Residents’ Demands for Urban Retail: Heterogeneity in Housing Structure Characteristics, Price Quantile, and Space

Round 1

Reviewer 1 Report

Dear authors,

I find your paper truly interesting. The methodology and data is properly used, the introduction is well structured, and the discussion of results is fascinating. The only question that should be addressed in the paper is the employability of the analysis to rural areas, which are lately in the verge of disappearance. 

Author Response

Point 1: I find your paper truly interesting. The methodology and data is properly used, the introduction is well structured, and the discussion of results is fascinating. The only question that should be addressed in the paper is the employability of the analysis to rural areas, which are lately in the verge of disappearance.

Response 1:Many thanks for the valuable suggestion.

Exploring the impact of UR development on employability of rural areas is great point. However, the real estate agency only contains the housing data in urban area. We have added the limitation in conclusion, and considered using the land price and other POI data to explore this important issue in future study.

Lines 459-462:

Finally, because of the limitation of data, this study does not further discuss the relationship between the residents’ demand for urban retail and the employment capacity created by retail, especially for rural areas. These may have some implications for governments’ retail layout plans.

Author Response File: Author Response.docx

Reviewer 2 Report

Notes:

-The main problem is the paper is not well structured, it contains a huge amount of flaws and it is a huge challenge for the reader to read the paper, because line of arguments is not identifiable. The introduction is a disaster, because the authors do not provide a good reasoning why their research is necessary. For me it remains unclear for what purpose this research is good. The problem is, not one of the quantitative results can be used for any purpose, because the preferences of the citizens are changing in time. If I read the conclusion I come to result that not one outcome of the paper is in some sense qualitatively novel. All outcomes are as expected by the reader. The authors themselves do not provide one idea how their outcomes can be used to improve the situation of citizens. Additionally, the authors do not explain, why they only used second-hand housing transactions. Then the authors exclude villas and townhouses to avoid a bias, but do not argue which bias they expect.

-The English, particularly the diction is very often flawed.

-The authors make fundamental mistakes, which create doubts, if they really understand the underlying issues. For example, they write (line 41) “Residents’ willingness to pay for housing is affected by their demands for UR”. That is nonsense, because the willingness to pay is represented by the demand function and this is course Economics 101 knowledge. That means the authors write about something what they do not understand.

-Further, instead of reasoning their statements, the authors put a reference at the end of statements so that the reader has to find out if the statement makes sense or not. For example, line 42-44; “At the macroeconomic level, the development of UR plays an important role in promoting economic development and improving the value of cities and regions [7,10]”. I am working as an economist since nearly 30 years and I have read hundred of textbooks and thousands of papers, but I have never noticed any theory in which the value of the city has an influence on macroeconomic variables. I also have read the two references, but the references do not confirm the statement of the authors.

Another problem is that I have no clue what the authors have in mind with “understanding the heterogeneity of residents’ demand”. Of course, that the demand functions of individuals are not identical is obvious, so what means the term above? Demand functions are derived from preferences and the authors want to understand the preferences? However, this paper does not explain the origins of the preferences. Anyways, economists take preferences as given Or do the authors mean the demand of heterogeneous residents? That is something total different, but that would make sense. The next problem is what do the authors have in mind with “Urban Retail” the capital letters indicate a name, but that makes no sense. I assume they have something in mind like where and how much shops and the like should be located in a city, and the term “distribution of urban retail” means how the urban retail system distributes the goods they sell, but that makes not much sense in the context of the authors.

-I have no clue, why the authors cite Elkington, (1998),"ACCOUNTING FOR THE TRIPLE BOTTOM LINE”. The reference does not fit in the context, because Elkington writes about accounting in companies. I would argue the authors’ definition of sustainable development is misleading. I also do not see what sustainability has to do with the research question. In the context of the paper the word “sustainable” is only used as buzz word like policy-maker abuse this term.

-The statement, “Currently, with the growth of urbanization, urban problems such as land shortage, traffic congestion, and environmental pollution have emerged [2].” What is growth of urbanization? Urbanization cannot grow. Problems like shortage of land, traffic congestion, and environmental pollution have currently emerged? I fear the authors have no knowledge of geography and history of cities. Since ancient times space is scarce in cities. For what is the reference (2) necessary? It is an abundant reference. Statements, which are common knowledge, do not require a reference. Nobody doubts that traffic jams occur in all cities around the planet.

-Or a statement like “Bayer [46] has provided empirical evidence on families from different social classes and indicated that the marginal willingness to pay increases with income. This effect cannot be observed directly using traditional HPM [47].” I do not think that any HPM is necessary to come to the conclusion that the willingness to pay for a normal good increases with income, that is common knowledge and this common knowledge can be found in any microeconomic textbook.

-In the same line are the research questions. Either they are trivial, like, “Is there heterogeneity in the demand for UR between the owners of high- and low-price houses? Or they will not be answered like “If so, why and how large is the gap?” or “Which locations are better able to meet the residents’ demands for UR and increase housing prices by improving UR?”

-Regarding the results, the authors do not provide an intuitive explanation or interpretation for their results.

-Regarding the variables sometimes the reader has no idea, what is meant. For example “property management” is either “need improve, low, mid or high”, but what does that mean precisely? The same with the variable “study area”. And then the authors get outcomes like “…which means that PM and YEAR have a direct capitalization effect on housing prices and that a 1% increase in PM and 1% decrease in YEAR will both achieve a 1% increase in housing price.” How can PM increase by 1%, if PM can only take 4 ordinal values? Such statements are nonsense. Such flaws raise the concern that the authors do not understand what they are doing.

-It is absolute annoying that the authors use always abbreviations in their statements, because the reader cannot remember all 20 or more abbreviations while reading the text.

-Furthermore, the authors applied different empirical methods, but they do not discuss in detail the advantages and disadvantages of each method. That creates the impression that the authors do not know them.

-The whole paper suffers from the fact that UR is a catch-all term for all kind of retail, from a small 20 square meter convenience store to a mega market with space of a few thousand square meters. Obviouly, it makes a difference if a small convenience store open in the neighborhood or a mega store. And then the authors measures UR by relative stores, where they do not define what a “relative store” is. Also the term “relevant store” is confusing, because what is relevant?

-The authors do not provide a straight proof for the inverted u-shape relationship between UR and housing prices.

Author Response

Point 1: The main problem is the paper is not well structured, it contains a huge amount of flaws and it is a huge challenge for the reader to read the paper, because line of arguments is not identifiable. The introduction is a disaster, because the authors do not provide a good reasoning why their research is necessary. For me it remains unclear for what purpose this research is good. The problem is, not one of the quantitative results can be used for any purpose, because the preferences of the citizens are changing in time. If I read the conclusion I come to result that not one outcome of the paper is in some sense qualitatively novel. All outcomes are as expected by the reader. The authors themselves do not provide one idea how their outcomes can be used to improve the situation of citizens. Additionally, the authors do not explain, why they only used second-hand housing transactions. Then the authors exclude villas and townhouses to avoid a bias, but do not argue which bias they expect.

Response 1: Many thanks for the valuable suggestion.

(1) To make the introduction is clearer, we reshape the introduction of our manuscript, and attempted to show the necessity of study by showing the obvious impact of UR on residents’ life quality and welfare, and briefly introducing the gap of existing studies. We hope these revisions can make readers easily understand our study.

Lines 27-81:

Overall Introduction section.

(2) To improve the values, based on the results of this study, we put some policy implications for the city plan department and real estate developer to find the opportunity to improve the existed UR layout.

Lines 427-446:

This study can provide several policy implications for the real-estate developers and the city planning department of the government. First, considering the negative impact of UR on residents’ life, the city planning department should prevent excessive urban retail development in developed areas, which, with a high UR intensity or older houses, focus on regional security and environmental management to offset the negative impacts of UR. Meanwhile, actively developing the UR in boundary areas that have lower UR intensity can improve the life quality of local residents and relieve the pressure on existing commercial areas. Second, real-estate developers should fully focus on the consumers’ characteristics and adopt different real estate development strategies. Specifically, for consumers with high incomes, real-estate developers should pay more attention to the establishment of a livable and private living environment, since these consumers are not sensitive to the convenience derived from UR. On the contrary, the convenience derived from UR and reducing the travel cost should be the focus of re-al-estate developers for consumers with low incomes as the major target consumer group. In addition, considering the direct and indirect positive impacts of suitable property management and construction technology on housing prices, real-estate developers can consider adopting better property management and construction technology on housing prices to get higher housing premiums, especially for areas with high UR intensity.

(3) Although, all outcomes are as expected by the reader. The quantitative results may provide a detailed reference for government and real estate developers. For example, the results of GWR provide a potential reference for the government to decide which area is more suitable for UR development. The results of quantile regression may help real estate developers to understand the demands of their object consumers with different incomes. Meanwhile, the results of the model with interaction items can help real estate developers decide whether to upgrade property management and construction technology by considering the premium and cost.

(4) We added the reason why only used second-hand housing transactions and exclude villas and townhouses.

Lines 175-176:

As the second-hand housing market exhibits a considerably more dispersed and large-scale housing supply compared with the new housing market [44].

Lines 180-183:

To make the data comparable, this study does not consider villas and townhouses, which command an obviously high housing price and only account for 2.3% of the total samples, and selects the multi-layer and high-rise housings as the major research object.

Point 2: The English, particularly the diction is very often flawed.

Response 2: Thank you for the reviewer’s valuable comment.

To improve the quality of language, we use the language edit services, the editing certificate is attached.

Point 3: The authors make fundamental mistakes, which create doubts if they really understand the underlying issues. For example, they write (line 41) “Residents’ willingness to pay for housing is affected by their demands for UR”. That is nonsense, because the willingness to pay is represented by the demand function and this is course Economics 101 knowledge. That means the authors write about something what they do not understand.

Response 3: Thank you for the reviewer’s valuable comment.

We deleted this mistake, and replace it by “UR to be an important determinant for residents’ expected housing prices [2-5].”

Lines 31:

have confirmed UR to be an important determinant for residents’ expected housing prices [2-5].

Point 4: Further, instead of reasoning their statements, the authors put a reference at the end of statements so that the reader has to find out if the statement makes sense or not. For example, lines 42-44; “At the macroeconomic level, the development of UR plays an important role in promoting economic development and improving the value of cities and regions [7,10]”. I am working as an economist since nearly 30 years and I have read hundred of textbooks and thousands of papers, but I have never noticed any theory in which the value of the city has an influence on macroeconomic variables. I also have read the two references, but the references do not confirm the statement of the authors.

Response 4: Thank you for the reviewer’s valuable comment.

We deleted the error cite and statement about city and region values, and only retain the other statement and ensure the validity of cites.

Lines 36-40:

The development of urban retail plays an increasingly important role in improving urban economic performance and residents’ welfare. At the urban level, UR can also drive the production activities of other sectors, improve the urban employment rate [6], promote the construction of urban support infrastructure, and optimize urban planning and layout [7, 8].

Point 5: Another problem is that I have no clue what the authors have in mind with “understanding the heterogeneity of residents’ demand”. Of course, that the demand functions of individuals are not identical is obvious, so what means the term above? Demand functions are derived from preferences and the authors want to understand the preferences? However, this paper does not explain the origins of the preferences. Anyways, economists take preferences as given Or do the authors mean the demand of heterogeneous residents? That is something total different, but that would make sense. The next problem is what do the authors have in mind with “Urban Retail” the capital letters indicate a name, but that makes no sense. I assume they have something in mind like where and how much shops and the like should be located in a city, and the term “distribution of urban retail” means how the urban retail system distributes the goods they sell, but that makes not much sense in the context of the authors.

Response 5: Thank you for the reviewer’s valuable comment.

The residents’ remand for urban retail in heterogeneous conditions (housing structure characteristics, space, and price quantile) is the major research object, not heterogeneity of residents’ demand. Therefore, we replaced heterogeneity of residents’ demand with residents’ remands for urban retail in heterogeneous conditions in overall context.

Point 6: I have no clue, why the authors cite Elkington, (1998),"ACCOUNTING FOR THE TRIPLE BOTTOM LINE”. The reference does not fit in the context, because Elkington writes about accounting in companies. I would argue the authors’ definition of sustainable development is misleading. I also do not see what sustainability has to do with the research question. In the context of the paper the word “sustainable” is only used as buzz word like policy-maker abuse this term.

Response 6: Thank you for the reviewer’s valuable comment.

In the original manuscript, we used sustainable development to present the viewpoint that only focusing on the economic benefits brought by UR is improper, the negative impacts of UR on the environment and society (e.g noise, waste pollution, traffic congestion, insecurity by strangers entry and so on) must need to be focused. And the residents’ demand and capitalization of UR on housing prices can reflect the degree of negative impacts. In now manuscript, to avoid the reader's confusion, we delete this statement.

Point 7: The statement, “Currently, with the growth of urbanization, urban problems such as land shortage, traffic congestion, and environmental pollution have emerged [2].” What is the growth of urbanization? Urbanization cannot grow. Problems like shortage of land, traffic congestion, and environmental pollution have currently emerged? I fear the authors have no knowledge of geography and history of cities. Since ancient times space is scarce in cities. For what is the reference (2) necessary? It is an abundant reference. Statements, which are common knowledge, do not require a reference. Nobody doubts that traffic jams occur in all cities around the planet.

Response 7: Thank you for the reviewer’s valuable comment.

The new introduction has deleted this statement.

Point 8: Or a statement like “Bayer [46] has provided empirical evidence on families from different social classes and indicated that the marginal willingness to pay increases with income. This effect cannot be observed directly using traditional HPM [47].” I do not think that any HPM is necessary to come to the conclusion that the willingness to pay for a normal good increases with income, that is common knowledge and this common knowledge can be found in any microeconomic textbook.

Response 8: Thank you for the reviewer’s valuable comment.

“Willingness to pay for a normal good increases with income” is a common knowledge. However, we want to use the limitation of traditional HPM to show why existing studies select quantile regression method to explore the relationship between willingness to pay and income.

Point 9: In the same line are the research questions. Either they are trivial, like, “Is there heterogeneity in the demand for UR between the owners of high- and low-price houses? Or they will not be answered like “If so, why and how large is the gap?” or “Which locations are better able to meet the residents’ demands for UR and increase housing prices by improving UR?”

Response 9: Thank you for the reviewer’s valuable comment.

We replace old research questions with two research questions:

Lines 65-69:

   How are the capitalization effects of UR on housing prices and residents’ demand for UR affected by heterogeneity in housing structure characteristics, price quantile, and space?

   How to adjust the UR layout of the city based on heterogeneity in housing structure characteristics, price quantile, and space?

(2) The results of this study answer the question 1, and we added some policy implications in conclusion to answer question 2.

Lines 427-446:

This study can provide several policy implications for the real-estate developers and the city planning department of the government. First, considering the negative impact of UR on residents’ life, the city planning department should prevent excessive urban retail development in developed areas, which, with a high UR intensity or older houses, focus on regional security and environmental management to offset the negative impacts of UR. Meanwhile, actively developing the UR in boundary areas that have lower UR intensity can improve the life quality of local residents and relieve the pressure on existing commercial areas. Second, real-estate developers should fully focus on the consumers’ characteristics and adopt different real estate development strategies. Specifically, for consumers with high incomes, real-estate developers should pay more attention to the establishment of a livable and private living environment, since these consumers are not sensitive to the convenience derived from UR. On the contrary, the convenience derived from UR and reducing the travel cost should be the focus of re-al-estate developers for consumers with low incomes as the major target consumer group. In addition, considering the direct and indirect positive impacts of suitable property management and construction technology on housing prices, real-estate developers can consider adopting better property management and construction technology on housing prices to get higher housing premiums, especially for areas with high UR intensity.

Point 10: Regarding the results, the authors do not provide an intuitive explanation or interpretation for their results.

Response 10: Thank you for the reviewer’s valuable comment.

We added some intuitive explanation or interpretation in each results parts, the added context is followed:

Lines 294-307:

The regression coefficient of UR is -0.051 at the 1% significance level, indicating that a 1% increase in UR is associated with a 0.051% decrease in housing prices. This result indicates that residents were more sensitive to the negative influences of UR compared to the convenience of UR, which makes them reject the increase in UR density.

For the results of Model (2), the regression coefficient of UR is similar to that of Model (1). The regression coefficient of the interaction item of UR and property management is -0.053 at the 10% significance level (P-value =0.092) and the value is 0.020.  This indicates that property management has a positive moderating effect on the capitalization effect of UR on housing prices. Generally, the negative capitalization effect of UR on housing prices decreases as the quality of property management increases. For the interaction item of UR and building age, the coefficient is -0.003 at the 10% significance level, indicating that the negative capitalization effect of UR on housing prices decreases with improvements in construction technologies (as the building age decreases).

Lines 316-320:

Meanwhile, the coefficient of PM is 0.083 at the 1% significance level (the results of Model 1 and Model 2 are approximately the same), which means that property management and building have a direct capitalization effect on housing prices. Further, the coefficient of Year is -0.014 at the 1% significance level, indicating that newer housing with better construction technologies is preferred by home-buyers.

Lines 331-334:

The pseudo R2 of all the quantiles is between 0.326 and 0.368, indicating that all models have adequate explanatory power. All the regression coefficients of LN(UR) are significant at the 1% level, indicating that UR has a capitalization effect on all levels of housing prices.

Lines 372-375:

The regression coefficients of LN(UR) show both positive and negative values, indicating that residents’ demands for UR are significantly different in different regions, and represent demand and rejection simultaneously.

Lines 400-404:

To directly confirm this deduction, we established a new hedonic price model, which introduced the square of LN(UR). The result was shown in Table A1. From Table A1, we can see that the regression coefficients of the square of LN(UR) were -0.015 at the 5% significance level, confirming the inverted U-shaped relationship between LN(UR) and housing prices

Point 11: Regarding the variables sometimes the reader has no idea, what is meant. For example “property management” is either “need improve, low, mid or high”, but what does that mean precisely? The same with the variable “study area”. And then the authors get outcomes like “…which means that PM and YEAR have a direct capitalization effect on housing prices and that a 1% increase in PM and 1% decrease in YEAR will both achieve a 1% increase in housing price.” How can PM increase by 1%, if PM can only take 4 ordinal values? Such statements are nonsense. Such flaws raise the concern that the authors do not understand what they are doing.

Response 11: Thank you for the reviewer’s valuable comment.

(1) the assessments of property management and school district come from one of the largest real estate agency websites (Fangtianxia), a higher number means higher quality in property management and school district. And we added this note in lines 222-224

lines 224-226:

Property management (PM) and study area (SA) are the scores assessed by the real estate agency website (Fangtianxia), with four levels and three levels, respectively. A higher number indicates a better quality of property management and study area.

(2) we revised the error statement (a 1% increase in PM and 1% decrease in YEAR will both achieve a 1% increase in housing price) and directly explained our result

Lines 316-320：

Meanwhile, the coefficient of PM is 0.083 at the 1% significance level (the results of Model 1 and Model 2 are approximately the same), which means that property management and building have a direct capitalization effect on housing prices. Further, the coefficient of Year is -0.014 at the 1% significance level, indicating that newer housing with better construction technologies is preferred by home-buyers.

Point 12: It is absolutely annoying that the authors use always abbreviations in their statements because the reader cannot remember all 20 or more abbreviations while reading the text.

Response 12: Thank you for the reviewer’s valuable comment.

For the convenience of the readers, we try our best to reduce the abbreviations, and only remain the abbreviation of urban retail since urban retail is the major research object and is mentioned many times.

Point 13: Furthermore, the authors applied different empirical methods, but they do not discuss in detail the advantages and disadvantages of each method. That creates the impression that the authors do not know them.

Response 13: Thank you for the reviewer’s valuable comment.

In the method part, we added the advantages and disadvantages of each method to show the reason why this method is appropriate for the study target.

Lines 254-258:

Compared with the hedonic price model, quantile regression has some advantages: 1) it does not require strong assumptions for the error terms and the estimation results are robust to outliers; and 2) it describes the whole conditional distribution of explained variables more comprehensively [36].

Lines 263-274:

An urban housing market, which usually comprises various submarkets, is too complex to be described as a spatially homogeneous unit [13, 49]. Tobler's First Law of Geography indicates there are more similarities between adjacent geographical entities. Due to the uneven distribution of urban retail resources and other housing characteristics, there may be spatial heterogeneity in the resident’s demands for UR. The global regression of the hedonic pricing model is not detailed enough to explain the local conditions. The geographically weighted regression model uses the local smooth pro-cessing method to solve the problem of spatial heterogeneity. Considering spatial heterogeneity, geographic coordinates and core functions are utilized to carry out local regression estimation on the adjacent individuals of each group. Therefore, this study tests the spatial heterogeneity of the capitalization effects of UR on housing prices based on the result of the geographically weighted regression model, as follows

Point 14: The whole paper suffers from the fact that UR is a catch-all term for all kinds of retail, from a small 20 square meter convenience store to a mega-market with space of a few thousand square meters. Obviouly, it makes a difference if a small convenience store open in the neighborhood or a mega store. And then the authors measures UR by relative stores, where they do not define what a “relative store” is. Also the term “relevant store” is confusing, because what is relevant?

Response 14: Thank you for the reviewer’s valuable comment.

We added the interpretation about urban retail stores. In this study, we select eight urban retail stores including catering stores, convenience stores, entertainment stores, life services stories, sport stories, clothing stores, cosmetics stores and other stores. This study selected the number of these stores within 500 meters of the house as UR variable. In addition, mega markets (e.g shopping malls) are broken up into multiple independent urban retail stores to separately count.

Lines 199-204

Urban retail refers to all consumer-related activities. Based on the classification rules of the Gaode Map, this study selected eight urban retail store categories, including catering stores, convenience stores, entertainment stores, life services stories, sport stories, clothing stores and cosmetics stores, and other stores (Figure 2a) and uses a number of these stores within 500 meters of the house. Meanwhile, mega markets (e.g. shopping malls) are broken up into multiple independent urban retail stores, to be counted separately.

Point 15: The authors do not provide a straight proof for the inverted u-shape relationship between UR and housing prices.

Response 15: Thank you for the reviewer’s valuable comment.

We added a model with the square of UR in Appendix to directly confirm the inverted u-shape relationship.

Lines 400-404

To directly confirm this deduction, we established a new hedonic price model, which introduced the square of LN(UR). The result was shown in Table A1. From Table A1, we can see that the regression coefficient of the square of LN(UR) was -0.015 at the 5% significance level, confirming the inverted U-shaped relationship between LN(UR) and housing prices.

Table A1

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors use multiple models to explain Heterogeneity of Residents’ Demands for Urban Retail. I think this is a good practice and can give the housing price researchers more inspiration. But I still have some small confusions in the process of reading:

1. Regarding the selection of variables, I would like to see a clearer description. In addition, the collinearity between these variables has not been seen in the analysis before model regressions. In the three models, especially the GWR model, if there are collinearity or insignificant variables (such as the distance to the green space), they should be excluded from the model, otherwise it may cause deviations.
2. In the comparison of Figure 3, I think it is more accurate to use similar rather than the same spatial distribution characteristics. The two figures can clearly see greater differences.
3. Too much to explain the impact of variables other than UR on housing prices. This is not very relevant to the topic, so it is recommended to streamline it.

Author Response

Point 1: Regarding the selection of variables, I would like to see a clearer description. In addition, the collinearity between these variables has not been seen in the analysis before model regressions. In the three models, especially the GWR model, if there are collinearity or insignificant variables (such as the distance to the green space), they should be excluded from the model, otherwise it may cause deviations.

Response 1: Many thanks for the valuable suggestion.

We added the collinearity in HPM model and establish new GWR model without insignificant variables. You can see it in table 2 (line289) and table 4 (line 384), respectively.

Point 2: In the comparison of Figure 3, I think it is more accurate to use similar rather than the same spatial distribution characteristics. The two figures can clearly see greater differences.

Response 2: Many thanks for the valuable suggestion.

We have replaced same by similar.

Lines 391-392:

We find that the LN(UR) coefficients have a similar spatial distribution as LN(UR).

Point 3: Too much to explain the impact of variables other than UR on housing prices. This is not very relevant to the topic, so it is recommended to streamline it.

Response 3: Many thanks for the valuable suggestion.

We have deleted the explanation on the impact of other variables.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

In general, the paper has improved regarding its readability and now I understand the paper better than the former version.

-However, I have some doubt if the used research methodology to treat UR as continuous variable does really make sense. The point is, if one more shop opens it a discontinuous or discrete change, and that may induce drastic changes in the neighborhood of the shop. Thus, to consider incremental changes is in my view not really convincing, I think it is unavoidable to consider discrete changes, otherwise no policy implications can be derived. To make that clear, in a neighborhood with on shop, and additional shop means an increase of 100% UR, but in a neighborhood with thousand shops an additional shop means an increase of 0.1% UR. According to the authors, in the former case the value of apartments will decline by 5% and in the latter case a decline by 0.005%. In latter case, I would agree with the result, but in the former case I have strong doubt regarding the outcome. From the view of planning the planner has to know how much UR measured in number of shops is optimal.

-Another methodological problem is, that the authors do not differentiate between internal and external variables, e.g. the property management is internal variable, which is determined solely by property owners. In contrast, the effects caused by UR are spillovers or externalities which cannot be influenced by the property owners.

-In general, it is questionable to do empirical research without having a coherent and consistent theory. And because of the missing theory, the explanations of the results are vague if not to say sometimes arbitrarily.

-A main problem is still, that I cannot detect any real novelty in the paper, all results coincide with what the reader expects, nothing is really new. Therefore, to read the paper is to some extent simply boring.

-The policy proposals are so general, that they also can be recommended without the empirical investigation.

-The authors argue now “Property management (PM) and school district (SD) are the scores assessed by real estate agency website (Fangtianxia), with four levels and three levels, respectively. A higher number indicates a better quality of property management and school district.” The problem here is, that the authors use scores as variables without knowing how the scores are determined. Maybe, the scores can be scientifically justified, but it is also be possible that the scores do make sense from a scientific point of view. It would be good idea to ask the company how they determine the scores, because to use data without knowing how the data is generated does not make sense. I remember that I have also asked few times people of the World Bank how some of their indicators were generated in detail. That may cost some time, but it is the only way to do reliable research.

-Still the authors use the term “relative store” in table 1. What is a “relative store”? I have no clue, and what is then an “absolute store”?

-In table 2, “SE” and “VIF” are not defined. 

-The term “construction technology” is the wrongly used, what the authors mean is noise mitigation measures and noise prevention for example by using Thermopane windows, or thicker walls. And to assume that newer buildings protect better against external and internal noise in an apartment does not coincide with the assumption of the authors. I have lived in apartments which were older than 150 years and they prevented much more noise than apartments build in the early 2000s. In so far, the assumption must be validated by a reference, which has investigated the relationship between noise prevention and age of buildings, otherwise it is simply a hypothesis.

Author Response

Point 1: However, I have some doubt if the used research methodology to treat UR as continuous variable does really make sense. The point is, if one more shop opens it a discontinuous or discrete change, and that may induce drastic changes in the neighborhood of the shop. Thus, to consider incremental changes is in my view not really convincing, I think it is unavoidable to consider discrete changes, otherwise no policy implications can be derived. To make that clear, in a neighborhood with on shop, and additional shop means an increase of 100% UR, but in a neighborhood with thousand shops an additional shop means an increase of 0.1% UR. According to the authors, in the former case the value of apartments will decline by 5% and in the latter case a decline by 0.005%. In latter case, I would agree with the result, but in the former case I have strong doubt regarding the outcome. From the view of planning the planner has to know how much UR measured in number of shops is optimal.

Response 1: Many thanks for the valuable suggestion.

UR ranges from 20 to 1400, therefore we regard UR as a continuous variable.

“in a neighborhood with thousand shops, an additional shop means an increase of 0.1% UR. According to the authors, in the former case the value of apartments will decline by 5% and in the latter case a decline by 0.005%” is the result of global regression. In other words, from a global view, the increase of UR will lead to a decrease in housing prices. The global regression ignores the difference among different sub-markets. Therefore, we very agree with the point of the reviewer “From the view of planning the planner has to know how much UR measured in the number of shops is optimal”, and try to explore the residents’ demand for UR in heterogeneous space, housing structure characteristics and price quantile. For example, local regression of geographically weighted regression can directly be told planer where is better for further UR development. Furthermore, according to matching the other characteristics of the house, we can find a limitation of UR development of the house with similar characteristics. Meanwhile, the inverted U curve represents the capitalization of UR on housing price in different UR degrees, which can help the planer to understand what extent is UR development appropriate.

Point 2: Another methodological problem is, that the authors do not differentiate between internal and external variables, e.g. the property management is internal variable, which is determined solely by property owners. In contrast, the effects caused by UR are spillovers or externalities which cannot be influenced by the property owners.

Response 2: Many thanks for the valuable suggestion.

Referring to many previous studies, this study divided all independent variables into three categories: structural variables, location variables, and environment variables. Structural variables usually are decided by real estate developers. Meanwhile, in China, the property management company usually is selected by real estate developers. Location variables and environment variables usually are decided by the external environment and cannot be influenced by the property owners.

Point 3: In general, it is questionable to do empirical research without having a coherent and consistent theory. And because of the missing theory, the explanations of the results are vague if not to say sometimes arbitrarily.

Response 3: Many thanks for the valuable suggestion.

Based on the theory of Demand and Supply, we added some explanation.

Lines 366-369

The above reasons made the demand curve of high-price house owners more to the left than the demand curve of low-price house owners. In other words, high-price house owners are willing to pay a lower premium for UR.

Lines 317-319; 325-328

The negative impact of UR on residents’ life quality usually decreases residents' preference for UR, which leads the demand curve to move to the left and the premium of UR decrease.

Therefore, compared with the owners who have housing with bad performance of building sound insulation, the demand curve of owners having houses with a great performance of building sound insulation is more to right. In other words, they are willing to pay a higher premium for UR.

Point 4: A main problem is still, that I cannot detect any real novelty in the paper, all results coincide with what the reader expects, nothing is really new. Therefore, to read the paper is to some extent simply boring

Response 4: Many thanks for the valuable suggestion.

We admit this study does not produce anti-common-sense results and all results coincide with what the reader expects. This study still has its academic value and practical significance. For academic value, firstly, because of the double impact of UR, this study assumed and confirmed an inverted U relationship between UR and housing price. Secondly, this study explores the moderating effect of building age and property management on the relationship between UR and housing prices. Few previous studies explore the relationship between house characteristics and housing prices from the above view. And above view may provide some reference for similar research, for example, the capitalization of transportation infrastructure on housing price. For practice, the results of geographically weighted regression can provide the planer a direct reference to where is appropriate to do UR development. And the result of the inverted U relationship benefits knowing what extent is UR development appropriate.

We added these academic and practical values in lines

Lines: 444-473

In contrast to previous studies, this study is the first to discuss the inverted U-shaped relationship between housing prices and UR, and analysis the moderating effect of housing structure characteristics on the resident’s demand for UR. These re-search perspectives can provide some reference for other study on housing prices, such as the relationship between transportation infrastructure and housing prices. For prac-tice, this study also provides several policy implications for the real-estate developers and the city planning department of the government. First, considering the negative impact of UR on residents’ life, the city planning department should prevent excessive urban retail development in developed areas, which, with a high UR intensity or older houses, focuses on regional security and environmental management to offset the negative impacts of UR. Specifically, the results of geographically weighted regression indicate Wenjian and Xindu is more appropriate to do UR development. On contrary, Wuhou, Jingjiang and Jinniu should try to decrease the density. Furthermore, because of the local regression of geographically weighted regression, according to refer the regression coefficients of houses with similar environment and location variables, planer can find the UR development limitation of specific region. The inverted U-shaped curve indicates the UR intensity of dwelling district shouldn’t more than 125/km2 (exp(3.441)/0.52). Meanwhile, actively developing the UR in boundary areas that have lower UR intensity can improve the life quality of local residents and relieve the pressure on existing commercial areas. Second, real-estate developers should fully focus on the consumers’ characteristics and adopt different real estate development strategies. Specifically, for consumers with high incomes, real-estate developers should pay more attention to the establishment of a livable and private living environment, since these consumers are not sensitive to the convenience derived from UR. On the contrary, the convenience derived from UR and reducing the travel cost should be the focus of real-estate developers for consumers with low incomes as the major target consumer group. In addition, considering the direct and indirect positive impacts of suitable property management and performance of building sound insulation on housing prices, real-estate developers can consider adopting better property manage-ment and performance of building sound insulation on housing prices to get higher housing premiums, especially for areas with high UR intensity.

Point 5: The policy proposals are so general, that they also can be recommended without the empirical investigation.

Response 5: Many thanks for the valuable suggestion.

We added some specific policy proposals

Lines 454-461

Specifically, the results of geographically weighted regression indicate Wenjian and Xindu is more appropriate to do UR development. On contrary, Wuhou, Jingjiang and Jinniu should try to decrease the density. Furthermore, because of the local regression of geographically weighted regression, according to refer the regression coefficients of houses with similar environment and location variables, planer can find the UR de-velopment limitation of specific region. The inverted U-shaped curve indicates the UR intensity of dwelling district shouldn’t more than 125/km2 (exp(3.441)/0.5²).

Point 6: The authors argue now “Property management (PM) and school district (SD) are the scores assessed by real estate agency website (Fangtianxia), with four levels and three levels, respectively. A higher number indicates a better quality of property management and school district.” The problem here is, that the authors use scores as variables without knowing how the scores are determined. Maybe, the scores can be scientifically justified, but it is also be possible that the scores do make sense from a scientific point of view. It would be good idea to ask the company how they determine the scores, because to use data without knowing how the data is generated does not make sense. I remember that I have also asked few times people of the World Bank how some of their indicators were generated in detail. That may cost some time, but it is the only way to do reliable research.

Response 6: Many thanks for the valuable suggestion.

We have consulted Fangtianxia about their assessment method for Property management (PM) and school district (SD) and added the statement in Lines 223-228

Lines 223-228

The assessment of property management follows the standard of property management service grade of the residential building. which classifies property management into four grades through the degree of building management, maintenance of shared facilities and equipment, maintenance of public order, and cleaning service. The assessment of the school district is decided by the rank of the supporting primary school and junior middle school of the house.

Point 7: Still the authors use the term “relative store” in table 1. What is a “relative store”? I have no clue, and what is then an “absolute store”?

Response 7: Many thanks for the valuable suggestion.

We are very sorry to make this spelling mistake, and replace relative store with relevant store. The definition of relevant store can be seen in line

Lines 198-204:

Urban retail refers to all consumer-related activities. Based on the classification rules of the Gaode Map, this study selected eight urban retail store categories, including catering stores, convenience stores, entertainment stores, life services stories, sport stories, clothing stores and cosmetics stores, and other stores (Figure 2a) and uses a number of these stores within 500 meters of the house. Meanwhile, mega markets (e.g. shopping malls) are broken up into multiple independent urban retail stores, to be counted separately.

Point 9: In table 2, “SE” and “VIF” are not defined.

Response 8: Many thanks for the valuable suggestion.

We have added the definition of “SE” and “VIF” in the note of table 2.

Lines: 297-299

SE represents the standard error of the regression coefficient. VIF represents the variance inflation factor, which is used to quantify the degree of multicollinearity.

Point 10: The term “construction technology” is the wrongly used, what the authors mean is noise mitigation measures and noise prevention for example by using Thermopane windows, or thicker walls. And to assume that newer buildings protect better against external and internal noise in an apartment does not coincide with the assumption of the authors. I have lived in apartments which were older than 150 years and they prevented much more noise than apartments build in the early 2000s. In so far, the assumption must be validated by a reference, which has investigated the relationship between noise prevention and age of buildings, otherwise it is simply a hypothesis.

Response 10: Many thanks for the valuable suggestion.

We have replaced construction technology with performance of building sound insulation. And set assumptions according to state the continuously raising sound insulation design standard of Chinese urban residential buildings.

Lines 248-253

In Chinese, the design of urban residential buildings must follow the standards of sound insulation. These standards of sound insulation have been revised many times to improve sound insulation performance. For example, non-standard before 2000, Code for sound insulation design of residential buildings in 2010 and 2020. Therefore, we assume building age has a positive relationship with building performance of sound insulation and use YEAR to characterize building performance of sound insulation.

Author Response File: Author Response.pdf