play a role in housing decisions (Cole, 2000). In the West, it is found that people have a higher willingness to pay for accessibility to jobs, cleanliness and security of the neighbourhood, for avoiding to live near landfills, and for higher air quality, than for gyms and cultural services (Hite, 2009; Torres et al., 2013). It indicates that consumers also consider the living environment when they purchase a house, probably for the health related reasons. In mainland China, people are willing to pay more for neighbourhood safety and accessibility to jobs in home purchase decisions than for regular dwelling attributes such as house size, number of bedrooms, and orientation (Wang and Li, 2004, 2006).

Although green dwelling attributes are found to be important, the relative weight of them in making a housing purchase has not been systematically analysed. Evidence from New Zealand shows that, although the governments have introduced both the voluntary and mandatory green measures, the most important determinant in housing decisions are still the location and price (Eves and Kippes, 2010). Some studies show that in the US rental market, households are not willing to pay for improved energy efficiency (Calcagni, 2012), and people in Hong Kong are willing to pay for improved public green space that provides better air quality, but are less willing to pay for green building attributes (Lo and Jim, 2010; Yau, 2012a). In mainland China no research has been done on trade-offs between attributes of green dwellings and a good living environment.

Upscaling green buildings is not only a technocratic exercise but part of a socio-ecological process (Portney, 2002). The socio-economic background of consumers would affect their demands of living quality (Jenks, 2000). Consumers’ willingness to pay for housing attributes is constrained by their purchase power and individual characteristics. With the deepening of China’s market reform, households with low income, low education level, and limited social capital can hardly afford good quality housing (Chen et al., 2001; Li, 2012; Man et al., 2011; Ng, 2002). For instance, the market value of residential housing varies greatly between different income groups; the difference between the richest and poorest 10% of urban households is about 400% (Man et al., 2011). Although green buildings contribute to living quality and sustainability, the green construction may spur environmental gentrification and the displacement of low-income residents (Curran and Hamilton, 2012; Hald, 2009; Wang, 2003). The housing affordability issue should be taken into account when upscaling green buildings.

Overview of study area

Our study area is the city of Nanjing, located in the Yangtze River Delta (Figure 1). It has been the most important heavy industrial production centre of East China for more than 50 years. As a consequence, air pollution has been a longstanding problem. Since 2000, the Nanjing government has attempted to change the image of the city by transforming it from an industrial to a post-industrial city. It restructured its commercial activities related to industrial production and Nanjing became a regional service centre. Master plans were formulated to modify land use. As a consequence, the old urban area has been renewed with the introduction of cleaner land uses such as commercial land, and most industries have been relocated to the northern part of the Yangtze River. Land use in the central districts is mainly allocated for commercial, governmental, and residential purposes. There are a few light industries. The southern districts are primarily residential, commercial, and light industrial with some concentration of heavy industrial activities. The northern districts are mainly residential and heavy industrial areas, with several concentrated commercial areas.

Figure 1.

The location and the land use map of Nanjing, China.

The three identified areas are differentiated not just in land use but also in other aspects. Since jobs and metro lines are concentrated in the city centre, apartments there have much better accessibility to jobs and public transport as compared to the newly-developed surrounding districts. In the southern districts, accessibility to metros and jobs is relatively better than in the northern districts which have no metro line – three bridges and one tunnel connect the northern and the southern parts of the city. The city centre has fully fledged services compared to the newly built districts. For instance, good schools are not evenly distributed over the city and the very good schools are located in the city centre only. Air quality varies as well within the city, with heavier air pollution in the northern districts. These differences have affected house prices: the average house price in the central districts exceeds the surrounding districts; houses in the southern districts are usually more expensive than those in the northern districts.

The segmentation of housing markets influences the housing choices of the different socio-economic groups. When people with different socio-economic backgrounds choose to buy an apartment, their choice set is restricted by their purchasing power. An increasing proportion of the lower-middle class live in the northern districts, while the middle and upper-middle classes concentrate in the southern and central districts. As the price of green apartments is high, developers only target the upper-middle class market, and locate green apartment buildings in the newly developed areas near the city centre and in the southern districts. To upscale green construction means to enlarge the market and satisfy more general consumer preferences.

Methodology and data

Experimental design

There are revealed and stated preference methods to evaluate housing preferences. We have applied stated preference method because the green market in China is just emerging, so no reliable market information is available yet. The stated preference method can be used to create a hypothetical market in which the willingness to pay can be considered as housing attribute values directly (Howie et al., 2010; Wang and Li, 2006). The conjoint model or stated preference method was used to estimate the willingness to pay for green housing attributes relative to other attributes and to distinguish between different market segments. The conjoint model assumes that homebuyers make trade-offs between housing characteristics, and in doing so they choose the housing bundles with the most benefits. These characteristics include both the housing attributes and their levels or values (the quality of the attributes).

As stated before, there are three distinct market segments in Nanjing: central districts, southern districts and northern districts. According to literature, influential housing attributes can be divided into five categories: house price, physical quality of the neighbourhood, social quality of the neighbourhood, accessibility to metros and jobs, and dwelling attributes (Howie et al., 2010; Visser et al., 2008). Dwelling attributes include both general and green attributes. To simplify the conjoint model and focus on the key attributes, the general dwelling attributes such as size and number of bedrooms were measured in the questionnaire but not included in the conjoint model. The green attributes are the key variables found in the green preference research discussed in the literature review. Selected attributes in other categories are those showing strong impact on the housing decisions in both the West and China (e.g. Hagoort, 2006; Hamilton and Morgan, 2010; Haurin and Brasington, 1996; Kockelman, 1997; Munroe, 2007; Pan and Zhang, 2008).

Thirteen attributes and their values were identified in each submarket (Table 1). Attribute levels in each market segment are slightly different. The average house price in the central districts can be twice that of the southern districts and three times more than the northern districts. The quality of schools is not evenly distributed either, and the very good schools are only located in the city centre. As the majority of residents in Nanjing rely on public transport, this study defines accessibility in terms of time cost to access a metro stop or a job by public transport. Since a metro line has yet to be built in the northern districts in 2011, the time cost to metro stops and work places (jobs) is much longer in northern districts than in other areas of the city. Moreover, while jobs are concentrated in the central districts, access to jobs is easier in the central districts than in the surrounding areas. Green apartments are situated in both the central districts and southern districts, which makes the quality of green attributes in these two areas better than in the northern districts. Other neighbourhood attribute levels might be the same in different market segments, including clean and unpolluted environment, safety, social status, and neighbours.

Table 1.

Attributes and their levels in the three market segments and an example of a choice-set.

Category	Attribute	Level (central districts)		Level (southern districts)	Level (northern districts)
House price		20000–25000 yuan/m2 25000–30000 yuan/m2 >30000 yuan/m2		10000–15000 yuan/m2 15000–20000 yuan/m2 20000–25000 yuan/m2	6000–8000 yuan/m2 8000–10000 yuan/m2 10000–15000 yuan/m2
Physical quality of the neighbourhood	School quality	Average Good Very good		Poor Average Good	Poor Average Good
	Environmental pollution	A lot Some No pollution		A lot Some No pollution	A lot Some No pollution
Social quality of the neighbourhood	Safety from crime	Poor Average Good		Poor Average Good	Poor Average Good
	Social status of residents	Low Middle High		Low Middle High	Low Middle High
	Neighbours	Do not know any neighbours Know some neighbours Have some family and friends		Do not know any neighbours Know some neighbours Have some family and friends	Do not know any neighbours Know some neighbours Have some family and friends
Accessibility	Accessibility to metro stop by public transport	20–30mins 10–20mins < 10mins		20–30mins 10–20mins < 10mins	> 50mins 40–50mins 30–40mins
	Accessibility to job by public transport	30–40mins 20–30mins < 20mins		> 40mins 30–40mins 20–30mins	> 40mins 30–40mins 20–30mins
Green attributes	Energy and water costs	High Average Low		High Average Low	High More than average Average
	Construction materials	Harmful Somewhat non-toxic Non-toxic		Harmful Somewhat non-toxic Non-toxic	Harmful Somewhat harmful Somewhat non-toxic
	Thermal insulation	Poor Average Good		Poor Average Good	Poor Below average Average
	Sound insulation	Poor Average Good		Poor Average Good	Poor Below average Average
	Ventilation	Poor Average Good		Poor Average Good	None Average Above average
An example of a choice-set
Choice-set No.1	Attribute	A (central districts)	B (southern districts)	C (northern districts)	None of these
House price		> 30000 yuan/m2	15000–20000 yuan/m2	6000–8000 yuan/m2
Physical quality of the neighbourhood	School quality	Very good	Poor	Good
	Environmental pollution	Some	No pollution	Some
Social quality of the neighbourhood	Safety	Average	Average	Average
	Social status of residents	Low	Middle	Middle
	Neighbours	Do not know any neighbours	Have some family and friends	Know some neighbours
Accessibility	Accessibility to metro stop	10–20mins by public transport	< 10mins by public transport	30–40mins by public transport
	Accessibility to job	20–30minsby public transport	20–30mins by public transport	20–30mins by public transport
Green attributes	Energy and water costs	Low	High	High
	Construction materials	Non-toxic	Somewhat non-toxic	Somewhat harmful
	Thermal insulation	Poor	Good	Poor
	Sound insulation	Good	Poor	Below average
	Ventilation	Good	Good	Average
Your choice

To investigate homebuyers’ willingness to pay for these attributes in different market segments, respondents were asked to make a choice from a choice-set which offered three fixed alternatives (central, southern and northern housing markets) and a base alternative (Table 1). To estimate the relative importance of each attribute level, the choice-sets should provide sufficient differentiation within the attribute values in the three market segments. In addition, the choice-sets should also present each level of each attribute the same number of times for respondents to choose. This has been achieved with the help of the orthogonal design method (Wang et al., 2000). In terms of the number of attributes and levels in Table 1, a total of 81 choice-sets were generated.

Utility function

The conjoint model assumes that homebuyers choose an alternative in a choice-set which will provide them with the most benefits or utility. With all the observed responses (choices), an estimate can be made of the utilities (relative importance) of various housing attributes through a utility function. The utility function indicates the extent to which each attribute contributes to the overall utility.

It is assumed that the overall utility consists of a structural part that can be explained by the estimated model, and an error component, the part of the utility that cannot be explained. The structural part of the utility is a linear summation of the part-worth utility contributions of the attributes (Molin, 2011). The utility function can be expressed as follows

U_{i} = \sum β_{k} x_{i k} + ε_{i}

where β_k is the coefficient of attribute k, x_ik is the value of alternative i with respect to attribute k, and ε_i is an error term representing the alternative features that are not specified in the model and variations of taste among respondents; β_kx_ik is part-worth utility of attribute k to the overall utility of alternative i.

The attribute levels are categorical. To include them in the utility function, the attribute levels need to be coded. Therefore, an orthogonal coding scheme was used (Wang et al., 2000). The three levels in each attribute are coded by two vectors which represent the linear and quadratic effects of the attributes. The indicators for attribute level one are −1 and 1; the indicators for attribute level two are 0 and −2; the indicators for attribute level three are 1 and 1.

When respondents make a choice, it is assumed that they choose the alternative with the highest utility. Therefore, utility can be linked to the probability that an alternative will be chosen. It is expressed as

P_{i} = \frac{e^{U_{i}}}{\sum e^{U_{j}}}

where P_i represents the probability of choosing alternative i, and U_i is the utility of alternative i.

By computing the probability of selecting a particular alternative from a choice-set and computing the utilities provided by this particular alternative, the trade-off decisions with regard to different housing attributes made by the respondents can be calculated. We use the multinomial logit (MNL) model to conduct regression on utilities.

Data collection

The conjoint surveys were conducted from September to December 2011 at 21 real estate sales offices spread across the three market segments in Nanjing. The respondents were potential homebuyers visiting the real estate sales offices at the time. Information regarding respondents’ demographic status, affordability, preferences on dwelling attributes (size, number of bedrooms, number of bathrooms, floor level) and hypothetical housing choices were collected.

To avoid fatigue and hence unreliable responses, respondents did not receive all 81 choice sets. Rather, we divided the 81 choice-sets into nine groups with nine randomly selected choice-sets in each group. Each respondent was asked to make choices within one group only, comprising nine choice-sets. Respondents had to make a choice from housing in the central, the southern, and the northern districts, or none from each choice-set. A total of 1373 valid surveys were collected, and each choice-set was chosen by 150 or more respondents.

Results

Taking the respondents’ socio-economic status into consideration, three socio-economic groups have been differentiated: lower-middle class, middle class, and upper-middle class (Table 2). The lower-middle class mainly consists of young people who are just starting their careers and are new to Nanjing; almost half of them have not received their household registration permit (hukou). They have a good educational background. They are either singles or couples without children. A relatively low budget of less than one million yuan only allows them to buy an apartment of around 100 square metres in the northern districts or a much smaller one in the southern districts. The central districts are beyond their means.

Table 2.

Demographic characteristics of three socio-economic groups.

Group	Budget (yuan)	Age	Live in Nanjing	Nanjing hukou	Family composition	Education level
Lower-middle class (31.4%)	< 1million	< 30 (64.7%)30–45 (27.4%)> 45 (7.9%)	< 4 years (30.6%)4–10 years (39.0%)> 10 years (30.4%)	Yes (54.2%)No (45.8%)	Single or young couple with no child (61.0%)Couple with a school-aged child (24.5%)Others (14.5%)	High school or lower (13.1%)Graduate (75.5%)Post-graduate (11.4%)
Middle class (36.7%)	1–2 million	< 30 (38.7%)30–45 (45.5%)> 45 (15.8%)	< 4 years (20.6%)4–10 years (31.5%)> 10 years (47.9%)	Yes (76.0%)No (24.0%)	Single or young couple with no child (45.5%)Couple with a school-aged child (35.1%)Others (19.4%)	High school or lower (10.7%)Graduate (72.3%)Post-graduate (17.0%)
Upper-middle class (31.9%)	> 2 million	< 30 (18.9%)30–45(54.2%)> 45 (26.9%)	< 4 years (9.7%)4–10 years (17.9%)> 10 years (72.4%)	Yes (87.6%)No (12.4%)	Single or young couple with no child (31.5%)Couple with a school-aged child (38.4%)Others (30.1%)	High school or lower (8.3%)Graduate (69.4%)Post-graduate (22.3%)

The middle class is a larger group comprising people older than thirty years who have lived in the city for a relatively long period of time. Three-quarters of them possess a Nanjing-registered hukou. A substantial proportion has received tertiary education (graduate degree) and established their career. Although most of them do not have children, the number of couples with school-aged children is higher. Their budget allows them to purchase an apartment in both the southern and northern districts. They are also able to buy an apartment in the central districts if they trade dwelling size for location. In short, the middle class has more housing options than the lower-middle class.

The upper-middle class are mainly those aged 30 and above, and most of them have lived in Nanjing for more than 10 years. As such, most of them already possess a Nanjing hukou. Their educational level is higher than that of the other two socio-economic classes and many have a well-established career. The proportion of couples with school-aged children within this group is high. Their budget allows them a choice of apartment in any of the three market segments, which gives the upper-middle class the most flexibility where housing is concerned.

Table 3 shows the housing choice model for the total sample and the choice models for three social groups. Generally speaking, a positive sign of a linear component indicates positive impact on housing choice when attribute level improves; a negative sign of a quadratic component indicates declining positive effect on housing choice when attribute level improves; and a positive sign of a quadratic component indicates increasingly positive effect. The adjusted R2 of the whole housing model is smaller than that of the three sub-models, indicating that the sub-models perform better and that housing market in Nanjing is segmented.

Table 3.

The housing choice model for the total sample and the choice models for three social groups.

Central districts	Coefficient				Southern districts	Coefficient				Northern districts	Coefficient
Attribute	Whole	Lower	Middle	Upper	Attribute	Whole	Lower	Middle	Upper	Attribute	Whole	Lower	Middle	Upper
Constant1	−1.124(0.000)	−2.664(0.000)	−1.403(0.000)	−0.412(0.000)	Constant 2	−0.248(0.000)	−1.055(0.000)	−0.153(0.000)	−0.189(0.000)	Constant 3	−0.702(0.000)	−0.009(0.814)	−0.769(0.000)	−6.511(1.000)
House price(yuan/m2)					House price(yuan/m2)					House price(yuan/m2)
Linear	−0.518(0.000)	−0.485(0.000)	−0.534(0.000)	−0.672(0.000)	Linear	−0.668(0.000)	−0.804(0.000)	−1.017(0.000)	−0.333(0.000)	Linear	−0.565(0.000)	−0.952(0.000)	−0.336(0.000)	−0.130(0.411)
Quadratic	−0.028(0.165)	−0.084(0.217)	0.066(0.111)	−0.045(0.108)	Quadratic	−0.064(0.000)	−0.032(0.345)	−0.080(0.000)	−0.077(0.003)	Quadratic	−0.060(0.000)	−0.082(0.000)	−0.058(0.049)	−0.051(0.557)
School quality					School quality					School quality
Linear	0.173(0.000)	0.240(0.054)	0.165(0.017)	0.243(0.000)	Linear	0.132(0.000)	0.156(0.006)	0.098(0.016)	0.195(0.000)	Linear	0.045(0.120)	0.069(0.114)	0.064(0.212)	−1.295(1.000)
Quadratic	−0.018(0.380)	−0.068(0.311)	−0.020(0.621)	0.010(0.718)	Quadratic	−0.014(0.350)	−0.006(0.858)	−0.030(0.202)	−0.005(0.843)	Quadratic	0.004(0.819)	−0.903(0.997)	0.009(0.759)	1.824(1.000)
Environmental pollution					Environmental pollution					Environmental pollution
Linear	0.408(0.000)	0.190(0.099)	0.125(0.063)	0.696(0.000)	Linear	0.356(0.000)	0.172(0.003)	0.291(0.000)	0.671(0.000)	Linear	−0.302(0.000)	0.336(0.000)	0.356(0.000)	5.458(1.000)
Quadratic	−0.006(0.757)	0.036(0.613)	0.033(0.419)	−0.008(0.766)	Quadratic	0.026(0.077)	0.042(0.209)	0.050(0.034)	0.024(0.349)	Quadratic	−0.014(0.390)	−0.037(0.150)	−0.007(0.811)	−1.670(1.000)
Safety from crime					Safety from crime					Safety from crime
Linear	0.286(0.000)	0.302(0.010)	0.260(0.000)	0.365(0.000)	Linear	0.192(0.000)	0.107(0.062)	0.145(0.000)	0.324(0.000)	Linear	0.027(0.361)	0.060(0.173)	0.027(0.615)	1.846(1.000)
Quadratic	−0.100(0.000)	0.095(0.189)	−0.089(0.020)	−0.148(0.000)	Quadratic	−0.077(0.000)	−0.026(0.418)	−0.059(0.012)	−0.151(0.000)	Quadratic	−0.030(0.071)	−0.036(0.160)	−0.034(0.259)	−1.741(1.000)
Social status					Social status					Social status
Linear	0.039(0.272)	0.087(0.481)	0.075(0.280)	−0.011(0.822)	Linear	0.031(0.227)	0.034(0.546)	0.048(0.241)	0.029(0.527)	Linear	−0.021(0.464)	−0.113(0.010)	0.091(0.086)	−1.721(1.000)
Quadratic	−0.023(0.244)	−0.064(0.344)	−0.019(0.634)	−0.044(0.124)	Quadratic	−0.023(0.116)	0.024(0.473)	0.014(0.554)	−0.082(0.002)	Quadratic	−0.017(0.300)	−0.018(0.479)	−0.047(0.116)	−1.693(1.000)
Neighbours					Neighbours					Neighbours
Linear	0.045(0.197)	0.191(0.127)	−0.015(0.831)	0.098(0.043)	Linear	0.047(0.063)	0.086(0.135)	0.054(0.187)	0.035(0.440)	Linear	−0.039(0.174)	−0.003(0.948)	−0.094(0.072)	0.265(0.205)
Quadratic	−0.001(0.944)	−0.080(0.228)	0.026(0.516)	0.006(0.836)	Quadratic	−0.007(0.631)	0.005(0.873)	−0.003(0.913)	−0.006(0.807)	Quadratic	0.006(0.725)	0.014(0.595)	0.003(0.912)	−0.106(0.433)
Accessibility to metro stop					Accessibility to metro stop					Accessibility to metro stop
Linear	0.001(0.977)	−0.247(0.038)	−0.013(0.844)	0.072(0.142)	Linear	0.068(0.007)	0.032(0.577)	0.116(0.005)	0.071(0.117)	Linear	0.073(0.012)	0.092(0.036)	0.082(0.111)	1.866(1.000)
Quadratic	0.014(0.485)	−0.009(0.894)	0.061(0.136)	0.029(0.300)	Quadratic	−0.008(0.606)	−0.031(0.356)	0.003(0.896)	0.000(0.991)	Quadratic	0.006(0.725)	−0.028(0.275)	0.031(0.317)	1.741(1.000)
Accessibility to job					Accessibility to job					Accessibility to job
Linear	0.016(0.647)	−0.007(0.950)	0.159(0.023)	−0.009(0.855)	Linear	0.067(0.008)	0.111(0.057)	0.021(0.606)	0.132(0.004)	Linear	0.163(0.000)	0.224(0.000)	0.159(0.003)	1.465(1.000)
Quadratic	−0.029(0.151)	0.101(0.137)	−0.054(0.168)	−0.084(0.003)	Quadratic	0.008(0.572)	−0.009(0.781)	0.021(0.370)	0.003(0.894)	Quadratic	−0.029(0.082)	−0.065(0.010)	−0.019(0.529)	−1.631(1.000)
Energy and water costs					Energy and water costs					Energy and water costs
Linear	0.028(0.429)	−0.071(0.546)	−0.081(0.249)	0.126(0.010)	Linear	−0.019(0.455)	−0.055(0.345)	−0.045(0.274)	0.040(0.377)	Linear	−0.007(0.804)	−0.005(0.903)	−0.054(0.294)	0.296(0.170)
Quadratic	0.007(0.740)	0.054(0.428)	−0.063(0.106)	0.009(0.747)	Quadratic	0.000(0.976)	0.015(0.645)	−0.006(0.808)	−0.010(0.705)	Quadratic	0.011(0.525)	0.023(0.363)	0.010(0.740)	−0.042(0.747)
Construction materials					Construction materials					Construction materials
Linear	0.083(0.014)	0.115(0.347)	0.040(0.544)	0.110(0.022)	Linear	0.084(0.001)	0.023(0.694)	0.081(0.047)	0.103(0.023)	Linear	−0.015(0.598)	0.028(0.516)	−0.046(0.380)	−1.453(1.000)
Quadratic	−0.008(0.667)	−0.130(0.043)	−0.022(0.565)	−0.001(0.976)	Quadratic	−0.011(0.437)	−0.037(0.259)	0.010(0.681)	−0.022(0.393)	Quadratic	0.001(0.970)	−0.011(0.659)	−0.012(0.691)	1.759(1.000)
Thermal insulation					Thermal insulation					Thermal insulation
Linear	−0.004(0.911)	0.006(0.963)	−0.035(0.593)	0.047(0.331)	Linear	0.010(0.683)	0.035(0.542)	0.015(0.708)	0.018(0.698)	Linear	0.036(0.221)	−0.013(0.775)	0.184(0.001)	1.818(1.000)
Quadratic	−0.037(0.051)	−0.146(0.026)	−0.013(0.735)	−0.033(0.225)	Quadratic	−0.008(0.577)	0.035(0.308)	−0.019(0.409)	−0.001(0.958)	Quadratic	0.011(0.499)	0.025(0.336)	−0.014(0.631)	−1.733(1.000)
Sound insulation					Sound insulation					Sound insulation
Linear	0.063(0.064)	0.114(0.354)	0.041(0.531)	0.086(0.074)	Linear	−0.007(0.783)	−0.090(0.120)	0.034(0.412)	−0.011(0.813)	Linear	0.039(0.179)	0.041(0.346)	0.045(0.387)	−1.694(1.000)
Quadratic	−0.038(0.051)	−0.147(0.021)	−0.028(0.453)	−0.007(0.799)	Quadratic	0.010(0.476)	−0.009(0.778)	0.018(0.434)	0.023(0.369)	Quadratic	−0.007(0.676)	−0.001(0.956)	−0.016(0.589)	1.514(1.000)
Ventilation					Ventilation					Ventilation
Linear	0.033(0.332)	0.049(0.685)	−0.105(0.115)	0.076(0.114)	Linear	0.036(0.158)	0.102(0.077)	−0.027(0.502)	0.031(0.487)	Linear	−0.004(0.887)	−0.007(0.868)	−0.022(0.667)	−3.176(1.000)
Quadratic	−0.019(0.319)	−0.077(0.222)	−0.053(0.155)	0.017(0.536)	Quadratic	0.005(0.724)	−0.005(0.876)	0.007(0.761)	0.027(0.298)	Quadratic	−0.002(0.914)	−0.019(0.458)	0.016(0.603)	0.028(0.844)
	Whole model			Model for lower-middle class		Model for middle class			Model for upper-middle class
Chi-squared	2067.667			864.293		988.180			982.751
RsqAdj	0.062			0.092		0.082			0.102

Note: figures in parentheses are significance levels.

In the total sample model, the constant is more negative in the model for the central districts than for the northern districts, and it is the least negative in the southern districts model. This implies that, in general, consumers are less likely to choose to live in the central districts compared to the northern districts, and that the southern districts are the most attractive housing sites. The coefficient of house price has a very strong negative linear sign and also a negative quadratic sign in all three submarkets, which means that the choice probability decreases more when price increases from level 2 to level 3 than from level 1 to level 2. The strong negative effects indicate that house price is one of the determinants of housing choice. Environmental quality in the neighbourhood shows strong a positive linear sign in all three submarkets. This indicates environmental pollution is also the major concern in housing choice. If the neighbourhood environment is polluted, the choice probability decreases. The quadratic sign is negative in the central districts and the northern districts, but positive in the southern districts. It indicates in the central districts and the northern districts that the choice probability increases less from some to no pollution than from a lot to some, while in the southern districts the choice probability increases more when environment quality improves. Consumers are tolerant on some pollution in the central districts and the northern districts, but demand higher environmental standards in the southern districts. School quality and safety also show significant positive coefficients on housing choice in linear and negative coefficients in quadratic components only in the central and southern districts. It means the choice probability increases less when school quality and safety are above the average level. Accessibility to metro and to jobs shows significant positive coefficients in linear components on housing choice only in the southern and northern districts. It might be because the central districts have relatively good accessibility, and consumers do not count the difference between 30–40 minutes and 20–30 minutes. In the southern districts, the quadratic sign of accessibility to the metro is negative, indicating that choice probability increases less when time cost to metro is within 20 minutes. In the northern districts, the quadratic sign of accessibility to metro is positive, indicating that choice probability increases more when accessibility level to metro improves. In the same vein, choice probability increases less when time cost to job is within 30 minutes in the southern districts, and choice probability increases more when accessibility level to job improves in the northern districts. Among the green dwelling attributes, only construction materials show small positive coefficients in linear and quadratic components only in the central and southern districts. It means consumers demand at least somewhat non-toxic construction material in the central and southern districts.

The housing choice models for the lower-middle class show that they mainly consider housing in northern districts. House price has very strong negative coefficients in the three submarkets. Environmental pollution also has relatively strong influences on their housing choice. Looking at Table 4, lower-middle class would pay 1841 yuan/m2 more for an unpolluted environment compared to some pollution and 292 yuan/m2 more for some pollution compared to a lot of pollution in the southern districts. But in the northern districts, lower-middle class would pay 593 yuan/m2 more for unpolluted environments compared to some pollution and 1171 yuan/m2 more for some pollution compared to a lot of pollution. It might be because the lower-middle class understands that some air pollution in this area is unavoidable and accept the relatively bad air quality. Job accessibility impacts housing choice as well. In Table 4, lower-middle class would pay 71 yuan/m2 more for access to their job within 20–30 minutes compared to a time of 30–40 minutes and 1103 yuan/m2 more for access to their job with 30–40 minutes compared to a time of more than 40 minutes in the northern districts. Understandably, green attributes in the northern districts are not significant on their list of priorities. To quote a respondent, ‘The green attributes are the last thing we are willing to pay for given our limited budgets. We chose this area for a temporary stay, so we do not expect a high-quality dwelling’.

Table 4.

Willingness to pay for housing attributes by three socio-economic groups (yuan/m2).

Central districts					Southern districts					Northern districts
Attribute	Whole	Lower	Middle	Upper	Attribute	Whole	Lower	Middle	Upper	Attribute	Whole	Lower	Middle	Upper
School quality					School quality					School quality
Average→Good	2189*	4598	2099*	1586*	Poor→Average	1291*	1082*	925*	3153*	Poor →Average	146	7293	261	0
Good→Very good	1154*	361	993*	2040*	Average→Good	679*	864*	39*	2703*	Average→Good	250	−6928	686	0
Environmental pollution					Environmental pollution					Environmental pollution
A lot→Some	4118*	856	262	5369*	A lot→Some	2085*	292*	694*	8979*	A lot→Some	1528*	1171*	2817*	0
Some→No pollution	3754*	3052	2090	4996*	Some→No pollution	3253*	1841*	2169*	11171*	Some→No pollution	1146*	593*	2496*	0
Safety from crime					Safety from crime					Safety from crime
Poor→Average	5647*	165*	4920*	6024*	Poor→Average	3166*	1163	1579*	−1922*	Poor→Average	521	441	954	0
Average→Good	−134*	6072*	−56*	−588*	Average→Good	−290*	174	−152*	11667*	Average→Good	−283	−126	−559	0
Social status					Social status					Social status
Low→Middle	1051	2879	1228	901	Low→Middle	746	−236	30	4129*	Low→Middle	135	−155*	1721	0
Middle→High	−301	−1086	187	−1065	Middle→High	−285	659	443	−3273*	Middle→High	−325	−439*	−358	0
Neighbours					Neighbours					Neighbours
Do not know→Know	477	4454	−853	596*	Do not know→Know	512	435	300	811	Do not know→Know	−253	−113	−775	0
Know→Have friends	395	−515	581	864*	Know→Have friends	196	641	226	240	Know→Have friends	−96	100	−626	0
Accessibility to metro stop					Accessibility to metro stop					Accessibility to metro stop
20–30mins→10–20mins	−403	−2268*	−1828	−112	20–30mins→ 10–20mins	679*	771	526*	1036	>50mins→40–50mins	242*	462*	−75	0
10–20mins→ <10mins	422	−2825*	1584	1191	10–20mins→ <10mins	341*	−367	615*	1081	40–50mins→30–40mins	404*	21*	1297	0
Accessibility to job					Accessibility to job					Accessibility to job
30–40mins→20–30mins	985	−3216	3008*	1809*	>40mins→30–40mins	316*	858	−207	1832*	>40mins→30–40mins	1107*	1103*	1610*	0
20–30mins→<20mins	−671	3062	−28*	−1943*	30–40mins→20–30mins	689*	516	413	2147*	30–40mins→20–30mins	338*	71*	760*	0
Energy and water costs					Energy and water costs					Energy and water costs
High→Average	73	−2402	1022	737*	High→Average	−152	−628	−138	1051	High→More than average	−174	−194	−626	0
Average→Low	461	938	−2530	1139*	Average→Low	−132	−50	−300	150	More than average→Average	110	168	−179	0
Construction materials					Construction materials					Construction materials
Harmful→Somewhat non-toxic	1042*	5196*	984	834*	Harmful→Somewhat non-toxic	884*	833	256*	556*	Harmful→Somewhat harmful	76	165	−75	0
Somewhat non-toxic→Non-toxic	561*	−2835*	−234	797*	Somewhat non-toxic→Non-toxic	375*	−547	541*	2538*	Somewhat harmful→Somewhat non-toxic	59	−16	−611	0
Thermal insulation					Thermal insulation					Thermal insulation
Poor→Average	1047	4557	28	1087	Poor→Average	260	−429	364	330	Poor→Below average	7	−226	1699*	0
Average→Good	−1120	−4443	−684	−387	Average→Good	−106	858	−212	195	Below average→Average	310	160	1051*	0
Sound insulation					Sound insulation					Sound insulation
Poor→Average	1696	5701	1172	797	Poor→Average	−287	−386	−108	−1216	Poor→Below average	267	121	708	0
Average→Good	−485	−3361	−403	484	Average→Good	182	−740	438	901	Below average→Average	80	97	−30	0
Ventilation					Ventilation					Ventilation
Poor→Average	873	2887	506	186	Poor→Average	151	728	30	−766	None→Average	6	129	−252	0
Average→Good	−236	−1876	−2474	946	Average→Good	384	541	241	1712	Average→Above average	−42	−168	194	0

Note: * = attribute is significant at 0.05 level.

The housing choice models for the middle class show that the coefficients of the constants are negative in the central and northern districts, but positive in the southern districts. Coefficients of house price for middle class are less strong than for lower-middle class, implying house price is a less constraining factor for housing choice by the middle class. The coefficients of school quality and safety are similar to the total model, meaning that middle class demands at least average school quality and safety. This might be because the family composition of this class makes them care more about neighbourhood quality. Environmental quality also has strong effects on their housing choice. Looking at Table 4, the middle class would pay 2169 yuan/m2 more for an unpolluted environment compared to an area with some pollution and 694 yuan/m2 more for some pollution compared to a lot of pollution in the southern districts. In the northern districts, they prefer a pollution free environment. Middle class would pay 2496 yuan/m2 more for an unpolluted area compared to some pollution and 2817 yuan/m2 more for some pollution compared to a lot of pollution. Job accessibility impacts housing choice as well. Table 4 shows that the middle class would pay 3008 yuan/m2 more for access to their job within 20–30 minutes compared to a time of 30–40 minutes but are not willing to pay for a further reduction within 20 minutes in the central districts. And they would pay 760 yuan/m2 more for access to their job within 20–30 minutes compared to a time of 30–40 minutes and 1610 yuan/m2 more in 30–40 minutes compared to more than 40 minutes in the northern districts. This group shows some interests in green attributes, and they would pay 541 yuan/m2 more for non-toxic construction materials compared to somewhat non-toxic materials, and 256 yuan/m2 more for somewhat non-toxic materials compared to harmful materials in the southern districts.

The housing choice models for the upper-middle class show that the constants are negative in the central and southern districts, and extremely negative in the northern districts, indicating that the upper-middle class does not consider living in the northern districts. Coefficients of house price for the upper-middle class are less negative than for middle class in the southern and northern districts. The upper-middle class does not consider house price as an important factor in making housing choices. Coefficients of environmental quality show strong positive effects. The upper-middle class would pay 4996 yuan/m2 more for unpolluted areas compared to some pollution and 5369 yuan/m2 more for some pollution compared to a lot of pollution in the central districts. Due to their family composition, the upper-middle class also cares about the school quality, neighbourhood safety, and their neighbours. This group favours low energy costs and non-toxic construction materials. They would pay 797 yuan/m2 more for non-toxic construction materials compared to somewhat non-toxic materials, and 834 yuan/m2 more for somewhat non-toxic materials compared to harmful materials in the central districts, and 556 yuan/m2 and 2538 yuan/m2 respectively in the southern districts. Job accessibility in the southern districts has some impact on housing choice as well. They would pay 2147 yuan/m2 more for access to their job within 20–30 minutes compared to a time of 30–40 minutes and 1832 yuan/m2 more for 30–40 minutes compared to more than 40 minutes.

Conclusions and discussion

This article attempts to value the willingness to pay for green attributes relative to the regular housing attributes by different socio-economic groups. Our results show that rather than green dwelling attributes, house price, environmental pollution, and accessibility to metros and jobs are the important factors when homebuyers choose an apartment in Nanjing. The socio-economic status of homebuyers determines their purchasing power and willingness to pay for green attributes. Only the rich are prepared to pay for green apartments to improve their living comfort. To all, the notion of health is appealing as consumers are willing to pay for an unpolluted environment and for non-toxic construction materials used in buildings in good locations. Clearly, health plays a pivotal role in planning sustainable cities (Duhl, 2005; Kenzer, 1998; Stephens, 1996).

To be specific, the lower-middle class cannot afford dwelling units in the central districts, and the southern districts are just about attainable if they compromise substantially on their living space. As a consequence, their housing choices are narrowly confined to the relatively poor living environment found in the northern districts. Relatively clean air and acceptable job accessibility are important to them. They are not willing to pay for the green attributes as they cannot afford them. In contrast, the middle class is willing to pay for better living conditions in the southern districts. They demand a higher quality of living and are prepared to pay for unpolluted air and easy job accessibility. Given their family composition – many are small families with young children – they are concerned about the quality of schools in the neighbourhood. This group shows some interest in green attributes such as non-toxic construction materials. Finally, the upper-middle class shows the greatest flexibility to choose the best quality of housing. They prefer to live in the southern districts or even in the central districts but not in the northern parts at all. They are willing to pay a lot more to improve their living comfort compared to the other two socio-economic groups. Their choice of apartment is governed by attributes such as no pollution, good school quality, safety, social status of their neighbours, closeness of family and friends, and good job accessibility. Other important considerations are low energy costs and non-toxic construction materials.

Green apartment buildings are only attractive when they are located in a good neighbourhood which offers clean air and good job accessibility (i.e. central and well-developed surrounding districts). At present, green apartments command a very limited market share and lack a clear driving force, because of their high selling price that fails to match the affordability of major consumer groups. As a consequence, the green market itself will not be able to develop automatically. Our results contribute to the understanding of why current policies and practices seem to be ineffective in upscaling green building. If affordability is in the way, subsidising developers seems superfluous given the current price premium in the niche market for comfortable apartments. Several alternatives could be considered. The first might be to raise standards for non-toxic building materials and energy performance in all building constructions. In that case, the production of healthy construction materials and energy-efficient equipment will be obligated. This might raise the house prices in the short term. However, in the long term, the house price will be stable or even decline. Due to scale advantages, a bigger supply of these green materials might push the production price down in the market, which in turn would reduce the costs of green building. Besides, to stimulate the consumer market, the government could reallocate developers’ subsidies to means-tested subsidies for homebuyers and green production companies (material production, equipment production, technology production, etc.) to address the affordability constraints.

Our results also confirm that consumers’ green housing choices are influenced by both green dwelling attributes and the broad living environment. In fact, consumers care more about a cleaner environment than about saving energy. However, one fifth of Chinese cities are facing severe air pollution (UN-Habitat et al., 2012). The current policy of building green apartments in Nanjing does not contribute directly to air quality. Traffic and the heavy industry are the most prominent sources of air pollution. Restructuring old industrial sites can contribute to both cleaner air and rising land prices as central city locations become available for residential construction. The rapid extension of the metro system in combination with high-density residential (re)development is instrumental to reduce car use and pollution.

This article provides a socio-economic analysis of green housing preferences in Nanjing, China. Given similar economic and social restructuring processes elsewhere in the country, the same methodology can be applied. However, it should be noted that the stated preferences only indicate the homebuyers’ probability of choice. In practice, their actual consumption behaviour may differ.

Footnotes

Acknowledgements

The authors would like to thank Dick Ettema, Faculty of Geosciences, Utrecht University for his help in building models and designing questionnaires.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Ali

Al Nsairat

(2009) Developing a green building assessment tool for developing countries: Case of Jordan. Building and Environment 44(5): 1053–1064.

Cai

Zhong

. (2009) China building energy consumption: Situation, challenges and corresponding measures. Energy Policy 37(6): 2054–2059.

Cai

Yin

Wennerstern

(2013) From energy efficiency to integrated sustainability in housing development in China: A case study in a hot-summer/cold-winter zone in China. Journal of Housing and the Built Environment 28(2): 329–344.

Calcagni

(2012) Household willingness to pay for improved energy efficiency in the US rental housing market: Implications for Rocky Mountain Institute’s superefficient housing initiative. Unpublished Master’s Thesis, The Sanford School of Public Policy, Duke University, NC.

Chau

Tse

Chung

(2010) A choice experiment to estimate the effect of green experience on references and willingness-to-pay for green building attributes. Building and Environment 45(11): 2553–2561.

Chen

Clark

Gottschang

. (2001) China Urban: Ethnographies of Contemporary Culture. Durham, NC: Duke University Press.

Cole

(2000) Editorial: Cost and value in building green. Building Research & Information 28(5–6): 304–309.

Curran

Hamilton

(2012) Just green enough: Contesting environmental gentrification in Greenpoint, Brooklyn. Local Environment 17(9): 1027–1042.

Duhl

(2005) Healthy cities and the built environment. Built Environment 31(4): 356–361.

10.

Eves

Kippes

(2010) Public awareness of ‘green’ and ‘energy efficient’ residential property: An empirical survey based on data from New Zealand. Property Management 28(3): 193–208.

11.

Guo

Yue

. (2010) Landscape design method for a green community based on green building design theory. Journal of Zhejiang University: Science A 11(9): 691–700.

12.

Hagoort

(2006) The neighbourhood rules: Landuse interactions, urban dynamics and cellular automata modeling. PhD Dissertation, Utrecht University, The Netherlands.

13.

Hald

(2009) Sustainable urban development and the Chinese eco-city: Concepts, strategies, policies and assessments. Fridtjof Nansen Institute 2009(5): 1–93.

14.

Hamilton

Morgan

(2010) Integrating lidar, GIS and hedonic price modeling to measure amenity values in urban beach residential property markets. Computers, Environment and Urban System 34(2): 133–141.

15.

Haurin

Brasington

(1996) School quality and real house prices: Inter- and intrametropolitan effects. Journal of Housing Economics 5(4): 351–368.

16.

Hite

(2009) Factors influencing differences between survey and market-based environmental value measures. Urban Studies 46(1): 117–138.

17.

Howie

Murphy

Wicks

(2010) An application of a stated preference method to value urban amenities. Urban Studies 47: 235–256.

18.

Jenks

(2000) The acceptability of urban intensification. In: Williams

Burton

Jenks

(eds) Achieving Sustainable Urban Form. London: E & FN Spon, pp. 242–250.

19.

Kenzer

(1998) Healthy cities: A guide to the literature. Environment and Urbanization 11(1): 201–220.

20.

Kibert

Grosskopf

(2007) Envisioning next generation green buildings. Journal of Land Use and Environmental Law 23: 145–160.

21.

Kientzel

Kok

(2011) Environmental assessment methodologies for commercial buildings: An elicitation study of US building professionals’ beliefs on leadership in energy and environmental design (LEED). Sustainability 3(12): 2392–2412.

22.

Kockelman

(1997) Effects of location elements on home purchase prices and rents in San Francisco Bay Area. Transportation Research Record 1606: 40–50.

23.

Leaman

Bordass

(2007) Are users ore tolerant of ‘green’ buildings? Building Research and Information 35(6): 662–673.

24.

S-m

(2012) Housing inequalities under market deepening: the case of Guangzhou, China. Environment and Planning A 44(12): 2852–2866.

25.

Jim

(2010) Willingness of residents to pay and motives for conservation of urban green spaces in the compact city of Hong Kong. Urban Forestry & Urban Greening 9(2): 113–120.

26.

Malmqvist

(2008) Environmental rating methods: Selecting indoor environmental quality (IEQ) aspects and indicators. Building Research and Information 36(5): 466–485.

27.

Man

Zheng

Ren

(2011) Housing policy and housing markets: Trends, patterns, and affordability. In: Man

(ed.) China’s Housing Reform and Outcomes. Cambridge, MA: Lincoln Institute of Land Policy, pp. 3–18.

28.

McKinley

(2009) The price is not right. Sustainable Business 157: 30–31.

29.

Molin

EJE

(2011) Conjoint analysis. In: Jansen

SJT

Coolen

HCCH

Goetgeluk

(eds) The Measurement and Analysis of Housing Preference and Choice. Dordrecht: Springer, pp. 127–156.

30.

Munroe

(2007) Exploring the determinants of spatial pattern in residential land markets: Amenities and disamenities in Charlotte, NC, USA. Environment and Planning B: Planning and Design 34(2): 336–354.

31.

(2002) Sustainable urban development issues in Chinese transitional cities: Hong Kong and Shenzhen. International Planning Studies 7(1): 7–36.

32.

Pan

Zhang

(2008) Rail transit impacts on land use: Evidence from Shanghai, China. Transportation Research Record 2048: 16–25.

33.

Portney

(2002) Taking sustainable cities seriously: A comparative analysis of twenty-four US cities. Local Environment 7(4): 363–380.

34.

Retzlaff

(2009) Green buildings and building assessment systems: A new area of interest for planners. Journal of Planning Literature 24(1): 3–21.

35.

Schweber

(2013) The effect of BREEAM on clients and construction professionals. Building Research and Information 41(2): 129–145.

36.

Stephens

(1996) Healthy cities or unhealthy islands? The health and social implications of urban inequality. Environment and Urbanization 8(2): 9–29.

37.

Torres

Greene

Ortúzar

JDD

(2013) Valuation of housing and neighbourhood attributes for city centre location: A case study in Santiago. Habitat International 39: 62–74.

38.

UN-Habitat, China Science Centre of International Eurasian Academy of Sciences, China Association of Mayors and Chinese Society of Urban Planning (2012) The State of China’s Cities 2012/2013. Beijing: UNHabitat and Foreign Languages Press.

39.

Vallance

Perkins

Dixon

(2011) What is social sustainability? A clarification of concepts. Geoforum 42(3): 342–348.

40.

Visser

van Dam

Hooimeijer

(2008) Residential environment and spatial variation in house prices in The Netherlands. Tijdschrift Voor Economische En Sociale Geografie 99(3): 348–360.

41.

Wang

S-m

(2004) Housing preferences in a transitional housing system: The case of Beijing, China. Environment and Planning A 36(1): 69–87.

42.

Wang

S-m

(2006) Socio-economic differentials and stated housing preferences in Guangzhou, China. Habitat International 30(2): 305–326.

43.

Wang

Borgers

Oppewal

. (2000) A stated choice approach to developing multi-faceted models of activity behavior. Transportation Research Part A: Policy and Practice 34(8): 625–643.

44.

Wang

(2003) Living conditions of migrants in inland Chinese cities. Journal of Comparative Asian Development 2(1): 47–69.

45.

White

Gatersleben

(2011) Greenery on residential buildings: Does it affect preferences and perceptions of beauty? Journal of Environmental Psychology 31(1): 89–98.

46.

Yau

(2012a) Eco-labels and willingness-to-pay: A Hong Kong study. Smart and Sustainable Built Environment 1(3): 277–290.

47.

Yau

(2012b) Willingness to pay and preferences for green housing attributes in Hong Kong. Journal of Green Building 7(2): 137–152.

48.

Yuan

Wang

Zuo

(2013) Renewable energy in buildings in China: A review. Renewable and Sustainable Energy Reviews 24: 1–8.