Skip to main content

Advertisement

SpringerLink
Log in

The Value of Scattered Greenery in Urban Areas: A Hedonic Analysis in Japan

  • Published:
Environmental and Resource Economics Aims and scope Submit manuscript

Abstract

This study investigates the impact of scattered greenery (street trees and yard bushes), rather than cohesive greenery (parks and forests), on housing prices. We identify urban green space from high-resolution satellite images and combine these data with data on both condominium sales and rentals to estimate hedonic pricing models. We find that scattered urban greenery within 100 m significantly increases housing prices, while more distant scattered greenery does not. Scattered greenery is highly valued near highways, and the prices of inexpensive and small for-sale and for-rent properties are less affected by scattered greenery. These results indicate that there is significant heterogeneity in urban greenery preferences by property characteristics and location. This heterogeneity in preferences for greenery could lead to environmental gentrification since the number of more expensive properties increases in areas with more green amenities.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3

Similar content being viewed by others

Data Availability

The data on green coverage can be provided upon request. The real estate data are restricted and were used under license for this study. Other geographic data are publicly available and can be freely obtained from official government websites.

Notes

  1. Technically, these passages are not streets but are called “cho-chos.” A cho-cho is the smallest geographical unit in Japan and is similar in concept to a street in the U.S. For simplicity, this paper uses the term “street.”.

  2. We assume that using April data from one year and October data from another year does not cause serious problems because the region does not experience significant changes in plant conditions except during the winter (December-February). However, given the concern that the difference in green cover between 2008 and 2013 is due to the month of observation, this study does not focus on the increase or decrease in green cover from 2008 to 2013 but only on the change in the impact of green cover on the real estate market in each year. Due to budget constraints, other data were not available, and this study is limited by the inability to consider changes in vegetation due to seasonal differences.

  3. The green coverage identified using only NDVI images contains misclassified objects. Therefore, we confirmed and corrected these misclassified areas with the support of JAPAN SPACE IMAGING CORPORATION, a company specializing in satellite image manipulation.

  4. In the 2009 Urban Area Land Use Subdivision Mesh Data, forests within parks are classified as “parks,” but in 2014, they are classified as “forests.” This is because the category classification was changed by the MLIT and not because the actual land use has changed. Since almost all forests in the area are within a parks, parks and forests are treated the same as when creating the variables.

  5. To facilitate comparison with recent related studies (e.g., Wu and Rowe 2022), 100-m intervals are used. To account for errors caused by the longitude and latitude information of the property and the shape of the building, the nearest greenery is defined as within 100 m. The upper limit is set at 500 m, since the living distance on foot in urban areas in Japan is generally approximately 500 m (Hoshino 2011). To consider the validity of our buffer intervals, we also performed an analysis using 50-m intervals, and the results were consistent (The results can be provided upon request).

  6. Apartments (condominiums) are important when effectively using small, densely populated areas, such as those in Tokyo, and are the main option for residential housing. Our data include detached properties, but the number of transactions is very small, and the transaction prices are extremely high. Additionally, detached houses are able to have more greenery in their own yards, causing endogeneity problems in the estimation. Thus, we focus on the price of or rent for apartments.

  7. Our original property dataset covers the entire Tokyo area, with 146,494 and 895,394 properties for sale and rent, respectively, during the analysis period. Extracting properties from the original dataset for which the exact longitude and latitude can be determined from the address and the property name, the sample size is 142,482 (97.3%) for sales and 744,167 (83.1%) for rentals. Of that sample, 17,847 and 144,534 for sales and rentals, respectively, are located within our satellite coverage. Therefore, the substantial sample survival rates are 98.3% (from 17,847 to 17,552) and 91.1% (from 144,534 to 131,713) for sales and rentals, respectively.

  8. The zones of a location define the types of buildings that can be constructed in these areas (low-rise residential, high-rise residential, commercial, industrial, etc.), and the building-to-land ratio and floor-area ratio are also defined for each zone. By controlling for the fixed effects of the zones, the estimation considers the effects of confounders such as the size of the yard and the height of the building.

  9. We also performed an estimation with price/rent per square meter as the explained variable, and the results were very similar to the main results. The results table can be made available upon request.

  10. The study area is a well-developed urban area, and as Figs. 1 and 2 show, the other types of greenery (e.g., parks and waterfront greenery) are scarce and unevenly distributed. Therefore, this study uses green spaces other than scattered greenery as a control variable only and does not provide a detailed interpretation of the corresponding impact.

  11. The results for sales properties with scattered greenery in 2008 indicate that scattered greenery within 300–400 m hurts sales prices. This negative effect is still observed after several robustness checks, but it is not consistent over varying distances or through the analysis years and is of low statistical significance. For scattered greenery away from home, the degree and frequency of contact vary greatly depending on people's living areas and commuting routes. Therefore, data such as visibility and frequency of use are needed to provide robust evidence of the impact of scattered greenery at a distance. Hence, we do not interpret the effect of the far distance band and leave it as a limitation of this study and as a topic for future work.

  12. Approximately 34% and 28% of scattered greenery was excluded in 2008 and 2013, respectively.

  13. We have confirmed that the parks, forests, and rivers in our study area have not changed significantly in the last 10–15 years. It should be noted, however, that different measurement errors can occur than in the main analysis.

  14. Based on the 2013 Housing and Land Survey, the density of floor space of sales apartments in the entire Setagaya and Suginami wards is approximately 9.26%.

  15. Since the cost of greenery in Setagaya is not available, only the values for Suginami are used here. Costs vary widely depending on the type of tree or grass, but average values are used here. Additionally, since we know only the cost per tree for street trees, we assume, based on our data, that approximately 25 square meters of green coverage is associated with one street tree. According to the 2018 Tokyo Greening White Paper, the average additional and maintenance costs per square meter of street trees (planted strips) in Suginami are 1,140 JPY (1,072 JPY) and 569 JPY (208 JPY), respectively. Since the ratios of the area of street trees and planted strips in Suginami Ward is 72% and 28%, respectively, we estimate that the average additional and maintenance cost per square meter of scattered greenery would be approximately 1,121 JPY and 468 JPY, respectively.

  16. Those who are concerned about risks, such as problems with neighbors or damage from disasters, could live in rental properties that are easy to move out of.

  17. The singles who most commonly live in rental properties in urban areas in Japan are university students. In Japan, universities are concentrated in large cities; thus, many university students leave their hometowns to live alone. Therefore, many students reside near the university only for four years while completing their studies and move out when they graduate.

References

Download references

Acknowledgements

We deeply appreciate the helpful comments and suggestions provided by Kentaro Nakajima, Yuta Uchiyama, Michio Naoi, and two anonymous referees. We would also like to thank the participants at the Annual Conference of the Society for Environmental Economics and Policy Studies in October 2022, Annual Meeting of the Applied Regional Science Conference in December 2022, and the Special Lecture at Nihon University in December 2022. This work benefited from a project funded by the Housing Research and Advancement Foundation of Japan. Satellite images and vegetation data were collected and generated in cooperation with JAPAN SPACE IMAGING CORPORATION. The views expressed are those of the authors and do not necessarily reflect those of any organizations with which the authors are affiliated.

Funding

This study was funded by the Housing Research and Advancement Foundation of Japan.

Author information

Authors and Affiliations

  1. Graduate School of Economics, Osaka Metropolitan University, 3-3-138, Sugimoto, Sumiyoshi-ku, Osaka, 558-8585, Japan

    Yuta Kuroda

  2. Housing Research and Advancement Foundation of Japan, 6-3 Nibancho, Chiyoda-Ku, Tokyo, 102-0084, Japan

    Takeru Sugasawa

Authors
  1. Yuta Kuroda
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Takeru Sugasawa
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

All authors contributed to the study conception and design. Data collection and analysis were performed by YK, and TS. The first draft of the manuscript was written by YK and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Yuta Kuroda.

Ethics declarations

Conflict of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Additional figures and tables

See Figs.

Fig. 4
figure 4

Comparison of aerial photography and NDVI green coverage areas around residential areas. The aerial imagery is based on open data created in 2016, published by Setagaya Ward

Full size image

4,

Fig. 5
figure 5

Comparison of aerial photography and NDVI green coverage areas around the park, including sports fields. The aerial imagery is based on open data created in 2016, published by Setagaya Ward

Full size image

5,

Fig. 6
figure 6

Green coverage by type in 2008 (classification by digital map). The location and amount of greenery are based on satellite images from 2008. The classification of green spaces is based on the 2022 edition of the digital map published by the Geospatial Information Authority of Japan

Full size image

6,

Fig. 7
figure 7

Green coverage by type in 2013 (classification by digital map). The location and amount of greenery are based on satellite images from 2013. The classification of green spaces is based on the 2022 edition of the digital map published by the Geospatial Information Authority of Japan

Full size image

7 and

Fig. 8
figure 8

Subsample analysis by quartiles. In each panel, each point represents the coefficient of scattered greenery within 100 m, estimated using subsamples from the first through fourth quartiles, respectively, as indicated by the horizontal axis. The confidence intervals around the point estimates reflect 95% confidence intervals, and standard errors are adjusted for clustering at the street level. The independent variables are the same as those listed in Appendix Table 10

Full size image

8 and Tables

Table 6 Summary statistics
Full size table

6,

Table 7 Nonlinear effects of scattered greenery
Full size table

7,

Table 8 Correlations among scattered greenery by distance band
Full size table

8 and

Table 9 Individual estimations by distance bands of scattered greenery
Full size table

9.

Full Results of the Main Estimations

See Tables

Table 10 Effects of scattered greenery on property prices and rents (full results)
Full size table

10,

Table 11 Robustness checks
Full size table

11,

Table 12 Alternative definitions of scattered greenery
Full size table

12,

Table 13 Subsample analysis
Full size table

13 and

Table 14 Effects of scattered greenery on property quality and surrounding amenities
Full size table

14.

Comparison of Sales and Rental Properties

See Tables

Table 15 Estimated results using rental properties near sales properties
Full size table

15,

Table 16 Number of households by household type and income (Setagaya and Suginami, 2013)
Full size table

16,

Table 17 Number of households by head count (Setagaya and Suginami, 2013)
Full size table

17 and

Table 18 Number of households by years living in the current house (within 10 km of CBD)
Full size table

18.

Our findings indicate that sales and rental properties are heterogeneously affected by scattered greenery. This appendix provides some analysis and discussion of the causes. First, the locations of the sales and rental properties could be different. Rental properties tend to be located near commercial areas because they are preferred by single people and students, who are more likely to move within short periods. In contrast, sales properties tend to be in quiet residential areas because they are more likely to be owned by family households that remain in place for a long time. To address this concern, we created a subsample of the closest rental apartments to each of the sales apartments in our data. If the closest rental property overlapped, the second, third, and so on were matched, and all properties were matched on a one-to-one correspondence. If location is an important cause, then rental properties that are similar in environment to sales properties could be significantly affected by scattered greenery.

Columns (3) and (4) of Appendix Table 15 present the estimation results using a subsample of rental properties in a similar environment to the sales properties, and the results are almost the same as the main results. To focus on rental properties that are more similar in the surrounding environment to the sales property, we also conducted an analysis using only rental properties within 100 m of the corresponding sales property. The results are shown in columns (5) and (6) and indicate that the rental properties are not significantly affected by scattered greenery. Additionally, we estimated using rental properties included in buildings where rooms have been marketed as sales properties. Columns (7) and (8) present the results, showing that even if the surrounding environment is the same as that of a sales property, rental property is not affected by scattered greenery. The results indicate that location does not explain the heterogeneous responses between sales and rentals. Additionally, since rooms in the same building respond differently to sales and rentals, it is unlikely that the difference between sales and rentals is caused by the surrounding environment or interior design. The results in Appendix Table 15 also indicate that the amount of scattered greenery and the quality of housing are associated with both sales and rental properties. Therefore, the heterogeneous response of sales and rental properties could be due to the characteristics of the residents.

Residents of sales and rental properties differ greatly in income, age, number of family members, and other characteristics, resulting in marked variations in the number of years lived on the properties. Unfortunately, our property data do not provide information on resident characteristics. As an alternative, we attempt to explain the causes of the difference in response between sales and rental properties based on the average resident demographics of the study area and the findings of previous studies. Some previous studies focused on the heterogeneity of residential environment preferences.

According to Hoshino (2011), who conducted a survey of Tokyo residents, accessibility to commercial areas is preferred, on average, but 30% of respondents did not want to live in commercial areas, suggesting strong heterogeneity in residential location preferences. People's preferences are heterogeneous by socioeconomic characteristics, with people from higher socioeconomic backgrounds tending to have a higher willingness to pay for urban green spaces (Schindler et al. 2018) and preservation of greenery (Tian et al 2020). Łaszkiewicz et al. (2019) also suggested that green space is a luxury good and that individuals with higher incomes are likely to be more environmentally oriented. In our study area, residents of sales properties have higher incomes, on average, than residents of rental properties. Appendix Table 16 shows that the ratio of households with more than 5 million JPY income is approximately 60% for sales properties but approximately 37% for rental properties. Therefore, the income of residents could be the source of heterogeneity in their response to scattered greenery.

The composition of households in sales and rental properties differs considerably. In Japan, couples and families with children tend to live in sales properties, while singles and university students tend to live in rental properties. Appendix Table 16 shows that in the study area, approximately 65% of family households live in sales properties, in contrast to approximately 25% of singles who live in sales properties. Additionally, approximately 40% of households living in sales properties have three or more members, while only approximately 15% of households living in rental properties have three or more persons (Appendix Table 17). Hammitt and Haninger (2017) indicate that the willingness to pay to reduce the risk of others in the household is significantly greater than the willingness to pay to reduce one's own risk. It has also been suggested that elderly people and children, who are physically weaker and more concerned about health risks, tend to value greenery that improves air quality (Cameron et al. 2010; Liu, Hanley, and Cambpell, 2020). Therefore, the family structure of the residents could also be a factor explaining the heterogeneity of responses to scattered greenery. The results of the subsample analysis, which showed that even for sales properties, single rooms are not affected by scattered greenery, suggest that not being a single person could be an important cause of the difference. The finding that only property buyers, not renters, appreciate scattered greenery near highways is also consistent with the fact that elderly people and children, who are more concerned about health risks, tend to live in sales properties.

There are several reasons for such differences in resident characteristics between sales and rentals, but the mortgage tax break could be one reason. In Japan, if one purchases a house with a loan, 0.7% of the outstanding loan balance each year is deducted from income tax for up to 13 years. Therefore, it is more beneficial to buy a residence than to rent one if one lives in the same location for many years. In contrast, if one is likely to move within several years or does not have sufficient income to qualify for a loan, one chooses to live in a rental property.Footnote 16 Therefore, families with children and elderly people who do not frequently relocate tend to live on sales properties; conversely, students or singles tend to live on rental properties.

Appendix Table 18 shows the number of households within 10 km of the CBD by years of residence in the current house, indicating that the number of years of residence for owned and rented households is quite different. More than 60% of households living in sales properties have lived in their current home for more than 13 years, and approximately 35% have lived in their current home for more than 28 years. In contrast, approximately 40% of households living in rental properties have lived in their current homes for less than 2 years, and approximately 80% have lived in their current homes for less than 12 years.Footnote 17 While residents of rental properties can easily move out if they encounter undesirable surroundings, this is not the case for sales properties. Thus, residents of sales properties are likely to value the surrounding environment more. Additionally, the surrounding environment, such as good air quality and beautiful landscapes, affects people's physical and mental health over time. Therefore, the expected years of residence could lead to heterogeneity in the valuation of the surrounding environment.

We note that the explanations given above are only suggestive evidence. These factors, such as socioeconomic background, income, number of household members, and years of residence, are correlated with each other. For example, people from higher socioeconomic backgrounds may have higher annual incomes and be therefore more likely to marry and have children, resulting in longer residence in larger homes. It is also important to note that the difference in response between sales and rental properties does not necessarily indicate people's potential preferences. Because the hedonic pricing approach focuses on the value realized in the market, it cannot identify whether residents are not interested in scattered greenery or do not have the ability to pay or whether homes with the combination of desired characteristics do not exist in the market (Sander and Zhao 2015). Therefore, the mechanisms behind the heterogeneity of sales and rental properties need to be analyzed using more detailed data and precise methodologies.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kuroda, Y., Sugasawa, T. The Value of Scattered Greenery in Urban Areas: A Hedonic Analysis in Japan. Environ Resource Econ 85, 523–586 (2023). https://doi.org/10.1007/s10640-023-00775-5

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10640-023-00775-5

Keywords

JEL Classification

Search

Navigation