The Impacts in Real Estate of Landscape Values: Evidence from Tuscany (Italy)

Round 1

Reviewer 1 Report

This paper discusses an interesting issue. I believe it will be a valuable study.

1. The abstract is too simple. The author can revise it clearly, especially on the findings.
2. As we know the housing price is very high, why we can’t direct consumption? Why does do we need smart real estate? The paper could be more explain in details.

Author Response

We thank the reviewer for the valuable comments. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The study explores the relationships between real estate values in Tuscany with the 2 territorial variables (proximity to urban centres and roads) and an experimental variable relating to the aesthetic perception of different types of landscape. The variable perception of the landscape is measured through interviews on the aesthetic-visual perceptions of citizens of the different types of land use. Data were analysed using spatial econometric techniques and local spatial statistics. The results demonstrate the presence of spatial relationships between the aesthetic values of the landscape and the prices of properties, highlighting the importance of extrinsic characteristics in determining their prices.

I add some specific revisions below:

• Figure 1 is too generic, it would be better to introduce here a map of the study area, of the territorial units and of the analysed data.
• Formula (2), such as (1) and (3) must be described in its entirety in the text after the equation.
• Paragraph 3. Analysis of the variables: All the 4 variables analysed are correctly represented in the various subsections, but as stated (lines 236-238), the average values of each variable referred to the OMI area as territorial unit were used in the analyses. It is therefore suggested to put a "descriptive" map of each analysed variable (property prices, distance from roads, distance from built-up areas and aesthetic value of the landscape) referring to the territorial research units which is the OMI area, perhaps in addition to current maps. Also, it would be useful to add a table of descriptive statistics of the data sample.
• In paragraph 3.4 all the statistics on the interviews refer to previous publications, but to better clarify the results and facilitate reading it would be better if the authors added here a summary of those statistics on the data sample focusing on the section 2 of the questionnaire. What are the questions in the questionnaire? What are the points on the Linkert scale? What are the aesthetic evaluations of landscapes? How do you calculate the average aesthetic value?
• Figure 6: it is not clear on which variables the Multivariate Local Change was performed (all?), Please clarify this in the text. Furthermore, since significant and insignificant clusters are mentioned in the text, with the relative p-values, it would be better to add the corresponding map in Figure 6.
• Lines 281 and 292 refer to the relationship between property values and services, but since there is no direct link to services in this paper, it might add some reference to the literature (e.g. on some case study in Italy).
• Line 327-330 “Parameter p is significant and positive” seems not true for the variable “aesthetic value” please check the sentence. Furthermore, the good fit of an OLS model is verified by other tests, e. g. L M test, residual autocorrelation, AIC, etc., could you clarify this point better?
• some of the statements made in lines 381-388 should be better justified, particularly those regarding people's preferences for home buying and where to live. Please clarify these points. Also, the limitations section of the study is totally missing, which is good practice to add.

Author Response

We thank the reviewer for the valuable comments. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This article assesses the impact of landscape values on real estate prices applying spatial econometric techniques to an empirical case study in Tuscany. The methodology is built to assess the marginal impact of the quality of environment (using a proxy of satisfaction) on housing price (per square metre) controlling for proximity to urban services and roads.

I would agree that the spatial econometric procedure is well presented and likely appropriate despite relatively poor justification. However, from my point of view, the overall article is flawed with several major deficiencies. Following is a description of some of these deficiencies.

• The article is not rooted with an appropriate literature review on the topic (valuation of the environment and real estate prices). In fact, there is not a literature review section and the reference about real estate prices are very general and relate to other topics like transportation, access to amenities, etc. That is not bad, but clearly insufficient. There is a very well-developed body of literature dealing with the topic of landscape perception and valuation effect on house prices. Here are some examples in publication order:
  • Orland et al. 1992. The effect of street trees on perceived values of residential property. Environment and Behavior.
  • Dombrow J. et al. 2000. The market value of mature trees in single-family housing markets. The Appraisal Journal.
  • Luttik J. 2000. The value of trees, water and open space as reflected by house price in the Netherlands. Landscape and Urban Planning. (590 citations)
  • Des Rosiers F. et al. 2002. Landscaping and house values : An empirical investigation. J. of Real Estate Research.
  • Bourassa S.C. et al. 2003. The price of aesthetic externalities. SSRN Electronic Journal.
  • Cavailhès J. et al. 2005. The landscape from home: seeing and being seen. A GIS-based hedonic price valuation.
  • Tangerini A. 2005. Bringing the hedonic price method into fashion when valuing landscape quality.
  • Kong et al. 2007. Using GIS and landscape metrics in hedonic price modelling of the amenity value of green space. Landscape and Urban Planning.
  • Cavailhès et al. 2007. Pricing the homebuyer’s countryside view.
  • Kumagai Y. et al. 2008. Green space relations with residential values in downtown Tokyo. Implications for urban biodiversity conservation. Local Environment.
  • Tritz C. 2009. A quantitative approach to the visual evaluation of the landscape. Annals. Assoc. American Geographers.
  • Hilal M. 2009. Landscape metrics for determining landscape prices.
  • Donovan G. et al. 2010. Trees in the City: Valuing street trees in Portland, Oregon. Landscape and Urban Planning.
  • Biao Z. et al. 2012. The effects of public green spaces on residential property value in Beijing.
  • Li W. et al. 2012. A spatial hedonic analysis of the value of urban land cover in the multifamily housing market in Los Angeles, CA. Urban Studies.
  • Pham TTH. Et al. 2012. Spatial distribution of vegetation in Montreal: An uneven distribution or environmental inequity? Landscape and Urban Planning.
  • Hussain MRH. Et al. 2014. The impact of landscape design on house prices and values in residential development in urban areas.
  • Escobero FJ. Et al. 2015. Urban forest structure effects on property values.
  • Glasener ML. et al. 2015. Neighbourhood green and services diversity effect on land prices: evidence from a multilevel hedonic analysis in Luxembourg. Landscape and Urban Planning.
  • Trojanek R. et al. 2018. The effect of urban green spaces on house prices in Warsaw.
  • Donovan G. et al. 2019. Urban trees, house price, and redevelopment pressure in Tampa, Florida.
• Without a sound literature review, the findings of this article cannot be compared to similar work. But more importantly, most of the studies cited above rely on the price and attributes (intrinsic and extrinsic) of individual properties sold at various locations in a city (housing market) or region (sets of housing markets). That is in accordance with Sherwin Rosen (1974) hedonic theory presented in an article entitled: Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition, Journal of political economy. That means that the value of a property is “determined by the intrinsic and extrinsic characteristics” (line 33 of this article) compared with other properties in the market (differentiation in pure competition). This article is based on aggregated data based on submarket zones (OMI) where only one value is used for each zone (maximum price per square metre). Therefore, there is a need for a strong justification because that is totally incompatible with the hedonic theory and the present state of the literature in the domain. Going that way would imply developing a new theory dealing with several issues like ecological fallacy (Robinson, 1950) and multiple area unit problem (Openshaw, 1984) and, if intrinsic (property-level) attributes are not controlled for, the omitted-variable problem. In their analysis, the authors consider only extrinsic attributes because they do not model at the property-level but at an aggregated level (like municipalities in rural areas) that imply an assumption of homogeneity within each unit. That is obviously not the case if the hedonic theory is correct. Thus, my main point is that the level at which the analysis is carried out and the methodology to select variables are not justified and most likely faulty.
• Let us now look at the dependent variable: the maximum sale price by OMI in the second half of 2016 for civil housing. Why did the author decided to use the maximum market value remains unjustified and what is the square metre basis (floor space, lot size, with or without the garage, etc.) is not specified. Using the maximum value per zone means that only the higher segment of the housing market is considered. Why not use the median or the average? Here again there is a choice but no justification and no discussion of the consequences of this choice on the reliability of the results. The authors give the reference to the geodetic system used to locate the information. That is good practice. But they do not explain how they locate the prices and other variables associated to each data unit. What is the resolution of the location of points? What are the consequences on the distance they measure to service centres and the road network. Since they use metric variables at the local level, that is relevant information that must be justified and discussed.
• For me, the main problem with this analysis is the total absence of property intrinsic attributes. My knowledge of the literature in real estate analysis of housing prices let me estimate the contribution of intrinsic attributes to approximately 60% of the value, leaving 40% to the location-based attributes (mostly the externalities). Among these intrinsic attributes are its age (depreciation and renovation), building quality, property facilities (e.g. garden, plot size, garage, parking, landscaping on the premises, etc.), property aesthetics, etc. All these intrinsic attributes influence the value and, more importantly, are not homogeneous within an OMI. That obviously leads to the omitted variable problem that bias an analysis based only on extrinsic attributes.
• Let us now look at the extrinsic attributes used. The model control for distance to cities and to the road network. Those are good choices. However, looking at Figure 2, it is obvious that a view to the see impacts the maximum price of housing in the coastal OMI. This kind of impact is known since the article of Luttik (2000) and widely documented in the real estate literature. Therefore, there is also omitted variables in the extrinsic side of the model. I would argue about the same point with terrain because higher altitudes are widely associated with higher values thanks to the regional view provided (see Cavailhès 2007). Thus, a variable that assesses the quality of the landscape based only on land use is highly debatable and must be justified. To the best of my knowledge, Tuscany had a rather mountainous countryside. Thus, distance alone (Voronoi polygons) seems an extremely poor way to assess the quality of the landscape even at a regional level. Why not integrate a terrain model in the estimate of landscape value?
• Next issue is with the way the quality of the landscape is assessed. According to the authors, they have done 250 face-to-face interviews with forest users in Tuscany. The authors provide descriptive statistics of the pool of respondents. Among others, there are 29% of students. Are they good candidates to estimate the appreciation of the landscape among home buyers? My point is that the authors provide no discussion of such critical points. The presentation of the methodology to convert the Likert scales of the survey to a continuous variable for the regression model is explained but weakly justified. Are average values sufficient for that purpose? Do indices for various categories sum up efficiently? Is it not better to compute indices for various land use types? Is Corrine land use map at a sufficient resolution to do that type of analysis? They explain a nice GIS procedure based on Voronoi diagrams to relate the “aesthetic value” of the landscape to housing prices. I am not sure I understand how Voronoi polygons relate to OMI areas and how distances were estimated to populate the spatial weight matrices. In fact, is there one or two weight matrices in equation 4? If there is only one, how can it be used with two different land tessellations (OMI for price and Voronoi for Esthval)?
• A minor point, there are many figures in this article. Some of them are useful, others must be removed (e.g. Figure 1, Figure 4 – The distances are remarkably similar except for very small zones.). The legend of Figure 5 is not appropriate. The figure shows the land uses with a constant for aesthetic values. The title must refer to land use.
• Looking at Figures 6 and 7, I am not sure if the analysis presented in Table 2 (Spatial Durbin model) was made using OMI zones or Voronoi polygons.
• At lines 358-363, the authors write:” This work could represent a useful tool for decision-makers to planning territorial strategies considering the social, economic, and environmental aspects. Indeed, thanks to this spatial analysis, the policymakers could be assisted to help people to search for housing based on the services/characteristics of different geographical areas. Our results could also be used to analyse how the real estate situation is directly or indirectly related to the quality of life of citizens.” I am sorry, but I do not see how the very technical aggregate analysis presented can guide planners who are working with land features like buildings, amenities, and roads to decide based on Voronoi diagrams. Moreover, I do not see the link with the social aspects and the quality of life, even the economic side is disturbed considering only the high segment of the housing market while the price distribution is likely highly skewed to the right.
• There is an obvious lack of self-criticism of the methodology, the results and the findings of this article and almost no discussion that relates the findings to previous research in the similar topics. Therefore, I am sorry to recommend that it should be rejected because I consider that there are too many flaws for a major revision in an acceptable time.

Author Response

We thank the reviewer for the valuable comments. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Although the article describes one of the regions of Italy, in my opinion it is universal in nature through the notion of the subject of assessing the impact of the features of inefficiency on its value.

In fact, the authors emphasized that their research was based on two features - distance from the center and road access, I think it was worthwhile to assess how these features compare with others. I mean such a comparison whether, for example, the type of soil, or the type of buildings, will not dominate the features mentioned by the authors with their influence.

I assume that since these features are the focus, they have a significant impact on the value of the real estate. I will be very pleased if the authors devote a few sentences to this subject, I am sending you some examples of literature:

I would also like to ask the authors to mention other methods used in such analyzes, especially here I am inclined to the classification trees that are close to me, which would broaden the analyzed case. I do not suggest a revision of the article, but I think it is worth mentioning such analyzes, if they were in the first part of the work - as a literature study. 

Fan, Gang-Zhi, et al. “Determinants of House Price: A Decision Tree Approach.” Urban Studies, vol. 43, no. 12, 2006, pp. 2301–2315. JSTOR, www.jstor.org/stable/43198334. Accessed 3 Jan. 2021.

Jasińska, E.; Preweda, E. Determining the cadastral-tax areas for the real estate premises based on the model of qualitative and quantitative. In Proceedings of the Environmental Engineering 10th International Conference, Vilnius, Lithuania, 27–28 April 2017

Jasińska, E.; Preweda, E. The Use of Regression Trees to the Analysis of Real Estate Market of Housing. In Proceedings of the GeoConference on Informatics, Geoinformatics and Remote Sensing: 13 International
Multidisciplinary Scientific Geoconference, Albena, Bulgaria, 16–22 June 2013; Volume 2, DOI:10.5593/SGEM2013/BB2.V2/S09.065

Case, B., Clapp, J., Dubin, R. et al. Modeling Spatial and Temporal House Price Patterns: A Comparison of Four Models. The Journal of Real Estate Finance and Economics 29, 167–191 (2004). https://0-doi-org.brum.beds.ac.uk/10.1023/B:REAL.0000035309.60607.53

Bencure, J.C.; Tripathi, N.K.; Miyazaki, H.; Ninsawat, S.; Kim, S.M. Development of an Innovative Land Valuation Model (iLVM) for Mass Appraisal Application in Sub-Urban Areas Using AHP: An Integration of Theoretical and Practical Approaches. Sustainability 2019, 11, 3731

I have no further comments.
Thank you

Author Response

We thank the reviewer for the valuable comments. Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

I have read the reviewers reports and the author response. I am globally satisfied with the modifications done to the article. The revised objectives are more realistic and state clearly the purpose of the paper with appropriate references to previous work at different scales. The revised article emphasizes the spatial autoregressive structure and provides a clearer statement of the limits and originality. The methodological part is appropriate and better justified than the previous version. Therefore, I now recommend acceptance. For me, it is a spatial analysis paper dealing with real estate market, but not a real estate value research that would need more specific data and a different scale of analysis.